A Variant Mode of Mammalian Olfaction

Permanent linkhttp://nrs.harvard.edu/urn-3:HUL.InstRepos:37944989

Terms of UseThis article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA

Share Your StoryThe Harvard community has made this article openly available.Please share how this access benefits you. Submit a story .

Accessibility

A Variant Mode of Mammalian Olfaction

A dissertation presented

by

Daniel Marcus Bear

to

The Division of Medical Sciences

in partial fulfillment of the requirements

for the degree of

Doctor of Philosophy

in the subject of

Neurobiology

Harvard University

Cambridge, Massachusetts

November 2016

© 2016 Daniel Marcus Bear All rights reserved.

iii

Dissertation Advisor: Dr. Sandeep R. Datta Daniel Marcus Bear

A Variant Mode of Mammalian Olfaction

Abstract

Olfaction – the sense of smell – informs animals about food, mates, threats, and

other chemical signals. The odors most relevant to survival and reproduction vary

across species and ecology; for this reason, olfactory nervous systems have evolved to

perceive select stimuli and to elicit adaptive responses. The mammalian olfactory

system accomplishes this with families of olfactory receptors – proteins that bind

characteristic odor molecules and signal through parallel neural pathways. Each animal

expresses a diverse repertoire of receptors, geared to its chemical environment, in a

canonical mode of one receptor gene per olfactory sensory neuron. This arrangement

produces precise neural representations for individual odors, which are capable of

driving a range of behaviors. However, it is not known whether this conserved

architecture is suited for all aspects of odor sensation. Here I report that part of the

olfactory system breaks from this predominant organization. An atypical sensory

subsystem, the so-called olfactory “necklace,” does not express traditional olfactory

receptors, but instead detects several classes of chemical stimuli with the previously

unknown MS4A receptor family. MS4A proteins are unrelated to other olfactory

receptors, suggesting that they have evolved to detect distinct odors and to transduce

signals by a different mechanism. Moreover, multiple members of the Ms4a family are

iv

expressed in each necklace neuron. The violation of the “one receptor per neuron” rule

endows this subsystem with broader sensory responses than granted by a single

receptor. By way of this variant organization, the necklace may play a unique role in

shaping odor perception.

v  

TABLE OF CONTENTS

Abstract iii

Acknowledgements vi

Chapter 1. General Introduction: Selective Sensation 1

References 5

Chapter 2. The Evolving Architecture of Vertebrate Olfaction 7

Summary 9

References 44

Chapter 3. A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction

55

Summary 57

Introduction 58

Results 62

Discussion 106

Experimental Procedures 113

References 149

Chapter 4. General Discussion and Conclusion 157

General Discussion 158

Conclusion 172

References 175

vi  

Acknowledgments

The work presented here is the result of extraordinary mentorship and friendship.

It traces back to a sequence of extreme luck during my first year of college: I was forced

to find a research lab for summer work because I had missed the deadline for a fancy

trip. I don’t know where I would be now, a decade later, if I hadn’t first written Mike

Greenberg, or if Mike hadn’t suggested I work with Steve Flavell (then a graduate

student in his lab). My time working with Steve convinced me to become a scientist; and

it also introduced to two more of Mike’s former students, Paul Greer and Bob Datta,

whose roles in my life should be obvious to anyone who has spoken to me since then,

or to anyone who reads this work.

Mike and Steve are among the most rigorous, insightful, and imaginative

scientists I’ve known. However impressive their professional qualities were to a 19-year-

old, though, their strongest influences on me have come through their patience and

guru-like guidance. In ten years, neither has ever offered advice or answered my

millions of questions with anything less than the greatest care and sincerity. I did not

realize, in the beginning, how rare these traits are in professional scientists, in one of

the most competitive environments in the world no less. I still have plenty to learn from

their mentorship. Steve and Mike are something much more impressive than first-rate

scientists: they are wonderful people, and I hope lifelong friends.

Paul is a brilliant scientist and, to me, a unique combination of colleague, mentor,

and friend. I would not be able to express my joy and gratitude for our relationship

without doubling the length of this dissertation. Like Mike and Steve, he has been

endlessly patient and kind over the years in giving me advice about my work, my career,

  vii  

and most importantly my life – which as for many scientists is all too inseparable from

work and career. But Paul has been consistent in his conviction that the relationships

one has with other people should always be the highest priority. He has said it and he

has lived it. I expect that after another decade of his friendship and wisdom, I will be

even more thankful.

Sandeep “Bob” Datta, my Ph.D. advisor, expresses more than anyone I’ve met

the excitement and wonder of being a researcher. He is open to new scientific ideas

and to the world. At the same time he demands only the highest quality work from

himself and from others. During the first few years of graduate school, when Bob, Paul,

and I were trying to start a research program from scratch, Bob’s uncompromising

nature seemed sometimes like more of a hindrance than an advantage. My feelings

have changed. I have a better sense, now, of how difficult research is under even the

best conditions; I have never met a graduate student, postdoctoral fellow, or young

investigator who did not at some point buckle under the stresses of science. Many days

begin with the thought of how long it’s been since you found something new and worth

your extreme effort to pursue; others end with the worry that you’ve missed something

simple, which would negate your prior work. Bob has always confronted these anxieties

head-on. I am impressed and proud that as his lab has matured, he has learned to allow

others to deal with the difficulties of research in their own lives and in their own ways,

while never wavering in his belief that we are all enormously privileged to be scientists.

He has passed this mindset on to me and my fellow students – Stan Pashkovski, Tari

Tan, Alex Wiltschko, and Maria Bloom – who, through our growing together, have

become permanent mentors and friends.

  viii  

I could not have survived graduate school without the help of other members of

the Datta Lab and the Department of Neurobiology. This is true of all graduate students

– science is not a solo endeavor – but it is especially true of me. I apologize for my

many (though hopefully all minor) lapses in responsible behavior. In particular, thank

you to the Datta Lab managers – Alexandra Nowlan, Allison Petrosino, and Neha

Bhagat – for keeping our workplace away from the edge of Chaos. Similarly, thank you

to my dissertation advisory committee – Rachel Wilson, Josh Kaplan, and David Corey

– for keeping my work away from the edge of Chaos. I would also like to acknowledge

the outstanding and sometimes thankless guidance of Karen Harmin, the administrator

for the Program in Neuroscience: if not for her patient reminders, the happy missed

deadline that landed me in the department would have been undone by many more.

Finally, I have neither the space nor the words to truly thank the people who

know me best. Fortunately I have no plans to inflict this dissertation on them. They have

absorbed my worst torrents of frustration and self-doubt with their stoic support, and

they have given me what science cannot: the reassurance that their love depends not

on what I learn or what I accomplish, but only on who I am. I haven’t always made it

easy for them, but I hope they feel my respect and my love all the same. I think they do.

Andrew, Eric, Dann, Dan, Lisa, Adam, Dad, and Mom – thank you for being there

behind all of this.

  ix  

“The Brain, within its Groove Runs evenly – and true – ”

Emily Dickinson (1863)

Chapter 1

General Introduction: Selective Sensation

  2  

General Introduction

Every animal has a unique view of the world. For each species, natural selection

has shaped the senses to detect, interpret, and remember critical features of the

physical environment that aid its survival and reproduction. The brain extracts, in

stages, progressively more complex information from sensory stimuli – about visual

edges, contours, shapes, and abstract objects, for instance (DiCarlo et al., 2012).

Neural representations magnify some facets of the external world to discern them more

carefully, while ignoring others that have less adaptive value.

The senses therefore detect and process a fraction of natural stimuli. Selective

perception originates in sensory neurons themselves. The limited scope of sensation is

easiest to illustrate in the case of vision: photoreceptor cells in the eye respond to only a

tiny segment of the electromagnetic spectrum, with photon wavelengths outside this

narrow span going undetected. Restricted interaction with the environment is a result of

both biophysical constraints – the thermal noise of retinal opsin photopigments makes

them poor sensors of infrared light – as well as the cost-to-benefit ratio of constructing a

functional sensory system (Laughlin, 2001; Luo et al., 2011). For example, many insects

use patterns of ultraviolet light for foraging and mating, but mammals do not; ultraviolet-

sensitive opsins were not advantageous enough to persist in mammalian lineages (Hunt

et al., 2001). More generally, individual opsin genes have often duplicated or acquired

mutations that shift their spectral sensitivity (Neitz et al., 1991; Yokoyama, 2002). The

resulting patterns of molecular and perceptual evolution show that many of these

changes were adaptive in the animal linages where they arose (Dominy and Lucas,

2001; Frentiu et al., 2007). Thus the genes that encode sensory receptors, molecules

  3  

that transduce physical into neural signals, are primary sites of functional evolution. By

shaping the repertoire of receptors, natural selection hones the set of stimuli that can be

detected by sensory neurons.

Sensory systems are split into parallel pathways for processing different aspects

of the same physical phenomena, such as the vertebrate retinal rod system for dark

vision existing alongside the cone system for daytime color vision. Within each

subsystem important functions are divided even further, with some circuits specialized

for identifying edges, others for movement, and so on (Gollisch and Meister, 2010). This

is another way in which the nervous system has evolved for selective signal extraction:

computations necessary for detecting certain features are hardwired into the

transduction mechanisms and neuronal architectures of each subsystem; different types

of information remain segregated as they are sent through the central nervous system

and used to influence different behaviors (Do and Yau, 2010; Kim et al., 2008; Zhang et

al., 2012).

The structures of receptors and neural pathways therefore reflect the various

ways natural selection has solved problems of extracting sensory information and

interpreting the external world. Here I examine the genetic and neuronal modes of

chemoreception that have evolved for mammalian olfaction, the sense of smell.

Olfaction detects signals in the form of volatile or labile molecules, such as the odors

emitted by food, predators, and potential mates. In Chapter 2, I describe the structure of

the major vertebrate olfactory subsystems, which are responsible for sensing a wide

variety of odors and driving a range of innate and learned behaviors; I then review the

mechanisms through which olfactory chemoreceptors and the sensory neurons that

  4  

express them have adapted to each species’ chemical environment. The main mode of

vertebrate olfaction includes multiple hardwired pathways that detect and interpret

different chemical cues; however, it develops flexibly enough that the repertoire of

olfactory receptors has shifted rapidly in response to selective pressures.

In Chapter 3, I report that one mammalian olfactory subsystem – dubbed the

olfactory “necklace” because of its necklace-shaped projections to the olfactory bulb –

represents a variant mode of chemoreception. Necklace sensory neurons employ a

distinct and previously unrecognized type of chemoreceptor protein, encoded by the

gene family Membrane-spanning, 4-pass A (Ms4a). The functional “logic” of the

necklace also breaks from the rest of olfactory system in that each necklace sensory

neuron expresses multiple types of receptor, rather than a single receptor (Axel, 1995;

Greer et al., 2016). This grants necklace cells with broader and more homogeneous

sensory tuning properties than conventional olfactory neurons. Finally, in Chapter 4, I

discuss how this atypical set of receptors and pattern of organization could be adapted

for conveying feeding-related chemical signals; this contrasts with the major olfactory

subsystems’ role in encoding precise odor identities. This work therefore opens a new

path to understanding how sensory receptors and subsystems have evolved to detect

different features of the physical world.

  5  

References Axel, R. (1995). The molecular logic of smell. Sci Am 273, 154-159.

DiCarlo, J.J., Zoccolan, D., and Rust, N.C. (2012). How does the brain solve visual object recognition? Neuron 73, 415-434.

Do, M.T., and Yau, K.W. (2010). Intrinsically photosensitive retinal ganglion cells. Physiol Rev 90, 1547-1581.

Dominy, N.J., and Lucas, P.W. (2001). Ecological importance of trichromatic vision to primates. Nature 410, 363-366.

Frentiu, F.D., Bernard, G.D., Cuevas, C.I., Sison-Mangus, M.P., Prudic, K.L., and Briscoe, A.D. (2007). Adaptive evolution of color vision as seen through the eyes of butterflies. Proc Natl Acad Sci U S A 104 Suppl 1, 8634-8640.

Gollisch, T., and Meister, M. (2010). Eye smarter than scientists believed: neural computations in circuits of the retina. Neuron 65, 150-164.

Greer, P.L., Bear, D.M., Lassance, J.M., Bloom, M.L., Tsukahara, T., Pashkovski, S.L., Masuda, F.K., Nowlan, A.C., Kirchner, R., Hoekstra, H.E., et al. (2016). A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction. Cell 165, 1734-1748.

Hunt, D.M., Wilkie, S.E., Bowmaker, J.K., and Poopalasundaram, S. (2001). Vision in the ultraviolet. Cell Mol Life Sci 58, 1583-1598.

Kim, I.J., Zhang, Y., Yamagata, M., Meister, M., and Sanes, J.R. (2008). Molecular identification of a retinal cell type that responds to upward motion. Nature 452, 478-482.

Laughlin, S.B. (2001). Energy as a constraint on the coding and processing of sensory information. Curr Opin Neurobiol 11, 475-480.

Luo, D.G., Yue, W.W., Ala-Laurila, P., and Yau, K.W. (2011). Activation of visual pigments by light and heat. Science 332, 1307-1312.

  6  

Neitz, M., Neitz, J., and Jacobs, G.H. (1991). Spectral tuning of pigments underlying red-green color vision. Science 252, 971-974.

Yokoyama, S. (2002). Molecular evolution of color vision in vertebrates. Gene 300, 69-78.

Zhang, Y., Kim, I.J., Sanes, J.R., and Meister, M. (2012). The most numerous ganglion cell type of the mouse retina is a selective feature detector. Proc Natl Acad Sci U S A 109, E2391-2398.

Chapter 2

The Evolving Architecture of Vertebrate Olfaction

8  

Attribution:

Chapter 2 has been formatted from the following published work, for which I am the

primary author:

Bear, D.M., Lassance, J.M., Hoekstra, H.E., Datta, S.R. (2016). Evolution of the genetic and neural architecture for vertebrate odor perception. Current Biology (in press). Author Contributions: DMB and SRD wrote the manuscript and made the figures, except for Figure 2.2. JML

and HEH commented on and edited the manuscript and made Figure 2.2. I would also

like to thank Taralyn Marie Tan for creating the design for Figure 2.1 and Hannah

Somhegyi for help editing the figures.

9  

Summary

Evolution sculpts the olfactory nervous system in response to the unique sensory

challenges facing each species. In vertebrates, dramatic and diverse adaptations to the

chemical environment are possible because of the hierarchical structure of the olfactory

receptor (OR) gene superfamily: expansion or contraction of the OR gene tree

accompanies major changes in habitat and lifestyle; independent selection on OR

subfamilies reflects local adaptation and conserved chemical communication; and

genetic variation in single OR genes among thousands alters odor percepts and

behaviors driven by precise chemical cues. However, this genetic flexibility contrasts

with the relatively fixed neural architecture of the vertebrate olfactory system, which

mandates that new olfactory receptors integrate into segregated and functionally distinct

neural pathways. This organization allows evolution to couple critical chemical signals

with selectively advantageous responses, but also constrains relationships between

olfactory receptors and behavior. The coevolution of the OR repertoire and the structure

of the olfactory system therefore reveals principles of how the brain solves specific

sensory problems and how it adapts to new ones.

10  

Introduction

Olfaction, the sense of smell, has evolved to detect signals from the chemical

environment, which contains clues about where to move, what to eat, when to

reproduce, and which stimuli to remember as rewarding or dangerous (Meister, 2015;

Wyatt, 2014). How an animal responds to chemical cues is in part learned and in part

innate, depending on how the olfactory nervous system has been shaped by both

experience across a lifetime and evolution across generations.

Interacting with the chemical world presents a challenge unlike other sensory

tasks. In contrast to light and sound waves, which vary continuously in wavelength and

amplitude, chemical compounds differ discretely in an enormous number of dimensions.

Compared to vision and audition, olfaction requires many more receptors – proteins

expressed in sensory organs that convert a physical event into an electrochemical

signal carried by neurons. A single photoreceptor pigment may interact with a range of

wavelengths that is sufficiently informative for an animal, as with our dark-vision

rhodopsin; however, biophysical constraints restrict chemoreceptor proteins to interact

with only subsets of chemical space. The high dimensionality of chemical space and the

diversity of the chemical stimuli an animal might encounter have therefore selected for

genomes that encode enormous receptor repertoires, containing hundreds to thousands

of olfactory receptor (OR) genes (Bargmann, 2006; Hansson and Stensmyr, 2011;

Touhara and Vosshall, 2009; Zhang and Firestein, 2002).

Here we examine how the olfactory receptor repertoire and the structure of the

olfactory nervous system evolve in concert to sense and interpret chemical information.

Adaptations in the genetic and neural architecture of olfaction reflect the unique

11  

chemosensory challenges faced along a species’ lineage — detecting novel odorants,

discerning especially critical odors with high acuity, and responding behaviorally to

molecules that acquire new meaning. We argue that the receptors underlying vertebrate

olfaction possess two properties essential for the range of adaptations seen in

vertebrate olfactory systems: a flexible and hierarchical pattern of evolution that allows

receptor adaptation to both dramatic and subtle changes in the chemical environment;

and access, through specific expression patterns, to a diverse array of neural pathways

that govern both hardwired, instinctual behaviors as well as more flexible odor learning.

In this way, OR families and subfamilies have evolved on larger scales to inform the

animal about wide swaths of chemical space, while at the same time some individual

receptors have become highly tuned to key odorants that elicit innate responses. We

speculate that the unique ability to incorporate new and evolving OR genes has

produced the substantial genetic and structural diversity of olfactory systems seen

across the vertebrate lineage (Finlay and Darlington, 1995; Meisami and Bhatnagar,

1998; Yopak et al., 2015).

The Molecular, Cellular and Neural Architecture of Vertebrate Olfaction

Smell begins when odor molecules bind to OR proteins on the endings of

sensory neurons. The set of odors an animal can detect therefore depends on the

expression pattern and the protein structure encoded by each of its OR genes. In

vertebrates, nearly all OR genes encode seven-pass transmembrane G-protein coupled

receptors (GPCRs) that, upon ligand binding, signal through G-proteins and intracellular

second messengers to ultimately open membrane ion channels; this depolarizes the

12  

sensory neuron to drive action potentials that are conducted along its axon into the

olfactory bulb of the brain (Figure 2.1) (Manzini and Korsching, 2011).

In vertebrates, most olfactory sensory neurons express a single olfactory

receptor gene from one of five major GPCR families (Dalton and Lomvardas, 2015).

The evolutionary history of each family relates roughly to the region of chemical space it

probes, as the odorant structures bound by family members are reminiscent of those

bound by the specific ancestral non-olfactory GPCR from which each family derived

(detailed below.) In the rodent olfactory system where they were first discovered,

members of each OR family are expressed in a Pointilistic pattern within spatially

restricted zones of the nasal epithelia. Most ciliated sensory neurons of the main

olfactory epithelium (MOE) express receptor genes of the largest family, the “classical”

mammalian olfactory receptors (mORs) (Buck and Axel, 1991; Ressler et al., 1993),

while a minority of neurons instead express members of the much smaller trace amino-

acid receptor (TAAR) family (Johnson et al., 2012). In the vomeronasal organ (VNO),

microvillar sensory neurons in the apical epithelial layer express Vomeronasal type 1

Receptors (V1Rs) (Dulac and Axel, 1995), but neurons in the basal layer each express

one Vomeronasal type 2 Receptor (V2R) plus a chaperone V2R in characteristic

combinations (Herrada and Dulac, 1997; Ryba and Tirindelli, 1997) (Silvotti et al., 2007).

Finally, some neurons in both layers of the VNO instead express members of the small

Formyl Peptide Receptor-related family (FPRs) (Liberles et al., 2009; Riviere et al.,

2009).

The one-receptor-per-neuron organization shapes how odor representations

propagate through the olfactory system. The axons of primary sensory neurons

13  

Figure 2.1

Organization of the mouse olfactory system (partially adapted from Dulac & Torello,

2003). Odors are blends of molecular compounds inhaled into the nasal cavity, where

they interact with olfactory receptor proteins expressed in the main olfactory epithelium

(medium gray), the vomeronasal organ (dark gray), or one of several smaller sensory

structures not pictured (upper). Each sensory neuron expresses a single olfactory

receptor denoted by its color, and neurons expressing the same receptor project to

insular structures in the olfactory bulb called glomeruli (lower). Glomeruli

corresponding to a given receptor have stereotyped spatial positions across animals.

Mitral cells in the olfactory bulb each send a dendrite into one glomerulus (main

olfactory bulb) or multiple glomeruli corresponding to the same olfactory receptor

(accessory olfactory bulb, not pictured), and project axons to a variety of central brain

regions that mediate odor learning and innate odor-driven behaviors. The targets of

main and accessory bulb mitral cell projections are largely distinct.

14  

Figure 2.1 (Continued)

15  

expressing the same receptor gene coalesce onto insular structures called glomeruli in

the outer layer of the olfactory bulb, with the positions of the glomeruli corresponding to

a given OR approximately fixed from one animal to the next (Mombaerts, 2004) (Figure

2.1). A given odorant binds to a stereotyped subset of olfactory receptors and thereby

activates a characteristic subset of glomeruli – like keys pressed together as a chord on

a piano – such that odors elicit unique spatiotemporal activity patterns across the

olfactory bulb (Lin da et al., 2006).

Odor responses in the olfactory bulb drive innate and learned behaviors via

neural projections to more central brain regions. The largest of these bulbar targets is

the piriform cortex, a three-layered paleocortical structure in which odor identity is

encoded in the patterned activity of distributed neuronal ensembles. Odor

representations are less well understood in other targets of the olfactory bulb, including

the cortical amygdala and the olfactory tubercle, a subdivision of the ventral striatum.

Mitral and tufted cells of the main olfactory bulb each send a single dendrite into one

glomerulus and extend parallel axon branches into each of these regions in

characteristic patterns, which is thought to underlie the distinct roles of each of these

areas in olfactory processing. Spatially diffuse projections to piriform cortex orchestrate

the generation of odor-specific neural ensembles that appear optimized for odor

learning. In contrast, the cortical amygdala receives stereotyped and spatially restricted

innervation from individual glomeruli and exhibits spatially biased responses to innately

relevant odors; consistent with this apparent hardwiring, the cortical amygdala mediates

instinctual responses to some predator odors (Figure 2.1) (Choi et al., 2011; Root et al.,

2014; Sosulski et al., 2011). In contrast, mitral cells in the accessory olfactory bulb each

16  

extend a dendrite into multiple glomeruli (often corresponding to the same vomeronasal

receptor) and project, by way of the medial amygdala, to hypothalamic nuclei involved in

reproductive and stress responses (Figure 2.1) (Mohedano-Moriano et al., 2007). The

targets of main and accessory bulb mitral cells are mostly distinct, and thus higher-order

olfactory brain regions have likely adapted to receive information about odors sensed by

different receptors that trigger distinct behaviors.

Olfactory brain regions, especially the olfactory bulb, vary tremendously in

relative size across vertebrate species; furthermore, in many species, the accessory

olfactory bulb and the vomeronasal organ are absent (Finlay and Darlington, 1995;

Meisami and Bhatnagar, 1998). We therefore focus primarily on the conserved genetic

architecture of vertebrate olfaction, which likely arose in the common ancestor of all ray-

fined fishes and tetrapods (lived ~460-430 MYA). The genome of the zebrafish contains

multiple members of four of the five major GPCR OR families (ORs, TAARs, V2Rs, and

several V1Rs) (Figure 2.2) (Hussain et al., 2009; Saraiva et al., 2015; Saraiva and

Korsching, 2007) and its sensory neurons generally express these receptors in the

same “one receptor, one sensory neuron” pattern as in mice (Oka et al., 2012; Sato et

al., 2007; Weth et al., 1996). Zebrafish sensory neurons project to insular glomeruli

within the olfactory bulb, where innervation from different sensory neuron types

(expressing distinct OR families) is at least partially segregated (Braubach et al., 2012;

Sato et al., 2005). Although the pattern of convergence from sensory neurons to

glomeruli has not been assessed for most species, this conservation across zebrafish

and rodents suggests that the neural organization of the olfactory system forms a stable

background on which to compare the changes in OR gene repertoire. The molecular

17  

Figure 2.2

Phylogenetic relationships of the vertebrate GPCR olfactory receptor repertoire. A

representative sample of ~2700 GPCR functional OR genes from fish (Danio rerio), frog

(Xenopus laevis), mouse (Mus musculus), and human (Homo sapiens) genomes was

used to construct a phylogenetic tree (cf. Manzini & Korsching 2011). Notable features

are visible, including large expansions of Class II ORs (red) in tetrapods, V2Rs (purple)

in amphibians, TAARs (blue) in fishes, and V1Rs (teal) in mammals; significant losses

of functional ORs in humans (blue dots) compared to mice (green dots); the monophyly

of Class I (orange) and Class II (red) tetrapod ORs; and the subfamily structure of the

“classical” ORs, with numeric labels corresponding to the subfamilies defined in Hayden

et al. (2010). A Class II OR subfamily (2,13) itself contains a smaller subfamily, the

“OR37” receptors, which has atypically expanded and been conserved at the protein

sequence level in humans. Note that all mammalian and some frog Class I ORs form a

monophyletic clade within the larger set of Class I ORs and other fish OR subfamilies

that are neither Class I nor Class II.

18  

Figure 2.2 (Continued)

19  

mechanism for choosing a single OR to be expressed in each sensory neuron – which

engages both cell type-specific transcription factors as well as stochastic DNA-DNA

interactions (Dalton and Lomvardas, 2015) – seems to have directly linked the genetic

evolution of OR families with the functional evolution of odor perception. This process

has allowed the olfactory system to flexibly incorporate or discard chemosensors, as

each OR gene stakes out its own population of sensory neurons in the olfactory

epithelium and its own glomerular territory in the olfactory bulb without disrupting the

expression or function of other receptors.

The Birth-and-Death Mechanism of Gene Family Evolution Allows Specific Adaptation to Chemical Space Singular OR expression potentially affords each newly arising OR gene a neural

pathway to inform the brain about the odorants it binds. Evolution of the OR gene

repertoire should therefore reflect the salient features of an animal’s chemical

environment. To test this hypothesis, genomic observations of OR evolution must be

combined with knowledge of what molecules individual receptors bind and how different

animals respond to these odors. Unfortunately, because of the enormous number of OR

proteins and the difficulty of measuring their molecular tuning in vitro, we know only a

small fraction of OR-odorant interactions (Saito et al., 2009).

Nevertheless, thanks to the mechanism of OR repertoire evolution, it is possible

to compare finer portions of the OR gene tree as they grow or die off on multiple

evolutionary timescales. The membership of olfactory receptor families evolves through

a characteristic birth-and-death process. Changes in the number of OR genes result

20  

from duplications of contiguous family members during unequal meiotic crossing over,

resulting in a redundancy and functional independence of the new genes that allows

them to mutate and acquire new ligand-binding properties (Nei and Rooney, 2005). A

side-effect of the relaxed selection pressure on recently duplicated genes is the

accumulation of nonsense or frameshift mutations that terminate receptor function,

producing pseudogenes embedded in rapidly duplicating OR gene clusters (Zhang and

Firestein, 2009). This birth-and-death process further explains the low proportion of

orthologous receptors found among species, as many new OR genes can arise in the

time since the divergence of even closely related lineages (Niimura et al., 2014). Other

modes of gene family evolution have occasionally homogenized the OR repertoire, such

as gene conversion between tightly-linked subfamily members and across-species

conservation of single ORs that function outside the olfactory system (Neuhaus et al.,

2009; Niimura et al., 2014; Sharon et al., 1999); however it is clear that birth-and-death

and relaxed selection are the dominant processes by which different species evolve

markedly different sets of olfactory receptors (Adipietro et al., 2012; Niimura and Nei,

2006).

The birth-and-death mechanism allows natural selection to act independently at

multiple levels of the OR gene tree, including each of the five olfactory GPCR families,

more recently diverged subfamilies within each OR family, and individual OR genes

(Figure 2.2). We argue below that progressively more subtle adaptations in the habitat,

lifestyle, and behavior of vertebrate lineages correspond to increasingly fine-scale

growth and death on branches of the OR tree. We then discuss how the olfactory

21  

nervous system might take advantage of these changes in the receptor repertoire to

generate new innate and learned olfactory behaviors.

Olfactory Overhaul: Widespread Reshaping of the OR Repertoire in New Sensory

Environments

As some vertebrates have adapted to new habitats or shifted to lifestyles that rely

less on smell, entire OR families and olfactory sensory organs have become

dispensable. For example, as its visual system has expanded dramatically, the old

world primate lineage has seen a massive loss of functional olfactory receptor genes, as

well as a near total loss of the vomeronasal receptor repertoire, accompanied by

pseudogenization of an important VNO chemotransduction channel gene TrpC2 (Figure

2.2) (Liman, 2012; Liman and Innan, 2003). Selection on primate OR genes may have

relaxed following the duplication of the opsin gene encoding the long-wavelength retinal

photopigment, as the genomes of trichromat primates – all old world monkeys and one

species of new world monkey, the howler monkey – have significantly higher ratios of

pseudogenized to intact OR genes than dichromat new world monkeys (Gilad et al.,

2004). Trichromacy may have allowed color vision to take over some olfactory tasks,

like discriminating the ripeness of leaves by their appearance along the red-green axis

(Dominy and Lucas, 2001; Lucas et al., 2003). This view is controversial, however,

because the proportion of OR pseudogenes in a mammalian genome does not correlate

with the number of intact ORs and thus poorly estimates olfactory function (Niimura et

al., 2014); moreover, loss of intact ORs in primates may have begun before the

divergence of new world monkeys and old world monkeys (Matsui et al., 2010).

22  

Nevertheless, the growth of the visual nervous system in old world primates likely

reduced reliance on olfaction for many social and foraging behaviors, which may

represent a general pattern: in multiple cases the enhancement of other senses like

vision, echolocation, and electroreception coincides with a smaller OR repertoire (Zhao

et al., 2011) (Niimura and Nei, 2007).

Changes in habitat that broadly alter chemical ecology appear to have also

resized and reshaped the OR families dramatically, though directly demonstrating

causal relationships remains a challenge. The divergence of the tetrapod lineage from

its aquatic relatives ~420 MYA was followed by an order-of-magnitude expansion in the

major vertebrate OR gene family (represented by the mORs in mammals), as hundreds

to thousands of these “classical” OR genes are found in individual amphibian and

terrestrial vertebrate genomes, whereas fish genomes have, at most, 150 (Figure 2.2)

(Niimura and Nei, 2005). The idea that having more OR genes is adaptive on land is

strengthened by converse evidence from multiple independent mammalian lineages that

returned to aquatic life. In these animals (e.g., whales, seals, and manatees), the

number of intact OR genes has dropped since they diverged from their terrestrial

relatives (e.g. cattle, dogs, and elephants) (Hayden et al., 2010).

The classical OR gene family has two major branches, the so-called Class I and

Class II ORs; however, only the Class II branch shows a massive increase in

membership as a possible result of adapting to land. This receptor subfamily is

expansive in the genomes of tetrapods (amphibians and terrestrial vertebrates) but is

essentially absent in fishes, including one more closely related to tetrapods than

teleosts – the coelacanth (Figure 2.2) (Freitag et al., 1998; Niimura, 2009). Why might a

23  

particular OR subfamily have expanded in a given vertebrate lineage? The simplest

hypothesis is that the ancestral family members were useful for sensing type of

chemical structure common or critical in that animal’s chemical environment. Having

more OR genes with somewhat diversified ligand-binding regions could have provided a

selective advantage – for instance by increasing perceptual range, sensitivity, or

discriminative power in this region of chemical space. The evolutionary history and

functional tests of the distinction between Class I and Class II ORs support this model.

First identified in the amphibian Xenopus laevis genome, Class I genes cluster

phylogenetically with fish ORs, while Class II genes cluster with the majority of

mammalian ORs (Figure 2.2) (Freitag et al., 1995). Furthermore, Class I genes are

expressed exclusively in the lateral diverticulum of the frog nasal cavity, which senses

waterborne but not airborne odorants from the outside world, while Class II genes are

expressed only in the medial diverticulum, which opens only when the frog is in air

(Freitag et al., 1995). Functional characterization supports the hypothesis that Xenopus

Class I genes are “fish-like”, as when expressed in heterologous cells they detect highly

water-soluble ligands like charged amino acids – important water-borne chemosignals

for fish. Conversely, Class II Xenopus ORs detect some airborne volatile compounds

like the aroma of coffee (Mezler et al., 2001). The ligands of mammalian Class I ORs

are also on average more polar (and thus more water-soluble) than Class II ligands

(Saito et al., 2009); but the relative size of the Class I repertoire has remained stable

across mammals, suggesting that even in purely terrestrial environments these

receptors bind odorants of ecological significance (Niimura et al., 2014).

24  

Skipping the Classics: Particular Habitats May Have Favored Expansion of Distinct OR

Families

The presence of smaller “classical” OR families in aquatic animals does not

necessarily mean that olfaction is less important in water than in air. Similar to the

above cases of Class I and Class II ORs, the aquatic environment may have instead

favored the use of other OR families because their ancestral ligand-binding properties

gave them a head-start in detecting more common aqueous odorants. For instance, the

TAAR family has expanded rapidly in teleost fishes (Figure 2.2). The zebrafish genome

encodes over 100 functional TAARs, almost an order of magnitude more than any

vertebrate outside the teleost clade. This expansion includes the derivation of an

entirely new TAAR subfamily under strong positive selection in the teleost lineage,

which was first thought not to encode amine receptors (like the other TAARs) because a

canonical amine-binding motif is lost in many subfamily members (Hussain et al., 2009).

It has now been shown, however, that these teleost TAARs initially evolved a second

amine-binding motif, changing their specificity to diamines over monoamines; later, a

subset of these secondarily lost the original motif, resulting in TAARs that use a distinct

mechanism for amine binding (Li et al., 2015). Together this pattern of TAAR duplication

and diversification suggests that the outsized importance of amine chemosensation

drove this particular OR family to prominence in the teleost lineage. This notion is

consistent with the important role known to be played by amino acids and their diamine

decarboxylation products in the chemical ecology of zebrafish (Hussain et al., 2013).

The ancestral TAAR was likely the orthologue of the mammalian TAAR1 and was

therefore probably a receptor for internally produced trace monoamines like TAAR1

25  

ligands tyramine, octopamine, and tryptamine (Liberles, 2015). Thus, the expansion of

the TAAR family in teleosts may have been biased from the start, since only a few

mutations in new TAAR genes would allow better sensing of amines, a region of

chemical space apparently important for the behavior of these fishes. For example,

cadaverine, a diamine signature of putrefaction, is detected by the zebrafish-specific

receptor TAAR13c (Hussain et al., 2013) and elicits innate aversion. This link between a

novel receptor and amine-driven behavior may be a lineage-specific adaptation, as

another teleost, the goldfish, is attracted to cadaverine (perhaps via the action of distinct

TAAR receptors) (Hussain et al., 2013; Li et al., 2015; Rolen et al., 2003).

The evolutionary trajectory of the TAARs raises the possibility that that along

different lineages, specific ancestral receptors have “gotten lucky” by binding chemical

structures similar to those with wider ecological significance. This line of reasoning

could explain why the V2R family expanded dramatically and reached its largest

membership in the amphibian lineage (Figure 2.2) (Nei et al., 2008). The V2Rs are

derived from an ancestral GPCR that senses the amino acid glutamate and have

diversified to allow broader detection of amino acids and longer polypeptides (Ryba and

Tirindelli, 1997). The amphibian V2R family is expressed in the amphibian vomeronasal

organ, which plays an essential role in sensing waterborne peptide sex signals and

coordinating complex mating behavior (Kikuyama et al., 2002; Syed et al., 2013).

Given these examples suggesting that the ecological importance of a given class

of chemicals is reflected by the size of the OR family that detects that chemical class, it

is intriguing to consider its converse: whether the apparent expansion of a specific OR

family can reveal the types of chemical cues that are most relevant to a given species.

26  

Consider the case of the semiaquatic platypus, one of only two extant members of the

monotreme lineage, which was initially thought not to rely on strong olfactory abilities

given the high proportion of pseudogenes in its classical OR family and its use of a

unique electroreceptive “bill sense” for locating prey in brackish water; however, it was

later found that the platypus genome contains more functional V1R family members

than any other vertebrate, reflecting an unparalleled expansion of this particular gene

family in the monotreme lineage (Grus et al., 2007). Indeed, the platypus possesses

one of the most complex vomeronasal organs described, suggesting that odor detection

through the “accessory” olfactory system has become its main mode of chemosensation.

Because V1Rs detect many steroid hormones that regulate reproductive and mating

behavior (Isogai et al., 2011), combing this region of chemical space may reveal cues

critical to the behavior of the platypus and ecological reasons that this particular

receptor family expanded.

The Finer Things in Life: Olfactory Specialization Through OR Subfamily Selection

OR birth and death operates locally within the genome, producing closely-related

receptor subfamilies (Figure 2.2) (Adipietro et al., 2012; Zhang and Firestein, 2002,

2009) that may evolve independently of one another. There is evidence that natural

selection can operate on the OR repertoire with this finer-scale flexibility to probe niche-

specific chemical cues with more precision. For example, Hayden et al. (2010) found

that the relative sizes of each mammalian OR subfamily were more representative of a

species’ ecotype – aquatic, semi-aquatic, terrestrial, or flying – than of its ancestry

27  

within the phylogenetic tree of mammals (Hayden et al., 2010). These results suggest

that particular receptor subfamilies are better suited to probe different environments.

Testing this hypothesis requires identifying the odors sensed by these receptors

and determining whether larger subfamilies actually confer a selective advantage in

discerning these cues. Observations of subfamily selection on shorter timescales,

correlated with even finer niche adaptation, may help hone in on the sensory demands

driving receptor evolution. For example, in an analysis of OR evolution in bats, the

relative sizes of two subfamilies correlate positively and negatively with independent

adaptations to a purely frugivorous diet (Hayden et al., 2014). A similar investigation of

bird and reptile OR repertoires identified a subfamily associated with carnivory and

several Class I and Class II subfamilies enlarged in aquatic and terrestrial feeders,

respectively (Khan et al., 2015). Members of these OR subfamilies may have become

tuned to volatile compounds found in the foods sought by individual species, and their

expression in the olfactory system may reveal how selective neural pathways adapt to

influence foraging and food selection.

Similarly, distinct subfamilies of the mouse V2R tree appear tuned to labile,

innate fear-producing chemosignals (“kairomones”) derived from different types of

mouse predators (e.g. snakes, birds, rats, and cats) (Isogai et al., 2011). The

observation that the family of Major Urinary Proteins (MUP) acts through V2Rs to

control both interspecies defensive behaviors and intraspecies aggressive behaviors

suggests that V2R-MUP pairing or binding affinities may be particular targets of lineage-

specific subfamily expansion and selection (Chamero et al., 2007; Papes et al., 2010;

Yang et al., 2005). Furthermore, if different V2R subfamilies are in fact tuned to the

28  

chemical signatures of particular predators, then their sensory signals could remain

segregated as they propagate through the nervous system – which might allow unique

defensive responses to evolve for each.

Finally, the modular nature of receptor gene evolution allows small subfamilies to

retain critical chemosensory roles, even against a backdrop of widespread receptor loss.

One peculiar mammalian subfamily, called “OR37”, is surprisingly conserved in both

mice and humans, even though humans have lost more than half of their functional OR

genes overall (Figure 2.2) (Hoppe et al., 2006; Niimura and Nei, 2007). Indeed, this

subfamily has expanded independently in the human lineage, suggesting it has not lost

its utility like other human subfamilies (Figure 2.2) (Hoppe et al., 2003). Together these

data suggest that humans may still detect a set of informative odors through the OR37

family, such as the long-chain fatty aldehyde ligands sensed by mouse OR37 members

(Bautze et al., 2012). Intriguingly, mouse mitral cell projections from OR37 glomeruli

target several hypothalamic nuclei rather than the cortical regions typical of most main

olfactory bulb projections (Bader et al., 2012; Strotmann et al., 2000). Thus the OR37

family may have evolved to link the sensation of a certain chemical class to regulation of

innate behaviors or internal state in a way that has remained crucial across species.

The overall pattern of birth-and-death OR evolution therefore demonstrates lineage-

specific adaptation to the chemical environment.

The Evolution of Single OR Genes: Direct Links Between Sensation and Action

Adaptation through change in the number of OR genes is only possible if the

olfactory nervous system has ways to employ the new receptor genes that emerge as

29  

particular subfamilies grow and diversify. Because natural odor blends activate only a

subset of receptors (Lin da et al., 2006), expressing newly duplicated and diversified

subfamily members may provide a more nuanced neural representation of the regions

of chemical space they sense. This could in turn allow finer resolution in discriminating

the makeup or concentrations of the odors sensed by the ancestral receptors.

What this model does not explain is how an animal’s interpretation of a given

odor may evolve. Many vertebrates exhibit innate patterns of behavior that can be

released by precise chemical stimuli, including even some monomolecular odorants

outside the context of their natural odor blends (Figure 2.3) (Dulac and Torello, 2003;

Hussain et al., 2013; Kikuyama et al., 2002; Lin et al., 2005). It is not obvious how

evolution could couple these blend constituents to appropriate innate behaviors via

changes in olfactory receptor number. Animals may instead interpret these

chemosignals through single, precisely- and sensitively-tuned olfactory receptors that

tap into hardwired, behavior-evoking neural circuits. This pattern of receptor tuning and

neural organization is prominent in insects, perhaps because their olfactory receptor

repertoires are much smaller (and likely less redundant) than in vertebrates. The

compounds that elicit innate responses in both insects and vertebrates are essential for

survival and reproduction – sex pheromones, attractive or aversive food odors, and

predator signatures – explaining why olfactory neurons have evolved to detect them

with high sensitivity and to couple directly to neural circuits that drive innate behaviors

(Figure 2.3) (Dekker et al., 2006; Kurtovic et al., 2007; Sakurai et al., 2015; Stensmyr et

al., 2012). For instance, the exocrine-secreted peptides ESP1 and ESP22 act

exclusively through V2Rp5 and an unknown V2R to promote female sexual receptivity

30  

Figure 2.3 A sample of monomolecular odorants that drive innate, species-specific behaviors

across vertebrates and insects. Instinctual reproductive, protective, and feeding

behaviors are released by a variety of chemical structures, including volatile airborne

small molecules in rodents (top row); waterborne acids, bases, and peptides in fishes

and amphibians (middle row); and volatile hydrocarbons in insects (bottom row).

Chemical structures (receptor responsible for behavior, if known), left to right: top row

2,4,5-trimethythiazoline, 2-sec-butyl-4,5- dihydrothiazole [SBT], 2,3-dehydro-exo-

brevicomin [DHB] (Dulac and Torello, 2003), phenethylamine (Mus musculus TAAR4)

(Dewan et al., 2013), trimethylamine (Mus musculus TAAR5) (Ferrero et al., 2011);

middle row 4-hydroxyphenylacetic acid (Danio rerio ORA1 [a V1R]) (Behrens et al.,

2014), cadaverine (Danio rerio TAAR13c) (Hussain et al., 2013), splendiferin (peptide

pheromone for Litoria splendida), sodefrin (peptide pheromone for Cynops

pyrrhogaster) (Kikuyama et al., 2002); bottom row methylhexanoate (D. sechellia

Or22a), octanoic acid (Dekker et al., 2006), cis-11-vaccenyl acetate (D. melanogaster

Or67d) (Kurtovic et al., 2007), (10E,12Z)-hexadeca-10,12-dien-1-al [upper] and

(10E,12Z)-hexadeca-10,12-dien-1-ol [lower] (aliases bombykal, receptor Bombyx mori

BmOr3, and bombykol, receptor B. mori BmOr1, respectively) (Sakurai et al., 2015).

Behavioral responses are sometimes enhanced by combinations of odorants, as with

SBT and DHB in mice and bombykal and bombykol in the silk moth. Moreover,

responses to the same compound(s) may differ drastically between species or between

sexes of the same species.

31  

Figure 2.3 (Continued)

32  

(the “lordosis” posture) and to block male sexual activity toward juveniles below

reproductive age, respectively (Ferrero et al., 2013; Haga et al., 2010). Similarly

phenethylamine found in the urine of carnivores, acts by binding to the receptor TAAR4

to promote avoidance behavior (Dewan et al., 2013; Ferrero et al., 2011). These

individual receptors therefore play non-redundant roles in shaping odor perception and

interpretation, suggesting that even in olfactory systems that use hundreds or

thousands of receptors, single members of these gene families may evolve outsized

ecological significance and privileged neural access to specific central circuits (Dewan

et al., 2013).

This model also predicts that, in animal lineages in which the role of olfaction in

driving stereotyped reproductive and defensive behaviors has waned, there will be few

cases where individual receptor genes are under strong selection pressure. This seems

to be borne out in humans, for whom there is almost no evidence for positive selection

in the coding sequences of single OR genes since the human-chimpanzee divergence

~7 MYA. It is therefore unlikely that particular OR genes have conferred a selective

advantage during human evolution (Gimelbrant et al., 2004).

That said, genetic variation in single human ORs – under selective pressure or

not – can produce large differences in odor perception. Detection thresholds and

(un)pleasantness ratings of many odors have been linked to common polymorphisms in

and around human OR genes, including a receptor coding mutation that alters

sensitivity to and valence of the male hormones androstenone and androstenedione

(Keller et al., 2007); a mutation that accounts for nearly all observed variation in the

perception of b-ionone and dramatically alters preference for foods emiting this odor

33  

(Jaeger et al., 2013); and a polymorphism closely linked to a cluster of OR genes that

tracks with the aversive “soapiness” of cilantro (Eriksson et al., 2012). Furthermore, as

in other mammals, gland secretions can in a laboratory setting can bias sexual

attraction and nipple-orienting behavior in adult and infant humans, respectively, raising

the possibility that we employ unidentified pheromones (Gelstein et al., 2011) (Varendi

and Porter, 2001). If these chemosignals follow the pattern of other social and suckling

pheromones, they are likely to include monomolecular odorants acting through high-

affinity, specialist receptors (Lin et al., 2005; Schaal et al., 2003). Even if they are not

under strong selection, then, it remains possible that single human ORs contribute

meaningfully to behavior through unknown mechanisms.

Segregated Olfactory Pathways As Substrates For Innate Olfactory Behaviors

Defined odorants can drive innate responses or preferences through a few

receptors, even against a background of hundreds or thousands – many of which may

be active simultaneously during typical sensation of complex odor blends (Rokni et al.,

2014). How, then, are particular chemical cues coupled precisely to appropriate odor-

driven behaviors? It appears that the olfactory system begins to sort odor information in

the peripheral sense organs. Several hardwired, parallel pathways – from sensory

neurons to subregions of the olfactory bulb to distinct central brain targets – arose early

in the evolution of vertebrates, using distinct receptor families and subfamilies to convey

different categories of chemical signal. Although zebrafish, unlike mice, have only a

single olfactory epithelium, the primary sensory neurons that express ORs and TAARs

are molecularly, morphologically, and anatomically distinct from those that express

34  

amino acid-sensing V2Rs; these two cell types (ciliated and microvillar sensory neurons,

respectively) are found in different layers of the olfactory epithelium and project to

separate regions of the olfactory bulb, presaging the separation of mammalian ciliated

neurons (in the MOE) and microvillar neurons (in the VNO) and their projections to

distinct main and accessory olfactory bulbs (Braubach et al., 2012; Sato et al., 2005).

The zebrafish epithelium also contains a third cell type (crypt neurons) that expresses

one of the few zebrafish V1R genes in most cells and projects to a unique glomerulus

(Ahuja et al., 2013; Oka et al., 2012; Saraiva and Korsching, 2007). In addition, a

closely related receptor has recently been found to detect a pheromone that releases

egg-laying behavior, reminiscent of the reproductive behaviors regulated by mammalian

V1Rs (Figure 2.3) (Behrens et al., 2014). This “ancestral” organization of the olfactory

periphery therefore suggests that discrete subsystems and the receptors they express

are hardwired to yield different behavioral responses to odor (Figure 2.4 top).

Consistent with this possibility, a major olfactory adaptation along the amphibian

and terrestrial branches of the vertebrate tree includes the establishment of segregated

neuroanatomical structures that express more recently-derived ORs. Unlike fish,

amphibians have evolved a separate vomeronasal organ that detects attractive peptide

sex pheromones (specific to amphibians) through unusually large and recently

expanded subfamilies of V2Rs, as well as an air-filled MOE compartment the expresses

the massively enlarged Class II OR repertoire described above; evolutionarily more

ancient V2Rs tuned to single amino acids remain expressed in the water-filled MOE

compartment along with V1Rs, Class I ORs, and TAARs (Date-Ito et al., 2008; Syed et

al., 2013; Syed et al., 2015) (Figure 2.4 middle). The observation that newer V2R genes

35  

Figure 2.4 Peripheral organization of OR family expression across vertebrates. Top: A horizontal

section of a quarter of one olfactory rosette from the zebrafish (Danio rerio). Sensory

neurons are embedded in a main olfactory epithelium (MOE) with a water-filled lumen.

Members of four of the five major GPCR OR families are expressed in different sensory

neuron types: TAARs and Class I “classical” ORs [here denoting also fish ORs that

belong to neither Class I nor Class II] are expressed in ciliated cells occupying the basal

layers of the epithelium; V2Rs in microvillar cells in the middle layer; and at least one

V1R in “crypt cells” of the upper layer. Axons of the three cell types project to different,

spatially segregated, and stereotyped glomeruli of the zebrafish olfactory bulb. Middle:

A coronal section of one half of the main olfactory epithelium and the vomeronasal

organ (VNO) of the western clawed frog (Xenopus laevis). An air-filled medial

diverticulum (MD) houses neurons expressing Class II “classical” ORs and several

V1Rs; a water-filled lateral diverticulum (LD), Class I “classical” ORs, several TAARs,

and ancestral V2Rs; the water-filled (VNO), newer members of the expanded V2R

family. Bottom: A coronal section of the MOE and VNO of the house mouse (Mus

musculus). Receptor family expression is segregated (several exceptions not pictured),

with Class I ORs and TAARs expressed in the dorsal MOE, Class II ORs in the ventral

MOE (and some in the dorsal MOE, not pictured), V1Rs in the apical VNO, V2Rs in the

basal VNO, and FPRs in both VNO layers. Some sensory neurons in the “cul-de-sacs”

of the MOE express the receptor guanylate cyclase GC-D and multiple members of a

non-GPCR family of four-pass transmembrane chemoreceptors, MS4A.

36  

Figure 2.4 (Continued)

37  

have acquired both a novel expression pattern and a novel role in innate reproductive

behaviors suggests that odor interpretation depends not only on what ligands are

detected, but also on the developmental identities and specialized sensory pathways of

the cells that detect them (Gliem et al., 2013).

The segregation of the olfactory periphery has proceeded even further in rodents.

Mice, for example, have separate epithelial layers for V1Rs and V2Rs within the VNO

(Figure 2.4 bottom); distinct dorsoventral zones and molecular machinery for the

expression of TAARs, Class I ORs, and Class II ORs in the MOE (Bozza et al., 2009); a

group of neurons at the tip of the nose, the Grueneberg ganglion, which is thought to

detect alarm pheromones and structurally related predator odorants (Brechbuhl et al.,

2008; Brechbuhl et al., 2013); a patch of olfactory neurons on the nasal septum (the

“septal organ”) that largely express a single broadly-tuned OR and may convey

information about breathing (Grosmaitre et al., 2009; Grosmaitre et al., 2007); and

within cul-de-sacs of the MOE, a set of neurons which are thought to mediate the social

acquisition of food preference and which express both the receptor guanylate cyclase

GC-D and a recently-described family of non-GPCR olfactory receptors (Figure 2.4

bottom) (Greer et al., 2016; Hu et al., 2007; Munger et al., 2010).

The olfactory bulb appears to organize this variety of incoming sensory

information into domains larger than a single glomerulus. For instance, the sensory

neurons that express class I OR-, class II OR-, TAAR-, or GC-D are developmentally

constrained to innervate circumscribed and nonintersecting regions of the olfactory bulb

(Figure 2.5) (Bozza et al., 2009) (Pacifico et al., 2012). Furthermore, there is substantial

evidence that activity in spatially and molecularly segregated glomeruli drives different

38  

Figure 2.5

Spatial and molecular organization of projection targets and behavioral responses

downstream of distinct mouse olfactory bulb glomeruli. Circumscribed zones of

glomeruli are innervated by sensory neurons expressing each of the four largest GPCR

OR families or the MS4As. In addition, Class I and Class II OR glomeruli occupy the

dorsal-most and more ventral zones, respectively; and projections from Grueneberg

Ganglion sensory neurons form their own “necklace” (light blue) anterior to the GC-

D/MS4A “necklace” (green). Dorsal Class II OR glomeruli form a molecularly defined

zone, which includes glomeruli necessary for innate aversion of some predator odorants.

Within the ventral zone of Class II OR glomeruli, some regions are enriched for the

glomeruli of TrpM5- or OR37 subfamily-expressing sensory neurons, which have

second-order projections to the “vomeronasal” amygdala and the hypothalamus,

respectively.

39  

Figure 2.5 (Continued)

40  

patterns of stereotyped behavior via hardwired projections to central brain regions. First,

an “innate aversion” domain for predator odor sensation can be found at the ventral

subdivision of the dorsal olfactory bulb, while the dorsal-most Class I-expressing

domain is instead required for aversive responses to spoiled food odorants

(Kobayakawa et al., 2007). Next, the ventral olfactory bulb houses the glomeruli of

molecularly distinct neurons (expressing components of an alternative, TrpM5-

dependent chemotransduction pathway) that preferentially respond to urine-derived

pheromones and project to the “vomeronasal amygdala”, as well as the cluster of OR37

glomeruli and their unusual mitral cell projections to the hypothalamus (Bader et al.,

2012; Lin et al., 2005; Lin et al., 2007; Thompson et al., 2012). Finally, the TAAR

domain of the olfactory bulb is required for innate behavioral responses to several

ecologically important amines, even though these same odorants can be detected by

receptors in the main olfactory system (Figure 2.5) (Dewan et al., 2013). Recent data

also suggest a global organization to the main olfactory bulb, in which glomeruli that

reside more rostrally evoke investigatory behaviors, while those that are more caudal

evoke aversion (Kermen et al., 2016). Characterizing the mitral cell projections

downstream of specific glomeruli may reveal different capacities to influence a variety of

central processing or behavioral centers. This “switchboard” organization could allow

each OR gene to adapt its chemosensory tuning according to its domain’s function

without altering the circuits downstream of the receptor.

However, the interpretation of an odor is not determined simply by whether it

recruits neural activity within a specific glomerular domain. For instance within the

TAAR domain, the TAAR4 glomerulus is required for aversion to phenethylamine, but

41  

the nearby TAAR5 glomerulus is highly tuned to the attractive mouse pheromone

trimethylamine (Dewan et al., 2013; Li et al., 2013). Thus even closely related receptors

and their closely apposed glomeruli — within a similar subdomain of the olfactory bulb –

may drive opposite behaviors. Combinatorial effects of multiple (TAAR and non-TAAR)

glomeruli may produce the different behavioral responses to high-affinity TAAR4 and

TAAR5 ligands. Consistent with this possibility, the effect of TAAR5 activation appears

to be inverted in rats, for which trimethylamine is aversive (Figure 2.3) (Li et al., 2013).

However it is also possible that, within the TAAR domain of the bulb, TAAR4 and

TAAR5 glomeruli are connected to distinct downstream circuits that drive fundamentally

different behaviors. Determining whether the differential effects of TAAR4 and TAAR5

ligands in mice relies upon glomerulus-specific projections to higher brain regions, the

coordinated activity of groups of glomeruli, or on circuit activity within the TAAR domain

will therefore reveal key principles of how odor representations in the olfactory bulb

govern perception and behavior.

The appeal of the “domain” hypothesis for coupling odors to behavior lies in its

relative developmental simplicity. In this model the olfactory system could take

advantage of developmentally-specific gradients of morphogenetic and axon guidance

molecules to build sub-circuits within the olfactory bulb that connect to distinct higher

brain regions. How might the olfactory system create a system in which individual

glomeruli within a given domain are differentially connected to higher brain regions?

One highly speculative possibility is that the olfactory bulb’s output layer of mitral cells is,

like the retinal ganglion cell layer of the retina, tiled with molecularly distinct cell types

that convey different types of information via their projection or spiking patterns.

42  

Hardwired synaptic connections to these mitral cell types might then be biased by the

complement of cell surface proteins on OSN axons, which are themselves determined

by both their developmental identity and the OR gene stochastically chosen for

expression (Serizawa et al., 2006). This sort of coupling between ORs and central

circuits could explain why many odors appear to have a stereotyped attractive or

aversive valence in mice and humans, even when the ORs responsible for their

detection are evolving neutrally (Haddad et al., 2010; Saraiva et al., 2016). In other

words, behavioral biases could be substrates for adaptation (e.g. through changes in

the expression of cell surface proteins given a particular OR choice) but would also

necessarily exist by chance. In this model, the multiglomerular domains of the olfactory

bulb may determine which downstream targets are available (e.g. medial amygdala

from the accessory bulb and cortical amygdala from the main bulb) but leave enough

flexibility for receptor-specific behaviors to evolve. The loss of several subsystems in

humans may explain why single OR genes are not under strong selective pressure in

our lineage, as a highly-tuned receptor would have no specialized circuitry to tap into to

generate an adaptive behavioral response.

Conclusion

How evolution sculpts the olfactory system is a response to both the general

problem of probing a wide swath of chemical space, as well as the unique sensory

challenges facing each animal species. In vertebrates, both dramatic and subtle

adaptations to the chemical environment are possible because of the vast, hierarchical

structure of the olfactory receptor (OR) gene families. First, rapid growth or pruning of

43  

large OR families accommodate major changes in habitat and lifestyle, as with ocean-

to-terrestrial shifts and derived primacy of the visual sense. Second, selection on

individual OR families or subfamilies permits niche adaptation or conserved chemical

communication between conspecifics. Third, genetic variation in single OR genes,

among thousands, can alter behaviors driven by precise chemical cues. We speculate

that the molecular logic and neuroanatomical architecture of the vertebrate olfactory

nervous system, in which newly arising olfactory receptor genes can tap into a number

of segregated pathways that influence both innate and learned behaviors, may have

favored this dramatic mode of olfactory receptor evolution. In this model OR genes form

the marble from which the vertebrate olfactory system is sculpted, specifying the ability

both to detect odors and to implement appropriate survival- and reproduction-promoting

responses. Peering into this dynamic process reveals a direct link between the evolution

of an animal’s genome and the specific tasks solved by its nervous system.

44  

References

Adipietro, K.A., Mainland, J.D., and Matsunami, H. (2012). Functional evolution of mammalian odorant receptors. PLoS Genet 8, e1002821.

Ahuja, G., Ivandic, I., Salturk, M., Oka, Y., Nadler, W., and Korsching, S.I. (2013). Zebrafish crypt neurons project to a single, identified mediodorsal glomerulus. Sci Rep 3, 2063.

Bader, A., Klein, B., Breer, H., and Strotmann, J. (2012). Connectivity from OR37 expressing olfactory sensory neurons to distinct cell types in the hypothalamus. Front Neural Circuits 6, 84.

Bargmann, C.I. (2006). Comparative chemosensation from receptors to ecology. Nature 444, 295-301.

Bautze, V., Bar, R., Fissler, B., Trapp, M., Schmidt, D., Beifuss, U., Bufe, B., Zufall, F., Breer, H., and Strotmann, J. (2012). Mammalian-specific OR37 receptors are differentially activated by distinct odorous fatty aldehydes. Chem Senses 37, 479-493.

Behrens, M., Frank, O., Rawel, H., Ahuja, G., Potting, C., Hofmann, T., Meyerhof, W., and Korsching, S. (2014). ORA1, a zebrafish olfactory receptor ancestral to all mammalian V1R genes, recognizes 4-hydroxyphenylacetic acid, a putative reproductive pheromone. J Biol Chem 289, 19778-19788.

Bozza, T., Vassalli, A., Fuss, S., Zhang, J.J., Weiland, B., Pacifico, R., Feinstein, P., and Mombaerts, P. (2009). Mapping of class I and class II odorant receptors to glomerular domains by two distinct types of olfactory sensory neurons in the mouse. Neuron 61, 220-233.

Braubach, O.R., Fine, A., and Croll, R.P. (2012). Distribution and functional organization of glomeruli in the olfactory bulbs of zebrafish (Danio rerio). J Comp Neurol 520, 2317-2339, Spc2311.

Brechbuhl, J., Klaey, M., and Broillet, M.C. (2008). Grueneberg ganglion cells mediate alarm pheromone detection in mice. Science 321, 1092-1095.

Brechbuhl, J., Moine, F., Klaey, M., Nenniger-Tosato, M., Hurni, N., Sporkert, F., Giroud, C., and Broillet, M.C. (2013). Mouse alarm pheromone shares structural similarity with predator scents. Proc Natl Acad Sci U S A 110, 4762-4767.

45  

Buck, L., and Axel, R. (1991). A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell 65, 175-187.

Chamero, P., Marton, T.F., Logan, D.W., Flanagan, K., Cruz, J.R., Saghatelian, A., Cravatt, B.F., and Stowers, L. (2007). Identification of protein pheromones that promote aggressive behaviour. Nature 450, 899-902.

Choi, G.B., Stettler, D.D., Kallman, B.R., Bhaskar, S.T., Fleischmann, A., and Axel, R. (2011). Driving opposing behaviors with ensembles of piriform neurons. Cell 146, 1004-1015.

Dalton, R.P., and Lomvardas, S. (2015). Chemosensory receptor specificity and regulation. Annu Rev Neurosci 38, 331-349.

Date-Ito, A., Ohara, H., Ichikawa, M., Mori, Y., and Hagino-Yamagishi, K. (2008). Xenopus V1R vomeronasal receptor family is expressed in the main olfactory system. Chem Senses 33, 339-346.

Dekker, T., Ibba, I., Siju, K.P., Stensmyr, M.C., and Hansson, B.S. (2006). Olfactory shifts parallel superspecialism for toxic fruit in Drosophila melanogaster sibling, D. sechellia. Curr Biol 16, 101-109.

Dewan, A., Pacifico, R., Zhan, R., Rinberg, D., and Bozza, T. (2013). Non-redundant coding of aversive odours in the main olfactory pathway. Nature 497, 486-489.

Dominy, N.J., and Lucas, P.W. (2001). Ecological importance of trichromatic vision to primates. Nature 410, 363-366.

Dulac, C., and Axel, R. (1995). A novel family of genes encoding putative pheromone receptors in mammals. Cell 83, 195-206.

Dulac, C., and Torello, A.T. (2003). Molecular detection of pheromone signals in mammals: from genes to behaviour. Nat Rev Neurosci 4, 551-562.

Eriksson, N., Wu, S., Do, C.B., Kiefer, A.K., Tung, J.Y., Mountain, J.L., Hinds, D.A., and Francke, U. (2012). A genetic variant near olfactory receptor genes influences cilantro preference. Flavour 1, 1.

46  

Ferrero, D.M., Lemon, J.K., Fluegge, D., Pashkovski, S.L., Korzan, W.J., Datta, S.R., Spehr, M., Fendt, M., and Liberles, S.D. (2011). Detection and avoidance of a carnivore odor by prey. Proc Natl Acad Sci U S A 108, 11235-11240.

Ferrero, D.M., Moeller, L.M., Osakada, T., Horio, N., Li, Q., Roy, D.S., Cichy, A., Spehr, M., Touhara, K., and Liberles, S.D. (2013). A juvenile mouse pheromone inhibits sexual behaviour through the vomeronasal system. Nature 502, 368-371.

Finlay, B.L., and Darlington, R.B. (1995). Linked regularities in the development and evolution of mammalian brains. Science 268, 1578-1584.

Freitag, J., Krieger, J., Strotmann, J., and Breer, H. (1995). Two classes of olfactory receptors in Xenopus laevis. Neuron 15, 1383-1392.

Freitag, J., Ludwig, G., Andreini, I., Rossler, P., and Breer, H. (1998). Olfactory receptors in aquatic and terrestrial vertebrates. J Comp Physiol A 183, 635-650.

Gelstein, S., Yeshurun, Y., Rozenkrantz, L., Shushan, S., Frumin, I., Roth, Y., and Sobel, N. (2011). Human tears contain a chemosignal. Science 331, 226-230.

Gilad, Y., Przeworski, M., and Lancet, D. (2004). Loss of olfactory receptor genes coincides with the acquisition of full trichromatic vision in primates. PLoS Biol 2, E5.

Gimelbrant, A.A., Skaletsky, H., and Chess, A. (2004). Selective pressures on the olfactory receptor repertoire since the human-chimpanzee divergence. Proc Natl Acad Sci U S A 101, 9019-9022.

Gliem, S., Syed, A.S., Sansone, A., Kludt, E., Tantalaki, E., Hassenklover, T., Korsching, S.I., and Manzini, I. (2013). Bimodal processing of olfactory information in an amphibian nose: odor responses segregate into a medial and a lateral stream. Cell Mol Life Sci 70, 1965-1984.

Greer, P.L., Bear, D.M., Lassance, J.M., Bloom, M.L., Tsukahara, T., Pashkovski, S.L., Masuda, F.K., Nowlan, A.C., Kirchner, R., Hoekstra, H.E., et al. (2016). A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction. Cell 165, 1734-1748.

47  

Grosmaitre, X., Fuss, S.H., Lee, A.C., Adipietro, K.A., Matsunami, H., Mombaerts, P., and Ma, M. (2009). SR1, a mouse odorant receptor with an unusually broad response profile. J Neurosci 29, 14545-14552.

Grosmaitre, X., Santarelli, L.C., Tan, J., Luo, M., and Ma, M. (2007). Dual functions of mammalian olfactory sensory neurons as odor detectors and mechanical sensors. Nat Neurosci 10, 348-354.

Grus, W.E., Shi, P., and Zhang, J. (2007). Largest vertebrate vomeronasal type 1 receptor gene repertoire in the semiaquatic platypus. Mol Biol Evol 24, 2153-2157.

Haddad, R., Medhanie, A., Roth, Y., Harel, D., and Sobel, N. (2010). Predicting odor pleasantness with an electronic nose. PLoS Comput Biol 6, e1000740.

Haga, S., Hattori, T., Sato, T., Sato, K., Matsuda, S., Kobayakawa, R., Sakano, H., Yoshihara, Y., Kikusui, T., and Touhara, K. (2010). The male mouse pheromone ESP1 enhances female sexual receptive behaviour through a specific vomeronasal receptor. Nature 466, 118-122.

Hansson, B.S., and Stensmyr, M.C. (2011). Evolution of insect olfaction. Neuron 72, 698-711.

Hayden, S., Bekaert, M., Crider, T.A., Mariani, S., Murphy, W.J., and Teeling, E.C. (2010). Ecological adaptation determines functional mammalian olfactory subgenomes. Genome Res 20, 1-9.

Hayden, S., Bekaert, M., Goodbla, A., Murphy, W.J., Davalos, L.M., and Teeling, E.C. (2014). A cluster of olfactory receptor genes linked to frugivory in bats. Mol Biol Evol 31, 917-927.

Herrada, G., and Dulac, C. (1997). A novel family of putative pheromone receptors in mammals with a topographically organized and sexually dimorphic distribution. Cell 90, 763-773.

Hoppe, R., Breer, H., and Strotmann, J. (2003). Organization and evolutionary relatedness of OR37 olfactory receptor genes in mouse and human. Genomics 82, 355-364.

48  

Hoppe, R., Lambert, T.D., Samollow, P.B., Breer, H., and Strotmann, J. (2006). Evolution of the "OR37" subfamily of olfactory receptors: a cross-species comparison. J Mol Evol 62, 460-472.

Hu, J., Zhong, C., Ding, C., Chi, Q., Walz, A., Mombaerts, P., Matsunami, H., and Luo, M. (2007). Detection of near-atmospheric concentrations of CO2 by an olfactory subsystem in the mouse. Science 317, 953-957.

Hussain, A., Saraiva, L.R., Ferrero, D.M., Ahuja, G., Krishna, V.S., Liberles, S.D., and Korsching, S.I. (2013). High-affinity olfactory receptor for the death-associated odor cadaverine. Proc Natl Acad Sci U S A 110, 19579-19584.

Hussain, A., Saraiva, L.R., and Korsching, S.I. (2009). Positive Darwinian selection and the birth of an olfactory receptor clade in teleosts. Proc Natl Acad Sci U S A 106, 4313-4318.

Isogai, Y., Si, S., Pont-Lezica, L., Tan, T., Kapoor, V., Murthy, V.N., and Dulac, C. (2011). Molecular organization of vomeronasal chemoreception. Nature 478, 241-245.

Jaeger, S.R., McRae, J.F., Bava, C.M., Beresford, M.K., Hunter, D., Jia, Y., Chheang, S.L., Jin, D., Peng, M., Gamble, J.C., et al. (2013). A Mendelian trait for olfactory sensitivity affects odor experience and food selection. Curr Biol 23, 1601-1605.

Johnson, M.A., Tsai, L., Roy, D.S., Valenzuela, D.H., Mosley, C., Magklara, A., Lomvardas, S., Liberles, S.D., and Barnea, G. (2012). Neurons expressing trace amine-associated receptors project to discrete glomeruli and constitute an olfactory subsystem. Proc Natl Acad Sci U S A 109, 13410-13415.

Keller, A., Zhuang, H., Chi, Q., Vosshall, L.B., and Matsunami, H. (2007). Genetic variation in a human odorant receptor alters odour perception. Nature 449, 468-472.

Kermen, F., Midroit, M., Kuczewski, N., Forest, J., Thevenet, M., Sacquet, J., Benetollo, C., Richard, M., Didier, A., and Mandairon, N. (2016). Topographical representation of odor hedonics in the olfactory bulb. Nat Neurosci 19, 876-878.

Khan, I., Yang, Z., Maldonado, E., Li, C., Zhang, G., Gilbert, M.T., Jarvis, E.D., O'Brien, S.J., Johnson, W.E., and Antunes, A. (2015). Olfactory Receptor Subgenomes Linked with Broad Ecological Adaptations in Sauropsida. Mol Biol Evol 32, 2832-2843.

49  

Kikuyama, S., Yamamoto, K., Iwata, T., and Toyoda, F. (2002). Peptide and protein pheromones in amphibians. Comp Biochem Physiol B Biochem Mol Biol 132, 69-74.

Kobayakawa, K., Kobayakawa, R., Matsumoto, H., Oka, Y., Imai, T., Ikawa, M., Okabe, M., Ikeda, T., Itohara, S., Kikusui, T., et al. (2007). Innate versus learned odour processing in the mouse olfactory bulb. Nature 450, 503-508.

Kurtovic, A., Widmer, A., and Dickson, B.J. (2007). A single class of olfactory neurons mediates behavioural responses to a Drosophila sex pheromone. Nature 446, 542-546.

Li, Q., Korzan, W.J., Ferrero, D.M., Chang, R.B., Roy, D.S., Buchi, M., Lemon, J.K., Kaur, A.W., Stowers, L., Fendt, M., et al. (2013). Synchronous evolution of an odor biosynthesis pathway and behavioral response. Curr Biol 23, 11-20.

Li, Q., Tachie-Baffour, Y., Liu, Z., Baldwin, M.W., Kruse, A.C., and Liberles, S.D. (2015). Non-classical amine recognition evolved in a large clade of olfactory receptors. Elife 4, e10441.

Liberles, S.D. (2015). Trace amine-associated receptors: ligands, neural circuits, and behaviors. Curr Opin Neurobiol 34, 1-7.

Liberles, S.D., Horowitz, L.F., Kuang, D., Contos, J.J., Wilson, K.L., Siltberg-Liberles, J., Liberles, D.A., and Buck, L.B. (2009). Formyl peptide receptors are candidate chemosensory receptors in the vomeronasal organ. Proc Natl Acad Sci U S A 106, 9842-9847.

Liman, E.R. (2012). Changing senses: chemosensory signaling and primate evolution. Adv Exp Med Biol 739, 206-217.

Liman, E.R., and Innan, H. (2003). Relaxed selective pressure on an essential component of pheromone transduction in primate evolution. Proc Natl Acad Sci U S A 100, 3328-3332.

Lin da, Y., Shea, S.D., and Katz, L.C. (2006). Representation of natural stimuli in the rodent main olfactory bulb. Neuron 50, 937-949.

Lin, D.Y., Zhang, S.Z., Block, E., and Katz, L.C. (2005). Encoding social signals in the mouse main olfactory bulb. Nature 434, 470-477.

50  

Lin, W., Margolskee, R., Donnert, G., Hell, S.W., and Restrepo, D. (2007). Olfactory neurons expressing transient receptor potential channel M5 (TRPM5) are involved in sensing semiochemicals. Proc Natl Acad Sci U S A 104, 2471-2476.

Lucas, P.W., Dominy, N.J., Riba-Hernandez, P., Stoner, K.E., Yamashita, N., Loria-Calderon, E., Petersen-Pereira, W., Rojas-Duran, Y., Salas-Pena, R., Solis-Madrigal, S., et al. (2003). Evolution and function of routine trichromatic vision in primates. Evolution 57, 2636-2643.

Manzini, I., and Korsching, S. (2011). The peripheral olfactory system of vertebrates: molecular, structural and functional basics of the sense of smell. e-Neuroforum 2, 68-77.

Matsui, A., Go, Y., and Niimura, Y. (2010). Degeneration of olfactory receptor gene repertories in primates: no direct link to full trichromatic vision. Mol Biol Evol 27, 1192-1200.

Meisami, E., and Bhatnagar, K.P. (1998). Structure and diversity in mammalian accessory olfactory bulb. Microsc Res Tech 43, 476-499.

Meister, M. (2015). On the dimensionality of odor space. Elife 4, e07865.

Mezler, M., Fleischer, J., and Breer, H. (2001). Characteristic features and ligand specificity of the two olfactory receptor classes from Xenopus laevis. J Exp Biol 204, 2987-2997.

Mohedano-Moriano, A., Pro-Sistiaga, P., Ubeda-Banon, I., Crespo, C., Insausti, R., and Martinez-Marcos, A. (2007). Segregated pathways to the vomeronasal amygdala: differential projections from the anterior and posterior divisions of the accessory olfactory bulb. Eur J Neurosci 25, 2065-2080.

Mombaerts, P. (2004). Genes and ligands for odorant, vomeronasal and taste receptors. Nat Rev Neurosci 5, 263-278.

Munger, S.D., Leinders-Zufall, T., McDougall, L.M., Cockerham, R.E., Schmid, A., Wandernoth, P., Wennemuth, G., Biel, M., Zufall, F., and Kelliher, K.R. (2010). An olfactory subsystem that detects carbon disulfide and mediates food-related social learning. Curr Biol 20, 1438-1444.

51  

Nei, M., Niimura, Y., and Nozawa, M. (2008). The evolution of animal chemosensory receptor gene repertoires: roles of chance and necessity. Nat Rev Genet 9, 951-963.

Nei, M., and Rooney, A.P. (2005). Concerted and birth-and-death evolution of multigene families. Annu Rev Genet 39, 121-152.

Neuhaus, E.M., Zhang, W., Gelis, L., Deng, Y., Noldus, J., and Hatt, H. (2009). Activation of an olfactory receptor inhibits proliferation of prostate cancer cells. J Biol Chem 284, 16218-16225.

Niimura, Y. (2009). Evolutionary dynamics of olfactory receptor genes in chordates: interaction between environments and genomic contents. Hum Genomics 4, 107-118.

Niimura, Y., Matsui, A., and Touhara, K. (2014). Extreme expansion of the olfactory receptor gene repertoire in African elephants and evolutionary dynamics of orthologous gene groups in 13 placental mammals. Genome Res 24, 1485-1496.

Niimura, Y., and Nei, M. (2005). Evolutionary dynamics of olfactory receptor genes in fishes and tetrapods. Proc Natl Acad Sci U S A 102, 6039-6044.

Niimura, Y., and Nei, M. (2006). Evolutionary dynamics of olfactory and other chemosensory receptor genes in vertebrates. J Hum Genet 51, 505-517.

Niimura, Y., and Nei, M. (2007). Extensive gains and losses of olfactory receptor genes in mammalian evolution. PLoS One 2, e708.

Oka, Y., Saraiva, L.R., and Korsching, S.I. (2012). Crypt neurons express a single V1R-related ora gene. Chem Senses 37, 219-227.

Pacifico, R., Dewan, A., Cawley, D., Guo, C., and Bozza, T. (2012). An olfactory subsystem that mediates high-sensitivity detection of volatile amines. Cell Rep 2, 76-88.

Papes, F., Logan, D.W., and Stowers, L. (2010). The vomeronasal organ mediates interspecies defensive behaviors through detection of protein pheromone homologs. Cell 141, 692-703.

Ressler, K.J., Sullivan, S.L., and Buck, L.B. (1993). A zonal organization of odorant receptor gene expression in the olfactory epithelium. Cell 73, 597-609.

52  

Riviere, S., Challet, L., Fluegge, D., Spehr, M., and Rodriguez, I. (2009). Formyl peptide receptor-like proteins are a novel family of vomeronasal chemosensors. Nature 459, 574-577.

Rokni, D., Hemmelder, V., Kapoor, V., and Murthy, V.N. (2014). An olfactory cocktail party: figure-ground segregation of odorants in rodents. Nat Neurosci 17, 1225-1232.

Rolen, S.H., Sorensen, P.W., Mattson, D., and Caprio, J. (2003). Polyamines as olfactory stimuli in the goldfish Carassius auratus. J Exp Biol 206, 1683-1696.

Root, C.M., Denny, C.A., Hen, R., and Axel, R. (2014). The participation of cortical amygdala in innate, odour-driven behaviour. Nature 515, 269-273.

Ryba, N.J., and Tirindelli, R. (1997). A new multigene family of putative pheromone receptors. Neuron 19, 371-379.

Saito, H., Chi, Q., Zhuang, H., Matsunami, H., and Mainland, J.D. (2009). Odor coding by a Mammalian receptor repertoire. Sci Signal 2, ra9.

Sakurai, T., Mitsuno, H., Mikami, A., Uchino, K., Tabuchi, M., Zhang, F., Sezutsu, H., and Kanzaki, R. (2015). Targeted disruption of a single sex pheromone receptor gene completely abolishes in vivo pheromone response in the silkmoth. Sci Rep 5, 11001.

Saraiva, L.R., Ahuja, G., Ivandic, I., Syed, A.S., Marioni, J.C., Korsching, S.I., and Logan, D.W. (2015). Molecular and neuronal homology between the olfactory systems of zebrafish and mouse. Sci Rep 5, 11487.

Saraiva, L.R., Kondoh, K., Ye, X., Yoon, K.H., Hernandez, M., and Buck, L.B. (2016). Combinatorial effects of odorants on mouse behavior. Proc Natl Acad Sci U S A 113, E3300-3306.

Saraiva, L.R., and Korsching, S.I. (2007). A novel olfactory receptor gene family in teleost fish. Genome Res 17, 1448-1457.

Sato, Y., Miyasaka, N., and Yoshihara, Y. (2005). Mutually exclusive glomerular innervation by two distinct types of olfactory sensory neurons revealed in transgenic zebrafish. J Neurosci 25, 4889-4897.

53  

Sato, Y., Miyasaka, N., and Yoshihara, Y. (2007). Hierarchical regulation of odorant receptor gene choice and subsequent axonal projection of olfactory sensory neurons in zebrafish. J Neurosci 27, 1606-1615.

Schaal, B., Coureaud, G., Langlois, D., Ginies, C., Semon, E., and Perrier, G. (2003). Chemical and behavioural characterization of the rabbit mammary pheromone. Nature 424, 68-72.

Serizawa, S., Miyamichi, K., Takeuchi, H., Yamagishi, Y., Suzuki, M., and Sakano, H. (2006). A neuronal identity code for the odorant receptor-specific and activity-dependent axon sorting. Cell 127, 1057-1069.

Sharon, D., Glusman, G., Pilpel, Y., Khen, M., Gruetzner, F., Haaf, T., and Lancet, D. (1999). Primate evolution of an olfactory receptor cluster: diversification by gene conversion and recent emergence of pseudogenes. Genomics 61, 24-36.

Silvotti, L., Moiani, A., Gatti, R., and Tirindelli, R. (2007). Combinatorial co-expression of pheromone receptors, V2Rs. J Neurochem 103, 1753-1763.

Sosulski, D.L., Bloom, M.L., Cutforth, T., Axel, R., and Datta, S.R. (2011). Distinct representations of olfactory information in different cortical centres. Nature 472, 213-216.

Stensmyr, M.C., Dweck, H.K., Farhan, A., Ibba, I., Strutz, A., Mukunda, L., Linz, J., Grabe, V., Steck, K., Lavista-Llanos, S., et al. (2012). A conserved dedicated olfactory circuit for detecting harmful microbes in Drosophila. Cell 151, 1345-1357.

Strotmann, J., Conzelmann, S., Beck, A., Feinstein, P., Breer, H., and Mombaerts, P. (2000). Local permutations in the glomerular array of the mouse olfactory bulb. J Neurosci 20, 6927-6938.

Syed, A.S., Sansone, A., Nadler, W., Manzini, I., and Korsching, S.I. (2013). Ancestral amphibian v2rs are expressed in the main olfactory epithelium. Proc Natl Acad Sci U S A 110, 7714-7719.

Syed, A.S., Sansone, A., Roner, S., Bozorg Nia, S., Manzini, I., and Korsching, S.I. (2015). Different expression domains for two closely related amphibian TAARs generate a bimodal distribution similar to neuronal responses to amine odors. Sci Rep 5, 13935.

54  

Thompson, J.A., Salcedo, E., Restrepo, D., and Finger, T.E. (2012). Second-order input to the medial amygdala from olfactory sensory neurons expressing the transduction channel TRPM5. J Comp Neurol 520, 1819-1830.

Touhara, K., and Vosshall, L.B. (2009). Sensing odorants and pheromones with chemosensory receptors. Annu Rev Physiol 71, 307-332.

Varendi, H., and Porter, R.H. (2001). Breast odour as the only maternal stimulus elicits crawling towards the odour source. Acta Paediatr 90, 372-375.

Weth, F., Nadler, W., and Korsching, S. (1996). Nested expression domains for odorant receptors in zebrafish olfactory epithelium. Proc Natl Acad Sci U S A 93, 13321-13326.

Wyatt, T.D. (2014). Introduction to Chemical Signaling in Vertebrates and Invertebrates. In Neurobiology of Chemical Communication, C. Mucignat-Caretta, ed. (Boca Raton (FL)).

Yang, H., Shi, P., Zhang, Y.P., and Zhang, J. (2005). Composition and evolution of the V2r vomeronasal receptor gene repertoire in mice and rats. Genomics 86, 306-315.

Yopak, K.E., Lisney, T.J., and Collin, S.P. (2015). Not all sharks are "swimming noses": variation in olfactory bulb size in cartilaginous fishes. Brain Struct Funct 220, 1127-1143.

Zhang, X., and Firestein, S. (2002). The olfactory receptor gene superfamily of the mouse. Nat Neurosci 5, 124-133.

Zhang, X., and Firestein, S. (2009). Genomics of olfactory receptors. Results Probl Cell Differ 47, 25-36.

Zhao, H., Xu, D., Zhang, S., and Zhang, J. (2011). Widespread losses of vomeronasal signal transduction in bats. Mol Biol Evol 28, 7-12.

Chapter 3

A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction

56

Attribution

Chapter 3 has been formatted from the following published work, for which I am a

primary author:

Greer, P.L.*, Bear, D.M.*, Lassance, J.M., Bloom, M.L., Tsukahara, T., Pashkovski, S.L., Masuda, F.K., Nowlan, A.C., Kirchner, R., Hoekstra, H.E., et al. (2016). A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction. Cell 165, 1734-1748.

*These authors contributed equally to this work.

Author Contributions DMB, PLG, and SRD conceived experiments and wrote the manuscript. DMB, PLG,

MLB, TT, SP, and ACN carried out all wet experiments; evolutionary analysis was

performed by JML, FKM, and HH. RK performed RNAseq read quality control and

alignment.

57

Summary

Odor perception in mammals is mediated by parallel sensory pathways that

convey distinct information about the olfactory world. Multiple olfactory subsystems

express characteristic seven-transmembrane G-protein coupled receptors (GPCRs) in a

one-receptor-per-neuron pattern that facilitates odor discrimination. Sensory neurons of

the “necklace” subsystem are nestled within the recesses of the olfactory epithelium and

detect diverse odorants; however, they do not express known GPCR odor receptors.

Here we report that members of the four-pass transmembrane MS4A protein family are

chemosensors expressed within necklace sensory neurons. These receptors localize to

sensory endings and confer responses to ethologically-relevant ligands including

pheromones and fatty acids in vitro and in vivo. Individual necklace neurons co-express

many MS4A proteins and are activated by multiple MS4A ligands; this pooling of

information suggests that the necklace is organized more like subsystems for taste than

for smell. The MS4As therefore define a distinct mechanism and functional logic for

mammalian olfaction.

58

Introduction

As animals navigate the natural world they encounter an unending variety of

small molecules, which are rich sources of information that signify the presence of

organisms and salient objects in the environment. The olfactory system detects many of

these molecules through odorant receptor proteins expressed by peripheral olfactory

sensory neurons (OSNs), which are coupled to higher brain circuits mediating odor

perception (Axel, 1995; Ihara et al., 2013). In mammals, the olfactory system is divided

into multiple, parallel processing streams made up of anatomically and molecularly

distinct sensory neuron populations. The largest subdivision, the main olfactory system,

is capable of detecting nearly all volatile odorants and plays key roles in odor

discrimination and learning. Smaller subsystems (such as the vomeronasal system) are

thought to play a more specialized role in odor perception, discriminating odors of innate

significance and releasing specific patterns of reproductive, agonistic, or defensive

behavior (Munger et al., 2009).

The main and vomeronasal olfactory systems each express characteristic

odorant receptor families that belong to the G-protein coupled receptor (GPCR)

superfamily; these receptor families define the specific receptive fields and therefore the

function of each subsystem (Buck and Axel, 1991; Dulac and Axel, 1995; Herrada and

Dulac, 1997; Liberles and Buck, 2006; Liberles et al., 2009; Matsunami and Buck, 1997;

Riviere et al., 2009; Ryba and Tirindelli, 1997). The identification of these receptor

genes (including the Odorant Receptors (ORs), Vomeronasal type 1 Receptors,

Vomeronasal type 2 Receptors (V2Rs), Formyl Peptide Receptors and the Trace

Amine-Associated Receptors) has revealed a key organizational principle: each mature

59

olfactory sensory neuron (with the exception of those within the basal subdivision of the

vomeronasal system) expresses just a single receptor gene of the hundreds encoded in

the genome (Dalton et al., 2015). This pattern of expression defines specific information

channels in the olfactory system, as the axons of those sensory neurons that express

the same odorant receptor converge on a small number of insular structures within the

olfactory bulb called glomeruli; these glomeruli are differentially recruited as animals

sense distinct smells, enabling the brain to discriminate odors detected by the nose

(Mori and Sakano, 2011). Individual basal vomeronasal sensory neurons also target

specific bulb glomeruli, but express two V2Rs instead of a single receptor (Martini et al.,

2001).

While the identification of odorant receptor genes has led to deep insight into the

sensory tuning and functional architecture of the main and vomeronasal subsystems,

there are additional mammalian subsystems whose modes of odor detection — and

therefore function — are less clear. Particularly mysterious is the “necklace” subsystem,

which is distinguished by its unusual anatomy: OSNs within this subsystem are

concentrated in the recesses of the olfactory epithelium (the “cul-de-sac” regions), and

project axons to a ring of 12-40 apparently interconnected glomeruli that encircle the

caudal olfactory bulb like beads on a necklace (Juilfs et al., 1997; Shinoda et al., 1989).

Necklace sensory neurons and glomeruli respond to a diverse range of chemical stimuli,

including gases (such as carbon disulfide and carbon dioxide), pheromones (such as

2,5-dimethylpyrazine (2,5-DMP), 2-heptanone, and E-farnesene), plant-derived

odorants, and urinary peptides (Fulle et al., 1995; Hu et al., 2007; Juilfs et al., 1997;

Leinders-Zufall et al., 2007; Meyer et al., 2000; Munger et al., 2010; Sun et al., 2009).

60

While many of these ligands have innate significance for the mouse, the specific role of

the necklace in mediating odor perception remains unclear.

Intriguingly, necklace OSNs do not express the signaling proteins known to

mediate GPCR-based chemotransduction in the rest of the main olfactory epithelium

(Juilfs et al., 1997; Meyer et al., 2000). While the ability of the necklace system to detect

and behaviorally respond to gases and peptides requires the single pass

transmembrane protein Guanylate Cyclase-D (GC-D), which is specifically expressed in

all necklace neurons, the remainder of the diverse sensory responses observed in this

system are unexplained (Guo et al., 2009; Leinders-Zufall et al., 2007; Sun et al., 2009).

These observations suggest that necklace OSNs harbor an as-yet unrecognized

receptor type, whose identification could reveal key features of the functional

organization and neural logic that governs the necklace system.

Here we show that necklace OSNs express a previously-unidentified class of

chemoreceptor encoded by the Ms4a gene family. Each MS4A protein detects a

specific subset of ethologically-relevant odorants — including fatty acids and the

putative mouse pheromone 2,5-DMP — that stimulate necklace sensory neurons in

vivo. Ectopic expression of MS4A proteins is sufficient to confer responses to MS4A

ligands upon conventional olfactory neurons. However, unlike all known mammalian

olfactory receptors, the Ms4a genes do not belong to the GPCR superfamily and are not

expressed in the conventional one-receptor-one-neuron pattern; instead, each Ms4a

gene encodes a four-pass transmembrane protein and many Ms4a family members are

expressed in every necklace sensory neuron. Taken together, this work defines a new

mechanism for mammalian olfaction and identifies a population of atypical olfactory

61

sensory neurons that each express many members of a receptor gene family,

suggesting a distinct perceptual role for odor information coursing through the necklace

subsystem. Because MS4A proteins are also expressed in chemosensory cells that

reside outside the nasal epithelium, these findings suggest a broader role for the MS4A

proteins in the detection of chemical cues.

62

Results

To identify candidate receptor gene families specific to the necklace olfactory

system, RNAseq was used to compare transcripts expressed by FACS-isolated GC-D-

expressing and conventional OSNs (Figures 3.1A, 3.1B, 3.2A, and 3.2B). This analysis

failed to reveal expression of known odorant receptor families or enrichment of other

GPCR subfamilies within necklace sensory neurons. Consistent with this and prior

reports, Golf, Adcy3, Cnga2, Cnga4, Trpc2, and Trpm5 — gene products required for

odor-related signal transduction in conventional OSNs — also were not expressed in

GC-D cells (Munger et al., 2009). To screen for potential non-GPCR odorant receptors,

the RNAseq data were filtered to identify transmembrane protein families that exhibit

sufficient molecular diversity to potentially interact with a wide range of ligands. As

detailed below, this screen identified the Membrane Spanning, 4-pass A (Ms4a) genes,

which encode a class of four-transmembrane (4TM) spanning proteins that are

structurally distinct from GPCRs (Eon Kuek et al., 2015).

RNAseq revealed transcripts for several Ms4a family members in GC-D-

expressing cells (Figure 3.1B). Because Ms4a transcripts had low RNAseq read counts,

the Nanostring single-molecule detection technique was used to unambiguously

determine the presence of every member of the Ms4a gene family in GC-D cells (Khan

et al., 2011). This analysis revealed the reproducible expression of 12 Ms4a family

members (of the 17 members annotated in the mouse genome), demonstrating that

GC-D cells express low levels of a specific subset of Ms4a genes (Figure 3.1C). None

of these 12 genes was detected in conventional OSNs above background (data not

shown). Notably absent from GC-D cells are the two best-studied Ms4a genes, Ms4a1

63

Figure 3.1

Expression of Ms4a genes in necklace sensory neurons. A. (Left) The Gucy2d-IRES-

TauGFP allele marks necklace sensory neurons expressing PDE2A (blue) and GC-D

(red). Grayscale is nuclear counterstain (DAPI). (Right) Pde2a+ necklace sensory

neurons reside in epithelial “cul-de-sacs” and do not express the Omp-IRES-GFP allele

or the conventional OR signal transduction protein Adenylyl Cyclase3 (red). Scale bars

10 µM. B. Enrichment versus expression plot for every sequenced mRNA in GC-D+ and

OMP+ sensory neurons. Each point is a transcript with detectable RNAseq reads, with

marker genes associated with GC-D cells (Car2, Pde2a, and Cnga3) and OMP cells

(Golf, Cnga2, and Cnga4) labeled in red and green, respectively; reliably detected Ms4a

family members are highlighted in blue. C. Nanostring-based mRNA quantification of

GC-D cells relative to OMP cells. Marker genes for OMP and GC-D cells are shown in

green and red, respectively. Twelve of the 17 annotated Ms4a genes are significantly

enriched in GC-D cells relative to OMP cells (blue bars). * p < 0.05, paired t-test.

64

Figure 3.1 (Continued)

65

Figure 3.2

A. Scatter plots of FAC sorted, dissociated olfactory epithelial cells from wild type mice

(left), mice harboring the Gucy2d-IRES-TauGFP allele (middle), or mice expressing the

Omp-IRES-GFP allele (right). The gate used to isolate ~100% pure populations of

fluorescent necklace OSNs or canonical OSNs is indicated. B. Heat map of the

correlation of gene expression between RNAseq samples, with warmer colors

corresponding to more highly correlated gene expression.

66

Figure 3.2 (Continued)

67

and Ms4a2, which have been implicated in calcium signaling downstream of the B-cell

receptor and high-affinity Fc-Epsilon receptor, respectively, but whose precise function

remains unclear (Bubien et al., 1993; Dombrowicz et al., 1998; Koslowski et al., 2008;

Lin et al., 1996; Polyak et al., 2008).

To assess the molecular diversity of the MS4A family, Ms4a genes were

identified from representative species of all major mammalian lineages. We then asked

how different these genes were within a given species, as amino acid differences are a

prerequisite for individual MS4As to interact with distinct odors. Multiple sequence

alignments revealed substantial intraspecies diversity amongst the MS4As, particularly

within the extracellular domains of the protein, whose length is variable (Figures 3.3A

and 3.3B). This diversity is comparable to that observed in the third through seventh

transmembrane domains in conventional ORs, the regions thought to form ligand-

binding pockets that give rise to odorant specificity (Buck and Axel, 1991; Man et al.,

2004; Singer, 2000). These findings raise the possibility that each MS4A within a given

organism may interact with a distinct set of extracellular cues.

Ms4a genes are found in a single genomic cluster in all queried mammals; this

organization, suggestive of tandem duplication, is also found in known chemoreceptor

gene families and is thought to facilitate diversification of family members (Figure 3.4A)

(Nei et al., 2008). We therefore also assessed differences between Ms4a genes across

species, to identify those regions of the MS4A protein subject to diversifying or purifying

selection. MS4A proteins were significantly divergent across evolution (Figure 3.3C)

(Nei et al., 2008); the most rapidly evolving amino acid residues in the MS4A proteins

are highly enriched in the predicted extracellular loops, where contact with

68

Figure 3.3

A. Multiple sequence alignment of the mouse MS4A proteins expressed in GC-D cells.

Residues that are more conserved are shown in warmer colors, whereas residues that

are less conserved are depicted in colder colors (conservation scores were determined

using PRALINE, see methods). The intracellular (IC), extracellular (EC), and

transmembrane (TM) regions of the proteins are indicated, and reveal the greatest

sequence diversity in the extracellular domains (with additional diversity in the

intracellular C-terminus). B. Multispecies alignments of MS4A proteins from either the

MS4A4 or MS4A6 subfamilies. Amino acid conservation was determined and heat-

mapped as in A. As with alignments of all GC-D-expressed MS4As, the extracellular

domains within these subfamilies are more diverse than other regions of the MS4As. C.

A phylogenetic tree of the mammalian Ms4a gene family was generated using every

Ms4a gene found in 37 representative taxa, which were selected to cover all major

mammalian lineages (Supplementary table 1). Every Ms4a gene was assigned to an

MS4A subfamily (see methods), and each subfamily is represented with a unique color

to facilitate visualization within the circular phylogenetic tree. Each Ms4a gene is

represented as a line within this plot where the length of the line corresponds to the

degree of evolutionary change within a lineage over time (the scale bar represents the

number of substitutions per site). The Ms4a gene family cluster diversified through

tandem duplication early in the evolution of mammals as illustrated by the presence of

10 homologs in the monotreme (platypus, light blue lines) and marsupial (Tasmanian

devil, red lines) representatives, which contrasts the single copy of MS4A15 found in

69

Figure 3.3 (Continued) bird genomes (Zuccolo et al., 2010). Further extension of the family occurred during the

evolution of placental mammals, with human and mouse genomes harboring 16 and 17

genes, respectively. The majority of MS4A subfamilies exhibit one-to-one orthologous

pairs across species. By contrast, the MS4A4 and MS4A6 subfamilies, which are highly

enriched in GCD neurons, demonstrate complex one-to-many and many-to-many

paralogous relationships between species. It is noteworthy that 50% of the genes

present in the bovid representatives are either lost or pseudogenized in cetacean

lineages suggesting rapid gene turnover throughout evolution.

70

Figure 3.3 (Continued)

71

Figure 3.4

Genomic clustering and positive selection of the Ms4a gene family. A. Graphical

representation of Chromosome 19 of Mus musculus, including the entire Ms4a gene

family (red); immediately telomeric to the Ms4a gene cluster resides a large group of

conventional mammalian odorant receptor genes (blue). B. Topographical

representations of the primary sequences of Mus musculus MS4A4A (top, left),

MS4A6B (top, right), ORAI1 (bottom, left), and TAS2R1 (bottom, right). Amino acid

residues under strong purifying selection are shown in blue, whereas those under

positive selection are shown in red (posterior probability > 0.90, see Experimental

Procedures); residues under positive selection are enriched within extracellular loops of

MS4A proteins and bitter taste receptors (p = 5.06 X 10-16 and p = 6.76 X 10-7,

respectively, hypergeometric test).

72

Figure 3.4 (Continued)

73

environmental chemical stimuli could occur (Figure 3.4B). Bitter taste receptors, which

accommodate the specific diet of their host organism, exhibit a similar pattern of

diversifying selection (Figure 3.4B) (Hayakawa et al., 2014; Wooding, 2011). In contrast,

members of the Orai family, which encode 4TM proteins and are not thought to detect

environmental chemical stimuli, show no signs of diversifying selection (Figure 3.4B)

(Amcheslavsky et al., 2015). The MS4A family therefore exhibits a pattern of expansion

and divergence similar to other chemosensors, although the observation that both

within- and between-species MS4A sequence variability is enriched within extracellular

loops — rather than traditional hydrophobic binding pockets — suggests that if the

MS4As detect odors they do so through a distinct domain from conventional GPCR

ORs.

MS4A Proteins Confer Odor Responses In Vitro

Although no endogenous or natural ligands have been identified for any member

of the MS4A family, the specific expression of a molecularly diverse complement of

MS4As within olfactory sensory neurons — taken with prior evidence suggesting

involvement in calcium signaling — raised the possibility that Ms4a genes encode a

novel class of chemoreceptor (Bubien et al., 1993; Dombrowicz et al., 1998; Koslowski

et al., 2008; Lin et al., 1996; Polyak et al., 2008). To test whether specific interactions

between odors and MS4A proteins induce calcium influx in cells, we heterologously

expressed individual MS4A proteins together with the genetically-encoded fluorescent

calcium indicator GCaMP6s in HEK293 cells (Figures 3.5A and 3.5B); expression of

MS4A proteins did not increase the baseline rate of calcium transients (Figures 3.5C

74

Figure 3.5

A. Representative confocal images of HEK293 cells transfected with plasmids encoding

GCaMP6S (green) and N-terminal mCherry-fusion proteins of the indicated MS4A

protein (red), revealing the presence of mCherry-MS4A fusions at the plasma

membrane. B. HEK293 cells transfected with GCaMP6S (green) and either mCherry

alone or mCherry-MS4A6C (red) were immunostained under non-permeablizing

conditons with an extracellularly-directed anti-MS4A6C antibody, revealing specific

labeling of MS4A6C proteins (blue) indicating that MS4A6C is efficiently trafficked to the

plasma membrane and adopts the predicted topology. C. The spontaneous activity rate

of HEK293 cells expressing MOR9-1 or the indicated MS4A was calculated and then

normalized to control cells that expressed GCaMP6s alone (see methods). Data are

presented as mean +/- SEM from 24 independent coverslips per condition. The

normalized spontaneous rate of 1 for GCaMP corresponds to a response percentage of

1.4%. No significant differences were observed amongst conditions (unpaired Student’s

T-test). D. As in C except HEK293 cells expressing any MS4A protein are considered

as a single group.

75

Figure 3.5 (Continued)

76

and 3.5D). Six MS4A proteins (selected for their structural diversity) were exposed to 11

compound mixtures, each of whose constituents shared similar chemical structure.

These mixes were designed to cover a broad swath of odor space, and included known

ligands for conventional and necklace glomeruli (Gao et al., 2010; Saito et al., 2009).

Increases in intracellular calcium were observed during odor exposure for specific

MS4A protein/odor mixture pairs, demonstrating that MS4A proteins transduce signals

reflecting the presence of extracellular small molecule ligands (Figures 3.6A and 3.6B,

Table 3.1).

MS4A sensory responses were specifically tuned to particular odor categories,

with responses enriched for long chain fatty acids, steroids, and heterocyclic

compounds. To identify individual MS4A ligands, the odorant mixture that evoked the

largest response for each MS4A was broken down into its monomolecular constituents

(Figure 3.6C). Individual MS4As conferred responses to a specific subset of odor

molecules within a functional class. For example, cells expressing MS4A6C responded

to a known ligand for necklace glomeruli, 2,5-dimethylpyrazine (2,5-DMP) as well as to

a molecule not previously known to activate the necklace, 2,3-dimethylpyrazine (2,3-

DMP) – but only weakly to 2,6-dimethylpyrazine, (2,6-DMP) and not at all to other

structurally similar molecules in the parent mixture (Figure 3.6C). Several additional

ligand-receptor relationships were identified, including MS4A4B/alpha-linolenic acid

(ALA), MS4A6D/oleic acid (OA), MS4A6D/arachidonic acid (AA), MS4A4D/4-pregnan-

11B,21-diol-3,20-dione 21-sulphate, MS4A7/5-pregnan-3A-ol-20-one sulphate, and

MS4A8A/4-pregnan-11B,21-diol-3,20-dione,21-sulphate (Figures 3.6C, 3.7A, and 3.7B).

77

Figure 3.6

MS4A proteins confer responses to odorants. A. GCaMP6 fluorescence in response to

indicated chemical mixtures in representative HEK293 cells expressing either the

indicated MS4A protein or GPCR mOR + G-protein (odor delivery indicated by gray bars

after accounting for line and mixing delays, see Experimental Procedures). B.

Responses of expressed MS4A protein/odor mixture pairs performed as in A (10 µM per

odor, see Table 3.1 for mixture definitions, 97 total compounds). Color code indicates

percentage of cells responding (n=3, total cells in experiment > 50,000) after

thresholding statistically significant responses (see Experimental Procedures).

Deconvolved mixture-MS4A pairs indicated with red circles. C. Deconvolution identifies

monomolecular odors that activate each MS4A receptor. Individual odors delivered at

50 µM in liquid phase (n=3, total cells in experiment >68,000) to cells co-expressing

GCaMP6s and the indicated MS4A receptor (bottom) or GCaMP6s alone (top). The

aggregate percent of cells that responded to each chemical across three independent

experiments is color-mapped as in B. SFA: saturated fatty acids, UFA: unsaturated fatty

acids. D. Dose response curves reveal low micromolar EC50s for MS4A4B/ALA,

MS4A6C/2,3-DMP, and MS4A6D/OA, which are similar to the EC50 observed for

MOR9-1/vanillin and the reported EC50s for mammalian ORs reported in the literature

(Saito et al., 2009; Mainland et al., 2015). Each data point represents the mean +/- SEM

from at least four independent coverslips.

78

Figure 3.6 (Continued)

79

Table 3.1. Odorants used for functional imaging experiments

Alcohols 1-butanol, 2,5-dimethylphenol, eugenol, guaicol, 1-hexanol,

isoeugenol, 1-nonanol, 1-octanol, 2-phenylethanol, thymol

Ketones acetylanilone, acetophenone, 2-butanone, cyclohexanone, 3-

decanone, dodecanolactone, 4-heptanone, 2-octanone, 2-

pentanone, vanillin

Sulfurs 2,4,5-trimethyl thiazole, TMT, thiophene, tetrahydrothiophene

Acids formic acid, hexanoic acid, heptanoic acid, ocantoic acid, tiglic

acid, valeric acid, isovaleric acid

Esters allyl cinnamate, amyl acetate, benzyl acetate, cycohexyl

aceatate, ethyl benzoate, ethyl propionate, ethyl valerate,

ethyl tiglate, piperidine, propyl butyrate

Aldehydes p-anise aldehyde, butyl formate, butyraldehyde,

benzaldehyde, cinnamaldehyde, ethyl formate, heptanal,

octanal, propionaldehyde, heptaldehdye

80

Table 3.1 (Continued). Odorants used for functional imaging experiments

Nitrogenous 2,5-dimethylpyrazine, 2,6-dimethylpyrazine, 2,3-

dimethylpyrazine, indole, nicotine, pyrrolidine, pyridine,

quinolone

Steroids 4-Androsten-17alpha-ol-3-one sulphate, 5-Androsten-3Beta

17Beta-diol disulphate, 1,3,5(10)-Estratrien-3 17Beta-diol

disulphate, 1,3,5(10)-Estratrien-3 17alpha-diol 3-sulphate,

5alpha-pregnen-3alpha-ol-20-one sulphate, 5beta-pregnen-

3beta-ol-20-one sulphate, 4-pregnan-11beta 21-diol-3 20-

dione 21-sulphate, 4-pregnen-21-ol-3 20-ione

glucosiduronate, 1,3,5(10)-Estratrien-3 17Beta-diol 3-

sulphate, 4-pregnen-11beta 17,21-triol3 20-dione 21-sulphate

PUFAs arachidonic acid, docosohexanoic acid, linoleic acid, linolenic

acid, nervonic acid, oleic acid, petroselenic acid

Saturated fatty

acids

decanoic acid, docosanoic acid, dodecanoic acid, eicosanoic

acid, hexanoic acid, myristic acid, octadecanoic acid, octanoic

acid, palmitic acid

Terpenes R-carvone, 1,4-cineole, citral, cintronellal, R-fenchone, E-beta

farnesene, geraniol, alpha-ionone, linalool, +-menthone,

gamma-terpinene, 1,3-minus-verbenone

81

Figure 3.7

A. Deconvolution of selected odorant mixtures reveals monomolecular compounds that

specifically activate a conventional odorant receptor (MOR9-1) or MS4A4D. Individual

odors were delivered at 50 µM in liquid phase to cells co-expressing GCaMP6s and the

indicated MS4A receptor or mOR (bottom) or GCaMP6s alone (top). The aggregate

percent of cells that responded to each chemical across three independent experiments

is color-mapped as indicated, where only statistically significant responses are plotted

(see Experimental Procedures). B. Traces of dF/F averaged across all cells that

responded to the best monomolecular odorant for each MS4A/mixture pair. C. The

timing of responses of four specific MS4A/odor pairs was assessed by comparing time

of response onset (red), time to half-maximal response (green), and time to peak

response (blue) to the same parameters for cells expressing the control odorant

receptor MOR9-1 and its optimal ligand (vanillin). No statistically-significant differences

were observed between any of the MS4A/odor pairs and MOR9-1/vanillin (Student’s T-

test, Bonferroni corrected for multiple comparisons). Data are presented as mean +/-

SEM. The black and blue dashed lines indicate the estimated time when odor

concentration reaches 75% and 90% of maximum, respectively (determined by dye

experiments, see methods). Between 29 and 65 responding cells across three

independent coverslips were analyzed per condition. D. As in C except calcium was

reported using GCaMP6f, images were acquired at 2 Hz, and odorants were delivered

with a stimulus pencil that was positioned directly above the cells. Between 129 and 700

responding cells from at least 4 independent coverslips were analyzed per condition. * p

< 0.01, Student’s T-test. E. Traces of dF/F from individual cells that responded to

82

Figure 3.7 (Continued) stimulus pencil-delivered odorant for each MS4A/ligand pair for which dose response

curves were generated. F. The requirement of extracellular calcium for MS4A ligand

responses was assessed by stimulating HEK293 cells expressing GCaMP6s alone or

co-expressing MS4A6C and GCaMP6s with 2,3-DMP in the presence or absence of

extracellular calcium. Data are presented as mean +/- SEM from at least 4 coverslips

per condition. * p < 0.01, Z-test of proportions.

83

Figure 3.7 (Continued)

84

We also asked whether the MS4As and conventional ORs responded to odors

with similar kinetics in vitro. No statistically significant differences were observed in time

to response onset, time to half-maximal response, or time to peak response either

between four queried MS4As or between these MS4As and MOR9-1 (Figure 3.7C).

However, because the bulk calcium imaging assay used to screen MS4A ligands is not

optimized for assessing response timing, we verified these results using GCaMP6f (a

more rapidly-responding calcium indicator than GCaMP6s), a faster imaging rate, and a

stimulus pencil to focally deliver odorants directly above the imaged cells. These

experiments revealed that MS4A responses occur seconds after odor presentation, with

similar or slightly faster response dynamics to those observed with MOR9-1 (Figures

3.7D and 3.7E, see Experimental Procedures). Dose-response curves revealed low

micromolar EC50s for three specific MS4A/odor pairs, similar to the EC50 observed for

MOR9-1/vanillin (Figure 3.6D), and well within the range of EC50s typically observed for

conventional odorant receptor/ligand pairs in vitro (Saito et al., 2009; Mainland et al.,

2015). Depleting extracellular calcium abolished MS4A ligand-dependent calcium

transients (Figure 3.7F). Taken together, these results demonstrate that individual

MS4A proteins enable calcium influx in response to specific monomolecular odorants in

heterologous cells, with different MS4A proteins conferring responses to different

ligands. The simplest explanation for this result is that the MS4A proteins are odorant

receptors.

85

Many Ms4a Genes And Proteins Are Expressed In Each Necklace Sensory Neuron

The functional organization of the mammalian olfactory system depends on each

mature OSN expressing just one (or two) of the hundreds of possible olfactory receptor

genes (Dalton et al., 2015). We therefore asked whether the MS4As are expressed in a

one-receptor-per-neuron pattern within GC-D cells, which would suggest that the

necklace system follows a similar functional logic to the main and accessory olfactory

systems. Target Ms4a mRNA molecules within dissociated GC-D cells were labeled

using a single molecule detection approach (RNAScope), in which messages are

detected as diffraction-limited fluorescent puncta whose abundance reflects transcript

levels (Wang et al., 2012). RNAScope probes generated multiple puncta in individual

GC-D cells for each of the 12 Ms4a genes found by RNA profiling, consistent with the

presence of these Ms4a messages in GC-D cells (Figure 3.8A). Conversely, RNAScope

failed to identify Ms4a1, Ms4a2, and Ms4a5 puncta in GC-D cells above the background

detection rate, consistent with the absence of these specific Ms4a genes in our earlier

RNA profiling experiments (Figures 3.8B and 3.9A).

Using a stringent criterion (in which a cell is counted positive only if it harbors

two or more puncta), transcripts for each of the 12 Ms4a genes expressed in GC-D cells

were found in 5-30% of GC-D cells (Figure 3.8B). Moreover, employing a less stringent

criterion for Ms4a positivity — in which cells with any Ms4a puncta are considered

positive — reveals that individual Ms4a family members may be expressed in >50% of

GC-D cells (Figure 3.9B). Under both of these analyses, the proportions of cells

expressing each Ms4a message sum to significantly greater than 100%, raising the

surprising possibility that each GC-D cell expresses more than one Ms4a gene. To

86

Figure 3.8 Multiple Ms4a genes are expressed in each necklace sensory neuron. A. RNAScope

single molecule fluorescent in situ hybridization of dissociated olfactory epithelial cells

detects Ms4a family members (red). Necklace cells identified via co-labeling with an

antibody against Car2 (blue), GFP from the Gucy2d-IRES-TauGFP allele (green, top

left) or an RNAScope probe against a necklace marker gene (green, all panels except

the top left). Necklace cells are not marked by a probe against the conventional OR

gene Olfr151 (top right). Nuclei marked by DAPI (grayscale); cytoplasmic anti-CAR2

signal was saturated to demarcate the entire volume of each GC-D cell. Scale bar 5 µM.

B. Proportion of Car2+ necklace OSNs with two or more detected puncta for each Ms4a

(blue bars) and Olfr probe (red bars, including Ms4a puncta in OR174-9-IRES-GFP

expressing cells; n=3 experiments, between 150-750 cells/probe, error bars are

standard error of the proportion). Dashed red line represents the average value of

negative controls. All Ms4a probes (other than negative controls Ms4a1, Ms4a2, Ms4a5)

give a significantly higher proportion of positive cells than negative controls (p < .01,

one-tailed Z test on population proportions). C. Representative images of Car2+ (blue)

cells labeled with probes against the indicated Ms4a family members (top panels). The

proportion and total number of cells with multiple puncta for neither, one, or both colors

was quantified (bottom panels, total cell number in parentheses). Each pair shows

significantly more double-positive cells (yellow) than expected (p < .05, Fisher’s Exact

Test on 2x2 table). Scale bar 2 µM.

87

Figure 3.8 (Continued)

88

Figure 3.9

A. Representative images from RNAscope assays of dissociated olfactory epithelial

cells. Necklace cells were identified with an antibody against Car2 (blue), and puncta

from probes against an Ms4a or Olfr family member are in red. DAPI stain is

represented as grayscale. Note that the Car2 signal was intentionally saturated to

identify the volumes of individual GC-D cells, enabling unambiguous assignment of red

puncta to individual GC-D cells. DAPI stain is represented as grayscale. Ms4a6c puncta

are not found in GFP+ cells from dissociated OR174-IRES-GFP epithelia (bottom

right). B. Proportion of necklace OSNs (identified as Car2+) with one or more

fluorescent puncta for each Ms4a and Olfr probe (n=3 experiments, between 150-750

cells/probe, error bars are standard error of the proportion). Dashed red line represents

the average value of negative controls (Ms4a1, Ms4a2, Ms4a5, and the Olfr genes). C.

Representative images of Car2+ (blue) cells co-labeled with additional Ms4a probe

pairs.

89

Figure 3.9 (Continued)

90

directly test whether Ms4a genes are co-expressed, we simultaneously labeled cells in

two different colors with RNAScope probes recognizing distinct Ms4a genes. Every

tested pair revealed a significant rate of cells that were positive for more than one Ms4a

gene, demonstrating that unlike conventional ORs, multiple Ms4a genes are co-

expressed in individual necklace sensory neurons (Figures 3.8C, 3.9C and data not

shown). These data are consistent with a model in which each GC-D cell expresses

multiple Ms4a genes.

Because different probes could have distinct false negative rates (and because

Ms4a transcript levels are low), RNAScope could not be used to definitively determine

the number of unique Ms4a genes expressed per necklace OSN (Kim et al., 2015). We

therefore asked whether the Ms4a expression pattern was more apparent at the protein

level. Peptide antibodies were raised against several MS4As and used to stain the

olfactory epithelium. Consistent with every necklace cell expressing all members of the

MS4A family (of the subset expressed in GC-D cells), each of five different anti-MS4A

antibodies labeled > 95% of GC-D neurons (Figures 3.10A and 3.10B). As expected,

these antibodies did not label conventional OMP+ OSNs (Figure 3.10A), and control

antibodies against MS4As not detected in necklace neurons by RNA profiling —

MS4A1, MS4A2, and MS4A5 — did not label GC-D cells (Figure 3.10B and data not

shown). Purified anti-MS4A antibodies were specific in vivo as assessed by peptide

competition, and in vitro as assessed by staining HEK 293T cells overexpressing

individual MS4A proteins, although minor cross-reactivity was observed between pairs

of highly homologous MS4A proteins (e.g., MS4A6B/MS4A6C and MS4A4B/MS4A4C)

(Figures 3.11A and 3.11B).

91

Figure 3.10

Multiple MS4A proteins are expressed within necklace sensory endings and glomeruli.

A. Anti-MS4A4B antibody stains every anti-PDE2A+ cell but no OMP-IRES-GFP+ cells

in sections of the olfactory epithelium. Scale bar 10 µM. B. Representative images of

immunostaining with antibodies against five different MS4A family members, each of

which stains >95% of anti-PDE2A+ necklace cells in epithelial cul-de-sacs; control

antibody against MS4A5, which is not detected at the mRNA level in GC-D cells, does

not label necklace cells (bottom right). Scale bars 10 µM. C. Anti-MS4A antibodies

label dendritic knobs. Many MS4A antibodies also appear to stain perinuclear and

nuclear regions; although the origin of this staining, which is eliminated by peptide

competition (Figure 3.11B), is unknown, it may represent MS4A protein trapped within

the endoplasmic reticulum or MS4A protein fragments (Cruse et al., 2013). Scale bars 5

µM. D. Anti-MS4A6D staining overlaps with all GCD-IRES-TauGFP+ necklace glomeruli

in sections of the olfactory bulb (left). Blue arrows mark non-necklace glomeruli, which

are not stained by anti-MS4A6D antibody. Similarly, anti-MS4A4B and anti-MS4A7

antibodies label each necklace glomerulus (right panels). Scale bars 20 µM.

92

Figure 3.10 (Continued)

93

Figure 3.11

A. Representative images of HEK293T cells transfected with a plasmid encoding a

single mCherry-MS4A fusion protein and stained with the indicated anti-MS4A antibody;

antibodies are specific, although under conditions of overexpression anti-MS4A4B

cross-reacts modestly with the closely related MS4A4C, as does anti-MS4A6C with

MS4A6B. anti-Ms4A immunofluorescence (red) overlaps almost completely with

appropriate mCherry fluorescence (blue) and is not seen in cells that do not express the

plasmid at appreciable levels, visible by nuclear counterstain (grayscale.) B.

Representative images of cul-de-sac tissue sections immunostained in the presence of

peptide competitor (~1000-fold molar excess, see Experimental Procedures). Only the

antigenic peptide, and not a peptide from a different MS4A protein, blocks staining of

necklace cells by a given antibody.

94

Figure 3.11 (Continued)

95

If MS4As function as necklace chemoreceptors they must be present at sensory

endings where transduction of odorant binding occurs (Barnea et al., 2004). Consistent

with this possibility, high resolution imaging demonstrated that each MS4A antibody

strongly labeled the dendritic knobs of GC-D cells, with some staining apparent in the

cilia as well (Figure 3.10C). Sensory neurons in the main olfactory epithelium also traffic

receptors to their axonal endings, which terminate in glomeruli in the olfactory bulb. To

test whether MS4A proteins are similarly localized to axonal endings, olfactory bulb

tissue sections were probed with anti-MS4A antibodies, which revealed that each MS4A

antibody labeled every necklace glomerulus — but failed to label any conventional

glomeruli — within the olfactory bulb (Figure 3.10D). Taken together, these data

demonstrate that MS4A proteins are appropriately positioned within sensory endings to

detect chemical cues in the environment, and that every glomerulus in the necklace

system receives input from sensory afferents potentially representing information pooled

from all the MS4As expressed in the necklace system; this pattern of organization

differs sharply from that apparent in the rest of the olfactory system, where individual

receptors (or pairs of receptors) define specific glomerular information channels.

Necklace Olfactory Neurons Respond to MS4A Ligands In Vivo

The expression of multiple MS4A family members in GC-D cells predicts that the

chemoreceptive fields of individual necklace olfactory neurons in vivo should include the

ligands identified for different MS4A proteins in vitro. To address this possibility, the

intact olfactory epithelium was explanted and functional responses were assessed by

multiphoton microscopy as odorants were delivered in liquid phase. Necklace neurons

96

were labeled with the fluorescent calcium reporter molecule GCaMP3 using the Emx1-

Cre driver line; the Emx1 gene was enriched in RNASeq and Nanostring analyses of

GC-D neurons, and distinguishes necklace cells from conventional OSNs, which

express the related protein EMX2 (Figure 3.12A and data not shown) (Hirota and

Mombaerts 2004). Because Emx1 labels a number of non-GC-D cells in the nasal

epithelium (whose identity is unclear), we specifically imaged cul-de-sac regions, where

nearly all GCaMP3-positive cells belong to the necklace (Figure 3.12B), and

heuristically identified necklace neurons as those that responded to carbon disulfide, a

known necklace ligand whose detection requires the GC-D protein (Munger et al.,

2010).

Necklace neurons were activated by a mixture of the unsaturated fatty acids oleic

acid and alpha-linolenic acid (UFAs), and by a mixture of 2,3- and 2,5-dimethylpyrazine

(DMPs) — two chemical classes that stimulated MS4A proteins in vitro — but rarely by

mixtures of ketones, esters, or alcohols, which were not identified as MS4A ligands

(Figures 3.13A, 3.13B, and 3.6B). Importantly, UFAs and DMPs do not activate adjacent

conventional OSNs, (i.e., those that responded to the control odor mixtures but not to

CS2 (data not shown)); necklace cells are therefore specifically tuned to MS4A-

activating compounds. These data also demonstrate that single necklace cells respond

to multiple compounds that individually activate different MS4A proteins in vitro as oleic

acid and alpha-linolenic acid are ligands for MS4A6D and MS4A4B, respectively, and

the dimethylpyrazines are ligands for MS4A6C (Figures 3.13A, 3.13B, and 3.6C).

Because MS4A proteins exhibit partially overlapping tuning properties to mixtures in

vitro, we confirmed that individual necklace cells responded to multiple monomolecular

97

Figure 3.12

A. Quantification of mRNA expression in GC-D cells relative to OMP cells using the

single-molecule detection method Nanostring. Marker genes for OMP cells such as

Adcy3 (green bars) and GC-D cells like Car2 (red) are enriched in the appropriate

populations. This analysis revealed that whereas OMP cells express the transcription

factor Emx2, GC-D sensory cells exclusively express Emx1 * p < 0.05, paired t-test. B.

Immunohistochemical analysis of sections prepared from the nasal epithelium of mice

co-expressing an Emx1-cre allele and a Cre-dependent GCaMP3 reporter using

antibodies against GCaMP (green) and the necklace marker CAR2 (red) reveals that a

large fraction of GCaMP-expressing cells are necklace cells; note that CAR2 staining

tends to be enriched in nuclei whereas GCaMP is enriched in cytoplasm.

98

Figure 3.12 (Continued)

99

Figure 3.13 Multiple in vitro MS4A ligands activate single necklace cells in vivo. A. Functional

responses of individual necklace cells (labeled with Emx1-cre/GCaMP3) to the indicated

odorants were heat-mapped (top row) and dF/F traces for six cells (drawn from the

heatmapped image) were plotted (bottom rows). Colored bars represent the 10-second

odor delivery period (after accounting for line and mixing delays, see Experimental

Procedures). All mixtures were at 100 µM total odorant concentration (DMP: 2,3-DMP

and 2,5-DMP, UFA: oleic acid and alpha-linolenic acid, ketones, esters, and alcohols as

in Table 3.1); based on testing with fluorescent dye, stimuli are diluted approximately

1:10 in the cul-de-sac lumen (see Experimental Procedures). Scale bar 10 µM. B. The

odor tuning properties of 74 CS2-responsive cells (columns) across eight experiments

were quantified, with color coding as in the top panel. DMPs and UFAs each activated

significantly more cells (response > 25% dF/F) than each negative control (P < 0.0001,

Fisher’s Exact Test, corrected for multiple comparisons). Significantly more cells

responded to both UFAs and DMPs than expected by chance (P < 0.01, Fisher’s Exact

Test).

100

Figure 3.13 (Continued)

101

MS4A ligands (data not shown). These results strongly suggest that co-expression of

Ms4a genes confers upon each GC-D cell a tuning profile that is broader than that

associated with any single MS4A protein.

Given that MS4A ligands posess a range of volatilities, we wished to directly

demonstrate that these odorants can activate necklace sensory neurons in intact mice.

Freely behaving mice were exposed to MS4A ligands in gas phase, and then activation

of GC-D cells was measured using an antibody against phosphorylated ribosomal S6

protein, an established marker of prior OSN activity (Jiang et al., 2015). These

experiments revealed that the MS4A6C ligands 2,5-DMP and 2,3-DMP, the MS4A4B

ligand alpha-linolenic acid, and the MS4A6D ligands oleic acid and arachidonic acid

reliably activated GC-D cells in vivo to a similar extent as the positive control carbon

disulfide (Figure 3.14A). Conversely, general odorants (including acetophenone and

eugenol) and compounds that only weakly activated MS4A-expressing HEK293 cells

(such as 2,6-DMP) did not activate necklace cells above the background rate of plain air

(Figure 3.14A). While prior work had implicated 2,5-DMP as a necklace ligand, 2,3-

DMP, arachidonic acid and the fatty acids had not been previously shown to activate

this olfactory subsystem. These experiments demonstrate that, in the context of active

exploration, necklace sensory neurons respond to ligands for MS4A receptors.

It is not clear how soluble ligands such as urinary peptides gain access to

necklace sensory neurons within the olfactory epithelium; nevertheless, one highly

soluble class of MS4A ligand — the sulfated steroids — also activated the necklace,

albeit more weakly than observed for other odorants (control DMSO = 8.7 ± .6 vs.

sulfated steroids 1,3,5(10)-estratrien-3,17-beta-diol disulphate/1,3,5(10)-estratrien-

102

Figure 3.14

MS4A ligands activate necklace sensory neurons, and MS4A proteins confer MS4A

ligand responses to conventional olfactory sensory neurons in awake, behaving mice.

A. Example images of cul-de-sacs from mice exposed to the indicated odorant,

immunostained for the necklace cell marker PDE2A (blue) and the neuronal activity

marker phospho-S6 (pSerine240/244) (red) (left panels). Quantification of the

proportion of pS6+ necklace cells in odor-exposed mice (right panel, mean +/- SEM, n

>= 3 independent experiments, * indicates p < 0.05, ** indicates p < 0.01, and ***

indicates p < 0.0001, unpaired t-test compared to null exposure). Scale bar 10 µM. B.

Olfactory epithelial sections of mice infected with adenovirus carrying an Ms4a6c-IRES-

Gfp expression cassette reveal a subset of virally infected cells (green) that also

express MS4A6C protein (red). Scale bars 5 µM. C. Representative images (left

panels) and quantification (right panel) of phospho-S6 positive, OSNs infected with

Ms4a6c-IRES-Gfp or Ms4a6d-IRES-Gfp-expressing adenovirus and exposed to the

indicated odorant (DMP = dimethylpyrizine, OA = oleic acid). Gray bars: GFP-positive/

MS4A6C-negative, red bars: GFP-positive/ MS4A6C or MS4A6D-positive cells (n >= 3

animals per odor, ** indicates p < .001, *** p < .0001, Fisher’s Exact Test comparing

MS4A-positive to MS4A-negative cells for each odorant). Scale bars 5 µM.

103

Figure 3.14 (Continued)

104

3,17alpha-diol 3-sulphate 12.9 ± 1.5 percent cells positive, n=4, p<.05 unpaired T test).

It is notable that both in explants and in vivo the MS4A ligands and the control ligand

carbon disulfide activated a fraction of the necklace sensory neurons. While these

partial responses (also previously observed for urinary peptides) (Leinders-Zufall et al.,

2007) may reflect specific technical features of these experiments, the consistency of

this observation across ligands and preparations raises the possibility that cellular

responses in the intact necklace system may be context- or state-dependent.

Altogether these results support a model in which MS4A receptors bind inhaled

odorants and induce calcium influx into necklace OSNs. This model predicts that

ectopically expressed MS4A protein will confer its in vitro chemosensitivity onto the

receptive fields of conventional olfactory neurons. To test this hypothesis directly,

mouse nares were irrigated with adenovirus encoding bicistronic Ms4a6c-IRES-GFP

transcript, yielding sparse populations of GFP+ OSNs. This approach generated

neurons that expressed MS4A6C protein as well as a subset of cells that were infected

but did not express MS4A6C, which served as an internal control (Figure 3.14B); the

failure of MS4A6C protein expression in some GFP positive cells may reflect stochastic

(and potentially cell type-specific) effects related to ectopic chemosensor expression

within mature OSNs (Tsai and Barnea, 2014). Ectopic MS4A6C protein was localized to

dendritic endings in infected neurons, suggesting that MS4A proteins associate

intrinsically with sensory structures even outside the molecular milieu of the necklace

(Figure 3.14B and data not shown). Quantitative analysis of neural activity (assessed by

phosphorylated S6 protein levels, Figure 3.14C, blue channel) after exposing awake,

behaving animals to odors in gas phase revealed that ectopic MS4A6C confers

105

responses to its in vitro ligands 2,3-DMP and 2,5-DMP, but not to the control odorants

eugenol and acetophenone (Figure 3.14C). Similarly, adenovirally-mediated ectopic

expression resulted in effective targeting of MS4A6D to sensory endings in conventional

olfactory sensory neurons, and conferred specific responses to the MS4A6D ligand

oleic acid, but not to the control odorant eugenol (Figure 3.14C). These results

demonstrate that an MS4A protein can directly impart its chemoreceptive properties to

generic olfactory neurons in vivo, and furthermore that signaling downstream of MS4A

proteins does not require necklace-specific molecular components such as GC-D,

PDE2A or CNGA3. The concordance of necklace cell responses with the

chemoreceptive fields of MS4A proteins expressed ectopically — both in vitro and in

vivo — implies that the necklace subsystem uses the Ms4a family to sense odors.

106

Discussion

Insects and mammals use similar molecular mechanisms to detect light, heat and

several gases, suggesting that solutions to common sensory problems are often

conserved (Caterina, 2007; DeLuca et al., 2012; Dhaka et al., 2006; Terakita, 2005).

However, peripheral mechanisms for odor detection differ amongst phyla — insects like

Drosophila melanogaster deploy several structurally-distinct ionotropic odorant receptor

classes to interrogate the chemical world, whereas mammals were thought to detect

smells exclusively through metabotropic GPCRs (Silbering and Benton, 2010). Our

identification of a non-GPCR family of odorant receptor reveals an unexpected similarity

between the mammalian olfactory system and that of insects: both use multiple

unrelated receptor types to transduce chemosensory cues into intracellular signals.

Multiple lines of evidence indicate that Ms4a genes encode a novel family of

chemoreceptors. Mammalian MS4A proteins are localized to the dendritic endings of

olfactory sensory neurons, the site of odorant chemotransduction, and contain

hypervariable regions that can potentially interact with diverse extracellular cues.

Expression of MS4As in both HEK293 cells in vitro and conventional OSNs in vivo

confers specific responses to odorants, indicating that MS4A proteins are sufficient to

transduce the binding of extracellular ligands into intracellular signals. Furthermore,

individual necklace sensory neurons, which co-express multiple MS4A proteins, each

respond in vivo to multiple MS4A ligands identified in vitro. These data demonstrate that

the MS4As define a new mechanism and logic for mammalian chemosensation, and are

likely responsible for endowing the necklace olfactory subsystem with specific sensory

odor tuning properties.

107

An Alternative Logic Organizes the Necklace Olfactory System

We find that MS4A ligands are enriched for molecules with innate significance for

the mouse, including the aversive pheromone 2,5-DMP and several appetitive long

chain fatty acids; interestingly, seeds and nuts, which are a major food source for mice

in the wild, have high concentrations of those specific fatty acids detected by the MS4As

(Sabir et al., 2012). Like the MS4A ligands, many of the molecules previously shown to

trigger activity within the necklace olfactory system also have innate meaning for mice;

these include carbon dioxide, which mice robustly avoid, and carbon disulfide and the

peptides guanylin and uroguanlyin, each of which promotes the social transmission of

food preferences (STFP) between mice (Arakawa et al., 2013; Hu et al., 2007; Munger

et al., 2010).

The tuning properties of the necklace are determined, in part, by the co-

expression of genes encoding multiple MS4A receptors within individual necklace

sensory neurons. This pattern of odor receptor gene expression stands in stark contrast

to the canonical one-receptor-per-neuron rule that (to a first approximation) organizes

the remainder of the mammalian and the entirety of the Drosophila olfactory systems. In

those systems, each OR is associated with one or a small number of glomeruli in the

brain; in the necklace system, each sensory neuron and glomerulus is, in principle,

capable of conveying odor information detected by all of the MS4A receptors expressed

in the nose.

The ability of individual necklace neurons to detect multiple innately relevant

signals (including gases, peptides, and volatile odors) of widely differing valences

suggests two broad models for the perceptual function of the necklace subsystem. First,

108

the necklace could discriminate between odors, albeit through distinct mechanisms from

the conventional olfactory system. Odor-evoked differences in activity in individual

necklace glomeruli could be generated by subtle differences in MS4A expression within

subsets of sensory neurons (but see Figure 3.10); such differences could also be

caused by conventional OSN axons, which may co-mingle with necklace sensory axons

in necklace glomeruli (Cockerham et al., 2009; Secundo et al., 2014). Stimulus

discrimination could alternatively be achieved through linear or non-linear interactions

between intracellular signals downstream of individual MS4A receptors, which could

allow the integration or gating of signals arising from distinct odorants and gases

detected by single necklace neurons (van Giesen et al., 2016); this process could

generate a range of firing rates (depending, for example, on the specific odor

components present in a blend) that could then be differentially read-out by the brain.

Second, the main function of the necklace system could be odor detection rather

than discrimination. In this model, the necklace system may be organized in a similar

manner to the mammalian bitter taste and C. elegans olfactory systems, which co-

express receptors with widely-divergent receptive fields in sensory neurons coupled to

circuits mediating stereotypical behavioral outputs, such as attraction or aversion (Adler

et al., 2000; McCarroll et al., 2005; Troemel et al., 1995). In the case of the necklace

system, this singular behavioral output could be aversion, as many of the MS4A ligands

have been shown to be contextually aversive, including (despite their nutritional value)

fatty acids at high concentrations (Galindo et al., 2012). Alternatively, the necklace

could act as an alert system, signaling the presence of salient cues in the environment

(whose identity would be disentangled by other sensory systems); such a notification

109

system could be useful for modulating the internal state of the animal in response to

particularly relevant external cues, thereby facilitating both innate responses to odors

and odor-related learning processes (potentially including STFP).

In either the discrimination or detection model, the physiological function of the

necklace could depend upon interactions between intracellular signals downstream of

the MS4A receptors and GC-D. Although the MS4As do not obligately require GC-D to

promote calcium influx, GC-D enzymatic activity is sensitive to calcium levels in vitro

(Duda and Sharma, 2008). This observation raises the possibility that the MS4A and

GC-D pathways intersect in some manner, perhaps facilitating coincidence detection

between gases and peptides on the one hand, and volatile odors on the other.

Dissecting MS4A Function

The availability of ligands for the MS4As now renders accessible key questions

about the biophysics and biochemistry of MS4A function. Previous biochemical studies

relied upon clustering antibodies to trigger MS4A1-mediated calcium flux (Janas et al.,

2005); this and related work found that the MS4As facilitate increases in intracellular

calcium by acting as co-receptors for associated ligand-binding proteins such as the B-

cell receptors (Dombrowicz et al., 1998; Eon Kuek et al., 2015). Interpreting these

experiments has been difficult in light of possible gains-of-function imposed by the use

of clustering reagents; here we show that small molecule ligands can cause an MS4A-

dependent influx of extracellular calcium, demonstrating that the MS4A molecules

themselves have a receptor function. It is not clear, however, whether the MS4A

proteins themselves form a calcium-permeable channel (similar to the Drosophila IR

110

odorant receptors), or whether the MS4As act as ligand-binding co-receptors for an ion

channel that is expressed in many cell types (Abuin et al., 2011; Benton et al., 2009).

The underlying molecular mechanisms through which the MS4A proteins interact

with ligands are not known. The observation that the molecular diversity within the

MS4A family is largely found within extracellular domains — rather than transmembrane

domains — suggests that the MS4As interact with odor ligands through biophysical

mechanisms that are distinct from those used by conventional ORs. These mechanisms

may be similar to those used by mammalian bitter taste receptors, whose diversity is

also found largely in extracellular domains (Hayakawa et al., 2014; Woodling, 2011). It

is also unclear whether the MS4As, despite being co-expressed in single cells, primarily

interact with odorants as homomers (like the bitter taste receptors) or whether they

heteromerize in a manner that alters their tuning properties (Howie et al., 2009).

The Four-Pass Transmembrane MS4As: Chemical Detectors Across Cell Types and

Species?

Although a number of cellular roles have been suggested for individual MS4A

proteins (largely in the context of the immune system), no clear picture has emerged of

the core function of the MS4A family across cell types. The finding that multiple MS4As

encode olfactory receptors suggests that they act as chemosensors in a range of

tissues; this role may be both exteroceptive, serving to detect small molecules from the

outside world, and interoceptive, as several of the ligands for MS4A proteins (e.g., oleic

acid and arachidonic acid) are used as signaling molecules physiologically. Consistent

with this hypothesis, members of the MS4A6/7 subfamily are expressed in microglia,

111

neuroimmune cells responsible for sensing and responding to a variety of endogenous

protein and lipid chemosignals in the mammalian brain; human MS4A8B and MS4A12

are expressed in epithelial cells at the lumenal surface of the small and large intestine,

tissues that both sense dietary lipids; human MS4A8B is localized specifically in

chemosensory cilia in lung cells tasked with probing the sensory environment; and

MS4A5 is expressed in mammalian spermatocytes, which chemotax to oocytes (Eon

Kuek et al., 2015; Koslowski et al., 2008; Wang et al., 2015).

MS4A homologs are present in all mammalian lineages and in many sequenced

deuterostomes including echinoderms (Figure 3.3C and data not shown), implying that

the evolution of the Ms4a genes antedates the advent of the mammalian receptors for

taste and for pheromones (Grus and Zhang, 2009). Taken with the finding that the

MS4As detect a variety of innately-relevant cues, these observations invite the

speculation that the MS4A molecules represent an ancient mechanism for sensing

ethologically-salient small molecules in the environment. Testing this hypothesis will

require establishing chemosensory roles for Ms4a genes in other species and better

definition of those natural ligands that optimally activate the MS4As (given the limits of

the synthetic odor panel explored here). It is important to note that all MS4A ligands

thus far identified are also detected by other receptor molecules in the smell and taste

systems (Isogai et al., 2011; Mamasuew et al., 2011; Oberland et al., 2015).

Nevertheless, the maintenance of the MS4A receptor repertoire for more than 500

million years, especially given the evolutionary success of G-protein coupled odor

receptors, argues that MS4As play an important — and non-redundant — role in

sensory physiology (Zuccolo et al., 2010).

112

Although the Ms4a genes are conserved across vertebrates, Gucy2d is itself

pseudogenized in most primates (Young et al., 2007). It is unclear if the absence of GC-

D reflects a disappearance of the necklace system in primates or merely that GC-D

became unnecessary for the tasks required of the primate necklace, given the

persistence of the Ms4a genes (Young et al., 2007). Similarly, the vomeronasal organ,

the sensory epithelium thought to mediate the majority of pheromone responses in

rodents, became vestigial approximately 25 million years ago (Zhang and Webb, 2003).

Future investigation of the functional role of Ms4a gene families across chordates —

and the relevance of interactions between MS4A proteins and ethologically-relevant

cues like pheromones and fatty acids — will reveal both common and species-specific

roles of the MS4As in processing information from the chemical environment.

113

Experimental Procedures Summary Graphical representations of the data are presented as the mean +/- standard

error of the mean unless otherwise noted. All mice were obtained from the Jackson

Laboratory with the exception of OR174-9-IRES-tauGFP, which was obtained from the

Axel lab. For deep sequencing, olfactory epithelia were dissociated using papain and

individual cells were FAC sorted; RNA was then isolated using Trizol (Invitrogen) before

SmartSeq2-based amplification and Illumina-based sequencing. Nanostring-based RNA

quantitation was performed using a modified protocol to optimize for recovery of low-

abundance transcripts. MS4A sequences were extracted from the Ensembl and NCBI

databases. MS4A proteins were expressed in HEK293 cells using a Tet-regulatable

system to control protein expression (Invitrogen); odor delivery was either achieved via

bulk exchange or through a focal stimulus pencil placed over specific fields of view, and

odor delays were determined using Rhodamine B dye as a surrogate. Odor responses

were imaged in both configurations using an Andor Neo sCMOS camera mounted on an

Olympus IX83. Single molecule in situ RNA detection was perfomed using RNAScope

(Advanced Cell Diagnostics); this protocol was adapted to simultaneously immunostain

with either anti-CAR2 or anti-GFP antibodies. Custom antibodies against the MS4A

proteins were developed in rabbits (Covance) and purified using protein A-sepharose

and bead-conjugated peptides. Multiphoton-based functional imaging (Prairie

Technologies) of explanted olfactory epithelia was performed in superfused

carboygenated modified Ringer’s solution, with odor delivery in liquid phase controlled

by a solenoid valve system (Warner Instruments); image alignment was achieved using

114

custom feature-tracking and homography algorithms. Human adenoviruses were

generated by the UNC Viral Core facility. See supplementary information for details

regarding experimental methods, reagent sources, and statistical procedures.

Mice:

GCD-IRES-tauGFP and OMP-IRES-GFP mice were obtained from the Jackson

Laboratory (strain B6;129P2-Gucy2dtm2Mom/MomJ, stock number 006704 and strain

B6;129P2-Omptm3Mom/MomJ, stock number 006667, respectively). OR174-9-IRES-

tauGFP mice were obtained from the Axel Laboratory (Sosulski et al., 2011). Emx1-

IRES-Cre and GCaMP3 mice were obtained from the Jackson Laboratory (strain

B6.129S2-Emx1tm1(cre)Krj/J, stock number 005628 and strain B6;129S-

Gt(ROSA)26Sortm38(CAG-GCaMP3)Hze/J and stock number 014538, respectively). Unless

otherwise noted, all experiments on wild type mice were performed on 6-8 week old

C57/BL6 male mice (Jackson Laboratory). All mouse husbandry and experiments were

performed following institutional and federal guidelines and approved by Harvard

Medical School’s Institutional Animal Care and Use Committee.

Epithelial single cell dissociation:

Mice were euthanized with a lethal dose of Xylazine (~50 mg/mouse, Lloyd).

Using 1X phosphate-buffered saline (PBS, VWR) to keep the epithelium moist during

dissection, the head of the mouse was severed and the olfactory epithelium was rapidly

dissected and placed in PBS. The epithelium was then transferred to a round-bottomed

glass dish containing 1 mL of papain solution (one vial of Papain, (Worthington)

115

dissolved in 5 mL of Earle’s Balanced Salt Solution (Worthington) and then equilibrated

10 minutes at 37 °C, 5% CO2) and 100 µL of DNAse solution (one vial of DNAse

(Worthington) dissolved in 1 mL of Earle’s Balanced Salt Solution (Worthington)). Bone

was removed from the epithelium under a dissecting microscope (Leica MZ75) and the

resultant tissue was placed in a 5 mL Falcon tube (Becton Dickinson) with an additional

1 mL of Papain solution and 100 µL of DNA solution and rocked gently for 30 minutes at

37 °C. The tissue was then gently triturated with a 5 mL stripette 10-15 times and the

non-dissociated pieces of tissue were allowed to sediment by gravity for approximately

3 minutes. The supernatant was transferred to a fresh 5 mL Falcon tube and

centrifuged 5 minutes at 300 x g. The supernatant was decanted, the cells were

washed twice with 5 mL DMEM, high glucose, minus glutamine (Life Technologies)/10%

heat inactivated Fetal Bovine Serum (FBS, Life Technologies) and resuspended in 1 mL

DMEM/10% FBS to use for experimentation.

FACS:

Olfactory epithelial cells were dissociated as described above. To label

hematopoetic cells, cell suspensions were incubated with PE-Cy5-conjugated rat anti-

mouse CD45 antibody (1:2000, BD Pharmingen) for one hour at room temperature and

then were washed twice with 1 mL DMEM/10% FBS. The cells were then incubated

with Hoescht to allow the differential labeling of dead cells, before being placed on ice

until they were sorted. Cells were sorted using either an Astrios EQ (Beckman Coulter,

Fort Collins, CO) or a FACSAriaII (Becton Dickinson, San Jose, CA). Following the

exclusion of dead cells and CD45 positive immune cells, GFP positive cells were then

116

sorted to >99% purity by setting gates based on the fluorescent profile of cells derived

from wild type, non-fluorescent mice. Cells were sorted directly into Trizol (Invitrogen)

prior to RNA extraction (see below).

RNA isolation:

Fluorescent cells were sorted directly into 750 ul Trizol (Invitrogen) and samples

were then vortexed to homogeneity before being placed on ice for the duration of the

FACS session. All subsequent RNA work was performed in a RNA-dedicated PCR

workstation (Air Clean Systems) that was UV-irradiated for 20 minutes prior to use and

then was subsequently decontaminated with RNAseZap wipes (Ambion). Samples were

passed through a 1 mL insulin syringe (Westnet) four times and the samples were

allowed to sit for 5 minutes at room temperature. 200 µL of chloroform (Sigma) was

then added to each sample, and the tubes were shaken vigorously by hand for 15

seconds and then placed at room temperature for 3 minutes. Samples were centrifuged

at 12,000 x g for 15 minutes at 4 °C. The aqueous phase was removed to a nuclease

free microcentrifuge tube (Ambion) and 10 ug of RNAse-free glycogen (Life

Technologies) was added as a carrier along with 500 ul of 100% 2-propanol (Sigma).

The samples were inverted several times to homogenize and incubated for 10 minutes

at room temperature. The samples were then centrifuged at 12,000 x g for 10 minutes

at 4 °C and the supernatant was removed with a pipette. The RNA pellet was washed

with 1 mL of 75% ethanol and then centrifuged for five minutes at 4 °C at 7,500 x g. The

supernatant was pipetted off, the pellet was allowed to air dry for five minutes and was

117

resuspended in 5-10 µL of nuclease free water (Ambion) and stored at -80 °C until

further use.

cDNA library generation:

cDNA libraries were generated from isolated RNA using a modified version of the

Smart-Seq2 protocol that was optimized for reliability and sensitivity using “spike” RNAs

(Ambion) (Picelli et al., 2014). Briefly, 0.6 µL of RNA was added to 8 µL lysis buffer (1.6

mM dNTPs, 2.5 µM oligo(dT)-VN primer (5’-

AAGCAGTGGTATCAACGCAGAGTACT30VN-3’, where ‘N’ is any base and ‘V’ is A, C,

or G), 0.2 ul RNAse inhibitor (Ambion), 0.1% TX-100 v/v) in a nuclease free PCR tube

and the samples were incubated in a Master Cycler ProS thermocycler (Eppendorf) for

3 minutes at 72 °C before being held at 4 °C for > 1 minute. Next, 11.4 µL of reverse

transcriptase mix (10 µL Superscript II (Invitrogen), 5 µL RNAse inhibitor (Ambion), 40

µL 5X first strand buffer (Invitrogen), 10 µL 0.1M DTT (Invitrogen), 8 µL 5 M Betaine,

37.8 µL nuclease free water (Ambion), 1.2 µL 1M MgCl2, 2 µL of TSO (5’-

AAGCAGTGGTATCAACGCAGAGTACATrGrG+G-3’, which contains two

riboguanisines (rG) and one LNA-modified guanisine (+G) to facilitate template

switching) was added to each sample and mixed by pipetting. The samples were

incubated in the thermocycler for 90 minutes at 42 °C and then were subjected to 10

cycles of 2 minutes at 50 °C followed by 2 minutes at 42 °C, one 15 minute incubation

at 70 °C, and a final hold at 4 °C lasting at least 1 minute. Next the samples were

amplified by adding 30 µL of PCR reaction mix (300 µL Kappa HiFi HotStart ReadyMix

(2X KAPA Biosystems), 6 µL ISPCR primer (5’-AAGCAGTGGTATCAACGCAGAGT-3’),

118

and 54 µL nuclease free water). The reactions were mixed by pipetting and incubated at

98 °C for 4 minutes followed by 24 cycles of 98 °C for 20 seconds, 67 °C for 15

seconds, and 72 °C for 6 minutes. A final incubation was performed at 72 °C for 5

minutes before the samples were placed at 4 °C for > 1 minute. The resultant cDNA

was isolated using an Agencourt Ampure XP kit (Beckman Coulter) following the

manufacturer’s instructions and was subsequently stored at -80 °C until further use.

RNA sequencing:

cDNA libraries were sheared to a size of approximately 250 bases using a

Covaris S2. The sheared cDNA was run on a Wafergen Apollo machine using the Kapa

Genomic Library Construction Kit (Kapa Biosystems, KK8234). Ten cycles of PCR were

run after the samples were removed from the robot and the reactions were cleaned

using magnetic beads. Quality control was performed using an Agilent 2200 Tape

Station with a D1000 High Sensitivity Tape with ladder provided. The morphology and

overall concentration of the samples were assessed and those samples that passed a

concentration cutoff were subjected to qPCR to more accurately determine

concentration. qPCR was run with SYBR green using the KAPA SYBR FAST Universal

2X qPCR Master Mix reagent (Kapa Biosystems, KK4602) and primers complementary

to the P5 and P7 regions of the adapter sequences. Serial dilutions of PhiX were used

to generate a standard curve, which was, in turn, used to determine the concentration of

cDNA in each sample prior to sequencing. Samples were sequenced on an Illumina

2500 in rapid mode on a single lane of SR50.

119

Alignment and differential expression:

RNA-sequencing reads were processed using the RNA-seq pipeline

implemented in version 0.8.3a-9483413 of the bcbio-nextgen analysis project. Briefly,

poor quality bases with PHRED scores less than five (Macmanes, 2014), contaminant

adapter sequences, and polyA tails were trimmed from the ends of reads with cutadapt

version 1.4.2, discarding reads shorter than twenty bases. A STAR (Dobin et al., 2013)

index was created from a combination of the Mus musculus version 10 (mm10) build of

the mouse genome and the Ensembl release 75 gene annotation. Trimmed reads were

aligned to the STAR index, discarding reads with ten or more multiple matches to the

genome. Quality metrics including mapping percentage, rRNA contamination, average

coverage across the length of the genes, read quality, adapter contamination and others

were calculated using a combination of FastQC, RNA-SeQC (DeLuca et al., 2012), and

custom functions from bcbio-nextgen and bcbio.rnaseq (available upon request). Read

mapping to genes were counted using featureCounts (Liao et al., 2014) version 1.4.4,

excluding reads mapping multiple times to the genome and reads that could not be

uniquely assigned to a gene. Counts were normalized and differential expression

between cell types was called at the level of the gene using DESeq2 (Love et al., 2014)

version 1.6.3. To identify gene families that were expressed in GC-D cells, the data set

was filtered to identify transmembrane protein-encoding genes that were at least ten-

fold enriched in GC-D cells with an adjusted p-value of < 0.05 after correcting for

multiple comparisons using the methods described by Benjamini and colleagues

(Benjamini et al., 2001). Despite using these stringent filtering criteria, with very large

gene families one might still observe the expression of an occasional family member by

120

chance and thus the data were filtered further to only consider gene families in which at

least three members were ten-fold enriched with an adjusted p-value of < 0.05. From

this list of gene families, the Ms4a genes were selected for further study as they

exhibited significant molecular diversity to encode chemoreceptors (see Figure 3.4) and

had relatively uncharacterized function.

Nanostring:

10,000 GFP positive cells from Gucy2d-IRES-GFP or Omp-GFP mice were

sorted into Trizol and the RNA was isolated as described above. Three biological

replicate RNA samples were hybridized to Nanostring probes using nCounter Elements

reagents according to the manufacturer’s specifications. The protocol was modified to

perform the hybridization step at 67 °C for 48 hours to maximize the detection of low

abundance transcripts. RNA molecules that hybridized to probe were captured and

quantified using an automated Nanostring prep station following the manufacturer’s

instructions. The resultant data were analyzed using nSolver software. Briefly, the

average number of detected molecules for six internal negative control probes (whose

complementary sequences are not present in the mouse genome) was used to

calculate a rate of non-specific hybridization. After subtracting the amount of binding

resulting from non-specific interactions, the number of molecules of each RNA transcript

found in GC-D samples and OMP samples was compared using Student’s t-test. See

Table 3.2 for probe sequences used in these experiments.

121

Table 3.2. Probe sequences used in Nanostring experiments

Ms4a1 GCAACCTGCTCCAAAAGTGAACCTCAAAAGGACATCTTCACT

GGTGGGCCCCACACAAAGCTTCTTCATGAGGGAATCAAAGG

CTTTGGGGGCTGTCCAA

Ms4a2 ACAGAAAATAGGAGCAGAGCAGATCTTGCTCTCCCAAATCCA

CAAGAATCCTCCAGTGCACCTGACATTGAACTCTTGGAAGCA

TCTCCTGCCAAAGCAG

Ms4a3 CCAGGCTTTCAAGGGTTGCCAATCTTCACCGTCACCTGATGT

CTGCATTTCCCTGGGTTCCTCATCAGATGGCCTGGTGTCTTT

AATGCTGATTCTCACC

Ms4a4a AACCCAAAATCCTTGGGATTGTGCAGATTGTAATCGCCATCA

TGAACCTCAGCATAGGAATTATGATGATAATTGCCACTGTGT

CGACCGGTGAAATACC

Ms4a4b CCTAGGATATTAACACTTCATTGCACTGGCTTTTGAGGTGAA

TATTAGATTTACTGTAAGTATGTAAGTCAAGCACTTATTAGGT

CAACAACACTTCAAC

Ms4a4c TGGCAAATCTATCTTCTGAACCACTCATTTCTGTGGTCTTAAT

GGCTCCAATTTGGGGACCAATAATGTTCATTGTCTCAGGATC

CCTGTCAATTGCAGC

122

Table 3.2 (Continued). Probe sequences used in Nanostring experiments

Ms4a4d ACAACTGGCACTACCATCGTGGTGAAAACCCAGCTCAAGCAT

ACCCACAAATAGAGTCCCACATCGAAACTCCACCACATTACT

CAAGGATACTGTTTCT

Ms4a5 TGAATTTACTTAGTGCTCTGGGAGCAGCAGCTGGAATCATTC

TCCTCATATTTGGCTTCCTTCTAGATGGGGAATTCATCTGTG

GCTATTCTCCAGATGG

Ms4a6b AAACAAAACTAAATACCACAAAAACAAATGGAACTATACCGC

AGAAGATATGTCTTCATGATAATGCAGAAATTCCAACCATCAC

AGGGTAGCAATGCTT

Ms4a6c CATGATTCCACAGGTAGTGACCAATGAGACCATCACAACGAT

TTCACCAAATGGAATCAACTTTCCCCAAAAAGACGAGTCCCA

GCCTACCCAACAGAGG

Ms4a6d AGTTTGGCTGCTTTAGAGCCTGCCTTGCAGCAATGTAAGCTG

GCTTTCACACAACTAGACACAACCCAAGATGCTTATCATTTCT

TTAGCCCTGAGCCAT

Ms4a7 GCCTCCAATGTAGCAAGCTCTGTTGTTGCCGTCATTGGCCTC

TTCCTCTTCACCTATTGTCTGATAGCCCTGGGGAGTGCTTTC

CCACACTGTAACTCAG

Ms4a8a TGTCACTACAACAATCCAGGTGTGGTCATTCCAAATGTCTAT

GCAGCAAACCCAGTGGTCATCCCAGAACCACCAAACCCAAT

ACCAAGTTATTCCGAAG

123

Table 3.2 (Continued). Probe sequences used in Nanostring experiments

Ms4a10 CCTAAGACCTCTCTGAAGGTTCTCTGTGTGATAGCCAACGTT

ATCAGCTTGTTCTGCGCACTGGCCGGCTTCTTTGTCATTGCC

AAGGACCTCTTCCTGG

Ms4a13 TTTCATGGCTGCTAACACCTGATGTAGGTGCCCATGAGATTC

CCATATAACAAGGCACACCTCATGCATTTTGTGCAAAAGGAA

ATTCACAACAAGGTGA

Ms4a15 GTGGGAAATCTTGGCTTCGCAGAGGTTTCGGAGGTTTGTCTT

CAAGATCATTAAGCACGGAGAACTCAGAATGTTCCAGAATAG

ACTGGCATTTCAGAGG

Ms4a18 GAATTCATCCTCACCTGCATAGCCTCACATTTTGGATGCCAG

GCTGTCTGCTGCGCCCATTTTCAGAACATGACAATGTTCCCA

ACCATATTTGGTGGCA

Pde1c GCTGTAATCGATGCATTGAAGGATGTGGATACGTGGTCCTTC

GATGTCTTTTCCCTCAATGAGGCCAGTGGAGATCATGCACTG

AAGTTCATTTTCTATG

Adcy3 CAACAACGGCGGCATCGAGTGTCTACGCTTCCTCAATGAGA

TCATCTCTGATTTTGACTCTCTCCTGGACAATCCCAAATTCCG

GGTCATCACCAAGATC

124

Table 3.2 (Continued). Probe sequences used in Nanostring experiments

Cnga2 GCTTGTGGATAATGGAGATCATGTGGGTTGAATTTCTAAGAG

CGTGACCTCCTAAGTCTCACAAGGAATCAGAGAATAGCTAAA

TTGTCCTTCCTGAGGC

Actb CAGGTCATCACTATTGGCAACGAGCGGTTCCGATGCCCTGA

GGCTCTTTTCCAGCCTTCCTTCTTGGGTATGGAATCCTGTGG

CATCCATGAAACTACAT

Gapdh AGGTTGTCTCCTGCGACTTCAACAGCAACTCCCACTCTTCCA

CCTTCGATGCCGGGGCTGGCATTGCTCTCAATGACAACTTTG

TCAAGCTCATTTCCTG

Pde2a CCACTAGCTTCTCTTCTGTTTTGTTCCCTATGTGTCGTGGGT

GGGGGAGGGGGCCACCTGCCTTACCTACTCTGAGTTGCCTT

TAGAGAGATGCATTTTT

Car2 TGCCCAGCATGACCCTGCCCTACAGCCTCTGCTCATATCTTA

TGATAAAGCTGCGTCCAAGAGCATTGTCAACAACGGCCACTC

CTTTAACGTTGAGTTT

Golf ATCGAAGACTATTTCCCGGAGTATGCCAATTATACTGTCCCT

GAAGATGCAACACCAGATGCGGGAGAAGATCCCAAAGTTAC

AAGAGCAAAGTTCTTTA

Emx1 CAGGCAAGCGACGTTCCCCAGGACGGGCTGCTTTTGCACGG

GCCCTTCGCACGCAAGCCCAAGCGGATTCGCACAGCCTTCT

CGCCCTCGCAGCTGCTGC

125

Multi species alignment of Mus musculus MS4A proteins:

FASTA format sequences of the indicated Mus musculus MS4A proteins were

downloaded from the NCBI protein database and aligned using the PRALINE sequence

alignment program on the Centre for Integrative Bioinformatics VU website using the

default settings. Amino acid conservation across family members was scored using the

PRALINE default settings where the least conserved amino acids were given a 0 score

and the most conserved amino acids were assigned a 10 (Simossis and Heringa, 2005).

TOPCONS was used to determine the predicted topology of the MS4A family member

that was used on the top line of the alignment. All topographical representations were

generated using the Protter program and manually entering the topographical

orientation of the MS4A protein as predicted by TOPCONS.

Phylogenetic and selection analyses with MS4A genes:

MS4A sequences were retrieved from both Ensembl and NCBI databases and

imported into Geneious v8 (Biomatters Ltd). We chose 37 representative taxa from all

the major mammalian lineages (see Table 3.3 for full list). When a gene had more than

one predicted isoform, the sequence that contained the longest open-reading frame was

selected. Coding DNA sequences were translated, aligned with MAFFT v7.017 (Katoh

and Standley, b) using the E-INS-i algorithm, the BLOSUM80 scoring matrix, and a gap-

opening penalty of 1. Sequences were then back-translated into codons. For

phylogenetic reconstruction of the multigene family tree, the OpenMPI version of

MrBayes v3.2.1 (Ronquist et al., 2012) and the GTR+I+G model as determined by

jModelTest 2.1.7 (Darriba et al., 2012) were used. The final dataset consisted of 411

126

Table 3.3. List of taxa used for phylogenetic reconstructions

Group Order Family Species Common name

Monotremata Ornithorhynchidae

Ornithorhynchus

anatinus platypus

Marsupalia Dasyuridae Sarcophilus harrisii Tasmanian devil

Placentalia

Soricomorpha Soricidae Sorex araneus common shrew

Carnivora

Mustelidae Mustela putorius furo ferret

Ursidae

Ailuropoda

melanoleuca giant panda

Canidae Canis lupus familiaris dog

Chiroptera Vespertilionidae Myotis lucifugus little brown bat

Perissodactyla Equidae Equus ferus caballus horse

Artiodactyla

Suidae Sus scrofa domesticus pig

Bovidae Bos taurus cow

Ovis aries sheep

Cetacea Delphinidae Tursiops truncatus bottlenose dolphin

Lipotidae Lipotes vexillifer baiji

Macroscelidea Macroscelididae Elephantulus edwardii elephant shrew

Afrosoricida Tenrecidae Echinops telfairi

lesser hedgehog

tenrec

Hyracoidea Procaviidae Procavia capensis rock hyrax

Cingulata Dasypodidae

Dasypus

novemcinctus

nine-banded

armadillo

127

Table 3.3 (Continued) List of taxa used for phylogenetic reconstructions

Primates

Galagidae Otolemur garnetti

northern greater

galago

Callitrichidae Callithrix jacchus

common

marmoset

Hominidae

Gorilla gorilla gorilla

Pan troglodytes chimpanzee

Homo sapiens human

Scandentia Tupaiidae Tupaia chinensis tree shrew

Lagomorpha Ochotonidae Ochotona princeps American pika

Leporidae Oryctolagus cuniculus rabbit

Rodentia

Heteromyidae Dipodomys ordii kangaroo rat

Dipodidae Jaculus jaculus jerboa

Muridae Mus musculus mouse

Rattus norvegicus rat

Cricetidae

Cricetulus griseus Chinese hamster

Mesocricetus auratus golden hamster

Microtus ochrogaster prairie vole

Sciuridae Spermophilus

tridecemlineatus

thirteen-lined

ground squirrel

Bathyergidae Heterocephalus glaber naked mole-rat

Chinchillidae Chinchilla lanigera chinchilla

Octodontidae Octodon degus degu

Caviidae Cavia porcellus guinea pig

128

sequences and 447 characters corresponding to sites present in at least 75 percent of

the aligned sequences. For individual gene tree reconstructions and evolutionary

analyses, sequence subsets were extracted based on their group membership as

predicted based on the multigene family tree. Sequences were then realigned

corresponding to each subset as above, the resulting alignments were trimmed to

remove positions that contained gaps in the majority of sequences. The phylogenetic

reconstruction was carried out using the OpenMPI version of RAxML v8 (Stamatakis,

2014).

To identify branches under episodic positive selection, the random-effects

likelihood branch-site method (BS-REL) (Kosakovsky Pond et al., 2011) was used as

implemented in the HyPhy package (Pond et al., 2005). The branch-site models allow

the nonsynonymous to synonymous substitution rate ratio ω (dN/dS) to vary both among

amino acid sites in the protein and across branches on the tree to detect positive

selection affecting specific sites along particular lineages (Anisimova and Yang, 2007).

We identified evidence for site-specific positive selection in MS4A homologs using the

codeml program in the PAML v4.8 software package (Anisimova and Yang, 2007). In

brief, we compared different site models, in which the evolutionary rate ω is allowed to

vary among sites. Here, we focused on comparison of model pairs: M1a (neutral; codon

values of ω fitted into two discrete site classes between 0 and 1) versus M2a (positive

selection; similar to M1a but with one additional class allowing ω>1); M7 (neutral; values

of ω following a beta distribution with ω=1 maximum) versus M8 (positive selection;

similar to M7 but with one additional class allowing ω>1); and M8a (neutral; similar to

M7 but with one fixed class with ω=1) versus M8. Multiple starting values of ω were

129

used, and either the F3x4 or F61 model of codon frequencies. To evaluate whether the

models allowing positive selection provided a significantly better fit to the data,

likelihood ratio tests were used. Notably, the M1a-M2a comparison is more stringent

and can lack power to detect signatures of diversifying selection compared to the M7-

M8 models, which impose less constraints on the distribution of ω. Finally, the M8a vs.

M8 comparison can be used to contrast the potential role of reduced purifying selection

(or relaxation) versus positive selection. When the null model is rejected, the empirical

Bayes procedure was implemented under model M8 to identify sites under positive

selection (posterior probability ≥ 0.90). To identify sites that have experienced purifying

selection (posterior probability ≥ 0.90), the Fast Unconstrained Bayesian

AppRoximation (FUBAR) (Murrell et al., 2013) was used as implemented in the HyPhy

package. Consensus topology predictions were made using a standalone version of

TOPCONS2.0 (Tsirigos et al., 2015). All computational analyses were run on the

Odyssey cluster supported by the FAS Division of Science, Research Computing Group

at Harvard University.

Plasmids:

pCI-MOR9-1 was a gift from Hiroaki Matsunami (Addgene plasmid # 22331)

(Saito et al., 2009). pGP-CMV-GCaMP6s was a gift from Douglas Kim (Addgene

plasmid # 40753) (Chen et al., 2013). The GNAI15 expression plasmid was kindly

provided by Steve Liberles. DNA sequences encoding mCherry-MS4A were cloned

into the tetracycline inducible mammalian expression plasmid, pcDNA5-FRT-TO using

standard molecular biology methods.

130

Heterologous expression of MS4A proteins:

Flp-In-T-Rex HEK293 cells (Invitrogen R78007) were maintained in complete

media (DMEM, high glucose, no glutamine (Life Technologies) supplemented with 10%

tetracycline free Fetal Bovine Serum (Clontech), penicillin, streptomycin, and glutamine

(Life Technologies)) with 5% CO2 in a 37 °C humidified tissue culture incubator

(NuAire) on 10 cm tissue culture plates (BD #430167). Approximately four hours prior to

transfection, cells were washed once with 10 mL plain DMEM and then incubated with 2

mL of Trypsin-EDTA solution (ATCC) for approximately 3 minutes at 37 °C. 10 mL of

complete media was then added to the plate and the cells were triturated 5-10 times

vigorously to generate single cell suspensions. After centrifugation at 300 x g for 5

minutes, the supernatant was aspirated and the cells were resuspended in complete

media before they were plated on Round German Glass 15 mm coverslips (Bellco

Biotechnology) in 12 well plates, which had been incubated for at least 24 hours with

0.2 mg/mL poly-d-lysine hydrobomide (Sigma) before being washed twice with ddH20.

After four hours, cells were transfected with calcium phosphate. For each coverslip, a

50ul reaction mix consisting of 250 mM CaCl2, approximately 2.5 ug GCaMP6s

encoding plasmid, and 1 µg MS4A encoding plasmid. This mix was homogenized by

pipetting 4 times. To this reaction mix, 50 µL of 2X HeBs (274 mM NaCl, 10 mM KCl,

1.4 mM Na2PO4.7H20, 15 mM D-glucose, 42 mM Hepes (free acid)) pH 7.04-7.10 was

added in a swirling motion from the bottom of the tube and bubbled briefly with air. This

mixture was incubated for five minutes at room temperature, pipetted once to mix, and

added to the well in a drop-wise manner. The 12 well plate was then shaken 5 times

131

along each major axis before being placed in the incubator. Approximately one hour

later tetracycline (Sigma #T7760) was added to a final concentration of 1 µg /mL and

cells were allowed to express MS4A proteins overnight.

MS4A6C surface expression:

HEK293 cells were transfected with a plasmid encoding GCaMP6s and either a

control plasmid or a plasmid encoding a mCherry-MS4A6C fusion protein as described

above. 16 hours following transfection, 1 µg of purified anti-MS4A6C antibody was

added to the cells. The cells were incubated with antibody for 1 hour at room

temperature in the dark, and then washed 3X with complete media. Cells were

subsequently fixed with 4% paraformaldehyde/1X PBS for ten minutes and then washed

three times with 1X PBS. Cells were re-blocked with PBS/0.3% Triton X-100/5%

donkey serum for 30 minutes and then incubated with Alexa633 goat anti-rabbit

secondary antibody (1:300) for 45 minutes. The fixed and stained cells were washed

three times with block, and the coverslips were mounted on slides using VECTASHIELD

Mounting Medicum with DAPI (Vector laboratories).

Odors:

Steroids were purchased from Steraloids (Newport, Rhode Island). Additional

compounds were purchased from Sigma and were obtained at the highest purity

possible.

132

In vitro odor screen:

HEK293 cells were co-transfected as described above with a plasmid encoding

GCaMP6s and either a plasmid encoding one of the MS4A proteins or plasmids

encoding the odorant receptor MOR9-1 and the G protein GNAI15 (which couples the

exogenous GPCR to intracellular calcium stores, see Ukhanov et al., 2014). Coverslips

were washed with Ringers solution and then secured in a perfusion/imaging chamber

(Warner Instruments) using High Vacuum Grease, Dow Corning (VWR). Ringers

solution was constantly perfused over the cells at a rate of ~10 mL/minute using an 8-

channel valve-controlled gravity-driven perfusion system (Warner Instruments) and

images were acquired using the MetaFluor software package using an Olympus IX83

microscope, a Sutter Lambda DG4 Light System with a Xenon arc lamp, and an Andor

Neo 5.5 sCMOS camera. A 20X NA 0.45 air objective (Olympus) was used to acquire

images at ~ 0.67 Hz at 3 x 3 pixel binning to reduce the exposure times required to

obtain images thereby limiting photobleaching and phototoxicity. Each imaging epoch

consisted of 900 seconds. For the first 300 seconds images were acquired but not

saved as a significant amount of GCaMP6s signal decay occurred within this window.

Subsequently the experiment consisted of 150 seconds of Ringers, valve switch and 75

seconds of Ringers (from a separate reservoir to mimic odor stimulation), valve switch

and 150 seconds of Ringers, valve switch and 75 seconds of odorant, followed by a

final valve switch and 150 seconds of Ringers. Odorants were delivered as mixtures of

chemicals diluted in Ringers solution such that each individual constituent was present

at 10 µM final concentration. Each coverslip was exposed to a single imaging epoch to

control for response adaptation. To assess the time required to clear dead volumes and

133

to estimate mixing delays, 100 nM RhodamineB dye was flowed instead of odor and

dye saturation kinetics were determined. In Figures 3.6A and S4B stimulus bars begin

when odor is estimated to reach 90% saturation.

To identify monomolecular compounds that activate HEK293 cells expressing

individual MS4A family members, several odorant mixes that elicited statistically

significant responses in cells expressing a particular MS4A protein were selected for

deconvolution analysis. The final concentration of each odorant in the deconvolution

analysis was 50 µM except for the polyunsaturated fatty acid constituents, which were

delivered at 10 µM, as at higher concentrations some of these molecules non-

specifically activated HEK293 cells.

Analysis of in vitro screen data:

Analysis was performed using fully automated custom scripts in iPython that

employed NumPy, SciPy, Pymorph, Pandas, and Mahotas source code packages. Each

fluorescent image of the time series was locally smoothed and downsampled 3-fold in

both dimensions to facilitate subsequent processing. The downsampled image stack

was segmented into cell areas using a watershed transform, which were then used to

extract the fluorescence values of individual cell areas over time. The resulting traces

were baseline-corrected using a wide (100 frames) moving median filter to remove long-

timescale drifts such as fluorescence decay due to photobleaching and smoothed using

a local (9 frames) moving average filter. A total of seven coverslips from the mixture

screen and five coverslips from the deconvolution screen were removed from the

analysis for failing to meet quality control.

134

Processed traces were normalized by calculating an average fluorescence value

across the period prior to odor stimulation and then dividing the entire trace by this

value (thereby generating a dF/F trace). Responses were identified as fluorescent

peaks within a twenty frame window centered at the time point at which maximal odor

concentration occurred (determined empirically using RhodamineB dye) that were five

standard deviations above the noise observed in the baseline period. The spontaneous

activity response rate (determined as responses observed during twenty frames of the

Ringers only period) was subtracted from the odor-evoked response rate; this rate did

not change based upon the specific position of the window chosen in the trace (as long

as it did not overlap with odor delivery). The proportion of cells co-expressing a

chemoreceptor and GCaMP6s that responded to a given stimulus was compared by Z-

test to the proportion of cells expressing only GCaMP6s that responded to the same

stimulus. Only response tallies with a P-value of < 0.05 following FDR correction for

multiple comparisons are depicted.

In vitro functional experiments with focal stimulus delivery pencil:

To characterize ligand-induced response kinetics with greater temporal precision

than in the bulk calcium assay, HEK293 cells were transfected with a plasmid encoding

the fast version of GCaMP6 (GCaMP6f) and plasmids encoding either an MS4A or

MOR9-1/GNAI15. The cells were perfused with Ringer’s solution and odorants were

delivered in liquid phase with a six channel gravity-controlled perfusion manifold

(Warner Instruments) through a custom-made “stimulus pencil” made of quartz tubing

focally positioned relative to the field of view. Images were acquired as described for the

135

in vitro odor screen with an imaging rate of 2 Hz. The delay between valve switch and

odor delivery was determining using RhodamineB dye as in the bulk calcium assay.

Dose response curves and EC50 calculations:

Dose response curves were determined as in (Mainland 2015). HEK293 cells co-

expressing GCaMP6f and MOR9-1/GNAI15 or an MS4A were odor stimulated with a

focal stimulus delivery pencil as described earlier. Each field of view was stimulated

using the stimulus pencil with odorant spanning six log orders of odorant (from 10 nM to

1 mM) starting with the lowest concentration. Fluorescent traces were extracted and

normalized as described above for the in vitro odor screen. Because of the tighter

stimulus control afforded by this configuration, cells were considered to have responded

if they exhibited fluorescent peaks greater than 4 standard deviations above baseline

within a 15 second window centered around the time of maximal odor responses. A

delay of 150 seconds was included between stimulus presentations. The cumulative

fraction of cells that had responded by each odor concentration was determined for both

receptor-expressing and control cells. To account for non-specific activation that might

occur at higher concentrations of odorant the fraction of control cells (expressing

GCaMP only) that was activated was subtracted from the fraction of receptor-

expressing cells and sigmoidal dose response curves were then fit to the receptor data.

Between four and twenty coverslips for both control and receptor-expressing cells were

imaged to construct each dose response curve and each data point represents the

mean +/- SEM of the relative cumulative fraction of cells that responded to that odor

concentration.

136

Analysis of calcium transient temporal dynamics:

Analysis was performed by applying fully automated custom iPython scripts to

extracted fluorescence traces. Response onset, half-rise time and peak time were

identified using a peak and trough finding package in SciPy

(SciPy.signalfind_peaks_cwt) (Du et al., 2006). Only cells that had response peaks

greater than 4 standard deviations above baseline were included for analysis. Two-

tailed Student’s T-tests were run on the distributions of data and were Bonferroni

corrected for multiple comparisons.

Requirement of extracellular calcium for MS4A ligand responses:

Experiments were performed similarly to those described using the stimulus

pencil delivery system. Cells were either perfused with Ringers supplemented with 1

mM calcium chloride (plus calcium) or 1 mM EGTA (minus calcium) to chelate calcium.

The percent of cells that responded was determined as described above.

RNAscope Fluorescent In Situ Hybridization:

Fluorescent in situ hybridization was performed on dissociated olfactory epithelial

cells adhered to glass coverslips. Nasal epithelia from 8-12 week old C57/BL6 male

mice, OR174-9-IRES-tauGFP mice, or GC-D-IRES-tauGFP mice were dissociated with

the same method used for FACS, except that they were incubated in papain + DNAse I

for 90 minutes. The final cell pellet from a single mouse epithelium was resuspended in

900 µL of DMEM + 10% FBS, and 75 µL was placed on each of twelve 12 mm #1 glass

137

coverslips (Bellco Biotechnology #1943-10012) that had been coated with 30 µL of 5

mg/mL poly-D-lysine (Sigma Aldrich #P6407), dried, and washed three times with

deionized water. Cells were allowed to settle on coverslips for 30 minutes at room

temperature, rinsed once with 1X PBS, and fixed with 4% paraformaldehyde (PFA)/1X

PBS (diluted from 20% PFA, Electron Microscopy Sciences #15713) for 30 minutes at

room temperature. Coverslips were washed three times with 1X PBS and passed

through a dehydration series of 50%, 70%, and 100% ethanol (v/v in water) for 5

minutes each. The last wash was replaced with fresh 100% ethanol and cells were

stored at -20 °C overnight.

On the second day, coverslips were stained for specific mRNA targets by the

RNAscope protocol (Advanced Cell Diagnostics, RNAscope Multiplex Fluorescent

Assay For Fresh Frozen Tissues User Manual rev. 20121003). Several modifications to

the RNAscope protocol were made to combine FISH for low abundance targets with

immunohistochemistry: after rehydration, coverslips were treated with protease

(“Pretreat 4”) diluted 1:30 in 1X PBS; RNAscope target probes, which are supplied at

50X concentration, were diluted to final 2X concentration in blank probe solution;

hybridization lasted for 4 hrs at 40 °C in the HybEZ oven (Advanced Cell Diagnostics

#310010); and for experiments using type C1 and C2 probes (see below) the

fluorescent reagent Amp4-altB was used, whereas for experiments using C1 and C3

probes we used Amp4-altC (Advanced Cell Diagnostics #320850).

To combine FISH with immunostaining, at the end of the RNAscope protocol,

instead of being counterstained and mounted, the coverslips were rinsed in 1X PBS

twice and blocked in 1X DAKO Serum-Free Protein Block (DAKO X0909) in PBS for 20

138

minutes at room temperature. Coverslips were then incubated with primary antibody

(rabbit anti-Car2 or chicken anti-GFP as below) in blocking solution for 1 hour at room

temperature, washed three times for 5 minutes in PBS, and stained with secondary

antibody in PBS for 30 minutes at room temperature. After three final washes in PBS

the coverslips were counterstained with ProLong Diamond Antifade Mountant + DAPI

(Life Technologies #P36961), fixed to slides with nail polish, and imaged by

epifluorescent or confocal microscopy as described below.

To quantify RNAscope signal, coverslips were viewed under epifluorescence with

a 63X/NA1.4 oil immersion objective lens (Zeiss Plan-Apochromat 63X/1.40 OIL DIC

440762-9904-000) and fluorescent puncta visible by eye were counted for each

immunopositive cell. In initial experiments puncta in random immunonegative cells were

also quantified and the background levels for MS4A probes were found to be ~ 1

punctum/5-10 cells (data not shown). In approximately 10% of experiments much higher

background (3+ puncta/cell) was observed, possibly due to sample drying. These

coverslips were discarded from the analysis. For the remainder of the analysis puncta

were quantified only in identifiable cells (i.e., GC-D/CAR2 or OR174-9 expressing cells.)

The target probes used in this assay are custom reagents designed by Advanced

Cell Diagnostics and are currently available in the ACD catalog. Mouse target probes

used are as follows: Ms4a1-c1 #318671, Ms4a2-c1 #438171, Ms4a3-c1 #438181,

Ms4a4A-c1 #427391, Ms4a4B-c1 #314611, Ms4a4C-c1 #426371, Ms4a4D-c1 #427401,

Ms4a5-c1 #318641, Ms4a6B-c1 #313801, Ms4a6C-c1 #314581, Ms4a6D-c1 #314591,

Ms4a7-c1 #314601, Ms4a8A-c1 #426361, Ms4a10-c1 #438151, Ms4a15-c1 #427381,

Olfr173-c1 #313771, Olfr151-c1 #431161, Olfr66-c1 #431171, Ms4a6C-c3 314581-C3,

139

Ms4a4B-c3 #314611-C3, Car2-c2 #313781-C2, Gucy2d-c2 #425451-C2, Pde2a-c3

#426381-C3.

MS4A Antisera:

Peptides derived from mouse MS4A protein sequences were synthesized by

Covance, Inc. (Denver, PA) as follows: 6C-pep: CKQSKELSLIEHDYYQ; 6D-pep:

SQNSKNKSSVSSESLC; 7-pep: HKREKTGHTYEKEDD; 4B-pep:

HQGTNVPGNVYKNHPC; 8A-pep: TAKSWEPEQERLTWC; 5-pep:

TTQEYQTTELTATAYNC. These peptides were coupled by terminal cysteine residues

to KLH and used to raise an immunoglobulin response in rabbits and guinea pig (6C-

pep only). Sera collected by Covance, Inc. from rabbits and guinea pig were tested for

ability to stain 293T cells expressing MS4A protein and GC-D cells (guinea pig anti-6C

and rabbit anti-6C antibodies yielded the same pattern of staining in vitro and in vivo);

the best lots of serum were purified by passing over protein A/sepharose (Life

Technologies #10-1042) and eluted with 100 mM Glycine, pH 2.7 to collect

immunoglobulin. A subset of these (anti-6C, 6D, and 7) were further affinity purified by

passing over resin cysteine-coupled to the respective antigenic peptides (Thermo

Scientific #44999) and fractions eluted with 100 mM glycine, pH 2.7 were tested for

greatest specificity on 293T cells. We also raised antisera against GC-D protein with the

peptide (Ac-QRIRTDGKGRRLAC), which was purified by passing over protein

A/sepharose and a peptide affinity column as above. For peptide competition

experiments, antibody concentration was estimated by reacting with protein assay dye

reagent (Bio-Rad #5000006) and measuring 595 nm absorbance with a Cary 60 UV-Vis

140

spectrophotometer (Agilent Technologies). Sample concentrations were estimated by

fitting to a line generated with IgG standards; most purified antibodies were between 1.0

and 5.0 mg/mL. Peptides were resuspended in 1X TBS at 1 mg/mL and mixed with

diluted antibodies in block at a 10 µg peptide: 1 µg antibody ratio; as the peptides are

~15 amino acid residues and IgGs are ~1500 residues total, it was estimated that this

gives ~1000-fold molar excess of peptide. Similar results were obtained with lower

peptide concentration (data not shown).

Primary antibodies/concentrations used were as follows: goat anti-Pde2a (1:50,

Santa Cruz Biotechnology sc-17227), chicken anti-GFP (1:1000, Abcam #ab13970),

rabbit anti-AC3 (1:100, Santa Cruz Biotechnology sc-588), rabbit anti-GC-D (serum

7444, affinity fraction 2, 1:1000), rabbit anti-Car2 (1:500, Abcam #191343), goat anti-

CD20 (1:50, Santa Cruz Biotechnology #sc-7735), rabbit anti-MS4A4B (serum 7512,

PAS fraction, 1:2000), rabbit anti-MS4A6C (serum 7277, affinity fraction 3, 1:500), anti-

MS4A6D (serum 7284, affinity fraction 3, 1:500), anti-MS4A7 (serum 7286, affinity

fraction 2, 1:500), anti-MS4A8A (serum 7508, PAS fraction, 1:2000), anti-MS4A5

(serum 7503, PAS fraction, 1:2000), guinea pig anti-MS4A6C (serum 7630, PAS

fraction, 1:1000), anti-S6 phosphoSerine240/244 (1:200, Cell Signaling Technologies

#22155).

Secondary antibodies/concentrations used were as follows: donkey anti-rabbit-

Cy3 (1:300, Jackson Immunoresearch, #711-167-003), donkey anti-rabbit-CF633

(1:300, Biotium #20125), donkey anti-goat-Alexa633 (1:300, Invitrogen #A21082),

donkey anti-goat-Alexa488 (1:300, Jackson Immunoresearch #705-546-147), donkey

141

anti-guinea pig-Alexa647 (1:300, Jackson Immunoresearch #706-606-148), donkey

anti-chicken-Alexa488 (1:300, Jackson Immunoresearch #703-545-155).

Cultured Cell Preparation for Immunostaining:

293T cells adhered to 12 mm coverslips and transfected with mCherry-MS4A

expression plasmids (see above) were fixed 24-48 hours post-transfection with 4%

PFA/1X PBS (Electron Microscopy Sciences) for 10 minutes at room temperature.

Coverslips were washed three times with 1X PBS and stored at 4 °C in 1X PBS until

immunostaining as described below.

Tissue Preparation for Immunostaining:

All preparation of nasal epithelia and olfactory bulbs was performed as follows,

except for assays involving phospho-S6: animals were euthanized as above and

olfactory epithelia were dissected out from the skull with olfactory bulbs attached and

fixed overnight in 4% PFA/1X PBS at RT. After washing 5 minutes with 1X PBS three

times, noses were decalcified overnight in 0.45M EDTA/1X PBS at 4 °C. Noses were

washed once more with 1X PBS, then sunk in 20% sucrose for 3 hrs at 4 °C, and finally

embedded in Tissue Freezing Medium (VWR #15146-025). 15 micron cryosections

were cut onto Superfrost Plus glass slides (VWR #48311703) and stored at -80 °C until

staining. For experiments involving anti-phosphoS6 staining, nasal epithelia were

instead fixed overnight in 4% PFA/1X PBS at 4 °C.

142

Immunostaining:

For experiments without anti-phosphoS6, cryosections on slides were removed

from the freezer, air dried for 10 minutes, and antigen retrieved by immersion for 10

minutes in a 98 °C solution of 10 mM Sodium Citrate, 0.1% Tween-20, pH 6.5. Slides

were rinsed in 1X Tris-buffered Saline (TBS: 50 mM Tris-Cl, 150 mM NaCl, pH 7.5) at

room temperature and blocked in 5% Normal Donkey Serum (Jackson Immunoresearch

#017-000-121)/0.1% Triton-X100/1XTBS for 30 minutes at room temperature. All

immunostaining was done at 4 °C overnight with antisera diluted in blocking solution.

On the following day, slides were washed three times for 10 minutes in 1X

TBS/0.1% TritonX100 at room temperature, and then incubated in secondary antibody

solution (antibodies listed above in blocking solution) for 45 minutes at room

temperature. Finally, slides were washed three times for 10 minutes in 1X TBS/0.1%

TritonX100, counterstained with Vectashield + DAPI (VWR #101098-044), and

coverslipped before confocal microscopic imaging.

For experiments involving anti-phosphoS6, staining was performed as above with

the following changes: there was no antigen retrieval step before blocking; blocking

solution was 1X PBS/0.1% TritonX100/3% Normal Donkey Serum/3% BSA; all washes

were done in 1X PBS/0.1% TritonX100; and the primary antibody solution included 50

nM of a phosphopeptide (“3P”) derived from the S6 sequence (Knight et al., 2012). This

peptide, which contains phosphorylated Serine235/236/240, is meant to compete away

nonspecific antibody binding to S6 protein not phosphorylated at its neuronal activity-

dependent sites.

143

Finally, assaying Ms4a6d-IRES-GFP-infected noses required costaining with two

rabbit primary antisera, one to MS4A6D and the other to phosphorylated S6. To prevent

cross-reactivity with secondary antibodies, phospho-S6 staining was performed first as

described above; after application of fluorescent secondary antibody, remaining sites on

the rabbit primary antibody were blocked by incubation with 5 ug/mL unconjugated

donkey anti-rabbit Fab fragments (Jackson Immunoresearch) in washing solution at

room temperature for 1 hour. After washing out unbound Fab fragments three times with

wash solution, tissue sections were stained with anti-MS4A6D overnight as described

above and detected with a CF633-conjugated secondary antibody (Biotium). No

fluorescence to indicate cross-reactivity was observed, and control experiments without

anti-MS4A6D antiserum showed that the later secondary antibody did not interact with

the Fab-blocked phospho-S6 primary antibody.

Explant calcium imaging:

Adult offspring of Emx1-IRES-Cre and ROSA-GCaMP3 mice were sacrificed and

dissected to expose the lateral olfactory epithelium, adjacent to the olfactory bulb. The

epithelium and adjacent skull (including the adjacent olfactory bulb) was embedded in a

custom-made perfusion chamber with 5% low-melt agarose (Invitrogen) made from

modified Ringer’s solution (115 mM NaCl, 5 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 25 mM

NaHCO3, 10 mM HEPES, pH 7.4). Small slits were cut anterior to the cul-de-sac regions

of exposed olfactory epithelium to allow fluid flow, and for the remainder of the

experiment the tissue was superfused with carboxygenated modified Ringer’s solution

(95% O2, 5% CO2). Odorants were delivered to the tissue in liquid phase via an 8-to-1

144

perfusion manifold controlled by solenoid valves (Warner Instruments) through a

custom-made “stimulus pencil” positioned near the olfactory epithelium. Each odorant or

mixture was diluted in 0.1% DMSO in modified Ringer’s solution to a concentration of

100 µM.

Cul-de-sac regions of the tissue containing GCaMP3+ cells were imaged through

the epithelial cartilage by standard multiphoton microscopy (Prairie Technologies) with a

Ti-sapphire laser (Coherent) at 965 nm through a 25X/NA1.05 water-immersion

objective lens (Olympus XL Plan N). 512x512 fluorescent images were acquired at 1

Hz, with each odor trial consisting of 20 seconds of Ringers, 10 seconds of odorant, and

finally 50 more seconds of Ringers. After imaging, a high resolution image of the field of

view was acquired to aid cell identification. As was done for the in vitro assays, the

delay time for odor delivery was assessed using RhodamineB dye; these experiments

revealed that the maximal lumenal concentration of dye at equilibrium was

approximately 10 percent of the initial dye concentration (as assessed via fluorescence

intensity). Image registration was accomplished using custom code, and publicly

available Python image processing and OpenCV packages. Time-series images were

aligned using a feature-based approach that is robust to regional fluorescence intensity

fluctuations over the course of the experiment. Images (for alignment purposes) were

first contrast enhanced to enable feature detection, and then all frames from a single

experiment were registered to a manually chosen target frame in a pairwise manner.

For each frame-target pair, positional features, typically corresponding to cell bodies

and blood vessels, were then automatically identified using Harris corner detection.

Corresponding features were then used to obtain an optimal homography (i.e.,

145

projective transformation) between frame and target, and this transform was then

applied to the raw images.

Cells from aligned movies of the raw images were identified using a semi-

automated approach. First, the centroids of putative cells (i.e., rounded and convex

GCaMP3+ objects, sometimes with nuclear exclusion of GCaMP3) in a time-series

average projection were manually specified. Cell masks were then generated using a

combination of morphological filtering and region-growing. Each cell mask was further

refined based on co-fluctuations in the fluorescence of pixels in the cell’s vicinity.

Independently covarying groups of pixels were first identified using nonnegative matrix

factorization. Pixels associated with the cell of interest were then assigned to the current

mask; pixels that correspond to adjacent cells were excluded. In the resultant binary

mask, each connected component therefore corresponded to a cell; fluorescence time-

courses for each cell were then obtained by averaging the pixels in each connected

component on a frame-by-frame basis.

Putative necklace cells were defined as cells within the cul-de-sacs that

responded to carbon disulfide with at least a 25% increase in average fluorescence and

did not respond to DMSO alone. A cell was then categorized as responding to an odor

stimulus if its average fluorescence increased at least 25% relative to the previous 10

frames and reached a peak after odor presentation, but not before. For display, plotted

traces were smoothed by convolution with a 3-frame rectangular window. For

generating the top panels of Figure 3.13A, the fluorescent signal following vehicle

stimulation was subtracted from the fluorescent signal following odorant stimulation, and

the resultant image was heat-mapped and overlaid onto a reference image using a

146

custom Python script. For the purposes of data representation these heatmaps are

shown as raw change in fluorescence; 100% on this fluorescence scale corresponds to

the full dynamic range of image intensity in this acquisition.

Confocal Microscopy and Image Processing:

Slides were imaged under a 63X/NA1.4 oil immersion planar apochromatic

objective lens (Zeiss). Digital images were acquired on a Zeiss LSM 510 Meta confocal

microscope (Harvard Neurodiscovery Imaging Center). Cyanine and Alexa fluorophores

were excited in sequence with an argon laser (488 nm line), a HeNe laser (543 nm),

and a second HeNe laser (633 nm). DAPI was imaged with a tuneable Coherent

Chameleon Laser at 740 nm in two-photon excitation mode. Emission was detected

with standard dichroic mirrors and filter sets. Using Imaris 8.1 software (Bitplane Inc.),

multi-channel z-stacks were maximum-intensity-projected into two dimensions and

passed through a median filter to remove debris much smaller than structures being

assayed.

Adenoviral Infection:

Six week old male mice (Jackson) were anesthetized with 100 mg/kg of

Ketamine, 10 mg/kg of Xylazine, and 3 mg/kg of Acepromazine. Infusions of 3~6 x 108

FFU of human adenovirus serotype 5 encoding Ms4a6c-IRES-hrGFP or Ms4a6d-IRES-

hrGFP in the right nostril of mice were performed as previously described (Holtmaat et

al., 1996). Mice were subjected to the odor exposure 6-8 days after injections.

147

Odor Exposure for Phospho-S6 Immunostaining and Quantification of Positive Cells:

8-10 week old C57/BL6 male mice (Jackson Laboratory) were group housed

overnight on a reverse light-dark cycle with 3-5 mice/cage. On the day of the

experiment, single mice were habituated in fresh cages for 3 hours in the dark. The odor

stimulus was then introduced into each cage as 100 µL of neat odorant blotted onto

Whatman paper in 10 cm petri dishes with slits cut into the lid. After 4 hours of odor

exposure, animals were sacrificed and their nasal epithelia were dissected and fixed as

described above.

To quantify phospho-S6 immunopositive cells, 10-15 images of Pde2a-positive

cell enriched cul-de-sacs were acquired with a confocal microscope (see above) for

each slide of odor-exposed olfactory epithelial sections. Laser and acquisition

parameters were held constant across each experiment. pS6 immunopositive cells were

counted manually from the resulting images; a cell was called positive if it showed

smooth pS6 fluorescent signal filling up the soma at levels visibly above adjacent tissue

and Pde2a-negative cells. Counts from each image were summed to give an estimate

of the proportion of activated necklace cells for each animal. In initial experiments with

each odorant, the person imaging and counting cells was blind to the identity of the

odorant. Because only sulfated steroids have a vehicle control (DMSO alone) these

data were analyzed separately from the data in Figure 3.14A.

The same method was used to quantify MS4A6C-ires-GFP and MS4A6D-ires-

GFP virus-infected cells, except that GFP+ cells were sparse and therefore called

positive for pS6 and MS4A6C (guinea pig antibody) by finding a GFP+ cell, imaging it

under laser excitation, and recording the cell as positive or negative manually. 20 or

148

more GFP+ cells were counted for 3-5 epithelia per odorant, and images of

representative cells were acquired and processed as described above.

149

References

Abuin, L., Bargeton, B., Ulbrich, M.H., Isacoff, E.Y., Kellenberger, S., and Benton, R. (2011). Functional architecture of olfactory ionotropic glutamate receptors. Neuron 69, 44-60.

Adler, E., Hoon, M.A., Mueller, K.L., Chandrashekar, J., Ryba, N.J., and Zuker, C.S. (2000). A novel family of mammalian taste receptors. Cell 100, 693-702.

Amcheslavsky, A., Wood, M.L., Yeromin, A.V., Parker, I., Freites, J.A., Tobias, D.J., and Cahalan, M.D. (2015). Molecular biophysics of Orai store-operated Ca2+ channels. Biophys J 108, 237-246.

Anisimova, M., and Yang, Z. (2007). Multiple Hypothesis Testing to Detect Lineages under Positive Selection that Affects Only a Few Sites. Molecular Biology and Evolution 24, 1219-1228.

Arakawa, H., Kelliher, K.R., Zufall, F., and Munger, S.D. (2013). The receptor guanylyl cyclase type D (GC-D) ligand uroguanylin promotes the acquisition of food preferences in mice. Chem Senses 38, 391-397.

Axel, R. (1995). The molecular logic of smell. Sci Am 273, 154-159.

Benjamini, Y., Drai, D., Elmer, G., Kafkafi, N., and Golani, I. (2001). Controlling the false discovery rate in behavior genetics research. Behav Brain Res 125, 279-284.

Benton, R., Vannice, K.S., Gomez-Diaz, C., and Vosshall, L.B. (2009). Variant ionotropic glutamate receptors as chemosensory receptors in Drosophila. Cell 136, 149-162.

Bubien, J.K., Zhou, L.J., Bell, P.D., Frizzell, R.A., and Tedder, T.F. (1993). Transfection of the CD20 cell surface molecule into ectopic cell types generates a Ca2+ conductance found constitutively in B lymphocytes. J Cell Biol 121, 1121-1132.

Buck, L., and Axel, R. (1991). A novel multigene family may encode odorant receptors: a molecular basis for odor recognition. Cell 65, 175-187.

150

Caterina, M.J. (2007). Transient receptor potential ion channels as participants in thermosensation and thermoregulation. Am J Physiol Regul Integr Comp Physiol 292, R64-76.

Chen, T.W., Wardill, T.J., Sun, Y., Pulver, S.R., Renninger, S.L., Baohan, A., Schreiter, E.R., Kerr, R.A., Orger, M.B., Jayaraman, V., et al. (2013). Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499, 295-300.

Cockerham, R.E., Puche, A.C., and Munger, S.D. (2009). Heterogeneous sensory innervation and extensive intrabulbar connections of olfactory necklace glomeruli. PLoS One 4, e4657.

Cruse, G., Beaven, M.A., Ashmole, I., Bradding, P., Gilfillan, A.M., and Metcalfe, D.D. (2013). A truncated splice-variant of the FcepsilonRIbeta receptor subunit is critical for microtubule formation and degranulation in mast cells. Immunity 38, 906-917.

Darriba, D., Taboada, G.L., Doallo, R., and Posada, D. (2012). jModelTest 2: more models, new heuristics and parallel computing. Nat Meth 9, 772-772.

DeLuca, D.S., Levin, J.Z., Sivachenko, A., Fennell, T., Nazaire, M.D., Williams, C., Reich, M., Winckler, W., and Getz, G. (2012). RNA-SeQC: RNA-seq metrics for quality control and process optimization. Bioinformatics 28, 1530-1532.

Dhaka, A., Viswanath, V., and Patapoutian, A. (2006). Trp ion channels and temperature sensation. Annu Rev Neurosci 29, 135-161.

Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, M., and Gingeras, T.R. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21.

Dombrowicz, D., Lin, S., Flamand, V., Brini, A.T., Koller, B.H., and Kinet, J.P. (1998). Allergy-associated FcRbeta is a molecular amplifier of IgE- and IgG-mediated in vivo responses. Immunity 8, 517-529.

Dulac, C., and Axel, R. (1995). A novel family of genes encoding putative pheromone receptors in mammals. Cell 83, 195-206.

Eon Kuek, L., Leffler, M., Mackay, G.A., and Hulett, M.D. (2015). The MS4A family: counting past 1, 2 and 3. Immunol Cell Biol.

151

Fulle, H.J., Vassar, R., Foster, D.C., Yang, R.B., Axel, R., and Garbers, D.L. (1995). A receptor guanylyl cyclase expressed specifically in olfactory sensory neurons. Proc Natl Acad Sci U S A 92, 3571-3575.

Galindo, M.M., Voigt, N., Stein, J., van Lengerich, J., Raguse, J.D., Hofmann, T., Meyerhof, W., and Behrens, M. (2012). G protein-coupled receptors in human fat taste perception. Chem Senses 37, 123-139.

Gao, L., Hu, J., Zhong, C., and Luo, M. (2010). Integration of CO2 and odorant signals in the mouse olfactory bulb. Neuroscience 170, 881-892.

Grus, W.E., and Zhang, J. (2009). Origin of the genetic components of the vomeronasal system in the common ancestor of all extant vertebrates. Mol Biol Evol 26, 407-419.

Guo, D., Zhang, J.J., and Huang, X.Y. (2009). Stimulation of guanylyl cyclase-D by bicarbonate. Biochemistry 48, 4417-4422.

Hayakawa, T., Suzuki-Hashido, N., Matsui, A., and Go, Y. (2014). Frequent expansions of the bitter taste receptor gene repertoire during evolution of mammals in the Euarchontoglires clade. Mol Biol Evol 31, 2018-2031.

Herrada, G., and Dulac, C. (1997). A novel family of putative pheromone receptors in mammals with a topographically organized and sexually dimorphic distribution. Cell 90, 763-773.

Holtmaat, A.J., Hermens, W.T., Oestreicher, A.B., Gispen, W.H., Kaplitt, M.G., and Verhaagen, J. (1996). Efficient adenoviral vector-directed expression of a foreign gene to neurons and sustentacular cells in the mouse olfactory neuroepithelium. Brain Res Mol Brain Res 41, 148-156.

Hu, J., Zhong, C., Ding, C., Chi, Q., Walz, A., Mombaerts, P., Matsunami, H., and Luo, M. (2007). Detection of near-atmospheric concentrations of CO2 by an olfactory subsystem in the mouse. Science 317, 953-957.

Ihara, S., Yoshikawa, K., and Touhara, K. (2013). Chemosensory signals and their receptors in the olfactory neural system. Neuroscience 254, 45-60.

Isogai, Y., Si, S., Pont-Lezica, L., Tan, T., Kapoor, V., Murthy, V.N., and Dulac, C. (2011). Molecular organization of vomeronasal chemoreception. Nature 478, 241-245.

152

Janas, E., Priest, R., Wilde, J.I., White, J.H., and Malhotra, R. (2005). Rituxan (anti-CD20 antibody)-induced translocation of CD20 into lipid rafts is crucial for calcium influx and apoptosis. Clin Exp Immunol 139, 439-446.

Jiang, Y., Gong, N.N., Hu, X.S., Ni, M.J., Pasi, R., and Matsunami, H. (2015). Molecular profiling of activated olfactory neurons identifies odorant receptors for odors in vivo. Nat Neurosci 18, 1446-1454.

Juilfs, D.M., Fulle, H.J., Zhao, A.Z., Houslay, M.D., Garbers, D.L., and Beavo, J.A. (1997). A subset of olfactory neurons that selectively express cGMP-stimulated phosphodiesterase (PDE2) and guanylyl cyclase-D define a unique olfactory signal transduction pathway. Proc Natl Acad Sci U S A 94, 3388-3395.

Katoh, K., and Standley, D.M. (2013a). MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Molecular Biology and Evolution 30, 772-780.

Katoh, K., and Standley, D.M. (2013b). MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol 30, 772-780.

Knight, Z.A., Tan, K., Birsoy, K., Schmidt, S., Garrison, J.L., Wysocki, R.W., Emiliano, A., Ekstrand, M.I., and Friedman, J.M. (2012). Molecular profiling of activated neurons by phosphorylated ribosome capture. Cell 151, 1126-1137.

Kosakovsky Pond, S.L., Murrell, B., Fourment, M., Frost, S.D.W., Delport, W., and Scheffler, K. (2011). A Random Effects Branch-Site Model for Detecting Episodic Diversifying Selection. Molecular Biology and Evolution 28, 3033-3043.

Koslowski, M., Sahin, U., Dhaene, K., Huber, C., and Tureci, O. (2008). MS4A12 is a colon-selective store-operated calcium channel promoting malignant cell processes. Cancer Res 68, 3458-3466.

Leinders-Zufall, T., Cockerham, R.E., Michalakis, S., Biel, M., Garbers, D.L., Reed, R.R., Zufall, F., and Munger, S.D. (2007). Contribution of the receptor guanylyl cyclase GC-D to chemosensory function in the olfactory epithelium. Proc Natl Acad Sci U S A 104, 14507-14512.

Liao, Y., Smyth, G.K., and Shi, W. (2014). featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30, 923-930.

153

Liberles, S.D., and Buck, L.B. (2006). A second class of chemosensory receptors in the olfactory epithelium. Nature 442, 645-650.

Liberles, S.D., Horowitz, L.F., Kuang, D., Contos, J.J., Wilson, K.L., Siltberg-Liberles, J., Liberles, D.A., and Buck, L.B. (2009). Formyl peptide receptors are candidate chemosensory receptors in the vomeronasal organ. Proc Natl Acad Sci U S A 106, 9842-9847.

Lin, S., Cicala, C., Scharenberg, A.M., and Kinet, J.P. (1996). The Fc(epsilon)RIbeta subunit functions as an amplifier of Fc(epsilon)RIgamma-mediated cell activation signals. Cell 85, 985-995.

Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550.

Macmanes, M.D. (2014). On the optimal trimming of high-throughput mRNA sequence data. Front Genet 5, 13.

Mamasuew, K., Hofmann, N., Breer, H., and Fleischer, J. (2011). Grueneberg ganglion neurons are activated by a defined set of odorants. Chem Senses 36, 271-282.

Man, O., Gilad, Y., and Lancet, D. (2004). Prediction of the odorant binding site of olfactory receptor proteins by human-mouse comparisons. Protein Sci 13, 240-254.

Matsunami, H., and Buck, L.B. (1997). A multigene family encoding a diverse array of putative pheromone receptors in mammals. Cell 90, 775-784.

McCarroll, S.A., Li, H., and Bargmann, C.I. (2005). Identification of transcriptional regulatory elements in chemosensory receptor genes by probabilistic segmentation. Curr Biol 15, 347-352.

Meyer, M.R., Angele, A., Kremmer, E., Kaupp, U.B., and Muller, F. (2000). A cGMP-signaling pathway in a subset of olfactory sensory neurons. Proc Natl Acad Sci U S A 97, 10595-10600.

Munger, S.D., Leinders-Zufall, T., McDougall, L.M., Cockerham, R.E., Schmid, A., Wandernoth, P., Wennemuth, G., Biel, M., Zufall, F., and Kelliher, K.R. (2010). An olfactory subsystem that detects carbon disulfide and mediates food-related social learning. Curr Biol 20, 1438-1444.

154

Munger, S.D., Leinders-Zufall, T., and Zufall, F. (2009). Subsystem organization of the mammalian sense of smell. Annu Rev Physiol 71, 115-140.

Murrell, B., Moola, S., Mabona, A., Weighill, T., Sheward, D., Kosakovsky Pond, S.L., and Scheffler, K. (2013). FUBAR: A Fast, Unconstrained Bayesian AppRoximation for Inferring Selection. Molecular Biology and Evolution 30, 1196-1205.

Nei, M., Niimura, Y., and Nozawa, M. (2008). The evolution of animal chemosensory receptor gene repertoires: roles of chance and necessity. Nat Rev Genet 9, 951-963.

Oberland, S., Ackels, T., Gaab, S., Pelz, T., Spehr, J., Spehr, M., and Neuhaus, E.M. (2015). CD36 is involved in oleic acid detection by the murine olfactory system. Front Cell Neurosci 9, 366.

Picelli, S., Faridani, O.R., Bjorklund, A.K., Winberg, G., Sagasser, S., and Sandberg, R. (2014). Full-length RNA-seq from single cells using Smart-seq2. Nat Protoc 9, 171-181.

Polyak, M.J., Li, H., Shariat, N., and Deans, J.P. (2008). CD20 homo-oligomers physically associate with the B cell antigen receptor. Dissociation upon receptor engagement and recruitment of phosphoproteins and calmodulin-binding proteins. J Biol Chem 283, 18545-18552.

Pond, S.L.K., Frost, S.D.W., and Muse, S.V. (2005). HyPhy: hypothesis testing using phylogenies. Bioinformatics 21, 676-679.

Riviere, S., Challet, L., Fluegge, D., Spehr, M., and Rodriguez, I. (2009). Formyl peptide receptor-like proteins are a novel family of vomeronasal chemosensors. Nature 459, 574-577.

Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D.L., Darling, A., Höhna, S., Larget, B., Liu, L., Suchard, M.A., and Huelsenbeck, J.P. (2012). MrBayes 3.2: Efficient Bayesian Phylogenetic Inference and Model Choice across a Large Model Space. Systematic Biology.

Ryba, N.J., and Tirindelli, R. (1997). A new multigene family of putative pheromone receptors. Neuron 19, 371-379.

Saito, H., Chi, Q., Zhuang, H., Matsunami, H., and Mainland, J.D. (2009). Odor coding by a Mammalian receptor repertoire. Sci Signal 2, ra9.

155

Secundo, L., Snitz, K., and Sobel, N. (2014). The perceptual logic of smell. Curr Opin Neurobiol 25, 107-115.

Shinoda, K., Shiotani, Y., and Osawa, Y. (1989). "Necklace olfactory glomeruli" form unique components of the rat primary olfactory system. J Comp Neurol 284, 362-373.

Simossis, V.A., and Heringa, J. (2005). PRALINE: a multiple sequence alignment toolbox that integrates homology-extended and secondary structure information. Nucleic Acids Res 33, W289-294.

Singer, M.S. (2000). Analysis of the molecular basis for octanal interactions in the expressed rat 17 olfactory receptor. Chem Senses 25, 155-165.

Sosulski, D.L., Bloom, M.L., Cutforth, T., Axel, R., and Datta, S.R. (2011). Distinct representations of olfactory information in different cortical centres. Nature 472, 213-216.

Sun, L., Wang, H., Hu, J., Han, J., Matsunami, H., and Luo, M. (2009). Guanylyl cyclase-D in the olfactory CO2 neurons is activated by bicarbonate. Proc Natl Acad Sci U S A 106, 2041-2046.

Terakita, A. (2005). The opsins. Genome Biol 6, 213.

Troemel, E.R., Chou, J.H., Dwyer, N.D., Colbert, H.A., and Bargmann, C.I. (1995). Divergent seven transmembrane receptors are candidate chemosensory receptors in C. elegans. Cell 83, 207-218.

Tsai, L., and Barnea, G. (2014). A critical period defined by axon-targeting mechanisms in the murine olfactory bulb. Science 344, 197-200.

Tsirigos, K.D., Peters, C., Shu, N., Käll, L., and Elofsson, A. (2015). The TOPCONS web server for consensus prediction of membrane protein topology and signal peptides. Nucleic Acids Research 43, W401-W407.

van Giesen, L., Hernandez-Nunez, L., Delasoie-Baranek, S., Colombo, M., Renaud, P., Bruggmann, R., Benton, R., Samuel, A.D., and Sprecher, S.G. (2016). Multimodal stimulus coding by a gustatory sensory neuron in Drosophila larvae. Nat Commun 7, 10687.

156

Wang, F., Flanagan, J., Su, N., Wang, L.C., Bui, S., Nielson, A., Wu, X., Vo, H.T., Ma, X.J., and Luo, Y. (2012). RNAscope: a novel in situ RNA analysis platform for formalin-fixed, paraffin-embedded tissues. J Mol Diagn 14, 22-29.

Wooding, S. (2011). Signatures of natural selection in a primate bitter taste receptor. J Mol Evol 73, 257-265.

Young, J.M., Waters, H., Dong, C., Fulle, H.J., and Liman, E.R. (2007). Degeneration of the olfactory guanylyl cyclase D gene during primate evolution. PLoS One 2, e884.

Zhang, J., and Webb, D.M. (2003). Evolutionary deterioration of the vomeronasal pheromone transduction pathway in catarrhine primates. Proc Natl Acad Sci U S A 100, 8337-8341.

Zuccolo, J., Bau, J., Childs, S.J., Goss, G.G., Sensen, C.W., and Deans, J.P. (2010). Phylogenetic analysis of the MS4A and TMEM176 gene families. PLoS One 5, e9369.

Chapter 4

General Discussion and Conclusion

  158  

General Discussion

Does The OR Expression Pattern Matter?

Most of the mammalian olfactory system obeys a “one receptor per neuron”

organization that arose in the common ancestor of jawed vertebrates. Why has the

necklace subsystem broken from this predominant molecular logic? Possible answers

depend, first, on considering why singular GPCR OR expression is so widespread.

While this pattern could reflect a developmental convenience or constraint, several

observations suggest that maintaining separate neuronal and glomerular populations for

each receptor has additional adaptive value for shaping odor perception and guiding

behavior.

The main reason to think the “one receptor per neuron” pattern is critical for

mammalian odor perception is that a variety of mechanisms have evolved to maintain it.

The biochemical process for selecting and maintaining OR choice has coopted at least

two pathways that tightly shut off the expression of hundreds or thousands of unchosen

genes (Dalton and Lomvardas, 2015). First, clusters of OR genes are packed into

constitutive heterochromatin that normally silences centromeric DNA; release from this

state requires DNA enhancers acting stochastically to remove the chosen OR locus to a

region of active transcription (Clowney et al., 2012; Magklara et al., 2011; Markenscoff-

Papadimitriou et al., 2014). This method differs markedly from singular OR expression

in insects, which is orchestrated by deterministic, cell-specific combinations of cis-acting

transcription factors (Ray et al., 2008). Second, nascent OR proteins in the endoplasmic

reticulum engage the unfolded protein response to block further protein expression and

sensory neuron development until the cells are capable of OR signaling; this culminates

  159  

in locking the chosen OR into permanent transcription and shutting off the choice

process, ensuring that one and only one functional receptor is expressed per cell

(Dalton et al., 2013).

Again, no signs of these processes have been found in insects, whose smaller

OR repertoires may achieve singular expression purely through cell identity

determinants. In contrast, complex mechanisms for regulating OR choice may be

essential in mammals express as many as tenfold more OR genes than insect and fish.

Comparing a hypothetical species with 100 OR genes to one with 1000 genes illustrates

the problem: if inadvertent OR expression were fixed at some low rate – say, an

expression probability of .001 per gene per cell – the sensory neurons of the 100-gene

species would express an extra receptor less than 10% of the time; however, the

neurons of the 1000-gene species would do so nearly two thirds of the time. In this case

the majority of the cells that chose a given receptor and innervated a given glomerulus

would have a random component to their sensory tuning due to unintentional receptor

expression. The precision of odor coding would drop dramatically with larger OR

repertoires, and innate behaviors gated, for example, by pheromone ligation of single

receptors might also be triggered by higher concentrations of unrelated odors. Thus

additional mechanisms for repressing OR expression seem to be required for precise

odor representation in animals that use both large receptor repertoires and a “one

receptor per neuron” organization. The fact that they have evolved in mammals, along

with the expansion of GPCR OR families to contain hundreds or thousands of genes,

strongly suggests that the predominant OR expression pattern carries fundamental

  160  

adaptive value to vertebrate olfaction, and cannot be discarded without sacrificing

important properties of odor perception.

Consistent with this explanation, OR choice appears to be less precise in

vertebrates with smaller receptor repertoires. As described in Chapter 2, the zebrafish

olfactory system uses the same families of GPCR ORs and the same glomerular

organization in its olfactory bulb as mammals; however, its genome contains ~250

functional receptors and its bulb only ~140 glomeruli, suggesting that many sensory

neurons and glomeruli receive odor signals through multiple ORs (Braubach et al.,

2012; Saraiva et al., 2015). Zebrafish cells expressing two ORs from the same gene

cluster have been observed directly (Sato et al., 2007). Together these data raise the

possibility that fish have not evolved the most stringent mechanisms for ensuring

singular OR expression. With smaller OR gene families, a low rate of aberrant

expression may be more tolerable in fish than in mammals.

There is further circumstantial evidence that extraneous OR expression would

disrupt odor coding: the mechanisms that repress unchosen ORs in sensory neurons

also target additional gene families. The first report that OR clusters are confined to

constitutive heterochromatin also showed this silencing chromatin signature affects

other types of chemoreceptors, such as bitter taste receptors, receptors for internal

signaling molecules, and non-GPCR immune cell receptors; the set of marked families

also includes the Ms4a genes, an observation that I confirmed is not due to chromatin

repression spilling over from an adjacent OR cluster (Magklara et al., 2011) (data not

shown.)

  161  

Why would the mammalian singular choice process repress genes that are not

normally expressed in typical olfactory sensory neurons? The simplest explanation is

that these gene families would, like ORs, scramble odor coding if expressed even at low

levels in the same cells as a chosen OR. This underscores that the widespread use of

GPCR ORs in vertebrate olfaction may be an evolutionary accident: other receptors with

enough structural diversity to bind a wide range of molecular structures could, in

principle, detect odors to drive neuronal chemotransduction. The treatment of the MS4A

family as a “threat” to odor coding in typical sensory neurons supports my finding that

these proteins can act as chemoreceptors in a range of cell types, outside the context of

necklace neurons.

In summary, the “one receptor per neuron” logic is not merely a developmental

convenience in mammals. Rather, this organization appears to be an adaptation for

odor perception that enhanced the survival and reproduction of mammalian species; the

most restrictive molecular pathways for repressing chemoreceptor expression may have

evolved when vertebrate genomes began to encode more olfactory receptors, for

example as tetrapods transitioned from water to land (Niimura and Nei, 2005).

How Does Singular OR Expression Aid Odor Perception?

What does the “one receptor per neuron” organization contribute to olfaction that

makes it such a widespread adaptation? As discussed in Chapter 2 the presence of a

“private line” into the brain for each olfactory receptor gene allows precise chemical

cues, such as pheromones that bind only a single receptor at their natural

concentrations, to elicit patterns of neural activity distinct from any other odorant or odor

  162  

blend. These unique single-receptor representations may drive equally precise

behaviors, such as the “lordosis” mating posture triggered in mice by ESP1 ligation of

the vomeronasal receptor V2Rp5. Given this organization, the olfactory system cannot

be “tricked” into driving a particular behavior through the activity of unrelated receptors.

This interpretation is supported by the independent evolution of the “one receptor

per neuron” expression pattern in insects. Olfactory sensory neurons of the fruit fly

Drosophila melanogaster generally each express a single functional olfactory receptor

from a repertoire of about 75, and their axons converge in a “one receptor per

glomerulus” pattern onto the antennal lobe of the brain (Hansson and Stensmyr, 2011;

Rytz et al., 2013). This initial odor processing structure has therefore evolved a

neuroanatomy similar to the olfactory bulb (Vosshall et al., 1999). A number of single

Drosophila receptors sense monomolecular odorants that drive feeding, reproductive,

and defensive behaviors (Dekker et al., 2006; Kurtovic et al., 2007; Stensmyr et al.,

2012); at the same time, natural odor blends are distinguished by more complex

multiglomerular activity patterns, and the recruitment of new glomeruli as odors or

concentrations shift can alter the intrinsic attractive or aversive valence of a stimulus

(Bell and Wilson, 2016; Hallem et al., 2004; Semmelhack and Wang, 2009). Thus

Drosophila olfactory behavior makes use of precise odor representations that include

anywhere from a single to a larger specified set of glomeruli, illustrating an advantage of

the “one receptor per neuron and glomerulus” organization: if each sensory neuron

expressed multiple receptors, animals might not develop or learn unique responses to

the activation of a particular receptor.

  163  

In insects like Drosophila that use fewer than 100 receptors and glomeruli, some

patterns of innate behavior may be driven only by the action of a particular receptor

channel. In vertebrates, understanding in general the relationship between odor activity

in the olfactory bulb and resulting changes in behavior may reveal specific cases where

a single receptor is both necessary and sufficient for an instinctual response (Dewan et

al., 2013). Thus the “one receptor per neuron” logic allows single receptors to play

innate and non-redundant roles in olfaction.

This organization may also help in discriminating complex multimolecular odors,

such as the volatile signatures of natural foods or other animals. The evoked patterns of

neural activity are probably unique for any given odor blend at a particular

concentration, determined by the binding affinities of its constituents with each of

hundreds of ORs (Lin da et al., 2006). Some mammals can discern extremely fine odor

features, such as the inclusion of one enantiomer or another within an odor blend

(Uchida and Mainen, 2003); odor discrimination ability may correlate with number of

functional olfactory receptors (Niimura et al., 2014). Fine-scale odor sensing is

important for inferring the value or threat of natural stimuli, such as the chemical

characteristics of fruit odors as they change with ripening (Hayden et al., 2014;

Hodgkison et al., 2007). In an extreme case, the African Elephant, which encodes more

identified ORs than any other mammal, can discriminate the garments worn by two

Kenyan ethnic groups: through hardwiring or learning, only the scent of the group that

hunts elephants provokes a flight and hiding response (Bates et al., 2007). Although

there are other OR expression patterns through which the brain could distinguish similar

odors, an array of single-receptor channels may have been the simplest and most

  164  

flexible way for this olfactory function to evolve and develop. As discussed in Chapter 2,

the “one receptor per neuron” organization abets the dramatic evolution of vertebrate

OR repertoires. The genetic and functional independence of different ORs may have

allowed adaptations for finer odor detection or discernment only in the regions of

chemical space most relevant to a given species (Hayden et al., 2010). It may be

possible to test this model directly using current genetic methods, for example by

demonstrating that mutant animals with fewer ORs or scrambled receptor expression

cannot discriminate natural odors as readily. Together these observations suggest that

the standard mode of mammalian olfaction has evolved because it produces precise

odor representations in the sensory periphery and the olfactory bulb. The distinct activity

patterns evoked by different odors are then available to drive unique responses.

What Does The Variant Necklace System Provide For Olfaction?

If the main and accessory olfactory systems mediate both innate and learned

behaviors with a standard OR repertoire and organization, why do mammals have a

necklace system with variant receptors and molecular logic? GC-D sensory neurons

could detect odors not sensed by the rest of the olfactory system or could, like

specialized retinal ganglion cell types in the visual system, extract distinct features of

the environment important for particular physiological responses or behaviors (Do and

Yau, 2010; Kim et al., 2008; Zhang et al., 2012).

Previous research suggests both of these roles for GC-D cells. These are the

only olfactory sensory neurons known to detect the volatile liquid carbon disulfide (CS2)

and the water-soluble peptides guanylin (G) and uroguanylin (UG) at their natural

  165  

concentrations in rodent breath, urine, and feces, respectively (Arakawa et al., 2013;

Leinders-Zufall et al., 2007; Munger et al., 2010). Electrophysiological responses to

these ligands require the transduction channel subunit CNGA3 and GC-D itself, which is

a receptor for G and UG; binding of these peptides directly stimulates the guanylate

cyclase catalytic activity to elevate intracellular cGMP, which gates CNGA3-containing

cation channels (Leinders-Zufall et al., 2007). These properties may represent a

molecular specialization for detecting ligands that do not bind GPCR ORs, which in the

rest of the main olfactory system signal through CNGA2-containing, cAMP-regulated

transduction channels (Touhara and Vosshall, 2009). Along these lines, dissociated

GC-D cells in vitro and necklace glomerulus-innervating olfactory bulb neurons in vivo

also respond to near-atmospheric levels of carbon dioxide (Hu et al., 2007); this likely

proceeds through its intracellular conversion to bicarbonate by the highly expressed

enzyme Carbonic Anhydrase II (CAR2) (Sun et al., 2009). Bicarbonate has been shown

to directly stimulate the catalytic activity of GC-D (Guo et al., 2009). Finally, cells that

innervate Pde2a+ necklace glomeruli were previously found to respond to 2,5-

dimethylpyrazine (2,5-DMP) and a surprisingly broad range of other molecular

structures characterized as “plant volatiles;” however, these compounds had not been

generally tested for their ability to stimulate GC-D sensory neurons directly (Gao et al.,

2010; Lin et al., 2004). Because these odors were known to activate conventional

OMP+ sensory neurons, some of which may co-innervate necklace glomeruli, it

remained unclear whether GC-D cells detected a wider set of volatile organic

compounds (Cockerham et al., 2009). Overall, the chemical tuning of GC-D cells

  166  

suggested that the necklace has a non-redundant role in sensing some stimuli but also

responds to chemicals that activate other olfactory subsystems.

Consistent with its sensory responses, the necklace appears to play perceptual

and behavioral roles distinct from the major GPCR-expressing systems. Innate

avoidance of elevated carbon dioxide concentrations has been suggested to require

GC-D cells because it is lost in CAR2-null mutant mice; CAR2 is expressed in other

non-olfactory neurons, though, so it is not certain that this phenotype results from

disrupting the necklace (Hu et al., 2007). Furthermore, trigeminal sensory fibers and

some amygdalar neurons respond to carbon dioxide and promote aversive and fear

responses, respectively (Komai and Bryant, 1993; Ziemann et al., 2009).

A better-established and more circumscribed role for the necklace is mediating a

specific form of olfactory learning, the social transmission of food preference (STFP)

(Munger et al., 2010). “Observer” mice in the presence of a “demonstrator” mouse that

has recently eaten food scented with some odor (such as cocoa or cinnamon) will, when

given a later choice, prefer food scented with the same odor. This effect, found in both

mice and rats, is thought to be an adaptation for avoiding unsafe food that would have

killed conspecifics (Galef et al., 1988). Carbon disulfide and uroguanylin at the

concentrations found in rodent breath and feces, respectively, can substitute for a

demonstrator mouse in this paradigm; but in GC-D-null mice the effect is lost (Arakawa

et al., 2013; Munger et al., 2010). Furthermore, STFP requires the activity of cholinergic

forebrain nuclei that are direct downstream targets of GC-D necklace glomeruli (Ross et

al., 2005; Tari Tan, unpublished data.) It is therefore possible that the STFP is

controlled by GC-D cell-gated cholinergic modulation, which both feeds back onto the

  167  

olfactory bulb to reshape sensory responses and also targets central brain regions

involved in general learning (Fletcher and Wilson, 2002; Ross and Eichenbaum, 2006).

Together these data suggest that GC-D neurons convey the unconditioned stimulus in

this form of classical conditioning: necklace activation is sufficient and necessary to

form a preference for a simultaneously sensed food odor. This is distinct from forms of

odor learning that proceed through the main olfactory system and one of its major

targets, the piriform cortex. In those paradigms, the unconditioned stimuli (foot shocks,

water rewards) are not olfactory and must converge on odor representations by other

pathways (Choi et al., 2011). Thus the role of GC-D cells in altering food-odor

preference may be unique and essential within the olfactory system, explaining their

presence alongside conventional sensory neurons.

How Is MS4A Chemoreception Suited For Necklace Function?

In the context of necklace-driven behavior, what are the possible reasons that

GC-D cells express a family of non-GPCR ORs in a “many receptors, every neuron”

pattern? The use of non-GPCRs and the pattern of expressing multiple receptors per

cell may serve independent purposes. The one-per-neuron expression of non-GPCR

ORs in insects demonstrates that this organizing mode does not depend completely on

receptor protein structure (Silbering and Benton, 2010); and the coexpression of dozens

or hundreds of GPCRs in other sensory cell types, ranging from C. elegans olfactory

neurons to mammalian bitter taste cells, demonstrates that “one receptor per neuron” is

not the only useful pattern of chemoreceptor expression (Chandrashekar et al., 2000;

Sengupta et al., 1996; Troemel et al., 1995). These other chemosensory neurons

  168  

illustrate a functional reason for expressing many receptors: because they relay signals

that trigger innate aversive responses, any ligand that signifies immediate danger will

set off the same “alarm” without having to be decoded by central neural circuits

(Bargmann et al., 1993; Mueller et al., 2005; Troemel et al., 1997). The same principle

defines another atypical olfactory subsystem, the Gruneberg Ganglion of the mouse.

These neurons at the tip of the nose detect both predator odorants and alarm

pheromones released by stressed mice to elicit innate fear behavior (Brechbuhl et al.,

2008; Brechbuhl et al., 2013). While it is not known which receptors bind these various

compounds, the Grueneberg Ganglion appears to pool several distinct, danger-related

odor signals.

The coexpression of multiple MS4A proteins may similarly integrate a variety of

stimuli in each GC-D neuron. Our data suggest that individual GC-D cells have broader

tuning than what is granted by any single MS4A in vitro, responding for example to the

MS4A6C ligands 2,3- and 2,5-DMP, the MS4A6D ligand oleic acid, and the MS4A4B

ligand alpha-linolenic acid (Greer et al., 2016). Three chemical classes in particular –

heteroaromatic compounds, steroid derivatives, and long-chain fatty acids – appear to

bind different MS4A proteins in vitro. However, the relationship between the tuning of

the MS4As and the tuning of GC-D cells remains unclear for several reasons. First, only

a limited variety of compounds were screened against a partial set of heterologously

expressed MS4As (see Chapter 3.) Broader screening may reveal some other chemical

class, such as one that combines a ring-shaped head group with a long-chain fatty acid

tail, that binds all MS4A proteins with high affinity. This scenario would still be consistent

with GC-D cells having adapted to detect a large range of ligands; however it would

  169  

argue against multiple MS4As having evolved to integrate different chemical classes.

On the other hand, it is already clear that multiple atypical receptors function in single

necklace cells – GC-D, CAR2 (a “receptor” for carbon dioxide), and MS4A proteins –

setting these cells apart from all other mammalian olfactory neurons.

It is also unknown whether distinct MS4A proteins form functional heteromers in

vivo. Although our data imply that single MS4As form chemoreceptors in both 293 cells

and virally-infected olfactory neurons, other reports have shown that overexpressed

members of the MS4A family can interact both with the same and with other MS4A

proteins (e.g., MS4A6B with MS4A4B) (Howie et al., 2009). Moreover, while some

family members are expressed alone in chemosensory but non-olfactory tissue (MS4A5

in testes; MS4A8A in small intestine epithelium; MS4A12 in colon epithelium), others

are routinely coexpressed (all the MS4A6 and MS4A7 proteins, along with MS4A4A, in

microglia) (Eon Kuek et al., 2015; Koslowski et al., 2008; PLG and DMB, unpublished

data). Thus different proteins could interact to form a wider variety of chemoreceptors

than those made from only one MS4A. One intriguing possibility is that GC-D cells

coexpress a large subset of MS4As in order to form a diverse range of heteromeric

receptors, which could give GC-D cells broader tuning than is currently appreciated.

Additional screening for MS4A ligands in vitro and GC-D cell ligands in vivo, along with

further characterization of MS4A receptor biochemistry, will provide evidence for or

against this hypothesis.

Finally, the coexpression of multiple MS4As in each necklace cell raises the

question of how MS4A signaling interacts, biochemically and perceptually, with the

other sensory functions of GC-D cells. Specifically, does the activation of these neurons

  170  

by MS4A ligands require or cooperate with the signaling pathways for sensing carbon

dioxide, carbon disulfide, and G/UG? Examining necklace neuron responses in the

context of GC-D and CNGA3-null mutations, compared with Ms4a deletions, may reveal

which components are needed to obtain an electrophysiological response or the large

calcium influx observed in the in vivo explant imaging assay. Is there a plausible

pathway through which MS4A-ligand binding could feed into the GC-D/cGMP/CNGA3

signaling mechanism? This is difficult to answer without a better understanding of MS4A

receptor structure and function, for example knowing whether MS4A proteins form

ligand-gated ion channels; however, the observation of MS4A-dependent, odorant-

evoked extracellular calcium entry in a 293 cell-based assay suggests that MS4A

ligands could produce a local calcium signal in GC-D cell dendrites. Minor elevations of

intracellular calcium produce large increases in purified GC-D catalytic activity, a unique

feature of this receptor guanylate cyclase (Duda and Sharma, 2008). This represents a

speculative route from MS4A-ligand binding to cGMP production and an

electrophysiological response. It is also possible that MS4A receptors activate the

necklace by a completely independent mechanism. As circumstantial evidence favoring

this model, MS4A ligands induce S6 phosphorylation in virally-infected conventional

sensory neurons that lack GC-D, and other tissues that express MS4A family members

are not known to express this particular calcium-sensitive enzyme – though they could

respond to ligand-gated calcium transients by other signaling pathways (Greer et al.,

2016).

It is harder to hypothesize about perceptual interactions between MS4A and GC-

D or CAR2 chemoreception. The MS4A ligands identified here have not been tested for

  171  

behavioral effects in either valence assays or the STFP paradigm; however it is not

difficult to imagine long chain fatty acids, for instance, providing an unconditioned

stimulus for food preference in the same way as carbon disulfide or uroguanylin. A

future set of experiments will address this possibility and test its dependence on

necklace function, compared to other olfactory neurons capable of detecting the same

ligands through conventional receptors. These results may also suggest whether MS4A

proteins have evolved specifically to bind food- or feeding-related molecular structures.

  172  

Conclusion

In identifying a variant mode of chemoreception, I have tried to explore the

evolutionary reasons that an atypical sensory subsystem has evolved alongside the rest

of the mammalian olfactory system. As described in Chapter 2, the partitioning of the

olfactory system into parallel pathways may have arisen in the ancestor of all

vertebrates as a way of extracting and separating distinct features of the chemical

environment. Signals passing along different pathways reach different regions of the

deeper brain, conveying different types of chemical information and regulating a variety

of innate and learned behaviors. Prior to this work, however, it was thought that all

vertebrate olfactory subsystems sensed odors through GPCR families expressed in a

characteristic “one receptor per neuron” pattern. The discovery of MS4A

chemoreceptors appearing to combine chemical signals, both across different MS4A

proteins and with the receptor guanylate cyclase GC-D, implies that an alternative

olfactory organization provided a selective advantage during the evolution of some

mammals. Finding a family of chemoreceptors that governs the tuning of GC-D cells will

aid in understanding the particular role of the necklace subsystem in the overall context

of olfaction.

The coexpression of chemoreceptors in GC-D cells is reminiscent of C. elegans

olfactory sensory neurons, several of which detect structurally distinct odorants through

multiple GPCR ORs to modulate chemotaxis (Bargmann et al., 1993; Sengupta et al.,

1996; Troemel et al., 1997). In converting a variety of chemical signals into a smaller

number of innate behavioral responses, these C. elegans cells may provide a partial

model for the function of the mammalian necklace subsystem. Previous work on the

  173  

GC-D cells suggests that they too produce an innately relevant signal, likely one that

gives context for feeding behavior. Thus it may have been adaptive for these cells to

express a variety of chemoreceptors that could detect stimuli related to food quality, like

carbon disulfide and uroguanylin. Since this function does not require decoding the

precise identity of an odor or influencing a large set of behaviors, the “one receptor per

neuron” pattern could have been abandoned in the necklace subsystem. This model of

necklace evolution does not require that STFP is its only major function; it explains the

integration of multiple, atypical receptor signals as long as all necklace odorants play a

common role in odor perception and this role is distinct from that of the main and

accessory olfactory systems.

The use of GPCR olfactory receptors is “canonical” within vertebrate olfaction,

but not from a wider view of animal chemosensation. Known insect olfactory receptors

belong to two families of ligand-gated ion channel, one an evolutionary offshoot of the

ionotropic glutamate receptors (Croset et al., 2010; Silbering and Benton, 2010). The

mammalian receptors for salty and acidic taste are likely ion-sensitive ion channels, and

TRP channels involved in nociception and other sensory modalities are selectively

modulated by chemical compounds like capsaicin and menthol (Clapham, 2003; Huang

et al., 2006). The general pattern that emerges, then, is not of GPCRs playing an

exclusive role in chemosensation, but of particular protein structures being co-opted by

natural selection when their ligand-binding properties are useful. In animal olfaction,

where a wide range of chemical structures must be detected and interpreted, the result

of this co-opting was the expansion of ancestral genes into large and diverse

chemoreceptor gene families. This signature of expansion and diversifying selection

  174  

provided a major clue that MS4A proteins might form chemoreceptors, as their structure

is not similar enough to any other proteins to indicate their function. Combining

molecular evolution and tissue expression patterns could be a generally useful way to

identify new types of chemoreceptor. For example, a new branch of the ionotropic

acetylcholine receptor family was recently found in the chemosensory tentacles of the

octopus, an invertebrate with an unknown receptor repertoire (Albertin et al., 2015).

These proteins have lost their acetylcholine binding residues, suggesting that they have

expanded beyond their ancestral function; if they are expressed in chemosensory

neurons and have diversified enough to bind a wide range of environmental odors, it

would suggest that they too have been co-opted as another form of olfactory receptor.

The chemical world holds innumerable signs for animals to use to their

advantage. Like other senses, mammalian olfaction employs multiple strategies to

extract this information and to enact the right response. The findings presented here

show that distinct modes of chemoreception arose in separate olfactory pathways; they

follow a general pattern in which diverse neural arrangements have evolved to detect

different features of the environment. Across species and sensory system, varied ways

to depict the world are carved within the brain.

  175  

References Albertin, C.B., Simakov, O., Mitros, T., Wang, Z.Y., Pungor, J.R., Edsinger-Gonzales, E., Brenner, S., Ragsdale, C.W., and Rokhsar, D.S. (2015). The octopus genome and the evolution of cephalopod neural and morphological novelties. Nature 524, 220-224.

Arakawa, H., Kelliher, K.R., Zufall, F., and Munger, S.D. (2013). The receptor guanylyl cyclase type D (GC-D) ligand uroguanylin promotes the acquisition of food preferences in mice. Chem Senses 38, 391-397.

Bargmann, C.I., Hartwieg, E., and Horvitz, H.R. (1993). Odorant-selective genes and neurons mediate olfaction in C. elegans. Cell 74, 515-527.

Bates, L.A., Sayialel, K.N., Njiraini, N.W., Moss, C.J., Poole, J.H., and Byrne, R.W. (2007). Elephants classify human ethnic groups by odor and garment color. Curr Biol 17, 1938-1942.

Bell, J.S., and Wilson, R.I. (2016). Behavior Reveals Selective Summation and Max Pooling among Olfactory Processing Channels. Neuron 91, 425-438.

Braubach, O.R., Fine, A., and Croll, R.P. (2012). Distribution and functional organization of glomeruli in the olfactory bulbs of zebrafish (Danio rerio). J Comp Neurol 520, 2317-2339, Spc2311.

Brechbuhl, J., Klaey, M., and Broillet, M.C. (2008). Grueneberg ganglion cells mediate alarm pheromone detection in mice. Science 321, 1092-1095.

Brechbuhl, J., Moine, F., Klaey, M., Nenniger-Tosato, M., Hurni, N., Sporkert, F., Giroud, C., and Broillet, M.C. (2013). Mouse alarm pheromone shares structural similarity with predator scents. Proc Natl Acad Sci U S A 110, 4762-4767.

Chandrashekar, J., Mueller, K.L., Hoon, M.A., Adler, E., Feng, L., Guo, W., Zuker, C.S., and Ryba, N.J. (2000). T2Rs function as bitter taste receptors. Cell 100, 703-711.

Choi, G.B., Stettler, D.D., Kallman, B.R., Bhaskar, S.T., Fleischmann, A., and Axel, R. (2011). Driving opposing behaviors with ensembles of piriform neurons. Cell 146, 1004-1015.

Clapham, D.E. (2003). TRP channels as cellular sensors. Nature 426, 517-524.

  176  

Clowney, E.J., LeGros, M.A., Mosley, C.P., Clowney, F.G., Markenskoff-Papadimitriou, E.C., Myllys, M., Barnea, G., Larabell, C.A., and Lomvardas, S. (2012). Nuclear aggregation of olfactory receptor genes governs their monogenic expression. Cell 151, 724-737.

Cockerham, R.E., Puche, A.C., and Munger, S.D. (2009). Heterogeneous sensory innervation and extensive intrabulbar connections of olfactory necklace glomeruli. PLoS One 4, e4657.

Croset, V., Rytz, R., Cummins, S.F., Budd, A., Brawand, D., Kaessmann, H., Gibson, T.J., and Benton, R. (2010). Ancient protostome origin of chemosensory ionotropic glutamate receptors and the evolution of insect taste and olfaction. PLoS Genet 6, e1001064.

Dalton, R.P., and Lomvardas, S. (2015). Chemosensory receptor specificity and regulation. Annu Rev Neurosci 38, 331-349.

Dalton, R.P., Lyons, D.B., and Lomvardas, S. (2013). Co-opting the unfolded protein response to elicit olfactory receptor feedback. Cell 155, 321-332.

Dekker, T., Ibba, I., Siju, K.P., Stensmyr, M.C., and Hansson, B.S. (2006). Olfactory shifts parallel superspecialism for toxic fruit in Drosophila melanogaster sibling, D. sechellia. Curr Biol 16, 101-109.

Dewan, A., Pacifico, R., Zhan, R., Rinberg, D., and Bozza, T. (2013). Non-redundant coding of aversive odours in the main olfactory pathway. Nature 497, 486-489.

Do, M.T., and Yau, K.W. (2010). Intrinsically photosensitive retinal ganglion cells. Physiol Rev 90, 1547-1581.

Duda, T., and Sharma, R.K. (2008). ONE-GC membrane guanylate cyclase, a trimodal odorant signal transducer. Biochem Biophys Res Commun 367, 440-445.

Eon Kuek, L., Leffler, M., Mackay, G.A., and Hulett, M.D. (2015). The MS4A family: counting past 1, 2 and 3. Immunol Cell Biol.

Fletcher, M.L., and Wilson, D.A. (2002). Experience modifies olfactory acuity: acetylcholine-dependent learning decreases behavioral generalization between similar odorants. J Neurosci 22, RC201.

  177  

Galef, B.G., Jr., Mason, J.R., Preti, G., and Bean, N.J. (1988). Carbon disulfide: a semiochemical mediating socially-induced diet choice in rats. Physiol Behav 42, 119-124.

Gao, L., Hu, J., Zhong, C., and Luo, M. (2010). Integration of CO2 and odorant signals in the mouse olfactory bulb. Neuroscience 170, 881-892.

Greer, P.L., Bear, D.M., Lassance, J.M., Bloom, M.L., Tsukahara, T., Pashkovski, S.L., Masuda, F.K., Nowlan, A.C., Kirchner, R., Hoekstra, H.E., et al. (2016). A Family of non-GPCR Chemosensors Defines an Alternative Logic for Mammalian Olfaction. Cell 165, 1734-1748.

Guo, D., Zhang, J.J., and Huang, X.Y. (2009). Stimulation of guanylyl cyclase-D by bicarbonate. Biochemistry 48, 4417-4422.

Hallem, E.A., Ho, M.G., and Carlson, J.R. (2004). The molecular basis of odor coding in the Drosophila antenna. Cell 117, 965-979.

Hansson, B.S., and Stensmyr, M.C. (2011). Evolution of insect olfaction. Neuron 72, 698-711.

Hayden, S., Bekaert, M., Crider, T.A., Mariani, S., Murphy, W.J., and Teeling, E.C. (2010). Ecological adaptation determines functional mammalian olfactory subgenomes. Genome Res 20, 1-9.

Hayden, S., Bekaert, M., Goodbla, A., Murphy, W.J., Davalos, L.M., and Teeling, E.C. (2014). A cluster of olfactory receptor genes linked to frugivory in bats. Mol Biol Evol 31, 917-927.

Hodgkison, R., Ayasse, M., Kalko, E.K., Haberlein, C., Schulz, S., Mustapha, W.A., Zubaid, A., and Kunz, T.H. (2007). Chemical ecology of fruit bat foraging behavior in relation to the fruit odors of two species of paleotropical bat-dispersed figs (Ficus hispida and Ficus scortechinii). J Chem Ecol 33, 2097-2110.

Howie, D., Nolan, K.F., Daley, S., Butterfield, E., Adams, E., Garcia-Rueda, H., Thompson, C., Saunders, N.J., Cobbold, S.P., Tone, Y., et al. (2009). MS4A4B is a GITR-associated membrane adapter, expressed by regulatory T cells, which modulates T cell activation. J Immunol 183, 4197-4204.

  178  

Hu, J., Zhong, C., Ding, C., Chi, Q., Walz, A., Mombaerts, P., Matsunami, H., and Luo, M. (2007). Detection of near-atmospheric concentrations of CO2 by an olfactory subsystem in the mouse. Science 317, 953-957.

Huang, A.L., Chen, X., Hoon, M.A., Chandrashekar, J., Guo, W., Trankner, D., Ryba, N.J., and Zuker, C.S. (2006). The cells and logic for mammalian sour taste detection. Nature 442, 934-938.

Kim, I.J., Zhang, Y., Yamagata, M., Meister, M., and Sanes, J.R. (2008). Molecular identification of a retinal cell type that responds to upward motion. Nature 452, 478-482.

Komai, M., and Bryant, B.P. (1993). Acetazolamide specifically inhibits lingual trigeminal nerve responses to carbon dioxide. Brain Res 612, 122-129.

Koslowski, M., Sahin, U., Dhaene, K., Huber, C., and Tureci, O. (2008). MS4A12 is a colon-selective store-operated calcium channel promoting malignant cell processes. Cancer Res 68, 3458-3466.

Kurtovic, A., Widmer, A., and Dickson, B.J. (2007). A single class of olfactory neurons mediates behavioural responses to a Drosophila sex pheromone. Nature 446, 542-546.

Leinders-Zufall, T., Cockerham, R.E., Michalakis, S., Biel, M., Garbers, D.L., Reed, R.R., Zufall, F., and Munger, S.D. (2007). Contribution of the receptor guanylyl cyclase GC-D to chemosensory function in the olfactory epithelium. Proc Natl Acad Sci U S A 104, 14507-14512.

Lin da, Y., Shea, S.D., and Katz, L.C. (2006). Representation of natural stimuli in the rodent main olfactory bulb. Neuron 50, 937-949.

Lin, W., Arellano, J., Slotnick, B., and Restrepo, D. (2004). Odors detected by mice deficient in cyclic nucleotide-gated channel subunit A2 stimulate the main olfactory system. J Neurosci 24, 3703-3710.

Magklara, A., Yen, A., Colquitt, B.M., Clowney, E.J., Allen, W., Markenscoff-Papadimitriou, E., Evans, Z.A., Kheradpour, P., Mountoufaris, G., Carey, C., et al. (2011). An epigenetic signature for monoallelic olfactory receptor expression. Cell 145, 555-570.

  179  

Markenscoff-Papadimitriou, E., Allen, W.E., Colquitt, B.M., Goh, T., Murphy, K.K., Monahan, K., Mosley, C.P., Ahituv, N., and Lomvardas, S. (2014). Enhancer interaction networks as a means for singular olfactory receptor expression. Cell 159, 543-557.

Mueller, K.L., Hoon, M.A., Erlenbach, I., Chandrashekar, J., Zuker, C.S., and Ryba, N.J. (2005). The receptors and coding logic for bitter taste. Nature 434, 225-229.

Munger, S.D., Leinders-Zufall, T., McDougall, L.M., Cockerham, R.E., Schmid, A., Wandernoth, P., Wennemuth, G., Biel, M., Zufall, F., and Kelliher, K.R. (2010). An olfactory subsystem that detects carbon disulfide and mediates food-related social learning. Curr Biol 20, 1438-1444.

Niimura, Y., Matsui, A., and Touhara, K. (2014). Extreme expansion of the olfactory receptor gene repertoire in African elephants and evolutionary dynamics of orthologous gene groups in 13 placental mammals. Genome Res 24, 1485-1496.

Niimura, Y., and Nei, M. (2005). Evolutionary dynamics of olfactory receptor genes in fishes and tetrapods. Proc Natl Acad Sci U S A 102, 6039-6044.

Ray, A., van der Goes van Naters, W., and Carlson, J.R. (2008). A regulatory code for neuron-specific odor receptor expression. PLoS Biol 6, e125.

Ross, R.S., and Eichenbaum, H. (2006). Dynamics of hippocampal and cortical activation during consolidation of a nonspatial memory. J Neurosci 26, 4852-4859.

Ross, R.S., McGaughy, J., and Eichenbaum, H. (2005). Acetylcholine in the orbitofrontal cortex is necessary for the acquisition of a socially transmitted food preference. Learn Mem 12, 302-306.

Rytz, R., Croset, V., and Benton, R. (2013). Ionotropic receptors (IRs): chemosensory ionotropic glutamate receptors in Drosophila and beyond. Insect Biochem Mol Biol 43, 888-897.

Saraiva, L.R., Ahuja, G., Ivandic, I., Syed, A.S., Marioni, J.C., Korsching, S.I., and Logan, D.W. (2015). Molecular and neuronal homology between the olfactory systems of zebrafish and mouse. Sci Rep 5, 11487.

  180  

Sato, Y., Miyasaka, N., and Yoshihara, Y. (2007). Hierarchical regulation of odorant receptor gene choice and subsequent axonal projection of olfactory sensory neurons in zebrafish. J Neurosci 27, 1606-1615.

Semmelhack, J.L., and Wang, J.W. (2009). Select Drosophila glomeruli mediate innate olfactory attraction and aversion. Nature 459, 218-223.

Sengupta, P., Chou, J.H., and Bargmann, C.I. (1996). odr-10 encodes a seven transmembrane domain olfactory receptor required for responses to the odorant diacetyl. Cell 84, 899-909.

Silbering, A.F., and Benton, R. (2010). Ionotropic and metabotropic mechanisms in chemoreception: 'chance or design'? EMBO Rep 11, 173-179.

Stensmyr, M.C., Dweck, H.K., Farhan, A., Ibba, I., Strutz, A., Mukunda, L., Linz, J., Grabe, V., Steck, K., Lavista-Llanos, S., et al. (2012). A conserved dedicated olfactory circuit for detecting harmful microbes in Drosophila. Cell 151, 1345-1357.

Sun, L., Wang, H., Hu, J., Han, J., Matsunami, H., and Luo, M. (2009). Guanylyl cyclase-D in the olfactory CO2 neurons is activated by bicarbonate. Proc Natl Acad Sci U S A 106, 2041-2046.

Touhara, K., and Vosshall, L.B. (2009). Sensing odorants and pheromones with chemosensory receptors. Annu Rev Physiol 71, 307-332.

Troemel, E.R., Chou, J.H., Dwyer, N.D., Colbert, H.A., and Bargmann, C.I. (1995). Divergent seven transmembrane receptors are candidate chemosensory receptors in C. elegans. Cell 83, 207-218.

Troemel, E.R., Kimmel, B.E., and Bargmann, C.I. (1997). Reprogramming chemotaxis responses: sensory neurons define olfactory preferences in C. elegans. Cell 91, 161-169.

Uchida, N., and Mainen, Z.F. (2003). Speed and accuracy of olfactory discrimination in the rat. Nat Neurosci 6, 1224-1229.

Vosshall, L.B., Amrein, H., Morozov, P.S., Rzhetsky, A., and Axel, R. (1999). A spatial map of olfactory receptor expression in the Drosophila antenna. Cell 96, 725-736.

  181  

Zhang, Y., Kim, I.J., Sanes, J.R., and Meister, M. (2012). The most numerous ganglion cell type of the mouse retina is a selective feature detector. Proc Natl Acad Sci U S A 109, E2391-2398.

Ziemann, A.E., Allen, J.E., Dahdaleh, N.S., Drebot, II, Coryell, M.W., Wunsch, A.M., Lynch, C.M., Faraci, F.M., Howard, M.A., 3rd, Welsh, M.J., et al. (2009). The amygdala is a chemosensor that detects carbon dioxide and acidosis to elicit fear behavior. Cell 139, 1012-1021.