Upload
danielle-newman
View
220
Download
0
Embed Size (px)
Citation preview
8/10/2019 Gawande PQE 5_6_11.pdf
1/16
FUNCTIONAL CONSEQUENCES OF VARIABILITY IN
MYCOBACTERIUM TUBERCULOSISAT THE HOST-
PATHOGEN INTERFACE
Richa GawandePreliminary Qualifying Exam
Biological Sciences in Public Health
05.06.11
8/10/2019 Gawande PQE 5_6_11.pdf
2/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
1
This proposal investigates variability at the host-pathogen interface and associated functional consequences
duringMycobacteriumtuberculosis(Mtb) infection.
Abstract and Significant AimsThe course of Tuberculosis (TB) disease is highly variable: most infected people never experience
active disease, while some develop progressive, primary disease soon after infection and others develop activedisease long after infection. While heterogeneity in host susceptibility contributes to the spectrum of outcomes
in TB infection, the trigger for reactivation, in particular, is often unknown1. Patients inhale only a few bacilli
at the time of infection, and variability in the virulence properties of these bacilli could modulate early events in
infection2. This is particularly significant if this variability translates to a functionally different early immune
response, as it could help explain differences in clinical outcome.
I propose that there exist bacterial components at the host-pathogen interface that modulate early eventsin infection. I will focus on features of the bacterial outer cell wall that are known to physically interact with
the host. In particular, a family of large, complex lipids of the outer cell wall is unique to pathogenicmycobacteria and has been associated with strain-based differences in virulence
3. Two members of the family
phthiocerol dimycocerosate (PDIM) and the Mtb-specific phenolic glycolipid (PGL) are important forvirulence in mouse models
4,5,6. For PDIM, we have characterized a frameshift mutation in its biosynthetic
locus. This mutation abrogates PDIM expression and occurs at high frequencies in vitro7. I propose that thisevent also occurs in vivo, and may translate to a heterogeneous population of bacilli that gets transmitted from
host to host. Here, I aim to further characterize heterogeneity of PDIM expression in infected hosts, and toforge a link between cell-to-cell heterogeneity in PDIM and heterogeneity in the immune response.
More stable genetic differences among Mtb strains may result in strain-based differences intransmission, lethality, immune modulation, and drug resistance8. For example, some strains associated with
hypervirulence in the East Asian lineage produce PGL and some strains, including those in the Euro-Americanlineage, do not
9. While PGL variability is one functional consequence of strain-based genetic diversity, many
interacting bacterial genetic factors may lead to functional variability in clinical presentation. Here, I propose acomprehensive approach to identify genetic requirements for growth for different clinical strains during mouse
infection. Further, I will focus on variability in requirements for, and production of, surface lipids at early times
in infection as a possible functional consequence for strain-based genetic diversity.Together, these experiments aim to determine the effects of genetic heterogeneity in Mtb virulence
factors on early events in infection.
Aim I: To test the hypothesis that Mtb can tune its early interactions with the host through high
frequency PDIM variability.
(1) I will identify a PDIM-specific cytokine fingerprint in infected macrophages that distinguishes between
PDIM(+) and PDIM(-) strains. (2) I will determine the single-cell immune response to PDIM(+), PDIM(-), andmixed infections, using fluorescent TB strains and fluorescent macrophage reporters. (3) I will measure PDIM
variation in single Mtb colonies derived from patient sputum by PCR resequencing and TLC analysis.
Aim II: To test the hypothesis that clinical strains from different lineages possess functionally differentrequirements for growth in vivo.
(1) I will conduct transposon-site capture and sequencing (TraCS) to identify genes that are differentiallyrequired for growth in mouse infections across clinical strains. (2) By lipidomics, I will compare the in vitro
lipid profiles in PGL-producing and -lacking strains. Together with the TraCS data, I will determine whetherPGL loss predicts in vivo essentiality and in vitro production of other surface lipids.
8/10/2019 Gawande PQE 5_6_11.pdf
3/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
2
Background
MTb, the causative agent of TB, is a unique pathogen highly divergent from model systems.Characterized by a thick, waxy cell wall, slow growth, and unique modes of virulence, MTb infection leads to
complex disease progression, which can be extremely difficult to diagnose10,11
. Infection can progress to activedisease, wherein bacteria can be aerosolized and efficiently passed on to a new host; upon inhalation of a few
bacilli, MTb is taken up by alveolar macrophages, and can establish its long-term phagosomal niche12,13.Mouse models have helped reveal many of the immunologic parameters required to control TB
infection. Upon exposure to Mtb, dendritic cells (DCs) migrate to the nearest draining lymph node and activateT cells
14,15. Infected macrophages activated by T cells deploy numerous innate immune effectors to quell
bacterial replication, including reactive oxygen species, antimicrobial peptides and TNFalpha, which, incoordination with the adaptive immune response, contributes to formation of the granuloma
16. In most
individuals, this process is efficient at controlling Mtb replication, which slows upon induction of the adaptiveimmune response, but doesnt result in sterilizing immunity, suggesting that Mtb possesses strategies to subvert
the immune response17
. Indeed, Mtb exhibits an exquisite capacity to evade the early macrophage assault,blocking phagolysosomal fusion, and forcing a balance of pro- and anti-inflammatory factors
18,19,20. Although
the exact molecular mechanisms underlying these host-pathogen interactions are not well-characterized, thecomplex lipids located in the outer cell wall of Mtb are thought to play a role in important events such as entry
and blocking phagosomal maturation21,22,23. The high lipid content of the mycobacterial cell wall makes itunique among other bacterial pathogens, and many of these lipids are critical determinants of mycobacterial
virulence4,5,24,25
.Here, I focus on two cell wall lipids important at the host-pathogen interface, PDIM and PGL, whose
production is restricted to pathogenic mycobacteria8,26,27
. PGL and PDIM share the same lipid core, and theirbiosynthesis and transport genes, encoding several multi-functional polyketide synthases, are clustered together
in the TB genome12,28
. PDIM mutants are attenuated from early on in mouse infections and reach a lowerplateau in the chronic phase of infection compared to wild-type Mtb
5,6. In a competition experiment, the
PDIM(-) strain exhibits lower fitness compared to the PDIM(+) strain, though not all PDIM(-) mutants arecleared (Figure 2, our lab). Infection of murine bone-marrow derived macrophages (BMDMs) with
PDIM(-) strains provokes an increased pro-inflammatory response compared to wild-type strains, suggesting
that PDIM helps Mtb evade the innate immune response29. Production of PGL is also unique to pathogenicmycobacteria such as Mycobacterium leprae, but is only produced in a limited number of Mtb strains
30.
Infection of mice with PGL-producing Mtb results in significantly reduced survival, and in BMDMs, reduction
in pro-inflammatory cytokines31,4
. This experimental data has been corroborated in clinical settings, where W-Beijing strains producing PGL are associated with epidemic spread, disseminated disease, and increased drug
resistance32,33,34
.The loss of PGL production through a 7-bp deletion anchors the divergence of the Euro-American lineage
from the remaining lineages in the Mtb complex (MTC) (Figure S1)9. The MTC includes all strains of Mtb andseven related mycobacterial species. Despite the ~99.9% genomic similarity characterizing the MTC, the globa
population of Mtb shows significant functional diversity in host tropism, clinical presentation of disease, anddrug sensitivity profile
,35. The globally dominant W-Beijing strains produce PGL; these lipids may contribute
to functional consequences in these strains. The population structure of the MTC, defined by genetic patterns ofinterspersed repetitive units, large sequence polymorphisms, and SNPs, consists of six main lineages, which are
geographically structured36,37. A sub-phylogeny of 100 diverse Mtb isolates suggests strong, stable associationsbetween Mtb strains and distinct geographic regions and host populations
38,39. As such, different host
environments may have contributed to the genetic divergence and evolution of these strains, resulting in geneticdifferences that are linked to unique mechanisms of survival.
The propensity to create genetic heterogeneity is limited in mycobacteria, which lack a mismatch repair(MMR) homolog but seem to compensate for this deficiency, exhibiting a very low base substitution rate
(similar to the MMR-positiveEscherichia coli)40,41
. One way mycobacteria may compensate is by reducing
8/10/2019 Gawande PQE 5_6_11.pdf
4/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
3
repetitive regions of the genome which are thought to be structurally unstable and prone to frameshift mutationswhich are repaired by MMR in other organisms
42. However, we still find such repetitive regions in the genome,
in particular, in the pps operon which encodes the phthiocerol portion of PDIM (Figure 1). Using an in vitroframeshift reporter, we measured the in vitro mutation rate of this sequence, and other repetitive sequences. In
the closely related Mycobacterium smegmatis, we found that the ppsA sequence mutates by the predicted +1mechanism at a rate 40X higher than the base substitution rate (Figure 3). I propose that this repetitive region
may be a high frequency mechanism for generating TB bacilli with differing capacities for in vivo virulence.The ability to diversify virulence determinants on a cell-to-cell and inter-strain basis is an important
strategy for many human pathogens43,44,45
. Very little is known about the mechanisms by which Mtb achievesfunctional diversity, and whether and how this diversity arises within an infected host. Through the proposed
work, I hope to understand how clinically-derived populations of Mtb generate functional variation in virulence,in an effort to understand the wide spectrum of TB disease.
Preliminary Data
__________________________________________________________________________________________Figure 1. ppsA mutation occurs in a repetitive region of the genome. The PDIM biosynthetic locus of
H37Rv. The ppsA repeat region occurs roughly in the middle of the ppsA gene, and the +1 mutation abrogatesPDIM expression.
__________________________________________________________________________________________
Figure 2. In a mixed competition, PDIM-negative strains are attenuated, but persist through 9 weeks.
MTb isolates from murine lung infections were sequenced for ppsA at varying weeks post infection. Ratio of
ppsA mutant/WT decreased significantly over time, though mutants persisted at 12% at 9 weeks. N indicatestotal number of lung isolates sequenced.
8/10/2019 Gawande PQE 5_6_11.pdf
5/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
4
Construct tested
__________________________________________________________________________________________
Figure 3. ppsA repeat has a 40X higher mutation rate than the base substitution rate, as calculated by an
in vitro frameshift reporter. Sequences of varying length were inserted upstream of the Kanamycin-resistance gene such that an insertion of +1 restores Kan resistance, otherwise the cell remains susceptible.
Fluctuation analysis was conducted to determine mutation rate.46
Research Design and Methods
Aim I: To test the hypothesis that Mtb can tune its early interactions with the host through high frequency
PDIM variability.
Aim 1A: To characterize the immune fingerprint which distinguishes between PDIM(+) and PDIM(-) at
early times in infection.
Previous studies suggest that PDIM helps Mtb evade the innate immune system29
. To determine howthe innate immune response differs in the absence or presence of PDIM, I will measure bulk cytokine
production of macrophages in response to PDIM(+) and wild-type H37Rv, and PDIM(-) H37Rv which containsthe ppsA frameshift mutation, at varying times post-infection.
I will use the RAW 246.7 murine macrophage-like cell line to measure cytokine response to PDIM(+)and PDIM(-) H37Rv infection. RAW 246.7 cells are widely used to study of signaling pathways and
inflammation47
. I will conduct pilot infection studies to optimize this system to capture the initial cytokineresponse to H37Rv infection. Based on the literature, I will infect RAW 246.7 cells at MOIs of 1:1 to 5:1
bacteria per cell with PDIM(+), PDIM(-) H37Rv in triplicate. I will potentially stimulate infected macrophageswith IFNgamma based on pilot studies. I will also maintain uninfected, activated cultures to control for
baseline cytokine expression, as well as cultures infected with an H37Rv mce mutant known to be
8/10/2019 Gawande PQE 5_6_11.pdf
6/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
5
hypostimulatory, to control for specific cytokine response. I will collect and filter culture supernatants at 4hours post infection, and 1, 3, and 5 days post infection. Macrophages will be lysed for CFU enumeration at
these times.I will measure cytokine production by standard sandwich ELISA assay. Measuring not just pro-
inflammatory cytokines but also immunomodulatory cytokines will allow me explore how PDIM helps silencethe immune response. I will measure the proinflammatory cytokines TNFalpha, Il-6 and IL12p40, and IL-18,
immunomodulatory cytokines such as IFNalpha, IFNbeta and IL-10, the inflammasome-mediated IL-1beta, andchemokines important for DC migration such as CCL3.
Anticipated Results and Limitations
I will determine whether PDIM(+) and PDIM(-) strains can differentially stimulate the innate immuneresponse in RAW 264.7 cells, compared to the uninfected, activated macrophages, and macrophages infected
with a hypostimulatory strain of Mtb as a control. The cytokines whose expression is most significantlydifferent between PDIM(+) and PDIM(-) infections (calculated by students t test), will be selected as the
PDIM-specific signature. Previous results from BMDMs indicate that TNFalpha may be a strong part of thissignature; the data from the above experiments will confirm and expand upon this result. A decreased release
of TNFalpha in PDIM(+) cells may be an important strategy for escaping initial host control of the infection.PDIM has been implicated in bacterial cell wall permeability; as such, loss of PDIM may change the
composition of the cell wall, exposing different molecules at the host-pathogen interface51. On the other hand,PDIM may be actively suppressing a pro-inflammatory response, for example, by skewing the receptor that
PDIM(+) bacilli use for entry. Production of anti-inflammatory cytokines will indicate whether PDIM may beactively inducing these cytokines. Upon measuring the cytokines listed above, I could use a semi-global
cytokine array to achieve better resolution of the full macrophage response to PDIM. Finally, it is possible thatPDIMs role is not to interact with the host at all, but to preserve Mtb in macrophages; in its absence, Mtb may
significantly weakened on the whole, leading to better control by macrophages.I anticipate that the technical details of the proposed experiments will not present any major challenges,
as the macrophage culture, infection, and immune assays are established methods in our lab. It may benecessary to switch or confirm results in other cell types, such as murine BMDMs or human monocyte-derived
macrophages.
Aim IB: To test the hypothesis that PDIM is a tuner of the host immune response at a single cell level.
Single cells lose PDIM over the course of infection through the slip-stranding mechanism proposed
above, thus PDIM variability is most likely to occur on the cell-to-cell level in infected hosts. Previous studieshave shown that macrophages respond differently to virulent and avirulent strains48. Similarly, the experiments
proposed above provide a rough-grain picture of whether PDIM can module innate immune cytokines in bulk.However, dominant signals released by a few activated macrophages may influence the cytokine response of the
entire macrophage population, and may not reflect the cell-to-cell variability associated with both bacterial andhost factors. While deconstructing dynamics of these activation signals is beyond the scope of this proposal, it
is important to examine additional dimensions when predicting how a mixed infection may modulate theimmune response. To characterize the differences in immune response to PDIM(+) and PDIM(-) bacteria, I aim
to answer the following questions. 1) Can the single-cell immune response distinguish between PDIM(+) andPDIM(-) infections? 2) Is the PDIM(+) response dominant? That is, is the single-cell response to a mixed
infection skewed towards a purely PDIM(+) or PDIM(-) infection, or does it reflect an intermediate level?I will use transcriptional reporters for innate immune genes to determine the immune response to
PDIM(-) and PDIM(+) infections at the single-cell level. Beaulieu et al have constructed fluorescence-basedpromoter-reporter constructs to measure single-cell responses to Mtb infection in RAW 264.7 cells
49. These
reporters include AmCyan-tagged TNFalpha promoter and EYFP-tagged IL-6 promoter. I propose to use thesereporters to measure the PDIM-specific single-cell response to Mtb infection. My bulk analyses will indicate
8/10/2019 Gawande PQE 5_6_11.pdf
7/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
6
which of the existing immune reporters has the potential to distinguish between PDIM(+) and PDIM(-)infection. I will select the best among these to establish the baseline single-cell system.
PDIM(+) and PDIM(-) strains will be marked with constitutively expressing fluorescent proteins such asGFP and mCherry, which we routinely use in our lab. Image acquisition and analysis will be done using our
DeltaVision wide-field fluorescence microscope and appropriate software, which others and I in the lab haveoptimized.
To answer the question of whether PDIM can modulate the cytokine response at a single cell level, I willinfect the reporter cells with fluorescently marked PDIM(+), PDIM(-), or an mce mutant of H37Rv. Based on
and using conditions developed in Aim IA, I will assess a range of activation methods to characterize the singlecell response to macrophages infected with these strains. Cells will be fixed at appropriate timepoints and
fluorescence intensity measured to ascertain per-cell cytokine transcription as it relates to bacterial type andburden. I predict that there will be underlying heterogeneity in cytokine transcription independent of PDIM
status, whose effect I will estimate from the uninfected, activated macrophage distribution. I will use thisdistribution to threshold and normalize the infected distributions to ensure consistent comparisons across
samples.To determine whether the PDIM(-) immune response is dominant in a PDIM(+)/PDIM(-) mixed
infection, I will infect macrophages with equal amounts of GFP-marked PDIM(+) and mCherry-markedPDIM(-) H37Rv, at an MOI of roughly 2:1 bacteria/macrophage. Based on the analyses above, I will determine
whether the PDIM (+) or PDIM(-) immune signature dominates in these mixed infections. Macrophages in thesame culture dish infected with only one strain will serve as a reference in addition to the experiments above.
As above, the analysis here must include subtraction of background heterogeneity in uninfected, activatedmacrophages, as well as normalization for number of bacteria per cell.
Anticipated Results and Limitations
The single-cell level data will provide one of the first insights into whether heterogeneity in the bacterialpopulation can direct heterogeneity of the early innate immune response. I anticipate that pure populations of
PDIM(+) or PDIM(-) Mtb will elicit a similar response trend on the single-macrophage level as they did on thepopulation level. Analysis of the single-cell distributions will reveal additional dimensions. For a given
infection, variance in fluorescence intensity will reflect how much variability exists in the cytokine response
across a population of macrophages. Temporal shifts in fluorescence will reveal kinetics and signal processingof cytokine transcriptioni.e., whether macrophages produce cytokine at once, or in a staggered manner. Ipredict that the kinetics will reflect some combination of intracellular and extracellular signals, but may be more
intracellularly driven for PDIM(-) compared to PDIM(+) infections due to the defect in immune evasionreported of PDIM(-) strains.
The mixed infection experiment could help determine the role of PDIM in infection. For example, ifPDIM is inhibiting a signaling pathway or receptor, its presence may be sufficient to shield a PDIM(-) bacillus
in the same macrophage. However, the intracellular dynamics are likely to be complex due to differentialphagosome localization and different receptors used. If necessary, I will stain with LysoTracker to obtain
information about phagosomal localization.The use of macrophage reporters has been well-established in other systems, and I dont anticipate
significant challenges in adapting these reporters to our system. However, underlying noise will be the majorchallenge in extracting a clean, PDIM-specific signal. It will be important to limit our immune readout to one or
two major cytokines, and to threshold our results based on the uninfected, activated macrophages as a negativecontrol. If necessary, I will apply dimension reduction techniques such as principal component analysis to
identify which condition(s) best explain the variance in my data.
8/10/2019 Gawande PQE 5_6_11.pdf
8/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
7
Aim 1C: To measure PDIM variation in Mtb derived from sputum of infected hosts.
The above experiments lay out a role for PDIM at the early host-pathogen interface, which is currently
unknown. In addition, the single-cell experiments may reveal a relationship between Mtb populationheterogeneity and host response heterogeneity. The unstable region in the PDIM biosynthesis locus, the high
energy cost involved in synthesizing PDIM, and the possibility that requirements for PDIM may be temporallyrestricted, suggest that a mixed population of bacilli may exist in an infected host. The PDIM biosynthetic
locus contains an unstable repeat region, which is prone to frameshift mutations (Figure 1). From workdescribed in Figure 2, the ppsA region offers a genetically tractable method for measuring PDIM variation in an
infected host. From these measurements, we found that indeed, PDIM(-) bacilli persist in infected mice. Todetermine whether PDIM variation exists in infected hosts, particularly at the transmission bottleneck, I propose
to measure PDIM expression in Mtb directly from patient sputa.I propose to directly plate smear-positive human sputum onto 7H10 plates to allow growth of single
mycobacterial colonies. I will plate undiluted sputum and serial dilutions to assure single colony resolution. Iwill include uninfected sputum spiked with PDIM(+) and PDIM(-) Mtb as controls. I will then pick a single
colony and conduct single-colony PCR to amplify an approximately 600bp region of ppsA encompassing therepeat region. I have successfully PCR amplified and sequenced the ppsA region from ~ 300 mycobacterial
colonies from mouse lung isolates using this approach (Figure 2). Multiple sequence alignment of the regionsurrounding the ppsA repeat in sequenced clinical strains show that this region is well-conserved, and thus I
anticipate that the H37Rv ppsA primers will also work for the strains isolated from human sputum. Colonieswhose sequencing traces contain the frameshift mutation in the ppsA repeat region will be considered evidence
for PDIM loss inside the infected host. For a subset of these colonies, I will perform chloroform-methanolextractions to isolate cell wall lipids, and will measure PDIM production by thin-layer chromatography (TLC)
using standard methods.We have an ongoing collaboration with investigators in South Africa and Tanzania. Our collaborators in
Tanzania have generously provided us with 12 smear- and culture-positive frozen sputum samples, for whichour lab has obtained approval from the Harvard Committee on Microbiological Safety to use for research
purposes. I will conduct pilot studies on this limited supply of sputum. If I am able to successfully amplifyppsA from these sputum-derived colonies, I hope to travel to Durban, South Africa to have access to a greater
number of smear-positive sputum samples.
Anticipated Results and Limitations
PCR of the ppsA repeat region from colonies derived from human sputum will indicate whether
heterogeneity in the ppsA region arose within the infected host. Sequencing traces will show whether a ppsAmutant (ppsA(mut)), ppsA wild-type (ppsA(WT)), or mixed population existed in the colony which was PCR
amplified. From mutant accumulation studies from our lab, the likelihood of a frameshift mutation occurringduring the growth of a single PDIM(+) bacterium into one colony (requiring approximately 10 replications) is
nearly zero. In any case, the maximum percentage of revertants (either ppsA(mut) to ppsA(wt) or vice versa)arising from a single bacterium growing into a colony is 50%, which can be detected by PCR followed by
sequencing (preliminary data, our lab). Therefore, I will consider any colonies with a mixed ppsA sequence asan ambiguous result, and any colonies with a ppsA(mut) sequence as evidence that the ppsA(mut) isolate arose
within the host, and not on the plate.If I dont recover any ppsA(mut) populations, it may be because the ppsA mutation could arise at
frequencies too low to be represented in a single sputum sample. Alternatively, plating may impose a selection.To overcome these challenges, I could use highly sensitive methods to detect ppsA(mut) directly from sputum,
without plating. I could use real time PCR using specific molecular beacons to detect the ppsA mutationdirectly from sputum, as has been recently pioneered in the GeneXpert Mtb diagnostic
50.
Together, these studies may for the first time demonstrate that high frequency genetic variation of avirulence factor occurs in infected hosts. The proposed macrophage studies will indicate whether this variation
can functionally modulate the early host pathogen interaction.
8/10/2019 Gawande PQE 5_6_11.pdf
9/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
8
Aim II: To test the hypothesis that clinical strains from different lineages possess functionally different
requirements for growth in vivo.
Large sequence polymorphisms (LSPs), SNPs, and other genetic patterns define distinct lineages of the
MTC, which was once thought to be a genetically homogeneous population36
. These genetic differences havebeen associated with wide host tropism, spectrum of disease, and drug sensitivity profiles, yet the functional
consequences of genetic diversity have not been studied in a comprehensive manner at the molecular level. Ihypothesize that different clinical lineages use overlapping but distinct strategies for in vivo growth and
survival. I propose to investigate the genetic requirements for in vivo infection in different Mtb lineages tounderstand how these lineages differ in their manipulation and interaction with the host environment.
Aim IIA: To test the hypothesis that clinical strains exhibit heterogeneity in genetic requirements for
survival during infection.
I propose to test the hypothesis that clinical strains from different lineages possess different genetic
requirements for growth. I propose to test this hypothesis using TraCS, a forward genetics approach based onthe TraSH (transposon-site hybridization) method pioneered in Mtb 10 years ago
51. TraSH utilizes phage
transduction to deliver a mariner transposon which randomly integrates into TA dinucleotides across thegenome, generating an Mtb library of ~10
5transposon mutants. This library can then be subjected to a variety
of growth conditions. In TraSH, surviving mutants are used to generate DNA probes complementary to thejunction of the transposon and the chromosome, and are then hybridized to an Mtb microarray. TraCS allows
direct Illumina sequencing of the transposon-chromosome junctions, which affords greater resolution oftransposon insertion sites and a much larger dynamic range in estimating the fitness of each mutant.
TraSH has been instrumental in revealing the genetic requirements for growth in H37Rv. Specifically,214 genes were differentially required for growth in mouse spleen at varying times post infection compared to
growth on a 7H10 agar plate52,53
. Although many of the 214 genes have unknown functions, some encodeproducts which are known to vary between lineages, such as PGL. Here, I propose to conduct strain-based
TraCS to determine whether clinical strains from divergent lineages have different requirements for growth in amouse.
To conduct strain-based TraCS, I will create a transposon library of mutants for each clinical strain
included in the study (see Table 1). The methodology used to create the transposon library has been well-established. We will use existing protocols to introduce the transposon into every gene of the Mtb genome. Wewill recover surviving mutants by plating on rich media and scraping single colonies together to form at once,
the transposon library and the in vitro pool. Libraries already exist in our lab for H37Rv PDIM(+), H37RvPDIM(-), and the clinical Euro-American strain, Erdman. Upon extracting genomic DNA from this pool, we
will construct in vitro probes for Illumina sequencing. Construction of probes has been validated in our lab, andinvolves, briefly, shearing of chromosomal DNA, end repair and adapter ligation. We will PCR amplify
transposon junctions using one primer complementary to the adapter with an additional portion homologous tothe Illumina sequencing chip, and a second primer complementary to the transposon junction.
We will align our sequencing reads to the closest reference genome, which may need to be optimizeddepending on the composition of our clinical strain panel. We will filter our reads to ensure that the transposon
junction is present in each of them, and will then count the number of reads that correspond to each TA site inthe genome, obtaining the read depth. Read depth at a given TA site will correspond to the fitness of that
transposon insertion mutant in the pool. The second useful metric is percent of TA sites that have transposoninsertions for given region of the genome. This metric allows both a gene-level analysis and a sub-gene,
domain-level analysis. Using a non-parametric rank sum test, we will combine these two metrics to determinewhich genes have significantly underrepresented TA hits in the in vivo pool compared to the in vitro pool.
Genes that are underrepresented in the in vivo pool will be considered essential for growth in a mouse.I will select strains to ensure diversity across the Euro American, Indo Oceanic, West African, East
Asian, and East African/Indian lineages. Selection of strains will be subject to public availability. I will
8/10/2019 Gawande PQE 5_6_11.pdf
10/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
9
attempt to balance the number of sequenced, well-characterized, in vitro-passaged strains with clinical isolatesabout which relatively little is known. A candidate panel of strains is indicated in Table 1. PGL production
varies across these strains, allowing a sub-analysis (see Aim IIB) to correlate PGL production with requirementsfor growth in vitro and in vivo.
Strain name Sequenced? PGL status Lineage37
Reason for including
HN878 Yes + East Asian Reference
621 No ? East Asian Immunologic data, our lab
NHN5 Yes - East Asian No PGL
K85 Yes ? West Africa 2 Reference
668 No ? East African Indian Immunologic data, our lab
947/01 No + Indo Oceanic Transcriptome/growth54
K21 Yes + Indo Oceanic Reference
581 No - Euro American Immunologic data, our lab
Erdman Yes - Euro American Reference
Table 1. Candidate strains to be included in TraCS.
To create the in vivo pool, we will infect C57Bl/6 mice with the clinical strain libraries by high-doseaerosol. The strains tested here have emerged because of successful pathogenesis strategies in the context of the
human host. Within the context of the experimental mouse model, aerosol infection imposes selection whichmost closely recapitulates that of the human host, compared to the intravenous route. This would be the first
TraSH/TraCS experiment to use the aerosol route of infection, and may reveal important differences betweenthese two in vivo environments as they impose selection the bacterium. Aerosol doses with 20,000 CFU per
mouse have been successfully performed in our lab, thus I propose to distribute the TraSH library of 105
mutants equally across five mice per strain, in triplicate. At two and four weeks post infection, I will pool
organs from each set of 5 mice, and will treat this pool as a single sample per time point.I first propose to do TraCS on existing H37Rv PDIM(+) and PDIM(-) TraSH libraries. This experiment
will allow a comprehensive characterization of the role of PDIM in infection, and will be an essential pilot
study to validate our infection parameters. I propose to infect PDIM(+) and PDIM(-) strains by high doseaerosol. We will harvest lungs at two and four weeks post-infection to reveal relative fitness before and after
onset of the adaptive immune response. We will titer the infected organs, plate mutants, isolate genomic DNA,create TraCS probes as described above, and send probes for Illumina sequencing. These pilot experiments will
help us determine the feasibility of the aerosol route of infection, which we will assess by mouse survival,recovery of mutant pools, and complexity of mutant pools as compared to TraSH by intravenous infection.
Furthermore, this experiment will be a natural extension of Aim I discussed above, revealing genes whoseproducts compensate for PDIM loss, potentially revealing roles for PDIM during infection.
Upon successful optimization of infection parameters and analysis methods, we will repeat the TraCSexperiment using clinical strain libraries as indicated in Table 1. For each clinical strain, we will recover
surviving mutants to form an in vivo pool as described above. Upon sequencing and analysis as described
above, we will measure in vivo fitness for viable mutants at two weeks and four weeks post-infection (or asinformed by the pilot experiment) as compared to each mutants in vitro fitness. Mutations that result insignificant change in fitness in vivo compared to in vitro will be selected for further analysis.
Anticipated Results and Limitations
Generation of the strain-based TraCS dataset will be a significant contribution to our knowledge ofglobal strain diversity in Mtb. Each step of the experimental process proposed above is a significant and novel
contribution to the field. The pilot experiment allows a comparison of genes required for growth of H37Rvduring aerosol infection in which the lung carries a large bacterial burden versus the previously established
8/10/2019 Gawande PQE 5_6_11.pdf
11/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
10
intravenous route which results in a large bacterial burden in the spleen. I predict there will be significantdifferences in the subset of required genes during the initial two weeks of infection, as aerosol infection may
select for different cell surface molecules on infecting bacilli. Furthermore, these experiments may show therole(s) for PDIM during infection, by revealing essential genes in its absence.
Our approach to find functional consequences of strain-based genetic differences would be anunparalleled characterization of strain diversity in Mtb, at the molecular level. I predict there will be a
significant core set of genes required for growth in all strains. This core set may represent a reduced subset ofthe genes required for infection as defined previously for H37Rv. Furthermore, I predict the in vitro core set
will be different from in vivo set, the latter being enriched for genes specific to intracellular survival, such asstress response, hypoxia, iron limitation, and cell wall remodeling. Finally, the pairwise comparisons of strains
within and between lineages, as well as multiple strain comparisons will reveal lineage and strain-specificessential genes (by ANOVA and the Tukey-Kramer HSD and post-hoc tests). Transcriptional profiling of
clinical strains and single-strain studies suggest that the gene expression profile for strains within a lineage aremore correlated than strains from different lineages53. For example, the mce4 locus, encoding an important
cholesterol uptake system, was upregulated, while nitrate reductase, important for in vivo survival, wasdownregulated in ancient strains compared to modern strains. Despite these transcriptional data, we may find
that the genes essential for surviving in a mouse are the same across clinical strains. This finding may indicatethat there is a highly conserved core genome across the strains tested, and that a high degree of functional
redundancy in Mtb genomes may compensate for the genetic differences between strains. This finding mayalso indicate that host variability is a more dominant driver of functional diversity in strains than bacterial
variability, a hypothesis we could test by conducting TraCS in mutant strains of mice.There are technical limitations to our experimental setup that may skew our ability to answer the
questions posed above. Some of these limitations are specific to the TraCS methodology, such as biasintroduced by type of PCR used to amplify the TraCS library. We can measure some of this bias by conducting
quality checks on our data to measure the ratio of specific to non-specific PCR products amplified. In addition,despite its advances, Illumina sequencing affords poor resolution of repetitive regions of the genome; we will
analyze our data in light of this knowledge. There are also technical challenges posed by clinical strainsthemselves, including differential ability to grow on 7H10 plates and unknown in vivo growth kinetics leading
to different courses of infection. While we will be able to gauge some of these growth differences by assessing
growth curves and comparing O.D. to CFU, using previously characterized clinical isolates will allow us todraw on published data to predict the growth of these strains. Finally, high-dose aerosol infection may present atechnical challenge, as we may find from our pilot experiments that PDIM(+) H37Rv delivered by high dose
aerosol causes excessive pathology and reduced survival for mice. In this case, we will conduct theexperiments proposed above using intravenous injection. While we will not capture important events
surrounding the initial colonization of the lung, we will still be able to identify significant differences inrequirements for intracellular growth across clinical strains.
Importantly, I recognize that this is a large undertaking, and thus propose to do the work describedabove in collaboration with a post-doctoral fellow in the Fortune Lab.
Aim IIB: To test the hypothesis that strains that produce PGL have functionally different requirements
for surface lipids during in vivo growth.The data generated from the TraCS experiment will reveal which genes are differentially required for in
vivo growth between clinical strains. I propose to investigate a portion of the TraCS data to establish thefunctional consequences of a known source of inter-strain heterogeneity, namely PGL production. Both PGL-
producing and PGL-lacking strains survive and persist globally. I hypothesize that loss of PGL production,particularly in the Euro-American strains, resulted in compensatory evolution of other surface lipids in these
strains. Indeed, there is in vitro evidence that loss of PGL-related lipids results in increased sulfolipid (SL)production
55. In vivo, PGL modulates the early host-pathogen interface. Some studies also position PGL and
other surface lipids as a physical shield from host defenses and a metabolic sink for storing carbon units during
8/10/2019 Gawande PQE 5_6_11.pdf
12/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
11
cholesterol metabolism55,56,57
. Strains that dont produce PGL may rely more heavily on other surface lipids toperform these actions.
To test the hypothesis that differential PGL production creates different requirements for other surfacelipids during mouse infections, I propose to cluster the strain-based TraCS data by PGL production. I will
identify genes differentially required for mouse infection in PGL(+) versus PGL(-) strains, and determinewhether surface lipids are enriched in this set, using the hypergeometric test. I will furthermore test the
functional consequences of differential PGL production by measuring in vitro production of known surfacelipids such as PDIM and SL, as well as unknown lipids. I will accomplish this by conducting mass-
spectrometry-based lipidomics for the strains included in the TraCS study, in collaboration with the laboratoryof Dr. Branch Moody, where this methodology is well-established. Recognizing that different in vitro
conditions can alter a strains lipid profile, I will grow the clinical strains, a PGL(-) version of HN878 as anisogenic control, and a PDIM(-) H37Rv mutant, to equal O.D.s in standard glycerol-containing 7H9 media, and
7H9 containing propionate, known to foster lipid production that is more reflective of growth in vivo54
. Ipredict that this analysis will reveal lipid-based differences across clinical strains. Together, the TraCS and
lipidomics data will help me assess whether PGL production is an important predictor of a clinical strains lipidprofile and its requirements for surface lipids during infection.
Anticipated Results and Limitations
The experiment proposed above is a tractable way to parse trends from the large TraCS dataset, andestablish functional consequences for specific genetic requirements for growth. I predict that PGLs metabolic
and immunomodulatory niche is filled by other lipids in PGL-lacking strains, resulting in different lipid profilesacross these strains. We may be able to identify some of these lipids based on differential essentiality in TraCS
data, and a complementing approach, in vitro lipidomics. In particular, I predict that the genes contributing toPDIM production and transport will be more essential in vivo in PGL-lacking strains, and that this will
correspond to increased in vitro PDIM production. However, there may be significant strain-to-strainheterogeneity in PDIM and PGL production that does not correlate with lineage; our lab has observed similar
heterogeneity in ESX1 substrate secretion in a small panel of clinical strains. As such, I may find that PGLproduction is not a significant predictor of strain-based lipid profiles. In this case, I will look to the lipidomics
data for other candidates whose production may track with lineage, and the TraCS data for other cell wall
virulence factors (such as ESX1 substrates) that may be better predictors of strain-based lipid profiles.
Strain-based heterogeneity must be incorporated into the existing paradigm of Mtb pathogenesis. Strains
differing in their modes of virulence may require specialized treatment programs, may possess unique antigenicand drug resistance signatures, requiring more specificity in diagnostics, and may result in heterogeneous
immune outcome. Furthermore, examining the relationship between Mtb phylogeny and functional diversity inpathogenesis may shed light on the long co-evolution of Mtb and its hosts, revealing valuable information about
differing host environments. The experiments proposed here examine functional consequences of bothunstable, cell-to-cell variability and stable, lineage-driven variability, at the host-pathogen interface.
8/10/2019 Gawande PQE 5_6_11.pdf
13/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
12
Supplemental Information
Figure S1. The MTC phylogeny, as determined by large sequence polymorphisms, exhibits a tightgeographical structure. The Euro-American lineage diverged partially as a result of a 7bp deletion leading to
loss of PGL production37
References
1Mycobacterium tuberculosis, macrophages, and the innate immune response: does common variation matter?. (2007) pp. 1-20
2Riley et al. Aerial dissemination of pulmonary tuberculosis. A two-year study of contagion in a tuberculosis ward. 1959. American
Journal of Epidemiology (1995) vol. 142 (1) pp. 3-14
3Daff and Draper. The envelope layers of mycobacteria with reference to their pathogenicity. Adv Microb Physiol (1998) vol. 39 pp.
131-203
4Reed et al. A glycolipid of hypervirulent tuberculosis strains that inhibits the innate immune response. Nature (2004) vol. 431 (7004)
pp. 84-7
8/10/2019 Gawande PQE 5_6_11.pdf
14/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
13
5Cox et al. Complex lipid determines tissue-specific replication of Mycobacterium tuberculosis in mice. Nature (1999) vol. 402
(6757) pp. 79-83
6Camacho et al. Identification of a virulence gene cluster of Mycobacterium tuberculosis by signature-tagged transposon
mutagenesis. Mol Microbiol (1999) vol. 34 (2) pp. 257-67
7Domenech and Reed. Rapid and spontaneous loss of phthiocerol dimycocerosate (PDIM) from Mycobacterium tuberculosis grownin vitro: implications for virulence studies. Microbiology (2009) vol. 155 (11) pp. 3532
8Hershberg et al. High Functional Diversity in Mycobacterium tuberculosis Driven by Genetic Drift and Human Demography. PLoS
Biol (2008) vol. 6 (12) pp. e31
9Reed et al. The W-Beijing Lineage of Mycobacterium tuberculosis Overproduces Triglycerides and Has the DosR Dormancy
Regulon Constitutively Upregulated. Journal of Bacteriology (2007) vol. 189 (7) pp. 2583-2589
10Gomez and McKinney. M. tuberculosis persistence, latency, and drug tolerance. Tuberculosis (Edinb) (2004) vol. 84 (1-2) pp. 29-44
11Stop TB Partnership. Geneva: WHO; 2006. The global plan to stop TB 2006201
12Smith. Mycobacterium tuberculosis pathogenesis and molecular determinants of virulence. Clin Microbiol Rev (2003) vol. 16 (3)
pp. 463-96
13Pieters. Mycobacterium tuberculosis and the macrophage: maintaining a balance. Cell Host Microbe (2008) vol. 3 (6) pp. 399-407
14Jiao et al. Dendritic cells are host cells for mycobacteria in vivo that trigger innate and acquired immunity. J Immunol (2002) vol.168 (3) pp. 1294-301
15Tascon et al. Mycobacterium tuberculosis-activated dendritic cells induce protective immunity in mice. Immunology (2000) vol. 99
(3) pp. 473-80
16North and Jung. I mmunity toT uberculosis. Annu. Rev. Immunol. (2004) vol. 22 (1) pp. 599-623
17Saunders and Britton. Life and death in the granuloma: immunopathology of tuberculosis. Immunol Cell Biol (2007) vol. 85 (2) pp.
103-111
18Russell. Mycobacterium and the coat of many lipids. The Journal of Cell Biology (2002) vol. 158 (3) pp. 421-426
19Malik et al. Cutting edge: Mycobacterium tuberculosis blocks Ca2+ signaling and phagosome maturation in human macrophages
via specific inhibition of sphingosine kinase. J Immunol (2003) vol. 170 (6) pp. 2811-5
20Russell. Who puts the tubercle in tuberculosis?. Nat Rev Micro (2007) vol. 5 (1) pp. 39-47
21Villeneuve. Mycobacteria use their surface-exposed glycolipids to infect human macrophages through a receptor-dependent process
The Journal of Lipid Research (2004) vol. 46 (3) pp. 475-483
22Fratti et al. Mycobacterium tuberculosis glycosylated phosphatidylinositol causes phagosome maturation arrest. Proc Natl Acad Sci
USA (2003) vol. 100 (9) pp. 5437-42
23Guenin-Mac et al. Lipids of Pathogenic Mycobacteria: Contributions to Virulence and Host Immune Suppression. Transboundaryand Emerging Diseases (2009) vol. 56 (6-7) pp. 255-268
24Daff and Draper. The envelope layers of mycobacteria with reference to their pathogenicity. Adv Microb Physiol (1998) vol. 39
pp. 131-203
25Daffe et al. Predominant structural features of the cell wall arabinogalactan of Mycobacterium tuberculosis as revealed through
characterization of oligoglycosyl alditol fragments by gas chromatography/mass spectrometry and by 1H and 13C NMR analyses. J
Biol Chem (1990) vol. 265 (12) pp. 6734-43
8/10/2019 Gawande PQE 5_6_11.pdf
15/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
14
26Brennan and Nikaido. The envelope of mycobacteria. Annu Rev Biochem (1995) vol. 64 pp. 29-63
27Daff and Laneelle. Distribution of phthiocerol diester, phenolic mycosides and related compounds in mycobacteria. J Gen
Microbiol (1988) vol. 134 (7) pp. 2049-55
28Cole et al. Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature (1998) vol. 393
(6685) pp. 537-44
29Rousseau et al. Production of phthiocerol dimycocerosates protects Mycobacterium tuberculosis from the cidal activity of reactive
nitrogen intermediates produced by macrophages and modulates the early immune response to infection. Cell Microbiol (2004) vol. 6
(3) pp. 277-287
30Muoz et al. Distribution of surface-exposed antigenic glycolipids in recent clinical isolates of Mycobacterium tuberculosis. Res
Microbiol (1997) vol. 148 (5) pp. 405-12
31Manca et al. Differential monocyte activation underlies strain-specific Mycobacterium tuberculosis pathogenesis. Infection and
Immunity (2004) vol. 72 (9) pp. 5511-4
32
Buu et al. The Beijing genotype is associated with young age and multidrug-resistant tuberculosis in rural Vietnam. Int J TubercLung Dis (2009) vol. 13 (7) pp. 900-6
33Sun et al. Association of Mycobacterium tuberculosis Beijing genotype with tuberculosis relapse in Singapore. Epidemiol. Infect.
(2005) vol. 134 (02) pp. 329
34Kong et al. Association between Mycobacterium tuberculosis Beijing/W Lineage Strain Infection and Extrathoracic Tuberculosis:
Insights from Epidemiologic and Clinical Characterization of the Three Principal Genetic Groups of M. tuberculosis Clinical Isolates.
J Clin Microbiol (2007) vol. 45 (2) pp. 409-414
35Hershberg et al. High Functional Diversity in Mycobacterium tuberculosis Driven by Genetic Drift and Human Demography. PLoS
Biol (2008) vol. 6 (12) pp. e311
36Brosch et al. A new evolutionary scenario for the Mycobacterium tuberculosis complex. Proc Natl Acad Sci USA (2002) vol. 99 (6)
pp. 3684-9
37Gagneux et al. Variable host-pathogen compatibility in Mycobacterium tuberculosis. Proc Natl Acad Sci USA (2006) vol. 103 (8)
pp. 2869-73
38Hirsh et al. Stable association between strains of Mycobacterium tuberculosis and their human host populations. Proc Natl Acad Sci
USA (2004) vol. 101 (14) pp. 4871-6
39Wirth et al. Origin, Spread and Demography of the Mycobacterium tuberculosis Complex. PLoS Pathog (2008) vol. 4 (9) pp.
e1000160
40David. Probability distribution of drug-resistant mutants in unselected populations of Mycobacterium tuberculosis. Appl Microbiol
(1970) vol. 20 (5) pp. 810-4
41Unpublished data, Fortune Laboratory.
42Wanner et al. Stabilization of the genome of the mismatch repair deficient Mycobacterium tuberculosis by context-dependent codon
choice. BMC Genomics 2008 9:249 (2008) vol. 9 (1) pp. 249
43van der Woude and Bumler. Phase and antigenic variation in bacteria. Clin Microbiol Rev (2004) vol. 17 (3) pp. 581-611, table of
contents
44Ghosh et al. Pathogenic consequences of Neisseria gonorrhoeae pilin glycan variation. Microbes Infect (2004) vol. 6 (7) pp. 693-
701
8/10/2019 Gawande PQE 5_6_11.pdf
16/16
Variability at the Host-Pathogen Interface in Mtb Richa Gawande // 05.06.11
15
45Recker et al. Antigenic Variation in Plasmodium falciparum Malaria Involves a Highly Structured Switching Pattern. PLoS Pathog
(2011) vol. 7 (3) pp. e1001306
46Rosche and Foster. Determining mutation rates in bacterial populations. Methods (2000) vol. 20 (1) pp. 4-17
47Wang et al. Pattern of proinflammatory cytokine induction in RAW264.7 mouse macrophages is identical for virulent and
attenuated Borrelia burgdorferi. J Immunol (2008) vol. 180 (12) pp. 8306-15
48Beisiegel et al. Combination of host susceptibility and Mycobacterium tuberculosisvirulence define gene expression profile in the
host. Eur. J. Immunol. (2009) vol. 39 (12) pp. 3369-3384
49Beaulieu et al. Genome-Wide Screen for Mycobacterium tuberculosis Genes That Regulate Host Immunity. PLoS ONE (2010) vol.
5 (12) pp. e15120
50Helb et al. Rapid detection of Mycobacterium tuberculosis and rifampin resistance by use of on-demand, near-patient technology. J
Clin Microbiol (2010) vol. 48 (1) pp. 229-37
51Sassetti et al. Comprehensive identification of conditionally essential genes in mycobacteria. Proc Natl Acad Sci USA (2001) vol.
98 (22) pp. 12712-7
52Sassetti and Rubin. Genetic requirements for mycobacterial survival during infection. Proc Natl Acad Sci USA (2003) vol. 100 (22)pp. 12989-94
53Sassetti et al. Genes required for mycobacterial growth defined by high density mutagenesis. Mol Microbiol (2003) vol. 48 (1) pp.
77-84
54Homolka et al. Functional Genetic Diversity among Mycobacterium tuberculosis Complex Clinical Isolates: Delineation of
Conserved Core and Lineage-Specific Transcriptomes during Intracellular Survival. PLoS Pathog (2010) vol. 6 (7) pp. e1000988
55Jain et al. Lipidomics reveals control of Mycobacterium tuberculosis virulence lipids via metabolic coupling. Proc Natl Acad Sci
USA (2007) vol. 104 (12) pp. 5133-8
56Camacho et al. Analysis of the phthiocerol dimycocerosate locus of Mycobacterium tuberculosis. Evidence that this lipid is
involved in the cell wall permeability barrier. J Biol Chem (2001) vol. 276 (23) pp. 19845-54
57Yang et al. Cholesterol Metabolism Increases the Metabolic Pool of Propionate in Mycobacterium tuberculosis. Biochemistry
(2009) vol. 48 (18) pp. 3819-3821