www.elsevier.com/locate/brainres

Brain Research 1000 (2004) 60–63

Mini review

Connectional neuroanatomy: the changing scene

Kathleen S. Rockland*

Laboratory for Cortical Organization and Systematics, Brain Science Institute, RIKEN, 2-1 Hirosawa, Wako-shi, Saitama 351-0198, Japan

Accepted 11 December 2003

Abstract

This article gives a brief overview of cortical connectivity, as this has been investigated by anatomic tracer studies over the past 25 years.

Open questions are discussed, from the perspective that connectivity studies are in a transition period when both techniques and conceptual

frameworks are rapidly changing.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Area specialization; Connectional convergence; Pyramidal cell type; Tracer study

1. Introduction

On the occasion of the BRES volume 1000, I would like

to briefly comment on the current state of cortical connec-

tivity studies. More and more clearly, these seem poised to

enter a new era, characterized by new techniques from

molecular biology, and a greater attention to cellular and

regional diversity [4,5,12,22]. The promise is that this will

lead, not to overwhelming detail, but to a more complete

synthesis across the molecular, cellular, and functional

domains.

It is interesting to look back and take as reference point

an early paper I co-authored with D.N. Pandya [25]. In that

paper, by using the newly developed HRP tracer, we were

able to demonstrate neurons of origin for several sets of

cortical connections. By also using complementary injec-

tions of anterograde tracers, we demonstrated, for the early

visual areas in macaque monkeys, a certain laminar signa-

ture of what became known as ‘‘feedforward’ and ‘‘feed-

back’’ connections. Subsequent tracer experiments by many

groups contributed data on laminae of origin and termina-

tion for different connections in different areas and species.

These were often referenced to three categories, feed-

forward, feedback, and lateral, in what was to be an

enormously influential model of hierarchical architecture

of cortical areas [9,20]. In retrospect, this model was almost

too successful, in that it tended to overshadow many other

0006-8993/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.brainres.2003.12.036

* Tel.: +81-48-462-1111x7111; fax: +81-48-467-6420.

E-mail address: rockland@brain.riken.go.jp (K.S. Rockland).

questions related to microcircuitry and network organization

[2,24]. Some of these are recapitulated in the next two

sections, along with comments on limitations and open

questions in the field.

2. Background

Conventional retrograde and anterograde tracers are well

suited for establishing the basic ‘‘wiring diagram,’’ or what

is connected to what; but connections are not just routes by

which ‘‘area A projects to area B.’’ Factors that need to be

considered include convergence and divergence, connec-

tional efficacy, and anatomical and functional subtypes.

Tracer studies have been less effective in addressing these

more detailed factors; but examples of some results are

given below.

Retrograde labeling in vivo [19] or combined with

intracellular fills in slice [16] demonstrated that neurons

giving rise to different projections had distinctive patterns of

dendritic and axonal arbors. Whether the patterns are

stereotyped within each projection system requires further

criteria, and remains actively debated [4,17]. Retrograde

tracers combined with other markers were another approach

to identifying projection-specific characteristics. In contrast

with interneurons, however, relatively few markers have

been available that differentiate among pyramidal cells. One

study of visual cortical connections combined immunohis-

tochemistry for neurofilament protein, labeled by antibody

to SMI32, with retrograde labeling of several groups of

efferent neurons [14]. A distinct distribution of neurofila-

K.S. Rockland / Brain Research 1000 (2004) 60–63 61

ment protein was indicated for some feedforward and

feedback projections.

The high-resolution anterograde tracers (PHA-L, biocytin,

and the biotinylated dextran amines) produced Golgi-

like images, which allowed further assessment and quanti-

fication of connections. Here, too, the evidence pointed

to further subtypes [23,24]. For example, the distinction

between feedforward and feedback connections was rein-

forced by differences in arbor size (Fig. 1): feedback con-

nections to layer 1 are widely divergent (>1.0 mm), while

individual arbors of feedforward axons are smaller (f 0.25

mm each). Correlative information on parent neurons would

be important.

In a few instances, anterograde labeling was able to show

terminations with signature distinct morphology. The clearest

example to date may be the two types of corticopulvinar ter-

minations [23]. So far, morphological criteria alone have not

been widely successful in this regard. There is some indica-

tion, however, that pathway-specific differences may be de-

finable by other means, such as receptor combinations [18].

Intracellular [15] or juxtacellular [33] filling improved on

extracellular injections in that both revealed the cell of origin,

as well as axon arborization. This allows fuller evaluation of

diversity or uniformity among neurons within a defined

projection, or even between neighboring neurons. Further

advances in this respect might be expected from newer

techniques producing Golgi-like images, such as ‘‘DiOlis-

tics’’ [10] or viral tracing techniques [7,31].

Fig. 1. (A) and, higher magnification from arrow, (B) Feedback terminations in are

adjacent sections have been merged in Adobe Photoshop (at the asterisk in (A). No

dextran amine in area V1 shows anterogradely labeled terminations in layer 4 of V

feedback-projecting neurons. Arrows emphasize the bistratified distribution of the n

bars = 200 Am (A, C), 50 Am (B).

Other data from conventional tracers are as follows.

Estimates can be made of connectional density, either by

the number of retrogradely filled neurons [6], or by the

density per unit area of anterogradely labeled terminals.

These are a useful estimate of connectional weight (or

efficacy), but need to be corroborated by other criteria.

Double retrograde tracer experiments provide evidence

for a subpopulation of neurons that branch to several areas.

This is confirmed by single axon reconstruction, which also

shows the occurrence of collaterals to several areas [23,30].

These ‘‘manifold neurons’’ [29] presumably make-up a

subpopulation of many if not all projection systems; and

their other characteristics and functional roles need to be

more fully investigated.

Laminar patterns, especially outside the early visual areas,

actually show major inter-area variability. Studies using

double anterograde tracers dramatically show that converg-

ing connections have differential overlapping or interdigitat-

ing patterns in different target areas [26]. Comparison of

single tracers across areas supports the same conclusion. To

give one example, corticothalamic terminations are common-

ly considered to arise from neurons in layer 6, but there area

specific ‘‘disjunctive features’’ (Fig. 12 in Ref. [11]). Further

systematization of these patterns is needed, especially in

correlation with other markers, such as area-specific gene

expression.

Retrograde tracers have been used to label pyramidal

neurons in combination with electron microscopy, in order

a V1, anterogradely labeled by an injection of fluoro-ruby in area V 4. Two

te divergent arbor, measuring over 0.8 mm. (C) An injection of biotinylated

2 and (acting as a bi-directional tracer for this pathway) retrogradely labeled

eurons, primarily in layers 2, upper 3, and 6. Modified from Ref. [24]. Scale

K.S. Rockland / Brain Research 1000 (2004) 60–6362

to map aspects of their synaptic connectivity. Populations of

efferent cells receive rather characteristic numbers of axoax-

onic or axo-somatic inhibitory inputs [8]. More complete

maps of excitatory as well as inhibitory inputs are still

needed.

2.1. Open questions

Despite years of work, the field of cortical connectivity is

at an early stage. There are open questions remaining at

almost every level. Why do many connections originate

from neurons in different layers? There have not even been

laminar-specific anterograde injections to clarify the likely

differences in terminal axons (Fig. 1).

What are the postsynaptic pools for different connection-

al systems in different areas? How do these compare in

different layers, especially when axons have collaterals in

multiple layers? Is the same neuron contacted twice on

different portions of its dendritic tree (routinely? some-

times? never?)?

Is the glutaminergic pyramidal cell population really

more homogeneous than GABAergic interneurons, or is

the relative homogeneity largely apparent, a consequence

of the small number of differentiating markers available for

pyramidal cells?

How do connections converge and interact at the level of

single neurons? This requires two or more connections to be

simultaneously labeled, in combination with reliable Golgi-

like filling of postsynaptic targets. The first requirement is,

in principle, now feasible, especially with dextran amines

tagged with different chromogens or fluorochromes. The

reliable simultaneous filling of postsynaptic neurons is more

difficult and is something still to be attained.

Another problem is more related to experimental design;

that is, which connections to emphasize among the large set

that converges on any area. The early concentration on a

feedback–feedforward dichotomy has arguably been too

restrictive (see also Refs. [2,23,24]). Coincidentally or not,

there have been years of neglect of the thalamocortical

connectivity [28] and some over-emphasis on a duality of

‘‘driving’’ vs. ‘‘modulatory’’ inputs [2,23].

3. Conclusions

A recent review states, with a large amount of plausi-

bility, ‘‘The field of interneuron research has come of

age’’ [21].

The same cannot in any way be said for cortical con-

nectivity; but after years of some decline, the field does

seem on the verge of an exciting renaissance. In part under

the impetus of functional imaging studies in humans, there

is great interest in cortical areas and regionalization. Impor-

tant results have already been contributed by neuroanatom-

ical studies of area characteristics and organization [35]. The

developments in DT MRI, one of the few noninvasive

techniques suitable for investigating connectivity in human

subjects, may give further impetus to a badly needed new

generation of studies in human neuroanatomy [3].

A distinct cause for optimism is the development of new

techniques and markers from molecular biology, and their

application to neuroanatomical questions. Electroporation,

one of several means for introducing exogenous genes in the

intact animal, permits imaging of detailed cell morphology

using GFP fluorescence [13]. In addition, patterns of gene

expression are being correlated with architectonic areas,

with some instances of area-specific expression [32]. In

the rodent, latexin, a carboxypeptidase A inhibitor, has been

shown to be expressed in the infragranular layers. More

particularly, expression is selective for corticocortical rather

than corticosubcortical neurons [1].

Additional advances are likely to be rapid with the

sponsorship of large-scale screens to create atlases of gene

expression at the cellular level. These have the self-pro-

claimed mission of providing vectors and transgenic mouse

lines to offer experimental access to CNS regions, cell

classes, and pathways [12].

One technique with immense impact would be an intra-

cellular transneuronal retrograde or anterograde marker.

This would dramatically demonstrate the actual composition

of neuronal groups, in terms of numbers, subtypes, and

laminar distribution of interconnected populations, and

directly advance information on neuronal networks. This

may still be in the future, but genetic approaches are already

available to visualize multisynaptic neural pathways in ro-

dents [34].

Finally, there have been important changes in the re-

search climate as regards connectional neuroanatomy. One

is a recent emphasis on the role of intrinsic factors in cortical

areas, layers, and modules [22,27]. The emerging evidence

for molecular gradients and signaling centers provides new

interpretations for some of the connectivity questions dis-

cussed here. A second change is increased attention to

species diversity and to diversity of cell types and areas.

This opens an immense challenge and opportunity for new

integration under the rubric of molecular neuroanatomy.

References

[1] Y. Arimatsu, M. Ishida, M. Sato, M. Kojima, Corticocortical associa-

tive neurons expressing latexin: Specific cortical connectivity formed

in vivo and in vitro, Cereb. Cortex 9 (1999) 569–576.

[2] J. Bullier, Hierarchies of cortical areas, in: J.H. Kaas, C.E. Collins

(Eds.), The Primate Visual System, CRC Press, Boca Raton, 2003,

pp. 181–204.

[3] F. Crick, E.G. Jones, Backwardness of human neuroanatomy, Nature

361 (1993) 109–110.

[4] J. DeFelipe, Preface to the special issue, J. Neurocytol. 31 (2002) 181.

[5] J. DeFelipe, L. Alonso-Nanclares, J.I. Arellano, Microstructure of the

neocortex: comparative aspects, J. Neurocytol. 31 (2002) 299–316.

[6] S.M. Dombrowski, C.C. Hilgetag, H. Barbas, Quantitative architec-

ture distinguishes prefrontal cortical systems in the rhesus monkey,

Cereb. Cortex 11 (2001) 975–988.

K.S. Rockland / Brain Research 1000 (2004) 60–63 63

[7] L.W. Enquist, J.P. Card, Recent advances in the use of neurotropic

viruses for circuit analysis, Curr. Opin. Neurobiol. 13 (2003) 1–4.

[8] I. Farinas, J. DeFelipe, Patterns of synaptic input on corticocortical

and corticothalamic cells in cat visual cortex: II. The axon initial

segment, J. Comp. Neurol. 304 (1991) 70–77.

[9] D.J. Felleman, D.C. Van Essen, Distributed hierarchical processing in

the primate cerebral cortex, Cereb. Cortex 1 (1991) 1–47.

[10] W.-B. Gan, J. Grutzendler, W.T. Wong, R.G.L. Wong, J.W. Lichtman,

Multicolor ‘‘DiOlistic’’ labeling of the nervous system using lipophil-

ic dye combinations, Neuron 27 (2000) 115–219.

[11] M. Giguere, P.S. Goldman-Rakic, Mediodorsal nucleus: areal, lami-

nar, and tangential distribution of afferents and efferents in the frontal

lobe of rhesus monkeys, J. Comp. Neurol. 277 (1988) 195–213.

[12] S. Gong, C. Zheng, M.L. Doughty, et al., A gene expression atlas of

the central nervous system based on bacterial artificial chromosomes,

Nature 425 (2003) 917–925.

[13] K. Haas, W.-C. Sin, A. Javaherian, Z. Li, H.T. Cline, Single-cell

electroporation for gene transfer in vivo, Neuron 29 (2001) 583–591.

[14] P.R. Hof, L.G. Ungerleider, M.J. Webster, R. Gattass, M.M. Adams,

C.A. Sailstad, J.H. Morrison, Neurofilament protein is differentially

distributed in subpopulations of corticocortical projection neurons in

the macaque monkey visual pathways, J. Comp. Neurol. 376 (1996)

112–127.

[15] D.M.G. Johnson, K.R. Illig, M. Behan, L.B. Haberly, New features of

connectivity in piriform cortex visualized by intracellular injection of

pyramidal cells suggest that ‘‘primary’’ olfactory cortex functions like

‘‘association’’ cortex in other sensory systems, J. Neurosci. 20 (2000)

6974–6982.

[16] L.C. Katz, Local circuitry of identified projection neurons in cat

visual cortex brain slices, J. Neurosci. 7 (1987) 1223–1249.

[17] J. Kozloski, F. Hamzei-Sichani, R. Yuste, Stereotyped position of

local synaptic targets in neocortex, Science 293 (2001) 868–872.

[18] S.S. Kumar, J.R. Huguenard, Pathway-specific differences in subunit

composition of synaptic NMDA receptors on pyramidal neurons in

neocortex, J. Neurosci. 23 (2003) 10074–10083.

[19] J.S. Lund, R.D. Lund, A.E. Hendrickson, A.H. Bunt, A.F. Fuchs, The

origin of efferent pathways from the primary visual cortex, area 17, of

the Macaque monkey as shown by retrograde transport of horseradish

peroxidase, J. Comp. Neurol. 164 (1975) 287–304.

[20] J.H.R. Maunsell, D.C. Van Essen, The connections of the middle

temporal visual area (MT) and their relationship to a cortical hierarchy

in the macaque monkey, J. Neurosci. 3 (1983) 2563–2586.

[21] D.D. Mott, R. Dingledine, Interneuron Diversity series: interneuron

research—challenges and strategies, TINS 26 (2003) 484–488.

[22] D.D.M. O’Leary, Y. Nakagawa, Patterning centers, regulatory genes

and extrinsic mechanisms controlling a realization of the neocortex,

Curr. Opin. Neurobiol. 12 (2002) 14–25.

[23] K.S. Rockland, Visual cortical organization at the single axon level: a

beginning, Neurosci. Res. 42 (2002) 155–166.

[24] K.S. Rockland, Feedback connections: splitting the arrow, in: J.H.

Kaas, C.E. Collins (Eds.), The Primate Visual System, CRC Press,

Boca Raton, 2003, pp. 387–405.

[25] K.S. Rockland, D.N. Pandya, Laminar origins and terminations of

cortical connections of the occipital lobe in the rhesus monkey, Brain

Res. 179 (1979) 3–20.

[26] L.D. Selemon, P.S. Goldman-Rakic, Common cortical and subcortical

targets of dorsolateral prefrontal and posterior parietal cortices in the

rhesus monkey: evidence for a distributed neural network subserving

spatially guided behavior, J. Neurosci. 8 (1988) 4049–4068.

[27] N. Sestan, P. Rakic, M.J. Donoghue, Independent parcellation of the

embryonic visual cortex and thalamus revealed by combinatorial Eph/

ephrin gene expression, Curr. Biol. 11 (2001) 39–43.

[28] S.M. Sherman, R.W. Guillery, The role of the thalamus in the flow of

information to the cortex, Philos. Trans. R. Soc. Lond., B 357 (2002)

1695–1708.

[29] L.C. Sincich, J.C. Horton, Independent projection streams from ma-

caque striate cortex to the second visual area and middle temporal

area, J. Neurosci. 23 (2003) 5684–5692.

[30] W. Suzuki, K.S. Saleem, K. Tanaka, Divergent backward projections

from area TE of the macaque inferotemporal cortex, J. Comp. Neurol.

222 (2000) 206–228.

[31] N. Tamamaki, K. Nakamura, T. Furuta, K. Asamoto, T. Kaneko,

Neurons in Golgi-stain-like images revealed by GFP-adenovirus in-

fection in vivo, Neurosci. Res. 38 (2000) 231–236.

[32] S. Tochitani, F. Liang, A. Watakabe, T. Hashikawa, T. Yamamori, The

occ1 gene is preferentially expressed in the primary visual cortex in

an activity-dependent manner: a pattern of gene expression related to

the cytoarchitecture area in adult macaque neocortex, Eur. J. Neuro-

sci. 13 (2001) 297–307.

[33] P. Veinante, M. Deschenes, Single-cell study of motor cortex projec-

tions to the barrel field in rats, J. Comp. Neurol. 464 (2003) 98–103.

[34] Y. Yoshihara, Visualizing selective neural pathways with WGA trans-

gene: combination of neuroanatomy with gene technology, Neurosci.

Res. 44 (2002) 133–140.

[35] K. Zilles, N. Palomero-Gallagher, C. Grefkes, F. Scheperjans, C. Boy,

K. Amunts, A. Schleicher, Architectonics of the human cortex and

transmitter receptor fingerprints: reconciling functional neuroanatomy

and neurochemistry, Eur. neuropsychopharmacol. 12 (2002) 587–599.