Download Intrinsic laminar lattice connections in primate visual cortex

THE JOURNAL OF COMPARATIVE NEUROLOGY 216303-318 (1983)
Intrinsic Laminar Lattice Connections
in Primate Visual Cortex
KATHLEEN S. ROCKLAND AND JENNIFER S. LUND
Department of Ophthalmology, Medical University of South Carolina,
Charleston. South Carolina 29425
ABSTRACT
Intracortical injections of horseradish peroxidase (HRP)reveal a system of periodically organized intrinsic connections in primate striate cortex.
In layers 2 and 3 these connections form a reticular or latticelike pattern, extending for about 1.5-2.0 mm around an injection. This connectional lattice
is composed of HRP-labeled walls (350-450 pm apart in Saimin' and about
500-600 pm in macaque) surrounding unlabeled central lacunae. Within the
lattice walls there are regularly arranged punctate loci of particularly dense
HRP label, appearing as isolated patches as the lattice wall labeling thins
further from the injection site. Aperiodic organization has also been demonstrated for the intrinsic connections in layer 4B, which are apparently in register with the supragranular periodicities, although separated from these by
a thin unlabeled region. The 4B lattice is particularly prominent in squirrel
monkey, extending for 2-3 mm from an injection. In both layers, these intrinsic connections are demonstrated by orthogradely and retrogradely
transported HRP and seem to reflect a system of neurons with long horizontal axon collaterals, presumably with arborizations at regularly spaced intervals. The intrinsic connectional lattice in layers 2 and 3 resembles the repetitive array of cytochrome oxidase activity in these layers; but despite
similarities of dimension and pattern, the two systems do not appear identical. In primate, as previously described in tree shrews (Rockland et al., '82),
the HRP-labeled anatomical connections resemble the pattern of 2-deoxyglucose accumulation resulting from stimulation with oriented lines, although the functional importance of these connections remains obscure.
Key words: intrinsic connections, visual cortex (area 171, neocortical lamination,
cytochrome oxidase, primate
Striate cortex, like other neocortical areas, contains an
intricate neuropil composed of extrinsic and intrinsic connections. Intrinsic connections, made by neurons within
striate cortex, establish a variety of linkages, both vertically in depth between laminae and horizontally within
laminae (Cajal, '11; Lorente de NO, '38; Lund, '73; Fisken
et al., '75; Lund and Boothe, '75; Creutzfeldt et al., '77;
Tigges et al., '77). Recently, we have described in striate
cortex of tree shrews aperiodically organized system of intrinsic connections apparently originating from pyramidal
neurons in layers 2 and 3A (Rockland and Lund, '81, '82c;
Rockland et al., '82). These regularly arranged anatomical
connections within laminae 2 and 3A constitute a previously unsuspected cortical substructure, distinct from the
well-organized cortical inhomogeneities imposed by extrinsic thalamic or corticocortical interareal connections.
Moreover, in the tree shrew this anatomically labeled pat-
tern resembles the physiologically mapped ordering of orientation specific units (Humphrey and Norton, '80). It also
resembles patterns of 2-deoxyglucose uptake in striate cortex following stimulation with oriented lines (Humphrey
et al., '80), and thus may be important for specific cortical
processes, related to orientation specificity or the elaboration of complex unit characteristics.
In the visual cortex of primates functionally induced
patterns of high metabolic activity result from a variety of
different stimulus parameters, including oriented lines
(Hubel et al., '78; Hendrickson and Wilson, '79) and
gratings of different spatial frequency (Tootellet al., '81).
In view of these reports, it seemed important to reexamine
the intrinsic connectivity of primate striate cortex, with
Accepted January 12,1983.
Address reprint requests to K.S. Rockland at her present
address: Southard Laboratory of Neuropathology, E.K. Shriver
Center, 200 Trapelo Rd., Waltham, MA 02154.
0 1983 ALAN R. LISS. INC.
Figure 1
INTRINSIC LATTICE CONNECTIONS IN STRIATE CORTEX
the particular goal of ascertaining whether, as in the tree
shrew, any periodic anatomical connections could be detected. As described here, intracortical injections of horseradish peroxidase in fact reveal a pattern of periodic intrinsic connections in both squirrel and macaque monkeys.
Some of these findings have been briefly reported elsewhere (Rockland and Lund, '82a,b).
METHODS
Seven squirrel monkeys and three macaques (Macaca
fasciculuris) received intracortical injections of horseradish peroxidase (HRP). Anesthesia was induced by a
mixture of ketamine and xylazine, supplemented for
macaques with N,O and halothane gas. Two or three craniotomies were made bilaterally over striate cortex, and one
to three small injections (0.01-0.03 pl of 20% HRP) were
delivered per opening into cortical gray matter. Survival
times (42-52 hours) and solvent mediums for the HRP (distilled H,O with polyornithine, lysolecithin, or proline) were
varied slightly in an attempt to optimize local HRP uptake
and transport. While some interanimal variability was evident in the quality of transport, we were not able to determine satisfactorily the critical variables, although adding
'H-proline to the HRP frequently improved results (as was
seen previously in the tree shrew-Rockland et al., '82).
Best results were also associated with fairly large injections where the uptake zone was judged to be 0.5-1.0 pm in
diameter (for further details, on defining the injection site
and other technical matters, cf. Rockland et al., '82). In one
animal (SM9),we placed unilateral injections in two prestriate regions-in area 18 of one hemisphere and in the
middle temporal area (MT)of the contralateral hemisphere.
Animals were allowed to survive 2-3 days postoperatively and then were anesthetized and perfused with a
warm saline rinse, followed by a solution of mixed aldehydes (1.5%paraformaldehyde and 1.75% gluteraldehyde
in phosphate buffer), and a final 20% sucrose buffer postwash (Rosene and Mesulam, '78). Tissue was subsequently
cut at 40 pn on a freezing microtome either in the horizontal plane or tangential to the pial surface. Generally, each
injection site was trimmed as a separate tissue block and
sections saved in series of three to be reacted for cytochrome oxidase or HRP (using the diaminobenzidine
(DAB),Adams, '81; or tetramethylbenzidine (TMB),Mesulam, '78, procedures). See Rockland et al. ('82)for further
details concerning methods.
Cytochrome oxidase reactions were carried out according to the method described in Wong-Riley ('79a). Sections
were collected in 10% sucrose buffer and subsequently incubated for 3-18 hours at 37°C.
305
RESULTS
Lamination
The lamination of squirrel monkey striate. cortex has
been previously described according to several schemes
(for example: Tigges et al.,'77; Colonnier and Sas, '78;
Hendrickson et al., '78),differing primarily in how layers 3
and 4 are subdivided. Our terminology, illustrated in F i g
ure 1A and B, was derived by comparing sections stained
for cell bodies, myelin, and cytochrome oxidase activity.
Cytochrome oxidase preparations of tissue.sectioned radially to the pia clearly reveal two densely staining bands,
apparently corresponding to the thalamic-recipient layers
4A ( - 50 p m wide) and 4C ( - 300 pm wide). These two
bands are separated by a lighter, athalmic zone, 4B ( - 100
pm wide).Myelin stains show up two fiber bands -a superficial plexus ( - 200 pn wide) situated within layer 3, and a
deeper plexus ( - 150 pm wide) lying within layer 4B and
the upper part of 4C. In Nissl stains, layer 4B corresponds
to a cell-sparse zone which blends inconspicuously with
layer 4C.
The general lamination scheme used for macaque striate
cortex has been described in previous Golgi studies (Lund,
'73; Lund and Boothe, '75) and is generally analogous to
the above.
Intrastriate injections in squirrel monkeys
In three squirrel monkeys (yielding five hemispheres)
HRP injections were made in the lateral convexity of striate cortex. Generally two injections (each estimated as
500-750 pn in diameter) were placed close together, sometimes merging into one large injection site (for instance,
SM4 in Figure 1C-G, where the injection is estimated as
1.2 pm across). Injections were confined to the gray matter
in these animals, and in one (SM6-Fig. 2) the injection remained confined to the supragranular layers above 4B.
In these animals, dense periodic accumulations of transported HRP, comprised of both retrogradely labeled cell
bodies and axon terminals, are visible surrounding the injection site in the supragranular layers (2 and 3). A second
set of periodicities occurs in layer 4B and the upper part of
4C. The patches in layers 2 and 3 measure about 200 pn
across, with a center-to-center spacing of 350-450 pn.
Those in 4B have similar center-to-center spacing and are
in register with the more superficial patches, although
there is an intervening gap in lower 3 and 4A (Figs. 1E-G,
2A). The band of label in 4B is 150-200 pm wide, with a
lower border actually extending down into the upper part
of 4C. At further distances from the injection site, the labeled band in 4B typically narrows and remains within the
athalamic zone of 4B proper (see Fig. IB). The wide HRP
band is prominent only when layer 4B itself is involved in
the injection. Otherwise, if the injection is restricted to layers 1-3 as in SM6 (Fig. 2B-D), there is only a narrow accuFig. 1. A,B. Pia to white matter sections of Sairniri striate cortex ( a d mulation of HRP at the upper border of layer 4B immedimal SM4) stained for cell bodies with cresyl violet (A)or for cytochrome oxidase activity (B)to demonstrate laminar boundaries described in text. The ately under the injection. The more superficial patches (in
darkly stained, cytochrome-rich zones in €3 (laminae 6,4C, 4A and patches 2 and 3) can be followed up to 1.5-2.0 mm from the edges of
in 2 and 3) are coincident with zones of thalamic inputs (Fitzpatricket al., the injection site. Those in 4B seem to continue further and
'83). C-G. Photomicrographs of five sequential sections (spacedabout 250 can be mapped for 2-3 mm from the injection. In addition
pm apart) depicting an HRP injection site (INJ)and resulting patches of
transported HRP in Suimiri SM4. Patches of HRP occur in layers 2 and 3 faint periodic accumulation of HRP can be seen in lamina
(examplesmarked with arrows in F) and in layer 4B (examplesdenoted hy 6, particularly close to the injection site (Fig ID-E).
underlying asterisks in F).There is some suggestion of periodicity in layer 6
As the HRP patches in these three cases were cut perclose to the injection. H. Higher magnificationphotomicrograph showing
pendicular
to the surface, it was not clear whether they
two patches of transported HRP in layers 2 and 3. Note the retrogradely laheled pyramidal neurons and orthogradely transported terminal label. were in fact isolated patches or whether they would line up
in a regular pattern -for example, in a stripelike fashion.
Labeled axon trunks pass between the two patches.
306
Fig. 2. A. A further example of HRP-labeled patches resulting from intracortical injection of HRP. Striate cortex of Saimiri (animal SM1).
Patches occur in layers 2 and 3 indicated by arrows and in lamina 4B (indicated by underlying asterisks). The injection site (not shown) extended to
lamina 6. B-D. In this animal (Saimiri SM6) the HRP injection (asterisk)
K.S. ROCKLAND AND J.S. LUND
remained superficial to layer 4B. Resultant HRP patches occur in layers 2
and 3 (solid arrows). There is also a subjacent band of HRP label immediately below the injection site in layer 4B (see C)but no widespread periodic
label in this layer. A band of label occurs in lamina 5 (open arrow)immediately below the injection site.
In one animal (SMl),where serial HRP sections had been more distant patches contain mostly terminals and not
saved, reconstructions indicated that the patches did labeled cells.
align, but formed a complex radiate, rather than stripelike
Intrastriate injections in macaque
geometry.
HRP
injections
were made in three macaque monkeys
In order to explore the conformation of the HRP patches
further over the cortical surface, tangentially cut sections (bilaterally in two animals), using the same procedures as
were prepared in three animals (SM2, SM3, SM8) for a to- described above. Three hemispheres were cut radially to
tal of four tissue blocks. Again, in this tissue plane, peri- the pia, while tangential sections were prepared of two
odic densities of transported HRP are evident. Tangential other hemispheres (each containing three injection sites).
sections, furthermore, clearly demonstrate a radiating or Periodic accumulations of HRP reaction product were
latticelike organization of HRP label, extending from the again evident (Fig. 4A-D) although compared to the squirinjection site as 12-14 heavily labeled spokelike walls rel monkey it proved more difficult to demonstrate the lataround unlabeled central spaces or lacunae (Figs. 3, 7). tice consistently with each injection. As in squirrel monEach heavily labeled wall measures about 200 p m across key, these connections consist of HRP-labeled cells and
and can be followed as a solid segment for about 0.5-0.7 terminals in the supragranular layers (2 and 3). Periodimm. Beyond this distance, HRP label can still be seen but cally organized connections can be found in layer 4B,but
it breaks up into regularly spaced discrete patches (200pn so far they have not been as convincingly demonstrated as
in diameter), placed 350-450 pn apart (Figs. 3,6C, 7). The in the squirrel monkey. Periodic label in layer 6 has not
INTRINSICLATTICE CONNECTIONS IN STRIATECORTEX
Fig. 3. A-E. Sequential 40-pm tangential sections illustrating an HRP
injection site and surrounding latticelike label in Saimin’ (SM2).
Asterisks
in B (and in higher-power micrograph F) denote HRP free lacunae sur-
307
rounded by walls of the HRP-labeled lattice. Section A is most superficial
and sections A-E are spaced about 120 pm apart.
Figure 4
INTRINSICLATTICE CONNECTIONS IN STRIATECORTEX
309
Fig. 5. Photomicrograph of HRP lattice in layer 4B of Suirniri in a section cut perpendicularto pia. The edges of two patches are indicated by 85terisks. Mainly spiny stellate neurons are labeled but also a few pyramidal
neurons (open arrows).Note the large stellate cell with long horizontal axon
collateral reaching back into layer 4B (solid arrow).
been clearly demonstrated in the macaque. In radial sections, the supragranular periodic HRP label appears as
patches, spaced 500-600 pm apart (slightly further apart
than in squirrel monkey), and extending about 1.5 mm
from the edge of an injection. Again, tangential sections
reveal that these connections are organized in an overall
latticelike conformation, which becomes more distinctly
punctate further from the injected area (Fig. 10A).
Tangential sections reveal axons crossing through the unlabeled central lacunae (betweenpatches), but the majority
seem to course within the HRP-labeled walls (Figs. 3A-F;
4B-C). The appearance of solid HRP-labeled walls surrounding unlabeled lacunae close to the injection site and
their reduction to patchlike HRP label further from the injection suggests the existence of two axon populations:
one locally projecting system that contributes terminals
to the walls and one long distance system that terminates
in patchlike regions within the walls.
The HRP label in 4B, unlike that in the upper layers, occurs in stellate neurons, which can be seen to bear HRPfilled spines on their dendrites, as well as in pyramidal cells
(Fig 5).Long axon segments of individual stellate cells can
frequently be followed for 0.5-mm lengths within 4B itself
and, occasionally, down into layer 6 before being lost. In
layer 4B, labeled axons are quite numerous between
patches and there is no consistent orientation of axons
traveling within or across HRP-labeled walls.
Neuronal composition of HRP-labeled connections
The neuronal components of the HRP-labeled connections were frequently well delineated in the DAB-reacted
sections. The layer 2 and 3 patches typically include a
small number of retrogradely labeled pyramidal cells embedded in a dense granular reaction product (Figs. l H , 4).
The dendritic spread of these neurons is visible over about
200 pn, a figure commensurate with the cross-sectional
width of the labeled patches. HRP-positive neurons occur
both in the center and near the edges of a labeled zone, with
some bias toward the center. In radially cut sections, axon
fibers are apparent crossing between patches (Fig. 1H).
Fig. 4. A,B. Tangential sections in theregion of an HRPinjection sitein
striate cortex of macaque monkey (animal H-16).The sections illustrate the
periodic nature of the surrounding HRP label in layers 2 and 3; section A is
most superficial. C,D. Higher-powerphotomicrographs showing the periodic accumulations of terminal label together with HRP-filled pyramidal
neurons. Note the abundance of HRP-filled axons traveling in all directions
between the patches.
Cytochrome oxidase and HRP patches
In both squirrel monkeys and macaques, cytochrome
oxidase staining reveals intense laminar labeling in layers
4A, 4C, and 6 (Fig lB),along with a distinct pattern of cytochrome-rich patches in layers 2 and 3 (Figs. lB, 6B,E).In
squirrel monkeys, the cytochrome patches in layers 2 and
3 typically measure 200 pm X 200 Fm and are arranged in a
regularly spaced array (Fig. 6B).The center-tocenter spacing between patches varies from 300 to 500 pn. In ma-
.
K.S. ROCKLAND AND J.S. LUND
310
Figure 6
INTRINSIC LATTICECONNECTIONS IN STRIATE CORTEX
caque, the organization of cytochrome patches in layer 3 is
somewhat more complex (Fig. 6E). As reported by others
(Horton and Hubel, '81; Humphrey and Hendrickson, '83),
these patches are characteristically oval (150 X 200 pm)
and aligned in rows, which are spaced 550 pm apart in one
direction (parallel to the long axis of the ovals) and 350 pm
in the other. In our material, however, there were frequent
irregularities in the shape of patches and their spacing, as
well as occasional suggestions of cross-links spanning the
rows in both directions. Irregularities were particularly
pronounced in the more superficial part of layer 3.
As shown in Figures 3, 4, 6, and 7, the pattern of HRP
connections assumes a distinctly punctuate appearance
with distance from the injection site. Typically, these
patches measure 100-200 pm across and are fairly regularly spaced with a center-to-center spacing of 350-450 pm
in squirrel monkeys and 500-600 pm in macaques. These
measurements are slightly smaller than the dimensions of
the cytochrome patches but are sufficiently similar to suggest that the two systems may be identical. Nevertheless,
as described below, this does not appear to be the case.
In order to investigate the relationship between these
two labels, alternate 40-pm sections were reacted for HRP
and cytochrome oxidase activity in six squirrel monkeys
and three macaques. Since the HRP injection site stains
intensely in the cytochrome material, this comparison is
based on matched comparisons outside the immediate region of the HRP injection. The tangential plane of section
was mainly used in these comparisons as the relatively
constant course of the radial cerebral blood vessels facilitates the alignment of sequential sections. The results of
such comparisons, illustrated in Figures 6 and 8-10, indicate that the HRP and cytochrome patches are not identical in anatomical locus. Frequently, the two labels appear
to interdigitate, but instances of overlap are not uncommon (Figs. 9, IOB). It may be significant that both
the HRP and cytochrome patches appear to be arranged
around common central lacunae or cores, as if these regions served as common reference points for both systems.
This relationship is particularly clear where the HRP reaction forms solid, latticelike walls around central openings.
In this case, the cytochrome patches fall within the HRP
walls but avoid the central lacunae (Fig. 9).
Within layer 4B there are no regions of heightened cytochrome activity, despite the presence of HRP periodicities
in this layer. I t should be noted, however, that near the injection site (within 1 mm), the HRP-labeled connections
are quite dense, even extending into 4C. Thus, they must
partially overlap with the cytochrome-rich band in 4C.
Striate efferent connections
to areas V2 and MT in Saimiri
In one squirrel monkey (SM9), a unilateral injection of
HRP was made in area V2 of one hemisphere and into area
MT of the contralateral hemisphere in order to investigate
the possible relationship between intrinsic periodicities in
area 17 and the populations of striate neurons projecting
Fig. 6. Photomicrographs of tangential sections comparing the HRP intrinsic lattice ( ASuimiri; D: macaque)with the pattern of cytochromeoxidase activity (B:Suimiri; E: macaque) in laminae 2 and 3 of striate cortex.
Arrows indicate common blood vessels in D and E of macaque material.
C. Higher-magnification photomicrograph of HRP lattice in tangential section of Suimiri (laminae 2 and 3).
311
500pm
Fig. 7. Map of pattern of HRP label (black)from tangential sections of
Saimin' (SM2) striate cortex in layers 2 and 3 (see Fig. 3A-E). The injection
site (solid black area) is surrounded by a region containing unlabeled
lacunae, which thins to isolated labeled punctuate loci further from the injection site.
to prestriate areas. After the injection in MT, filled neurons occurred mainly in layer 4B of area 17, with a few neurons in layer 6 (Fig. 11B). In agreement with other studies
(Tigges et al., %1), we find these neurons to be both pyramidal and stellate in shape and to be concentrated in the upper part of 4B. No periodicity of these efferent neurons
could be detected, but, owing to their position in lamina
4B, they may overlap with at least some of the intrinsically
projecting neurons in this layer.
Injections in area 18 demonstrate striate efferent neurons in layer 3, at approximately the same depth as intrinsically projecting neurons (Fig. 11A).Efferent neurons are
pyramidal in shape and after small injections in area V2
tend to be grouped in clusters about 200 p m across (as also
described by Tigges et al., '81).These dimensions are similar to the cytochrome-rich loci described above, but comparison of several pairs of adjacent 40-pm sections reacted
for HRP or cytochrome oxidase indicate that the two patterns are not identical. These efferent clusters also differ in
several points from the intrinsic periodicities: There are
more labeled neurons per cluster, there are more labeled
neurons in the spaces between clusters, and there are numerically more and thicker descending axons which can be
traced at least into layer 5. Moreover, with large injections
the periodicity is lost and a continuous band of efferent
neurons is seen (Tigges et al., '81). Further experiments,
using double retrograde labels, are, however, needed to establish whether there is in fact any common identity between intrinsically projecting neurons and those projecting to area V2. Similar findings of patchy distribution of
area 17 neurons efferent to V2 have been reported in macaque (Rockland and Pandya, '79; Maunsell et al., 'SO;
Tigges et al., %l),
so that the same observations probably
apply to macaques as well.
Both these prestriate injections in Saimiri also reveal periodically organized connections within areas V2 and MT.
Unlike the bilaminar arrangement of intrastriate connections, the HRP patches surrounding injection sites in
areas 18 and MT consist of accumulations of HRP in
layers 1-5 which appear slightly larger than the intrastri-
K.S. ROCKLANDAND J.S. LUND
3 12
A
B
Fig. 8. Diagrammatic reconstructions of the pattern of HRP (solid
black)-and cytochromeoxidase (dashedcircled-labeledloci in laminae 2 and
3 of Suimin striate cortex. Each was derived by comparing 12 consecutive
40-p.m tangential sections alternately stained for HRP or cytochrome oxi-
dase and matched by blood vessel patterns. In these sections there appears
to be no consistent relationship between the two sets of loci (however, see
Fig. 9).
INTRINSIC LATTICE CONNECTIONS IN STRIATE CORTEX
313
0
0
A
Fig. 10. A. Reconstruction of HRP-labeled loci (black)around an injection site in macaquemonkey striate cortex, layers 2 and 3. in tangential section. B.The same reconstruction of HRP-labeled loci as shown in A but
with the position of cytochrome-rich loci indicated as hatched profiles, (The
alternately stained section series were matched by blood vessel patterns.)
As in Figure 8, the two labels clearly do not coincide but both may surround
common lacunae as in Figure 9.
K.S. ROCKLAND AND J.S. LUND
314
Fig. 11. A. Two patches (arrows)of striate efferent pyramidal neurons
in layers 2 and 3 of Suimiricortex labeled retrogradely by an HRP injection
in area V2. More retrogradely labeled cells occur in these efferent patches
than within intrinsically labeled loci. Note the numerous vertically de-
scending axons and conspicuous band of label in layer 5. B. HRP-labeled
neurons in the upper part of layer 4B of Suiniri, retrogradely labeled after
an HRP injection in area MT. No periodicity was found in this efferent
population.
ate patches. As previously observed, intrinsic periodic connections also occur in macaque V2, where they measure
about 350 pm across (Fig. 3C in Rockland and Pandya, '79)
and in more anterior prestriate areas such as V4, where
they are typically larger (about 600 pm across). The general tangential conformation of these prestriate periodicities has not yet been examined.
laminae 2 and 3 form stripelike arrays, the periodic connections in the primate appear to have a more complex configuration. Within about 1mm of the injection site, these connections form a reticular or latticelike pattern, made up of
HRP-labeled walls surrounding central unlabeled lacunar
regions. The lacunae do not contain labeled cells or terminals (as detectable at the light microscopic level), but are
frequently traversed by HRP-labeled fibers crisscrossing
in all directions. Further from the injection site, there is a
general decrease in the density of label, and the HRP walls
typically break up into a more punctate pattern, spaced
about 350 pn across (Fig. 3C in Rockland and Pandya, '79)
wider (500-600 pn) in macaque.
Several laboratories have recently reported a set of regularly arranged extrinsic thalamic axon terminations in the
supragranular layers of primate visual cortex (Hubel and
Livingstone, Bl), arising from the intercalated laminae of
the lateral geniculate nucleus (Fitzpatrick et al., '83). These
thalamic termination zones, which can also be detected
histochemically by their high cytochrome oxidase (Horton
and Hubel, '81; Humphrey and Hendrickson, '83) or glutamate decarboxylase (GAD) activity (Hendrickson et al.,
'81), show a remarkable resemblance to the intrinsic ana-
DISCUSSION
This study describes a periodically organized system of
intrinsic connections in layers 2 and 3 of the primate visual
cortex, which extends at least 2 mm from the edges of an
HRP injection. These connections, like those recently reported in the tree shrew (Rockland et al., '82),are composed
of a small number of retrogradely filled pyramidal neurons
embedded in a dense matrix of orthogradely and perhaps
collaterally transported HRP presumed to be in axon terminals. Judging from the density of reaction product, the
HRP is probably transported throughout all axon collaterals of labeled neurons, especially closer t o the injection
site.
Unlike the tree shrew, where the intrinsic connections in
INTRINSIC!LATTICE CONNECTIONS IN STRIATE CORTEX
tomical periodicities described in this study in terms of
laminar disposition, frequency of occurrence, dimensions,
and punctate pattern. Despite these apparent similarities,
however, as we ascertained by comparing cytochrome-rich
and HRP-labeled loci in alternate sections (Figs. 6, 8, and
lo), the zones of thalamic fiber terminations are not identical in locus with the intrinsic connections, nor do the two
systems strictly interdigitate. Both systems do, however,
appear to surround common lacunar regions. Another histochemical marker, butyrylcholinesterase (BuChE), has
been shown to have a periodic distribution within the supragranular layers (Graybiel and Ragsdale, '82). It is presently uncertain whether or not these several periodically
organized systems are interrelated in an orderly fashion.
The periodic organization of the intrinsic anatomical
connections shown in the present study is somewhat puzzling, especially in view of the apparently large size of the
HRP uptake zone. One explanation, as proposed earlier
(Rockland et al., '82), is that there is a regularly arranged
but discontinuous subpopulation of neurons within the
cortical neuropil with long horizontal axon collaterals.
These axon collaterals appear to have periodically arranged terminal arbors that are in register with the cell
bodies and dendrites of the same widely connected neurons (but not necessarily directly synapsing with them).
Pyramidal neurons with long horizontal axon collaterals
have been described in various Golgi studies (Lorente de
NO, '38; Lund, '73; Lund and Boothe, '75) and are seen
following intracellular injections of HRP (Gilbert and
Wiesel, '79). However, the high degree of connectional
order implied by our results, although hypothesized (Gilbert and Wiesel, '79; Lund, '81), has not previously been
demonstrated .
An equally orderly distribution of neurons with very local axon collaterals can also be proposed to explain another
aspect of the intrinsic lattice; namely, the presence of unlabeled lacunae even close to the HRP injection site. Presumably, such unlabeled loci contain neurons with very
local collaterals that do not extend within the zone of HRP
uptake. In addition, the characteristic walls of the HRP
lattice near an injection (as opposed to the more punctate
pattern further away) may result from a combination of local axon collaterals, arborizing within the lattice walls, and
the longdistance punctate connectional system.
The patterned distribution of HRP label around the injection site can, however, be interpreted in other ways than
on the basis of subsets of uniquely connected neurons. For
example, HRP uptake may vary significantly according to
differential metabolic rates of individual neurons. Thus,
patterned intrinsic connections may reflect a substrate of
particularly active neurons embedded in a continuum of
similarly organized but less active neurons (or vice versa).
Some evidence against this possibility is provided by the
apparently random correlation between HRP- and cytochrome oxidase-labeled loci. That is, combined experiments with both 2-deoxyglucose (2-DG)and cytochrome
oxidase suggest that the cytochrome-rich regions have
high metabolic rates, and yet these regions do not appear
to superimpose regularly with or exclude the punctate
HRP uptake zones (Horton and Hubel, '81; Humphrey and
Hendrickson, '83). Another possibility is that HRP is
taken up and transported only by neurons with a particular biochemical behavior, perhaps associated with transmitter specificity (LeVay and Sherk, '81). In this regard,
the recently described (Graybiel and Ragsdale, '82) latticelike compartmentalization of the enzyme BuChE bears an
315
interesting and perhaps significant resemblance both to
the HRP intrinsic lattice in laminae 2 and 3 and to the lattice pattern often shown by 2-DG uptake experiments using specific line orientations.
The relationship between the HRP-labeled intrinsic connections and the cytochrome-marked thalamic terminations around common lacunae provides a possible answer
to the question of whether the HRP-labeled system represents a fixed position lattice or a continuum of shifting anatomical connections. As discussed above (and also previously, Rockland et al., '82), we have suggested that such
connections in the tree shrew could reflect a fixed-position
subpopulation of neurons with long horizontal collaterals
which are periodically organized. These neurons may interleave regularly with neurons with very local intrinsic connections or, alternately, these neurons may be superimposed against a continuum of such locally projecting
neurons (Fig. 11in Rockland et al., '82). I t is possible, however, as recently hypothesized by Mitchison and Crick
('82), that the long-collateral neurons also constitute a continuum. In this case, the apparent periodicity may be derived from specific connectional rules. According to Mitchison and Crick, for instance, stripelike connections could
result from specifically oriented axons in a continuum all
across the cortex. The orientation of these axons may be
correlated with the functional orientation preference of the
parent neurons to visual stimuli. As these authors point
out, two strategically placed injections, or experiments
with double labels, could resolve this issue: The "continuum" interpretation predicts at least a partial filling of the
periodic pattern between two appropriately spaced injections. While this issue clearly requires further experimental work, data available in the monkey for the cytochrome
oxidase system (Horton and Hubel, '81; Humphrey and
Hendrickson. '82; Fitzpatrick et al., '82) show that at least
the cytochrome oxidase pattern occupies a fixed anatomical position, coincident with patches of geniculate axon
terminations in layer 3. These are centered over the ocular
dominance columns of layer 4 in macaque. Given that both
the HRP-labeled loci and cytochrome-rich regions of thalamic axon terminations surround common lacunae, it is
difficult to see how the HRP-labeled system can be a continuum in the primate. I t seems likely that other periodicities revealed in laminae 2 and 3 -by HRP (present report),
by BuChE (Graybiel and Ragsdale, '82), and by 2-DG
(Hubel et al., '78; Hendrickson and Wilson, '79; Horton and
Hubel, '81; Humphrey and Hendrickson, '83)-will also be
found to demonstrate different aspects of a single fixedposition connectional lattice. Perhaps, as suggested here,
the unlabeled regions shown in sections processed by each
technique may prove to be common lacunae in a general
lattice structure. Additional support for a fixed-position
intrinsic anatomical network is suggested by the existence
of other periodically organized connections within striate
cortex which are known to be underlain by fixed-position
anatomical substrates. These include thalamic terminations in layer 4C (Hubel and Wiesel, '72), pulvinar terminations in layers l and 2 (Ogren and Hendrickson, '77), or
clusters of striate neurons projecting to areaV2 (Rockland
and Pandya, '79; Wong-Riley, '79b; Gilbert and Wiesel, '80;
Maunsell et al., '80; Tigges et al., '81). This interesting
question, however, is by no means solved and needs further investigation.
In addition to the regularly arranged anatomical connections in layers 2 and 3, HRP injections have also revealed a
periodic organization of intrinsic connections in layer 4B,
316
K.S. ROCKLAND AND J.S. LUND
most clearly demonstrated in the squirrel monkey. These
connections originate from both stellate and pyramidal
neurons, unlike those in the supragranular layers, which
they otherwise resemble in terms of spacing and overall
configuration. In squirrel monkey these 4B connections
are frequently denser than those in layers 2 and 3, extend
further (2-3 mm) from the injection area, and appear to be
mainly intrinsic to lamina 4B.I t is interesting that enzymatic markers have not as yet revealed any compartmentalization in layer 4B,although there is some evidence of a
periodic accumulation of 2-DG in this layer after stimulation with uniformly oriented lines (Hubel et al., '78;
Hendrickson and Wilson, '79; Horton and Hubel, '81;Humphrey and Hendrickson, '83).
The functional significance of the anatomical connectional lattice described here remains to be ascertained. It is
possible that the axon terminals of the labeled neurons
synapse with inhibitory neurons or some combination of
inhibitory and excitatory postsynaptic targets. One functional possibility, discussed in the context of our previous
tree shrew work, is that periodic intrinsic connections
might be related to orientation selectivity, either directly
or indirectly. In primate, as in the tree shrew, there is an intriguing resemblance of dimension, laminar emphasis, and
overall configuration between the HRP-labeled anatomical
pattern and the patterns of 2-DG accumulation resulting
from stimulation with oriented lines (Humphrey and Nor-
ton, '80; Humphrey et al., '80; Rockland et al., '82). It is particularly interesting that the intrinsic connectivity pattern and 2-DG pattern are similar within each animal but
the overall patterning between tree shrew and macaque is
different. The geometry of iso-orientation bands over the
cortical surface is more complex in primate than in tree
shrew. Physiological results reveal frequent reversals and
breaks in the orderly sequence of orientation preference
(Hubel and Wiese1,'74);and 2-DG experiments (with single
orientation line stimuli) produce a complicated picture of
both latticelike and linear arrays (Fig. 12). The linear arrays, in accord with physiological results, often form
whorls and blind ends, rather than following a consistent
direction across the cortex, as in tree shrew (Hubel et al.,
'78; Humphrey et al., '80). Moreover, recent work from several laboratories (Horton and Hubel, '81; Humphrey and
Hendrickson, '83) suggests that the 2-DG "orientation"
pattern in primates actually includes at least two dissociable systems. One, coincident with the cytochrome-rich
patches in the supragranular layers (corresponding also to
thalamic terminations, Fitzpatrick et al., '83),contains
neurons with little orientation selectivity (Hubel and Livingstone, '81) and shows up as a punctate 2-DG pattern
after stimulation with a ganzfeldt or with lines of all orientations. The other system (appearing as a herringbone z i g
zag array in squirrel monkey (Fig. 12; Humphrey and
Hendrickson, '83) and a swirling stripelike array in the
Fig. 12. Darkfield photomicrograph of Bdeoxyglucose label seen in a
tangential section of Sairniri striate cortex, lamina 2-3. The animal was
stimulated binocularly with vertical stripes. Material courtesy of Drs.
Allen Humphrey and Anita E. Hendrickson. Note that both linear stripe
patterns and reticular patterning can be discerned. See Humphrey and
Hendrickson ('83)for further discussion of 2deoxyglucose labeling in primate visual cortex.
INTRINSICLATTICE CONNECTIONS IN STRIATE CORTEX
macaque (Hubel et al., '78)) presumably corresponds to
physiologically described loci of high specific orientation
selectivity (Hubel and Wiesel, '74) and probably includes
labeling of the nonorientation selective system as well.
Thus, combined labeling of oriented and nonoriented systems probably underlies the 2-DG patterns resulting from
stimulation with single orientations. In view of these findings, one might hypothesize that the HRP- and cytochrome-labeled loci mark the position of anatomical substrates for different components of these 2-DG patterns.
For example the HRP-labeled system may create a crosslinked pattern which includes the cytochrome-rich regions
in its walls. The potential association of intrinsic connectional lattices and orientation selectivity, however,
requires further experimental evidence; the possibility
remains that these anatomical periodicities may be important for other aspects of visual processing or may be involved in other components of unit characteristics besides
orientation selectivity (for example, the elaboration of
complex or hypercomplex properties; Hubel and Wiesel,
'77). Moreover, it should be pointed out that intrinsic patterned connections occur in nonprimary visual cortical
areas and in cortex unrelated to vision, which again raises
questions as to their functional significance (Kunzle, '76;
Jones et al., '78).
In terms of the morphological organization of striate cortex, it is interesting that several characteristics of the periodic intrinsic connections do not immediately conform to
the idea of the cortical column or slab as narrowly defined
(e.g., a 250-500-pm-wide cylinder or slab, extending from
pia to white matter and sharing common functional properties). The intrinsic lattices in layers 2 and 3, and 4B, for
example, although they are in register, do not form a solid,
interconnected unit, but rather seem to comprise two
largely separate, widespread laminar systems. There is a
slight gap between them, and the 4B lattice exhibits preferentially horizontal rather than vertical fiber connections,
failing to label when injections do not extend into layer
IVB itself. These studies thus imply an important role for
horizontal laminar-specific processing in cortical organization, in addition to the vertical linkages revealed by earlier
work (e.g., Lorente de NO, '38; Hubel and Wiesel, '72, '77;
Lund and Boothe, '75). Further experiments will be necessary to ascertain precisely how these various periodic connections may be interrelated and the nature of their functional correlates.
ACKNOWLEDGMENTS
This work was supported by NIH grants EY-03321,
EY04064, and EY04946 and a fellowship to Kathleen S.
Rockland awarded from S.C. State Appropriations. We
would like to thank Kathy Cowart, Kathy Ludden, and
David Whittaker for their technical assistance and Diane
Ashworth for typing. The helpful discussions and critical
comments of Drs. Gary Blasdel, David Fitzpatrick, and
Raymond Lund are gratefully acknowledged.
LITERATURE CITED
Adams, J.C. (1981) Heavy metal intensification of DAB-based HRP reaction product. 3. Histochem. Cytochem. 29:775.
Cajal, S.R. (1911)Histologie de Systeme Nerveux de 1'Homme et des Vertebrbs, Vol. 11. Paris: Maloine. Reimpress., Madrid: Instito Cajal, 1955.
Colonnier. M., and E. Sas (1978)An anterograde degeneration study of the
tangential spread of axons in cortical areas 17 and 18 of the squirrel
monkey (Sairniri sciureus). J. Comp. Neurol. 179:245-262.
317
Creutzfeldt, O.D., L.J. Garey, R. Kuroda, and J.-R. Wolff (1977)The distribution of degenerating axons after small lesions in the intact and isolated visual cortex of the cat. Exp. Brain Res. 27:419-440.
Fisken, R.A., L.J. Garey, andT.P.S. Powell (1975)Theintrinsic, association
and commissural connections of area 17 of the visual cortex. Philos.
Trans. R. SOC.miol.] 272487-536.
Fitzpatrick, D., K. Itoh, and I.T. Diamond (1983)The laminar organization
of the lateral geniculate body and the striate cortex in the squirrel monkey (Sairniri sciureus). J. Neurosci. 3:673-702.
Gilbert, C.D., and T.N. Wiesel(1979)Morphology and intracortical projections of functionally characterized neurons in cat visual cortex. Nature
280:lZO- 125.
Gilbert, C.D., and T.N. Wiesel(1980)Interleaving projection bands in corticocortical connections. Proc. SOC.
Neurosci. (Abstr.)6;315.
Graybiel, A.M., and C.W. Ragsdale (1982)Pseudocholinesterase stainingin
the primary visual pathway of the macaque monkey. Nature 299;439442.
Hendrickson, A.E., S.P. Hunt, and J:Y. Wu (1981) Immunocytochemical
localization of glutamic acid decarboxylase in monkey striate cortex.
Nature 292605-607.
Hendrickson. A.E., and J.R. Wilson (1979) A difference in 14C deoxyglucose autoradiographic patterns in striate cortex between Macaca and
Sairniri monkeys following monocular stimulation. Brain Res. 170:
353-358.
Hendrickson. A.E., J.R. Wilson, and M.P. Ogren (1978)The neuroanatomical organization of pathways between the dorsal lateral geniculate nucleus and visual cortex in Old World and New World primates. J. Comp.
Neurol. 182:123-136.
Horton, J.C., and D.H. Hubel (1981) Regular patchy distribution of cytochrome-oxidase staining in primary visual cortex of macaque monkey.
Nature 292762-764,
Hubel. D.H., and M.S. Livingstone (1981)Regions of poor orientation tuning coincide with patches of cytochrome oxidase staining in monkey
Neurosci. (Abstr.) 7;357.
striate cortex. Proc. SOC.
Hubel, D.H., and T.N. Wiesel(1972)Laminar and columnar distribution of
geniculocortical fibers in the macaque monkey. J. Comp. Neurol. 146:
421-450.
Hubel, D.H., and T.N. Wiesel(1974)Sequence regularity and geometry of
orientation columns in the monkey striate cortex. J. Comp. Neurol. 158:
267-294.
Hubel, D.H., and T.N. Wiesel (1977) Functional architecture of macaque
monkey visual cortex. Proc. R. SOC.Lond [Biol.] 198:1-59.
Hubel, D.H., T.N. Wiesel, and M.P. Stryker (1978)Anatomical demonstration of orientation columns in macaque monkey. J. Comp. Neurol. 177:
361-379.
Humphrey, A.L., and A.E. Hendrickson (1983)Background and stimulusinduced patterns of high metabolic activity in the visual cortex (area 17)
of the squirrel and macaque monkey. J. Neurosci. 3;345-358.
Humphrey, A.L., and T.T. Norton (1980)Topographic organization of the
orientation column system in the striate cortex of the tree shrew
(Tupaia glis). I. Microelectrode recording. J. Comp. Neurol. 192531548.
Humphrey, A.L., L.C. Skeen, and T.T. Norton (1980)Topographic organization of the orientation column system in the striate cortex of the tree
shrew (Tupaiaglis). 11. Deoxyglucose mapping. J . Comp. Neurol. 192:
549-566.
Jones, E.G., J.D. Coulter, and S.H.C. Hendry (1978) Intracortical connectivity of architectonic fields in the somatic sensory, motor and parietal
cortex of monkeys. J. Comp. Neurol. 181:291-348.
Kunzle, H. (1976)Alternating afferent zones of high and low axon terminal
density within the macaque motor cortex. Brain Res. 106:365-370.
LeVay, S., and H. Sherk (1981)Retrograde transport of L-3H-prolineby cortical layer V pyramidal cells. Proc. Soc. Neurosci. (Abstr.) 7;112.
Lorente de N6, R. (1938) The cerebral cortex: Architecture, intracortical
connections and motor projections. In J.F. Fulton (ed):Physiology of
the Nervous System, 2nd Ed. New York Oxford University Press,pp.
29 1-32],
Lund, J.S. (1973)Organization of neurons in the visual cortex, area 17. of
the monkey (Macaca rnulatta).J. Comp. Neurol. 147;455-496.
Lund, J.S. (1981) Intrinsic organization of the primate visual cortex, area
17, as seen in Golgi preparations. In F.O. Schmitt, F.G. Worden, G.
Adelman, and S.G. Dennis (eds):The Organization of the Cerebral Cortex. Proceedings of a Neuroscience Research Program Colloquium.
Cambridge, Massachusetts: MIT Press, pp. 105-124.
K.S. ROCKLAND AND J.S. LUND
318
Lund, J.S., and R.G. Boothe (1975) Interlaminar connections and pyramidal neuron organization in the visual cortex, area 17, of the macaque
monkey. J. Comp. Neurol. 153:305-334.
Maunsell, J.H.R., W.T. Newsome, and D.C. Van Essen (1980)The spatial
organization of connections between V1 and V2 in the macaque: Patchy
Neurosci. (Abstr.) 6580.
and non-patchy projections. Roc. SOC.
Mesulam, M.-M. (1978) Tetramethylbenzidine for horseradish peroxidase
neurochemistry: A noncarcinogenic blue reaction-product with superior sensitivity for visualizing neural afferents and efferents. J. Histochem. Cytochem. 26:106-117.
Mitchison, G., and F. Crick (1982) Long axons within the striate cortex:
Their distribution, orientation, and patterns of connection. Roc. Natl.
Acad. Sci. USA 79:3661-3665.
Ogren, M.P., and A.E. Hendrickson (1977)The distribution of pulvinar terminals in visual areas 17 and 18of the monkey. Brain Res. 137:343-350.
Rockland, K.S., and J.S. Lund (1981)Anatomical banding of intrinsic connections in tree shrew striate cortex. ARVO Abstract Suppl. Invest.
Ophthalmol. 20:176.
Rockland, K.S., and J.S. Lund (1982a)Lattice-like intrinsic neural connections in primate striate visual cortex. Roc. Soc. Neurosci. (Abstr.)8:
706.
Rockland, K.S., and J.S. Lund (1982b)Periodically organized intrinsic connections in squirrel monkey striate cortex. ARVO Abstract Suppl. Invest. Ophthalmol. 2211.
Rockland. K.S., and J.S. Lund (1982~)
Widespread periodic intrinsic connections in tree shrew visual cortex (area 17).Science 215:1532-1534.
Rockland, K.S., J.S. Lund, and A.L. Humphrey (1982)Anatomical banding
of intrinsic connections in striate cortex of tree shrews (Tupaia glis).J.
Comp. Neurol. 209:41-58.
Rockland, K.S., and D.N. Pandya (1979)Laminar origins and terminations
of cortical connections of the occipital lobe in the rhesus monkey. Brain
Res. 179:3-20.
Rosene, D.L., and M.M. Mesulam (1978)Fixation variables in horseradish
peroxidase neurohistochemistry: Effects of perfusion and postfixation
on sensitivity. J. Histochem. Cytochem. 26%-39.
Tigges, J., M. Tigges, and A.A. Perachio (1977) Complementary laminar
terminations of afferents to area 17 originatingin area 18 andin the lateral geniculate nucleus in squirrel monkey. J. Comp. Neurol. 176:
87-100.
Tigges, J., M. Tigges, S. Anschel, N.A. Cross, W.D. Letbetter, andR.L. McBride (1981)Areal and laminar distribution of neurons interconnecting
the central visual cortical areas 17, 18. 19, and MT in squirrel monkey
(Sairniri).J. Comp. Neurol. 202:539-560.
Tootell, R.B.. M.S. Silverman, and R.L. DeValois (1981)Spatial frequency
columns in primary visual cortex. Science 214:813-815.
Wong-Riley. M. (1979a) Changes in the visual system of monocularly sutured or enucleated cats demonstrable with cytochrome oxidase histochemistry. Brain Res. 171:ll-28.
Wong-Riley, M. (1979b)Columnar corticocortical interconnections within
the visual system of the squirrel and macaque monkey. Brain Res. 162:
201-217.