THE JOURNAL OF COMPARATIVE NEUROLOGY 324~37-66 (1992)
Cortical Connections of Subdivisions of Inferior Temporal Cortex in Squirrel Monkeys R.E. WELLER
AND
G.E. STEELE
Department of Psychology, University of Alabama at Birmingham, Birmingham, Alabama 35294
ABSTRACT Patterns of cortical connections and architectonics were used to determine subdivisions of inferior temporal (IT) cortex of squirrel monkeys. Single or multiple injections of the tracers wheat germ agglutinin-horseradishperoxidase, Fast Blue, Diamidino Yellow, Fluoro-Gold, and 3H-aminoacids were placed into IT cortex. Most injections were placed in caudal IT cortex in the region previously shown to receive input from the caudal subdivision of the Dorsolateral Area, DLc; additional injections were placed in more rostral IT cortex. The results indicate the presence of two major regions: a caudal region, ITc, and a rostral region, ITR.An intermediate region of cortex along the ITC-ITRborder that displays some connections of ITc and some connections of ITRmay be another area. ITc contains a more myelinated dorsal area, ITCd, and a larger ventral area, ITCv. Both ITCd and ITCv receive a major projection from DLC; additional input from DLR,MT, and V 11; and send strong projections to ITR,the lateral bank of the superior temporal sulcus, and dorsolateral prefrontal cortex. Only ITCd has strong connections with DLR and cortex in the depths of the superior temporal sulcus, and only ITCv has connections with lateral orbital cortex. The overall pattern of connections between ITc and DLc suggests that ITc has a crude topographic organization, with dorsal cortex representing the lower field and ventral cortex representing the upper field. ITRdiffers from ITc by receiving little if any input from DLc; projecting to inferior temporal polar cortex, the rostral Sylvian fissure, and medial orbital cortex; and having a less distinct layer IV. Comparison of subdivisions of inferior temporal cortex defined in the present study in squirrel monkeys and those reported in other primates suggests that ITc of squirrel monkeys may correspond to area TEO of macaque monkeys. o 1992 Wiley-Liss, Inc. Key words: dorsolateral area, V4, TE, TEO, primates
Humans with damage to inferior occipito-temporal cortex exhibit a deficit in the visual identification and recognition of objects called visual agnosia (for review, see Damasio et al., '82; Levine, '82; Farah, '90). A comparable visual memory deficit is found in macaque monkeys with damage to inferior temporal (IT) cortex co-extensive with architectonic area TE of von Bonin and Bailey ('47; Mishkin, '54; Mishkin and Pribram, '54;Iwai and Mishkin, '68, '69;Cowey and Gross, '70; for review, see Dean, '82; Mishkin, '82). An increasing amount of information gained from a number of species of primates suggests that IT contains more than one visual area (e.g., Seltzer and Pandya, '78; Desimone and Gross, '79;Weller and Kaas, '87; Yaginuma, '90; Van Essen et al., '90). However, the total number and borders of such subdivisions are not well established in any primate. Most previous experiments on IT cortex have used macaque monkeys. Such experiments have been compli-
o 1992 WILEY-LISS, INC.
cated by the many cortical sulci and ventrolateral location of IT cortex in macaques. To minimize these problems, Weller and Kaas ('85, '87) investigated IT cortex in the relatively lissencephalic brain of New World owl monkeys. However, because owl monkeys are not diurnal like the majority of primates, subdivisions of IT cortex in owl monkeys may be less differentiated or otherwise different than in other primates. Thus, to further investigate the organization of IT cortex in primates, we have turned to another New World primate, the squirrel monkey, as a model. Squirrel monkeys, like owl monkeys, have a relatively unfissured cortex, but are diurnal and thus more representative of primates. Indeed, previous studies of visual cortex in squirrel monkeys have revealed an emphasis on central vision more comparable to that in macaque Accepted June 3,1992
38
R.E. WELLER
monkeys and humans than that in owl monkeys (Cowey, '64;Cusick and Kaas, '88; Krubitzer and Kaas, '90; for review, see Jacobs, '85). In a previous investigation in squirrel monkeys (Steele et al., '91a), we found that the major layer IV or "feedforward" projection (Tigges et al., '73, '74, '81; Rockland and Pandya, '79) of the caudal subdivision of the Dorsolateral Visual Area (DLc),the probable homologue of at least part of V4 of macaque monkeys (see Steele et al., '91a, for review), was confined to a caudal region of IT, ITc (Fig 1).This provided the first evidence that IT cortex of squirrel monkeys contained subdivisions. Some questions remained unanswered from this study, however. The rostral border of ITc was not clearly established (R of Fig. 1).Single injections in DLc sometimes labeled separate locations in ITc, and the laminar distribution of cells retrogradely labeled in the very dorsal portion of ITc (d of Fig. 1)was different from that of cells labeled in more ventral ITc, suggesting that dorsal ITc might be a separate area. Projections from DLc also hinted at a crude topographic organization for caudal IT. Results from one case suggested that DLR, unlike DLc, projected to both caudal and rostral IT (Steele et al., '91a). Finally, projections to IT cortex from areas other than DLc and DLR have been occasionally reported in squirrel monkeys (e.g., V 11, Cusick and Kaas, '88; MT, Krubitzer and Kaas, '901, but needed confirmation. To answer questions about the subdivisions and cortical connections of IT cortex of squirrel monkeys, we made injections of neuroanatomical tracers into different locations in IT. Most injections were placed in the previously identified DLc-receptive region of IT, ITc (Steele et al., '91a), with additional injections placed in more rostral IT. IT cortex was also architectonically analyzed. The results provide evidence for at least two anteroposterior regions of IT cortex, ITc and IT,; and possibly a third, intermediate region, ITI. ITc contains separate dorsal and ventral areas, ITCd and ITCv. Some of these results have been briefly described previously (Weller et al., '89).
Abbreviations
D DI DL DLc DLR DM DPL ER FEF FST FV IT ITr
ITCd ITCv IT1
IT, ITS LIP MST MT PR STS V VI
v I1
VOT
dorsal area rostral t o V I1 dorsointermediatearea dorsolateralarea caudal subdivision of DL rostral subdivision of DL dorsomedial area dorsal prelunate area of macaques entorhinal cortex frontal eye field fundal superior temporal area frontal visual area lateral to the FEF inferior temporal caudal subdivision of IT cortex; the DLc-projectionzone the dorsal subdivision of IT? the ventral subdivision of 1Tr possible intermediate subdivision of IT cortex the rostral subdivision of IT cortex; the ITr:-projec:tion zone inferior temporal sulcus lateral intraparietal area of macaques medial superior temporal area middle temporal area perirhinal cortex superior temporal sulcus ventral cortex rostral toV I1 primary visual cortex or striate cortex second visual area ventral occipito-temporal area of macaques
AND
G.E. STEELE
2mm
Fig. 1. Subdivisions of visual cortex previously identified in squirrel monkeys. Lateral (A) and ventral (B) views of the left hemisphere are shown, with the banks of the superior and inferior temporal sulci (STS and ITS) opened to scale. Borders shown with broken lines are less well established than those shown by solid lines. V I, lateral V 11, and the frontal eye field have been electrophysiologically defined (Cowey, '64; Huerta et al., '87). Other areas have been identified by architectonic appearance and patterns of connections: MT (Tigges et al., '74, '81; Martinez-Millan and Hollander, '75; Wong-Riley, '79; Cusick and Kaas, '88; Krubitzer and Kaas, '90); MST and FST (Krubitzer and Kaas, '90); caudal (C) and rostral (R) subdivisions of DL (Cusick and Kaas, '88; Steele et al., '91a); a dorsal area or areas (D; Weller et al., '91; or DM and DI of Krubitzer and Kaas, '90); a medial area (not shown; MartinexMillan and Hollander, '75); and a frontal ventral area (FV; Huerta et al., '87). Rostra1borders are not shown for MST and FST because they have only been identified in flattened sections (Krubitzer and Kaas, '90). Feedforward projections from DLc defined a caudal subdivision of inferior temporal cortex, ITc, as the projection zone of DLc (Steele et al., '91a), but its rostral extent (R?) remained unclear. In most of IT(;, cells retrogradely labeled after injections in DLc; were in superficial and deep layers of cortex, but in dorsal IT(; (d),they were almost exclusively in deep layers. See text for abhreviations and details.
METHODS Cortical connections of IT cortex were revealed by injections of the neuroanatomical tracers wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP), 3Hamino acids, and fluorescent dyes in various locations in IT cortex in 19 hemispheres of 13 squirrel monkeys (Saimiri sciureus). Methods of surgery, injection of tracers, histology, and data analysis were similar to those used previously (Steele et al., '91a; Weller et al., '91). Many animals received injections of more than one tracer (Table 1).Two sterile surgeries were often used in these animals to accommodate the different transport periods needed for the fluorescent tracers (7-10 days; Keizer et al., '83) and WGA-HRP (2 days; Mesulam, '78). "-amino acids were injected at the
CONNECTIONS OF IT
39
TABLE 1. Summaw of Injections Made in the Present Study Location DLc Projection Zone, or ITc Dorsal
Case no.’
89-7L (1) 89-9 ( 2 ) 8 9 - E 4 (3) Dorsocentrd 90-8L (4) 90-5 (5) 89-19 (6) Ventral 89-7L ( 7 ) 89-a4 ( 8 ) 89-fj4 (9) 90-8R (101 90-7R(ll) 90-4L (12) 90-1L (13) 90-4L (14) Central 89-7R (15) 89-14 (16) 89-8 (17) 90-7L(18) Rostral ( I T ~ I T R 90-3R (19) border repionl 90-3L (201 90-3Ll21) 89-14(22) ITc ProJeCtiOnZone, 90-5 (23) or ITR 90-14(24) 90-4R(25,
Tracer injected2
FB 3H WGA-HRP FB DY WGA-HRP DY FB FB WGA-HRP WGA-HRP DY 3H 3H WGA-HRP FB, 3H FGold FB WGA-HRP 3H FB DY FB WGA-HRP WGA-HRP
Size (rnm2I3 Figure no.
1.04 4.24 4.71 .58 23 .92 1.05 2.48 5.64 9.60 10.90 .70 1.87 3.07 14.15 1.14 4 41 .68 6.40 3.03 1.30 .51 1.08 8.00 6.57
5 6 7 8
5 9 10
10 11 12 13 14 14 12 15 16
’Numbers in parentheses refer to the numbered injection sites in Figure 4. ‘DY, Diamidino Yellow; FB, Fast Blue; FGold, Fluoro-Gold; 3H, tritiated proline-leucine; WGA-HRP, wheat germ agglutinin conjugated to horseradish peroxidase. ”Refers to the center, or core, of the injection only. The size of the core plus halo was 35-19894 larger. ‘Injections that invaded underlying white matter and could have resulted in artifactual labeling.
same time as the fluorescent dyes. Surgery was performed on animals tranquilized with acepromazine maleate (0.55 mg/kg, i.m.1 and anesthetized with ketamine hydrochloride (50 mgikg, i.m.; White et al., ’82). Supplemental doses of acepromazine (0.55 mg/kg), ketamine (12.5-1 7 mg/kg), and, if necessary, short-acting barbiturates were given throughout surgery to maintain surgical levels of anesthesia. To expose IT cortex, two incisions were made in the scalp: one along the dorsal midline; and a second, shorter one, perpendicular to the first,just in front of the ear. Underlying fascia and the temporalis muscle were reflected. An opening was made in the skull over the intended injection site and the dura cut and reflected. Injections were placed relative to landmarks of the exposed cortex, such as the superior temporal sulcus and the visible edges of the temporal lobe. Concentrations of the tracers injected were 2-3% WGAHRP (Sigma, prepared from type VI peroxidase), 133 pCi/pl of a 1:l mixture of 3H-proline (112.0-126.9 Ci/ mmol; New England Nuclear) and 3H-leucine (140.0-143.7 Ci/mmol;New England Nuclear), and 2-3.5% of the fluorescent dyes Fast Blue, Diamidino Yellow (Illing GmbH & Co. KG, Germany) and Fluoro-Gold (Fluorochrome, Inc., Englewood, CO) in sterile saline. Injection volumes were 0.1-0.4 pl, except for Diamidino Yellow, for which larger volumes (0.5-1 ~ 1 were ) used. Tracers were delivered over a ten minute period through separate 1 p1 Hamilton syringes. After the injections were completed, the dura was folded back in place, the opening in the skull closed with a cap of dental acrylic, and the muscle and skin sutured closed. Postoperatively, the animal received an analgesic (Nalbuphine; 15 mg/kg, i.m.1. After a survival period appropriate for the tracer(s1 injected, the animal was anesthetized with ketamine hydro-
chloride, sacrificed with a lethal dose of sodium pentobarbital (i.p.), and perfused intracardially with 0.9% heparinized saline, followed by 3%buffered paraformaldehyde and 10% sucrose in phosphate buffer. The brain was removed, blocked, photographed in lateral and ventral views, and stored in 30% sucrose in phosphate buffer at 4°C for one day. Brains were cut in the coronal plane at 50 pm on a freezing microtome. Series of sections (2 or 3 out of 10) were processed for each type of injected tracer (fluorescent dyes, WGA-HRP, 3H-amino acids). Sections processed for WGA-HRP used tetramethylbenzidine (Mesulam, ’78) in a modification of the procedure of Gibson et al. (’84; see Steele et al., ’91a). Sections were processed for autoradiography according to Cowan et al. (’721,with 12 or 16 weeks of exposure. Sections examined for fluorescent label were mounted, dried on a warming tray, and stored, uncoverslipped, at 4°C. Fluorescent material was photographed from a separate series of slides coverslipped with DPX (Gallard Schlessinger) and stored at 4°C. Sections containing WGA-HRP or fluorescent label were stained with cresyl violet after the label was drawn and photographed. These sections were used for identifying the laminar locations of labeled cells and terminations. Additional 1 in 10 series of sections were stained for cytoarchitectonics with cresyl violet or for myeloarchitectonics with a modification of the Gallyas (’79) silver stain (Steele et al., ’91a). The latter sections were used for delineation of the architectonic borders of V I, MT, and subdivisions of the inferior temporal lobe. Because architectonic borders of several other subdivisions of visual cortex are most readily identified in squirrel monkeys in tangential, not coronal, sections (e.g., the rostral border of V 11,the border between DLc and D L R , the border between DLc and the dorsal area; Cusick and Kaas, ’88;Krubitzer and Kaas, ’90; Steele et al., ’91a), the locations of these borders were estimated in the present study. The WGA-HRP and autoradiographic sections were examined under dark-field optics, and the fluorescent sections were examined under the appropriate epifluorescent illumination. The locations of injection sites, labeled cells and terminations, and architectonic borders were drawn onto enlarged drawings of brain sections and reconstructed onto outlines of the cortex made from the photographs. When label was found within the superior and inferior temporal sulci, the banks of these sulci were “opened” in drawings to scale and the label shown on these drawings. Summary diagrams were prepared by transferring injection sites and label onto standard brain views, using sulci and architectonic borders for reference. A limited architectonic analysis of IT cortex was made from the experimental brains and several normal squirrel monkey brains sectioned in parasagittal, horizontal, or tangential planes. Sections from these brains were stained with cresyl violet or silver nitrate (Gallyas, ’791, as described above. In addition, sections were examined from squirrel monkey brains that had been embedded in celloidin, sectioned in the coronal or horizontal plane, and stained with thionin or hematoxylin.
RESULTS Results of the architectonic analysis of inferior temporal cortex will be described first. Described next are the ipsilatera1 cortical connections of IT cortex, the contralateral
R.E. WELLER
40
connections of IT cortex and the commissural pathways they traverse, and laminar patterns of connections.
Architectonic subdivisions of inferior temporal cortex Much of the lateral and ventrolateral inferior temporal lobe of squirrel monkeys appears similar to cortex identified in macaque monkeys as TE by von Bonin and Bailey (’47). The most pronounced cellular features of TE in squirrel monkeys are a broad, dense layer IV; large pyramidal cells at the bottom of layer I11 that are sometimes separated by a slight gap from the smaller cells of layer IV; and relatively light staining of layers V and VI (Figs. 2A-D, 3B-D). Additional features are a smooth border between a broad layer I and a narrow, darkly staining layer 11; the vertical alignment of cells in layers 111-V in “palisades” (Brodmann, ’09);and the slightly denser appearance of cells in layer VI than in layer V. TE is more heavily myelinated than most adjacent cortical areas, with prominent inner and outer bands of Baillarger (Figs. 2E-F, 3A). The outer band, in the location of layer IV, is narrow, while the inner band, extending from the middle of layer V to underlying white matter, is broad. The most obvious border of TE in squirrel monkeys is its ventral border with an area we identified as TF, after von Bonin and Bailey (’47). The TE-TF border is usually deep in the lateral bank of the inferior temporal sulcus (Figs. 2A, E-F, 3B, 4C). TF extends across the medial bank of the inferior temporal sulcus onto adjacent cortex of the parahippocampal gyrus. In contrast to TE, TF is characterized by a thinner layer IV and more prominent layers V and especially VI that contain larger pyramidal cells. In addition, TF is paler than TE myeloarchitectonically, with a much thinner outer band of myelination and virtually no inner band (Fig. 2E-F). Defining the precise location of the border between TE and TF was sometimes complicated by the compressed appearance of cortex in the inferior temporal sulcus and the lack of an exact correspondence between the decrease in thickness of layer IV and the increase in cell size and darkness of staining of layers V and VI in the transition from TE to TF. As a further impediment, TF and rostral TE are less dissimilar (see below), making it more difficult to distinguish their border. Not all of neocortex ventromedial to TE was TF. Two other regions could be identified medial to TF, TL and TH (Figs. 2,4C),also named after areas identified in macaques. According to von Bonin and Bailey (’47),in macaques, TF is replaced immediately caudal to entorhinal cortex by area TH. In contrast to TF, TH lacks a layer Iv, has little distinction between layers I1 and 111, and has many large pyramidal cells in undifferentiated layers V and VI. Area TL, not recognized by von Bonin and Bailey (’471, was described by Rosene and Pandya (’83)as an area between TF and TH with intermediate cytoarchitectonic features (a faint layer IV and slightly distinguishable layers I1 and 111). Although we observed cortex with features of TF, TL, and TH in squirrel monkeys, we found it difficult to distinguish their borders. Instead, we often noted a gradual lateromedial decrease in the thickness of layer IV, increase in the darkness of staining of layers V and VI, and decrease in myelination (Fig. 2). In primates with a well-developed rhinal sulcus, such as macaque monkeys, two types of transitional cortex have been described between ventral inferior temporal neocortex
AND
G.E. STEELE
and more medial entorhinal cortex: lateral perirhinal cortex and medial prorhinal cortex (van Hoesen and Pandya, ’75). Perirhinal cortex is distinguished by large, darkly staining layer V pyramidal cells below a pale, acellular “layer IV,” while prorhinal cortex has large, dark stellate cells in layer I1 and a fusion of layers 111-VI. Sanides (unpublished observations cited in Van Hoesen and Pandya, ’75) suggested that perirhinal cortex, or area 35 of Brodmann (’091, actually consisted of two areas, a lateral area 36 and a medial area 35. Among other differences, area 36 has a weak, but distinct, granular layer IV, while area 35 does not (Amaral et al., ’90). Based on these descriptions, squirrel monkeys have a narrow strip of transitional cortex that consists of lateral and medial subdivisions of perirhinal cortex, or areas 36 and 35, but do not appear to have prorhinal cortex (although the existence of an extremely narrow prorhinal cortex is difficult to rule out). Another usually discernible border of TE in squirrel monkeys is the anterior border of TE with cortex of the inferior temporal pole (Fig. 4). Following terminology of von Bonin and Bailey (’47), we refer to this polar cortex as TG. In comparison to TE, TG is thicker, has a much reduced layer IV, broader and more prominent infragranular layers, lacks large pyramidal cells in layer 111, a.nd is paler in sections stained for myelinated axons (Fig. 3A-B). The dorsal border of TE in the superior temporal sulcus is difficult to see because of the compressed appearance of this tissue. Although the remaining border of TE is with the caudal Dorsolateral Areas, which are pale in sections stained for myeloarchitecture (Cusick and Kaas, ’88;Krubitzer and Kaas, ’90; Steele et al., ’91a) and should thus contrast with the more myelinated TE, we were unable to distinguish this border in our largely coronal material. There are regional variations in the appearance of ‘I’E in squirrel monkeys that suggest the existence of subdivisions. The most obvious architectonic variation is between rostral TE and caudal TE (Fig. 3). Compared to caudal TE, rostral TE has a thinner layer IV, lacks the gap between the small cells of layer IV and the large pyramidal cells of‘deep layer 111, and is slightly less myelinated. In spite of these differences, the transition from caudal to rostral TE appeared to be gradual rather than sharp. Dorsocaudal TE along the lip of the superior temporal sulcus could be distinguished from ventrocaudal TE by its increased myelination (Fig. 2F) and scattered population of large and darkly staining pyramidal cells in layer 111. Occasionally, a narrow (2 mm) strip of cortex between dorsocaudal and ventrocaudal TE could be distinguished by its reduced layer IV, lack of a gap between the cells of layer IV and layer 111, and decreased myelination. However, because the strip was usually beneath a major surface blood vessel, its dissimilar appearance could be an artifact of compression.
Summary of injection sites Twenty-five injections of neuroanatomical tracers were made in IT cortex (Fig. 4 and Table 1).In relation to the previously defined projection zone of DLc, ITc (Steele et al., ’91a; Fig. 11, 18 injections were in caudal cortex that was clearly part of this zone (sites 1-18 of Fig. 41, four injections were near the rostral border of ITc or in the ITc-ITRborder region (sites 19-22), and three injections were in cortex clearly rostral to ITc (sites 23-25). For descriptive purposes, the injections in ITc have been divided into dorsal
CONNECTIONS OF IT
Fig. 2. Architectonic subdivisions of inferior temporal cortex in squirrel monkeys. Nomenclature is from von Bonin and Bailey ('471, except for TL, which is from Rosene and Pandya ('83). A is a photomicrograph of part of a coronal section, stained with thionin, through the right inferior temporal lobe. Dorsal (D) and lateral (L) are indicated. The superior temporal sulcus is shown at upper right, and the inferior temporal sulcus is shown below. The arrow in the lateral bank of the inferior temporal sulcus indicates the border of TE with TF. The section shown is from the comparative mammalian brain collection of the University of Wisconsin. Regions enclosed by brackets in A are shown
41
at higher power in B-D. For B-D, tissue is oriented so that the pial surface of cortex is at top, and layers of cortex are indicated at left. E is a photomicrograph of a coronal section through caudal inferior temporal cortex, stained with the Gallyas ('79) silver stain for myeloarchitecture. The large arrow in the inferior temporal sulcus (also shown in F) points to the border between TE and TF, and the smaller arrow in E points to the medial border of TF. F is a higher power view of the lateral portion of the section shown in E. Note the darker appearance of dorsal cortex between the arrows. Scale bar in A and E is 1 mm, and in B and F, it is 500 bm, C and D are at the same scale as B.
R.E. WELLER AND G.E. STEELE
42
Fig. 3. Rostrocaudal differences in the architectonic appearance of inferior temporal cortex of squirrel monkey. A and B are horizontal sections through the inferior temporal lobe of two different brains. Rostra1 (R) and lateral (L) are indicated. The section in A was stained with hematoxylin for myelinated axons, and the section in B was stained with thionin for cell bodies. A shows that caudal IT cortex is more myelinated than rostral cortex. The regions of cortex enclosed by brackets in B are shown at higher power in C (left bracket) and D (right
bracket). Layers of cortex are numbered in C and D after Brodmann ('09). In comparison to rostral IT (0,caudal IT (D) has a more darkly staining layer IV. The arrows in A and B point to the border between TE and TF. Scale bar in B is 1 mm. and A is at the same magnification. Scale bar in D is 500 km, and C is at the same magnification. Sections shown are from the comparative mammalian brain collection of the University of Wisconsin.
(sites 1-6), ventral (sites 7-14), and central (sites 15-18) groups.
numerous in its dorsal than ventral portion, while cells labeled in DLR were equally distributed in dorsal and ventral regions. In this and many other cases, dense labd surrounded the injection site in IT (e.g., Figs. 5-81, In the present case, many labeled cells were also found immediately caudal and dorsal to the injection, including in the ventral bank of the superior temporal sulcus and depths of the dorsal bank. By location, some of these cells may be in the fundal area of the superior temporal sulcus, FST, and possibly also the medial area of the superior temporal area, MST (see Fig. 1). Ventral ITc and central IT cortex displayed little label. A broad region of rostral IT cortex was strongly labeled, including much of the ventral bank of the superior temporal sulcus. Both banks of the rostral inferior temporal sulcus were labeled more sparsely, as was parahippocampal cortex medial to this sulcus. Two regions of prefrontal
Ipsilateral connections of the DLc-projection zone, ITc Dorsal injections. Six injections were made in the dorsal aspect of ITc (injection sites 1-6 of Fig. 4; see also site 15 of the central injections). In the first case, an injection of the retrograde tracer Fast Blue was made close to the superior temporal sulcus (injection site 1 of Fig. 4; case 89-7L; Fig. 5A-D; the results of a second injection made in the same case in ventral cortex are described later). For this and subsequent injections, the size of the core of the injection is given in Table 1. Cells were labeled over an extensive region of lateral extrastriate cortex estimated as rostral V 11, DLc, DLR, cortex dorsal to DLR, and perhaps M T (compare with Fig. 1).Cells labeled in DLc were more
CONNECTIONS OF IT
Fig. 4. Summary of the locations of the 25 injections of neuroanatomical tracers made in IT cortex in the present study. Architectonic subdivisions of the inferior temporal lobe are also shown (TE, TG, TF, TL, TH, PR, ER; see text). Lateral (A and B)and ventral (C) views of a standard left hemisphere are shown. The banks of the inferior temporal sulcus are opened in C . Individual injection site centers were placed on the summary in reference to the superior temporal sulcus, other limits of the inferior temporal lobe, and architectonic borders or their estimated locations. Note that injections 1-3 involved the dorsal region (D) of ITc that had a different laminar distribution of retrograde label following injections in DLc than the remainder of ITc, and injections 19-22 involved the questionable rostral (R) portion of ITc (Fig. 1)or the ITc-ITRborder region. See Table 1for details about the injections, and text for abbreviations.
cortex were labeled, one in dorsolateral cortex, lateral to the inferior arcuate sulcus; and the other in more lateral cortex, close to the lateral apex of cortex. In a second dorsal case (89-91, a larger injection of the anterograde tracer 3H-proline-leucine was made in a location similar to that of the previous case (site 2 of Fig. 4;Fig. 6). The injection strongly labeled the regions of dorsal DLc and DLR. Sparser label was in lateral DLc; cortex dorsal to DLR, approaching the Sylvian fissure; and caudal MT, or
43
the portion thought to represent central vision, based on connections with V I and V I1 (Tigges et al., '74, '81; Krubitzer and Kaas, '90). As in case 89-72, (Fig. 5 ) , much of ventral ITc was unlabeled, except for scattered foci close t o and within the inferior temporal sulcus. In the superior temporal sulcus, patches of label were scattered along much of the ventral bank. Rostroventral IT cortex, including the lateral bank of the inferior temporal sulcus, was densely labeled. There was a small focus of label in dorsolateral prefrontal cortex. A large injection of WGA-HRP was made in dorsal ITc in a third case, 89-15 (site 3 of Fig. 4; Fig. 7). The injection extended 0.6 mm into the white matter, including a 0.2 mm incursion of the syringe track. Dense label occupied the region of lateral to dorsal DLc, most of DLR, and cortex dorsorostral to the latter areas, extending to the Sylvian fissure. Dorsocaudal MT was weakly labeled, and lateroventral V I1 contained a few scattered labeled cells. Cortex dorsal and rostral to the injection was heavily labeled, including cortex in the inferior and superior temporal sulci. In the latter sulcus, cortex labeled included much of the ventral bank and foci in the dorsal bank. The rostral parahippocampal gyrus contained primarily retrograde label. In prefrontal cortex, labeled cortex extended from dorsolateral to lateral orbital. An occasional cell was labeled in entorhinal cortex, the superior temporal gyrus, and rostral cingulate cortex. While the pattern of label observed in this case was basically similar to that in the previous cases, the injection in 89-15 produced much more label, some of which was in areas unlabeled in the other cases (i.e., lateral orbital cortex, cortex dorsal to DLc and DLR, and entorhinal cortex). This additional label may be due to more effective transport from the larger injection or the syringe track's slight encroachment into the white matter and transport of WGA-HRP by cut fibers of passage (Mesulam, '82). The next three injections were placed in dorsocentral rather than dorsal ITc. A small injection of the retrograde tracer Fast Blue was made in an anterior location in case 90-8L (site 4 of Fig. 4;Fig. 8). Unlike results observed in the previous cases, cells labeled in DLc and DLRin 90-8L were primarily in lateroventral rather than dorsal cortex. Fewer cells were labeled in MT and the ventral bank of the caudal Sylvian fissure. Many cells were labeled in rostral and caudal IT cortex, including in the superior and inferior temporal sulci. An extensive region of the parahippocampal gyrus was strongly labeled. In prefrontal cortex, foci of labeled cells were in dorsolateral, lateral, lateral orbital, and, quite sparsely, medial orbital cortex. A few cells were labeled in rostral cingulate cortex. Another small injection of Diamidino Yellow was made in dorsocentral ITc in case 90-5 (site 5 of Fig. 4; not illustrated). Although this case produced considerably fewer labeled cells in comparison to previous cases, this could be related to the limited size of the injection. Cells labeled in DLc were in rostroventral DLc, if in DLc at all, rather than cortex anterior to DLc. Ventral DLR contained a small cluster of cells, and ventrorostral MT, a single cell. Besides IT cortex labeled in the vicinity of the injection, including the ventral bank of the superior temporal sulcus, label in IT was limited to a small number of cells in rostral IT and the lateral bank of the inferior temporal sulcus. A few cells were labeled in lateral prefrontal cortex. A final dorsocentral injection of WGA-HRP (case 89-19, site 6 of Fig. 4;not illustrated) was quite caudal. Although,
TS
a9 -
Fig. 5. Neurons retrogradely labeled by an injection of Fast Blue in dorsocaudal IT cortex (A-D) in comparison to those labeled by an injection of Diamidino Yellow in ventrocaudal IT cortex (E-H) of the same animal (case 89-7L, injection sites 1and 7 of Fig. 4). A and E are lateral views and C and H are ventral views of the left hemisphere. The banks of the superior temporal sulcus (STS) in the regions enclosed by rectangles in A and E are opened in B and F, and the banks of the inferior temporal sulcus (ITS) enclosed by rectangles in C and H are opened in D and G. Injection centers are black and halos are shaded. Dots indxate labeled cells, with the number
V
Fast Blue
Diarnidino Yellow
2,:.*
+
- -..c . .. .. ...,F..:
.a.
2mm
$5
-
::... .
G
ITS
of dots representing the relative density of labeling, except where isolated dots appear, indicating individual cells. The star in A indicates where the injection site of Diamidino Yellow obscured detection of cells that might have been labeled by Fast Blue; the reverse is shown in E by the star for the Fast Blue injection site. Areal borders indicated by solid lines were architectonically determined in the present case, while borders represented with dashed lines were estimated. See list in text for abbreviations.
7~
M
F
H M M
r/)
$ U
?3
tF M
r:M
4
4
45
CONNECTIONS OF IT 3H
89 - 9 - Proline - Leucine
ITS
Fig. 6. Projections observed as the result of an injection of 3H-proline-leucine in dorsocaudal IT cortex in case 89-9 (injection site 2' of Fig. 4). A is a lateral view and C is a ventral view of the left hemisphere. The banks of the STS are opened in B, and those of the ITS are opened in D. Dots indicate the density of autoradiographic silver grains; other conventions as in previous figures.
by location, this injection could have involved DLR, the pattern of label it produced was largely unlike that described for DLR (i.e., it failed to label MT and the dorsal area; Cusick and Kaas, '88; Steele et al., '91a), but similar to that found for nearby IT injections in other cases (sites 5 and 18 of Fig. 4).Most label in DLc was rostroventral. The ventral bank of the superior temporal sulcus and separate central and rostral regions of lateral IT were labeled. The only label on the ventral surface of the hemisphere was a small patch of anterograde label in the lateral bank of the inferior temporal sulcus several millimeters rostral to V 11. Label in prefrontal cortex was limited to dorsolateral cortex. To briefly compare the results of the six dorsal injections, the most dorsal of the injections (sites 1-3) produced a somewhat different pattern of label than that produced by the other, more dorsocentral injections (sites 4-6). The three dorsal injections strongly labeled DLc, DLR,superficial and deep portions of the ventral bank of the superior
temporal sulcus, and rostral IT; and more weakly labeled MT, V IT, cortex dorsorostral to DLR,the deep portion of the dorsal bank of the superior temporal sulcus, the inferior temporal sulcus, parahippocampal gyrus, inferior temporal pole, rostral cingulate cortex, and dorsolateral to lateral prefrontal cortex. The dorsocentral injections also labeled many of the same areas, but, unlike the dorsal injections, the dorsocentral injections labeled dorsolateral DLc only sparsely, but labeled rostroventral DLc or cortex just anterior to this portion of DLc more strongly; only weakly labeled DLRand the depths of the superior temporal sulcus; and did not label V I1 or cortex dorsal to DLR;but did label the caudal Sylvian fissure and lateral orbital cortex (Table 2).
Ventral injections. Eight injections were made in ventral ITc (injection sites 7-14 of Fig. 4).The pattew of label observed in case 89-7L, in which a small injection of the retrograde tracer Diamidino Yellow was made in ventral cortex (site 7 of Fig. 4; Fig. 5 ) , was representative of that
- V9M Sl - 6 8
d8H
CONNECTIONS OF IT
47
90
- 8L
Fast Blue
r
. ....... D . ,
Fig. 8. Cells labeled by an injection of Fast Blue in dorsocentral IT cortex in case 90-8L (site 4 of Fig. 4). A, lateral view; C, ventral view; B, the opened STS; and D, the opened ITS. Open arrows indicate where labeled cells were found in the Sylvian fissure; other conventions and abbreviations as in previous figures.
observed in most of the other cases. Although the ventral injection in 89-7L labeled cells throughout most of DLc, cells were densest in lateral-to-ventral DLc. Labeled cells in DLRwere few in number and scattered. Cells were labeled in rostrolateral V 11, or the part estimated to represent central vision (Cowey, '64). A few cells were labeled in the region of MT, in cortex dorsal to DLc, along the lateral apex of prefrontal cortex, and in rostral cingulate cortex. Most cells labeled in caudal IT were in the vicinity of the injection, except for cells in the ventral bank of the superior temporal sulcus and close to the inferior temporal sulcus. A moderate number of cells were labeled in rostral IT, including the rostral tip of the inferior temporal sulcus; and in parahippocampal cortex caudal to entorhinal cortex. The existence of injections in dorsal and ventral portions of ITc in case 89-7L permitted direct comparison of the resulting patterns of labeled cells. While both injections labeled similar regions and numbers of cells in the inferior temporal sulcus, parahippocampal gyrus, and rostroventral IT, they differed in labeling complementary portions of DLc
and in the extent of labeling they produced in the superior temporal sulcus (Fig. 5). Two other cases with injections of retrograde tracers in locations similar to that of the ventral injection in 89-7L (injection sites 8 and 9 of Fig. 4;not illustrated) labeled cells in locations similar to those labeled in 89-7L, with the exception of some additional label that could be due to the larger sizes of the injections or, more likely, the fact that syringe tracks in both cases involved 0.7 mm of underlying white matter. A large injection of WGA-HRP was made in caudoventral IT in case 90-8R (site 10 of Fig. 4; Fig. 9). Although this injection extended quite caudally, it appeared to spare DLc, based on its location relative to the borders of MT and V I. Lateral to ventrolateral V I1 was strongly labeled. Although much of DLRand DLc were labeled, lateral DLc was labeled most strongly, and dorsolateral DLc contained primarily retrograde label. Caudal and ventral portions of MT were labeled, and there were two small foci of label in the caudal Sylvian fissure. Label in rostral IT cortex was limited, and
R.E. WELLER
48 TABLE 2. Summary of Connections Between Different Parts of IT Cortex and Other Areas of Cortex in Squirrel Monkeys' Area injected Area of Cortex labeled
v I1 DL
Cortical Connections of Subdivisions of Inferior Temporal Cortex in Squirrel Monkeys R.E. WELLER
AND
G.E. STEELE
Department of Psychology, University of Alabama at Birmingham, Birmingham, Alabama 35294
ABSTRACT Patterns of cortical connections and architectonics were used to determine subdivisions of inferior temporal (IT) cortex of squirrel monkeys. Single or multiple injections of the tracers wheat germ agglutinin-horseradishperoxidase, Fast Blue, Diamidino Yellow, Fluoro-Gold, and 3H-aminoacids were placed into IT cortex. Most injections were placed in caudal IT cortex in the region previously shown to receive input from the caudal subdivision of the Dorsolateral Area, DLc; additional injections were placed in more rostral IT cortex. The results indicate the presence of two major regions: a caudal region, ITc, and a rostral region, ITR.An intermediate region of cortex along the ITC-ITRborder that displays some connections of ITc and some connections of ITRmay be another area. ITc contains a more myelinated dorsal area, ITCd, and a larger ventral area, ITCv. Both ITCd and ITCv receive a major projection from DLC; additional input from DLR,MT, and V 11; and send strong projections to ITR,the lateral bank of the superior temporal sulcus, and dorsolateral prefrontal cortex. Only ITCd has strong connections with DLR and cortex in the depths of the superior temporal sulcus, and only ITCv has connections with lateral orbital cortex. The overall pattern of connections between ITc and DLc suggests that ITc has a crude topographic organization, with dorsal cortex representing the lower field and ventral cortex representing the upper field. ITRdiffers from ITc by receiving little if any input from DLc; projecting to inferior temporal polar cortex, the rostral Sylvian fissure, and medial orbital cortex; and having a less distinct layer IV. Comparison of subdivisions of inferior temporal cortex defined in the present study in squirrel monkeys and those reported in other primates suggests that ITc of squirrel monkeys may correspond to area TEO of macaque monkeys. o 1992 Wiley-Liss, Inc. Key words: dorsolateral area, V4, TE, TEO, primates
Humans with damage to inferior occipito-temporal cortex exhibit a deficit in the visual identification and recognition of objects called visual agnosia (for review, see Damasio et al., '82; Levine, '82; Farah, '90). A comparable visual memory deficit is found in macaque monkeys with damage to inferior temporal (IT) cortex co-extensive with architectonic area TE of von Bonin and Bailey ('47; Mishkin, '54; Mishkin and Pribram, '54;Iwai and Mishkin, '68, '69;Cowey and Gross, '70; for review, see Dean, '82; Mishkin, '82). An increasing amount of information gained from a number of species of primates suggests that IT contains more than one visual area (e.g., Seltzer and Pandya, '78; Desimone and Gross, '79;Weller and Kaas, '87; Yaginuma, '90; Van Essen et al., '90). However, the total number and borders of such subdivisions are not well established in any primate. Most previous experiments on IT cortex have used macaque monkeys. Such experiments have been compli-
o 1992 WILEY-LISS, INC.
cated by the many cortical sulci and ventrolateral location of IT cortex in macaques. To minimize these problems, Weller and Kaas ('85, '87) investigated IT cortex in the relatively lissencephalic brain of New World owl monkeys. However, because owl monkeys are not diurnal like the majority of primates, subdivisions of IT cortex in owl monkeys may be less differentiated or otherwise different than in other primates. Thus, to further investigate the organization of IT cortex in primates, we have turned to another New World primate, the squirrel monkey, as a model. Squirrel monkeys, like owl monkeys, have a relatively unfissured cortex, but are diurnal and thus more representative of primates. Indeed, previous studies of visual cortex in squirrel monkeys have revealed an emphasis on central vision more comparable to that in macaque Accepted June 3,1992
38
R.E. WELLER
monkeys and humans than that in owl monkeys (Cowey, '64;Cusick and Kaas, '88; Krubitzer and Kaas, '90; for review, see Jacobs, '85). In a previous investigation in squirrel monkeys (Steele et al., '91a), we found that the major layer IV or "feedforward" projection (Tigges et al., '73, '74, '81; Rockland and Pandya, '79) of the caudal subdivision of the Dorsolateral Visual Area (DLc),the probable homologue of at least part of V4 of macaque monkeys (see Steele et al., '91a, for review), was confined to a caudal region of IT, ITc (Fig 1).This provided the first evidence that IT cortex of squirrel monkeys contained subdivisions. Some questions remained unanswered from this study, however. The rostral border of ITc was not clearly established (R of Fig. 1).Single injections in DLc sometimes labeled separate locations in ITc, and the laminar distribution of cells retrogradely labeled in the very dorsal portion of ITc (d of Fig. 1)was different from that of cells labeled in more ventral ITc, suggesting that dorsal ITc might be a separate area. Projections from DLc also hinted at a crude topographic organization for caudal IT. Results from one case suggested that DLR, unlike DLc, projected to both caudal and rostral IT (Steele et al., '91a). Finally, projections to IT cortex from areas other than DLc and DLR have been occasionally reported in squirrel monkeys (e.g., V 11, Cusick and Kaas, '88; MT, Krubitzer and Kaas, '901, but needed confirmation. To answer questions about the subdivisions and cortical connections of IT cortex of squirrel monkeys, we made injections of neuroanatomical tracers into different locations in IT. Most injections were placed in the previously identified DLc-receptive region of IT, ITc (Steele et al., '91a), with additional injections placed in more rostral IT. IT cortex was also architectonically analyzed. The results provide evidence for at least two anteroposterior regions of IT cortex, ITc and IT,; and possibly a third, intermediate region, ITI. ITc contains separate dorsal and ventral areas, ITCd and ITCv. Some of these results have been briefly described previously (Weller et al., '89).
Abbreviations
D DI DL DLc DLR DM DPL ER FEF FST FV IT ITr
ITCd ITCv IT1
IT, ITS LIP MST MT PR STS V VI
v I1
VOT
dorsal area rostral t o V I1 dorsointermediatearea dorsolateralarea caudal subdivision of DL rostral subdivision of DL dorsomedial area dorsal prelunate area of macaques entorhinal cortex frontal eye field fundal superior temporal area frontal visual area lateral to the FEF inferior temporal caudal subdivision of IT cortex; the DLc-projectionzone the dorsal subdivision of IT? the ventral subdivision of 1Tr possible intermediate subdivision of IT cortex the rostral subdivision of IT cortex; the ITr:-projec:tion zone inferior temporal sulcus lateral intraparietal area of macaques medial superior temporal area middle temporal area perirhinal cortex superior temporal sulcus ventral cortex rostral toV I1 primary visual cortex or striate cortex second visual area ventral occipito-temporal area of macaques
AND
G.E. STEELE
2mm
Fig. 1. Subdivisions of visual cortex previously identified in squirrel monkeys. Lateral (A) and ventral (B) views of the left hemisphere are shown, with the banks of the superior and inferior temporal sulci (STS and ITS) opened to scale. Borders shown with broken lines are less well established than those shown by solid lines. V I, lateral V 11, and the frontal eye field have been electrophysiologically defined (Cowey, '64; Huerta et al., '87). Other areas have been identified by architectonic appearance and patterns of connections: MT (Tigges et al., '74, '81; Martinez-Millan and Hollander, '75; Wong-Riley, '79; Cusick and Kaas, '88; Krubitzer and Kaas, '90); MST and FST (Krubitzer and Kaas, '90); caudal (C) and rostral (R) subdivisions of DL (Cusick and Kaas, '88; Steele et al., '91a); a dorsal area or areas (D; Weller et al., '91; or DM and DI of Krubitzer and Kaas, '90); a medial area (not shown; MartinexMillan and Hollander, '75); and a frontal ventral area (FV; Huerta et al., '87). Rostra1borders are not shown for MST and FST because they have only been identified in flattened sections (Krubitzer and Kaas, '90). Feedforward projections from DLc defined a caudal subdivision of inferior temporal cortex, ITc, as the projection zone of DLc (Steele et al., '91a), but its rostral extent (R?) remained unclear. In most of IT(;, cells retrogradely labeled after injections in DLc; were in superficial and deep layers of cortex, but in dorsal IT(; (d),they were almost exclusively in deep layers. See text for abhreviations and details.
METHODS Cortical connections of IT cortex were revealed by injections of the neuroanatomical tracers wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP), 3Hamino acids, and fluorescent dyes in various locations in IT cortex in 19 hemispheres of 13 squirrel monkeys (Saimiri sciureus). Methods of surgery, injection of tracers, histology, and data analysis were similar to those used previously (Steele et al., '91a; Weller et al., '91). Many animals received injections of more than one tracer (Table 1).Two sterile surgeries were often used in these animals to accommodate the different transport periods needed for the fluorescent tracers (7-10 days; Keizer et al., '83) and WGA-HRP (2 days; Mesulam, '78). "-amino acids were injected at the
CONNECTIONS OF IT
39
TABLE 1. Summaw of Injections Made in the Present Study Location DLc Projection Zone, or ITc Dorsal
Case no.’
89-7L (1) 89-9 ( 2 ) 8 9 - E 4 (3) Dorsocentrd 90-8L (4) 90-5 (5) 89-19 (6) Ventral 89-7L ( 7 ) 89-a4 ( 8 ) 89-fj4 (9) 90-8R (101 90-7R(ll) 90-4L (12) 90-1L (13) 90-4L (14) Central 89-7R (15) 89-14 (16) 89-8 (17) 90-7L(18) Rostral ( I T ~ I T R 90-3R (19) border repionl 90-3L (201 90-3Ll21) 89-14(22) ITc ProJeCtiOnZone, 90-5 (23) or ITR 90-14(24) 90-4R(25,
Tracer injected2
FB 3H WGA-HRP FB DY WGA-HRP DY FB FB WGA-HRP WGA-HRP DY 3H 3H WGA-HRP FB, 3H FGold FB WGA-HRP 3H FB DY FB WGA-HRP WGA-HRP
Size (rnm2I3 Figure no.
1.04 4.24 4.71 .58 23 .92 1.05 2.48 5.64 9.60 10.90 .70 1.87 3.07 14.15 1.14 4 41 .68 6.40 3.03 1.30 .51 1.08 8.00 6.57
5 6 7 8
5 9 10
10 11 12 13 14 14 12 15 16
’Numbers in parentheses refer to the numbered injection sites in Figure 4. ‘DY, Diamidino Yellow; FB, Fast Blue; FGold, Fluoro-Gold; 3H, tritiated proline-leucine; WGA-HRP, wheat germ agglutinin conjugated to horseradish peroxidase. ”Refers to the center, or core, of the injection only. The size of the core plus halo was 35-19894 larger. ‘Injections that invaded underlying white matter and could have resulted in artifactual labeling.
same time as the fluorescent dyes. Surgery was performed on animals tranquilized with acepromazine maleate (0.55 mg/kg, i.m.1 and anesthetized with ketamine hydrochloride (50 mgikg, i.m.; White et al., ’82). Supplemental doses of acepromazine (0.55 mg/kg), ketamine (12.5-1 7 mg/kg), and, if necessary, short-acting barbiturates were given throughout surgery to maintain surgical levels of anesthesia. To expose IT cortex, two incisions were made in the scalp: one along the dorsal midline; and a second, shorter one, perpendicular to the first,just in front of the ear. Underlying fascia and the temporalis muscle were reflected. An opening was made in the skull over the intended injection site and the dura cut and reflected. Injections were placed relative to landmarks of the exposed cortex, such as the superior temporal sulcus and the visible edges of the temporal lobe. Concentrations of the tracers injected were 2-3% WGAHRP (Sigma, prepared from type VI peroxidase), 133 pCi/pl of a 1:l mixture of 3H-proline (112.0-126.9 Ci/ mmol; New England Nuclear) and 3H-leucine (140.0-143.7 Ci/mmol;New England Nuclear), and 2-3.5% of the fluorescent dyes Fast Blue, Diamidino Yellow (Illing GmbH & Co. KG, Germany) and Fluoro-Gold (Fluorochrome, Inc., Englewood, CO) in sterile saline. Injection volumes were 0.1-0.4 pl, except for Diamidino Yellow, for which larger volumes (0.5-1 ~ 1 were ) used. Tracers were delivered over a ten minute period through separate 1 p1 Hamilton syringes. After the injections were completed, the dura was folded back in place, the opening in the skull closed with a cap of dental acrylic, and the muscle and skin sutured closed. Postoperatively, the animal received an analgesic (Nalbuphine; 15 mg/kg, i.m.1. After a survival period appropriate for the tracer(s1 injected, the animal was anesthetized with ketamine hydro-
chloride, sacrificed with a lethal dose of sodium pentobarbital (i.p.), and perfused intracardially with 0.9% heparinized saline, followed by 3%buffered paraformaldehyde and 10% sucrose in phosphate buffer. The brain was removed, blocked, photographed in lateral and ventral views, and stored in 30% sucrose in phosphate buffer at 4°C for one day. Brains were cut in the coronal plane at 50 pm on a freezing microtome. Series of sections (2 or 3 out of 10) were processed for each type of injected tracer (fluorescent dyes, WGA-HRP, 3H-amino acids). Sections processed for WGA-HRP used tetramethylbenzidine (Mesulam, ’78) in a modification of the procedure of Gibson et al. (’84; see Steele et al., ’91a). Sections were processed for autoradiography according to Cowan et al. (’721,with 12 or 16 weeks of exposure. Sections examined for fluorescent label were mounted, dried on a warming tray, and stored, uncoverslipped, at 4°C. Fluorescent material was photographed from a separate series of slides coverslipped with DPX (Gallard Schlessinger) and stored at 4°C. Sections containing WGA-HRP or fluorescent label were stained with cresyl violet after the label was drawn and photographed. These sections were used for identifying the laminar locations of labeled cells and terminations. Additional 1 in 10 series of sections were stained for cytoarchitectonics with cresyl violet or for myeloarchitectonics with a modification of the Gallyas (’79) silver stain (Steele et al., ’91a). The latter sections were used for delineation of the architectonic borders of V I, MT, and subdivisions of the inferior temporal lobe. Because architectonic borders of several other subdivisions of visual cortex are most readily identified in squirrel monkeys in tangential, not coronal, sections (e.g., the rostral border of V 11,the border between DLc and D L R , the border between DLc and the dorsal area; Cusick and Kaas, ’88;Krubitzer and Kaas, ’90; Steele et al., ’91a), the locations of these borders were estimated in the present study. The WGA-HRP and autoradiographic sections were examined under dark-field optics, and the fluorescent sections were examined under the appropriate epifluorescent illumination. The locations of injection sites, labeled cells and terminations, and architectonic borders were drawn onto enlarged drawings of brain sections and reconstructed onto outlines of the cortex made from the photographs. When label was found within the superior and inferior temporal sulci, the banks of these sulci were “opened” in drawings to scale and the label shown on these drawings. Summary diagrams were prepared by transferring injection sites and label onto standard brain views, using sulci and architectonic borders for reference. A limited architectonic analysis of IT cortex was made from the experimental brains and several normal squirrel monkey brains sectioned in parasagittal, horizontal, or tangential planes. Sections from these brains were stained with cresyl violet or silver nitrate (Gallyas, ’791, as described above. In addition, sections were examined from squirrel monkey brains that had been embedded in celloidin, sectioned in the coronal or horizontal plane, and stained with thionin or hematoxylin.
RESULTS Results of the architectonic analysis of inferior temporal cortex will be described first. Described next are the ipsilatera1 cortical connections of IT cortex, the contralateral
R.E. WELLER
40
connections of IT cortex and the commissural pathways they traverse, and laminar patterns of connections.
Architectonic subdivisions of inferior temporal cortex Much of the lateral and ventrolateral inferior temporal lobe of squirrel monkeys appears similar to cortex identified in macaque monkeys as TE by von Bonin and Bailey (’47). The most pronounced cellular features of TE in squirrel monkeys are a broad, dense layer IV; large pyramidal cells at the bottom of layer I11 that are sometimes separated by a slight gap from the smaller cells of layer IV; and relatively light staining of layers V and VI (Figs. 2A-D, 3B-D). Additional features are a smooth border between a broad layer I and a narrow, darkly staining layer 11; the vertical alignment of cells in layers 111-V in “palisades” (Brodmann, ’09);and the slightly denser appearance of cells in layer VI than in layer V. TE is more heavily myelinated than most adjacent cortical areas, with prominent inner and outer bands of Baillarger (Figs. 2E-F, 3A). The outer band, in the location of layer IV, is narrow, while the inner band, extending from the middle of layer V to underlying white matter, is broad. The most obvious border of TE in squirrel monkeys is its ventral border with an area we identified as TF, after von Bonin and Bailey (’47). The TE-TF border is usually deep in the lateral bank of the inferior temporal sulcus (Figs. 2A, E-F, 3B, 4C). TF extends across the medial bank of the inferior temporal sulcus onto adjacent cortex of the parahippocampal gyrus. In contrast to TE, TF is characterized by a thinner layer IV and more prominent layers V and especially VI that contain larger pyramidal cells. In addition, TF is paler than TE myeloarchitectonically, with a much thinner outer band of myelination and virtually no inner band (Fig. 2E-F). Defining the precise location of the border between TE and TF was sometimes complicated by the compressed appearance of cortex in the inferior temporal sulcus and the lack of an exact correspondence between the decrease in thickness of layer IV and the increase in cell size and darkness of staining of layers V and VI in the transition from TE to TF. As a further impediment, TF and rostral TE are less dissimilar (see below), making it more difficult to distinguish their border. Not all of neocortex ventromedial to TE was TF. Two other regions could be identified medial to TF, TL and TH (Figs. 2,4C),also named after areas identified in macaques. According to von Bonin and Bailey (’47),in macaques, TF is replaced immediately caudal to entorhinal cortex by area TH. In contrast to TF, TH lacks a layer Iv, has little distinction between layers I1 and 111, and has many large pyramidal cells in undifferentiated layers V and VI. Area TL, not recognized by von Bonin and Bailey (’471, was described by Rosene and Pandya (’83)as an area between TF and TH with intermediate cytoarchitectonic features (a faint layer IV and slightly distinguishable layers I1 and 111). Although we observed cortex with features of TF, TL, and TH in squirrel monkeys, we found it difficult to distinguish their borders. Instead, we often noted a gradual lateromedial decrease in the thickness of layer IV, increase in the darkness of staining of layers V and VI, and decrease in myelination (Fig. 2). In primates with a well-developed rhinal sulcus, such as macaque monkeys, two types of transitional cortex have been described between ventral inferior temporal neocortex
AND
G.E. STEELE
and more medial entorhinal cortex: lateral perirhinal cortex and medial prorhinal cortex (van Hoesen and Pandya, ’75). Perirhinal cortex is distinguished by large, darkly staining layer V pyramidal cells below a pale, acellular “layer IV,” while prorhinal cortex has large, dark stellate cells in layer I1 and a fusion of layers 111-VI. Sanides (unpublished observations cited in Van Hoesen and Pandya, ’75) suggested that perirhinal cortex, or area 35 of Brodmann (’091, actually consisted of two areas, a lateral area 36 and a medial area 35. Among other differences, area 36 has a weak, but distinct, granular layer IV, while area 35 does not (Amaral et al., ’90). Based on these descriptions, squirrel monkeys have a narrow strip of transitional cortex that consists of lateral and medial subdivisions of perirhinal cortex, or areas 36 and 35, but do not appear to have prorhinal cortex (although the existence of an extremely narrow prorhinal cortex is difficult to rule out). Another usually discernible border of TE in squirrel monkeys is the anterior border of TE with cortex of the inferior temporal pole (Fig. 4). Following terminology of von Bonin and Bailey (’47), we refer to this polar cortex as TG. In comparison to TE, TG is thicker, has a much reduced layer IV, broader and more prominent infragranular layers, lacks large pyramidal cells in layer 111, a.nd is paler in sections stained for myelinated axons (Fig. 3A-B). The dorsal border of TE in the superior temporal sulcus is difficult to see because of the compressed appearance of this tissue. Although the remaining border of TE is with the caudal Dorsolateral Areas, which are pale in sections stained for myeloarchitecture (Cusick and Kaas, ’88;Krubitzer and Kaas, ’90; Steele et al., ’91a) and should thus contrast with the more myelinated TE, we were unable to distinguish this border in our largely coronal material. There are regional variations in the appearance of ‘I’E in squirrel monkeys that suggest the existence of subdivisions. The most obvious architectonic variation is between rostral TE and caudal TE (Fig. 3). Compared to caudal TE, rostral TE has a thinner layer IV, lacks the gap between the small cells of layer IV and the large pyramidal cells of‘deep layer 111, and is slightly less myelinated. In spite of these differences, the transition from caudal to rostral TE appeared to be gradual rather than sharp. Dorsocaudal TE along the lip of the superior temporal sulcus could be distinguished from ventrocaudal TE by its increased myelination (Fig. 2F) and scattered population of large and darkly staining pyramidal cells in layer 111. Occasionally, a narrow (2 mm) strip of cortex between dorsocaudal and ventrocaudal TE could be distinguished by its reduced layer IV, lack of a gap between the cells of layer IV and layer 111, and decreased myelination. However, because the strip was usually beneath a major surface blood vessel, its dissimilar appearance could be an artifact of compression.
Summary of injection sites Twenty-five injections of neuroanatomical tracers were made in IT cortex (Fig. 4 and Table 1).In relation to the previously defined projection zone of DLc, ITc (Steele et al., ’91a; Fig. 11, 18 injections were in caudal cortex that was clearly part of this zone (sites 1-18 of Fig. 41, four injections were near the rostral border of ITc or in the ITc-ITRborder region (sites 19-22), and three injections were in cortex clearly rostral to ITc (sites 23-25). For descriptive purposes, the injections in ITc have been divided into dorsal
CONNECTIONS OF IT
Fig. 2. Architectonic subdivisions of inferior temporal cortex in squirrel monkeys. Nomenclature is from von Bonin and Bailey ('471, except for TL, which is from Rosene and Pandya ('83). A is a photomicrograph of part of a coronal section, stained with thionin, through the right inferior temporal lobe. Dorsal (D) and lateral (L) are indicated. The superior temporal sulcus is shown at upper right, and the inferior temporal sulcus is shown below. The arrow in the lateral bank of the inferior temporal sulcus indicates the border of TE with TF. The section shown is from the comparative mammalian brain collection of the University of Wisconsin. Regions enclosed by brackets in A are shown
41
at higher power in B-D. For B-D, tissue is oriented so that the pial surface of cortex is at top, and layers of cortex are indicated at left. E is a photomicrograph of a coronal section through caudal inferior temporal cortex, stained with the Gallyas ('79) silver stain for myeloarchitecture. The large arrow in the inferior temporal sulcus (also shown in F) points to the border between TE and TF, and the smaller arrow in E points to the medial border of TF. F is a higher power view of the lateral portion of the section shown in E. Note the darker appearance of dorsal cortex between the arrows. Scale bar in A and E is 1 mm, and in B and F, it is 500 bm, C and D are at the same scale as B.
R.E. WELLER AND G.E. STEELE
42
Fig. 3. Rostrocaudal differences in the architectonic appearance of inferior temporal cortex of squirrel monkey. A and B are horizontal sections through the inferior temporal lobe of two different brains. Rostra1 (R) and lateral (L) are indicated. The section in A was stained with hematoxylin for myelinated axons, and the section in B was stained with thionin for cell bodies. A shows that caudal IT cortex is more myelinated than rostral cortex. The regions of cortex enclosed by brackets in B are shown at higher power in C (left bracket) and D (right
bracket). Layers of cortex are numbered in C and D after Brodmann ('09). In comparison to rostral IT (0,caudal IT (D) has a more darkly staining layer IV. The arrows in A and B point to the border between TE and TF. Scale bar in B is 1 mm. and A is at the same magnification. Scale bar in D is 500 km, and C is at the same magnification. Sections shown are from the comparative mammalian brain collection of the University of Wisconsin.
(sites 1-6), ventral (sites 7-14), and central (sites 15-18) groups.
numerous in its dorsal than ventral portion, while cells labeled in DLR were equally distributed in dorsal and ventral regions. In this and many other cases, dense labd surrounded the injection site in IT (e.g., Figs. 5-81, In the present case, many labeled cells were also found immediately caudal and dorsal to the injection, including in the ventral bank of the superior temporal sulcus and depths of the dorsal bank. By location, some of these cells may be in the fundal area of the superior temporal sulcus, FST, and possibly also the medial area of the superior temporal area, MST (see Fig. 1). Ventral ITc and central IT cortex displayed little label. A broad region of rostral IT cortex was strongly labeled, including much of the ventral bank of the superior temporal sulcus. Both banks of the rostral inferior temporal sulcus were labeled more sparsely, as was parahippocampal cortex medial to this sulcus. Two regions of prefrontal
Ipsilateral connections of the DLc-projection zone, ITc Dorsal injections. Six injections were made in the dorsal aspect of ITc (injection sites 1-6 of Fig. 4; see also site 15 of the central injections). In the first case, an injection of the retrograde tracer Fast Blue was made close to the superior temporal sulcus (injection site 1 of Fig. 4; case 89-7L; Fig. 5A-D; the results of a second injection made in the same case in ventral cortex are described later). For this and subsequent injections, the size of the core of the injection is given in Table 1. Cells were labeled over an extensive region of lateral extrastriate cortex estimated as rostral V 11, DLc, DLR, cortex dorsal to DLR, and perhaps M T (compare with Fig. 1).Cells labeled in DLc were more
CONNECTIONS OF IT
Fig. 4. Summary of the locations of the 25 injections of neuroanatomical tracers made in IT cortex in the present study. Architectonic subdivisions of the inferior temporal lobe are also shown (TE, TG, TF, TL, TH, PR, ER; see text). Lateral (A and B)and ventral (C) views of a standard left hemisphere are shown. The banks of the inferior temporal sulcus are opened in C . Individual injection site centers were placed on the summary in reference to the superior temporal sulcus, other limits of the inferior temporal lobe, and architectonic borders or their estimated locations. Note that injections 1-3 involved the dorsal region (D) of ITc that had a different laminar distribution of retrograde label following injections in DLc than the remainder of ITc, and injections 19-22 involved the questionable rostral (R) portion of ITc (Fig. 1)or the ITc-ITRborder region. See Table 1for details about the injections, and text for abbreviations.
cortex were labeled, one in dorsolateral cortex, lateral to the inferior arcuate sulcus; and the other in more lateral cortex, close to the lateral apex of cortex. In a second dorsal case (89-91, a larger injection of the anterograde tracer 3H-proline-leucine was made in a location similar to that of the previous case (site 2 of Fig. 4;Fig. 6). The injection strongly labeled the regions of dorsal DLc and DLR. Sparser label was in lateral DLc; cortex dorsal to DLR, approaching the Sylvian fissure; and caudal MT, or
43
the portion thought to represent central vision, based on connections with V I and V I1 (Tigges et al., '74, '81; Krubitzer and Kaas, '90). As in case 89-72, (Fig. 5 ) , much of ventral ITc was unlabeled, except for scattered foci close t o and within the inferior temporal sulcus. In the superior temporal sulcus, patches of label were scattered along much of the ventral bank. Rostroventral IT cortex, including the lateral bank of the inferior temporal sulcus, was densely labeled. There was a small focus of label in dorsolateral prefrontal cortex. A large injection of WGA-HRP was made in dorsal ITc in a third case, 89-15 (site 3 of Fig. 4; Fig. 7). The injection extended 0.6 mm into the white matter, including a 0.2 mm incursion of the syringe track. Dense label occupied the region of lateral to dorsal DLc, most of DLR, and cortex dorsorostral to the latter areas, extending to the Sylvian fissure. Dorsocaudal MT was weakly labeled, and lateroventral V I1 contained a few scattered labeled cells. Cortex dorsal and rostral to the injection was heavily labeled, including cortex in the inferior and superior temporal sulci. In the latter sulcus, cortex labeled included much of the ventral bank and foci in the dorsal bank. The rostral parahippocampal gyrus contained primarily retrograde label. In prefrontal cortex, labeled cortex extended from dorsolateral to lateral orbital. An occasional cell was labeled in entorhinal cortex, the superior temporal gyrus, and rostral cingulate cortex. While the pattern of label observed in this case was basically similar to that in the previous cases, the injection in 89-15 produced much more label, some of which was in areas unlabeled in the other cases (i.e., lateral orbital cortex, cortex dorsal to DLc and DLR, and entorhinal cortex). This additional label may be due to more effective transport from the larger injection or the syringe track's slight encroachment into the white matter and transport of WGA-HRP by cut fibers of passage (Mesulam, '82). The next three injections were placed in dorsocentral rather than dorsal ITc. A small injection of the retrograde tracer Fast Blue was made in an anterior location in case 90-8L (site 4 of Fig. 4;Fig. 8). Unlike results observed in the previous cases, cells labeled in DLc and DLRin 90-8L were primarily in lateroventral rather than dorsal cortex. Fewer cells were labeled in MT and the ventral bank of the caudal Sylvian fissure. Many cells were labeled in rostral and caudal IT cortex, including in the superior and inferior temporal sulci. An extensive region of the parahippocampal gyrus was strongly labeled. In prefrontal cortex, foci of labeled cells were in dorsolateral, lateral, lateral orbital, and, quite sparsely, medial orbital cortex. A few cells were labeled in rostral cingulate cortex. Another small injection of Diamidino Yellow was made in dorsocentral ITc in case 90-5 (site 5 of Fig. 4; not illustrated). Although this case produced considerably fewer labeled cells in comparison to previous cases, this could be related to the limited size of the injection. Cells labeled in DLc were in rostroventral DLc, if in DLc at all, rather than cortex anterior to DLc. Ventral DLR contained a small cluster of cells, and ventrorostral MT, a single cell. Besides IT cortex labeled in the vicinity of the injection, including the ventral bank of the superior temporal sulcus, label in IT was limited to a small number of cells in rostral IT and the lateral bank of the inferior temporal sulcus. A few cells were labeled in lateral prefrontal cortex. A final dorsocentral injection of WGA-HRP (case 89-19, site 6 of Fig. 4;not illustrated) was quite caudal. Although,
TS
a9 -
Fig. 5. Neurons retrogradely labeled by an injection of Fast Blue in dorsocaudal IT cortex (A-D) in comparison to those labeled by an injection of Diamidino Yellow in ventrocaudal IT cortex (E-H) of the same animal (case 89-7L, injection sites 1and 7 of Fig. 4). A and E are lateral views and C and H are ventral views of the left hemisphere. The banks of the superior temporal sulcus (STS) in the regions enclosed by rectangles in A and E are opened in B and F, and the banks of the inferior temporal sulcus (ITS) enclosed by rectangles in C and H are opened in D and G. Injection centers are black and halos are shaded. Dots indxate labeled cells, with the number
V
Fast Blue
Diarnidino Yellow
2,:.*
+
- -..c . .. .. ...,F..:
.a.
2mm
$5
-
::... .
G
ITS
of dots representing the relative density of labeling, except where isolated dots appear, indicating individual cells. The star in A indicates where the injection site of Diamidino Yellow obscured detection of cells that might have been labeled by Fast Blue; the reverse is shown in E by the star for the Fast Blue injection site. Areal borders indicated by solid lines were architectonically determined in the present case, while borders represented with dashed lines were estimated. See list in text for abbreviations.
7~
M
F
H M M
r/)
$ U
?3
tF M
r:M
4
4
45
CONNECTIONS OF IT 3H
89 - 9 - Proline - Leucine
ITS
Fig. 6. Projections observed as the result of an injection of 3H-proline-leucine in dorsocaudal IT cortex in case 89-9 (injection site 2' of Fig. 4). A is a lateral view and C is a ventral view of the left hemisphere. The banks of the STS are opened in B, and those of the ITS are opened in D. Dots indicate the density of autoradiographic silver grains; other conventions as in previous figures.
by location, this injection could have involved DLR, the pattern of label it produced was largely unlike that described for DLR (i.e., it failed to label MT and the dorsal area; Cusick and Kaas, '88; Steele et al., '91a), but similar to that found for nearby IT injections in other cases (sites 5 and 18 of Fig. 4).Most label in DLc was rostroventral. The ventral bank of the superior temporal sulcus and separate central and rostral regions of lateral IT were labeled. The only label on the ventral surface of the hemisphere was a small patch of anterograde label in the lateral bank of the inferior temporal sulcus several millimeters rostral to V 11. Label in prefrontal cortex was limited to dorsolateral cortex. To briefly compare the results of the six dorsal injections, the most dorsal of the injections (sites 1-3) produced a somewhat different pattern of label than that produced by the other, more dorsocentral injections (sites 4-6). The three dorsal injections strongly labeled DLc, DLR,superficial and deep portions of the ventral bank of the superior
temporal sulcus, and rostral IT; and more weakly labeled MT, V IT, cortex dorsorostral to DLR,the deep portion of the dorsal bank of the superior temporal sulcus, the inferior temporal sulcus, parahippocampal gyrus, inferior temporal pole, rostral cingulate cortex, and dorsolateral to lateral prefrontal cortex. The dorsocentral injections also labeled many of the same areas, but, unlike the dorsal injections, the dorsocentral injections labeled dorsolateral DLc only sparsely, but labeled rostroventral DLc or cortex just anterior to this portion of DLc more strongly; only weakly labeled DLRand the depths of the superior temporal sulcus; and did not label V I1 or cortex dorsal to DLR;but did label the caudal Sylvian fissure and lateral orbital cortex (Table 2).
Ventral injections. Eight injections were made in ventral ITc (injection sites 7-14 of Fig. 4).The pattew of label observed in case 89-7L, in which a small injection of the retrograde tracer Diamidino Yellow was made in ventral cortex (site 7 of Fig. 4; Fig. 5 ) , was representative of that
- V9M Sl - 6 8
d8H
CONNECTIONS OF IT
47
90
- 8L
Fast Blue
r
. ....... D . ,
Fig. 8. Cells labeled by an injection of Fast Blue in dorsocentral IT cortex in case 90-8L (site 4 of Fig. 4). A, lateral view; C, ventral view; B, the opened STS; and D, the opened ITS. Open arrows indicate where labeled cells were found in the Sylvian fissure; other conventions and abbreviations as in previous figures.
observed in most of the other cases. Although the ventral injection in 89-7L labeled cells throughout most of DLc, cells were densest in lateral-to-ventral DLc. Labeled cells in DLRwere few in number and scattered. Cells were labeled in rostrolateral V 11, or the part estimated to represent central vision (Cowey, '64). A few cells were labeled in the region of MT, in cortex dorsal to DLc, along the lateral apex of prefrontal cortex, and in rostral cingulate cortex. Most cells labeled in caudal IT were in the vicinity of the injection, except for cells in the ventral bank of the superior temporal sulcus and close to the inferior temporal sulcus. A moderate number of cells were labeled in rostral IT, including the rostral tip of the inferior temporal sulcus; and in parahippocampal cortex caudal to entorhinal cortex. The existence of injections in dorsal and ventral portions of ITc in case 89-7L permitted direct comparison of the resulting patterns of labeled cells. While both injections labeled similar regions and numbers of cells in the inferior temporal sulcus, parahippocampal gyrus, and rostroventral IT, they differed in labeling complementary portions of DLc
and in the extent of labeling they produced in the superior temporal sulcus (Fig. 5). Two other cases with injections of retrograde tracers in locations similar to that of the ventral injection in 89-7L (injection sites 8 and 9 of Fig. 4;not illustrated) labeled cells in locations similar to those labeled in 89-7L, with the exception of some additional label that could be due to the larger sizes of the injections or, more likely, the fact that syringe tracks in both cases involved 0.7 mm of underlying white matter. A large injection of WGA-HRP was made in caudoventral IT in case 90-8R (site 10 of Fig. 4; Fig. 9). Although this injection extended quite caudally, it appeared to spare DLc, based on its location relative to the borders of MT and V I. Lateral to ventrolateral V I1 was strongly labeled. Although much of DLRand DLc were labeled, lateral DLc was labeled most strongly, and dorsolateral DLc contained primarily retrograde label. Caudal and ventral portions of MT were labeled, and there were two small foci of label in the caudal Sylvian fissure. Label in rostral IT cortex was limited, and
R.E. WELLER
48 TABLE 2. Summary of Connections Between Different Parts of IT Cortex and Other Areas of Cortex in Squirrel Monkeys' Area injected Area of Cortex labeled
v I1 DL
