Organization of Intrinsic Connections in Owl Monkey Area MT

Transcription

1 Organization of Intrinsic Connections in Owl Monkey Area MT R. Malach, T. D. Schirman, M. Harel, R. B. H. Tootell 1 and D. Malonek Department of Neurobiology, Weizmann Institute of Science, Rehovot, Israel and 1 Nuclear Magnetic Resonance Center, Massachusetts General Hospital, th Street, Charlestown, MA 02129, USA Area MT (middle temporal) is a well-defined visual representation common to all primates, which shows a clear selectivity to the analysis of visual motion. In the present study we examined the architecture of the intrinsic connections in area MT in an attempt to reveal its organizing principles and its potential relationship to the functional domains in area MT. Intrinsic connections were studied by placing small injections of the tracer biocytin in area MT of seven adult owl monkeys (Aotus nancymae). The injections were targeted at well-defined orientation domains revealed using optical imaging of intrinsic signals. The distribution of axons labeled by these injections was related both to the cytochrome oxidase histochemistry and to the layout of functional domains in area MT and surrounding tissue. Tracer injections in the superficial layers of area MT produced a complex network of extrinsic and intrinsic axonal connections. Clear instances of extrinsic connections were observed between area MT proper and the MT crescent situated postero-medially to it. The intrinsic connections were laterally spread and organized in patch-like clusters with an average distance from injection center to the furthest patch of 1.8 ± 0.55 mm (± SD, n = 9). The overall axonal distribution tended to be anisotropic, i.e. the patches were distributed within an elongated ellipse [average anisotropy ratio: 1.86 ± 0.66 (± SD)] and were asymmetrically distributed about either side of the injection site [average asymmetry ratio: 2.3 ± 0.7 (± SD)]. Finally, there was a tendency for the intrinsic connections to connect to functional domains of similar orientation preference in area MT. However, this tendency varied substantially between individual cases. The highly specific nature of MT lateral connections puts clear constraints on models of surround influences in the receptive fields of MT neurons. Introduction A central feature of primate visual cortex is its subdivision into a set of functionally distinct cortical areas. One of the best characterized examples of these is area MT (middle temporal), which is situated posteriorly to the superior temporal sulcus (STS). Area MT (also termed V5) is one of the few cortical areas that have been recognized across all primate species including humans (e.g. Van Essen, 1979; Bixby et al., 1980; Krubitzer and Kaas, 1990; Tootell et al., 1985, 1995). Area MT contains a well-defined myeloarchitecture (Allman and Kass, 1971; Van Essen et al., 1981), it has a typical cytochrome oxidase (CO) staining pattern (Tootell et al., 1985) and a unique set of extrinsic cortico-cortical connections, including a direct projection to and from area V1 (Ungerleider and Desimone, 1986; Weller et al., 1984). Surrounding MT on the dorsal and caudal aspects there is a narrow crescent-shaped cortical region, which can be differentiated from MT by its distinct CO and myelin staining pattern (Tootell et al., 1985). This region has been termed the MT crescent (Kaas and Morel, 1993). Functionally, area MT has a complete representation of the contralateral hemifield, and shows a clear sensitivity to visual motion. Most neurons in area MT exhibit some preference to a specific direction of visual motion. Area MT contains a complex compartmental architecture. In owl monkey MT, neurons sensitive to global and local motion are segregated in a band/interband architecture (Born and Tootell, 1992). Within each compartment, neurons showing preference to a particular line orientation are clustered in a columnar fashion (Malonek et al., 1994). A further subdivision of these orientation (or axis of motion) columns into opposing directionality columns has been proposed based on singleneuron studies in the macaque (Albright et al., 1984) and optical imaging studies in owl monkeys (Malonek et al., 1994). One aspect of area MT that received relatively little attention is the organization of its intrinsic connectivity. In the owl monkey, Weller et al. (1984) have described the intrinsic connections as being widely distributed within MT, connecting disparate retinotopic points, and forming band-like structures. The fact that the intrinsic connections are not distributed evenly across MT suggests that they may be related to the underlying functional compartments. In the present study we explored this question by relating the organization of the intrinsic connections to underlying functional domains in owl monkey area MT revealed through optical imaging of intrinsic signals. Materials and Methods This study is based on experiments performed in seven owl monkeys (Aotus nancymae). Two of the monkeys were used for purely anatomical studies including tracer injections and CO histochemistry, while the other five were subjects in experiments that combined optical imaging of intrinsic signals with anatomy. The procedures for optical imaging, tracer injections and histology were described extensively in previous publications (Malach et al., 1993, 1994; Malonek et al., 1994). Consequently, only a brief description of the methodology will be provided here. Optical Imaging of Intrinsic Signals Animals were anesthetized with ketamine hydrochloride (10 20 mg/kg i.m.) followed by sodium penthotal (20 mg/kg i.v.). A headset was implanted over the anterior part of the skull to allow head immobilization with minimal trauma. Animals were then placed in the stereotactic apparatus using the headset attachment. A round opening was cut in the skull overlying area MT and the dura was carefully reflected. To reduce brain pulsation during the optical imaging stage, a chamber filled with inert solution and sealed with a transparent disk was placed over the opening and anchored to the skull with screws and dental cement. The animals were paralyzed with vecuronium bromide (Norcuron, 0.25 mg/kg/h) and were artificially respirated. The animals received anesthetics and paralytics throughout the experiment and their heart rate, electroencephalogram, end-tidal CO 2 and temperature were constantly monitored. Before visual stimulation the corneas were fitted with contact lenses, and accommodation paralyzed by topical application of atropine (1%). The eyes were focused on a tangent screen at a distance of cm. Receptive fields of V1 neurons were recorded through a small hole in the Cerebral Cortex Jun 1997;7: ; /97$4.00

2 skull. Guided by receptive field loci, the two eyes were converged using a Risley prism. Visual stimuli included mainly rectangular wave gratings of high contrast (0.5 c/deg and 0.2 duty cycle). A slow scan CCD camera attached to a special macro lens arrangement (Ratzlaff and Grinvald, 1991) was positioned over the chamber and recorded the intrinsic optical signals produced by the visual stimuli. Biocytin Injections Targeted at Optically Identified Functional Domains To study the connections between functional domains, we obtained on-line functional maps, and then placed injections of the tracer biocytin (King et al., 1989; Amir et al., 1993; Malach et al., 1993) (2% in Tris ph 7.6, 4 µa/4 min DC current) at the centers of clearly defined domains. Area MT was identified by its location on the postero-medial side of the superior-temporal sulcus (STS), and its borders (posterior, medial and lateral) were delimited from the functional maps: clearly defined orientation domains were observed within it and none in its immediate neighborhood (Malonek et al.,1994, and see Fig. 1 below). Injection pipettes (12 µm tip diameter ) were inserted using a four-axis micro-manipulator to a depth of mm into the cortex. The tracer was iontophoretically ejected (tip positive). Injection-site diameters varied from 150 to 350 µm. To avoid an overlap of biocytin patches corresponding to different injection sites, no more than three injections were placed in each mapped optical region. Following the injections, animals were allowed to survive for 8 12 h under constant monitoring of physiological states and levels of anesthesia. Histology After the survival period, animals were deeply anesthetized with Nembutal, received units of heparin IV, and were perfused transcardially with phosphate-buffered saline (PBS) followed by light fixation with 2.5% paraformaldehyde 0.5% glutaraldehyde in PBS containing 5% sucrose. The brains were removed, and the cortical tissue surrounding the STS was cut out, flattened and post-fixed in the same fixative as used for the perfusion but containing 15% sucrose. To maintain maximal correlation of the histological sections to the optically derived maps, no attempt was made to unfold the STS. After post-fixation, the flattened cortices were frozen and prepared for sectioning using the method of Tootell and Silverman (1985). The tissue was cut tangentially to the cortical surface into µm thick sections. Sections were processed for biocytin and for every third section for CO histochemistry. Staining for biocytin was performed using a modified avidin-biotin horseradish peroxidase (HRP) complex protocol including the following main steps. Sections were first rinsed thoroughly and incubated overnight in 0.5% Triton-X-100 (Sigma Chemical Co., MO, USA). The following day they were rinsed and incubated for 3 h in a solution prepared from Vector Labs Elite ABC Kit (Vectastain, ABC Kit, Vector Labs, Burlingame, CA, USA). Staining was done using diaminobenzidine (DAB) as the chromogen for 25 min. The procedure for CO staining was as described by Tootell et al. (1985). Anatomical Data Analysis The stained sections were viewed at low power with a Wild M5 macroscope equipped with light- and dark-field illumination, a drawing tube, a 35 mm camera and a color video camera attached to a computer. Neighboring brain sections were aligned by the numerous vascular landmarks and various marking holes. Higher-power viewing of the sections was obtained by a Universal Zeiss microscope equipped with a special three-dimensional computer mapping system which allowed efficient and accurate registration and quantification of labeled axons and cell bodies (Amir et al., 1993). In order to match the optically derived maps and the histological staining patterns, the cortical tissue was cut tangentially so as to maximally preserve the pattern of the superficial blood vessels. The same pattern of blood vessels was photographed during the optical recording session and was used to align the histological and optical maps (see Malach et al., 1994 for details). Deeper sections were aligned using the perpendicularly running blood vessels. For quantitative analysis of the data, the cortical section containing the most clearly visible patches of connections was chosen from each Figure 1. Correspondence between histochemically and functionally defined area MT. (A) Schematic drawing of the CO staining in area MT. In this figure, top is anterior, and lateral is to the right. The dense CO staining of area MT is indicated by light gray in this drawing, while the CO tongues in the MT crescent are indicated by dark gray. The MT border on its postero-medial side is indicated by a dotted line. The antero-lateral border was difficult to assign in our material. The area which was optically imaged to reveal intrinsic signals is indicated by the four corners. Scale bar = 1 mm. (B) Functional map of area MT. This is a differential orientation map showing cortical domains preferentially activated by vertically oriented gratings in black, and domains preferentially activated by horizontal gratings in white. The histochemically defined border of MT from A is depicted by a dotted line. Note that the orientation-selective domains are fully confined within the architectonically defined MT border. hemisphere. The borders of the axonal patches as well as the injection site were estimated, subjectively, as regions showing a clear increase in axonal density. The patch borders were drawn using with a drawing tube. Patches were qualitatively categorized in two classes according to their axonal density. The anatomical borders were superimposed on the functional maps using a computer graphics program (Adobe Photoshop 3.0). The distribution of functionally defined zones at the injection sites and within the patches of connections was quantitatively analyzed using image-processing software (NIH-image 1.59). The data derived from the different injection sites was then summarized and normalized (Malach et al., 1993 and see further details below). Results Relationship of Functional Domains to MT Borders Defined by Histochemistry We compared the CO staining patterns of the tangential sections to the optically derived functional maps of area MT. In agreement with previous reports (Tootell et al., 1985; Kaas and Morel, 1993) area MT appears in CO staining as an oval of dense Cerebral Cortex Jun 1997, V 7 N 4 387

3 Figure 2. Intrinsic horizontal connections in area MT. This is a dark-field photo-montage of a tangential section through layer 2 3. Injection site is indicated by the black arrow. The axonal distribution shows a clear patchy organization. Patches intrinsic to MT, shown at higher power in Figure 3A, are indicated by filled arrowheads. Extrinsic patches located in the MT crescent, shown at higher power in Figure 3B, are indicated by empty arrowheads, Note the anisotropy and asymmetry of the axonal projections. In this figure top is medial and anterior is to the right. Scale bar = 250 µm. staining bordering the medial end of the STS. This dense oval is surrounded on its postero-medial aspect by the MT crescent (Tootell et al., 1985; Kaas and Morel, 1993) (MTc), a region of lower CO density containing a series of CO-dense patches. A schematic drawing of this CO pattern is shown in Figure 1A to illustrate the relationship of MT proper and the MTc. The oval 388 Optical Imaging and Connections of Owl Monkey Area MT Malach et al.

4 Figure 4. Schematic drawing of the patches shown in Figure 2 superimposed on CO-defined area MT. Injection site is delineated by a yellow contour. Densely innervated patches are indicated by a red contour. Weakly innervated patches are indicted by a green contour. Note the clear patchy projections to the MTc. Same orientation as in Figure 1. Scale bar = 1 mm. borders. Note that several of the patches are clearly located inside the MTc. At least some of these connections could be seen to traverse directly, through the gray matter, from area MT to the MTc. The patches in MTc, unlike those in MT, could be found throughout the cortical depth, suggesting that the two areas are linked by hierarchically lateral type of connections (Felleman and Van Essen, 1991). Figure 3. Higher power view of selected patches shown in Figure 2. Patches shown in (A) are enlargements of intrinsically connected patches indicated by filled arrowheads in Figure 2. (B) Enlargements of patches in the MTc indicated by empty arrowheads in Figure 2. shape of area MT can be clearly seen as well as the tongues of labeling protruding into the MTc. The border of MT around the STS is ill-defined due to the cortical folding in this region. When the CO-defined MT border was superimposed on the functional maps obtained from optical imaging, there was a good match between the area showing high orientation selectivity to moving gratings and the histochemically defined MT. This is shown in Figure 1B. Note that the orientationselective patches are strictly confined to MT, while the MTc and the cortex further posteriorly do not show such clear selectivity. Connectivity Patterns in Area MT Tracer injections in superficial layers of area MT produced a large and complex network of horizontal connections. An example of such a network is shown in Figure 2. Around the injection site there was a massive and fairly uniform halo of emanating fibers. At further distance, within area MT, the fibers tended to form axonal clusters or patches (e.g. filled arrowheads, Figs 2 and 3A). These patches were largely confined to the superficial layers. Further away, axonal clusters could be clearly seen within the MTc (empty arrowheads Figs 2 and 3B) thus indicating a direct link between the two areas. The connections to the MTc were seen unambiguously in five of the injections studied, and an example is shown in Figure 4 which depicts the distribution of the axonal patches relative to the MT and MTc Global Asymmetries Close inspection of the lateral connections within MT across different injection sites revealed two additional characteristics of the connectional pattern. First, the global distribution of axons was typified by major anisotropies. The connections were not distributed evenly around the injection sites, but rather tended to spread more along certain preferred axes. It was interesting to determine whether small and large injection sites label connections with different degrees of anisotropy (see Discussion). Figure 5A and 5B show examples of a small and large MT injection, respectively. It is clear that although the injection in Figure 5B was 3-fold larger than that in Figure 5A its connectional distribution remained highly anisotropic. These results suggest that the connectional anisotropy is, at least partially, invariant to the local properties of the injected column, but rather reflects a more global characteristic of area MT. To see how the anisotropies were distributed across area MT we estimated the major axis for each injection by applying a Gaussian filter to schematic pictures of the clusters of patches. This blurred the ensemble of individual patches into one elliptical cloud whose major axis was determined by the presence of the majority of patches along it. This axis was taken as the anisotropy axis of the connectional pattern. The distribution of the entire set of MT injections and their corresponding anisotropies is depicted in Figure 6. In this figure arrows represent the anisotropy axis for each injection point as determined above. The length of the arrow corresponds to the distance between the two furthermost patches situated along the anisotropy axis. For the nine injections studied, the average distance was found to be 1.8 ± 0.55 mm (± SD). To estimate the level of anisotropy, we divided the distance between the two furthest patches along the anisotropy axis, by the distance Cerebral Cortex Jun 1997, V 7 N 4 389

5 Figure 6. Anisotropy of intrinsic connections across area MT. This is a summary scheme showing the anisotropy axis for nine of the injections studied. Each injection site and its size are indicated by the colored circles, while the colored arrows show the axis of anisotropy for each injection. The length of the arrows indicates the size of the cloud of patches produced by each injection. The numbers associated with each injection indicate the anisotropy ratio, i.e. how elongated is the cloud of patches produced by each injection (see text for details). Dotted and dashed lines show the estimated vertical and horizontal meridian representation, while the red star and gray lines show the estimated location of the foveal representation and iso-eccentricity lines respectively (Allman and Kaas, 1976). Same orientation as in Figure 1. Scale = 1 mm. Figure 5. Anisotropy of intrinsic connections. Two dark-field, low-power macrophotographs of a small (A) and a large (B) injection in MT. The center of injection is indicated by a curved arrow. Note the clear anisotropy of connections that exists regardless of injection size. Same orientation as in Figure 1. Scale bar = 0.5 mm. between the furthest patches along the orthogonal axis. This ratio is indicated by the numbers near each injection in Figure 6. The average anisotropy ratio across the injections was 1.86 ± 0.66 (± SD). The circle and ring near each arrow indicate the size of the injection site and the halo respectively. From the figure, it appears that there was a trend for injections close to the vertical meridian representation to have an anisotropy axis which parallels the iso-eccentricity lines, while injections at more peripheral locations appear to run more perpendicularly to the iso-eccentricity lines. Another type of asymmetry observed was a tendency for the patches to lie more to one side of the injection site than the other. We estimated this trend by drawing an imaginary line through each injection site so as to bisect the cluster of patches at the most unbalanced position. We then counted the number of patches on either side of this imaginary line, and divided the two numbers to obtain an asymmetry ratio. The results of such calculations on 11 injections gave an averaged asymmetry ratio of 2.3 ± 0.7 (± SD), indicating that at the maximal imbalance, the number of patches on one side of the injection site was more than twice that on the other side. Injection Size and Patchy Connections In a previous study of lateral connections in macaque visual cortex we have found that the size of axonal patches tended to increase somewhat with increased size of injection, but that beyond a certain patch limit there was a tendency for the generation of new patches rather than further increase in patch size (Amir et al., 1993). To see if an upper limit on patch size was also a feature of area MT, we analyzed the relationship between injection size and the lateral connections. Figure 7A shows a plot of the relationship between injection size and the combined area of all the patches emanating from the injection. Despite the substantial variability among injection sites, there was a clear positive correlation between injection size and the total cortical area targeted by the lateral connections. Thus, there was an 6-fold increase in the total patch area from the smallest to the largest injection sites. In contrast, average patch width increased only moderately, and with injection sites >0.07 mm 2 it leveled off and remained in the narrow range of µm. Relationship of Patchy Connections to Functional Architecture The organization of the lateral connections in a patchy manner suggests that they are correlated in some fashion to the functional architecture in area MT. To explore this relationship we superimposed the connectional patterns on the optically derived orientation maps obtained from the same experiments. An example of such a relationship is given in Figure 8 which depicts, at low (A) and high (B) power the distribution of axonal patches relative to a functional map showing activation by vertically oriented moving gratings (black regions) versus horizontally oriented gratings (white regions). Note that the injection site was targeted in this case onto a region selective for a close to horizontal (white) orientation domain, and that the patches show a tendency to avoid the orthogonal (black) orientation domains. However, as can be observed in this figure, 390 Optical Imaging and Connections of Owl Monkey Area MT Malach et al.

6 Figure 7. Relationship between injection size and patch dimensions. The graph in (A) plots the relationship between injection site area and total patch area for all injections studied. There was a clear increase in total cortical area connected by axonal patches as the injection site grew larger. (B) The relationship of average patch width (the minor axis in each patch) to injection size. Here it appears that patch width levels off at µm. this was far from a strict rule, and many exceptions could be found where patches occupy domains of dissimilar orientation. To obtain a quantitative estimate of this tendency, the correlation between the distribution of orientation preference at each injection site and its corresponding patches was calculated from high-resolution orientation angle maps (Blasdel and Salama, 1986; Grinvald et al., 1986) of the injected area. The global orientation distribution in the entire MT region was also calculated. The latter distribution is expected if the axonal patches were randomly located throughout area MT and thus served as a control. Each histogram was normalized so that the total area under it was equal to 100. The global-mt orientation distribution was then subtracted from the injection site and patch distributions. This resulted in histograms whose total area equaled zero. In these histograms, orientation domains that are connected to the injection site area are given positive values, and orientation domains lacking connections to the injection site, are given negative values. To compare across injection sites the orientation at the center of each injection site was set to a value of zero. With this normalization, distance along the x-axis reflects the extent of deviation of the orientation found at the Figure 8. Relationship of axonal patches to orientation domains in area MT. The axonal patches were superimposed on a differential orientation map obtained from the same patch of cortex. (A) Superposition of an injection onto a differential vertical (black) and horizontal (white) orientation map. (B) An enlargement of the same picture shown in (A). This was an injection targeted close to the horizontal (white) orientation domain. Note the tendency of the axonal patches to avoid the vertical (dark) orientation domains. Scale bar = 1 mm. axonal patches from the orientation preference found at the center of the injection site. The results of this analysis for 10 injections are shown in Figure 9. Figure 9A shows the distribution of orientation preferences within the injection sites and indicates that they were confined to a range of ±20. The distribution of the patches shows that there was indeed a trend for orientation domains of similar orientation preference to be connected since all positive values in this histogram clustered around zero orientation deviation. However, this trend was quite weak and variable as indicated by the relatively large error bars in Figure 9B. Discussion Connections to the MT Crescent Our results indicate clear connectivity between area MT and the MTc. Previously the MTc has been shown to be associated with area FSTv and the temporal stream (Kaas and Morel, 1993). Given this, the MT-to-MTc connectivity should be taken as another link between the temporal and parietal streams, similar to the connection between area DL and MT proper. The fact that Cerebral Cortex Jun 1997, V 7 N 4 391

7 Figure 9. Relationship of axonal patches to orientation domains in area MT overall distribution. This is a summary histogram from 10 injections. The horizontal axis shows the deviation (measured in degrees) from the preferred orientation at the center of the injection site. The vertical axis shows the percentage area, relative to control, at a particular orientation selectivity deviation from the center of the injection site. Positive and negative values indicate axonal connectivity above and below, respectively, that expected from a random distribution. (A) shows the spread of orientation preferences within the injection sites; (B) shows the spread of orientation preferences among the axonal patches. The injection site histogram is narrowly confined around zero, indicating that each injection site contained a narrow range of orientations. The connected patches also show positive values around zero, indicating a tendency for similar orientation domains to be connected. However, the small amplitude of the distribution in (B), and the large error-bars indicate a highly variable trend. the MT MTc connections are of the hierarchically lateral type suggests that the two areas are likely to be on a similar hierarchical level (Felleman and Van Essen, 1991). It will be interesting to see if this is indeed confirmed by connections of the MTc to other cortical areas. Asymmetry and Anisotropy of Connections One of the striking features of MT intrinsic connections is their tendency to spread further along preferred axes and to lie preferentially on one side of the injection site. In macaque V1 it has been found that lateral connections tend to elongate in a direction perpendicular to borders of ocular dominance columns a phenomenon most likely related to the global retinotopic anisotropy imposed by the interdigitating OD columns (Blasdel et al., 1992; Malach et al., 1993; Yoshioka et al., 1996). Even more interesting is the recent finding of a striking local anisotropy of lateral connections in area V1 of the tree shrew, in which the axis of elongation of the lateral connections was directly related to the orientation preference of the projecting neurons (Fitzpatrick et al., 1994). These results raise the possibility that the connectional asymmetries and anisotropies observed in MT are also related somehow to the functional organization of area MT. If the source of the asymmetries is local, one would expect the connections to be more asymmetric for small injections that label small cortical sites, as opposed to large injections which trace the connections from regions which are more likely to include a full range of local cortical functions. The fact that in our study large injections remained anisotropic even though they trace the connections from regions that are likely to include a full range of local functional modules, indirectly suggests that the asymmetries relate, at least partially, to a more global functional attribute within area MT such as retinotopic location. The functional implications of these anisotropies and asymmetries are as yet unclear, but a simple explanation might be magnification factor distortions similar to those encountered in macaque V1. One of the intriguing questions related to the anisotropy issue is the question whether the mapping of the lateral connections is continuous or not, i.e. whether two sets of patches emanating from two nearby cortical sites are overlapping. It was argued previously that discontinuous mapping maximizes the connectional diversity of cortical neurons (Malach, 1994). One specific prediction of the maximal diversity hypothesis is that as injection size increases, patch width should increase up to a limit and with further increases in injection size, additional separate patches will be labeled. It was further proposed that the upper limit on patch width should be correlated to the average dendritic spread of the neurons targeted by the axonal patches. This relation was in fact observed experimentally (Lund et al., 1993). It should be noted that our data indeed suggest that in area MT patch width has an upper limit of µm (Figure 7). It will be interesting to see if the average dendritic spread of layer 2 3 neurons in MT in fact corresponds to this value. Overall, the clear clustering into axonal patches observed in MT and the MTc is consistent with numerous observations of clustering into patches or stripes that by now appear to span the entire primate cortex from V1 to prefrontal frontal cortex (Levitt et al., 1993; Kritzer and Goldman-Rakic, 1995). Relationship to Functional Domains Quantitative analysis of the relationship between the connectivity patterns in area MT and orientation domains showed a trend for orientation domains of similar orientation preference to be connected. This trend appears to be a common feature across species and cortical areas. Thus, it was shown in cat V1 (Gilbert and Wiesel, 1989) and macaque V1 both between orientation domains (Blasdel et al., 1992; Malach et al., 1993), same eye ocular dominance columns (Yoshioka et al., 1992; Malach et al., 1993), and CO-blobs (Livingstone and Hubel, 1984; Yoshioka et al., 1996). In V2 of the squirrel monkey there was again a similar, albeit quite weak, trend (Malach et al., 1994). The present results show that in owl monkey area MT as well, the tendency of similar orientation domains to be connected was quite weak and variable (see Figure 9). Our data are too limited at present to reveal the source of this variability, which may be due to a true diversity in the targets of axonal connections. Alternatively, the variability may reflect a heterogeneity in the population of the injection sites which may 392 Optical Imaging and Connections of Owl Monkey Area MT Malach et al.

8 belong to different subcompartments within MT such as the band inter-band architecture (Born and Tootell, 1992), or even to different direction of motion patches. An intriguing question is whether the organization of the patches is also correlated to the underlying direction of motion subdomains. Unfortunately, this functional organization was more difficult to detect using optical imaging, and consequently we could not obtain sufficient data to answer this question. Viewed in relation to other studies of lateral connections, it does appear that as one moves away from area V1 to higher-order areas, the precision with which lateral connections relate to simple feature selectivities of cortical neurons, such as orientation selectivity, is reduced. This may be an indication that the computations carried out by the lateral connections in high-order areas involve more complex attributes of the visual scene. The central theme emerging from the present study is that lateral connections in area MT, far from being randomly distributed, show both a global and local organizational specificity. Considering that area MT contains a well-organized retinotopic map (Allman and Kaas, 1976; Sereno et al., 1994; Shmuel et al., 1996), these highly specific patterns of connections ref lect clear constraints on the types of relationships between different visual field locations that may be processed by MT neurons. Notes We thank Dr A. Grinvald for making available his optical imaging set up and T. Kenet for going over the manuscript. This research was supported by US Israel BSF grant to R.M and R.T., the ISF, Mijan de Salonique and Wolfson foundations to A.G, and Minerva Foundation grant 8322 to R.M. and A.G. Correspondence should be addressed to Rafael Malach, Brain Research, Weizmann Institute, Rehovot, Israel References Albright TD, Desimone R, Gross CG (1984) Columnar organization of directionally selective cells in visual area MT of the macaque. J Neurophysiol 51: Allman JM, Kass JH (1971) A representation of the visual field in the caudal third of the middle tempral gyrus of the owl monkey (Aotus trivirgatus). Brain Res 31: Allman JM, Kaas JH (1976) Representation of the visual field on the medical wall of occipital- parietal cortex in the owl monkey. Science 191: Amir Y, Harel M, Malach R (1993) Cortical heirarchy reflected in the organization of intrinsic connections in Macaque monkey visual cortex. J Comp Neurol 334: Bixby JL, Maunsell JH, Van Essen DC (1980) Effects of motor unit size on innervation patterns in neonatal mammals. Exp Neurol 70: Blasdel GG, Salama G (1986) Voltage-sensitive dyes reveal a modular organization in monkey striate cortex. Nature 321: Blasdel GG, Yoshioka T, Levitt JB, Lund JS (1992) Correlation between patterns of lateral connectivity and patterns of orientation preference in monkey striate cortex. Soc Neurosci Abstr 169.4:(Abstract) Born RT, Tootell RBH (1992) Segregation of global and local motion processing in primate middle temporal visual area. Nature 357: Felleman DJ, Van Essen DC (1991) Distributed hierarchical processing in the primate cerebral cortex. Cereb Cortex 1:1 47. Fitzpatrick D, Schofield BR, Strote J (1994) Spatial organization and connections of ISO-orientation domains in tree shrew striate cortex. Soc Neurosci Abstr 20:349.4(Abstract) Gilbert CD, Wiesel TN (1989) Columnar specificity of intrinsic horizontal and corticocortical connections in cat visual cortex. J Neurosci 9: Grinvald A, Lieke E, Frostig RD, Gilbert CD, Wiesel TN (1986) Functional architecture of cortex revealed by optical imaging of intrinsic signals. Nature 324: Kaas JH, Morel A (1993) Connections of visual areas of the upper temporal lobe of owl monkeys: the MT crescent and dorsal and ventral subdivisions of FST. J Neurosci 13: King MA, Lois PM, Hunter BE, Walker DW (1989) Biocytin: a versatile anterograde neuroanatomical tract-tracing alternative. Brain Res 497: Kritzer MF, Goldman-Rakic PS (1995) Intrinsic circuit organization of the major layers and sublayers of the dorsolateral prefrontal cortex in the rhesus monkey. J Comp Neurol 359: Krubitzer L, Kaas JH (1990) Cortical connections of MT in four species of primates: areal, modular and retinotopic patterns. Vis Neurosci 5: Levitt JB, Lewis DA, Yoshioka T, Lund JS (1993) Topography of pyramidal neuron intrinsic connections in macaque monkey prefrontal cortex (areas 9 and 46). J Comp Neurol 338: Livingstone MS, Hubel DH (1984) Specificity of intrinsic connections in primate primary visual cortex. J Neurosci 4: Lund JS, Yoshioka T, Levitt JB (1993) Comparison of intrinsic connectivity in different areas of macaque monkey cerebral cortex. Cereb Cortex 3: Malach R (1994) Cortical columns as devices for maximizing neuronal diversity. Trends Neurosci 17: Malach R, Amir Y, Harel M, Grinvald A (1993) Relationship between intrinsic connections and functional architecture revealed by optical imaging and in vivo targeted biocytin injections in primate striate cortex. Proc Natl Acad Sci USA 90: Malach R, Tootell RB, Malonek D (1994) Relationship between orientation domains, cytochrome oxidase stripes, and intrinsic horizontal connections in squirrel monkey area V2. Cereb Cortex 4: Malonek D, Tootell RB, Grinvald A (1994) Optical imaging reveals the functional architecture of neurons processing shape and motion in owl monkey area MT. Proc R Soc Lond B Biol Sci 258: Ratzlaff EH, Grinvald A (1991) A tandem-lens epifluorescence macroscope: hundred-fold brightness advantage for wide-field imaging. J Neurosci Methods 36: Sereno MI, McDonald CT, Allman JM (1994) Analysis of retinotopic maps in extrastriate cortex. Cereb Cortex 4: Shmuel A, Schirman TD, Harel M, Malonek D, Grinvald A, Malach R (1996) Optical cartography: simultaneous retinotopic and functional mapping of several areas in primate visual cortex using optical imaging. Soc Neurosci Abstr 22:760.11(Abstract) Tootell RB, Silverman MS (1985) Two methods for flat-mounting cortical tissue. J Neurosci Methods 15: Tootell RB, Hamilton SL, Silverman MS (1985) Topography of cytochrome oxidase activity in owl monkey cortex. J Neurosci 5: Tootell RB, Reppas JB, Kwong KK, Malach R, Born RT, Brady TJ, Rosen BR, Belliveau JW (1995) Functional analysis of human MT and related visual cortical areas using magnetic resonance imaging. J Neurosci 15: Ungerleider LG, Desimone R (1986) Cortical connections of visual area MT in the macaque. J Comp Neurol 248: Van Essen DC (1979) Visual areas of the mammalian cerebral cortex. Annu Rev Neurosci 2: Van Essen DC, Maunsell JH, Bixby JL (1981) The middle temporal visual area in the macaque: myeloarchitecture, connections, functional properties and topographic organization. J Comp Neurol 199: Weller RE, Wall JT, Kaas JH (1984) Cortical connections of the middle temporal visual area (MT) and the superior temporal cortex in owl monkeys. J Comp Neurol 228: Yoshioka T, Blasdel GG, Levitt JB, Lund JS (1992) Patterns of lateral connections in macaque visual area V1 revealed by biocytin histocvhemistry and functional imaging. Soc Neurosci Abstr 18:134.13(Abstract) Yoshioka T, Blasdel GG, Levitt JB, Lund JS (1996) Relation between patterns of intrinsic lateral connectivity, ocular dominanace, and cytochrome oxidase reactive regions in macaque monkey striate cortex. Cereb Cortex 6: Cerebral Cortex Jun 1997, V 7 N 4 393