| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Departments of Psychology, 3Cell and Developmental Biology, and 4Ophthalmology and Visual Sciences, Vanderbilt University, Nashville, Tennessee; and 2Departments of Neurobiology and 5Community and Family Medicine, Duke University, Durham, North Carolina
Submitted 4 April 2005; accepted in final form 4 June 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Blasdel and Salama 1986; Bonhoeffer et al. 1995; Hubel and Wiesel 1977; Landisman and Ts'o 2002; LeVay and Nelson 1991). In addition to columns related to these stimulus features, staining visual cortex for the activity of the metabolic enzyme cytochrome oxidase (CO) reveals an additional columnar structure: CO blobs or patches (Horton 1984; Horton and Hubel 1981; Livingstone and Hubel 1984). Thus, viewed from the cortical surface, the superficial layers of V1 contain at least four distinct but overlapping columnar systems: a map of visual space, a map of orientation preference, a map of ocular dominance, and a mosaic of patches that are alternately rich and poor in CO.
Specific relationships between these columnar systems have been observed in some cases. The most striking map relationships have been observed in macaque V1, where, for example, singularities in the orientation preference map tend to lie near the center of ocular dominance domains and iso-orientation contours tend to run orthogonal to ocular dominance boundaries (Bartfield and Grinvald 1992; Blasdel and Salama 1986; Obermayer and Blasdel 1993). In addition, CO blobs are found aligned over the centers of ocular dominance bands (Bartfeld and Grinvald 1992; Blasdel and Salama 1986; Horton and Hubel 1981). Although not as strong, similar relationships among ocular dominance domains, orientation preference maps, and cytochrome oxidase blobs have been reported in the cat area 17/18 (Crair et al. 1997a,b; Hübener et al. 1997; Kim et al. 1999; Löwel et al. 1998; Murphy et al. 1995). A relationship between compartments defined by CO and columnar organization for orientation selectivity has also been observed in V2 of the macaque (Livingstone and Hubel 1984; Roe and Ts'o 1995; Ts'o et al. 1990), squirrel monkey (Malach et al. 1994), and owl monkey (Xu et al. 2004a).
Given the striking relationships among columnar systems found in both macaque V1 and cat area 17, and the evolutionary distance between the primate and carnivore lines of descent, one is tempted to conclude that systematic relationships among maps are a general feature of cortical organization and of fundamental importance for the representation of multiple features within the same functional area. Indeed, it has been argued that systematic map relationships may be important for achieving uniform coverage, which is ensuring that all combinations of stimulus features are represented uniformly for all regions of visual space (Swindale 2000; Swindale et al. 2000). There are, however, reasons to question the idea that map relationships are common to all species that exhibit these columnar systems and thus are critical for achieving uniform coverage. For example, in another carnivore, the ferret, little evidence was found for a relationship between the map of orientation preference and the map of ocular dominance (White et al. 2001). Similarly, in a different primate, the squirrel monkey, CO domains do not adhere strictly to the center of ocular dominance domains as they do in the macaque (Horton and Hocking 1996). Currently, however, no study has provided quantitative evidence on the structure and relationships of all four of these columnar systems in any primate other than the macaque.
The goal of the present study was to examine the map of visual space, the map of orientation preference, the map of ocular dominance, and CO staining in another primate, the prosimian bush baby. We found robust spatial organization for each of these columnar systems in V1 but little evidence for strong relationships among maps. Our results suggest that precise relationships among stimulus feature maps in V1 are not a general feature of cortical organization in primates and that the relationships observed in the cat and the macaque may have evolved independently. Moreover, assuming the properties of extant prosimians can be taken to resemble those of early ancestral primates (Purvis 1995), map relationships appear to be a more recently acquired trait, one that is not essential for either the development or proper function of circuits in V1. Some of the results reported here were presented previously in abstract form (Bosking et al. 1996; Emeric et al. 2003).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Twenty bush babies of both sexes were handled according to the approved protocols from the Institutional Animal Care and Use Committees at Vanderbilt University and Duke University. Twelve animals were used for experiments at Duke University, and the other eight were used at Vanderbilt University. In these animals, V1 was successfully imaged in 16 hemispheres; V2 was imaged in 11 hemispheres.
Additional data from two owl monkeys imaged as part of a previous study (Xu et al. 2004a) were used to compare the organization of V2 in bush babies and owl monkeys. The owl monkeys and bush babies were imaged in the same setup under identical conditions at Vanderbilt.
General preparation
The general preparation differed slightly for experiments conducted at the two institutions. At Duke University, anesthesia was induced with a mixture of ketamine and xylazine and maintained at surgical levels with a 2:1 mixture of NO2 and O2 and 1–3% halothane. Animals were placed in a stereotaxic frame, intubated, paralyzed (intravenous pancuronium bromide, 0.2 mg · kg–1 · h–1), and artificially ventilated with the gas mixture at a depth and rate that kept the expired CO2 near 4.0%. Skin and muscle were reflected, and a craniotomy was performed exposing parts of V1, V2, and adjacent visual areas. A stainless steel chamber was secured to the skull using dental acrylic, the dura was removed, and then the chamber was filled with saline and sealed with a glass top. Anesthesia was then lowered for optical imaging to a 1:1 mixture of NO2 and O2 and 0.5–1% halothane.
At Vanderbilt University, the animals were initially anesthetized with isofluorane (2–4% in O2) and, after tracheal intubation and implantation with two femoral catheters, were mounted in a stereotaxic apparatus. Neuromuscular blockade was initiated by intravenous injection of 0.7–1.0 mg/kg vecuronium bromide. Animals then were artificially ventilated with a 3:1 mixture of NO2 and O2 delivered at a rate sufficient to maintain the peak end tidal CO2 level at around 4%. Paralysis was maintained by intravenous infusion of vecuronium bromide (0.1–0.3 mg · kg–1 · h–1) mixed in 5% dextrose lactated Ringer. Propofol (2,6-di-isopropylphenol: 10 mg · kg–1 · h–1) was used for anesthesia during these experiments. An opening was made in the skull over visual cortex and was sealed with 1% agarose in saline under a cover glass.
Pupils were dilated with 1% atropine eye drops, and clear contact lenses with artificial pupils 3 mm in diameter were used to render the retina conjugate with the viewing screen 28.5 or 57 cm distant. During experiments the optic disks and areae centralii (ACs) were plotted on the screen using either a reversible ophthalmoscope or by back reflection of the retinal image from the surface of the tapetum using a fiber optic light source.
Optical imaging and visual stimuli
Intrinsic optical imaging signals were acquired with the differential video-enhancement imaging system and data-acquisition software from Optical Imaging (Mountainside, NJ). Similar imaging procedures have been described in detail previously (Bosking et al. 1997, 2002; Xu et al. 2004a,b).
Reference images of cortical vasculature were acquired with a 540-nm green light. The cortex was illuminated with a 611- or 689-nm light during data acquisition. We obtained consistent results or equivalent image quality with either a 611- or a 689-nm light, although 689 nm tended to reduce blood vessel artifacts in some cases. To map the visuotopic organization of V1, topographically limited horizontal or vertical grating stimuli were presented monocularly through the contralateral eye either within narrow rectangular windows [1–2° width x full screen height (54°) or full screen width (74°) x 1–2° height] or 2° circular patches (see Figs. 1 and 2, insets) at eccentricities ranging from 0 to 15°. High-contrast square wave gratings (fundamental spatial frequency: 0.5–1.0 cycle/°, drift velocity: 2 Hz, duty cycle: 20%) were moved back and forth within the windows at different locations. At each location mapped, stimuli consisted of two orthogonally oriented gratings. A single trial consisted of data acquisition during continued presentation of the drifting grating stimulus for 8 s and an interstimulus interval of 12 s using a blank screen of mean luminance (
30 cd/m2). Stimulus sets were repeated 10–20 times.
|
|
; Das and Gilbert 1995; Grinvald et al. 1994). To study the organization of orientation preference, high-contrast rectangular gratings (fundamental spatial frequency: 0.5 cycle/°, drift velocity: 2 Hz, duty cycle: 20%) of four orientations at intervals of 45° were displayed on the full screen and presented binocularly. In some blocks, all orientation conditions and a blank control were presented in a randomized order. On other blocks, only orthogonal pairs of orientations (e.g., 0/90°, 45/135°) were presented in separate trials. Stimulus sets were made up of 20–40 blocks.
To determine the map of ocular dominance, four or eight oriented gratings were presented while either the left or right eye was blocked using a hand-held occluder or electrical eye shutters. Data acquired during presentation of all orientations to the left eye were summed as were data acquired during presentation of all orientations to the right eye, and the two data sets were then subtracted from each other or divided by each other to form an ocular dominance map.
Video images were acquired at a rate of 30 frames/s, but all frames acquired for each condition during the 8-s period of stimulation were summed together for eight data frames before further analysis. Individual data frames were either 744 x 480 pixels, with a resolution of either 87 or 174 pixels/mm or 655 x 480 pixels with a resolution of 75 pixels/mm.
Visuotopic map analysis
To create visuotopic maps, all images associated with stimuli of the same orientation were summed and divided by the "pure blank" obtained by summing the images of the blank control to create single-condition maps using Winmix or TVMIX software (Optical Imaging), unless specified otherwise. The resulting maps were routinely "clipped" at 1–2 SD, smoothed by using a mean-filter kernel or Gaussian-filter kernel at a radius of 3–7 pixels and scaled in the range of 0–255 gray levels for appropriate display, using Matlab (MathWorks, Natick, MA). In addition, to quantify the activation pattern sizes in response to topographically restricted stimuli, the distinctive modular foci of activation on the resultant topographical map that contained pixels >1 SD from the image mean of mask-filtered maps (see following text) were detected by a custom Matlab program. We then outlined activation contours manually and measured across the extracted activation contour to get the width, orthogonal to the major axis of cortical activation.
Visual field maps were constructed based on a series of activation patterns produced by topographically limited horizontal and vertical grating stimuli in different locations of the visual field. The midlines of these activation patterns were plotted to create iso-azimuth or iso-elevation lines using the following procedure: 1) the low-pass filtered single condition image was thresholded to include only the top 20% darkest pixels to show the central points of each activation patch. 2) The same image was also used to create contours of the total activation zone by reducing the number of gray levels to 8 or 16 discrete levels. The latter (4th or 8th gray level) helped to determine the outer edges of the activation zone (see Blasdel and Campbell 2001). 3) Data obtained from the center points of activation (1) and the outermost contour of the total activation zone (2) were combined to determine the midline of the activated area in cortex. The intersections of the iso-azimuth and iso-elevation midlines were used to determine individual retinotopic points. Some of these points were also verified by examining the location of activation patterns produced by the 2° circles of stimulation presented at different eccentricities.
The cortical magnification factor (CMF) was measured based on the visual field maps and calculated as the average linear distance in cortex representing 1° in visual space (mm/°) at a particular eccentricity (Daniel and Whitteridge 1961; Van Essen et al. 1984). Specifically, the distance between two adjacent intersection points of iso-azimuth and iso-elevation lines on the topographical maps of cortex (in mm) was divided by the corresponding distance in degrees of visual space.
Orientation map analysis
Data acquired during presentation of one orientation were divided by or subtracted from data acquired during presentation of the orthogonal orientation to produce difference images using Winmix or TVMIX software. Custom programs written in National Institutes of Health Image (as an extension to the public domain National Institutes of Health Image program, at http://rsb.info.nih.gov/nih-image/) and Matlab were used to further process the data. In most cases, to reduce vascular artifacts, we used a stack of images from the same data set to create a mask indicating the location of the major blood vessels. The grayscale value for each pixel in the data images that was located in the mask was replaced by the mean of the grayscale values of the appropriate surrounding pixels (5–35 pixels) outside the mask (Blasdel and Campbell 2001; Bosking et al. 2000). In most cases, this process altered about 10% of the pixel values in the region of V1 that we imaged. Quantitative measures were made in regions outside of the masked areas. Resulting difference images were smoothed using a 7 x 7-pixel mean filter kernel. Low-frequency noise was reduced by convolving the image with a 80 x 80 pixel mean filter kernel and subtracting the result from the original image. These difference images were "clipped" at 1.5–2 SDs around the mean of the image pixel distribution and scaled in the range of 0–255 gray levels. Difference images were also combined by vector summation to produce orientation preference maps. In this paper, orientation preference maps where we show only the angle information from this vector summation we term angle maps; those maps where only the magnitude of the vector is depicted we term magnitude maps; and those maps where angle and magnitude information are combined we term polar maps. Maps showing rate of change of orientation preference were also constructed for the same regions of cortex with the highest rate of change of orientation represented as white and the lowest as black.
Ocular dominance map analysis
As described earlier, ocular dominance maps were constructed by comparing the images acquired during the stimulation of one eye (with full field gratings at 4 or 8 orientations) with images acquired during the stimulation of the other eye so that regions dominated by different eyes appear dark and light, respectively. The ocular dominance maps were band-pass filtered in the same way as described above for orientation difference maps. The same images were also used to create contours of ocular dominance by reducing the number of gray levels from 256 to 8–16 discrete levels and subsequently determining the geometric centers of ocular dominance domains.
Quantitative analyses of map relationships
ORIENTATION PREFERENCE VERSUS CO. The outlines of CO blobs (manually drawn or determined by a custom program) were transferred (see next section) to maps of orientation and ocular dominance. Orientation pinwheel centers were identified based on high rate of change regions in the map of orientation preference, and confirmed by visual inspection of orientation preference maps and difference images. Pinwheel locations were then transferred to ocular dominance maps and CO maps from the same cortical region. We then quantified the number of pinwheel centers found in different regions of the ocular dominance map and in the blob and interblob regions. To quantitatively examine the degree of orientation response selectivity between compartments (i.e., V1 CO blobs and interblobs), we calculated the index of orientation response selectivity on the 8-bit magnitude map by dividing the gray scale value of each pixel (the resultant vector strength of 4 stimulus orientations) by the maximum gray value, 255. The orientation selectivity index ranged from 0 to 1 (lowest to highest).
ORIENTATION PREFERENCE VERSUS OCULAR DOMINANCE. Color-coded iso-orientation contours were created from orientation preference maps and superimposed over contours of the ocular dominance map to examine how iso-orientation contours intersected with ocular dominance contours. In addition, we quantified the relationship between orientation preference pinwheels and ocular dominance centers from two cases by performing a nearest neighbor analysis. In each case, we first examined the distribution of distances to the nearest neighboring pinwheel for each pinwheel in the analysis region. We also computed nearest neighbor distances using a random placement of the same number of pinwheels in the same cortical area. This procedure was repeated for 1,000 sets of randomly placed pinwheels and the mean cumulative probability distribution and 95% confidence intervals on this distribution were calculated. The same procedure was carried out for ocular dominance centers. In addition, we calculated the minimum distance between each pinwheel and the nearest ocular dominance center. We compared this distribution to the same distribution calculated using random placement of pinwheel centers. Again this procedure was performed for 1,000 sets of random pinwheel placements and the mean and 95% confidence interval were calculated.
Histology and alignment
At the termination of each experiment, the animal was killed with an overdose of sodium pentobarbital and perfused transcardically with a saline rinse followed by 2% paraformaldehyde in 0.1 M phosphate buffer. The brain was removed and the imaged area of cortex was separated and flattened. The imaged piece of cortex was then frozen and cut with the surface vascular pattern preserved in the first 100- to 150-µm section. Subsequent sections were cut at 50 µm. CO staining was performed using methods described previously (Boyd and Matsubara 1996).
Surface and radial blood vessels were the primary landmarks used to align histological sections to the optical reference images of the imaged area. Differences between images and sections due to distortion or tissue shrinkage (10–15%) were handled by global scaling and rotation. After the optical images were aligned with the histological data, we examined the corresponding locations or patterns of the imaged brain areas to compare CO-staining patterns with map structures of orientation preference and ocular dominance and to see if the anatomical delineation of each visual area was compatible with the optical imaging results.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Visuotopic organization
Our goal was to use optical imaging with stimuli restricted to small regions of the visual field to examine the population response to simple stimuli and to examine the structure of the map of visual space. We focused on the most accessible portion of bush baby V1 for optical imaging, which represents approximately the central 10° of the visual field (Rosa et al. 1997).
Topographically restricted stimuli produced large, patchy, activation patterns (Figs. 1 and 2). The patchy nature of the activation pattern was expected because we used oriented gratings within topographically restricted windows. For stimulus gratings confined to 2° windows, this activation pattern was 2.8 ± 0.36 (SD) mm in width (n = 5 cases). Stimuli as small as 0.5° produced activation patterns that were 1.34 ± 0.2 mm in width (n = 2 cases, data not shown). Movement of the stimulus window by as little as 1° on the screen produced highly overlapping but discretely shifted cortical activation patterns. For example, as the stimulus was moved from the vertical meridian to more eccentric positions, the stripe of activation moved from anterior to posterior in V1 (Figs. 1 and 2). As the stimulus was moved from above the horizontal meridian into the lower visual field, the band of activation moved from lateral to medial in V1. The same size stimuli appeared to activate slightly more cortical territory when presented close to the area centralis (e.g., Figs. 1H and 2D).
By using a series of stimuli in both the vertical and horizontal axes spaced at 1° we were able to construct complete maps of visual space for central V1 in two cases (e.g., Fig. 1K). Although in some cases we could see a band of activation in V2 (
, Fig. 1E), the strength of the signal in V2 was not sufficient to construct a map of visual space for this area. The validity of our maps produced using this technique was confirmed by examining the activation foci produced by small circular patch stimuli presented at specified locations. For example, for the case shown in Fig. 2, a 2° circular stimulus patch located 4° from the vertical meridian (VM) and 1° below the horizontal meridian (HM), produced cortical activation at the location predicted by the intersection of the contours shown in Fig. 2H.
The location of the V1/V2 border predicted by our imaging with topographically restricted stimuli was consistent with the border determined by differences in CO staining between V1 and V2 (see following text). The V1/V2 border also could be distinguished by differences in structure in the orientation preference map such as larger domains in V2 (see following text).
We obtained the cortical magnification factor (CMF) by measuring the distance (mm) between two adjacent intersection points of iso-azimuth and iso-elevation lines of the visual field maps and dividing this number by the corresponding distance in degrees of visual space (see Fig. 1K). The CMF values we obtained for eccentricities of 1–8° using optical imaging (Fig. 3 A) matched the CMF versus eccentricity function previously generated by physiological recordings {Fig. 3A, dashed line, CMF = 2.36 · [E (eccentricity) + 0.73]–0.8} (Rosa et al. 1997). Our measurement of the average CMF at eccentricities of 1–8° (n = 4 cases) was 0.77 mm/°, which matches the average extrapolated value (0.76 mm/°) of Rosa et al. (1997) at the same eccentricities. The CMF for the representation of central vision in bush babies is much less than that reported in macaque monkeys (Fig. 3B, dotted line).
|
). CMF measured along the iso-azimuth dimension (parallel to the vertical meridian) was greater than that measured along the iso-elevation dimension (parallel to the horizontal meridian), which was indicated by the rectangular (not square) pattern of grids seen in the visual field map (Figs. 1K and 2H) defined by the iso-elevation and -azimuth lines. The mean ratio of the iso-azimuth versus iso-elevation dimension for eccentricities between 0 and 8° was 1.16 ± 0.03 (n = 4). Orientation preference in V1 and V2
In V1, optical imaging of intrinsic signals revealed a highly regular and continuous map of orientation preference (Fig. 4). The orientation preference map was similar to those found in other species and contained regions where orientation cycles around point singularities (pinwheel centers), regions where orientation preference changes in a straight line (linear zones), and regions where orientation preference changes slowly (saddle points) (e.g., cat, Bonhoeffer and Grinvald 1991; Shmuel and Grinvald 2000; macaque, Blasdel 1992; squirrel monkey, Malach et al. 1994; ferret, Chapman et al. 1996; tree shrew, Bosking et al. 1997; owl monkey, Xu et al. 2004a). The average center-to-center spacing of iso-orientation domains in V1 is 0.69 ± 0.1 mm. Areas of high rate of change in the orientation preference map are largely confined to pinwheel centers (Fig. 4D). We found both pinwheel centers with clockwise progression of orientation preference (Fig. 4E, blue squares), and pinwheel centers with counter-clockwise progression of orientation preference (Fig. 4E, red squares). Regions with low magnitude of orientation selectivity (dark areas in Fig. 4B) occupy only a small fraction of the map and largely correspond to the centers of pinwheels and to short thin bands that interconnect adjacent pinwheel centers (Fig. 4D).
|
; Malach et al. 1994; Roe and Ts'o 1995; Ts'o et al. 1990; Xu et al. 2004a). Surprisingly no such compartments were seen in V2 of bush baby. Figures 5 and 6 compare the organization of V2 in bush babies and owl monkeys imaged in the same setup under identical conditions. As can be seen in Fig. 5, the map of orientation preference in bush baby V2 was largely uniform, whereas in owl monkey (Fig. 6), there were clear stripes of low and high orientation selectivity in V2 (see also Xu et al. 2004a). As reported by others, we also saw no evidence of CO bands in bush baby V2 (see Fig. 5H) (also see Collins et al. 2001; Condo and Casagrande 1990). Owl monkey V2, however, clearly has bands in CO-stained sections (Fig. 6D).
|
|
; Ts'o et al. 1990; Xu et al. 2004a). Consequently, the density of pinwheel centers was lower in V2 than in V1 (Fig. 5G). The mean pinwheel density in V2 was 3.7 ± 0.5 mm–2, whereas it is 6.4 ± 0.4 mm–2 for V1. Ocular dominance map in V1
Consistent with both previous anatomical and physiological evidence for ocular dominance columns in bush baby V1 (DeBruyn et al. 1993; Glendenning et al. 1976), we found evidence for the existence of ocular dominance domains in eight of nine cases. In the ocular dominance map shown in Fig. 7, dark regions correspond to one eye activation and light regions to the other eye activation; white and black dots indicate the geometric centers of ocular dominance domains. Ocular dominance domains found in the bush baby tended to be less regular, more ovoid, and less stripe-like than those observed in the macaque (e.g., see Blasdel 1992; Blasdel and Campbell 2001; Shtoyerman et al. 2000). The average distance (center to center) between adjacent left and right eye columns was 0.53 ± 0.12 mm. The ocular dominance patterns that we observed were confined to V1 (Fig. 7, C and D).
|
The relationship between the map of orientation preference and the map of ocular dominance was investigated in detail for two cases. We used three methods to examine this relationship. The first method was to obtain iso-orientation contours and compare these to the map of ocular dominance (Fig. 8 A). A majority of iso-orientation contours tended to run orthogonal to the borders of ocular dominance domains (Fig. 8A). This impression was quantified by creating frequency histograms of the distribution of intersection angles between iso-orientation contours and ocular dominance contours (Fig. 8B). The mean intersection angle, pooled from two cases, was 64.4 ± 29.9, with around 55% of the intersection angles falling in the range of 68–90°. The relationship between orientation and ocular dominance contours is less regular than reported in the macaque and more similar to that seen in cats (Bartfeld and Grinvald 1992; Blasdel and Salama 1986; Hübener et al. 1997; Kim et al. 1999; Obermayer and Blasdel 1993).
|
; Blasdel and Salama 1986; Obermayer and Blasdel 1993), pinwheels did not appear to be preferentially located over the darkest or lightest regions of the ocular dominance map (e.g., see Fig. 8A). Overall, only 31.5 ± 6.3% of pinwheel centers (mean ± SD, n = 3 cases) were located in the center of ocular dominance domains; ocular dominance centers as measured here occupied 20% of the measured V1 area.
Finally, the third method we used to quantitatively assess the relationship between the map of orientation preference and the map of ocular dominance was to examine the distribution of distances between pinwheel centers and ocular dominance centers. For each case, we first examined the cumulative distribution of distances between neighboring pinwheel centers (Fig. 9 A). The cumulative distribution for our real data for pinwheel centers was significantly shifted to the right compared with a similar distribution generated by calculating nearest neighbor distances after random placement of pinwheels. This indicates that the placement of pinwheels is highly nonrandom. Pinwheel centers tended to be spaced more uniformly than would be expected for a random arrangement with many nearest neighbors found at distances of 300 µm. A similar result for the distribution of pinwheel centers has been obtained in the cat and the ferret (Müller et al. 2000).
|
500 µm. Although based on this nearest neighbor analysis each individual map was arranged in a highly nonrandom fashion, we found no evidence for a relationship between the two maps. The relationship between maps was assessed by calculating the distance between each pinwheel center and the nearest ocular dominance center. This procedure was also performed for data sets where pinwheels were placed randomly. In this case, we found no difference between the cumulative distribution generated from our real data and the one generated using random pinwheel placement (Fig. 9C). Had there been a strong tendency for pinwheel centers to lie near the center of ocular dominance domains, we would have expected a leftward shift of the cumulative distribution.
Relationship between CO blobs and ocular dominance and orientation preference maps
The pattern of CO staining was examined in tangential sections in seven bush babies. As seen in the macaque, and previously reported for the bush babies (Condo and Casagrande 1990), dark CO staining in superficial layers of V1 appeared as a periodic series of "blobs" (see Figs. 5 and 10). Figure 10 shows the CO-staining pattern, angle map, magnitude map, and ocular dominance map for the same region of cortex from the same case with the stained tissue aligned to the optical reference image. We first analyzed the number of pinwheels found in blob and interblob areas. In Fig. 10, B and C, for this case, 33.6% of cortex was covered by blobs, and 31.5% of the pinwheels were found in blobs, which suggested that there is no tendency for pinwheels to be located preferentially in the blob areas. Similar relationships were obtained in two other animals. We also examined the magnitude of orientation selectivity in blob and interblob regions. The average orientation selectivity index from 24 CO blobs pooled from two cases was 0.51 ± 0.01 (mean ± SE); the average index from 20 interblobs was 0.45 ± 0.03. These two values were not significantly different.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Visuotopic organization
We found that even stimuli restricted to small portions of the visual field evoked large patterns of activity over a millimeter in width in V1. This result is similar to the results from previous studies that have attempted to determine the point image or the amount of cortex activated by restricted stimuli in other species (Blasdel and Campbell 2001; Bosking et al. 2002; Grinvald et al. 1994). The systematic changes in the V1 activation pattern we observed with small movements in the stimulus in either the horizontal or vertical axis suggest that the map of visual space is nevertheless organized in a smooth and continuous fashion. This result is also similar to results from other primates (Blasdel and Campbell 2001; Tootell et al. 1988) and to results obtained in the tree shrew (Bosking et al. 2002) but differs from at least some reports for the cat (Das and Gilbert 1995). The combination of large population responses to spatially restricted stimuli and an orderly representation of visual space has important consequences for considering how coverage is obtained in visual cortex (see following text).
The cortical magnification factor for central vision (eccentricities of 0–10°) was much smaller in the bush baby (0.8 mm/°) than that observed in diurnal primates such as the macaque monkey (2.2 mm/°) and the squirrel monkey (2.3 mm/°) (Blasdell and Campbell 2001) but is comparable to that found in other nocturnal primate and non-primate species such as the owl monkey (0.9 mm/°, unpublished observations) and the cat (0.9 mm/°) (Tusa et al. 1978). This result is not surprising given the much smaller amount of retinal specialization for central vision found in the retina of the bush baby as compared with diurnal species such as the macaque monkey (DeBruyn et al. 1980; Perry and Cowey 1985; Wässle et al. 1990).
We found a small amount of overall anisotropy in the cortical magnification factors for the two axes of visual space with 1.2 times as much cortical space devoted to representing 1° of visual space along an iso-azimuth line as compared with 1° of visual space along an iso-elevation line. This type of asymmetry in the mapping of visual space has also been found in macaque and squirrel monkeys, and in these species, it has been suggested that this expansion of the map of the visual field on one axis is related to the layout of ocular dominance bands or domains (Blasdel and Campbell 2001; Tootell et al. 1988). We did not attempt to resolve whether this is the source of the magnification factor anisotropy in the bush baby.
Spatial organization for orientation, ocular dominance, and CO
An organized orientation preference map is a robust feature of bush baby V1 as it is in all carnivores (Bonhoeffer and Grinvald 1991; Chapman et al. 1996; Shmuel and Grinvald 2000; White et al. 2001) and primates (Blasdel 1992; Malach et al. 1994; Xu et al. 2004a) that have been examined as well as for some other mammalian species such as the tree shrew (Bosking et al. 1997). The map of orientation preference is organized according to similar principles in all of these species despite large differences in the overall functional organization observed in V1 of different mammals including differences in the size of V1, the cortical magnification factor, the number of other stimulus features represented in V1, and the presence or absence of ocular dominance domains and CO blobs. There are also differences in the actual sets of connections that are likely to provide the anatomical basis for orientation tuning in each species. For example, orientation selectivity first arises in layer 4 of the cat and bush baby (DeBruyn et al. 1993; Hubel and Wiesel 1962) but is not found until more superficial layers in the macaque and tree shrew (Blasdel and Fitzpatrick 1984; Chisum and Fitzpatrick 2004; Hubel and Wiesel 1968). Overall, these observations demonstrate striking similarities among species in the layout of the map of orientation preference in upper layers of V1 despite possible variation in the neural mechanisms that first instantiate orientation selectivity in the cortical network. These findings also suggest that cortical maps of orientation preference in V1 may be a natural consequence of activity-dependent mechanisms of cortical development that are constrained by orientation bias elsewhere in the retino-geniculostriate pathway (Chisum and Fitzpatrick 2004; Xu et al. 2002). It seems unlikely that one specific genetic plan for generating an orientation preference map in V1 has been conserved across all the mammals that exhibit these maps. Nevertheless, the universal presence of orientation selectivity at the first telencephalic level in both birds and mammals suggests that this feature is essential to higher order processing of visual features (see Medina and Reiner 2000).
We found no evidence for regional variation in the magnitude of orientation selectivity in V1. Areas of poor selectivity and high rate of change in orientation preference were largely confined to pinwheel centers. This result is consistent with observations from the cat (Bonhoeffer and Grinvald 1991; Shmuel and Grinvald 2000), tree shrew (Bosking et al. 1997), macaque (Blasdel 1992; Blasdel and Salama 1986), and owl monkey (Xu et al. 2004a). However, a recent study has noted that when orientation selectivity is evaluated by examining the height and width of tuning curves instead of using the magnitude of a vector addition that larger areas of cortex are found that have a low height of the orientation tuning curve (Swindale et al. 2003). We did not examine this issue in the bush baby, but it is clear in the bush baby and in other species that orientation tuning width appears to remain at near normal values within small distances of pinwheel centers and that the map of orientation selectivity remains continuous across any area of low tuning height that may exist (Maldonado et al. 1997; Swindale et al. 2003).
We found a robust spatial organization for ocular dominance in the bush baby. In primates, much more variation is seen in the strength and organization of the ocular dominance map than is observed for orientation selectivity (Horton and Hocking 1996). For example owl monkeys do not have robust ocular dominance columns (Blasdel and Campbell 2001; Kaas et al. 1976; O'Keefe et al. 1998; unpublished optical imaging results from our lab). Individual squirrel monkeys show strong variation in ocular dominance organization (Horton and Hocking 1996). In marmosets, it has been possible to demonstrate ocular dominance columns only in animals that have had long periods of monocular suture or in neonates (DeBruyn and Casagrande 1981; Sengpiel et al. 1996; Spatz 1989). The organization of the ocular dominance patterns that we observed in the bush baby is more similar to the patchy pattern found in cats (Hübener et al. 1997; Issa et al. 1999; Löwel et al. 1998) than to the more stripe-like pattern found in macaque monkeys (Blasdel 1992; Blasdel and Campbell 2001; Shtoyerman et al. 2000), although the density of anatomically defined dominance columns in cats is lower (2/mm2) (Hubel and Wiesel 1977; Shatz et al.1977) than found in bush babies (4/mm2) (Glendenning et al. 1976). It is not clear why different primate species exhibit differences in the strength and pattern of ocular dominance organization (see also Adams and Horton 2003).
Finally, in this report, as in other reports for the bush baby, we found a robust and highly regular pattern of staining for CO in V1 (Collins et al. 2001; Condo and Casagrande 1990). With small differences in the shape and distribution of CO blobs (see Condo and Casagrande 1990), the pattern of CO staining in bush baby V1 is similar to that observed in macaque monkeys (Horton and Hubel 1981; Livingstone and Hubel 1984) and squirrel monkeys (Malach et al. 1994).
Map relationships
Despite the robust and regular pattern of each individual columnar system in the bush baby, we found very little evidence for precise relationships between maps. The only robust relationship that we did find was a tendency for contours in the map of orientation preference to cross the borders of ocular dominance domains at right angles. This particular result is similar to results obtained for quantification of intersections between contours in these two maps in the cat and macaque (Bartfeld and Grinvald 1992; Blasdel and Salama 1986; Hübener et al. 1997; Kim et al. 1999; Obermayer and Blasdel 1993). However, it is clear that the relationship between these two maps is not as constrained in the bush baby as it is in the macaque or cat because when we examined the location of pinwheel centers relative to ocular dominance centers, we found no systematic relationship.
Although we only examined the relationship between ocular dominance and CO domains using qualitative methods, it is clear that the relationship between these columnar systems is also not the same as in the macaque. In the macaque monkey, CO blobs have been found to precisely align with the center of ocular dominance bands (Bartfeld and Grinvald 1992; Blasdel and Salama 1986). In the bush baby, we find instead that individual CO blobs can lie entirely within one ocular dominance domain or can be found directly spanning the border between left and right eye domains. This type of organization has also been observed in the squirrel monkey (Horton and Hocking 1996). A lack of relationship between these two maps poses an interesting problem in the bush baby given that the direct LGN input to CO blobs is all monocular, coming from K cells related either to the left or right eye (Ding and Casagrande 1997; Lachica and Casagrande 1992; Lachica et al. 1992). A preliminary study in bush babies, nevertheless, showed that even though ocular dominance columns are clear, injection of tracer into one eye labeled every CO blob (Olivaria et al. 1997). This would suggest that all CO blobs receive K monocular inputs from both eyes in bush babies even though ocular dominance columns innervated by either P or M cells remain monocularly segregated (Florence and Casagrande 1987, 1990). Presumably this is not the case for macaque monkeys since each blob appears to be dominated by the monocular input represented in the ocular dominance column in which it lies (Bartfeld and Grinvald 1992; Blasdel and Salama 1986; Horton and Hubel 1981).
Finally we found no evidence for a relationship between compartments defined by CO in V1 and the map of orientation selectivity. Our first measure of this relationship was an assessment of the number of pinwheels in blob and interblob areas as compared with the percentage of V1 occupied by these two regions. As some have questioned the accuracy and precision of optical imaging for addressing this question (Polimeni et al. 2005), we also evaluated the magnitude of orientation selectivity in blob versus interblob regions. By this measure, we also found no difference between blob and interblob regions. This result is in accord with previous physiology from the bush baby (DeBruyn et al. 1993). Whether orientation selectivity is lower in the blobs in the macaque monkey has been somewhat controversial (Lennie et al. 1990; Leventhal et al. 1995; Livingstone and Hubel 1984; Ts'o and Gilbert 1988). What has been clear in both species, however, is that the orientation preference map is continuous across blob boundaries, and that no large areas of poor orientation tuning width exist in V1 (Bartfeld and Grinvald 1992; Blasdel 1992; Blasdel and Salama 1986). Optical imaging has been used to demonstrate variation in the magnitude of orientation selectivity in V2 of owl monkeys and macaque monkeys (Roe and Ts'o 1995; Ts'o et al. 1990; Xu et al. 2004a), so we have reason to believe that we should be able to see these regions in V1 if they exist. Thus CO blobs do not serve to strictly segregate V1 into regions that either exhibit orientation selectivity or fail to exhibit orientation selectivity.
Compartments in V2
In the macaque monkey, squirrel monkey, and owl monkey, relationships between columnar systems for orientation selectivity and CO staining are found in V2 (Bartfeld and Grinvald 1992; Blasdel and Salama 1986; Livingstone and Hubel 1984; Malach et al. 1994; Roe and Ts'o 1995; Ts'o et al. 1990; Xu et al. 2004a). In the bush baby, we found no evidence for functional compartments in V2 as defined by either CO staining or regions of poor orientation selectivity. This suggests that compartments of this type are not a common feature of all primates and that they are not required for generation of the basic response properties found in V2. This result, however, should not be construed as providing evidence that there are no functional subdivisions of any sort in bush baby V2. Stripe-like regional variation in the input and output connections of V2 is still observed in the bush baby, and it is likely that this reflects real differences in response properties across the layout of V2 (Boyd et al. 1997; Collins et al. 2001).
Lack of map relationships
A lack of relationships between columnar systems, or feature maps in V1, had already been shown for particular combinations of maps in the ferret (White et al. 2001), squirrel monkey (Horton and Hocking 1996), and tree shrew (Bosking et al. 2002). Our results in the bush baby extend this observation to another primate, one that has particularly striking representation of each individual columnar system. Thus our results demonstrate that precise relationships between feature maps are not a common element of functional organization in all primates. Furthermore it seems likely that such relationships are no
, band of activation in V2. K: visual field map of V1 constructed based on the series of activation patterns shown in B–J, superimposed on a cortical surface image from the imaged region. For the visual field map, H –2° and H +2° are 2° below and above the horizontal meridian (HM) representation, respectively; V +2° and V +4° are 2 and 4° eccentric from the vertical meridian (VM) representation, respectively. A, anterior; M, medial.
