2004

Farley, B.J, and Sur, M.
Retinotopic maps in rat visual cortex revealed by temporal decoding of optically-imaged intrinsic signals.
Soc. Neurosci., 2004.

We have explored the complete retinotopic organization of rat visual cortex using optical imaging of intrinsic signals. Long Evans rats were anesthetized with isoflurane gas in a mixture of nitrous oxide and oxygen. Imaging was performed through the dura using a 700 nm filter. As a stimulus, we used a single vertical or horizontal bar that drifted across the visual field periodically, in a direction orthogonal to its orientation. Light reflectance data was continuously captured, and retinotopic maps generated by analyzing the optical response at the frequency of visual stimulation. Using this technique we have generated robust maps of retinotopy in rat visual cortex. Iso-azimuth contours in V1 run nearly parallel to the antero-posterior axis of cortex, whereas iso-elevation contours run obliquely to this axis and extend from antero-lateral to postero-medial cortex. This layout is very similar to that found in the mouse. Near the center of V1, the magnification factor for vertical space is approximately 0.05 mm/deg, while that for horizontal space is 0.03 mm/deg. Our data additionally indicate the existence of multiple visual areas in rat visual cortex, as evidenced by regions outside of V1 displaying retinotopy patterned distinctly from that within V1.

Lyckman, A., Horng, S.H., Oray, S. and Sur, M.
Layer-IV specific expression of cardiac troponin c: developmental and activity-dependent regulation of a novel calcium binding protein during ocular dominance plasticity in mouse visual cortex.
Soc. Neurosci., 2004.

Ocular dominance plasticity (ODP) in the developing visual cortex is a model of activity-dependent cortical synaptic plasticity. Calcium appears to be a key signaling molecule in this system. Here, we show that cardiac troponin C (CTC), an EF-hand calcium-binding protein typically associated with regulation of actin-myosin interactions in heart muscle, may have a prominent role in mediating ODP. CTC mRNA is strongly upregulated in mouse visual cortex at the peak of ODP, as demonstrated by both DNA microarray analysis and semi-quantitative PCR. Immunofluorescence shows that CTC protein is widely expressed throughout the early postnatal mouse brain, but becomes largely restricted to layer IV cortical excitatory neurons by P15. CTC protein levels fall in response to brief (3d) monocular deprivation during the critical period for ODP. CTC in layer IV neurons is largely nuclear or perinuclear suggesting that it transits between the nucleus and cytoplasm. Western blotting shows that heart and cortex coexpress isoforms of 50 and 25kD, while isoforms of 18kD and 41kD are specific to heart and cortex, respectively. These data demonstrate the discovery of a layer IV specific calcium-binding regulatory protein whose expression is development- and activity-dependent, and that is potentially involved in physiological and structural synaptic plasticity during the critical period.

Lyon, D.C., Schummers, J., Sharma, J. and Sur, M.
Effects of stimulus sequence and abrupt transitions in luminance on V1 neurons in awake monkey.
Soc. Neurosci., 2004.

Studies of contrast adaptation in V1 of anaesthetized animals have focused primarily on the effects on a cell’s firing rate to a single contrast presented several times in rapid succession or successive presentations of gradually increasing/decreasing contrasts. This approach has been useful for defining the time course of contrast or pattern adaptation. Yet, in awake monkeys, different levels of awareness or expectation may modify these adaptation effects. For example, sudden changes in luminance can invoke strong bottom-up or stimulus-driven processes that capture attention, while a sequence of stimuli can set up an expectation that is altered with an unexpected transition. To test the interaction of abrupt changes in luminance with adaptation we presented sequences of white or gray squares on a slightly darker gray background. The squares were presented for 1sec every 3sec to the fixating monkey, on the receptive fields of V1 neurons. In general cells showed moderate to weak adaptation when the same square was presented several times in a row, in that firing rate gradually decreased over repeated presentation of the same square. At points in the sequence where contrast increased “unexpectedly”, neurons fired as if the adaptation effect was extinguished. To further explore effects of expectation the monkey viewed sequences of gray – white squares, still spaced 2sec apart. Periodically, two white or two gray squares would appear in succession. In most of these cells general adaptation effects were apparent. Interestingly, contrary to what would be expected of adaptation, in some cells the average firing rate was higher following the second successive (oddball) presentation of a white or gray square. We conclude that the monkey was not only adapting to the luminance of the stimulus, but also to the repetitive sequence. The occurrence of an unexpected change in the stimulus sequence served to eliminate the neuron’s bottom-up adaptation.

Majewska, A., Newton, J.R. and Sur, M.
Dendritic spine and axon terminal dynamics in visual, somatosensory and auditory cortices in vivo.
Soc. Neurosci., 2004.

Remodeling of the primary visual cortex during ocular dominance plasticity is thought to progress from functional alterations in the response properties of single neurons to large anatomical shifts in axonal arborizations. Moreover, it has been suggested that the functional changes are a direct result of altered synaptic efficacy which is first apparent in the extragranular layers of the cortex (layers II/III, V and VI). We have examined the structural correlates of this functional modification at the level of single synapses by using two-photon microscopy to visualize the dynamic properties of dendritic spines in V1 both in vivo and in acute slices. In transgenic mice expressing GFP in a subset of their layer 5 pyramidal neurons, spine dynamics were elevated in extragranular layers following brief (2-3 day) monocular lid suture during the critical period. This increase in spine dynamics was not accompanied by other morphological changes in average spine length, head diameter or neck diameter. In order to explore potential mechanisms that might promote structural dynamics, we examined the effects of extracellular matrix degradation through the tissue plasminogen activator (tPA) / plasmin proteolytic cascade. In acute slices, both tPA and plasmin significantly upregulated spine dynamics in normal, non-deprived visual cortex. Further, brief monocular deprivation occluded any subsequent effects of the tPA/plasmin cascade in a lamina-specific manner, indicating that these processes share common mechanisms. These data are consistent with the hypothesis that rapid changes in synaptic efficacy following monocular deprivation are accompanied by increased extracellular matrix degradation, which can then promote structural remodeling and may potentially lead to significant anatomical reorganization.
Support Contributed By: Grants from the NIH (MS) and a Whiteman Fellowship (AM).

Newton, J.R., Sharma, J., Yu, H. and Sur, M.
Intrinsic signal optical imaging reveals the distinct effects of thalamic and cortical mechanisms on patterning and arealization in V1 and V2 of ephrin A2/A5 knock-out mice.
Soc. Neurosci., 2004.

The Eph receptor tyrosine kinases and their ephrin ligands have distinct complementary gradients in the thalamus and cortex. Ephrins are known to play an important role in topographic mapping; however their role in establishing the organization and patterning of cortical areas is unknown. We have examined the size, location and eye-specific driving of visual cortex in adult normal and ephrin A2/A5 knock-out mice using optical imaging of intrinsic signals. Visual field elevation and azimuth were mapped by monocularly presenting a single high-contrast bar drifting in periodic fashion. Images were analyzed by extracting the Fourier component of the intrinsic signal at the periodicity of the drifting bar. We find that although their size remains similar, the location of V1 and V2 in ephrin A2/A5 knock-out mice is shifted significantly medial and anterior relative to their location in normal mice. There is also an expansion of the ipsilateral representation within V1 and V2 of the knock-out mice. The shift in the location of V1 and V2 is consistent with a cortical ephrin A5 gradient (Bolz et al., 2004). The expansion of the ipsilateral representation is consistent with expanded retinogeniculate projections in ephrin A2/A5 knock-out mice (Ellsworth et al., 2004), which is due to removal of an ephrin gradient in the LGN (Feldheim et al., 1998). Thus, the organization of visual cortex reflects two distinct roles of ephrins: a cortical gradient that influences location and regionalization of V1 and V2, and a thalamic gradient that influences eye-specific patterning within these areas.

Support: NIH grants EY13900 (JRN), EY11512 and NS39022 (MS).

Oray, S., Majewska, A. and Sur, M.
Dendritic spine motility in visual cortex is regulated by brief monocular deprivation and extracellular matrix degradation.
Soc. Neurosci., 2004.

Remodeling of the primary visual cortex during ocular dominance plasticity is thought to progress from functional alterations in the response properties of single neurons to large anatomical shifts in axonal arborizations. Moreover, it has been suggested that the functional changes are a direct result of altered synaptic efficacy which is first apparent in the extragranular layers of the cortex (layers II/III, V and VI). We have examined the structural correlates of this functional modification at the level of single synapses by using two-photon microscopy to visualize the dynamic properties of dendritic spines in V1 both in vivo and in acute slices. In transgenic mice expressing GFP in a subset of their layer 5 pyramidal neurons, spine dynamics were elevated in extragranular layers following brief (2-3 day) monocular lid suture during the critical period. This increase in spine dynamics was not accompanied by other morphological changes in average spine length, head diameter or neck diameter. In order to explore potential mechanisms that might promote structural dynamics, we examined the effects of extracellular matrix degradation through the tissue plasminogen activator (tPA) / plasmin proteolytic cascade. In acute slices, both tPA and plasmin significantly upregulated spine dynamics in normal, non-deprived visual cortex. Further, brief monocular deprivation occluded any subsequent effects of the tPA/plasmin cascade in a lamina-specific manner, indicating that these processes share common mechanisms. These data are consistent with the hypothesis that rapid changes in synaptic efficacy following monocular deprivation are accompanied by increased extracellular matrix degradation, which can then promote structural remodeling and may potentially lead to significant anatomical reorganization.
Support Contributed By: Grants from the NIH (MS) and a Whiteman Fellowship (AM).

Sharma, J., Lyon, D.C., Schummers, J. and Sur, M.
Spatial and Temporal effects of directed attention in alert monkey V1.
Soc. Neurosci., 2004.

J. Schummers, J. Mariño and M. Sur.Picower Center for Learning and Memory, MIT, Cambridge, MA 02139 and Neurocom, Universidade da Coruña Spain

The mechanisms of synaptic integration involved in computing orientation selectivity are not fully understood. We have previously shown that neurons near pinwheel centers (PWs) in the orientation preference map receive broadly tuned inputs, yet produce sharply tuned outputs1. Here, we have examined the role of rapid membrane potential (Vm) fluctuations (“synaptic noise”) in shaping the spike tuning. It has been shown that Vm fluctuations in the gamma band preferentially lead to spikes, and that the tuning of power in the gamma range is more sharp than the mean Vm 2,3. One possibility, therefore, is that larger Vm fluctuations in the gamma (or other) range at preferred orientations compared to orthogonal generate sharp spike tuning in neurons near PWs. We combined optical imaging of intrinsic signals with whole cell recordings of Vm in adult cat area 17, and analyzed the tuning of different frequency bands, including gamma band Vm fluctuations, in neurons near PWs and in orientation domains. We find that gamma band power is generally, but not always, more sharply tuned than mean Vm, but less sharp than spike tuning. Cells in orientation domains show sharper tuning of power than cells at PWs, not only in the gamma range but also in frequency ranges below and above the gamma band. In some pinwheel neurons, gamma power is only weakly tuned. Overall, our analysis suggests that it is unlikely that differences in the temporal structure of Vm fluctuations between different stimulus orientations play a dominant role in producing sharp spike tuning in neurons near PWs. Other mechanisms, including strong orthogonal inhibition, may be more important in generating sharp tuning near PWs.

Supported by EY07023 (MS), MECD (JM) and HHMI (JS)

Schummers, J., Marino, J. and Sur, M.
Orientation tuning of rapid membrane potential fluctuations in neurons near and far from pinwheel centers.
Soc. Neurosci., 2004.

The mechanisms of synaptic integration involved in computing orientation selectivity are not fully understood. We have previously shown that neurons near pinwheel centers (PWs) in the orientation preference map receive broadly tuned inputs, yet produce sharply tuned outputs1. Here, we have examined the role of rapid membrane potential (Vm) fluctuations (“synaptic noise”) in shaping the spike tuning. It has been shown that Vm fluctuations in the gamma band preferentially lead to spikes, and that the tuning of power in the gamma range is more sharp than the mean Vm 2,3. One possibility, therefore, is that larger Vm fluctuations in the gamma (or other) range at preferred orientations compared to orthogonal generate sharp spike tuning in neurons near PWs. We combined optical imaging of intrinsic signals with whole cell recordings of Vm in adult cat area 17, and analyzed the tuning of different frequency bands, including gamma band Vm fluctuations, in neurons near PWs and in orientation domains. We find that gamma band power is generally, but not always, more sharply tuned than mean Vm, but less sharp than spike tuning. Cells in orientation domains show sharper tuning of power than cells at PWs, not only in the gamma range but also in frequency ranges below and above the gamma band. In some pinwheel neurons, gamma power is only weakly tuned. Overall, our analysis suggests that it is unlikely that differences in the temporal structure of Vm fluctuations between different stimulus orientations play a dominant role in producing sharp spike tuning in neurons near PWs. Other mechanisms, including strong orthogonal inhibition, may be more important in generating sharp tuning near PWs.

Supported by EY07023 (MS), MECD (JM) and HHMI (JS)

Tropea, D., Lyckman, A., Kreiman, G., Mukherjee, S. and Sur, M.
Gene expression in mouse visual cortex after visual input deprivation: gene systems mediating distinct effects of activity.
Soc. Neurosci., 2004.

Functional plasticity in visual cortex after visual deprivation is a powerful model system for understanding how activity and experience shape connections in cortical circuits. It is as yet unknown whether mechanisms that require the presence of activity are different from those that mediate competitive effects of activity. Here, we ask whether specific deprivation paradigms have distinct effects on gene expression in visual cortex. We extracted mRNA from the visual cortex of three groups of mice at P27: control (non-deprived); dark-reared (DR, P0-P27); and, monocularly deprived (MD, P11-P27). RNA samples were run on the MG-U74 v2 ABC microarray chip set (Affymetrix) covering 36701 cDNA target sequences. The effects of the two sensory deprivation regimens (DR and MD) were determined by comparing their average gene expression levels with the average control levels. Using a criterion of p<0.001 (Student’s t-test), 177 genes were down-regulated in DR, 85 genes were down-regulated in MD, 597 genes were up-regulated in DR and 22 genes were up-regulated in MD. Expression profiles of selected genes were confirmed by semi-quantitative PCR. For the known genes that showed a significant change in gene expression (either up- or down-regulated), we studied the enrichment of Gene Ontology (GO) categories in the set. We found that although genes affecting multiple cellular processes are affected by both MD and DR (including cell communication, metabolism, and secretion), some biological processes are more influenced by one of the deprivation paradigms: cellular and organismal physiological processes by MD, and cell maintenance, death and motility by DR. This is consistent with the hypothesis that vision per se is required for certain activity dependent mechanisms, whereas an imbalance of visual activity initiates distinct mechanisms that lead to a re-shaping of functional connections.
Supported by EY14134 and EY15068.

Wilson, N., Ty, M., Sur, M. and Liu, G.
An inverse relationship between quantal synaptic strength and network size.
Soc. Neurosci., 2004.

In highly plastic regions of the brain such as the hippocampus and cerebral cortex, excitatory synapses are believed to adjust their strength dynamically in order to ensure the effective propagation of information through the network. If network activity levels are low, neurons will scale up the strengths of their synapses to become more sensitive to sparse transmission events. If activity levels are high, neurons will scale down the strengths of their synapses to maintain firing rates within a functional range that is more physiologically meaningful. We hypothesized that a related “inverse rule” might be invoked to tune the strength of a neuron’s inputs as a function of the size of the network in which it is situated. For example, if the aggregate synaptic drive to or from a cell were a “finite resource” that is distributed among each of its synaptic partners, then a large network in which a cell has more synaptic partners would incur weaker inputs than a comparable network of fewer partners. To directly test this prediction, we made use of photolithographic microstructures to construct cultured neuronal networks of precisely controlled sizes. Larger networks that contained more neurons and involved more synapses exhibited a reduced unitary conductance per synapse than smaller networks. Furthermore, microlithographs of different sizes maintained a constant cell density and excitatory / inhibitory ratio across networks. These observations indicate that neurons in plastic circuits will trade off a few strong connections with a small number of neighbors for more numerous, weaker connections when additional neighbors are introduced. In this framework, negative feedback forms of plasticity could provide a fundamental mechanism for network processing, by enabling cells to alter the strengths of their inputs in proportion to the number of synapses and cells in their local microcircuit.

Yu, H.B., Majewska, A. and Sur, M.
Monitoring dendritic spine structure during monocular deprivation in the primary visual cortex of the ferret in vivo.
Soc, Neurosci., 2004.

Dendritic spines are small protrusions on dendritic processes. They contain the postsynaptic domains of most of the excitatory synapses in cortex, and are therefore well poised to subserve functional neuronal plasticity. Previous studies have revealed that dendritic spine morphology is very dynamic and plastic, indicating that spines have the potential to consolidate the functional plasticity at the micro-structural level. In this study, we examine whether changes in visually-driven activity elicit the modification of dendritic spine morphology in ferret visual cortex. In order to compare changes in subcellular structure to changes in network activity over time, structural two-photon imaging and functional intrinsic optical imaging are performed in the same piece of cortex in vivo. A modified Sindbis virus (Jeromin et al 2003) delivering Green Fluorescence Protein (GFP) is focally injected into the superficial layers of ferret primary visual cortex, and a transparent window is made above the injected cortex to allow chronic two-photon and optical imaging. We find that GFP labeled neuronal structures, including dendritic spines, can be clearly imaged two days after virus injection, and the label remains stable and can be repeatedly imaged for a couple of days. Optical imaging demonstrates a strong ocular dominance and orientation map in both injected and non-injected regions. These maps remain unchanged over a month in control animals. In order to examine visually-driven functional and structural changes, we monocularly deprive P40-50 ferrets by suturing the contralateral eyelids, and ocular dominance significantly shifts to the ipsilateral eye in a matter of days. During this time, the same groups of dendritic spines are repeatedly examined day by day in the deprived and non-deprived eye dominated regions, and the spine dynamics are compared with the simultaneous functional shift in ocular dominance.

(Supported by NIH grants EY07023, EY15068 & Whiteman Fund).

2003

Dragoi, V. and Sur, M.
Orientation Discrimination in visual cortex and the statistics of natural stimuli.
Soc. Neurosci., 2003.

A key issue in the function of the visual cortex is whether its properties are adapted to the statistics of natural scenes. How these statistics are influenced by eye movements during vision is not well understood. We analyzed
the eye movements of three rhesus monkeys freely viewing natural scenes and found a significant anisotropy in the stimulus statistics at the center of gaze. We found that fixation to an image patch is more likely to be followed by a saccade to a nearby patch of similar orientation structure or by a saccade to a more distant patch of largely dissimilar orientation structure. Dynamic changes
in the orientation tuning of primary visual cortex (V1) neurons can enable the visual system to take advantage of these statistics. We recorded from awake monkey V1 neurons and found that brief adaptation to either nearby or orthogonal orientations can increase the dynamic range of single neurons near their peak orientation, by altering the slope of the orientation tuning curve. This would improve orientation discrimination performance following nearby or orthogonal adaptation, as is indeed observed psychophysically.
Support Contributed By: McDonnell-Pew (VD) and NIH (MS)

Citation:
V. Dragoi, M. Sur. ORIENTATION DISCRIMINATION IN VISUAL CORTEX AND THE STATISTICS OF NATURAL STIMULI. Program No. 439.12. 2003 Abstract Viewer/Itinerary Planner. Washington, DC: Society for Neuroscience, 2003. Online.

Jin, D.Z., Dragoi, V., Sur, M. and Seung, H.S.
The tilt aftereffect is quantitatively consistent with adaptation changes of
orientation tuning observed in visual cortex.
Soc. Neurosci., 2003.

The tilt aftereffect (TAE) is a striking visual illusion in which prolonged adaptation to an oriented visual stimulus causes subsequent stimuli to appear rotated away from the adapting orientation. Explaining this and other aftereffects in terms of neural mechanisms has been an important outstanding problem. Historically, a popular explanation of the TAE has been a hypothesized relative suppression of neurons tuned to the adapting orientation. Recent physiological studies have identified another important factor: the preferred orientations of V1 neurons repulsively shift away from the adapting orientation. Here we construct a population coding model that includes both factors, and show that the repulsive shift of the preferred orientations is necessary for quantitatively explaining the TAE. According to the model, the TAE is indeed caused by the relative suppression of the neural responses. However, it is substantially weakened by the preferred orientation shift. The relative suppression and the repulsive shift together lead to a TAE that is quantitatively consistent with the psychophysical data. Suppression alone would produce a TAE that is more pronounced than that observed. We suggest that the visual system uses the repulsive shift of the preferred orientations to reduce the perceptual error in orientation that could be induced by neural response suppression. Our population model of the TAE suggests a strategy that can also be useful for understanding other perceptual aftereffects, such as the motion aftereffect and the spatial frequency aftereffect.
Support Contributed By: Howard Hughes Medical Institute (D.Z.J and H.S.S), the McDonnell-Pew Foundation (V.D.) and the NIH (M.S.)

Leamey, C.A., Kang, N., Croisier, E., Wang, K.H., Brandeau, O., Fassier, R., Tonegawa, S. and Sur, M.
Ten_m3 is expressed in the developing visual pathway.
Soc. Neurosci., 2003.

Cortical areas are characterised by their unique patterns of connectivity and cytoarchitecture. Recently, there has been increasing evidence that intrinsic molecular cues may play an important role in the control of cortical regionalization, yet little is known about the substrates that underlie the development of the appropriate patterns of connectivity. Using microarrays, we identified a number of genes that are differentially expressed between rostral and caudal cortical regions in the neonatal mouse. One of the genes, Ten_m3, is highly expressed in caudal cortex in layer V of a region that corresponds well to visual cortex during the first postnatal week. Ten_m3 is also expressed in the developing lateral geniculate nucleus and in the ganglion cells of the ventral retina. Retrograde tracing studies from the superior colliculus showed that corticocollicular neurons expressed Ten_m3, whereas injections of tracer into the contralateral cortex showed that callosal neurons do not express the gene. Ten_m3 encodes a type II transmembrane protein, and antibody staining revealed that the protein is present not only in the somata of layer V neurons, but also along the pathway of the corticofugal axons as they descend from the cortex towards the thalamus. We have also examined the embryonic expression pattern of the gene and found that it is expressed in the developing cortical plate by embryonic day (E)16. The expression of a transmembrane protein in the axons of a specific subset of cortical projection neurons in the developing visual system is intriguing and suggests that Ten_m3 may play a regulatory role during the formation of this pathway.
Support Contributed By: NIH EY14134 (MS)

Lyon, D.C., Schummers, J., Marino, J. and Sur, M.
Distribution of inhibitory and excitatory inputs to pinwheel centers and orientation domains.
Soc. Neurosci., 2003.

Orientation selectivity in cat V1 can be refined by excitatory and inhibitory intracortical inputs. At pinwheel centers, cells preferring one orientation are in close proximity to cells with very different orientation preferences.One might predict that locally a greater range of orientations would impinge upon a pinwheel cell. Whole-cell recordings show that they do receive a wider range of subthreshold inputs, yet their spike tuning remains sharp. Thus, heterogeneous excitatory inputs are perhaps balanced by the spread and/or strength of inhibitory inputs. To determine whether the local connectivity is responsible for this effect, we imaged orientation maps in cat V1 and placed small injections of fluorescent CTB at pinwheels and domains. Following a survival of 2 days the brain was cut tangential to the dorsal lateral V1 surface and GABA immunofluorescence was used to identify inhibitory neurons. Digital photos of the retrogradely transported label and GABA positive neurons were merged and the distributions of both the excitatory (tracer transport) and inhibitory (double labeled) neurons were plotted.The pattern of labeled cells was aligned to the orientation map using injection sites, and the blood vessels as landmarks. As predicted, within a 400um radius of the injection site, injections in pinwheel centers labeled neurons over a wide range of orientations, whereas domain injections predominantly labeled cells within the same domain. Furthermore, there was a greater proportion of GABA positive double labeled cells around pinwheel centers compared to domains. When the sampling radius was extended to 800um, the distributions of cells labeled from the injections were less orientation specific for both domains and pinwheels and the proportion of double labeled cells was reduced. Thus, local inhibition (within 400-500 um) may balance the broad excitation to produce narrow spike tuning near pinwheels.
Support Contributed By: NRSA (DL), HHMI (JS), MEC (JM), NIH (MS).

Lyckman, A., Ellsworth, C., Horng, S., Leamey, C. and Sur, M.
Gene expression during the critical period for ocular dominance plasticity in mouse V1.
Soc. Neurosci., 2003.

Majewska, A., Newton, J. and Sur, M.
Comparison of in vivo dendritic spine motility in different sensory cortical areas during development.
Soc. Neurosci., 2003.

Marino, J., Schummers, J. and Sur, M.
Input conductance at different locations within V1 orientation map.
Soc. Neurosci., 2003.

The mechanisms by which neurons in primary visual cortex (V1) generate orientation selectivity appear to be heterogeneous. Intracellular recordings have shown that subthreshold activity changes with cortical depth and with the position in the orientation map. Precise measurements of inhibitory and excitatory input conductances during visual stimulation have indicated a variety of combinations that may reflect structural inhomogeneities. Here we have calculated the changes in input conductances for neurons at known locations within the orientation map.
Experiments were conducted in adult cats. Optical imaging of intrinsic signals was performed to obtain the precise pattern of orientation domains and singularities (pinwheels). Next, we made whole-cell recordings at different locations in the map. The injection of negative and positive holding currents allowed us to measure changes in total conductance (Gt), excitatory conductance (Ge) and inhibitory conductance (Gi) during visual stimulation with drifting gratings. We selected for analysis 15 domain and 15 pinwheel neurons. Subthreshold activity, but not spike response, was significantly different between orientation domains and pinwheel centers. The averaged tuning curves of cells located at pinwheel centers, when compared with orientation domains, showed: a) a more flat profile, with increase in Gt for all stimulus orientations; b) a larger increase in both Ge and Gi for orientations orthogonal to the preferred. At the preferred orientation the mean value of the ratio Gi/Ge was > 1 for both types of neurons, but at orthogonal orientations, the ratio was > 1 for pinwheel cells and 1 for domain cells.
These results indicate that neurons located at specific locations in the orientation map receive very different combinations of excitatory and inhibitory inputs. In particular, inhibition sharpens orientation tuning much more at pinwheels than at orientation domains.
Support Contributed By: NIH Grant EY07023, MECD (Spain) and HHMI

Newton, J., Sharma, H., Yu, H. and Sur, M.
Optical imaging of intrinsic signals reveals visual organization in mouse V1 and V2.
Soc. Neurosci., 2003.

Examining the detailed organization of mouse primary visual cortex (V1) and secondary visual cortex (V2) is important for identifying the role of activity in shaping maps and modules. We used optical imaging of intrinsic signals to examine the retinotopy and possible modular organization of V1 and V2. Adult mice (2 months old) were anesthetized with a combination of urethane (1.5g/kg) and the sedative chlorprothixene (0.2mg). The skull was thinned over the visual cortex and illuminated with 605 nm light. Visual field elevation and azimuth were mapped by presenting binocularly a single high-contrast bar drifting in periodic fashion. Ocular dominance regions were explored by presenting either periodic drifting bars or drifting gratings of various orientations to the ipsilateral or contralateral eye in interleaved monocular trials. Orientation domains were examined by presenting periodic drifting bars or gratings of various orientations binocularly. Images in response to drifting bars were analyzed by extracting the Fourier component of the intrinsic signal at the periodicity of the drifting bar. Confirming previous work, we find that the vertical meridian lies along the V1/V2 border, and that there is an anisotropy of representation in the elevation axis. Specifically, the cortical magnification factor is approximately 2x greater along the elevation axis relative to the azimuth axis. Similar to higher mammals, this may suggest the presence of finer ocular dominance slabs that tend to run into the V1/V2 border and distort retinotopy by representing the same location in visual space separately for each eye. Compared to a stimulus consisting of drifting gratings, the presentation of a periodic single bar provides a better signal to noise ratio, and may help resolve whether or not modular organizations for ocular dominance and orientation exist in mouse V1 and V2.
Support Contributed By: EY13900 (JRN), EY11512 and NS39022 (MS)

Oray, S., Majewska, A. and Sur, M.
Compartmental differences in spine motility of layer V neurons in visual cortex following brief monocular deprivation.
Soc. Neurosci., 2003.

Dendritic spines receive the large majority of excitatory inputs in the mammalian cortex and changes in spine dynamics appear to be correlated with the restructuring of synaptic connections. One system in which synaptic rearrangements are well-characterized is that of monocular deprivation in the visual cortex. Monocular lid suture during a plastic period of development causes a rapid electrophysiological shift of responses towards the open eye. Moreover, it has been suggested that these changes occur first in the superficial and deep layers of the cortex (layers 2/3, 5 and 6), followed by changes in thalamic-recipient layer 4. We have studied spine motility in this system following brief periods of monocular deprivation using transgenic mice expressing GFP in a subset of layer 5 pyramidal neurons. Using two-photon microscopy, spines were imaged in the binocular visual cortex in acute brain slices following short (2-3 day) monocular lid suture during the height of the critical period (starting at P26). We find that spine dynamics on layer 5 pyramidal neurons, quantified as a change in spine length per unit time, exhibit different motilities based on their laminar distribution. Spines representing putatively intracortical synapses, such as those on the distal parts of the apical dendrite (in layer 2/3), or near the soma on proximal dendrites (in layer 5), show increased motility in deprived cortex as compared to control spines at matching dendritic locations in nondeprived cortex. This is in contrast to spines in a middle region of the apical dendrite (in layer 4), which show no change in spine motility following brief monocular deprivation. These data are consistent with the hypothesis that rapid changes in synaptic efficacy following activity imbalance occur first at intracortical synapses, and are a precursor to changes in feedforward connections during critical period plasticity.
Support Contributed By: Grants from the NIH (MS) and a Whiteman Fellowship (AM).

Schummers, J., Marino, J. and Sur, M.
Temporal dynamics of orientation tuning in conductance-based model neurons on the location within orientation map.
Soc. Neurosci., 2003.

Within the orientation preference map in V1, the orientation composition of the local connections impinging on any point in the map depends strongly on the local orientation gradient at that particular location. We have previously shown that the subthreshold synaptic inputs to individual neurons reflect the distribution of orientations in the local cortical patch surrounding them. However, the spike outputs do not depend on location in the map, indicating some form of non-linear input-output transformation. We have examined the timecourse of responses to flashed stimuli to determine whether the dynamics of responses might shed some light on the integration of inputs. Using reverse-correlation analysis of spike responses to rapidly changing orientations, we have compared the tuning curves at different times relative to stimulus onset between neurons located near and far from pinwheel centers. We find that the earliest spike responses in pinwheel neurons are less selective than in orientation domain neurons. Near pinwheel centers, there is an initial increase in response to all orientations, relative to a blank stimulus, which is subsequently sculpted by suppression at non-optimal orientations. At later timepoints, there is no difference in the tuning curves of pinwheel and orientation domain neurons, as in the steady state responses to drifting gratings. We have also made whole cell recordings of responses to flashed gratings. Preliminary analysis is consistent with several aspects of the extracellular reverse-correlation analysis. Overall, our results indicate that the global and/or orientation specific suppressive mechanisms may play an important role in shaping the input-output transformation performed by V1 neurons.
Support Contributed By: EY07023 (MS), MECD (JM) and HHMI (JS)

Sharma, J., Lyon, D. and Sur, M.
Influence of attention on center-surround interactions in alert monkey V1.
Soc. Neurosci., 2003.

One of the remarkable properties of neurons in primary visual cortex (V1) is that their responses are modulated by the presence of stimuli outside the classical receptive field (CRF) only when a stimulus is placed within their CRF. A key issue is whether contextual modulation in V1 is a neural correlate of perception or merely a low-level competitive interaction of feed-forward inputs that vie for a neurons response. If contextual modulation aids perception, then this modulation should be influenced by the behavioral state of the animal, such as attention. Here we have investigated the spatial and temporal effects of the extra-classical surround on center responses and the influence of attentional state. Monkeys were trained to attend to a small peripheral spot while fixating on a centrally placed spot. They had to leave a bar as soon as the attention spot was extinguished while maintaining fixation in order to earn juice reward. After determining the cells receptive field and its preferred orientation, a center patch consisting of drifting sinusoidal gratings was placed within its CRF. A surround grating patch extending 5 degrees from the center, and of similar or orthogonal orientation to the CRF, was also presented. The attention spot was located within the cells CRF or in the opposite hemi-field. In cases where the attentional state was not manipulated, the iso-oriented surround suppressed center responses while the orthogonal surround either facilitated responses or had little influence. The suppressive effect of the surround was almost always preceded by an initial facilitation that lasted between 150-200 ms. These responses were compared with those when the monkey attended towards the CRF. Our preliminary results show that the influence of surround was delayed in time when attention was directed towards CRF. Thus center-surround interactions in V1 can be modulated by the attentional state of the animal.
Support Contributed By: NIH grant EY07023.

Yu, H., Farley, B. and Sur, M.
Relationships between orientation and ocular dominance maps in cat area 18.
Soc. Neurosci., 2003.

Multiple stimulus feature maps exist in primary visual cortex, and specific relationships between them help to achieve a more complete representation of all possible feature combinations. In cat area 17 iso-orientation and ocular dominance domains are isotropic, but in cat area 18 orientation domains are highly elongated. Using optical imaging of intrinsic signals, we determined whether there is a corresponding change in the pattern of ocular dominance in area 18, and whether spatial relationships between orientation and ocular dominance maps are conserved across areas. We found that, despite the orientation map having domains that are elongated, ocular dominance domains in cat area 18 remained isotropic. Spatial relationships between orientation and ocular dominance maps were similar in both area 17 and 18: there was a tendency for contours of the two maps to intersect at near orthogonal angles, and there was a negative correlation between the gradients of the two maps. Further investigation demonstrated that the strength of each of these spatial relationships differed in local areas of cortex: in the region with the strongest orthogonal intersection relationship, the negative correlation between orientation and ocular dominance gradients was the weakest, while the region with near parallel contours demonstrated the strongest negative correlation between gradients. The differential distribution across cortex of each relationship may reflect the need for multiple strategies to achieve uniform coverage of features whose mapping patterns can be heterogeneous. Displacing one map artificially with respect to the other led to the deterioration of these two relationships. Thus precise relationships between these maps are conserved across area 17 and 18, even though the overall structure of the orientation map differs greatly between the two areas.
Support Contributed By: NIH grant EY07023

2002

Croisier, E., Leamey, C., Tonegawa, S. and Sur, M.
Novel gene transcripts expressed in developing sensory neocortex in the mouse.
Soc. Neurosci. Abst. 28, 2002.

The connective patterns of mammalian neocortical areas play a fundamental role in defining their characteristic functions. The mechanisms that underlie the development of these connectivity patterns are not well understood however. We are interested in identifying novel genes and gene products that may be involved in the development of neocortical areas. To do this we used a high-density DNA microarray to detect expressed sequence tags (ESTs) that are differentially expressed between the somatosensory and visual cortices in early postnatal mice. Total RNA was extracted from tissue removed from the presumptive somatosensory and visual cortices of postnatal day (P)0-1 Black 6 mice. Biotinylated cRNA was then produced via cDNA synthesis and in vitro transcription reactions. Four separately prepared samples from each area were hybridized to a microarray containing multiple oligos corresponding to 12,000 ESTs. In analysis of the results, we identified ESTs that fulfilled 3 criteria: apppropriate absence/presence call, difference call, and had a fold change greater than 2 for at least 3 of the 4 pairs of samples examined. We found 19 ESTs in which these criteria were met, 10 of which are more highly expressed in the visual cortex than in the somatosensory cortex, and 9 of which are more highly expressed in the somatosensory cortex than in the visual cortex. The differential expression of many of these genes between the cortical regions has been confirmed by semi-quantitative PCR. We are currently examining the temporal and areal expression patterns of the genes by in situ hybridization.
Supported by: NIH EY11512 and NS39022 to MS and MH58880 and NS32925 to ST.

Dragoi, V., Miller, E. and Sur, M.
Effect of reward expectation on response selectivity in monkey V.
Soc. Neurosci. Abst. 28, 2002.

Neurons in early visual cortex encode features of the sensory input. However, it is unknown whether extra-sensory inputs that exert influences on the behavioral state of the organism can also influence feature encoding. We asked whether changes in behavioral performance induced by the expectation of reward during an orientation discrimination task influence response selectivity in V1 of awake monkeys. Reward expectation affects responses in prefrontal cortex, raising the possibility that feedback projections to visual cortex would carry reward-related signals. We trained monkeys to discriminate orientations during a match-to-sample task in which V1 neurons were stimulated using gratings of random orientation briefly flashed after the presentation of a grating of fixed orientation. We varied the probability of reward at the end of each correct trial between 1 and 0.5. We found that low reward probability impairs orientation discrimination by broadening V1 orientation tuning curves, by reducing responses, and by decreasing the S/N ratio, whereas high reward probability has the opposite effect. To examine whether these changes in responses are indeed related to the expectation of reward and not to changes in attention, we increased the difficulty of the orientation discrimination task. A significant effect of reward expectation was still seen in the high-attention condition. These results indicate that reward-induced influences act to modulate V1 circuits mediating visual performance, and that this modulation is expressed through different channels than those involved in attentional effects.
Supported by: McDonnell-Pew (VD), MIT-Riken NS Ctr (EKM), and NIH (MS).

Ellsworth, C., Lyckman, A. and Sur, M.
Ephrin-A mediates eye-specific patterning of retinogeniculate projections in rewired mice.
Soc. Neurosci. Abst. 28, 2002.

In neonatal mice, deafferentation of the medial geniculate nucleus (MGN) induces its innervation by retinal ganglion cells. Previous work in mice demonstrated that these rewired visual projections show eye-specific segregation in the MGN. To examine factors that control the patterning of rewired projections, we injected red and green Alexa Fluor conjugated CtB into the right and left eyes, respectively, of rewired adult mice. Projections were examined using confocal microscopy. The rewired ipsilateral projection is biased toward the posterior quarter of the MGN. Ephrin-A2 and -A5 are expressed in a high anterolateral, low posteromedial gradient in the LGN and MGN. Ipsilateral projections, which arise from the ventro-temporal retina, innervate areas of low ephrin expression in the LGN. We rewired ephrin-A2/A5 double knockout mice and examined the eye-specific projections in the MGN. Rewiring is quantitatively enhanced in the double knockout. Moreover, 1) the proportion of ipsilateral projections in the MGN is preferentially enhanced compared to wildtype mice; and 2) there is an equal representation of ipsilateral projections in the anterior and posterior MGN. These findings suggest that ephrin ligands contribute to the patterning of visual projections in a novel target.
Supported by: March of Dimes and NIH EY11512.

Leamey, C., Wang, K., Tonegawa, S. and Sur, M.
Differential gene expression by specific sets of neo cortical neurons during development
in the mouse.
Soc. Neurosci. Abst. 28, 2002.

Little is known about how the specific sets of extrinsic and intrinsic connections that characterise the mammalian neocortex arise during development. We used high-density DNA microarrays to identify genes that are differentially expressed between the somatosensory and visual cortices in neonatal mice. The arrays used contain oligos corresponding to 36,000 genes and ESTs. Fify-two genes and ESTs consistently showed differential expression between somatosensory and visual areas. We are examining the expression patterns of these genes. Two have been studied in detail: Ten_m3 and BCL6. Ten_m3 encodes a type II transmembrane protein, it is highly expressed near the caudal pole of the cortex in a region corresponds well to visual cortex. Strikingly, expression is largely confined to layer V cells. To identify the cells which express Ten_m3, in situ hybridisation was combined with retrograde tracing. Tracer injections into the superior colliculus labelled cortical layer V neurons which expressed Ten_m3, whereas injections of tracer into the contralateral cortex labelled layer V neurons that did not express the gene. BCL6 is a transcription repressor. It is expressed in layer V neurons in the rostral 2/3 of cortex corresponding to the somatosensory and motor cortices but avoids the visual cortex. Retrograde tracing studies show that BCL6 is expressed in corticospinal neurons but not in callosal neurons. The expression patterns of these genes in specific populations of projection neurons suggest they may play a regulatory role during development of these projections.
Supported by: NIH EY11512 and NS39022 to MS and MH58880 and NS32925 to ST.

Lyckman, A. and Sur, M.
Visual activation of immediate early genes in rewired auditory cortex after early
induction of retinal projections into the auditory thalamus.
Soc. Neurosci. Abst. 28, 2002.

Neonatal ablation of the inferior colliculus, the predominant auditory input to the medial geniculate nucleus (MGN), causes retinal axons to innervate the MGN. This manipulation alters the flow of stimulus-driven activity to, and within, the developing cortex. To examine the interplay of genes, input activity, and plasticity in the cortex, we have asked: 1) Does visual stimulation activate immediate early gene transcription in rewired A1? 2) Is the pattern of visual activity-dependent stimulus transduction similar in rewired A1 versus normal V1?

Mice were rewired on the day of birth and reared to adulthood. Normal mice and rewired mice were dark-housed for 2 days, and exposed to 20 min or 2 h of strong visual stimulation without auditory stimulation. The brains were processed for immunohistochemistry of immediate early gene products, including c-fos, egr-1, p-Elk-1, p-c-Jun and p-CREB. Unilaterally rewired mice showed strong visual activation of c-fos that was stronger in rewired A1 than in normal A1. Bilaterally rewired mice showed strong, but asymmetric, activation of c-fos in right and left A1’s. The laminar pattern of c-fos activation in rewired A1 was similar to that in primary visual cortex (V1). Egr-1 expression was strong in V1 and rewired A1, showing a complex multilaminar pattern. p-CREB expression was uniformly strong and did not predict c-fos or egr-1 expression. p-c-Jun and p-Elk-1 patterns were weak. These data suggest that visual activity can regulate the expression of certain immediate early genes in rewired A1, and in a manner similar to that in V1.
Supported by: NIH grant NS39022.

Majewska, A. and Sur, M.
Motility of dendritic spines in the developing mouse visual cortex in vivo.
Soc. Neurosci. Abst. 28, 2002.

Dendritic spines are the postsynaptic elements of most excitatory synapses in the mammalian cortex. Spine morphology allows the restriction of biochemical environments to single synapses, allowing synapse-specific regulation. Spine shape, however, is highly dynamic. Morphological changes in spines are developmentally regulated and sensitive to sensory deprivation. This suggests that spine motility may be important for synaptic plasticity and circuit development. We have examined spine motility in vivo in the visual cortex of transgenic mice expressing GFP. Spines on apical dendrites of layer 5 neurons were imaged through a craniotomy using a two-photon microscope in anaesthetized animals. To examine the effect of age and sensory experience, spine shape was monitored by time lapse imaging for a period of 2 hours. At the ages we studied (P21-P42), we observed significant changes in spine shape on a time scale of tens of minutes. At these ages most of the dendritic protrusions have spine-like morphologies, although some filopodia are also observed. We did not observe significant changes in spine number, i.e., the appearance of new spines or the disappearance of existing spines, during the imaging period. We are also examining the effects of binocular visual deprivation on spine motility during the critical period for ocular dominance plasticity in the visual cortex. Our data to date suggest that spines in visually deprived animals also undergo substantial motility.
Supported by: Whiteman Fellowship (AM) and NIH Grant EY11512.

Newton, J., Ellsworth, C., Miyakawa, T., Tonegawa, S. and Sur, M.
Retinal axons directed to the auditory pathway accelerate visual cued fear conditioning
in mice.
Soc. Neurosci. Abst. 28, 2002.

Retinal axons can be directed to the auditory thalamus by neonatal manipulations, and provide visual driving of cells in auditory thalamus and cortex. However, the extent to which these novel inputs influence behavior is largely unknown. Mice express fear to a tone presented alone after a few tone-shock pairings whereas they need many more light-shock pairings before they express fear to a light alone (Heldt et al., 2000). The current study explores whether visual inputs routed to the auditory pathway influence fear conditioning in rewired mice. Retinal axons were induced to project to the MGN of the thalamus through bilateral ablation of the inferior colliculus at p0. As adults (> 8 weeks) the mice underwent 3 sessions of fear conditioning and behavioral testing. A conditioning session consisted of 10 minutes of habituation followed by 3 cue-shock pairings (30 sec ISI). The cue (65 dB noise or 1 Hz flickering diodes) was presented for 5 seconds co-terminating with a shock (2 sec, 0.3 mA). Freezing was then measured in either the conditioning chamber (contextual fear, 24 h) or an altered context chamber (cued fear, 48 h). Rewired mice froze more during the light presentation in the altered context chamber after just one conditioning session whereas most sham lesion mice required several sessions. Contextual fear was comparable for both groups of mice. Our results indicate that retinal projections directed to the auditory thalamus accelerate cued fear conditioning to a visual stimulus and that this pathway conveys novel information capable of mediating behavior.
Supported by: NIH EY13900-01 (JRN), EY11512 (MS), RIKEN (TM,ST) and HHMI (ST).

Oray, S., Majewska, A, and Sur, M.
Glutamate receptor expression correlates with dendritic spine motility in layer 5 cortical neurons.
Soc. Neurosci. Abst. 28, 2002.

Most excitatory synapses in the mammalian cortex are made onto dendritic spines. Even though the functional properties of spines are not completely understood, it is increasingly clear that they are important for the compartmentalization of post-synaptic signals and that their dynamic morphology has functional consequences. It has been shown that spine motility is regulated developmentally and through the activation of glutamatergic receptors. This suggests that the motility of spines may play an important role in regulating synaptic efficacy and may be correlated with the presence of AMPA and NMDA receptors. To examine this issue, we used transgenic mice expressing GFP in a subset of layer 5 pyramidal neurons. Using two-photon microscopy, neurons from developing mice were imaged in cortical slices and spines were analyzed for motility. In addition, by fixing and resectioning the slices after imaging, we were able to perform standard immunolabeling to quantify the expression of AMPA and NMDA receptors in these neurons. Using the unique morphologies of the imaged neurons, it was possible to find the same spines which had been imaged for motility after fixation and to assess the degree of AMPA and NMDA receptor expression on them. In individual spines, we found a strong correlation between motility and NMDA receptor expression, very little correlation between motility and AMPA receptor expression, and a very strong, positive correlation between AMPA and NMDA receptor expression. These results indicate that glutamatergic synapses may be involved with dynamic structural changes in dendritic spines.
Supported by: NIH Grant EY 11512 and Whiteman Fellowship (AM).

Schummers, J. and Sur, M.
Reverse-correlation analysis of orientation selectivity in cat area 17.
Soc. Neurosci. Abst. 28, 2002.

We have used the reverse-correlation procedure to characterize the responses of neurons in cat area 17 to a rapid sequence of randomly oriented gratings. In general, neurons demonstrated simple dynamics, with responses beginning typically between 30 and 40 msec, reaching peak selectivity between 45-55 msec, and relaxing back to baseline. A significant portion of neurons did not relax smoothly to baseline, but rather showed an inflexion point, or a second, smaller peak in selectivity between latencies of 80-120 msec. Thus, some neurons were selective for latencies up to 140 msec, while others were only selective up to ~60 msec. This behavior may reflect sustained network activity or an excitatory rebound from inhibition. Large shifts in the preferred orientation during the response were extremely rare. As reported in the primate (Ringach et al., 2002) robust suppression of spike probability by non-optimal (usually orthogonal) orientations was common in our population. This suppression may be indicative of inhibitory inputs at non-preferred orientations. In a subset of experiments, both the cortical depth and position with regard to pinwheels in the optically imaged orientation map were recorded, allowing a 3-D characterization of the dynamics of cortical responses. Further analysis will be directed at detecting difference in the dynamics of tuning as a function of lamina and location within the orientation map. In particular, the heterogeneity in the orientation composition of the local circuit near pinwheel centers predicts that interesting dynamics may be pronounced in these locations.
Supported by: EY07023 to MS and an HHMI predoctoral fellowship to JS.

Sharma, J., Dragoi, V. and Sur, M.
Modulation of V1 responses by an internal model of stimulus location.
Soc. Neurosci. Abst. 28, 2002.

Neurons in primary visual cortex (V1) are modulated by direction of gaze in monkeys trained to fixate at different locations in space (Sharma et al, SFN 1999). We have now examined the progressive dependence of responses on the sequential appearance of a fixation locus. Two monkeys were trained to fixate at one of three separate locations, while single unit recordings were made from V1 neurons in response to sinusoidal gratings presented in their receptive fields. In Condition 1, the spot appeared repeatedly at the same location for a number of trials. This resulted in progressively shorter latencies to achieve fixation as the trials progressed. No such systematic change in fixation latency was observed in Condition 2, where the fixation spot varied randomly from trial to trial. In a subset of neurons, there was a significant modulation in spike responses to gratings of the preferred orientation presented at a particular location in the sequential condition, compared to responses to gratings of the same orientation presented at the same location but in the random condition. Importantly, in the sequential condition, the magnitude of responses increased or decreased with each successive stimulus presentation, particularly in the early part of the sequence. The change in neuron responses paralleled the progressive shortening of fixation latencies in this condition. These data suggest that visuomotor behavior is guided by progressive acquisition of an internal model of stimulus location, and that cortical responses as early as V1 can be modulated by such internal constructs.
Supported by: NIH grant EY07023 and McDonnell_Pew fellowship(VD).

Yu, H.B., Farley, B. and Sur, M.
Spatial relationships between retinotopy and four other feature maps in ferret visual cortex.
Soc. Neurosci. Abst. 28, 2002.

Dept Brain & Cognitive Sci, MIT, Cambridge, MA, USA
In the primary visual cortex, spatial relationships between some combinations of stimulus feature maps have been described, while others are less well understood. In particular, how retinotopy, one of the most fundamental features mapped, relates to other maps has remained elusive. Using optical imaging of intrinsic signals, we have analyzed spatial relationships between functional maps of retinotopy, orientation, ocular dominance, spatial frequency, and direction. We find that the magnification factor of visual space is anisotropic in ferret primary visual cortex. Specifically, the representation of visual space is more compressed along the azimuth axis than along the elevation axis. This distorted magnification factor appears to be reflected in the orientation map: orientation fracture lines have a significant tendency to lie along the low magnification axis of retinotopy. Furthermore, orientation and retinotopic gradient angles tend to be orthogonal. These relationships suggest that along the axis where receptive field location changes more slowly, orientation changes more rapidly. Similarly, we find that high rate-of-change regions in the orientation map colocalize with low rate-of-change regions of ocular dominance and spatial frequency maps. However, most direction fracture lines overlap with high rate-of-change regions of the orientation map, except for some which bisect iso-orientation domains. The preferred direction of motion is orthogonal to the preferred orientation over most of cortex. It appears that the spatial layouts of each of these five functional maps, including the retinotopic map, are highly interdependent.
Supported by: NIH grant EY07023.

2001

Dragoi, V. and Sur, M.
Inhomogeneities in the structure of V1 orientation maps and their consequences for
cortical function.
Soc. Neurosci. Abst. 27: 619.18, 2001.

It is generally assumed that adult V1 is uniform in its capacity for plasticity, and that the response properties of neurons are relatively uniform within a given layer. Using a combination of intrinsic signal imaging, multiple-electrode recording, and visual psychophysics, we investigate here the effect of two types of inhomogeneities in the layout of V1 orientation maps: the inhomogeneous rate of change of orientation preference across the cortex (pinwheel centers vs. orientation domains) and the asymmetry in the size of orientation domains (cardinal vs. oblique domains). We used short-term adaptation to a stimulus of fixed orientation to induce changes in the tuning properties of neurons at selected cortical locations. We found a significant correlation between the orientation distribution of local inputs (<500um) and the adaptation-induced changes in preferred orientation and firing rate. These changes are pronounced when neurons integrate broadly tuned oriented inputs (recording site close to a pinwheel center or within an oblique orientation domain), whereas neurons retain stable response properties when the local orientation distribution is tuned (recording site in the middle of a cardinal orientation domain). We have tested the implications of these results for perception by demonstrating that human observers report different adaptation-induced changes in orientation tuning between cardinal and oblique axes. We suggest that preferential cortical locations for adaptive changes may be a strategy that the visual cortex employs to enable plasticity as well as stability in the face of adaptation to the statistics of natural images. Supported by McDonnell-Pew and Merck fellowships to V.D. and by NIH grant to M.S.

Ellsworth, C.A., Lyckman, A. and Sur, M.
Eye-specific patterning of retino-geniculate terminations in the medial geniculate
nucleus of rewired mice.
Soc. Neurosci. Abst., 27: 27.13, 2001.

Previous work in mice (Lyckman et al, Soc Neurosci Abstr 25:2263) has shown that denervation of the medial geniculate nucleus (MGN) at birth induces retinal fibers to innervate the MGN (rewiring). Rewiring offers a novel opportunity to explore the relative contributions of afferents and targets to patterning. We have examined whether eye-specific zones form in the MGN as they do in a normal thalamic target of retinal projections, the lateral geniculate nucleus (LGN). Cholera toxin B-subunit conjugated to AF-488 or to AF-594 was injected into the ipsilateral or contralateral eyes, respectively, of unilaterally rewired, adult mice. By confocal microscopy, retinal projections from the two eyes were found to occupy separate zones in the MGN. The contralateral eye contributed most of the retinal innervation of MGN, though in 4 of 5 cases the relative contribution from the ipsilateral eye was greater than that found in the normal LGN. The eye-specific segregation in the MGN of rewired mice resembles that in rewired ferrets (Angelucci et al, J Neurosci 17:2040). Eye-specific segregation suggests that afferent-dependent mechanisms are important to patterning. The targets, the MGN and LGN, appear to differ in their receptivity towards ipsilateral versus contralateral retinal innervation.
Supported by: March of Dimes & NIH EY11512

Leamey, C.A., Wang, K.H., Tonegawa, S. and Sur, M.
Differential gene expression between neo cortical areas in the developing mouse.
Soc. Neurosci. Abst. 27: 252.14, 2001.

Neocortical areas are characterised by their unique patterns of afferent, efferent and intrinsic connections. We are interested in understanding the mechanisms which underlie the development of these patterns of connectivity. To do this we have made use of a commercially available high density DNA microarray to identify genes which are differentially expressed between sensory neocortical areas in early postnatal mice. Tissue from the presumptive somatosensory and visual cortices of postnatal day (P)0-1 Black 6 mice was removed and total RNA was extracted. cDNA synthesis and in vitro transciption reactions were then performed to produce biotinylated cRNA. Three separately prepared pairs of samples were then hybridised to the microarray which contains multiple oligos corresponding to over 12,000 known genes and expressed sequence tags (ESTs). Results were analysed according to specific criteria which required that the genes pass an absence/presence call, a difference call and have a fold change of at least 2 for all 3 of the pairs of samples used. Using these criteria we have found 8 genes and ESTs which are expressed at higher levels in the visual cortex than in the somatosensory cortex, and 7 genes and ESTs which are more highly expressed in the somatosensory cortex than visual cortex. The differentially expressed genes include a protein tyrosine phosphatase receptor, a growth factor and a transcription factor. We are currently confirming the differential expression of these genes by hybridising additional samples to the microarray and by in situ hybridisation.
Supported by: NIH grant EY11512 and NS39022 to MS and MH58880 and NS32925 to ST

Oray, S., Leamey, C. and Sur, M.
Co regulation of AMPA and NMDA receptors during activity blockade in the developing ferret LGN.
Soc. Neurosci. Abst. 27: 27.14, 2001.

In the lateral geniculate nucleus (LGN) of the developing ferret, retinal axons segregate into eye specific laminae during the first two postnatal weeks and subsequently into ON/OFF sublaminae during the third and fourth postnatal weeks. During this period of sublamination, AMPA and NMDA receptors maintain stable electrophysiological properties as well as a constant AMPA to NMDA ratio (Hohnke, Oray, and Sur, J Neurosci 20:8051, 2000). However, on a short time scale, the expression of AMPA and NMDA receptors can be dramatically altered by activity blockade. To further investigate this short-term change in receptor expression, pharmacological manipulations of the NMDA receptor (NMDA and AP5) were performed. Receptor expression for GluR1 (AMPA) and NR1 (NMDA) subunits was quantified with immunocytochemistry and confocal laser microscopy. While there is significant variation in the effects of each drug, the degree to which AMPA and NMDA receptors were colocalized remained remarkably consistent irrespective of drug treatment. Additionally, both receptor subunits increased or decreased together in response to drug application. These effects were seen throughout the period of sublamination (postnatal days 14, 21 and 28). To specifically examine receptors at retinogeniculate synapses, the anterograde tracer cholera toxin subunit B (CTB) was injected into the eyes and CTB, NR1, and GluR1 was visualized in triple-labeling experiments. In these experiments, NR1 and GluR1 subunits were again coregulated regardless of drug treatment. These data argue that the expression of AMPA and NMDA receptors throughout the LGN, and at retinogeniculate terminals in particular, is tightly coupled.
Supported by: NIH Grant EY 11512

Sharma, J., Dragoi, V. and Sur, M.
Temporal influences of the receptive field surround on center responses in primary visual cortex of behaving macaque.
Soc. Neurosci. Abst. 27: 619.20, 2001.

The responses of V1 neurons to stimuli presented within their classical receptive field are influenced by stimuli present outside the center. Furthermore, prior exposure to a stimulus in the surround affects the center response.We have previously reported that center responses are influenced in an orientation-specific manner by an adapting surround stimulus: an iso-oriented surround suppresses responses, whereas an orthogonal surround either facilitates or suppresses responses, though to a lesser extent (Sharma et al.,2000). We have now investigated the temporal effects of the adapting surround. We first determined orientation tuning and time course of response of a cell to drifting sinusoidal gratings(3 degree patch covering the CRF). Next, the surround was adapted by a grating for 1 second (10deg., center blocked out), followed by the center stimulus, with or without stimulus in the surround. In many cases, the suppressive effect of the surround was preceded by an initial facilitation, lasting up to 200 msec. In addition,in a subset of cells that we recorded (18/63),the presence of a surround influences the temporal modulation of the cell’s response by the drifting gratings. In general,the ratio of the spatial phase specific component (F1)to the spatial phase invariant component (F0) for these cells is reduced. The spatial phase selective responses are orientation specific and the surround modulation is pronounced in cells with high direction selectivity.Thus, the influence of the surround on the center can be quite specific in time and space, and the surround can cause a reduction not only in overall center responses but also in their temporal modulation.
Supported by: Supported by NIH grants EY07023 and NS39022

Schummers, J., Marino, J. and Sur, M.
Orientation tuning of intracellular potentials and spike responses at pinwheel centers and iso-orientation domains in primary visual cortex.
Soc. Neurosci. Abst. 27: 619.19, 2001.

The representation of orientation across the primary visual cortex is inhomogeneous. That is, the orientation map contains foci in which neurons preferring many different orientation are situated in close proximity (pinwheel centers) as well as wider regions where neurons share similar orientation preference (iso-orientation domains). Suprisingly, neurons located in pinwheel centers are sharply tuned for orientation despite the broad orientation representation surrounding them. However, there is no evidence that the dense network of local connections respects the inhomogeneities in the orientation representation.
We have investigated whether the membrane potential responses to oriented stimuli, which should reveal synaptic inputs from the local network, differ between neurons in pinwheel centers and iso-orientation domains. We have found substantial variability in the orientation selectivity of the synaptic response, regardless of map location. Many neurons show significant depolarization in response to only a narrow range of stimulus orientations. However, many neurons respond with robust subthreshold activity to all orientations, and their spiking tuning curve is dramatically sharpened by an iceberg effect. We have detected a tendency for the latter class of neurons to be found near pinwheel centers, and for the former to be found in iso-orientation domains. We conclude that the orientation representation in the local region of the map may profoundly influence the tuning properties of synaptic inputs to a neuron.
Supported by: NIH grants EY07023 & NS39022;JS is an HHMI predoctoral fellow

Yu, H.B., Farley, B., Sharma, J. and Sur, M.
Relationship between multiple stimulus feature maps in ferret visual cortex.
Soc. Neurosci. Abst. 27: 619.17, 2001.

We used optical imaging of intrinsic signals in ferret visual cortex to simultaneously examine stimulus feature maps of ocular dominance, orientation, spatial frequency and direction. Visual cortex was segregated into unusually broad ocular dominance bands that tended to be elongated mediolaterally. The particular shape and extent of these bands varied appreciably between animals (cf White et al., J. Neurosci. 19:7089, 1999). Within each eye’s ocular dominance band, multiple orientation pinwheel centers were represented. Iso-orientation lines tended to cross the borders of ocular dominance domains at right angles, and orientation vector magnitude fracture lines did not run along ocular dominance borders. This suggests that the pattern of ocular dominace bands does not disrupt the continuity of orientation representation. Cortex activated by high spatial frequency appeared restricted to near the occipital pole, whereas lower spatial frequencies caused activation to spread rostrally. Within iso-orientation domains, activation regions elicited by opposite directions tended to be displaced, and regions representing different spatial frequencies appeared shifted. Our results indicate both local and large-scale regulation of feature maps in ferret visual cortex. Orientation, direction and spatial frequency appear to be continuously mapped over local cortical regions. The segregation of cortex into broad ocular dominance bands requires that coverage of visual space by the two eyes include large regions of cortex. Yet, the regular relationship between ocular dominance borders and orientation columns suggests a uniform mapping rule linking the two and possibly other columnar systems.
Supported by: NIH grant 000011284

2000

Dragoi, V., Miller, E.K. and Sur, M.
Reward-induced changes in orientation tuning in monkey primary visual cortex.
Soc. Neurosci. Abst. 26: 408.6, 2000.

Visual performance is enhanced by increasing response-contingent reward probability and impaired when the reward probability decreases. These changes in the expectation of reward are correlated with changes in the response of neurons in higher cortical areas. Given the massive feedback projection to the primary visual cortex (V1), we have investigated whether signaled changes in reward probability can alter performance during orientation discrimination while simultaneously modulating the orientation tuning of V1 neurons. One monkey was trained to maintain fixation for 2 s during an orientation discrimination delayed match-to-sample task while V1 neurons were stimulated using high-contrast sinewave gratings of random orientation (flashed for 500 ms after the presentation of a 1-s grating of fixed orientation). Reward was given at the end of each successful fixation followed by a correct response; reward probability (p) was either 1 or < 1, and in the partially reinforced trials was reduced gradually to 0.5 at the end of training. We derived orientation tuning curves psychophysically for the monkey and physiologically for V1 neurons. During the partially reinforced trials, psychophysical performance (percent correct discriminations) dropped from 91 to 79% for a 0° discrimination and from 88 to 72% for a 90° discrimination. From a total of 31 orientation selective V1 neurons recorded with p stabilized at 0.5, we found that 23 reduced their orientation selectivity index (OSI) by at least 20%, whereas only 2 showed an increase in their OSI by at most 10%. A temporal analysis of responses in 5 neurons showing strong reward-induced effects revealed that orientation tuning emerges 40-50 ms after stimulus presentation, remains stable for about 20 ms, and then broadens on both flanks of the tuning curve. We suggest that the initially sharp orientation tuning is due to orientation-specific feedforward or local cortical mechanisms. Subsequently, cortical feedback from higher cortical areas provides nonspecific inputs that broaden orientation selectivity. These results indicate that reward-induced top-down influences act to modulate V1 circuits that mediate visual performance.
Supported by: NIH Grant EY07023 (MS) and MIT-Riken NS Ctr. (EKM)

Oray, S., Leamey, C., and Sur, M.
Rapid regulation of AMPA and NMDA receptors in the ferret LGN during the
development of on/off sublamination.
Soc. Neurosci. Abst. 26: 119.10, 2000.

Retinal axons innervating the ferret lateral geniculate nucleus (LGN) segregate into ON/OFF sublaminae during the third and fourth postnatal weeks. Over this two week period, AMPA and NMDA receptors maintain stable electrophysiological properties as well as a constant AMPA to NMDA ratio (Hohnke et al., 2000). However, this long-term stability may conceal short-term changes in receptors underlying the sublamination process. To investigate this possibility, pharmacological manipulations of the NMDA receptor were performed to assess their effect on AMPA and NMDA receptor expression. After brief application of AP5 or NMDA, antibodies against the GluR1 (AMPA) and NR1 (NMDA) receptor subunits were used to asses the degree of receptor expression with scanning laser confocal microscopy. Early in sublamination (postnatal day 14), NMDA treatment showed rapid regulation of GluR1 and NR1 expression: NMDA increased receptor expression relative to control and AP5 conditions. Most notably, the NR1 and GluR1 expression covaried, suggesting that expression of NR1 and GluR1 was tightly coupled on this short time scale. After sublamination is complete (postnatal day 28), treatment with AP5 or NMDA had minimal impact upon receptor expression. At both ages, the colocalization of NR1 and GluR1 subunits remained relatively constant, indicating that the up- or down-regulation of receptors occurred together in close spatial proximity. These results suggest that the expression of AMPA and NMDA receptors can be rapidly regulated on a short time scale during the period of ON/OFF sublamination by activity of the NMDA receptor, and that this effect diminishes by the end of sublamination.
Supported by: NIH Grant EY 11512.

Lyckman, A., G. Fan, Rios,M., Jaenisch, R. and Sur, M.
The role of BDNF in eye-specific patterning of visual connections.
Soc. Neurosci. Abst., 26: 406.7, 2000.

Refinement of eye-specific segregation of visual connections is thought to involve activity-dependent retraction of overlapping (binocular) inputs. Remodeling of the terminal arbors of retinal fibers in the lateral geniculate nucleus (LGN) may require brain-derived neurotrophic factor (BDNF), a neurotrophin whose expression is modulated by activity in the LGN and the visual cortex. We have asked whether eye-specific segregation of retinal inputs to the LGN occurs in the absence of BDNF using two transgenic mouse lines: BDNF-knockouts (KO), carrying a homozygous, germ-line deletion of the BDNF gene; and, brain-specific mutants (BM), in which the BDNF gene is deleted from daughter cells of CNS progenitor cells in mid-gestation. In wild type mice, eye-specific segregation in LGN is complete by P8. KO mice (examined at P16) and BM mice (examined at P11) received intraocular injections of cholera toxin B-subunit (CTB) coupled to fluorescein in one eye, and CTB coupled to tetramethylrhodamine in the other eye. The next day, mice were euthanized and transcardially perfused with 4% paraformaldehyde; vibratome sections of brain were examined by confocal microscopy. Projections from the two eyes were just as well segregated in the KO and BM mice as in their litter mate controls or wild type mice. No additional overlap between axons from the two eyes was detected in the BDNF-deficient mice. These data suggest that BDNF may not be required for refinement of eye-specific segregation in the LGN. Whether binocular overlap of thalamocortical projections is altered in BDNF-deficient mice is under examination.
Supported by: EY11512 (MS) & R35 CA44339 (RJ).

Rivadulla, C. and Sur, M.
Contribution of corticocortical connections to the generation of orientation maps in V1.
Soc. Neurosci. Abst. 26: 53.11, 2000.

Primary visual cortex is considered to receive driving inputs from the LGN and modulatory inputs from corticocortical connections. To examine the contribution of these systems to the generation of cortical properties such as orientation selectivity, we performed optical imaging of intrinsic signals in V1 of adult cats, combined with extracellular recording and pharmacological blockade in the LGN. Seven barrel pipettes were used for extracellular recording and ejection of GABA in the A layer of the LGN. Tungsten electrodes were attached to the pipette in order to ascertain the extent of blockade. Orientation maps were imaged through the contralateral eye. Extracellular recordings were also made in V1 in order to identify the region where receptive fields overlapped those recorded in the LGN. In one set of experiments, visual stimuli were restricted to the area blocked in LGN layer A, so that layer C provided the only input to V1. Blockade of the LGN totally abolished cortical maps. Thus, layer C inputs, by themselves, appear unable to generate orientation maps in V1. In a different set of experiments, we used full field stimulation so that the cortical area affected by the blockade could receive information from other cortical regions. The maps obtained during LGN blockade (as wide as 800 um) were indistinguishable in their orientation structure from control maps. Thus, intracortical connections can provide substantial input and build a cortical map that is similar to normal. The fact that a restricted stimulus could not evoke a map during blockade shows that the map is constructed through orientation-specific corticocortical connections by an active filling-in process
Supported by: NIH grant EY07023

Schummers, J. and Sur, M.
Rules of functional connectivity in primary visual cortex.
Soc. Neurosci. Abst. 26: 53.12, 2000.

The map of orientation in V1 is characterized by domains in which orientation preference changes gradually, punctuated by singularities at which orientation preference changes rapidly. Neurons with similar orientation preferences tend to show correlated firing, even when separated by distances up to several mm. We have investigated whether local functional connectivity spans a wider range of orientation preferences than does long-range connectivity, particularly near orientation singularities, and hence the rules for short and long-range connections in V1. Using a combination of optical imaging and single-unit recordings, we have examined the stimulus-dependence of correlated firing between pairs of neurons (n = 143 pairs) as a function of their location in the orientation map. Nearby neurons (<700mm apart) with widely differing orientation preferences show correlated firing; correlations do not only occur between neurons with similar orientation selectively, as is seen with neurons further apart. Furthermore, the stimulus selectivity of correlated firing is strongly dependent on map location. Using a simple model of the orientation selectivity of local inputs to neurons, derived from the orientation map, we can predict the stimulus specificity of the correlated firing for a neuron pair. These data indicate that the lateral input to V1 neurons arises from two distinct sources: spatially homogenous local connections, and patchy, orientation specific long-range connections. Furthermore, the local connections do not respect features of the orientation map, and thus neurons can receive homogeneously or inhomogeneously oriented inputs, depending on their location in the orientation map.
Supported by: EY 07023.

Sharma, J., Dragoi, V. and Sur, M.
Dynamics of center-surround interactions in alert macaque V1.
Soc. Neurosci. Abst. 26: 53.10, 2000.

The response of neurons in primary visual cortex to the stimuli presented within their classical receptive field (CRF)can be modulated by stimuli presented well outside the CRF.This contextual modulation depends on interactions between the orientation and contrast of center and surround stimuli.We have previously reported that prior exposure to oriented stimuli alters the orientation specificity and response magnitude of V1 neurons.We have now investigated the effect of oriented high-contrast sinusoidal gratings presented in the surround on neuronal response to gratings presented within the CRF of cells in monkey V1.Initially, the orientation tuning and time course of response for each cell was determined by presenting gratings within a 3deg patch centered on the CRF.Next,surround stimuli were presented in a 10deg window without the central patch for a period of 1sec, followed by gratings within the central patch.Following such short-term surround adaptation,most cells show a shift in the preferred orientation of CRF responses, majority of them by >10degs, these include repulsive and attractive changes in orientation preference. Neuronal responses to CRF stimuli are in general suppressed under the influence of an adapting surround; interestingly, in some cases,there is an initial facilitation followed by suppression.Such facilitatory effects are more common in cells that show small shifts in orientation preference.These results suggest a dynamic modification in responses of V1 neurons as a result of interactions between the previous history of surround stimulation and the current stimulus within the receptive field of the cell.
Supported by: NIH grant EY07023