Exp Brain Res (1991) 85:235-239

Experimental Brain Research �9 Springer-Verlag 1991

Research Note

The contribution of GABA-ergic neurons to horizontal intrinsic connections in upper layers of the cat's striate cortex

K. Albus 1, P. Wahle 1, J. Liibke 1, and C. Matute z

1 Department of Neurobiology, Max-Planck-Institute of Biophysical Chemistry, W-3400 G6ttingen, Federal Republic of Germany z Department of Neurosciences, Faculty of Medicine, University of Pals Vasco, E-489340 Leioa, Spain

Received July 10, 1990 / Accepted December 31, 1990

Summary. The contribution of neurons containing Y- aminobutyric acid (GABA) to horizontal intrinsic projections in layers I - I I I of cat's striate cortex was investigated by combining GABA-immunohistochemis- try with axonal tracing. After intracortical injections of Rhodamine-labelled latex microspheres Rhodamine- labelled neurons form patch- or bandlike aggregations (clusters) separated from each other by regions contain- ing fewer, evenly distributed or no labelled neurons. Of the Rhodamine-labelled neurons about 5% display GABA-immunoreactive material (double labelled = DL- neurons). Approximately 70% of the DL-neurons occur at distances of less than 1 mm, and the remaining 30% at distances between 1 mm and 2.5 mm from the injec- tion. About 60 % of the DL-neurons reside within clusters and 40% are located in regions between clusters; the respective percentages of the Rhodamine labelled GABA- negative neurons are about 85 and 15. Considering their small number and their spatial distribution inhibitory interneurons seem to make only minor contributions to the clustered pattern of intrinsic connections. Our results demonstrate that the topographical organization of neurons giving origin to lateral inhibitory interactions in upper layers of cat's striate cortex is different from that of neurons mediating excitatory functions.

Key words: Retrograde t r a c i n g - GABA-immunohis- tochemistry - Double labelled neurons - Horizontal in- trinsic connections - Cat area 17

Controversial results have been published concerning the types of neurons contributing to clustered intrinsic con- nections in the cat's striate cortex. In area 18, a few

Offprint requests to: K. Albus (address see above)

GABA-positive neurons have been found in cell clusters as far as 1 mm away from the injection (Matsubara 1988). In another report (LeVay 1988), GABA-positive neurons were seen only in the fringes of the injection zone and did not contribute to more distant clusters. In the present report we demonstrate that in upper layers of area 17 of the cat the GABA-positive and thus inhibitory interneurons labelled by a small injection of tracer occur in both cluster and intercluster regions and are arranged rather in a diffuse than clusterlike fashion. In addition our findings indicate that the distances over which lateral inhibitory interactions may occur are greater (up to 2.5 mm) than previously reported.

The results are based on 9 intracortical tracer injec- tions into area 17 of 5 cats; of these 4 injections have been quantitatively evaluated so far. Injections were placed into the medial part of area 17 by slightly retracting one hemisphere and inserting a glass pipette (tip diameter 15 gm-30 gm) in the contralateral hemisphere through small holes cut into the falx cerebri. Pressure injections of Rhodamine-labelled latex microspheres (Luma Fluor, N.Y.) were made 200-700 gm deep into cortex. After survival times between 2-10 days, animals were perfused with 0.9% NaC1 in 50 m M sodium phosphate buffer pH 7.4, followed by a mixture of 4% paraformaldehyde and 0.1-0.2 % glutaraldehyde and finally by 4 % parafor- maldehyde in 100 mM sodium phosphate buffer at pH 7.4. Immunohistochemistry. GABA-immunohistochemis- try was carried out on 60 gm thick parasagittal vibra- tome sections. Sections were rinsed in TBS (50 mM Tris-HCL, pH 7.4, 150 mM NaC1, pH 7.4) treated with 0.2% Triton X-100 in TBS for 1 h, rinsed, blocked in 5% bovine serum albumin (BSA, fraction V, Sigma; lh), and incubated overnight at room temperature in the mono- clonal ant i-GABA-antibody (dilution 1:1000 in 5% BSA, for specification see Matute and Streit, 1986). After several rinses in TBS, sections were processed with bioti-

236

nylated rabbit anti-mouse IgG (Dakopatts) diluted 1 : 300 in TBS for 3 h at room temperature, followed by an avidin-biotin-peroxidase complex used according to the manufacturers protocol. After several rinses, the peroxidase reaction product was developed with 0.02% diaminobenzidine in 50 mM Tris-HC1 buffer pH 7.6 and 0.001% H202. The sections were mounted on gelatinized slides, air dried, briefly passed through absolute ethanol and xylene, and coverslipped with Eukitt R. In situ hy- bridization. The 2.3 kilobase complementary DNA for glutamic acid decarboxylase, provided by Dr. A.J. Tobin (Kaufman et al. 1986) was inserted in antisense and sense orientation into transcription-vector pSP65. Full length antisense and sense riboprobes were transcribed from 1 gg template DNA (linearized with BamHI) in the presence of digoxigenin-UTP (Boehringer). Riboprobes were dissolved in 100 gl 10 mM Tris-HCL, 5 mM EDTA pH 7.5, 100 gg/ml yeast tRNA and 50 U/ml RNAsin (Pharmacia). For the double labelling 60 ~tm thick vibra- tome sections were. cut under sterile phospate buffer, collected in 2 x SSC and placed into prehybridization solution (50 % formamide, 250 gg/ml denaturated salmon sperm DNA, 100 gg/ml yeast tRNA, 50 mM sodium phosphate pH 6.5, 4 x SSC, 5% dextransulfate, 1 x Den-

hardts:0.02% polyvinylpyrrolidone, 0.02% BSA, 0.02% ficoll400) in 30-well glass plates. After prehybridization for 3 h at 42 ~ C the hybridization solution was applied (riboprobe diluted 1:100 to 1:400 in prehybridization solution). After overnight hybridization at 42 ~ C, sec- tions were rinsed in 2x SSC at room temperature, followed by three 15 min washes at 42 ~ C in 2 x SSC/50% formamide, 0.1 x SSC/50% formamide and 0.1 x SSC. Sections were equilibrated to TBS (100 mM Tris-HCL, 150 mM NaC1 pH 7.4) and hybrid molecules were de- tected with anti-digoxigenin-antibodies tagged with alkaline phosphatase diluted 1 : 1000 for 2 h (Boehringers nucleic acid detection kit). The blue alkaline phosphatase reaction product was developed with nitroblue tetra- zolium and 5-bromo-4-chloro-3-indolyl-phosphate (0.35 mg/ml and 0.18 mg/ml, respectively), in TBS-Mg buffer pH 9.5.

Sections were examined under Rhodamine epifluo- rescence to visualize the Rhodamine beads and under brightfield for either the brown peroxidase reaction product indicating the presence of GABA or the blue reaction product indicating the presence of GAD mRNA. The number of single labelled (containing only Rhodamine beads) and double labelled (containing both

Fig. 1A-F. Horizontal projections of GABA-ergic neurons in cat's striate cor- tex. A-C Photomicrographs of neurons containing Rhodamine beads or G A D mRNA, or both. D-F Photomicrographs of neurons containing Rhodamine beads or GABA-like immunoreactiv- ity, or both. A, F) bright- field, B, E) epifluorescence and weak brightfield, C, D) epilluorescence. Double labelled neurons are mark- ed by arrows. Stars in- dicate blood vessels. Scale: 25 gm

237

GABA-immunoreactivity and Rhodamine beads) neu- rons were counted by using an object micrometer grid with a sidelength of 150 gm. The sections were scanned at a magnification of xl00 and the number and location of labelled neurons were mapped by using a X/Y plotter connected to the stage of the microscope.

Inhibitory neurons are characterized by the presence of GABA-immunoreactive material and the expression of glutamic acid decarboxylase (GAD), the GABA syn- thesizing enzyme. As expected they contain the mRNA for GAD. When probed with either antisense riboprobes specific for GAD mRNA (Fig. 1AC) or anti-GABA antibodies (Fig. 1D-F) some of the Rhodamin labelled neurons were found to contain GABA and/or GAD mRNA which marks them as being inhibitory neurons. DL-neurons of various soma sizes were observed. Small- er (Fig. 1A-C, D F bottom left) and larger (Fig. 1D F top right) neurons occur closer to the injection site whereas DL-neurons seen further away (more than 500 gm distance) generally have larger somata with diameters of about 20 gm. DL-neurons with larger somata were also found when another marker for inhibitory neurons was used, the plant lectin from Vieia Villosa (data not shown). It specifically binds to a GalNac-epitope present on certain types of GABA-ergic cells, among them the large basket neurons (Naegele and Katz 1990). Our find- ing that these DL-neurons are located as far as 2 mm from the injection supports the view that the large basket neurons participate in rather widespread lateral inhib- itory interactions (Somogyi et al. 1983). The morphology of the DL-neurons with smaller somata has to be deter- mined; they could correspond to interneurons with long horizontal axon collaterals as characterized by Meyer and Wahle (1988) and Wahle and Meyer (1987).

An injection of 200 nl Rhodamine beads into the upper layers (I III) of cat's area 17 reveals the charac- teristic distribution of retrogradely labelled neurons in a plane parallel to the cortical surface (Fig. 2A). Rhoda- mine-labelled neurons form patch-or bandlike aggrega- tions separated from each other by regions containing either a few evenly distributed or no labelled neurons. The cell density within clusters as well as the size of clusters decreases with increasing distance from the injec- tion. In the case shown in Fig. 2 6048 neurons were labelled with Rhodamine and 223 of these cells (3.7%) contained GABA-immunoreactive material (13 sections evaluated). In other cases (not shown) the respective proportions of DL-neurons ranged between 3 % and 8 %. Figure 2 compares in one representative section the spa- tial distribution of Rhodamin-labelled GABA-negative neurons (Fig. 2A) with that of DL-neurons (Fig. 2B). As can be seen, the single labelled neurons clearly form clusters but the DL-neurons obviously do not do so. In addition, DL-neurons seem to be more concentrated around the injection than the single labelled neurons. Both observations are corroborated by combining the data in layers I-III from all sections of the case shown in Fig. 2 as well as considering the data from other cases not shown here. In the case shown in Fig. 2, at a distance of more than 160 gm from the injection approximately 85 % of the single labelled neurons reside in clusters and

the remaining 15% are located in between clusters. The respective numbers for the DL-neurons are 60% and 40%, suggesting only a weak tendency if at all, of the DL-neurons to associate with clusters. Considering all labelled neurons from the thirteen sections evaluated in the case shown in Fig. 2 it was found that 49 % of the DL-neurons but only about 30% of the single labelled neurons occur at a distance of less than 660 gm from the injection; for locations at distances between 660 gm and 1320 I~m, and 1320 gm and 1980 gm, the proportions are 36% and 12% for the DL-neurons, and 45% and 21% for the single labelled neurons. Four percent of the single labelled and about 3 % of the DL-neurons were located more than 2 mm from the injection; the largest distance from the injection measured was 3.5 mm for a single labelled and 2.5 mm for a double labelled neuron.

Our finding that the GABA-positive neurons labelled by an intracortical injection of a tracer are arranged only loosely in a clusterlike fashion is in accordance with observations that most presumed inhibitory interneurons lack the clustered axonal projections characteristic for many pyramidal neurons (Gilbert and Wiesel 1983 ; Mey- er 1983; Somogyi et al. 1983; but see Martin et al. 1983). On the other hand, our findings concerning the lateral spread of GABA-ergic axons differ from some of the earlier reports which have found the axons of most presumed inhibitory neurons to spread only a few hun- dreds of microns away from the soma (Meyer 1983). An exception are the large multipolar and basket neurons found in layers 2/3 and 5/6 which possess axons occa- sionally projecting as far as 1 mm-1.5 mm away from the soma (Somogyi et al. 1983; Meyer and Ferres-Torres 1984; Wahle and Meyer 1987; Meyer and Wahle 1988). One would expect on the basis of these studies that an. intracortical injection does not label significant numbers of GABA-positive neurons at distances of more than 1 mm. In our cases however, a substantial proportion (about 30 %) of all labelled GABA-positive neurons were located at distances between 1 mm and 2.5 mm from the injection. The shorter range of axonal projections as seen in the earlier morphological studies is probably due to incomplete impregnation (Golgi technique) and incom- plete filling and/or reconstruction (intracellular injection technique) because a study using anterograde degenera- tion and EM-analysis has reported ranges of horizontal inhibitory projections which are very similar to our data. For example, 31% of degenerating symmetrical axonal terminals were located at distances between 1 mm and 2 mm and about 12% at distances between 2 mm and 3 mm from a cortical lesion (Fisken et al. 1975).

A number of physiological studies have emphasized the importance of inhibitory lateral interactions for generating and maintaining the functional properties of visual cortical neurons, in particular their orientation sensitivity. The strength and the mode of operation of the inhibitory interactions are, however, still a matter of dispute (Ferster and Koch 1987). Clustered horizontal connections run preferentially between regions of similar orientation preference (Gilbert and Wiesel 1989). Since we have found that at least half of the inhibitory inter- neurons are associated with clusters it is likely that a

A �9

O e e � 9 �9

" 0 - �9 �9 �9 �9 �9 �9 �9 �9 �9 �9

O

e~

. 1

�9 2 - 4

�9 5 - 9

o 1 0 - 1 4

O 3 5 - 1 9

B

Q �9 e 0

o

Fig. 2A, B. Horizontal arrangement of intrinsic projection neurons in upper layers of cat's striate cortex as shown in one representative parasagittal section (60 Ixm). A Rhodamine-labelled neurons. The number of neurons counted per square (see Methods) is in~'eatecl by dots of different sizes. The code is given at the right hand side of the figure. Dots surrounded by thin lines are cell clusters defined on the basis of two criteria: a) a minimum of 6 neurons in 1-3 squares adjacent to each other and b) a minimum difference in cell density between cluster and non-cluster region, of at least 7(1% at three sites (e.g. anterior, dorsal and ventral) and of 313% at the remaining fourth side (e.g. posterior) of a cluster border. B Rhoda- mine-labelled GABA-immunoreact ive neurons. Same section as in A. Each dot marks the actual position of one DL-neuron in the

Irnm

zSL

_ 0 a n t .

section. The regions demarcated by the thin lines are the cluster regions defined in A. The position of cell clusters and DL-neurons in the striate cortex is shown by the insert at the right hand side of the figure. The neurons labelled with Rhodamine and either G A D m R N A or Vicia Villosa lectin were found at distances up to 2 mm from the injection; the histoehemical procedure in these preliminary experiments, however, seemed to have caused some loss of Rhoda- mine beads. Cortical layers were determined on the basis of Nissl stains. The injection center is marked by a white star; neurons counted within 160 ixm from the injection center were not con- sidered for quantitative evaluation. Abbreviations, LG: lateral gy- ms ; SSPS" suprasplenial sulcus; S SPG: suprasplenial gyrus; dors: dorsal; ant: anterior

239

significant c o m p o n e n t of the intr insic inh ib i tory connec- t ions each n e u r o n receives has the same or ien ta t ion tun ing characteristics as its intr insic excitatory inputs ( i so-or ienta t ion inhibi t ion) .

Our f indings of the ana tomica l o rgan iza t ion of G A - BA-ergic connec t ions so far suggest an or ien ta t ion tun ing of inh ib i t ion in the striate cortex with a pre- ponderance of i soor ien ta t ion inh ib i t ion ; oblique- and crossor ienta t ion inh ib i t ion as well as o r ien ta t ion un- specific inh ib i t ion are no t excluded on the basis of our findings, however. Physiological evidence for all these types of inh ib i tory o r i en ta t ion t un ing has been presented (Creutzfeldt et al. 1974; Sillito et al. 1980; Fers ter 1986; Bonds 1989; Albus and Baumfa lk 1989).

Acknowledgement. This work received financial support from the Deutsche Forschungsgemeinschaft (SPP Dynamik und Stabilisie- rung neuronaler Strukturen).

References

Albus K, Baumfalk U (1989) Bicuculline induced changes in ex- citability and orientation selectivity of striate cortical neurons. Soc Neurosci Abstr 15 : 324

Bonds AB (1989) Role of inhibition in the specification of orienta- tion selectivity of cells in the cat striate cortex. Vis Neurosci 2: 41-55

Creutzfeldt OD, Kuhnt U, Benevento LA (1974) An intracellular analysis of visual cortical neurons to moving stimuli: responses in a cooperative neuronal network. Exp Brain Res 21 : 251-274

Ferster D (1986) Orientation selectivity of synaptic potentials in neurons of the cat primary visual cortex. J Neurosci 6:1284-1301

Ferster D, Koch C (1987) Neuronal connections underlying orientation selectivity in the cat visual cortex. TINS 10:487-492

Fisken RA, Garey L J, Powell TPS (1975) The intrinsic, association and commissural connections of area 17 of the visual cortex. Proc R Soc Lond 272:48%536

Gilbert CD, Wiesel TN (1983) Clustered intrinsic connections in cat visual cortex. J Neurosci 3 : 1116-1133

Gilbert CD, Wiesel TN (1989) Columnar specificity of intrinsic horizontal and corticocortical connections in the cat visual cor- tex. J Neurosci 9:2432-2442

Kaufman DL, McGinnis JF, Krieger NR, Tobin AJ (1986) Brain glutamate decarboxylase cloned in gt-l l : fusion protein produces y-aminobutyric acid. Science 232:1138-1140

LeVay S (1989) Patchy intrinsic projections in visual cortex, area 18, of the cat: morphological and immunocytochemical ev- idence for an excitatory function. J Comp Neurol 269: 265 274

Martin KAC, Somogyi P, Whitteridge D (1983) Physiological and morphological properties of identified basket cells in the cat's visual cortex. Exp Brain Res 50:193-200

Matsubara JA (1988) Local, horizontal connections within area 18 of the cat. Progr Brain Res 75:265-274

Matute C, Streit P (1986) Monoclonal antibodies demonstrating GABA-like immunoreactivity. Histochemistry 86: 142157

Meyer G (1983) Axonal patterns and topography of short axon neurons in visual areas 17, 18, and 19 of the cat. J Comp Neurol 220 : 405M38

Meyer G, Ferres-Torres R (1984) Postnatal maturation of nonpyra- midal neurons in the visual cortex of the cat. J Comp Neurol 228 : 226-244

Meyer G, Wahle P (1988) Early postnatal development of Cholecys- tokinin-immunoreactive structures in the visual cortex of the cat. J Comp Neurol 276:360-386

Naegele JR, Katz LC (1990) Cell surface molecules containing N-Acetylgalactosamine are associated with basket cells and neurogliaform cells in cat visual cortex. J Neurosci 10:540-557

Sillito AM, Kemp JA, Milson JA, Berardi N (1980) A reevalution of the mechanisms underlying simple cell orientation selectivity. Brain Res 194:517-520

WaNe P, Meyer G (1987) Morphology and quantitative changes of transient NPY-ir neuronal populations during early postnatal development of the cat visual cortex. J Comp Neurol 261 : 165-192

Somogyi P, Kisvarday ZF, Martin KAC, Whitteridge D (1983) Synaptic connections of morphologically identified and physio- logically characterized large basket cells in the striate cortex of the cat. Neuroscience 10: 261-294