Specific Patterns of Intrinsic Connections between

Specific Patterns of Intrinsic Connections between Representation Zones in the Rat Motor Cortex

The organization of intrinsic connections in rat motor cortex was studied by combining microstimulation and tract-tracing techniques. Maps of forelimb and vibrissaI movements were constructed from the distribution of cortical sites from which movements were evoked in response to intracortical microstimulation. Then, a sin­ gle injection of a fluorescent dextran was placed into either a vibrissal or a wrist representation zone, or into a region bordering these zones, resulting in anterograde labeling of long intrinsic, horizontal axons. Following injection into the vibrissal area, axons were largely re­ stricted to the whisker representation zone and to the border region with the forelimb representation. Injec­ tions into a wrist zone labeled projections largely re­ stricted to the forelimb area and to the border with the vibrissal area. Injections into a border region labeled dense projections throughout most of the forelimb and vibrissal areas. These findings indicate that intrinsic axon collaterals in the motor cortex form specific and extensive connections among representation zones re­ lated to movements of the same body pan. These con­ nections may be involved in the coordination of activity in different representation zones for the execution of complex movement patterns. The projection of axon collaterals into border regions may be the anatomical substrate for the rapid reorganization of motor cortical maps that occurs following various experimental ma­ nipulations.

David S. Weiss and Asaf Keller Depanment of Anatomy and Cell Biology. and Program in Neuroscience, Uniformed Services University of the Health Sciences, Bethesda, Maryland 20814

The primary motor conex contains a somatotopic rep­ resentation of the major subdivisions of the body mus­ culature, such as the hindlimb. forelimb, and face_ Within each of these subdivisions, groups of neurons related to movements about individual joints are or­ ganized as discrete represemation zones (Humphrey, 1986; Asanuma, 1989; Lemon, 1990). Little is known about the mechanisms by which neurons in different representation zones interact to produce movement patterns that require temporal coordination of differ­ ent muscles. Furthermore, a single jOint may be ac­ tivated by several spatially segregated zones in the motor conex, and zones related to a panlcular group of muscles are intermixed with zones related to dif­ ferent muscles (Asanuma and Sakata, 1967;Jankowska et aI., 1975; Cheney and Fetz, 1985; Donoghue et aI., 1992; Keller. 1993b). The spatial separation of motor zones related to the same movement suggests that a mechanism exists by which these zones interact for the activation of their related muscles. Recent studies have provided evidence suggesting that the spatio­ temporal coordination of activity within individual representation zones in the motor conex may be me­ diated by intrinSiC, or intracortical, pathways (Huntley and Jones, 1991; Aroniadou and Keller, 1993; Keller, 1993a,b). Intrinsic connections have also been im­ plicated in the rapid reorganization of representation zones in the rat motor conex that occurs following nerve lesions (Donoghue et aI., 1990), and in re­ sponse to changes in motor activity (Donoghue and Sanes, 1991; Sanes et aI., 1992). Donoghue and col­ laborators (Jacobs and Donoghue, 1991;Jacobs et aI., 1991) have suggested that this pliability in the func­ tional organization of the motor cortex can be attrib­ uted to the existence of extensive intrinsic connec­ tions linking heterotypical representation zones. These connections are normally masked by inhibi­ tory, GABAergic processes, such that rapid reorgani­ zation of representation zones can occur by disinhi­ bition of these latent pathways (Jacobs and Donoghue, 1991). Intrinsic, horizontal connections have been dem­ onstrated to occur in different cortical areas of several species (e.g., Gilben, 1992). In the motor cortex, these intrinsic connections are formed by the long, hori­ zontal axon collaterals of pyramidal cells in layers JI­ VI (Jacobs et aI., 1991; Keller, 1993a; Keller and Asa­ numa, 1993). It is not known whether in the rat motor cortex these intrinsic connections form specific patC�r�br:ll Cnn�x Mar/Apr 1994;4:205-214; 1047-3211/94/'4.00

terns of connections among neurons located in spe­ cific represemation zones. The goal of the present study is to examine the hypothesis that intracortical connections provide a substrate for the coordination of movement patterns and for the pliability of motor cortical maps. For this purpose, representation zones of the mystacial vibrissa and forelimb movements were identified by the use of intracortical microstimulation, and the intrinsic connections between these repre· sentation zones were identified by neuroanatomical tract·tracing methods. A preliminary report of some of these findings has appeared elsewhere (Keller, 1993a).

Materials and Methods Surgical Procedures

Data were obtained from 13 adult Wister rats of both sexes, weighing between 167 and 480 gm (297 ± 23 gm, mean ± SO). Surgical procedures were con· ducted using sterile techniques. Rats were anesthe· tized with ketamine HCI (100 mg/kg, i.p.) and "y' lazine HCI (0.5-1.0 mg, Lm.). Dexamethasone (0.4 mg, Lm.) was administered to prevent cortical edema, and Bupiracaine HCI was applied locally to surgical wounds. In order [Q maintain constant levels of an· esthesia, ketamine was administered through an in· travenous c:mnula placed in the femoral vein 00-20 mg/hr). Body temperature was maintained at 37"C using a heating pad. The motor cortex was exposed through a craniot· omy, the dura was cut and reflected, and the exposed cortex was covered with warm mineral oil. The cis· terna magna was opened to prevent cortical pulsation and edema. Upon completion of the experimental procedures, the wounds were sutured and the animals were returned to their cages for a recovery period lasting 7-9 d.

Intracortlcal Microstimuilltion

The functional organization of the rat motor cortex was determined by monitoring movemems about in· dividual joims in response to intracortical microstim· ulation OCMS). Trains of stimulation (II cathodal pulses. 100 lISec pulse width, 30 msec duration) were delivered at a rate of 0.2-0.5 Hz through glass·insu· lated platinum-iridium microelectrodes (5-10 "m tip diameters, 0.7-2.0 MO). Electrode penetrations were made in a grid pattern at 500 "m steps, and the pen· etration sites were marked on a photograph of the cortical surface and on a two·dimension:1i graph hav· ing a coordinate system with bregma as its origin. At each penetration site, the microelectrode was low· ered approximately 1 mm deep [Q the pial surface, and ICMS at current intensity of 40 IlA was delivered. Then, the current intensity was gradually reduced and the penetration depth manipulated until a threshold current for eliciting reproducible movements of an individual joint was attained. In order to minimize cortical damage that may imerfere with the subse· quent tract·tracing procedure, [CMS mapping was reo

206

Inlrimi(' Connl'l'I;on in R:II MOlor

CUrll'X

•

Wl'iss and K..-II"r

stricted 10 the forelimb and vibrissal areas of the motor cortex. Damage to the cerebral cortex can be caused by the relatively long exposure of Ihe cortical surface during the experimental procedures, and by the mul· tiple penetrations with the stimulating electrode. Upon completion of the cortical mapping, three electrolytiC lesions (10 #lA, 30 sec) were made through the stimulating electrodes at depths of 0.5 and 1.0 mm to aid in the reconstruction of the ICMS maps. Neuroanatomlcal Tract Tracing

After the ICMS mapping was completed, a fluorescent dextran tracer (10% in saline, Molecular Probes)­ either fluorescein·dextran (10,000 MW) or tetrameth· ylrhodamine·dextran (Hfluoro ruby")-was injected iontophoretically (8 #lA, 8 min, positive alternating current) through a micropipette (35 "m tip diameter) into a single site in the motor cortex. Since the pattern of axonal labeling was similar follOWing tetramethyl· rhodamine·dextran or fluorescein·dextran injections, data obtained using both procedures will not be dis· tinguished. Injections were made into the core of a wrist representation area (n 3 rats) or a vibrissal representation area (n = 7), or into a border region between wrist and vibrissal representations (n 3). =

=

Histology and Data Analysis

After a survival time of 7-9 d, the animals were anes· thetized and perfused transcardially with buffered sa· line followed by a solution containing 4% buffered paraformaldehyde. The brains were then removed and the motor cortex was Oauened between two glass slides and stored in the dark for 48 hr at 4°C. The flauened cortex was CUt on a vibratome into 150·pm·thick hor· izontal sections. These sections were examined with a fluorescent microscope to determine the center of the injection site, the patterning of labeled axons, and the locations of the electrolytic lesions. Several con· secutive sections containing the injection site and labeled neuronal elements were selected for further analyses. Digital images of these sections were ac· quired through a 20 x air objective using a laser·scan· ning confocal microscope (Bio·Rad MRC·600). Thin optical sections of each digital image were scanned at ..j "m depth increments. filtered using a Kalman algorithm (7-20 iterations), and reconstructed as an extended focus image. This allowed the visualization of focused images throughout the thickness of each 150·/lm·thick section. A montage of extended· focus images, containing the injection site and axonal la· beling throughout the region mapped by ICMS, was constructed digitally for each 150·"m· thick section. Two to four sections from each brain were digitized and reconstructed. The distribution of the labeled axons and somata was traced from the full set of extended·focus images lIsing a computer· aided morphometry system (Neu· rolucida , MicroBrightField). These traces were cor· related with the ICMS maps by aligning the recon· stflH.:Il:d tr aci ngs with the stereotaxic 10(:;11 ions of the lhrt't' liducial points-thl' ckl,trolylil.' lesions.

A

B

() -

3

3

2

2

Bregma

(

%

(_)

Bre

:I

)

-

() -

4

-3

-2

0

-1

2

C

-3

-2

0

-1

2

D (

)

-

-

-

(-)

3

{-}

()()

() -

() -

3

2

Bre

() -

2

4

-3

-2 MoL (mm)

-1

o

-3

MoL (mm)

-2

-1

o

Figura 10 Somatotopic representation maps of the lat Il1lJtClr CCltU obtained from four animals by inllacuniCllI microstirnulaticn (ICMSI. Symbols indicate sites of microelectlllde peretration. V. vibrissae: W. Mist; E. elbow: So shoulder: ( J. no lesponse. Simultaneous movemeI1IS of I'AV joints elicited by threshold ICMS currenu are illlicared by syniJDJs sep8lared by a hyphen. The shaded areas indicate legions flCllll Mlich identical movements were elicited: hindlimb IBllfe5ent8tion is indicated by the h8rc1red IM«em. -

Cert:hral COrlex Mar/Apr

1994, V 4 N 2 ZD7

Figura 2. Digital images obtained Irom tlte laser-scanning confocal microscope: iontop/loletic injecrion 01 conj�ated Hooresceirrdextran (AI or tetramethylrftOdaminHextran (81 into 'IITist movement aleas 01 lite rat motOI cortex. In C. reuDgladely laheled somata were located aJlllol ximately 8UO pm anteralateralto an injecrion site in tlte wrist area_ Arrolt3' inditate electrode penetration sites wflere rrovements about the elhow (EI Of ",rist (WI were elicited hy tlttesltold leMS currents. Mearal is to 1111! right and anterior is toward the top 01 each panel.

Results Representation Maps

Examples of typical representation maps obtained with the intraconicai microstimulation (ICMS) technique are shown in Figure 1. The borders separating indi­ vidual representation zones were defined as the mid­ point between ICMS sites that elicited different move­ ments. Electrode penetrations were restricted to the vibrissal and forelimb movement regions to avoid ex­ cessive damage to the motor conex that might inter­ fere with the transpon of anatomical tracers. The cor­ tical area mapped by ICMS was approximately 2.53.5 mm anterior, 1.5-2.5 mm posterior, and 4.0-4.5 mm lateral to bregma. Thresholds for eliciting movements about specific joints varied from animal to animal, and between elec­ trode sites. Although the upper limit for accepting an evoked movement was 40 p.A, movements were com­ monly elicited at lower currents, typically within a range of 8-30 p.A (mean ± so 22.3 ± 10.1 p.A). Vibrissal movements were evoked in 25.8% of the ICMS sites. At these sites, ICMS typically elicited a single cycle of protraction-retraction of the mystacial vibrissae, primarily in the horizontal plane. A large group of whiskers was recruited at higher stimulation currents, while a smaller number of whiskers (one to three) were activated as stimulation currents ap­ proached threshold values. Movements were primar­ ilycontralateral to the stimulated conex (89.5% of the =

208 tntrinsic COllllectioll in (l;\t Mmor Cortex· Weiss and Keller

whisker sites); however, ipsilateral (8.5%) and bilat· eral (2.0%) movements were also evoked. The most common forelimb movements were ""ist extension (45.7% of all sites) and elbow flexion ( 1 1.8%). In addition, movements of the neck and shoulder also occurred in 3.2% of the sites. Hindlimb movements, elicited by stimulation at regions caudal to the forelimb and vibrissal movement areas, con­ sisted of ankle, knee, foot, and tail movements. In a few ICMS sites (9.6%), simultaneous move­ ments of two joints were elicited at threshold ICMS currents (e.g., Fig. IA,D). The most common over· lapping movements were between the vibrissae/wrist (3.2%) and wrist/elbow (2.2%) joints. Although there were similarities between ICMS maps obtained from different animals, a unique or­ ganization of movement zones characterized each an· imal. For example, (here was a single, contiguous vibrissal representation along the medial aspect of the motor cortex in all animals. Yet, the shape of the whisker movement area varied from animal to animal (Fig. 1). Similarly, although the forelimb represen­ t:ltion area was always lateral to the whisker area, sev­ eral noncontiguous zones for individual forelimb movements-such as wrist extension or elbow flex· ion-were organized in different patterns for each animal. For example, as shown in Figure lA, two non­ contiguous wrist areas, one lateral, the other caudal to the vibrissal area, are separated by elbow and hind­ limb representation zones. To aid in comparing reo

Figure 3. Digital images obtained from Ihe laser·scanning confocal microscope, depicling intrinsic alan collalerals and Iheir varicosilies. presumably alan terminals. In A. labe!ed arons projecl from an injection site localed within a region bordering a vibrissae-wrist movement area. These intrinsic arons praject toward the vibrissa movement area: note cells on leh that are irnmediate� adjacent 10 the injection site, The alon collaterals in 8 are located in an elbow region and wele labeled followirg an injection into a wrist rB!liM. The aron rollaterals in C lWle located in a vibrissa area, I.B mm anterior to an i,jection site in a ....fist-vibrissa border area. The a.on collatercl in 0 originates trom an injectm sile in a Mist-yibrissa border regitn and projects 10 bolh elbow and ....fist regions. Scale bar, 50 pm.

Fluorescent tracers were injected illlo regions of mo· tor cortex defined functionally as representation zones corresponding (0 movements of the vibrissae (11 7 rats), wrist (n 3), and the border region between whisker and forelimb areas (n 3).

were located 800-1200 I'm from the pial surface, plac­ ing them in layer V of the motor cortex. None of the injections encompassed the subcortical white matter, which in this cortical area is located at least 1800 I'm below the pial surface. In two rats, the injection sites were located 600-750 /olin from the pia, placing them at the layer III/V border. Despite these variabilities, the patterning of labeled axons was similar regardless of the size of the injection site or the cortical layer in which they were placed.

Injection Sites The ability to obtain thin optical sections with the laser·scanning confocal microscope permits direct vi· sualization of the injection site core. Unlike the image obtained with a standard fluorescence microscope, where the borders of the injection site are obscured by fluorescence ariSing superficial and deep to the injection site, the confocal image of an injection site is characterized by a clearly defined central core (Fig. 2A,B). The borders orthe injection site were defined as the boundary between this acellul;tr zone of fluo· rescence and the labeled somata surrounding this zone (see Fig. 2A,B). The injection sites were either cir· cular or oval, ranging in diameter from 68 x 83 I-lln to 250 x 350 I'm. I n 11 of 13 rats, the injection sites

AlIterogradely Labeled Axons Numerous labeled axons projected radially from each injection site, forming dense clusters of bOUlons­ presumably axon terminals-in the vicinity of the in· jection sites. A small number of cells were retrograde· ly labeled, usually in close proximity to the injection site (Figs. 2A,B; 3A), although clusters of labeled cells were occasionally found at greater distances (Fig. 2C). These cells were rarely labeled beyond their proximal dendrites, and in no case was the axon of these cells visible. Many anterogradely labeled axons projected for long distances (up to 4 mm) distal to the injection site, primarily in the anteroposterior plane (Fig. 3). The diameters of the larger axons ranged from 0.79 to 1.64 I'm, and those of thinner

suits from different animals, individual forelimb movement zones were unified into a combined fore· limb representation area. Neuroanatomlcal Tract Tracing

=

=

=

Cer ..bral Conex Mar/Apr 199'1.

V 4

N 2 209

A

Vibrissae � Forelimb I"C>! Hindlimb D

A

//

L.J 1.0mm

Figure 4. Oi$lIibution 01 fluoresceilH!!lIran-la�eled axons lollowil1!J iontophoretic injection imo deep layer V 01 vihtissal representation ZIl!IeS. The paltems nl aronaI proiecttons are correlated wnh rnovemem repleseRt3tion mnes obtained USil1!J leMS. Horizontal 3xons. as tMy awear in Haltened sections. are dl!llicted in A and B; varicosities. plesumably axon lerminals. are Sho\\l1 in A. {J denotes the location of bregma.

axonal branches ranged from 0.22 to 0.68 szm. Both bOlltons en passant and bot/tons lermirwlIx were dis· tributed along these axons ( Fig. 3D). These \'aricos· ities. presumably axon terminals, ranged in diameter from 0.48 (0 2.22 szm.

Labelingjollollling Injections into

Labelingjol/ow;ng Injections into

Forelimb Represe1tfations

Vibrissa/ Representations

Fluorescent tracers were injected into various sites within core regions of the vibrissal movement zones in seven rats. Representative examples of the pattern· ing of labeled axons are depicted in Figure 4. The most prominent feature of the labeling fol· lowing injection into a vibrissal zone was that the vast majority of the axons were restricted to the vibrissal representation area ( Fig. 4). Most of these axons pro· jected in the anteroposterior plane, parallel to the midline. with shorter axons projecting laterally, to­ ward the border region with the forelimb represen­ tation. Axons and axon terminals were distributed along this border region, invading the medial border of the forelimb representation (Fig. 4). In addition. bundles of thick·diameter axons projected posteriorly through the hindlimb representation zone. These ax­ ons were often traced back to the primary somato­ sensory cortex, where they terminated within the pos­ teromedial barrel subfield-the somatosensory representation of the mystacial vibrissae (see also Mi· yashita et al 1992; Poner and Izraeli, 1992). In sev· eral rats (Fig. 4A), long, horizontal axons projected toward the rostrolateral pole of the frontal cortex, a .•

:l10

region containing representarions of the face and low· er jaw, or toward the rostromedial region, correspond· ing to the medial agranular field (AGm) of the mOlOr cortex (Neafsey, 1990).

Imrinsk Conru"clion in Ral Mnlllr Cnnl'x \X'dss •

and Kdh .. r

Representative examples of axonal labeling following injection into the core of the forelimb representation zone are shown in Figure 5. Axons originating from these injection sites projected radially in all directions from the injection site and were largely restricted to the forelimb representation area (Fig. 5). In addition, a number of axons projected to and terminated within the border region with the vibrissal representation, encroaching into the lateral aspect of these regions (Fig. 5). Labelitlg jollowing Injections inlo Border Regions

Figure 6 depicts representative examples of axonal labeling following injection of fluorescent tracer into a region bordering the vibrissaI and forelimb move· ment areas. Injections placed precisely over the bor­ der region (Fig. 6B) or immediately adjacent to it (Fig. 6A) resulted in labeling of horizontal axons throughout the vibrissal and the forelimb represen· tation zones. Similar (0 the patte rn of laheling follow· ing d ther vibrissa I or forelimb injections, Iabelt'd axons projeclt'd also into lhe AGIII fidd and toward the somatosensory cortex.

B

Vibrissae

..,

Forelimb

EETI

Hindlimb

0

A

L.J

Figure 5. Combined represemation maps and drawings of labeled axon cnllaterals following injection imo layer V of wrist representa.ion zones. See Figure 4 101 fuoher de.ails.

Summary

The distribu.ion panerns of intrinsic axons labeled following injections into v ibrissal, forelimb, or border regions were consistem for each category of injection site. Vibrissal injections invariably resulted in an an· isotropic d istribution of axons, oriented in the amer· oposterior plane, whereas forelimb injections always revealed a radial d istribution of labeled axons. Injec· tion sites in border regions produced a third pattern of labeling, consisting of widespread radial projec· tions extending throughout the motor cortex. Thus, small variations in the location of the injection sites produced strikingly different labeling patterns. This demonstrates that neurons located within d ifferent functional subdivisions of the motor cortex form dif· ferent patterns of intrinsic, horizontal projections.

Disl:ussion Representation Maps

The results of our intracortical microstimulation (leMS) experimems are in agreement with those of previous studies on the functional organization of the rat mOlor cortex (Donoghue and W ise, 1982; Gioanni and Lamarche, 1985; Neafsey, 1990). Specifically, we have shown that the rat motor cortex comains a so· matotopic representation of movements about indio vidual joints and that, although the sizes and shapes of individual representation zones differ among ani· mals, some general features of the mOlOr map are consistem. These include an elongated vibrissal rep· resentation oriented in the anteroposterior domain along the medial portion of area AGI, and a more laterally located representation of various forelimb movements. Within the forelimb representations, there occur several noncomiguous representations of the same joint, and there are also sites where movement of more than one joint can be elicited by threshold stimulation. Within the vibrissaI representation zone,

threshold stimulation usually elicits movemeOlS of several adjacent whiskers on the contralateral whisker pad, although movements of individual vibrissae and movements of ipsilateral vibrissae were also repre· sented.

Identity of Labeled Axons

The main objective of this study was to determine the patterns of intrinsic connections among individual representation zones in the rat motor cortex. For this purpose, minute injections of lluorescent dextrans were made within identified representation zones in the motor cortex to label intrinsic axon collaterals. Several lines of evidence suggest that the axonal la­ beling obtained represents primarily axon collaterals of neurons whose somata are located at the injection site. In the rodent, fluorescent dextrans have been previously demonstrated to be transported primarily anterogradely from neurons at the injection site, pro­ ducing little retrograde labeling or labeling of axons of passage (Nance and Burns, 1990; Schmued et aI., 1990). In the present study, labeled somata were 10' cated preferentially at or in close proximity to the injection sites, suggesting that they were labeled pri· marily by transport of tracers from their dendritic trees that traversed the injection site. In none ofthe labeled somata located more distal to the injection sites were the main axons or their intrinsic collaterals visible, further supporting the conclusion that most axonal labeling originated from neurons at the injection sites. Finally, the possibility that some of the long, hori­ zontal axons represent intracortical branches of af­ ferents to the motor cortex is also unlikely, since both thalamocortical (Deschenes and Hammond, 1980; Shinoda and Kakei, 1989) and cortlcocortical ( Isseroff et al.. 1984; Porter and Izraeli, 1992) alferents to the motor cortex terminate preferentially in vertically ori­ ented clusters and have relatively few long, horizontal branches. CNcbral Conex Mar/Apr 1994. Voj N

2 211

B

Vibrissae

Iai!J

Forelimb

1:1

llindlimb

0

A

Figure 6. Combined !eJ1!l!Sentalinn maIlS and drawings of lalnlled alan mUalerals fullowlIlg an injet1ion. in layel V. invnediately adjatent

10

L.J

IAIIlf precisely over 181 a vibrissa­

Imelimb burlier region.

Horl%ontal Connections In Motor Cortex

A striking feature of the intracortical axon collaterals revealed in the present study is their long, horizontal branches that extend as far as 4 mm from the injection site. These long, horizontal axon collaterals originate most probably from pyramidal neurons located throughout layers II-VI of the rat motor conex (Kel­ ler, 1993a). Thus, for example, pyramidal tract neu­ rons in layer V have horizontal axon collaterals that project for long distances within layers V and VI (Lan· dry et aI., 1980; Ghosh et aI., 1988). The horizontal axon collaterals belonging to pyramidal neurons in layers II-III of the cat motor cortex also project for long distances, commonly forming clusters of axon terminals (Landry et aI., 1980: Keller and Asanuma, 1993). This finding is in agreement with other evi· dence for long, horizontal intrinsic connections in the motor cortex. For example, applications of minute quantities of anterograde or retrograde tracers in a single representation zone of the cat and monkey motor cortex reveal the existence of long, horizontal a.'Con collaterals that terminate in clusters of axon ter­ minals (Huntley andjones, 1991; Keller, 1993b). Oth­ er studies analyzing the distribution of intrinsic axon terminals in the motor cortex labeled by lesion-in· duced degeneration (Gatter and Powell, 1978) or by the application of tracers (Jones et aI., 1978: DeFelipe et aI., 1986; Lund et aJ., 1993) have also provided evidence for long, horizontal axons belonging to in­ trinsic axons in the motor cortex. The existence of long, horizontal axon collaterals that terminate in clusters of axon terminals is a com· mon feature of pyramidal neurons in many cortical areas (Feldman, 1984). For example, these patterns have been demonstrated in the visual cortex (Gilbert and Wiesel, 1979: Fitzpatrick et aJ., 1985: Burkhalter, 1989). in the somatosensory cortex (DeFelipe et aI., 1986: Schwark and jones, 1989; Bernardo et aI., 1990: Lund et aI., 1993), and in the auditory cortex (Wallace et aI., 1991). Nonpyramidal, inhibitory neurons in the 212

Intrinsic Connt"ctinn in Kat Motor Cnrlex

•

Wdss ;md Kt'llt'r

motor cortex are less likely to form long, horizontal axon collaterals since the axons of inhibitory neurons in the motor cortex are largely restricted to a narrow columnar domain and do not project extensively par· allel to the pial surface (DeFelipe and Jones, 1985). However, a contribution to horizontal intrinsic con· nections can be made by the GABAergic basket cells whose horizontal axon collaterals may extend for more than 1 mm Oones, 1975; DeIlelipe and jones, 1985). Recent studies in the visual cortex have shown that a subset of GABAergic neurons contribute to long-range, inhibitory connections (Kritzer et aI., 1992: Matsubara and Boyd, 1992; McDonald and Burkhalter, 1993). Role of Intrinsic Connections In Coordinating Movement Patterns

The combined electrophysiological and anatomical data in the present study reveal that intrinsic con­ nections in the rat motor cortex are specific in that they link neurons within representation zones related to the same movement patterns. Thus, for example, intrinsic connections occur between various repre· sentation zones of the forelimb musculature, with rel­ :uively few connections occurring between forelimb and vibrissa I representations. This is in agreement with findings in the monkey motor cortex, where cells in the forelimb representations (digits, elbow, wrist) are reciprocally connected via intrinsic axon collat­ erals, but have virtually no intrinsic connections with neurons in the face area (Huntley and jones. 1991). Similarly, profuse intrinsic connections link neurons in the forelimb representation zones of the cat motor cortex, but only sparse connections are found with the hindlimb or the face representation (Keller, 1993b). These specific intracortical connections may be involved in coordinating movements about differ· em joints during the execution of complex, multi­ jointed motor acts, as well as in the integration of different aspects of a single movement pattern (Kel· ler, 1993a).

Comparable conclusions are suggested from stud· ies examining the roles of intrinsic connections in other cortical areas. Intrinsic, excitatory connections occur primarily between groups of neurons sharing similar receptive field properties in the visual cortex (Ts'o et aI., 1986; Gilbert and Wiesel, 1989; Schwarz and Bolz, 1991), the auditory cortex (Matsubara and Phillips, 1988), and the somatosensory cortex (Me· therate and Dykes, 1985). Role of Intrinsic Connections In the Pliability Of Motor Maps

Several recent studies have demonstrated that rep· resentation zones in the motor cortex can be dynamo ically modulated. For example, motor maps can be modulated in response to repetitive conical stimu· lation (Nudo et aI., 1990), intracortical application of pharmacological agents Qacobs and Donoghue, 1991), training in a motor task (Woody and B1ack·Cleworth, 1973; Donoghue and Sanes, 1991; Milliken et aI., 1992), or following motor nerve lesions (Donoghue et aI., 1990). Donoghue and collaborators Qacobs and Dono· ghue, 1991;Jacobs et aI., 1991) have suggested that the rapid reorganization of motor maps may result from the activation of intrinsic connections between heterotypical representation zones, connections that are normally masked by inhibition. In the present study, axon collaterals of neurons in both vibrissal and the forelimb representation zones were shown to terminate within the border region separating the forelimb and vibrissal representations. Furthermore, tracer injections into the border regions produced extensive :txonal labeling within both the vibrissa and forelimb zones. These heterotypic:tl patterns of in· trinsic connections may form the anatomical substrate for the r:tpid pliability of motor maps. Thus, enhance· ment of the synaptic efficacies of these "inappropri· ate" intraconical pathways, either by Hebbian plas· ticity (Iriki et aI., 1989; Keller et aI., 1990) or by modulation of inhibition Qacobs and Donoghue, 1991), may mediate the rapid reorganization of rep· resentation zones in the motor cortex.

Notes The expen lechnical supporl of Mr. D;lIIid Weimr;tub is gralefull y :Il'knowle