Anatomical and physiological definition of the motor cortex of the marmoset monkey.

Abstract

We used a combination of anatomical and physiological techniques to define the primary motor cortex (M1) of the marmoset monkey and its relationship to adjacent cortical fields. Area M1, defined as a region containing a representation of the entire body and showing the highest excitability to intracortical microstimulation, is architecturally heterogeneous: it encompasses both the caudal part of the densely myelinated "gigantopyramidal" cortex (field 4) and a lateral region, corresponding to the face representation, which is less myelinated and has smaller layer 5 pyramidal cells (field 4c). Rostral to M1 is a field that is strongly reminiscent of field 4 in terms of cyto- and myeloarchitecture but that in the marmoset is poorly responsive to microstimulation. Anatomical tracing experiments revealed that this rostral field is interconnected with visual areas of the posterior parietal cortex, whereas M1 itself has no such connections. For these reasons, we considered this field to be best described as part of the dorsal premotor cortex and adopted the designation 6Dc. Histological criteria were used to define other fields adjacent to M1, including medial and ventral subdivisions of the premotor cortex (fields 6M and 6V) and the rostral somatosensory field (area 3a), as well as a rostral subdivision of the dorsal premotor area (field 6Dr). These results suggest a basic plan underlying the histological organization of the caudal frontal cortex in different simian species, which has been elaborated during the evolution of larger species of primate by creation of further morphological and functional subdivisions.

Full text

Page 1

Anatomical and Physiological Definition
of the Motor Cortex of the Marmoset
Monkey
KATHLEEN J. BURMAN,1SUSAN M. PALMER,1MICHELA GAMBERINI,2
MATTHEW W. SPITZER,3
1Department of Physiology, Monash University, Clayton, Victoria 3800, Australia
2Dipartimento di Fisiologia Umana e Generale, Universita ´ di Bologna,
Bologna 40127, Italy
3School of Psychology, Psychiatry and Psychological Medicine, Monash University,
Clayton, Victoria 3800, Australia
AND MARCELLO G.P. ROSA1*
ABSTRACT
We used a combination of anatomical and physiological techniques to define the primary
motor cortex (M1) of the marmoset monkey and its relationship to adjacent cortical fields.
Area M1, defined as a region containing a representation of the entire body and showing the
highest excitability to intracortical microstimulation, is architecturally heterogeneous: it
encompasses both the caudal part of the densely myelinated “gigantopyramidal” cortex (field
4) and a lateral region, corresponding to the face representation, which is less myelinated and
has smaller layer 5 pyramidal cells (field 4c). Rostral to M1 is a field that is strongly
reminiscent of field 4 in terms of cyto- and myeloarchitecture but that in the marmoset is
poorly responsive to microstimulation. Anatomical tracing experiments revealed that this
rostral field is interconnected with visual areas of the posterior parietal cortex, whereas M1
itself has no such connections. For these reasons, we considered this field to be best described
as part of the dorsal premotor cortex and adopted the designation 6Dc. Histological criteria
were used to define other fields adjacent to M1, including medial and ventral subdivisions of
the premotor cortex (fields 6M and 6V) and the rostral somatosensory field (area 3a), as well
as a rostral subdivision of the dorsal premotor area (field 6Dr). These results suggest a basic
plan underlying the histological organization of the caudal frontal cortex in different simian
species, which has been elaborated during the evolution of larger species of primate by
creation of further morphological and functional subdivisions. J. Comp. Neurol. 506:860–876,
2008.
© 2007 Wiley-Liss, Inc.
Indexing terms: frontal lobe; somatotopic organization; cytoarchitecture; area 4; premotor
cortex; New World monkey
This study forms part of an ongoing project aimed at
understanding the organization of cortical areas in the
marmoset (Callithrix jacchus), a species of New World
monkey that has become an important model for the study
of neurological and neuropsychological conditions (for re-
cent examples, see Eslamboli et al., 2005; Brevard et al.,
2006; Walker et al., 2006; Blezer et al., 2007). Here we use
a combination of histological analyses, electrophysiology,
and anatomical tracing to clarify the extent of the primary
motor cortex (M1) relative to adjacent fields.
The relationship between the stimulation of sites in M1
and the excitation of specific muscle groups is consistent
among primates: facial movements are represented most
laterally, followed by those of the forelimb, trunk, and
hindlimb/tail as the midline is approached (e.g., Penfield
and Boldrey, 1937; Woolsey et al., 1952; Strick and Pres-
ton, 1978; Stepniewska et al., 1993). The presence of a
Grant sponsor: National Health and Medical Research Council; Grant
numbers: 334094 and 384116.
The first two authors contributed equally to this work.
*Correspondence to: Marcello Rosa, Ph. D., Department of Physiology,
Monash University, Clayton, VIC 3800, Australia.
E-mail: Marcello.Rosa@med.monash.edu.au
Received 5 September 2007; Revised 3 October 2007; Accepted 24 Octo-
ber 2007
DOI 10.1002/cne.21580
Published online in Wiley InterScience (www.interscience.wiley.com).
THE JOURNAL OF COMPARATIVE NEUROLOGY 506:860–876 (2008)
© 2007 WILEY-LISS, INC.

Page 2

complete skeletomotor representation, with movements
elicited by relatively low currents in comparison with ad-
jacent cortex, is therefore a key criterion for the physio-
logical definition of M1. At the same time, the primary
motor cortex has been traditionally equated with a histo-
logical field (Brodmann’s area 4), which is characterized
by features such as low cellular density, poor lamination,
absence of layer 4, and presence of very large pyramidal
cells in layer 5 (Betz, 1874; Lewis, 1878; Campbell, 1905;
Brodmann, 1909; see Geyer et al., 2000, for review). How-
ever, studies in Old World monkeys have acknowledged
that the histological character of M1 may change, with
layer 5 pyramidal cells becoming smaller from medial to
lateral portions of this area (e.g., Vogt and Vogt, 1919;
Matelli et al., 1985). For example, the primary motor
cortex of baboons has been shown to include three archi-
tectural subdivisions in mediolateral succession (4a, 4b,
and 4c), which can be distinguished not only by the size of
layer 5 pyramidal cells but also by the presence/absence of
an incipient layer 4 (Watanabe-Sawaguchi et al., 1991).
The complexities involved in defining M1 have been
further highlighted by studies in the New World monkeys,
which added another dimension to the puzzle by describ-
ing caudal (M1c) and rostral (M1r) histological subdivi-
sions in M1 (Stepniewska et al., 1993). These authors
reported that “M1r” and “M1c” differ not only by the size
of layer 5 neurons but also in terms of electrical excitabil-
ity and connections. An analogous rostrocaudal subdivi-
sion of M1 has since been reported also in the macaque
(Preuss et al., 1997). Given these caveats, one may ques-
tion to what extent the physiologically defined M1 adheres
to the traditional expectation of a cortical area being a
physiologically, histologically, and connectionally well-
defined “tile” of the cortical mosaic. Indeed, recent studies
have questioned the validity of these expectations, with
reference to other regions of the neocortex (cf. Rosa and
Tweedale, 2005; Palmer and Rosa, 2006b; Graziano and
Aflalo, 2007). The origin of corticospinal axons adds little
light to the issue of defining M1, given that cells forming
these projections are known to originate in many areas,
including premotor and parietal cortices (Dum and Strick,
1991; He et al., 1993; Galea and Darian-Smith, 1994;
Luppino et al., 1994; Galea, 1997; Wu et al., 2000).
We explored the relationship between histological fields
and the somatotopic organization of electrically elicited
movements and then compared the extent of the putative
M1 with the sources of frontal projections to anterior and
posterior parietal fields (regions that have major roles in
somatic sensation and vision, respectively). Taken to-
gether, these results clarify the organization of the caudal
frontal cortex of the marmoset and the homologies be-
tween fields in New and Old World monkeys.
MATERIALS AND METHODS
Experiments were approved by the Monash University
Animal Experimentation Ethics Committee, which also
monitored the welfare of the animals. Guidelines of the
Australian Code of Practice for the Care and Use of Ani-
mals for Scientific Purposes were followed.
Microstimulation experiments
Intracortical microstimulation experiments were per-
formed on one female and two male adult marmosets
(Callithrix jacchus), with the objective of defining the rep-
resentation of different body parts in M1. The animals
were premedicated with intramuscular (i.m.) injections of
diazepam (3.0 mg/kg) and atropine (0.2 mg/kg) and, after
30 minutes, anesthetized with ketamine (50 mg/kg),
which provided deep anesthesia for the duration of the
surgical procedures (?2 hours). A tracheotomy was per-
formed, and a cannula was inserted into the saphenous
vein of the leg ipsilateral to the cortical hemisphere under
study, in order to allow a continuous infusion of a solution
consisting of sufentanil (6 ?g/kg/h) and dexamethasone
(0.4 mg/kg/h) in a saline/glucose solution. The animal was
placed in a stereotaxic frame and artificially ventilated
with a mixture of oxygen and nitrous oxide (2:3). This
anesthetic protocol was chosen as the one most compatible
with microstimulation experiments in this species, after
attempts using halothane, ketamine (i.m.), and telazol
(i.m.), all of which produced disappointing results. When-
ever movements were elicited, results from these early
experiments revealed a topographic organization similar
to the one observed in the cases reported here; however,
these observations were not included in the present paper.
A craniotomy was performed and the dura mater folded
back to expose the frontal cortex. The cortical surface was
covered with warm agar to avoid desiccation. Using ste-
reotaxic coordinates obtained in the course of previous
studies of the marmoset frontal cortex (Burman et al.,
2006), the cortex in the likely location of cytoarchitectural
fields 3, 4, and 6 was stimulated, with penetrations form-
ing rostrocaudal or mediolateral rows while avoiding sur-
face blood vessels. Low-resistance (?1 M?) parylene-
coated tungsten microelectrodes were used, penetrating to
depths of up to 2,400 ?m along the vertical stereotaxic
axis. Given that in early experiments we found that the
lowest thresholds were typically obtained at a depth of
2,100 ?m, this depth was used for microstimulation in
most penetrations (because the vertical plane is not per-
pendicular to the cortical layers in this part of the mar-
moset cortex, this depth corresponded to the cortical in-
fragranular layers).
The microstimulation currents were delivered in 60-
msec trains, with a pulse duration of 0.2 msec and a pulse
frequency of 300 Hz. Penetration sites were stimulated
initially with currents ranging from 50 to 200 ?A in order
to locate the topography of muscle representations, al-
though at unresponsive sites higher currents (of up to 300
?A) were used. Once a motor response was elicited, the
current was gradually reduced to identify the threshold
current required to create a movement. Movements
evoked by microstimulation were identified by at least two
observers and recorded in digital video.
Tracer injections
In three animals we examined the relationship between
histological subdivisions of the frontal cortex and the lo-
cation of neurons with projections to the posterior parietal
cortex. One of these animals (CJ45; see Fig. 8) was also
part of the study of Burman et al. (2006); however, the
present injections were not reported in the earlier study
and were located in the opposite cerebral hemisphere.
These animals were premedicated as described above and
kept under ketamine anesthesia for the duration of the
procedure (50 mg/kg initial dose, 10 mg/kg additional
doses as needed). They were placed in a stereotaxic frame,
and the parietal cortex was exposed. The retrograde fluo-
rescent tracers diamidino yellow (DY) and fast blue (FB)
The Journal of Comparative Neurology. DOI 10.1002/cne
861 PRIMARY MOTOR AREA IN MARMOSETS

Page 3

were directly applied into the cortex as crystals (approxi-
mately 200 ?m in diameter) with the aid of tungsten
wires; Fluoro Ruby (FR) and Fluoro Emerald (FE) were
injected (0.25 ?l) via 1-?l syringes (Scientific Glass Engi-
neering, Melbourne, Australia). The placement of the trac-
ers was guided by stereotaxic coordinates obtained in the
course of previous studies of marmoset visual cortex (Rosa
and Schmid, 1995; Rosa et al., 2005), and their exact
location in relation to cortical layers and areal boundaries
was later assessed by histological reconstruction (e.g.,
Palmer and Rosa, 2006a,b). The cortex was covered with
saline-soaked ophthalmic gelfilm, the piece of bone re-
moved during the craniotomy was fixed back in place with
dental acrylic, and the wound was closed in anatomical
layers. Analgesics (oral paracetamol, 3 drops every 6
hours) and antibiotics (Norocillin, Norbrook, Newry,
Northern Ireland; 0.1 ml) were given routinely for the first
24 hours post surgery. A survival time of 2 weeks was
allowed.
Perfusion, fixation, and tissue processing
At the end of the microstimulation experiments, or after
the survival time for retrograde transport of the tracers,
the animals were overdosed with sodium pentobarbitone
(100 mg/kg, i.v.) and perfused transcardially with 1 liter of
heparinized saline followed by 1 liter of 4% paraformalde-
hyde in 0.1 M phosphate buffer (pH 7.4). The brains were
postfixed in the same medium for a minimum of 24 hours
before being cryoprotected by immersion in buffered para-
formaldehyde solution containing increasing concentra-
tions of sucrose (10%, 20%, 30%). Frozen 40-?m-thick
sections were cut parasagittally or coronally using a cry-
ostat. One series of sections was mounted onto gelatinized
slides and air-dried for later staining with cresyl violet,
whereas a second series was kept in phosphate buffer and
stained free floating for cytochrome oxidase activity
(Wong-Riley, 1979). A third series was stored in 10% for-
malin for at least 2 weeks before staining for myelin, with
either the Gallyas (1979) stain or the gold chloride stain
(Schmued, 1990). In the anatomical experiments a final
series of sections was mounted unstained on gelatinized
slides for the analysis of cells labeled with fluorescent
tracers, after quick dehydration in 100% ethanol, clearing
in xylene, and coverslipping with DPX.
Documentation of results
Dorsal reconstructions of the frontal cortex were pre-
pared by aligning magnified drawings of sections 200 ?m
apart, using fiducial holes as references (created by intro-
ducing tungsten rods in the brain, prior to freezing and
sectioning), and then graphically stacking projections of
the dorsal surface onto a horizontal plane. The exact lo-
cation of each microstimulation site was deduced by his-
tological reconstruction of the electrode tracks relative to
landmarks such as dimples on the cortical surface pro-
duced by large blood vessels (which were compared with
photographs of the pattern of blood vessels obtained dur-
ing the microstimulation experiments). Histological bor-
ders (see Results) were plotted as transition zones of var-
ious widths, which reflect sources of uncertainty such as
the separation between sections used in the analysis, the
test-retest variability (assessed by repeated plotting by
the same observer on different days), the slightly different
estimates provided by different architectural methods,
and the interference of histology artifacts. Photomicro-
graphs were obtained through Zeiss AxioCam and Axio-
Vision v4.2 software (Carl Zeiss, Oberkochen, Germany).
Low-power views typically involved the merging of multi-
ple images, from adjacent portions of a same section, using
Adobe Photoshop 8.0 (Adobe Systems, San Jose, CA). In
such cases, the contrast and exposure of adjacent shots
was matched by using standard software features, but no
other adjustments were made. For lettering and figure
composition, the images were imported into Canvas X
(ACD Systems, Victoria, BC, Canada). Two-dimensional
reconstructions of the cortical surface were also obtained
by graphically “unfolding” contours of layer 4 of sections
400 ?m apart, in such a way as to keep the neighborhood
relationships between and within sections (Van Essen and
Maunsell, 1980).
RESULTS
We will start the description of the results by illustrat-
ing the architectural characteristics of the cortex within
and around M1, followed by a correlation between archi-
tectural fields and the somatotopic maps derived from
intracortical microstimulation. We will then demonstrate
the results of anatomical tracing experiments revealing
interconnections between the agranular frontal areas and
subdivisions of the posterior parietal cortex, which ad-
dress the issue of the distinction between primary motor
and premotor areas.
Architecture of the posterior frontal cortex
Architectural definition of the posterior and ventral
borders of M1.
In agreement with the primate norm,
the marmoset frontal cortex reveals a caudal to rostral
succession of motor, premotor, and granular frontal (pre-
frontal) fields, characterized by reduction in the size of
pyramidal cells in layer 5 and increasing definition of
layer 4 (Peden and Von Bonin, 1947; Burman et al., 2006).
Figure 1 illustrates parasagittal sections approximately
5 mm from the medial surface. (As indicated by the mi-
crostimulation results below, this level typically corre-
sponds to the representation of the forelimb in M1.) Cau-
dal to the frontal cortex are somatosensory fields 3b and
3a, which form successive narrow bands elongated in the
posteromedial to anterolateral direction (Huffman and
Krubitzer, 2001; Burman et al., 2006). In Nissl-stained
sections field 3b is conspicuous by the presence of a thick
and densely populated layer 4, whereas in field 3a this
layer is present but is thinner and overall less defined. As
the border of architectural field 4 is crossed, layer 4 dis-
appears completely. There are large, conspicuous cells in
layer 5 of field 3a but not in 3b (Fig. 1). Although the layer
5 pyramidal cells in field 3a are on average smaller than
those found in field 4, there is sufficient overlap in the
distribution of sizes to demand the use of additional cri-
teria in defining the caudal border of the motor cortex. In
myelin-stained sections throughout most of the mediolat-
eral extent of M1, field 3a can be easily seen as a less
myelinated “gap” separating fields 3b and 4 (Figs. 1, 2A).
The architectural characteristics of M1 change accord-
ing to mediolateral level, making these distinctions less
clear in the most lateral regions, from which facial move-
ments are elicited (Figs. 2B–D, 3). “Giant” isolated pyra-
midal cells (Betz neurons) are conspicuous near the mid-
line (Fig. 2B) but become gradually smaller in the forelimb
representation (Figs. 1, 2C). Given the likely homology
The Journal of Comparative Neurology. DOI 10.1002/cne
862 K.J. BURMAN ET AL.

Page 4

with cytoarchitectural fields 4a and 4b of other primate
species, and the gradient-like change in architectural fea-
tures, we refer to the main portion of the marmoset M1 as
field 4.
Furthermore, in agreement with studies in Old World
monkeys, we found that in the lateral part of M1 the
pyramidal cells in layer 5 are relatively small, being com-
parable in size to the analogous neurons in premotor ar-
eas, and that a thin layer 4 can be usually distinguished
(Fig. 2D). In addition, the lateral part of M1 is slightly less
myelinated than the medial part, while retaining the sub-
tle separation between the inner and outer bands of Bail-
larger that distinguishes this area from the dorsal premo-
tor cortex (Fig. 3; see also below). We adopted the
designation field 4c to refer to this subregion of M1 (Vogt
and Vogt, 1919; Barbas and Pandya, 1987; Watanabe-
Sawaguchi et al., 1991). The exact boundaries between
architectural fields are difficult to determine in this lat-
eral region, resulting in relatively large regions of uncer-
tainty, particularly in specimens sectioned in the parasag-
ittal plane. In our experience, the myelin stains provide
the best definition of the extent of M1.
Architectural fields along the rostral border of M1.
Figure 1 shows that in the marmoset the cortex populated
by large pyramidal cells in layer 5 is wide, extending to at
least 3–4 mm anterior to field 3a. However, there are
differences between caudal and rostral parts of this re-
gion: in the caudal region layer 5 is thicker, and the
Fig. 1.
dal frontal cortices of a marmoset (ML level approximately ?5.0),
illustrating the cytoarchitectural (top) and myeloarchitectural (bot-
tom) characteristics of somatosensory (3a, 3b) and motor (4, 6Dc, 6Dr)
fields. The bottom section was stained by using Schmued (1990) gold
chloride stain; similar patterns were observed in Gallyas-stained ma-
terials. Note the sharply defined layer 4 and heavy myelination of
Parasagittal sections through the rostral parietal and cau-
field 3b, as well as the lower myelination and relatively large layer 5
cells in area 3a. Fields 4 and 6Dc both appear as densely myelinated,
although the bands of Baillarger are more defined in field 4. The
rostral border of field 6Dc is marked by a concomitant sudden reduc-
tion in the size of layer 5 pyramidal cells and myelination. Rostral is
to the left. Scale bar ? 1 mm.
The Journal of Comparative Neurology. DOI 10.1002/cne
863 PRIMARY MOTOR AREA IN MARMOSETS

Page 5

pyramidal neurons on average larger, than in the rostral
region. A correspondingly subtle transition can be ob-
served in myelin-stained sections (Fig. 1): the caudal part
tends to show an incipient separation between the inner
and outer bands of Baillarger, whereas the rostral part
shows little evidence of lamination. Based on the results of
physiological and anatomical tracing experiments de-
scribed below, we refer to the rostral subdivision as field
6Dc (Barbas and Pandya, 1987). As reported earlier (Bur-
man et al., 2006) field 6Dc is separated from the granular
frontal cortex (architectural fields 8B, 8Ad, and 8Av) by
another architectural field (6Dr), which lacks the deeply
Fig. 2.
lobe of a marmoset (approximately at AP ?9.5; medial to the left),
stained for myelin using the Gallyas (1979) technique. Note the my-
elination “gap” corresponding to area 3a, which defines the ventral
border of M1 at the level of representation of the axial and limb
musculatures. B–D: Mediolateral differences in the cytoarchitecture
A: Coronal section through the caudal portion of the frontal
of M1. B: Medial portion. C: Intermediate portion. D: Lateral portion
(field 4c). The asterisks in A indicate the locations from which strips
B and C were photographed, in the adjacent Nissl-stained coronal
section. The location of strip D is indicated in a similar manner in
Figure 3. Scale bar ? 1 mm in A; 200 ?m in D (applies to B–D).
The Journal of Comparative Neurology. DOI 10.1002/cne
864K.J. BURMAN ET AL.