Cortical connections of auditory cortex in marmoset monkeys: lateral belt and parabelt regions

Abstract

The current working model of primate auditory cortex is constructed from a number of studies of both New and Old World monkeys. It includes three levels of processing. A primary level, the core region, is surrounded both medially and laterally by a secondary belt region. A third level of processing, the parabelt region, is located lateral to the belt. The marmoset monkey (Callithrix jacchus jacchus) has become an important model system to study auditory processing, but its anatomical organization has not been fully established. In previous studies, we focused on the architecture and connections of the core and medial belt areas (de la Mothe et al., 2006a,b). In the current study the corticocortical connections of the lateral belt and parabelt were examined in the marmoset. Tracers were injected into both rostral and caudal portions of the lateral belt and parabelt. Both regions revealed topographic connections along the rostrocaudal axis, where caudal areas of injection had stronger connections with caudal areas, and rostral areas of injection with rostral areas. The lateral belt had strong connections with the core, belt and parabelt, whereas the parabelt had strong connections with the belt but not the core. Label in the core from injections in the parabelt was significantly reduced or absent, consistent with the idea that the parabelt relies mainly on the belt for its cortical input. In addition, the present and previous studies indicate hierarchical principles of anatomical organization in the marmoset that are consistent with those observed in other primates.

Introduction

Based on studies conducted in Old World and New World monkeys over more than 35 years, a working model of the primate auditory cortex has emerged (for reviews, see Aitkin, 1990; Hackett, 2002; 2010; Jones, 2003; Kaas and Hackett, 1998; Kaas et al., 1999; 2000; Pandya, 1995; Rauschecker, 1998) (Fig. 1). In this model, auditory cortex is defined as the regions of cortex that receive preferential input from one or more divisions of the medial geniculate complex (MGC) (Hackett, 2011). While other areas of cortex are responsive to auditory stimuli (rostral superior temporal gyrus (STGr), temporal pole, superior temporal sulcus (STS), posterior parietal, and prefrontal cortex (Bruce et al., 1991; Cohen et al., 2004; Grunewald et al., 1999; Linden et al., 1999; Poremba et al., 2004; Romanski and Goldman-Rakic 2002), they do not receive significant input from the MGC and are reliant on auditory cortex for auditory input. These areas are referred to as auditory-related areas.

Figure 1.

Schematic of the marmoset auditory cortex. (A) Lateral view of the marmoset brain with the relative location of auditory cortex. (B) Enlarged schematic of auditory cortex in the marmoset denoted by a dashed line in panel A. The lateral sulcus (LS) of the left hemisphere was graphically opened (cut) to reveal the locations of auditory cortical areas (medial belt and portions of the core region) otherwise hidden. Approximately one half-two thirds of the core region as well as the lateral belt and parabelt regions are located on the superior temporal gyrus (STG). The three regions of auditory cortex are identified with varying degrees of shading: core (dark shading), belt (light shading), parabelt (medium shading). Abbreviations: ITG, Inferior Temporal Gyrus; LS, lateral sulcus; PV, Parietoventral area; Pro, Proisocortical area; Ri, Retroinsular area; S2, Somatosensory area 2; STGr, Rostral superior temporal gyrus; STS, Superior temporal sulcus; Tpt, Temporal parietotemporal area. Scale bars: A=10mm, B=5mm.

Based on this definition, three regions of auditory cortex are identified; core, belt and parabelt. The core is made up of three areas (Auditory area 1 (A1), Rostral area (R), Rostrotemporal area (RT)) and is surrounded by a belt region both medially and laterally. With recent anatomical and functional confirmation of the Middle medial area (MM) as a distinct area (de la Mothe et al., 2006a,b; Kusmierek and Rauschecker, 2009; Rauschecker, 1998; Petkov eta al., 2006; Woods et al., 2006), the medial belt is divided into four areas (Caudomedial (CM), MM, Rostromedial (RM), Rostrotemporal medial (RTM)), as is the lateral belt (Caudolateral (CL), Middle lateral (ML), anterolateral (AL), and Rostrotemporal lateral (RTL)). Located laterally adjacent is the parabelt which is divided into rostral (RPB) and caudal (CPB) areas.

An important aspect of the model is that information is thought to be processed both serially and in parallel (Hackett and Kaas, 2004; Hackett et al. 1998a,b; Kaas and Hackett, 1998; Kaas et al. 1999; Rauschecker, 1998; Rauschecker et al. 1997). Evidence for a processing hierarchy comes from both anatomical and physiological studies (Kaas and Hackett, 2000; Hackett, 2010). Connectional data show that neurons in the core project to the surrounding belt region (Aitkin et al., 1988; Cipolloni and Pandya, 1989, Fitzpatrick and Imig, 1980; Galaburda & Pandya, 1983; Jones et al., 1995; Leuthke et al., 1989; Morel et al., 1993; Morel and Kaas, 1992; Pandya and Sanides, 1973; Pandya et al., 1969), but do not project to the parabelt (Cipolloni and Pandya, 1989; Hackett et al., 1998a). The core does, however, receive feedback projections from the parabelt (de la Mothe et al., 2006a). Additionally it has been demonstrated physiologically that neuron responses in the belt area CM are at least partially reliant on the core area A1 for activation (Rauschecker et al., 1997) and various physiological gradients have been observed across the regional hierarchy (Aitkin et al. 1986; Crum et al., 2008; Imig et al., 1977; Kajikawa et al., 2005; Kosaki et al. 1997; Kusmierek et al., 2009; Merzenich and Brugge, 1973; Morel et al., 1993; Rauschecker and Tian, 2004; Rauschecker et al., 1995; 1997; Recanzone et al., 2000a, b). Thus information is thought to flow from the core to the belt, and then from the belt to the parabelt in a serial manner. The core, belt and parabelt are all made up of multiple areas and information is also thought to be processed in parallel within the various subdivisions of a region. This occurs via thalamocortical projections in which one division of the MGC projects to multiple areas in a region (e.g., MGv projects to core areas A1, R, RT), as well as via corticocortical projections in which a cortical area projects to multiple areas within a region (e.g., A1 projects to multiple caudal belt fields: CM, MM, CL, ML, among others).

While the absence of projections from the core to the parabelt has been established in the macaque, it has yet to be determined if a similar pattern is present in other primates. In previous studies of the medial belt and core areas in the marmoset, we found that connection patterns were comparable to those reported in other monkeys and the primate model in general (de la Mothe et al., 2006a, b). Accordingly, it is reasonable to hypothesize that the connections of the lateral belt and parabelt in the marmoset would be similar to other primates. In addition, the previous studies of the medial belt and core also established topographic differences between rostral and caudal areas suggesting that distinct connection patterns will be found among the rostral and caudal areas of lateral belt and parabelt, as well. These patterns would be expected to correlate with architectonic differences among areas of the lateral belt and parabelt in the marmoset (de la Mothe 2006a) and would be consistent with connectional studies of other primates that identified distinct patterns between rostral and caudal lateral belt areas (Morel and Kaas, 1992; Romanski et al., 1999) as well as between rostral and caudal parabelt areas (Hackett et al., 1998; Romanski et al., 1999).

In addition, neurophysiological recordings have revealed significant differences in response properties between rostral and caudal areas of both the belt and parabelt. Three areas (CL, ML, and AL) have been identified physiologically in the lateral belt based on reversals of tonotopic organization (Rauschecker et al., 1995; Rauschecker and Tian, 2004), and supported in fMRI (Petkov et al., 2006). Functional differences have also been reported between the caudal and rostral areas of the lateral belt (Tian et al., 2001; Tian and Rauschecker, 2004).

The main purpose of the present study was to determine whether the cortical connections of the lateral belt and adjacent parabelt in the marmoset monkey are comparable to other primates, complementing the connections of the core and medial belt regions established in our previous studies conducted in the same cases (de la Mothe et al., 2006a, b). The combined results have added significance, as the marmoset has become an important neurophysiological model for auditory cortex (Bartlett & Wang (2005, 2007, 2011); Bendor & Wang (2005, 2008, 2010); Kajikawa et al., (2005, 2008, 2011); Lu & Wang (2004); Wang et al., (1995)), but its anatomical organization remains incompletely studied. To address these questions, tracer injections were made in rostral and caudal lateral belt as well as rostral and caudal parabelt to facilitate a comparison of areas within regions (caudal vs. rostral), as well as between regions (belt vs. parabelt). Specifically, the following predictions of the model were tested: (1) The marmoset auditory cortex includes lateral belt and parabelt regions that have distinct connections; (2) The lateral belt and parabelt regions contain subdivisions that have topographically distinct patterns of connections. The results of the present study, combined with those of our previous studies of the core and medial belt regions, suggest that the basic principles of primate auditory cortex organization are highly conserved in New World and Old World monkeys, including the marmoset, owl monkey (Fitzpatrick and Imig, 1980; Imig et al., 1977; Morel and Kaas, 1992) and macaque monkey.

Materials and Methods

The experiments proposed in this report were conducted in accordance with the Vanderbilt University Institutional Animal Care and Use Committee Guidelines and the Animal Welfare Act under a protocol approved by the Vanderbilt University Institutional Animal Care and Use Committee. Four adult marmosets (Callithrix jacchus jacchus) served as animal subjects in the present study. The experimental history of each animal is included in Table 1. Since the current companion papers on the lateral belt and parabelt are extensions of the previous medial belt and core studies experimental procedures have been reported elsewhere (de la Mothe et al., 2006a, b), and are only briefly described below. The current studies involve the same cases as previously reported and thus share common architectonic borders while investigating injections of additional regions. Full descriptions as well as detailed anatomical figures are available in the original manuscripts (de la Mothe et al., 2006a,b).

Table 1.

Experimental history of animal subjects. Post-injection survival time is listed for each animal. Areas of tracer injections (CL, caudolateral belt; CM, caudomedial belt; RTL, rostrotemporolateral belt; CPB, caudal parabelt; RPB, rostral parabelt). Neuroanatomical tracers (CTB-red, cholera toxin subunit B 594; FR, fluororuby; DY; diamidino yellow), aqueous concentration and volume injected are listed for each tracer.

Case #	Case	Sex	Weight	Survival Time	Areas Injected	Tracer	%	Volume (μl)
1	(01-89)	M	300g	10 days	CL/CM CPB	FR DY	10 10	0.30 0.30
2	(03-59)	M	394g	14 days	CPB	CTB-red	1	0.35
3	(04-40)	M	334g	11 days	RPB	CTB-red	1	0.35
4	(04-51)	M	295g	10 days	RTL	FR	10	0.30

General Surgical Procedures

Microinjections of anatomical tracers were made into the lateral belt and parabelt of auditory cortex in marmoset monkeys under aseptic conditions. Anesthesia was maintained by intravenous administration of ketamine hydrochloride (10 mg/kg) supplemented by intramuscular injections of xylazine (0.4 mg/kg) or maintained with (2-3%) isoflurane. Vital signs (heart rate, expiratory CO₂, and O₂ saturation, body temperature) were continuously monitored throughout the surgery and were used to adjust the levels of anesthesia.

The head of the monkey was stabilized using a stereotaxic instrument (David Kopf Instruments, Tujunga, CA), and a midline incision was made exposing the skull, followed by the retraction of the left temporal muscle. A craniotomy was performed exposing the superior temporal gyrus and the lateral sulcus, followed by the cutting and retraction of the dura. Upon completion of the injections, the exposed area of the brain was covered with softened gelfilm, the craniotomy was closed with dental acrylic, and the overlying temporal muscle and skin sutured back into place. Injections of penicillin G (10 000 units i.m.) were given daily for 5 to 7 days after surgery, along with Banamine (1 mg/kg) as needed for analgesia.

Tracer Injections

Lateral belt and parabelt regions were located by using landmarks and blood vessels, on the lateral surface of the superior temporal gyrus (STG), and confirmed postmortem by architectonic analyses. Injections of tracers were made into target areas using 1-2 μl syringes, with a pulled glass pipette tip, attached to a hydraulic microdrive. In all cases the injections made were manual pressure injections of various amounts (Table 1), directly into the auditory areas, after which the syringe remained for approximately 10 minutes under continuous observation to maximize uptake and minimize leakage. The tracers used were Alexa Fluor 594 conjugated cholera toxin-B (CTB-red) (Invitrogen Corp); fluororuby (FR); and diamidino yellow (DY). While some of the tracers are bidirectional, anterograde transport was not examined in this study. Due to the various levels of sensitivity of the tracers the amounts and solution concentrations were varied accordingly as shown in table 1.

Perfusion

Upon completion of an experiment (between ten to fourteen days following the injections), a lethal dose of pentobarbital was administered. Just before cardiac arrest the animal was perfused through the heart with warm saline followed by cold 4% paraformaldehyde dissolved in 0.1 M phosphate buffer. Immediately following the perfusion the brain was removed and photographed, the two hemispheres and the brainstem were separated and placed in 30% sucrose for several days and then blocked.

Histology and Data Analysis

The hemispheres were cut perpendicular to the lateral sulcus at 40 μm. Depending on the case, series of sections were processed for: (i) fluorescent microscopy; (ii) CTB; (iii) myelin (Gallyas, 1979); (iv) acetylcholinesterase (AChE) (Geneser-Jensen and Blackstad, 1971); (v) stained for Nissl substance with thionin (vi) cytochrome oxidase (CO) (Wong-Riley, 1979); or (vii) parvalbumin immunohistochemistry. Cells labeled with fluorescent tracers were all plotted on an X-Y plotter (Neurolucida; MBF Bioscience, Williston, VT) coupled to a Leitz microscope under ultraviolet illumination.

Architectonic reconstructions of all cases included in this study were previously completed in our earlier studies of the core and medial belt regions in the same cases (de la Mothe et al., 2006a). Detailed descriptions of the architectonic features used to identify regional and areal borders were provided in those papers. Examples of those features are illustrated in Fig. 2, and major characteristics are briefly summarized here.

Figure 2.

Architecture of the marmoset auditory cortex. Low magnification images of corresponding sections of nissl, myelin, and AChE with borders drawn to identify the various regions based on respective architectonic criteria.

Along the core-belt-parabelt axis, the core region was identified by the presence of a broad layer IV and a cell sparse layer V with a densely packed population of small granule cells, astriate dense myelination where bands through layers IV and Vb were not visible, and dense expression of AChE and CO through the middle layers. The belt region expressed an increase in large pyramidal cells in lower layer III, bistriate myelination where bands through layers IV and Vb were visible, and a decreased expression of AChE and CO. The parabelt exhibited pronounced columnar organization of cells, a general decrease in overall myelin density with particular weakening of the infragranular layers, decreased expression CO and AChE compared to the adjacent lateral belt.

As the brains were sectioned in a near-coronal plane, the borders between adjacent areas along the caudal-rostral axis could not be visualized. Therefore, the delineation of areas along this axis was based on the identification of architectonic differences between sequential sections. Descriptions of these characteristics were reported in the previous study of the same brains (de la Mothe et al., 2006a). Overall, it can be stated that within a region (core, belt parabelt), the density of AChE expression and myelinated fiber density is systematically reduced from caudal to rostral, especially in the layer IV band. This is accompanied by broader columnar spacing and reduced cortical thickness. Whether these shifts are abrupt or transitional could not be determined in the plane of section used here.

Injection sites were identified and compared to architectonic reconstructions in order to verify location of injections. Composite drawings were made from adjacent sections processed for tracer label, AChE, myelin, CO, and Nissl by aligning common architectonic borders and blood vessels. Reconstructions of the composite images were created using Canvas 8.0 software (Deneba software, Miami, FL, USA). The final composites were analyzed to reveal the individual connection patterns of the retrograde labeled cells and the connection patterns of injections at similar or dissimilar locations. In general, every other section was selected for illustration purposes in the figures. Due to variability in tracer sensitivity, cell counts were performed on all auditory areas and converted to percentages of total labeled cells in order to better compare the general connection patterns between tracers. While sensitivity of the tracers did vary the general patterns revealed by injections of specific regions were maintained, regardless of the actual number of cells labeled.

Results

Ipsilateral connections of the lateral belt

In the previous study of the marmoset medial belt region (de la Mothe 2006a) strong rostrocaudal connections were revealed with the core and the belt region (medial and lateral). Injections were made into caudal and rostral areas of the lateral belt to determine if this general belt organization is consistent in the marmoset. In case 1, an injection was made caudal to A1 that involved both CM and CL (Fig. 3). Caudally, analysis of the retrograde labeled cells revealed weak connections with presumptive Tpt (section #52) with stronger connections in supragranular and infragranular layers of CM, CL and Ri (#64-76). Just posterior to A1, near the injection site, the connection pattern in CM and CL continued with denser label in CM, but cells were absent from Ri (#88). As A1 emerged, there were strong connections in A1, MM, ML, and Ri (#100-148). The overall strength of connections decreased rostrally with distance from the injection site, but labeled cells were consistently denser in MM. From the caudal emergence and throughout CPB there were labeled cells in both supragranular and infragranular layers, with the strongest connections more caudal in the area, closer to the injection site. Transitioning into rostral auditory cortex, connections were weak overall and mainly involved AL and RPB (#160-184). Label was sparse or absent from RM and R.

Figure 3.

Ipsilateral cortical connections of areas CM/CL and CPB, case 1. Series of serial sections are arranged from caudal (upper left) to rostral (lower right), and continue onto the next page. FR labeled cells (red filled triangles) and DY labeled cells (black filled circles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Black shading illustrates injection site. Dashed line in section 100 denotes where the label is artificially cut off dorsally due to interference from FR injection. Inset, schematic of marmoset auditory cortex showing location of FR injection into CM/CL, and DY into CPB.

In case 4, the injection was made into RTL rostral to the core area RT (Fig. 4). In the most caudal sections there were weak connections in the supragranular layers of MM and the lateral division of A1 and both the supragranular and infragranular layers of ML (#130-118). Connections strengthened at the A1/R border (#106) with the transition into rostral auditory cortex and with the pattern of predominantly supragranular connections continuing in R and RM (#106-70). In AL connections were also present in both the supragranular and infragranular layers in caudal R (#106-94), but shifted to predominantly supragranular layers more rostral in the area (#82) and weakened significantly right before the transition into RT (#70). Strength of connections increased with the beginning of RT in areas RTL, RT and RTM, although connections were strongest in RT (#58). Label was present in both the supragranular and infragranular layers of RT and RTL, while confined predominantly to the supragranular layers in RTM. At the rostral extent of auditory cortex where RT is no longer present, there were strong connections in RTL with weaker connections in RTM, RPB and Pro (#34). Until this point label was virtually absent from the parabelt, with the exception of a few sparse cells in RPB (#106, 82-58). This pattern of weak connections was also present in Pro throughout R and RT.

Figure 4.

Ipsilateral cortical connections of area RTL, case 4. Series of serial sections are arranged from caudal (upper left) to rostral (lower right). FR labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Black shading illustrates injection site. Inset, schematic of marmoset auditory cortex showing location of FR injection into RTL.

Contralateral connections of the lateral belt

The contralateral connections with the lateral belt region were concentrated in layer III of the corresponding area of injection in the contralateral hemisphere (Figs. 5, 6). In case 1, the injection of FR into CM and CL labeled cells primarily in the belt (medial belt and lateral belt), as well as the core region of caudal auditory cortex (Fig. 5). Caudally there was strong labeling in CM and CL with sparse labeling in CPB (#100-112). Label in MM strengthened as A1 emerged and moderate label was present in A1, ML, and CPB (#124-136), with sparse label in the most rostral section (#148). In case 4, cells labeled by the RTL injection were concentrated in the rostral core and lateral belt regions with sparse cells in RPB (Fig. 6). Caudally there was sparse label in R (#100) and moving rostrally, label was also present in AL with the densest label at the R/RT border (#88-70). Label continued in RT and RTL (#58-46) and was concentrated in RTL, rostral to RT (#28).

Figure 5.

Interhemispheric cortical connections of CM/CL and CPB, case 1. Series of serial sections are arranged from caudal (upper right) to rostral (lower left). FR labeled cells (red filled triangles) and DY labeled cells (black filled circles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Inset, schematic of marmoset auditory cortex showing location of FR injection in CM/CL, DY in CPB in the contralateral hemisphere.

Figure 6.

Interhemispheric cortical connections of area RTL, case 4. Series of serial sections are arranged from caudal (upper right) to rostral (lower left). FR labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Inset, schematic of marmoset auditory cortex showing location of FR injection in RTL in the contralateral hemisphere.

Summary of lateral belt connections

Outside the areas of injection, CM and CL, the strongest connections were with CPB followed by A1, R and ML (summarized in Fig. 7). There were weaker connections with the more rostral areas RM, R, AL, and RPB, but no connections with the most rostral extent of auditory cortex. Label in the contralateral hemisphere was consistent with the area of injection with label predominantly in CM as well as CL, A1, ML, and CPB. Judging from the location of the injection site in the ipsilateral hemisphere as well as the contralateral label, it appears, that the injection, which borders CM and CL, may have involved more of CM than CL.

Figure 7.

Summary of ipsilateral (left) and interhemispheric (right) connections of CM/CL. Top panels illustrate connections (arrows) of CM/CL on schematic diagram of marmoset auditory cortex. Arrow size is proportional to connection strength, as indicated in the histograms below each panel. Single arrows indicate unidirectional projections. Bottom left, white bar indicates that cell counts for ipsilateral CM/CL may be inaccurate (deflated) due to masking by the tracer injection.

Other than the area of injection, RTL, label was found predominantly in the rostral core areas R and RT as well as AL (summarized in Fig. 8). There were also moderate connections with RTM and RPB, and weak connections with ML, RM and Pro, and the rostral portions of A1 and CM. Label in A1 was almost exclusively in the lateral division. Both the core areas A1 and R had label predominantly in the superficial layers. The injection into RTL exhibited a pattern of rostrocaudal topography. Contralateral label was present predominantly in RTL and the adjacent core area RT. There were also connections from R and AL with weak label in RPB.

Figure 8.

Summary of ipsilateral (left) and interhemispheric (right) connections of RTL. Top panels illustrate connections (arrows) of RTL on schematic diagram of marmoset auditory cortex. Arrow size is proportional to connection strength, as indicated in the histograms below each panel. Single arrows indicate unidirectional projections. Bottom left, white bar indicates that cell counts for ipsilateral RTL may be inaccurate (deflated) due to masking by the tracer injection.

Ipsilateral connections of the parabelt

In Old World primates, the parabelt region has strong rostrocaudal connections with the belt region (medial and lateral) (Hackett et al., 1998). Injections were made into caudal and rostral areas of the parabelt to determine if a similar organization exists in the marmoset. In case 1, the injection was made into the caudal portion of CPB just before the boundary of A1 (Fig. 3). Caudally there was weak labeling in both the supragranular and infragranular layers of CM and CL, with similar labeling present in Tpt (#52). This pattern of label continued and connections strengthened rostrally within the defined areas CM and CL (#64-88) and was present throughout area ML (#100-148), with the strongest connections in the most caudal ML (#100-112), closer to the CPB injection. Beginning with the most caudal CPB section and continuing throughout the CPB there was dense label in both the supragranular and infragranular layers (#76-148). In addition, label was present ventral to the CPB in the STS with the strongest connections in the caudal portion (#76-124). Connections weakened rostrally and shifted to predominantly infragranular layers (#136-172). Area A1 had only sparse label throughout its rostrocaudal extent (#100-148) with even weaker connections in the more rostral core area, R (#160-196). While label was present in caudal CM, with the emergence of A1, the label in the medial belt, area MM, was absent, or if present was very sparse (#100-136). As the transition occurs between caudal to rostral auditory cortex (A1/R border), there was an increase in labeled cells in the medial belt area RM (#148-172), but label was confined to the infragranular layers of RM in the most rostral sections (#184-196). This pattern of predominantly, or only infragranular connections was also present in the rostral lateral belt and parabelt areas AL and RPB (#160-196).

In case 2, the injection was made into the rostral portion of CPB close to the A1/R border (Fig. 9). In the most caudal section (#267), label was in both the supragranular and infragranular layers of Tpt, with some weak label in the most caudal portion of CL. Connections in the caudal belt areas (CM/CL) strengthened rostrally, and weak connections were present in the supragranular layers of the adjacent Tpt (#255). As CPB emerged, supragranular label was present and label in the caudal belt areas weakened (#243). At the level of A1, connections strengthened in the lateral belt, area ML, as well as the CPB, but label in MM was virtually absent. The strong connections persisted until the A1/R border (#231-159), which coincided with the end of the area of diffusion from the injection. As A1 was shifting into medial and lateral divisions, there appeared to be an area of transition both in the architecture and the connections along the ML/A1 border (#207). The shift from both supragranular and infragranular label to supragranular only appears to correspond with the emergence of the lateral division of A1. Due to lack of core-like architectonic features (i.e. lack of astriate myelination) we have defined the label as part of ML, but indicate that this is a zone of transition with the dashed line. While there was label in the most caudal medial belt section where CM and CL are adjacent (#255-#243), connections from the medial belt area MM, were absent. With the transition from caudal auditory cortex to rostral auditory cortex, label was present in RM, mainly in the supragranular layers (#159-111). Connections were present in the supragranular layers beginning with the most caudal section of CPB (#243), but throughout the remaining CPB as well as the RPB, label was present in both supragranular and infragranular layers (#231-111). Labeled cells were also found ventral to CPB in the STS, but decreased in numbers rostrally, adjacent to RPB (#219-111). In case 3, the injection was made into the rostral portion of RPB (Fig. 10). The injection appears to have infringed on the underlying white matter, which could potentially involve fiber tracts resulting in corresponding labeled cells. While there is a large number of cells that are labeled due to this injection the overall pattern is consistent with other injections into the parabelt region, and thus if contamination did occur the effect on the results does not appear to alter the overall impression. In the most caudal section connections were mainly in the infragranular layers of CL with weak infragranular connections in CM. With the emergence of A1, the majority of the label was present in supragranular and infragranular layers in the lateral belt, ML, and parabelt, CPB (#148-184). In the same sections there were only weak connections in A1, MM, and Ri. There was a sudden surge of labeled cells in the medial belt at the transition from caudal auditory cortex to rostral auditory cortex. Labeled cells were present in RM, AL, and RPB with absent or weak connections in R (#196-232), and continued throughout the auditory cortex until the R/RT border. At the transition into RT connections were mainly in the supragranular layers with some weak connections in the infragranular layers (#244-280). Label in RTM, RTL, and RPB was still present in the supragranular and infragranular layers but was more dense in the superficial layers in the medial and lateral belt areas (#244-280). In the most rostral section where the core area RT was no longer present and the medial and lateral belt areas have joined together, the majority of the labeled cells were present in RTL, as well as RTM and RPB, in both the supragranular and infragranular layers (#292). Label was present in Pro beginning at the transition of the A1/R border where area Pro emerges medial to RM, and persisted throughout the extent of rostral auditory cortex, decreasing with the rostral border of the core, the end of area RT (#196-292). Connections were present ventral to the parabelt in the STS, beginning with the transition into rostral auditory cortex. Initially connections were weak (#196-208), but strengthened rostrally (#220-268) and were found predominantly in the infragranular layers in the most rostral sections (#280-292).

Figure 9.

Ipsilateral cortical connections of area CPB, case 2. Series of serial sections are arranged from caudal (upper left) to rostral (lower right), and continue onto the next page. CTB labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Black shading illustrates injection site. Inset, schematic of marmoset auditory cortex showing location of FR injection into CM/CL, and DY into CPB.

Figure 10.

Ipsilateral cortical connections of area RPB, case 3. Series of serial sections are arranged from caudal (upper left) to rostral (lower right), and continue onto the next page. CTB-red labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Black shading illustrates injection site. Dashed line indicates infragranular bias. Inset, schematic of marmoset auditory cortex showing location of CTB-red injection into RPB.

Contralateral connections of the parabelt

Similar to results of the lateral belt, contralateral connections of the parabelt region were concentrated in layer III of the corresponding area of injection in the contralateral hemisphere (Figs. 5, 11, 12). In case 1, label was concentrated in the supragranular layers of the caudal parabelt and lateral belt regions (Fig. 5). Most caudally, label was present in CM/CL (#100), but shifted to strictly area CL and CPB moving rostrally (#112). The densest label was present with the emergence of A1 and was confined to ML and CPB (#124). This pattern continued rostrally but connections weakened (#136). Sparse label was found in A1 in the most rostral section (#148).

Figure 11.

Interhemispheric cortical connections of area CPB, case 2. Series of serial sections are arranged from caudal (upper right) to rostral (lower left). CTB-red labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Inset, schematic of marmoset auditory cortex showing location of CTB-red injection in CPB in the contralateral hemisphere.

Figure 12.

Interhemispheric cortical connections of area RPB, case 3. Series of serial sections are arranged from caudal (upper right) to rostral (lower left). CTB-red labeled cells (filled triangles) are drawn onto each section, showing borders between areas identified by architectonic criteria. Inset, schematic of marmoset auditory cortex showing location of CTB-red injection in RPB in the contralateral hemisphere. Summary of RPB connections.

In case 2, similar to case 1, label was concentrated in the lateral belt and parabelt regions of caudal auditory cortex (Fig. 11). In the most caudal section there was some sparse label in CL (#195) that strengthened rostrally in ML with the emergence of A1 (#189). Continuing rostral there was strong label in ML and CPB (#177-165) that shifts to predominantly CPB (#153). At the rostral extent of the label, cells were concentrated in ML (#147).

In case 3, label was present in each of the different rostral regions of auditory cortex, but was concentrated in the parabelt and the lateral belt in the most rostral sections (Fig. 12). Caudally label was present in RPB (#112) and some sparse label was present in RM (#124). Closer to the R/RT border, label was stronger in RPB and is also present in AL, RM, and Pro, with some weak label present in R (#136-148). Label in RPB strengthens rostrally with the presence of RT, and label was also present in RTM with sparse label in RTL and RT (#172). Moving rostrally this pattern continued with an increase in labeled cells in RTL (#184-196). In the most rostral section the strongest label was in RPB and RTL (#208).

Summary of parabelt connections

Regarding connections of the CPB, other than the area of injection, the strongest connections were with the adjacent lateral belt area, ML (summarized in Fig. 13). Connections were also present with the neighboring areas RPB, CL and the STS, as well as CM. Label in CM was present caudally but absent in MM in one case, and in the other case, greatly reduced causing a gap in the continuous rostrocaudal label in the medial belt. Outside the area of injection the strongest connections with RPB were with the bordering areas RTL, AL, and the STS (summarized in Fig. 14). Additional connections were also present with Pro, ML, RTM, RM and CPB as well as weak connection with the caudal areas Ri, CL, CM, MM, and the core areas A1 and R. Compared to the other core areas where label was sparse, there was increased label in RT, predominantly in the supragranular layers. Strong label was present in the medial belt once the transition was made into rostral auditory cortex, accompanied by strong label in the lateral belt and parabelt. The absence of label in R created a pronounced gap (Fig. 15) that was visible throughout R until the RT border. Proportionally the RPB had the most even distribution of label in the ipsilateral hemisphere of all of the lateral belt and parabelt injections. For all parabelt injections, contralateral label was strongest in the homotopic area of injection.

Figure 13.

Summary of ipsilateral (left) and interhemispheric (right) connections of CPB. Top panels illustrate connections (arrows) of CPB on schematic diagram of marmoset auditory cortex. Arrow size is proportional to connection strength, as indicated in the histograms below each panel. Single arrows indicate unidirectional projections. Bottom left, white bar indicates that cell counts for ipsilateral CPB may be inaccurate (deflated) due to masking by the tracer injection.

Figure 14.

Summary of ipsilateral (left) and interhemispheric (right) connections of RPB. Top panels illustrate connections (arrows) of RPB on schematic diagram of marmoset auditory cortex. Arrow size is proportional to connection strength, as indicated in the histograms below each panel. Single arrows indicate unidirectional projections. Bottom left, white bar indicates that cell counts for ipsilateral RPB may be inaccurate (deflated) due to masking by the tracer injection.

Figure 15.

Image of the rostral areas from an injection of FR into RPB with the core area, R void of labeled cells creating a gap pattern.

Laminar distribution of labeled cells

The plane of section, perpendicular to the lateral sulcus, permits the analysis of the laminar distribution of retrograde labeled cells with less distortion of columnar organization than would be imposed by a coronal plane of sections. Several patterns were identified. The most common pattern, ipsilaterally, was labeled pyramidal cells in both the supragranular (layersII/III) and infragranular layers (layersV/VI), with densest labeling closer to the injection site. In addition to the increased labeling around the injection site, in a few instances a second pattern was revealed in which the labeled cells were present throughout the cortical layers, including supragranular (II/III), granular (layer IV), and infragranular layers (V/VI) (Fig. 3).

A third laminar pattern was observed from both CPB (Fig. 3) and RPB (Fig. 10) injections, increasingly infragranular, or infragranular only label with increased distance form the injection site. In both cases injections were made at the rostral or caudal extremes of the respective areas, and the pattern of label mainly involved the sections furthest from the injection site (e.g. most rostral sections from a caudal CPB injection).

An additional fourth pattern of supragranular only label was present in several instances. First, in the ipsilateral hemisphere, an injection into rostral RTL revealed mainly supragranular label with increasing caudal distance (Fig. 4). Second, from the CPB injection near the RPB/CPB border, labeled cells in the medial belt were only observed in the supragranular layers (Fig. 9). Third, this pattern of supragranular only label was the predominant pattern in the contralateral hemisphere, however it appears to be concentrated in layer III.

Discussion

The current study examined the cortical connections of the lateral belt and parabelt regions of the marmoset auditory cortex, and included injections into both rostral (RTL, RPB) and caudal areas (CM/CL, CPB). With the goal of providing a comprehensive framework for the organization of the auditory cortex of the marmoset monkey, substantial injections were made throughout the auditory cortex into each region (core, belt, parabelt). The results from injections into the medial belt and core regions were published previously (de la Mothe et al., 2006a, b) with remaining injections for the lateral belt and the parabelt reported here, and while the injections were fewer in number, overall the results were consistent with previous studies in macaques and in line with two major features of the primate model (Hackett et al., 1998; Imig et al., 1977; Morel and Kaas, 1992; Romanski et al., 1999).

First, the lateral belt was strongly connected with the core, belt, and parabelt regions, whereas the parabelt was strongly connected with the belt, but had only weak connections with the core. This is consistent with a serial processing model in which the core projects strongly to the belt, but not the parabelt, and the belt projects to the parabelt. Secondly, there was rostrocaudal topography in the connections between the lateral belt and the parabelt. In combination with our previous studies of the medial belt and core regions, conducted in the same animal subjects (de la Mothe et al., 2006a,b), this work expands our understanding of the anatomical organization of the marmoset monkey auditory cortex, and supports the hypothesis that there is a common pattern of organization in primates. These results are discussed in further detail below.

Regional distinctions between the lateral belt and the parabelt

Two main differences in connection patterns were identified between the lateral belt and parabelt regions. First, injections into the lateral belt revealed connections with the core, belt, and parabelt, while injections into the parabelt resulted in connections with the belt and parabelt regions, but only sparse label in the core areas A1 and R (fig. 16). The core area RT did have connections with RPB (fig. 16). Secondly, connections with the STS were mainly from the parabelt, and were strongest with proximity to the rostrocaudal level of injection.

Figure 16.

Summary of regional connections of the lateral belt and parabelt. Dotted line denotes moderate connections from area RT of the core to the parabelt, not present in the other two core areas (A1, R). All connections between regions were found to be topographic (possible exception RM of the medial belt). Core region (dark grey), belt region (light grey), parabelt region (white). Region of injections and connectional study are denoted by thick black outline.

The lack of label throughout most of the core from the parabelt injections is consistent with evidence of serial processing in auditory cortex. This hypothesis comes in part from a study by Rauschecker et al., 1997, in which following the ablation of core area A1, responses in R were unaffected; however the responses in belt area CM to tonal stimuli were abolished, while some responses to complex stimuli remained. These results suggested a dependence of CM on A1 for cortical activation and support the hypothesis that the belt region surrounding the core represents a second level of processing in auditory cortex (Kaas and Hackett 1998; Rauschecker 1998). Neurophysiological findings consistent with this hypothesis include increased response latencies, increased spectral tuning, and decreased temporal precision along the core, belt, parabelt hierarchy (Aitkin et al. 1986; Crum et al., submitted; Imig et al., 1977; Kajikawa et al., 2005; Kosaki et al. 1997; Kusmierek et al., 2009; Merzenich and Brugge, 1973; Morel et al., 1993; Rauschecker and Tian, 2004; Rauschecker et al., 1995; 1997; Recanzone et al., 2000a, b).

The main anatomical support for serial processing in Old World primates comes from Hackett et al., 1998, where injections along the rostrocaudal extent of the parabelt in macaque monkeys revealed strong connections between the belt and parabelt, but a lack of connections with the core. In this study of the marmoset, this pattern was also observed, except in the core area, RT, which contained labeled cells from an injection into RPB. This is in line with the rostrocaudal topography of the parabelt, but highlights unique features of RT that was hinted at in previous studies. While cells were present in this most rostral core area, they were still reduced in number compared to the dense label of the surrounding belt areas. Hackett et al. (1998) reported similar findings where injections into RPB resulted in label rostral to R, but in numbers that were lower than in the surrounding belt. Similarly, Reser et al. (2009) reported connections between RT and the parabelt following injections into RT. Because its connections and architecture share features of both core and belt areas, RT has been considered the least certain member of the core (see Hackett et al., 1998a).

While RT exhibits architectonic characteristics of the other two primary areas (koniocellular, dense expression of CO, Pv, and dense myelination), these features tend to be muted in comparison to A1 and R. This is consistent with anatomical gradients that change systematically along the caudorostral axis of the core (de la Mothe et al 2006a; Galaburda and Pandya, 1983; Imig et al., 1977; Jones et al., 1995; Merzenich and Brugge, 1973; Morel and Kaas, 1992; Pandya and Sanides, 1973; Hackett, 2010). These gradients included tonotopic reversals at the borders between each area (Bendor and Wang, 2005, 2008; Petkov et al., 2006; Tanjii et al 2010), and systematic increases in response latency and stimulus selectivity from A1 to R to RT (Bendor and Wang, 2008; Kikuchi et al 2010). Given that RT is the least typical member of the core region, perhaps it is worth revisiting the possibility put forth by Morel and Kaas (1992), and echoed by Reser et al. (2009) that RT is also an intermediate or hybrid area, one that combines core-like and belt-like features. Similar proposals have been advanced for area CM, which has both core and belt features (de la Mothe et al., 2006a,b).

Another main difference between the lateral belt and the parabelt regions were the connections with the STS. The parabelt had strong connections with the STS that extended into the fundus, possibly corresponding with the location of the superior temporal polysensory area (STP) in macaque monkeys. STP is a higher-order area, responsive to auditory, somatosensory and visual stimuli (Bruce et al., 1981; Luppino et al., 2001) that encompasses the temporal parietal occipital area (TPO) (Padberg et al., 2003). The lateral belt had comparatively weaker connections with this region of cortex. The label was strongest in the STS at the rostrocaudal level of the injection and weakened with distance from the injection site; so that the CPB was strongly connected with caudal STS and rostral parabelt with the rostral STS. Additional studies have reported similar results in which injections into the parabelt revealed strong connections with the STS in macaques (Barnes and Pandya, 1992; Cipolloni and Pandya, 1989; Cusick et al., 1995; Galaburda and Pandya, 1983; Hackett et al., 1998; Morel et al., 1993; Seltzer and Pandya, 1978). Other studies have shown connections between the lateral belt and the STS (Jones and Powell, 1970; Hackett et al., 2007; Morel et al., 1993; Seltzer and Pandya 1978; Smiley et al., 2007); however, only Morel et al. (1993) reported results from an injection that, like the current study, did not encroach on the parabelt. This is in contrast to Hackett et al., (1998) in which an injection confined to the lateral belt revealed no label in the upper bank of the STS. Injections into the upper bank of the STS consistently labeled cells in the parabelt and in some cases label extended to the lateral and medial belt (Seltzer and Pandya, 1991; Seltzer and Pandya, 1994). In general, it appears that the core is strongly connected to the belt, the belt to the parabelt, and the parabelt to the STS, consistent with a serial flow of information that extends beyond auditory cortex ventrally.

Topographic connections of the rostral and caudal areas

The main connectional pattern revealed between caudal and rostral areas of the lateral belt and parabelt was that caudal areas were most strongly connected with caudal areas and rostral with rostral areas. This extended to the callosal connections as wellconsistent with other findings in auditory cortex (Aitkin et al., 1989: de la Mothe et al., 2006a; Fitzpatrick and Imig, 1980; Hackett et al., 1999; Leuthke et al., 1989; Morel et al., 1993). The exception to this rostrocaudal trend appeared to be area RM which was more strongly connected with CPB than the adjacent MM.

Label from the caudal belt injection was found in the core, medial and lateral belt, and parabelt areas predominantly in caudal auditory cortex. While connections with RTL also followed this rostrocaudal topography, one difference was the decrease of label in the parabelt (RPB 5%) in comparison to the strong connections from the caudal belt injection (CPB 32%). It is unclear why there is such a discrepancy between the rostral and caudal belt areas with regards to the input from the parabelt, however the connections were consistent with rostrocaudal topography in auditory cortex (Cipolloni and Pandya, 1989; de la Mothe et al. 2006a; Fitzpatrick and Imig, 1980; Galaburda and Pandya, 1983; Hackett et al., 1998a; Jones et al. 1995; Leuthke et al. 1989; Morel and Kaas, 1992; Morel et al. 1993).

As well as connections with the caudal auditory cortex, injection of the caudal belt revealed moderate connections with Ri, consistent with previous studies involving CL (Hackett et al., 2007; Smiley et al., 2007), and other caudal belt areas (Cipolloni and Pandya, 1989; de la Mothe et al., 2006a; Galaburda and Pandya, 1983). Additional studies with injections involving the caudal lateral belt have had mixed results, reporting label in a corresponding location to Ri (also identified as ventral somatosensory area) in some, but not all of the cases (Cappe et al., 2005; Morel & Kaas, 1992; Kaas et al., 1993). In general, however, there does appear to be a substantial connection between Ri and the caudal belt areas. While the belt area, CM, has been the focus of several studies due to the discovery of somatosensory responses in this area (Fu et al., 2003; Kayser et al., 2005; Robinson and Burton 1980a,b; Schroeder and Foxe, 2002; Schroeder et al., 2001), neurons in CL have also demonstrated responsiveness to somatosensory stimuli ( Kayser et al., 2005). Area Ri, which is located in the fundus of the lateral sulcus, medial to areas CM and MM, and bordered by S2 on the upper bank, is somatotopically organized with strong somatosensory connections (see Hackett 2007 for review), is considered a likely source of this somatosensory input. Unlike the current study in which label was weak or absent in Ri from injections into CPB, connections between Ri and the CPB have been reported (Cipolloni and Pandya, 1989; Galaburda and Pandya, 1983; Hackett et al., 1998). The injections in these studies, however, do not appear to be confined to CPB and most likely involved adjacent areas (ML, CL, Tpt).

Injections into CPB labeled cells in the medial belt caudal to A1 (CM), and in RM, but label was sparse or absent in the portion of the belt medial to A1, corresponding to area MM. This is in contrast to injections made into MM which showed reciprocal connections with CPB (de la Mothe et al., 2006a), as well as findings from Hackett et al. (1998a), which reported connections between CPB throughout the caudal medial belt, including the portion medial to A1. It is unclear why MM has only a weak, if not completely absent, projection to CPB when the medial belt areas flanking its rostral and caudal borders have moderate projections, and considering the strong input it receives from CPB.

The RPB exhibited a topographically distinct pattern of connections with the medial belt compared to CPB. At the transition into rostral from caudal auditory cortex there is a significant and sudden surge of label in the rostral medial belt as well as the adjacent area Pro, from injection into RPB. Both CPB and RPB had connections with RM, consistent with findings in the macaque (Hackett et al., 1998a). Thus, RM appears to be the main exception to the rostrocaudal topography typical of the connections between most other auditory areas.

Consistency with the working model of primate auditory cortex

The present study of the lateral belt and parabelt regions complement and complete our earlier studies of the core and medial belt regions in the marmoset. Together, these studies indicate that the overall organization of the marmoset auditory cortex is highly similar to other monkeys, and in line with the working model of primate auditory cortex. Regional findings from the current study, summarized in figure 16 reveal the major patterns of connectivity observed. First, the data provide strong support for a serially organized system of projections in which the lateral belt is strongly connected with the belt, core, and parabelt, but the parabelt has no significant connections with core areas A1 and R. Thus, the belt acts as a secondary, intermediate stage as information flows from the core to the belt and then parabelt. The core area RT appears to be the exception to this, as connections were observed between RT and RPB. Additional studies of RT should be conducted in order to provide a better understanding of the position of this area in the working model. Secondly, rostrocaudal topography is evident in the connections of the lateral belt and parabelt regions, consistent with connections of the medial belt and core regions described previously (de la Mothe 2006a). The main exception to this pattern is area RM, which is broadly connected with rostral and caudal areas.

Thus, at the present level of analysis, the organization of the auditory cortex of marmosets, owl monkeys and the macaque auditory cortex is very similar. However, it should be mentioned that some differences were noted that may impact future studies. At the gross anatomical level, the depth of the lateral and superior temporal sulci are greatly reduced in the marmoset compared to owl monkeys and macaques. The absence of the STS in some marmosets made identification of STS areas especially challenging and somewhat uncertain, although this region was consistently connected with the parabelt. At the architectonic level, we observed that the cytoarchitectonic and myeloarchitectonic borders between the lateral belt and parabelt were less distinct compared to macaques. AChE expression was also weaker overall in auditory cortex of the marmoset, and therefore less useful as a regional marker. In contrast, parvalbumin, vesicular glutamate 2, and SMI-32 have regional expression patterns that closely match that observed in macaques (Crum et al 2008; Paxinos atlas), suggesting that our AChE findings may reflect a methodological exception rather than a difference in neurochemical expression. As ongoing and future work in the marmoset may reveal other differences between species, it will be important to carefully document them as a means to better understand the precise relationships between structure and function at different levels of organization within the system.