- Top of page
- MATERIALS AND METHODS
- SUMMARY AND CONCLUSIONS
- CONFLICT OF INTEREST STATEMENT
- ROLE OF AUTHORS
- LITERATURE CITED
The corticobulbar projection to the hypoglossal nucleus was studied from the frontal, parietal, cingulate, and insular cortices in the rhesus monkey by using high-resolution anterograde tracers and stereology. The hypoglossal nucleus received bilateral input from the face/head region of the primary (M1), ventrolateral pre- (LPMCv), supplementary (M2), rostral cingulate (M3), and caudal cingulate (M4) motor cortices. Additional bilateral corticohypoglossal projections were found from the dorsolateral premotor cortex (LPMCd), ventrolateral proisocortical motor area (ProM), ventrolateral primary somatosensory cortex (S1), rostral insula, and pregenual region of the anterior cingulate gyrus (areas 24/32). Dense terminal projections arose from the ventral region of M1, and moderate projections from LPMCv and rostral part of M2, with considerably fewer hypoglossal projections arising from the other cortical regions. These findings demonstrate that extensive regions of the non-human primate cerebral cortex innervate the hypoglossal nucleus. The widespread and bilateral nature of this corticobulbar connection suggests recovery of tongue movement after cortical injury that compromises a subset of these areas, may occur from spared corticohypoglossal projection areas located on the lateral, as well as medial surfaces of both hemispheres. Since functional imaging studies have shown that homologous cortical areas are activated in humans during tongue movement tasks, these corticobulbar projections may exist in the human brain. J. Comp. Neurol. 522:3456–3484, 2014. © 2014 Wiley Periodicals, Inc.
Vocalization, chewing, swallowing, and respiration are extremely complex motor activities requiring a significant and timely contribution of tongue movements (Miller, 2002). It has long been recognized that tongue movements are comprehensively influenced by assemblies of integrated neuronal circuits located within the lower brainstem (Holstege and Kuypers, 1977; Holstege et al., 1977, 1983; Lowe, 1980; Jean, 1984, 2001; Miller, 1999; Sawczuk and Mosier, 2001; Jürgens, 2002; Gestreau et al., 2005; Hannig and Jürgens, 2005; Lund and Kolata, 2006; Yamada et al., 2005). However, very little is known about the role of the cerebral cortex in mediating tongue movements, which is quite surprising, because clinical observations have long noted that oromotor deficits occur in patients sustaining localized cortical injury (Meadows, 1973; Willoughby and Anderson, 1984; Horner et al., 1988; Robbins and Levin, 1988; Martin and Sessle, 1993; Robbins et al., 1993; Daniels and Foundas, 1997; Hamdy et al., 1997). Indeed, tongue weakness, dysarthria, dysphagia, and aspiration are common in patients suffering the most frequently occurring stroke, middle cerebral artery (MCA) occlusion (Miller, 1999; Umapathi et al., 2000; Falsetti et al., 2009; for review, see Hamdy et al., 2000; Singh and Hamdy, 2006; Michou and Hamdy, 2009). Reinforcing the existence of a cortical influence on swallowing and tongue movements are seminal reports showing that surface stimulation of the lateral precentral cortex in humans evokes swallowing, tongue, and mastication movements (Penfield and Boldery, 1937; Penfield and Welch, 1949; Woolsey et al., 1979), whereas more recent studies show similar effects following transcranial magnetic stimulation (TMS) of the precentral region (Meyer et al., 1997; Krings et al., 1997; Muellbacher et al., 1998, 1999, 2001; Rödel et al., 2003; D'Ausilio et al., 2009; Boudreau et al., 2013.
Although the TMS findings have been enlightening, renewed interest in the role of the cerebral cortex in mediating oromotor movements has largely transpired from functional imaging studies showing that multiple cortical sites are active when healthy human subjects perform tongue movements in isolation, or during integrated movements such as swallowing and speaking (Birn et al., 1998; Corfield et al., 1999; Hamdy et al., 1999; Mosier et al., 1999; Zald and Pardo, 1999; Kern et al., 2001; Martin et al., 2001, 2004; Mosier and Bereznaya, 2001; Malandraki et al., 2009; Sörös et al., 2009; Grabski et al., 2012). From these observations, a common neural network of cortical areas appears to influence tongue/orofacial movement including the ventral portion of the primary motor and somatosensory cortices, insula, anterior cingulate gyrus, supplementary motor cortex, and parietal cortex (Humbert and Robbins, 2007; Miller, 2008; Michou and Hamdy, 2009).
Of these brain areas, the cortex forming the ventral extension of the primary motor cortex (M1) and adjacent ventral premotor region has frequently been associated with mediation of tongue movements in non-human primates. Aspiration resection of both cortical areas has long been known to induce immediate postsurgical tongue dysfunction (Green and Walker, 1938; Luschei and Goodwin, 1975; Larson et al., 1980). Complementing these observations are physiological studies in non-human primates demonstrating that face, tongue, and mandibular movements occur following stimulation of this cortex (Horsley and Schäfer, 1888; Leyton and Sherrington, 1917; Walker and Green, 1938; Woolsey et al., 1952; Cure and Rasmussen, 1954; Luschei et al., 1971; Luschei and Goodwin, 1974; McGuinness et al., 1980; Sessle and Wiesendanger, 1982; Huang et al., 1988; Murray and Sessle, 1992; Hatanaka et al., 2005). These physiological observations are grounded, in part, by anatomical findings in monkeys showing that the ventral region of M1 projects to multiple cranial nerve motor nuclei including the hypoglossal nucleus (Kuypers, 1958a; Kuypers and Lawrence, 1967; Jürgens and Alipour, 2002), facial nucleus (Kuypers, 1958a; Kuypers and Lawrence, 1967; Jenny and Saper, 1987; Morecraft et al., 2001), and trigeminal motor nucleus (Kuypers, 1958a; Kuypers and Lawrence, 1967).
As for the other constituents of the proposed cortical tongue/swallowing network, far less is known about their role in mediating tongue movements, but some observations are suggestive. Foremost are stimulation studies showing that orofacial responses occur following stimulation of the rostral region of M2 in monkey (Woolsey et al., 1952; Mitz and Wise, 1987; Luppino et al., 1991; Godschalk et al., 1995) and human (Penfield and Welch, 1951; Talairach and Bancaud, 1966; Fried et al., 1991). In monkey, orofacial movements have also been observed following microstimulation of the cortex lining the lower bank of the cingulate sulcus, possibly corresponding to the rostral part of the rostral cingulate motor cortex (Godschalk et al., 1995) and rostral part of the caudal cingulate motor cortex (Luppino et al., 1991; Godschalk et al., 1995). Previous monkey tracer studies have been less convincing, indicating that M2 (Künzle, 1978; Jürgens, 1984; Wiesendanger and Wiesendanger, ) and the ventrolateral premotor cortex may not project to the hypoglossal nucleus (Künzle, 1978; Simonyan and Jürgens, 2003), which, in part, prompted the current investigation using high-resolution tract tracing methodology.
What has attracted our attention is that a number of homologous regions in the monkey cortex, which have been shown in human neuroimaging work to be associated with tongue movements, have been shown to project to the facial nucleus of the lower pons (Jenny and Saper, 1987; Morecraft et al., 2001; Gong et al., 2005). Based on these findings, physiological observations (Woolsey et al., 1952, Mitz and Wise, 1987; Luppino et al., 1991; Godschalk et al., 1995), and topographical patterns of cortical interconnections with the face region of M1 (Muakkassa and Strick, 1979; Morecraft and Van Hoesen, 1992; Morecraft et al., 1996; Tokuno et al., 1997), we proposed the existence of an integrated network of face regions including the ventral part of the M1and lateral premotor cortex (LPMCv), as well as the rostral part of the supplementary (M2), rostral cingulate (M3), and caudal cingulate (M4) motor cortices that may uniquely influence volitional and emotional components of facial expression (Morecraft et al., 1996, 2001, 2004b, 2014). Whether or not these cortical face regions constitute orofacial representations, or craniocervical representations, has yet to be determined.
Given that the neuroscientific cornerstones characterizing the M1 face/head region include the demonstration of physiologically evoked cranial movements and direct corticobulbar projections to multiple cranial nerve motor nuclei, it is possible that other frontal and cingulate regions shown physiologically to elicit orofacial movements, and anatomically to issue corticofacial projections, may also project to other cranial nerve motor nuclei much like the ventral M1. This could reinforce perspectives on cortical face/head representation in the debate on somatotopic motor organization currently dominated by viewpoints focusing on cortical forelimb and hindlimb representation. Such findings could broaden our current approach toward investigating the potential role of cortical areas, in addition to the ventral M1, that may contribute to neurological disorders adversely affecting cranial nerve motor function. Clinical intuition would also suggest that this information may assist in expanding our initiative of localizing cortical areas that may favorably contribute to the recovery of orofacial movements following supratentorial brain injury. Such considerations prompted the present investigation to explore the possibility that corticobulbar projections to the hypoglossal nucleus arise from regions of the monkey cerebral cortex that innervate the facial nucleus and correspond to homologous components of the human cortical swallowing network indicated in the functional imaging literature. Because the insula, frontal operculum, anterior cingulate gyrus, and lateral parietal cortex have also been implicated in this network, these areas were included in our investigative analysis.
- Top of page
- MATERIALS AND METHODS
- SUMMARY AND CONCLUSIONS
- CONFLICT OF INTEREST STATEMENT
- ROLE OF AUTHORS
- LITERATURE CITED
Dating back to the seminal works of Hans Kuypers on descending cortical projections to the lower brainstem (Kuypers 1956, 1958a, 1958b), it was recognized that the hypoglossal nucleus receives a direct, and massive, cortical projection from the ventral region of the precentral cortex in the human and non-human primate brain. Our findings reinforce this classical discovery in the monkey model and show for the first time that the corticohypoglossal projection is not unique to the ventral precentral area, as it arises, at much lesser intensity, from many parts of the cortical mantle (Fig. 14). Although not as prominent as the M1 projection, we found a fairly robust corticohypoglossal projection from the ventral part of the LPMCv and rostral part of M2. A comparatively weaker, but potentially influential projection arose from area ProM, the periarcuate genu region of the LPMCv and LPMCd, the rostral part of M3, and the rostral part of the M4, including an unsuspected projection from the pregenual region of the anterior cingulate gyrus. A small projection was also found to the hypoglossal nucleus from the face/head area of the primary somatosensory cortex and rostral region of the insula. From the perspective of corticomotor somatotopy, the sizable hypoglossal projection from the ventral part of the M1 and LPMCv, as well as notable projections from rostral parts of the medial motor cortices (M2, M3, and M4), strengthens the proposed concept of five orofacial motor representations located in the frontal and cingulate motor cortices in the non-human primate brain (Morecraft et al., 2001, 2004b). Our experimental findings additionally show that the cortical orofacial representations are positioned ventral (for lateral motor areas) and anterior (for medial motor areas) to adjacent arm representations. Finally, our experiments indicate that the cortex, thought to predominantly regulate arm movements in the dorsolateral premotor region (LPMCd), may, to some degree, influence hypoglossal motor mechanisms.
Figure 14. Summary diagram illustrating the main findings of the present study. We localized bilateral projections to the hypoglossal nucleus from 11 different regions of the cortical mantle involving frontal, cingulate, parietal, and insular territories of the telencephalon. Given the highly diverse and specialized functions regulated by these cortical regions, our findings suggest that hypoglossal mechanisms are likely to be influenced by motor- and sensory-related parts of the cerebral cortex, as well as multimodal association and limbic-related regions of the cortex. The bilateral nature and widespread origin of these projections suggests that cortical input to the non-human primate hypoglossal nucleus is highly protected, and the redundancy of this connection may underlie favorable recovery of tongue movement following localized supratentorial brain injury. For abbreviations, see list.
Download figure to PowerPoint
Corticohypoglossal projections from the primary motor cortex (M1)
The corticohypoglossal projection from the ventral region of M1 was by far the most intense corticofugal projection found in our study (Figs. 4B, 4D, 7, Table 4). The finding of a powerful and bilateral projection from the ventral region of M1 in the monkey is consistent with previous studies (Kuypers, 1958a; Jürgens and Alipour, 2002). In terms of laterality, we found a slightly predominant contralateral projection in four of the five experiments conducted (Fig. 6A, Table 4), which is in agreement with Kuypers's observations in monkey and chimpanzee (Kuypers, 1958a). These observations correlate with human TMS studies showing that the corticohypoglossal projection from M1 is bilateral (Urban et al., 1994, 1996; Meyer et al., 1997; Ghezzi and Baldini, 1998; Muellbacher et al., 1998, 1999, 2001; Rödel et al., 2003; Bawa et al., 2004), with more contralateral influence in most individuals, characterized physiologically by lower motor thresholds and larger mean evoked motor potentials (Ghezzi and Baldini, 1988; Muellbacher et al., 1998, 1999, 2001; Rödel et al., 2003; Bawa et al., 2004). Our finding of a slight contralateral predominance in most experiments may correlate with recent findings that a small angle of tongue deviation occurs in the majority of healthy individuals (Wei et al., 2012), although hemispheric dominance, handedness, and direction of tongue deviation in healthy controls were not considered.
Our series of neurological experiments were partly designed to gain insight into possible connectional differences in the projection from dorsal versus ventral regions of the M1 orofacial representation. We found that the corticohypoglossal projection is much stronger from the ventral part of the M1 orofacial representation than from the dorsal part. For example, in case SDM32, an equal amount of a compatible tract tracer was injected in a single location in the dorsal (SDM32-BDA) and ventral sectors (SDM32-FD) of M1 (Figs. 1, 4A,B). The results demonstrated a greater number of estimated boutons from the ventral site, on both the ipsilateral and contralateral sides of termination (Table 4). Likewise, large injection sites of LYD were made dorsally in case SDM35, and ventrally in case SDM66, with the latter being slightly less in tracer volume (Fig. 1, Table 1). Both injection sites gave rise to the highest number of estimated boutons found in all cases with the ventral injection site (SDM66-LYD) giving rise to double the number of ipsilateral and contralateral boutons compared with the dorsal injection site (SDM35-LYD), even though more tracer was injected into SDM35. This finding would correlate with reports in monkeys showing that electrophysiologically evoked facial movements occur more commonly in the dorsal M1 orofacial region, with tongue responses occurring more frequently ventrally (McGuinness et al., 1980; Huang et al., 1988; Murray and Sessle, 1992; Hatanaka et al., 2005). This finding closely corresponds to human TMS work showing overlap of M1 face and tongue movements with the primary lingual area located laterally (Rödel et al., 2003).
Another interesting feature of the M1 projection is the relatively heavy labeling found in many regions of the nucleus (i.e., medial, lateral, dorsal, and ventral) (Figs. 4B, 4D, 7), in contrast to the ventral M1 projection to the facial nucleus, which is topographically specific. For instance, in the monkey the M1 projection dominantly innervates the ventrolateral region of the contralateral facial nucleus (Kuypers, 1958a; Jenny and Saper, 1987; Morecraft et al., 2001), which contains motor neurons supplying the lower facial muscles (Jenny and Saper, 1987; Morecraft et al., 2001). Thus, injury to ventral M1 in non-human primates produces lower facial paralysis, with sparing of the upper facial muscles (Green and Walker, 1938; Symon et al., 1975; Larson et al., 1980), a condition that characteristically occurs following unilateral MCA infarction in stroke patients (Afifi and Bergman, 1980; Brodal, 1981; Adams et al., 1997). Given this example, the widespread nature of M1 labeled terminals occurring within the hypoglossal nucleus (Figs. 4B, 4D, 7) does not readily foster predictions regarding which tongue muscles may be more adversely affected following ventral M1 damage. However, Sokoloff and Deacon (1992) have shown in Macaca fascicularis that the hypoglossal nucleus is musculotopically organized. Thus, future studies should be pursued to determine whether terminal bouton density varies within these musculotopic subsectors and relates to preferential M1 innervation of some intrinsic and extrinsic tongue muscles.
In our efforts to further localize the corticohypoglossal projection from M1, we examined potential hypoglossal nucleus labeling from injection sites located in the arm and leg representations of M1 (Fig. 3). We were surprised to find evidence of very few labeled corticohypoglossal terminals following injections into the electrophysiologically defined arm area of M1, a finding that is in extreme contrast to the estimated 218,847 corticospinal terminal boutons located in the cervical enlargement (C5-T1) estimated for the SDM61 injection site (Morecraft et al., 2013; SDM61-LYD, Table 2). The virtual absence of a projection to the hypoglossal nucleus from the arm area of M1 in our material correlates well with the observations of Kuypers (1958a) and Leichnitz (1958c), who found no evidence of a corticohypoglossal projection from the arm region using older tract tracing methodology. Similarly, in one M1 hindlimb injection case (Fig. 3, top), we were unable to find terminal labeling in the hypoglossal nucleus, in sound agreement with Kuypers (1958a). Overall, these findings provide strong connectional support for the general view of a somatotopically organized M1 simiusculus in monkey (Woolsey et al., 1952) and, by association, an ordered homunculus in the human primary motor cortex (Penfield and Boldery 1937; Penfield and Welch, 1951; Woolsey et al., 1979). However, our findings of an extremely sparse corticohypoglossal projection from the arm/hand region of M1 may have scientific value in predicting that a few labeled cells may occur in the arm/hand representation area of M1 following injections of retrograde transneuronal viral tracer into the non-human primate tongue, or following injection of retrograde tract tracer into the hypoglossal nucleus.
Corticohypoglossal projections from the lateral premotor cortex and ProM
We localized a significant corticohypoglossal projection from the ventral region of the lateral premotor cortex (area 6Vb), which accounted for the second strongest corticobulbar projection in our investigation (Table 4). Several previous studies have examined the possibility of a corticohypoglossal projection from the ventrolateral premotor region in the monkey (Kuypers, 1958a, Künzle, 1978; Simonyan and Jürgens, 2003; Borra et al., 2010). Among these, Kuypers's (1958a) report was the only investigation to indicate that the cortex corresponding to ventral area 6 (area FBA and area FCBm) may project to the hypoglossal nucleus, as he inconclusively stated that this cortical region “does not contribute substantially to the projections” innervating the hypoglossal nucleus when compared with the ventral area 4 (M1). With respect to the report of Simonyan and Jürgens (2003), their BDA injection sites were placed into an electrophysiologically defined laryngeal region of area 6Vb (Simonyan and Jürgens, 2003; see their Fig. 1). As a result, they found heavy terminal labeling in the reticular formation of the lower medulla indicating an adequate postinjection survival interval for BDA transport to the lower brainstem, but found no terminal labeling in the adjacent hypoglossal nucleus. Several explanations may account for this incongruity.
First, the injection site in the Simonyan and Jürgens (2003) work was very close in proximity to the area 47L cytoarchitectonic border of the prefrontal cortex, thus being located slightly anterior and inferior to our ventral LPMCv injection site (SDM22-BDA) (see their Fig. 1 compared with our Fig. 1). Likewise, our ProM injection site (SDM40-BDA) was located on the lateral opercular convexity of the frontal cortex (Figs. 1, 4G), slightly more ventral than the three injection sites analyzed by Simonyan and Jürgens (2003). In Künzle's comprehensive study (1978), the lack of terminal hypoglossal labeling from his ventrolateral premotor injection cases may be due to technical/methodological limitations. For instance, Künzle used the autoradiographic tract tracing procedure, which is less sensitive than the dextran tract tracing method (Morecraft et al., 2014), and the short postinjection survival period (2–5 days) may not have been long enough for sufficient long-distance axonal transport of the injected amino acids. Lastly, Borra and colleagues (2010) investigated brainstem projections from the dorsal part of area 6Va (area F5 according to their nomenclature) and did not report a hypoglossal projection, but injection site location may account for this discrepancy. Our injection site (SDM57-BDA, Fig. 1) was located on the gyral convexity of dorsal area 6Va (Fig. 1), whereas the three injection cases studied by Borra and co-workers (2010) were located in the cortex lining the posterior bank of the arcuate sulcus, directly below the arcuate spur. From this, it can be deduced that the corticohypoglossal projection arises from cortex on the gyral surface of dorsal area 6Va and not from cortex lining the adjacent arcuate sulcus region. Our observation is reinforced by intracortical microstimulation work demonstrating that orofacial movements are evoked from this cortical location (Gentilucci et al., 1988; Godschalk et al., 1995; Graziano et al., 2002). The finding of corticohypoglossal projection neurons in this area is also supported by the observation that the dorsal region of area 6Va may be involved with the tongue deviation and protrusion that accompanies the action of bringing a grasped hand to the mouth (Graziano et al., 2002). However, it is likely that the long-duration stimulus trains used to evoke these movements would evoke temporal summation in multiple frontal cortical regions that receive cortical projections from the LPMCv and also innervate the hypoglossal nucleus, including ventral M1.
We also encountered an unanticipated corticohypoglossal projection from the lateral premotor cortex located just dorsal to the genu/spur region of the arcuate sulcus (SDM23-FD and SDM57-FD; Fig. 1, Table 4). Traditionally, this frontal region is recognized for its involvement in the motor preparation and execution of voluntary upper extremity movements (Davare et al., 2006; Hoshi and Tanji, 2007). Interestingly this cortex has been shown to issue corticofugal projections to the spinal cord (Martino and Strick, 1987; Dum and Strick, 1991; He et al., 1993) as well as the facial nucleus (Morecraft et al., 2001; SDM23-FD, Fig. 12). Taken jointly, these structural observations show that the dorsal periarcuate region contains either a mixture of corticospinal and corticobulbar projection neurons, neurons that give rise to collateral projections to multiple motor targets in the brainstem and spinal cord, or a combination of these potential neuronal phenotypes. Understanding the functional significance of these orofacial projection neurons in this physiologically characterized arm-related cortex in the non-human primate would appear to be of considerable interest as they may be spared following ventrolateral brain injury and may play a role in the recovery of tongue and facial movements.
Corticohypoglossal projections from the supplementary motor cortex
In the early 1950s, a somatotopically organized supplementary motor cortex, anchored by a face/head representation rostrally, was described in the human (Penfield and Welch, 1951) and monkey (Woolsey et al., 1952) brain. However, the available neuroanatomical literature concerned with studying the brainstem projections from the supplementary motor cortex did not reveal a corticohypoglossal projection (DeVito and Smith, 1959; Künzle, 1978; Jürgens, 1984; Wiesendanger and Wiesendanger, 1984), which is in contrast to the bilateral projection found in our study (Figs. 5B, 10, Table 5). This difference is likely due to the less sensitive tract tracing methods available at the time these previous investigations were carried out (Morecraft et al., 2014). In terms of pinpointing the origin of this projection, our comparative analysis of corticofugal projections from multiple injection sites allowed us to localize the corticohypoglossal projection to the rostral (face/head) region of M2. For example, the injection site in the preSMA (SDM27-BDA), did not give rise to hypoglossal afferents, and the injection site in the physiologically defined arm region of M2 gave rise to very few hypoglossal projections and numerous corticospinal labeled terminals (McNeal et al., 2010; Table 2, SDM54, SDM77). In contrast, a considerable portion of the SDM35-FD injection site was located anterior to the coronal plane through the anterior commissure (Fig. 2), and gave rise to numerous hypoglossal labeled terminals as well as corticospinal labeling, indicating involvement of both the rostral part of the M2 arm representation and its adjacent orofacial representation. Furthermore, the injection in case SDM35-BDA was positioned immediately anterior to the SDM35-FD injection (Fig. 2) and produced very few hypoglossal labeled terminals (Table 5). Collectively, this set of experiments provides strong support for a somatotopically organized supplementary motor cortex with face/head representation being positioned rostral to the arm representation (Woolsey et al., 1952; Mitz and Wise, 1987; Godschalk et al., 1995; Luppino et al., 1991).
Corticohypoglossal projections from the cingulate cortex
To our knowledge, this is the first study to investigate the corticohypoglossal projection from the anterior cingulate cortex. We found three distinct cingulate regions giving rise to a corticohypoglossal projection including the rostral part of M3, the rostral part of M4, and the pregenual region (area 24/32) of the gyral component of this cortex. Although these are considerably weaker than the ventral frontal lobe projections, their functional contribution to tongue movement may be related to motivational and other higher order behaviors, particularly when one considers the widespread and diverse cortical inputs to these cingulate regions (Vogt and Pandya, 1987; Carmichael and Price, 1995, 1996; Barbas and Blatt, 1995; Morecraft and Van Hoesen, 1998; Barbas et al., 1999; Morecraft et al., 2004a, 2012) versus the highly localized cortical inputs converging on the ventrolateral precentral and premotor regions (Matelli et al., 1986; Morecraft and Van Hoesen, 1993; Tokuno et al., 1997). Indeed, the functions mediated by the anterior cingulate include the integration of negative affect, pain and cognitive control (Shackman et al., 2011), set-shifting (Bissonette et al., 2013), conflict monitoring, and decision making (Carter and van Veen, 2007; Shenhav et al., 2013), as well as self-regulation (Posner et al., 2007). Anatomically, we have shown that the cortex lining the lower bank of the cingulate sulcus, corresponding to the orofacial region of M3 and M4, receives an extensive set of higher order cortical inputs including multimodal association and limbic cortical afferents from prefrontal, orbitofrontal, and medial temporal sources (Morecraft et al., 2004a [see case 6 for M4], 2012 [see cases 5 and 6 for M3]), which link this cortex to such complex associative cingulate functions. Similarly, the pregenual region of the anterior cingulate gyrus receives strong medial temporal lobe inputs that are coupled to higher order memory functions (Barbas and Blatt, 1995, Carmichael and Price, 1995, 1996; Barbas et al., 1999; Morecraft et al., 2012). Potentially adding to this rich complexity of cortical behavioral control is the fact that both M3 and the pregenual cingulate region receive extensive amygdala inputs (Van Hoesen, 1981; Amaral and Price, 1984; Barbas and De Olomos, 1990; Carmichael and Price, 1995; Morecraft et al., 2007a), implying a particularly strong emotionally related influence on these cingulate corticohypoglossal projection areas.
Corticohypoglossal projections from the parietal cortex and insula
In one experimental case we found a light corticohypoglossal projection from the ventral region of the gyral portion of the somatosensory cortex (areas 1 and 2) (Table 4). The significance of this weak projection is unclear, but it is interesting to note that the paucity of a strong projection to this brainstem “motor” nucleus parallels, in principle, the paucity of direct corticospinal projections to motor neurons in lamina IX of the cervical spinal cord following injections of tract tracer into the gyral part of the arm region of S1 (areas 1 and 2) and upper part of the cortex lining the posterior bank of the central sulcus (areas 3b and 1) (Coulter and Jones, 1977). However, our injection site in the face/head region of S1 in case SDM66-FR did not involve the cortex lining the depths of the central sulcus, which could potentially increase the likelihood of detecting more significant corticohypoglossal projections from S1 than found in the current study because the anterior bank of the central sulcus harbors the majority of descending monosynaptic projections to spinal motor neurons compared with the gyral portion of the frontal motor cortex (Cheney et al., 1991; Rathelot and Strick, 2006, 2009). Indeed, future experiments are required to more thoroughly address this issue with injections of dextran tracers placed into cortex forming the posterior bank of the central sulcus (areas 3 and 1). It is possible that S1 (and M1) projections to the hypoglossal nucleus mediate hypoglossal interneurons and first-order axons from tongue muscle afferents that project to the hypoglossal nucleus for sensory influences on control of tongue motion.
To our knowledge, a projection from the insula to the hypoglossal nucleus has not been reported in the monkey model. In our experiment, the projection was found to be light (Table 1, case SDM66-BDA). However, this must be interpreted cautiously, as the BDA injection site itself was very small and the resultant labeling may not be fully representative of the entire insular–hypoglossal projection. However, our finding does demonstrate that the non-human primate insula has a direct projection to the hypoglossal nucleus and that further study is justified to reinvestigate this issue with larger injection sites of high-resolution anterograde tracers. Functionally, it has been shown with magnetoencephalography (MEG) that cingulate and insular activation begin 2,000 ms prior to the initiation of swallowing and continue throughout swallowing, perhaps being essential for the initiation of swallowing (Watanabe et al., 2004). This is supported by the finding that dysphasia occurs in patients with an isolated stroke confined to the anterior insula, but not posterior insula (Daniels and Foundas, 1997).
One goal of the current study was to investigate the possibility that regions of the monkey frontal and cingulate cortex issuing corticobulbar projections to the facial motor nucleus of cranial nerve VII (Morecraft et al., 2001) also give rise to corticobulbar projections to the hypoglossal motor nucleus of cranial nerve XII. Our incentive to investigate this possibility originated partly from the idea that demonstrating common cortical regions projecting to multiple cranial nerve motor nuclei could reinforce perspectives on cortical face/head representation in the debate on somatotopic corticomotor organization, which is dominated by investigations and viewpoints focusing on forelimb and hindlimb representation (Murray and Coulter, 1981; Hutchins et al., 1988; Luppino et al., 1994; He et al., 1993, 1995). We found that the same general regions of the frontal lobe and cingulate cortex known to project to the facial nucleus also issued projections to the hypoglossal nucleus, thus underscoring their recognition of cortical orofacial representations. Furthermore, as discussed earlier, these regions are located ventral to the motor arm representations on the lateral surface (i.e., M1 and LPMCv), and anterior to arm representations on the medial cortical surface (i.e., M2, M3 and M4). However, future studies are needed to examine whether a general topography for these corticobulbar projections exists within each orofacial representation, and whether there are projections to the trigeminal motor nucleus of cranial nerve V.
Clinical considerations of potential widespread corticohypoglossal projections
Although the most common form of stroke involves the middle cerebral artery (MCA), which supplies the orofacial representation of M1, the occurrence of clinically detectable tongue paresis following peri-Rolandic injury in humans is rare (Urban et al., 1996, 1997; Muellbacher et al., 1998). However, this clinical outcome has been described following small, localized unilateral M1 injury (Willoughby and Anderson, 1984; Ghika et al., 1989; Durieu and Leys, 1994; Urban et al., 1996, 1997; Muellbacher et al., 1998) and following transient ischemic attacks (Wei et al., 2012). Experimentally, it has been shown in stroke patients with contralesional lingual paresis that TMS stimulation of the lesioned hemisphere did not evoke lingual movement in the five patients studied, but stimulation of the nonlesioned hemisphere did evoke bilateral genioglossus muscle activity and tongue movement despite the clinical presence of paresis (Muellbacher et al., 1998). In a follow-up study, after complete recovery of tongue movement, stimulation of the lesioned hemisphere again did not produce lingual movements in four of the five patients (Muellbacher et al., 1999). However, TMS of the nonlesioned hemisphere elicited symmetric lingual movement, demonstrating a potential role for uncrossed corticobulbar projections from the intact hemisphere in the recovery process.
Our findings are consistent with this evidence, as we found very prominent ipsilateral terminal labeling from M1 in all of our experimental cases (Table 4). Our work also suggests that spared cortical areas on the dorsolateral surface (LPMCd,) as well as the medial surface (M2, M3, M4, and the anterior cingulate gyrus) may support tongue motor recovery following unilateral cortical stroke, which needs to be assessed in future patient studies. Indeed, if they are present in the human brain, our findings of medial wall corticohypoglossal projections may be of particular importance for favorable recovery of tongue movement control in patients with a history of MCA strokes that initially affect one hemisphere, followed by a second stroke affecting the other hemisphere (Umapathi et al., 2000), or in patients sustaining bilateral opercular MCA occlusion (Ferrari et al., 1979; Grattan-Smith et al., 1989; Nisipeanu et al., 1997; Suresh and Deepa, 2004)
Conversely, damage to the medial wall of the hemisphere from anterior cerebral artery (ACA) stroke is a rare clinical phenomenon (Bogousslavsky and Regli, 1990; Kazui et al., 1993; Kumral et al., 2002; Carrera et al., 2007, for review, see Brust, 1992), as are alterations in clinically detectable simple voluntary tongue movements after such injury (Critchley, 1930). However, some observations indicate that ACA stroke results in higher order oromotor complications including speech disturbances, deficits in articulation (dysarthria), and swallowing difficulties (dysphagia) (Critchley, 1930; Bogousslavsky and Regli, 1990; Chamorro et al., 1997). Thus, it is possible that the hypoglossal projections from the medial wall, if present in the human brain, may subserve a complex or supportive role in articulation and swallowing compared with a more movement-specific related role for the lateral corticohypoglossal projections.
Study limitations and future research
The present study attempted to estimate the number of terminal-like immunoreactive particles (boutons) within the anatomically defined boundaries of the hypoglossal nucleus. It is likely that many of these terminals interact with cell bodies and dendrites of hypoglossal motor neurons that are confined to this anatomical region, but some caution should be exercised in this interpretation. Specifically, we do not know from the present work whether these axon projections terminate on hypoglossal motor neuron somas or their dendritic processes, or on hypoglossal interneurons (Cooper, 1981; Sokoloff and Deacon, 1992). Furthermore, it is possible that corticohypoglossal terminals form synaptic contacts with dendrites located within the anatomical confines of the hypoglossal nucleus that arise from cell bodies residing in the opposite hypoglossal nucleus (Cooper, 1981) or adjacent reticular formation. Additionally, we do not know from our material whether these terminals are excitatory or inhibitory. Indeed, future studies are required to address these intriguing synaptic possibilities.
As pointed out for the M1 projection, it will be of significant interest to determine whether the projections arising from each cortical area preferentially terminate in distinct musculotopic subsectors of the hypoglossal nucleus based on the monkey musculotopic template mapped out by Sokoloff and Deacon (1992). Indeed, in contrast to the heavy and widespread projections from M1, as discussed previously, that included peripheral and central targets of the nucleus, many of the non-M1 projections tended to innervate the peripheral regions of the nucleus, which has been shown by Sokoloff and Deacon (1992) to harbor motor neurons innervating the intrinsic tongue muscles.
As we have stressed, it is not likely that any of the non-M1 projections will rival the anatomical density and physiological influence of the M1 hypoglossal projection. However, it is important to note that the terminal density of many of the non-M1 hypoglossal projections may be underestimated in our study, including the insular, cingulate motor area, and pregenual cingulate projections, because these values were derived from experimental cases that had very small injection sites (Figs. 1, 2, 4G, 5C,E,G, Tables 1, 2). The small, localized nature of these injection sites was intended to avoid tracer spread into adjacent cortical architectonic areas and nearby subcortical gray matter structures. Indeed, the core of nearly all of our cingulate motor area injection sites did not involve the cortex lining the upper and lower parts of the cingulate sulcus fundus, which, if involved, would have likely increased the total number of estimated terminal boutons for these pathways. Future studies with large cingulate and insular injection sites could appropriately address this issue. However, a distinct advantage of these small injection sites lies in the fact that they were highly effective in pinpointing the specific cytoarchitectonic origin of these nonfrontal corticohypoglossal projections.
Finally, it is important to point out that there are many indirect pathways of cortical origin that are likely to have a significant influence on tongue movements including corticostriate, corticopontine, and corticoreticular projections, for example. In fact, the corticoreticular projection from the ventral region of M1 and adjacent ventrolateral premotor cortex in the monkey has been shown to be quite extensive (Kuypers, 1958a). Other related work has shown that premotor neurons in the pontine and medullary parvocellular reticular formation project directly to the hypoglossal nucleus (Holstege and Kuypers, 1977; Holstege et al., 1977, 1983). Specifically, the lateral propriobulbar projection system is derived primarily from the lateral part of the lateral tegmental field and innervates the ipsilateral hypoglossal nucleus, whereas the medial propriobulbar projection system originates from the medial part of the lateral tegmentum and tends to project bilaterally to the hypoglossal nucleus (Holstege and Kuypers, 1977; Holstege et al., 1977). In addition to contributing to this multisynaptic projection system, it is possible that direct cortical projections to the lower bulbar reticular formation may terminate on dendrites originating from cell bodies residing in the hypoglossal nucleus that extend out into the nearby reticular formation (Cooper, 1981). Thus, future studies are required to investigate the anatomical characteristics of the indirect, corticoreticular projection from the non-M1 regions found to directly innervate the hypoglossal nucleus. The possibility that these descending corticoreticular projections may be quite strong compared with their corticohypoglossal counterparts could significantly add to our current level of understanding of cortical contributions to hypoglossal motor mechanisms.