Disynaptic cerebrocerebellar pathways originating from multiple functionally distinct cortical areas
Abstract
The cerebral cortex and cerebellum both play important roles in sensorimotor processing, however, precise connections between these major brain structures remain elusive. Using anterograde mono-trans-synaptic tracing, we elucidate cerebrocerebellar pathways originating from primary motor, sensory, and association cortex. We confirm a highly organized topography of corticopontine projections in mice; however, we found no corticopontine projections originating from primary auditory cortex and detail several potential extra-pontine cerebrocerebellar pathways. The cerebellar hemispheres were the major target of resulting disynaptic mossy fiber terminals, but we also found at least sparse cerebrocerebellar projections to every lobule of the cerebellum. Notably, projections originating from association cortex resulted in less laterality than primary sensory/motor cortices. Within molecularly defined cerebellar modules we found spatial overlap of mossy fiber terminals, originating from functionally distinct cortical areas, within crus I, paraflocculus, and vermal regions IV/V and VI - highlighting these regions as potential hubs for multimodal cortical influence.
Introduction
The environment provides constantly updating sensory signals that must be acted upon for an animal to perform basic behaviors necessary for survival, including navigating through the environment, feeding/foraging, and other goal-directed behaviors. These perception and action loops require various sensory modalities (i.e. somatosensation, audition, and vision) to be integrated and translated into directed motor output. Both the cerebral cortex (Stein and Stanford, 2008) and the cerebellum (Snider and Stowell, 1944; Rondi-Reig et al., 2014; Baumann et al., 2015) are major brain regions involved in this sensorimotor integration and translation. Connections between these two brain regions form one of the largest projection pathways in the brain and selective expansion of this cortico-cerebellar system occurs across evolution (Gutiérrez-Ibáñez et al., 2018; Smaers and Vanier, 2019). This reflects the importance of cerebrocerebellar communication, however, the precise functions of these pathways are not fully understood (Apps and Watson, 2013). A vital initial step in understanding the role of cerebrocerebellar communication is to have a comprehensive map of the pathways linking these two structures as well as the precise organization of the termination of these pathways within the cerebellum.
Due to the indirect nature of cerebrocerebellar connections, it has been difficult to study their organization in precise detail. While a large body of research has mapped out the corticopontine projections (Brodal and Bjaalie, 1997; Glickstein, 1997; Leergaard and Bjaalie, 2007; Proville et al., 2014) or the pontocerebellar projections (Pijpers and Ruigrok, 2006; Pakan et al., 2010; Proville et al., 2014; Biswas et al., 2019), few studies have utilized neurotropic viruses (Kelly and Strick, 2003; Suzuki et al., 2012). Therefore, the precise routes for information flow from the cortex to the cerebellum, including detailed terminal organization in the highly modular cerebellar cortex, have remained largely inferred. The integration of multimodal inputs to the cerebellum is a fundamental operation that would allow for the precise coordination of sensory-driven movements within this brain region. Anatomically, the potential for a single granule cell to receive multimodal input via both descending motor cortex and ascending proprioceptive pathways has been shown in mice (Huang et al., 2013). However, the potential for co-innervation originating from various cortical areas spanning multiple modalities is unknown, even on the regional level in the cerebellum, and has consequences for the role of cerebro-cerebellar-cerebro feedback loops in learning and predictive motor control (e.g. Chabrol et al., 2019).
Using a mono-trans-synaptic anterograde viral tracer (Zingg et al., 2017; Zingg et al., 2020), we investigated the precise cerebrocerebellar pathways from key sensory, motor, and association regions of the cortex via the pontine and other intermediate precerebellar nuclei to all regions of the cerebellar cortex (Figure 1A–C); ultimately providing a map of the potential pathways linking various functionally specific cortical regions with the cerebellum. Following injections into the primary motor (M1), somatosensory (S1), visual (V1), auditory (A1), posterior parietal association cortex (PPC), and the dorsal field of auditory cortex (AuD), we found a highly organized topography of labeled pontine cells, with the notable exception that injections into A1 produced only terminal labeling in the pons; indicating the lack of a A1-ponto-cerebellar pathway in mice. We quantified the number of resulting mossy fiber terminals and described their relationship to the internal organization of the cerebellum. The majority of labeled mossy fiber terminals from the primary sensory and motor cortical regions were in the contralateral cerebellar hemisphere, whereas from association cortices this laterality was less evident. Cortical influences were not restricted to the cerebellar hemispheres, as terminals spanned all regions of the cerebellar cortex, with biases depending on the cortical modality. Cerebellar subdivisions with the highest regional co-innervation of multimodal inputs were crus I, the paraflocculus (PFl), vermal lobule VI and lobules IV/V, highlighting the potential for modular multimodal processing of information originating from the cerebral cortex.

Anterograde tracing of indirect cerebrocerebellar pathways using a mono-trans-synaptic adeno-associated virus (AAV).
(A) Schematic outlining cortical target areas for mono-trans-synaptic anterograde tracer injections: primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), and dorsal auditory (AuD) cortex. (B) Principle of mono-trans-synaptic anterograde tracing (e.g cortico-pontine-cerebellar pathway) using a specific adeno-associated virus (AAV1.cre). (C) Schematic of gross anatomical divisions of the unfolded mouse cerebellum (according to Marani and Voogd, 1979). (D) Images of coronal sections showing representative injection sites into M1, S1, PPC, V1, A1, and AuD (from left to right). Arrowheads indicate regional borders. Distance from bregma is indicated based on the mouse stereotaxic atlas (Franklin and Paxinos, 2007). (E) Images of coronal sections illustrating the mono-trans-synaptic labeling in the pontine nuclei (i.e. corticopontine fibers, postsynaptic pontine cells, pontine fibers) following injections into M1, S1, PPC, V1, A1, and AuD. Note that the medial-lateral topography of pontine labeling correlates with the rostral-caudal localization of these cortical regions. Injections into A1 resulted in labeled fibers within the pons, but no labeled cells. (F) Images of coronal section showing representative mossy fiber labeling following an injection into M1 at different magnifications. Scale bars 1 mm (D, E, F left), 20 µm (insets in E), 200 µm (F middle), 50 µm (F right). Retrosplenial Cortex (RSD); secondary visual cortex (V2).
Results
To examine the topography of cerebrocerebellar pathways from primary sensory and motor cortical regions as well as sensory association areas, we utilized an AAV1.cre construct that has been shown to act as a trans-synaptic anterograde tracer that crosses a single functional synapse (mono-trans-synaptic; Zingg et al., 2017; Zingg et al., 2020). We injected AAV1.cre into various cortical regions in tdTomato reporter mice and quantified the resulting labeled precerebellar pathways and mossy fiber terminals in the cerebellum (Figure 1). Target cortical areas included M1 (forelimb/hindlimb regions), S1 (forelimb/hindlimb regions), PPC, V1, A1, and AuD (Figure 1A,D; three mice per target region; see Supplementary file 1A).
Anterograde tracing of indirect cerebrocerebellar pathways
Following cortical injections, mono-trans-synaptically labeled cells were found in the ipsilateral basal pontine nuclei (referred to as pontine nuclei throughout) from all cortical target regions, with the exception of A1 (Figure 1E). In contrast, after injections in the more dorsal secondary auditory region, AuD, a corticopontine pathway was observed (Figure 1E). The lack of corticopontine projections from A1, was confirmed by injecting CAV.cre into the pontine nuclei; this viral vector is preferentially taken up by axon terminals, resulting in retrograde labeling. Here we found retrogradely labeled cells in AuD and more ventral auditory cortex (AuV), but no labeled cells in A1 (Figure 1—figure supplement 1). Although anterogradely labeled cell bodies were not observed in the pontine nuclei following AAV1.cre injections in A1, labeled fibers were found in the dorsomedial portion of the ipsilateral pons (Figure 1E; Supplementary file 1B). These fibers could be either traversing through the pons, potentially towards lower brainstem structures (e.g., cochlear and vestibular nuclei see Figure 2A; see also Budinger et al., 2000), or disynaptic terminals resulting from labeled indirect pathways from A1 to the pontine nuclei, likely via the inferior colliculus (Schuller et al., 1991b; Caicedo and Herbert, 1993). Indeed, anterogradely labeled neurons were consistently observed in the inferior colliculus after A1 injections (Figure 2B) and following injection of AAV1.cre in the inferior colliculus, anterogradely labeled cells were found in the pontine nuclei as well as mossy fiber terminals in the cerebellum, demonstrating the potential for a trisynaptic A1-collicular-cerebellar pathway via the pontine nuclei (Figure 1—figure supplement 2).

Extra-pontine labeling in key precerebellar nuclei following mono-trans-synaptic adeno-associated virus (AAV) injections in target cerebral cortical regions.
(A) Representative labeling in precerebellar nuclei listed in Supplementary file 1B, after injections of AAV1.cre into primary motor (M1), primary somatosensory (S1), primary auditory (A1), and dorsal auditory (AuD) cortex. Images show labeled cells in interpolar part of the spinal trigeminal nucleus (Sp5I), lateral reticular nucleus (LRt), matrix region x (Mx), superior vestibular nucleus (SuVe), red nucleus (RPC, RMC), pontine reticulotegmental nucleus (RtTg), dorsal cochlear nucleus (DC) and ventral cochlear nucleus, posterior part (VCP), which provide alternative extra-pontine mossy fiber pathways for indirect cortical input to the cerebellum. Descending labeled fiber tracts can also be observed in the longitudinal fasciculus of the pons (lfp) and the pyramidal tracts (py). Further, fibers travelling to the cerebellar cortex in the middle and inferior cerebellar peduncles (mcp and icp, respectively) after S1 injections. (B) Labeled cells in the inferior colliculus (central [CIC], external cortex [ECIC] and dorsal cortex [DCIC] inferior colliculus) following A1 injections. (C) Labeled fibers in the ipsilateral inferior olive (IO) following S1 injections. For all, distance from bregma is indicated based on the mouse stereotaxic atlas (Franklin and Paxinos, 2007). Scale Bars 500 µm (A), 1 mm (B, C), 100 µm all insets. 4th ventricle (4V), vestibular nerve (8vn), nucleus of Darkschewitsch (Dk), lateral periaqueductal gray (LPAG), medial lemniscus (ml), medullary reticular nucleus, ventral (MdV), raphe obscurus nucleus (Rob).
In agreement with previous studies (Leergaard and Bjaalie, 2007; for review see, Kratochwil et al., 2017), resulting pontine labeling after cortical injections in mice was topographically organized with more rostral cortical regions projecting more medially in the pons and caudal cortical areas projecting towards the lateral extent of the pontine nuclei (Figure 1E, see also Figure 3B). While most pontine labeling was strictly ipsilateral, M1 was the only cortical target region that resulted in bilateral pontine labeling (average of 118 ± 36 neurons ipsilateral to 8 ± 3 neurons contralateral; Figure 1E). Previous anterograde tracer studies have observed corticopontine fibers largely in the ipsilateral pontine nuclei with relatively sparse labeling in the contralateral pons (Mihailoff et al., 1985; Leergaard and Bjaalie, 2007). However, it was unclear if these sparse contralateral projections were fibers of passage or axon terminals; our results suggest the later for M1 projections and the former for all other cortical targets regions described here. Interestingly, the existence of functional synapses for contralateral corticopontine projections was also reported in primates, however, also exclusively after M1 injections (Morecraft et al., 2018).

Organization of indirect cerebrocerebellar mossy fiber terminals in the cerebellum from target cerebral cortical regions.
(A) Schematic of cortical target areas as outlined on the cortical surface (according to Franklin and Paxinos, 2007; Kirkcaldie, 2012) including core (black) and halo (grey) regions of mono-trans-synaptic anterograde tracer injections: primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), and dorsal auditory (AuD) cortex. Scale bar is 1 mm. (B) Schematic illustrating the topographical location of pontine labeling following injections. Colored lines indicate regions where labeled pontine neurons were found and dots (following A1 injections) indicate regions where terminals were observed. (C) Percentage of mossy fiber labeling in the ipsilateral and contralateral cerebellum (left) and in its gross divisions (right): vermis, hemispheres and vestibulocerebellum (VbC). (D) Percentage of labeled mossy fiber terminals observed in the anatomical lobules (simplex lobule [Sim], crus I/II ansiform lobule [crus I/II], paramedian lobule [PM], copula pyramis [Cop], paraflocculus [PFI], flocculus [FL], vermal cerebellar lobules [I-X]) of the ipsilateral (left) and contralateral (right) cerebellum (% of the total cerebellar input).
We observed anterogradely labeled cells in various other precerebellar nuclei (Figure 2A; Supplementary file 1B; see also Ruigrok et al., 2015). In fact, after S1 injections we found more labeled cells in extra-pontine precerebellar nuclei than in pontine nuclei (70 ± 8% of total labeling) and after M1 injections, just under half (40 ± 3%). This labeling spanned several precerebellar nuclei, including the red nucleus and reticulotegmental nucleus, as well as more caudal brainstem regions including the spinal trigeminal nucleus, lateral reticular nucleus, the matrix region x and the vestibular nuclei (Figure 2, Supplementary file 1B). While the reticulotegmental nucleus is often grouped together with the basal pontine nuclei, these regions in the pons appear to play different functional roles (Cicirata et al., 2005), hence, in this study we classify the reticulotegmental nucleus as a separate extra-pontine region. Additionally, we observed labeled fibers in both the medial and inferior cerebellar peduncles (see Figure 2A), indicating disynaptic tracts projecting to the cerebellum from pontine regions and more caudal extra-pontine precerebellar nuclei, respectively. Since there was no labeling observed in the pontine nuclei after A1 injections, the resulting mossy fiber terminals must reach the cerebellum through other precerebellar nuclei, of which we found labeling in the vestibular as well as the cochlear nuclei (see Figure 2A).
We did not observe labeled cells from any target cortical region in the spinal or lateral vestibular nuclei, the external cuneate nucleus, or the inferior olive. We did, however, observe fibers and terminal labeling in the inferior olive following injections in S1 (Figure 2C) and to a lesser extent after injections in M1. Interestingly, also only after S1 and M1 injections, anterogradely labeled cells were observed in the matrix region x, which has been suggested as a candidate preolivary relay nucleus (Ackerley et al., 2006). There is some controversy regarding the existence of direct cerebro-olivary connections in rodents, particularly originating from M1 (Swenson et al., 1989; Baker et al., 2001; Ackerley et al., 2006); for review see Watson and Apps, 2019); our results support disynaptic input from the cerebral cortex to inferior olivary neurons. Sparse fibers were also observed in the lateral cerebellar nucleus following S1 and M1 injections, likely from pontocerebellar collaterals (Cicirata et al., 2005; Biswas et al., 2019).
Regional organization of cerebrocerebellar mossy fiber terminals
Cerebrocerebellar mossy fiber terminal labeling in the granule cell layer of the cerebellum was systematically present, regardless of the cortical origin (e.g. Figure 1F, Supplementary file 1B). We did not find AAV1.cre-induced tdTomato expressing granule cells or Purkinje cells in any of the cases, indicating that the AAV construct did not transfer to higher-order downstream structures beyond mono-synaptic connections (see also Zingg et al., 2017; Zingg et al., 2020). We quantified the number of resulting mossy fiber terminals throughout the cerebellum and found that this was highly variable across cortical regions, with S1 and M1 resulting in the largest number of labeled terminals (~10,000) and AuD the least (~580; see Supplementary file 1B). Although the absolute number of mossy fiber terminals was variable, the resulting regional pattern of labeling was highly consistent within, and specific to, the cortical region injected (Figure 3). To examine the underlying organization of cerebrocerebellar projections, we quantified the proportion of mossy fiber terminals within each side (Figure 3C), division (Figure 3C), and lobule (Figure 3D) of the cerebellum.
Cerebrocerebellar projections were largely, but not exclusively, to the contralateral cerebellum (average across all animals: contralateral 70 ± 3%, ipsilateral 30 ± 3%). This was especially prominent after S1 injections, which resulted in 85 ± 3% of mossy fiber terminals in the contralateral cerebellum. However, this laterality varied across functionally distinct cerebrocerebellar input, such that the ratio of the number of mossy fiber terminals on each side was closer to equivalent for association/secondary cortical regions compared to that of primary cortical areas (ratio of ipsilateral: contralateral = 0.73 ± 0.13 versus 0.35 ± 0.06, respectively; p=0.008; two-sample t-test). Although M1 injections resulted in bilateral pontine labeling (Figure 3B, see also Figure 1E), the resulting mossy fiber terminals were still largely located in the contralateral cerebellum (74 ± 2%; Figure 3D, see also Figure 1F).
With respect to gross cerebellar divisions (Figure 3C), input from M1 and S1 was biased towards the cerebellar hemispheres (83 ± 3% and 55 ± 5%, respectively), with much smaller contributions to the vestibulocerebellum (8 ± 3% and 13 ± 3% respectively). The proportion of cortical input to the vermis was quite similar across cortical regions (~30%), with the exception of V1, which had very few vermal projections (14 ± 1%) and A1, which showed a comparatively strong bias towards vermal projections (47 ± 7%). Interestingly, projections from AuD deviated from A1 with a high proportion to the cerebellar hemispheres and without the vermal bias. Projections from V1 were biased towards the vestibulocerebellum (53 ± 5%) and PPC also showed a larger proportion of inputs to the vestibulocerebellum in comparison to other cortical areas. In fact, PPC was the only area that showed largely equal proportions of projections across the cerebellar subdivisions with no clear majority (hemispheres: 40 ± 3%, vermis: 30 ± 7%, vestibulocerebellum: 30 ± 3%).
We then quantified the proportion of mossy fiber terminals in each cerebellar lobule across both the ipsilateral and contralateral cerebellum (Figure 3D). After injections into M1, most resulting mossy fiber terminals were in contralateral crus II ansiform lobule (crus II), followed by the simple lobule (Sim), paramedian lobule (PM), and the crus I ansiform lobule (crus I). Although the absolute number of terminals from M1 was higher in the contralateral cerebellum, the relative proportion of labeling across the lobules was strikingly similar for both sides, resulting in a highly symmetrical pattern of labeling (Figure 3D, see also e.g. Figure 1F). Injections into S1 resulted in mossy fiber terminals largely in contralateral crus I, PM, and the copula pyramis (Cop). Indeed, S1 was the only cerebral cortical region with substantial input to the Cop (14 ± 5% contralateral projections; no other regions were >1.7%). Additionally, there was a relatively high proportion of terminals in contralateral vermal lobule IV/V after S1 injections. In contrast to M1 projections, S1 showed relatively few terminals in the Sim and crus II. Although the terminal pattern of S1 labeling was also largely symmetrical, the Cop was a clear exception, with only a small relative proportion of ipsilateral labeling (0.7 ± 0.3%) compared to the contralateral side.
Injections in the association area, PPC, resulted in less laterality of terminal labeling within the cerebellum, as previously mentioned, however, there were substantial differences in the pattern of labeling between the ipsilateral and contralateral sides (Figure 3D). On the contralateral side, the relatively equal balance of mossy fiber projections to the cerebellar divisions is reflected in the high proportion of terminals to crus I (19 ± 6%; i.e. hemispheres), lobules IV-VI (10 ± 1%, i.e. vermis), and paraflocculus (PFl; 19 ± 6%, i.e. vestibulocerebellum). In contrast, the ipsilateral side showed less prevalent crus I labeling (4 ± 0.5%) but a relatively large proportion of projections to vermal lobule VI (13 ± 7%) and crus II (5 ± 2%). Therefore, cerebrocerebellar input from the PPC results in the lowest laterality, but also the least symmetrical, terminal organization.
With regard to the other sensory areas, after injections into V1, most labeled mossy fiber terminals were found in the contralateral PFl (44 ± 5%); while ipsilaterally the PFl also had a relatively high proportion of projections compared to other lobules (7 ± 2%), this remained less than a third of that to the contralateral PFl (Figure 3D). Crus I was the only region in the hemispheres with substantial terminal labeling from V1 (16 ± 6%, contralateral). Labeling in the vermis was sparse but highly symmetric, with lobule VI having the most prominent labeling (5 ± 2%), similar to M1 and PPC. Following A1 injections, a large proportion of the resulting mossy fiber terminals were found in the vermis (specifically lobules III [9 ± 5%], IV/V [12 ± 2%], and VI [8 ± 2%]), and in the cerebellar hemispheres there was a fairly equal distribution of terminals in crus I (8 ± 1%) and crus II (7 ± 1%). In general, A1 was the only cortical region with a similar proportion of projections specifically to the cerebellar hemispheres bilaterally (ipsilateral: 18 ± 2%; contralateral: 24 ± 2%). Finally, after injections into AuD, the most prominent mossy fiber labeling was in crus I (15 ± 3%) and crus II (8 ± 1%) followed by vermal lobules IV/V (7 ± 0.4%) and VI (7 ± 1%). In contrast to projections from A1, very few terminals were observed in lobule III (0.9 ± 0.1%). Although the absolute number of terminals from AuD was slightly higher in the contralateral cerebellum, the relative proportion of labeling across the lobules was strikingly similar bilaterally, resulting in a highly symmetrical pattern of labeling (Figure 3D).
Lobule co-innervation of cerebrocerebellar inputs from distinct cortical areas
To determine the spatial overlap of cerebrocerebellar input from various cortical regions we examined the distribution of the total number of mossy fiber terminals within each lobule for each side of the cerebellum. All cerebellar lobules, on both ipsilateral and contralateral sides, received some cerebrocerebellar projections, however, the proportion of terminals in each lobule varied widely (ranging from 0.1–19.3% of the total ipsilateral terminal labeling and 0.1–16.9% of the total contralateral labeling; Figure 4A,B). The cerebellar hemispheres (Sim, crus I, crus II, PM, and Cop) received the majority of the total cerebrocerebellar mossy fiber input (68.2% of ipsilateral terminals, 67.1% of contralateral terminals), followed by vermal lobules IV/V and VI together (18.1% of ipsilateral terminals, 17.7% of contralateral terminals). Interestingly, we found a notable scarcity of mossy fiber terminals in posterior vermis, especially vermal lobule VII (1.2% of ipsilateral terminals, 0.4% contralateral terminals), which is contrary to studies of pontocerebellar projections alone (Serapide et al., 1994). Considering its relatively smaller volume, the PFl also received a large proportion of inputs (7.9% of ipsilateral terminals, 11.6% contralateral terminals; Figure 4A,B).

Regional convergence of cerebrocerebellar terminals originating across functionally distinct regions of cerebral cortex.
(A–B) Proportion of labeled mossy fiber terminals in the ipsilateral (A) and contralateral (B) cerebellum following AAV injections into primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), or dorsal auditory (AuD) cortex. Values were normalized to the total number of labeled mossy fibers found within each lobule (simplex lobule [Sim], crus I/II ansiform lobule [crus I/II], paramedian lobule [PM], copula pyramis [Cop], paraflocculus [PFI], flocculus [FL], vermal lobules [I-X]) across all animals (right x-axis). (C) Density of mossy fiber (MF) terminals in the ipsilateral (left) and contralateral (right) cerebellum. Each dot represents the average density (number of MFs/mm3) after each cortical injection M1, S1, PPC, V1, A1 and AuD. See also Supplementary file 1C. (D) Schematic of the unfolded mouse cerebellum summarizing the topography of mossy fiber inputs. Multimodal lobules receive at least 4% of the total mossy fiber input for that side (see A-B) and are highly (solid color) or moderately (striped color) multimodal based on input from at least two different functional cortical regions in addition to the dominant one (>15% and<15%, respectively).
All cerebellar lobules also received mossy fiber inputs from two or more cortical regions; however, the proportion of cerebrocerebellar projections to a single lobule from each functionally distinct cortical region again varied substantially (Figure 4A,B). To examine the dominance of these projections from each target cortical region, we calculated the density of terminal labeling for each cerebellar subdivision according to the number of mossy fibers and the volume based on 16.4T MRT data (Ullmann et al., 2012; Figure 4C; Supplementary file 1C). In the cerebellar hemispheres, we found the highest density of terminals from motor input (M1), especially ipsilaterally, including Sim, crus I, crus II, and PM, but excluding the Cop (Supplementary file 1C). Bilaterally, the Cop had the highest density from somatosensory input; to the extent that 96% of the mossy fiber terminals found in the contralateral Cop originated in S1 (Figure 4B).
Lobules in the vermis had the highest density from S1 input (Figure 4C), except for ipsilateral lobule VI, which had a relatively large number of mossy fibers from M1 injections, and the most posterior vermal region, vermal lobules IX and X, which are part of the vestibulocerebellum (Figure 4A,B; see also Figure 3D). In general, the vestibulocerebellum had diverse cerebrocerebellar input. Bilaterally, the PFL and Fl had the highest density of terminals from S1 injections, although both also had substantial input from most other cortical regions (Figure 4A–C). Lobule IX and X also had sparse input from most cortical regions, however, IX had a proportionally large number of terminals from auditory regions (especially A1 but also AuD), while M1 was largely represented in lobule X (Figure 4A,B). The representation of inputs from A1 and AuD in vermal vestibulocerebellar regions is interesting considering the large number of labeled cells observed in the vestibular nuclei relative to other precerebellar nuclei from auditory cortical regions (see Supplementary file 1B). It is important to note, however, that most regions of the vestibulocerebellum (with the exception of the PFl) had a small density of mossy fiber terminals (Figure 4C, Supplementary file 1C); hence, while still representative, caution must be taken in interpreting the pattern of labeling within these regions.
We then classified the topography of mossy fiber inputs to the various cerebellar lobules according to the degree of regional overlap, as this would represent a prerequisite for potential multimodal processing between corticocerebellar inputs (Figure 4D). Since all cerebellar lobules received input from two or more cortical regions but some lobules had a low density of total mossy fiber terminals, we included subdivisions that received a minimum proportion of the total mossy fiber input for each side (>4%; see Figure 4A,B) and classified lobules as: ‘highly co-innervated’ if >15% of mossy fiber labeling was from at least two different functional cortical regions in addition to the dominant one, and ‘moderately co-innervated’ if this value was <15% (Figure 4D). Using these criteria, bilateral crus I, PFl, vermal lobule VI, and IV/V had the greatest potential for multimodal co-innervation; followed by bilateral Sim, crus II, PM, and contralateral Cop.
Spatial organization of mossy fiber terminals in molecularly defined cerebellar modules
Cerebellar lobules contain highly organized parasagittally oriented modules based on anatomical, physiological and molecular subdivisions (for review see Apps and Hawkes, 2009). Therefore, to determine more precisely the potential for spatial overlap of multimodal information, we examined the distribution of mossy fiber terminals across the mediolateral and rostrocaudal extent of the identified multimodal lobules. To do this, we used the parasagittal expression pattern of aldolase C (or zebrin Brochu et al., 1990; Ahn et al., 1994) as a molecular marker, which is highly conserved across individuals. This allowed us to align the pattern of labeling observed in cerebellar regions across animals and, hence, quantify the spatial relationship between mossy fiber terminals originating from functionally distinct cortical areas (Figure 5, Figure 6, see also Figure 5—figure supplement 1, including validation of alignment across animals Figure 5—figure supplement 1J).

Spatial distribution of mossy fiber terminals within divisions of the cerebellar hemispheres with multimodal regional convergence.
(A) Schematic reconstruction of the parasagittal distribution of labeled mossy fiber terminals in regions of the contralateral cerebellar hemispheres, after injections of mono-trans-synaptic AAV into target cortical regions: primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), or dorsal auditory (AuD) cortex. Data taken from 120 μm thick representative sections (three coronal sections) for rostral (top) to caudal (bottom) regions of the indicated lobules; distance from bregma is indicated based on the mouse stereotaxic atlas (Franklin and Paxinos, 2007). Data were aligned across animals using the parasagittal pattern of Purkinje cell molecular marker aldolase C (grey represents high expression, AldoC+; white represents low expression, AldoC-; see also Figure 5—figure supplement 1). For high density projections, M1 and S1, each dot represents 5–10 mossy fiber terminals, for all other lower density projections each dot represents 1–5 mossy fiber terminals. Simplex lobule (Sim), crus I/II ansiform lobule (crus I/II), paramedian lobule (PM), and copula pyramis (Cop). (B) Average pairwise distance between all terminals originating from the same cortical injection site (e.g. M1–M1) or from different cortical injection sites (e.g. M1-Other, where ‘Other’ is represented by all injected sites except M1 and S1). Statistical significance represents terminals being closer to like-terminals than to terminals from other cortical regions (**p<0.01, ***p<0.001, for exact p-values see Supplementary file 1E). (C) Proportion of the five nearest neighbors of each terminal location following each cortical injection site (i.e. seed origin: M1, S1 and Other; indicated on x-axis) that originate from M1, S1 or other cortical regions (i.e. neighbor origin: indicated on legend).

Spatial distribution of mossy fiber terminals within divisions of the vestibulocerebellum and vermis with multimodal regional convergence.
(A) Schematic reconstruction of the parasagittal distribution of labeled mossy fiber terminals in contralateral cerebellar lobules spanning the vestibulocerebellum (VbC) and the vermis, after injections of mono-trans-synaptic AAV into target cortical regions: primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), or dorsal auditory (AuD) cortex. Data taken from 120 μm thick representative sections (three coronal sections) for rostral (top) to caudal (bottom) regions of the indicated lobules; distance from bregma is indicated based on the mouse stereotaxic atlas (Franklin and Paxinos, 2007). Data were aligned across animals using the parasagittal pattern of Purkinje cell molecular marker aldolase C (grey represents high expression, AldoC+; white represents low expression, AldoC-; see also Figure 5—figure supplement 1). For high density projections, M1 and S1, each dot represents 5–10 mossy fiber terminals, for all other lower density projections each dot represents 1–5 mossy fiber terminals. Dotted lines represent midline and dashed line represents division between paraflocculus (PFl) and copula pyramis (Cop). Vermal cerebellar lobules: lobule IV/V and lobule IV). (B) Average pairwise distance between all terminals originating from the same cortical injection site (e.g. M1–M1) or from different cortical injection sites (e.g. M1-Other, where ‘Other’ is represented by all injected sites except M1 and S1). Statistical significance represents terminals being closer to like-terminals than to terminals from other cortical regions (*p<0.05, for exact p-values see Supplementary file 1E). (C) Proportion of the five nearest neighbors of each terminal location following each cortical injection site (i.e. seed origin: M1, S1 and Other; indicated on x-axis) that originate from M1, S1 or other cortical regions (i.e. neighbor origin: indicated on legend).
After alignment, for identified lobules with substantial co-innervation in the cerebellar hemispheres, terminals from both sensory and motor cortical areas were spatially overlapping (Figure 5) - with the exception of crus II, where M1 injections resulted in terminal labeling largely in dorsal regions and S1 in ventral regions (Figure 5A). Consequently, crus II had diverse functional cerebrocerebellar input but labeling was more spatially segregated within the lobule, with the distance between M1 terminals to themselves being significantly shorter than the distance between M1 terminals and those originating from other cortical injection sites (Figure 5B; M1-M1: 258 ± 22 µm, M1-Other: 305 ± 10 µm, p=0.007, two-sample t-test), this was also true for S1 (S1-S1: 190 ± 13 µm, S1-Other: 316 ± 15 µm, p<0.001, two-sample t-test). Hence, when identifying the nearest neighbors of M1 terminal locations in crus II, over 80% of these were also of M1 origin (Figure 5C); therefore, crus II has lower potential for multimodal spatial overlap at the modular level. In contrast, crus I showed highly spatially overlapping patterns, that is, the distance between both M1 and S1 terminal locations to that of other cortical origins was not significantly different to the distance between themselves (Figure 5B, M1–M1: 222 ± 8 µm, M1-Other: 220 ± 7 µm, p=0.953; S1-S1: 246 ± 10 µm, S1-Other: 233 ± 7 µm, p<0.001, two-sample t-test), and a higher proportion of their nearest neighbors came from extrinsic cortical origins (Figure 5C). Throughout the hemispheres, terminals from primary sensory regions V1 and A1 as well as association cortices (PPC and AuD) were generally located more on the apex of the folia, where they were intermingled with M1 terminals, whereas S1 terminals tended to be more at the base of the folia, also intermingled with M1 terminals.
This same apical/basal pattern was also observed in the PFl (Figure 6A); hence, the ventral PFl (part of the vestibulocerebellum) was highly co-innervated with terminal labeling from sensory and association cortices and the dorsal PFl largely contained S1 and M1 terminals. In contrast, in vermal lobules IV/V and VI terminals from all cortical regions were overlapping in more medial zones and S1 and M1 terminals were additionally in more paravermal zones, with little labeling from other modalities (Figure 6). Therefore, S1 and M1 terminals were quite widely distributed parasagittally, whereas terminals from other primary sensory and association areas tended to be in medial-vermal and lateral-hemispheric regions. In general, the pairwise distance between terminal locations was not significantly different between M1 and other cortical origins, however, S1 terminals were more tightly clustered to each other in both the PFl and lobule VI (Figure 6B).
Generally throughout the cerebellum, S1 injections resulted in the patchiest distribution with clusters of mossy fiber terminals in certain lobules aligning to the AldoC expression pattern; this was especially the case in Cop and to a lesser extent in the PM, where mossy fiber clusters aligned with AldoC+ stripes (Figure 5—figure supplement 1D). Interestingly, this pattern of projections is in agreement with results from electrophysiological recordings in the PM in rats, demonstrating alternating patches of somatosensory responses to hindlimb/forelimb stimulation (Shambes et al., 1978). However, distributed mossy fiber labeling was also present in both AldoC+ and AldoC- regions in the Cop and PM (Figure 5—figure supplement 1D), so that no specific bias was seen on average (Figure 5—figure supplement 1G–I). Conversely, in crus I clustered S1 mossy fiber terminals were more biased towards AldoC- regions (Figure 5—figure supplement 1B,G), although not strictly so.
To test for spatial randomness of terminal locations, we calculating the Ripley’s K-function for each lobule and compared these distributions with simulated distributions of complete spatial randomness (CSR; see Materials and methods); for S1, all lobules except crus II (p=0.260) and lobule IV (p=0.642) showed significant deviation from CSR (p≤0.028; t = 60 µm, 100 simulated CSR distributions; 3D Ripley’s K-function; Hansson et al., 2013). Mossy fiber terminals from M1 injections generally had a more distributed, less clustered, pattern throughout the lobules and did not appear to consistently follow particular AldoC expression boundaries (Figure 5, Figure 6, Figure 5—figure supplement 1E,F); however, with the exception of lobule VI (p=0.391), M1 terminal locations still showed significant deviation from CSR (p≤0.015; t = 60 µm, 100 simulated CSR distributions; 3D Ripley’s K-function; Hansson et al., 2013). Injections in the other cortical regions also resulted in distributed and/or sparse terminal labeling which could not be assigned to a particular AldoC expression pattern and only deviated from CSR in crus I, sim, PM and Cop, and the PFl (p≤0.001; t = 60 µm, 100 simulated CSR distributions; 3D Ripley’s K-function; Hansson et al., 2013). Therefore, mono-trans-synaptic cerebrocerebellar mossy fiber terminals did not adhere to a specific pattern of zonal organization with respect to AldoC expression boundaries, however, many patterns of expression significantly deviated from spatial randomness. Since these terminals potentially relay in a number of different precerebellar nuclei (see Supplementary file 1B and Figure 2) our results do not preclude the existence of a finer-scale relationship with respect to individual cerebrocerebellar pathways and/or in a lobule specific manner.
Intermediate cerebrocerebellar brainstem pathways
Based on the results of the mono-trans-synaptic tracer, there are a number of precerebellar nuclei, both pontine and extra-pontine, that have the potential to act as intermediate nuclei for cerebrocerebellar projections (summarized in Figure 7A; see also Supplementary file 1B). To gain insight into the likelihood that these various precerebellar nuclei act as intermediate sources of mossy fiber input to the lobules identified as having high co-innervation, we performed additional tracing experiments with injections of the retrograde tracer Cholera-toxin-B (CTB) into key vermal regions (lobules IV/V, IV) as well as cerebellar hemispheres (Sim and crus I; Figure 7—figure supplement 1). Additionally, we performed dual injections of CTB into lobule IV/V or crus I and the trans-mono-synaptic AAV1.cre into either M1 or S1, respectively (Figure 7B). Our results show that the majority of retrogradely labeled cells were located in the pons (across all cases: pontine = 6429 cells [73% of total]; extra-pontine = 2378 cells [27% of total]; Figure 7—figure supplement 1). With our dual injections, we also observed some double labeled cells in the pontine nuclei (e.g. Figure 7B), demonstrating direct confirmation that these pontine cells receive cerebrocerebellar disynaptic input. We note that the probability of double labeling was likely low due to the spatially restricted injection sites.

Summary of intermediate brainstem nuclei potentially supporting disynaptic cerebrocerebellar pathways.
(A) Schematic summary of pathways to precerebellar nuclei from target regions of cerebral cortex: primary motor (M1), primary somatosensory (S1), posterior parietal association cortex (PPC), primary visual (V1), primary auditory (A1), or dorsal auditory (AuD) cortex. (B) Representative labeling in pontine (top) and precerebellar nuclei (middle, bottom) after combined injections of AAV1.cre (red) into cortical regions and CTB (green) into cerebellar regions (M1 + lobule lV/V; S1 + Crus I). Note the double labeled cells within the pontine nuclei (Pn; total 7 double labelled/52 anterogradely labeled cells from M1), and spatially overlapping cells within the reticulotegmental nucleus (RtTg) and lateral reticular nucleus (LRt). Distance from bregma is indicated based on the mouse stereotaxic atlas (Franklin and Paxinos, 2007). Scale bars 500 µm and 50 µm for insets. (C) Proportion of the five nearest neighboring cells for the cortical injection sites in (B) that are anterogradely labeled cells originating from the same cortical injection site (e.g. M1) or retrogradely labeled cells originating from injections in the cerebellar cortex (e.g. lobule IV/V; indicated by legend) within the indicated precerebellar nuclei. (D) Correlation of anterogradely labeled precerebellar cells with the total number of mossy fiber terminals per animal for either pontine only cells (left; R = 0.632, p=0.012, n = 18, Pearson’s correlation) or pontine plus extra-pontine precerebellar cells (right; R = 0.837, p<0.001, n = 18, Pearson’s correlation). Cochlear nuclei (CN), longitudinal fasciculus of the pons (lfp), matrix region x (Mx), medial lemniscus (ml), interpolar part of the spinal trigeminal nucleus (Sp5I), red nucleus (Ru), and vestibular nuclei (VN).
We also observed that many extra-pontine precerebellar nuclei identified as receiving cortical projections (Figure 2; Supplementary file 1B) contained retrograde labeling (e.g. lateral reticular nucleus, reticulotegmental nucleus, vestibular nuclei, interpolar part of the spinal trigeminal nucleus, and matrix region x; Figure 7B; Figure 7—figure supplement 1; Supplementary file 1D). Although double labeled cells were not directly observed in extra-pontine precerebellar nuclei, the labeling was spatially overlapping (Figure 7B), and quantification of the identity of the five nearest neighbors to cells originating from anterograde cortical injections, reveled that 13–45% of these were retrogradely labeled cells from the cerebellar cortex (Figure 7C; see also Figure 7—figure supplement 2).
From retrograde injections into the cerebellar cortex, we found that, while the majority of retrograde labeling was observed in the pontine nuclei, lobules IV/V and VI showed relatively higher proportions of extra-pontine retrograde labeling (IV/V, 44%; VI, 39%) in comparison to the cerebellar hemispheres (sim, 17%; crus 1, 21%; Figure 7—figure supplement 1). Lastly, although we did not observe significant cerebrocerebellar projections to lobule VII (e.g. see Figure 3), we confirmed that this lobule does receive pontocerebellar projections, but these were also proportionally few in comparison to other injected cerebellar regions (52% pontine vs 48% extra-pontine retrograde labeling, Figure 7—figure supplement 3), and originated largely from lateral regions of the pons, an area that topographically has been shown to receive corticopontine projections from the RSC (Suzuki et al., 2012; see also Figure 1—figure supplement 1).
Although, the proportion of observed mossy fiber terminals that relay through extra-pontine pathways cannot be precisely specified, we found a stronger positive correlation with the total number of mossy fiber terminals when all precerebellar cells are accounted for (R = 0.837, p<0.001, n = 18, Pearson correlation; Figure 7D) compared to when only pontine labeled cells are included (R = 0.632, p=0.012, n = 18, Pearson correlation). Additionally, we found lower variability in the total number of mossy fibers per labeled cell when all precerebellar cells are included (all cells: 50 ± 12 mossy fibers per cell; pontine cells only: 120 ± 63 mossy fibers per cell). Whether this ratio of ~50 mossy fiber terminals per precerebellar neuron can be generalized across individual cortico-precerebellar pathways remains to be determined, however, using single neuron tracing techniques, previous studies have reported a comparable 67 ± 7 (mice: Biswas et al., 2019) and 46 ± 9 (rats: Na et al., 2019) mossy fiber terminals in the cerebellum per pontine projecting neuron.
Discussion
In this study we elucidated the cerebrocerebellar projections from primary motor, sensory, and secondary/association cortical areas to various regions of the mouse cerebellum. By using a mono-trans-synaptic anterogradely transported AAV (Zingg et al., 2017; Zingg et al., 2020), we found a highly organized topography of labeled corticopontine cells and also observed a number of potential extra-pontine cerebrocerebellar pathways. Notably, injections into A1 produced only terminal labeling in the pons, whereas injections into more dorsal secondary auditory cortex showed evidence of an auditory cortico-ponto-cerebellar pathway. We quantified the proportion and detailed the precise organization of cerebrocerebellar terminals from each injected cerebral cortical region. The majority of labeled mossy fiber terminals from primary sensory and motor cortical regions were in the contralateral cerebellum, whereas projections from the secondary/association areas PPC and AuD resulted in less laterality in the cerebellum. Cerebellar subdivisions with the highest spatial overlap of multimodal cerebrocerebellar inputs within molecularly defined modules were bilateral crus I, PFl, vermal lobules VI, and lobule IV/V indicating that these regions may act as hubs for the integration of multimodal information originating from the cerebral cortex.
Pontine and extra-pontine cerebrocerebellar pathways
Although the use of anterograde trans-synaptic tracers has not been extensive, our results in mice generally confirmed previous findings regarding the topographic organization of corticopontine pathways examined using anterogradely labeled terminals or retrograde pontine injections for mapping across other species; with more rostral cortical regions projecting more medially and caudal cortical areas projecting towards the lateral extent of the basal pontine nuclei (rats, Odeh et al., 2005; Leergaard and Bjaalie, 2007; cat, Bjaalie et al., 1997; monkey, Glickstein et al., 1985; Schmahmann and Pandya, 1997). Our results are also in agreement with the topographical organization of corticopontine anterograde terminal labeling in mice following S1 and M1 (Proville et al., 2014) as well as V1 (Inoue et al., 1991) injections. However, some differences may exist across species. For instance, although the current study did not compare various somatotopic regions within S1, following injections centered around the hindlimb/forelimb regions in mice, we did not find a substantial caudal bias in the pontine nuclei, as has previously been reported in rats from injections into similar cortical regions (Leergaard et al., 2003; Odeh et al., 2005); whether this reflects a species difference requires finer detailed somatotopic mapping in mouse S1 using these techniques.
Additionally, we found that the primary auditory cortex was unique in that it was the only cortical target region we observed with no direct corticopontine pathway. Following A1 injections, we found labeled mossy fibers bilaterally in the cerebellum (largely lobules III-V of vermis, PFI, crus I and II) but there were no intermediate cells trans-synaptically labeled in the pontine nuclei; indicating that this disynaptic mossy fiber input must reach the cerebellum through other precerebellar nuclei. In contrast, more dorsal auditory cortex (i.e. AuD) did have projections directly to the pontine nuclei. This is also consistent with our observations that mossy fiber terminals were more biased to vermal regions after A1 injections and included more projections to the cerebellar hemispheres (especially crus I) after AuD injections. There are substantial conflicting results in many species describing either the presence of an A1-pontine pathway (cats: Perales et al., 2006; rabbits: Knowlton et al., 1993; gerbils: Budinger et al., 2000 and rats: Wiesendanger and Wiesendanger, 1982; Legg et al., 1989), while others found either no major A1-pontine projection or projections only from secondary/association auditory cortex (cats: Brodal, 1972; monkeys: Glickstein et al., 1985; rats: Azizi et al., 1985; bats: Schuller et al., 1991a), and studies in mice have not been extensive. However, through both anterograde and retrograde tracing, here we found no major A1-pontine pathway in mice.
After A1 injections, we did observe labeled fibers in the dorsolateral pontine nuclei, with apparent synaptic boutons - indicating a disynaptic functional connection with pontine cells. There are a few possibilities as to the origin of these fibers. First, they may be terminals labeled from trans-synaptically labeled cells in secondary auditory cortex; however, the pattern of pontine cell labeling we observed following AuD injections was topographically distinct from the location of these labeled fibers. Secondly, disynaptic A1-cerebellar pathways may travel via other extrapontine nuclei and only trisynaptically through the pons. For instance, here we confirmed direct projections from the IC to dorsolateral regions of the pons (Caicedo and Herbert, 1993). Interestingly, anterograde labeling in the IC from auditory cortex was largely in the DCIC subregion, in which corticocollicular projections have been suggested to play a modulatory role on auditory processing (Barnstedt et al., 2015) compared to the sharper frequency tuning in CIC (Stiebler and Ehret, 1985; Yu et al., 2005). Further, A1 in mouse has been shown to be tonotopically organized, whereas the dorsal belt of auditory cortex is not (Stiebler et al., 1997; Guo et al., 2012), and auditory corticopontine projections in cats are also not tonotopically organized (Perales et al., 2006; whether this functional principle is a determinant for auditory corticopontine projections requires further consideration. Finally, another potential relay nucleus for these A1-pontine fibers is the cochlear nuclear complex, which is known to project directly to the pontine nuclei (Kandler and Herbert, 1991) as well as the cerebellum as mossy fibers (Huang et al., 1982; Fu et al., 2011). Thus, this precerebellar nucleus may serve as a disynaptic and potentially trisynaptic (via the pontine nuclei) auditory cerebrocerebellar intermediate, which may be functionally relevant in terms of temporal processing of auditory information. Future studies are needed to further delineate these polysynaptic cerebrocerebellar auditory pathways, including their functional significance in cerebellar processing.
Beyond corticopontine projections, we found a large - and in some cases equal - proportion of anterogradely labeled extra-pontine precerebellar cells in various brainstem nuclei, especially from S1, M1 and auditory regions. While the nature of the mono-trans-synaptic tracing does not permit the proportion of observed mossy fiber terminals that relay through pontine versus extra-pontine pathways to be precisely segregated, our observations from retrograde tracer injections into key cerebellar lobules, as well as evidence of observed fibers tracts travelling through the inferior cerebellar peduncle, suggest a number of precerebellar nuclei identified as receiving direct cortical input also have projections to cerebellar lobules with spatial overlap of multimodal cerebrocerebellar input. Therefore, it is likely that both pontine and extra-pontine nuclei act as sources of the observed mossy fiber input through disynaptic cerebrocerebellar pathways (see also Watson and Apps, 2019), and these precerebellar nuclei may play an important functional role to relay/integrate multimodal information from the cerebral cortex to the cerebellum (Ruigrok, 2004; Fu et al., 2011; Sillitoe et al., 2019). Since mossy fiber terminals were observed in the cerebellum but no pontine labeled cells were found after A1 injections, one can conclude that, at least for this disynaptic pathway, extra-pontine intermediate nuclei play a large role. The nature of relay versus integration of cortical signals along each of these precerebellar pathways will make for interesting future studies.
Organization of cerebrocerebellar input to the cerebellum
Previous studies on pontocerebellar projections have found that the majority of mossy fibers terminate in the contralateral cerebellum (e.g. Serapide et al., 2002). Here we found that mono-trans-synaptically labeled mossy fibers terminals from the cerebral cortex (through both pontine and extra-pontine pathways) project to a varying degree to the ipsilateral and contralateral cerebellum. Presumably these projections can cross either at the level of the pontine nuclei or the cerebellar peduncles, or both, as was recently found to be the case for some reconstructions of individual pontocerebellar neurons (Biswas et al., 2019; Na et al., 2019). We found that cerebrocerebellar pathways from sensory association and secondary cortices show less laterality than those from both primary sensory and motor cortices. Interestingly, using retrograde rabies tracing, Suzuki et al., 2012 also found relatively large bilateral cerebrocerebellar pathways from orbitofrontal and retrosplenial cortical regions to the posterior cerebellar cortex. This pattern of laterality suggests that information coming from ‘higher-order’ prefrontal and cortical association areas may have a more global effect on cerebellar processing than more targeted ‘unisensory’ or primary motor inputs; whether this reflects a broader functional implication for the influence of the cerebral cortex in learning and predictive processing in the cerebellum is yet to be determined.
The cerebellar hemispheres as well as lobules VI and VII are traditionally described as cerebrocerebellar receiving areas (Serapide et al., 1994; Kandel et al., 2000); although see Coffman et al., 2011). We confirm the cerebellar hemispheres as the major target for cerebrocerebellar input in mice and detail the pattern of projections from primary motor and sensory areas, but also emphasize the diversity of disynaptic connections between the cerebral cortex and cerebellar cortex. Using this anterograde tracing technique, we found at least sparse cerebrocerebellar projections to every lobule of the cerebellum from motor and sensory cortices. Additionally, we found a relatively large number of labeled mossy fibers in lobules IV/V but few in more posterior vermal regions, including lobule VII. Lobule IV/V is functionally considered part of the spinocerebellum and implicated in motor function, but not traditionally described as receiving extensive pontocerebellar or cortical input. However, recent studies have emphasized how this region is modulated by behavioral state and locomotive behaviors (Jelitai et al., 2016; Muzzu et al., 2018). Specifically, Muzzu et al., 2018 suggested that predictive motor responses observed in mouse lobules IV/V may carry a motor efference copy coming from higher cortical areas. This is in agreement with our results, which confirmed disynaptic cerebrocerebellar as well as direct pontocerebellar input to lobule IV/V in mice. Conversely, our results of very few observed cerebrocerebellar terminals in lobule VII are seemingly in contradiction to previous findings of large pontocerebellar pathways to this region, which have been assumed to carry cerebrocerebellar information (Päällysaho et al., 1991; Thielert and Thier, 1993; Biswas et al., 2019). However, our results in mice are in agreement with the suggestion in other species that pontocerebellar inputs to lobule VII may relay cortical information of higher order than the primary motor and sensory areas, such as the retrosplenial cortex (rats, Suzuki et al., 2012) and prefrontal regions (rats, Watson et al., 2009; Suzuki et al., 2012); monkey, Kelly and Strick, 2003), or subcortical brain regions such as the superior colliculus and the accessory optic system (rats, Mihailoff et al., 1989; birds, Pakan and Wylie, 2006; monkeys, Kralj-Hans et al., 2007; for review see Voogd and Barmack, 2006).
Functional implications
Our results show that cerebrocerebellar information originating from different functional cortical regions shows significant spatial overlap within molecularly defined modules in crus I, PFL, and vermal lobules IV/V and VI. Therefore, all gross divisions of the cerebellar cortex (hemispheres, vestibulocerebellum and vermis) have the potential for cortical multimodal influence, that is, the influence of the cerebral cortex is not strictly limited to the cerebellar hemispheres. These spatial overlaps constitute the fundamental anatomical basis for multimodal integration processes of cortical information at a modular and potentially cellular level (see also Proville et al., 2014). Functionally speaking, in vivo electrophysiological and imaging studies have shown that in rat (Ishikawa et al., 2015), mouse (Chabrol et al., 2015; Giovannucci et al., 2017; Markwalter et al., 2019) and mormyrid fish (Sawtell, 2010) single granule cells can respond to multiple modalities. For instance, Ishikawa et al., 2015 found responses to sensory stimuli in half of the granular cells recorded in crus I/II and many showed responses to at least two different sensory modalities. Further, Markwalter et al., 2019 also found that just over half of granular cells in vermal lobules VI respond to sensorimotor stimuli, with the majority exhibiting responses to multiple sources of distinct-modality stimulation. Although in functional studies examining granule cell integrative responses to date, the specific origin of mossy fiber inputs was either undetermined or originated from bottom up brainstem centers (e.g. Chabrol et al., 2015), it has also been shown that in decerebrated cats (Jörntell and Ekerot, 2006; Bengtsson and Jörntell, 2009) granule cells respond to only one modality. Therefore, the potential for multimodal convergence of cerebrocerebellar inputs in the granule cell layer that we have observed in this study may be an important determinant of granule cell function.
Alternatively, since the cerebellum receives both ascending and descending sensory and motor input, which has been shown to be anatomically integrated at the level of individual granule cells (Huang et al., 2013), it is also possible that the majority of basic sensory information is carried by ascending pathways and these descending cerebrocerebellar pathways are carrying action-related information from various cortical regions; including from primary sensory areas, in which a growing body of studies have demonstrated the influence of motor behaviors, such as locomotion, on sensory responses (Niell and Stryker, 2010; Schneider et al., 2014; Pakan et al., 2016; Pakan et al., 2018; Ayaz et al., 2019; Poulet and Crochet, 2019). In this way, the cerebellum may use these cortical signals in closed feedback loops to regulate and adjust ongoing predictive responses, as suggested by a feed forward model for motor control (Kelly and Strick, 2003; Shadmehr et al., 2010; Gao et al., 2018; Chabrol et al., 2019). Future studies examining the precise nature of the information carried by the cerebrocerebellar pathways identified here will further elucidate the functional influence of the cerebral cortex on cerebellar processing.
Assessing the full functional contribution of cerebrocerebellar input, and subsequent cortico-cerebellar loops, will require both advanced anatomical delineation of polysynaptic circuits combined with simultaneous observation and circuit manipulation in behaving animals. Mouse models, with access to advanced genetic tools and the establishment of increasingly complex behavioral paradigms, provide many advantages in this regard. The results of the current study demonstrate conserved principles of organization between mice and other mammalian species, as well as provide foundational insight into the diversity, and potential for spatial multimodal convergence, within cerebrocerebellar pathways in mice. These findings will guide future research to delineate the precise nature of sensorimotor integration at multiple circuit levels.
Materials and methods
Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
---|---|---|---|---|
Strain, strain background Mus musculus | Strain: Ai9; Rosa26-CAG-LSL-tdTomato Cre reporter mice line Background:C57BL/6J | The Jackson Laboratory | RRID:IMSR_JAX:007909 | male and female |
Recombinant DNA reagent | AAV1.hSyn.Cre.WPRE.hGH, | Addgene, | RRID:Addgene_105553 | titer 3.0 × 10 13 GC/ml |
Recombinant DNA reagent | CAV.cre | PVM IGMM Montpellier, | titer 5.5 × 10 12 GC/ml | |
Peptide, recombinant protein | Cholera-toxin-B 488 | ThermoFisher Scientific | Cat no. C34775 | |
Antibody | anti-Aldolase C (H-11) (mouse monoclonal) | Santa Cruz Biotechnoloy Host species: mouse | RRID:AB_10659113 | 1:100 |
Antibody | anti-RFP (polyclonal) | Rockland Host species: rabbit | RRID:AB_2209751 | 1:2000 |
Antibody | Anti-Mouse IgG (H+L) Secondary, Alexa Fluor 488 Conjugated (polyclonal) | ThermoFisher Scientific Host species: donkey | RRID:AB_141607 | 1:200 |
Antibody | Cy3-AffiniPure Goat Anti-Rabbit IgG (H+L) (polyclonal) | Jackson Immuno-Research Host species: goat | RRID:AB_2338006 | 1:200 |
Software, algorithm | MATLAB 2013/2017a | Mathworks | RRID:SCR_001622 | |
Software, algorithm | ImageJ (Fiji) | NIH – public domain | http://fiji.sc; RRID:SCR_002285 |
Experiments were performed with 42 mice (9–12 weeks; both males and females) of the Rosa26-CAG-LSL-tdTomato Cre reporter mice line (Ai9; RRID:IMSR_JAX:007909). Note that this tdTomato reporter line is known to have increased levels of auto-fluorescence (Hahn et al., 2019), particularly in Purkinje cells (e.g. Figure 1F), however, this was easily distinguishable from cre-induced tdTomato expression under the microscope and in merged images. Animals were housed in standard laboratory cages (22°C, 12 hr light-dark cycle) with food and water available ad libitum. All experiments were performed according to the NIH Guide for the Care and Use of Laboratory animals (2011) and the Directive of the European Communities Parliament and Council on the protection of animals used for scientific purposes (2010/63/EU) and were approved by the animal care committee of Sachsen-Anhalt, Germany (42502-2-1479 DZNE).
Neuroanatomical injections
Request a detailed protocolFor mono-trans-synaptic tracing of the cerebrocerebellar pathways we used an adeno-associated virus (AAV1.cre; AAV1.hSyn.Cre.WPRE.hGH, Addgene, USA, RRID:Addgene_105553) that expresses cre protein, which then drives tdTomato expression in the Ai9 reporter mouse line. The AAV1.cre is transported anterogradely to mono-synaptically connected neurons (Zhao et al., 2017; Zingg et al., 2017; Zingg et al., 2020) and subsequently also drives cre expression in anterograde target neurons (e.g. pontine cells); tdTomato reporter expression then fills these target neurons including axon terminals (e.g. mossy fiber terminals; Figure 1; see also Zingg et al., 2017). This AAV1.cre construct has been shown to spread exclusively to synaptically connected neurons with trans-synaptic spread nearly abolished by tetanus toxin light chain (i.e. inhibition of presynaptic vesicle fusion; Zingg et al., 2020). Limitations in transport may exist through neuromodulatory projections (e.g. noradrenergic, cholinergic, serotonergic), however, efficient transport has been demonstrated for glutamatergic and GABAergic synapses to both excitatory and inhibitory neurons as well as long-distance projections (Zingg et al., 2020). Retrograde transport has also been shown with this AAV1.cre construct - although with lower efficiency (Tervo et al., 2016; Zingg et al., 2017). Therefore, using this technique, reciprocally connected regions may include cells that are both anterogradely and, to a lesser extent, retrogradely labeled; a clear advantage of using this technique to examine cerebrocerebellar projections in this study is that these pathways are largely descending and unidirectional (for review see Léna and Popa, 2015), which removes this as a confounding factor.
A total of 42 mice were used in this study. For mono-trans-synaptic tracer experiments, AAV1.cre injections (18 mice) were made into target cortical regions spanning multiple modalities, including primary sensory areas, primary motor regions and association cortex (3 mice per target): M1 (forelimb/hindlimb), S1 (forelimb/hindlimb), PPC (lateral and medial parietal association cortex), V1 (monocular), A1, and AuD. Further circuit investigations were performed as follows. An injection of CAV.cre (PVM IGMM Montpellier, France) was made into the pons (1 mouse) and an injection of AAV1.cre into the IC (2 mice). In a series of combined anterograde plus retrograde experiments (total 7 mice), injections of AAV1.cre into a target cortical region as well as injections of conjugated Cholera-toxin-B 488 (CTB; Thermo Fisher, USA) into individual cerebellar lobules were made (e.g. M1 + IV/V; S1 + crus I). Lastly, 14 mice were used for retrograde experiments alone, where conjugated CTB 488 (Thermo Fisher, USA) was injected into individual cerebellar lobules (IV/V, VI, VII, Sim and crus I) in order to examine the brainstem origin of mossy fiber projections to these key cerebellar regions.
All injections were performed as previously described (Henschke et al., 2015; Henschke et al., 2018). Briefly, animals were anesthetized with isoflurane (4%; Baxter, Germany), the cranial skin was incised, the skull exposed by a displacement of the skin and muscles, and a small hole was drilled into the skull. Cerebral cortical injections were always performed on the left side (in order to minimize potential variation due to lateralization effects), precluding the identification of lateralized projections. We used the following stereotaxic coordinates derived from the mouse brain atlas in reference to bregma (Franklin and Paxinos, 2007) and targeted cortical layer 5: M1: 1.7 mm rostral and 2 mm lateral; S1: −0.2 mm caudal and 2.2 mm lateral; PPC −2 mm caudal and 2 mm lateral; V1: −2.92 mm caudal and 2.5 mm lateral; A1: −2.8 mm caudal and 4.5 mm lateral; AuD: −2.3 mm caudal and 4 mm lateral. The following coordinates were used for the left IC: −5.02 mm caudal and 1 mm lateral, 1 mm deep. For all above injections, 60 nl of the AAV1.cre (titer 3.0 × 1013 GC/ml) in 6 × 10 nl steps. The left side pontine nuclei were also injected: 6 mm caudal, 1 mm lateral and with an angle of 21°, 5.4 mm deep; with 100 nl of the CAV.cre (titer 5.5 × 1012 GC/ml) in 10 × 10 nl steps. Since the majority of projections travel contralaterally from the cerebral cortex to the cerebellar cortex, we performed retrograde tracer injection on the right side cerebellar cortex targeting the granule cell layer: lobule IV/V: −5.9 mm caudal, 1 mm lateral and 0.7 mm deep; lobule VI: −6.6 mm caudal, 1 mm lateral and 0.5 mm deep; lobule VII: −7.6 mm caudal, 1 mm lateral and 1 mm deep; crus I: −6.8 mm caudal, 2.5 mm lateral and 1 mm deep; sim: −6.1 mm caudal; 2 mm lateral and 1.5 mm deep, with 200 nl of CTB in 10 × 20 nl steps.
All injections were made over 10 min into these areas via a nanoliter delivery system (World Precision Instruments, Germany) and fine glass micropipettes (tip diameter 20 µm). The speed of injection is important as trans-synaptic transport of the AAV1.cre construct is highly dependent on the titer within individual neurons (Zingg et al., 2017; Zingg et al., 2020); hence, small/slow volume injections where viral particles are not quickly dispersed through tissue were done to concentrate viral particles and optimize trans-synaptic activity. After injections, craniotomies were sealed with bone wax (Ethicon, Johnson and Johnson, Germany) and the skin was closed with tissue adhesive (Histoacryl; B/Braun, Germany). Subsequently, the animals were returned to their home cages for 2 weeks to allow for the expression of the viral constructs (both AAV1.cre and CAV.cre), and 5 days for experiments involving CTB injections alone. Note, for trans-synaptic labeling with the AAV1.cre construct, time periods longer than 2 weeks do not label additional orders of connections, that is labeling remains mono-trans-synaptic even over extended periods of time (Zingg et al., 2017).
Histological processing
Request a detailed protocolAnimals were deeply anesthetized with ketamine (20 mg/100 g body weight, ip) and xylazine (1 mg/100 g body weight, ip) and perfused transcardially with 20 ml of 0.1 M phosphate-buffered saline (PBS, pH 7.4) followed by 200 ml of 4% paraformaldehyde. The brains were removed, post-fixed overnight in 4% paraformaldehyde at 4°C, and then cryoprotected in 30% sucrose in PBS for 48 hr. Brains were cut on a cryostat (CryoStar NX70, Thermo Scientific, USA) into 40 µm (cerebellum, brainstem) and 50 µm (cortex and underlying structures) thick coronal sections. Sections containing the cerebral cortex were directly mounted on gelatin-coated glass slides, sections containing the cerebellum were collected in PBS (free-floating) and blocked in normal donkey serum (10% and 0.4% triton in PBS) for 1 hr. Sections were then incubated in primary antibodies overnight at 4°C to visualize aldolase C expression (anti-AldoC; mouse, 1:100, Santa Cruz Biotechnology Cat# sc-271593, RRID:AB_10659113, USA) and red fluorescent protein (anti-RFP, rabbit, 1:2000, Rockland Cat# 600-401-379, RRID:AB_2209751, USA) for enhancement of the tdTomato signal. After rinsing in PBS, sections were incubated with the respective secondary antibodies (anti-mouse Alexa 488, 1:200, Thermo Fisher Scientific Cat# A-21202, RRID:AB_141607, USA; anti-rabbit Cy3, 1:200, Jackson ImmunoResearch Labs Cat# 111-165-144, RRID:AB_2338006, United Kingdom) for 2 hr. Finally, sections were rinsed again in PBS, mounted on gelatin-coated slides, and coverslipped with MOWIOL (Fluka, Germany).
Quantification and data analysis
Request a detailed protocolSections of 40 µm thickness were examined using a confocal microscope (Zeiss LSM 700, Germany) equipped with a 2.5x objective (NA 0.085, Zeiss, Germany). For each section throughout the extent of the cerebellum two high-resolution images (2048 × 2048 Pixel) were acquired, one for each hemisphere. The two tiles were merged using inbuilt functions for feature based panoramic image stitching in MatLab (Mathworks, MA, USA; resulting in a ~ 4000×4000 pixel image of the entire cerebellum); brightness and contrast were adjusted as necessary using Adobe Photoshop software (v. 13.0.6 for Windows).
Only experimental cases where AAV1.cre injections were verified to be located within the aimed target areas M1 (forelimb/hindlimb regions), S1 (forelimb/hindlimb regions), PPC, V1, A1, and AuD were included in the analysis. The injection sites had a roughly cylindrical shape. All sections that covered the central core but not surrounding halo of each injection site were included in the analysis (see also Figure 1—figure supplement 3). We then calculated injection site volume as follows: , where h corresponds to the maximum value across sections for the height of the injection site [dorso-ventral direction], a corresponds to the maximum value across sections for the radius of the injection site parallel to the cortical layers [medio-lateral direction], and b corresponds to the rostrocaudal extent of the core injection site. The Line Measure Tool of the Zen software (Zeiss, Germany) was used to measure h and a, and b was calculated by counting the numbers of sections covering the core injection site multiplied by their thickness (40 µm).
The number of labeled mossy fiber terminals was evaluated by point marking mossy fiber rosettes in the reconstructed images, in all sections of each experimental animal using the Cell Counter plugin in ImageJ (Fiji, v. 1.43 r, NIH, Bethesda, MD, USA, RRID:SCR_002285). Counted mossy fiber terminals were assigned to either AldoC+/-cerebellar stripes using the ROI manager plugin of ImageJ. Unless otherwise stated, all data are reported as mean or, where applicable, mean ± standard error of the mean (s.e.m). The density D of labeling for the ipsilateral as well as the contralateral side within each cerebellar lobule was calculated by taking the total number of mossy fiber terminals NMF for each side within each lobule and dividing by half the total volume V of each lobule (Vi), in order to represent each hemisphere separately (D = NMF/(Vi*½)). Total volume measurements V for each lobule were based on 16.4T MRT data (Ullmann et al., 2012; VI = 0.14 mm3, VII = 1.16 mm3, VIII = 1.79 mm3, VIV/V=5.61mm3, VVI = 2.5 mm3, VVII = 0.67 mm3, VVIII = 1.52 mm3, VIX = 2.81 mm3, VX = 1.26 mm3, VSim = 4.67 mm3, VCrusI = 4.17 mm3, VCrusII = 4.18 mm3, VPM = 3.59 mm3, VCop = 2 mm3, VPFI = 3.35 mm3, VFI = 0.79 mm3).
For internal validation of AldoC+/-alignment across animals, we calculated Pearson’s correlation for the frequency of mossy fiber terminals in AldoC+ and AldoC- regions of each lobule across animals. The pattern of labeling across lobules for animals with the same cortical injection sites (e.g. M1 - animal 1, 2, and 3) should be significantly correlated with each other if the AldoC alignment is consistent across animals.
For nearest neighbor analyses for both mossy fiber terminal locations as well as the location of anterogradely (from cortex) and retrogradely (from cerebellum) labeled cells, the five nearest events to each individual event (i.e. terminal location or labeled cell) was identified in turn and indexed according to their origin (i.e. cortical or cerebellar injection site) in order to quantify the proportion of nearest neighbors from the appropriate categories. Pairwise Euclidean distances between the identified nearest neighbours were calculated for each event and then averaged across events. Further, pairwise Euclidean distanced were calculated for all individual events in relation to all other events according to their origin (i.e. cortical or cerebellar injection site) and within a 500 µm radius.
Point-process analyses to test for spatial randomness of terminal locations were performed using a three-dimensional Ripley’s K-function (K(t); Marani and Voogd, 1979; RipleyGUI: Hansson et al., 2013), which can be simply stated as:
in which a constant total point intensity is estimated using , where n is the observed number of points in a region V (Jafari-Mamaghani et al., 2010). Edge correction was employed (Jafari-Mamaghani et al., 2010) and a maximum distance of t = 60 µm was used. To quantify the deviation of a distribution from complete spatial randomness (CSR), a comparison set of 100 CSR distributions was created with the same properties (size and intensity) as the test distribution, the corresponding estimates of the K-functions are then processed to return a probability density function for each t. P-values were then acquired by locating K(t) on the corresponding probability density function and calculating the area under the curve (Hansson et al., 2013).
Data availability
Data generated during this study (e.g cell counts, etc) are included in the manuscript and supporting files.
References
-
The cloning of zebrin II reveals its identity with aldolase CDevelopment 120:2081–2090.
-
Cerebellar cortical organization: a one-map hypothesisNature Reviews Neuroscience 10:670–681.https://doi.org/10.1038/nrn2698
-
BookCerebro-cerebellar connectionsIn: Manto M, Gruol D. L, Schmahmann J, editors. Handbook of the Cerebellum and Cerebellar Disorders. Dordrecht. Netherlands: Springer. pp. 1131–1154.https://doi.org/10.1007/978-94-007-1333-8_48
-
Divergent projections of single pontocerebellar axons to multiple cerebellar lobules in the mouseJournal of Comparative Neurology 527:1966–1985.https://doi.org/10.1002/cne.24662
-
Zebrin II: a polypeptide antigen expressed selectively by purkinje cells reveals compartments in rat and fish cerebellumThe Journal of Comparative Neurology 291:538–552.https://doi.org/10.1002/cne.902910405
-
Salient anatomic features of the cortico-ponto-cerebellar pathwayProgress in Brain Research 114:227–250.https://doi.org/10.1016/s0079-6123(08)63367-1
-
Topography of descending projections from the inferior colliculus to auditory brainstem nuclei in the ratThe Journal of Comparative Neurology 328:377–392.https://doi.org/10.1002/cne.903280305
-
Mossy-fibre sensory input to the cerebellumProgress in Brain Research 114:251–262.https://doi.org/10.1016/s0079-6123(08)63368-3
-
High-resolution imaging of fluorescent whole mouse brains using stabilised organic media (sDISCO)Journal of Biophotonics 12:e201800368.https://doi.org/10.1002/jbio.201800368
-
RipleyGUI: software for analyzing spatial patterns in 3D cell distributionsFrontiers in Neuroinformatics 7:5.https://doi.org/10.3389/fninf.2013.00005
-
Possible anatomical pathways for short-latency multisensory integration processes in primary sensory corticesBrain Structure and Function 220:955–977.https://doi.org/10.1007/s00429-013-0694-4
-
Crossmodal connections of primary sensory cortices largely vanish during normal agingFrontiers in Aging Neuroscience 10:1–14.https://doi.org/10.3389/fnagi.2018.00052
-
Postnatal development of the pontine projections from the visual cortex of the mouseOkajimas Folia Anatomica Japonica 67:479–492.https://doi.org/10.2535/ofaj1936.67.6_479
-
Spatial point pattern analysis of neurons using Ripley's K-Function in 3DFrontiers in Neuroinformatics 4:9.https://doi.org/10.3389/fninf.2010.00009
-
Properties of somatosensory synaptic integration in cerebellar granule cells in vivoJournal of Neuroscience 26:11786–11797.https://doi.org/10.1523/JNEUROSCI.2939-06.2006
-
Cerebellar loops with motor cortex and prefrontal cortex of a nonhuman primateThe Journal of Neuroscience 23:8432–8444.https://doi.org/10.1523/JNEUROSCI.23-23-08432.2003
-
BookNeocortexIn: Watson C, Paxinos G, Puelles L, editors. The Mouse Nervous System. Academic Press. pp. 52–111.https://doi.org/10.1016/B978-0-12-369497-3.10004-4
-
Projections from the auditory cortex to the pontine nuclei in the rabbitBehavioural Brain Research 56:23–30.https://doi.org/10.1016/0166-4328(93)90019-M
-
Independent roles for the dorsal paraflocculus and vermal lobule VII of the cerebellum in visuomotor coordinationExperimental Brain Research 177:209–222.https://doi.org/10.1007/s00221-006-0661-x
-
Topography of the complete corticopontine projection: from experiments to principal mapsFrontiers in Neuroscience 1:211–223.https://doi.org/10.3389/neuro.01.1.1.016.2007
-
Corticopontine projection in the rat: the distribution of labelled cortical cells after large injections of horseradish peroxidase in the pontine nucleiThe Journal of Comparative Neurology 286:427–441.https://doi.org/10.1002/cne.902860403
-
BookCerebrocerebellar Loops in the Rodent BrainIn: Heck DH, editors. The Neuronal Codes of the Cerebellum. Academic Press. pp. 135–153.https://doi.org/10.1016/B978-0-12-801386-1.00006-X
-
The morphology of the mouse cerebellumActa Morphologica Neerlando-Scandinavica 17:33–52.
-
Sensorimotor coding of vermal granule neurons in the developing mammalian cerebellumThe Journal of Neuroscience 39:6626–6643.https://doi.org/10.1523/JNEUROSCI.0086-19.2019
-
Projections to the basilar pontine nuclei from face sensory and motor regions of the cerebral cortex in the ratThe Journal of Comparative Neurology 237:251–263.https://doi.org/10.1002/cne.902370209
-
Survey of noncortical afferent projections to the basilar pontine nuclei: a retrograde tracing study in the ratThe Journal of Comparative Neurology 282:617–643.https://doi.org/10.1002/cne.902820411
-
Organization of visual mossy fiber projections and zebrin expression in the pigeon vestibulocerebellumThe Journal of Comparative Neurology 518:175–198.https://doi.org/10.1002/cne.22192
-
Action and learning shape the activity of neuronal circuits in the visual cortexCurrent Opinion in Neurobiology 52:88–97.https://doi.org/10.1016/j.conb.2018.04.020
-
Two optic flow pathways from the pretectal nucleus lentiformis mesencephali to the cerebellum in pigeons (Columba livia)The Journal of Comparative Neurology 499:732–744.https://doi.org/10.1002/cne.21108
-
Focal projections of cat auditory cortex to the pontine nucleiThe Journal of Comparative Neurology 497:959–980.https://doi.org/10.1002/cne.20988
-
Organization of pontocerebellar projections to identified climbing fiber zones in the ratThe Journal of Comparative Neurology 496:513–528.https://doi.org/10.1002/cne.20940
-
The cortical states of wakefulnessFrontiers in Systems Neuroscience 12:64.https://doi.org/10.3389/fnsys.2018.00064
-
Cerebellum involvement in cortical sensorimotor circuits for the control of voluntary movementsNature Neuroscience 17:1233–1239.https://doi.org/10.1038/nn.3773
-
How the cerebellum may monitor sensory information for spatial representationFrontiers in Systems Neuroscience 8:1–13.https://doi.org/10.3389/fnsys.2014.00205
-
BookPrecerebellar Nuclei and Red NucleusIn: Paxinos G, editors. The Rat Nervous System. Academic Press. pp. 167–204.https://doi.org/10.1016/B978-012547638-6/50009-2
-
BookCerebellum and cerebellar connectionsIn: Paxinos G, editors. The Rat Nervous System. Academic Press. pp. 133–205.https://doi.org/10.1016/B978-0-12-374245-2.00009-7
-
Anatomic organization of the basilar pontine projections from prefrontal cortices in rhesus monkeyThe Journal of Neuroscience 17:438–458.https://doi.org/10.1523/JNEUROSCI.17-01-00438.1997
-
Auditory pontine grey: connections and response properties in the horseshoe batEuropean Journal of Neuroscience 3:648–662.https://doi.org/10.1111/j.1460-9568.1991.tb00851.x
-
Facilitation and delay sensitivity of auditory cortex neurons in CF - FM bats, Rhinolophus rouxi and Pteronotus p.parnelliiEuropean Journal of Neuroscience 3:1165–1181.https://doi.org/10.1111/j.1460-9568.1991.tb00051.x
-
Laterality of the pontocerebellar projections in the ratEuropean Journal of Neuroscience 15:1551–1556.https://doi.org/10.1046/j.1460-9568.2002.01993.x
-
Error correction, sensory prediction, and adaptation in motor controlAnnual Review of Neuroscience 33:89–108.https://doi.org/10.1146/annurev-neuro-060909-153135
-
Receiving Areas of the tactile, auditory, and visual systems in the cerebellumJournal of Neurophysiology 7:331–357.https://doi.org/10.1152/jn.1944.7.6.331
-
Multisensory integration: current issues from the perspective of the single neuronNature Reviews Neuroscience 9:255–266.https://doi.org/10.1038/nrn2331
-
The auditory cortex of the house mouse: left-right differences, tonotopic organization and quantitative analysis of frequency representationJournal of Comparative Physiology A: Sensory, Neural, and Behavioral Physiology 181:559–571.https://doi.org/10.1007/s003590050140
-
Organization of cerebral projections to identified cerebellar zones in the posterior cerebellum of the ratJournal of Neuroscience 32:10854–10869.https://doi.org/10.1523/JNEUROSCI.0857-12.2012
-
Organization of cerebral cortico-olivary projections in the ratNeuroscience Research 7:43–54.https://doi.org/10.1016/0168-0102(89)90036-9
-
Oculomotor cerebellumProgress in Brain Research 151:231–268.https://doi.org/10.1016/S0079-6123(05)51008-2
-
Electrophysiological mapping of novel prefrontal - cerebellar pathwaysFrontiers in Integrative Neuroscience 3:18.https://doi.org/10.3389/neuro.07.018.2009
-
BookCerebro-Cerebellar ConnectionsIn: Manto M, Schmahmann J. D, Rossi F, Gruol D. L, Koibuchi N, editors. Handbook of the Cerebellum and Cerebellar Disorders. Cham: Springer International Publishing. pp. 1–26.
-
The corticopontine system in the rat. I. mapping of corticopontine neuronsThe Journal of Comparative Neurology 208:215–226.https://doi.org/10.1002/cne.902080302
-
In vivo auditory brain mapping in mice with Mn-enhanced MRINature Neuroscience 8:961–968.https://doi.org/10.1038/nn1477
-
Anterograde Trans-Synaptic tagging mediated by Adeno-Associated virusNeuroscience Bulletin 33:348–350.https://doi.org/10.1007/s12264-017-0099-0
-
Synaptic specificity and application of anterograde transsynaptic AAV for probing neural circuitryThe Journal of Neuroscience 40:3250–3267.https://doi.org/10.1523/JNEUROSCI.2158-19.2020
Decision letter
-
Megan R CareyReviewing Editor; Champalimaud Foundation, Portugal
-
Richard B IvrySenior Editor; University of California, Berkeley, United States
-
Roy V SillitoeReviewer; Baylor College of Medicine, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Acceptance summary:
The authors have used a new mono-trans-synaptic tracing technique to characterize disynaptic cerebrocerebellar pathways. The results suggest a substantial amount of overlap within the cerebellar cortex of projections from distinct regions of cerebral cortex. The comprehensive results are beautifully presented and provide an important foundation for our understanding of how efference copies and other cortical information might be used by cerebellum to perform accurate sensory motor actions.
Decision letter after peer review:
[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]
Thank you for submitting your work entitled "Convergence of multimodal inputs connecting the cerebral cortex to the cerebellum" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers. The following individuals involved in review of your submission have agreed to reveal their identity: Richard Apps (Reviewer #2); Roy V Sillitoe (Reviewer #3).
Our decision has been reached after extensive consultation between the reviewers and the Reviewing Editor. Based on these discussions and the individual reviews below, we regret to inform you that we are not prepared to move forward with publication in eLife at this time. Although the reviewers agreed that the manuscript has the potential to serve as a thorough atlas for cortical projections to precerebellar nuclei, it was agreed that this goal was not achieved by the current manuscript. There were concerns regarding specific biological conclusions that can be reached at this stage, based on the data presented. However, the Editors and reviewers were enthusiastic about several aspects of this study. We have therefore prepared a summary of suggestions brought up during the consultation process, which involve new data and substantial revisions to the manuscript. Should these points, as well as the individual comments of the reviewers, be addressed in a reconfigured version of the paper, we would be willing to consider such a future submission for publication at eLife.
1) Overall, the reviewers felt that packaging this paper as a Short Report made it difficult to evaluate the findings, and we suggest expanding the manuscript to strengthen the evidence for the claims that are made. All reviewers wanted more documentation of injection sites (volumes and cell counts), and more evidence for truly comprehensive mapping. This was seen as critical for supporting the main potential strength of this paper as a comprehensive atlas study.
2) While all three reviewers supported the paper in principle, and agreed that it presents an interesting angle on topography and convergence of cortico-cerebellar connections, they each struggled with the data interpretation in light of limitations of the technique. The fundamental problem was that the experiments are unable to indicate the brainstem origin of the MF projection patterns. They may be mainly from the pons, but it is not possible to say if any of the other brainstem structures labelled are also sources. The reviewers came up with two suggested experiments to address this concern:
a) In the spirit of an "atlas paper," the authors could confirm pontine projections to the labeled and unlabelled (A1) regions of cerebellum using retrograde AAV (to rule out extrapontine projections). This could validate the method and strengthen novel findings such as pontine projections to lobes IV/V.
b) Another possibility could be to do dual viral injections: (transsynaptic) AAV-cre in cortex, and AAV-floxed-XFP in pons. If feasible, this could confirm the presence of certain cortical-pontine-cerebellar projections.
3) It was agreed that the conclusions regarding potential multimodal integration in individual granule cells were not adequately supported by the data. Rather, this was seen as one possible (and reasonable) interpretation based on a general overlap of MF labelling. Although reviewers (particularly #1) suggested specific experiments that could strengthen this claim, we do not think that it is critical to demonstrate this kind of convergence in the current manuscript. The demonstration of broad overlap itself is interesting. However, the conclusions regarding ultimate convergence should be toned down. The current findings can be presented as broadly consistent with the idea of multimodal integration in individual granule cells, which has been previously suggested by earlier studies (eg Huang, Ishikawa…).
We hope that you will find these comments and the reviews below helpful in preparing a revised version of the manuscript.
Reviewer #1:
In this manuscript, Henschke and Pakan describe a beautiful set of experiments in which they used an AAV2.1-based anterograde, fluorescence protein-based tracing technique (Zingg et al., 2017) to elucidate cerebro-pontine-cerebellar pathways. They focused on primary sensory (S1, V1 and A1), motor and one association area (PPC). They observed, as in previous studies, that these regions target ipsilateral basal pons in a spatially segregated manner. In turn, cerebellar mossy fiber (MFs) are found to be labelled in specific regions cerebellum including both in the vermis and lateral hemispheres. These results are consistent with other tracer and electrophysiological studies. Interestingly they find that A1 does not provide pontine-cerebellar projections, but dorsal auditory cortex does. This contrasts previous results in other species. Finally, and most novel, the authors examined the spatial overlap of MF projections from different cortical regions and show that in some regions multimodal processing is likely. While this has been shown for single granule cells in vivo in Crus I (Ishikawa et al., 2016), a survey of what cerebellar regions are likely to receive multimodal cerebro-pontine input had not been shown previously. In summary, the authors provide provoking evidence of a strategy to specifically label and manipulate different that pontine-cerebellar circuits will be possible.
The technique and results described in this manuscript are illuminating, and the data thoroughly analyzed. The manuscript, therefore, has the potential to serve as a thorough atlas for cortical projections to precerebellar nuclei. This study paves the way for future studies to refine our understanding of how efference copies and other cortical information might be used by cerebellum to perform accurate sensory motor actions. Unfortunately, limitations in the method preclude a clear biological conclusion from the current dataset, beyond the notion that cortical-pontine-cerebellar wiring is similar in mice as other species:
1) The authors show that AAV1-cre injections in most cortical regions tested indeed label the pons, but also extra-pontine precerebellar nuclei. This highlights the fundamental problem that we cannot be sure that the MF projection patterns in cerebellum are solely a result of pontine projections -as implied in the manuscript. One example of a new biological conclusion is that cerebro-pontine projections were largely thought to be restricted to posterior and lateral cerebellum, but here the authors seem to show prominent projections in lobes 4 and 5. However, we cannot be sure of this conclusion because the MF labelling in lobes 4 and 5 could be from another precerebellar nucleus. One possible experiment to confirm that pons projects to a given region of cerebellum would be to perform retrograde labelling from different cerebellar regions identified in the anterograde transynaptic method, and verify that cell bodies in pons are labeled.
2) Another conclusion proposed by the authors is that multimodal integration is likely in certain cerebellar regions. While overlapping domains of MF projections is necessary to achieve multimodal integration. A critical question is whether single granule cells receive different cortical inputs. Such multimodal combinations have been described for motor and proprioception (Huang et al., 2015), for visual, somatosensory and auditory (Ishikawa et al., 2015), as well as visual and vestibular information (Chabrol et al., 2015). Understanding the role of multimodal processing in cerebellum requires knowing whether cortical information is "mixed" at the level of single granule cells, as proposed by Marr and Albus. Are there specific wiring rules for cerebro-ponto-cerebellar projections? For example, perhaps corollary information only pairs with its principal sensory modality partner. Unfortunately, these experiments are more challenging and would require two-color labelling using another conditional method (e.g. Flp), in addition to a strategy to sparsely label GCs (electroporation, patch or mouse lines (Huang et al., 2013)). Nevertheless, this would significantly advance our understanding of how cortical information is processed at the input layer of cerebellum.
3) The lack of cortico-pontine projections into cerebellum from A1 is surprising, in light of previous work. Thus, it is necessary to independently confirm this result using retrograde labelling strategies. If true, a better discussion of why it might be different than previous results is merited. Finally, another concern about the A1 experiments, is that in Figure 1—figure supplement1 AAV1-cre injections there is very little labeling of inferior colliculus. As I understand the method, retrograde labelling is also expected. As the exact mechanism of transynaptic propagation of the virus or Cre is not known, is it possible that the method failed for this brain region?
I also have additional specific concerns.
1) When examining table 2, I noticed that similar numbers of labelled precerebellar cell bodies can produce differences in the number of MFs by nearly a factor of 10. Does this discrepancy limit the ability normalize for labelling efficiency and thus limit quantification of the results? How could this happen?
2) The conditions used to perform imaging should be better described. For example, how was high-resolution imaging performed with a 2.5 x objective? What was the NA of this objective? Given the accuracy of the galvanometer mirrors, necessary zooms may not be possible to achieve true diffraction limit resolution. The pixel and image sizes should be mentioned.
3) How was the injection volume measured? The width of the column size measure was presumably estimated from a line profile along an image, but what width parameter was taken as a width. The images shown are clearly saturated. Is this the same image in which the size of the infected region was estimated?
Reviewer #2:
This is a short report using a trans mono synaptic anterograde tracer method to chart cerebro-cerebellar anatomical projections in mice. Injections into a number of different neocortical regions (M1, S1, V1, A1, posterior parietal association cortex or the dorsal field of auditory cortex) revealed topographically organized projections to the pons and mossy fibre terminals in the cerebellar cortex. The report is generally well written and provides a valuable 'road map' for future studies of the functional significance of cerebro-cerebellar projections in mice.
Essential revisions:
1) The report needs to make clear what are the novel findings.
2) How did differences in cell counts and other measures of injection site affect the pattern and extent of anterogradely labelled cell and mossy fibre terminals? It was not clear to me if Figure 2A shows the total extent of injection sites into different neocortical areas or just the defined regions of interest. Either way the area clearly varies considerably between targets which will affect projection density and possibly topography.
3) The histogram axes in Figure 2D are not all drawn on the same scale making comparisons difficult. Also percentage values are potentially misleading given the large variation in cell counts and mossy fibre terminals listed in Figure 1—figure supplement 1B.
4) Given that the pons is the primary source of mossy fibre inputs, how did the authors determine that other brainstem nuclei also containing labelled cells in the same experiments necessarily provided mossy fibre projections?
5) The topographical organization within the pons should be compared to the principles of organization defined in previous studies (eg Leergaard, 2003; Odeh et al., 2005).
6) The discussion should consider the extent to which cerebro-cerebellar pathways are conserved across mammalian species. How useful is the mouse as a model for study of cerebro-cerebellar projections?
Reviewer #3:
Henschke and Pakan have submitted a manuscript entitled "Convergence of multimodal inputs connecting the cerebral cortex to the cerebellum" in which they use modern neuronal tracing methods to map circuit projections into the mouse cerebellum. The authors traced the cerebellar inputs originating from sensory, motor, and association cortices. They argue that diverse functional connections project from the cortex to cerebellum, where they ultimately exhibit a convergent terminal field distribution within specific lobules. The manuscript is well written and technically, it offers a much-needed approach to better understand the organization and trajectories of major cortico-cerebellar projections. In addition, the figures are well made, schematics very helpful, and overall the panels are easy to follow. However, below we provide a number of comments and concerns that we hope will improve the clarity and impact of the study.
1) Throughout the text, the discussion of A1 and its unique projection through non-pontine pre-cerebellar nuclei is often highlighted, however the significance of this point is not well articulated. Further clarification in the abstract and in the body is needed if the authors wish to highlight the uniqueness of this pathway. This is especially true given the findings in the text that S1 and M1 both have substantial, if not even majority in the case of S1, projections to non-pontine pre-cerebellar nuclei.
2) Introduction. Please clarify why the identified cortical regions were selected. In addition, while we agree that the characterization of the injections is rather comprehensive, there is a limitation in this study in that either the number of cortical regions chosen or the breadth of the characterization of the selected regions (ie: why weren't at least two neighboring sites in a given cortical region analyzed) prevent the present study from providing a fully comprehensive map, as stated by the authors.
3) Figure 1A. Please indicate the circumscribed subregion in each cortical area where the injection was made.
4) Subsection “Anterograde tracing of indirect cerebrocerebellar pathways”. Please clarify percentages for A1 projections in order to make the second sentence "This pattern of extra-pontine labeling was similar…" clearer. Further, in Figure 2B, while the scarcity of the projections to the pontine nuclei is similar, they seem to be qualitatively different in that A1 projects terminals, while the AuD experiment labels cell bodies. This distinction is not clear in the text. Moreover, the fact you see terminals in one region and then cell bodies another makes it a challenge to piece together what the actual pathway(s) might look like. That is, how do you substantiate the claim that the cell bodies in the pons (the unexpected labeling) actually project to the cerebellum? And on the flip side, for the majority of S1 labeling, which is in the precerebellar non-pontine nuclei, can we really assume that these all project to cerebellum? If not, what percentage?
5) The Supplementary file 1 is a bit difficult to interpret. Please include the percentages referenced in the text in the table. In addition, it is unclear why in some injection sites the sum of the individual pre-cerebellar nuclei cells equals the summed total, while in other cases the sum is less than or greater than the listed total. (ie M1-1 52 vs 55; V1-1 8 vs 8; A1-1 10 vs 8)
6) The significance of terminals in pontine/pre-cerebellar nuclei without cell body labelling is unclear and there should be an explanation of this in the text. For instance, in regard to A1/AuD does this indicate inefficiency in trans-synaptic labelling, that there is an alternative monosynaptic projection from A1/AuD to the pontine nuclei, or is there an alternative explanation? This is mentioned for M1 projections to the olivary nuclei (Subsection “Anterograde tracing of indirect cerebrocerebellar pathways”).
7) In the subsection "Convergence of multimodal cerebrocerebellar input within lobules…" the metric for dominant representation (> or < 55%) is problematic given the relative sparsity of the injections (either across or within cortical regions). Furthermore, given the variability of numbered terminals across individual animals in a given cortical area (ie from 2406 to 25616 from the M1 injections), this type of quantification and conclusions about convergence seem premature.
8) While the discussion about convergence and integration is quite interesting to speculate upon, it is hard to grasp whether the current data is sufficient to reach such a conclusion. Given the vast number of granule cells and the fine scale nature of cerebellar microzones, conclusions about integration of multimodal afferent information by single units (whether microzone or granule cell) cannot be easily drawn. To do this, a method distinct from the approach used in this paper would be needed whereby labelling from two distinct cortical areas could be distinguished in the cerebellum in a single animal. In the end, I am having trouble fully appreciating how integrated the "multimodal" and "convergence" nature of the pathways actually is, based on the data provided.
9) Finally, there is a core assumption at play, which is that cortical injection - pontine/precerebellar cell body labeling - mossy fiber terminal marking. In the case of cortico-ponto-cerebellar pathways, the abundance of literature makes this assumption sound. However, in the case of the precerebellar projections (especially S1 which has a majority non-pontine projections), sagittal sections (for example) demonstrating the axonal projections, and elucidating the monosynaptic pathway would provide a helpful piece of evidence to support the current content.
1) In the Abstract, it is not clear what is actually meant by "Prominent terminal overlap across motor and sensory modalities…". And, how was this measured?
2) It would be very helpful to include the species studied in the Title and/or the Abstract.
3) It is critical that the authors fully outline the limitations of the mono-trans synaptic anterograde viral tracer, in their hands, and specifically for the purpose of this study.
4) In Figure 1B, it is not clear what the red arrows are representing and pointing towards. Aren't the arrows pointing the wrong way if this is to show anterograde tracing direction?
5) Is there a time dependence on the tracing? That is, do longer periods of survival after injection label additional projections?
6) In Figure 1F, M1 should project ipsilateral, but in the cerebellum one can see many terminals on both sides. Please explain.
7) The authors state "…extent M1, indicating the presence of a polysynaptic cerebro-olivary pathway…" What is the literature to support this?
8) The authors state "We quantified the number of resulting mossy fiber terminals throughout the extent of the cerebellum and found that this was highly variable across cortical regions…" How variable is the size of the infected cell population in the cortex?
9) The authors state "There was a clear and consistent topography of…" It would be helpful to fully define what you mean by topography here. There are many levels of topography, in different brain regions, and this could mean very different things to different researchers.
10) In Figure 3E, what does each data point represent? A single terminal?
11) Along the same lines as above, it would be helpful to see more actual data with the combined zebrin labeling. Mossy fiber patterns, by nature of their clustered appearance, are not so easily appreciated and therefore additional examples, at higher power magnification, would be welcomed.
12) The authors state "This allowed for a detailed comparison of cerebellar regions across brains and, hence, across functional cortical regions representing multiple modalities (see Figure 3—figure supplement 1)." However, I don't see a full description or explanation of the data in the text.
13) The number of animals used in the study seems quite low.
14) In Figure 3—figure supplement 1, what is the red staining in Purkinje cells?
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Thank you for resubmitting your work entitled "Cerebral cortical connections to the cerebellum across multiple modalities" for further consideration by eLife. Your revised article has been evaluated by Richard Ivry (Senior Editor) and a Reviewing Editor.
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:
Summary:
Henschke and Pakan have provided a substantially revised version of a paper describing the connectivity between different areas of the cerebral cortex with the cerebellum. The authors have used a virus based transsynaptic tracing approach to show that multi-modal inputs may converge upon specific topographical domains within the cerebellar vermis and hemispheres. The study is anatomical in nature and takes an in-depth analysis of mossy fiber terminal field distributions in the cerebellum, but the authors also examine the different intermediate projections of the cerebral-cerebellar projections that are located within various brainstem nuclei. The paper is well-written and the images are high quality. reviewers particularly appreciated the addition of important controls (characterization of extra-pontine sources of MFs, confirmation of pontine sources using retrograde labeling, and identification of an A1-colliculo-cerebellar pathway), which significantly increase confidence in their original findings and propel the study towards the authors' original objective as a seminal reference, or atlas, for the characterization of cerebrocerebellar mossy fiber (MF) pathways.
Essential revisions:
The major remaining concern – shared by both reviewers – is the overemphasis/ speculation on 'multimodality' for this essentially anatomical study. Specifically:
1) I was delighted to see a more detailed examination of the location of MFs from different modalities within each lobule, since a prerequisite for multimodal processing is that the information project to the same subregion of a lobule. I also found rather striking the case in Crus II where there was still segregation between two MF types within the same lobe. This latter result highlights the notion that multimodal innervation within lobes is necessary but not sufficient to conclude multimodal processing. I therefore encourage the authors to soften their choice of words of "highly multimodal" and "moderately multimodal" as a conclusion of Figure 4. Perhaps the use of the term "lobule co-innervation" is more appropriate. Then after Figure 5, the argument for putative multimodal processing is much stronger, but still only suggestive.
2) Indeed, Figure 5 shows an essential set of analyses suggestive of multimodal processing, but I was left wanting for some clear quantification of "intermingled" and a simple overall summary.
a) Would it be possible to analyze the distribution of nearest-neighbor distances between terminals of different origins? Those that co-innervate would have shorter distances and those that do not, e.g. Crus II, would show a shifted distribution. Perhaps a more sophisticated point-process analysis, such as the application of the bivariate Ripley K-function, could be used to test for spatial randomness between two species of points and provide statistical confidence.
b) To add some statistical rigor to the alignment with adolase C staining, one could compare the frequency of AldoC positive and negative terminals for each cortical injection. Statistical tests for the frequency of events can then be used.
3) The title continues to hint at a more functional set of analyses, which the paper really does not provide. Please revise.
4) Similarly, the Abstract could present the key anatomical distinctions that were uncovered rather than the speculative discussion about multimodal sensory input – this functional idea should remain, but it should be toned down to make space to better clarify the actual anatomical findings.
5) The main Figure 6 merits a more thorough quantification of retrograde labeling.
6) Quantification and presentation of co-labeled and spatially intermingled neurons following dual viral tracing injections would be appropriate.
7) I like the estimation of the number of MFs per pontine neuron and would like to see it as a main figure.
8) The Discussion section rambles on a bit. I believe it can be shortened and used to more clearly summarize the major findings, the similarities and differences with previous results, and the implications for cerebellar function.
9) The section titles are rather vague. For example, "Organization of cerebrocerebellar mossy fiber terminals" is essentially duplicated in the Results and Discussion, and perhaps needs more specific wording. "Spatial organization of cerebrocerebellar mossy fiber types". I am not really sure why there is a Discussion subsection on "Diversity of cerebrocerebellar pathways" as you define the pathways by viral injection. Perhaps something like: "Cerebellar cortical target diversity of cerebrocerebellar pathways?"
https://doi.org/10.7554/eLife.59148.sa1Author response
[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]
Our decision has been reached after extensive consultation between the reviewers and the Reviewing Editor. Based on these discussions and the individual reviews below, we regret to inform you that we are not prepared to move forward with publication in eLife at this time. Although the reviewers agreed that the manuscript has the potential to serve as a thorough atlas for cortical projections to precerebellar nuclei, it was agreed that this goal was not achieved by the current manuscript. There were concerns regarding specific biological conclusions that can be reached at this stage, based on the data presented. However, the Editors and reviewers were enthusiastic about several aspects of this study. We have therefore prepared a summary of suggestions brought up during the consultation process, which involve new data and substantial revisions to the manuscript that are likely to take more than two months. Should these points, as well as the individual comments of the reviewers, be addressed in a reconfigured version of the paper, we would be willing to consider such a future submission for publication at eLife.
1) Overall, the reviewers felt that packaging this paper as a Short Report made it difficult to evaluate the findings, and we suggest expanding the manuscript to strengthen the evidence for the claims that are made. All reviewers wanted more documentation of injection sites (volumes and cell counts), and more evidence for truly comprehensive mapping. This was seen as critical for supporting the main potential strength of this paper as a comprehensive atlas study.
As suggested, we have now structured the manuscript as a full Research Article adding additional experiments, main and supplementary Figures, and agree that this allows us to more fully elaborate on the many implications that follow from our results.
2) While all three reviewers supported the paper in principle, and agreed that it presents an interesting angle on topography and convergence of cortico-cerebellar connections, they each struggled with the data interpretation in light of limitations of the technique. The fundamental problem was that the experiments are unable to indicate the brainstem origin of the MF projection patterns. They may be mainly from the pons, but it is not possible to say if any of the other brainstem structures labelled are also sources. The reviewers came up with two suggested experiments to address this concern:
We have performed a number of new experiments that provide additional information regarding the brainstem origin of the mossy fiber projections. These are described in detail below, and also further in the specific responses to the three reviewers.
a) In the spirit of an "atlas paper," the authors could confirm pontine projections to the labeled and unlabelled (A1) regions of cerebellum using retrograde AAV (to rule out extrapontine projections). This could validate the method and strengthen novel findings such as pontine projections to lobes IV/V.
We have confirmed the finding of pontine projections to lobules IV/V by using retrograde tracers as suggested (see new Figure 6 and Figure 6—figure supplement 1).
We have further completed retrograde injections in the key cerebellar regions identified as hubs for multimodal spatial convergence (see new Figure 6 and related supplementary Figures). However, It is difficult to know specifically which A1 receiving cerebellar regions (‘unlabelled’ through pons) would not actually receive pontine projections via other cortical areas and hence still result in substantial pontine labelling via retrograde injections (e.g. see lobule VII injections and Figure 6 – supplementary Figure 2). Pontine and extrapontine pathways may have overlapping terminal fields in the cerebellum – hence, success in separating them spatially with retrograde injections would be unlikely. This is also evident from our retrograde injections in Sim, Crus I, and lobules IV/V, VI, and VII where in each case we found retrogradely labelled cells in both pontine and numerous extrapontine nuclei (see new Figure 6 and Figure 6—figure supplement 1 and Figure 6—figure supplement 2).
Furthermore, we do not think that extrapontine projections should be ‘ruled out’ in this regard – but rather that these pathways are actually likely to contribute (in combination with pontine pathways) to disynaptic cerebrocerebellar input (see Discussion section; reviewer 1 point 1). Of course, this makes these pathways more complicated to separate on the individual level, and the feasibility of making anatomical injections in each separate precerebellar nuclei to trace individual mossy fibers is quite low (e.g. due to difficulty in targeting these brainstem nuclei without approaching through the cerebellum, see also point 2b below). However, we feel that the addition of our new experiments as a whole have added substantial information regarding the potential for both pontine and extra-pontine cerebrocerebellar pathways and the implications of this are also discussed in more detail in the revised Discussion section.
b) Another possibility could be to do dual viral injections: (transsynaptic) AAV-cre in cortex, and AAV-floxed-XFP in pons. If feasible, this could confirm the presence of certain cortical-pontine-cerebellar projections.
Due to the technical challenges inherent in this approach as well as some limitations specified below, we opted for the strategy described above (see point 2a). Specifically, to reach the pontine nuclei you must either go through the cerebellum or cortex – creating potential confounds through the chance for AAV leakage of the floxed-XFP construct. Additionally, one would need 100% transfection rate in the pontine nuclei to be able to conclude that any RFP labelled mossy fiber terminals were from extra-pontine sources. If even one or two pontine cells expressed RFP but not the floxed-XFP, then we would still be unable to conclude the specificity of this pathway; 100% transfection rates with AAVs are rarely feasible – given the rostral caudal extent of labelling possible in the pons, even less so. Finally, to accurately determine the likelihood of each separate precerebellar nuclei projecting as intermediate connections between the cerebral cortex and cerebellar cortex we would have to complete this strategy (AAV-floxed-XFP) for all noted individual precerebellar nuclei. This strategy was not feasible – due to the same issues as stated above, as well as the need for targeting each nuclei without spreading to surrounding structures but still resulting in enough labelling to double label mossy fibers.
In short, we fully agree that the precise brainstem origin of these mossy fiber terminals is of great interest. The most feasible strategy (retrograde injections in the cerebellar cortex, see point 2a) allowed us to determine with greater precision the potential for the specific extrapontine precerebellar nuclei that receive corticofugal input to, in turn, provide input to key cerebellar regions of multimodal spatial convergence. We believe that this additional information provided by our study will be vital to guide many future studies specifically targeting individual pathways for more in depth polysynaptic circuit analysis. This has now been discussed throughout the manuscript and many potential avenues for future research pointed out in the Discussion section.
3) It was agreed that the conclusions regarding potential multimodal integration in individual granule cells were not adequately supported by the data. Rather, this was seen as one possible (and reasonable) interpretation based on a general overlap of MF labelling. Although reviewers (particularly #1) suggested specific experiments that could strengthen this claim, we do not think that it is critical to demonstrate this kind of convergence in the current manuscript. The demonstration of broad overlap itself is interesting. However, the conclusions regarding ultimate convergence should be toned down. The current findings can be presented as broadly consistent with the idea of multimodal integration in individual granule cells, which has been previously suggested by earlier studies (eg Huang, Ishikawa…).
We agree that this is a specific point of interest and also should be explored in future studies – particularly with further development of circuit tracing tools combined with electrophysiological techniques. Here, we have been more careful to describe convergence as regional or spatial in nature, rather than at the cellular-integration level, and have expanded our discussion with regard to the individual granule cell level with reference to previously published literature which was already cited (e.g. Huang, Ishikawa, Hantman, etc) and including additional evidence from Markwalter et al., 2019 (subsection “Neuroanatomical injections”).
Reviewer #1:
In this manuscript, Henschke and Pakan describe a beautiful set of experiments in which they used an AAV2.1-based anterograde, fluorescence protein-based tracing technique (Zingg et al., 2017) to elucidate cerebro-pontine-cerebellar pathways. They focused on primary sensory (S1, V1 and A1), motor and one association area (PPC). They observed, as in previous studies, that these regions target ipsilateral basal pons in a spatially segregated manner. In turn, cerebellar mossy fiber (MFs) are found to be labelled in specific regions cerebellum including both in the vermis and lateral hemispheres. These results are consistent with other tracer and electrophysiological studies. Interestingly they find that A1 does not provide pontine-cerebellar projections, but dorsal auditory cortex does. This contrasts previous results in other species. Finally, and most novel, the authors examined the spatial overlap of MF projections from different cortical regions and show that in some regions multimodal processing is likely. While this has been shown for single granule cells in vivo in Crus I (Ishikawa et al., 2016), a survey of what cerebellar regions are likely to receive multimodal cerebro-pontine input had not been shown previously. In summary, the authors provide provoking evidence of a strategy to specifically label and manipulate different that pontine-cerebellar circuits will be possible.
The technique and results described in this manuscript are illuminating, and the data thoroughly analyzed. The manuscript, therefore, has the potential to serve as a thorough atlas for cortical projections to precerebellar nuclei. This study paves the way for future studies to refine our understanding of how efference copies and other cortical information might be used by cerebellum to perform accurate sensory motor actions. Unfortunately, limitations in the method preclude a clear biological conclusion from the current dataset, beyond the notion that cortical-pontine-cerebellar wiring is similar in mice as other species:
1) The authors show that AAV1-cre injections in most cortical regions tested indeed label the pons, but also extra-pontine precerebellar nuclei. This highlights the fundamental problem that we cannot be sure that the MF projection patterns in cerebellum are solely a result of pontine projections -as implied in the mansucript. One example of a new biological conclusion is that cerebro-pontine projections were largely thought to be restricted to posterior and lateral cerebellum, but here the authors seem to show prominent projections in lobes 4 and 5. However, we cannot be sure of this conclusion because the MF labelling in lobes 4 and 5 could be from another precerebellar nucleus. One possible experiment to confirm that pons projects to a given region of cerebellum would be to perform retrograde labelling from different cerebellar regions identified in the anterograde transynaptic method, and verify that cell bodies in pons are labeled.
We performed additional tracing experiments with injections of the retrograde tracer Choleratoxin-B (CTB) into lobes IV/V, VI, VII, Sim and crus I of the cerebellum, as well as using a dual injection approach as a proof of principle to show double labelled cells in the pontine nuclei from injections of the trans-mono-synaptic AAV1.cre into M1 and retrograde CTB injections in lobule IV/V (Figure 6 and Figure 6—figure supplement 1, Figure 6—figure supplement 2). Our results show that the majority of retrogradely labeled cells were located in the pons, with some cells also double labeled, i.e. direct confirmation that these pontine cells receive disynaptic input from the specified cortical regions. We also note, however, that many precerebellar nuclei identified as receiving cortical projections also contained retrograde labelling in these key regions (e.g. lateral reticular nucleus (LRt), reticulotegmental nucleus (RtTg), vestibular nucleus (VN), interpolar part of the spinal trigeminal nucleus (Sp5I) and matrix region x (Mx) (see new Figure 6 and Figure 6—figure supplement 1). Although double labelled cells were not directly observed in extra-pontine nuclei, the labelling was spatially overlapping (Figure 6B) and the proportion of retrogradely labeled cells in extra-pontine nuclei were particularly substantial from lobule IV/V and VI (Figure 6—figure supplement 1). Given the low density of trans-synaptic mossy fiber labelling and the spatially restricted injection sites, we believe the probability of observing double labelled cells within these precerebellar nuclei was very low; however, importantly, these results indicate the potential for specific precerebellar nuclei to either relay or integrate cerebrocerebellar information. Therefore, we conclude that cortical information is likely to reach the cerebellum through both pontine as well as extra-pontine intermediate nuclei. This is now discussed in detail in the revised manuscript (e.g. Discussion section).
2) Another conclusion proposed by the authors is that multimodal integration is likely in certain cerebellar regions. While overlapping domains of MF projections is necessary to achieve multimodal integration. A critical question is whether single granule cells receive different cortical inputs. Such multimodal combinations have been described for motor and proprioception (Huang et al., 2015), for visual, somatosensory and auditory (Ishikawa et al., 2015), as well as visual and vestibular information (Chabrol et al., 2015). Understanding the role of multimodal processing in cerebellum requires knowing whether cortical information is "mixed" at the level of single granule cells, as proposed by Marr and Albus. Are there specific wiring rules for cerebro-ponto-cerebellar projections? For example, perhaps corollary information only pairs with its principal sensory modality partner. Unfortunately, these experiments are more challenging and would require two-color labelling using another conditional method (e.g. Flp), in addition to a strategy to sparsely label GCs (electroporation, patch or mouse lines (Huang et al., 2013)). Nevertheless, this would significantly advance our understanding of how cortical information is processed at the input layer of cerebellum.
We agree that this is a very interesting question, and indeed, difficult to address fully at this time with our current tools. A very recent proof of principle paper has just been published (Zingg et al., 2020) using similar techniques to that suggested, with Flp and cre dual labeling, so this may be a strategy we can also pursue in the future given the availability of AAVs, mouse lines, etc. Regardless, for any future in vivo study of multimodal cellular integration in this regard, targeting in the cerebellum will also be difficult since the density of labelled mossy fiber projections is limited, one would need to maximize the likelihood of targeting investigations to a region (and even a specific module) with high spatial overlap of cerebrocerebellar terminals. This was a large motivation for the current study and is where we see a main strength – as a vital first step/road-map towards future in vivo experiments examining the convergence of cortical input in the cerebellum; making the current manuscript a unique resource moving forward for the field as a whole (see also response to editor, point 3).
3) The lack of cortico-pontine projections into cerebellum from A1 is surprising, in light of previous work. Thus, it is necessary to independently confirm this result using retrograde labelling strategies. If true, a better discussion of why it might be different than previous results is merited. Finally, another concern about the A1 experiments, is that in Figure 1—figure supplement 1 AAV1-cre injections there is very little labeling of inferior colliculus. As I understand the method, retrograde labelling is also expected. As the exact mechanism of transynaptic propagation of the virus or Cre is not known, is it possible that the method failed for this brain region?
To further support our finding of a lack of cortico-pontine projections from A1 we performed two additional sets of experiments. First, we injected a retrograde CAV.cre virus into the pons in the tdTomato reporter mice and found numerous retrogradely labeled cells in the dorsal (AuD) and ventral (AuV) auditory cortex but none in the primary auditory cortex (A1). This finding confirms our previous results that there is no mono-synaptic cerebro-pontine pathway originating in mouse A1.
However, there is a disynaptic pathway connecting A1 with the IC and the pons. This was already evident from our previous experiments, where we found mono-trans-synaptic labeled cells (most likely not retrogradely labeled cells; see below) in the IC. These cells were mainly found in the ECIC and DCIC but rarely in the CIC, as shown in Figure 2 of the revised version of the manuscript, which better illustrates the large amount of labeling in the ECIC/DCIC; therefore, we do not feel as though this method failed along this particular pathway and this is now further demonstrated. Additionally, in order to provide further support for a disynaptic A1-IC-pons pathway, we injected the anterograde trans-monosynaptic AAV1.cre virus into the IC. Following these injections, we found labeled cells in the pons and subsequent MFs in the cerebellum (Figure 1—figure supplement 2).
Based on the results of these two new sets of experiments we conclude that there is indeed not a direct cerebro-pontine pathway to the cerebellum from A1, but instead likely a trisynaptic pathway via the IC. Although the extent of this trisynaptic pathway is yet to be precisely determined, we discuss these findings and outline the potential pathways in detail in the revised manuscript (subsection “Anterograde tracing of indirect cerebrocerebellar pathways”; Discussion section).
Regarding the potential retrograde transport of the AAV1.cre virus, we note that there is no reported projection from IC directly to A1 in mice, this pathway is ‘top-down’ from A1 to IC and not reciprocal, rather information from IC travels to A1 via the thalamus (see also Zingg et al., 2020 where this pathway is chosen specifically for its unidirectional nature). We also note that following our AAV1.cre injections into the IC, we found only a couple labeled cells in the auditory cortex of each experimental animal – very likely representing the extent of retrograde capability of this AAV. This is also in agreement with the original Zingg and colleagues papers (2017; 2020), demonstrating the retrograde transport of this virus is not substantial and far less efficient than in the anterograde direction. These considerations have been clarified in the revised manuscript (see subsection “Neuroanatomical injections”).
4) When examining table 2, I noticed that similar numbers of labelled precerebellar cell bodies can produce differences in the number of MFs by nearly a factor of 10. Does this discrepancy limit the ability normalize for labelling efficiency and thus limit quantification of the results? How could this happen?
We do not wish to downplay the variability in the quantity of labelling we see after viral injections, and hence have mentioned this is in a number of places (including providing the data in the revised Supplementary file 1). However, we do see a significant correlation between the number of anterogradely labeled cells within precerebellar nuclei and the number of mossy fiber terminals (this has been made more clear in the revised manuscript, see subsection “Intermediate cerebrocerebellar brainstem pathways”; Figure 6—figure supplement 3) and a surprising amount of consistency across our cases in the pattern of observed labelling. We have made efforts to present the data in both normalized (scaled [Figure 3] and unscaled [Figure 4]) and raw (Supplementary file 1) format to avoid any misleading conclusions based on normalization.
We believe the variability in absolute numbers is a property of the virus/trans-synaptic transport which is highly dependent on the titre, not only of the injected solution, but subsequently that gets taken up into individual cells; which, in turn, is dependent on a number of factors: volume, spread, speed of injection, etc. For instance, an individual cell that contains a high virus titre will result in transport, however, a neighboring cell that still contains virus at the injection site, but at a low titre, will not contribute to the trans-synaptic quantification. We believe this introduces some inherent variability across cases, even with similar injection site total volume. We have expanded our description of the potential factors to be aware of during injection in this regard (see subsection “Neuroanatomical injections”).
5) The conditions used to perform imaging should be better described. For example, how was high-resolution imaging performed with a 2.5 x objective? What was the NA of this objective? Given the accuracy of the galvanometer mirrors, necessary zooms may not be possible to achieve true diffraction limit resolution. The pixel and image sizes should be mentioned.
We have added the appropriate information into the revised manuscript: Sections of 40µm thickness were examined using a confocal microscope (Zeiss LSM 700, Germany) equipped with a 2.5x objective (NA 0.085, Zeiss, Germany). For each section throughout the extent of the cerebellum two high-resolution images (2048 x 2048 Pixel) were acquired, i.e. one for each hemisphere. The two tiles were merged using custom written MatLab scripts (Mathworks, MA, USA; resulting in a ~4000 x 4000 pixel image of the cerebellum). See subsection “Data analysis”.
6) How was the injection volume measured? The width of the column size measure was presumably estimated from a line profile along an image, but what width parameter was taken as a width. The images shown are clearly saturated. Is this the same image in which the size of the infected region was estimated?
We have added further information regarding the exact parameters used to determine the injection site volume, including a supporting Supplementary file 2A: The injection sites had a roughly cylindrical shape. All sections that covered the central core but not surrounding halo of each injection site were included in the analysis (see Supplementary file 2). We then calculated injection site volume as follows: V = π*a*b*h, where h corresponds to the maximum value across sections for the height of the injection site [dorso-ventral direction], a corresponds to the maximum value across sections for the radius of the injection site parallel to the cortical layers [medio-lateral direction], and b corresponds to the rostrocaudal extent of the core injection site. The Line Measure Tool of the Zen software (Zeiss, Germany) was used to measure h and a, and b was calculated by counting the numbers of sections covering the core injection site multiplied by their thickness (40 µm). See subsection “Data analysis”.
Reviewer #2:
This is a short report using a trans mono synaptic anterograde tracer method to chart cerebro-cerebellar anatomical projections in mice. Injections into a number of different neocortical regions (M1, S1, V1, A1, posterior parietal association cortex or the dorsal field of auditory cortex) revealed topographically organized projections to the pons and mossy fibre terminals in the cerebellar cortex. The report is generally well written and provides a valuable 'road map' for future studies of the functional significance of cerebro-cerebellar projections in mice.
Essential revisions:
1) The report needs to make clear what are the novel findings.
The main strength of our work is the description of the multiple disynaptic pathways, in all their complexity, in a single study directly comparable across brain regions. Here we go beyond a focus simply on traditional pontine projections and also detail the various precerebellar nuclei that receive both direct cortical input and project to the cerebellum, for corticocerebellar pathways originating from multiple sensory and motor processing brain regions.
Sorting out all the specific disynaptic and trisynaptic pathways observed in our findings and the functional implications is now perhaps a years-long endeavor – building on the vast amount of solid anatomical work that has come before us, with the ever advancing circuit tracing tools and functional markers now at our disposal. We have confirmed previous hypothesized results, removing ambiguity of terminal fields where functional synaptic connections were unclear, and also provided data from mice that had only previously been investigated in other mammalian species. We strongly believe that this study provides a critical starting point to further investigations and we detail a number of implications from our findings that should be pursued. While this strength of our paper is perhaps not a traditional ‘novel finding’ that can be easily pinpointed by bullet points, we have also uncovered novel findings along the way and have made efforts to highlight these in the revised manuscript, particularly throughout the discussion.
2) How did differences in cell counts and other measures of injection site affect the pattern and extent of anterogradely labelled cell and mossy fibre terminals? It was not clear to me if Figure 2A shows the total extent of injection sites into different neocortical areas or just the defined regions of interest. Either way the area clearly varies considerably between targets which will affect projection density and possibly topography.
We have provided further clarification of the injection site extent (revised Figure 3A; additional Supplementary file 2A), relationship between the injection site volume and mossy fiber terminals (additional Supplementary file 2B) as well as cell counts and mossy fiber terminals (Figure 6—figure supplement 3); including explanation throughout the text (e.g. subsection “Intermediate cerebrocerebellar brainstem pathways”). In short, we found that the number of all anterogradely labelled cells correlated most strongly with the number of mossy fiber terminals and measures of injections site volume were a poor predictor of general extent of mossy fiber labeling.
This may be due to a number of factors related to the fact that the trans-synaptic properties of the virus are heavily reliant on having a high titer concentration and, therefore, a high concentration of virus taken up by individual cells; the number of viral particles within each cell is of course difficult to assess and may be reliant on other properties such as the speed of injection, properties of diffusion within the cortical layers, and proximity to the pipette tip, for example – making the number of cells that are transfected to a level amenable to transsynaptic transport variable, even across injection sites with equal volumes. In fact, one may argue that, given the same volume of injected solution, a smaller injection site would indicate a higher concentration of viral particles and therefore predict a negative correlation between injection site and resulting trans-synaptic labeling.
In reality, there are a number of factors at play that may lead to the variability we see here. Of course, although the absolute number of resulting anterograde and terminal labelling is still of interest, this is one reason why for this study we have opted to inject in a limited number of pre-defined key cortical regions, and to normalize the mossy fiber labelling we have observed across regions to present the data in a way that the pattern of resulting labelling can be appreciated independently of the variability we see across injection sites. We have made considerable effort to make this clear in the revised manuscript with the new Figures noted above and further discussion (e.g. noting the need for further targeted exploration of neighboring somatotopically organized regions, subsection “Diversity of cerebrocerebellar pathways”).
3) The histogram axes in Figure 2D are not all drawn on the same scale making comparisons difficult. Also, percentage values are potentially misleading given the large variation in cell counts and mossy fibre terminals listed in supplementary Figure 1B.
We understand the reasoning of the reviewer, it would be ideal to have the same scaling in all graphs for the revised Figure 3D. However, the percentages of labeled mossy fiber terminals varied substantially across the individual cortical areas and we felt that putting them on the same absolute scale would detract from subtle, but still appreciable, differences observed in the pattern of labelling across various lobules for the less well studied cortical targets (i.e. beyond M1 and S1). To overcome this issue and to directly compare the magnitude of inputs of various modalities into the different lobules we choose the format depicted in the revised Figure 4A,B, as well as including the density measures in Figure 4C (and Supplementary file 1C for reference), which are more scalable reflections of the input given by each cortical target region. In this way, both the pattern of expression across all areas and the biases for cortical contribution to individual cerebellar regions can be appreciated. We have also made a note in the manuscript regarding the interpretation of data from regions with very low-density terminal labelling (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”).
4) Given that the pons is the primary source of mossy fibre inputs, how did the authors determine that other brainstem nuclei also containing labelled cells in the same experiments necessarily provided mossy fibre projections?
While further investigation is required in relation to individual pathways in this regard, we now provide different lines of evidence showing the potential for specific extra-pontine precerebellar nuclei to act as cerebrocerebellar intermediaries:
- Since no pontine labeled cells were observed after A1 injections, but mossy fiber terminals were still found in the cerebellum, one can conclude that at least for this disynaptic pathway, extra-pontine intermediate nuclei play a large role.
- We observed labelling in the inferior cerebellar peduncle as well as the middle cerebellar peduncle, which is now shown in revised Figure 2.
- Our observations from retrograde tracer injections into key cerebellar lobules show a spatial overlap between trans-synaptically labelled precerebellar cells and retrogradely labelled cells from cerebellar injections.
We have detailed this evidence throughout the Results section and discussed in detail in the revised Discussion section (e.g. subsection “Diversity of cerebrocerebellar pathways”; see also reviewer 1, point 1) and note that further investigations into the role of these precerebellar nuclei as relay versus integration centers for cortical signals will make for interesting future studies.
5) The topographical organization within the pons should be compared to the principles of organization defined in previous studies (eg Leergaard, 2003; Odeh et al., 2005).
We significantly expanded our discussion in this regard in the revised version of the manuscript in both the Results section and the Discussion section.
6) The discussion should consider the extent to which cerebro-cerebellar pathways are conserved across mammalian species. How useful is the mouse as a model for study of cerebro-cerebellar projections?
We have added comparison to other species in various applicable places throughout the discussion (e.g. subsection “Diversity of cerebrocerebellar pathways”) and included a statement regarding the role of the mouse as a model for cerebrocerebellar projections (subsection “Convergence of functionally distinct cerebrocerebellar input”).
Reviewer #3:
Henschke and Pakan have submitted a manuscript entitled "Convergence of multimodal inputs connecting the cerebral cortex to the cerebellum" in which they use modern neuronal tracing methods to map circuit projections into the mouse cerebellum. The authors traced the cerebellar inputs originating from sensory, motor, and association cortices. They argue that diverse functional connections project from the cortex to cerebellum, where they ultimately exhibit a convergent terminal field distribution within specific lobules. The manuscript is well written and technically, it offers a much-needed approach to better understand the organization and trajectories of major cortico-cerebellar projections. In addition, the figures are well made, schematics very helpful, and overall the panels are easy to follow. However, below we provide a number of comments and concerns that we hope will improve the clarity and impact of the study.
1) Throughout the text, the discussion of A1 and its unique projection through non-pontine pre-cerebellar nuclei is often highlighted, however the significance of this point is not well articulated. Further clarification in the abstract and in the body is needed if the authors wish to highlight the uniqueness of this pathway. This is especially true given the findings in the text that S1 and M1 both have substantial, if not even majority in the case of S1, projections to non-pontine pre-cerebellar nuclei.
We have significantly expanded our discussion in regard to the unique A1 pathways and added additional experiments to support our findings. See subsection “Anterograde tracing of indirect cerebrocerebellar pathways”, subsection “Diversity of cerebrocerebellar pathways” (see also reviewer 1 point 3 and point 4 below).
2) Introduction. Please clarify why the identified cortical regions were selected. In addition, while we agree that the characterization of the injections is rather comprehensive, there is a limitation in this study in that either the number of cortical regions chosen or the breadth of the characterization of the selected regions (ie: why weren't at least two neighboring sites in a given cortical region analyzed) prevent the present study from providing a fully comprehensive map, as stated by the authors.
We agree that the term ‘fully comprehensive map’ was misleading, as the goal of the study was to describe, in detail, only key sensory and motor cerebrocerebellar pathways – consequently, we have toned down these terms throughout. Indeed, two separate strategies could be to (1) briefly characterize many regions without much breadth, or (2) to fully characterize immediately neighboring regions with great detail; however, we opted for a strategy in between, so that we could devote enough space to fully explore the potential for multimodal overlap along with the implications that arise from our findings.
We strongly feel that this type of experimental design is advantageous as it allows us to compare within a single study a number of key functional regions without creating a dataset that acts as only a resource and must be fully interpreted by the reader alone. There is clearly a tradeoff between the number of cortical sites that could be investigated and the detail in which those specific pathways can be highlighted and discussed in depth. We believe that the necessity still exists for more detailed topographic mapping within single modalities, as well as large scale fully comprehensive mapping of pathways (as is provided, for instance, by resources such as the Allen brain Institute), but the strength of our study allows for both the presentation of a large dataset spanning key cortical regions as well as detailed quantification and interpretation of both the results, and identification of key avenues of future research in light of the findings. We believe that with the expansion of the manuscript into a full Research Article format we have achieved this.
3) Figure 1A. Please indicate the circumscribed subregion in each cortical area where the injection was made.
We have further specified the core and halo region of the injection site within the cortical subregions in the revised Figure 3.
4) Subsection “Anterograde tracing of indirect cerebrocerebellar pathways”. Please clarify percentages for A1 projections in order to make the second sentence "This pattern of extra-pontine labeling was similar…" clearer. Further, in Figure 2B, while the scarcity of the projections to the pontine nuclei is similar, they seem to be qualitatively different in that A1 projects terminals, while the AuD experiment labels cell bodies. This distinction is not clear in the text. Moreover, the fact you see terminals in one region and then cell bodies another makes it a challenge to piece together what the actual pathway(s) might look like. That is, how do you substantiate the claim that the cell bodies in the pons (the unexpected labeling) actually project to the cerebellum? And on the flip side, for the majority of S1 labeling, which is in the precerebellar non-pontine nuclei, can we really assume that these all project to cerebellum? If not, what percentage?
The sentence in question has been revised in conjunction with the expanded discussion of the projections from A1 throughout, specifically clarifying the difference between the terminal labelling we see in the pons after A1 injections and the cell bodies observed after AuD injections (with supporting evidence from retrograde labelling from the pontine nuclei [Figure 1—figure supplement 1] and mono-trans-synaptic anterograde labelling in the pons following IC injection [Figure 1—figure supplement 2]). See subsection “Anterograde tracing of indirect cerebrocerebellar pathways” and subsection “Diversity of cerebrocerebellar pathways“ (see also reviewer 1 point 3). We have also further quantified the percentage of extra-pontine versus pontine labeling observed after targeted retrograde injections (new Figure 6—figure supplement 1).
5) The supplementary file 1 is a bit difficult to interpret. Please include the percentages referenced in the text in the table. In addition, it is unclear why in some injection sites the sum of the individual pre-cerebellar nuclei cells equals the summed total, while in other cases the sum is less than or greater than the listed total. (ie M1-1 52 vs 55; V1-1 8 vs 8; A1-1 10 vs 8)
We have included the percentages in the revised Supplementary file 1B (total cell count and mean ± s.e.m. for the percentage of extra-pontine labeled cells for each target cortical region, as discussed in the text). We have also amended the sums, as these were typos based on a previous version of the Table.
6) The significance of terminals in pontine/pre-cerebellar nuclei without cell body labelling is unclear and there should be an explanation of this in the text. For instance, in regard to A1/AuD does this indicate inefficiency in trans-synaptic labelling, that there is an alternative monosynaptic projection from A1/AuD to the pontine nuclei, or is there an alternative explanation? This is mentioned for M1 projections to the olivary nuclei (Subsection “Anterograde tracing of indirect cerebrocerebellar pathways”).
We have performed additional experiments (e.g. IC injection) and expanded the discussion regarding these points in the text (in the subsection “Anterograde tracing of indirect cerebrocerebellar pathways” as well as the subsection “Diversity of cerebrocerebellar pathways”). We indicate the potential for polysynaptic pathways and discuss likely routes for cerebral information from A1 to get to the cerebellar cortex through higher-order polysynaptic pathways. Clearly, deciphering all trisynaptic pathways from all cortical target regions is beyond the scope of the current study, but here we provide not only a starting road-map for these targeted investigations, but also functional ties and implications in the discussion (e.g. in relation to A1/AuD, subsection “Diversity of cerebrocerebellar pathways”; in relation to olivary projections, subsection “Diversity of cerebrocerebellar pathways”).
7) In the subsection "Convergence of multimodal cerebrocerebellar input within lobules" the metric for dominant representation (> or < 55%) is problematic given the relative sparsity of the injections (either across or within cortical regions). Furthermore, given the variability of numbered terminals across individual animals in a given cortical area (ie from 2406 to 25616 from the M1 injections), this type of quantification and conclusions about convergence seem premature.
We agree with the reviewer and have replaced this figure with a more quantitative assessment of density (Figure 4C and Supplementary file 1C). This still demonstrates the point that motor and somatosensory cortical projections in general dominate throughout the cerebellar cortex, without setting arbitrary thresholds or presenting misleading summaries per lobule based on narrow cortical target regions. Additionally, we have included a note in the manuscript regarding the interpretation of data from regions with sparse terminal labelling (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”).
8) While the discussion about convergence and integration is quite interesting to speculate upon, it is hard to grasp whether the current data is sufficient to reach such a conclusion. Given the vast number of granule cells and the fine scale nature of cerebellar microzones, conclusions about integration of multimodal afferent information by single units (whether microzone or granule cell) cannot be easily drawn. To do this, a method distinct from the approach used in this paper would be needed whereby labelling from two distinct cortical areas could be distinguished in the cerebellum in a single animal. In the end, I am having trouble fully appreciating how integrated the "multimodal" and "convergence" nature of the pathways actually is, based on the data provided.
The concept of ‘convergence’ in this study has been clarified throughout to mean regional spatial overlap and the text has been revised throughout to limit speculation about integration and highlight the potential for modular multimodal processing (see also response to editor point 3). Additionally, we have expanded our discussion regarding convergence on the single cell level with more extensive reference to previously published literature (e.g. see subsection “Convergence of functionally distinct cerebrocerebellar input”).
9) Finally, there is a core assumption at play, which is that cortical injection – pontine/precerebellar cell body labeling – mossy fiber terminal marking. In the case of cortico-ponto-cerebellar pathways, the abundance of literature makes this assumption sound. However, in the case of the precerebellar projections (especially S1 which has a majority non-pontine projections), sagittal sections (for example) demonstrating the axonal projections, and elucidating the monosynaptic pathway would provide a helpful piece of evidence to support the current content.
We thank the reviewer for the suggestion and note that we were able to observe clear labelling in the inferior cerebellar peduncle (even in coronal sections), which has now been included in the revised manuscript (Figure 2; subsection “Anterograde tracing of indirect cerebrocerebellar pathways”). Please also see response to reviewer 2 point 4.
10) In the Abstract, it is not clear what is actually meant by "Prominent terminal overlap across motor and sensory modalities…". And, how was this measured?
We have altered this sentence in the Abstract to read: ‘Within molecularly defined cerebellar modules we found spatial overlap of mossy fiber terminals across motor and sensory modalities…’ and in combination with edits to the last sentence: ‘…the regional convergence of multimodal inputs.’ we have clarified this.
11) It would be very helpful to include the species studied in the Title and/or the Abstract.
The species has now been included in the Abstract.
12) It is critical that the authors fully outline the limitations of the mono-trans synaptic anterograde viral tracer, in their hands, and specifically for the purpose of this study.
This has been made more explicitly clear in the various places in the manuscript (e.g. subsection “Neuroanatomical injections”).
13) In Figure 1B, it is not clear what the red arrows are representing and pointing towards. Aren't the arrows pointing the wrong way if this is to show anterograde tracing direction?
Here, the red ‘arrows’ were indicative of labeled projection neurons (i.e. layer 5 pyramidal cell bodies). These were, however, quite small illustrations, and we have revised this figure for clarity and changed the symbols to circles (Figure 1B).
14) Is there a time dependence on the tracing? That is, do longer periods of survival after injection label additional projections?
Previous studies using this AAV as a mono-trans-synaptic tracer (Zingg et al., 2017) have systematically investigated the time dependence of this technique. They found no significant difference in the number of anterogradely labelled cells between 2, 3, or 4 weeks expression. Indeed, in pilot studies we also confirmed their results by testing different survival times after the injections, up to 4 weeks and, additionally, never observed labelled granule cell/Purkinje cells or other identifiable non-specific labeling – which would indicate transmission beyond mono-synaptically connected neurons. We did not test survival times for shorter than 2 weeks, but this is a fairly standardized time period for any AAV transfection. Hence, 2 weeks expression was used for all the data collected for the main study. We have added a note in the methods to this effect (subsection “Neuroanatomical injections”).
15) In Figure 1F, M1 should project ipsilateral, but in the cerebellum one can see many terminals on both sides. Please explain.
We are unsure as to the specific confusion, but we can clarify that after M1 injections, pontine labeling was mostly located on the ipsilateral side, with few cells also labelled on the contralateral side (Figure 1E), as has been previously suggested in rodents examining terminal labelling in the pons (Mihailoff et al., 1985; Leergaard and Bjaalie, 2007). This conclusion is also supported by studies in primates, by using anterograde tracer to confirm additional corticopontine fibers from M1 on the contralateral side with functional synapses (Morecraft et al., 2018). This is now shown here in mice and with trans-synaptic anterograde labelling without confusion regarding the presence of axon terminals or fibers of passage – a significant advantage of our study. This information has been further clarified in the revised text (see subsection “Anterograde tracing of indirect cerebrocerebellar pathways”). However, in the cerebellum, the majority of mossy fiber labeling after all injections into the cerebral cortices (including M1) was located on the contralateral site (e.g. M1: ~25% ipsilateral and 75% contralateral; Figure 3C). The prominent contralateral ponto-cerebellar pathways are well documented (e.g. Serapide et al., 2002) but recent studies examining the projections of single pontine cells in rodents have found both ipsilateral and contralateral collaterals, with some even crossing twice (Biswas et al., 2019; this is also documented in the revised manuscript, subsection “Organization of cerebrocerebellar terminals”). The image shown in Figure 1F is representative of the proportion of ipsilateral and contralateral labelling observed in our study (see also Figure 3C,D), as well as in agreement with previous literature.
16) The authors state "…extent M1, indicating the presence of a polysynaptic cerebro-olivary pathway…" What is the literature to support this?
We have expanded on this in subsection “Diversity of cerebrocerebellar pathways”.
17) The authors state "We quantified the number of resulting mossy fiber terminals throughout the extent of the cerebellum and found that this was highly variable across cortical regions…" How variable is the size of the infected cell population in the cortex?
It is difficult to determine the precise number of transfected cells within the injection site due to the trans-synaptic nature of the AAV and the density of labeling in the core region. This is why we have opted to report the volume of the injection site core (Supplementary file 1A and Supplementary file 2A). We additionally added a new supplementary file (Supplementary file 2) to illustrate the quantification of the injection site, and further characterize the relationship with the resulting mossy fiber labeling.
18) The authors state "There was a clear and consistent topography of…" It would be helpful to fully define what you mean by topography here. There are many levels of topography, in different brain regions, and this could mean very different things to different researchers.
We have rephrased this for clarity: ‘In agreement with previous studies (Leergaard and Bjaalie, 2007; for review see, Kratochwil et al., 2017), resulting pontine labeling after cortical injections was topographically organized with more rostral cortical regions projecting more medially in the pons and caudal cortical areas projecting towards the lateral extent of the pontine nuclei (Figure 1E, see also Figure 3B).’ (subsection “Anterograde tracing of indirect cerebrocerebellar pathways”).
19) In Figure 3E, what does each data point represent? A single terminal?
For the high density projections, M1 and S1, each dot represents 5-10 mossy fiber terminals, for all other lower density projections each dot represents 1-5 mossy fiber terminals. Because this is a representative schematic to show the modular pattern of labelling, we refer readers to Figure 3 and Figure 4 for quantification of results across cerebellar regions. We have clarified this in the figure legend (revised Figure 5).
20) Along the same lines as above, it would be helpful to see more actual data with the combined zebrin labeling. Mossy fiber patterns, by nature of their clustered appearance, are not so easily appreciated and therefore additional examples, at higher power magnification, would be welcomed.
We revised the current Figure 5 and have added the new Figure 5—figure supplement 1 to the revised manuscript, presenting more AldoC related data. We have also added additional text to the results, specifically detailing the pattern of distributions in relation to the molecular zones (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”).
21) The authors state "This allowed for a detailed comparison of cerebellar regions across brains and, hence, across functional cortical regions representing multiple modalities (see Figure 3—figure supplement 1)." However, I don't see a full description or explanation of the data in the text.
We have rephrased this to clarify the goal of using the aldolase c expression pattern as a tool to align cerebellar regional maps across target cortical injection sites (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”). Additionally, as noted above, we have added further description of the pattern of labelling in the text (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”). Our results were relatively surprising, in that we did not see clear patterns of sagittal organization on a larger scale (which we were anticipating). We believe this is partially due to the fact that we are labelling multiple cerebrocerebellar pathways, and finer scale projections may show more defined sagittal organization. This is now explicitly noted in the revised manuscript (subsection “Regional convergence of multimodal cerebrocerebellar input within lobules”).
22) The number of animals used in the study seems quite low.
We have added a substantial number of additional animals through various new experiments and have almost tripled the number of animals used in this study; we now use 42 animals. Although somewhat subjective for anatomical studies, we would argue that for our experimental design, descriptive statistics, and multiple target regions investigated, the animal numbers we use are in the normal range (for example, Zingg et al., 2017, n = 4 for main anatomical experiments examining a single pathway).
23) In Figure 3—figure supplement 1, what is the red staining in Purkinje cells?
We use the tdTomato Ai9 mice line, which tends to have a level of background fluorescence (see https://www.jax.org/strain/00790, and also e.g. Hahn et al., 2019), as reported by the donating investigator to the Jackson Laboratory: “Importantly, the donating investigator reports that very low levels of tdTomato expression may be present prior to introduction of Cre recombinase – but the tdTomato expression levels after Cre recombination are significantly greater than those baseline levels”.
In our hands, this background staining is very clearly distinguishable, particularly under the microscope, and qualitatively distinct from the AAV1.cre-induced tdTomato expression in cells and terminals. We have added an explanation to this effect in the revised manuscript (Materials and methods section, and revised Figure 5—figure supplement 1 legend).
[Editors’ note: what follows is the authors’ response to the second round of review.]
Essential revisions:
The major remaining concern – shared by both reviewers – is the overemphasis/ speculation on 'multimodality' for this essentially anatomical study. Specifically:
1) I was delighted to see a more detailed examination of the location of MFs from different modalities within each lobule, since a prerequisite for multimodal processing is that the information project to the same subregion of a lobule. I also found rather striking the case in Crus II where there was still segregation between two MF types within the same lobe. This latter result highlights the notion that multimodal innervation within lobes is necessary but not sufficient to conclude multimodal processing. I therefore encourage the authors to soften their choice of words of "highly multimodal" and "moderately multimodal" as a conclusion of Figure 4. Perhaps the use of the term "lobule co-innervation" is more appropriate. Then after Figure 5, the argument for putative multimodal processing is much stronger, but still only suggestive.
In reference to Figure 4, we have changed the terms here as suggested to utilize ‘coinnervated’ and we thank the reviewer for this suggestion (e.g. subsection “Lobule co-innervation of cerebrocerebellar inputs from distinct cortical areas”, Figure 4).
We have additionally further reworked our concept of multimodal ‘integration’ for this more anatomically relevant term throughout the manuscript where appropriate. Therefore, we have toned down the functional ‘multimodal’ emphasis and highlighted the strength of the work as an anatomical study (e.g. Title; Introduction).
2) Indeed, Figure 5 shows an essential set of analyses suggestive of multimodal processing, but I was left wanting for some clear quantification of "intermingled" and a simple overall summary.
a) Would it be possible to analyze the distribution of nearest-neighbor distances between terminals of different origins? Those that co-innervate would have shorter distances and those that do not, e.g. Crus II, would show a shifted distribution. Perhaps a more sophisticated point-process analysis, such as the application of the bivariate Ripley K-function, could be used to test for spatial randomness between two species of points and provide statistical confidence.
b) To add some statistical rigor to the alignment with adolase C staining, one could compare the frequency of AldoC positive and negative terminals for each cortical injection. Statistical tests for the frequency of events can then be used.
a) As suggested, we have analyzed the distribution of nearest neighbor distances (Figure 5B and Figure 6B) as well as the identity of the cortical origin of nearest neighbors (Figure 5C and Figure 6C) and included appropriate statistical tests for significance (see also exact pvalues in Supplementary file 1E). Consequently, we have split the previous Figure 5 into two figures due to space constraints (now Figure 5 and Figure 6). The results have been amended to reflect these changes and incorporate the main points from the quantification (e.g. subsection “Spatial organization of mossy fiber terminals in molecularly defined cerebellar modules”). We have additionally included analysis using the Ripley K function to test for spatial randomness and included the results in the text (subsection “Spatial organization of mossy fiber terminals in molecularly defined cerebellar modules”). Of course, the methods section has also been updated to reflect the new analysis (subsection “Quantification and data analysis”). We believe these additional analyses have indeed increased the statistical confidence in the qualitative pattern of data previously reported and thank the reviewer for their detailed suggestions.
b) We have added statistical rigor and internal validation of the AldoC+/- alignment across animals by correlating the frequency of mossy fiber terminals across + and – regions of each lobule. See new Figure 5—figure supplement 1J where we present the correlation matrix across all animals as well as the associated p-values; demonstrating the high correlation across animals with injections within the same cortical target region (referenced in the figure legend as well as subsection “Spatial organization of mossy fiber terminals in molecularly defined cerebellar modules” and subsection “Quantification and data analysis”).
3) The title continues to hint at a more functional set of analyses, which the paper really does not provide. Please revise.
We have revised the title accordingly: ‘Disynaptic cerebrocerebellar pathways originating from multiple functionally distinct cortical areas’. We think this title preserves the concept of functional specificity of cortical inputs, but that it is now also clear that the study is based on an anatomical premise (see also point 1).
4) Similarly, the Abstract could present the key anatomical distinctions that were uncovered rather than the speculative discussion about multimodal sensory input – this functional idea should remain, but it should be toned down to make space to better clarify the actual anatomical findings.
We have rewritten the Abstract to downplay the multimodal integration further and to highlight the anatomical strengths. We agree that it is now much more focused.
5) The main Figure 6 merits a more thorough quantification of retrograde labeling.
We have added further quantification to (now) Figure 7, including a quantitative assessment of the origin of nearest neighbors for anterogradely labelled cells (Figure 7C) as well as the distance between anterograde and retrograde labelled cells from dual injections (Figure 7—figure supplement 2; subsection “Intermediate cerebrocerebellar brainstem pathways”). We also refer to supplementary file 1B which includes a table of the quantification of retrograde labelling and Figure 7—figure supplement 1 and Figure 7—figure supplement 3 as qualitative as well as quantitative representation of this data.
6) Quantification and presentation of co-labeled and spatially intermingled neurons following dual viral tracing injections would be appropriate.
We have added this information in conjunction to the points listed above (point 5), including further quantification and description regarding co-labeled and spatially intermingled neurons (see also Figure 7 figure legend).
7) I like the estimation of the number of MFs per pontine neuron and would like to see it as a main figure.
We have including this into the updated main Figure 7 (see Figure 7D).
8) The Discussion section rambles on a bit. I believe it can be shortened and used to more clearly summarize the major findings, the similarities and differences with previous results, and the implications for cerebellar function.
We have edited the Discussion section throughout to be more concise and clearer (reducing the length by ~500 words), highlighting the major findings while still maintaining a thorough discussion of the nuances of this extensive dataset.
9) The section titles are rather vague. For example, "Organization of cerebrocerebellar mossy fiber terminals" is essentially duplicated in the Results and Discussion, and perhaps needs more specific wording. "Spatial organization of cerebrocerebellar mossy fiber types". I am not really sure why there is a Discussion subsection on "Diversity of cerebrocerebellar pathways" as you define the pathways by viral injection. Perhaps something like: "Cerebellar cortical target diversity of cerebrocerebellar pathways?"
We have amended and refined most section titles. In the Results section, we now move from the more general trans-synaptic anterograde tracing technique and general pathway results, to the regional organization of terminals in the cerebellum, to the co-innervation of inputs at the level of the lobule, to the precise spatial organization of inputs on the level of molecularly defined modules within each lobule, and finally to the summary examining in detail the entire cerebrocerebellar pathway including the intermediate brainstem nuclei.
Likewise, the Discussion section titles have been amended to explore the stages of the cerebrocerebellar pathways, first cortex to precerebellar nuclei (subsection “Pontine and extra-pontine cerebrocerebellar pathways”), then the terminal organization in the cerebellum (subsection “Organization of cerebrocerebellar input to the cerebellum”), and finally the functional implications of the study (subsection “Functional implications”). We believe the general flow is now improved.
https://doi.org/10.7554/eLife.59148.sa2Article and author information
Author details
Funding
European Regional Development Fund (ZS/2016/04/78113)
- Julia U Henschke
- Janelle MP Pakan
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We are grateful to Doug Wylie and Eike Budinger for helpful comments on an earlier version of the manuscript as well as Cathleen Knape for expert technical assistance. This work was funded by the European Regional Development Fund (ERDF: Center for Behavioral Brain Sciences ZS/2016/04/78113).
Ethics
Animal experimentation: All experiments were performed according to the NIH Guide for the Care and Use of Laboratory animals (2011) and the Directive of the European Communities Parliament and Council on the protection of animals used for scientific purposes (2010/63/EU) and were approved by the animal care committee of Sachsen-Anhalt, Germany (42502-2-1479 DZNE).
Senior Editor
- Richard B Ivry, University of California, Berkeley, United States
Reviewing Editor
- Megan R Carey, Champalimaud Foundation, Portugal
Reviewer
- Roy V Sillitoe, Baylor College of Medicine, United States
Version history
- Received: May 20, 2020
- Accepted: July 28, 2020
- Version of Record published: August 14, 2020 (version 1)
Copyright
© 2020, Henschke and Pakan
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 2,430
- Page views
-
- 325
- Downloads
-
- 25
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Chromosomes and Gene Expression
- Neuroscience
Mathys et al. conducted the first single-nucleus RNA-seq (snRNA-seq) study of Alzheimer’s disease (AD) (Mathys et al., 2019). With bulk RNA-seq, changes in gene expression across cell types can be lost, potentially masking the differentially expressed genes (DEGs) across different cell types. Through the use of single-cell techniques, the authors benefitted from increased resolution with the potential to uncover cell type-specific DEGs in AD for the first time. However, there were limitations in both their data processing and quality control and their differential expression analysis. Here, we correct these issues and use best-practice approaches to snRNA-seq differential expression, resulting in 549 times fewer DEGs at a false discovery rate of 0.05. Thus, this study highlights the impact of quality control and differential analysis methods on the discovery of disease-associated genes and aims to refocus the AD research field away from spuriously identified genes.
-
- Neuroscience
The strength of a fear memory significantly influences whether it drives adaptive or maladaptive behavior in the future. Yet, how mild and strong fear memories differ in underlying biology is not well understood. We hypothesized that this distinction may not be exclusively the result of changes within specific brain regions, but rather the outcome of collective changes in connectivity across multiple regions within the neural network. To test this, rats were fear conditioned in protocols of varying intensities to generate mild or strong memories. Neuronal activation driven by recall was measured using c-fos immunohistochemistry in 12 brain regions implicated in fear learning and memory. The interregional coordinated brain activity was computed and graph-based functional networks were generated to compare how mild and strong fear memories differ at the systems level. Our results show that mild fear recall is supported by a well-connected brain network with small-world properties in which the amygdala is well-positioned to be modulated by other regions. In contrast, this connectivity is disrupted in strong fear memories and the amygdala is isolated from other regions. These findings indicate that the neural systems underlying mild and strong fear memories differ, with implications for understanding and treating disorders of fear dysregulation.