1. Neuroscience
Download icon

Diverse inhibitory projections from the cerebellar interposed nucleus

  1. Elena N Judd
  2. Samantha M Lewis
  3. Abigail L Person  Is a corresponding author
  1. Department of Physiology and Biophysics, University of Colorado School of Medicine, Anschutz Medical Campus, United States
Research Article
  • Cited 0
  • Views 490
  • Annotations
Cite this article as: eLife 2021;10:e66231 doi: 10.7554/eLife.66231

Abstract

The cerebellum consists of parallel circuit modules that contribute to diverse behaviors, spanning motor to cognitive. Recent work employing cell-type-specific tracing has identified circumscribed output channels of the cerebellar nuclei (CbN) that could confer tight functional specificity. These studies have largely focused on excitatory projections of the CbN, however, leaving open the question of whether inhibitory neurons also constitute multiple output modules. We mapped output and input patterns to intersectionally restricted cell types of the interposed and adjacent interstitial nuclei in mice. In contrast to the widespread assumption of primarily excitatory outputs and restricted inferior olive-targeting inhibitory output, we found that inhibitory neurons from this region ramified widely within the brainstem, targeting both motor- and sensory-related nuclei, distinct from excitatory output targets. Despite differences in output targeting, monosynaptic rabies tracing revealed largely shared afferents to both cell classes. We discuss the potential novel functional roles for inhibitory outputs in the context of cerebellar theory.

Introduction

The cerebellum plays a critical role in refining motor control through learning. The cerebellar nuclei (CbN), which constitute the major outputs of the cerebellum, are proposed to relay predictive computations of the cerebellar cortex and store well-learned patterns, placing them in a central position to implement cerebellar control (Eccles and Szentágothai, 1967Ohyama et al., 2003; Chan-Palay, 1977). The CbN are a collection of nuclei that house diverse neuronal subtypes that differ in their targets. Recent studies have greatly expanded our understanding of this diversity, using approaches such as genomic profiling and projection specific tracing (Bagnall et al., 2009; Low et al., 2018; Fujita et al., 2020; Kebschull et al., 2020; Uusisaari et al., 2007; Uusisaari and Knöpfel, 2010; Uusisaari and Knöpfel, 2011; Husson et al., 2014; Ankri et al., 2015; Canto et al., 2016). Through these studies, we know that multiple diverse output channels intermingle (Fujita et al., 2020; Low et al., 2018; Sathyamurthy et al., 2020), widespread collateralization is common, and genetic diversity of excitatory projection neurons varies systematically along the mediolateral extent of the CbN which encompasses the medial (fastigial), interposed, lateral (dentate), interstitial, and vestibular nuclei (Kebschull et al., 2020).

The mouse cerebellar interposed nucleus has received recent attention at the anatomical and functional levels with studies identifying specific projection patterns and functional roles for neuronal subtypes within the structure. Interposed excitatory neurons project to a variety of motor-related spinal cord and brainstem targets, as well as collateralize to motor thalamus (Low et al., 2018; Sathyamurthy et al., 2020; Kebschull et al., 2020). Ablation of a subset of anterior interposed (IntA) glutamatergic cells that express Urocortin3, for example, disrupts accurate limb positioning and timing during a reach to grasp task and locomotion (Low et al., 2018). Chemogenetic silencing of excitatory neurons that project ipsilaterally to the cervical spinal cord also impaired reach success in mice (Sathyamurthy et al., 2020). Moreover, closed-loop manipulation of IntA disrupts reach endpoint in real time (Becker and Person, 2019). The interposed nucleus also mediates conditioned eyelid responses, sculpts reach and gait kinematics, and is responsive to tactile stimulation (Darmohray et al., 2019; Ten Brinke et al., 2017; Rowland and Jaeger, 2005). How anatomical organization of the structure confers such functions is an open question.

Functional consequences of cell-type-specific manipulations have not been limited to excitatory neurons. Ablation of inhibitory nucleo-olivary cells demarcated with Sox14 expression also resulted in motor coordination deficits (Prekop et al., 2018). These cells were traced from the lateral nucleus and suggested to project solely to the inferior olive (IO), consistent with conclusions from experiments using dual labeling methods (Ruigrok and Teune, 2014). Nevertheless, older reports of inhibitory projections from the CbN that target regions other than the IO raise the question of whether inhibitory outputs might also play a role in regulating brainstem nuclei outside the olivocerebellar system. Combined immunostaining with horseradish peroxidase tracing from the basilar pontine nuclei (i.e., pontine grey [PG]) in rats and cats showed GABA immunopositive neurons in the lateral nucleus (Aas and Brodal, 1989; Border et al., 1986), although the literature is inconsistent (Schwarz and Schmitz, 1997). Glycinergic output projections from the medial nucleus (fastigial) inhibitory output population include large glycinergic neurons that project to ipsilateral brainstem targets outside the IO (Bagnall et al., 2009), unlike its Gad2-expressing neurons which exclusively target the IO (Fujita et al., 2020). In aggregate, these various observations indicate that better understanding of whether the interposed nucleus houses inhibitory output neurons that project to targets outside IO is needed.

Here, we use a range of viral tracing methods to isolate and map projections from and to inhibitory and excitatory neurons of the intermediate cerebellar nuclear groups, defined through intersectional labeling methods using single or multiple recombinases coupled with pathway-specific labeling (Fenno et al., 2014). This method permitted analysis of collateralization more specific than traditional dual-retrograde labeling strategies since it leverages genetic specification and projection specificity and permits entire axonal fields to be traced. We elucidate the projection ‘fingerprints’ of genetic- and projection-defined cell groups. Surprisingly, we observed widespread inhibitory outputs, comprised at least in part of putative collaterals of some IO-projecting neurons, that target both ipsilateral and contralateral brainstem and midbrain structures. Monosynaptic rabies transsynaptic tracing (Kim et al., 2016; Wickersham et al., 2010) restricted to excitatory premotor neuron populations through the selective expression of Cre recombinase under the Slc17a6 (Vglut2) promoter (Gong et al., 2007) and inhibitory neurons through Cre expression controlled under the Slc32a1 (Vgat) promoter revealed reproducible patterns of presynaptic inputs largely shared across cell types. Taken together, these experiments provide new insight into input/output organization of the intermediate cerebellum, suggest potential functional diversity of parallel channels, and provide anatomical targets for functional studies aimed at evaluating these putative roles.

Results

Anterograde tracing of Int-Vgat neurons

To determine projection patterns of inhibitory neurons of the interposed nucleus, we stereotaxically injected AAV2-EF1a-DIO-YFP into Vgat-Cre transgenic mice, ‘Int-Vgat,’ (N=5, Figure 1A). We mapped and scored the extent and density of terminal varicosities on a 4-point scale and recorded injection sites, plotted for all experiments (Figure 1—figure supplement 1; see Materials and methods, Projection quantification).

Figure 1 with 5 supplements see all
Anterograde tracing of Int-Vgat neurons.

(A) Schematic of injection scheme. (B) Example injection site of AAV2-EF1a-DIO-eYFP. The three main CbN are outlined in white (lateral nucleus (LN), interposed (IN), and medial nucleus (MN) from left to right). Images oriented so that the dorsal-ventral axis runs up/down and the medial/lateral axis runs right/left; right of midline is contralateral. (C) Location of labeled cells by injection into Int of Vgat-Cre mice. Specimens are color-coded by the proportion of cells labeled in anterior interposed (IntA) where the highest proportion corresponds to darkest color. (D) Mapping of terminal fields based on restriction of injection site to IntA. The highest unilateral relative projection strength (RPS) in each region is plotted for all specimens included in analysis. The values are always assigned as integers but are offset here so overlap can be better appreciated. (E) YFP-positive terminals (green) in inferior olive (IO), spinal trigeminal nuclei (SPVc), pontine grey (PG), and red nucleus (RN) are stained for antibodies against Gad65/67 (top) and Vglut2 (bottom; magenta). Dashed circles indicate colocalized terminals while solid lines indicate a lack of colocalization observed in the two channels. Scale bars=20 µm. (F) Example cells from in situ hybridization showing clear overlap with an mRNA probe against Slc32a1 (Vgat) and no overlap with an mRNA probe against Slc17a6 (Vglut2). Scale bars=10 µm. (G) Projection targets in caudal cerebellum and brainstem (B-7.45). Boxes expanded in (i–iii) (top) or (i–iii’) (bottom). (H) Projection targets within the intermediate cerebellum (B-6.35). Injection site depicted in (C). (I) Projection targets within rostral brainstem (B-4.95). (J) Projection targets in the caudal midbrain (B-3.93). (K) Projection targets to the rostral midbrain (B-3.93). Scale bars (C, G–K) =1 mm and (i–iii) 200 µm. The inset (black border) depicts the location of coronal sections shown in (G–K) along a parasagittal axis. Cuneate nucleus (CU), gigantocellular reticular nucleus (GRN), hypoglossal nucleus (XII), intermediate reticular nucleus (IRN), interstitial cell groups (icgs), lateral reticular nucleus (LRN), lateral vestibular nucleus (LAV), midbrain reticular nucleus (MRN), motor nucleus of the trigeminal (V), nucleus prepositus (PRP), Nucleus Y (Y), oculomotor nucleus (III), parabrachial (PB), paraflocculus (PFl), paragigantocellualr reticular nucleus (PGRN), periaqueductal grey (PAG), principle sensory nucleus of the trigeminal (PSV), pontine reticular nucleus (PRN), posterior interposed (IntP), spinal trigeminal nucleus, caudal/interpolar subdivision (SPVc/i), spinal vestibular nucleus (SPIV), superior vestibular nucleus (SUV), and tegmental reticular nucleus (TRN).

As expected, injections labeled neurons that densely innervated the contralateral dorsal accessory IO (Figure 1D, E and G), with less dense but consistent innervation of ipsilateral IO (Ruigrok and Voogd, 1990; Balaban and Beryozkin, 1994; Fredette and Mugnaini, 1991; Prekop et al., 2018Ruigrok and Voogd, 1990; Ruigrok and Voogd, 2000 Van der Want et al., 1989). Surprisingly these injections also consistently labeled terminal fields outside IO, within the brainstem, even when injection sites were completely restricted to the anterior interposed nucleus. Viral expression of Int-Vgat neurons labeled axonal varicosities which were immunopositive for antibodies against Gad65/67, but never Vglut2, consistent with a GABAergic phenotype for these projections (Figure 1E, Figure 1—figure supplement 2; analyzed in the IO, spinal trigeminal nucleus, interpolar (SPVi), PG, red nucleus (RN), and vestibular nuclei). In situ hybridization (ISH) revealed that 98% of virally labeled cells co-expressed the Vgat gene Slc32a1 (230/234 cells from two mice), while 4/234 cells overlapped the Vglut2 gene Slc17a6 (Figure 1F, Figure 1—figure supplement 3, Supplementary file 2). A Gad1-Cre driver line (Higo et al., 2009) was tested but not used owing to non-specific label (Figure 1—figure supplement 4; see Materials and methods;Supplementary file 2).

Most Int-Vgat injections included both interposed and interstitial cell groups slightly ventral to the interposed nucleus, plotted in Figure 1B, color-coded for the proportion of the injection site contained within IntA. Although injection site spillover into interstitial cell groups (Sugihara and Shinoda, 2007) was common, injection site spillover into the main vestibular groups ventral to the fourth ventricle was minimal to absent. Following these injections, terminal label within the brainstem was extensive, and invariably also included beaded varicosities within the cerebellar cortex characteristic of the inhibitory nucleocortical pathway (Ankri et al., 2015). Modestly dense but spatially extensive terminal fields ramified in the posterior medulla along the anterior-posterior axis (Figures 1 and 2D); no retrogradely labeled neurons were observed. Among sensory brainstem structures, terminal fields ramified within the ipsilateral external cuneate nucleus (ECU), cuneate nucleus (CU), nucleus of the solitary tract (NTS), SPVi, especially the lateral edge, parabrachial nuclei (PB), principal sensory nuclei of the trigeminal (PSV), and all vestibular nuclei. Int-Vgat axons extended through the pontine reticular nuclei (PRN) to innervate the tegmental reticular nuclei (TRN; commonly abbreviated NRTP) and the PG (i.e., basilar pontine nuclei; Figure 1D and J), which are themselves major sources of cerebellar mossy fibers. Int-Vgat neurons also innervated the medial magnocellular RN (Figure 1K) bilaterally. Rarely, Int-Vgat axons progressed to the caudal diencephalon, very sparsely targeting the ipsilateral zona incerta ZI in 2/6 mice (Supplementary file 3). Axonal varicosities were vanishingly sparse or non-existent within the spinal cord following Int-Vgat injections (data not shown).

Figure 2 with 1 supplement see all
Intersectional labeling of IO-projecting Int-Vgat neurons (IntIO-Vgat) and comparison with Int-Vgat.

(A) Schematic of experiment. (B) Example injection site of AAV8.hSyn.Con/Fon.hChR2.EYFP in a Vgat-Cre mouse. The three main CbN are outlined in white (lateral nucleus (LN), interposed (IN), and medial nucleus (MN) from left to right). Images oriented as in Figure 1. Scale bars=1 mm. (C) Location of labeled cells by injection of Retro-Flp to the contralateral inferior olive (IO) and Con/Fon-YFP into Int of Vgat-Cre mice. Specimens are color-coded by the proportion of cells labeled in anterior interposed (IntA) where the highest proportion corresponds to darkest color. (D) Graphical representation of average projection strength in all targeted regions for IntIO-Vgat (n=5; maroon) and Int-Vgat (n=6; white) mice. See the list of abbreviations for complete listing. (E) Mapping of terminal fields based on restriction of injection site to IntA. The highest unilateral RPS in each region is plotted for all specimens included in analysis. The values are always assigned as integers but are offset here so overlap can be better appreciated. (F) Example terminal fields within the (IO) and red nucleus (RN) bilaterally, lateral reticular nucleus (LRN), pontine grey (PG), tegmental reticular nucleus (TRN), and spinal vestibular nucleus (SPIV). Scale bars=200 µm. External cuneate nucleus (ECU), hypoglossal nucleus (XII), interstitial cell groups (icgs), lateral vestibular nucleus (LAV), magnocellular reticular nucleus (MARN), Nucleus Y (Y), oculomotor nucleus (III), parabrachial (PB), periaqueductal grey (PAG), pontine reticular nucleus (PRN), posterior interposed (IntP), spinal trigeminal nuclei, interpolar (SPVi), spinal vestibular nucleus (SPIV), superior vestibular nucleus (SUV), and zona incerta (ZI).

Beaded nucleocortical fibers from Int-Vgat injections were reliably labeled if the injection site included interstitial cell groups (Figures 1G, H, 2D; Ankri et al., 2015). Int-Vgat neurons targeted all cerebellar lobules, even extending contralaterally.

Some targets noted were sensitive to injection site restriction (Figure 1D). However, labeling of varicosities outside IO was not attributable solely to injection site leakage outside Int. The smallest Int-Vgat injection, contained entirely within IntA, labeled fine caliber axons that ramified within the ipsilateral superior and spinal vestibular nuclei (Figure 1—figure supplement 5). Labeled fibers coursed in the superior cerebellar peduncle, decussating at the level of the pontine nuclei (–4 mm Bregma). As they coursed ventrally, they produced numerous varicosities in the pontine nuclei, specifically the tegmental reticular nucleus and PG, before turning caudally, labeling dense terminals fields in the contralateral IO (DAO) and modestly dense fields in the ipsilateral IO. Very sparse varicosities were also noted in the parabrachial nucleus and magnocellular RN. Despite the presence of these terminal fields, no nucleocortical fibers were seen following the most restricted Int-Vgat injection, suggesting these may originate from interstitial cell groups and/or other CbN. To summarize, Int-Vgat injections labeled fibers that innervated numerous brainstem nuclei outside IO, even following highly restricted injections.

Projection-specific Int-Vgat neuron tracing

The terminals observed in the brainstem and midbrain from Int-Vgat labeling suggested the existence of inhibitory channels from the intermediate cerebellum beyond those targeting the IO.

Next, to restrict label to genetic- and projection-specific Int neurons (Fenno et al., 2014), we used a two-recombinase-dependent reporter virus (AAV8-hsyn-Con/Fon-eYFP) injected into Int in conjunction with Flp recombinase retrogradely introduced via the contralateral IO with AAVretro-EF1a-Flp (Figure 2A; N=5). The fluorescent reporter will only express in the presence of both Cre and Flp recombinases. This Cre-on Flp-on approach, termed ‘Con/Fon,’ was used to isolate IO-projecting Int-Vgat neurons. Specificity was determined via injections in wild-type C57/Bl6 mice (N=2) and off-target injections in Cre mice (N=3), which did not yield YFP positive neurons in the CbN (Figure 2—figure supplement 1).

IntIO-Vgat neurons had more restricted terminations than most direct Int-Vgat injections. Varicosities were consistently observed in dorsal PG, PRN, TRN, IO, and the vestibular complex. Less consistent and sparser label occurred in other brainstem nuclei (Figure 2). These data are consistent with either of two non-mutually exclusive possibilities: that at least some IO-projecting cells collateralize to a subset of targets relative to the constellation of regions targeted by all Int-Vgat neurons, typically excluding nucleocortical projections, modulatory/affective regions, and sensory nuclei; and/or these intersectional methods restrict the range of neurons infected owing to IO targeting, which restricts the other axonal fields labeled. We also note that this result does not preclude the existence of IO-only projecting neurons, which was not directly examined.

Anterograde tracing from excitatory output neurons

To compare Int-Vgat projections more directly to excitatory outputs, we injected Int of Ntsr1-Cre mice with AAV1-CAG-flex-GFP (N=2) or AAV2-DIO-EF1a-eYFP (N=3) (Figure 3). Int-Ntsr1 terminal varicosities consistently colocalized with Vglut2 immunolabel, but never Vgat, consistent with a glutamatergic phenotype of Ntsr1 output neurons (Figure 3E, Figure 3—figure supplement 1), and somata overlapped predominantly with the glutamatergic marker Slc17a6 (Figure 3F; Figure 1—figure supplement 3). Dense and consistent terminal varicosities labeled by Int-Ntsr1 neurons occurred in patches within the caudal medulla, midbrain, and thalamus, which are known targets of Vglut2-Cre and Ucn3-Cre neurons (Figure 3D and G–K, Low et al., 2018; Kebschull et al., 2020; Sathyamurthy et al., 2020). Varicosities filled the ipsilateral parvicellular reticular nucleus PARN (commonly abbreviated PCRt) which extended rostrally to blend into the spinal nucleus of the trigeminal (SPV), known forelimb control structures (Esposito et al., 2014), and ipsilateral terminals ramified in the motor nucleus of the trigeminal (V). Bilateral patches of terminals were seen in the lateral reticular nucleus (LRN) and all four subdivisions of the vestibular nuclei. At the level of the decussation of the superior cerebellar peduncle, axons turned ventrally and produced dense Vglut2-positive varicosities in the TRN (commonly abbreviated NRTP) and sparsely in PG (Cicirata et al., 2005; Schwarz and Schmitz, 1997). Axons also ramified within the magnocellular RN and the deep layers of the superior colliculus (SC). Diencephalic projections were densely targeted to thalamic nuclei and more sparsely targeted to ZI. All specimens exhibited dense terminal fields in the ventromedial (VM) and anterior ventrolateral (VAL) nuclei of the thalamus (Teune et al., 2000; Aumann et al., 1994; Houck and Person, 2015; Kalil, 1981; Low et al., 2018; Stanton, 1980). Additionally, we observed terminals in intralaminar thalamic structures including centromedial (CM), paracentral (PCN), mediodorsal (MD), parafascicular (PF), ventral posterior (VP), and posterior (PO) nuclei (Teune et al., 2000; Chen et al., 2014; Dumas et al., 2019). Int-Ntsr1 neurons formed nucleocortical mossy fibers in multiple lobules across the cortex (Figure 3G–I; Gao et al., 2016; Houck and Person, 2015; Tolbert et al., 1978; Low et al., 2018; Sathyamurthy et al., 2020).

Figure 3 with 2 supplements see all
Anterograde tracing of Int-Ntsr1 neurons.

(A) Schematic representation of injection scheme. (B) Example injection site of AAV2-EF1a-DIO-eYFP. The three main CbN are outlined in white (Lateral nucleus (LN), interposed (IN), and medial nucleus (MN) from left to right). Images oriented as in Figure 1. (C) Distribution of labeled cells by injection into Int of Ntsr1-Cre mice. Specimens are color-coded by the proportion of cells labeled in anterior interposed (IntA) where the highest proportion corresponds to darkest color. (D) Mapping of terminal fields based on restriction of injection site to IntA. The highest unilateral RPS in each region is plotted for all specimens included in analysis. The values are always assigned as integers but are offset here so overlap can be better appreciated. (E) YFP-positive terminals (green) in parvicellular reticular nucleus (PARN), red nucleus (RN), and ventral anterior-lateral complex of the thalamus (VAL) are stained for antibodies against Gad65/67 (top; magenta) and Vglut2 (bottom; magenta). Dashed circles indicate colocalized terminals while solid lines indicate a lack of colocalization observed in the two channels. Scale bars=20 µm. (F) Example cells from in situ hybridization showing overlap with both an mRNA probe against Slc32a1 (Vgat) and Slc17a6 (Vglut2). Scale bars=10 µm. (G) Projection targets in caudal cerebellum and brainstem (B-7.05). Boxes expanded in (i–iv). (H) Projection targets within the intermediate cerebellum (B-6.35). Injection site depicted in (C). (I) Projection targets within and ventral to the anterior cerebellum (B-5.65). (J) Projection targets to pontine nuclei (B-4.25). (K) Projection targets in the rostral midbrain (B-3.38). Note the dense terminals in RN. (L) Projection targets to the caudal thalamus (B-1.65). (M) Projection targets to the rostral thalamus (B-1.35). Scale bars (C, G–M) =1 mm and (i–iv) 200 µm. The inset (black border) depicts the location of coronal sections shown in (G–M) along a parasagittal axis. Centromedial nucleus of the thalamus (CM), cuneate nucleus (CU), gigantocellular reticular nucleus (GRN), inferior olive (IO), intermediate reticular nucleus (IRN), interstitial cell groups (icgs), lateral reticular nucleus (LRN), lateral vestibular nucleus (LAV), mediodorsal nucleus of the thalamus (MD), medullary reticular nucleus, dorsal/ventral subdivision (MDRNd/v), midbrain reticular nucleus (MRN), nucleus raphe magnus (RM), nucleus X (X), nucleus Y (Y), parabrachial (PB), paracentral nucleus of the thalamus (PCN), parafascicular nucleus (PF), periaqueductal grey (PAG), pontine reticular nucleus (PRN), posterior interposed (IntP), simplex lobule (Sim), superior colliculus (SC), superior vestibular nucleus (SUV), supratrigeminal nucleus (SUT), spinal cord (SpC), tegmental reticular nucleus (TRN), ventromedial nucleus (VM), and zona incerta (ZI).

Beyond the major targets described above, Int-Ntsr1 projected sparsely to a variety of other regions. In three out of five animals, we observed a small patch of terminals within the contralateral dorsal subnucleus of IO that were positive for Vglut2 (Figure 3G, Figure 3—figure supplement 1). Near the dense terminal field within the contralateral RNm, fine caliber axons bearing varicosities spilled over into the ventral tegmental area, VTA (Figure 3—figure supplement 2; Carta et al., 2019; Teune et al., 2000) and extended dorsally through the contralateral midbrain/mesencephalic reticular nucleus (MRN; Ferreira-Pinto et al., 2021) to innervate the caudal anterior pretectal nucleus (APN) anterior ventrolateral periaqueductal grey (PAG) (Vaaga et al., 2020; Sugimoto et al., 1982; Gayer and Faull, 1988; Low et al., 2018; Teune et al., 2000). To summarize, Int-Ntsr1 neurons targeted regions well known to receive excitatory input from the interposed nucleus, as well as a previously unappreciated sparse Vglut2+ afferent to the IO.

Projection-specific Int-Ntsr1 neuron tracing

We next used the Con/Fon intersectional approach described above to restrict labeling to RN-projecting Ntsr1-Cre neurons (IntRN-Ntsr1, Figure 4; N=4), asking whether projection-specific labeling recapitulated data from direct label of Int-Ntsr1 cells, as would be expected if RN projecting neurons collateralize to other targets. The projection pattern of IntRN-Ntsr1 was almost identical to the pattern observed in Int-Ntsr1 injections, with a few notable exceptions. Namely, only Int-Ntsr1 neurons projected to lobule 8, anterior pretectal nucleus (APN), IO, and pedunculopontine nuclei (PPN). Terminal fields in the contralateral thalamus, especially VAL, VM, and CM/ PCN as well as layers 7/8 of the contralateral cervical spinal cord (2/3 specimens with spinal cords available) support the observation in Sathyamurthy et al., 2020 that contralaterally projecting cerebellospinal neurons collateralize to both RN and thalamus. We conclude that it is likely that Int-Ntsr1 neurons reliably project to RN and collateralize to a restricted collection of other targets, although these data do not distinguish between broad versus restricted collateralization of IntRN-Ntsr1 neurons.

Figure 4 with 2 supplements see all
Intersectional labeling of RN-projecting Int-Ntsr1 neurons (IntRN-Ntsr1).

(A) Schematic of experiment. (B) Example injection site of AAV8.hSyn.Con/Fon.hChR2.EYFP in an Ntsr1-Cre mouse. The three main CbN are outlined in white (lateral nucleus (LN), interposed (IN), and medial nucleus (MN) from left to right). Images oriented so right of midline is contralateral. Scale bars=1 mm. (C) Location of labeled cells by injection of Retro-Flp to the contralateral red nucleus (RN) and Con/Fon-YFP into Int of Ntsr1-Cre mice. Specimens are color-coded by the proportion of cells labeled in anterior interposed (IntA) where the highest proportion corresponds to darkest color. (D) Graphical representation of average projection strength in all targeted regions for IntRN-Ntsr1 (n=4; navy) and Int-Ntsr1 (n=6; white) mice. See the list of abbreviations for complete listing. (E) Mapping of terminal fields based on restriction of injection site to IntA. The highest unilateral RPS in each region is plotted for all specimens included in analysis. The values are always assigned as integers but are offset here so overlap can be better appreciated. (F) Example terminal fields within the red nucleus (RN), spinal cord (SpC), parvicellular reticular nucleus (PARN), tegmental reticular nucleus (TRN), superior colliculus (SC), and ventral anterior-lateral complex of the thalamus (VAL). Scale bars=200 µm. Centromedial nucleus of the thalamus (CM), gigantocellular reticular nucleus (GRN), inferior olive (IO), intermediate reticular nucleus (IRN), interstitial cell groups (icgs), lateral reticular nucleus (LRN), lateral vestibular nucleus (LAV), mediodorsal nucleus of the thalamus (MD), midbrain reticular nucleus (MRN), motor nucleus of the trigeminal (V), nucleus raphe magnus (RM), nucleus X (X), nucleus Y (Y), parabrachial (PB), paracentral nucleus of the thalamus (PCN), parafascicular nucleus (PF), periaqueductal grey (PAG), pontine grey (PG), pontine reticular nucleus (PRN), posterior complex of the thalamus (PO), posterior interposed (IntP), superior vestibular nucleus (SUV), supratrigeminal nucleus (SUT), tegmental reticular nucleus (TRN), ventral tegmental area (VTA), ventromedial nucleus (VM), ventral posterolateral nucleus of the thalamus (VPL), and zona incerta (ZI).

Projections of IntARN neurons traced with AAVretro-Cre

As described above, we noted that both Int-Vgat and Int-Ntsr1 labeled varicosities within RN. This presented a target that we could exploit to test whether Int neurons collateralize to both RN and IO independent of genetic Cre label. We retrogradely expressed Cre in RN-projecting neurons, injecting AAV2retro-Cre into RN and a flexed reporter virus into Int (AAV1-CAG-flex-GFP/ RFP) of wild-type C57/Bl6 mice (Figure 4—figure supplement 1; N=4). Following these injections, we observed label in both IO and RN contralateral to the Int injection (Figure 4—figure supplement 1; Supplementary files 1 and 3). We also observed terminals in other locations consistently targeted by either Int-Ntsr1 (MRN, VAL, VPM, VM, PF, MD, PO, SC, and ZI) or Int-Vgat (Lob 9, IO, lateral SPV, ipsilateral PRN, and ECU). Following these injections, terminal varicosities in IO, and subsets in TRN and PG expressed Gad65/67 while varicosities in SPV, RN, PG, VAL, and TRN were positive for Vglut2 (Figure 4—figure supplement 2; N=2). We conclude that retrograde uptake of Cre from synaptic terminals in RN results in reporter expression of both glutamatergic and GABAergic neurons in Int that both project to RN.

Comparison of projection patterns across labeling methods

Across the distinct labeling methods, we observed a variety of notable patterns that differentiated them. First, Int cell sizes differed by Cre driver lines. We measured the cross-sectional area and elliptical diameter of somata of virally labeled cells. Int-Vgat neurons tended to be small with tortured dendrites (Figure 5A–C; 14.4±0.5 µm diameter, 95% confidence interval [CI]=[12.99, 15.85]; 109.3±7.8 µm 2 area, 95% CI=[87.54, 131.1]; N=5 mice; n=316 neurons). By contrast, Int-Ntsr1 neurons were characteristically large with smooth dendrites (Figure 5A–C; 22.2±0.8 µm diameter, 95% CI=[20.03, 24.37]; 224.7± 13.6 µm 2 area, 95% CI=[186.9, 262.5]; N=5 mice; n=229 neurons). Similarly, IntIO-Vgat neurons were small (Figure 5B–C; 14.5±0.2 µm diameter, 95% CI=[13.92, 15.07]; 103.4±2.2 µm2 area, 95% CI=[97.42, 109.5]; N=5 mice; n=404 neurons) and IntRN-Ntsr1 neurons larger (Figure 5B–C; 22.1±1.0 µm diameter, 95% CI=[19.02, 25.27]; 238.2±18.5 µm 2 area, 95% CI=[179.4, 297.1]; N=4 mice; n=125 neurons). We compared these groups statistically and found that the Int-Vgat and IntIO-Vgat cells were not significantly different from one another but were significantly smaller than Int-Ntsr1 and IntRN-Ntsr1 neurons (one-way analysis of variance [ANOVA]; Tukey’s multiple comparison test; F (3,15)=46.99, p<0.0001 for all cross-genotype comparisons of means across specimens; p>0.99 for all within genotype comparisons of means across specimens).

Comparison Int-Vgat and Int-Ntsr1 cell sizes and projection patterns.

(A) Example YFP+ cells in a Vgat-Cre (top) and Ntsr1-Cre (bottom) specimen. Scale bars=50 µm. (B) Differences in soma diameter of neurons based on isolation method. Grand mean ± SEM is plotted; per animal mean is denoted with colored circles (Int-Vgat=red, IntIO-Vgat=maroon, Ntsr1=blue, and IntRN-Ntsr1=navy). Int-Vgat (n=316 cells, 5 mice) or IntIO-Vgat neurons (n=404 cells, 5 mice) are smaller than Int-Ntsr1(n=229 cells, 5 mice) or IntRN-Ntsr1 neurons (n=125 cells, 4 mice; one-way ANOVA; Tukey’s multiple comparison’s test, p<0.0001****). (Note that Int includes all subdivisions of the interposed nucleus and icgs.) (C) Cumulative frequency distribution of measured cell diameter for all specimens. (D) The average proportion of the total (summed) RPS value that is derived from projections to motor, sensory, or modulatory extracerebellar brain regions. Mean and SEM are plotted. Welch’s t-test with FDR correction of 1%, p=0.035 (ns), 0.0014 (**), 0.00023 (***), 0.045 (ns), respectively. (E) Same as (D) but showing the contribution of ipsilateral or contralateral projections to total RPS per transgenic line. Welch’s t-test with FDR correction of 1%, p=0.017 (ns) and 0.16 (ns), respectively. (F) Schematic of projection signatures from Ntsr1-Cre (blue) and Vgat-Cre (red). (G) Axons from Int-Vgat and Int-Ntsr1 follow unique paths through the pontine reticular nuclei (PRN). (H) Morphology differences in terminal contacts within the cerebellar cortex (top; boutons observed within the granule cell (GrC) layer; dotted white line in Nstr1 image denotes Purkinje Cell layer) and red nucleus (RN; bottom). Note mossy fiber nucleocortical terminals seen in Ntrs1-Cre mice but not Vgat-Cre mice. Scale bars=50 µm. Centromedial nucleus of the thalamus (CM), copula (Cop), Crus1 (Cr1) cuneate nucleus (CU), external cuneate nucleus (ECU), flocculus (Fl), gigantocellular reticular nucleus (GRN), hypoglossal nucleus (XII), inferior olive (IO), intermediate reticular nucleus (IRN), lateral reticular nucleus (LRN), lateral vestibular nucleus (LAV), medullary reticular nucleus, dorsal/ventral subdivision (MDRNd/v), midbrain reticular nucleus (MRN), Nucleus Y (Y), nucleus prepositus (PRP), oculomotor nucleus (III), parabrachial (PB), paracentral nucleus of the thalamus (PCN) parafasicular nucleus of the thalamus (PF), paraflocculus (PFl), paragigantocellular reticular nucleus, dorsal (PGRNd), paramedian lobule (PM), parvicellular reticular nucleus (PARN), periaqueductal grey (PAG), pontine grey (PG), pontine reticular nucleus (PRN), principal sensory nucleus of the trigeminal (PSV), simplex lobule (Sim), spinal trigeminal nucleus, caudal/interpolar subdivision (SPVc/i), spinal vestibular nucleus (SPIV), superior colliculus (SC), superior vestibular nucleus (SUV), supratrigeminal nucleus (SUT), superior vestibular nucleus (SUV), tegmental reticular nucleus (TRN), trigeminal motor nucleus (V), ventral tegmental area (VTA), ventromedial nucleus (VM), vestibular nuclei (VEST), and zona incerta (ZI).

Figure 5—source data 1

Raw data for Figure 5B,C containing the size (in pixels and microns) of thresholded YFP labeled soma in Int per specimen, data is segregated by transgenic mouse line.

https://cdn.elifesciences.org/articles/66231/elife-66231-fig5-data1-v2.xlsx

Second, we noted that many targets were distinct between genotypes and projection classes. We classified extracerebellar target regions as motor, sensory, and modulatory, based in part on groupings of the Allen Brain Atlas (see Materials and methods). Notably, aggregate projection strength analyses indicated that on average, Int-Vgat neurons targeted sensory structures more densely than Int-Ntsr1 neurons (Figure 5D; t(7.6)=4.9, p=0.001, unpaired Welch’s t-test). By contrast, we observed significantly stronger innervation of modulatory regions by Int-Ntsr1 than Int-Vgat (Figure 5D; p=0.0002, t(5.4) = 8.6, unpaired Welch’s t-test). Additionally, Int-Ntsr1 projections showed a contralateral bias and Int-Vgat an ipsilateral bias, but these trends were not significant when accounting for false positive discovery rates (Figure 5E and F; t(6.9)=3.1, p=0.02, unpaired Welch’s t-test).

Third, qualitative assessment showed that axons tended to ramify in distinct subdivisions within the subset of targets shared by Int-Vgat and Int-Ntsr1. For example, Int-Vgat neurons projected to more lateral regions of the caudal spinal nucleus of the trigeminal (SPVc), and to more lateral and anterior divisions of the principle sensory nucleus of the trigeminal (PSV). Int-Ntsr1 projected to the medial edge of SPVc near the border with MDRNd/ PARN and to PSV near the border of the trigeminal (V). While both Int-Vgat and Int-Ntsr1 projected to the vestibular nuclei, Int-Vgat projections ramified more caudally in the spinal and medial nuclei than Int-Ntsr1. Int-Vgat projections to the SC were absent. We also noted striking distinctions in the midbrain, where fibers from the two genotypes coursed in distinct locations. After decussating, Int-Vgat axons coursed farther lateral before turning ventrally toward the pontine nuclei. By contrast, Int-Ntsr1 axons turned ventrally at more medial levels after decussation, near the medial tracts through the pontine reticular nucleus (Figure 5G). Because injection sites did not differ qualitatively across injection types, we interpret these distinctions to reflect targeting differences across cell classes.

Finally, as has been noted in previous studies, nucleocortical fiber morphology differs between excitatory and inhibitory neurons (Ankri et al., 2015; Houck and Person, 2015; Gao et al., 2016; Batini et al., 1992). Int-Vgat injections labeled beaded varicosities devoid of mossy fiber morphological specializations (Figure 5H, top panels). Int-Ntsr1 labeled terminals with typical mossy fiber endings, large excrescences with fine filopodial extensions, and these predominantly targeted more intermediate lobules. Additionally, terminals in RN from Int-Vgat were very fine caliber while those from Int-Ntsr1 had thicker axons (Figure 5H, bottom panels). While these observations are qualitative in nature, they align with the small cellular morphology of Int-Vgat neurons relative to Int-Ntsr1 neurons.

Cell-type-specific input tracing using monosynaptic rabies virus

Having mapped pathways from diverse cell types of the intermediate CbN, we next investigated afferents to these cells (Figure 6A). As described above, ISH in Vgat-Cre mice validated Vgat somatic expression in YFP labeled cells within the interposed nucleus, thus these mice were used for input tracing to inhibitory neurons (n=3). However, although output tracing from Int-Ntsr1 was validated with immunolabeling of terminals varicosities against Vglut2, ISH analysis of Ntsr1-Cre revealed 89% of YFP-positive cell bodies expressed Slc17a6 (Vglut2) probes (Figure 3F, Figure 1—figure supplement 3; 119/132 cells in two mice) but some labeled cells, possibly interneurons, expressed Slc32a1 (Vgat) (15/132 cells in two mice). Thus, to ensure input mapping specific to excitatory neurons, we tested mRNA probe specificity of Vglut2-Cre mice (Figure 6B, Figure 1—figure supplement 3, Supplementary file 2): 178/179 YFP expressing cells (99%) expressed Slc17a6 (Vglut2) mRNA and 3/179 expressed Slc32a1 (Vgat) probes. Therefore, Vglut2-Cre (n=3) mice were used to isolate inputs to glutamatergic Int populations. These mice were used in conjunction with modified rabies (EnvA-∆G-Rabies-GFP/mCherry) and Cre-dependent receptor and transcomplementation helper viruses (Figure 6A; see Materials and methods; Kim et al., 2016; Wall et al., 2010; Watabe-Uchida et al., 2012; Wickersham et al., 2007; Wickersham et al., 2010). Direct rabies virus infection was limited to cells that expressed the receptor, TVA; transsynaptic jump was restricted by complementation of optimized rabies glycoprotein (oG). In a subset of experiments, oG was restricted to TVA-expressing neurons (Liu et al., 2017). 72.9±9.6% of starter cells in Vglut2-Cre specimens were mapped to Int (Figure 6C and D); with the remaining starter cells located in the lateral (15%), medial (2%), and superior vestibular nucleus (5%). Similarly, 80.8±4.7% starter cells in Vgat-Cre mice were in Int, with the remainder in superior vestibular nuclei (7%), lateral (5%), medial (5%), and parabrachial (1%) nuclei (Figure 6D, Supplementary file 4). Total numbers of starter cell estimates (Doykos et al., 2020), defined by presence of rabies and TVA (Figure 6C) averaged 156±131 in Vglut2-Cre and 307±132 neurons in Vgat-Cre. TVA expression was not observed in cortex of Vgat-Cre or Vglut2-Cre mice, minimizing concerns of tracing contaminated by projections to cortical neurons.

Figure 6 with 1 supplement see all
Monosynpatic tracing of inputs to the interposed nucleus.

(A) Schematic of viral experiment. Cells labeled by this method provide monosynaptic input to Int. (B) Example Slc17a6-Cre driven YFP cell following in situ hybridization showing overlap with an mRNA probe against Slc17a6 (Vglut2) and no overlap with an mRNA probe against Slc32a1 (Vgat). Scale bars=10 µm. (C) Example starter cells from both transgenic mouse lines in IntA. Scale bars=1 mm. Insets to the right show rabies (magenta, top), TVA (green channel, top), and overlay (bottom). Scale bars=50 µm. (D) Locations of putative starter cells largely overlap for both cell types (mean + SEM). Note starter label in both IntA and IntP (IN). (E) Location of retrogradely labeled ipsilateral PCs by lobule. (F) Example extracerebellar rabies positive cells in motor (spinal vestibular nuclei, SPIV), sensory (parabrachial, PB), modulatory (raphe magnus, RM), and mixed (zona incerta, ZI) brain regions for both mouse lines. (G) Proportion of non-PC inputs to Vglut2-Cre or Vgat-Cre starter cells separated by modality. Simplex lobule (sim), Crus1 (Cr1), Crus 2 (Cr2), Copula (cop), paramedian lobule (PM), Paraflocculus (PFl), and Flocculus (FL).

The CbN receive a massive projection from Purkinje cells (PCs). The location of retrogradely labeled PCs was similar between specimens (Figure 6E), regardless of genotype. PCs in ipsilateral Lobules 4/5, Crus 1, and Simplex were most densely labeled following rabies starting in both cell types. No contralateral PC label was observed in any specimen.

Extracerebellar input to Vglut2-Cre and Vgat-Cre cells was diverse and wide-ranging (Figure 6—figure supplement 1). Both cell types receive input from brain regions related to motor, sensory, or modulatory functions (Figure 6F), corroborating previous observations with traditional tracers (Fu et al., 2011; but see Barmack, 2003). For a complete list of brain regions that provide input to Vglut2-Cre and Vgat-Cre Int neurons, see Supplementary file 4 and Figure 6—figure supplement 1. Vglut2-Cre cells received a majority of inputs from ipsilateral sources, but not by large margins, with 64% of inputs originating in ipsilateral regions (95% CI=[39, 90]; 36% contralateral, 95% CI=[10, 61]). For Vgat-Cre cells, 54% (95% CI=[50, 60]) of non-PC label was from ipsilateral sources (46% contralateral, 95% CI=[40, 50])). These differences were not significant (t(2.2)=1.6, p=0.2; unpaired Welch’s t-test). No extracerebellar region accounted for more than 10% of the total cells, suggesting widespread integration within Int. Of note, significantly more LRN neurons were retrogradely labeled following Vgat-Cre injections (5.9% of non-PC rabies labeled cells (95% CI=[3, 9]; >300 cells/specimen)) than Vglut2-Cre (0.4% (95% CI=[–1.2, 1.9] of total non-PC rabies label cells (<40 cells/specimen; t(3.1)=7.6, p=0.004, unpaired Welch’s t-test)), suggesting a more extensive input to Int-Vgat neurons from LRN. Aside from this difference, extracerebellar projections to both cell types came from medial vestibular nuclei, TRN, and other reticular formation nuclei. We observed retrograde label in the contralateral medial cerebellar nucleus from both Vgat and Vglut2-Cre mice.

Many canonical sources of mossy fibers, such as ECU, PRN, TRN, PG, and LRN (Parenti et al., 1996), were identified as sources of nuclear input as well as recipients of a projection from at least one cell type within Int (Figure 7A–B; Tsukahara et al., 1983; Murakami et al., 1981). Figure 7B summarizes the inputs and outputs of both cell types ranked by average proportion of total rabies labeled cells within a given region (for retrograde tracing, excluding Purkinje cells) or average projection strength (for anterograde tracing), excluding the weakest and least consistent projections, but including those that may originate from interstitial cell groups. The only brain regions which received a projection but were not also retrogradely labeled were the thalamic nuclei. In converse, the only brain regions with retrogradely labeled cells but not anterograde projections were motor cortex, somatosensory cortex, subthalamic nucleus, and lateral hypothalamus, among other minor inputs (Supplementary file 4).

Reciprocal loops between Int and extracerebellar targets, for both Vglut2 and Vgat expressing cells.

(A) Images depicting rabies labeled cells (columns 1 and 3, rabies + cells circled if singular or very small) and projections that included axon varicosities to the same regions at the same coordinates relative to bregma (columns 2 and 4). Medial vestibular nuclei (MV), tegmental reticular nucleus (TRN), lateral reticular nucleus (LRN), pontine reticular nuclei (PRN), spinal trigeminal nucleus, interpolar subdivision (SPVi), superior colliculus (SC), red nucleus (RN), principal sensory nucleus of the trigeminal (PSV), nd motor nucleus of the trigeminal (V). White dotted line denotes boundary between PSV and V. (B) Inputs and outputs listed in order of percent of non-PC rabies labeled cells (left) and relative projection strength (right). Only inputs with greater than 1% of the total extracerebellar rabies labeled cells and regions with mean relative projection strengths greater than 1 are listed. Asterisks denote regions that constituted a major afferent (>1% of the total input) and received a major projection (an RPS>1 in Slc32a1-Cre mice and >1.2 in Ntsr1-Cre mice). Anterior pretectal nucleus (APN), centromedial nucleus of the thalamus (CM), cuneate nucleus (CU), external cuneate nucleus (ECU), gigantocellular reticular nucleus (GRN), hypoglossal nucleus (XII), inferior olive (IO), intermediate reticular nucleus (IRN), interposed nucleus (IN), interstitial cell groups (icgs), lateral nucleus (LN), lateral reticular nucleus (LRN), lateral vestibular nucleus (LAV), magnocellular reticular nucleus (MARN), medial nucleus (MN), medial vestibular nuclei (MV), mediodorsal nucleus of the thalamus (MD), medullary reticular nucleus, dorsal/ventral subdivision (MDRNd/v), midbrain reticular nucleus (MRN), motor nucleus of the trigeminal (V), nucleus of the solitary tract (NTS), nucleus raphe magnus (RM), nucleus X (X), Nucleus Y (Y), nucleus prepositus (PRP), oculomotor nucleus (III), parabrachial (PB), paracentral nucleus of the thalamus (PCN), parafasicular nucleus of the thalamus (PF), paragigantocellular reticular nucleus, dorsal (PGRNd), paramedian lobule (PM), parvicellular reticular nucleus (PARN), periaqueductal grey (PAG), pontine grey (PG), pontine reticular nucleus (PRN), principal sensory nucleus of the trigeminal (PSV), red nucleus (RN), spinal trigeminal nucleus, interpolar/oral subdivision (SPVi/o), spinal vestibular nucleus (SPIV), superior colliculus (SC), superior vestibular nucleus (SUV), supratrigeminal nucleus (SUT), superior vestibular nucleus (SUV), tegmental reticular nucleus (TRN), trigeminal motor nucleus (Y), ventral anterior-lateral complex of the thalamus (VAL), ventral posteromedial nucleus of the thalamus (VPM), ventral tegmental area (VTA), ventromedial nucleus (VM), and zona incerta (ZI).

Discussion

Here, we systematically examined the input and output patterns of diverse cell populations of the interposed cerebellar nucleus, Int, using intersectional viral tracing techniques. Consistent with previous work, we found that the putative excitatory output neurons of Int collateralize to regions of the contralateral brainstem, spinal cord, and thalamus and more restrictedly to the caudal ipsilateral brainstem, including to regions recently shown to control forelimb musculature. However, we also found that Int GABAergic projection neurons innervate brainstem regions other than IO, including the pontine nuclei, medullary reticular nuclei, and sensory brainstem structures. Interestingly, at least some IO-projecting neurons collateralize to comprise a subset of these projections. Inputs to these distinct cell types were also mapped using monosynaptic rabies tracing. We found that inputs to glutamatergic and Vgat neurons of the intermediate cerebellum are largely similar with only the LRN standing out as preferentially targeting Vgat neurons. Merging anterograde and retrograde datasets, region-level reciprocal loops between Int and brainstem targets were similar across both cell types.

The most surprising results were the diverse projections of GABAergic neurons of Int. To address concerns that these projections may be the result of a methodological artifact, we note a variety of data that support our interpretation. First, ISH and immunolabel support the view that Vgat-Cre is restricted to Slc32a1 expressing neurons that express Gad65/67 in terminal boutons. Second, projection patterns of excitatory neurons were distinct, particularly within the ipsilateral caudal brainstem and diencephalon, thus non-specific viral label cannot account for the data. Third, AAV-retroCre injections into RN—a putative target of both Int-Vgat and Int-Ntsr—labeled targets matching mixed projections of excitatory and inhibitory neurons, including terminal label in IO. Finally, we used an intersectional approach, targeting Vgat-Cre expressing neurons that project to the IO. This method of isolating Int inhibitory neurons also consistently labeled terminals elsewhere in the brainstem. Taken together, leak of Cre cannot explain the sum of these observations.

Another study restricting tracer to lateral (dentate) nucleus Sox14-Cre expressing neurons, a transcription factor marking nucleo-olivary neurons, showed terminal label in the IO as well as the oculomotor nucleus (III). Based on retrograde tracing from III, terminals there were interpreted to reflect virus uptake by nucleus Y near the injection site (Prekop et al., 2018). This finding raised the question of whether brainstem and midbrain targets of Int-Vgat neurons described in the present study are merely a consequence of viral uptake in regions neighboring the interposed nucleus. Although projections were more extensive following larger injections in Vgat-Cre mice, we observed axon varicosities outside IO following injections that were completely restricted to the interposed nucleus. As has been noted in previous studies, the ventral border of the interposed nucleus is poorly distinguished but houses numerous islets of cells within the white matter tracts (Sugihara and Shinoda, 2007; Sugihara, 2011). These regions receive Purkinje input from zebrin negative zones, and have been proposed to be distinct subregions of the CbN. A medial population, named the interstitial cell group, resides ventrally between the medial and interposed nuclei. An anterior extension, the anterior interstitial cell group, resides ventral to the interposed nucleus, and more posterior and laterally, the parvocellular interposed and lateral cell groups, neighboring nucleus Y, complete this constellation of loosely organized cell groups. Our data, which include injections of these regions, hint that these areas may house Vgat neurons that produce more extensive extra-IO projections, including nucleocortical beaded axons distinct from nucleocortical mossy fibers, although this conjecture will remain speculative until methodological advances permit cell-type-specific tracing from such minute regions to be carried out.

While these inhibitory projections were unknown, these data, combined with previous literature from the medial nucleus, suggest that inhibitory projections from the CbN may be a more prominent circuit motif than is widely appreciated. The medial nucleus contains glycinergic projection neurons that innervate ipsilateral brainstem nuclei matching contralateral targets of excitatory neurons (Bagnall et al., 2009). Additional evidence of inhibitory outputs includes dual retrograde tracing suggesting that nucleo-olivary projections from medial nucleus and the vestibular complex collateralize to the VM hypothalamic nucleus (Diagne et al., 2001; Li et al., 2017). Studies combining retrograde horse radish peroxidase tracing from the basilar pontine nuclei (i.e., PG) with immunohistochemistry observed double-labeled GABA immunopositive neurons in the lateral nucleus of rats and cats (Aas and Brodal, 1989), although the literature is inconsistent (Schwarz and Schmitz, 1997). More recent work in mice tracing Vgat-Cre neurons of the lateral nucleus listed projections to a variety of brainstem structures as well as IO (Locke et al., 2018), but these results were not discussed.

Despite these corroborating experimental results, we note that our data may appear to contradict conclusions drawn from a dual-retrograde tracing study, in which only minor dual retrograde label was observed in the lateral and interposed nuclei following tracer injections into IO and RN or IO and TRN (Ruigrok and Teune, 2014). This study concluded that two distinct populations exist within the CbN: one which projects widely to several regions and one which projects exclusively to IO. However, this study did report a small number of cells colabeled by retrograde injections to IO and TRN as well as IO and RN. This observation may account for the present finding that a population of neurons that projects to both IO and premotor nuclei exists in smaller numbers, and that topographic specificity may have precluded previous methods from fully detecting the collateralization of inhibitory populations. Importantly, our results focus on all Int-Vgat neurons, and thus may label subsets of neurons that project exclusively to IO or exclusively outside of IO, which is not resolvable with dual retrograde methods.

Projection patterns of glycinergic medial and vestibular nucleus neurons have an ipsilateral bias relative to excitatory contralateral projections. (Bagnall et al., 2009; Prekop et al., 2018; Sekirnjak et al., 2003; Shin et al., 2011). This organizational structure has been proposed to potentially mediate axial muscular opponency. While there was also a trend for an ipsilateral targeting of Int-Vgat neurons, this bias was not significant when accounting for false discovery rates (p=0.02), with both excitatory and inhibitory cells projecting bilaterally. Future studies investigating the functional roles of these projections may explore agonist/antagonist opponency in motor targets of these projections, which remain lateralized for limb musculature. Additionally, the widespread observation of Purkinje neurons that increase rates during cerebellar dependent behaviors may suggest the potential for a double disinhibitory pathway through the CbN, if these Purkinje neurons converged on inhibitory nuclear output neurons (De Zeeuw and Berrebi, 1995; Zeeuw and Chris, 2020).

What might be the role of inhibitory projections from the CbN? Two intriguing patterns emerged that are suggestive of potential function. First, inhibitory projections targeted more sensory brainstem structures than excitatory outputs. Predicting sensory consequences of self-generated movement, termed forward models, is a leading hypothesis for the role of cerebellum in sensorimotor behaviors. While populations of Purkinje neurons may perform this computation, it is unknown how forward models are used by downstream targets. Inhibitory projections from cerebellum to sensory areas would seem to be ideally situated to modulate the sensory gain of predicted sensory consequences of movement (Brooks et al., 2015; Shadmehr, 2020). Moreover, negative sensory prediction error could be used to actively cancel predicted sensory reafference (Kim et al., 2020; Requarth and Sawtell, 2014; Shadmehr, 2020; Conner et al., 2021), raising implications for a combined role of negative sensory prediction error in guiding learning both through modulation of climbing fiber signaling in IO and through modulation of sensory signals reaching the cerebellum upon which associative learning is built. Second, GABAergic projections to the pontine nuclei, which are themselves a major source of mossy fiber inputs to the cerebellum, suggest a regulatory feedback pathway that could operate as a homeostat akin to the feedback loops through the IO (Medina et al., 2002). The pontine nuclei are a major relay of cortical information into the cerebellum. Thus, through inhibitory feedback, cortical information could potentially be gated to facilitate strategic (i.e., cortical) control for novel skill learning or turned down to facilitate automatic (i.e., ascending/non-cortical) control of movements (Schwarz and Thier, 1999).

The present study compliments a recent collection of papers examining cerebellar nuclear cell types. Transcriptomics analyses of the CbN identified three distinct excitatory cell types within IntA. These classes included two broad projection types: those that target a wide array of brainstem nuclei and those that target the ZI (Kebschull et al., 2020). Another recent study identified two distinct interposed cell types based on projection patterns to the spinal cord, which were shown to constitute a minority of neurons (<20%). Nevertheless, these spinal-projecting neurons collateralized to many other targets, including the MDRNv, RN, and the VAL (Sathyamurthy et al., 2020). Inhibitory projections were not examined in these studies, thus it will be interesting to examine how the inhibitory projection neurons identified in the present study map onto transcript clusters of the inhibitory cell types, five total across the CbN. At a minimum, these clusters would include IO-projecting neurons, interneurons, MN glycinergic projection neurons, and a collateralizing population of inhibitory neurons identified here (Ankri et al., 2015; Bagnall et al., 2009; Fujita et al., 2020; Husson et al., 2014; Kebschull et al., 2020; Sathyamurthy et al., 2020).

Inputs to these neuronal populations were largely similar, though we observed minor differences in the input signatures of Int-Vglut2 and Int-Vgat. Many more neurons in the LRN were labeled following Vgat-Cre starting cells for monosynaptic rabies tracing, suggesting a predominant innervation of inhibitory neurons by LRN. It remains unclear if there are differences in input connectivity between Vgat subgroups, specifically interneurons and projection neurons, or whether Gad65/67 expressing neurons co-express GlyT2. In comparing input and outputs to diverse cell types, we noticed that reciprocal loops were common, broadening themes of reciprocal loops demonstrated previously (Tsukahara et al., 1983; Beitzel et al., 2017; Murakami et al., 1981), to also include inhibitory neurons. Such loops resemble neural integrators used in gaze maintenance or postural limb stabilization (Albert et al., 2020; Cannon and Robinson, 1987), another potential functional role of the anatomy presented here. Interestingly, we observed a few neocortical inputs to the intermediate/interstitial groups of the CbN. We speculate that these regions may conform to the reciprocal loop motif, albeit polysynaptically, predicting that thalamic targets innervate neocortical areas that project back to the CbN.

In conclusion, the anatomical observations presented here open the door to many potential functional studies that could explore the roles of inhibitory projections in real-time motor control, sensory prediction and cancellation, and dynamic cerebellar gain control. Taken together, the present results suggest distinct computational modules within the interposed CbN based on cell types and shared, but likely distinct, participation in motor execution.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Mus musculus)C57BL/6JCharles RiverStock
Genetic reagent (M. musculus)Gad1-CreGift from Dr. Diego Restrepo, recv’d frozen embryos from Tamamaki groupRRID: IMSR CARD:2065PMID:19915725
Genetic reagent (M. musculus)Ntsr1-CreMutant Mouse Regional Resource CenterRRID: MMRRC_030648-UCD
Stock, Tg(Ntsr1-cre) GN220
Gsat/Mmucd
PMID:17855595
Genetic reagent (M. musculus)Vgat-ires-cre knock-in (C57BL/6J)Jackson LabsRRID: IMSR_JAX:028862Stock, #028862PMID:21745644
Genetic reagent (M. musculus)Vglut2-ires-cre knock-in (C57BL/6J)Jackson LabsRRID: IMSR_JAX:028862Stock, #028863PMID:21745644
Recombinant DNA reagentAAV1.CAG.flex.GFP (virus)AddgeneRRID: Addgene_51502Lot #: V41177Titer: 2.0×1013 (GFP)1.2×1013 (RFP)
Recombinant DNA reagentAAV1.CAG.flex.RFP (virus)AddgeneRRID: Addgene_28306Lot #: V5282
Recombinant DNA reagentrAAV2.EF1a.DIO.
eYFP.WPRE.pA (virus)
UNCRRID: Addgene_27056Lot #: AV4842FTiter: 4.5×1012
Recombinant DNA reagentAAV8.hysn.ConFon.
eYFP (virus)
AddgeneRRID: Addgene_55650Lot #: V15284PMID:24908100Titer: 2.97×1013
Recombinant DNA reagentAAVretro.EF1a.
FlpO (virus)
AddgeneRRID: Addgene_55637Lot # V56725PMID:24908100 Titer: 1.02×1013
Recombinant DNA reagentAAV2.retro.hSyn.
NLS.GFP.Cre (virus)
Viral preparations were a gift of Dr. Jason AotoRRID: Addgene_175381PMID:23827676
Recombinant DNA reagentAAV9.EF1a.FLEX.
H2B.GFP.2A.oG (virus)
Salk InstituteRRID: Addgene_74289Titer: 2.41×1012
Recombinant DNA reagentAAV1.EF1a.FLEX.
TVA.mCherry (virus)
UNCRRID: Addgene_38044PMID:22681690
Recombinant DNA reagentAAV1.syn.FLEX.split
TVA.EGFP.tTA (virus)
AddgeneRRID: Addgene_100798PMID:28847002
Recombinant DNA reagentAAV1.TREtight.mTag
BFP2.B19G (virus)
AddgeneRRID: Addgene_100799PMID:28847002
Recombinant DNA reagentEnvA.Gdeleted.
EGFP (virus)
Salk InstituteRRID: Addgene_32635
Recombinant DNA reagentEnvA.Gdeleted.
mCherry (virus)
Salk InstituteRRID: Addgene_32636
AntibodyAnti-Vglut2
(Rabbit monoclonal)
AbcamRRID: AB_2893024Cat #: a FP1487001KT b216463Lot #: GR3249111-2(1:250)
AntibodyAnti-Gad65/67
(Rabbit polyclonal)
Sigma-AldrichRRID: AB_2893025Cat#: ABN904Lot#: 3384833(1:200)
AntibodyAnti-Tyrosine Hydroxylase
(Sheep polyclonal)
MilliporeSigmaRRID: AB_90755Cat#: AB1542(1:200)
AntibodyAnti-GFP-Alexa Fluor 488
conjugate (Rabbit polyclonal)
InvitrogenRRID: AB_221477Cat#: A21311Lot #: 2017366(1:400)
AntibodyAnti-Rabbit DyL594
(Goat polyclonal)
BethylRRID:AB_10631380Cat#: A120-601D4(1:400)
AntibodyAnti-Mouse AF555
(Goat polyclonal)
Life TechnologiesRRID:AB_141596Cat#: A21127(1:400)
AntibodyAnti-Sheep AF 568
(Donkey polyclonal)
Life TechnologiesRRID:AB_2535753Cat#: A21099(1:400)
Commercial Assay KitRNAscope Intro Pack
for Multiplex Fluorescent
Reagent Kit v2
Advanced Cell DiagnosticsCat#: 323,136
Sequence-based reagentEYFP-C1Advanced Cell DiagnosticsCat#: 312,131mRNA probe
Sequence-based reagentMm-Slc32a1-C2Advanced Cell DiagnosticsCat#: 319,191mRNA probe
Sequence-based reagentMm-Slc17a6-C3Advanced Cell DiagnosticsCat#: 319,171mRNA probe
OtherOpal Fluorophore
Reagent Pack 520
Akoya BiosciencesCat #: FP1487001KT
OtherOpal Fluorophore
Reagent Pack 570
Akoya BiosciencesCat #: FP1488001KT
OtherOpal Fluorophore
Reagent Pack 690
Akoya BiosciencesCat #: FP1497001KT
OtherDAPI stainAdvanced Cell DiagnosticsCat#: 323,108

Animals

All procedures followed the National Institutes of Health Guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus. Animals were housed in an environmentally controlled room, kept on a 12 hr light/dark cycle, and had ad libitum access to food and water. Adult mice of either sex were used in all experiments. Genotypes used were C57BL/6 (Charles River Laboratories), Neurotensin receptor1-Cre [Ntsr1-Cre; Mutant Mouse Regional Resource Center, STOCK Tg(Ntsr1-Cre) GN220Gsat/Mmucd], Gad1-Cre (Higo et al., 2009); Slc32a1 (Vgat)-Cre[#028862; Jackson Labs], and Slc17a6 (Vglut2)-Cre [#028863; Jackson Labs]. All transgenic animals were bred on a C57BL/6 background. Gad1 and Ntsr1-Cre mice were maintained as heterozygotes and were genotyped for Cre (Transnetyx). For all surgical procedures, mice were anesthetized with intraperitoneal injections of a ketamine hydrochloride (100 mg/kg) and xylazine (10 mg/kg) cocktail, placed in a stereotaxic apparatus, and prepared for surgery with a scalp incision.

Viral injections

Request a detailed protocol

Injections were administered using a pulled glass pipette. Unilateral pressure injections of 70–200 nl of Cre-dependent reporter viruses (AAV1-CAG-flex-GFP; AAV2-DIO-EF1a-eYFP; AAV8-hysn-ConFon-eYFP, see Key Resources Table) were made into Int. Injections were centered on IntA, with minor but unavoidable somatic label appearing in posterior interposed (IntP), lateral nucleus (LN), interstitial cell groups (icgs), and the dorsal region of the vestibular (VEST) nuclei, including dorsal portions of the superior vestibular nucleus (SUV), lateral vestibular nucleus (LAV), and Nucleus Y (Y). We occasionally observed minor somatic label in the parabrachial nucleus (PB) and the cerebellar cortex (Cb-Ctx) anterior or dorsal, respectively, to Int in injections into Vgat-Cre mice. In control injections (n=3; virus in C57/Bl6 mice or off-target injection into Ntsr-1 Cre mice), viral expression was not detected. We did not see appreciable somatic label in the medial nucleus (MN) of any specimens. For RN injections, craniotomies were made unilaterally above RN (from bregma: 3.5 mm posterior, 0.5 mm lateral, and 3.6 mm ventral). For Int injections, unilateral injections were made at lambda: 1.9 mm posterior, 1.6 mm lateral, and 2.1 mm ventral. For IO injections, the mouse’s head was clamped facing downward, an incision was made near the occipital ridge, muscle and other tissue was removed just under the occipital ridge, and unilateral injections were made at 0.2 mm lateral, and 2.1 mm ventral with the pipet tilted 10° from the Obex. This method consistently labeled IO and had the advantage of avoiding accidental cerebellar label via pipette leakage. To achieve restricted injection sites, smaller volumes were required in Vgat-Cre mice compared to Ntsr1-Cre mice (40–100 nl vs. 150–200 nl, respectively). The smallest Vgat-Cre injection was made iontophoretically using 2 M NaCl. Current (5 μA) was applied for 10 min, the current was removed and after a waiting period of 5 min, the pipet was retracted. Retrograde labeling of RN-projecting IntA neurons was achieved through AAVretro-EF1a-cre (Tervo et al., 2016) Retrograde injections of RN were performed simultaneously with flex-GFP injections of IntA. Retrograde virus (AAVretro-EF1a-Flp) was injected into IO 1 week before reporter viruses because of the different targeting scheme and mice were allowed to heal 1 week prior to the reporter virus injection. All mice injected with AAVs were housed postoperatively for ~6 weeks before perfusion to allow for viral expression throughout the entirety of the axonal arbor. Control injections were performed where Cre or Flp expression was omitted, either by performing the injections in wild-type mice or in transgenic mice without the Retro-flp injection into IO or RN, confirming the necessity of recombinase presence in reporter expression (Fenno et al., 2017).

For monosynaptic rabies retrograde tracing, AAV1-syn-FLEX-splitTVA-EGFP-tTA and AAV1-TREtight-mTagBFP2-B19G (Addgene; Liu et al., 2017) were diluted 1:200 and 1:20, respectively, and mixed 1:1 before co-injecting (100 nl of each; vortexed together) unilaterally into IN of Slc32a1-IRES-Cre (n=3; “Vgat”) and Slc17a6-IRES-Cre (n=1; “Vglut2”) mice. Two additional Vglut2-Cre mice were prepared using AAV1-EF1a-Flex-TVA-mCherry (University of North Carolina Vector Core; Watabe-Uchida et al., 2012) and AAV9-Flex-H2B-GFP-2A-oG (Salk Gene Transfer, Targeting and Therapeutics Core; Kim et al., 2016). After a 4–6 week incubation period, a second injection of EnvA-SAD∆G-eGFP virus (150–200 nl) was made at the same location (Salk Gene Transfer, Targeting and Therapeutics Core; Kim et al., 2016; Wall et al., 2010; Wickersham et al., 2007). Mice were sacrificed 1 week following the rabies injection and prepared for histological examination. Control mice (C57Bl/6; n=1) were injected in the same manner, however, without Cre, very little putative Rabies expression was driven, though eight cells were noted near the injection site. No cells were identified outside this region (Figure 2—figure supplement 1).

Tissue preparation and imaging

Request a detailed protocol

Mice were overdosed with an intraperitoneal injection of a sodium pentobarbital solution, Fatal Plus (MWI), and perfused transcardially with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer. Brains were removed and postfixed for at least 24 hr then cryoprotected in 30% sucrose for at least 24 hr. Tissue was sliced in 40 μm consecutive coronal sections using a freezing microtome and stored in 0.1 M phosphate buffer. Sections used for immunohistochemistry were floated in phosphate-buffered saline (PBS), permeabilized using 0.1–0.3% Triton X-100, placed in blocking solution (2–10% Normal Goat serum depending on antibody) for 1–2 hr, washed in PBS, and incubated in primary antibodies GFP (1:400), Gad65/67 (1:200), Vglut2(1:250), and TH (1:200) for 24–72 hr. Sections were then washed in PBS three times for a total of 30 min before incubation in secondary antibodies (Goat anti-rabbit DyL594, Goat anti-mouse AF555 (1:400), or Goat anti-sheep AF568 (1:400), see Key Resources Table) for 60–90 min. Finally, immunostained tissue was washed in PBS and mounted in Fluoromount G (SouthernBiotech).

Every section for rabies experiments and every third section for anterograde tracing experiments was mounted onto slides and imaged. Spinal cord sections were also sliced in 40 μm consecutive coronal sections with every fourth section mounted. Slides were imaged at 10× using a Keyence BZX-800 epifluorescence microscope or a slide-scanning microscope (Leica DM6000B Epifluorescence & Brightfield Slide Scanner; Leica HC PL APO 10× Objective with a 0.4 numerical aperture; Objective Imaging Surveyor, V7.0.0.9 MT). Images were converted to TIFFs (OIViewer Application V9.0.2.0) and analyzed or adjusted via pixel intensity histograms in ImageJ. We inverted fluorescence images using greyscale lookup tables in order to illustrate results more clearly. YFP terminals stained for neurotransmitter transport proteins (Gad65/67 or Vglut2) were imaged using a 100× oil objective on a Marianas Inverted Spinning Disc confocal microscope (3I). We imaged in a single focal plane of 0.2 μm depth to analyze colocalization of single terminal endings. Images were analyzed in ImageJ.

Analysis of overlap by genetically defined neurons

Request a detailed protocol

To distinguish overlap of Cre expression with transmitter markers, we performed ISH. For ISHs, RNAse-free PBS was used for perfusions and the tissue was cryoprotected by serial applications of 10%, 20%, and 30% Sucrose for 24–48 hr each. The brain tissue was then embedded in OCT medium and sliced to 14 μm thick sections on a cryostat (Leica HM 505 E). Tissue sections were collected directly onto SuperFrostPlus slides and stored at –80°C for up to 3 months until RNA ISH for EYFP (virally driven), Slc32a1 (Vgat), and Slc17a6 (Vglut2) from RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics). The slides were defrosted, washed in PBS, and baked for 45 min at 60°C in the HybEZ oven (Advanced Cell Diagnostics) prior to post-fixation in 4% PFA for 15 min at 4°C and dehydration in ethanol. The sections were then incubated at room temperature in hydrogen peroxide for 10 min before performing target retrieval in boiling 1× target retrieval buffer (Advanced Cell Diagnostics) for 5 min. The slides were dried overnight before pretreating in protease III (Advanced Cell Diagnostics) at 40°C for 30 min. The RNAscope probes #312131, #319191, and #319171 were applied and incubated at 40°C for 2 hr. Sections were then treated with preamplifier and amplifier probes by applying AMP1, AMP2 at 40°C for 30 min and AMP3 at 40°C for 15 min. The HRP signals were developed using Opal dyes (Akoya Biosciences): 520 (EYFP probe), 570 (Slc17a6 probe), and 690 (Slc32a1 probe) and blocked with HRP blocker for 30 min each.

The CbN were stained using DAPI for 30 s before mounting in Prolong Gold (Thermo Fisher Scientific). Washes were performed two times between steps using 1× wash buffer (Advanced Cell Diagnostics). Fluorescence was imaged for YFP, Slc17a6 (Vglut2), Slc32a1 (Vgat), and DAPI using a Zeiss LSM780 microscope. Each image was captured using a 34-Channel GaAsP QUASAR Detection Unit (Zeiss) at 40× magnification in water from 14 µm sections. Images were stitched using ZEN2011 software and analyzed in ImageJ. Cre-expressing cells were identified by somatic labeling in the YFP channel; colocalization with the Slc32a1 (Vgat) or Slc17a6 (Vglut2) channel was determined by eye using a single composite image and the ‘channels tool.’ Positive (Advanced Cell Diagnostics, PN 320881) and negative control probes (Advanced Cell Diagnostics, PN 32087) resulted in the expected fluorescent patterns (Figure 1—figure supplement 3).

We analyzed the fidelity of our transgenic lines using virally mediated YFP somatic label and DAPI staining to identify cells expressing Cre and analyzed the colocalization of Slc32a1 or Slc17a6 mRNA within the bounds of a YFP cell. The YFP signal was often less punctate than other endogenous mRNA probes, thus we restricted our analysis to cells largely filled by YFP signal that contained DAPI stained nuclei. Due to the high expression patterns of Slc32a1 and Slc17a6, analyzing by eye was reasonable. Only a total of four cells across all analyzed sections appeared to have ISH-dots in both the Slc32a1 and Slc17a6 channels. This may be due to poor focus on these individual cells or background fluorescence. Two sections per animal and one (Vglut2-Cre and Gad1-Cre) or two animals (Ntsr1-Cre and Vgat-Cre) per transgenic line were counted.

In preliminary studies, we tested a Gad1-Cre driver line (Higo et al., 2009) for specificity since Gad1 was recently identified as a marker of inhibitory neurons within the CbN (Kebschull et al., 2020). However, in this line, we observed clear instances of both Gad65/67 and Vglut2- immunoreactivity in YFP labeled terminal varicosities as well as some Slc17a6 mRNA (Vglut2) expression in YFP expressing somata (Supplementary file 2, Figure 1—figure supplements 34). 87 % YFP expressing cells colocalized with Slc32a1 (60/69 cells) and 13% colocalized with Slc17a6 (9/60 cells). The clear instances of promiscuity in Gad1-Cre mice precluded further use of these mice in the present study.

Cell size analysis

Request a detailed protocol

We imaged cells within IntA at 20× with the epifluorescent Keyence microscope, then used the ‘Measure’ tool in ImageJ to gather the cross-sectional area and the ‘Fit ellipse’ measurement to gather minimum and maximum diameter which we converted from pixels to microns using reference scale bars. We report the maximum diameter. We analyzed 15–110 well focused and isolated cells for each specimen. Statistical analyses were conducted using a repeated-measures ANOVA on the per-animal and grand means of cell diameter per experimental condition.

Brain region classification

Request a detailed protocol

We used a combination of the Allen Mouse Brain Reference Atlas and the Mouse Brain in Stereotaxic Coordinates by Franklin and Paxinos to identify brain regions, while noting that there were minor differences in location, shape, and naming of the brain regions between these reference sources (Lein et al., 2007; Paxinos, 2008). We use the term IntA to refer to the anterior and dorsolateral interposed nucleus; IntP to refer to the posterior interposed nucleus; IN in the transsynaptic tracing section to refer to all subdivisions of the interposed nucleus; and Int to refer to the intermediate region of the deep CbN including all subdivisions of the interposed nucleus and the more ventral intermediate interstitial cell groups. We grouped the dorsolateral and anterior subdivisions of Int because they were often co-labeled, are difficult to confidently distinguish, and occur at similar anterior-posterior (AP) coordinates. For monosynaptic rabies tracing and difficulty in targeting multiple viruses to the exact same location—we grouped all subdivisions of the interposed nucleus (IN; anterior, posterior, and dorsolateral). In general, we followed nomenclature and coordinates respective to bregma of the Allen Mouse Brain Reference Atlas including its classification conventions of motor, sensory, modulatory/affective sources from the 2008 version. Thalamic regions were classified as motor if they project to motor cortices; sensory if they project to sensory cortices, with intralaminar thalamic nuclei classified as modulatory/affective. The intermediate and deep layers of the SC harbored terminal fields and retrogradely labeled neurons and is thus classified as motor.

Projection quantification

Request a detailed protocol

All specimens were included in data sets if their injection sites were centered on IntA, with spillover described in each figure. If no label was observed or injections were mistargeted, specimen were excluded from further consideration. We mapped terminals to a collection of extracerebellar targets spanning the anterior-posterior (A-P) axis from the posterior medulla to the thalamus. We assigned terminal fields a semi-quantitative relative projection strength (RPS) of 0–4 based on the density and anterior-posterior spread (Supplementary file 1) made by a single observer and verified by a second. The values were assigned relative to the highest density projection target for each genotype: All Ntsr1-Cre projection fields were assessed relative to the density of terminals in RN whereas Vgat-Cre specimens were assessed relative to the density of IO terminals (Figure 1—figure supplement 1). Briefly, a terminal field that was both dense and broad (in spanning the anterior-posterior axis) was assigned a RPS of 4, semi-dense and semi-broad assigned a 3, semi-dense and/or semi-broad a 2, and fields determined to be sparse but nevertheless present, were assigned an RPS of 1. In addition, we compared our specimens to analogous preparations published in the Allen Mouse Brain Connectivity Atlas, specifically the histological profile of Cre-dependent labeling following injections into IntA of either Ntsr1-Cre or Vgat-Cre mice. These publicly available sources recapitulated projection signatures from lab specimens (Supplementary file 2). We included the Allen injection data in our analysis of average projection strength for Ntsr1-Cre (n=1) and Vgat-Cre (n=1) specimen but did not use the histological images of these injections here. The full histological profiles of genetically restricted GFP label from the Allen can be found at: 2011 Allen Institute for Brain Science. Mouse Brain Connectivity Atlas. Available from: http://connectivity.brain-map.org/, experiments #264096952, #304537794.

We determined the average proportion of the total RPS value that is derived from specific projections (to specific modalities or hemispheres; Figure 5D and E) by summating the RPS values to every region receiving a projection per specimen. We then divided this number by summated RPS values in the groupings of interest. We report the average proportion of total RPS values across all specimens in each experimental cohort (biological replicates). These measurements are therefore indicative of the strength of projection to certain modalities or hemispheres, and not simply a measure of the number of brain regions targeted.

Rabies quantification

Request a detailed protocol

We identified presumptive starter cells as rabies (mCherry) positive cells that also contained GFP (AAV1-syn-FLEX-splitTVA-EGFP-tTA). We used an antibody against EGFP (see Key Resources Table) to visualize TVA at these concentrations, but mBFP (AAV1-TREtight-mTagBFP2-B19G) could not be visualized. However, oG expression is restricted to cells expressing TVA due to the necessity of the tetracycline transactivator gene encoded by the virus delivering TVA (Liu et al., 2017). In two Vglut2-Cre mice used for rabies tracing, we identified presumptive starter cells as rabies positive cells within the CbN where both mCherry (AAV1-EF1a-Flex-TVA-mCherry) and GFP (AAV9-Flex-H2B-GFP-2A-oG-GFP/ EnvA-SAD∆G-eGFP) were expressed. We could not easily identify cells in which all three components were present due to overlapping fluorescence from the oG and modified rabies viruses, thus starter cell identification is an estimate (Doykos et al., 2020).

Data availability

All data analysis is included in the manuscript and supporting files.

References

  1. Book
    1. Chan-Palay V
    (1977) The Cerebellar Dentate Nucleus
    In: Chan-Palay V, editors. Cerebellar Dentate Nucleus. Springer. pp. 1–24.
    https://doi.org/10.1007/978-3-642-66498-4
  2. Book
    1. Paxinos G
    (2008)
    The Mouse Brain in Stereotaxic Coordinates, Compact: The Coronal Plates and Diagrams
    Elsevier Science.

Decision letter

  1. Roy V Sillitoe
    Reviewing Editor; Baylor College of Medicine, United States
  2. Ronald L Calabrese
    Senior Editor; Emory University, United States
  3. Roy V Sillitoe
    Reviewer; Baylor College of Medicine, United States
  4. Albert I Chen
    Reviewer; Nanyang Technological University, Singapore
  5. Hirofumi Fujita
    Reviewer; Johns Hopkins University, United States

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Acceptance summary:

There is a growing body of evidence demonstrating that the cerebellum is involved in motor and non-motor functions. In this paper, Judd and colleagues present compelling evidence that cerebellar output neurons are more diverse in their projections than previously appreciated. They show that excitatory and inhibitory cerebellar nuclei neurons innervate a range of extra-cerebellar loci. The anatomical maps uncovered in this work could support diverse cerebellar functions that control different behaviors.

Decision letter after peer review:

Thank you for submitting your article "Widespread inhibitory projections from the interposed cerebellar nucleus" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Roy V Sillitoe as Reviewing Editor and Reviewer #1, and Ronald Calabrese as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Albert I Chen (Reviewer #3); Hirofumi Fujita (Reviewer #4).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Summary:

In this study, Judd and colleagues use a combination of mouse genetics and viral marking to expand the extra-cerebellar map of projections. These data will impact our understanding of how the cerebellum contributes to behavior and in general how brain function is packaged at the anatomical level. These data will not only impact cerebellar scientists but also those workers interested in how inter-regional brain connectivity is organized and how fine input-output circuit relationships are structured.

Essential revisions:

1) The reviewers felt that further validation of the mouse Cre lines is necessary for the authors to make the major conclusion of the paper.

2) The reviewers pointed to a significant number of errors and omissions throughout the paper, including missing data and organizational problems between the text and figure panels.

3) The reviewers have identified instances in which additional and/or revised statistical analyses are necessary.

Reviewer #1 (Recommendations for the authors):

General Comments:

1. The major conclusions of this paper rely on the assumption that Gad1 and Vgat-Cre lines are restrictive to inhibitory neurons and the Ntsr1-Cre line to excitatory neurons. Did the authors confirm the expression of the Gad1-Cre and Vgat-Cre line are indeed restricted to those inhibitory neurons and the Ntsr1-Cre line only to the presumed excitatory neurons? The analyses presented in this paper do not demonstrate overlap or restricted expression between any of the three lines, as they rely on previous expression patterns of genes, not the Cre recombination capability and specificity in the Cre-line(s) itself. The nearly 20% pixel overlap between Ntsr1 and Vgat or Gad1 cells (supplemental figure 1) is also large enough to be a potential confound in the core findings of this paper.

We suggest to test the assumption that each of the Cre-lines is expressed in the intended cell types by co-labelling Cre and Ntsr1/Gad1/Vgat/Vglut2 in your mouse lines using in situ hybridization or co-staining of projections using antibodies for Vglut2 or Vgat. These experiments need to be performed on all three Cre-lines to validate the Cre expression.

2. It is surprising that there are Vgat-positive, Gad1-negative (or vice versa) cells with differential projection patterns (according to Figure 2). The authors need to provide a high-resolution image showing the co-labelling for Vgat and Gad1 neurons demonstrating this apparent incomplete overlap (Figure 1)? Also, were the authors able to find labeled neurons that are Vgat-positive and not Gad1-positive (or vice versa)? Could some of the observed difference be explained by a lower number of infected Gad1 positive neurons than Vgat neurons per injection? Please discuss in the text.

If not, the author may want to change their statements regarding their finding that Vgat and Gad1 populations are not completely overlapping neural subtypes.

3. How did authors differentiate from projection fibers (axons) and projection targets (synapses)? Similarly, how did authors differentiate between fibers and cell bodies?

4. When quantifying "distribution of labeled cells", did the authors observe more unique projections in mice with a lower proportion of labelled cells in the IntA? "Distribution of labeled cells" also may be an ambiguous phrase, as it suggests that the observed proportion of IntA (for example) neurons is labeled according to some map. Perhaps a better title would be "Location of labeled cells." Apologies if I am missing something, in the end it just seems like a leap in the way these data are presented and discussed.

5. It is unclear how input regions were identified, this is especially important given that twice as many Gad1-Cre mice were injected than Ntsr1-Cre mice and that the number of input cells appears a magnitude lower in the Ntsr1-Cre mice. How many rabies-infected cells were labelled for each injection (per mouse and per group)? Does the input region need to be found in all mice from one group, or just one? What is the variability in input region?

And how did the authors control for non-specific labelling? Did they perform any (specifically the rabies tracing) in Cre-negative mice? (methods suggest yes, but results are not show). Please clarify and add the details to the manuscript.

6. Authors should consider including some of the low magnification whole mount images of the supplemental data into the main text figures. They provide a better picture of the data than the cropped images.

Statistics comments:

1. N-number is of neurons is inflated in Figure 3. Statistics should be compared for averages on each mouse, not on total number of cells. The authors seem to perform multiple pairwise comparisons without correcting for multiple comparisons.

Textual comments:

1. Line 113: put abbreviation brackets direction behind "nucleus."

2. Line 114: define "RN."

3. Line 118: define "ZI."

4. Line 119: define "VM, VPM, VAL"

Define all abbreviations in the text first time they are used.

5. Line 116: "a morphological… excitatory neurons." This observation is best left to when figure 5A is discussed, as experimental approach or data describing excitatory neurons has not yet been discussed at this point.

6. Line 124-126: "Vgat neurons generally targeted these regions more strongly than Gad1 neurons." How does this relate to the number of infected neurons in the IntA? This is also a qualitative statement that is not backed by statistical tests.

7. Line 149-155: should authors refer to supplemental figures 2 and 3 here? I would suggest for the authors to order the supplemental figures in the order that they are discussed in the text.

8. Line 179-182: this method may warrant a little more explanation for readers that are not familiar with intersectional viral mapping.

9. Line 182-184: "Injections in wildtype…the nuclei." The authors need to include images of these control experiments in the manuscript.

10. Line 224-225: what tests do these p-values represent?

11. Line 234-235: "we validated…" This is not a validation of your Cre-lines.

12. Line 238-240: "was within the noise of presumptive non-overlapping neurons." Presumed by whom? Almost 20% overlap in signal seems pretty high when you are trying to define cell-type specific projection maps. Please clarify.

13. Line 255-257: "Together, these results…in the brainstem." I think that this conclusion is not warranted based on the data that is shown. See comments 11 and 12 and General comment 1.

14. Line 279: define "nIntA". Why was it chosen to name the putative excitatory neurons based on their Cre-line but the putative inhibitory neurons based on the assumption they are inhibitory (iIntA)? In the end, too many assumptions about the identity of the different neurons were made.

15. Line 289-291: "We currently are unable…neurons project to IO." This is another reason to validate the Cre-lines.

16. Line 291-292: "Exclusive retrograded labeling…ventral brainstem." If this is the case, why can you perform retrograde labeling from IO to iIntA? Are there no iIntA terminals just dorsal to the IO? The supplemental figures do not clarify this.

17. Cite Supplemental Figure 4 in text.

18. Line 333-334: "We conclude it….information broadly." An alternative conclusion is that the majority of Ntsr1-Cre positive cells projects to the RN. I do not think you can conclude that all Ntsr1 cells project everywhere (this is not tested).

19. Line 366-369: Figure 5B only shows proportion of projections. It does not show number of regions targeted or relative strength of these proportions. I think that a qualitative statement can be made that only Ntsr1-Cre neurons project to modulatory regions but the author sshould refrain from making quantitative statements based on proportions of injections that are inherently variable between cell types and injections.

20. Line 398-409: Why are these data not shown?

21. Line 438: how many starter cells did you observe in Gad1 vs Ntsr1 mice?

22. Line 450-451: did the authors test this statistically?

23. Line 441-462: why did the author change the way they referred to Gad1 neurons?

24. Line 460-461: "this patterned… cell types." In previous figures it was shown that iIntA do not heavily project to modulatory regions, whereas nIntA neurons do. Therefore, this pattern does not mirror targeted patterns.

25. Line 535-537: partial difference in projections do not preclude partial overlap in Gad1-Cre and Ntsr1-Cre expression.

Methods comments:

1. What are in the atlas coordinates for the injection sites. Please provide them all.

Figure comments:

1. Please put the abbreviations used in each figure in its figure legend.

2. Please be consistent in the use of "Cre" (capitalized and with an "e") in all figure legends.

3. Label IntA (and other cerebellar nuclei) in larger cross-sections for consistency.

4. Figure 1, use similar schematic of viral injection.

5. Figure 1, please be consistent in labeling of anatomical areas in the figure panels: "Vgat" and "Gad1" above figure panels, and abbreviations for anatomical locations within the figure panel.

6. Figure 1B and 2B, are the Vgat infected cells always more medial than Gad1 infected cells? Is this a problem of injection variability? If so, how do you control for this variability in the different projection maps?

7. Figure 1C-E, boxes in schematics are not same size but figure panels seem similar magnification. Boxes on schematics are a little confusing. It may help to label the anatomical location of interest using color and naming it.

8. Figure 1D, what is the difference between "PSV" and "V" in left images?

9. Figure 2, what does Con/Fon mean?

10. Figure 2, please keep order of Gad1 and Vgat injections consistent with figure 1 for comparison.

11. Figure 3, the statistics for this figure should be rerun, see statistical comments.

12. Figure 4E, see comment 6.

13. Figure 4E, 4th row, there are 3 boxes here.

14. Figure 5 B, please keep color-coding of groups consistently throughout the paper.

15. Figure 6C, provide legend.

16. Figure 6D, consider plotting this data on two images with separate left oriented Y-axis. This figure panel is very unclear.

17. Figure 6F, the leftward and rightwards histograms have been used in other figures to denote ipsi- and contra-lateral projections. Changing this is confusing. It may be beneficial to plot these in the same orientation.

18. Figure 6G, this figure is too small to see and the difference between top panels and bottom panels is unclear.

19. Figure 7, please keep the order in which the Gad1 and Ntsr1 labelled neurons are shown consistent with figure 6 for comparison (keep this order similar with supplemental figure 5 and all figure legends too).

20. Figure 7, retrograde rabies labeled cells are hard too to see, please increase the magnification for input neurons.

21. Figure 7, if I understand correctly, the results for each column are from 4 different injection paradigms/experiments that are also summarized in other figures. It would help the reader if the authors would include schematics for each experiment (include: mouse line, injection side, and type of virus used).

22. In supplemental figure 2, make lines of boxes in inset thicker. Box for Gvi is missing. Box for Hiv is missing. B has no scale bar.

23. In supplemental figure 3, make lines of boxes in inset thicker. Explain scale bars in Figure legend.

24. In supplemental figure 4, make lines of boxes in inset thicker. Label boxes in E-K. Also, B has no scale bar.

25. Table 1, please write out full name of each nucleus in column 1.

Reviewer #2 (Recommendations for the authors):

1. For Figure 1, please clarify how much the distinctions in projection distribution could be due to variability in specificity of initial targeting or recombination ability of Vgat-Cre and Gad1-Cre? The conclusions can be strengthened by analyzing the starter distribution of IntA neuron subtypes compared to vestibular or cerebellar cortex like Figure 2C.

The main concern here is that selective targeting of IntA using Cre-lines that recombine in all 3 subnuclei is a potential weakness in methodology since this relies entirely on precision of stereotaxic injection.

2. Questions about similarities or differences between Vgat-Cre and Gad1-Cre recombination or targeting by CN neurons cannot be resolved by in situ hybridization analysis of Vgat and Gad1 in the CN alone (Figure 1A). Perhaps more informative if authors could provide Vgat-Cre; reporter expression with Gad1 in situ and vice versa. Also, it is difficult to see signals in the small panels, and the overlay seems to indicate a lot of heterogeneity. Quantification of Gad1 and Vgat in situ would be informative.

3. Line 177, please describe how labeling of Purkinje cells might affect interpretation. Perhaps clarify that PC labeling might lead to VEST and PBN labeling from Purkinje cell axons directly projecting out of the cerebellum.

4. In Figure 3A, the prediction is that the Gad1-Cre more generally labels Gad1 projections and Gad1IO labels a more restricted subset since using IO retrolabeling. But in 3A, it seems that Gad1IO picked up neurons that were not picked up by Gad1 anterograde labeling, specifically the larger neurons. Text argues perhaps Gad1 labels more interneurons than Gad1IO, but should not exclude Gad1 picking up some of these large neurons (maybe just less). Please explain.

5. Within IntA, what % of vGluT2 neurons does Ntsr1 make up? If Ntsr1 is in almost all, then the result in Figure 4 says almost all nIntA neurons projections are collaterals of RN-projecting nIntA neurons. But if Ntsr1 is in a much smaller subset of IntA vG2 neurons, then this says something different. Analyzing the % of Ntsr1 neurons that are vg2 in the IntA is needed to resolve this.

6. Figure 5A, please provide quantification of the morphology to strengthen the reported differences (PCA analysis?). For IO, wondering if the morphology of CN terminals could fit into the morphological schemes proposed by Vrieler et al., 2019 Brain Structure and Function 224. In other words, do distinct IntA neurons project to specific subtypes of IO neurons.

7. Please show images of the cerebellar cortex in Ntsr1 and Gad1 to show starter cells to address the following questions. For Ntsr1, how much extra-CN recombination does Ntsr1-Cre mediate? For Gad1, I would like to get a sense of how many inhibitory neurons (interneurons and Purkinje cells) are labeled initially. Both are important to determine how to interpret input specifically into CN or whether the authors are looking at input into cell types elsewhere in the cerebellum.

Another way to ask this same question, in Figure 6D, what are the cerebellar cortex Ntsr1 starter cells? Granule cells or maybe even inhibitory neurons? Ntsr1-Cre recombines in excitatory neurons in the CN (Houck and Person 2015), but could more generally label excitatory and inhibitory in the cerebellar cortex. And for Gad1 starter cells in the cerebellar cortex, are they mostly Purkinje cells or do these include Golgi/stellate/basket cells? The concern here is that if the rabies virus receive G and TVA from granule/Purkinje/Golgi/stellate/basket cells, then in addition to CN input, authors are looking at climbing and mossy fibers that project to non-CN cells confounding interpretation.

8. This is important for analysis and interpretation: how does one dissociate the GFP from rabies or GFP from oG? The double GFP strategy is a major concern here.

9. Figure 6B, starter cells are defined as intersection between GFP and mCherry, but in B, there is very little overlap here mostly because red channel is so faint. For instance in Gad1, I can see 2 red cells and 4 green cells but with only 1 yellow cell. I am concerned about the small number of starter cells.

Same issue as Figure 6, in Figure 7B, worried that the input sources represent more than just input sources for CN, but for cell types in the cerebellar cortex especially given high percentage indicated in 6D.

10. For Figure 7A, please provide ways to distinguish axons from dendrites.

Reviewer #3 (Recommendations for the authors):

(1a) Interrelationships between the cell types used are not clear. The authors show overlap between Ntsr1+, Gad1+/Vgat+, and VGluT2+ neurons (FigS1B). Does this mean that Cre-dependent AAV tracing for Gad1Cre+/VgatCre+ neurons could also label VGluT2+ neurons? Does the Ntsr1Cre+ cell type contain inhibitory neurons? If so, how did the authors verify that the axonal projections are actually inhibitory or excitatory? Please discuss the technical limitations.

(1b) Relatedly, it would be helpful to make clear which inhibitory cell types are taken into account and discussed: Gad1-IO, Gad1-collateral, Gad1-GABA-local, Gad1-glycine-local, Gad1-glycine-projection? How did the authors allocate VgatCre+ neurons? What is the overlap between Gad1Cre+ and Vgat1Cre+ neurons (they look different because labeled VgatCre+ neurons are in the medial part of the IntA and labeled Gad1Cre+ neurons are in the central part of the IntA consistently throughout Figures)? Something like Venn diagrams for Gad1Cre+, VgatCre+, and Ntsr1Cre+ neurons, could help.

(2a) I see substantial and consistent injection leak to non-IntA and non-CN regions. How did the authors identify that the connections were really of IntA but not from other regions, for example, Vest? For example, inconsistent with our knowledge are that (1) IntA projections to DCK and PO of IO, (2) IntA projections to several targets (please compare Table 1 with Teune et al., 2000, Prog Brain Res), (3) CrI/PFL Purkinje projections to IntA neurons, which are known to IntP and LN, (4) sparse RN projection to Ntsr1Cre+ IntA neurons, and (5) very dense vestibular projections to Ntsr1Cre+ IntA neurons. Inconsistent within the manuscript is that projection specific tracing revealed more projection targets than non-specific tracing (Table 1). Please discuss these and other relevant technical issues.

(2b) Again, given the technical challenges, the Gad1+/Vgat+ axons labeled in non-IO regions could be explained by injection leak to non-CN regions or to excitatory neurons. On what evidence can the readers be convinced of "inhibitory projections" that are "from IntA"?

3) I see potential issues in viruses used: 1) efficiency for the rabies tracing was significantly different between NtsrCre (yielded ~150 cells) and Gad1Cre neurons (yielded ~4000 cells), 2) rabies tracing didn't identify known input from IO, 3) Cre-independent TVA expression from Cre-dependent AAV has been reported to be a significant problem in the rabies tracing scheme used in this study (Fagat et al., 2016, Cell Rep), and 4) how did the authors control retrograde infection of AAVs? Please discuss the technical limitations.

4) I don't see any evidence for input to VTA.

5) Table 2 is missing.

6) 'long-standing dogma' -- I don't think so. Although it has been 'widely assumed', the existence of collaterals of preolivary neurons has been looked for many times and remains an open question.

7) Difficult to see axonal terminals in photos in general.

8) Cerebellar nuclear outputs are powerfully modulated by their upstream Purkinje cells. In this sense, I think it's worth discussing the organization of PC inputs to inhibitory neurons. Can you discuss more about implication of broader Purkinje innervation of inhibitory IntA neurons? Are these different than the parasagittal modules? Or, does the difference in number/location of labeled PCs for inhibitory vs excitatory DCN neurons simply reflect the technical limitation in rabies tracing?

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Widespread inhibitory projections from the cerebellar interposed nucleus" for further consideration by eLife. Your revised article has been evaluated by Ronald Calabrese (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Please refer to the specific comments by each Reviewer. In general, the Reviewers have identified areas where the clarity of the figures could be improved. We feel that they will go a long way in helping our readers fully appreciate the data. Also, please address the issues regarding the nomenclature of mouse alleles and how they are presented (e.g. the use of italics). Importantly, all 3 Reviewers have identified a number of important pieces of information that are either missing and/or contain errors.

Reviewer #1 (Recommendations for the authors):

In this revised manuscript, Judd et al., have performed essential experiments to validate their Cre-mouse lines. They have incorporated additional mouse lines based on their findings (and previous reviewer comments) and repeated specific experiments in these mouse lines. The observation that inhibitory neurons in the interposed nuclei send widespread projections throughout the brainstem holds and indeed the data present an interesting and important set of findings. However, below are suggestions about the text and figures that will aid the reader in understanding the detailed anatomical data presented in this manuscript.

– The overlap in dots and lines in Figures 1D, 2E, 3D, and 4E make these figures hard to interpret. Some of the projections included in the results may be specific to the injection site outside of the interposed nucleus (off-target viral expression) and this possibility is discussed briefly in the text. In addition, all the regions in which projections are found, even if in just a few viral replicates with the least specificity, are included in the summary of Figure 7B. This should be clearly stated in the text. In addition, we suggest clarifying the panels in 1D, 2E, 3D, and 4E. One option would be to replace these plots with UpSet plots (for example) that show the overlap between the different data sets. Alternatively, the current graphs could be kept and the authors could instead include a ratio of the number of replicates in which the projection was observed. In the end, any way to help the reader more easily appreciate the data would be highly beneficial.

– The authors state "… dorsal to IntA, but nucleocortical terminals that were included in the projection analysis were not located in the same topographical area." Apologies, but I do not understand what you mean by this. What topographical area are you referring to? Please consider rephrasing.

– The authors state "Because injection sites did not differ systematically across…". What do mean? What is your measure that gives you confidence in regard to being systematic?

Reviewer #2 (Recommendations for the authors):

In the revised manuscript, the authors have more than sufficiently addressed all my questions and concerns (and those of the other reviewers) about the interpretations of the distinctions of input and output patterns by different Int neuronal subtypes and how the data was generated. The replacement of Ntsr-Cre/Gad1-Cre with vGluT2-Cre/vGAT-Cre mice for monosynaptic rabies tracing clarifies and strengthens the conclusions for the input-output connectivity by excitatory and inhibitory Int neuronal subtypes.

This is a carefully conducted study that describes exciting findings about the anatomical organization of cerebellar interposed anterior nucleus, especially the inhibitory subpopulation. These findings will undoubtedly provide important ground work for future investigation of the functional relevance of distinct Int neuronal subtypes and pathways in dexterous and locomotor movements, and beyond.

I very much enjoyed reading and reviewing this manuscript and congratulate the authors on a well conducted study.

Reviewer #3 (Recommendations for the authors):

The manuscript is significantly improved for clarity by the authors' thorough hard work. The existence of ramifying inhibitory projection neurons from Int CN became more convincing. Particularly, analyses on Gad/Vglut immunoreactivity of axonal terminals are helpful.

Now it became clearer that the manuscript adds the widespread ramifying cells to the existing repertoire of inhibitory cerebellar nucleus neurons. The authors establish this cell type by demonstrating the inhibitory signature of projected axons and then place them into context in comparison with well-known types of cells regarding outputs, inputs, and cell morphology. During these efforts, the authors also made clear outputs of a subset of excitatory cells (Ntsr1+ cells) and discovered monosynaptic inputs to the excitatory and 'inhibitory' cells, which themselves are novel and intriguing.

I only have one major comment for improving the clarity of conclusions.

The above efforts made me realize that the current experimental designs do not distinguish or identify the ramifying cells from the widely known IO-only cells (I tentatively call them like this though they could have their own collaterals), which may be important for one to incorporate this cell type into cerebellar theories, although distinction from a subset of excitatory cells is clear. As far as I can see, only sparse and highly restricted labeling in IO in Figure 4-S1D would suggest that IO projections from the ramifying cells are made with a different topographical rule than IO-only cells, which is assuring because these results show a potential distinction between inhibitory cells.

Specifically, in the current manuscript, results for cell morphology, axonal trajectory, and input circuits demonstrated in Figure 5, 6. and 7, which are currently treated as data for the ramifying cells, could simply reflect those of IO-only cells and could barely reflect the ramifying cells. This is because the injection strategies utilized do infect both IO-only and the ramifying cells (To selectively target ramifying cells by avoiding IO-only cells, something like Cre-dependent retrograde infection from RN/PG/TRN in Vgat-Cre mouse would be required). It must be made clear in the Result that these analyses do not distinguish them. This limitation should influence conclusions regarding Figure 5-7 and I recommend modifying them accordingly. Similarly, Line 83-84, Line 376 may also be misleading. It would be clearer to state that what this study identified is "inhibitory projection cell type(s) that (or, at least some of which) collateralizes to IO" rather than "IO projection cell type that also collateralizes to other areas".

https://doi.org/10.7554/eLife.66231.sa1

Author response

Essential revisions:

1) The reviewers felt that further validation of the mouse Cre lines is necessary for the authors to make the major conclusion of the paper.

We have used two independent methods to validate the transgenic Cre driver mouse lines used in the study – in situ hybridization and immunohistochemistry. Of critical importance, we learned that the Gad1-Cre line promiscuously expressed Cre in multiple cell types that project from the interposed nucleus. This promiscuity was observed both in immunohistochemistry, where we found terminal varicosities that expressed Vglut2 or Gad65/67, as well as in somata that expressed Vglut2 and not Vgat, or vice versa, inconsistent with specific label of inhibitory neurons. We could find no report of this promiscuity in the literature, despite the use of this line by other groups. Therefore, as a service to the field we include the description of this validation step in this report, while also entirely removing tracing datasets derived from the Gad1-Cre line from the manuscript. By contrast, we found the Vgat-Cre line to be well controlled with corroborating IHC and ISH data specific to inhibitory cell types. Importantly for the main focus of the manuscript, and as reported in the original submission, we observed that these cells innervated multiple targets outside of IO. The discovery that the Gad1-Cre line was not specific, necessitated replacing Gad1-Cre with Vgat-Cre mice for monosynaptic rabies tracing. We describe details of this discovery, decision, and consequences of this change more thoroughly in the main text of the revised manuscript. All terminal varicosities of Ntsr1-Cre derived projections were Vglut2 immunoreactive, mitigating concerns that these outputs were mixed. However, despite the specificity of immunolabel in terminals, in situ hybridization data in the cerebellar nuclei of Ntsr1-Cre mice revealed a small population of Vgat neurons, putative interneurons, that colocalized with Cre-driven YFP. Therefore, we replaced the Ntsr1-Cre dataset in the monosynaptic rabies experiments with Vglut2-Cre mice. The complete reworking of the monosynaptic rabies datasets in part accounts for the long delay between receiving reviews and this revision. We thank the reviewers for these critical suggestions, which we believe strengthened the main conclusions of the manuscript.

2) The reviewers pointed to a significant number of errors and omissions throughout the paper, including missing data and organizational problems between the text and figure panels.

We have thoroughly revised the manuscript, paying close attention to align text, figures, and figure legends. As described in the point-by-point responses, we have addressed noted errors as well as those not yet noted and are sincerely grateful to the referees for their careful reviews.

3) The reviewers have identified instances in which additional and/or revised statistical analyses are necessary.

We thank the reviewers for these corrections. In the cell size section, where statistical reporting was critiqued, we have (1) expanded the number of mice used, and compared means across animals rather than cells as independent measurements as suggested by the reviewer; and (2) performed statistical tests incorporating adjustments for multiple comparisons (Tukey’s multiple comparisons test), as suggested. The main result – that Vgat-Cre neurons are significantly smaller than Ntsr1-Cre neurons, regardless of direct or projection-specific labeling methods, holds, and is consistent with previously published reports.

Reviewer #1 (Recommendations for the authors):

General Comments:

1. The major conclusions of this paper rely on the assumption that Gad1 and Vgat-Cre lines are restrictive to inhibitory neurons and the Ntsr1-Cre line to excitatory neurons. Did the authors confirm the expression of the Gad1-Cre and Vgat-Cre line are indeed restricted to those inhibitory neurons and the Ntsr1-Cre line only to the presumed excitatory neurons? The analyses presented in this paper do not demonstrate overlap or restricted expression between any of the three lines, as they rely on previous expression patterns of genes, not the Cre recombination capability and specificity in the Cre-line(s) itself. The nearly 20% pixel overlap between Ntsr1 and Vgat or Gad1 cells (supplemental figure 1) is also large enough to be a potential confound in the core findings of this paper.

We suggest to test the assumption that each of the Cre-lines is expressed in the intended cell types by co-labelling Cre and Ntsr1/Gad1/Vgat/Vglut2 in your mouse lines using in situ hybridization or co-staining of projections using antibodies for Vglut2 or Vgat. These experiments need to be performed on all three Cre-lines to validate the Cre expression.

We thank the reviewer for this critical feedback; indeed, the concerns were shared amongst all three the reviewers. As summarized above under Essential Revisions, we have addressed Cre driver line specificity using immunohistochemistry and in situ hybridization in our own tissue, no longer relying on validation through public databases from the Allen Brain Atlas. To summarize changes to the revision that have stemmed from these new experiments: (1) We have added Cre line validation data within each section describing results from the respective Cre lines; (2) In situ hybridization data, along with a variety of hold-out controls, further test for viral specificity; (3) We have removed the Gad1-Cre driver line from the body of the experiments, but retain the characterization of this line as non-specific for the benefit of field; (4) We replace Gad1-Cre driver line with the Vgat-Cre driver line in all sections – direct Cre-dependent tracing, projection-specific tracing, and monosynaptic rabies input tracing; (5) We retain Ntsr1-Cre for output tracing since terminals labeled consistently expressed Vglut2, however we replace this line with Vglut2-Cre for input tracing.

2. It is surprising that there are Vgat-positive, Gad1-negative (or vice versa) cells with differential projection patterns (according to Figure 2). The authors need to provide a high-resolution image showing the co-labelling for Vgat and Gad1 neurons demonstrating this apparent incomplete overlap (Figure 1)? Also, were the authors able to find labeled neurons that are Vgat-positive and not Gad1-positive (or vice versa)? Could some of the observed difference be explained by a lower number of infected Gad1 positive neurons than Vgat neurons per injection? Please discuss in the text.

If not, the author may want to change their statements regarding their finding that Vgat and Gad1 populations are not completely overlapping neural subtypes.

The reviewer makes an important point that we think is fully addressed by the observation that the Gad1-Cre line was non-specific in its label of neurons. With this observation in hand, it is perhaps no longer surprising that these projection patterns of Gad1-Cre and Vgat-Cre labeled neurons did not overlap. As noted above, we have removed the Gad1-Cre tracing from the study which we hope fully mitigates the reviewer’s concern.

3. How did authors differentiate from projection fibers (axons) and projection targets (synapses)? Similarly, how did authors differentiate between fibers and cell bodies?

We now describe these criteria more clearly in the text and have added immunohistochemistry to bolster the interpretation of axonal varicosities as synaptic endings (Figures 1, Figure suppl 1-1,1-5 Figure 3, Figure 3 suppl 1). Morphological features, namely swellings along branches of coursing axons helped differentiate axons from synapses. These areas also colocalized markers for neurotransmitter transporters Vglut2 or Vgat. Fibers could be more easily differentiated from cell bodies based on morphological features. While automatic sorters based on fluorescence intensity may sometimes misattribute axons and somata, trained human observers, in my experience, rarely make such attribution errors, which was our approach.

4. When quantifying "distribution of labeled cells", did the authors observe more unique projections in mice with a lower proportion of labelled cells in the IntA? "Distribution of labeled cells" also may be an ambiguous phrase, as it suggests that the observed proportion of IntA (for example) neurons is labeled according to some map. Perhaps a better title would be "Location of labeled cells." Apologies if I am missing something, in the end it just seems like a leap in the way these data are presented and discussed.

This is a great point. We now plot where we saw terminal label as a function of injection site, with color coded plots in Figures 1, 2, 3, and 4. Not surprisingly, the location of terminal label was sensitive to injection site, which we discuss more in depth in the text. We note that even in most restricted injection to anterior interposed in Vgat-Cre mice, we observed axonal varicosities outside of IO. We also thank the reviewer for pointing out that our terminology was confusing. We calculate the fraction of injection site neurons that appear within the cerebellar nuclei and neighboring regions, retitled “Location of injection site” (Figures 1,2,3,4). We now describe more clearly how smaller and smaller injections localized most cleanly in IntA ramify relative to injections that include neurons nearby.

5. It is unclear how input regions were identified, this is especially important given that twice as many Gad1-Cre mice were injected than Ntsr1-Cre mice and that the number of input cells appears a magnitude lower in the Ntsr1-Cre mice.

Vgat-Cre (N=3) and Vglut2-Cre (N=3) replace Gad1-Cre and Ntsr1-Cre, and are N matched. Input regions are simply areas where retrogradely labeled neurons were observed. We clarify this now in the Methods. We think the large difference in input numbers derived from the low numbers of starter neurons in Ntsr1-Cre mice. Vglut2-Cre mice do not show such a stark difference and we now make only one observation about differential innervation of these cells relative to Vgat-Cre, namely that Vgat-Cre cells may be preferentially innervated by neurons of the lateral reticular nucleus.

How many rabies-infected cells were labelled for each injection (per mouse and per group)?

We now quantify the number of starter cells which averaged 156 in Vglut2-Cre and 307 in Vgat-Cre mice. Note that we have modified this experiment by replacing Gad1-Cre mice with Vgat-Cre and included Vglut2-Cre with Ntsr1-Cre to provide a firmer sense of convergence ratios.

Does the input region need to be found in all mice from one group, or just one? What is the variability in input region?

We report all data – so input regions seen in only one mouse are included; such variability would be reflected in the quantification of percent of inputs with error bars in Figure 4 -figure sup1. The raw data are provided in Supplementary file 4.

And how did the authors control for non-specific labelling? Did they perform any (specifically the rabies tracing) in Cre-negative mice? (methods suggest yes, but results are not show). Please clarify and add the details to the manuscript.

We thank the reviewer for pointing out that this was unclear. As alluded to in the methods and now better described and shown, we used rabies in Cre-negative mice as reported for controls. In brief, in Cre-negative mice, rabies (and other recombinase-dependent methods) show little to no label anywhere in the brain (0-8 cells at injection site, and analysis of sections across the brain showed no label anywhere else), bolstering our interpretation of cellular label as “real” signal. Injection of the full complement of viruses – starter viruses plus modified rabies – into wildtype mice resulted in vanishingly little label. This was not comparable to the Cre-dependent tracing which labeled ~60-100s of neurons.

6. Authors should consider including some of the low magnification whole mount images of the supplemental data into the main text figures. They provide a better picture of the data than the cropped images.

This is a great suggestion and indeed very nicely highlights the clear differences in projection patterns observed in the midbrain between Vgat-cre and Ntsr1-cre mice. Low magnification images are shown in Figure 1, 3, and 5. In 5G, side-by-side comparisons are provided for these two genotypes at low magnification, for the reader to appreciate the stark difference in axon tract patterns in the midbrain.

Statistics comments:

1. N-number is of neurons is inflated in Figure 3. Statistics should be compared for averages on each mouse, not on total number of cells. The authors seem to perform multiple pairwise comparisons without correcting for multiple comparisons.

We now add per-animal means for Figure 3 cell size data and perform comparisons on means of animals with posthoc tests that account for multiple comparisons.

Textual comments:

1. Line 113: put abbreviation brackets direction behind "nucleus."

done

2. Line 114: define "RN."

done

3. Line 118: define "ZI."

done

4. Line 119: define "VM, VPM, VAL"

done

Define all abbreviations in the text first time they are used.

done

5. Line 116: "a morphological… excitatory neurons." This observation is best left to when figure 5A is discussed, as experimental approach or data describing excitatory neurons has not yet been discussed at this point.

Reorganization has fixed this.

6. Line 124-126: "Vgat neurons generally targeted these regions more strongly than Gad1 neurons." How does this relate to the number of infected neurons in the IntA? This is also a qualitative statement that is not backed by statistical tests.

Gad1-Cre data has been removed, and point is taken.

7. Line 149-155: should authors refer to supplemental figures 2 and 3 here? I would suggest for the authors to order the supplemental figures in the order that they are discussed in the text.

Fixed.

8. Line 179-182: this method may warrant a little more explanation for readers that are not familiar with intersectional viral mapping.

Good point. We have added more explanation.

9. Line 182-184: "Injections in wildtype…the nuclei." The authors need to include images of these control experiments in the manuscript.

We now show injections into wildtype mice in Figure S10.

10. Line 224-225: what tests do these p-values represent?

Thank you for catching that this was only in the figure legend; We have re-run the statistics as suggested elsewhere and now report the test used in the main text as well as figure legend (One-way ANOVA; Tukey’s test for multiple comparisons.)

11. Line 234-235: "we validated…" This is not a validation of your Cre-lines.

Please see response to Major point 1 above.

12. Line 238-240: "was within the noise of presumptive non-overlapping neurons." Presumed by whom? Almost 20% overlap in signal seems pretty high when you are trying to define cell-type specific projection maps. Please clarify.

We have replaced analysis of publicly available data from the Allen Brain Atlas (which showed 10 micron brightest point projections which we think accounted for the “overlap” seen), with in situ hybridizations performed with our own mouse lines which show overlap between 1.4-14% of cross-phenotype label in each genotype. As noted above, we removed Gad1-Cre entirely from the study and replaced Ntsr1-Cre with Vglut2-Cre for monosynaptic rabies retrograde tracing.

13. Line 255-257: "Together, these results…in the brainstem." I think that this conclusion is not warranted based on the data that is shown. See comments 11 and 12 and General comment 1.

For points 11-13, we have addressed these limitations with RNAscope in situ hybridization and immunostaining as described more thoroughly in the main critique section. Axonal varicosities from Vgat-Cre mice colocalized with Gad65/67 in multiple targets besides IO. We found this line to be highly specific to Vgat neurons. As a result, we maintain our conclusion that Vgat neurons of interposed and nearby interstitial groups project to multiple targets in the brainstem.

14. Line 279: define "nIntA". Why was it chosen to name the putative excitatory neurons based on their Cre-line but the putative inhibitory neurons based on the assumption they are inhibitory (iIntA)? In the end, too many assumptions about the identity of the different neurons were made.

We have changed the nomenclature to reflect this excellent point. We now use the terms: Int-Vgat, Int-Ntsr1, IntIO-Vgat and IntRN-Ntsr1 to refer to the various datasets.

15. Line 289-291: "We currently are unable…neurons project to IO." This is another reason to validate the Cre-lines.

Done.

16. Line 291-292: "Exclusive retrograded labeling…ventral brainstem." If this is the case, why can you perform retrograde labeling from IO to iIntA? Are there no iIntA terminals just dorsal to the IO? The supplemental figures do not clarify this.

The targets of these IntIO-Vgat neurons are largely restricted to IO, BPN, RN and SPVi. We now draw this distinction out more clearly in the text.

17. Cite Supplemental Figure 4 in text.

Done

18. Line 333-334: "We conclude it….information broadly." An alternative conclusion is that the majority of Ntsr1-Cre positive cells projects to the RN. I do not think you can conclude that all Ntsr1 cells project everywhere (this is not tested).

We clarify that our data suggest that the Ntsr1-Cre neurons that project to RN also project to other locations, but not necessarily everywhere. We take the point that every neuron might not ramify widely and clarify this point in the text.

19. Line 366-369: Figure 5B only shows proportion of projections. It does not show number of regions targeted or relative strength of these proportions. I think that a qualitative statement can be made that only Ntsr1-Cre neurons project to modulatory regions but the author sshould refrain from making quantitative statements based on proportions of injections that are inherently variable between cell types and injections.

We now clarify that projection strength is a semiquantiative metric describing the density of projections, on which we perform statistical tests to support our claims that Int-Ntsr1 targets modulatory regions more than Int-Vgat neurons (See Methods, Projection Quantification section).

20. Line 398-409: Why are these data not shown?

These data, for the retro-Cre injection to RN are now included in a extended figure: Figure 3—figure supplement1.

21. Line 438: how many starter cells did you observe in Gad1 vs Ntsr1 mice?

See response to Point 5

22. Line 450-451: did the authors test this statistically?

We have removed this claim and analysis (referring originally to denser PC projections to Ntsr1-Cre cells).

23. Line 441-462: why did the author change the way they referred to Gad1 neurons?

Nomenclature has been fixed.

24. Line 460-461: "this patterned… cell types." In previous figures it was shown that iIntA do not heavily project to modulatory regions, whereas nIntA neurons do. Therefore, this pattern does not mirror targeted patterns.

With the replacement of the monosynaptic rabies datasets with new Cre driver lines, we find only one statistically significant difference in input patterns between Vgat and Vglut2-Cre cells – those from the lateral reticular nucleus. Thus, we have removed the specific details noted by the reviewer, but addressed the underlying concern with statistical analysis.

25. Line 535-537: partial difference in projections do not preclude partial overlap in Gad1-Cre and Ntsr1-Cre expression.

Indeed. Gad1-Cre has been eliminated and antibody validation has been used to validate Ntsr1-Cre projections.

Methods comments:

1. What are in the atlas coordinates for the injection sites. Please provide them all.

All injection coordinates were reported in the Animals section and may therefore have been difficult to find. We have moved them to the Viral injections section.

Figure comments:

1. Please put the abbreviations used in each figure in its figure legend.

Done

2. Please be consistent in the use of "Cre" (capitalized and with an "e") in all figure legends.

Done

3. Label IntA (and other cerebellar nuclei) in larger cross-sections for consistency.

We have added a panel to figure 1 indicating boundaries of nuclear subdivisions. Otherwise, as indicated in the text, we use Franklin and Paxinos and Sugihara and Shinoda 2007, to define boundaries.

4. Figure 1, use similar schematic of viral injection.

Done.

5. Figure 1, please be consistent in labeling of anatomical areas in the figure panels: "Vgat" and "Gad1" above figure panels, and abbreviations for anatomical locations within the figure panel.

We have done our best to unify styles of reporting.

6. Figure 1B and 2B, are the Vgat infected cells always more medial than Gad1 infected cells? Is this a problem of injection variability? If so, how do you control for this variability in the different projection maps?

We now plot the locations of the injection sites and have removed Gad1-Cre from the study. We did not observe systematic differences in the location of injection sites across types of tracing.

7. Figure 1C-E, boxes in schematics are not same size but figure panels seem similar magnification. Boxes on schematics are a little confusing. It may help to label the anatomical location of interest using color and naming it.

We have replaced the schematics with low magnification images and boxes now indicate regions with 20X views to the right. We have elected to refer to boxes with lowercase roman numerals. Region abbreviations are provided in all fields of view.

8. Figure 1D, what is the difference between "PSV" and "V" in left images?

We have added lines to denote boundaries between adjacent regions when there might be ambiguities, eg Figure 1Giv

9. Figure 2, what does Con/Fon mean?

This is now defined and described thoroughly in the text:

“Next, to restrict label to genetic- and projection-specific Int neurons (Fenno et al., 2018), we used a two-recombinase-dependent reporter virus (AAV8-hsyn-ConFon-eYFP) injected into Int in conjunction with Flp recombinase retrogradely introduced via IO with AAVretro-EF1a-Flp. The fluorescent reporter will only express in the presence of both Cre and Flp recombinases. This Cre-on Flp-on approach, termed “Con/Fon”, was used to isolate IO-projecting Int-Vgat neurons.”

10. Figure 2, please keep order of Gad1 and Vgat injections consistent with figure 1 for comparison.

Dealt with in revision.

11. Figure 3, the statistics for this figure should be rerun, see statistical comments.

Done.

12. Figure 4E, see comment 6.

Done.

13. Figure 4E, 4th row, there are 3 boxes here.

Fixed

14. Figure 5 B, please keep color-coding of groups consistently throughout the paper.

We have tried to address this.

15. Figure 6C, provide legend.

We have added a key to the panel (now 6D). Good point.

16. Figure 6D, consider plotting this data on two images with separate left oriented Y-axis. This figure panel is very unclear.

We have not included this panel in the revised manuscript.

17. Figure 6F, the leftward and rightwards histograms have been used in other figures to denote ipsi- and contra-lateral projections. Changing this is confusing. It may be beneficial to plot these in the same orientation.

With respect, we have retained the leftward-rightward plotting here but have made the distinction between left and right more transparent and easier to understand with clearer axis labels.

18. Figure 6G, this figure is too small to see and the difference between top panels and bottom panels is unclear.

We have removed this set of panels.

19. Figure 7, please keep the order in which the Gad1 and Ntsr1 labelled neurons are shown consistent with figure 6 for comparison (keep this order similar with supplemental figure 5 and all figure legends too).

We have done our best to improve parallelism.

20. Figure 7, retrograde rabies labeled cells are hard too to see, please increase the magnification for input neurons.

We have done our best to maintain comparable magnification across many panels and balance showing cell groups vs single cells.

21. Figure 7, if I understand correctly, the results for each column are from 4 different injection paradigms/experiments that are also summarized in other figures. It would help the reader if the authors would include schematics for each experiment (include: mouse line, injection side, and type of virus used).

With respect, we think adding the schematics makes this already very large figure more compressed and harder to see the data. We have worked to clarify what is shown better in the legend, and the remaining should orient the reader.

22. In supplemental figure 2, make lines of boxes in inset thicker. Box for Gvi is missing. Box for Hiv is missing. B has no scale bar.

We unfortunately don’t have a great deal of latitude with the boxes because they are added through the microscopy system at the time of capture. We hope the reproduction process maintains the legibility of the boxes. We have fixed remaining issues.

23. In supplemental figure 3, make lines of boxes in inset thicker. Explain scale bars in Figure legend.

Please see response to point 22.

24. In supplemental figure 4, make lines of boxes in inset thicker. Label boxes in E-K. Also, B has no scale bar.

Done, except see note 22.

25. Table 1, please write out full name of each nucleus in column 1.

Done.

Reviewer #2 (Recommendations for the authors):

1. For Figure 1, please clarify how much the distinctions in projection distribution could be due to variability in specificity of initial targeting or recombination ability of Vgat-Cre and Gad1-Cre? The conclusions can be strengthened by analyzing the starter distribution of IntA neuron subtypes compared to vestibular or cerebellar cortex like Figure 2C.

The main concern here is that selective targeting of IntA using Cre-lines that recombine in all 3 subnuclei is a potential weakness in methodology since this relies entirely on precision of stereotaxic injection.

We thank the reviewer for this keen insight and concern. For all injections, we quantify and plot “location of injection label”. We note in the text where the size of the injection site or specific details of the injection location may account for observed patterns. Of note, injections entirely contained within IntA, and devoid of Purkinje label, were found to project to the pontine nuclei (reticulotegmental and pontine gray), the vestibular nuclei and very sparsely to the parabrachial nucleus. Notably, these injections did not label nucleocortical Vgat terminals. Nucleocortical projections, along with many of the other sensory brainstem label, appeared when injections included neurons of the interstitial nuclei – a loosely organized set of cell groups ventral to the interposed nucleus. We now discuss this observation and draw connections with other previous literature on this topic in both the Results and Discussion.

2. Questions about similarities or differences between Vgat-Cre and Gad1-Cre recombination or targeting by CN neurons cannot be resolved by in situ hybridization analysis of Vgat and Gad1 in the CN alone (Figure 1A). Perhaps more informative if authors could provide Vgat-Cre; reporter expression with Gad1 in situ and vice versa. Also, it is difficult to see signals in the small panels, and the overlay seems to indicate a lot of heterogeneity. Quantification of Gad1 and Vgat in situ would be informative.

We agree with the reviewer have now performed a variety of validation steps that have led to the elimination of Gad1-Cre datasets from the study. We refer the reviewer to our response to Reviewer 1, Point 1 for a thorough description of our validation approaches and findings, prompted by these reviews, including quantitated in situ labeling.

3. Line 177, please describe how labeling of Purkinje cells might affect interpretation. Perhaps clarify that PC labeling might lead to VEST and PBN labeling from Purkinje cell axons directly projecting out of the cerebellum.

We thank the reviewer for raising this point and now include mention of these caveats in the text. Please see our response to Point 1. In addition, we note that in even the most restricted Vgat-Cre IN injection, with no Purkinje neurons labelled, we observed axons entering and ramifying in the medial vestibular nucleus and PB.

4. In Figure 3A, the prediction is that the Gad1-Cre more generally labels Gad1 projections and Gad1IO labels a more restricted subset since using IO retrolabeling. But in 3A, it seems that Gad1IO picked up neurons that were not picked up by Gad1 anterograde labeling, specifically the larger neurons. Text argues perhaps Gad1 labels more interneurons than Gad1IO, but should not exclude Gad1 picking up some of these large neurons (maybe just less). Please explain.

The reviewer is right that this was a puzzling observation. We have eliminated the Gad1-Cre datasets entirely and can only speculate that the peculiar behavior was a consequence of the promiscuous expression of Cre that we observed in this mouse line. We refer the reviewer to more intuitive outcomes of these approaches in Vgat-Cre and Ntsr1-Cre intersectional label.

5. Within IntA, what % of vGluT2 neurons does Ntsr1 make up? If Ntsr1 is in almost all, then the result in Figure 4 says almost all nIntA neurons projections are collaterals of RN-projecting nIntA neurons. But if Ntsr1 is in a much smaller subset of IntA vG2 neurons, then this says something different. Analyzing the % of Ntsr1 neurons that are vg2 in the IntA is needed to resolve this.

We thank the reviewer for this point. We can’t make any claims about the percent of Vlgut2 neurons that Ntsr1 neurons make up because we don’t expect that our virus labels all Ntsr1-Cre cells. We chose to focus on whether the viral labeling approach was specific to Vglut2 or Vgat neurons, and rather than on the fraction of Ntsr1 probe localized with Vglut2 mRNA probes. We therefore back off of claims that collateralization is widespread from individual cells but includes many regions when looked at in aggregate from RN-projecting Ntsr1-Cre cells.

6. Figure 5A, please provide quantification of the morphology to strengthen the reported differences (PCA analysis?).

We appreciate the reviewer’s points here and have addressed the roots of these issues in the following ways. First we better highlight morphological distinctions between projection types in RN and the cerebellar cortex between Vgat-Cre and Ntsr1-Cre projections with new images, but we do not use these as criteria for differentiation, just to point them out qualitatively. We are unaware of PCA methods that could be used to differentiate the morphological features. We note that mossy fiber boutons and the beaded varicosities of the inhibitory nucleocortical projection described by Ankri and observed here are fairly clear distinctions. We have also tried to provide clearer images differentiating the fine structure of the Vgat projection to RN and the differences between the fine caliber axons

For IO, wondering if the morphology of CN terminals could fit into the morphological schemes proposed by Vrieler et al., 2019 Brain Structure and Function 224. In other words, do distinct IntA neurons project to specific subtypes of IO neurons.

This is such an interesting point. However, because we do not fill IO neurons, addressing whether distinct IntA neurons project to specific subtypes of IO neurons is beyond the scope of this study. Importantly, we do not make any claims about this point.

7. Please show images of the cerebellar cortex in Ntsr1 and Gad1 to show starter cells to address the following questions. For Ntsr1, how much extra-CN recombination does Ntsr1-Cre mediate? For Gad1, I would like to get a sense of how many inhibitory neurons (interneurons and Purkinje cells) are labeled initially. Both are important to determine how to interpret input specifically into CN or whether the authors are looking at input into cell types elsewhere in the cerebellum.

We now show and quantify ‘starter’ label for Vglut2-Cre as well as Vgat-Cre mice. As you can appreciate the starter label is located in the cerebellar nuclei and not in the cerebellar cortex. There are no reported direct inputs to Purkinje cells from extracerebellar sources, partially mitigating this concern. However, we only rarely observe Golgi cells labeled with rabies, thus we think that the preponderance of extracerebellar label is from projections to the CN.

Another way to ask this same question, in Figure 6D, what are the cerebellar cortex Ntsr1 starter cells? Granule cells or maybe even inhibitory neurons? Ntsr1-Cre recombines in excitatory neurons in the CN (Houck and Person 2015), but could more generally label excitatory and inhibitory in the cerebellar cortex.

We have replaced Ntsr1-Cre with Vglut2-Cre in monosynaptic rabies tracing. There are not Vlgut2-positive cells in the cerebellar cortex.

And for Gad1 starter cells in the cerebellar cortex, are they mostly Purkinje cells or do these include Golgi/stellate/basket cells? The concern here is that if the rabies virus receive G and TVA from granule/Purkinje/Golgi/stellate/basket cells, then in addition to CN input, authors are looking at climbing and mossy fibers that project to non-CN cells confounding interpretation.

We address this question below Point 7. Beyond modulatory inputs, I am unaware of extracerebellar afferents that target Purkinje cells or MLIs. We noted vanishingly few Golgi cells that were rabies labeled. Please note too, that the differences in inputs between lines was not seen in the new Cre driver lines used for monosynaptic rabies tracing in this revision.

8. This is important for analysis and interpretation: how does one dissociate the GFP from rabies or GFP from oG? The double GFP strategy is a major concern here.

Originally, we count any neuron with both TVA and cell-filling rabies as a starter neuron, noting that it may or may not also express oG. We now include new data with rabies and oG components of monosynaptic labeling expressing different fluorophores, mitigating concern that we did not know if rabies labeled neurons were expressing oG or whether they were labeled via a monosynaptic jump. In the few experiments retaining the original method, oG is easily differentiated from GFP by being histone restricted, thus it labels only a small segment of the nucleus whereas GFP is a cell fill.

9. Figure 6B, starter cells are defined as intersection between GFP and mCherry, but in B, there is very little overlap here mostly because red channel is so faint. For instance in Gad1, I can see 2 red cells and 4 green cells but with only 1 yellow cell. I am concerned about the small number of starter cells.

Same issue as Figure 6, in Figure 7B, worried that the input sources represent more than just input sources for CN, but for cell types in the cerebellar cortex especially given high percentage indicated in 6D.

There was very little overlap of red and green channels in the cerebellar cortex, mitigating this concern. We now show better images to appreciate this observation, although the TVA and oG channels is not very photogenic. The large numbers of cells in the cerebellar cortex labeled in from injections (pointed out for 6D) is because Purkinje neurons constitute a major input to the CN.

10. For Figure 7A, please provide ways to distinguish axons from dendrites.

This is a good point and we recognize the reviewers’ concern. We now note in the figure text now that we show terminal fields that contain axonal varicosities. We show colocalization of terminals with markers for neurotransmitter transporters in (Figures 1, Figure suppl 1-1,1-5 Figure 3, Figure 3 suppl 1).

Reviewer #3 (Recommendations for the authors):

(1a) Interrelationships between the cell types used are not clear. The authors show overlap between Ntsr1+, Gad1+/Vgat+, and VGluT2+ neurons (FigS1B). Does this mean that Cre-dependent AAV tracing for Gad1Cre+/VgatCre+ neurons could also label VGluT2+ neurons? Does the Ntsr1Cre+ cell type contain inhibitory neurons? If so, how did the authors verify that the axonal projections are actually inhibitory or excitatory? Please discuss the technical limitations.

We thank the reviewer for pointing out these important caveats. We have added a variety of datasets to address the points raised here. Between in situ hybridization with Cre-dependent reporter viruses and immunolabelling of terminals against either Vglut2 or Gad65/67, we have strengthened the argument that our methods isolate specific subtypes of the cerebellar nuclei. Please see responses within Essential Revisions and Reviewer 1 Point 1 for more details.

(1b) Relatedly, it would be helpful to make clear which inhibitory cell types are taken into account and discussed: Gad1-IO, Gad1-collateral, Gad1-GABA-local, Gad1-glycine-local, Gad1-glycine-projection? How did the authors allocate VgatCre+ neurons? What is the overlap between Gad1Cre+ and Vgat1Cre+ neurons (they look different because labeled VgatCre+ neurons are in the medial part of the IntA and labeled Gad1Cre+ neurons are in the central part of the IntA consistently throughout Figures)? Something like Venn diagrams for Gad1Cre+, VgatCre+, and Ntsr1Cre+ neurons, could help.

As discussed above, we have eliminated the use of the Gad1-Cre line in this study, which may mitigate some concern. We use better terminology to differentiate Vgat neurons labeled from direct injections vs projection-specific label. We also plot and report location of injection sites across all injection types to ensure there are not systematic differences in location of injections across experiments. We do not focus on local inhibitory cell types nor specifically glycinergic cells, though our Vgat terminals colocalized with Gad65/67.

(2a) I see substantial and consistent injection leak to non-IntA and non-CN regions. How did the authors identify that the connections were really of IntA but not from other regions, for example, Vest? For example, inconsistent with our knowledge are that (1) IntA projections to DCK and PO of IO,

We now more directly relate injection site to targets observed, acknowledging the possibility, when appropriate, of projections that could derive from tracer injection location, plotted now in Figures 1,2,3,4. We note when results hold when injection sites are more restricted.

2) IntA projections to several targets (please compare Table 1 with Teune et al., 2000, Prog Brain Res),

We cannot account for all differences, other than to simply report what we observe. We see projections that differ from Teune et al., 2000 even in spectacularly clean injections into IntA (to ipsilateral IO, for example). For other differences, we speculate that label in the interstitial cell groups may account for differences, as this is seen to account for some differentiatinon in PIN targets (T82 in Teune). We now discuss the interstitial cell groups much more thoroughly as the possible source of some of the Vgat projections we observe, as well as the nucleocortical Vgat pathway, which we observed when interstitial cell groups were labeled but not when label was restricted to IntA.

3) CrI/PFL Purkinje projections to IntA neurons, which are known to IntP and LN,

We have de-emphasized IntA specifically and note most starter cells in interposed, with only a minority in LN and interstitial groups. We plot the location of starter cells.

4) sparse RN projection to Ntsr1Cre+ IntA neurons,

We see a sparse RN projection to Vglut2-Cre neurons. This was reported in Beitzel et al., 2017 as well, to Ntsr1-Cre cells.

and (5) very dense vestibular projections to Ntsr1Cre+ IntA neurons.

We now note and cite literature in the text that review vestibular projections that may contradict these findings. (Barmack 2003).

Inconsistent within the manuscript is that projection specific tracing revealed more projection targets than non-specific tracing (Table 1). Please discuss these and other relevant technical issues.

We saw this in the Gad1-Cre tracing (between Int-Gad1 and IntIO-Gad1) but not in any other projection specific tracing. In removing the Gad1-Cre driver line from the study we hope the reviewer’s excellent concern is mitigated.

(2b) Again, given the technical challenges, the Gad1+/Vgat+ axons labeled in non-IO regions could be explained by injection leak to non-CN regions or to excitatory neurons. On what evidence can the readers be convinced of "inhibitory projections" that are "from IntA"?

These are very important points. In addition to quantification of injection site label in each dataset, we have amended our claims to reflect the point that our injection sites at times include other cerebellar nuclei. On whole, we now do not emphasize anterior interposed, though we do discuss in sections when we observe terminals correlating with spillage into other nuclei. Of note, we see terminal varicosities in the tegmental reticular nucleus and pontine gray from Vgat-Cre injections even when the tracer is totally limited to IntA, thus we are confident that these projections are not simply a consequence of tracer leakage.

3) I see potential issues in viruses used:

1) efficiency for the rabies tracing was significantly different between NtsrCre (yielded ~150 cells) and Gad1Cre neurons (yielded ~4000 cells),

We have replaced both Cre driver lines in the revision. Nevertheless, there is about a 10x difference between number of starter cells in the Vgat-Cre than Vglut2-Cre line. We note this in the text. Given that with the new Cre lines there is very little difference in the pattern of inputs to the two cell classes, we are not concerned that the differences in starter number account for the data.

2) rabies tracing didn't identify known input from IO,

The lack of IO in some specimen is indeed a puzzling feature, and anecdotally is seen by other groups doing similar experiments, as discussed at meetings, possibly due to unknown tropism. With the newer driver lines, we do see IO label in 4/6 specimen. Because we do not see many cells, the input percent is low, thus these known reciprocal projections do not appear in the reciprocal projection map in Figure 7. As has been noted with monosynaptic rabies methods, negative results cannot be interpreted to suggest that specific areas do not project to a given target, but positive results have been validated (Wickersham et al., 2007).

3) Cre-independent TVA expression from Cre-dependent AAV has been reported to be a significant problem in the rabies tracing scheme used in this study (Fagat et al., 2016, Cell Rep),

We thank the reviewer for this concern. We now show control injections in wildtype mice with vanishingly little label suggesting that this caveat is not likely contributing to the main conclusions (Figure S10). We also now include linked TVA-oG construct. Unfortunately we could not find the referenced study to determine if there were other concerns associated with the specific virus used.

and (4) how did the authors control retrograde infection of AAVs? Please discuss the technical limitations.

Retrograde infection of AAVs is easily distinguished by somatic label outside the injection location. We very rarely observed retrogradely labeled neurons in the brainstem and midbrain following our virus injections, suggesting little capacity for terminal label to be misinterpreted. We now note this in the text.

4) I don't see any evidence for input to VTA.

We show this more clearly in Figure2—figure supplement 1, with varicosities amongst TH positive cells in the region of the VTA.

5) Table 2 is missing.

This may have been a typo referencing Supplementary File 3. We thank the reviewer for catching this. We have ensured the tables are all provided.

6) 'long-standing dogma' -- I don't think so. Although it has been 'widely assumed', the existence of collaterals of preolivary neurons has been looked for many times and remains an open question.

We thank the reviewer for this important point. We have removed reference to ‘long-standing dogma’ and replaced it with ‘widely assumed’. In addition, we introduce the question as one that has received empirical support but has nevertheless not been followed up on with cell type specific viral tracing methods (See beginning of Results).

7) Difficult to see axonal terminals in photos in general.

We appreciate the concern. We now include data showing colocalization with Vglut2 or Gad65/67 immunostaining, as described above for R1 and R2 in (Figures 1, Figure 1 figure suppl 1-5 Figure 3, Figure 3 suppl 1).

8) Cerebellar nuclear outputs are powerfully modulated by their upstream Purkinje cells. In this sense, I think it's worth discussing the organization of PC inputs to inhibitory neurons. Can you discuss more about implication of broader Purkinje innervation of inhibitory IntA neurons? Are these different than the parasagittal modules? Or, does the difference in number/location of labeled PCs for inhibitory vs excitatory DCN neurons simply reflect the technical limitation in rabies tracing?

We too were interested in this point, but after replacing the Gad1-Cre datasets with Vgat-Cre, we no longer observed clear evidence of broader innervation of Vgat-Cre cells relative to Vglut2-Cre cells. Thus, we do not have reason to suggest the integrative patterns would differ across cell types.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1 (Recommendations for the authors):

In this revised manuscript, Judd et al., have performed essential experiments to validate their Cre-mouse lines. They have incorporated additional mouse lines based on their findings (and previous reviewer comments) and repeated specific experiments in these mouse lines. The observation that inhibitory neurons in the interposed nuclei send widespread projections throughout the brainstem holds and indeed the data present an interesting and important set of findings. However, below are suggestions about the text and figures that will aid the reader in understanding the detailed anatomical data presented in this manuscript.

– The overlap in dots and lines in Figures 1D, 2E, 3D, and 4E make these figures hard to interpret. Some of the projections included in the results may be specific to the injection site outside of the interposed nucleus (off-target viral expression) and this possibility is discussed briefly in the text. In addition, all the regions in which projections are found, even if in just a few viral replicates with the least specificity, are included in the summary of Figure 7B. This should be clearly stated in the text.

We now point out the inclusion and exclusion criteria for regions listed in Figure 7B – reminding the reader that these include only the strongest and most consistently labeled regions, but nevertheless may reflect projections of interstitial cell groups or other Cb nuclei labeled as described in the earlier figures.

In addition, we suggest clarifying the panels in 1D, 2E, 3D, and 4E. One option would be to replace these plots with UpSet plots (for example) that show the overlap between the different data sets. Alternatively, the current graphs could be kept and the authors could instead include a ratio of the number of replicates in which the projection was observed. In the end, any way to help the reader more easily appreciate the data would be highly beneficial.

We completely agree with the reviewer’s critique of the figure panels that were intended to map the location of projections for each animal. The overlapping scores across replicates made it difficult to see/track the specific pattern of projection targets for individuals, however. Rather than doing an UpSet plot, which would discard information on projection strengths, we elected to offset each specimen slightly from one another and give each specimen a unique symbol. This allows the reader to track each individual while also keeping available the information on injection site precision and projection strengths. We think this edit should fully address the reviewer’s astute observation about the limitations of the previous iteration of these plots.

– The authors state "… dorsal to IntA, but nucleocortical terminals that were included in the projection analysis were not located in the same topographical area." Apologies, but I do not understand what you mean by this. What topographical area are you referring to? Please consider rephrasing.

Reviewer 3 also noted this confusing text and suggested it be removed, so this verbiage is now gone.

– The authors state "Because injection sites did not differ systematically across…". What do mean? What is your measure that gives you confidence in regards to being systematic?

We have replaced the term ‘systematically’ with ‘qualitatively’. The main idea is that if we look at the injection site plots across all the injection types, we do not see that they are categorically different – that is, we could not decipher what the injection scheme was just from looking at the location of the injection site. Nevertheless, we see the Reviewer’s point that the term ‘systematically’ could imply more than we intend and therefore hope the wording change clarifies our point.

Reviewer #2 (Recommendations for the authors):

In the revised manuscript, the authors have more than sufficiently addressed all my questions and concerns (and those of the other reviewers) about the interpretations of the distinctions of input and output patterns by different Int neuronal subtypes and how the data was generated. The replacement of Ntsr-Cre/Gad1-Cre with vGluT2-Cre/vGAT-Cre mice for monosynaptic rabies tracing clarifies and strengthens the conclusions for the input-output connectivity by excitatory and inhibitory Int neuronal subtypes.

This is a carefully conducted study that describes exciting findings about the anatomical organization of cerebellar interposed anterior nucleus, especially the inhibitory subpopulation. These findings will undoubtedly provide important ground work for future investigation of the functional relevance of distinct Int neuronal subtypes and pathways in dexterous and locomotor movements, and beyond.

I very much enjoyed reading and reviewing this manuscript and congratulate the authors on a well conducted study.

We sincerely appreciate these kind words and again express gratitude for the excellent suggestions.

Reviewer #3 (Recommendations for the authors):

The manuscript is significantly improved for clarity by the authors' thorough hard work. The existence of ramifying inhibitory projection neurons from Int CN became more convincing. Particularly, analyses on Gad/Vglut immunoreactivity of axonal terminals are helpful.

Now it became clearer that the manuscript adds the widespread ramifying cells to the existing repertoire of inhibitory cerebellar nucleus neurons. The authors establish this cell type by demonstrating the inhibitory signature of projected axons and then place them into context in comparison with well-known types of cells regarding outputs, inputs, and cell morphology. During these efforts, the authors also made clear outputs of a subset of excitatory cells (Ntsr1+ cells) and discovered monosynaptic inputs to the excitatory and 'inhibitory' cells, which themselves are novel and intriguing.

I only have one major comment for improving the clarity of conclusions.

The above efforts made me realize that the current experimental designs do not distinguish or identify the ramifying cells from the widely known IO-only cells (I tentatively call them like this though they could have their own collaterals), which may be important for one to incorporate this cell type into cerebellar theories, although distinction from a subset of excitatory cells is clear. As far as I can see, only sparse and highly restricted labeling in IO in Figure 4-S1D would suggest that IO projections from the ramifying cells are made with a different topographical rule than IO-only cells, which is assuring because these results show a potential distinction between inhibitory cells.

Specifically, in the current manuscript, results for cell morphology, axonal trajectory, and input circuits demonstrated in Figure 5, 6. and 7, which are currently treated as data for the ramifying cells, could simply reflect those of IO-only cells and could barely reflect the ramifying cells. This is because the injection strategies utilized do infect both IO-only and the ramifying cells (To selectively target ramifying cells by avoiding IO-only cells, something like Cre-dependent retrograde infection from RN/PG/TRN in Vgat-Cre mouse would be required). It must be made clear in the Result that these analyses do not distinguish them. This limitation should influence conclusions regarding Figure 5-7 and I recommend modifying them accordingly. Similarly, Line 83-84, Line 376 may also be misleading. It would be clearer to state that what this study identified is "inhibitory projection cell type(s) that (or, at least some of which) collateralizes to IO" rather than "IO projection cell type that also collateralizes to other areas".

We have now added modifiers and text in a variety of places that states more clearly that our data indicate that at least some IO-projecting neurons collateralize and that we cannot distinguish neurons that solely project to IO, and that our data do not preclude their existence. Eg: Lines 84; 177-178; 379; 444-445.

Also, upon reflecting on this point, we have elected to tone down the title a touch so as not to unintentionally mislead the reader that all inhibitory projections are widespread. To this point, we have replaced ‘widespread’ with ‘diverse’. We hope this minor edit also addresses the spirit of the concern raised here.

https://doi.org/10.7554/eLife.66231.sa2

Article and author information

Author details

  1. Elena N Judd

    Department of Physiology and Biophysics, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8211-5140
  2. Samantha M Lewis

    Department of Physiology and Biophysics, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, United States
    Contribution
    Data curation, Investigation, Methodology
    Competing interests
    No competing interests declared
  3. Abigail L Person

    Department of Physiology and Biophysics, University of Colorado School of Medicine, Anschutz Medical Campus, Aurora, United States
    Contribution
    Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing - original draft, Writing - review and editing
    For correspondence
    abigail.person@cuanschutz.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9805-7600

Funding

National Institute of Neurological Disorders and Stroke (114430)

  • Abigail L Person

National Science Foundation (1749568)

  • Abigail L Person

Simons Foundation

  • Abigail L Person

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Courtney Dobrott for sharing expertise in spinal cord removal, RNAscope methodology, and helpful comments on the manuscript. We thank Aya Miften for assistance in histology and Daniel Heck for work on preliminary datasets. We are grateful to Dr Jason Aoto for the preparation of AAVRetro viruses. This work was supported by NS114430, NSF CAREER 1749568, and by a grant from the Simons Foundation as part of the Simons-Emory International Consortium on Motor Control. Imaging of rabies label was performed in the Advanced Light Microscopy Core, part of the NeuroTechnology Center at University of Colorado Anschutz Medical Campus supported in part by Rocky Mountain Neurological Disorders Core Grant Number P30 NS048154 and by Diabetes Research Center Grant Number P30 DK116073 with the assistance of Dr. Radu Moldovan. Rabies viruses were obtained from the Salk GT3 Core Facility supported by NIH-NCI CCSG: P30 014195, an NINDS R24 Core Grant and funding from NEI.

Ethics

All procedures followed the National Institutes of Health Guidelines and were approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus under protocol #43, Laboratory of Abigail Person, re-approved 11/2020. Every effort was made to minimize suffering.

Senior Editor

  1. Ronald L Calabrese, Emory University, United States

Reviewing Editor

  1. Roy V Sillitoe, Baylor College of Medicine, United States

Reviewers

  1. Roy V Sillitoe, Baylor College of Medicine, United States
  2. Albert I Chen, Nanyang Technological University, Singapore
  3. Hirofumi Fujita, Johns Hopkins University, United States

Publication history

  1. Preprint posted: January 1, 2021 (view preprint)
  2. Received: January 4, 2021
  3. Accepted: September 19, 2021
  4. Accepted Manuscript published: September 20, 2021 (version 1)
  5. Version of Record published: September 30, 2021 (version 2)

Copyright

© 2021, Judd et al.

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

  • 490
    Page views
  • 81
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Roni O Maimon-Mor et al.
    Research Article Updated

    The study of artificial arms provides a unique opportunity to address long-standing questions on sensorimotor plasticity and development. Learning to use an artificial arm arguably depends on fundamental building blocks of body representation and would therefore be impacted by early life experience. We tested artificial arm motor-control in two adult populations with upper-limb deficiencies: a congenital group—individuals who were born with a partial arm, and an acquired group—who lost their arm following amputation in adulthood. Brain plasticity research teaches us that the earlier we train to acquire new skills (or use a new technology) the better we benefit from this practice as adults. Instead, we found that although the congenital group started using an artificial arm as toddlers, they produced increased error noise and directional errors when reaching to visual targets, relative to the acquired group who performed similarly to controls. However, the earlier an individual with a congenital limb difference was fitted with an artificial arm, the better their motor control was. Since we found no group differences when reaching without visual feedback, we suggest that the ability to perform efficient visual-based corrective movements is highly dependent on either biological or artificial arm experience at a very young age. Subsequently, opportunities for sensorimotor plasticity become more limited.

    1. Cell Biology
    2. Neuroscience
    Shahzad S Khan et al.
    Research Advance

    Activating LRRK2 mutations cause Parkinson's disease, and pathogenic LRRK2 kinase interferes with ciliogenesis. Previously, we showed that cholinergic interneurons of the dorsal striatum lose their cilia in R1441C LRRK2 mutant mice (Dhekne et al., 2018). Here, we show that cilia loss is seen as early as 10 weeks of age in these mice and also in two other mouse strains carrying the most common human G2019S LRRK2 mutation. Loss of the PPM1H phosphatase that is specific for LRRK2-phosphorylated Rab GTPases yields the same cilia loss phenotype seen in mice expressing pathogenic LRRK2 kinase, strongly supporting a connection between Rab GTPase phosphorylation and cilia loss. Moreover, astrocytes throughout the striatum show a ciliation defect in all LRRK2 and PPM1H mutant models examined. Hedgehog signaling requires cilia, and loss of cilia in LRRK2 mutant rodents correlates with dysregulation of Hedgehog signaling as monitored by in situ hybridization of Gli1 and Gdnf transcripts. Dopaminergic neurons of the substantia nigra secrete a Hedgehog signal that is sensed in the striatum to trigger neuroprotection; our data support a model in which LRRK2 and PPM1H mutant mice show altered responses to critical Hedgehog signals in the nigrostriatal pathway.