Skip to main content
NCBI home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2023 Oct 1.
Published in final edited form as: Nat Methods. 2023 Mar 2;20(4):580–589. doi: 10.1038/s41592-023-01787-1

Combining long-term circuit mapping and network transcriptomics with SiR-N2c

Hassal Lee 1,&,#, Ernesto Ciabatti 1,#,#, Ana González-Rueda 1, Elena Williams 1, Fiona Nugent 2, Souradip Mookerjee 3, Fabio Morgese 1, Marco Tripodi 1,#
PMCID: PMC7614628  EMSID: EMS176613  PMID: 36864202

Abstract

An exciting frontier in circuit neuroscience lies at the intersection between neural network mapping and single-cell genomics. Monosynaptic rabies viruses provide a promising platform to converge circuit mapping methods with -omics approaches. However, three key limitations have hindered the extraction of physiologically meaningful gene expression profiles from rabies-mapped circuits: 1. inherent viral cytotoxicity, 2. high viral immunogenicity and 3. virus-induced alteration of cellular transcriptional regulation. These factors alter the transcriptional and translational profiles of infected neurons and their neighbouring cells. To overcome these limitations, we applied a self-inactivating genomic modification to the less immunogenic rabies strain, CVS-N2c, generating a self-inactivating CVS-N2c rabies virus (SiR-N2c). SiR-N2c not only eliminates undesired cytotoxic effects but also substantially reduces gene expression alterations in infected neurons and dampens the recruitment of innate and acquired immune responses, thus enabling open-ended interventions on neural networks and their genetic characterization using single-cell genomic approaches.

Introduction

The anatomical organization of connected neurons into functional motives is one of the crucial aspects underpinning brain function1, 2 and substantial efforts have been made to reconstruct whole connectomes in various species3, 4. Equally important to the ontogenesis of animal behaviour is the understanding of the genetic diversity among individual neurons participating in neural circuits and how such diversity contributes to circuit function and specificity5. Recent efforts have been undertaken to elucidate the genetic composition of individual neurons to an unprecedented scale6, 7. However, one aspect still missing is the ability to link these two lines of research, e.g. whole-brain circuits mapping and single-cell genomics, and to do so at scale. One of the most promising approaches to bridge this gap would be to combine virus-based circuit tracing methods with single-cell RNA profiling. With respect to virus-based circuit mapping, monosynaptic tracing using glycoprotein-deleted rabies virus (ΔG-Rabies) 8, 9 is the most adopted approach as it allows the labelling of a large fraction of neurons presynaptic to a targeted neuronal population in the absence of prior hypotheses of the circuit topology 10, 11.

However, experimenters face three critical problems when trying to combine genomics and virus-based circuit mapping: the first is the inherent cytotoxicity of rabies-based methods 9, 10, 12; the second is the triggering of immune responses 13; the third is the direct effect that viral protein expression has on global endogenous transcription and translation 14, 15.

While most recent efforts have focused on solving the first of these three problems by attenuating viral cytotoxicity12, 16, 17, the other two points remain largely untouched. In particular, RNA viruses tend to elicit a strong and early innate immune response and affect the transcriptional machinery to sustain their transcription at the expense of endogenous transcripts 13–15, 18, affecting any attempt to collect meaningful physiological “-omics” data.

In this work, we report a method to address these three problems at once by developing an engineered Rabies vector. Different Rabies strains show various degrees of neuroinvasiveness, immunogenicity, and pathogenicity 18–20. A general trend indicates that more pathogenic strains exhibit increased neuroinvasiveness and decreased immunogenicity 20, 21. The SAD-B19 Rabies vaccine virus was the first strain to be engineered for monosynaptic spreading 8, 22, but its inherent toxicity has limited its use mostly to short-term experiments 8, 10–12. To reduce the toxicity of the G-deleted rabies SAD-B19 (ΔG-Rabies-B19), a self-inactivation genomic modification was introduced 16, 23, resulting in the Self-inactivating Rabies virus (SiR-B19), which enables long-term labelling of circuits in vivo 16, 23, 24.

Another approach pursued to limit cellular toxicity in rabies tracing experiments was to develop a G-deleted variant of a challenge strain of the rabies virus (CVS-N2c). By nature of being a challenge strain, ΔG-Rabies-N2c exhibited decreased cytotoxicity compared to the ΔG-Rabies-B19 12.

Here, we apply the SiR genomic modification 16, 23 to the CVS-N2c strain to generate a self-inactivating rabies virus (SiR-N2c). This virus shows improved neuronal survival in vivo and has a negligent effect on endogenous transcription, triggers a reduced innate immune response, and can be used to study the single-cell transcriptomic profile of a transsynaptically traced neural network.

Results

SiR-N2c enables the long-term survival of virally labelled neurons

In earlier work, we engineered a rabies vector containing a proteasome targeting domain fused to the rabies nucleoprotein that switches off its transcription shortly after primary infection 16. Here, we wanted to export this approach to the CVS-N2c strain, reasoning that using an inherently less toxic strain would further improve the technology.

We introduced a TEVs-PEST modification to the C-terminus of the nucleoprotein of the ΔG-Rabies-N2c virus (Fig. 1A) generating a self-inactivating version of the N2c challenge strain (SiR-N2c) from cDNA 16 (Fig. 1B). Given earlier reports of potential genomic instability of self-inactivating strains 25, we sequenced all produced viruses prior to their use in vivo. Sanger sequencing of individual genomic copies prior to injection confirmed that the engineered sequence remained unaltered during production. Subsequent RNA-sequencing of the viral genomes from infected brain tissue also confirmed SiR genomic stability in vivo, consistent with our earlier report 23 (Extended Data Fig. 1).

Fig. 1. SiR-N2c enables long-term survival of rabies virus-infected neurons.

Fig. 1

Open in a new tab

(A) Design of ΔG-Rabies-B19, SÌR-B19, ΔG-Rabies-N2c and SiR-N2c vectors. (B) Scheme of SiR-N2c production protocol from cDNA. (C) ΔG-Rabies-N2c or SiR-N2c viruses expressing CRE recombinase were injected into the CA1 of YFP CRE reporter mice followed by longitudinal monitoring of infected cell numbers. (D) YFP+ neurons survival over time normalized to 1 week time point (mean ± SEM, n = 3 animals per time point; SiR-N2c: one-way ANOVA, F = 0.2, P = 0.92; ΔG-Rabies-N2c: one-way ANOVA, F = 34, P = 0.0005). (E) Representative images of infected hippocampi with SiR-N2c or ΔG-Rabies-N2c at different times p.i‥ (F) Confocal image of hippocampus infected with SiR-N2c at 6 months p.i., with zoomed images of labelled neurons in CA1 (G) and CA3 (H). Scale bars 100 μm in E, 500 μm in F, 50 μm in G and H.

We then quantified virus-driven cytotoxicity in vivo by comparing the total number of SiR-N2c and canonical ΔG-Rabies-N2c rabies-infected neurons over time in the hippocampal CA1 region (Fig. 1C) 16. Injection of CRE recombinase-expressing SiR-N2c or ΔG-Rabies-N2c into reporter mice resulted in activation of YFP expression in hippocampal neurons at 1 week post injection (p.i.) (Fig. 1D-E). In line with our earlier results 16, 23, we found no significant decrease of YFP positive (YFP+) cells for up to six months p.i. in the hippocampi of SiR-injected animals (one-way ANOVA, F = 0.2, P = 0.92, Fig. 1D-H). Conversely, the unmodified ΔG-Rabies-N2c virus exhibits progressive cytotoxicity and loss of YFP+ neurons over one month (oneway ANOVA, F = 34, P = 0.0005, Fig. 1D-E).

The CVS-N2c strain induces lesser innate immune responses than the SAD-B19 strain in vivo

Rabies virus challenge strains display delayed cytotoxicity and increased neuroinvasiveness than vaccine strains by evading the host immune response12, 20, 21. To directly compare CVS-N2c and SAD-B19 strains, we checked the activation of the immune response elicited by SiR and ΔG-Rabies infection in cell culture. Viral sensor RIG-I 26 was upregulated in all conditions when compared with the SiR-N2c (Extended Data Fig. 2). Downstream effector interleukin 6 (IL6) induction levels suggest a trend in the degree of innate immune responses with ΔG-Rabies-B19 > SiR-B19 > ΔG-Rabies-N2c > SiR-N2c (Extended Data Fig. 2).

We compared the in vivo immune response of wild-type ΔG-Rabies strains by injecting them in CA1 of the hippocampus of mice. We assessed the transcriptional levels of RIG-I and two downstream interferon-stimulated genes, IFIT1 and Mx1 at 1week p.i. (Fig. 2A). Rabies infections led to upregulation of RIG-I-driven interferon responses when compared to PBS-injected controls, with SAD-B19 strain inducing a 3-fold higher RIG-I activation and almost double the response of downstream genes than CVS-N2c (Fig. 2B). We characterised the activation of RIG-I upon infection of all four viral strains by immunostaining, which shows an ordered degree of staining intensity with ΔG-Rabies-B19 > SiR-B19 > ΔG-Rabies-N2c > SiR-N2c (Fig. 2C, D), in line with the in vitro data. Notably, ΔG-Rabies-B19 injection triggers widespread RIG-I upregulation in neurons and glial cells compared to ΔG-Rabies-N2c (Fig. 2E-N). SiR viruses show significantly decreased RIG-I responses compared to their respective parental ΔG-Rabies (Fig. 2D, E-N).

Fig. 2. CVS-N2c strain attenuates activation of innate viral sensor RIG-I in vivo.

Fig. 2

Open in a new tab

(A) PBS, ΔG-Rabies-B19 or ΔG-Rabies-N2c were injected in CA1 and 1 week p.i. total RNA was extracted from the hippocampi for RT-qPCR. (B) mRNA fold change of immune response genes (RIG-I, IFIT1, MxP) normalised to ΔG-Rabies-N2c condition. (mean ± SEM, n = 3, one-way ANOVA with Tukey correction for multiple comparisons; left to right: p = ****4.1e-6, ****5.1e-5, **3.3e-3e-3, ***8.0e-4). (C) YFP CRE reporter mice were injected in CA1 with rabies viruses expressing CRE recombinase for immunofluorescence studies. (D) RIG-I activation in stained hippocampi at 1 week p.i. measured as the RIG-I-positive area normalised to PBS-injected controls. (mean ± SEM, n = 6, two-tailed, unpaired Student’s T-test with Welch’s correction; left to right: P = ***1.5e-4; *1.6e-2; *2.3e-2). (E-N) Representative images showing RIG-I (mΔGenta), YFP (green, virally infected cells) and DAPI (blue) stained hippocampi injected with PBS (E-J), ΔG-Rabies-B19 (F-K), SiR-B19 (G-L), ΔG-Rabies-N2c (H-M), SiR-N2c (I-N). (J-N) Higher resolution images. Scale bars: 200 μm top panels, 20 μm bottom panels.

SiR-N2c exhibits attenuated immunogenicity and no detectable alteration in synaptic genes expression

To characterise the physiological response to ΔG-Rabies-N2c or SiR-N2c infection in vivo, we injected ΔG-Rabies-N2c, SiR-N2c or PBS in the hippocampal CA1 region and collected the tissue 1 week p.i. for total RNA sequencing. We detected significant gene expression alteration upon ΔG-Rabies-N2c virus injection with 895 RNA transcripts upregulated and 173 transcripts downregulated in comparison to PBS-injected controls (Fig 3A-C, Extended Data Fig. 3A). Gene ontology of upregulated transcripts revealed immunoregulatory processes and activation of both the innate and adaptive immune response (Extended Data Fig. 3B-D). In line with our earlier analysis (Fig. 2), ΔG-Rabies-N2c infection upregulated viral detection molecules, such as pattern recognition receptors (PRRs) of the RIG-I like receptor (RLR), toll-like receptor (TLR) and NOD-like receptor (NLR) families and their downstream pro-inflammatory signalling pathways (e.g. IRF1, 5, 7, NFkB1, 2, IFNAR, IFNGR, Stat2-4 and Jak2, 3; Supplementary Table 1). IFN-induced cytokines were also upregulated (e.g. CXCL10, CCL5; Supplementary Table 1). At 1 week p.i., the enrichment of innate immune response genes was more prominent than the adaptive immune response ones (Fig.3D, Extended Data Fig. 4) in line with previous work 19–21, 27.

Fig. 3. SiR-N2c infection attenuates the immune response to viral infection and does not impair hippocampal synaptic function.

Fig. 3

Open in a new tab

(A) Table showing the number of up- or downregulated transcripts following ΔG-Rabies-N2c or SiR-N2c infection. (B) Volcano plot of genes found to be differentially regulated following ΔG-Rabies-N2c injection. Genes that are significantly misregulated following SiR-N2c injection are shown in red. (DESeq2 with cut-off point of p<0.05 after Benjamini and Hochberg multiple testing correction). (C) Number of biological processes, pathways, molecular functions, cellular components and protein classes GO categories that are statistically overrepresented in up- or downregulated gene lists upon ΔG-Rabies-N2c or SiR-N2c injection. (Fisher's exact test with Benjamini and Hochberg multiple testing correction) (D) Dot plot of the top pathways and protein classes enriched in up- and downregulated genes by ΔG-Rabies-N2c, with correspondent data for SiR-N2c (Fisher's exact test with Benjamini and Hochberg multiple testing correction). No dot is displayed when there is no statistical enrichment.

In addition, several genes involved in specific neuronal processes were significantly downregulated (Fig.3C-D, Extended Data Fig. 3B-D). Among them, we found genes involved in synaptic function, such as activity-regulated cytoskeleton-associated protein (ARC) and neurexin I (NRXN1) (Supplementary Table 2). These results show that, despite its inherent lower immunogenicity, ΔG-Rabies-N2c infection significantly alters the neuronal transcriptional profile, including critical synaptic components, as early as 1 week p.i‥

In contrast, SiR-N2c infection induced significantly fewer transcriptional changes in the infected area. Only 49 transcripts were upregulated and 4 downregulated (as opposed to 895 upregulated and 173 downregulated transcripts upon ΔG-Rabies-N2c infection) (Fig. 3A-B). Consequently, a reduced number of pathways and biological processes were significantly enriched after gene ontology analysis (Fig. 3C-D). Most of the upregulated transcripts were identified under the immune process ontology (37 out of 49) and were one order of magnitude fewer than those of the parental ΔG-Rab-N2c (37 versus 390) (Fig. 3A-B and Supplementary Table 1). We found a reduced number of major histocompatibility complex genes upregulated upon SiR-N2c versus ΔG-Rab-N2c infection (4 versus 13). Notably, GFAP and Iba1, two markers of astrocytic and microglial activation 28, 29, were upregulated following ΔG-Rabies-N2c but not SiR-N2c infection (Fig. 3B). Importantly, we did not identify any synaptic protein genes within SiR-N2c downregulated transcripts (Fig. 3A,B,D and Supplementary Table 2). This indicates that in addition to its reduced cytotoxicity, SiR-N2c also largely preserves the transcriptional landscape of infected cells in contrast to the parental ΔG-Rabies-N2c.

Glial immune responses are reduced upon SiR-N2c infection compared to ΔG-Rabies-N2c

We further investigated the glial immune response surrounding SiR-N2c infected hippocampal neurons by immunohistological analysis and compared it to ΔG-Rabies-N2c and ΔG-Rabies-B19 viruses (Fig. 4A). We used immunostainings against Iba1 and GFAP proteins to identify activated microglia and astrocytes, respectively (Fig. 4B-G). Quantification of Iba1+ and GFAP+ cells revealed that compared to the unmodified variants (ΔG-Rabies-N2c and ΔG-Rabies-B19), SiR-N2c infection resulted in a decreased number of activated glia in the hippocampus. Compared to ΔG-Rabies-N2c, the SiR counterpart showed a 17-fold reduction of activated microglia and 32-fold reduction of activated astrocytes. Compared to ΔG-Rabies-B19, the SiR-N2c virus resulted in a 234-fold decrease in activated astrocytes following infection of hippocampal cells. The fewer activated microglia and astrocytes following SiR-N2c infection were also smaller in area (4-fold reduction) (Fig. 4D,G). These results are in line with our RNA sequencing data (Fig. 3B) as Ibal and GFAP were significantly upregulated following ΔG-Rabies-N2c but not SiR-N2c infection.

Fig. 4. SiR-N2c infection induces lower glial immune response compared to two strains of ΔG-Rabies.

Fig. 4

Open in a new tab

(A) ΔG-Rabies-B19, ΔG-Rabies-N2c or SiR-N2c expressing CRE were injected in CA1 of reporter mice and animals sacrificed 1 week p.i. for immunofluorescence studies. (B) Hippocampal slices stained for microglial marker Iba1 (magenta, microglia) and YFP (green, virally infected cells). Graph displaying quantification of Iba1+ activated microglia number (C) and area (D) (mean ± SEM, n = 3, two-tailed, unpaired Student’s T-test with Welch’s correction; left to right, in C: P = n.s 0.45; **3.2e-3; n.s. 0. 21; left to right, in D: P = n.s 0.13; ****1.0e-5; n.s. 0.85). (E) Hippocampal slices stained for astrocyte marker GFAP (magenta, astrocytes) and YFP (green, virally infected cells). Graph displaying the quantification of GFAP+ activated astrocyte number (F) and area (G) (mean ± SEM, n = 3, two-tailed, unpaired Student’s T-test with Welch’s correction; left to right, in F: P = *4.5e-2; *4.3e-2; n.s. 0.67; left to right, in D: P = *2.9e-2; n.s 0.21; n.s. 0.54). Scale bars 100 μm.

SiR-N2c retains improved peripheral neurotropism and transsynaptic spreading capabilities

Rabies virus challenge strains, such as CVS-N2c, are more neuroinvasive than rabies vaccine strains12, 30. Nonetheless, ΔG-SAD-B19 vector neurotropism has recently been improved by complementing it with an optimized glycoprotein (oG)31. Hence, we compared the peripheral neurotropism of SiR-N2c-CRE and of oG-coated SiR-B19-CRE following injection into the quadriceps of postnatal day 3 (P3) pups (Extended Data Fig. 5A). SiR-N2c infected around 10-fold higher number of neurons (Extended Data Fig. 5B), indicating that SiR engineering does not affect the superior peripheral neurotropism of the CVS-N2c strain12.

We then tested SiR-N2c ability to transsynaptically spread from motor neurons following peripheral injection10, 11. We co-injected AAV-G-nucFLAG with SiR-N2c-CRE into the quadriceps of P3-5 CRE-reporter pups (Extended Data Fig. 5C). This led to extensive transsynaptic labelling (tdTomato+/nucFLAG- neurons; Extended Data Fig. 5D-G), with a ratio of premotor to motor neurons labelled in line with what was previously reported for the parental ΔG-Rabies-N2c12. We also confirmed a significant reduction of microglia activation compared to that induced by ΔG-Rabies-N2c (Extended Data Fig.6).

To enable SiR-N2c use for transsynaptic tracing in central neural circuits, we pseudotyped the virus with the chimeric EnvA glycoprotein8 and compared its performance to ΔG-Rabies-N2c. We first injected the nucleus accumbens (NAc) of CRE-reporter mice with an AAV expressing TVA and G. TVA expression allows the selective targeting of starter cells by an EnvA-pseudotyped ΔG-Rabies vector8, 12, 16. We retargeted the NAc with either SiR-N2c-CRE(EnvA) or ΔG-Rabies-N2c-CRE(EnvA) (Fig. 5A, Extended data Fig. 7A) to identify presynaptically labelled neurons in cortical and subcortical input regions 32, such as the basolateral amygdala (BLA) (Fig. 5B). We first confirmed the EnvA and G dependency of, respectively, infection and transsynaptic spreading (Extended data Fig. 7B-J). We then assessed the transsynaptic spreading performance of SiR-N2c over the parental virus. At 10 days p.i. we observed a similar number of traced neurons in ΔG-Rabies-N2c and SiR-N2c injected brains (Fig. 5B-D). Notably, by 4 weeks p.i. the number of detected labelled neurons was more than halved in the BLA of ΔG-Rabies-N2c injected animals (Fig. 5C-D), indicating cytotoxic effects by ΔG-Rabies-N2c at this time point (Fig. 5B). Conversely, the number of SiR-N2c retrogradely labelled neurons in the BLA increased by 4 weeks p.i. (Fig. 5C-D).

Fig. 5. SiR-N2c retains transsynaptic spreading capabilities in the brain without altering neuronal physiology.

Fig. 5

Open in a new tab

(A) CRE-reporter mice were injected in the NAc with AAV-TVA-G followed by re-targeting with either EnvA-pseudotyped SiR-N2c-CRE or ΔG-Rabies-N2c-CRE 3 weeks later. (B) Representative images of whole brains of Rosa-LoxPSTOPLoxP-tdTomato mice showing SiR-N2c and ΔG-Rabies-N2c-CRE transsynaptic labelling in presynaptic regions, with zoomed panels of tdTomato+ neurons in the BLA at 10 days and 4 weeks p.i‥ (C) Representative images of retrogradely traced neurons in BLA region of Rosa-LoxPSTOPLoxP-nucGFP mice with SiR-N2c and ΔG-Rabies-N2c-CRE at 10 days and 4 weeks p.i‥ (D) Quantification of nucGFP+ neurons in the BLA of injected mice at 10 days and 4 weeks p.i‥ (mean ± SEM, n = 4 for ΔG-Rabies-N2c, and SiR-N2c at 10 days p.i., n = 6 for SiR-N2c at 4 weeks p.i; two-tailed, unpaired Student’s T-test with Welch’s correction; left to right: P = *2.4e-2; **7.9e-3) (E). Scheme of the strategy for transsynaptic labelling of presynaptic cortical neurons projecting to the hippocampus for whole-cell patch clamp recording. (F) Membrane potential response to steps of positive and negative current of SiR+ and uninfected control neurons at 4 weeks p.i. (n = 15 and n = 16, respectively) (G-J) Action potential (AP) amplitude and half-width (G), resting membrane potential (RMP) and AP threshold (H), input resistance, and firing frequency oat increasing steps of positive current (J) for SiR+ neurons and uninfected control neurons at 4 weeks p.i. (mean ± SEM; two-tailed, unpaired Student’s T-test in (G-I), two-way ANOVA in (J); P = n.s 0.45 AP amplitude, n.s. 0.45 AP half-width in (G); n.s. 0.58 RMP, n.s. 0.48 AP threshold in (H); n.s. 0.67 in (I); F = 1.64, P = n.s. 0.20 in (J)). Scale bars: 1 mm for whole brain (B), 50 μm for zoomed images (B), 250 μm in (C).

Next, we transsynaptically traced cortical neurons from the hippocampus using SiR-N2c-CRE(EnvA) (Fig. 5E) and compared the electrophysiological properties of SiR-targeted and neighbouring uninfected cortical neurons at 4 weeks p.i. using whole-cell patch-clamp mode recordings. The two neuronal populations showed regular spiking profiles (Fig. 5F), with no significant difference between the two neuronal populations in the action potential amplitude and half-width (Fig. 5G), resting membrane potential and action potential threshold (Fig. 5H), input resistance (Fig. 5I), and instantaneous firing frequencies (Fig. 5J).

Transcriptomic profiling of transsynaptically traced neurons at single-cell resolution with SiR-N2c

Single-cell transcriptomics allows the investigation of the genetic heterogeneity of neuronal types within the nervous system33, 34 Despite several lines of evidence suggesting that genetics and synaptic connectivity properties conjunctively define neuronal subclasses35, 36, combining single-cell RNA sequencing with transsynaptic mapping remains challenging13. Therefore, we tested if SiR-N2c can be used to overcome current limitations. We focused on the spinal network to take advantage of the detailed recent molecular characterization of spinal cord neurons 37–39. We co-injected AAV-G-nucFLAG with SiR-N2c-CRE into the quadriceps of P3 CRE-dependent Sun1-GFP reporter pups to transsynaptically trace spinal premotor neurons40 (Fig. 6A). Labelled nuclei were recovered by fluorescent activated cell sorting (FACS) at 10 days p.i. for snRNA-seq. Neuronal transcriptomes were subclassified into neuronal subtypes using the SeqSeekClassify neural network41. SeqSeek combines single-cell transcriptomic data from several studies of the postnatal mouse spinal cord and provides a pipeline to map sequenced nuclei onto known spinal cord neuronal types that fall onto defined spinal cord lineages (e.g. dorsoentral; Fig. 6B-H). These neuronal transcriptomes are shown projected onto a dimension-reduced UMAP space along with the reference spinal cord data set (Fig. 6B-G).

Fig. 6. SiR-N2c allows single cell transcriptomic analysis of the transsynaptically labelled premotor network in the spinal cord.

Fig. 6

Open in a new tab

(A) Experimental design for snRNAseq of transsynaptically targeted neurons: SunlGFP-reporter mice were co-injected intramuscularly with SiR-N2c-CRE and AAV-G-nucFLAG and sacrificed 10 days p.i.; single nuclei were extracted from labelled spinal cords and subjected to FACS for positive vs. negative GFP expression followed by 10x Genomic library preparation and NGS. Recovered nuclei from (B) SiR-N2c infected GFP+ cells and (C) uninfected GFP-cells were mapped onto D) a SeqSeek reference scRNAseq spinal cord atlas using Seurat’s FindTransferAnchors and MapQuery functions, and projected onto the reference UMAP space. Neuronal transcriptomes from infected SiR GFP+ cells (E), uninfected GFP- cells (F) and the reference dataset (G) were then selected for further subclassification into neuronal subtypes using the neural network classifier provided with SeqSeek. (H) A heatmap showing the proportions of neuronal subgroups identified in the uninfected GFP- spinal cord sample, monosynaptically traced SiR GFP+ premotor circuit and the spinal cord reference atlas. (Two-sided chi-squared test, P = ****3.20e-13). Neuron subtypes are functionally ordered and aligned to a dendogram of neuronal cell type relationships adapted from 41. The dendogram is licensed under a Creative Commons Attribution 4.0 International License. Minor changes were made to enable figure alignment (http://creativecommons.org/licenses/by/4.0/). MN motoneuron, IN interneurons (and projection neurons), CSF-cN cerebrospinal fluid contacting neurons, DE dorsal excitatory, DI dorsal inhibitory ME mid excitatory, MI mid inhibitory, VE ventral excitatory, VI ventral inhibitory, center represents a group of 3 cell types located near lamina X–the center of the spinal cord.

We detected an enrichment for specific neuronal subtypes within the virally traced premotor circuit (Fig. 6H, Extended Data Fig. 8). In the ventral spinal cord, the neuronal subtypes Inhib-27 and Excit-33 were most over-represented, while in the dorsal spinal cord, Excit-13 and Inhib-08 were most over-represented (Fig. 6H, Extended Data Fig. 8). The majority of the under-represented neuronal subtypes in the premotor circuit (e.g. Excit-16, 10 and Inhib-04, 06) were functionally located in the dorsal spinal cord, consistent with the known distribution of spinal premotor neurons11. The identification of known premotor mid-excitatory Satb1-expressing neurons (Excit-26) 11, 42 and of V1/V2b neurons (Inhib-27)43 provides further validation of the suitability of our approach. In addition to confirming the presence of previously identified premotor neurons, our results also reveal the nature of previously unidentified neuronal subclasses that contribute to the premotor network, such as the Syt2 (Excit-31) and dI2.1 (Excit-33) classes.

Next, we compared single-nuclei datasets from SiR-N2c and parental ΔG-Rabies-N2c traced spinal circuits. In accordance with our results in the brain, SiR-N2C infected neurons show reduced transcriptional dysregulation of neuronal function and structure compared to ΔG-Rabies-N2c infected neurons (Extended Data Fig. 9). Thus, SÌR-N2C virus can be used in conjunction with 10x single nuclei sequencing to transcriptomically subclassify virally labelled neurons.

Discussion

A growing interest is building up around the idea of combining single-cell genomics34, 44, 45. with structural and functional connectomics. The use of viral tools, and the rabies virus in particular, seems a promising method for the combination of transsynaptic tracing with single-cell genomics. However, the cytotoxicity of the rabies virus, combined with its high immunogenicity and its perturbation of cellular gene expression, has largely limited its use for genomic studies13–15. Here, we developed and characterized a variant of self-inactivating rabies virus based on the challenge N2c strain that exhibits improved neuronal survival and triggers a reduced immune response with a negligible effect on endogenous cellular transcription.

Through RNA-seq analysis, we show that despite the known reduced immunogenicity of rabies challenge strains 19–21, ΔG-Rabies-N2c induces substantial transcriptional changes in the infected neurons, most of them immune-related. ΔG-Rabies-N2c also induces downregulation of genes associated with synaptic function, highlighting the importance of considering the perturbing effects of viral infection on physiological neuronal properties 46. Importantly, the Rabies- driven misregulation of normal gene expression can be present as early as 1 week p.i., implying that reducing the length of the infection period before functional investigation of rabies-infected neurons may not always be a viable option.

On the other hand, the SiR-N2c virus that we describe here, on top of recapitulating the improved neuronal survival of the first generation of SiRs 16, 23, provides two major advantages. First, the virally induced immune response is markedly reduced. Secondly, SiR-N2c does not substantially perturb synaptic gene expression even at a time when the virus is transcriptionally active and shows a negligible change in the transcriptional landscape of the cells. Furthermore, the addition of the self-inactivating modification to the N2C strain does not affect its ability to spread transsynaptically.

SiR-N2c can hence be used to characterise the single-cell transcriptomic profiles of transsynaptically traced neurons. While the pilot experiment presented here was not done to a sufficiently large scale to provide a full description of the heterogeneity of the entire spinal premotor network, it provides a first window into the genetic composition of the spinal premotor network. In the future, this method could be extended to study the transcriptomic profile of neural circuits innervating different muscles11, the transcriptional remodelling of the spinal network following spinal cord injuries47, 48 or the transcriptomic effects of genetic mutations associated with neurodegeneration on premotor networks49. Given the possibility to generate SiR-N2c EnvA-coated particles capable of transsynaptic spreading, the approach can also be extended beyond spinal networks.

In conclusion, the SiR-N2C-based approach described here will help to synergise the functional, structural and genetic characterization of whole-brain neural networks and to marry connectomic and transcriptomic studies, potentially opening a frontier for network genomics.

Methods

Mice

All animal procedures were conducted in accordance with the UK Animals (Scientific procedures) Act 1986 and European Community Council Directive on Animal Care under project license PPL PCDD85C8A and approved by The Animal Welfare and Ethical Review Body (AWERB) committee of the MRC-LMB. All mice were bred and maintained in pathogen and opportunistic ΔGents-free conditions and monitored quarterly in the ARES Animal Facility in the Babraham campus. All experimental procedures were performed in the MRC-LMB Animal Facility. Animals were group-housed in a 12 hours light/dark cycle (7 a.m. to 7 p.m.), with temperature controlled at 19-23° C, humidity controlled at 45-65%, and with food and water ad libitum. Experiments in adult animals were performed on mice aged between 8 and 12 weeks. Intramuscular injections were carried out in pups aged P3-P5. The following mouse lines were used: WT C57BL/6 (The Jackson Laboratories: 000644), Rosa-LoxP-STOP-LoxP -tdTomato (The Jackson Laboratories: 007914), Rosa-LoxP-STOP-LoxP -YFP (The Jackson Laboratories: 006148) and Rosa-LoxP-STOP-LoxP-Sun1GFP (The Jackson Laboratories: 030952).

Cell lines

HEK293T, BHK-21 and mouse Neuro2A cells were purchased from ATTC. All cell lines were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and 100units/mL penicillin and 100μg/mL streptomycin unless otherwise specified. The cells were maintained in a humidified incubator with 5% CO2.

AAV production

For each viral preparation, HEK-293T cells were seeded onto ten 15-cm dishes. At 80% confluency, cells were transfected with AAV helper plasmid (Rep/Cap), AAV vector (with expression cassette) and pHGTI-adenol (adenoviral helper) using PEI. 60h after transfection, cells were dislodged from the plate and collected by centrifugation at 800 x g for 10 minutes. The cell pellet was then washed with 15mL PBS and resuspended in 10mL lysis buffer (150mM NaCl, 20mM tris pH 8.0, filter sterilized and stored at 4°C). Cells were subsequently freeze-thawed three times using dry ice/ethanol and a 37°C water bath. Cell lysates were dounce homogenized 20 times to destroy cell nuclei and put into a new falcon tube. MgCl2 and benzonase (Sigma) was added to a final concentration of 1mM and 250U/mL respectively and the mixture was incubated at 37°C for 10-15 min to dissolve genomic DNA/protein aggregates formed during freeze-thaw cycles. Cell debris was spun down at 2500 x g, 4°C for 20min and the viral supernatant was purified further using 17, 25, 40, 60% Iodixanol gradients in PBS, 1mM MgCl2, 2.5mM KCl (Optiprep, Sigma). 0.0005% phenol red was added to alternate layers for visualisation. Viral lysate (14-15mL) was added to the top of the gradient in a 36mL Optiseal tube (Beckman 362183) and centrifuged for 90min at 21500 x g (Beckman VTi50 rotor) at 16°C. The 40% viral fraction was harvested, diluted in PBS and centrifuged for 30min 2500 x g at 4°C (Eppendorf 5810R) in Millipore Amicon 100K columns (UFC910008) for a total of 3 times for removal of iodixanol and concentration of virus. 150-250μL of virus was recuperated after the last spin, aliquoted and stored at -80°C. The titer was determined by calculating viral genome copies from RT-qPCR using primer pairs for the CMV promoter, and serial dilutions of linearized vector plasmid for the standard curve. Purified AAVs were treated with DNase to remove residual plasmid DNA from purification (15 min, 37°C, DNAseI Roche followed by 10min inactivation at 95°C) and then incubated with proteinase K (Sigma Aldrich, 37°C, 15 min) before quantification with RT-qPCR.

SiR and rabies virus production

SiR viruses were produced using HEK-TGG packaging cells expressing rabies G (B19 or N2c), and TEV protease using a similar protocol as we previously described16. Briefly, HEK-293T were infected with Lenti-GFP-2A-G(B19 or N2c) and Lenti-puro-2A-TEVp and selected for puromycin resistance (1-2 ug/ml) and high GFP fluorescence by fluorescent activated cell sorting (FACS). Low passage HEK-TGG cells were always used for SiR production to ensure high level of TEVp activity23. SiR viruses were recovered from cDNA by co-transfection of plasmids containing the SiR genome together with plasmids encoding rabies protein (B19 or N2c) N, P, L, G and T7 polymerase as previously described50. High titre SiR preparation were obtained using BHK-TGG (B19 or N2c) generated by infection of BHK-21 cells with Lenti-GFP-2A-G(B19 or N2c) and Lenti-puro-2A-TEVp and selected in the same way as HEK-TGG lines. BHK-TGG were seeded in 15-cm plates and when at ~70-80% confluence, transduced using ~5 ml of low passage infectious supernatant obtained from cDNA in 10% FBS/DMEM and cells incubated at 37°C and 5% CO2. The following day, the medium was exchanged to 2% FBS/DMEM and cells incubated at 35°C and 3% CO2. Once confluent, cells were split 1:2 into 15-cm plates in 10% FBS/DMEM and incubated at 37°C and 5% CO2. Once ~80% confluent the medium was changed to 2% FBS/DMEM and incubated for 48 hours at 35°C and 3% CO2 for supernatant collection. After collection of viral supernatant, cell debris were removed by centrifugation at 4000 rpm for 10 min followed by filtration with 0.45 mm filter (Millipore, SLHV033RS). The virus was concentrated by ultracentrifugation as previously described 50. Rabies viruses pseudotyped with EnvA were produced as previously described using BHK-T-EnvA cells instead of BHK-EnvA cells16.ΔG-Rabies-N2c or B19 viruses were produced similarly to SiR viruses using HEK-GG cells (HEK-293T expressing the B19 or N2c G protein and GFP) and BHK-GG (BHK-21 expressing the B19 or N2c G protein and GFP)

Rabies virus titration

Functional titration of rabies virus was performed through in vitro infections with serially diluted viral preparations of HEK293T CRE-reporter cell line similarly to what described in previous reports12, 16, 51. CRE-reporter HEK293T cells were plated onto 6-well plates and at 70-80% confluency infected with serially diluted virus. After 48h viral infection was assessed by quantifying the CRE-activation of mCherry reporter by FACS. The number of infectious units per ml was calculated referring to the dilution giving a low multiplicity of infection (1-10%) by multiplying the number of infected cells by the dilution factor.

Single Particle rabies virus sequencing

RNA viral genomes were extracted using a Direct-zol RNA Miniprep kit (Zymoresearch R2051). The purified RNA genomes were converted to cDNA (Superscript IV, Invitrogen) using a barcoded primer that annealed to the 3’ rabies virus leader sequence and contained a random 8-nucleotide sequence on its 5’end 25. Thus, a unique random index was incorporated onto the 5’ end of each cDNA molecule corresponding to each individual viral particle’s RNA genome. The cDNA sequence was then amplified from the 5’ end through to the first half of the rabies virus P gene using Q5 Hot Start High-Fidelity DNA Polymerase (NEB) giving a PCR amplicon of approximately 2kb length. The amplicons were extracted from an agarose gel using a QIAquick Gel Extraction Kit and inserted into a standard cloning vector pBluescript KS (+) opened with restriction enzymes Kpnl and KbaI (NEBuilder® HiFi DNA Assembly). The plasmids were transformed into Stbl3 competent E. coli, and 48 clones were isolated and purified for sequencing. The sequencing procedure is summarised in Extended Data Fig. 1A. For each clone, the 8-nucleotide index and 3’-end of the N gene were sequenced. The sequences were aligned using Snapgene 4.1.9 with the reference sequence based on the plasmid pSiR-N2c-iCRE-P2A-mCherryPEST used for viral production. The primer sequences are listed in Supplementary Table 3.

Stereotaxic intracerebral injections

Adult mice aged between 6-12 weeks were used for stereotaxic intracerebral injections. Mice were anaesthetised with isofluorane (induction at 3% in 2L/min of oxygen, maintenance at 1-2% in 2L/min oxygen) and mounted onto a stereotaxic frame (David Kopf Instruments). Rimadyl (2 mg/kg body weight) was administered subcutaneously for anti-inflammatory perioperative care prior to viral injection. Viruses were injected at the following coordinates relative to bregma: CA1 region of the hippocampus 2.3mm AP, 1.65mm ML, 1.48mm DC; NAc -1.3 mm AP, 1.33 mm ML, 4.7 mm DC. A small 500μm diameter hole was drilled in the skull to allow viral injection with a 35G bevelled Hamilton NanoFil needle. 800nL of virus was injected. The needle was left in the brain 5-10min post-injection to prevent viral backflow following needle withdrawal. CVS-N2c viruses were injected at titres of ~1x107, we suggest not exceeding these titres for the SiR as it might result in saturation of the proteasome capacity.

Transsynaptic retrograde tracing in the brain

To assess retrograde transsynaptic viral spreading efficiencies, 400 nL of AAV-TVAeGFP-G(N2c) or AAV-TVAmCherry-G(N2c) were injected in the NAc of Rosa-LoxP-STOP-LoxP-tdTomato or Rosa-LoxP-STOP-Sun1GFP animals, respectively. After 3 weeks, TVA expressing cells were re-targeted with 500 nL of same titer SiR-N2c-CRE or control ΔG-Rabies-N2c-CRE EnvA-pseudotyped. At 10 days and 4 weeks p.i. brains were collected and sectioned at the cryostat (40 μm). Transsynaptic spreading and neuronal survival were assessed by imaging SiR or ΔG-Rabies infected neurons in the NAc (acquiring one every 4 sections) using a robot assisted Nikon HCA microscope mounting a 10x (0.45NA) air objective and fluorescent neurons counted using Nikon HCA analysis software. Specificity of transsynaptic labelling was assessed either with no injection or by injecting AAV-nucGFP-TVA in the NAc of Rosa-LoxP-STOP-LoxP-tdTomato mice prior to retargeting with SiR-N2c-CRE EnvA-pseudotyped.

Intramuscular injections

C57BL/6J WT or transgenic pups were fostered onto CD-1 foster mothers shortly after birth. At P3, pups were anaesthetized with isofluorane (induction at 3% in 3L/min oxygen, maintenance at 1-2% in 1-2L/min oxygen). A small incision was made in the skin over the quadriceps muscle and a rabies virus preparation or a viral mixture of AAV-G and rabies viruses was injected with a picospritzer (Parker Hannifin). For the comparison of B19 versus N2c motor neuron tropism, SiR-N2c and SiR-B19 were injected at a titre of 0.5-1x108 i.u./ml and 1-2x109 i.u./ml, respectively. Transsynaptic spreading with SiR-N2c was achieved by injecting a viral mixture of SiR-N2c at 0.5-1x108 i.u./ml and AAV at 3-5x1012 genomic copies/ml (g.c./ml) expressing N2c-G protein and a nuclear H2B-FLAG epitope for staining.

Sample Preparation

For immunohistochemistry experiments, mice were terminally anaesthesized with Euthatal (0.2mL) and transcardially perfused with 20mL ice-cold PBS followed by 20mL 4% paraformaldehyde (PFA).

For analysis of brain samples, brains were dissected out and fixed overnight in 4% PFA at 4°C. Post-fixation, samples were dehydrated in 30% sucrose/PBS overnight for cryoprotection. Subsequently, free-floating coronal sections (35 μm thickness) were cut from the brain using a freezing sledge microtome for immunofluorescence.

For spinal cord samples, spinal cords were dissected out and fixed for 4-8 hours at 4°C. Post-fixation, samples were dehydrated in 30% sucrose/PBS overnight for cryoprotection. Spinal cords were subsequently frozen in O.C.T. compound (VWR, Radnor, PA) and sliced into 30 μm sections using a CM1950 cryostat (Leica, Wetzlar, Germany). Sections were mounted on glass slides immediately after cryosectioning using an anti-roll blade and stored at -80°C.

Antibody Staining

Primary antibodies used in this work were: rabbit anti-RFP (1:2000, ABIN129578), goat anti-RFP (1:100, AB0040-200), chicken anti-GFP (1:2000, Aves GFP 1020), goat anti-Iba1 (1:200, NB 100-1028), rabbit anti-GFAP (1:500, Dako-Z0334), rat anti-FLAG (1:500, 637304), rabbit anti-RIG-I (1:100, ab45428), rabbit anti-NF-kB (1:400, 8242), rabbit phospho p38 MAPK (1:500, 4511L) and Chat (1:1000, Chemicon AB144P). The secondary antibodies used in this work were: donkey anti-chicken AF488 (1:1000, 703-545-155), donkey anti-rat AF488 (1:1000, A21208), donkey anti-rabbit Cy3 (1:1000, 711-165-152), donkey anti-goat Cy3 (1:1000, 705-165-147), donkey anti-goat Cy5 (1:1000, 705-175-147), donkey anti-rabbit Cy5 (1:1000, 711-175-152), goat anti-rat Cy5 (1:1000, 712-005-153).

All incubation steps were conducted at 4°C on a rocking platform for 24-48h and all antibodies were diluted in blocking solution (1% bovine serum albumin and 0.3% Triton X-100 in PBS) except for rabbit RIG-I which was diluted in 3% BSA, 10% donkey serum, 0.3% Tx. Between primary and secondary antibody staining, samples were washed four times with PBS at room temperature.

Image Acquisition and analysis

For cell counting experiments, spinal cord or brain sections were automatically detected and imaged on a robot-assisted Nikon High Content Analysis microscope equipped with a 10X air objective (0.45 NA) operated by NIS-Elements HC software (Nikon, Tokyo, Japan). For higher resolution images, fluorescence was visualised using a Leica SP8 inverted confocal microscope.

For cell longevity experiments in the brain, the whole hippocampus was sampled. For spinal cord experiments, the whole region of T5-S5 was sampled. In both cases, every fourth 35μm or 30μm thick section respectively was acquired on a Nikon High Content Analysis microscope and RFP or YFP positive cells manually counted using Nikon HCA software.

For quantification of immune response markers, a total of 10 sections were analysed for each animal. These sections covered the maximally infected hippocampal area. Positively stained immune cells were then quantified for total cell number and cell area using Nikon General Analysis software.

Electrophysiology

For electrophysiological recordings, AAV-TVA-G(N2c) was injected in the CA1 of one month-old Rosa-LoxP-STOP-LoxP-tdTomato followed 3 weeks later by SiR-N2c-CRE EnvA-pseudotyped injection. Recordings were performed at 1 month p.i., as we previously described16. Briefly, coronal hippocampal slices (350 μm) were prepared using a vibrating microtome (7000smz-2, Campden Instruments LTD, Loughborough, UK) in ice-cold sucrose-based cutting solution oxygenated with carbogen gas (95% O2, 5% CO2) and with the following composition (in mM): KCl 3, NaH2PO4 1.25, MgSO4 2, MgCl2 1, CaCl2 1, NaHCO3 26.4, glucose 10, sucrose 206, ascorbic acid 0.40, kynurenic acid 1. Slices were incubated at 37°C for 30 min in a submerged-style holding chamber with oxygenated artificial cerebrospinal fluid (aCSF; in mM: NaCl 126, KCl 3, NaH2PO4 1.25, MgSO4 2, CaCl2 2, NaHCO3 26.4, glucose 10) with an osmolarity adjusted to 280-300 mOsm/L and stored thereafter in the same holding chamber at room temperature for at least 1 hr. Slices were then individually transferred to the recording chamber and were superfused with oxygenated aCSF at room temperature at a flow-rate of approximately 2 mL/min.

Whole-cell current-clamp recordings were obtained from cortical neurons using 6-9 MΩ pipettes pulled from borosilicate glass capillaries (1.5 mm OD x 0.86 mm ID). Pipettes were filled with artificial intracellular solution containing (in mM): K-gluconate 150, HEPES 10, NaCl 4, ATP-Mg 4, GTP-Na 0.3 and EGTA 0.2; adjusted to pH 7.2 and osmolarity 270-290 mOsm/L. Data were recorded using an Axon Multiclamp 700B amplifier (Molecular Devices, Union City, CA, USA) and signals were low-pass filtered at 2 kH and acquired at 5 kHz using a digitizer (Axon Digidata 1550A, Molecular Devices, Union City, CA, USA) on a PC running pClamp.

RNA Extraction and cDNA library preparation

Mice were injected in triplicates with ΔG-Rabies-N2c or SiR-N2c viruses. 1 week p.i., mice were culled and the injected hippocampi dissected out for homogenization (Tissuelyser II, QIAGEN) and total RNA extraction with direct-zol RNA Miniprep kit (Zymoresearch R2051). The quality and quantity of purified RNA was analysed using Agilent RNA 6000 Pico chips on an Agilent 2100 Bioanalyzer. Subsequently, cDNA libraries were generated from total RNA using a TruSeq® Stranded Total RNA Library Prep Kit (Illumina, San Diego, CA) using low throughput single adaptors for each individual sample. The samples and pooled library were quantified using the KAPA Library Quantification Kit for Illumina Platforms (KR0405) and sequenced on an Illumina HiSeq 4000 machine (paired end, 150bp read length).

RNAseq data processing and analysis

RNA-seq sequence reads were trimmed using Trim Galore (version 0.4.2) with default parameters to remove the standard Illumina adapter sequences.

For differential expression analysis of host genes, RNA-seq reads were mapped to the GRCm38v95 mouse genome assembly using STAR version 2.5.3amodified167 and reads with MAPQ scores <20 were discarded. Downstream quantitation was conducted using the RNA-seq quantitation pipeline in SeqMonk software (http://www.bioinformatics.babraham.ac.uk/projects/seqmonk/). Differentially expressed genes were identified using the DESeq2 statistical filter with a cut-off point of p<0.05 after Benjamini and Hochberg multiple testing correction. Gene ontology (GO) enrichment analysis was conducted using the panther classification system 52 with PBS injected control RNA-seq transcripts used as the background list against which the target list was compared. Hierarchy diagrams were generated using the Gene Ontology enRIchment anaLysis and visuaLizAtion tool (GOrilla) using a single ranked list of genes 53.

For viral mutation detection, RNA-seq reads were aligned to reference viral genomes using the Burrow-Wheeler Aligner 54 (BWA-0.7.17d). The reference viral genomes were taken from the viral genome starter plasmids used for viral productions. We then used the freeBayes software55 to detect polymorphisms present within the length of a short RNA-seq read.

Single nuclei isolation from spinal cords

10 days p.i. pups were culled by dislocation followed by decapitation then spinal cords were rapidly extracted by hydraulic extrusion using ice-cold PBS. The extracted tissue was immediately snap frozen in an isopentane bath and stored at -80°C until processing. All samples were processed in parallel to reduce likelihood of batch variation. The nuclear isolation protocol was adapted from56. Briefly, samples were dissociated on ice in a Dounce homogenizer, using 10 strokes of the Loose and 5 strokes of the Tight pestle, in ice-cold lysis buffer (10mM Tris pH 7.4, 10mM NaCl, 3mM MgCl2, 0.05% NP-40 (v/v)). Samples were incubated in lysis buffer for 5 minutes before quenching with wash buffer (5% BSA, 0.25% glycerol, 0.5X PBS, 40 units/ml Protector RNAse inhibitor) and filtering through a 30μm filter. Samples were then centrifuged at 4 degrees Celsius, 500 x g for 5 minutes. Supernatants were discarded, pellets resuspended in wash buffer and filtered then centrifuged a second time, as above. The pellets were resuspended in wash buffer, pooled (N=3 biological replicates per experimental condition) and centrifuged a final time before resuspending in PBS with 1ug/mL Hoechst stain for Fluorescence Activated Nuclei Sorting (FANS). Nuclei were sorted using a BD Biosciences FACSAria II flow cytometer, gating based on Forward and Size scatter to exclude debris, Hoechst staining to positively select for nuclei, and GFP (+/-) to select rabies-infected nuclei. For each sample both Hoechst +/GFP + and Hoechst +/ GFP – nuclei were collected and submitted separately for sequencing as GFP + and GFP – samples. After sorting, samples were centrifuged at 500 x g for 5 minutes before being resuspended to the specifications required by 10x technologies. Libraries were prepared using droplet-based snRNA-seq technology using the Chromium Single Cell 3' Reagent Kit v3.1 (dual index) according to the manufacturer’s protocol (10x Genomics). The generated snRNA-seq libraries were sequenced using the NovaSeq 6000 system in a single S1 lane.

snRNA-seq analysis

A custom reference transcriptome was generated by appending the sequence of SiR-N2c-CRE to the mm10 genome (Mus_Musculus.GRCm38/Ensembl98). Counts per gene were generated by aligning reads to the custom genome using CellRanger software (v6.1.2) (10x Genomics). The CellRanger count algorithm was run with the –include-introns parameter in order to capture nuclear pre-mRNA transcripts and the —expect-cells 10000 parameter to account for mixed populations of cells within our samples. Subsequent analyses on the filtered count matrices were performed using R studio v1.4.1717, R v4.1.2 and Seurat v4.1.0. 176,433 nuclei across all samples were combined and tagged with experimental conditions as metadata. The data were then normalized, variable features detected (using the vst selection method) and scaled. A reference spinal cord data set was also loaded into the analyses as provided by SeqSeek. A principal component analysis (PCA) and uniform manifold approximation and projection (UMAP) algorithms for dimension reduction were run on both the experimental dataset and the reference dataset. The FindTransferAnchors function from the Seurat package was then applied between the experimental dataset and reference dataset, enabling the use of an anchor set to map the experimental data onto the reference scRNAseq spinal cord atlas41. This enabled projection of the experimental data into the reference UMAP space and for filtering of neuronal transcriptomes. We used these stringent quality control criteria to ensure to select only for high quality neuronal transcriptomes from our mixed cell population. From 176K profiled nuclei, 37% were filtered out on import due to containing fewer than 200 features, 62% of the remaining were filtered out as doublets and unk, and 1% were identified as neuronal transcriptomes.

Subclassification of all cells identified as neurons by Seurat was performed by using the SeqSeek neural network as described in the respective github repository (SeqSeekClassify: https://github.com/ArielLevineLabNINDS)41. These annotations were loaded and attached back to the experimental Seurat object metadata. A two-tailed chi-squared test was performed to compare the proportions of neuronal subtypes between virally traced vs. non-virally traced spinal cord neurons.

For differential expression analyses, DP is defined as differential proportion genes. We considered genes with log2 expression greater than 1 as being expressed and otherwise as non-expressed. A chi-squared test was then performed to compare the proportion of expression of each gene between samples. Functional profiling of differentially expressed genes between samples was performed by running ordered gene ontology enrichment queries on g:Profiler57.

Statistical Analysis

Throughout the work, mean values are accompanied by SEM. Statistical analyses and graphical visualisation of data were performed using Prism 8 software (Graphpad). The significance of differences between the means of experimental conditions are depicted as follows: * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. Two-tailed, unpaired Student’s T-tests were used in experiments comparing two groups separated by one independent variable, and one-way or two-way ANOVA was used where more than two groups are separated by one or two independent variable, respectively. Tukey correction was performed for multiple comparisons where every group is compared to every other group and Welch correction was performed for comparisons of two groups with different standard deviations.

Extended Data

Extended Data Fig 1.

Extended Data Fig 1

Open in a new tab

Extended Data Fig 2.

Extended Data Fig 2

Open in a new tab

Extended Data Fig 3.

Extended Data Fig 3

Open in a new tab

Extended Data Fig 4.

Extended Data Fig 4

Open in a new tab

Extended Data Fig 5.

Extended Data Fig 5

Open in a new tab

Extended Data Fig 6.

Extended Data Fig 6

Open in a new tab

Extended Data Fig 7.

Extended Data Fig 7

Open in a new tab

Supplementary Material

Extended Data Fig 1
Extended Data Fig 2
Extended Data Fig 3
Extended Data Fig 4
Extended Data Fig 5
Extended Data Fig 6
Extended Data Fig 7
Supplementary Tables

Editorial Summary.

A self-inactivating variant of the CVS-N2c rabies virus enables both retrograde viral tracing and transcriptomic analyses, thereby allowing a combination of circuit mapping and molecular studies.

Acknowledgments

We would like to thank the creators of the SeqMonk program (Babraham Institute) and the SeqSeek atlas41 which were utilised in our analysis pipelines. We thank the members of the LMB Biological Service Group for their support with the in vivo work. This study was supported by the Medical Research Council (MC_UP_1201/2) to M.T., the European Research Council (STG 677029) to M.T., the Rosetrees Trust with an MBPhD fellowship to H.L. (M598), by the Cambridge Philosophical Society and St. Edmund’s College (University of Cambridge) with the Henslow Research Fellowship to A.G.R‥

Footnotes

Author Contributions

M.T., H.L and E.C. conceived the project and analysed the data. H.L. and E.C. conducted the experiments and wrote the manuscript with inputs from M.T‥ E.C. and H.L. co-developed the SiR-N2c technology. E.C. produced SiR viruses with the help of H.L. and F.M‥ H.L. led the in vivo characterisation of the virus and performed the immunology and bulk RNAseq experiments. H.L. performed the long-term survival experiments and peripheral neurotropism experiments with the help of E.C‥ E.C. performed the pseudotyped rabies tracing experiment. E.C., E.W. and F.N. performed single cell-genomic experiments. H.L. and S.M. analysed the bulk and single cell genomic data. A.G.R. performed electrophysiological recordings.

Declaration of Interests

E.C. and M.T. are inventors of a patent related to the SiR technology (WO2018203049A1). The remaining authors declare no competing interests.

Data Availability

All materials described in this paper can be obtained for noncommercial purposes after signing a material transfer agreement (MTA) with the UKRI. Data generated during this study are included in the manuscript and supporting files. The DNA constructs to generate the viral vectors used in this study are available from Addgene (99610, 194352 – 194356). The raw NGS dataset has been deposited into NCBI’s Sequence Read Archive (SRA) and is accessible through accession number PRJNA901288. Further requests for information, resources and reagents should be directed to the Lead Contact, Marco Tripodi.

References

  • 1.McCulloch WS, Pitts WA. logical calculus of the ideas immanent in nervous activity. Bull Math Biol. 1990;52:99–115. 1943 discussion 173-197. [PubMed] [Google Scholar]
  • 2.Sherrington CS. The Integrative Action of the Nervous System. Cambridge University Press; 1947. [Google Scholar]
  • 3.Li PH, et al. Automated Reconstruction of a Serial-Section EM Drosophila Brain with Flood-Filling Networks and Local Realignment. Microscopy and Microanalysis. 2019;25:1364–1365. [Google Scholar]
  • 4.White JG, Southgate E, Thomson JN, Brenner S. The structure of the nervous system of the nematode Caenorhabditis elegans. Philos Trans R Soc Lond B Biol Sci. 1986;314:1–340. doi: 10.1098/rstb.1986.0056. [DOI] [PubMed] [Google Scholar]
  • 5.Luo L, Callaway EM, Svoboda K. Genetic Dissection of Neural Circuits: A Decade of Progress. Neuron. 2018;98:865. doi: 10.1016/j.neuron.2018.05.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Iourov IY, Vorsanova SG, Yurov YB. Single cell genomics of the brain: focus on neuronal diversity and neuropsychiatric diseases. Curr Genomics. 2012;13:477–488. doi: 10.2174/138920212802510439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Rosenberg AB, et al. Single-cell profiling of the developing mouse brain and spinal cord with split-pool barcoding. Science. 2018;360:176–182. doi: 10.1126/science.aam8999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Wickersham IR, et al. Monosynaptic restriction of transsynaptic tracing from single, genetically targeted neurons. Neuron. 2007;53:639–647. doi: 10.1016/j.neuron.2007.01.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Wickersham IR, Finke S, Conzelmann KK, Callaway EM. Retrograde neuronal tracing with a deletion-mutant rabies virus. Nat Methods. 2007;4:47–49. doi: 10.1038/NMETH999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Stepien AE, Tripodi M, Arber S. Monosynaptic rabies virus reveals premotor network organization and synaptic specificity of cholinergic partition cells. Neuron. 2010;68:456–472. doi: 10.1016/j.neuron.2010.10.019. [DOI] [PubMed] [Google Scholar]
  • 11.Tripodi M, Stepien AE, Arber S. Motor antagonism exposed by spatial segregation and timing of neurogenesis. Nature. 2011;479:61–66. doi: 10.1038/nature10538. [DOI] [PubMed] [Google Scholar]
  • 12.Reardon TR, et al. Rabies Virus CVS-N2c(DeltaG) Strain Enhances Retrograde Synaptic Transfer and Neuronal Viability. Neuron. 2016;89:711–724. doi: 10.1016/j.neuron.2016.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Huang KW, Sabatini BL. Single-Cell Analysis of Neuroinflammatory Responses Following Intracranial Injection of G-Deleted Rabies Viruses. Front Cell Neurosci. 2020;14:65. doi: 10.3389/fncel.2020.00065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Zhao P, et al. Analysis of expression profiles of long noncoding RNAs and mRNAs in brains of mice infected by rabies virus by RNA sequencing. Sci Rep. 2018;8:11858. doi: 10.1038/s41598-018-30359-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Prosniak M, Hooper DC, Dietzschold B, Koprowski H. Effect of rabies virus infection on gene expression in mouse brain. Proc Natl Acad Sci U S A. 2001;98:2758–2763. doi: 10.1073/pnas.051630298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Ciabatti E, Gonzalez-Rueda A, Mariotti L, Morgese F, Tripodi M. Life-Long Genetic and Functional Access to Neural Circuits Using Self-Inactivating Rabies Virus. Cell. 2017;170:382–392.:e314. doi: 10.1016/j.cell.2017.06.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Chatterjee S, et al. Nontoxic, double-deletion-mutant rabies viral vectors for retrograde targeting of projection neurons. Nat Neurosci. 2018;21:638–646. doi: 10.1038/s41593-018-0091-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Zhang D, et al. Genome-Wide Transcriptional Profiling Reveals Two Distinct Outcomes in Central Nervous System Infections of Rabies Virus. Front Microbiol. 2016;7:751. doi: 10.3389/fmicb.2016.00751. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Miao FM, et al. Comparison of immune responses to attenuated rabies virus and street virus in mouse brain. Arch Virol. 2017;162:247–257. doi: 10.1007/s00705-016-3081-7. [DOI] [PubMed] [Google Scholar]
  • 20.Wang ZW, et al. Attenuated rabies virus activates, while pathogenic rabies virus evades, the host innate immune responses in the central nervous system. J Virol. 2005;79:12554–12565. doi: 10.1128/JVI.79.19.12554-12565.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Dietzschold B, Li J, Faber M, Schnell M. Concepts in the pathogenesis of rabies. Future Virol. 2008;3:481–490. doi: 10.2217/17460794.3.5.481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Conzelmann KK, Cox JH, Schneider LG, Thiel HJ. Molecular cloning and complete nucleotide sequence of the attenuated rabies virus SAD B19. Virology. 1990;175:485–499. doi: 10.1016/0042-6822(90)90433-r. [DOI] [PubMed] [Google Scholar]
  • 23.Ciabatti E, et al. Genomic stability of Self-inactivating Rabies. bioRxiv. 2020:2020.2009.2019.304683 [Google Scholar]
  • 24.Menegas W, Akiti K, Amo R, Uchida N, Watabe-Uchida M. Dopamine neurons projecting to the posterior striatum reinforce avoidance of threatening stimuli. Nat Neurosci. 2018;21:1421–1430. doi: 10.1038/s41593-018-0222-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Matsuyama M, et al. “Self-inactivating” rabies viruses are susceptible to loss of their intended attenuating modification. bioRxiv. 2020:550640. doi: 10.1073/pnas.2023481120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Takeuchi O, Akira S. MDA5/RIG-I and virus recognition. Curr Opin Immunol. 2008;20:17–22. doi: 10.1016/j.coi.2008.01.002. [DOI] [PubMed] [Google Scholar]
  • 27.Hooper DC, et al. Collaboration of antibody and inflammation in clearance of rabies virus from the central nervous system. J Virol. 1998;72:3711–3719. doi: 10.1128/jvi.72.5.3711-3719.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Eddleston M, Mucke L. Molecular profile of reactive astrocytes—Implications for their role in neurologic disease. Neuroscience. 1993;54:15–36. doi: 10.1016/0306-4522(93)90380-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Sasaki Y, Ohsawa K, Kanazawa H, Kohsaka S, Imai Y. Iba1 is an actin-cross-linking protein in macrophages/microglia. Biochem Biophys Res Commun. 2001;286:292–297. doi: 10.1006/bbrc.2001.5388. [DOI] [PubMed] [Google Scholar]
  • 30.Morimoto K, et al. Rabies virus quasispecies: implications for pathogenesis. Proceedings of the National Academy of Sciences of the United States of America. 1998;95:3152–3156. doi: 10.1073/pnas.95.6.3152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kim EJ, Jacobs MW, Ito-Cole T, Callaway EM. Improved Monosynaptic Neural Circuit Tracing Using Engineered Rabies Virus Glycoproteins. Cell Rep. 2016;15:692–699. doi: 10.1016/j.celrep.2016.03.067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Russo SJ, Nestler EJ. The brain reward circuitry in mood disorders. Nat Rev Neurosci. 2013;14:609–625. doi: 10.1038/nrn3381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Tasic B, et al. Adult mouse cortical cell taxonomy revealed by single cell transcriptomics. Nat Neurosci. 2016;19:335–346. doi: 10.1038/nn.4216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zeisel A, et al. Brain structure. Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science. 2015;347:1138–1142. doi: 10.1126/science.aaa1934. [DOI] [PubMed] [Google Scholar]
  • 35.Kohl J, et al. Functional circuit architecture underlying parental behaviour. Nature. 2018;556:326–331. doi: 10.1038/s41586-018-0027-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Kim EJ, et al. Extraction of Distinct Neuronal Cell Types from within a Genetically Continuous Population. Neuron. 2020;107:274–282.:e276. doi: 10.1016/j.neuron.2020.04.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Alkaslasi MR, et al. Single nucleus RNA-sequencing defines unexpected diversity of cholinergic neuron types in the adult mouse spinal cord. Nat Commun. 2021;12:2471. doi: 10.1038/s41467-021-22691-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Blum JA, et al. Single-cell transcriptomic analysis of the adult mouse spinal cord reveals molecular diversity of autonomic and skeletal motor neurons. Nat Neurosci. 2021;24:572–583. doi: 10.1038/s41593-020-00795-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Sathyamurthy A, et al. Massively Parallel Single Nucleus Transcriptional Profiling Defines Spinal Cord Neurons and Their Activity during Behavior. Cell Rep. 2018;22:2216–2225. doi: 10.1016/j.celrep.2018.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Mo A, et al. Epigenomic Signatures of Neuronal Diversity in the Mammalian Brain. Neuron. 2015;86:1369–1384. doi: 10.1016/j.neuron.2015.05.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Russ DE, et al. A harmonized atlas of mouse spinal cord cell types and their spatial organization. Nat Commun. 2021;12:5722. doi: 10.1038/s41467-021-25125-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Levine AJ, et al. Identification of a cellular node for motor control pathways. Nat Neurosci. 2014;17:586–593. doi: 10.1038/nn.3675. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Arber S. Motor circuits in action: specification, connectivity, and function. Neuron. 2012;74:975–989. doi: 10.1016/j.neuron.2012.05.011. [DOI] [PubMed] [Google Scholar]
  • 44.Lake BB, et al. Neuronal subtypes and diversity revealed by single-nucleus RNA sequencing of the human brain. Science. 2016;352:1586–1590. doi: 10.1126/science.aaf1204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Tiklova K, et al. Single-cell RNA sequencing reveals midbrain dopamine neuron diversity emerging during mouse brain development. Nat Commun. 2019;10:581. doi: 10.1038/s41467-019-08453-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Gradinaru V, et al. Targeting and readout strategies for fast optical neural control in vitro and in vivo. J Neurosci. 2007;27:14231–14238. doi: 10.1523/JNEUROSCI.3578-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Courtine G, Sofroniew MV. Spinal cord repair: advances in biology and technology. Nat Med. 2019;25:898–908. doi: 10.1038/s41591-019-0475-6. [DOI] [PubMed] [Google Scholar]
  • 48.Takeoka A, Vollenweider I, Courtine G, Arber S. Muscle spindle feedback directs locomotor recovery and circuit reorganization after spinal cord injury. Cell. 2014;159:1626–1639. doi: 10.1016/j.cell.2014.11.019. [DOI] [PubMed] [Google Scholar]
  • 49.Roselli F, Caroni P. From intrinsic firing properties to selective neuronal vulnerability in neurodegenerative diseases. Neuron. 2015;85:901–910. doi: 10.1016/j.neuron.2014.12.063. [DOI] [PubMed] [Google Scholar]
  • 50.Osakada F, Callaway EM. Design and generation of recombinant rabies virus vectors. Nature Protocols. 2013;8:1583–1601. doi: 10.1038/nprot.2013.094. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Wickersham IR, Sullivan HA, Seung HS. Production of glycoprotein-deleted rabies viruses for monosynaptic tracing and high-level gene expression in neurons. Nat Protoc. 2010;5:595–606. doi: 10.1038/nprot.2009.248. [DOI] [PubMed] [Google Scholar]
  • 52.Mi H, et al. Protocol Update for large-scale genome and gene function analysis with the PANTHER classification system (v.14.0) Nat Protoc. 2019;14:703–721. doi: 10.1038/s41596-019-0128-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Eden E, Navon R, Steinfeld I, Lipson D, Yakhini Z. GOrilla: a tool for discovery and visualization of enriched GO terms in ranked gene lists. BMC Bioinformatics. 2009;10:48. doi: 10.1186/1471-2105-10-48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26:589–595. doi: 10.1093/bioinformatics/btp698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Garrison E, Marth G. Haplotype-based variant detection from short-read sequencing. arXiv preprint. 2012:arXiv:1207.3907 [Google Scholar]
  • 56.Nagy C, et al. Single-nucleus transcriptomics of the prefrontal cortex in major depressive disorder implicates oligodendrocyte precursor cells and excitatory neurons. Nat Neurosci. 2020;23:771–781. doi: 10.1038/s41593-020-0621-y. [DOI] [PubMed] [Google Scholar]
  • 57.Raudvere U, et al. g:Profiler: a web server for functional enrichment analysis and conversions of gene lists (2019 update) Nucleic Acids Res. 2019;47:W191–W198. doi: 10.1093/nar/gkz369. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Extended Data Fig 1
EMS176613-supplement-Extended_Data_Fig_1.jpg (1.4MB, jpg)
Extended Data Fig 2
EMS176613-supplement-Extended_Data_Fig_2.jpg (955.9KB, jpg)
Extended Data Fig 3
EMS176613-supplement-Extended_Data_Fig_3.jpg (592.3KB, jpg)
Extended Data Fig 4
EMS176613-supplement-Extended_Data_Fig_4.jpg (546.9KB, jpg)
Extended Data Fig 5
EMS176613-supplement-Extended_Data_Fig_5.jpg (2.2MB, jpg)
Extended Data Fig 6
EMS176613-supplement-Extended_Data_Fig_6.jpg (1.4MB, jpg)
Extended Data Fig 7
EMS176613-supplement-Extended_Data_Fig_7.jpg (1.9MB, jpg)
Supplementary Tables
EMS176613-supplement-Supplementary_Tables.pdf (451.4KB, pdf)

Data Availability Statement

All materials described in this paper can be obtained for noncommercial purposes after signing a material transfer agreement (MTA) with the UKRI. Data generated during this study are included in the manuscript and supporting files. The DNA constructs to generate the viral vectors used in this study are available from Addgene (99610, 194352 – 194356). The raw NGS dataset has been deposited into NCBI’s Sequence Read Archive (SRA) and is accessible through accession number PRJNA901288. Further requests for information, resources and reagents should be directed to the Lead Contact, Marco Tripodi.

RESOURCES