Differences in neurotropism and neurotoxicity among retrograde viral tracers

Abstract

Background: Neurotropic virus-based tracers have been extensively applied in mapping and manipulation of neural circuits. However, their neurotropic and neurotoxic properties remain to be fully characterized.

Methods: Through neural circuit tracing, we systematically compared the neurotropism discrepancy among different multi-trans-synaptic and mono-synaptic retrograde viral tracers including pseudorabies virus (PRV), rabies virus (RV), and the newly engineered retro adeno-associated virus (rAAV2-retro) tracers. The (single-cell) RNA sequencing analysis was utilized for seeking possible attribution to neurotropism discrepancy and comparing cell toxicity caused by viral infection between glycoprotein-deleted RV (RV-∆G) and rAAV2-retro. Viral toxicity induced microglia activation and neuronal protein change were evaluated by immunohistochemistry.

Results: Multi-trans-synaptic retrograde viral tracers, PRV and RV, exhibit differential neurotropism when they were used for central neural circuit tracing from popliteal lymph nodes. Mono-synaptic retrograde tracers, including RV-∆G and rAAV2-retro, displayed discrepant neurotropic property, when they were applied to trace the inputs of lateral hypothalamic area and medial preoptic nucleus. rAAV2-retro demonstrated preference in cerebral cortex, whereas RV-∆G prefers to label basal ganglia and hypothalamus. Remarkably, we detected a distinct preference for specific cortical layer of rAAV2-retro in layer 5 and RV-∆G in layer 6 when they were injected into dorsal lateral geniculate nucleus to label corticothalamic neurons in primary visual cortex. Complementation of TVA receptor gene in RV-resistant neurons enabled EnvA-pseudotyped RV infection, supporting receptors attribution to viral neurotropism. Furthermore, both RV-∆G and rAAV2-retro exerted neurotoxic influence at the injection sites and retrogradely labeled sites, while the changes were more profound for RV-∆G infection. Finally, we demonstrated a proof-of-concept strategy for more comprehensive high-order circuit tracing of a specific target nucleus by combining rAAV2-retro, RV, and rAAV tracers.

Conclusions: Different multi-trans-synaptic and mono-synaptic retrograde viral tracers exhibited discrepant neurotropism within certain brain regions, even cortical layer preference. More neurotoxicity was observed under RV-∆G infection as compared with rAAV2-retro. By combining rAAV2-retro, RV, and rAAV tracers, high-order circuit tracing can be achieved. Our findings provide important reference for appropriate application of viral tracers to delineate the landscape and dissect the function of neural network.

Keywords: Mono-synaptic Tracing; Multi-trans-synaptic Tracing; Neurotoxicity; Neurotropic Virus; Neurotropism; RNA Sequencing; Retrograde tracing; pseudorabies virus (PRV); rAAV2-retro; rabies virus (RV).

Fig. 1

The structures of all viral tracers used in our study. (a) Structures of multi-trans-synaptic retrograde viruses including RV-B2C-EGFP and PRV-152 are illustrated. For RV-B2C-EGFP, EGFP gene was inserted between G- and L-coding sequences of CVS-B2C strain. For PRV-152, EGFP was driven by the CMV promoter following gG gene of the Bartha strain. (b) Structures of mono-synaptic retrograde viruses are illustrated. Because TK (thymidine kinase) gene is involved in viral replication and US9 is responsible for anterograde transport, a retrograde mono-synaptic tracing virus PRV-∆TK-∆US9-EGFP was obtained by deleting both TK and US9 genes, in which the location of US9 gene was replaced by CMV promoter and EGFP gene via homologous recombination on the BAC backbone of the Becker strain. For the RV-∆G mono-synaptic virus, EGFP, DsRed and Cre-T2A-tagBFP was respectively inserted in the rabies virus genome to replace the G gene. For the rAAV2-retro, the ef1a and CMV promotor was used to drive YFP and Cre-T2A-tagBFP expression, respectively. (c) Structures of AAV helper viruses. The hSyn and ef1a promoter were used to drive TVA, DIO-TVA and DIO-RVG elements

Fig. 2

Multi-trans-synaptic retrograde tracing via injection of RV-B2C-EGFP and PRV-152 into the lymph nodes. (a-d) RV-B2C-EGFP retrogradely labeled M1 (A), S1 (A), VPM (B), LPAG (C), and cerebellum (D), but not PVN, LC, or Bar when injected into the lymph nodes. (e-h) Retrograde labeling by PRV-152 from the lymph nodes was observed in M1 (E), S1 (E), PVN (F), LPAG (G), LC and Bar (H), but not VPM (F) and cerebellum (H). (i) The Map of inputs to LHA labeled by RV-B2C-EGFP (upper line) and PRV-152 (lower line) at coronal view are arranged along the rostrocaudal direction. Retrogradely labeled neurons are indicated as colored dots. Numbers indicate position of sections relative to bregma (mm). (j) The semi-quantified labeling discrepancy between RV-B2C-EGFP and PRV-152 in the whole brain. Three mice were analyzed for each viral tracer. Insets show magnified images of the boxed regions. Scale bars = 1 mm in lower magnification images; 100 μm in higher magnification insets

Fig. 3

Mono-synaptic retrograde labeling of LHA input circuits using rAAV2-retro-YFP, RV-∆G-EGFP, and green retrobeads. (a) Schematic diagram of tracer injection into the lateral hypothalamic area (LHA). (a’-a”’) The injection sites of rAAV2-retro-YFP, RV-∆G-EGFP, and green retrobeads were confirmed to be localized in the LHA. (b-b‴) Retrograde labeling of the MPA was observed for all three tracers. (c-d′) Retrogradely labeled neurons were observed in the mPFC following injection of rAAV2-retro-YFP (C, C′), while no such labeling was observed in the NAc (C, C”) or PVN (D, D’). (e-f’) Neurons retrogradely labeled with RV-∆G-EGFP were localized in the NAc (E, E”) and PVN (F, F’), but were barely observed in the mPFC (E, E’). (g-h′) Green retrobeads retrogradely labeled the NAc (G, G”) and PVN (H, H’), but not the mPFC (G, G’). (i) The number of labeled neurons in each upstream region of the LHA was compared among the three tracers. Three mice were analyzed for each viral tracer or retrobeads. Scale bars = 1 mm for A′-A‴, C, D, E, F, G, H; 200 μm for the magnified images

Fig. 4

Co-injection of rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP into LHA and MPO nuclei to trace input neurons. (a) Schematic diagram of virus co-injection into the LHA of Ai9 reporter mice and the injection sites confirmation. (b-e”) When the LHA was simultaneously injected with rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP, discrepancies in the input circuits retrogradely labeled by the two viruses were observed. The mPFC (Cg1+PrL), M2, PVA, and DI were exclusively labeled by rAAV2-retro-Cre-tagBFP, while the NAc and PVN were preferentially labeled by RV-∆G-EGFP, although overlapping labeling was observed in the MPA. (f) Schematic diagram and the injection sites of virus co-injection in the medial preoptic nucleus (MPO). (g-j”) Injection of rAAV2-retro-Cre-tagBFP in MPO was associated with preferential labeling in the mPFC, AI and PMV. In contrast, injection of RV-∆G-EGFP in MPO was associated with preferential labeling in the LSI. Overlapping signals were observed in the AHiPM. Red, tdTomato expressed by rAAV2-retro-Cre-tagBFP-labeled cells; green, GFP expressed by RV-∆G-EGFP-labeled cells; blue, DAPI. Scale bars = 1 mm for A-D and F-I; 200 μm for B’-D’, G’-I’, B”-D”and G”-I”; 100 μm for E-E” and J-J”

Fig. 5

Labeling preference of rAAV2-retro and RV-∆G viral tracers in certain brain region and cortical layer. (a-b) Whole-brain quantification analysis of the retrograde labeling of input circuits to the LHA (A) and MPO (B) following injection of rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP. The number of labeled neurons in each nucleus on the side ipsilateral to the injection site was quantified and normalized to the total number of ipsilateral labeled neurons for each mouse (n=3). (c) The percentage of labeled input brain regions to LHA and MPO was represented as pie chart to indicate the labeling preference of rAAV2-retro in cerebral cortex, and RV-∆G in basal ganglia and hypothalamus. (d) Retrograde tracers of rAAV2-retro-YFP, RV-∆G-EGFP, and green retrobeads were separately injected into the dLGN for retrograde labeling of upstream corticothalamic neurons in the primary visual cortex (V1). The corticothalamic neurons in V1 labeled by rAAV2-retro-YFP were predominately localized in the fifth layer (L5). In contrast, both RV-∆G-EGFP and green retrobeads mostly labeled corticothalamic neurons in the sixth layer (L6) of V1. (E) Retrograde tracers of rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP were co-injected into the dLGN of Ai9 mice for simultaneously labeling of upstream corticothalamic neurons in V1. The rAAV2-retro-Cre-tagBFP predominately labeled V1 corticothalamic neurons in the fifth layer (L5), while RV-∆G-EGFP mostly labeled corticothalamic neurons in the sixth layer (L6) of V1. (F) The injection sites of rAAV2-retro-YFP, RV-∆G-EGFP, green retrobeads, and co-injection of rAAV2-retro-Cre-tagBFP with RV-∆G-EGFP. Scale bars = 200 μm in D-E, 1 mm in F

Fig. 6

Complementing TVA receptor expression in the mPFC neurons enables EnvA pseudotyped RV infection. (a) Schematic diagram depicting the co-injection of rAAV2-retro-YFP and RV-∆G-DsRed into the LHA. (b-f) Co-injection of rAAV2-retro-YFP and RV-∆G-DsRed into the LHA labeled different input nuclei. The mPFC was labeled by rAAV2-retro-YFP and NAc was labeled by RV-∆G-DsRed. (g) The injection site for rAAV2-retro-YFP and RV-∆G-DsRed in LHA. (h) Schematic diagram depicting the injection of helper AAV expressing TVA (rAAV9-EGFP-TVA) to enable EnvA pseudotyped RV (EnvA-RV-∆G-DsRed) infection in the mPFC. (i-m) Complementation of TVA receptor expression in mPFC neurons enabled EnvA-RV-∆G-DsRed infection in the mPFC. No positive labeling of EnvA-RV-∆G-DsRed was detected in NAc. (n) The injection site for EnvA-RV-∆G-DsRed in LHA. (o) Schematic diagram depicting the injection of EnvA-RV-∆G-DsRed into the LHA without helper rAAV9-EGFP-TVA. (p-t) No signals were observed in the mPFC following injection of EnvA-RV-∆G-DsRed alone. (u) The injection site for EnvA-RV-∆G-DsRed in LHA. Scale bars =1 mm for B, I, P; 200 μm for magnification images

Fig. 7

Gene profiling analysis of rAAV2-retro-labeled and RV-∆G-labeled neurons via single-cell RNA sequencing. (a) Flow chart of single-cell isolation and RNA sequencing. (b) Gene cluster analysis of the neuronal groups infected by rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP. (c) Expression of neuronal marker genes including VLGUT1, VGLUT2, GAD65, GAD67, TH, ChAT, and Pet1 in the rAAV2-retro-Cre-tagBFP and RV-∆G-EGFP groups. (d) Expression heat-map of potential receptor candidates for rabies virus, including Ncam, Chrna, and Ngfr, in each sample from rAAV2-retro and RV-∆G labeled groups. (e) These are no significant difference of Ncam1 expression between rAAV2-retro and RV-∆G labeled groups, P=0.088. Chrna, nicotinic acetylcholine receptor gene; Ncam, neural cell adhesion molecule gene; Ngfr: Nerve growth factor receptor

Fig. 8

Analysis of neurotoxicity induced by rAAV2-retro and RV-ΔG at injection sites and retrogradely labeled nuclei. (a) Schematic diagram of virus injection and tissue extraction for RNA sequencing. (b) Injection of rAAV2-retro-YFP into the LHA caused alterations in gene expression profiles. (c) Injection of RV-∆G-EGFP into the LHA caused alterations in gene expression profiles. (d) Gene profile alterations in retrogradely labeled mPFC following injection of rAAV2-retro-YFP into the LHA. Red dots represent the genes with significant change, while the black dots represent the genes without significant change. (e) Gene profile alterations in retrogradely labeled NAc following injection of RV-∆G-EGFP into the LHA. (f-g) GO-term pathway analysis of differentially expressed genes at the site of RV-∆G-EGFP injection (LHA) and in retrogradely labeled nucleus (NAc). (h-i) Immunostaining for the microglial marker Iba1 in the rAAV2-retro-YFP group. Iba1-positive cells were observed on the injection site, but barely in the PBS injected (Mock) LHA. (j-k) No activated microglia was detected in the ipsilateral mPFC following injection of rAAV2-retro-YFP or PBS (Mock) into the LHA. (l-m) Intense microglial activation was observed at the injection site when RV-∆G-EGFP was injected into the LHA, while few signals were observed in the PBS injected (Mock) LHA. (n-o) Intense microglial activation was observed in the retrogradely labeled ipsilateral NAc following injection of RV-∆G-EGFP into the LHA, but not in the ipsilateral NAc of Mock control. (p) Quantification of mean fluorescence intensity (Mean±SEM) of Iba1 in LHA after virus injection. rAAV2-retro-YFP: 14.09±0.8373, Mock: 6.059±1.132, rAAV2-retro-YFP vs Mock: P=0.0013; RV-∆G-EGFP: 19.75±1.403, Mock: 7.885±1.265, RV-∆G-EGFP vs Mock: P = 0.0008; rAAV2-retro-YFP vs RV-∆G-EGFP: P = 0.0133; n = 4, n is mice number. (q) Quantification of mean fluorescence intensity (Mean±SEM) of Iba1 in mPFC injected with rAAV2-retro-YFP and PBS. rAAV2-retro-YFP: 6.438±1.114, PBS: 7.713±1.192, P = 0.4639, n = 4. (r) Quantification of mean fluorescence intensity (Mean±SEM) of Iba1 in NAc injected with RV-∆G-EGFP and PBS. RV-∆G-EGFP: 17.20±1.408, Mock: 6.883±0.8132, P = 0.0007; n = 4. (s) Immunostaining for oxytocin in PVN neurons 7 days following the injection of RV-∆G-EGFP into the LHA. (t, u) Boxed areas in S were magnified to show oxytocin-positive neurons in the ipsilateral and contralateral PVN to the virus injection side. (v) Higher-magnification images of the dashed box depicting the ipsilateral PVN in S with green color and merged color. Green, GFP expressed by RV-∆G-EGFP; red, oxytocin immunostaining signal. (w) Quantification of oxytocin-positive neurons (Mean±SEM) in the ipsilateral and contralateral PVN. Contra: 23.50±1.443, Ipsi: 62.00±3.136, P=0.0001, n = 4. MFI, mean fluorescence intensity; Ipsi, ipsilateral; Contra, contralateral. n is mice number. Scale bars =200 μm for S; 100 μm for H-O; 50 μm for T-V. Unpaired t-tests, *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 9

Higher-order circuit tracing using a combination of different tracing viruses. (a) Strategy for mapping the higher-order input circuits of target nuclei by combining different viral tracers. (b) Diagram of higher-order circuit mapping of the LHA by combining rAAV2-retro-Cre-tagBFP, rAAV9-DIO-EGFP-TVA, rAAV9-DIO-RVG, and EnvA pseudotyped RV-∆G-DsRed. (c) Injected site for rAAV2-retro-Cre-tagBFP in LHA. (d) Direct upstream neurons from the LHA (first order) were labeled with both green and red fluorescence in the PrL and IL of the mPFC (also called starter neurons). Green, TVA-positive neuron; red, EnvA-RV-∆G-DsRed-positive neuron; yellow, starter neurons. (e-g) Nuclei upstream from the starter neurons (second order) were labeled with only red fluorescence in the AI, LO, Pir (E), AM (F), and BLA (G). (h-i) Diagram of the higher-order circuit map of the LHA obtained by combining rAAV9-DIO-RVG and RV-∆G-Cre-tagBFP labeling in the PVN. (j) Injected site for RV-∆G-Cre-tagBFP. (k) RV-∆G-Cre-tagBFP retrogradely labeled only neurons in the ipsilateral PVN. (l) Complementing RV-G by injecting AAV-DIO-RVG into the ipsilateral PVN resulted in the labeling of second-order neurons in the contralateral PVN. Scale bars = 200 μm.