Identification of Two Classes of Somatosensory Neurons That Display Resistance to Retrograde Infection by Rabies Virus

Abstract

Glycoprotein-deleted rabies virus-mediated monosynaptic tracing has become a standard method for neuronal circuit mapping, and is applied to virtually all parts of the rodent nervous system, including the spinal cord and primary sensory neurons. Here we identified two classes of unmyelinated sensory neurons (nonpeptidergic and C-fiber low-threshold mechanoreceptor neurons) resistant to direct and trans-synaptic infection from the spinal cord with rabies viruses that carry glycoproteins in their envelopes and that are routinely used for infection of CNS neurons (SAD-G and N2C-G). However, the same neurons were susceptible to infection with EnvA-pseudotyped rabies virus in tumor virus A receptor transgenic mice, indicating that resistance to retrograde infection was due to impaired virus adsorption rather than to deficits in subsequent steps of infection. These results demonstrate an important limitation of rabies virus-based retrograde tracing of sensory neurons in adult mice, and may help to better understand the molecular machinery required for rabies virus spread in the nervous system. In this study, mice of both sexes were used.SIGNIFICANCE STATEMENT To understand the neuronal bases of behavior, it is important to identify the underlying neural circuitry. Rabies virus-based monosynaptic tracing has been used to identify neuronal circuits in various parts of the nervous system. This has included connections between peripheral sensory neurons and their spinal targets. These connections form the first synapse in the somatosensory pathway. Here we demonstrate that two classes of unmyelinated sensory neurons, which account for >40% of dorsal root ganglia neurons, display resistance to rabies infection. Our results are therefore critical for interpreting monosynaptic rabies-based tracing in the sensory system. In addition, identification of rabies-resistant neurons might provide a means for future studies addressing rabies pathobiology.

Keywords: neuronal circuits; rabies-mediated monosynaptic tracing; sensory system.

Figure 1.

Characterization of the prevalence of neuronal subtypes in lumbar DRGs. A–D, Immunohistochemical analysis of DRG neurons with antibodies against NeuN, CGRP, and NF200 (A); NeuN, TH, and IB4 (B); P2X3, PlxnC1, and IB4 (C); and CGRP and TrkA (D). E, The relative abundance of DRG neurons immunoreactive for the indicated marker(s). F, Schematic representation of which neuron classes are labeled by the antibodies or antibody combinations used in this study. Scale bar, 100 μm. Error bars represent the SD.

Figure 2.

DRG neurons are sensitive to prolonged incubation with SAD ΔG rabies. A, B, Top row, mCherry (red) expression at 5 (A) and 10 d (B) after the injection of SAD.RabiesΔG-mCherry (SAD-G). Bottom row, colabeling of mCherry and NeuN is depicted (green). C, D, Immunohistochemical analysis of SAD.RabiesΔG-mCherry-infected DRGs with antibodies against mCherry (red) and the macrophage marker IBA1 (cyan) at 5 and 10 d after spinal injection. mCherry expression is depicted in the top row; colabeling of mCherry and IBA1 is depicted in the bottom row. Note the reduced number of mCherry+ cell bodies at 10 d after injection and the strong upregulation of the macrophage marker IBA1. SAD.RabiesΔG-mCherry (SAD-G) was injected intraspinally at 2.4 × 10⁸ focus forming units per milliliter. Scale bar, 100 μm.

Figure 3.

Molecular profiling of SAD.RabiesΔG-eGFP (SAD-G)-infected and SAD.RabiesΔG-mCherry (N2C-G)-infected DRG neurons. A, Schematic representation of the experimental design. SAD-G-pseudotyped RabiesΔG-mCherry/eGFP or N2C-G-pseudotyped RabiesΔG-mCherry was injected at L3–L5 levels of the lumbar spinal cord of wild-type mice. mCherry/eGFP expression from the rabies genome enables tracing of infected neurons. B, eGFP expression in a representative image of a lumbar spinal cord section 5 d after infection. Note that infection of local interneurons predominantly occurs in laminae I–IV. The dotted white line outlines the spinal gray matter. C–E, G–I, Immunohistochemical analysis of infected DRG neurons with antibodies against NF200, mCherry, and TrkA (C, G), with antibodies against P2X3, mCherry, and Plxnc1 (D, H), and with antibodies against mCherry and TH (E, I). Left column in C–E and G–I shows expression of the indicated markers and the right column depicts the indicated markers together with the rabies-derived mCherry. Arrows indicate mCherry+ neurons labeled with two additional markers. Arrowheads indicate mCherry+ neurons labeled with one additional marker. Asterisks indicate mCherry+ neurons negative for the analyzed marker. F, J, Quantification of mCherry+ neurons characterized by the expression of the respective marker after spinal injection of either SAD.RabiesΔG-mCherry (SAD-G; F) or SAD.RabiesΔG-mCherry (N2C-G; J). SAD.RabiesΔG-mCherry (SAD-G) and SAD.RabiesΔG-mCherry (N2C-G) were adjusted to the same titer and subsequently injected at 2.4 × 10⁸ focus forming units per milliliter. Scale bars, 100 μm. Error bars represent the SD.

Figure 4.

Evaluation of the activity dependency of rabies virus infection. A, Schematic representation of the experimental design. SAD-G-pseudotyped rabiesΔG-eGFP was injected at L3–L5 levels of the lumbar spinal cord of wild-type mice followed immediately by the injection of formalin into the hindpaw that provides innervation to the spinal injection site. B, Quantification of the percentage of infected DRG neurons in naive and formalin-injected mice. C–E, Immunhistological stainings with antibodies against GFP, NF200, and TrkA (C); GFP, P2X3, and Plxnc1 (D); and GFP and TH (E). F, Quantification of the percentage of infected neurons expressing the indicated marker (**p < 0.01). SAD.RabiesΔG-eGFP (SAD-G) was injected at 4.5 × 10⁷ focus forming units per milliliter. Scale bars, 100 μm. Error bars represent the SD.

Figure 5.

Characterization of the molecular identity of DRG-expressing Cre in SNS::Cre mice. A–C, Immunohistochemical analysis of lumbar DRG neurons in SNS::Cre; Rosa26^lsl-Tomato mice. A, Analysis of Tomato+ DRG neurons with antibodies against NeuN and TH. B, Analysis of Tomato+ DRG neurons with antibodies against P2X3 and Plxnc1. C, Analysis of Tomato+ DRG neurons with antibodies against NF200 and TrkA. D, Termination zone of Tomato+ primary afferent fibers in the spinal cord. Dotted lines show the borders between LII and LIII and between LIII and deeper dorsal horn. E, Quantification of the proportion of Tomato+ DRG neurons of all DRG neurons (Tom/NeuN) and quantification of the percentage of the indicated markers among all Tomato+ neurons. Scale bar, 100 μm. Error bars represent the SD.

Figure 6.

Molecular profiling of SAD.RabiesΔG-eGFP (Enva)-infected DRG neurons from SNS::Cre; Rosa26^lsl-TVA mice. A, Schematic representation of the experimental design. SAD.RabiesΔG-eGFP (Enva) contains the same genome as SAD.RabiesΔG-eGFP (SAD-G), but is pseudotyped with the EnvA glycoprotein instead. SAD.RabiesΔG-eGFP (Enva) was injected into L3–L4 levels of the lumbar spinal cord of SNS::Cre; Rosa26^lsl-TVA mice. B, Characterization of eGFP+ neurons with antibodies directed against NF200 and TrkA. C, Detection of a P2X3+;Plxnc1+-infected DRG neuron (eGFP+). D, Analysis of infected DRGs with antibodies against GFP and TH. Arrows indicate eGFP+ neurons labeled with two additional markers. Asterisks indicate eGFP+ neurons negative for the analyzed marker. E, Quantification of the percentage of eGFP+ neurons expressing the indicated marker. SAD.RabiesΔG-eGFP (Enva) was injected at 3.5 × 10⁸ focus forming units per milliliter. Scale bars, 100 μm. Error bars represent the SD.

Figure 7.

Monosynaptic tracing initiated from spinal Grp::Cre neurons. A, A', Grp::Cre neurons are located in lamina II of the spinal cord. The neuropil and cell bodies of GRP::Cre neurons overlap with central terminals of IB4+ and TH+ primary afferent fibers. B, High-resolution imaging of a tdTomato+ Grp neuron (red) indicates the presence of Homer1+ (green) postsynapses on Grp neurons, which are near IB4+ (blue) terminals. C, High-resolution imaging of tdTomato+ Grp neuropil (red) indicates the presence of Homer1+ (blue) postsynapses on Grp neurons near vGlut3+ (green) primary afferent terminals. D, Rabies-mediated monosynaptic retrograde tracing reveals many primary (mCherry+ and eGFP+) and secondary (only eGFP+) infected spinal neurons. E, Secondary infected DRG neurons are mostly NF200+ (arrows). F, Quantification of eGFP-labeled DRG neuron subtypes after retrograde transduction mediated by either SAD-G or oG rabies glycoprotein. SAD.RabiesΔG-eGFP (Enva) was always injected at 3.5 × 10⁸ focus forming units per milliliter. AAV1.EF1a.flex.mCherry-2A-SADG and pAAV1.Ef1a.DIO.oG.WPRE.hGH were injected at 9.5 × 10¹² GC/ml. Scale bars: A, D, E, 100 μm; B, C, 5 μm. Error bars represent the SD.

Figure 8.

Monosynaptic retrograde tracing is limited to subsets of sensory neurons. The dorsal spinal horn is innervated by myelinated and unmyelinated primary afferent fibers [myelinated: Aβ-fibers and Aδ-fibers from NF and PEP neurons; unmyelinated, C-fibers from NP, PEP, and C-LTMRs (TH)]. G-deleted rabies virus pseudotyped with the rabies glycoproteins SAD19B-G or N2C-G are largely restricted from entering central terminals of NP and TH neurons. Central terminals of NP neurons show some limited susceptibility to infection with G-deleted rabies virus pseudotyped with the optimized rabies glycoprotein (oG).