Sensory Afferents Share Information With Diverse Array of Neurons
Pain signals enter the spinal cord via primary sensory afferent neurons, which pass information along to ascending projection neurons that extend to the brain. But pain signals undergo processing along the way, which is not well understood. Using cell labeling techniques, researchers have now identified neurons in neonatal mice that make connections with cutaneous primary sensory afferents, the nature of their putative synaptic inputs—including intraganglion, intraspinal, and supraspinal inputs—onto the afferents’ central terminals in the spinal cord, and the neurotransmitters the neurons release.
The work, led by Fan Wang, Duke University, Durham, US, was published August 31 in the Journal of Clinical Investigation.
“The work is technically impressive,” said Andrew Todd, University of Glasgow, UK, who was not involved in the study, but he cautioned that the interpretation that the labeled cells all made presynaptic inputs onto afferents may be premature. “There is no doubt that the authors labeled cells that form synapses on primary afferents,” Todd said, “but because they labeled some cells thought not to form synapses … it raises some questions.”
A method to map presynaptic inputs
The cell bodies of sensory neurons—including those that sense pain—are found in the dorsal root ganglia (DRG), from which one long process extends to the periphery while another central process extends into the spinal cord and connects with projection neurons and spinal interneurons (see PRF related news stories here, here, and here). But before (or perhaps while) passing along their message, those central terminals receive synaptic inputs. Previous electron microscopy studies have shown evidence of presynaptic inputs onto primary afferent axons in the spinal cord from axons (axo-axonic) and dendrites (dendro-axonic) of other neurons, but the source and identity of those inputs was unknown.
To investigate the identity of the presynaptic inputs—in this case, inputs onto cutaneous sensory neurons—first author Yi Zhang and colleagues used an unbiased mapping approach. They began with a deficient monosynaptic rabies virus (ΔG-RV), a genetically engineered virus in which the gene encoding the rabies glycoprotein (rabies-G), which is critical for viral replication, is replaced with a gene for a fluorescent protein. Live viruses injected into an animal can infect neurons and travel to central terminals, but they cannot replicate and spread without rabies-G. In order to enable the fluorescently labeled virus to spread only to presynaptic neurons, the researchers used a mouse line with Cre recombinase-dependent expression of rabies-G (RΦGT). They crossed RΦGT mice with advillin-Cre or Trpv1-Cre mice, which express Cre in all sensory neurons or only in peptidergic nociceptors, respectively, giving them the ability to make rabies-G. In these mice, the labeled virus infected sensory neurons, replicated, and spread to neurons that made presynaptic inputs to the afferents, labeling those cells as well. “The mechanisms for selective spreading into presynaptic neurons are not very clear, but this property has made rabies virus an ideal tool to trace presynaptic inputs,” Wang told PRF in an email.
Whether the rabies virus traveled from primary sensory neurons only to presynaptic contacts in this study is unclear. A recent study showed that rabies virus can transfer from proprioceptive sensory neurons to postsynaptic motor neurons (Zampieri et al., 2014). In the current study, the virus did not label projection neurons postsynaptic to the afferents, “but that doesn’t prove there isn‘t transport going in that direction,” Todd said.
Labeling goes viral in the DRG, spinal cord, and brain
In the DRG, the virus moved across cell-cell connections of an uncertain type. After injection of labeled virus into the plantar skin of RΦGT mice that did not express Cre, labeling was limited to about 500 infected neurons per animal, from which the virus could not replicate or spread. In advillin-Cre RΦGT and Trpv1-Cre RΦGT mice, however, the virus spread to other neurons in the ganglia through interneuronal connections, labeling several thousand neurons. “The ganglion labeling indicated that neurons within the ganglion can release transmitters and influence each other’s activity,” Wang said, which had been previously shown, “but the extensive viral spreading certainly highlights the existence of extensive intraganglion transmission.” These connections probably do not resemble conventional synaptic structures, but instead may represent poorly understood points of contact between tightly packed neurons. The connections might be relevant to the clinical phenomenon of secondary allodynia, Wang postulated, in which the sensation of pain spreads from an injury site to a neighboring uninjured area.
Green fluorescent protein (GFP)-labeled rabies virus infection of DRG neurons and spreading into presynaptic spinal interneurons. Credit: Yi Zhang and Fan Wang.
In the spinal cord, labeled neurons were found mostly in the dorsal layers ipsilateral to the site of injection. In advillin-Cre RΦGT mice, the infection of 3,600 sensory neurons resulted in labeling of about 1,700 spinal interneurons that made synaptic connections with the afferents. Although 2,300 Trpv1-positive neurons were infected, far fewer spinal neurons were labeled in Trpv1-Cre RΦGT mice—only 85—supporting previous reports that peptidergic nociceptors receive fewer spinal synaptic inputs than other types of sensory neurons.
Molecular markers used to characterize the spinal interneurons showed that the vast majority contained the inhibitory neurotransmitters γ-aminobutyric acid (GABA) and/or glycine. A small percentage also contained the peptides somatostatin (SST) or preproenkephalin (PENK1), from which endogenous opioids are derived. Surprisingly, about 20 percent contained a vesicular glutamate transporter, indicating they were glutamatergic. “There is no previous evidence of axo-axonic synapses of excitatory neurons onto primary afferent terminals,” Todd told PRF. “It raises the question about whether transport of the rabies virus is exclusively to presynaptic inputs.” The authors concede that glutamatergic neurons may have been labeled at extrasynaptic release sites, or that anterograde transfer of virus may have labeled neurons postsynaptic to the afferents.
In the brain, the investigators found labeled neurons in the rostroventral medulla (RVM), a source of descending modulation of pain signals. About 100 RVM neurons were labeled in advillin-Cre RΦGT mice and about five in Trpv1-Cre RΦGT mice—which the authors surmise may represent only a fraction of RVM neurons that contact afferents in the spinal cord, because of labeling inefficiency. Although 17 percent of the RVM neurons were serotoninergic, about 80 percent contained GABA, and most of those also contained PENK1, suggesting that the presynaptic neurons provide dual GABAergic and opioidergic input to sensory afferents.
GFP-labeled axons from RVM Penk1-expressing neurons densely innervating layers I, II, and V of the spinal cord. The red staining is anti-vGlut1, which labels touch sensory afferents that project to layers III, IV, and V. Credit: Yi Zhang and Fan Wang.
Behavioral consequences
The researchers next wanted to determine the functional consequences of activating—or silencing—the input neurons from the RVM. They used optogenetic technology to render PENK1-expressing RVM neurons sensitive to light. Photo-activation of those RVM neurons resulted in primary afferent depolarization, as would be expected from GABA release at central terminals. (Although release of GABA normally results in postsynaptic membrane hyperpolarization, the high intracellular chloride found in sensory afferent central terminals causes them to depolarize in response to GABA. The end result is still inhibitory, though, because the depolarizing current acts as a shunt that dampens incoming action potentials.) The authors say the finding suggests that the RVM neurons made functional synaptic connections with the afferents.
Todd, however, urged caution at that interpretation. Although the finding of primary afferent depolarization does indicate that RVM signaling affects afferents, he told PRF, “it does not completely tie in with what we know about the anatomical arrangement in the spinal cord. Either we have missed those axo-axonic inputs from RVM, or presynaptic transfer is not the only route for the rabies virus. There is a burden of proof to show more evidence that these are direct presynaptic inputs,” he said, which could be provided by using an anterograde tracer to label RVM neurons and then characterizing their ultrastructure with electron microscopy. In fact, a previous study (Antal et al., 1996) performed such an experiment; among 107 synapses, the authors found no evidence of axo-axonic RVM inputs in the spinal cord. However, Wang told PRF in an email, “they only labeled a small number of neurons located right at midline, whereas the cells we found can be localized more to the lateral side of the RVM.”
The researchers next measured pain behaviors in mice in which RVM neurons were prevented from releasing GABA, while the endogenous opioid enkephalin remained available. When RVM neurons were silenced, mice displayed shorter withdrawal times in hot plate and tail-flick measures of heat sensitivity, but mechanical sensitivity was unchanged in the mice, suggesting that descending GABAergic RVM inputs are important for dampening incoming heat but not mechanical pain signals. Ablation of Penk1-containing RVM neurons, however—removing both their enkephalin and GABA signals—increased animals’ sensitivity to both mechanical and heat stimuli. Conversely, activation of Penk1-containing RVM neurons that expressed a designer receptor exclusively activated by a designer drug (DREADD) increased animals’ withdrawal latency to mechanical and heat stimuli, confirming the anti-nociceptive nature of the RVM inputs.
The new findings further add to the complexity of pain processing in the spinal cord and even the ganglia. The exact nature of the connections described in the new research remains unclear, but the area is ripe for further investigation.
Stephani Sutherland, PhD, is a neuroscientist, yogi, and freelance writer in Southern California.
Article Contributors:
