How many potential analgesic drug targets identified in animal research have failed to reach clinical use? It’s an answer that’s hard to estimate, but the problem of mistranslation of drug targets has been at the heart of analgesic drug development for decades.
Now, single-cell transcriptomics analysis of human dorsal root ganglia (DRG) neurons provides a new way to identify therapeutic targets that are relevant for human disease. The study from Ted Price and colleagues at the University of Texas at Dallas, US, was published in Science Translational Medicine on 16 February 2022.
“This is ground-breaking work which essentially allows precision medicine to go forward in the pain field. It’s a hugely valuable resource for all groups who are working on ways to treat pain and somatosensory disorders. Scientists using animal models will also be able to confirm the relevance of their therapeutic targets, enhancing the focus on translational targets,” commented Steve Davidson, University of Cincinnati, US, who was not a part of the study.
The paper from Price’s group goes hand-in-hand with a similar paper, published in eLife on 26 November 2021, from Steve Davidson’s and Nicholas Ryba’s groups.
“It really impressed me how similar our findings were with the Davidson group’s paper. What excites me from our papers is that human physiologists, especially those who use microneurography, can use this dataset and do in vivo pharmacology in people, and test some very specific hypotheses. I’m hoping this will blow everything wide open so we can figure things out with more precision than before,” Price said.
Single-cell RNA-seq of human DRG neurons
As Price’s group writes in the introduction of the paper, “There has been an unsatisfying failure to translate preclinical work on peripheral pain mechanisms.” Therefore, the goal of the study was rather simple – create a database of the RNA profiles of human DRG neurons that can be used to mine for drug targets that are relevant for human pain disorders.
A seemingly straightforward goal, but not a simple endeavor. The group had previously published RNA sequencing (RNA-seq) data of human DRG neurons; however, this was “bulk RNA-seq,” or when investigators excise the whole DRG tissue and then extract RNA.
“With bulk RNA-seq, you get the full characterization of the DRG, but it isn’t cell specific. Our aim in this study was to have single-cell resolution to get a more accurate picture of the neuronal transcriptome. We needed new technologies for this, so we turned to spatial transcriptomics approaches with Visium Technologies,” co-first author Diana Tavares-Ferreira told PRF.
For samples, the group had a longstanding collaboration with the Southwest Transplant Alliance, who coordinated DRG tissue recovery following organ donation surgeries. Eight samples were processed for RNA-seq in this study, four females and four males. Next, DRG tissue was permeabilized, and mRNA was barcoded and processed for library preparation.
After data analysis, the group selected 12 clusters of neurons based on their overlapping gene expression of known neuronal markers. The process of cluster identification, Price described, was something of a “cat-and-mouse” game. Neuron clusters were identified based on unbiased statistical criteria, and then functionality was ascribed to each cluster based on knowledge from decades of research (often from animal model data). The overall RNA-seq dataset produced well-verified findings – TRPM8 marks a population of cold nociceptors, and PVALB/NTRK3 marks a population of slowly adapting low-threshold mechanoreceptors (LTMRs), and so forth.
Breaking dogmas: Do we need to rethink C-fibers in human physiology?
The overall architecture of human DRG single-cell transcriptomes were largely comparable to recently published mouse and macaque single-cell DRG transcriptomes. After all, we survive and thrive in similar environments, so it’s not surprising we share the same somatosensory repertoire to respond to them. However, the dataset did detect some subtle differences in the tuning of the system between species, particularly in C-fiber populations.
One major difference between the human and mouse transcriptomes was the expression of markers of peptidergic (CALCA, expressing CGRP) and non-peptidergic C-fibers (PTX3R, IB4). The Price group’s datasets showed a very high overlap between peptidergic and non-peptidergic markers, suggesting that the two populations are very intermingled in humans. Furthermore, the Davidson group found the same results in their RNA-seq dataset.
“The implications for this are huge,” Davidson said. “It means the breakdowns of C-fiber groups seen in mice might not make sense in human physiology. There have been thousands of papers based on the classification of peptidergic versus non-peptidergic nociceptors, and those papers now need to be evaluated in the context of these new human transcriptomics data.”
Another major difference was the expression of TRP channels in C-fibers.
“One thing we found here was that TRPV1 is expressed in most, if not all, nociceptors in humans. In mice it’s only ~30% of neurons, suggesting a major difference in the role of TRPV1 in nociception in humans versus mice,” said co-first author Stephanie Shiers.
Again, these findings overlap with evidence from the Davidson group.
“In mice, most TRPM8+ cold-sensing neurons don’t express the mechanoreceptor PIEZO2, but in our paper, we found that humans have two populations of TRPM8+ neurons – those with, and those without, PIEZO2. This suggests that there’s a greater prevalence of polymodality in humans caused by more expression of TRP channels across neuronal populations,” Davidson told PRF.
Where are all the C-LTMRs?
Evidence for greater levels of polymodality was also found in the RNA-seq datasets when trying to identify C-LTMRs, a class of C-fibers that have been described by some as having a direct line to affective and social touch.
“We had a hard time identifying C-LTMRs. We just couldn’t see markers from the mouse literature like TH [tyrosine hydroxylase] in either the RNA-seq or in RNAscope validation experiments. Instead, we identified this population with GFRA2 and PPB expression levels, but they were also expressed in the itch population, so it wasn’t very specific,” Tavares-Ferreira said.
Davidson’s group also struggled to identify a class of C-LTMR neurons, and instead grouped putative C-LTMRs with likely mechanically sensitive chemonociceptors with weak similarity to mice. This class predicts a less clearly defined role in affective and social touch due to potential responsivity to other types of stimuli than mechanical.
However, Davidson was keen to point out that this might also be attributed to incomplete knowledge of C-LTMR gene expression profiles.
“It was striking that in both papers we had a hard time identifying C-LTMRs. It’s possible that we don’t know the appropriate gene principal components that define C-LTMRs in humans. After all, we really can’t define neuronal populations simply by the expression of genes; we need to know what these cells are responsive to and correlate this with gene expression,” Davidson said.
Subtle sex differences
One of the goals of this study was to identify any sex differences in the expression of RNA molecules in human DRGs. While the authors were not anticipating any major differences between male and female neuron clusters, they did predict to see some specific translational differences, as per previous literature.
In total, only 44 genes were different between the four female and four male samples, but there were some key sex differences when the researchers looked at neuron subtypes. For one, pruritogenic receptor-enriched neurons had a high number of differently expressed genes, suggesting molecular differences in the mechanisms of itch between men and women. There was also a higher expression of CALCA in females, suggesting possible sex differences in CGRP-mediated nociception and pain.
“It was so important that they used equal numbers of males and females, and were able to compare differences. Unfortunately, we couldn’t do this in our paper. The Price group only had a group of eight, but it’s an important start to investigating sex differences,” Davidson said.
A resource that opens doors for drug discovery
One of the most important aspects of Price’s and Davidson’s studies is their utility as databases for future pain research. Previous transcriptomics studies have either been in model animals, like the mouse, or have been bulk RNA-seq studies that preclude identification of single-cell identities. Now, researchers have a way to screen for important molecules in pain and somatosensation, and potential therapeutic drug targets in human disease.
Price’s team show this potential here, too, by comparing expression of genes considered to be important pharmacological drug targets between their human DRG data and the publicly available mouse data. Here, they highlight several key differences between human and mouse in the expression of potential drug targets. For example, the G protein-coupled receptor LPAR3, associated with neuropathic pain, showed differential expression across nociceptor subtypes in humans and mice, suggesting potential differences of LPAR3 drug targets between species.
“We found this to be the case for lots of drug targets – the population expression levels were quite different between human and mouse DRG, which predicts failure of some drugs in development. Hopefully, we now have the starter information we need to do a better job of what targets might work clinically, and we can start this earlier,” Price said.
Similar work from the Price lab analyzing lung sensory neuron transcriptomes is already making an impact in the medical community.
“In 2020, we published a paper looking at interactions of ligands and receptors in bronchiolar lavage fluid from COVID-19 patients. Our screen highlighted increased expression in several molecules, such as the NMDA GluN2B receptor. From this, a clinical study had good success based off one of the interactions, suggesting an effective treatment for COVID-19. This is a good representation of what we can hopefully do with pain treatment in this current dataset,” Shiers said.
Fred Schwaller, PhD, is a freelance science writer based in Germany.
Featured image: Cover art for Science Translational Medicine (online version) – volume 14, issue 632, 16 February 2022. Credit: Stephanie I. Shiers/Science Translational Medicine