Single cell RNA sequencing of dorsal root ganglia identifies understudied druggable targets for osteoarthritis pain

Abstract

Purpose: Using surgical destabilization of the medial meniscus (DMM) to induce experimental murine knee osteoarthritis (OA), we have discovered that joint damage is accompanied by extensive anatomical and functional neuronal plasticity of intra-articular nociceptors innervating the knee. Nociceptors are characterized by the expression of the sodium channel, NaV1.8, and specific subsets of these neurons have been identified that are of particular interest for OA pain, because they are specialized in detection of noxious mechanical stimuli. These specialized subsets include Mrgprd+ fibers and TH+ fibers (the latter are known as C-low threshold mechanoreceptors, or C-LTMR), and we have shown that these nociceptor subsets are present in mouse knees with experimental OA. Here, we used sequencing of the entire transcriptome of single lumbar dorsal root ganglia (DRG) cells, which contain the cell bodies of knee afferents, in order to identify druggable targets expressed by mechanosensitive nociceptors. In particular, we sought to identify new targets by focusing on understudied proteins within the druggable genome (GPCRs, ion channels, kinases). Understudied proteins were defined by: (1) the protein has a low number of publications/citations, with a Jensen PubMed score of <50 and (2) investigation of the protein has minimal or no NIH funding. Methods: We collected (ipsilateral) L3-L5 DRG from 11 naïve male mice, aged 18 weeks. DRGs were acutely isolated from each mouse, pooled. and separated into a single cell solution following standard protocols. Single-cell RNA-seq (scRNA-seq) was then performed by using the 10x Chromium system for single cell sequencing library prep, followed by Illumina HiSeq sequencing (50bp, paired-end reads). This platform enables sequencing of the entire transcriptome of thousands of cells simultaneously. Bioinformatics can then be employed to cluster these cells into different groups based solely on similarities observed in patterns of gene transcription. Here, sequencing data were first processed using 10X Cell Ranger software, followed by further clustering analysis using the Seurat R package to identify lists of genes that serve as markers for different cell types. PANTHER classification software was used to annotate these gene lists in order to identify druggable targets. Results: The protocol yielded high cell viability (87%), and single cell sequencing data matched other reports in the literature (Fig 1). In addition to distinct subpopulations of neurons, including peptidergic and non peptidergic neurons, Mrgprd+ neurons and TH+ neurons, this approach also revealed groups of non-neuronal cells present in the DRG, such as glia, cells derived from blood vessels and immune cells (Fig.1). We then interrogated the data set to ask whether particular genes were expressed in any particular cell group. We focused on druggable targets: G-protein coupled receptors (GPCRs) because they are often promising drug targets, as well as kinases. Examples are shown in Fig. 2: it can be seen that the orphan GPCR, Gpr45, is expressed in many types of neurons, although hardly at all in non-neuronal cells. In contrast, the expression pattern of the GPCR Prokr1 (prokineticin receptor 1) is highly localized to the Mrgprd and TH neuronal populations, indicating that this receptor may be an interesting target to control pain signaling originating from these neurons.Examination of the data set also revealed several potential targets among the class of understudied kinases. For example, the kinase Hipk4 is hardly expressed at all in the DRG, while Cdc42bpa is highly expressed in virtually every population of neurons (Fig. 2). On the other hand, kinases which have highly selective expression patterns could represent interesting targets for treating OA pain- for example: Cdk15, which is expressed solely in non-peptidergic and TH neurons, Mapk4 which is expressed mostly in Mrgprd and TH neurons, and Stk32a which is virtually only expressed in the Mrgprd population. Conclusions: ScRNA-seq of the knee-innervating DRG provides an unbiased approach that allows us to access the complete biological potential of DRGs. Using this approach, we can identify novel druggable targets preferentially expressed by mechanosensitive Mrgprd+ and TH+ nociceptors. Future studies will employ genetic and pharmacological approaches to generate proof-of-concept data that targeting these selected gene products produces analgesic effects in mouse models of OA pain.View Large Image Figure ViewerDownload Hi-res image Download (PPT)