Chemicals and reagents were: diethylpyrocarbonate (DEPC), dimethylsufoxide (DMSO), buffers and salts were purchased from Sigma-Aldrich (Vienna, Austria); (1R,2R)-N-[(2R,3R)-2-amino-3-methoxybutyl]-N-[4-[4-(methoxymethyl)phenyl]phe-nyl]-2-pyridin-2-ylcyclopropane-1-carboxamide (RTI-13951-33) hydrochloride (MedChemTronica, #HY-112612A), 2-pyridin-2-yl-cyclopropanecarboxylic acid (2-PCCA) hydro-chloride (MedChemTronica, #HY-100013C), (1R,2R)-2-PCCA hydrochloride (MedChemTronica, #HY-100013A1), (1S,2S)-2-PCCA hydrochloride (MedChemTronica, #HY-100013B1), S-(4-nitrobenzyl)-6-thioinosine (= nitrobenzylmercaptopurine ribonucleosi-de/NBMPR) (Sigma-Aldrich, #N2255), pertussis toxin (PTX) (Sigma-Aldrich, #P7208), gallein (MedChemExpress, #HY-D0254).

Cell culture media and constituents were: Dulbecco’s modified Eagle medium (DMEM) with high glucose (4.5 g/L), heat-inactivated fetal bovine serum (Biowest, Riverside, MO, United States); collagenase, penicillin, streptomycin, cytosine-β-D-arabinofuranoside hydrochloride (Ara-C), insulin, poly-D-lysine, 4′,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich); nerve growth factor (Alomone Labs Ltd.); dispase II (Roche, Mannheim, Germany), trypsin (Worthington, Lakewood, NJ, United States).

Kits and reagents for molecular biology and RNAScope were: RNeasy Qiagen mini kit (Qiagen, Maryland, United States, #74104); NEB Next Poly(A) mRNA Magnetic Isolation Module and the NEB Next UltraExpress RNA Library Prep Kit (New England Biolabs); RNAscope Multiplex rat and RNAscope antisense probes against GPR88 (Rn-Gpr88, #445481) and RBFOX3 (Rn-RFOX3-C3, #436351-C3), NEFH (Rn-NEFH-C2, #474241-C2), P2RX33 (Rn-P2RX3-C3, #543401-C3), and CALCA (Rn-CALCA-C2, #317511-C2), Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics/Bio-Techne Ireland Limited, Dublin, United Kingdom, #323290); Fluoromount-G mounting medium (Thermo Fisher Scientific).

Primary cultures of DRG neurons were prepared as previously described with slight modifications (Ray et al., 2019). Pregnant Sprague-Dawley rats were purchased from Charles River (Sulzfeld, Germany) and housed in a Scantainer (Scanbur, Karlslunde, Denmark) under a 12-h light-dark cycle, at 20 °C–25 °C, 40%–70% humidity, and food as well as water ad libidum. Rat pups of either sex aged 10–15 days were killed by decapitation in accordance with the ARRIVE guidelines (Percie du Sert et al., 2020), the guidelines of Good Scientific Practice by the Medical University of Vienna (https://www.ris.bka.gv.at/Dokumente/BgblAuth/BGBLA_2004_I_118/BGBLA_2004_I_118.html) and the Austrian animal protection law (https://www.ris.bka.gv.at/Dokumente/BgblAuth/BGBLA_2012_I_114/BGBLA_2012_I_114.html). The experiments were performed ex vivo. Wild-type animals were sacrificed, and their dorsal root ganglia were used subsequently. According to §1 of the Austrian animal experimentation law (https://www.ris.bka.gv.at/NormDokument.wxe?Abfrage=Bundesnormen&Gesetzesnummer=20008142&FassungVom=2025-09-07&Artikel=&Paragraf=1&Anlage=&Uebergangsrecht=), an approval of an animal ethics committee is not required for ex vivo experiments.

Ganglia were dissected from all levels of the spinal column and digested in collagenase (1.5 mg/mL) and dispase II (3.0 mg/mL) for 30 min, followed by trypsin (0.25%) for 10 min at 37 °C, respectively. Subsequently, the tissue was washed twice with Ca²⁺-free Tyrode solution (NaCl 150 mM, KCl 4 mM, MgCl₂ 2 mM, glucose 10 mM, HEPES 10 mM, adjusted to pH 7.4 with NaOH) and triturated to produce a cell suspension in DMEM with high glucose (4.5 g/L) supplemented with 25,000 U/L penicillin, 25 mg/L streptomycin, 10 mg/L insulin, and 50 μg/L nerve growth factor. A total of 3,000 cells were seeded onto 35 mm culture dishes coated with poly-D-lysine. Cultures were maintained at 37 °C in a humidified 5% CO₂ atmosphere and supplemented with 5% heat-inactivated fetal bovine serum 2 h after plating. The medium was exchanged on day 1 after seeding. Ara-C (100 µM) was added to inhibit glial cell proliferation.

DRG neurons were harvested immediately after dissection (time point 0 h), and after 2 h, 24 h (=day 1), 72 h (=day 3) and 168 h (=day 7) in culture to capture time-dependent changes. Total RNA was isolated using the RNeasy Qiagen mini kit according to the manufacturer’s protocol. Sequencing libraries from total RNA of the samples were prepared at the genomics core facility of the Medical University of Vienna, using the NEB Next Poly(A) mRNA Magnetic Isolation Module and the NEB Next UltraExpress RNA Library Prep Kit for Illumina with Unique Dual Indices (UDIs) according to the manufacturer’s protocols. Quality control of the libraries was done on a Bioanalyzer 2100 (Agilent) using a High Sensitivity DNA Kit for monitoring correct insert size and quantified using Qubit dsDNA HS Assay (Invitrogen). Libraries were sequenced on a P2 flow cell of a NextSeq2000 instrument (Illumina) in 1 × 76 bp single-end sequencing mode. Reads in fastq format were generated using the Illumina bcl2fastq command line tool (v2.19.1.403) including trimming of the sequencing adapters. On average, 30 million reads per sample were generated. Reads in fastq format were aligned to the rat reference genome version rnor 6.0 with Ensembl 96 annotations using STAR aligner (Dobin et al., 2013) version 2.6.1a in 2-pass mode. Raw reads per gene were counted by STAR. Differential gene expression was calculated using DESeq2 (Love et al., 2014) version 1.44.0 (Zhu et al., 2019). Transcripts per million (TPM) were quantified with the RSEM (RNA-Seq by Expectation Maximization) software (Li and Dewey, 2011). Volcano plots were generated with the web application VolcaNoseR, based on differentially expressed genes (DEGs) extracted from DESeq2 analysis (Goedhart and Luijsterburg, 2020).

Rat DRG were excised and embedded in toto in Tissue-Tek® Optimal Cutting Temperature (O.C.T.) compound (Sakura Finetek purchased through Avantor - Austria VWR International GmbH, Vienna, Austria), frozen in liquid nitrogen, and stored at −80 °C until sectioning. Tissue sections (15 μm) were cut using a cryostat (CM3050S, Leica, Wetzlar, Germany) and mounted onto Superfrost Plus glass slides. Consecutive sections from each DRG were spaced at least 100 µm apart. The slides were stored at −80 °C until processing with the RNAscope Multiplex Fluorescent Reagent Kit v2. RNA detection was done with solutions provided in the kit according to the instructions of the manufacturer: sections were fixed in 4% paraformaldehyde (PFA) for 1 h at 4 °C, followed by two washes with phosphate-buffered saline and sequential dehydration in 50% ethanol for 5 min, 70% ethanol for 5 min, and 100% ethanol for 10 min at room temperature. The slides were then allowed to air-dry. Hydrophobic barriers were drawn around the tissue sections using a hydrophobic barrier pen. The sections were treated with hydrogen peroxide for 10 min, rinsed with sterile, diethylpyrocarbonate- (DEPC-) treated water and subsequently subjected to a protease IV digestion for 30 min at room temperature. After rinsing with DEPC-treated water, sections were covered with a solution containing rat RNAscope antisense probes against GPR88 and against the neuronal markers RBFOX3, NEFH, P2RX3 and CALCA in a 50:1 ratio (GPR88 probe to the other probes). The slides were placed into the HybEZ II oven (Advanced Cell Diagnostics) for 2 h at 40 °C. Subsequently, sections were rinsed twice in RNAscope wash buffer for 2 min and stored overnight in 5X SSC buffer (750 mM NaCl and 75 mM trisodium citrate, pH adjusted to 7.0 with HCl) at room temperature. On the next day, the slides were rinsed twice with RNAscope wash buffer, followed by amplification with Amp 1 for 30 min, Amp 2 for 30 min, and Amp 3 for 15 min at 40 °C, with additional rinses after each step. HRP-C1, HRP-C2, and HRP-C3 signals were developed by adding the corresponding RNAscope Multiplex FL v2 HRP reagents to the slides, followed by incubation in the HybEZ Oven at 40 °C for 15 min. Slides were then rinsed twice with wash buffer at room temperature. The sections were incubated with TSA Vivid Fluorophore 570 for C1-tagged probes and TSA Vivid Fluorophore 520 for C2- and C3-tagged probes for 30 min at 40 °C and then washed twice with wash buffer. Thereafter, a solution containing the HRP blocker was added and slides were incubated for 15 min at 40 °C before a final rinse with wash buffer. Nuclei were visualized by brief exposure to a solution containing 0.3 µM DAPI. The slides were then mounted with Fluoromount-G mounting medium, covered with a glass coverslip and stored at 4 °C. Images were captured by confocal laser scanning microscopy (Zeiss LSM 980 Airyscan 2) and processed using ImageJ. RNAscope signals were analysed qualitatively. According to criteria set forth by the manufacturer “A guide to RNAscope analysis” – http://fab.cba.mit.edu), cells are considered positive if they have ≥1 dot/cell. In our case, all neurons, which were scored positive had >2 dots/cell.

Twenty-four hours after seeding, neuronal cultures were incubated in the absence or presence of the indicated concentrations of RTI-13951-33 hydrochloride, 2-PCCA hydrochloride, 1R,2R-2-PCCA hydrochloride, 1S,2S-2-PCCA hydrochloride, or NBMPR for 24 h. In some instances, cultures were also incubated with 100 ng/mL PTX. Neurons and their neurite extensions were visualized by phase contrast microscopy (Nikon Eclipse TS100). Sholl analysis was performed by manually placing the neuronal soma into the center of the concentric circles, followed by measuring the distance (=radius) between the center and the most distant point reached by the longest neurite using the ImageJ (version 2.16.0) Neuroanatomy-plugin. This approach is conservative, because it underestimates total neurite length. However, it has the advantage to avoid ambiguities arising from tracing neurite extensions. Quantification was performed only on neurons, where neurite extensions were clearly identifiable. Neurons without neurite extension were not analyzed.

The peripheral nervous system has an inherent ability for self-repair, i.e., severed axons can sprout and re-innervate their target areas. Regenerating neurons activate a transcriptional program to support axon outgrowth. We surmised that this regenerative process was–at least in part–recapitulated in vitro, if sensory neurons were placed in culture: isolation of DRG neurons severs their axonal projections, but these begin to sprout again over a time course of several days. Accordingly, we monitored the evolution of transcriptional changes by RNA sequencing: RNA was isolated from freshly isolated DRG neurons (time point = 0 h) and 2, 24, 72, and 168 h after placing the neurons in culture. This time-resolved transcriptomic analysis aimed to capture the dynamic gene expression landscape underlying axonal regeneration. RNA sequencing revealed a total of 1,322 genes that were differentially expressed across all time points (DESeq2, adjusted p < 0.05) relative to the baseline (0 h). The most extensive transcriptional changes occurred after 2 h, with 386 genes upregulated and 678 downregulated, reflecting a rapid and robust early response to axonal injury (Figure 1A). This pattern reflects neuronal injury and the early phase of neuronal regeneration, which includes the rapid initiation of neurite outgrowth and gene activation. The gene expression remained significantly altered at 24 h (second set of bars in Figure 1A, 164 upregulated, 381 downregulated), suggesting continued transcriptional activity beyond the initial response. This stage likely represents the transition from the early to the late phase of regeneration. From 72 h onward, the number of DEGs declined markedly, consistent with a shift from an actively reprogramming to a more transcriptionally stable state that supports axonal elongation and maturation. We compared DEGs (DESeq2; adjusted p < 0.05, |log₂ fold change| ≥ 1) between day 1 (24 h) at later time points, i.e., day 3 (72 h) and day 7 (168 h), to capture the transcriptional shift from early reprogramming to the more stable late phase of regeneration. This analysis identified 791 and 1,654 DEGs on day 1 relative to day 3 and day 7, respectively. Here, we focused on GPCRs and SLC transporters due to their relevance as druggable targets. One GPCR gene and 15 SLC genes were consistently differentially expressed in both comparisons (day 1 vs. day 3 and day 1 vs. day 7): in this relative comparison, transcripts encoding the GPCR GPR88 and SLC29A1, the equilibrative nucleoside transporter-1, were upregulated and downregulated, respectively (Figure 1B). SLC29A1 is druggable: in fact, the inhibitors of SLC29A1 dilazep and dipyridamole are approved drugs. Furthermore, NBMPR is an experimental inhibitor of SLC29A1, which has been used for some 5 decades as a research tool to interrogate the role of SLC29A1 (Pickard and Paterson, 1972; Paterson et al., 1980). We examined, if inhibition of SLC29A1 affected neurite outgrowth. The incubations were done in the presence of NBMPR rather than dipyridamole, because dipyridamole does not have any appreciable affinity for rat ENT1/SLC29A1, whereas the transporter is inhibited by NBMPR with an affinity in the range of 5 nM (Yao et al., 1997). However, incubation in the presence of 500 nM NBMPR, a quasi-saturating concentration (=100*K_D), did not affect neurite outgrowth (Supplementary Figure S1). The other SLCs listed in Table 1 represent transporters, which translocate nutrients (glucose and amino acids), metabolites (monocarboxylates, citrate, uric acid), Zn²⁺, excitatory amino acids and choline. They are unlikely to represent suitable drug targets. Thus, we did not further pursue the solute carriers, but we focused on GPR88, which is also druggable: while the endogenous/orthosteric agonist of GPR88 is not known, GPR88 is activated by several synthetic compounds, which were reported to act as positive allosteric modulators, including 2-PCCA, RTI-13951-33 and RTI-122 (Jin et al., 2014; Jin et al., 2018; Chen et al., 2022; Rahman et al., 2023).

The volcano plots shown in Figure 1B are based on a comparison of transcript levels in early and later stage DRG cultures. It is also instructive to examine the changes in transcript levels over the entire time course: transcripts for GPR88 rapidly reached a peak, i.e. 24 h after the DRG neurons had been placed into culture (Figure 1C).

We verified our hits by mining the data compiled and deposited by Xu et al. (2022) in the DRGProfile database (http://121.41.67.1:3838/DRGProfile/) for time-dependent changes in transcript levels after peripheral nerve injury in vivo. Among the identified hits, GPR88 emerged as a particularly compelling candidate: GPR88 transcripts were consistently upregulated in rat DRG across various in vivo peripheral nerve injury models, i.e., crushing of (bars labeled ScNI_crush in Figure 2A) or ligation of the sciatic nerve (bars labeled ScNI_ligation in Figure 2A) and ligation (bars labeled SpNI_ligation, Figure 2A) and transection of the spinal nerves (bars labeled SpNI_transection, Figure 2A). In addition, induction of GPR88 appears to represent a general response to nerve injury: time-dependent accumulation of GPR88 transcripts can also be seen after spinal cord injury in the segments rostral and caudal to the lesion (Figure 2B; data extraction from the deposition by Yu et al., 2019). Finally, the transcriptomic data retrieved from the DRGProfile database (deposited by Yang et al., 2021) showed that gpr88 was subject to pronounced developmental regulation: transcript abundance is low during the early embryonic stages, on embryonic days 9 and 11 (E9d and E11d in Figure 2C), but rises substantially on embryonic day 14 (E14d in Figure 2C), i.e., prior to the formation of synaptic contacts, which is initiated between embryonic days 15 and 17 (May and Biscoe, 1975). Transcript levels of GPR88 peak between embryonic day 18 and postnatal day 3 (E18d and P3d in Figure 2C) and subsequently decline to reach a steady state in postpartum weeks 8 and 12 (P8w and P12w in Figure 2C).

DRG comprise functionally distinct neuronal subtypes specialized in transmitting different sensory modalities, including touch, temperature, proprioception, and pain. Transcriptomic profiling of single cells allowed for classifying these subtypes based on their characteristic gene expression patterns. Commonly defined groups include Aβ low-threshold mechanoreceptors (LTMRs), Aδ-LTMRs, C-LTMRs, peptidergic nociceptors (PEPs), non-peptidergic nociceptors (NPs), cold-sensitive neurons (TRPM8⁺), and proprioceptors (Usoskin et al., 2015; Wu et al., 2021). Among these, we selected the most prevalent subtypes to verify GPR88 expression across distinct DRG neuron populations using multiplex fluorescent mRNA in situ hybridization (RNAscope): the analysis confirmed robust GPR88 expression in all neurons, as identified by the marker RNA binding fox-1 homolog 3/neuronal nuclei (RBFOX3/NeuN; top row in Figures 3A,B). In line with this, GPR88 was consistently detected across distinct neuronal subtypes, including: (i) neurofilament heavy chain (NEFH) positive neurons, giving rise to large myelinated Aβ/Aα fibers involved in mechanoreception and proprioception (second row in Figures 3A,B), (ii) calcitonin gene-related peptide-α (CALCA) positive neurons, identifying small PEPs (giving rise to C-fibers) associated with pain and temperature sensing (third row in Figures 3A,B), and (iii) purinergic receptor 3 (P2RX3) positive neurons, labeling NPs (giving rise to C-fibers) involved in pain perception (bottom row in Figures 3A,B). Statistical analysis confirmed ubiquitous GPR88 expression in all neurons identified by the pertinent marker (Figure 3C). These results confirm that GPR88 is broadly expressed across major DRG neuronal subtypes. Conversely, it is evident from the images shown in Figure 3A that the non-neuronal cells, which were present in the slices prepared from the DRG, lacked any detectable reactivity for the GPR88 probe: their nuclei were visualized by DAPI (Figure 3A, third row) but were devoid of both, the fluorescence signal of the pertinent neuronal marker and of GPR88 (cf. merged images in the fourth row of Figure 3A).

Several compounds, including 2-PCCA and its congeners, were identified as synthetic agonists of GPR88 some 10 years ago (Jin et al., 2014). RTI-13951-33 is an analogue of 2-PCCA with improved selectivity and pharmacokinetics (Jin et al., 2018). The structure of GPR88 was elucidated at atomic resolution by cryo-electron microscopy and revealed 2-PCCA in an allosteric site rather than the canonical orthosteric ligand binding pocket (Chen et al., 2022). Accordingly, we tested whether these allosteric activators enhanced neurite outgrowth in cultured DRG neurons. Cultures of DRG neurons were treated with increasing concentrations of RTI-13951-33 or 2-PCCA for 24 h; thereafter microscopic images were captured: representative micrographs obtained from cultures incubated in the absence (Figure 4A top) and presence of a quasi-saturating concentration (3 µM) of RTI-13951-33 (Figure 4A middle) or of 2-PCCA (Figure 4A bottom) illustrate that incubation of DRG neurons with either compound for 24 h promoted neurite outgrowth. We generated concentration-response curves by quantifying the distance covered by the longest neurite in randomly selected neurons: incubation of DRG cultures in the presence of both, RTI-13951-33 (Figure 4B, top panel) and 2-PCCA (Figure 4B, middle panel) enhanced neurite growth in a concentration-dependent manner. A statistically significant increase in neurite extension was observed at concentrations ≥0.1 µM (Kruskal–Wallis test followed by Dunn’s post hoc test). From the concentration-response curve, we calculated EC₅₀ values of 22.3 ± 9.8 nM and 67.8 ± 44.8 nM for RTI-13951-33 and 2-PCCA, respectively.

Our assessment was based on randomly selecting neurons for measuring neurite outgrowth. We verified that there was no selection bias by comparing the distribution of soma sizes of the neurons selected for assessing the action of RTI-13951-33 and of 2-PCCA with those selected from the corresponding control cultures: the cumulative distributions were comparable (Figure 5A). We also plotted the neurite extension of neurons in control cultures and in cultures treated with RTI-13951-33 and of 2-PCCA (each at 1 µM) as a function of soma size and examined, if there was any correlation (Figure 5B). This analysis showed that neurite length was independent of soma size (R ² = 0.0066, 0.0071 and 0.0673 for neurons incubated under control, in the presence of RTI-13951-33 and of 2-PCCA, respectively). Thus, consistent with the RNAscope results, which indicated that all major subtypes of neurons expressed GPR88, the promotion of neurite outgrowth by RTI-13951-33 and of 2-PCCA was not confined to a subpopulation on neurons.

2-PCCA is a racemic mixture consisting of the 1R,2R- and 1S,2S-enantiomers. The 1R,2R-enantiomer was previously shown to be approximately 5-fold more potent in activating GPR88 than the 1S,2S-isomer (Jin et al., 2014). Accordingly, we capitalized on the availability of the enantiomers to further confirm that the neurite outgrowth promoting action of 2-PCCA and its analogue RTI-13951-33 were due to engaging GPR88 rather than to an off-target effect. 1R,2R-2PCCA was substantially more potent than 1S,2S-2PCCA (Figures 6A,B): the EC₅₀-values estimated from the concentration-response curves were 9.1 ± 5.0 nM and 138.3 ± 98.4 nM for the 1R,2R- and the 1S,2S-enantiomer, respectively.

Upon activation, GPR88 preferentially recruits G_i/o family members (Jin et al., 2014; Cheng et al., 2022; Watkins and Orlandi, 2021), coupling to other classes of heterotrimeric G proteins (G_s, G_q, G_12/13) was not detectable (Watkins and Orlandi, 2021). Thus, the action of the allosteric activators 2-PCCA and its analogue RTI-13951-33 are predicted to be blocked by pertussis toxin. This prediction was verified: if cultured DRG neurons were treated with pertussis toxin (100 ng/mL), 2-PCCA (1 µM) and RTI-13951-33 (1 µM) failed to promote neurite outgrowth (right-hand set of bars in Figure 7A). Importantly, PTX did not affect the neurite length in control cultures incubated in the absence of GPR88 allosteric agonists (cf. first two bars in Figure 7). We further probed the signaling pathway by exploring the effect of gallein, a small molecule high-affinity inhibitor of Gβγ-dependent signaling (Lehmann et al., 2008; Seneviratne et al., 2011): incubation in the presence of gallein (10 µM) did not affect neurite outgrowth under control conditions (cf. first two bars in Figure 7B), but it abolished the stimulation of neurite outgrowth by both, 2-PCCA (1 µM) and RTI-13951-33 (1 µM).