Authors: Gith Noes-Holt, Kathrine L. Jensen, Raquel Comaposada-Baró, Sara E. Jager, Mette Richner, Carolyn M. Goddard, Marco B.K. Kowenicki, Line Sivertsen, Lucía Jiménez-Fernández, Rita C. Andersen, Jamila H. Lilja, Andreas H. Larsen, Sofie P. Boesgaard, Grace A. Houser, Nikolaj R. Christensen, Antonio Marino, Anke Tappe-Theodor, Michael Wierer, Christian B. Vægter, Rohini Kuner, Kenneth L. Madsen, Andreas T. Sørensen
Categories: Article, PICK1, protein interacting with C-kinase 1, neuropathic pain, pain treatment, neuropathic pain treatment, gene therapy, AAV therapeutics
Source: Cell Reports Medicine
Authors: Gith Noes-Holt, Kathrine L. Jensen, Raquel Comaposada-Baró, Sara E. Jager, Mette Richner, Carolyn M. Goddard, Marco B.K. Kowenicki, Line Sivertsen, Lucía Jiménez-Fernández, Rita C. Andersen, Jamila H. Lilja, Andreas H. Larsen, Sofie P. Boesgaard, Grace A. Houser, Nikolaj R. Christensen, Antonio Marino, Anke Tappe-Theodor, Michael Wierer, Christian B. Vægter, Rohini Kuner, Kenneth L. Madsen, Andreas T. Sørensen
Chronic neuropathic pain’s impact, persistence, and limited treatments render it relevant for gene therapy. Here, we describe the development and application of self-assembling dimeric peptide inhibitors of the pain-associated scaffolding protein PICK1 (protein interacting with C-kinase 1), delivered by adeno-associated viral (AAV) vectors. In mice, these peptides prevent mechanical allodynia in inflammatory and neuropathic pain models and reverse neuropathic pain for up to 1 year. Targeting somatosensory pathways relieves pain without overt side effects, while selective transduction of dorsal root ganglion (DRG) neurons is sufficient to provide pain relief. Using proteomic and phosphoproteomic analysis of DRG tissue, we identify regulation of protein kinase C alpha (PRKCA) as a candidate that potentially shapes this pain-relieving phenotype. We finally confirm PICK1 expression and peptide target engagement in human donor tissue, supporting the potential of AAV-encoded PICK1 inhibitors as a clinically meaningful strategy for neuropathic pain conditions.
Exploiting gene therapies for refractory pain conditions holds promise for overcoming several adversities of current pain medicine, including low efficacy, dose-limiting side effects, and opioid addiction.^1^^,^^2^^,^^3^ Today’s drawbacks not only affect patients and clinical practice but also have severe socioeconomic consequences, as confirmed by the present opioid crisis.^4^ It is estimated that one in five adults in the US is affected by chronic pain,^5^ of whom approximately half suffer from chronic neuropathic pain.^6^ This condition arises from a lesion or disease of the somatosensory nervous system,^7^ and no single treatment is currently able to prevent or cure neuropathic pain.
The safest and most widely used gene therapy vector for the nervous system is the adeno-associated viral (AAV) vector. It is characterized by its replication-deficient, low-immunogenic, and long-lasting expression profile.^2^^,^^3^^,^^8^ Its innate ability to deliver gene expression locally and to a defined cell population adds powerful dimensions to AAV-tailoring for disease treatment. While available treatments for chronic neuropathic pain exclusively target ion channels, receptors, and transporters,^9^ we envisioned that targeting intracellular pathways might be more efficacious and circumvent systemic side effects.
Synaptic PDZ (PSD-95/DLG-1/ZO-1) domain scaffold proteins are vital for the integrity and modulation of neuronal signaling and, therefore, are clinically relevant pain targets. PDZ proteins regulate discrete, dynamic signaling functions, such as protein transport, ion channel signaling, and other signaling systems, through protein-protein interactions.^10^^,^^11^ Several synthetic peptide inhibitors that block specific PDZ-domain-mediated protein-protein interactions have been developed to modulate synaptic function.^12^^,^^13^^,^^14^^,^^15^^,^^16^ Such peptides are characterized by their C-terminal residues, typically 5–13 amino acids in length, with the identical motif of a known protein-interacting partner. Notably, peptides targeting PSD-95 (postsynaptic density protein 95) disrupt NMDA-type (N-methyl-D-aspartic acid) ionotropic glutamate receptor-dependent signaling and can attenuate animal pain-related behaviors.^17^^,^^18^ We previously addressed another PDZ protein, PICK1 (protein interacting with C-Kinase 1), as a target for pain modulation.^15^^,^^19^^,^^20^ This protein is conserved, expressed in neuronal tissue, and best known for its trafficking role of AMPA-type (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) ionotropic glutamate receptors.^21^^,^^22^^,^^23^ We initially found that a synthetic monovalent peptide inhibitor targeting PICK1 was not efficacious in pain models. To improve potency, we instead constructed bivalent analog peptides with two C-terminal HWLKV amino acid motifs fused by a polyethylene glycol (PEG) linker, conjugated to the trans-activator of transcription (Tat) peptide sequence (TPD5) or a lipid chain (mPD5), rendering the peptides cell-permeable. Administration of such PICK1-targeting peptides fully reversed mechanical allodynia for 3–4 h in the spared nerve injury (SNI) model of neuropathic pain.^15^^,^^20^
In the current study, we developed an AAV-based strategy that enables recombinant peptide inhibitors to post-translationally assemble into a dimeric configuration mimicking the bivalency of TPD5 and mPD5. Two C-terminal HWLKV motifs can thereby be freely positioned to interact with PICK1. We assessed the therapeutic efficacy of this approach in mouse pain models and harnessed AAVs' ability to deliver the peptide effectively and selectively to neurons. Our data revealed that various pain modalities in both inflammatory and neuropathic pain models were prevented or reversed without causing side effects. Non-specific targeting of dorsal root ganglion (DRG) neurons alone was sufficient to reverse pain hypersensitivity, whereas selectively targeting only nociceptive neurons proved ineffective. As putative mechanisms explaining the pain-relieving phenotype, we identified, within DRGs, phosphoregulation of substrates for kinases previously reported to be involved in pain conditions. We finally verified PICK1 expression and peptide target engagement in human tissues, reinforcing PICK1 as a promising target for AAV-based therapies tailored to address refractory neuropathic pain conditions.
Inspired by our previous work on synthetic bivalent peptide inhibitors of PICK1,^15^^,^^20^ we conceived an AAV-deliverable recombinant design enabling posttranslational self-assembly of monomeric peptides into a dimer. The gene cassette consisted of an N-terminal dimerization domain originating from the yeast basic-region leucine zipper (bZIP) transcription factor (GCN4, General Control Nondepressible 4), a flexible glycine linker region (GGGGS), and a C-terminal PICK1 PDZ-binding motif (HWLKV). We created two variants, one with the native GCN4 sequence containing aspartic acid (D) and serine (S) at positions 7 and 14, respectively, and another with two prolines (P) at the same positions (Figure 1A), predicted to render the homo-dimerization more dynamic (Figures S1A–S1C). Biochemical experiments with biotinylated peptides confirmed their self-assembly and PICK1 binding (GCN4-HWLKV, Ki,App = 119 nM; GCN4(7P14P)-HWLKV, Ki,App = 548 nM) (Figure S1D) and that GCN4-HWLKV, but not GCN4(7P14P)-HWLKV, drove PICK1 oligomerization (Figures S1E–S1H), similar to the bivalent peptides TPD5 and mPD5. A biotinylated monomeric peptide (troponin-HWLKV) failed to pull down PICK1 in mouse tissue. However, introducing leucine-zipper-promoting leucine residues into troponin to mediate dimerization resulted in PICK1 binding, similarly to GCN4-HWLKV and GCN4-(7P14P)-HWLKV (Figures S1I and S1J). Next, we confirmed peptide expression in primary neuronal cell cultures exposed to AAVs using the human synapsin (hSyn) promoter for selective, constitutive, pan-neuronal expression. AAV vectors were manufactured in-house, including two therapeutic AAV8-Di-C5 (AAV8-hSyn-GCN4-HWLKV) and AAV8-Di(7P14P)-C5 (AAV8-hSyn-GCN4(7P14P)-HWLKV). Primary mouse cortical cultures were transduced with either of these vectors or two negative control AAV8-tdTomato (AAV8-hSyn-tdTomato) and AAV8-Di-GS (AAV8-hSyn-GCN4-GGGGS-GGGGS), the latter designed as a PICK1 non-binding variant. Peptide expression, as determined by GCN4 immunocytochemistry, was confined to the cytoplasm and excluded from the nucleus (Figure 1B). We next evaluated the vectors’ analgesic efficacy in the complete Freund’s adjuvant (CFA) model of inflammatory pain. Mechanical allodynia was quantified as the paw withdrawal threshold to von Frey filament stimulation, measured repeatedly before and after a single intrathecal (i.t.) injection at the L5–L6 intervertebral disc level in wild-type (WT) male mice. After approx. 3 weeks, inflammatory pain was induced in the left hind paw by intraplantar CFA injection. Both AAV8-Di-C5 and AAV8-Di(7P14P)-C5 prevented allodynia, unlike the control, AAV8-tdTomato (Figure 1C). In addition, the pain threshold remained unaltered in the non-inflamed (contralateral) paw, regardless of treatment (Figure 1C), suggesting an analgesic effect without numbing (anesthetizing) mechanical sensitivity. After 11 days, the paw withdrawal threshold for AAV8-tdTomato returned to baseline levels, highlighting the transient nature of this model. Subsequently, the opioid receptor antagonist naltrexone (NTX) was administered once on days 11 and 40, based on the assumption that it would reinstate allodynia only in AAV8-tdTomato controls.^24^ Indeed, NTX selectively reinstated allodynia in this group (Figure 1C). In addition, two designs of the AAVs, with the peptide sequences tagged with enhanced green fluorescent protein (EGFP), were also efficacious in this CFA experiment (Figure S2A). Additionally, administration of AAV8-Di-C5, but not AAV8-tdTomato, once CFA-induced hypersensitivity had resolved, occluded re-instatement at subsequent NTX administrations on days 40 and 60 (Figure S2B). Together, these experiments indicate that the timing of treatment, whether before or after CFA administration, yields comparable effects. Next, we tested the effect of AAV8-Di-C5 and AAV8-tdTomato on heat-induced hyperalgesia. CFA lowered threshold temperatures for AAV8-tdTomato, while this effect was smaller and non-significant for AAV8-Di-C5. However, the threshold did not differ significantly between groups (Figure S2C), suggesting partial relief of heat hyperalgesia.Figure 1AAV-encoded dimeric peptides targeting PICK1 fully abolish CFA-induced mechanical allodynia(A) Conceptual design of AAV-encoded dimeric PICK1-targeting peptide inhibitors with HWLKV motifs. Two different amino acid variants within the GCN4 helical structure are shown.(B) Immunocytochemistry performed on cortical neuronal cultures transduced with different AAVs. GCN4 immunoreactivity (green), DAPI nuclei staining (blue), and tdTomato autofluorescence (red) (AAV8-tdTomato only). N = 2 mice. Scale bars, 5 μm (the first three) and 20 μm (the last).(C) von Frey-determined paw withdrawal thresholds in CFA male WT mice. AAV8-tdTomato was compared to AAV8-Di-C5 and AAV8-Di(7P14P)-C5, including the two EGFP-containing vectors shown in Figure S2A. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(32, 208) = 2.198; Dunnett’s multiple comparison. N = 6–8 mice. Also see Figures S2A and S2B.(D and E) Western blots of PICK1 (D) or PSD-95 (E) following pull-down with biotinylated peptides from brain stem, lumbar spinal cord, and L1-L5 DRGs. PICK1-targeting GCN4-HWLKV and GCN4(7P14P)-HWLKV (7P14P-HWLKV in figure). Negative GCN4-GS (GCN4-GGGGS in figure). PSD-95-targeting GCN4-IETDV. All conditions for each tissue type were run on the same blot. Irrelevant lanes were removed digitally as indicated by spaces between lanes. N = 2 mice (pooled).(F) von Frey-determined paw withdrawal thresholds in CFA male WT mice. AAV8-Di-C5, AAV8-SSO10a-C5, AAV8-Mdv1-C5, AAV8-Atg16-C5, and AAV8-Di-GS were compared. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(12, 57) = 2.449; Tukey’s multiple comparison. N = 4–6 mice.(G) Illustration of AAV-encoded recombinant peptides, including the non-binding GCN4-GS and three PICK1-binding variants (SSO10a, Atg16, Mdv1) having dimerization domains of different lengths and orientations.All data in (C) and (F) are shown as mean ± SEM. Abbreviations are as BL, baseline; CFA, complete Freund’s adjuvant; DAPI, 4',6-diamidino-2-phenylindole; DRG, dorsal root ganglion; GCN4, general control nondepressible 4; i.t., intrathecal; ITR, inverted terminal repeats; ns, non-significant; NTX, naltrexone; PICK1, protein interacting with C-kinase 1.
To further investigate target specificity, we performed another pull-down experiment using biotinylated GCN4 peptides, either containing the PICK1-targeting HWLKV motif or the IETDV motif previously used to target PSD-95.^14^ Western blots of mouse brain stem, spinal cord, and DRGs tissues demonstrated robust target engagement and clear selectivity toward PICK1 and PSD-95, respectively (Figures 1D and 1E). Finally, we asked whether the pain-relieving effects of the PICK1 inhibitors were dependent on both the GCN4 dimerization domain and the HWLKV motif. First, we exchanged GCN4 with alternative dimerization Atg16-C5, Mdv1-C5, and SSO10a-C5, which are longer, antiparallel, or both (Figures 1F and 1G). Atg16-C5 and Mdv1-C5 were too long for synthetic production, but biochemical experiments with SSO10a-C5 confirmed that antiparallel self-assembly by the SSO10a alpha helix conferred affinity for PICK1 binding and drove PICK1 oligomerization, similar to GCN4-C5 (Figures S3A–S3E). Second, we substituted the PDZ-binding sequence, HWLKV, with a second copy of the linker sequence, GGGGS, to yield GCN4-GS. Biochemical experiments validated intact helical and dimeric assembly of GCN4-GS, while PICK1 binding and oligomerization were abolished (Figures S3A–S3D and S3F). Mice administered i.t with AAV8-GCN4-GS, in contrast to AAV8-Di-C5, developed mechanical allodynia in the CFA model (Figure 1F), demonstrating the critical role of the PDZ-binding motif (Figures 1F and 1G). Injection of AAV8-Atg16-C5, AAV8-Mdv1-C5, or AAV8-SSO10a-C5 also relieved mechanical allodynia to varying extents, with AAV8-Atg16-C5 not being significantly different from AAV8-GCN4-GS (Figures 1F and 1G). This demonstrates that the observed effects were not dependent on the GCN4 domain itself. In contrast, the varying effect of substituting GCN4 might reflect differences in dimerization strength, steric hindrance, expression level, etc. To this end, we concluded that different high-affinity, PICK1-targeting, AAV-delivered recombinant dimeric peptides can prevent mechanical allodynia in the CFA model, even during NTX-induced hyperalgesic reinstatement, suggesting a non-opioid analgesic effect on inflammatory pain. We therefore decided to further characterize the expression and efficacy profiles of AAV8-Di-C5 and AAV8-Di(7P14P)-C5.
To determine which neurons contributed to the pain-relieving effect, we used the transgenic mouse line Hoxb8-Cre, which expresses Cre recombinase in neurons caudal to the cervical spinal cord level 3, including the DRGs.^25^ By exploiting a Cre-ON (AAV8-DIO[EGFP-P2A-Di-C5]ON) and Cre-OFF (AAV8-DIO[EGFP-P2A-Di-C5]OFF) approach (Figures S4A and S4B) in male and female CFA mice, the OFF variant in WT mice (i.e., peptide expression) fully prevented mechanical allodynia (Figures S4C and S4D), in agreement with our previous result (Figure 1C), while the ON variant in WT mice (i.e., no peptide expression) was without effect (Figures S4C and S4D). In Hoxb8-Cre mice, however, the ON variant (i.e., peptide expression caudal to spinal level C3) fully abolished mechanical allodynia. The OFF variant (i.e., peptide expression rostral to spinal level C3) displayed a non-significant trend toward a treatment effect (Figures S4C and S4D). From this, we concluded that transgene expression caudal to cervical level 3 is adequate to mitigate mechanical allodynia.
To obtain insight into the AAV transduction profile following i.t. delivery, we next used the Cre-reporter mouse line Ai14, which expresses tdTomato following Cre-mediated recombination (Figure S5A). For this purpose, we employed an AAV encoding iCre fused to EGFP (AAV8-iCre:EGFP-P2A-Di-C5). Tissues were examined 40–44 days after AAV delivery, and both cryostat-prepared sections and cleared whole-mount spinal cords were imaged. For the latter, we extended the iDISCO tissue-clearing technique with decalcification, rendering both soft and hard tissues transparent and suitable for visualization using light-sheet microscopy. Clear tdTomato fluorescence was detected within neuronal cell bodies of sacral, lumbar, and thoracic DRGs (Figures 2A–2C and S5B) and their ascending fibers confined to the spinal cord dorsal column (Figures 2D, S5B, and S5C), consistent with robust transduction of mechanosensitive Aβ fibers. This labeling pattern was accompanied by fibers extending into the dorsal horn laminae of the spinal cord (Figure 2D), although the nociceptive origin of these fibers remains unclear. TdTomato fluorescence along the spinal cord dorsal column terminated at the cuneate and gracile nuclei without extending rostrally beyond the pons (Figure 2E), though a few labeled neurons were seen in the brain, e.g., in the cerebellum and olfactory structures (Figure S5C). Little, if any, cell body expression was observed within the spinal cord at the lumbar level (Figures 2A, 2B, and 2D). This expression pattern suggests that AAV8, after i.t. injection at the L5–6 intervertebral level, predominantly transduces DRGs, with fibers extending into both local spinal circuits and supraspinal sites.Figure 2Peptide expression following intrathecal AAV8 injection is primarily within dorsal root ganglion neurons(A and B) Whole-mount imaging following i.t. injection of AAV8-iCre:EGFP-P2A-Di-C5 into Ai14 mice. A) Longitudinal view of lumbosacral spinal cord labeled by tdTomato. Top image (autofluorescence signal) outlines the spinal cord and pairs of DRGs (dotted lines). Bottom image shows tdTomato-labeled neurons within DRGs. Scale bars, 500 μm. B) Transverse view of the spinal cord displayed in A). DRGs on both sides are labeled, with no neuronal somas detected within the spinal cord. Non-neuronal tissue surrounding the vertebral column is also labeled by tdTomato. ∗ = NR. Scale bars, 300 μm. A and B) N = 8 mice. Also see Figure S5B.(C–E) Cryostat sections following i.t. injection of AAV8-iCre:EGFP-P2A-Di-C5 in Ai14 mice, showing C) tdTomato-labeled neurons within lumbar (L1–L5) and sacral (S1–S2) DRGs, D) tdTomato-labeled DRG neuron ascending fibers within the spinal cord dorsal column (Insert I), with no visible neuronal somas (insert II, gain adjusted), and E) tdTomato-positive fibers in brainstem (Insert III) terminating within cuneate and gracile nuclei (Insert IV). The bottom three panels are gain-adjusted. C–E) N = 9 mice. Also see Figure S5C. Scale bars, 100 μm (C) and 500 μm (D and E).(F) RT-qPCR-evaluated peptide mRNA expression levels in nervous and liver tissue from naive male WT mice after receiving i.t. injection of AAV8-Di-C5 or vehicle (DPBS). L3-L5 DRGs (t(6) = 13.20, p < 0.0001), sciatic nerve (t(6) = 2.549, p = 0.0435), lumbar spinal cord (t(6) = 3.084, p = 0.0216), cervical spinal cord (t(6) = 1.787, p = 0.1241), dorsal column (t(6) = 0.01795, p = 0.9863), pons (t(6) = 0.3108, p = 0.7665), cerebellum (t(6) = 1.552, p = 0.1716), and liver (t(6) = 43.62, p < 0.0001). ∗p < 0.05, ∗∗∗∗p < 0.0001. Unpaired t test for all comparisons. N = 4 mice. Data are shown as mean ± SEM. Also see Table S1.(G and H) Uniform manifold approximation and projections (UMAPs) visualizing G) neuronal and H) non-neuronal cell types obtained by single nucleus RNA sequencing (snRNA-seq) from L3–L5 DRGs. N = 2 mice, pooled. Also see Table S2.Abbreviations are as AP, area postrema; AQ, cerebral aqueduct; CLTMR, C-low threshold mechanoreceptors; CU, cuneate nucleus; cuf, cuneate fascicle; DPBS, Dulbecco's Phosphate Buffered Saline; DRG, dorsal root ganglion; DRG L1–5, lumbar 1–5 DRG; DRG S1–2, sacral 1–2 DRG; ECU, external cuneate nucleus; endo, endothelial cells; fibro (endon.), fibroblast endoneurium; fibro, fibroblasts; GR, gracile nucleus; NF, Nefh^+^ A-fiber low threshold mechanoreceptor; NP, non-peptidergic neurons; NR, nerve root (also indicated by asterisk [∗] in panel B); Pep, peptidergic neurons; SC, spinal cord; Schwann M, myelinating Schwann cells; Schwann N, non-myelinating Schwann cells; SGC, satellite glial cells; VSMC, vascular smooth muscle cells.
Male and female Ai14 mice did not develop mechanical allodynia in the CFA model following i.t. injection of AAV8-Di-C5, suggesting that these tdTomato-expressing neurons are sufficient for pain relief (Figure S5D). Further examination revealed that transduced tdTomato-positive DRG neurons, which were stained for IB4, CGRP, or NF200 markers, comprised only a fraction of all DRG neurons (Figures S5E and S5F) and were significantly shifted toward larger DRG neurons (Figure S5F). While smaller DRG neurons consisted mainly of IB4^+^ or CGRP^+^ neurons, larger DRG neurons comprised mainly CGRP^+^ or NF200^+^ neurons (Figure S5G). In a separate cohort of WT mice treated with AAV8-Di-C5, we detected Di-C5 mRNA in DRGs (Figure 2F; Table S1), consistent with the tdTomato signal seen in Ai14 mice (Figures 2A–2D). In the sciatic nerve and lumbar spinal cord, peptide mRNA levels were also detectable, but not equally prominent. In all other tissues examined, except for the liver, peptide mRNA was not identified (Figure 2F; Table S1). Moreover, single-nucleus RNA transcriptomic profiling (snRNA-seq) of DRG tissue from naive mice and from mice i.t. injected with AAV8-Di-D5 revealed robust recovery of mRNA transcripts and captured all major neuronal and non-neuronal cell types (Figures 2G and 2H; Table S2), demonstrating efficient recovery and annotation of the DRG transcriptome. Transcripts encoding the Di-C5 peptide were detected at low abundance across all neuronal subtypes (Table S2), consistent with broad but modest transgene expression. Notably, although completely absent in control samples, Di-C5 transcripts were also detected at low frequency in non-neuronal cells (Table S2), suggesting non-exclusive pan-neuronal expression. We concluded that i.t. delivery of AAV8 vectors gives rise to significant DRG expression, which appears sufficient to relieve mechanical allodynia.
We next explored the efficacy of AAVs in the SNI model of neuropathic pain in three different treatment paradigms. We first evaluated the ability of the recombinant peptides to prevent mechanical allodynia in a setup in which male WT mice were i.t. injected 14 days before SNI surgery. Following SNI surgery, robust and comparable levels of allodynia were evident until at least 28 days for AAV8-tdTomato and vehicle (DPBS, Dulbecco's Phosphate-Buffered Saline) but were fully prevented by AAV8-Di-C5 and AAV8-Di(7P14P)-C5 (Figure 3A). In a similar setup, AAV8-Di(7P14P)-C5 was equally effective in female mice (Figure 3B). In the second paradigm, we evaluated the therapeutic efficacy when AAVs were delivered 5 weeks after SNI surgery, in the chronic neuropathic pain phase. Here, AAV8-Di(7P14P)-C5 significantly attenuated mechanical allodynia from week 9 after SNI until the end of the experiment at day 173 (Figure 3C). Finally, we assessed the long-term durability of the intervention by monitoring male SNI mice for an entire year. Here, AAV8-Di(7P14P)-C5 significantly reversed mechanical allodynia for 365 days (Figure 3D). From day 322 onwards, the effect magnitude declined, although significant reversal was still observed at days 350 and 365. We concluded that the recombinant peptides could both prevent and reverse mechanical allodynia in both acute and chronic stages of neuropathic pain, maintaining efficacy for at least 1 year.Figure 3Long-term suppression of neuropathic pain by AAV-delivered PICK1 inhibitors(A and B) von Frey-determined paw withdrawal thresholds in SNI WT A) male or B) female mice. i.t. injections were performed 14–22 days before SNI surgery. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. A) Two-way repeated-measures ANOVA, F(18, 126) = 5.474; Tukey’s multiple comparisons. N = 5–8 mice. B) Two-way repeated measures ANOVA, F(8, 72) = 4.106; Tukey’s multiple comparisons. N = 6–8 mice.(C) von Frey-determined paw withdrawal threshold in SNI WT male mice. i.t. injections were performed 5 weeks after SNI surgery. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(9, 93) = 5.474; Holm-Šidák’s multiple comparisons. N = 7–8 mice.(D) von Frey-determined paw withdrawal thresholds in SNI male WT mice for 1 year. i.t. injections were performed 2 days after SNI surgery. ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(20, 360) = 8.305; Holm-Šidák’s multiple comparisons. N = 10 mice.(E–H) SNI or sham male WT mice underwent a battery of tests to assess different pain modalities on the ipsilateral hind paw. i.t. injections were performed 2 days after SNI/sham surgery. E) Brush. Paw withdrawal reaction scores following brushing. ∗∗∗p < 0.001. Two-way repeated-measures ANOVA, F(6, 38) = 9.971; Tukey’s multiple comparisons.(F) Cold plate. Latency for first paw withdrawal on a 2°C cold plate. ∗∗p < 0.01. Two-way repeated-measures ANOVA, F(9, 57) = 13.39; Tukey’s multiple comparisons.(G) Acetone test. Total time exhibiting nocifensive behavior during 60 s following acetone application. ∗∗p < 0.01. One-way repeated-measures ANOVA, F(3, 19) = 9.101; Tukey’s multiple comparisons.(H) Spontaneous paw lifts. Number of spontaneous paw lifts recorded over 15 min. ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(3, 19) = 5.013; Tukey’s multiple comparisons. E–H) N = 5–6 mice. The same mice were used. Also see Figures 4A, S6A, and S6B + E-Q.(I) von Frey-determined paw withdrawal thresholds in SNI Advillin-Cre male mice. i.t. AAV injection was performed 5 days after SNI surgery. ∗∗∗∗p < 0.0001. Mixed-model ANOVA, F(6, 53) = 7.252 (missing BL value on the first BL measurement on a single mouse); Tukey’s multiple comparisons. N = 5–8 mice.(J) von Frey-determined paw withdrawal threshold in SNI SNS-Cre female and male mice. i.t. AAV injection was performed 2 days after SNI surgery. ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(6, 39) = 10.06. N = 4–6 mice. Also see Figures S6C and S6D.Abbreviations are as BL, baseline; i.t., intrathecal; SNI, spared-nerve injury. All data in (A–J) are shown as mean ± SEM.
Neuropathic pain conditions typically manifest across several pain modalities. Therefore, we tested WT male mice receiving AAV8-Di(7P14P)-C5 or AAV8-tdTomato across other sensory perceptions, including both SNI and sham groups (Figure S6A). In accordance with previous results, injection of AAV8-Di(7P14P)-C5 completely attenuated mechanical allodynia in SNI mice, while sham mice did not show any change in their paw withdrawal threshold (Figure S6B). When using dynamic (brush) instead of static (von Frey hairs) touch, averaged response scores were increased after SNI surgery (compared to sham) and were significantly lower for AAV8-Di(7P14P)-C5 compared to AAV8-tdTomato at day 40, whereas responses in both sham groups were unchanged (Figure 3E). For cold allodynia tested on a 2°C cold plate, SNI mice displayed a much lower response latency 6 days following injury. Treatment with AAV8-Di(7P14P)-C5 caused a significant recovery in response latency on day 35, but not day 41, compared to mice injected with AAV8-tdTomato, albeit with considerable variations, which was not observed for the other groups (Figure 3F). Similarly, local paw application of acetone to probe cold allodynia caused aggravated nocifensive behaviors in both SNI groups, but with trending, yet non-significant, treatment differences seen for SNI mice injected with AAV8-Di(7P14P)-C5 and AAV8-tdTomato (Figure 3G). As an index of spontaneous pain, we quantified the number of spontaneous paw lifts (SPL)^26^ for 15 min on days 46 and 70. While sham mice, regardless of treatment, had almost no SPL, this behavior was prominent in the SNI groups but significantly reduced in mice receiving AAV8-Di(7P14P)-C5 compared to AAV8-tdTomato (Figure 3H). This series of behavioral assays supports that AAV-encoded PICK1 inhibitors can attenuate several sensory pain modalities caused by chronic nerve injury.
To conclusively determine whether DRGs can mediate the observed effects, we again applied a Cre-ON strategy, this time in Advillin-Cre mice, which restricts Cre recombinase expression almost exclusively to peripheral sensory neurons.^27^ As a positive control, AAV8-Di-C5 fully relieved mechanical allodynia in the SNI model, and the AAV8-DIO[EGFP-P2A-Di-C5]ON vector induced an equivalent effect, thereby confirming that expression in DRGs is sufficient (Figure 3I). In SNS-Cre mice, however, the Cre-dependent vector did not affect static (von Frey) (Figure 3J), dynamic (brush) touch, or cold allodynia (Figures S6C and S6D). This strain expresses Cre in Nav1.8-positive sensory neurons, which primarily comprise small-diameter C-fibers.^28^ The lack of efficacy in SNS-Cre mice indicates that nociceptors alone are not sufficient for the therapeutic effect and that the contribution of non-nociceptive DRG populations is critical, reinforcing the conclusion that the site of action involves DRGs broadly rather than nociceptors exclusively.
Along with the battery of sensory tests (Figures 3E–3H), the same SNI and sham groups of mice were automatically monitored for changes in voluntary behaviors, including rearing, grooming, climbing, immobility, and locomotion in single-housed cages. No differences were found between groups (Figures 4A and S6E–S6K). Next, we asked whether the therapeutic intervention could reverse any SNI-associated motor function deficits. In CatWalk, SNI mice had significantly lower max contact area and swing speed of the injured paw than sham mice, with no detected treatment differences (Figures S6L and S6M), while average run speed and duration were unaffected by both SNI surgery and treatments (Figures S6N and S6O). In voluntary wheel running, sham mice ran significantly longer than SNI mice, again with no treatment difference across groups (Figure S6P). These data support that gross motor deficits caused by peripheral nerve transection cannot be attenuated by recombinant PICK1 treatment, nor does it alter normal gross motor functions. In addition, body weight was monitored periodically over 91 days, with no clear changes seen over time across treatments (Figure S6Q). Other phenotypical changes (nest building, social behavior, muscle weakness, etc.) were also not observed by the experimenters throughout.Figure 4AAV treatment does not alter motor performance or trigger immune activation(A) Automatic cage monitoring of general behavior for 22 h in sham or SNI male WT mice following i.t. injection of AAVs using the Laboras system. Percentages of time spent on each classified behavior are shown. Individual scores and statistics are provided in Figures S7C–7I. N = 5–6 mice. The same mice were used in Figures 3E–3H, S6A, and S6B + E-Q.(B–F) Naive male WT mice receiving i.t. injection of AAVs or vehicle (DPBS) underwent B–D) a 2-h open field test and E–F) RotaRod tests.(B) Total distance traveled. One-way ANOVA, F(2, 13) = 0.979; Tukey’s multiple comparisons.(C) Time spent in the center zone. One-way ANOVA, F(2, 13) = 0.583; Tukey’s multiple comparisons.(D) Number of center entries. One-way ANOVA, F(2, 13) = 2.459; Tukey’s multiple comparisons.(E) Accelerating RotaRod, time to fall (left) and best speed obtained (right). Time to One-way ANOVA, F(2, 13) = 0.671; Tukey’s multiple comparisons. Best One-way ANOVA, F(2, 13) = 0.620; Tukey’s multiple comparisons. Average scores of 3 trials per mouse. F) Fixed-speed RotaRod. Fall is shown as a survival probability. AAV8-Di-C5 vs. AAV8-tdTomato: Log-rank (Mantel-Cox) test, Chi-square = 0.089, p = 0.765; AAV8-Di-C5 vs. Log rank (Mantel-Cox) test, Chi-square = 0.916, p = 0.339. Insert: best speed obtained. One-way ANOVA, F(2, 13) = 0.648; Tukey’s multiple comparisons. B–F) N = 5–6 mice; the same mice were used. Also see Figures S7A–S7C.(G–J) Flow cytometry. Number of immune cells per 0.5 cm ipsilateral or contralateral sciatic nerve 28–29 days after SNI surgery in female mice i.t. injected 21 days before SNI surgery (same mice as in Figure 3B).(G) Immune cells, defined as CD45^+^ cells. ∗∗p < 0.01. One-way ANOVA, F(5, 16) = 8.304; Dunnett’s multiple comparisons. H) Macrophages, defined as CD45^+^ and CD11b^+^ cells. ∗p < 0.05, ∗∗p < 0.01. One-way ANOVA, F(5, 16) = 7.428; Dunnett’s multiple comparisons.(I) Neutrophils, defined as CD45^+^ and Ly6G^+^ cells. One-way ANOVA, F(5, 16) = 2.210; Dunnett’s multiple comparisons.(J) T cells, defined as CD45^+^ and TCR^+^ cells. ∗p < 0.05, ∗∗p < 0.01. One-way ANOVA, F(5, 16) = 6.690; Dunnett’s multiple comparisons. G–J) N = 3–4 mice. Also see Figures S7E–S7L.(K–M) H&E staining of livers from naive female WT mice 3 weeks after i.t. injection. Infiltrates appear as small dense nuclei, shown by arrows. Scale bars, 100 μm.(N) Ratio of immune infiltrate area to entire liver tissue area. One-way ANOVA, F(2, 6) = 2.854. N = 3 mice; 5–6 liver sections per mouse.(O) Schematic of the lower spinal cord showing the injection site at the end of the spinal cord, and the L3 DRGs with their projections entering the L3 spinal cord.(P) Whole-mount spinal cord imaging of CD45^+^ immunostaining in female mice 28 days after i.t. injection with AAV8-Di-C5 or vehicle. N = 3 mice. Scale bars, AAV8-Di-C5, 300 μm (upper), 500 μm (middle), 700 μm (lower); vehicle, 200 μm (upper), 500 μm (middle), 500 μm (lower).All data in (B–J) and (N) are shown as mean ± SEM. Abbreviations are as DPBS, Dulbecco's Phosphate Buffered Saline; DRG, dorsal root ganglion; SC, spinal cord; SNI, spared-nerve injury.
To further probe potential side effects, naive male mice i.t. injected with AAV8-Di-C5, AAV8-tdTomato, or vehicle (DPBS) (Figure S7A) were tested in a 2-h open field test 35 days after injection. Total distance traveled, time spent in the center, and number of center entries were similar between groups (Figures 4B–4D and S7B), and using either accelerating or fixed speed in the RotaRod, latency to fall and best speed obtained were unaltered across groups (Figures 4E and 4F). Lastly, these mice were subjected to CFA, confirming that AAV8-Di-C5, but not AAV8-tdTomato or vehicle, fully prevented mechanical allodynia (Figure S7C). Together, these results argue that recombinant PICK1-targeting peptides selectively affect maladaptive pain signaling circuits and processing without altering gross motor or proprioceptive function in healthy, intact circuits.
Since AAV impurities can potentially elicit an immune response, all AAV preparations were qualitatively assessed before use. No protein bands, except capsid proteins VP1, VP2, and VP3, were detected (Figure S7D). Nonetheless, to evaluate potential local immune responses after AAV injection, immune cell populations were quantified by flow cytometry in 0.5 cm segments of ipsilateral and contralateral sciatic nerves from female SNI mice treated with AAV8-Di-C5, AAV8-tdTomato, or vehicle (Figures 4G–4J and S7E–S7L). This was done 49–50 days after the lesion. Immune cells were defined as CD45^+^ cells, macrophages as CD45^+^CD11b^+^, neutrophils as CD45^+^Ly6G^+^, and T cells as CD45^+^TCR^+^. Consistent with the injury, the number of immune cells was overall increased on the lesioned side relative to the contralateral nerve; however, the AAV treatments did not elicit any distinct changes compared to the vehicle control (Figures 4G–4J and S7E–S7L).
As peptide mRNA levels were prominent in the liver (Figure 2F; Table S1), we also evaluated potential immune cell infiltration in this tissue. Hematoxylin and eosin (H&E) staining was used to quantify cells with small, dark nuclei and revealed no difference among liver samples collected 28 days after injection from naive WT mice treated with AAV8-Di-C5, AAV8-tdTomato, or vehicle (Figures 4K–4N). Finally, we used whole-tissue clearing combined with CD45 immunostaining to visualize putative infiltrating immune cells in the same mice analyzed above, examining only AAV8-Di-C5 and vehicle. The entire spinal column, including vertebrae, DRGs, and surrounding connective tissues, was imaged from the i.t. injection site at the L5–L6 intervertebral disc up to the L3 spinal level, corresponding to the point where the L3 DRGs project into the spinal cord (Figures 4O and 4P). Across all examined tissues, no clear areas of infiltrating cells or differences across conditions were noticed. In summary, we concluded that the viral vectors did not elicit any detectable immune or histological changes across all examined tissues and conditions compared to vehicle.
AAV capsids such as AAVPHP.S and AAVPHP.eB can selectively transduce either the peripheral nervous system (PNS) or the central nervous system (CNS), respectively.^29^ Motivated by our previous observations on i.t. administration of AAV8 vectors, we wanted to test whether systemic AAV application would be equally efficacious. Mechanical allodynia was confirmed before AAVPHP.S or AAVPHP.eB vectors were intravenously (i.v.) administered to WT male mice 37 days after SNI surgery. Control groups receiving either vehicle (DPBS) or AAVPHP.S-iCre:EGFP displayed mechanical allodynia after SNI surgery and did not recover, while mice receiving AAVPHP.S-iCre:EGFP-P2A-Di-C5 or AAVPHP.eB-iCre:EGFP-P2A-Di-C5 recovered completely for 125 days after SNI surgery (Figure 5A), with no observable adverse effects. Transduction profiles of such AAVs were next examined in Ai14 mice, where the AAVPHP.S vector gave rise to clear tdTomato labeling of dorsal column fibers in the spinal cord (Figure 5B), DRG neuronal cell bodies (Figure 5C), and brainstem (Figure 5D), almost identical to the patterns seen previously with the i.t. injected AAV8 vector (Figures 2C–2E and S5C). Some scattered tdTomato-positive neurons were also observed within the CNS without region specificity (Figure 5D). In contrast, the AAVPHP.eB vector gave rise to robust tdTomato expression in neurons of the spinal cord (Figure 5E), brainstem, and brain (Figure 5G), while DRG neurons were not labeled (Figure 5F). EGFP expression was largely confined to spinal cord cell bodies and broadly distributed in the brain (Figure 5G). These results indicate that systemic delivery of recombinant PNS- or CNS-targeting peptides can effectively alleviate chronic neuropathic pain.Figure 5Full reversal of mechanical allodynia by systemic AAV-delivered PICK1 inhibitors(A) von Frey-determined paw withdrawal thresholds in SNI male WT mice. i.v. injections were performed 37 days after SNI surgery. ∗p < 0.05, ∗∗p < 0.01. Two-way repeated-measures ANOVA, F(5, 100) = 3.727; Dunnett’s multiple comparisons (AAVPHP.S-iCre:EGFP serving as control). N = 6 mice. Data are shown as mean ± SEM.(B–D) TdTomato signaling from Ai14 mice receiving the PNS-transducing AAVPHP.S-iCre:EGFP-P2A-Di-C5 vector, showing expression in B) dorsal column fibers, C) neuronal somas in DRGs, and D) ascending fibers terminating in the brainstem. EGFP expression is only visible in DRGs. N = 4 mice. Scale bars, 100 μm (C) and 500 μm (B and D).(E–G) TdTomato signaling from Ai14 mice receiving the CNS-transducing AAVPHP.eB-iCre:EGFP-P2A-Di-C5 vector, showing expression throughout the spinal cord and brain (E, G) and not detected in DRGs (F). EGFP expression is visible as green dots in the spinal cord and is distributed across the brain. N = 8 mice. Abbreviations are as CNS, central nervous system; DRG, dorsal root ganglion; i.v., intravenous; PNS, peripheral nervous system; SC, spinal cord; SNI, spared-nerve injury. Scale bars, 100 μm (F) and 500 μm (E and G).
To elucidate how AAV-encoded PICK1 inhibitors mediate pain relief in DRGs, we next performed mass spectrometry (MS) on L3–L5 DRGs from both lesioned (ipsilateral) and intact (contralateral) sides of SNI mice treated with either AAV8-Di-C5 or AAV8-tdTomato. DRGs were collected 84–85 days post-SNI surgery, when full pain relief was observed (Figure 6A). An average of 7,587 proteins were identified (Figures S8A–S8F), with more proteins upregulated than downregulated by injury in both groups (Figures 6B and 6C). Notably, several injury- and inflammation-associated proteins, such as Sprr1a, Lyz2, Lgals3, and GAP43, were significantly upregulated in ipsilateral versus contralateral DRGs in both groups (Figures 6B, 6C, and S8G), confirming lesion detection. Biological process enrichment analysis revealed an overrepresentation of upregulated proteins related to lipid and lipid-transport processes, and of downregulated proteins related to muscle contraction (Figures 6D and S8H–S8J). Notably, of the 97 proteins differentially regulated between ipsilateral and contralateral DRGs for AAV8-Di-C5, but not AAV8-tdTomato, only GRIA4 (glutamate ionotropic receptor AMPA subunit 4) has been previously linked to PICK1 PDZ interactions (Table S3).Figure 6Proteomic and phosphoproteomic profiling of DRGs treated with AAV8-Di-C5(A) von Frey-determined paw withdrawal threshold in SNI male WT mice. i.t. injections were performed 11 days after SNI surgery. ∗∗p < 0.01, ∗∗∗∗p < 0.0001. Two-way repeated-measures ANOVA, F(3, 18) = 12.97; Holm-Šidák’s multiple comparisons. N = 4 mice. Data are shown as mean ± SEM.(B–G) Volcano plots of (B–D) protein or (E–G) phosphorylation fold changes (log2FC) in L3–L5 DRGs. Protein-level fold changes between ipsi- and contralateral DRGs of (B) AAV8-TdTomato and (C) AAV8-Di-C5 treated mice, and (D) between ipsilateral DRGs of AAV8-TdTomato and AAV8-Di-C5 treated mice. Phosphorylation fold changes between ipsi- and contralateral DRGs of (E) AAV8-tdTomato and (F) AAV8-Di-C5 treated mice, and (G) between ipsilateral DRGs of AAV8-TdTomato and AAV8-Di-C5 treated mice, with enrichment of regulated kinases shown below. (B–D) Also see Figure S8 and Table S3. (E–G) Also see Figures S9. MS performed on mice from (A). N = 4 mice.(H) Western blot of PICK1 following pull-down with biotinylated peptides from human spinal cord and DRG tissue of both PICK1-targeting GCN4-HWLKV and GCN4(7P14P)-HWLKV (7P14P-HWLKV in figure). Negative GCN4-GS (GCN4-GGGGS in figure). PSD-95-targeting GCN4-IETDV. N = 3 technical replicates.Abbreviations are as contra, contralateral; DRG, dorsal root ganglion; ipsi, ipsilateral; i.t., intrathecal; SNI, spared-nerve injury.
Next, phosphorylation mapping and motif enrichment analysis were performed on an average of 8,895 annotated protein fractions across the four DRG groups (Figures S9A–S9F), showing differences both within and between groups (Figures 6E–6G). Intriguingly, phosphorylation of sites phosphorylated by PRKCA (protein kinase C alpha, PKCα), known to interact with the PICK1 PDZ domain, was downregulated in ipsilateral versus contralateral DRGs for AAV8-tdTomato (Figure 6E), a change not detected for AAV8-Di-C5 (Figure 6F). Direct comparison of ipsilateral DRGs between treatments identified downregulation of phosphorylation at SGK1 (serum- and glucocorticoid-inducible kinase 1) sites for AAV8-Di-C5 (Figure 6G). Taken together, these findings suggest that PICK1 inhibition alters the DRG phosphoproteome, likely by engaging multiple pathways rather than a single dominant mechanism, contributing to the behavioral phenotype.
Finally, as the concluding experiment, we performed a pull-down on tissue from deceased human donors of both sexes using the same synthetic biotinylated GCN4 peptides as for mouse tissue. This experiment confirmed PICK1 expression and engagement by the PICK1-targeting peptides in human DRG and spinal cord tissue (Figure 6H), underscoring the clinical relevance of PICK1 as a target for chronic pain and the translational potential of AAV-based PICK1 therapeutics.
AAV-based gene therapy is a new therapeutic frontier in medicine and has the potential to address significant unmet clinical needs,^1^^,^^2^^,^^8^ including pain conditions.^3^^,^^30^^,^^31^^,^^32^ In this study, we developed an AAV gene therapy approach utilizing recombinant dimeric peptides designed to target PICK1 with high affinity and selectivity. The analgesic efficacy of this approach was validated across three academic institutions, demonstrating robust and reproducible pain relief. We show that this intervention effectively reverses both inflammatory and neuropathic-induced hypersensitivity, is efficacious in both sexes, alleviates hypersensitivity across several sensory modalities affected by chronic pain, and is equally effective when administered before or after injury.
PICK1 is an interesting scaffold protein due to its association with proteins linked to pain mechanisms, including several glutamate receptors (GRIK1, GRIK2, GRIK4, GRM7), AMPA receptor subunits (GRIA2, GRIA3, and GRIA4), acid-sensing ion channels (ASIC1 and ASIC2), aquaporin (AQP1), and PKCα.^33^ Furthermore, PICK1 is highly conserved, and its mRNA is expressed in neurons along the somatosensory nociceptive pathway of both rodents and humans.^33^ While prior research had established the presence of PICK1 protein in mouse DRG neurons,^34^ this study extends this knowledge by demonstrating its expression in human DRG and spinal cord tissue of both sexes, as well as target engagement with recombinant PICK1-targeting peptides. Studies using inhibitory peptides, small interfering RNA (siRNA), and knockout mice have also shown that loss of PICK1 function reduces hyperexcitable pain conditions,^34^^,^^35^^,^^36^^,^^37^ consistent with the present observations.
The DRG proteomics analysis revealed no differential enrichment of known PICK1 PDZ-domain interaction partners, including PICK1-associated proteins linked to pain, except for GRIA4. Although GRIA4 is established in spinal pain processing, its role in DRG neurons is less clear, and its contribution to the observed analgesic effect remains to be determined in future studies. The phosphoproteomic motif enrichment analysis identified regulation of PKCα activity in response to SNI surgery and normalization in the AAV8-Di-C5 group. This kinase has been shown to interact with PICK1 in mouse DRGs, where its interaction enhances PKCα-mediated phosphorylation and promotes hypersensitivity in the CFA mouse model.^35^ SGK1 activity was further differentially regulated between the AAV8-Di-C5 and AAV8-tdTomato groups. This kinase is inactive in its primary form and requires phosphorylation by PDK1 (3-phosphoinositide-dependent kinase 1) and mTORC2 (mechanistic target of rapamycin complex 2) for its activation. Its reduced activity in ipsilateral DRGs treated with AAV8-Di-C5, compared to AAV8-tdTomato, is interesting, as the active kinase has been reported to be implicated in both the induction and maintenance of inflammatory and neuropathic pain in rats, acting in spinal cord and DRG neurons to modulate excitability.^38^^,^^39^ Although PDK1 and mTORC2 are not directly linked to established PICK1 signaling pathways, it can be speculated that indirect crosstalk between the PICK1 and SGK1 signaling axis may occur. It is important to note that the involvement of PKCα and SGK1 was inferred from motif enrichment analysis and represents an association with PICK1 inhibition. Whether these changes reflect long-term adaptations to viral expression of AAV8-Di-C5 cannot be determined from the current data.
The i.t. and i.v. delivery routes in Ai14 mice, using AAV8 and AAVPHP.S, respectively, demonstrated that these methods primarily transduced DRG neurons and their extensions, while neuronal cell bodies in the spinal cord and supraspinal regions were not targeted. This was further supported by the peptide mRNA profiles obtained from WT mice following i.t. injection of AAV8-Di-C5. In accordance, a study with i.t. injection of AAV8-GFP in rats resulted in labeling of the dorsal column fibers and DRG neurons, but with no neuronal cell bodies found outside the DRGs.^40^ This indicates that the therapeutic effect, as seen for AAV8-Di-C5 and AAV8-Di(7P14P)-C5 vectors, may involve ascending dorsal column-medial lemniscus (DCML) fibers. In agreement, mechanical allodynia was resolved in Advillin-Cre but not in SNS-Cre mice. Still, the latter result may reflect incomplete transduction of SNS-Cre-positive neurons or incomplete coverage of all nociceptive subtypes, as SNS-Cre primarily targets Nav1.8-positive neurons, rather than a definitive lack of contribution from nociceptors.
The DCML pathway consists of Aβ low-threshold mechanoreceptor fibers (Aβ-LTMRs) that normally transmit non-noxious mechanosensory information. However, nerve injury may cause a phenotypic switch in which Aβ fibers acquire nociceptive capacity, which is usually mediated by small unmyelinated C fibers or thinly myelinated Aδ fibers.^41^ Notably, tactile allodynia induced by spinal nerve L5/L6 ligation in rats was abolished following complete dorsal column lesion without causing motor deficits,^42^ and optogenetic activation of Aβ fibers following peripheral nerve injury drove tactile allodynia.^43^ Using refined intersectional transgenic strategies combined with targeted ablation or optogenetic manipulations, recent work also shows that mechanical activation of Aβ-LTMRs can drive nociception in both inflammatory and neuropathic conditions, whereas thermal nociception remains unaffected.^44^ Therefore, specific pain modalities processed by distinct sensory neurons appear to be altered under pathological conditions. While these studies match our observation of full reversal of mechanical allodynia, they do not account for the modest effects on cold and heat-induced hyperalgesia reported here. This likely reflects contributions from mixed DRG subtypes, with PICK1 mainly modulating Aβ-LTMRs. Still, if PICK1 inhibition translates to humans, AAV-based PICK1 gene therapy could provide a mechanism-based treatment for peripheral neuropathic pain that is both distinct from existing therapies and potentially safer.
Before advancing such AAV treatments for human application, it will be essential to thoroughly address potential translational barriers. While AAV therapeutics are generally well tolerated in humans, the dimeric scaffold, originating from a yeast-derived sequence, may carry a risk of eliciting an immunogenic response not detectable in mice. This could be mitigated by substituting the dimerization domain with a human equivalent. Our findings demonstrate that the dimerization domain can be substituted without compromising efficacy, indicating that alternative, potentially safer domains can be identified. Moreover, to further minimize the spread of AAV, a transforaminal epidural injection, used in clinical practice to deliver steroids for lumbar radicular pain,^45^ could be advantageous for targeted expression towards those DRGs involved in the process and maintenance of neuropathic pain. Another consideration involves the use of the hSyn promoter, selected to achieve pan-neuronal expression. Surprisingly, whole-mount imaging of Ai14 reporter mice revealed distinct tdTomato signals from what appear to be connective and muscle tissue surrounding the vertebrae, and within DRGs, the peptide transcript was detectable in non-neuronal cells. These findings underscore the necessity for thorough evaluation of AAV promoters beyond the intended target, even those considered cell-specific. When packaged with the AAVPHP.eB capsid, peptide expression was restricted to the CNS, spanning the entire spinal cord and supraspinal areas. Here, equally potent treatment efficacy for mechanical allodynia was observed compared with expression in peripheral neurons, suggesting that the approach may also have clinical relevance for central neuropathic pain. The therapeutic utility may even extend beyond AAV-based monotherapy, potentially achieving greater efficacy when combined with conventional pain medications, including peripheral nerve blocks.
In this study, we demonstrate durable efficacy lasting at least 1 year in mice, indicating that the underlying mechanism is largely resistant to desensitization. However, the magnitude of the effect was reduced at the latest time points. The cause of this decline is unclear and could reflect changes in peptide expression or age-related neuronal loss. Long-term expression data would be needed to clarify this. Nonetheless, our results show that AAV-encoded dimeric, PICK1-targeting peptide inhibitors can achieve long-term suppression of chronic pain and establish PICK1 as a clinically relevant target, positioning AAV-based interventions as a potential one-shot, lifelong treatment for chronic neuropathic pain.
While our study highlights the therapeutic potential of PICK1-based gene therapy, important limitations merit attention. It remains unclear whether the reduced neuronal excitability previously reported following TPD5 treatment in the mouse SNI model^15^ is recapitulated following AAV treatment. Addressing this question will require dedicated electrophysiology or calcium imaging studies. In addition, clinical translation poses further challenges beyond those discussed above, including the high cost of clinical-graded AAV, stringent regulatory requirements, and the need to assess neutralizing antibodies. Collectively, these limitations outline a roadmap for future mechanistic and translational investigations to advance this AAV-based strategy towards clinical application.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Andreas T. Sørensen (andreass@sund.ku.dk).
Requests for plasmids and AAVs produced specifically for this study should be addressed to and will be fulfilled by the lead contact.
Raw data from snRNA-seq have been deposited in the SRA database SRA: SRP645687 (SRA: SRX31117391, SRX31117392), whereas processed snRNA-seq data are accessible at the Zenodo repository (Zenodo: 17629531). MS proteomic (PRIDE: PXD074049) and phosphoproteomic (PRIDE: PXD073925) datasets are accessible at the PRIDE repository. The paper does not report original code. Any additional information required to reanalyze the data reported in this work is available from the lead contact upon request.
We thank Lone Rosenquist and Nabeela Khadim for excellent technical assistance and Ole Kiehn for generous access to transgenic animals. We acknowledge the Core Facility for Flow Cytometry and Single Cell Analysis for support with flow cytometry experiments; the Core Facility for Integrated Bioimaging; and the Rodent Metabolic Phenotyping Platform Histology Lab, Faculty of Health and Medical Sciences, University of Copenhagen. We also acknowledge the Danish Neuro Single Cell (NeuSiC) platform (supported by 10.13039/501100003554Lundbeckfonden, R433-2023-1585) and BRIC Core Facility for assistance with snRNA library preparation, sequencing, and data analysis. MS-based proteomics analyses were performed by the Proteomics Research Infrastructure (PRI) at the University of Copenhagen, supported by the 10.13039/501100009708Novo Nordisk Foundation (grant: NNF19SA0059305).
Funding: This work was supported by grants from the Independent Research Fund Denmark (2025-00028B) to G.N.-H., the 10.13039/501100003554Lundbeck Foundation (R322-2019-1816) to K.L.J., the 10.13039/501100003554Lundbeck Foundation (R389-2021-1596; Neuroscience Academy Denmark) to C.M.G.; the 10.13039/501100003554Lundbeck Foundation (R347-2020-2339) to A.H.L., the 10.13039/501100003554Lundbeck Foundation (R344-2020-1063) to K.L.M., 10.13039/501100009708Novo Nordisk Foundation pre-seed (DolorestBio) to K.L.M. and A.T.S., Augustinus Foundation (17-3517) to A.T.S., 10.13039/501100005747A.P. Møller Foundation (18-L-0213 and L-2021-0031) to A.T.S., and an 10.13039/100012774Innovation Fund Denmark grant (9122-00012B) to A.T.S. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
G.N.-H: conceptualization, methodology, validation, formal analysis, investigation, writing – original draft, writing – review and editing, visualization, supervision. K.L.J., M.R., M.B.K.K., L.J.-F., A.H.L., S.P.B., G.A.H.: investigation. R.C.-B., S.E.J., C.M.G., R.C.A., J.H.L., N.R.C.: investigation, formal analysis. L.S.: investigation, methodology. A.M.: methodology, formal analysis. A.T.-T., M.W., R.K.: supervision. C.B.V.: investigation, supervision. K.L.M.: conceptualization, validation, supervision, writing – original draft. A.T.S.: conceptualization, validation, supervision, writing – original draft, writing – review and editing. All authors critically evaluated and approved the final version of the manuscript.
The recombinant dimeric peptides and their utility are described in a patent pending with the European Patent Office (EPO) and the United States Patent and Trademark Office (USPTO). K.L.M. and A.T.S., co-founders of Zyneyro, hold ownership interests in the company, which has exclusive licensing rights to the patent owned by the University of Copenhagen, Denmark. G.N.-H., K.L.J., and R.C.-B. conducted the studies while employed at the University of Copenhagen and are now, together with M.B.K.K., employed at Zyneyro.
REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesMouse monoclonal anti-PICK1 [L20/8]Antibodies IncorporatedCat#75-040; RRID: AB_2164544Mouse monoclonal anti-PSD95 [K28/43]AbcamCat#ab192757; RRID: AB_2750929Rabbit monoclonal anti-GCN4 [C11L34]Absolute AntibodyCat# Ab00436–23.0; RRID: AB_3095603Rabbit anti-mCherryThermo Fisher ScientificCat#PA5-34974; RRID: AB_2552323IB4-647ThermoFisherCat#I32450; RRID: SCR_014365Goat anti-CGRPAbcamCat#ab36001; RRID: AB_725807Chicken anti-NF200MilliporeCat#Ab5539; RRID: AB_11212161RFP Antibody Pre-adsorbedRocklandCat#600-401-379; RRID: AB_2209751Alexa Fluor® 594 anti-mouse CD45 Antibody [30-F11]BioLegendCat#103144; RRID: AB_2563458Brilliant Violet 510™ anti-mouse CD11c [N418]BioLegendCat#117337; RRID: AB_2562010BD Horizon™ BUV395 Rat Anti-Mouse Ly-6G [1A8]BD BiosciencesCat#563978; RRID: AB_2716852Brilliant Violet 650™ anti-mouse Ly-6C [HK1.4]BioLegendCat#128049; RRID: AB_2800630Brilliant Violet 785™ anti-mouse F4/80 [BM8]BioLegendCat#123141; RRID: AB_2563667FITC anti-mouse CD45 [30-F11]BioLegendCat#103108; RRID: AB_312973PerCP/Cyanine5.5 anti-mouse I-A/I-E [M5/114.15.2]BioLegendCat#107625; RRID: AB_2191072PE/Cyanine7 anti-mouse/human CD11b [M1/70]BioLegendCat#101215; RRID: AB_312798PE anti-mouse TCR β chain [H57-597]BioLegendCat#109207; RRID: AB_313430APC anti-mouse CD19 [6D5]BioLegendCat#115511; RRID: AB_313646Alexa Fluor® 700 anti-mouse NK-1.1 [PK136]BioLegendCat#108729; RRID: AB_2074426Purified anti-mouse CD16/32 [93]BioLegendCat#101302; RRID: AB_312801Bacterial and virus strainspAAV2/8pAAV2/8 was a gift from James M. WilsonAddgene Cat# 112864; RRID: Addgene_112864pAdDeltaF6pAdDeltaF6 was a gift from James M. WilsonAddgene Cat# 112867; RRID: Addgene_112867AAV8-Di-C5Full AAV2.8-hSyn-HA-GCN4-GS4-HWLKVThis paperN/AAAV8-Di(7P14P)-C5Full AAV2.8-hSyn-HA-GCN4(7P14P)-GS4-HWLKVThis paperN/ApAAV-hSyn-tdTomatoProduced as AAV8-tdTomato (Full AAV2.8-hSyn-tdTomato) in-housepAAV-hSyn-tdTomato was a gift from Jonathan Ting, Allen Brain Institute for Brain Science, USAN/AAAV8-Di-SGGGGFull AAV2.8-hSyn-HA-GCN4-GS4-GGGGSThis paperN/AAAV8-SSO10a-C5Full AAV2.8-hSyn-HA-SSO10a-GS4-HWLKVThis paperN/AAAV8-Atg16-C5Full AAV2.8-hSyn-HA-Atg16-GS4-HWLKVThis paperN/AAAV8-MDV1-C5Full AAV2.8-hSyn-HA-MDV1-GS4-HWLKVThis paperN/AAAV8-EGFP-P2A-Di-C5Full AAV2.8-hSyn-EGFP:HA-GCN4-GS4-HWLKVThis paperN/AAAV8-EGFP-Di-C5Full AAV2.8-hSyn-EGFP:HA-GCN4-GS4-HWLKVThis paperN/AAAV8-iCre:EGFP-P2A-Di-C5Full AAV2.8-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKVThis paperN/AAAV8-DIO[EGFP-P2A-Di-C5]ONFull AAV2.8-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]ONThis paperN/AAAV8-DIO[EGFP-P2A-Di-C5]OFFFull AAV2.8-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]OFFThis paperN/AAAV8-DIO[EGFP:Di-C5]ONFull AAV2.8-hSyn-DIO[EGFP:HA-GCN4-GS4-HWLKV]ONThis paperN/ApUCmini-iCAP-PHP.SChan et al.^29^Addgene Cat# 103006; RRID: Addgene_103006pUCmini-iCAP-PHP.eBChan et al.^29^Addgene Cat# 103005; RRID: Addgene_103005AAVPHP.S-iCre:EGFP-P2A-Di-C5Full AAV2.PHP.S-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKVThis paperN/AAAVPHP.S-iCre:EGFPFull AAV2.PHP.S-hSyn-iCre:EGFPThis paperN/AAAVPHP.eB-iCre:EGFP-P2A-Di-C5Full AAV2.PHP.eB-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKVThis paperN/ABiological samplesHuman Female Lumbar DRG 3, 4, or 5AnaBios230116DHAHuman Male Lumbar DRG 3, 4, or 5AnaBios221001DHAHuman Female Lumbar Spinal Cord (L3-L5)AnaBios211117ScHAHuman Male Lumbar Spinal Cord (L3-L5)AnaBios201015ScHAChemicals, peptides, and recombinant proteinsGCN4-HWLKVSequence: biotin-ahx-RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-HWLKVTAG Copenhagen; This paperN/AGCN4(7P14P)-HWLKVSequence: biotin-ahx-RMKQLEPKVEELLPKNYHLENEVARLKKLVGGGGS-HWLKVTAG Copenhagen; This paperN/AGCN4-GGGGSSequence: biotin-ahx-RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-GGGGSTAG Copenhagen; This paperN/ASSO10a-HWLKVSequence: biotin-ahx-GEELLEDIRKFNEMRKNMDQLKEKINSVLSIRQGGGGS-HWLKVTAG Copenhagen; This paperN/AGCN4-IETDVSequence: biotin-ahx- RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-IETDVTAG Copenhagen; This paperN/ATroponin-HWLKVSequence: biotin-ahx-KEDAKGKSEEKEDAKGKSEEKEDAKGKSEEGGGGSHWLKVTAG Copenhagen; This paperN/ATroponinLeuZip-HWLKVSequence: biotin-ahx- KLDALGKSLEKLDAKLKSLEKELAKLKSELGGGGSHWLKVTAG Copenhagen; This paperN/ApET41-PICK1-WTMadsen et al.^46^N/ApET41-PICK1-A87LMadsen et al.^46^N/ACritical commercial assaysPierce™ BCA Protein Assay KitsInvitrogenCat# 23225Quant-iT™ PicoGreen™ dsDNA Assay KitInvitrogenCat# P11496RNeasy Lipid Tissue Mini KitQiagenCat# 74804Maxima First Strand cDNA Synthesis KitThermoFisherCat# K1641LIVE/DEAD™ Fixable Near IR (780) Viability Kit, for 633 nm excitationInvitrogenCat# L34992Deposited datasnRNAseq, raw dataThis paperSRA study SRA: SRP645687; SRA: SRX31117391, SRX31117392snRNAseq, processed dataThis paperZenodo: 17629531MS global proteomeThis paperPRIDE: PXD074049MS phosphoproteomeThis paperPRIDE: PXD073925Experimental Organisms/strainsMouse: Hoxb8^tm2.1(cre)Mrc^/JThe Jackson Laboratory; Witschi et al.^25^JAX: 035978; RRID: IMSR_JAX:035978Mouse: Ai14 (B6.Cg-Gt(ROSA)26Sor^tm14(CAG-tdTomato)Hze^/J)The Jackson LaboratoryJAX: 007914; RRID: IMSR_JAX:007914Mouse: Advillin-Cre (B6.129P2-Avil^tm2(cre)Fawa^/J)The Jackson Laboratory;Zhou et al.^27^JAX: 032536; RRID: IMSR_JAX:032536Mouse: SNS-Cre (C57BL/6-Tg(SCN10A-Cre)1Rkun/Uhg)Provided by Professor Rohini Kuner at Heidelberg University, Germany; Agarwal et al.^28^MGI: 3042874OligonucleotidesEef1a1 (forward): TGCTGGAGCCAAGTGCTAATEurofins Genomics;Košuth et al.^47^N/AEef1a1 (reverse): GTGCCAATGCCGCCAATTTTEurofins Genomics;Košuth et al.^47^N/AYhwaz (forward): GAAGCATTGGGGATCAAGAAEurofins Genomics;Sørensen et al.^48^N/AYhwaz (reverse): AGACGGAAGGTGCTGAGAAAEurofins Genomics;Sørensen et al.^48^N/AAAV-Di-C5 (forward): ATCCGTATGATGTGCCGGATTEurofins Genomics; This paperN/AAAV-Di-C5 (reverse): CGCCACTTCGTTTTCCAGATGEurofins Genomics; This paperN/ASoftware and algorithmsAlphafold (colab v2.1.0)Jumper et al.^49^https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynbGROMACS (v2021.4)Abraham et al.^50^https://manual.gromacs.org/documentation/Prism (v9.3.1 or newer)GraphPadhttps://www.graphpad.comImageJ FijiImageJhttps://imagej.net/software/fiji/downloadsZEN Microscopy SoftwareZEISShttps://www.zeiss.com/microscopy/en/service-support/downloads.htmlQuPath (v0.5.1)Bankhead et al.^51^https://qupath.github.ioFlowJo (v10)BDhttps://www.flowjo.com/flowjo/overviewThermo Proteome Discoverer (v3.2.0.450)ThermoFisher Scientifichttps://www.thermofisher.com/dk/en/home/industrial/mass-spectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/multi-omics-data-analysis/proteome-discoverer-software.htmlProteomelit (v1.1.0)Proteomics Research Infrastructure (PRI), University of Copenhagen; Santos et al.^52^N/AVolcaNoseRGoedhart & Luijsterburg^53^https://huygens.science.uva.nl/VolcaNoseR/Cell Ranger (v9.0.1)10x Genomicshttps://www.10xgenomics.com/support/software/cell-ranger/latestCellBender (v0.3.0)Fleming et al.^54^https://github.com/broadinstitute/CellBenderScrubletWolock et al.^55^https://github.com/AllonKleinLab/scrubletPagoda2 (v1.0.13)pagoda2: Single Cell Analysis and Differential Expression. R package version 1.0.13https://github.com/kharchenkolab/pagoda2ConosBarkas et al.^56^https://github.com/kharchenkolab/conos
Primary cultures were made from pooled cortices from postnatal day 1 (P1) pups (assuming mixed sexes) from C57BL6/J female mice (Charles River) and cultured in filter-sterilized Neurobasal A media (10888022 ThermoFisher) supplemented with 1x GlutaMAX (35050061 Gibco), 2% B-27 Plus 50x (A3582801 ThermoFisher), 0.12% Penicillin-Streptomycin (P0781 Sigma), and 0.1% Kynurenic acid solution (0.5M Kynurenic acid (K3375 Sigma) in 1M NaOH) and incubated at 37°C in CO2 incubator (5% CO2, 95% relative humidity) for at least two days before transduction with AAVs.
HEK293t cells were used for AAV production. The cells were cultured in 500 mm^2^ dishes (Corning) in high glucose DMEM with GlutaMAX and pyruvate (31966047 Gibco) supplemented with 10% fetal bovine serum and 1% Penicillin-Streptomycin (P0781 Sigma) until 12–24 h post-transfection with AAV plasmid DNA.
The cell lines have not been authenticated. All cell lines are routinely screened for mycoplasma contamination with MycoAlert Mycoplasma Detection Kit (Lonza).
Unless otherwise specified, male or female C57BL/6NRj, male C57BL/6JRj (Janvier, France), or male C57BL/6JBomTac (Taconic, Denmark) wild-type (WT) mice between 7 and 12 weeks of age were ordered externally. The Ai14 (homozygous, both sexes, JAX:007914) and Hoxb8-Cre (heterozygous, both sexes, JAX:035978)^25^ mouse strains were generously provided by Prof. Ole Kiehn and propagated within the animal facilities. Transgenic SNS-Cre (heterozygous, both sexes, MGI:3042874)^28^ mice were provided by Professor Rohini Kuner for experiments at Heidelberg University. Advillin-Cre (heterozygous, males, JAX:032536)^27^ mice were acquired from Jackson Laboratory.
Throughout our study, we did not find that sex influenced the results of behavior experiments. For some behavioral experiments (von Frey, brush, and coldplate) with genetically modified mice (Figures 3J, S4C, S4D, S5D, S6C, and S6D), we pooled the sexes to achieve a proper sample size. These experiments have sample sizes for each sex ranging from N = 2–6. Despite the small sample size, no statistical difference was found between the sexes, and the results did not warrant further investigation of sex differences. We used the sexes interchangeably across both behavioral and non-behavioral test methods, limiting the ability to uncover non-behavioral sex differences.
All mice were drug- and test-naive before the start of experiments. Mice were housed in individually ventilated cages (IVC) with access to water and chow food ad libitum, maintained on a 12-h light/dark cycle, and group-housed (2–8 mice per cage). Upon arrival at the animal facility, littermates of the same sex were randomly separated into cages. Animals were allowed at least 7 days of habituation to the animal facility before the initiation of the experiment. Cages were randomly allocated an AAV and/or surgery. For all CFA experiments, however, animals were randomly assigned an AAV irrespective of their cage littermates. All experiments involving animals were performed in AAALAC-accredited animal facilities and approved by the Danish Animal Experiments Inspectorate (#2016-15-0201-00976, #2019-15-0201-0160, #2021-15-0201-01036) or by the Regierungspräsidium, Karlsruhe, Germany (#G184/18).
Two DRG pairs from the lumbar region (L3-L5) from one female (46 years, Caucasian) and one male (52 years, Caucasian) and two spinal cord sections from the lumbar region (L3-L5) from one female (52 years, Caucasian) and one male (55 years, Caucasian) human donors were acquired from AnaBios, San Diego, US. All donor organ transfers to AnaBios are fully traceable and regularly reviewed by U.S federal authorities. Informed consent was obtained from each donor by AnaBios, specifying that the donor tissue can be utilized strictly for laboratory purposes only. None of the donors had a known history of neuropathic pain. Tissues were flash-frozen in liquid nitrogen after dissection and shipped under temperature-controlled conditions. All samples were anonymized, and demographic information was provided in a form that prevents any possibility of tracing them back to individual donors.
The tissue samples were used for detection of PICK1 protein and confirmation of target engagement in human DRG and spinal cord in both sexes. As a limitation, the sample size was N = 1 (with three technical replicates) for each tissue type and sex, and no conclusion can be made on any differences between the sexes.
The fluorophore-conjugated peptide, 5FAM-di-HWLKV, was conjugated in-house with the fluorophore, 5FAM, through solid-phase peptide synthesis, as described elsewhere.^15^ All other synthetic peptides were purchased from TAG Copenhagen, Denmark, and verified >95% pure through UPLC-MS. These synthetic peptides, if not otherwise stated, were biotin-conjugated through N-terminal Ahx-linkage. Synthesis of biotin-ahx-Atg16-HWLKV and biotin-ahx-MDV1-HWLKV was attempted but was unsuccessful due to complexity of these compounds. The following peptides (name: primary sequence) were
HWLKV: HWLKV.
5FAM-di-HWLKV: 5FAM-PEG4-(HWLKV)2
GCN4-HWLKV: biotin-ahx-RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-HWLKV.
GCN4(7P14P)-HWLKV: biotin-ahx-RMKQLEPKVEELLPKNYHLENEVARLKKLVGGGGS-HWLKV.
GCN4-GGGGS: biotin-ahx-RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-GGGGS.
SSO10a-HWLKV: biotin-ahx-GEELLEDIRKFNEMRKNMDQLKEKINSVLSIRQGGGGS-HWLKV.
GCN4-IETDV: biotin-ahx- RMKQLEDKVEELLSKNYHLENEVARLKKLVGGGGS-IETDV.
Troponin-HWLKV: biotin-ahx-KEDAKGKSEEKEDAKGKSEEKEDAKGKSEEGGGGS-HWLKV.
TroponinLeuZip-HWLKV: biotin-ahx-KLDALGKSLEKLDAKLKSLEKELAKLKSELGGGGS-HWLKV.
50 mL LB with kanamycin was inoculated with Escherichia coli (BL21-DE3-pLysS) containing a PICK1 or PICK1-A87L encoding pET41 plasmid^46^ and incubated at 37°C with shaking overnight. The pre-cultures were transferred to 1 L LB media containing kanamycin with continued incubation. When the optical density at 600 nm, OD600, reached 0.6, protein expression was induced with 1 mM IPTG, and the temperature was turned down to 20°C with shaking overnight. The cells were harvested through centrifugation (F9 rotor, 7,800 x g, 12 min, 4°C) and suspended in lysis buffer (50 mM Tris, 125 mM NaCl, 2 mM DTT, 1% Triton X-100, 20 μg/mL Dnase 1, one tablet of cOmplete Protease Inhibitor Cocktail (Roche) pr 200 mL lysis buffer, pH 7.4). After a −80°C/4°C freeze-thaw cycle, the proteins were extracted through centrifugation (F20 rotor 36,000 x g, 30 min, 4°C), and the supernatant was incubated with 750 μL PBS-washed Glutathione Sepharose 4B beads (GE Healthcare) for two hours with gentle turning. Three centrifugation-wash cycles were performed with gentle centrifugation (3,000 x g, 5 min, 4°C) and TBS wash buffer (50 mM Tris, 125 mM NaCl, 2 mM DTT, 0.01% Triton X-100, pH 7.4). The PICK1- or PICK1-A87L-beads complex was transferred to a PD 10 gravity column and further washed with three column volumes of TBS wash buffer before being collected and incubated with 5 μL thrombin (0.075U/μL stock) overnight at 4°C with gentle turning. The cleaved samples were eluted, and absorbance was measured on NanoDrop 2000 at 280 nM. The extinction coefficient of PICK1, εA280PICK1 = 32320 M^−1^∗cm^−1^. The samples were always kept at 4°C or on ice.
Fluorescence polarization (FP) binding saturation assay was carried out with an increasing concentration (Cmax = 30 μM) of recombinant PICK1 or PICK1-A87L and a fixed concentration of synthetic fluorophore-conjugated peptide, 5FAM-di-HWLKV (20 nM). FP binding competition assay was carried out with a fixed concentration of PICK1 or PICK1-A87L (75% of [PICK1] or [PICK1A87L] at maximum binding in saturation assay), a fixed concentration of fluorophore-conjugated peptide, 5FAM-di-HWLKV (20 nM), and an increasing concentration of an unconjugated test peptide (Cmax = 100 μM). Both assays were carried out in 96-well plates (Corning, half-area, black, non-binding) using TBS buffer (1x TBS, 2 mM DTT, 0.01% Triton X-100) and incubated at 4°C for 20 h after mixing to ensure equilibrium was reached. The plates were read on a POLARstar Omega plate reader using an excitation filter at 488 nm and an emission filter at 535 nm. The data were plotted using GraphPad Prism and fitted to a sigmoidal dose-response curve for the saturation assay or a one-site competition curve for the competition assay. Ki’s were automatically calculated by GraphPad Prism using the Cheng-Prusoff equation. Three technical replicates were performed simultaneously, and the assay was repeated 2–3 times.
Fast-protein liquid chromatography (FPLC) was performed with varying ratios of PICK1 to a synthetic peptide in sterile-filtered TBS buffer (1x TBS, 2 mM DTT, 0.01% Triton X-100) on a 24 mL size-exclusion chromatography column (GE Healthcare Superdex 200 increase 10/300 GL). Samples were mixed in a total volume of 500 μL in buffer, which was injected into the column, followed by 30 mL buffer. Samples with PICK1:peptide complexes were allowed a minimum of 20 h of incubation at 4°C to reach equilibrium. Samples containing either PICK1 or synthetic peptide were run after a minimum of 20 min incubation on ice. Absorbance was detected at 280 nm. FPLC curves were plotted using GraphPad Prism.
For circular dichroism (CD), peptides were diluted to 30 μM in 50 mM NaPi buffer (pH 8.0) and measured on a Jasco J1500 at 25°C using a quartz cell with a 1 mm pathlength cuvette. Spectra were recorded from 260 to 190 nm with 0.1 nm step resolution and 50 nm/min scan speed. The mDEG signal was converted to molar ellipticity θ (deg ∗ cm^2^ ∗ dmol) using the equation θ = (mDEG∗106)/(C∗N∗L), where mDEG is the measured signal, C is protein concentration (μM), N is the peptide residue length, and L is the cuvette pathlength (mm). CD spectra were plotted using GraphPad Prism.
Initial structures of GCN4-HWLKV and GCN4(7P14P)-HWLKV were generated in Alphafold (colab v2.1.0).^49^ Molecular dynamics (MD) simulations were run in GROMACS 2021.4^50^ using the Charmm36_ljpme force field,^50^ as this force field best reflected the circular secondary structure of the peptides, as measured with circular dichroism (Figure S1C). The peptides were aligned with the dimeric structure of the GCN4 basic region leucine zipper (PDB: 1YSA).^57^ The dimers were placed in a cubic box with periodic boundary conditions, with a min 2 nm distance from protein to edge (approx. 11 × 11 × 11 nm) and solvated in TIP3P water with 0.1 nM NaCl. The system was energy-minimized followed by an equilibration with a constant number of particles, volume, and temperature (NVT) for 100 ps with 2 fs timesteps. The temperature was fixed at 300 K using v-rescale temperature coupling.^58^ This was followed by 100 ps equilibration with a constant number of particles, pressure, and temperature (NPT). The pressure was kept at 1 bar using isotropic Parinello-Rahmen pressure coupling.^59^ One peptide was fixed using a harmonic position restraint with a force constant of 1,000 kJ/mol on all atoms. The other peptide was then pushed toward and then pulled away from the first peptide, using GROMACS pull code at a rate of 1 nm per ns and a harmonic potential with a force constant of 500 kJ/mol/nm.^2^ Frames with 0.2 nm distance were extracted from the push and pull trajectories. Each frame was simulated for 10 ns with constant NPT, with the first peptide kept in place by a position restraint and the second peptide fixed using a 500 kJ/mol/nm^2^ harmonic restraint. The potential of mean force was calculated using the weighted histogram algorithm method,^60^ where the first 100 ps of each window were omitted from the analysis. This process was repeated 5 times for each construct to obtain mean values and standard errors, which are reported. The free energy of dimerization was estimated as the difference between the minimum free energy value and the bulk value.
10 μL of undiluted AAV or vehicle (DPBS, #14190-144, Gibco) was mixed with 10 μL 2x Laemmli buffer, boiled (100°C, 6 min), and loaded on a 4–15% gradient (#4561083, Bio-Rad) SDS-PAGE gel (120 volts, 2 h). The gel was washed in water for 5 min, stained with Imperial Protein Stain (#24615, Thermo Fisher Scientific) (1 h, RT), and destained in water overnight. Imaging was performed using an Amersham Imagequant 800 imager.
DRGs L1-L5, lumbar spinal cord, and brainstem located beneath cerebellum were dissected and pooled from 2 naive male WT mice and immediately homogenized in 400 μL ice-cold lysis buffer (100 mM NaCl, 50 mM Tris-Cl, 1% sodium deoxycholate, protease inhibitor cocktail (10 μL in 1 mL lysis buffer, P8340, Sigma) followed by agitation on spinning wheel (4°C, 30 min). The lysates were centrifuged (10 min, 1,000–1,500 x g, 4°C), supernatant collected, re-homogenized in 400 μL ice-cold lysis buffer (only for spinal cord and brainstem), before another centrifugation (10 min, 1,000–1,500 x g, 4°C) and pooling of supernatants. The lysates were centrifuged (15 min, 4°C, 13,000 x g) and the supernatant collected.
The mouse lysates were pre-cleared with 3x PBS-washed streptavidin beads (15μL, Dynabeads MyOne Streptavidin T1, 65601, Invitrogen) on a spinning wheel (4°C, 1 h). The total amount of protein in the lysates was determined by a BCA assay (Invitrogen) and read on Omega POLARstar plate reader (BMG Labtech). Simultaneously, streptavidin beads (30 μL) were pre-incubated with 10 μM biotinylated peptide (GCN4-GGGGS, GCN4-HWLKV, GCN4(7P14P)-HWLKV, GCN4-IETDV, Troponin-HWLKV, or TroponinLeuZip-HWLKV) in 500 μL TBS buffer (1x TBS, 2 mM DTT, 0.01% Triton X-100) on a spinning wheel (4°C, 3 h), followed by removal of unbound peptide through 3x washes in TBS buffer. Pre-cleared lysate (500 μg protein for spinal cord and brainstem, 80 μg protein for DRGs) was added to pre-incubated bead-peptide mix and incubated overnight on spinning wheel (4°C). Unbound material was gently removed through 3x washes in TBS buffer.
The pulled peptides/proteins were eluted from the beads with 30 μL 2x Laemmli buffer and boiling (100°C, 6 min). Input (30 μg protein for spinal cord and brainstem, 20 μg protein for DRGs) was likewise boiled in 30 μL 2x Laemmli buffer. The resulting pulled peptides/proteins and input were run on a 4–15% gradient (#4561083, Bio-Rad) or a 12% (#4561044, Bio-Rad) SDS-PAGE gel and transferred to a membrane. The blot was blocked in 5% (w/v) non-fat dry milk (#9999, Cell Signaling Technology) in PBS-T buffer (PBS with 0.01% Tween 20, P9416, Sigma-Aldrich) and incubated in primary antibody against PICK1 (1:1,000, mouse monoclonal, Antibodies Incorporated #75-040, NeuroMab L20/8) or PSD-95 (1:500, mouse monoclonal, ab192757, Abcam) (4°C, overnight, gentle shaking). The blots were washed for 5 × 5 min in PBS-T buffer and incubated in secondary antibody (1:2,000, Goat anti-mouse IgG HRP, #31430, Pierce) in PBS-T buffer (1.5 h, gentle shaking) followed by 5 × 5 min washes in PBS-T buffer. The protein complexes were analyzed on Alpha Innotech (Flour H2D software) with SuperSignal ELISA Femto Substrate (#37075, Thermo Scientific) for PICK1 and ECL Prime Western Blotting Detection (#89168-782, GE Healthcare) for PSD-95.
Lysate preparation, pull-down, SDS-PAGE, and western blotting were performed as done for mouse tissue with the following exceptions.
Lysates from human tissue were made from DRGs (L3, L4 or L5) from one female and one male donor, and spinal cord section (from lumbar region L3-L5) from one female and one male donor.
50 μg protein was used for pull-down conditions and 20 μg protein for input.
Proteins were eluted with 20 μL Laemmli buffer, and Any kD SDS-PAGE gels (#4569035, Bio-Rad) were used. The protein complexes were analyzed on Amersham ImageQuant 800 (Cytiva) with SuperSignal ELISA Femto Substrate (#37075, Thermo Scientific).
The AAV plasmids (pAAV) were all designed using the same basic template. Transgene expression was regulated by the human synapsin 1 (hSyn) promoter to achieve stable, pan-neuronal expression.^61^ Downstream of the transgene, a woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) and human growth hormone polyadenylation (hGH polyA) were incorporated. The entire gene cassette was flanked by AAV2 inverted terminal repeats (ITR).
The therapeutic transgene consists of the DNA sequence for the basic leucine zipper region of the yeast transcription factor, GCN4-p1 (GCN4),^62^ a short flexible linker region, and the PICK1-binding motif, HWLKV.^46^^,^^63^ The GCN4(7P14P) variant has two prolines in positions 7 and 14. Alternative zippers are; SSO10a,^64^ Atg16,^65^ and MDV1.^66^ The linker region consists of 4 glycine’s and 1 serine residues (GGGGS). The C-terminus is composed of HWLKV for achieving PICK1 binding or exchanged with GGGGS, which serves as a negative control.
For all pAAVs containing GCN4, GCN4(7P14P), SSO10a, Atg16, or MDV1 sequences, a human influenza hemagglutinin (HA) epitope tag was inserted upstream of the zipper region. This HA-tag was also included in fluorescently tagged transgenes, i.e., EGFP or iCre:EGFP. Posttranslational self-cleavage was obtained by inserting a 2A peptide sequence derived from porcine teschovirus-1 polyprotein (P2A)^67^ between EGFP/iCre:EGFP and the transgene. To induce expression in Ai14 transgenic mice, codon-optimized Cre (iCre) was fused to EGFP (iCre:EGFP). To achieve Cre-dependent expression with the Cre ON/OFF system, the expression cassette was flanked by two sets of lox sites (loxP and lox2272).
DNA sequences of HA-GCN4-GS4-HWLKV, HA-GCN4(7P14P)-GS4-HWLKV, HA-GCN4-GS4-GGGGS, HA-SSO10a-GS4-HWLKV, HA-Atg16-GS4-HWLKV, and HA-MDV1-GS4-HWLKV were ordered in pEX plasmid vectors from Eurofins Genomics, Germany. Restriction sites Eco53kI and EcoRI were used to excise the transgene of one pEX vector and inserted into pAAV-hSyn-P2A-EYFP-WPREpA excised with HpaI and EcoRI resulting in pAAV-hSyn-HA-GCN4-GS4-HWLKV. The remaining vectors were made by using the restriction sites KpnI and EcoRI from pAAV-hSyn-HA-GCN4-GS4-HWLKV and the pEX vectors.
Vectors containing EGFP and/or iCre were cloned from the sequences BsrGI-P2A-HA-GCN4-HWLKV-HindIII and BsrGI-HA-GCN4-GS4-HWLKV-HindIII (ordered from Eurofins Genomics, Germany), cut with BsrGI and HindIII and inserted into the BsrGI and HindIII-cut backbones of pAAV-hSyn-EGFP-WPREpA and pAAV-hSyn-iCre:EGFP-WPREpA.
The DIO vectors, pAAV-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]ON and pAAV-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]OFF were made using In-Fusion HD Cloning kit (Takara) through PCR amplification of pAAV-hSyn-EGFP-P2A-HA-GCN4-HWLKV with the respective primer 5′-tgctagctcgactagatatccagcacagtggcggcc-3’ (Fw) and 5′-ggcgcgcccgccatattgcggccgcttacact-3’ (Rv), and 5′-ggcgcgcccgccatagccaccatggtgagcaagggcgaggag-3’ (Fw) and 5′-acgaagttatgctagttacactttcagccaatggctgcc-3’ (Rv), and inserted into the NdeI and SpeI-cut or NdeI and NheI-cut of the pAAV-hSyn-DIO-MSC-WPREpA backbone, respectively.
The vectors pAAV-hSyn-EGFP-WPREpA, pAAV-hSyn-iCre:EGFP-WPREpA, pAAV-hSyn-DIO-MSC-WPREpA, and pAAV-hSyn-TdTomato-WPREpA were already pre-made in the lab. All final AAV vector plasmids were sequenced and validated prior to AAV production.
Recombinant AAV vectors were produced in-house using a linear polyethylenimine (PEI, MW 25,000) triple-transfection protocol and purified with an iodixanol density gradient column. In brief, for each viral construct, 3 × 500 mm^2^ dishes (Corning) were seeded with 140 million HEK293t cells/dish the day before transfection in DMEM (Glutamax, 10% FBS, 1% P/S, 4.5g/L glucose). The cells were transfected with 400 μg DNA/3 dishes in a 4:2 ratio of pAAV: rep/ pHelper (Addgene plasmid #122867), and using linear PEI.^68^ The day following transfection, media was exchanged to low serum DMEM (Glutamax, 1% FBS, 1% P/S, 4.5g/L glucose) and incubated for 96 h before harvesting of AAV vector in both media and cells as previously described.^68^ In brief, cells were harvested (15 min, 4°C, 2000 x g) and separated from media. Cell pellet was PBS-washed (10 min, 4°C, 1,000 x g) and resuspended in 0.8 mL buffer A (50mM Tris, 150mM NaCl, pH 8.4) before three freeze-thaw cycles were performed. Cells were triturated through 23 G needle and incubated on ice. Media was mixed with 20% (v/v) of 40% PEG8000 solution, incubated for 2 h on ice, and centrifuged (swing bucket; 30 min, 4°C, 4,000 x g). PEG pellet was resuspended with 1 mL buffer A, combined with cell lysate, and incubated for 30 min at 37°C with 250 U benzonase. The AAVs were purified from the 44% layer of an iodixanol density gradient column with 17% (0.72 x PBS, 0.72 mM MgCl2, 1.8 mM KCl, 1 M NaCl), 28% (0.54 x PBS, 0.54 mM MgCl2, 1.34 mM KCl), 44% (0.26 x PBS, 0.26 mM MgCl2, 0.65 mM KCl) and 54% (1 x PBS, 1 mM MgCl2, 2.5 mM KCl) iodixanol following 15 h swing bucket ultracentrifugation (rotor SW28.1, 28,000 rpm/rmax = 150,000 rcf, 4°C). AAVs were washed with DPBS (14190 Gibco) two times, concentrated in spin columns (15 mL, 100 kDa cutoff, Milipore) (15 min, 4°C, 3,500 x g), and supplied with 0.005% (v/v) pluronic F68 (Gibco) before storage at −80 to −70°C. Purity was confirmed with SDS-PAGE, and titer (vg/mL) was determined with Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). All AAVs are made in serotype AAV2.8 unless otherwise specified. The serotypes AAV2.PHP.S and AAV2.PHP.eB were used for the transduction of PNS or CNS, respectively.^29^ All AAV vectors were produced within the lab at University of Copenhagen, and their abbreviated and full names are as
AAV8-Di-C5: AAV2.8-hSyn-HA-GCN4-GS4-HWLKV.
AAV8-Di(7P14P)-C5: AAV2.8-hSyn-HA-GCN4(7P14P)-GS4-HWLKV.
AAV8-tdTomato: AAV2.8-hSyn-tdTomato.
AAV8-Di-SGGGG: AAV2.8-hSyn-HA-GCN4-GS4-GGGGS.
AAV8-SSO10a-C5: AAV2.8-hSyn-HA-SSO10a-GS4-HWLKV.
AAV8-Atg16-C5: AAV2.8-hSyn-HA-Atg16-GS4-HWLKV.
AAV8-MDV1-C5: AAV2.8-hSyn-HA-MDV1-GS4-HWLKV.
AAV8-EGFP-P2A-Di-C5: AAV2.8-hSyn-EGFP-P2A-HA-GCN4-GS4-HWLKV.
AAV8-EGFP-Di-C5: AAV2.8-hSyn-EGFP:HA-GCN4-GS4-HWLKV.
AAV8-iCre:EGFP-P2A-Di-C5: AAV2.8-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKV.
AAV8-DIO[EGFP-P2A-Di-C5]ON: AAV2.8-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]ON.
AAV8-DIO[EGFP-P2A-Di-C5]OFF: AAV2.8-hSyn-DIO[EGFP-P2A-HA-GCN4-GS4-HWLKV]OFF.
AAV8-DIO[EGFP:Di-C5]ON: AAV2.8-hSyn-DIO[EGFP:HA-GCN4-GS4-HWLKV]ON.
AAVPHP.S-iCre:EGFP-P2A-Di-C5: AAV2.PHP.S-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKV.
AAVPHP.S-iCre:EGFP: AAV2.PHP.S-hSyn-iCre:EGFP.
AAVPHP.eB-iCre:EGFP-P2A-Di-C5: AAV2.PHP.eB-hSyn-iCre:EGFP-P2A-HA-GCN4-GS4-HWLKV.
Primary cultures of mouse postnatal cortical neurons were transduced with AAVs and incubated for a minimum of 6 days. The cells were fixed, permeabilized, and blocked (5% goat serum, 0.3% Triton X-100), and incubated with rabbit anti-GCN4 (1:275, C11L34, Absolute Antibodies) and A488 goat anti-rabbit (1:5,000, A11034, Invitrogen) before stained with mounting media containing DAPI (4',6'-diamidino-2-phenylindole). Cells were imaged with a Zeiss LSM 780 laser scanning confocal microscope with a 63x objective. Images were processed using ImageJ Fiji or ZEN (ZEISS software).
Ai14 reporter mice receiving 7 μL i.t injection of AAV2.8-iCre:EGFP-P2A-Di-C5 mice were perfused with PBS followed by 4% paraformaldehyde (PFA) 40–44 days post injection. Brain, spinal cord, and DRGs were dissected, and tissue was kept in 4% PFA overnight, stored in 30% sucrose for 2 days, and finally embedded in OCT (TissueTek). DRGs (10 μm), spinal cords (20 μm), and brains (40 μm) were cut with a cryostat (Leica CM3050 S). Tissue sections were mounted with media containing DAPI and imaged using AxioScan Z1 with 20x objective. Image visualization was performed using ZEN (ZEISS) and ImageJ Fiji software.
For DRG immunohistochemistry, 10 μm DRG sections were blocked in blocking buffer (PBS with 0.3% Triton X-100 and 10% donkey serum) for at least 1 h before they were incubated in primary antibodies diluted in blocking buffer overnight at room temperature. Next, they were washed in PBS with 0.3% Triton X-100 3 × 5 min, incubated in secondary antibodies diluted in blocking buffer for at least 4 h, washed 3 × 5 min, and mounted with mounting medium. The stained DRG sections were imaged on a ZEISS Axio Scan.Z1 slide scanner and analyzed using ZEISS ZEN software. Primary and secondary antibodies, including their dilution, tdTomato: Rabbit anti-mCherry (1:200, PA5-34974, Thermo Fisher Scientific); Goat anti-rabbit 568 (1:500, A11036, Invitrogen) or donkey anti-rabbit 568 (1:500, A10042, Life technologies); IB4: IB4-647 (1:500, I32450, ThermoFisher); CGRP: Goat anti-CGRP (1:400, ab36001, Abcam); Donkey anti-goat 647 (1:500, A21447, Invitrogen); NF200: Chicken anti-NF200 (1:500, Ab5539, Millipore); Goat anti-chicken 647 (1:500, A21449, Life technologies).
Female C57BL/6N WT mice (8 weeks old) receiving a 7 μL i.t. injection of AAV8-Di-C5, AAV8-tdTomato, or vehicle were perfused with PBS followed by 4% PFA 28 days after injection. The left liver lobe was dissected and post-fixed in 10% neutral buffered formalin for 3 days at 4°C followed by paraffin embedding, slicing, deparaffinization, and staining with filtered hematoxylin & eosin (H&E) with a Leica Auto stainer XL. Slices were imaged with an Axioscanner 7 microscope at 10× magnification, and immune cell infiltrations were quantified in QuPath^51^ and presented as a percentage of immune area of total tissue area. For each mouse, 5–6 liver sections, 3 μm thick and spaced at least 30 μm apart, were analyzed.
Ai14 mice received a 7 μL i.t. injection of AAV2.8-iCre:EGFP-P2A-Di-C5. 40 days post-injection, mice were euthanized and perfused with 20 mL of PBS followed by 20 mL of 4% PFA to ensure fixation. A tissue sample was dissected with the entire vertebrae and skull attached. Samples were post-fixed in 4% PFA for 24 h at 4°C, then washed with PBS containing sodium azide (NaN3) 0.02%.
The tissue-clearing and immunolabeling processes were adapted from the iDISCO protocol^69^ with several modifications to enhance tissue preparation and labeling efficiency. Initially, decalcification was conducted by immersing the samples in 20% EDTA (ED4SS, Sigma-Aldrich) at a pH of approximately 7.4, maintained at 37°C for four days. Following decalcification, the samples were thoroughly washed with water and dissected to remove the skull, preparing them for subsequent processing. Before initiating the standard iDISCO protocol, an additional delipidation and bleaching phase was incorporated. This involved treating the samples with 25% Quadrol (#122262, Sigma-Aldrich) for 48 h at 37°C, followed by 5% ammonium hydroxide (#35574, ThermoFischer) for 24 h at the same temperature. After these preparatory steps, the samples were processed according to the iDISCO method, including the standard immunolabeling procedures. The primary CD45 antibody conjugated to AF594 was incubated for 10 days (1:1,000, #103144, Biolegend) or the RFP antibody was incubated for 10 days (1:1,000, #600-401-379, Rockland) followed by the 647 donkey anti-rabbit (1:1,000, #711-605-152, Jackson ImmunoResearch) secondary antibody for 7 days at 37°C. Following the application of the secondary antibody, an extra post-fixation step was introduced to stabilize the immunocomplexes, involving overnight incubation in 2% PFA at 4°C. This additional step ensured enhanced preservation of tissue morphology and labeling fidelity prior to the final dehydration process. Ethyl cinnamate (W243000, Sigma-Aldrich) was used as the final clearing solvent following the two washes of dichloromethane.
The acquisitions were done on a ZEISS LS7 scanning Gaussian-beam light sheet microscope with orthogonal light paths for illumination and detection. Two 5x illumination objectives (ZEISS, NA 0.1) were used to generate dual side illumination of the sample, together with an EC Plan-NEO 5× detection objective (ZEISS, NA 0.16, 10.5 mm working distance). Images were acquired with two PCO Edge 4.2M sCMOS cameras using ZEISS ZEN Black software. The laser line used to capture the autofluorescence signal in the sample was 488 nm (Diode laser, 30 mW). The laser line used to capture the RFP signal was 638 nm (Diode laser, 75 mW). To refine the emission signal additional filters were used, i.e., BP (band pass) 505–545 nm and LP (long pass) 660 nm for the corresponding excitation.
Relative gene expression was examined through quantitative Real-Time PCR (qRT-PCR). Seven-week-old WT animals received i.t. injections of 7 μL AAV8-Di-C5 or DPBS (n = 4). Four weeks later, the mice were sacrificed, and sciatic nerve, DRGs (L3-L5 from both sides), lumbar spinal cord, dorsal column, cervical spinal cord, pons, cerebellum, and liver were dissected and immediately flash-frozen on dry ice. Hereafter, tissue samples were stored at −80° until RNA extraction. All surfaces and tools were pre-treated with 70% ethanol and/or RNaseZap Solution (ThermoFisher) to prevent RNAse activity. Each tissue sample was directly placed into 1mL of QIAzol Lysis Reagent (Qiagen), followed by homogenization with a handheld pestle and trituration through a 21G needle as previously described.^48^ Automated RNA extraction was performed with RNeasy Lipid Tissue Mini Kit (Qiagen) on a semi-automated QIAcube Connect (Qiagen) according to manufacturer’s instructions. Reverse transcription was performed with Maxima First Strand cDNA Synthesis Kit (ThermoFisher) using 500 ng RNA from each sample or 200 ng RNA from the sciatic nerve. The resulting complementary DNA (20mL reaction) was diluted with 125 mL EB buffer (Qiagen) and added onto a 384-well plate (0.4 μL per well) already loaded with 2 μL/well of Sybr green Mastermix (Roche Life Science) and 1.6 μL/well of primer solution (forward and reverse primers, 0.95 μM). Mixing was performed with a contactless liquid handler (I-DOT, Immediate Drop-on-demand Technology, Dispendix). Real-time quantitative PCR was performed on a LightCycler 480 II instrument (Roche Life Science) using a customized protocol^48^ consisting of initial denaturation (10 min, 95°C) followed by 40 cycles of denaturation (15 s, 95°C), annealing, elongation, and acquisition (30 s, 60°C). A specific primer for AAV8-Di-C5 mRNA was designed and validated before the experiment, using both transduced cell culture samples and non-transduced tissue samples. Eef1a1 and Yhwaz were used as housekeeping genes.^47^^,^^48^ The primer efficiencies were ascertained through a seven-step dilution curve. The corresponding Ct value for each well was calculated using the on-board software (Roche) with a maximum cut-off of 35 cycles. For each independent experiment, samples were run in technical triplicates, and averaged Ct values were used for calculations.^70^(Equation 1)ΔCt = mean Ctgene of interest- mean Cthousekeeping genes(Equation 2)ΔΔCt = mean ΔCtDPBS~~group- mean ΔCtgene of interest
The following primer pairs were
Eef1a1 (forward): TGCTGGAGCCAAGTGCTA AT; Eef1a1 (reverse): GTGCCAATGCCGCCAATTTT.
Yhwaz (forward): GAAGCATTGGGGATCAAGAA; Yhwaz (reverse): AGACGGAAGGTGCTGAGAAA.
AAV8-Di-C5 (forward): ATCCGTATGATGTGCCGGATT; AAV8-Di-C5 (reverse): CGCCACTTCGTTTTCCAGATG.
C57BL/6N male mice (12 weeks old) were either i.t. injected with 7 μL AAV8-Di-C5 or left uninjected as controls. Three weeks after injection, the mice were overdosed with pentobarbital, and PBS perfused before the DRGs were dissected, flash frozen on dry ice and stored at −70°C. For each condition, all DRGs from two mice were pooled. A previously described protocol was used.^71^ In brief, the DRGs were lysed in 2 mL ice-cold working buffer HB (see buffer list below) and homogenized with a Tissue-Tearor (D1000 handheld Homogenizer, Merck) for 10 s at a low-speed setting. The supernatants were transferred to a pre-cooled 7 mL Dounce homogenizer together with 3 mL ice-cold working buffer HB and bounced for 10–12 strokes with a tight pestle. 320 μL 5% IGEPAL (18896, Sigma-Aldrich) (see buffer list) was added and bounced 5 additional strokes. The sample was filtered through a 40 μm strainer (22363547, Fisher brand) and 5 mL ice-cold working solution (see buffer list) was added to reach a full volume of 10 mL with 25% iodixanol (Opti prep, 07820, Stem cell technologies). Next, the sample was loaded onto an iodixanol density gradient consisting of a 30% layer on top of a 40% layer and spun for 25 min at 10,000 g at 4°C with a low acceleration and deceleration using an SW 28 Swing bucket rotor (SW28, Beckman Coulter). Following centrifugation, the nuclei situated in the interface between the 30% and 40% iodixanol layers were extracted. The extracted nuclei were washed in 10 mL PBS with 0.05% BSA (A1470-100G, Sigma Aldrich), spun at 500 g for 7 min, and resuspended in 50 μL PBS with 0.05% BSA.
Buffer
Diluent 150 mM KCl (Thermo Fisher Scientific, AM9640G), 30 mM MgCl2 (Thermo Fisher Scientific, AM9530G), 120 mM Tricine-KOH (Sigma, T0377) (Sigma, 221465) (pH 7.8).
Stock buffer HB: 0.25 M Sucrose (Millipore, 573113), 25 mM KCl, 5 mM MgCl2, 20 mM Tricine-KOH (pH 7.8), 5 mg/mL actinomycin (Sigma Aldrich, A9415).
Working solution 50% Iodixanol: 15 mL Opti-Prep (Iodixanol) (Stem cell technologies, 07820), 6 mL Diluent buffer.
30% Iodixanol: 2.4 mL Working solution 50% Iodixanol, 1.6 mL Stock Buffer HB, 16.8 μL 10% BSA (diluted in PBS), 6.4 μL RNase Inhibitor (Promega, N2611).
40% Iodixanol: 4.8 mL Working solution 50% Iodixanol, 1.2 mL Stock buffer HB, 25.2 μL 10% BSA (Sigma-Aldrich, A1470) (diluted in PBS), 9.6 μL RNase Inhibitor (Promega, N2611), 20 μL Phenol Red (Supplier NA).
Working buffer HB: 12 mL Stock Buffer HB, 50.4 μL 10% BSA, 18 μL RNase Inhibitor.
5% IGEPAL: 40 μL IGEPAL CA-630 (Sigma-Aldrich, 18896), 760 μL Working buffer HB.
Chromium Single Cell 3′ Reagent Kit v3.1 Dual Index (10x Genomics, PN-1000268) was used for library preparation according to a standard protocol. In brief, nuclei were counted on a hemocytometer under a microscope, mixed with reverse transcription mix, and partitioned together with v3 Gel Beads on Chromium Chip B (10x Genomics, PN-1000120) into Gel Beads-in-emulsion (GEMs) using Chromium Controller (10x Genomics, PN-120223). The aim was to get 10,000 nuclei per sample. Following reverse transcription, samples were frozen at −20°C, and within a week, samples were processed for complementary DNA (cDNA) cleanup and preamplification [12 polymerase chain reaction (PCR) cycles]. After SPRIselect (Beckman Coulter, B23318) cleanup, cDNA was quantified, fragmentated, and end-repaired. Fragments were cleaned up using SPRIselect reagent and processed through steps of adapter ligation, SPRIselect cleanup, and sample index PCR [using Chromium i5 and i7 Sample Indices (10x Genomics, PN-1000215) for 13 PCR cycles]. Following, libraries were cleaned up with SPRIselect reagent and quantified using the Qubit HS dsDNA Assay Kit (Thermo Fisher Scientific, Q32854) and Qubit Fluorometer and using the High Sensitivity D5000 Reagents (Agilent, 5067–5593) and Agilent 4150 TapeStation. Libraries were pooled according to the expected number of nuclei per sample; pool was quantified and sequenced using NovaSeq 6000 SP reagent Kit v1.5 (20028401, Illumina) on Illumina NovaSeq 6000 System (Illumina 20012850) controlled by NovaSeq Control Software. Libraries were sequenced using 28 cycles for read 1, 10 cycles for i7 index, 10 cycles for i5 index, and 90 cycles for read 2. In average, there were 48,037.5 reads and 1,446 genes per nucleus.
Primary data processing was performed using Cell Ranger (10x Genomics) to generate FASTQ files (cellranger mkfastq) and count matrices (cellranger count). A custom reference transcriptome including the transgene GCN4 was built following the 10x Genomics guidelines. The mouse GRCm39 release 115 FASTA and GTF files were downloaded from Ensembl and filtered to retain only protein-coding genes. The GCN4 sequence was appended to both files, and the final reference was generated using cellranger mkref command. Cell filtering was carried out using a locally adapted version of CRMetrics, which bundles standard preprocessing steps and provides visual assessment of filtered cells. CellBender v0.3.0^54^ was applied to remove ambient RNA counts and droplets containing primarily ambient RNA. Scrublet^55^ was used to calculate doublet scores, and only cells below the sample-specific threshold were retained. In addition, cells with a mitochondrial gene fraction >5% or fewer than 200 unique molecular identifiers (UMIs) were excluded. Normalization was performed using Pagoda2 followed by sample integration with Conos^56^ (alignment strength = 0.1) to achieve optimal cross-sample alignment. UMAP embedding was generated, and clustering was performed using the Leiden community detection algorithm. The dataset was then subdivided into neuronal and non-neuronal cells based on clusters expressing at least one neuronal marker gene (Cacna1b, Camk2a, Kcnd2, Tafa4). Each subset was again integrated, embedded, and clustered. Final clusters were generated by merging some of the estimated clusters. The cell types were annotated based on marker gene expression.^72^ Clusters named “unidentified” either had too few detected transcripts per nuclei or had unexpected marker gene combinations.
The raw single nucleus RNA sequencing data have been deposited to The Sequence Read Archive (SRA) under SRA study SRP645687 (SRA: SRX31117391, SRX31117392), and the processed data to Zenodo (Zenodo: 17629531).
Female WT mice (same mice as in Figure 3B) received AAV8-Di(7P14P)-C5, AAV8-tdTomato, or vehicle with an i.t. injection 22 days before SNI surgery. Mice were anesthetized with an overdose of pentobarbital before transcardial perfusion with PBS 49–50 days after SNI surgery. From each mouse, 0.5 cm of the sciatic nerve was dissected from both the ipsilateral (proximal from suture) and contralateral side. The tissue was placed in ice-cold F12 medium (#N6658, Sigma Aldrich) in a 96-well PCR plate (#AB0600, Thermo Fisher). The F12 medium was removed, and the tissue was enzymatically digested in 50 μL of 6.25 mg/mL collagenase type IA (#C9891-1G, Sigma Aldrich), 0.4% wt/vol hyaluronidase (#H3506-100MG, Merck) and 0.2% wt/vol pronase (#53702-10KU, Merck Millipore) diluted in F12 medium. After adding the enzymatical mix, the tissue was cut in smaller pieces and incubated on a shaker at 37°C for 45 min. Samples were centrifuged, and enzymatic mix was removed before resuspension in 200 μL warm mFACS buffer containing Hank’s balanced salt solution (HBSS) (#14175095, Gibco) with 0.4% bovine serum albumin (#A3983, Sigma Aldrich), 15 mM HEPES (#H0887-100ML, Merck), and 2 mM EDTA (#324506, Merck). Tissue was mechanically dissociated by pipetting up and down 50 times and filtered through a 70-μm Flowmi cell strainer (BAH136800070, Merck) into a labeled U-bottom plate. Samples were centrifuged, and mFACS buffer was removed before resuspending in 50 μL of Live/dead stain (L34992, Thermo Fisher) diluted 1,500 in HBSS and incubated on ice protected from light for 15 min. For the rest of the protocol, the samples were protected from excessive light. Samples were resuspended in 50 μL antibody mix (see list below) and incubated on ice for 30 min. The cells were fixed with 50 μL 4% PFA for 5 min on ice, before resuspension in 200 μL mFACS buffer and storage at 4°C until flow cytometric analysis. The samples were run on a LSRFortessa 5 laser system from BD. Data analysis and fluorescent compensation were performed with the software FlowJo. All gates were placed based on fluorescence-minus-one (FMO) controls, and the gating strategy is shown in Figures S7F–S7L.
Antibody
Brilliant Violet 510 anti-mouse CD11c (Biosite, 117337).
BUV395 Rat anti-mouse Ly6G, (BD biosciences, 563978).
Brilliant violet 650 anti-mouse Ly6C, (Biosite, 128049).
Brilliant violet 785 anti-mouse F/80 (Biosite, 123141).
FITC anti-mouse CD45, (Biosite, 103108).
PerCP/Cyanine5.5 anti-mouse I-A/I-E (Biosite, 107625) (anti-MHCII).
PE/Cyanine7 anti-mouse/human CD11b (Biosite, 101215).
PE anti-mouse TCR beta chain (Biosite, 109207).
APC anti-mouse CD19 (Biosite, 115511).
Alexa Fluor 700 anti-mouse NK-1.1 (Biosite, 108729).
Purified anti-mouse CD16/32 (Biosite, 101302) (blocking).
Male SNI c57BL6/J WT mice (same mice used in Figure 6A) were anesthetized, and PBS perfused 73–74 days after i.t. injection of 7 μL AAV8-Di-C5 or AAV8-tdTomato. Ipsilateral and contralateral L3-L5 DRGs were dissected and snap frozen on dry ice until MS sample preparation. For each mouse, ipsilateral L3-L5 DRGs were pooled, and contralateral L3-L5 DRGs were pooled.
Tissues were lysed by adding preheated (95°C) lysis buffer (5% SDS, 100 mM Tris, pH 8.5) and homogenized with a BeatBox homogenizer (PreOmics) for 5 min at standard setting. Samples were incubated at 95°C for 10 min and lysed with a second round of BeatBox for 5 min at standard setting, followed by incubation at 95°C for 10 min. Protein concentrations were determined using the BCA assay (Pierce). Samples were reduced with 5 mM TCEP (final conc.) for 15 min at 55°C and alkylated with 20 mM chloroacetamide (CAA) (final conc.) for 30 min at room temperature. 233 μL acetonitrile (ACN) and 25 μL of pre-equilibrated (70% ACN) MagReSyn Hydroxyl beads (Resyn Biosciences) were added to 100 μL protein extract (100 μg protein). The samples were mixed and settled for 10 min at room temperature. Beads were washed three times with 1 mL 95% ACN, twice with 1 mL 70% EtOH on a KingFisher Apex (Thermo Fisher) instrument in 96-well format and digested overnight at 37°C with 100 μL digestion buffer (50 mM TEAB) containing Trypsin and Lys-C in a 50 (enzyme: protein) ratio. Peptides were acidified with 1% (final conc.) trifluoroacetate (TFA) and desalted using EasyPep peptide clean-up plates (Thermo Fisher) following the manufacturer’s instructions.
Peptides were resuspended at a final concentration of 100 mM HEPES before labeling. TMTpro-16plex non-deuterated reagents (Thermo Fisher Scientific, A44520) were reconstituted in 100% anhydrous acetonitrile. Labeling was carried out for 1 h at room temperature using a 5 sample-to-label ratio. After the incubation, samples were acidified by the addition of 5% hydroxylamine 20% formic acid for 15 min at RT before being combined and desalted using Thermo-desalting plate as described above.
Dried peptides were resuspended in 20 μL of 0.1% TFA containing 0.01% N-dodecyl-β-d-maltoside (DDM). Phosphorylated peptides were enriched using Fe(III)-NTA cartridges (Agilent Technologies) in an automated manner on the AssayMAP Bravo Platform (Agilent Technologies), following the standard protocol. Briefly, cartridges were primed with 100 μL of priming buffer (100% acetonitrile, 0.1% TFA) and equilibrated with 50 μL of loading buffer (80% acetonitrile, 0.1% TFA). After sample loading, cartridges were washed with the loading buffer while syringes were rinsed with the priming buffer. Phosphopeptides were eluted using 20 μL of 500 mM ammonium dihydrogen phosphate (NH4H2PO4) containing 0.01% DDM. Both the enriched phosphopeptides and flowthroughs were dried by vacuum centrifugation and resuspended in buffer A∗ (2% acetonitrile, 0.1% TFA).
Labeled desalted peptides were resuspended in buffer A∗ and fractionated into 16 fractions by high-pH fractionation. For this, 10 μg peptides were loaded onto a Kinetex 2.6u EVO C18 100 Å 150 × 0.3 mm column via a VanquishNEO (Thermo Fisher Scientific) in buffer AF (10 mM TEAB), and separated with a non-linear gradient of 3–60% buffer BF (10mM TEAB, 80% acetonitrile) at a flow rate of 2.0 μL/min over 66 min. Fractions were collected using the OT-II Robot (Opentron) liquid handler every 60 s with a concatenation scheme to reach 16 final fractions (e.g., fraction 17 was collected together with fraction 1, fraction 18 together with fraction 2, and so on).
Peptides were separated on a C18 PepMap trapping column, 300 μm inner diameter (Thermo Scientific) using Vanquish Neo UHPLC system (Thermo Scientific), followed by separation on an Aurora Ultimate 25 × 75 C18 column (IonOpticks), Peptide separation was performed using a 120 min stepped gradient of 2–17% solvent B (0.1% formic acid in acetonitrile) for 71 min, 17–25% solvent B for 26 min, 25–35% solvent B for 13 min, using a constant flow rate of 400 nL/min. Column temperature was controlled at 50°C. Upon elution, peptides were injected through a nano-electrospray source into an Orbitrap Ascend Tribrid mass spectrometer (Thermo Fisher Scientific). Samples were acquired using a Real Time Search (RTS) MS3 data acquisition where the Tribrid mass spectrometer was switching between a full scan (120K resolution, 50 ms max. injection time, AGC target 100%) in the Orbitrap analyzer, to a data-dependent MS/MS scans in the Ion Trap analyzer (Turbo scan rate, 23 ms max. injection time, AGC target 100% and CID activation type). Isolation window was set to 0.7 (m/z), and normalized collision energy to 34. Precursors were filtered by charge state of 2–5 and multiple sequencing of peptides was minimized by excluding the selected peptide candidates for 60 s. MS/MS spectra were searched in real time on the instrument control computer using the Comet search engine (UniProtKB #UP000000589 Mus musculus FASTA file), 1 max miss cleavage, carbamidomethylation on cysteine and TMTpro16plex on lysine and N-terminus were set as fixed modification, oxidation on methionine as variable modification. For post-translational modification (PTM) analysis, STY-phosphorylation was added, and 35 ms max search time with an Xcorr scoring threshold of 1.4 and 20 precursor ppm error. For phosphorylation, multistage activation was enabled with a 97.9763 neutral loss mass. MS/MS spectra resulting in a positive RTS identification were further analyzed in MS3 mode using the Orbitrap analyzer (90k resolution, 187 ms max. injection time, AGC target 300%, HCD collision energy 55 and SPS = 10). The total fixed cycle time, switching between all 3 MS scan types, was set to 3 s.
MS raw files were processed using Thermo Proteome Discoverer version 3.2.0.450. Briefly, UniprotKB UP000000589 (Taxa ID: 10090) Mus musculus FASTA database was used. Carbaminomethylation of cysteine was set as a fixed modification, whereas the oxidation of methionine and TMTpro labels on lysine and the protein N-terminus were set as variable modifications. For PTM analysis, phosphorylation of tyrosine, serine, and threonine was added as variable modifications. The maximum number of missed cleavages was 2, and the minimum and maximum peptide lengths were 7 and 30 amino acids. Sequest HT database search and INFERYS rescoring algorithm were used for peptides sequencing, and reporter ion intensities were extracted with a most confident centroid integration method with a 20 ppm integration tolerance, using a custom quant method that specified TMTpro reporter ion isotope distributions (leakage). The target false discovery rate (FDR) was set to 0.01. The PTM localization was activated, and the site probability threshold was set to 75.
All statistical analysis was performed using the Proteomic Research Infrastructure (PRI)-developed Python code (Proteomelit, version 1.1.0), based on the automated analysis pipeline of the Clinical Knowledge Graph.^52^ Intensity values were log2-transformed, and features with less than 70% valid values in at least one group were removed. For the phosphoproteome dataset, statistical and enrichment analyses were performed on phosphor-modified peptides exclusively. Missing values were imputed using the mixed imputation approach. In the mixed imputation approach, remaining missing values were replaced by mixed imputation, where kNN and MinProb^73^ (width = 0.3 and shift = 1.8) methods were used for values missing at random (MAR) and values missing not at random (MNAR), respectively. For normalised proteome results, modified peptide intensities were normalised to the protein abundance from the non-enriched global proteome. Differentially expressed features were identified by statistical unpaired t-tests, and permutation-based FDR correction for multiple hypothesis testing.^74^ An FDR threshold of 0.01 was applied for the global proteome and 0.05 for the phosphoproteome, with an absolute fold-change cutoff of 1, an s0 value of 1, and 250 permutations. Volcano plots display all features statistically tested, with up- and down-regulated features highlighted in red and blue, respectively. Volcano plots have been generated with the online tool VolcaNoseR.^53^
Functional enrichment analysis of significant features was performed by applying Fisher’s exact test (p < 0.01) on significant proteins as defined above. Gene ontology annotations were retrieved from UniProt^75^ and used for the enrichment testing. The analysis was performed at the protein level for both the global proteome and phosphoproteome datasets. Up- and down-regulated features were used as foreground and the remaining features in the dataset were considered background. Kinase-substrate and linear motif enrichments were performed using the same strategy (p < 0.05). Annotation data was obtained from PhosphoSitePlus^76^ and Phosida,^77^ and phosphorylated protein sites mapped to significantly regulated modified peptides were used as foreground.
The MS proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE repository, for the global proteome (PRIDE: PXD074049) and for the phosphorylation enrichment (PRIDE: PXD073925) (https://www.ebi.ac.uk/pride/archive). Each deposit datasets comprise two technical replicates, using either deuterated or non-deuterated labels. After confirming replication consistency, only samples with non-deuterated labels were retained for further analysis.
Mice were briefly anesthetized with 2% isoflurane (max. 60 s) and injected in the plantar surface of the right hind paw with 50 μL (5 μL in left hindpaw for Figure S2B) undiluted Complete Freund’s Adjuvant (CFA) (F5881, Sigma) with a single intraplantar (i.pl.) injection using an insulin needle (0.3 mL BD Micro-Fine). Analgesic treatment was not applied to CFA-injected animals. For CFA experiments in Figures 1C, S2A, and S2C, 3 mg/kg naltrexone (NTX) was subcutaneously injected. In Figure S2B, mice were intraperitoneally injected with 3mg/kg NTX.
Briefly, mice were deeply anesthetized with 2% isoflurane, and a small incision was made in the skin of the left hind leg between the thigh bone and knee joint. Using blunt dissection, the muscles were separated to expose the sciatic trifurcation point. The tibial and common peroneal nerve branches were ligated with a surgical knot, and a small piece of the nerve was cut out distally to the ligation which was left untouched. The muscles were tucked into place, the skin was closed with glue and/or sutures, and the health status of the mice was carefully monitored, especially in the immediate days following the surgery. Sham surgeries were performed similarly, exposing the sciatic trifurcation point, but without ligation of the nerve branches. Differences in surgical and analgesic procedures between study sites are highlighted
University of Copenhagen: The tibial and peroneal nerve branches were ligated with a single surgical knot (nonabsorbable polypropylene 6-0 sutures) (8660H, Ethicon). The skin was closed with glue (Tissue adhesive, Klinibond) (115 500, Klinion). Buprenorphine (Temgesic; 0.3 mg/mL, RB Pharmaceuticals, diluted in isotonic saline from Fresenius Kabi) was administered s.c. 5 min before surgery. A lidocaine/bupivacaine mixture (2.5 mg/mL/1.25 mg/mL) was topically applied to muscle layer during surgery. Rimadyl (Carprofen, 1% from 0.5 mg/mL stock) was injected s.c. after surgery. Buprenorhpine and Rimadyl treatment was repeated once daily for another 2 days.
Aarhus University: The tibial and peroneal nerve branches were ligated with a single surgical knot (nonabsorbable polypropylene 6-0 sutures) (8660H, Ethicon). The skin was closed with glue (Tissue adhesive, Klinibond) (115 500, Klinion). Lidocaine SAD (10 mg/mL; Amgros I/S) was applied once to the skin. Buprenorphine (Temgesic, 0.3 mg/mL; RB Pharmaceuticals) and ampicillin (Pentrexyl, 250 mg/mL; Bristol-Myers Squibb) were mixed and diluted 10 in isotonic saline (9 mg/mL; Fresenius Kabi), and 0.1 mL was injected s.c. following the surgery. Buprenorphine and Pentrexyl treatment were repeated once daily for another 2 days, and always given after behavioral assessments, if performed on Day 2.
Heidelberg University: The tibial and peroneal nerve branches were ligated with a single surgical knot (absorbable 5-0 sutures) (17241041, Catgut GmbH). The skin was closed with several surgical knots (absorbable 5-0 sutures) (17241041, Catgut GmbH) followed by glue (1050052, B. Braun Surgical). Only isoflurane anesthesia was used.
AAVs were administered either by intrathecal (i.t.) or intravenous (i.v.) injection. For i.t. injections, mice were anesthetized with 2% isoflurane (max. 2 min). Using a hand-held Hamilton syringe (30G, 22mm needle) inserted at the L5/L6 intervertebral space, 7 μL AAV (2.2–3.3 × 10^12^ vg/mL; total viral load 1.5–2.3 × 10^10^ vg/mouse) was injected once a Straub tail response was elicited. For i.v. injections, mice were placed in a warming chamber to allow the tail vein to swell, and 50 μL (PHP.S, 2.0 × 10^13^ vg/mL, total viral load 1.0 × 10^12^ vg/mouse; PHP.eB, 2.0 × 10^12^ vg/mL, total viral load 1.0 × 10^11^ vg/mouse) was injected using an insulin syringe.
Behavioral testing and following data analysis were performed blinded to experimental conditions, e.g., without knowing which AAVs were used. In general, prior to experimental onset, mice were allowed to habituate to the room for a minimum of 30 min or until mice were calm. Individual experiments were performed by the same female experimenter throughout the duration of the experiment.
Von Frey was performed at three different study sites. In common, all sites used the ascending von Frey method with manual von Frey filaments (Bioseb) ranging from 0.04 to 2.0g (0.04, 0.07, 0.16, 0.4, 0.6, 1.0, 1.4, 2.0g). Each filament was applied to the lateral plantar surface of each hind paw 5 times. Positive responses were used to determine the paw withdrawal threshold (PWT).
University of Copenhagen: Mice were placed on an elevated wire mesh in 8 (Ø) x 7.5 (h) cm red, transparent, plastic cylinders or in 11.5 (w) x 14 (d) cm PVC plastic boxes and allowed to habituate for a minimum of 20 min or until calm before the initiation of the experiment. Each filament was applied 5 times to the hind paw of the same mouse over approx. 30–60 s or longer resting periods if required. The paw withdrawal threshold was determined as 3 positive responses out of 5 for two consecutive filaments. A positive response was defined as a sudden paw withdrawal, flinching, and/or paw licking induced by the filament.
Aarhus University: Mice were placed on an elevated wire mesh in 8 (Ø) x 7.5 (h) cm red, transparent, plastic cylinders and allowed to habituate for a minimum of 15 min before the initiation of the experiment. Each filament was applied 5 times to the hind paw of the same mouse over 30 s (each stimulus lasting approx. 2 s), followed by application of the filament to the contralateral hind paw. The paw withdrawal threshold was determined as 3 positive responses out of 5 for the same filament. A positive response was defined as a sudden paw withdrawal, sudden flinching, or sudden paw licking.
University of Heidelberg: Mice were placed on an elevated wire mesh in 9.5 (w) x 9.5 (d) x 13.5 (h) cm plastic boxes and allowed to habituate for a minimum of 30 min or until calm before the initiation of the experiment. Each filament was applied to each hind paw of each mouse over 5 rounds, giving a total of 5 applications to each hind paw with a minimum of 5 min resting period between each application of the same filament. Between different filaments, a resting period of minimum 5 min resting period was used between each filament size. An additional resting period of 5 min was used between each filament size. The paw withdrawal threshold was determined as 3 positive responses out of 5 for two consecutive filaments. A positive response was defined as a sudden paw withdrawal, flinching, and/or paw licking induced by the filament.
Mice were placed on an elevated metal grid in 9.5 (w) x 9.5 (d) × 13.5 (h) cm chambers with lids, and left to habituate for 30 min. A soft, round, size 3 brush was applied in one dynamic motion in a heel-to-toe direction, and the paw withdrawal reaction was scored as 0 = no response, 1 = brief lifting, 2 = paw lift and guarding, 3 = paw flinch, 4 = paw flick or multiple paw flinches, 5 = paw licking, or flick and guarding. The brush was applied a total of 3 times for each paw. Data are shown as the average score of 3 applications.
Mice were placed on an elevated metal grid in 9.5 (w) x 9.5 (d) × 13.5 (h) cm chambers with lids, and left to habituate for 30 min. A single drop of acetone was applied to the ipsilateral paw with a syringe with a pipette tip. A camera recorded the behavior for 1 min, and the videos were manually assessed for the duration of the nocifensive behavior of the ipsilateral paw, including paw flick, licking, and guarding.
The mice were placed on a 2°C cold plate (Bioseb), and the latency for the first paw withdrawal was recorded. A paw withdrawal was defined as a sudden paw withdrawal, flinching, and/or paw licking. A cutoff of 30 s was used, and all mice without a paw withdrawal at the end of the experiment (30 s) got the value 30 s.
For the dynamic hotplate (Bioseb), the mice were placed on a hotplate with a temperature ramp starting at 30°C (±0.1) and rising at 5°C/min increments until the first nocifensive response (paw withdrawal, flick, or lick). A cutoff temperature was set to 50°C. The data show the temperature at which the nocifensive response occurred.
Mice were placed in a 19 (w) x 19 (d) x 13.5 (h) cm box with 9.5 (w) x 9.5 (d) cm chambers. The box was placed on a transparent surface, allowing a video camera to record the mice from below. Mice were recorded for 15 min. The videos were manually assessed for ipsilateral paw lifting in absence of exploratory behavior, including rearing, walking, grooming, and coordinated movements of other paws. Only paw lifts, when the mouse was standing still and no other paws were coordinately lifted, were counted.
Mice were placed in individual cages containing a wheel with free access to food and water. The number of wheel rotations was collected with AWM counter and AWM software (Lafayette Instruments, Louisiana, USA), using optical sensors to detect revolutions. 1 rotation equals 0.4 meters. The mice entered the cages during the light phase for habituation, and data were collected in 1-h intervals during the dark phase overnight (18:00-06:00). The data represent 3 technical replicates.
Mice were placed in individual Laboras-specified cages (Metris B.V., Netherlands) placed on a carbon fiber platform that detects a range of behaviors based on the vibration pattern, which was collected with the Laboras 2.6 software. Behaviors assessed consist of frequency and time courses of grooming, rearing, climbing, locomotion, and immobility. Mice were assessed over 22 h and had free access to food and water.
Mice were placed in the center of square white open arenas (40 × 40 × 40 cm) and monitored for 120 min. Open-field locomotion was recorded and analyzed (distance and position in the arena) using Ethovision video-tracking software (Noldus), with distance traveled further divided into 5 min bins.
Mice were placed on the CatWalk XT (Noldus) and allowed 3 runs with the following average run speed <30 cm/s and run variation <100. The data shown represents an average of 3 runs. The post-run analysis included the following run duration (s), average speed (cm/s), the maximum contact area of the ipsilateral hind paw (LH_MaxContactArea) (cm2), swing speed of the ipsilateral hind paw (LH_SwingSpeed) (cm/s). CatWalk settings during Camera Gain = 20.7 dB, Green Intensity Threshold = 0.12, Red Ceiling Light = 17.7 v, Green Walkway Light = 16.5 v.
RotaRod was performed over 3 days. On day 1, the mice were trained on the RotaRod. Once 3 min at 5 rpm was completed for each mouse, a 3-day accelerating paradigm was started. Over 5 min, the speed was increased from 4 to 40 rpm (1 rpm/8 s). 3 trials per mouse were performed on day 1–3. On day 3 following the accelerating paradigm, the mice were tested with a fixed speed paradigm. The mice were placed on the rod at 5, 10, 15, 20, 25, 30, 35, and 40 rpm for 5 min each time, or until the mouse fell off 3 times at the same speed. An inter-trial break of at least 10 min was used between each speed.
Behavioral experiments were performed by female experimenters at the University of Copenhagen (Denmark), Aarhus University (Denmark), and Heidelberg University (Germany). The list specifies the study site for each
Figures 1C + 1F: University of Copenhagen.
Figure 3A, 3C and 3D: Aarhus University.
Figure 3B + 3I: University of Copenhagen.
Figure 3E–3H +3J Heidelberg University.
Figure 4A: Heidelberg University.
Figure 4B–4F: University of Copenhagen.
Figure 5A: University of Copenhagen.
Figure 6A: University of Copenhagen.
Figures S2A–S2C: University of Copenhagen.
Figures S4C and S4D: University of Copenhagen.
Figure S5D: University of Copenhagen.
Figures S6B–S6Q: Heidelberg University.
Figures S7B and S7C: University of Copenhagen.
Statistical details are reported in the figure legends and were computed with GraphPad Prism 9 and 10. One-way ANOVA was used to compare group means between more than two groups, followed by either Dunnett’s multiple comparisons, where groups were compared with controls, or Tukey’s multiple comparisons, where means of every two groups were compared. Two-way ANOVA, or mixed-effect analysis, was used to compare group means when there were two or more variables, followed by either Tukey’s, Holm-Šidák’s, or Dunnett’s multiple comparisons (as noted in the figure legends). Depending on the experimental design, either two-way ANOVA or two-way repeated measures ANOVA was used. One-sample t test was used to compare group means. Survival probability was determined by Log rank (Mantel-Cox) test, whereas cumulative distribution was determined by Kolmogorov-Smirnov (KS) test. All comparisons are two-sided. Statistical significance was determined as ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, as indicated in the figures. Results show mean ± SEM, if not otherwise stated. No estimate of power was performed before experiments, but sample sizes were similar to those generally employed in the field.
Data analysis and statistical methods specific for snRNAseq and MS can be found under their respective STAR Methods section.