In silico design of 1.7iPAs from NaV1.7-IDRs

The candidate iPAs were designed through an a priori strategy aimed at defining the short linear functional disordered peptides from the intrinsically disordered domains (IDDs) (12), initially from NaV1.7 protein IDRs, on the hypothesis that NaV1.7 IDDs contain the functional sequences that modulate NaV1.7 channel function. We analyzed the full length of the rat NaV1.7 protein sequence using DisorderEd PredictIon CenTER (DEPICTER), which combines 10 popular algorithms for IDR predictions within the primary sequence based on amino acid (aa) biophysical features for the protein’s disordered ensemble (18). Results return a score between 0 and 1 for each residue, indicating the degree to which a given residue is part of an ordered or disordered region (residues with scores over 0.5 are considered as disordered). Results revealed clear order-to-disorder transitions where NaV1.7 transmembrane (TM) domains and intracellular portions join, and scores indicate a disordered nature of NaV1.7 intracellular and terminal regions (Figure 1, A–C). Specifically, the most extensive IDRs are in the intracellular loops (ICL), while protein TM domains are highly ordered.

In silico prediction of NaV1.7 IDRs and design of candidate NaV1.7iPAs. (A) Diagram of rat NaV1.7 protein, with white boxes labeling DI-DIV of NaV1.7 and (B) the red bars below showing position of the predicted iPAs. (C) Consensus prediction of IDRs by DEPICTER. (D) The phosphorylation sites were predicted by DEPP. (E) Nine candidate iPAs with their aa sequences and position in NaV1.7, and IDR scores. (F) A map showing each component of an AAV plasmid coding GFP-iPA with a black line pointing to iPAs. (G) The structure analysis of GFP-fused 1.7iPA1 by I-TASSER. The top image shows structure of free 1.7iPA1. (H) Images (GFP, left; phase, middle; and merged pictures, right) show expression of constructs carrying 1.7iPA1–4 and 6 after transfection to HEK cells. Scale bar: 25 μm. (I) GFP and Gapdh Western blots of the cell lysates after transfection with 1.7iPA1-4 and 6 to HEK cells. (J) Initial screening of 9 iPAs on INa by whole-cell patch-clamp recording as described in methods after transfection into HEK1.7 cells. *P < 0.05 and ***P < 0.001; 1-way ANOVA and Tukey’s post hoc.

Potential phosphorylation sites in the NaV1.7 sequence were identified using Disorder Enhanced Phosphorylation Predictor (DEPP) (19). Results showed that most potential phosphorylation residues (serine, threonine, and tyrosine with high DEPP scores) reside in NaV1.7-IDRs, particularly in the IDRs within the ICL1 and ICL2 (Figure 1D). NaV1.7-IDRs feature as potential protein-protein interaction (PPI) binding sites, suggesting these IDRs could contain key binding motifs or domains of the NaV1.7 regulatory signaling interactome (20). These observations predict that focusing on the NaV1.7-IDRs could be an avenue for identifying short peptides effective in modulating NaV1.7 channel function.

The potentially functional domains within the NaV1.7-IDRs (21) were further analyzed using SLiMPrints (22), which predict short linear motifs (SLiMs) based on strongly conserved primary aa sequences followed by filtering based on the prediction scores (22). The enumerated motifs predicted within NaV1.7-IDRs suggest many possible functional peptides as hot spots of functional IDDs, including proteolytic cleavage sites, ligand binding sites, posttranslational modification (PTM) sites, and subcellular targeting sites. Nine peptides were designed computationally based on IDR scores and phosphorylation sites and were the focus as 1.7iPA candidates for further testing (Figure 1, E and B).

Constructs of 1.7iPAs and transfection expression

AAV expression plasmids containing transgene expression cassettes encoding various GFP-1.7iPA chimeras were constructed. Specifically, the sequences for interchangeable iPA peptides were cloned with a linker sequence (GLRSRAQASNSAVDGTAGPGS), which we have described previously (23), to form a chimeric transgene in a GFP-linker-iPA orientation transcribed by a hybrid human cytomegalovirus (CMV) enhancer/chicken β-actin (CBA) promoter. This generated pAAV-CBA-GFP-1.7iPAs (pAAV-1.7iPA) expression plasmids in which the oligonucleotide encoding the interchangeable 1.7iPAs are inserted at the 3′ end of GFP (Figure 1F). The predicted protein structure analysis of GFP1.7iPA1 by I-TASSER tool (24) shows an unfolded and extended, highly flexible structural ensemble of linker-1.7iPA1 (Figure 1G) that is compatible with a well-exposed mode for potential binding to targets. Similar structures were also identified by I-TASSER for other GFP1.7iPAs (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI170813DS1).

Inhibition of NaV1.7 current in HEK1.7 cells by 1.7iPAs

The stable expression of each construct was verified by transfection into HEK293 cells stably expressing human WT NaV1.7 (HEK1.7 cells), followed by immunoblots (IBs). Representative tests for GFPlinker (GFP), 1.7iPAs (1, 2, 3, 4, 6) were shown (Figure 1, H and I). Initial screening experiments by whole-cell voltage clamp of inhibition of NaV1.7 current (INa1.7) in HEK1.7 cells transfected with plasmids encoding 9 1.7iPAs (1.7iPA1-9) were performed to characterize the INa1.7. The presence of 9 different 1.7iPAs in HEK1.7 cells on peak INa1.7 density (3 days after transfection) was summarized in Figure 1F, in which the data points recorded by at least 2 replicates were combined (Figure 1J). The results showed that 1.7iPA1, 4, and 6 produced approximately 68%, 59%, and 54% reduction of peak INa1.7 density, respectively, while 1.7iPA2 increased peak INa1.7 density (~35%). Transfection with plasmids expressing the GFPlinker and 1.7iPA3, 5, 7, 8, and 9 showed no significant effects on peak INa1.7 density, compared with sham-transfected HEK1.7 cells, which were transfected with PEI but no plasmid. These experiments thus identified 1.7iPA1 and 1.7iPA4, both of which were derived from ICL1, as well as 1.7iPA6, derived from ICL2, as effective iPAs (over 50% INa1.7 inhibition). We next focused on the validation of INa1.7 inhibition and channel kinetics by 1.7iPA1, 4, and 6 on HEK1.7 cells in new experiments. These results replicated the prior screening testing results of peak INa1.7 densities and showed that the steady-state activation and fast inactivation kinetics of NaV1.7 channels were not significantly affected in the presence of 1.7iPA1, 4, and 6 (Figure 2, A–E). The 1.7iPA1 peptide is polyampholytic, enriched with 38.6 % positively charged arginine or lysine (17 of 44), 22.7% of serine (10 of 44), and 18.1% acidic residues (8 of 44) and is highly conserved between rodents and humans (Figure 2F). Searching databases revealed that 2 serine phosphorylation and 2 lysine acetylation sites were assigned in high throughput (proteomic discovery mass spectrometry) studies (25) and a nuclear localization signal was predicted by SeqNLS (26). These analyses strongly suggest that 1.7iPA1 is a functional IDD peptide. Since 1.7iPA1 revealed higher inhibition of INa1.7 and was highly homologous to other TTXs NaV subtypes (see further), we selected it as a prototype and named NaViPA1 for further ‘hit to lead’ characterization.

Confirmation of INa1.7 inhibition by 1.7iPA1, 4, and 6 and gating kinetics. (A) Representative traces of INa1.7 by whole-cell patch-clamp recording from naive (transfection without plasmid), GFP, 1.7iPA3 (NP), 1.7iPA1, 1.7iPA4, and 1.7iPA6-transfected HEK1.7 cells. Inserts: recording protocol and current/time scales. (B) Summary of the confirmation tests of candidate iPAs expression in HEK1.7 cells (C) in comparison with corresponding mean peak current density-voltage (I/V) relationship from different constructs, as indicated and quantitative analysis of averaged peak INa1.7 density; **P < 0.01, ***P < 0.001, 1-way ANOVA and Tukey’s post hoc. (D)No effects of expression of GFPiPA1, GFPiPA4, and GFPiPA6 were observed on steady-state activation (inset: V1/2 activation) and (E) fast inactivation (inset: V1/2 inactivation), compared with naive and GFP or NP-transfected HEK1.7 cells. (F) NaViPA1 is highly conserved in rat, mouse, and human. Black and yellow asterisks at the bottom denote positively and negatively charged aa; the red and blue asterisks on the top denote known lysine acetylation and serine phosphorylation sites, and IDR scores and percent of positively (+) and negatively (–) charged aa were shown at the right sides of the alignment.

Specificity of NaViPA1 occupancy to various voltage-gated ion channels

Development of NaV1.8 stable expression system based on HEK cells. To assess the potential of NaViPA1 in affecting INa conducted by NaV1.8 channels, we developed stable expression of recombinant human NaV1.8 heterologous systems based on HEK cells (HEK1.8). Stable NaV1.8 expression was confirmed by IBs of NaV1.8α and Naβ2 in the cells after at least 10–20 rounds of G418 selection (400–800 μg/mL), followed by single-cell isolation using BIOCHIPS Single-cell Isolation Chip (Thermo Fisher Scientific). Both NaV1.8α and Naβ2 were found to be highly expressed in the cell membrane. Functional NaV1.8 expression was identified by the presence of slowly inactivating inward INa elicited by voltage steps from –140 mV to +80 mV during the whole-cell voltage-clamp recordings, and the averaged peak INa1.8 density in approximately 85% of the HEK1.8 was greater than 0.5 nA; INa1.8 was sensitive to a NaV1.8 channel blocker, A803467 (Alomone) and resistant to high concentration of TTX (5 μM, Tocris Bioscience). We used this HEK1.8 cell line for the initial screening tests of the NaViPA1 on INa1.8. In comparison, INa1.8 amplitudes in CHO-Nav1.8 cells were generally less than 100 pA, which was insufficient for our experimental needs (Supplemental Figure 2).

Selectivity of NaViPA1 on ion channel occupancy. NaV subtype stable cell lines based on HEK cells used for this experiment included HEK1.1, 1.3, 1.6, 1.5, and 1.8. Sequence alignments identified high-level homology of NaViPA1 with the corresponding sequences of TTXs NaV1.1, 1.3, and 1.6, but much less homologous to TTXr NaV1.5, 1.8, and 1.9 (Figure 3, A and B). Expression of NaViPA1 (fused to GFP) resulted in a significant block of INa conducted by fast-activating and inactivating NaV1.1, NaV1.3, and 1.6 (Figure 3, C–E). No effects on INa1.5 and INa1.8 were observed in the presence of NaViPA1 in the HEK1.5 and HEK1.8 cells (Figure 3, F and G) or in ND7/23 cells transiently transfected with Nav1.8 (Supplemental Figure 3). We did not test NaViPA1 against NaV1.9 channels as the expression cell line is unavailable; however, INa1.9 inhibition by 1.7iPA1 is not expected since there is no sequence homology of NaViPA1 to NaV1.9. The negative effects of NaViPA1 on potassium current (BK IKv) were found in NG108-15 cells, which naturally express potassium channels (12), and no effects on high-voltage activated (HVA) ICa were recorded on AAV-mediated NaViPA1 expression in DRG-PSNs. Potent INa1.7 inhibition by NaViPA1 was also confirmed in neuronal NG108-15 cells and F11 DRG-neuronal-like cells that naturally express NaV1.7. These experiments showed no pleiotropic effects of NaViPA1 on either BK potassium channels or HVA ICa (Supplemental Figure 4).

Sodium channel specificity of NaViPA1 (1.7iPA1) inhibition. (A) The aa sequence alignment of 1.7iPA1 with the corresponding sequences of TTXs NaV1.6, NaV1.3, NaV1.1, (B) as well as TTXr NaV1.5, NaV1.8, and NaV1.9 of rat proteins. The homologous aa (identity and similarity) was highlighted in heavy or light black shadows and percent of identical or similar aa shown at the right sides of the alignments. (C–G) Panels from left to right show the comparisons of INa traces in presence of 1.7iPA1 in HEK1.1, 1.3, 1.6, 1.5, and 1.8 cells (insert: pulse protocol and scale); peak INa density (**P < 0.01 and ***P < 0.001, 1-way ANOVA and Turkey’s post hoc), I/V curves, steady-state activation (insert: V1/2 activation) and fast inactivation kinetics (insert: V1/2 inactivation).

AAV6-mediated NaViPA1 expression in DRG-PSNs inhibits TTXs INa but not TTXr INa. Because no heterologous system or cell lines can fully mimic the in vivo conditions of sensory neurons, we further tested the functional inhibition of INa by NaViPA1 in DRG-PSNs. AAV6 vectors encoding GFP-fused NaViPA1 were generated and injected into lumbar vertebrae (L) 4 and/or 5 DRG of naive male rats, and acutely dissociated sensory neurons from DRG were tested 4 weeks after injection. AAV6 encoding GFPlinker and NP (1.7iPA3), which was derived from the N-terminus of NaV1.7 (Figure 1) and showed no effect on INa after being transfected into HEK1.7 (Figures 1 and 2), were used as the control. A voltage protocol was adopted that demonstrates successful separation of TTXr INa (Nav1.8-like) and TTXs INa in dissociated DRG neurons (27, 28), comparable to the recordings after addition of TTX (1.0 μM) in bath solution (Supplemental Figure 5, A and B). Whole-cell voltage-clamp recordings by the voltage protocol from small/medium-sized PSNs (less than 35 μm) showed that AAV-mediated expression of NaViPA1 produced significant inhibition of total and TTXs INa whereas it produced no significant inhibition on TTXr INa (Figure 4, A–C).

NaViPA1 on INa of rat DRG neurons (male) and hiPSC-SNs (female). (A–C) Panels from top to bottom illustrate representative traces and averaged peak INa densities of total INa (A), TTXs INa (B), and TTXr INa (C) recorded from sensory neurons (diameter < 35 μm) dissociated from naive male rats subjected to (panels from left to right) sham (surgical exposure without injection), and 4wk after L4/L5 DRG injected with AAV6-encoded GFP, GFPNP, and GFPNaViPA1. Inserts: representative PSN images (scale bars: 25 μm) of each group, current/time scales, and recording pulse protocol. (D and E) Representative montage ICC images illustrate hiPSC-SNs at DIV25 after transduction with LV-GFPNP (D) and LV-GFPNaViPA1 (E) at equal MOI = 5. Panels F–H illustrate representative traces and averaged peak INa densities of total INa (F), TTXs INa (G), and TTXr INa (H) recorded from hiPSC-SNs (DIV 25) of sham, expressing NP, and NaViPA1. Inserts: current/time scales and recording pulse protocol. *P < 0.05, **P < 0.01, and ***P < 0.001, 1-way ANOVA and Turkey’s post hoc.

Inhibition of TTXs INa by NaViPA1 in human iPSC-derived sensory neurons. To study the relevance of our findings in a human context, we used human induced pluripotent stem cell–derived (hiPSC-derived) sensory neurons (hiPSC-SNs, female, Anatomic) (29) to test whether inhibition of TTXs INa by NaViPA1 represents a meaningful and quantitative index of the functional lead in human sensory neurons. This also allowed examination NaViPA1 without potential overexpression effects in HEK-NaV cells. The hiPSC-SNs were differentiated to small-sized PSN morphology with a soma diameter around 20–25 μm and developed extensive neurites after 4–7 days in vitro (DIV) in differentiation cultures, indicating that these cells were sufficiently committed to the neuronal lineage. We used lentivector (LV) GFP (Supplemental Figure 6) to test hiPSC-SN transduction efficiency. We have succeeded in expressing NaViPA1 and 1.7NP (control) in the differentiated hiPSC-SNs by LV transduction at multiple of infection (MOI) equaling 5 (Figure 4, D and E). Electrophysiological recordings were performed on the hiPSC-SNs (DIV 25) with TTX (1 μΜ) in the bath solution, and TTXr/TTXs INa were separated by a subtraction protocol (27). To prevent the TTX effect, a voltage manipulation similar to DRG neuron recording was used. Additionally, a protocol was adopted to isolate somatic INa by a brief prepulse to voltage (40 mV) near to a spike inactivating voltage for hiPSC-SN axonal spikes, but not for somatic spikes (30). Results showed that NaViPA1 significantly inhibited TTXs INa but not TTXr INa in differentiated hiPSC-SNs (DIV 25) (Figure 4, F–H), comparable to rat DRG-PSNs. No effects were observed for BK IKv and HAV ICa recorded (DIV 21) from hiPSC-SNs in the presence of NaViPA1 (Supplemental Figure 7). Results indicate that inhibitory efficacy of NaViPA1 on TTXs INa defined in cell lines and rat DRG-PSNs are translatable to human PSNs.

Initial testing of molecular mechanisms of NaViPA1

We first validated the specificity of NaV1.7 antibody by IB using the cell lysates prepared from naive HEK cells, stable cell lines expressing different NaV isoforms, and 50B11 rat DRG neuronal cells. This NaV1.7 antibody (Alomone, ASC-008) was raised by an antigenic peptide corresponding to amino acid residues 446–460 of rat NaV1.7 and no significant sequence homologous with other NaV isoforms. Results showed that the Nav1.7 antibody detected full-length NaV1.7 only in HEK1.7 cells, but not other NaV isoforms and 50B11 cells that naturally do not express physical and functional NaV1.7 (31) (Figure 5A). By IHC on rat tissue sections, NaV1.7 expression was detected with high immunoreactive density in small/medium-sized PSNs using the NaV1.7 antibody, and NaV1.7 was also detected in spinal cord dorsal horn (SDH), sciatic nerve, and cutaneous terminals in hindpaws (Figure 5, B–E), with the patterns similar to the prior report (32). These results confirmed the specificity of the NaV1.7 antibody to detect NaV1.7 expression by IHC and IB.

NaViPA1 binds to full-length NaV1.7 protein and phosphoinositides. (A) IBs show selectivity of NaV1.7 antibody using cell lysates from naive HEK cells, HEK1.5, HEK1.7, HEK1.6, HEK1.1, HEK1.3, HEK1.8 cells, and 50B11 cells. (B–E) Representative IHC images show NaV1.7 detection (red) in SDH (red), sciatic nerve (green), DRG neurons (red), and cutaneous nerve fibers (red). Scale bars: 100 μm. (F) IBs of NaV1.7, GFP, NKA1α, and Gapdh in the cytosol and membrane samples extracted from HEK1.7 cells transfected with sham (transfection without plasmid), GFP, GFPNaViPA1, and GFP1.7iPA2. A vertical white line in GFP panel denotes that the lanes were run on the same gel but were noncontiguous. (G) NaV1.7 IB (left) and silver stain (right) of inputs (cell lysates, 20 μg for each lane) and pulldown beads (10 μL for each lane) prepared by a nondenaturing lysis buffer from HEK1.7 cells transfected with GFP, GFPNaViPA1, and GFP1.7iPA2. (G, right) Stained gel pieces ranging 100–300 kDa (G, red asterisk denotes NaV1.7 site) from GFPNP and GFPNaViPA1 excised for mass spectrometry. (H) Silver stain on 1D SDS-PAGE gel of GFP-affinity pulldown beads in the NG108-15 cells transfected with GFPNaViPA1 and GFP and cell lysates prepared using denaturing RIPA buffer (I) and the results of PIP strip analysis.

Since NaV1.7 is an integral membrane protein, we therefore tested whether NaViPA1 expression in the HEK1.7 cells would interrupt NaV1.7 intracellular trafficking. Our results do not support this mechanism since no clear reduction of membrane NaV1.7 protein was evident in the fractionized preparations, from HEK1.7 cells transfected with NaViPA1 and controls (Figure 5F). Studies have shown that IDRs in the membrane proteins engage in interactions with the membrane (33). To test whether NaViPA1 interference of NaV1.7 might be via direct block of NaV1.7, GFP affinity pull-down by ChromoTek GFP-Trap (ChromoTek) was performed after transfection of GFP-NaViPA1 in HEK1.7 cells using GFPlinker and GFP-1.7iPA2 (Figure 1) as the controls. Cell lysates were prepared by a lysis buffer containing 0.5% Nonidet p40, a nondenaturing mild lysis detergent, for preventing interaction breaking and maximizing the retention of NaViPA1-protein interactions (34). IBs verified full-length NaV1.7 protein trapped in the GFPNaViPA1 pull-down sample but not in controls (Figure 5G, NaViPA1 pull-down with NaV1.7 was confirmed in an additional experiment), and nano liquid chromatography mass spectrometry (nLC-MS/MS) detection of unique hNaV1.7 peptides (Table 1) confirmed hNaV1.7 on the excised band from silver-stained sodium dodecyl sulfate–polyacrylamide electrophoresis (SDS-PAGE) gel (Figure 5G, right panel) of GFPNaViPA1 affinity pull-down sample. These results indicate that NaViPA1 block of NaV1.7 channel activation could be via binding to the NaV1.7 protein, i.e., an intramolecular domain-domain interaction (intraDDI) (35). It has been reported that polybasic IDRs in TM proteins preferably bind to negatively charged lipids (36, 37). We reasoned that NaViPA1 might be able to bind phosphoinositides, and this hypothesis was tested by using phosphatidylinositol phosphate (PIP) strips (Echelon PIP Strip). GFPNaViPA1 and GFP (control) were transfected into neuronal NG108-15 cells, and cell lysates were prepared by a RIPA buffer containing 0.1% SDS with 1% Triton X100, both strong detergents, and 1% deoxycholate, an anionic detergent, for maximal denaturing to break NaViPA1 PPI complex formations. Silver stain after SDS-PAGE gel showed clean purification of GFP and GFPNaViPA1 (Figure 5H) and samples were applied to the PIP strips. Results (Figure 5I, repeat twice) showed that GFP NaViPA1 was efficiently bound to a number of anionic PIPs, PIP2, phosphatidic acid (PA), and phosphatidylserine (PS). In contrast, affinity pull-down GFP did not show clear binding to lipid spots, as previously reported (38). This is consistent with the reports that basic residues, often clustered in IDRs, can modulate membrane protein functions by binding via electrostatic interactions with lipids (39, 40).

Unique hNav1.7 peptides detected in GFPNaviPA1 pull-down sample

NaViPA1 is a polybasic arginine/lysine and serine-enriched peptide (Figure 2F). Protein-conserved polybasic domains with adjacent serine PTMs often play roles in protein function (41–43). We designed experiments to examine the role of polybasic NLS and multiple adjacent polyserine in the function of NaViPA1. Initial tests were performed by generating NaViPA1 mutant 1 (NaViPA1mt1) (GFP-fused) in which alanine substitution for ten serine residues within NaViPA1 was made, and NaViPA1mt2 (GFP-fused) was generated by alanine substitution for 9 arginine or lysine within the predicted polybasic NLS of NaViPA1 (Figure 6A). ICC showed that nuclear localization of NaViPA1 (HEK1.7 cell transfection) was observed in NaViPA1 and NaViPA1mt1 but diminished in NaViPA1mt2 (Figure 6, B–E). With a comparable transfection rate at approximately 40% for each construct, IBs revealed that NaViPA1 was detected in the extracted cytosol, membrane, and nuclear samples, and that the membrane-binding and nuclear entry signals in NaViPA1mt1 were comparable to NaViPA1 but both vanished in NaViPA1mt2. Full-length NaV1.7 was enriched in the membrane samples, as shown in Figure 5F, and the presence of NaViPA1, mt1 and mt2 in HEK1.7 cells appeared not to impede NaV1.7 protein membrane integration (Figure 6F). Whole-cell voltage-clamp recording showed that INa1.7 in the presence of NaViPA1mt1 and mt2 was comparable to naive and GFP-NP transfected HEK1.7 cells, suggesting that both polybasic arginine/lysine and multiple adjacent serine residues were required for NaViPA1 inhibitory effect on NaV1.7 current. To further map the critical serine sites, we generated additional NaViPA1mt3–mt6 with alanine substitution for dual or triple serine residues (Figure 6A). Whole-cell voltage-clamp recordings showed that NaViPA1mt3 and 5 with alanine substitution at different serine sites lost inhibitory effects on INa1.7 after transfection to HEK1.7 cells while mt4 and mt6 showed a significant block of INa1.7 (Figure 6, G and H). As expected, NaViPA1mt1 and mt2 did not change INa1.8 after being transfected to HEK1.8 cells, similar to NaViPA1 (Supplemental Figure 8). These data suggest that conserved polybasic NLS and multiple adjacent serine residues within the NaViPA1 are synergistic for INa1.7 inhibition. The polybasic motif determines the polar association with the plasma membrane and nuclear entry of disordered NaViPA1 peptide and multiple adjacent serine residues are required for NaViPA1 inhibitory effect to INa1.7. However, the full-length NaV1.7 membrane integration, which is determined by its TM domains but not intracellular sequences, was unaffected in the presence of NaViPA1 (Figure 6F).

Define polybasic NLS and adjacent serine in NaviPA1. (A) Sequence alignments of NaviPA1, mutant 1 (mt) with alanine substitution of ten serine residues, mt2 by alanine substitution of arginine/lysine (R/K) (mt2) within predicted NLS domain, and mt3–6 with alanine substitution of bi- or triserine residues at different serine sites, as indicated. (B–E) ICC comparison of GFP signals 48 hours after plasmids coding NaviPA1, GFPNP, mt1, and mt2 transfected into HEK1.7 cells. Scale bar: 100 μm. (F, left) Representative IBs of endogenous Nav1.7, as well as GFPNP, NaviPA1, mt1, and mt2, in extracted cytosol, membrane, and nuclear samples after transfection into HEK1.7 cells. Cytosol, membrane, and nuclear loading were indicated by GAPDH, NKA1α, and LamB1, respectively. (F, right) Quantitative (ImageJ gel analysis) comparison of membrane binding and nuclear entry of NaviPA1, mt1, and mt2 after transfection, *P < 0.05, **P < 0.01, 1-way ANOVA and Tukey’s post hoc. (G) Representative INa1.7 traces of HEK1.7 cells recorded from sham, GFPNP, NaviPA1, mt1–mt6 (3–4 days after transfection), as indicated. (H) Quantification summary of peak INa densities; *P < 0.05, **P < 0.01, and ***P < 0.001; 1-way ANOVA and Tukey’s post hoc.

Future delineation of (a) the properties of serine and other residue PTMs within NaViPA1 underlying inhibition of various TTXs INa in sensory neurons and (b) investigation of whether the presence of NaViPA1 might undermine TTXs Nav channel activity via decoying interaction, diminishing PTMs in the full-length protein, and/or altering intradomain effects are of interest from both pathophysiological and therapeutic perspectives. Our goal in this study is to develop a strategy of peripherally targeted analgesia via AAV-mediated sensory neuron-specific inhibition of sodium channels. Therefore, in the following in vivo experiments, we focused on testing whether DRG-PSN–targeted expression of NaViPA1 is effective in attenuating neuropathic pain behaviors.

Analgesia after intraganglionic delivery of AAV-NaViPA1 in rats after TNI

We first conducted a pilot in vivo analgesia testing. High-titer and high-purity of AAV6-GFPNaViPA1 (AAV6-NaViPA1) and control AAV6-GFPNP (AAV6-NP) were generated and injected into the L4/5 DRG of adult male rats. Three weeks after DRG-AAV injection, tibial nerve injury (TNI) was induced and subsequent sensory behavior evaluation was performed weekly for an additional 5 weeks, after which tissues were harvested for IHC characterization of transgene expression. Results (Supplemental Figure 9) showed that AAV6-NaViPA1 injection reduced TNI-induced mechanical and cold sensitization. IHC revealed efficient NaViPA1 (fused to GFP) expression in DRG neurons and their peripheral (cutaneous) and central terminals (SDH). These data indicate that sustained expression of the NaViPA1 selectively in the PSNs of the pathological DRG after TNI prevented development of pain behaviors.

Treatment of established neuropathic pain by DRG-AAV6-NaViPA1 in male rats

We next extended experiments to evaluate the effectiveness of DRG-AAV6-NaViPA1 in a more clinically relevant design for reversal of established pain behaviors, including both evoked responses as well as spontaneous ongoing pain following TNI. In the experimental design, the sensitivity to mechanical and thermal cutaneous stimulation was assessed at baseline and weekly after TNI for 2 weeks before AAV injection. Thereafter, rats were randomized to receive DRG injection of either AAV6-NaViPA1 or control AAV6-NP into the L4/L5 DRG ipsilateral to TNI, after which sensory behaviors were evaluated weekly for additional 6 weeks. As a terminal experiment, Gabapentin-induced (GBP-induced, 100 mg/kg, i.p.) conditioned place preference (CPP) test was performed in both groups to evaluate spontaneous pain (12, 44). Behavior measures before AAV injection on the 14th day after TNI were used as a treatment baseline (tBL) to evaluate effectiveness of vector treatments (Figure 7, A and B). Tissues were harvested for IHC characterization of transgene and target gene expression and for whole-cell current-clamp of neuronal excitability on dissociated DRG neurons.

Treatment of established neuropathic pain by DRG AAV6-NaViPA1 in male rats. (A) Silver stain of purified AAVs (a vertical white line denotes that the lanes were run on the same gel but were noncontiguous) were prepared for the experiment in an animal protocol schematically outlined in panel B. (C–F) The time courses (graphs on the left) of vF (C), Pin (D), Heat (E), and (F)Cold before and after DRG injection of either AAV6-NaViPA1 (n = 7) or AAV6-NP (control, n = 8). The measures on the 14th day after TNI and before AAV treatment (tBL) were converted as the peak pain intensity (100%), and the measures of each sensory modality after treatment were normalized to the measures at the tBL and the percentage of pain relief for each modality at multiple time points was calculated (graphs on the right). *P < 0.05, **P < 0.01 and ***P < 0.001 for comparisons to the tBL within group and #P < 0.05, ##P < 0.01, and ###P < 0.001 between groups. Repeated measures 2-way ANOVA for vF and Heat, and Tukey’s (within group) and Bonferroni’s (between groups) post hoc; and nonparametric Friedman ANOVA for Pin and Cold tests and Dunn’s post hoc. Summed average pain relief in the 6-week treatment course showed 52%, 49%, 69%, and 67% reduction of vF-, Pin-, Cold-, and Heat-stimulated mechanical and thermal pain behaviors, respectively (G). ** P < 0.01 *** P < 0.001, unpaired, 2-tailed student’s t test. (H) Results of CPP scores (seconds, s) of preconditioning chamber and of the GBP-paired chamber between AAV-NaViPA1 (n = 7) and AAV-NP (control, n = 8), ***P < 0.001 (unpaired, 2-tailed Student’s t test).

All rats developed multiple modalities of pain behaviors 2 weeks after TNI, including lowered threshold for withdrawal from mild mechanical stimuli using calibrated monofilaments (von Frey test, vF), more frequent hyperalgesic-type responses after noxious mechanical stimulation, by applying a 22 g spinal anesthesia needle to the plantar surface of the hind paw with enough force to indent, but not puncture the skin (Pin test), and hypersensitivity to heat and acetone stimulation. These behaviors persisted after injection of the control AAV6-NP during the 6 weeks of observation course. In contrast, rats injected with AAV6-NaViPA1 showed a gradual reversal of these changes, which were maintained throughout and predicted to outlast the observation period (Figure 7, C–F). For our protocol of treating existing pain, we converted the measures on the 14th day after TNI and before AAV treatment (tBL) as the peak pain intensity (100%), and the measures of each sensory modality after treatment were normalized to the measures at the tBL and the percentage of pain relief for each modality at multiple time points was calculated (Figure 7, C–F). Summed average pain relief in the 6-week treatment course showed 52%, 49%, 69%, and 67% reduction of vF-, Pin-, Cold-, and Heat-stimulated mechanical and thermal pain behaviors, respectively (Figure 7G). Using a biased CPP paradigm (45), the effect of AAV-NaViPA1 treatment on the affective aspect of spontaneous pain was evaluated. None of the animals in either group were excluded from study because of their baseline preference/avoidance for a chamber (45). A significant GBP-induced CPP effect was observed in the TNI rats injected with AAV6-NP, indicating ongoing pain, while there was no significant difference in the time spent in the initially nonpreferred chamber during baseline versus testing periods in AAV-NaViPA1–treated TNI animals, indicating that AAV-NaViPA1 treatment significantly relieved on going spontaneous pain (Figure 7H).

Histological examination (Figure 8) determined the in vivo transduction rate for AAV6-NaViPA1 in the 6th week after vector injection. The NaViPA1-positive neurons (GFP) comprised 37% ± 4 % (1,283 out of 3,447 total neuronal profiles) identified by a panneuronal marker β3-tubulin (n = 4 DRG, 3–4 sections per DRG, selected as every 5th section from the consecutive serial sections). Transduced DRG neurons included the full-size range of the PSNs that also expressed NaV1.7 and NaV1.6, and expression showed multiple subcellular localizations, preferably in PSN cytosol. Positive GFP signals were not detected in GFAP-positive perineuronal glial cells. GFP signals were also detected in the ipsilateral dorsal horn, sciatic nerve, and cutaneous afferent terminals.

IHC of GFP-NaViPA1 and target gene expression. (A–D) Representative IHC montage images (GFPNaViPA1 with Tubb3) show neuronal expression profile 6 weeks after AAV- NaViPA1 injection in TNI rats (A; Scale bar: 200 μm), colocalization of GFP-NaViPA1 with NaV1.7 and NaV1.6-positive neurons (B and C; Scale bar: 100 μm), but not with GFAP positive perineuronal glia (D; Scale bar: 100 μm). The square region was enlarged and montage images shown as the 4th image in row D). (E–G) Representative IHC montage images illustrate GFPNaViPA1 (green) and NaV1.7 (red) in PSN central terminals of ipsilateral spinal dorsal horn (E; Scale bar: 200 μm), GFPNaViPA1 (green) and Tubb3 (red) in sciatic nerve (F; Scale bar: 50 μm), and GFPNaViPA1 (green) and NF200 (red) in PSN peripheral terminals of skin section (G; Scale bar: 50 μm).

These findings together demonstrate that DRG injection of AAV6-encoded NaViPA1 induced NaViPA1 expression restricted to the PSNs of injected DRG and their peripheral and central processes. This strategy via AAV6-mediated expression of NaViPA1 selective in the sensory neurons of the anatomically segmental DRG responsible for pain pathophysiology has clear analgesic effectiveness in normalizing the established peripheral hypersensitivity for both evoked and spontaneous pain behavior in the rat model of peripheral injury–induced neuropathy.

Reversal of PSN hyperexcitability by AAV6-NaViPA1 treatment in male rats

Increased excitability of nociceptive PSNs is a fundamental process underlying neuropathic pain (46). We therefore examined whether AAV6-NaViPA1 treatment reverses the enhanced neuronal excitability of nociceptive PSNs following TNI (12, 47) using the whole-cell current-clamp AP recording of DRG dissociated neurons from rats after the treatment protocol shown in Figure 7B. Although TNI results in DRG containing injured and uninjured neurons, nerve injury can induce an increase of voltage-gated ion channel activity in both axotomized neurons and adjacent intact neurons, leading to similar electrophysiological (EP) changes and increased discharge frequency in axotomized and neighboring intact DRG neurons (48, 49), possibly through interneuronal signaling and coupling (50). We therefore performed current-clamp recordings from randomly chosen small-to-medium-sized neurons (under 35 μm in diameter) (51) in the cultures from dissociated L4 and L5 DRG. Transduced neurons were identified by GFP fluorescence, and excitability was evaluated by measuring rheobase and repetitive action potential (AP) firing during 250 ms current pulses stepping from 100 pA and 280 pA current injection. Results showed that the averaged rheobase in the neurons from TNI rats was significantly decreased and, in response to a step stimulus, the frequency of APs evoked in neurons from TNI rats was significantly increased compared with sham controls. These were normalized in the transduced neurons after AAV6-NaViPA1 treatment, whereas NP-transduced neurons had no significant effects (Figure 9). These findings indicate that reversal of nerve injury–induced sensory neuronal hyperexcitability by NaViPA1 may contribute to its analgesic effects in attenuation of neuropathic pain behaviors, i.e., conduction block of TTXs NaV ion channels selectively in PSNs leads to a decrease in neural excitability, resulting in mitigation of pain behaviors.

NaViPA1 expression on neuronal excitability of male rat PSNs. (A and B) Representative AP traces elicited by 250 ms depolarizing current of 180 pA (A) and 280 pA (B) (same cells) from RMP were recorded from DRG neurons dissociated from the rats of sham, TNI only, and GFP-expressing neurons in TNI treated with AAV6-NP or AAV6-NaViPA1, as indicated. (C) Comparison of responses (number of APs evoked by a 250 ms stimulus) for the populations of DRG neurons in different groups across a range of step current injections from 100 to 280 pA; ***P < 0.001, 2-way ANOVA of main effects of groups with Bonferroni’s post hoc. Scatter plots with bars show analysis of the (D) rheobases and (E and F) AP numbers evoked by input current at (E) 180 pA and (F) 280 pA from RMP. The number in each group is the number of analyzed neurons per group. *P < 0.05 and ***P < 0.001, 1-way ANOVA and Turkey’s post hoc.

Analgesia of DRG-AAV6-NaViPA1 treatment in female TNI rats

Sex differences exist in experimental and clinical pain and in responsivity to interventions (52). We therefore next tested whether DRG-AAV6-NaViPA1 treatment is also effective in attenuating hypersensitivity induced by TNI in female animals, using the protocol similar to the tests in male animals (Figure 7). The same batch preparation of AAV6-NaViPA1 and AAV6-NP tested in male rats was used for injection. Results showed that the female rats displayed similar phenotypic development of hypersensitivity after induction of TNI to male rats and that both evoked mechanical/thermal hypersensitivity and GBP-CPP responses were normalized after AAV6-NaViPA1 treatment, demonstrating comparable analgesic effects (Figure 10, A–E) with the male animals. IHC on the DRG sections from female TNI rats 6 weeks after AAV6-NaViPA1 injection also revealed GFP-NaViPA1 expression profile comparable with male rats (Figure 10, F–H ), and the in vivo transduction rate was 39% ± 8 % (766 out of 1,983 total Tubb3-positive neuronal profiles). Thus, although not rigorously compared, treatment effects were comparably concordant between the sexes, suggesting that sexual dimorphism seems not apparent for both pain behavior phenotypes after TNI and in response to DRG-AAV6-NaViPA1 treatment in our studies (12).

Analgesia of DRG-AAV6-NaViPA1 treatment in female TNI rats. (A–D) Analogous figures to those shown in Figure 7 show significant analgesia (left graphs) and % of pain reduction (middle graphs) after DRG delivery of AAV6-NaViPA1 in the established TNI pain behaviors of female rats. *P < 0.05, **P < 0.01 and ***P < 0.001 for comparisons to the treatment baseline (tBL) within group and #P < 0.05, ##P < 0.01, and ###P < 0.001 for comparisons between groups. Repeated measures parametric 2-way ANOVA for vF and Heat followed by Tukey’s (within group) and Bonferroni’s (between groups) post hoc; and nonparametric Friedman ANOVA for Pin and Cold tests and Dunn’s post hoc. Right graphs of A–D show average pain relief of each modality in 3.5-month treatment, ** P < 0.01 and *** P < 0.001 comparisons between groups (unpaired, 2-tailed Student’s t tests). (E) CPP difference scores (s) of preconditioning chamber and of the GBP-paired chamber between AAV- NaViPA1 (n = 8) and AAV-NP (control, n = 8), *P < 0.01 (unpaired, 2-tailed Student’s t test). (F–H) Representative montage IHC images colocalization of GFP-NaViPA1 with Tubb3 (F), NaV1.7 (G), and NaV1.6 (H) 6 weeks after AAV-NaViPA1 injection. Scale bar: 100 μm.