Female C57Bl/6J mice of 18 months of age (Jackson laboratory) were used in the study. All mice were maintained at MGH animal housing facility in a specific pathogen free (SPF) environment. The room temperatures were 19 °C–23 °C, the humidity was 40%–60%, and had a 12 h light/dark cycle. Three to four mice were housed in each ventilated cage. Food and water were provided ad libitum. Mice were randomly assigned to experimental groups. Gabapentin (100 mg/kg, Sigma-Aldrich, G154) and resveratrol (40 mg/kg, Fisher Scientific, R0071) were administered by intraperitoneal (i.p.) injection. Gabapentin was dissolved in saline (5 mg/mL), and resveratrol was dissolved in a vehicle solution (5% DMSO+30% PEG300 in water) to obtain 15 mg/mL solution. The respective dissolving solution was used as vehicle control. To reduce stress, animals received 1 day break after every 6 days of injection. The animal protocol was approved by Massachusetts General Hospital Institutional Animal Care and Use Committee.

Spared nerve injury (SNI) was produced as previously described (Decosterd and Woolf, 2000). Briefly, a skin incision on the left thigh was made on a fully anesthetized mouse (1%–2% isoflurane in 100% O₂) to expose the sciatic nerve. The common peroneal and tibial nerve branches were completely sectioned, leaving the sural nerve branch intact. Sham-operated mice were subjected to the same procedure without sectioning the nerves. The incision site was sutured, and mice were returned to home cages when fully recovered from anesthesia.

All behavioral tests were carried out by the investigators who were blinded to experimental groups. Mice were habituated to the test environment for two consecutive days (30 min per day) prior to testing.

To overexpress Sirt1 in the hippocampus, adeno-associated virus 9 (AAV9) vector carrying Sirt1 or eGFP was purchased from Vector Biolab (Malvern, PA, United States): AAV9-CMV-eGFP-2A-mSirt1 (Catalog number: 7000), AAV9-eGFP (Catalog number: 7007). The titer of viruses is 1 × 10¹³ gc/mL (genome copy/mL). The mice were anesthetized with isoflurane and secured in a stereotaxic frame (Kopf, Tujunga, CA, United States), and AAV was injected with the Nanoject III (Drummond Scientific Company, model 3-000-207) under sterile conditions. Bilateral injection was performed and holes of the size of the injection needle were drilled into the skull: 2.1 mm caudal to bregma, 1.5 mm ventral to pial surface, and 1.5 mm right of midline for right side injection and 1.5 mm left of midline for left side injection. Each side was injected with 1 μL of AAV vector (1 × 10¹³) at a rate of 0.2 μL/min.

The hippocampal tissue was harvested and stored at −80 °C until being processed. For Western blot analysis, tissue was homogenized in RIPA buffer (Cell Signaling 9806S) containing Protease inhibitor (Thermofisher A32953) and phosphatase inhibitor (1 mM sodium fluoride, 1 mM β- glycerophosphate, 0.5 mM sodium orthovanadate, and 0.5 mM sodium pyrophosphate). Protein (30 μg/lane) was separated on SDS-PAGE gel, transferred to PVDF membrane (Sigma Immobilon-P, IPVH00010), probed with antibodies, and detected using ECL substrates (SuperSignal West Pico Plus, ThermoFisher 34580) and X-ray films. PageRuler Plus Pretained protein standard (ThermoFisher 0026619) was used to estimate the protein size. X-ray films were scanned, and protein bands were quantified using ImageJ (NIH) (Schneider et al., 2012). p-Tau proteins were normalized to tau; Sirt1, CaMKIIα and tau were normalized to actin. In graphs, the values shown are relative to “sham + saline” or “AAV-eGFP + saline” or “saline + vehicle”. The antibodies used in this study are listed in Table 1.

Total RNA was isolated from brain tissue using Trizol (Thermosfisher 15596026). cDNA was synthesized using ProtoScript First Strand cDNA Syn Kit (E6300L) from New England BioLabs (Ipswich, MA, USA). qPCR was preformed using TaqMan probes (CamKII2a: Mm00437967_m1; Gapdh: Mm99999915_g1) and TAQman universal MMIX II (Cat # 4440040) from Life Technologies (CA, United States) on a Quant 3 QuantStudio™ 3 Real-Time PCR system (Applied Biosystems). The data were analyzed by the ΔΔCt method. The mRNA level of CamKII2a was normalized to GAPDH, and the value in the graph is expressed as relative to “sham + saline”.

Hippocampal CaMKII activity was analyzed using CycLex CaM-kinase II assay kit (CY-1173) and CaM-Kinase II positive control (CY-E1173) (MBL International Corporation, Woburn, MA) according to the protocol provided by the manufacture. Sirt1 activity was analyzed using Sirt1 assay kit (Sigma, Cat # 1040) according to the protocol provided by the manufacturer.

Data were analyzed using GraphPad Prism 8 software and expressed as mean ± SEM. Unless specified, one-way ANOVA with Tukey’s multiple comparisons was used for the analysis. Two-way ANOVA was used for repeated measurement of paw withdrawal threshold in von Frey test. T-test was used to compare paw withdrawal threshold before and after GBP administration. The significance level was set at 0.05.

Aged (18-month-old) mice with SNI or sham surgery (sham) were treated with GBP (100 mg/kg/daily, i. p.) for ∼ 2 months. The dose of GBP was based on previous studies (Kusunose et al., 2010; Aydin et al., 2012; Cheng and Chiou, 2006). The experimental design is illustrated in Figure 1. At 7 days after administering the last dose of GBP or saline, OFT was performed to determine whether nerve injury and/or GBP treatment affected motor activity in aged mice. In the open field, the four groups of mice traveled similar total distance, suggesting that motor activity was not significantly affected in aged mice (one-way ANOVA, p = 0.27, n = 10/group) (Figure 2A).

Mice were subsequently subjected to NOR test to examine short-term memory (Cohen and Stackman, 2015) as illustrated in Figure 2B. In training phase, both saline- and GBP-treated sham or SNI mice spent an equal amount of time exploring two identical objects, and had a similar recognition index (RI) for both objects (Supplementary Figure S1). In testing phase, saline-treated mice (SNI or sham) spent more time exploring a novel object than GBP-treated sham or SNI mice did. Therfore, GBP-treated mice had a lower RI for the novel object, showing an impairment in recognition memory (Figure 2C) (one-way ANOVA with Tukey’s multiple comparisions test. sham + saline vs. sham + GBP: ***p < 0.001; SNI + saline vs. SNI + GBP: **p < 0.01, n = 10/group).

Next, we performed FCT to assess associative fear learning and memory as depicted in Figure 3A. In training phase, the average freezing time was less than 10 s for all mice, and the four groups of mice did not differ significantly in freezing time (one-way ANOVA, p = 0.62, n = 10/group) (Figure 3B). Mice were tested on day 3 and day 7 for context- and tone-associated memory. In context test, GBP-treated mice had less freezing time than saline-treated mice (Figures 3C,E, day 3 and day 7; for sham or SNI groups, saline vs. GBP: *p < 0.05, **p < 0.01). In tone test, the freezing time was decreased in SNI mice when tested on day 3 (Figure 3D, SNI groups: saline vs. GBP, *p < 0.05), but not on day 7 (Figure 3F). We also compared the freezing time during the first 3 min in the chamber during context and tone test. Saline-treated mice had longer freezing time in the context test chamber (black/white stripes pattern) than in the tone test chamber (black/white checkered pattern), while GBP-treated mice did not exhibit differences (Supplementary Figures S2A,B), indicating that saline-treated mice could remember and distinguish the environment wherein they experienced shock, but GBP-treated mice could not. These data re-enforced our data presented in Figures 3B–E showing GBP-treatment induced cognitive impairments.

We assessed SNI-induced mechanical allodynia using von Frey test on day 3, 7, 33, and 63 prior to daily GBP administration. Mice with SNI exhibited prolonged mechanical allodynia compared to sham mice, as SNI mice had a lower threshold in response to von Frey fiber stimulation on the ipsilateral paw than sham mice (two-way ANOVA, p < 0.0001, n = 10/group (Supplementary Figure S3A). GBP treatment effectively attenuated nociception in SNI mice measured at 2 h after GBP administration (Supplementary Figure S3B), suggesting that signs of cognitive impairment in GBP-treated mice, as shown above, were not caused by SNI-induced pain.

Taken together, these data suggest that GBP treatment imparied cognitive function in aged mice with or without neuropathic pain.

Tau proteins, a group of six isoforms, are microtubule-associated proteins and are highly expressed in neurons in the hippocampus and cortex (Mandelkow and Mandelkow, 2012; Binder et al., 1985; Jameson et al., 1980). We examined if GBP treatment caused abnormal tau phosphorylation, as dysregulation of tau phosphorylation is associated with cogntive impairment. We analyzed the hippocampus tissue using Phospho-Tau Family Antibody Sampler Kit and Western blot. GBP-treated mice had increased levels of p-tau (S262) and p-tau (S416) in the hippocampus compared to saline-treated mice (Figures 4A–C) (**p < 0.01, *p < 0.05, n = 4/group). The levels of total tau, p-tau (S202), p-tau (S396), and p-tau (S404) did not significantly differ between GBP and saline treatment groups (Figures 4D–H). Our observation indicates that hippocampal tau hyperphosphorylation is associated with GBP-induced congntive impariment in aged mice.

Phosphoration of tau is regulated by several protein kinases. Ser/Thr protein kinase CaMKIIα preferentially phosphorylates Tau at S262 and S416 sites (Sironi et al., 1998; Yamamoto et al., 2005; Hector et al., 2020). CaMKIIα, a major isoform of CaMKII, is highly expressed in the hippocampus (Wang et al., 2013). Accordingly, we analyzed the expression levels and kinase activity of CaMKIIα. Western blot and q-RT-PCR analysis showed elevated CaMKIIα expression levels in the hippocampus of GBP-treated mice (Figures 5A,B,D). Moreover, the kinase activity of CaMKIIα in the hippocampus was also increased in GBP-treated mice (Figure 5E), which is consistent with the increase of CaMKIIα protein expression. To examine how GBP regulates CaMKIIα expression, we focused on Sirt1, a histone deacetylase, as Sirt1 transcriptionally regulates CaMKIIα expression via deacetylating histone H3 lysine 9 at the CaMKIIα promoter (Zhou et al., 2020). Sirt1 protein expression in the hippocampus was decreased in GBP-treated mice (Figure 5C). Taken together, GBP treatment led to differential expression of CaMKIIα and Sirt1 in the hippocampus of aged mice.

Since we identified that decreased Sirt1 level could contribute to tau pathology induced by GBP in aged mice, we examined if overexpression of Sirt1 rescued behavioral outcomes and tau pathology. Adeno-associated viral vector-mediated gene transfer and overexpression was used. AAV9-eGFP (control) or AAV9-eGFP-Sirt1 (10¹³ gc/mL, 1 µL/side, Vector Biolabs) were bilaterally infused into the hippocampus of 18-month-old (aged) mice. At 3 weeks after the AAV9 infusion, mice were subjected to SNI surgery, treated with saline or GBP for ∼ 2 months, and examined for behavioral outcomes as illustrated in Figure 1. AAV9 infusion did not affect motor function in mice (one-way ANOVA, p = 0.41, n = 10/group) (Figure 6A). Sirt1 or eGFP overexpression in the hippocampus did not affect SNI-induced nociception (Supplementary Figure S4). NOR and FCT tests showed that Sirt1 overexpression ameliorated cognitive impairment in GBP-treated mice (Figures 6B–G, GBP groups: eGFP vs. Sirt1, **p < 0.01, Supplementary Figures S2A,C, AAV-eGPF/saline and AAV-Sirt1/GBP: pre-context vs. pre-tone, *p < 0.05). Tissue analysis showed that AAV9-eGFP-Sirt1 infusion rescued Sirt1 expression levels and activity in GBP-treated mice, inhibited CaMKIIα expression and phosphorylation of p-tau (S416) and p-tau (S262) (Figure 7, ***p < 0.001, **p < 0.01, *p < 0.05, n = 4/group). As a control for AAV infusion, the expression of eGFP did not alter the adverse effect of GBP on cogntive function and tau phosphorylation in aged SNI mice (Figures 6, 7, AAV-eGFP groups: saline vs. GBP, **p < 0.01, *p < 0.05). Thus, overexpression of Sirt1 suppressed the adverse effects of GBP on cognitive activity and tau phosphorylation in aged mice.

To examine if enhancing Sirt1 activity could improve GBP-induced cogntive impairment, we treated mice with resveratrol, a Sirt1 activator. After SNI, mice were co-administered with GBP (100 mg/kg, i.p.) and resveratrol (RSV) (40 mg/kg, i.p.) for 60 days. Three groups of mice were included: saline + vehicle, GBP + vehicle, and GBP + RSV. Behaviral test was conducted as depicted in Figure 1. The dose of resveratrol was based on previous studies (Park and Pezzuto, 2015). Resveratrol treatment did not affect motor function (one-way ANOVA, p = 0.35, n = 10/group) (Figure 8A). In NOR test, resveratrol improved recognition index (RI) for the novel object in GBP-treated mice (GBP + vehicle vs. GBP + RSV: ***p < 0.001) (Figure 8B). In FCT, resveratrol treatment increased freezing time in GBP-treated mice in context and tone tests. Thus, resveratrol prevented cognitive impairments in GBP-treated mice (Figures 8A–H) (Supplementary Figures S2A,D, saline/vehicle and GBP/RSV: pre-context vs. pre-tone, ,**p < 0.01, *p < 0.05). Analysis of the hippocamal tissue indicated that reservatrol treatment restored Sirt1 activity (Figure 9A), inhibited CaMKIIa expression and hyperphosphoryaltion of tau (Figures 9B–F) in GBP-treated mice.