Expression in skin biopsies supports genetic evidence linking CAMKK2, P2X7R and P2X4R with HIV-associated sensory neuropathy

HIV-associated sensory neuropathy (HIV-SN) affects 14–38% of HIV+ individuals stable on therapy with no neurotoxic drugs. Polymorphisms in CAMKK2, P2X7R and P2X4R associated with altered risk of HIV-SN in Indonesian and South African patients. The role of CaMKK2 in neuronal repair makes this an attractive candidate, but a direct role for any protein is predicated on expression in affected tissues. Here, we describe expression of CaMKK2, P2X7R and P2X4R proteins in skin biopsies from the lower legs of HIV+ Indonesians with and without HIV-SN, and healthy controls (HC). HIV-SN was diagnosed using the Brief Peripheral Neuropathy Screen. Biopsies were stained to detect protein gene product 9.5 on nerve fibres and CaMKK2, P2X7R or P2X4R, and were examined using 3-colour sequential scanning confocal microscopy. Intraepidermal nerve fibre densities (IENFD) were lower in HIV+ donors than HC and correlated directly with nadir CD4 T-cell counts (r = 0.69, p = 0.004). However, IENFD counts were similar in HIV-SN+ and HIV-SN− donors (p = 0.19) and so did not define neuropathy. CaMKK2+ cells were located close to dermal and epidermal nerve fibres and were rare in HC and HIV-SN− donors, consistent with a role for the protein in nerve damage and/or repair. P2X7R was expressed by cells in blood vessels of HIV-SN− donors, but rarely in HC or HIV-SN+ donors. P2X4R expression by cells in the epidermal basal layer appeared greatest in HIV-SN+ donors. Overall, the differential expression of CaMKK2, P2X7R and P2X4R supports the genetic evidence of a role for these proteins in HIV-SN. Supplementary Information The online version contains supplementary material available at 10.1007/s13365-023-01134-2.


Introduction
HIV-associated sensory neuropathy (HIV-SN) is a debilitating complication of HIV infection and antiretroviral therapy (ART) (Affandi et al. 2008;Wadley et al. 2011;Octaviana et al. 2018;Pillay et al. 2019). Despite the use of ART that excludes drugs associated with neurotoxicity (usually stavudine) in the last decade, HIV-SN continues to impact up to 38% of people living with HIV (PLWH) Pillay et al. 2019). Symptoms of HIV-SN include numbness, "pins and needles", disordered sensation and neuropathic pain, which impact an individual's ability to work and their quality of life (Phillips et al. 2010). No interventions prevent or reverse HIV-SN progression (Phillips et al. 2010), so a better understanding of underlying mechanisms is sought.
Clinical features of HIV-SN include degeneration of long axons in a "die-back" manner, so that reduced intraepidermal nerve fibre densities (IENFD) have been observed in some studies describing PLWH with severe HIV disease and/or experiencing pain (Polydefkis et al. 2002;Phillips et al. 2014). Increased infiltration of mononuclear cells and cytokine expression in dorsal root sensory ganglia (DRG) (Jones et al. 2005;Hahn et al. 2008) and increased expression of chemokine receptors by inflammatory cells surrounding intraepidermal nerves (Mountford et al. 2018) suggest a central role for inflammation in the manifestation of HIV-SN. An inflammatory aetiology is supported by genetic associations with polymorphisms in a block of genes surrounding the TNF gene (Wadley et al. 2015;Gaff et al. 2020b), and in three neighbouring genes: P2X7R, P2X4R and CAMKK2 (the P2X-block) (Goullee et al. 2016;Gaff et al. 2019Gaff et al. , 2020aSafri et al. 2020). We linked polymorphisms and haplotypes from the P2X-block with HIV-SN in Indonesian and South African patients, implicating the encoded proteins in the pathogenesis of HIV-SN (Gaff et al. 2019(Gaff et al. , 2020a. Whilst the clearest associations were with CAMKK2, linkage disequilibrium makes a role for P2X4R or P2X7R plausible. These proteins are the focus of the present study. CAMKK2 encodes calcium/calmodulin-dependant protein kinase kinase 2 (CaMKK2) which phosphorylates adenosine monophosphate (AMP)-activated protein kinase (AMPK), sirtuin 1 (SIRT1), and calcium/calmodulin-dependant kinase 1 and 4 (CaMKIV and CaMKI) (Kokubo et al. 2009;Wen et al. 2013). AMPK activation mediates inflammation and apoptosis Zhang et al. 2019). SIRT1 activation regulates axonal regeneration, promotes dendrite arborisation and protects neurons from oxidative stress Codocedo et al. 2012;Liu et al. 2013). CaMKIV activation upregulates nuclear factor kappa B and cyclic AMP response element-binding protein which stimulates brain-derived neurotrophic factor (BDNF), promoting neuronal growth and survival (Cao and DeLeo 2008;Wayman et al. 2008;. Excessive BDNF expression is implicated in neuropathic pain (Ulmann et al. 2008). Activated CaMKI plays a role in axonal growth cone morphology and outgrowth, dendrite arborisation and synapse formation (Wayman et al. 2004;Ageta-Ishihara et al. 2009).
Experimental evidence implicates P2X7R, P2X4R and CaMKK2 expressed by cells in DRG, spine and brain in neurological outcomes affecting the periphery. It is also plausible that these proteins may contribute directly to the pathogenesis of HIV-SN via degeneration of sensory nerve terminals in the skin. P2X7R is expressed in the skin by keratinocytes, Langerhans cells, dermal dendritic cells, T-cells and macrophages (Geraghty et al. 2016). P2X4R expression has been detected in cultured keratinocytes, macrophages and sensory axon terminals (Inoue et al. 2007;Gaff et al. 2018;Moehring et al. 2018;Sadler et al. 2020), and CaMKK2 is expressed by monocytes/macrophages Gaff et al. 2018).
Here, we assess ex vivo expressions of CaMKK2, P2X7R and P2X4R and their association with nerve fibres in skin biopsies from the lower leg, donated by Indonesian PLWH with and without HIV-SN, and healthy controls (HC).

Participants and phenotyping
The study was approved by the Ethics Committee of the Faculty of Medicine, Universitas Indonesia (579/UN2.F1/ ETIK/2014), and validated by Curtin University (HR210-2015). Written and informed consent was obtained from all participants. HIV+ adults who had used ART for at least 12 months (median = 6.4 years; range = 1.2-11.7 years), but who had never been exposed to stavudine, were screened for neuropathy at POKDISUS HIV Care Clinic, Cipto Mangunkusumo Hospital, Jakarta, Indonesia, in 2016 . Individuals with a history of other conditions potentially associated with neuropathy or conditions preventing informed consent were excluded. Neuropathy was assessed using the Brief Peripheral Neuropathy Screen (BPNS). HIV-SN was defined as present with one or more lower limb neuropathic symptoms (pain, aching or burning, "pins and needles" or numbness) and reduced ankle reflexes or vibration sense at the great toe (128-Hz tuning fork vibration felt for 10 s or less). Neuropathy was not diagnosed in patients with only asymptomatic neuropathic signs, as the presence of both symptoms and signs better associates with impaired peripheral nerve function and pathology (Cherry et al. 2005;Lauria et al. 2010). Biopsies were collected from six individuals with HIV-SN (HIV-SN+) and six individuals without HIV-SN (HIV-SN−) to provide sufficient tissue to assess expression of P2X4R, P2X7R and CaMKK2 in triplicate from three individuals per group. Five control biopsies were provided by healthy, age-matched adults of South East Asian ancestry (HC) who declared no risk factors for HIV.

Microscopy
Samples were imaged using an inverted Nikon A1+ confocal microscope (Nikon Instruments, New York, NY, USA). Images were acquired using a digital scan resolution of 0.64 μM/pixel, pinhole of 1.2 airy units and pixel dwell 6.2 with 1024 resolution using a 20 × Plan Apo dry objective (N.A. 0.75; Olympus Corporation, Tokyo, Japan). Sequential scanning was completed using three lasers: 405 nm (450/50 filter), 561 nm (595/50 filter) and 640 nm (700/75 filter) to view nuclei, PGP9.5 + nerve fibres and P2X4R+ , P2X7R+ and CaMKK2+ cells, respectively. At least three images spanning 0.5 mm were acquired from every section across the full width of the epidermis in a z-series at 1.1-μm intervals (Nyquist settings) through the full thickness of the section. Z-series were combined for each section into a maximum intensity projection image with NIS-Elements Viewer (Nikon Instruments). Equivalent thresholds were applied across all images to assess PGP9.5+ nerve fibres (red) and P2X7R-, P2X4R-or CaMKK2-positive cells (green).
CaMKK2, P2X7R and P2X4R expression and intraepidermal nerve fibre density Expression patterns of CaMKK2, P2X7R or P2X4R and association with PGP9.5+ nerve fibres were investigated in a subset of HC (n = 3), HIV-SN− (n = 3) and HIV-SN+ (n = 3) donors selected randomly for each protein. A minimum of three images per section from three sections per donor were collected and expression was described and semi-qualitatively assessed by a single rater (JG) blinded to participant diagnoses. All CaMKK2+ and P2X7R+ cells were counted. Most cells in the basal layer of the epidermis were P2X4R+ and therefore were scored using a brightness scale: 1, very weak; 2, weak; 3, moderate; 4, strong; 5, very strong (Supplementary Table 1).
The IENFD was determined from three images from all HC (n = 5), HIV-SN− (n = 6) and HIV-SN+ (n = 6) donors. Images were available for an additional two HC, three HIV-SN− and two HIV-SN+ donors assessed for IENFD in 2017 using the same study criteria and protocols, and were included here (Mountford et al. 2018;Octaviana et al. 2018). All images were de-identified and PGP9.5+ nerve fibres were quantitated by three raters (JG, PK and PP) blinded to donor diagnoses. Quantitation was completed using standard IENFD protocols (Lauria et al. 2010) in which single fibres crossing the dermo-epidermal junction are counted and secondary branches excluded. Average counts were doubled to generate IENFD per square millimetre. Fisher's exact tests, Mann-Whitney tests and Spearman's correlations were used as appropriate to compare IENFD with donor characteristics using GraphPad Prism version 8.2.1 for Windows (Graphpad Software, La Jolla, CA, USA). Intraclass correlation coefficients were calculated (two-way random, single measures and absolute agreement) using the "irr" package (Gamer et al. 2010) in the R statistical environment (R Core Team 2010).

Intraepidermal nerve fibre densities correlated with nadir CD4 T-cells/μl but did not differ between donors with and without HIV-SN
Confocal images of PGP9.5+ intraepidermal nerve fibres were collected (Supplementary Fig. 2) and IENFD determined for all donors (Table 1). The intraclass correlation coefficient was 0.86 (95% confidence interval = 0.79-0.92), suggesting strong agreement between raters (data not shown). IENFD was significantly higher for HC than for those with HIV-SN (p = 0.02) and higher than for those without HIV-SN but this difference was not statistically significant (p = 0.07). IENFD was similar in HIV-SN+ and HIV-SN− donors (p = 0.19; Table 1). There were no differences in age, height, time on ART and current CD4 T-cells between HIV-SN− and HIV-SN+ donors (Table 1; Supplementary Table 2). However, nadir CD4 T-cell counts were lower in HIV-SN+ individuals than in those without HIV-SN (p = 0.05) and IENFD correlated positively with nadir CD4 T-cell counts (r = 0.67, p = 0.004; Supplementary Fig. 3). Age did not correlate significantly with IENFD (r = −0.36, p = 0.08). Thus, IEFND associates with the severity of HIV disease and is not a useful marker of neuropathy in our cohort.  Fig. 1c). CaMKK2+ cells were typically located close to dermal and epidermal nerves (yellow arrows) or co-located with dermal nerves (yellow box; Fig. 1c), so CaMKK2+ cells may interact with peripheral nerves in HIV+ patients.

P2X7R+ cells were observed in dermal blood vessels of HIV-SN− donors, but rarely in HC or HIV-SN+ donors
P2X7R was variably expressed across donor groups but was rare in sections from HC (medians of 1, 2 and 4 positive cells per donor; Fig. 2a; Supplementary Table 1). Expression was also rare in sections from HIV-SN+ donors, but a few P2X7R+ cells were observed in dermal blood vessels (yellow arrows) and some within the epidermis close to epidermal nerves (yellow box; medians of 4, 4 and 6 positive cells per donor; Fig. 2c). Sections from two of the three HIV-SN− donors exhibited an abundance of P2X7R+ cells in dermal blood vessels (medians of 18 and 20 positive cells per donor; Fig. 2b). Sections from the remaining HIV-SN− donor had moderately higher levels of P2X7R expression in blood vessels than HC and HIV-SN+ donors (Supplementary Tables 1 and 2, donor #11). P2X7R+ cells were observed close to epidermal nerve fibres in sections from all  (Fig. 2b). This may suggest upregulation by HIV disease, potentially as a mechanism that suppresses or prevents HIV-SN.  Table 1). P2X4R+ cells were often closely located to epidermal nerves and sub-epidermal nerve plexi (yellow arrows; Fig. 3a-c). Whilst P2X4R+ cells occurred in all donors, its expression in the basal layer of the epidermis was more abundant in sections from all three HIV-SN+ donors (median brightness scores of 4, 4 and 5; Fig. 3c). compared to HC (median brightness scores of 1, 2 and 3; Fig. 3a) and HIV-SN− donors (median brightness scores of 2, 2 and 3; Fig. 3b). The trend is consistent with a role for P2X4R in HIV-SN.

Discussion
We assessed PGP9.5+ IENFD and expression of P2X7R, P2X4R and CaMKK2 in skin biopsies from healthy controls, and HIV-SN− and HIV-SN+ Indonesians receiving stavudine-free ART for > 12 months. The median (range) IENFD for HC was 12.7 (7.4-17.3/mm 2 ), consistent with healthy individuals in published studies of other peripheral neuropathies (McArthur et al. 1998;Bakkers et al. 2009). IENFD was lower in HIV+ donors but similar in donors with and without HIV-SN (Table 1). We are not the first to show this. A longitudinal investigation of 150 HIV+ Thai individuals found no difference in IENFD between those with and without HIV-SN irrespective of the use of stavudine (Shikuma et al. 2015). Lower IENFD was recently associated with the presence of neuropathic pain at the biopsy site (Patel and Kamerman 2019). The prevalence of neuropathic pain in Indonesian HIV-SN+ patients is low (5/28; ) and pain was not reported by patients described here.
Age is negatively correlated with IENFD in healthy individuals (Bakkers et al. 2009) but has not been linked in HIV patients (Polydefkis et al. 2002;Phillips et al. 2014;Shikuma et al. 2015;Mountford et al. 2018). Here, HIV+ participants are relatively young (aged 23 to 47) and the age-associated decline in IENFD was not statistically significant (r = −0.36, p = 0.08). However, we identified a positive correlation between nadir CD4 T-cell counts and IENFD in the HIV+ groups (r = 0.67, p = 0.004; Supplementary Fig. 3). We previously associated lower CD4 T-cell counts at the time of assessment with greater risk of HIV-SN in Indonesians , and with lower nadir CD4 T-cells in HIV+ Africans (Gaff et al. 2020a). A lower CD4 T-cell count reflects greater severity of HIV disease, supporting a direct role for HIV itself in the degeneration of epidermal nerve fibres. Early initiation of ART is essential to minimise HIV disease, and may preserve IENFD. CaMKK2+ cells were identified in all HIV+ sections. CaMKK2+ cells were usually located close to or co-located with damaged nerve fibres and were more common in donors with HIV-SN (Fig. 1). CaMKK2 activates AMPK, a master regulator of cellular energy homeostasis. AMPK activation can replenish ATP supplies required for energy-intensive axonal growth by recruiting mitochondria to the site of repair (Sheng 2017). Therefore, close association between CaMKK2+ cells and nerve fibres may reflect CaMKK2-mediated neuronal growth and repair pathways. CaMKK2 is also expressed by macrophages and mediates inflammatory pathways Gaff et al. 2018). CaMKK2 activates CaM-KIV which in turn activates the p38-MAPK cascade and activation factor 1 (AP-1), inducing pro-inflammatory cytokines including TNFα, IL-1β and IL-6 (Ageta-Ishihara et al. 2009). It is plausible that CaMKK2+ cells may contribute directly to a reduced IENFD and HIV-SN via neuronal or inflammatory pathways.
P2X4R expression was observed in the basal layer of the epidermis, in blood vessels and closely located to epidermal nerves and sub-epidermal nerve plexi in all donors assessed (Fig. 3). However, P2X4R appeared upregulated in the basal layer of the epidermis in HIV-SN+ donors. Most cells in the basal layer proliferate and differentiate into keratinocytes which can express sensory receptors and produce neuroactive molecules that can illicit nociceptive responses in epidermal axon terminals of sensory neurons in response to noxious stimuli (Talagas et al. 2020). Rodent studies show that mechanical, cold and heat stimulation of keratinocytes produces ATP which activates P2X4R receptors on sensory neurons and results in behaviours associated with temperature stress and pain (Moehring et al. 2018;Sadler et al. 2020). As extracellular ATP can act on P2X4R in an "autocrine" fashion (Di Virgilio and Sarti 2018), the increased expression of P2X4R in the epidermis of HIV-SN+ donors could plausibly exacerbate this nociceptive pathway and contribute to neuropathic symptoms.
P2X4R is also a moderator of inflammation (Ulmann et al. 2008). Sciatic nerve injury in a rodent model of neuropathy upregulates P2X4R expression in spinal microglia leading to increased production of IL-1β, TNFα and IL-6  HIV-SN+ (c). P2X7R+ cells (yellow arrows) are rarely seen in sections from HC donors (a). P2X7R+ cells (yellow arrows) are abundant in dermal blood vessels and sometimes closely located to epidermal nerves of sections from HIV- SN-donors (b). P2X7R+ cells are occasionally seen in dermal blood vessels and closely located to epidermal nerves (yellow box) in HIV-SN+ donors (c). scale bar = 100 μm ◂ (Ulmann et al. 2008;Zhang et al. 2020). Activation of keratinocyte P2X4R is associated with production of IL-6 (Inoue et al. 2007). Like P2X4R, P2X7R is implicated in inflammation associated with neuropathic pain (Chessell et al. 2005). Interestingly, P2X7R was abundantly expressed in the blood vessels of HIV-SN− donors, but less in HC or HIV-SN+ individuals (Fig. 2). Expression of P2X7R may depend on the disease or stage, and tissue or cell type (Liu et al. 2016;Amadio et al. 2017). For example, P2X7R expression was upregulated in peripheral blood monocytes and lymphocytes from patients with neuropathic pain, but not patients with chronic nociceptive lower back pain (Luchting et al. 2016). Furthermore, P2X4R and P2X7R expression may be compensatory. P2X4R was upregulated in CD4 T-cells from P2X7R knockout mice in a model of heart transplantation (Vergani et al. 2013).
Our study has limitations. Firstly, we acknowledge the modest number of donors. However, a well-established protocol for determining IENFD (Cherry et al. 2005;Lauria et al. 2010) and assessment of IENFD by multiple raters blinded to donor diagnoses ensured reliable quantification. This was supported by an intraclass correlation coefficient of 0.86 (95% confidence interval = 0.79-0.92) indicating strong agreement between raters. Furthermore, expression of CaMKK2, P2X7R and P2X4R was sought in at least three sections per donor and in at least three images per section so at least nine images per donor were assessed. Further investigations in larger, longitudinal cohorts are required to confirm the expression patterns of CaMKK2, P2X7R and P2X4R but the findings described here were consistent between donors (Supplementary Table 1). Secondly, we utilised a simple clinical tool to diagnose HIV-SN. It is possible that HIV-SN− donors may have had sub-clinical peripheral nerve pathology or physiological malfunctions not detected by this tool. We have also investigated large and small fibre neuropathy diagnosed using established nerve conduction and stimulated skin wrinkling tests, respectively (Safri et al. 2020). IENFD did not differ between donors with and without large or small fibre neuropathies (0.28 and p = 0.62, respectively; Supplementary Table 3). No data pertaining to the duration of neuropathy were available due to the crosssectional design.
In conclusion, we have extended our genetic studies linking alleles of CAMKK2, P2X7R and P2X4R with HIV-SN, providing evidence of differential expression of the encoded proteins in affected tissues. Using multiple sections from three individuals per category, we showed that expression of P2X4R in epidermal basal layer cells was greater in HIV+ donors with HIV-SN, whereas P2X7R+ cells were more abundant in the blood vessels of HIV+ donors without HIV-SN. Moreover, tissue from donors with HIV-SN contained a few more CaMKK2+ cells than donors without HIV-SN, and most CaMKK2+ cells were located near or co-located with PGP9.5+ nerves. This would be consistent with a role for CaMKK2 in the repair of nerves damaged by HIV. Hence, the expression and location of CaMKK2+ , P2X7R+ and P2X4R+ cells support roles for these proteins in the pathogenesis of HIV-SN in patients treated with stavudine-free ART.
Our data suggests that CaMKK2 should be prioritised in studies of larger cohorts with follow-up to assess changes over time. Ideally, patients willing to undergo biopsies should also be selected by carriage of polymorphisms or haplotypes associated with risk or protection. The identification of proteins contributing to the development of HIV-SN is an important step in the development of targeted diagnostic and therapeutic strategies. We have provided some interesting candidates. Author contribution JG stained the sections and performed the microscopy, statistical analyses and preparation of the manuscript; FO recruited participants, collected medical data, conducted neuropathy assessments and collected biopsies; SL assisted with immunostaining and analyses; CJ provided sample preparation and microscopy expertise; PK provided statistical advice and assisted with analyses; JP provided histopathology expertise; JM assisted with the collection and processing of biopsies, and provided the additional images for IENFD analyses; PP assisted with analyses, checked the manuscript and coordinated the project.

Supplementary Information
Funding Open Access funding enabled and organized by CAUL and its Member Institutions.
Data availability Data are available from the authors upon request.

Declarations
Conflict of interest The project was funded by Curtin University and the Universitas Indonesia. No pharmaceutical grants were received.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are Fig. 3 Representative confocal images demonstrating intraepidermal expression of PGP9.5 (red) and P2X4R (green) in HC (a), HIV-SN-(b), HIV-SN+ (c). P2X4R expression is observed in the basal layer of the epidermis and dermal blood vessels of all donors but was increased in sections from donors with HIV-SN (c). P2X4R+ cells were observed in dermal blood vessels in all donors and located near dermal and epidermal nerves (yellow arrows). scale bar = 100 μm ◂ included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.