Controlling neuropathic pain by adeno-associated virus driven production of the anti-inflammatory cytokine, interleukin-10

Neuropathic pain is an especially difficult chronic pain syndrome to treat. No compounds are yet available that successfully resolve such pain [1‐3]. To date, drug therapies developed for human neuropathic pain have targeted neurons. However, evidence has recently accumulated that pathological pain, including neuropathic pain, is dynamically and dramatically amplified as a result of spinal cord glial activation [4‐6]. Spinal cord glia become activated as a consequence of inflammation and/or trauma to peripheral nerves [4, 6]. This raises the intriguing possibility that finding ways to target glial activation, or its downstream consequences, may provide a novel approach for neuropathic pain control [5, 7].

Several glial-selective drugs have been successful in preventing or reversing neuropathic pain in animals, but none is optimal for clinical applications. For example, fluorocitrate is a selective astrocyte inhibitor [8, 9]. While effective in blocking the induction of neuropathic pain in animals [10], fluorocitrate is inappropriate for human use due to inhibition of glial glutamate uptake and consequent seizures can occur [11]. Similarly, minocycline is a selective microglial inhibitor [12]. It too is effective in blocking the induction of neuropathic pain in animals [13, 14]. However, as minocycline fails to affect established neuropathic pain [13, 14], this compound does not appear to have therapeutic potential.

Other approaches have focused on the fact that glia are "immune cell-like". Upon activation, glia and immune cells each release pro-inflammatory substances, most notably pro-inflammatory cytokines (interleukin [IL]-1, tumor necrosis factor [TNF] and IL-6). The release of pro-inflammatory cytokines by activated spinal cord glia is key as these cytokines enhance pain and have been implicated in the initiation and maintenance of neuropathic pain [10, 15, 16]. Immunosuppressive (methotrexate) and immunomodulatory (propentofylline) drugs have been tested with the aim of suppressing neuropathic pain via suppression of glial-derived pro-inflammatory cytokines in spinal cord. While these drugs have proven effective in reducing enhanced pain responses [17, 18], they are not optimal for chronic use in humans, as their systemic administration would negatively impact the peripheral immune system. In addition, although selective pro-inflammatory cytokine antagonists have proven successful in resolving neuropathic pain [10, 15], their lack of CNS penetration negates systemic administration and their relatively short duration of action poses problems for chronic intrathecal administration in humans.

The purpose of the present series of studies was to explore a new approach to neuropathic pain control; that is, intrathecal gene therapy using an adeno-associated viral (serotype II; AAV2) vector to drive the production and release of interleukin-10 (IL-10), a powerful anti-inflammatory cytokine [19]. As IL-10 is known to suppress the production and release of all 3 pro-inflammatory cytokines [19], it would be predicted to be efficacious for the treatment of neuropathic pain. Two animal models of neuropathic pain were employed: (a) sciatic inflammatory neuropathy (SIN) induced by localized inflammation of the sciatic nerve in the absence of frank trauma, and (b) chronic constriction injury (CCI), a model involving both trauma to and inflammation of the sciatic nerve. Depending upon the intensity of sciatic nerve inflammation induced, the SIN model can produce either an ipsilateral or bilateral mechanical allodynia [20]. This allows examination of both ipsilateral and mirror-image pain. CCI was chosen for study both because it is a classic model of partial nerve injury [21] and because it induces thermal hyperalgesia in addition to mechanical allodynia [22‐24]. We attempted to prevent and reverse SIN- and CCI-induced pain changes, respectively. These studies demonstrate that intrathecal administration of an AAV2 vector that encodes for rat IL-10 (AAV2-r-IL-10) is effective in transiently resolving neuropathic pain in both models.

The present experiments document the efficacy of AAV2-r-IL-10 in preventing and reversing neuropathic pain. In addition, this work provides initial evidence that intrathecal gene therapy to express anti-inflammatory cytokines, such as IL-10, may be an approach worth pursuing for the treatment of chronic pain. When AAV2-r-IL-10 was administered intrathecally 3 days prior to chronic inflammation of the sciatic nerve (SIN), AAV2-r-IL-10 prevented the onset of both ipsilateral and mirror-image mechanical allodynia, as measured by the von Frey test. Blockade of these SIN-induced allodynias lasted for 8 days (11 days after intrathecal AAV2-r-IL-10), with allodynia developing by day 10 (13 days after intrathecal AAV2-r-IL-10). In contrast, rats receiving AAV2-GFP (Control) prior to induction of SIN exhibited strong ipsilateral and mirror-image allodynia throughout the timecourse tested. Such profound and prolonged allodynia with this chronic SIN procedure is in agreement with prior studies [10]. Neither AAV2-r-IL-10 nor AAV2-GFP altered mechanical response thresholds in sham-operated rats. AA2-r-IL-10 was also successful in reversing established CCI-induced thermal hyperalgesia. Intrathecal AVV2-r-IL-10 administered 10 days after CCI surgery returned hindpaw response latencies to radiant heat (Hargreaves test) to pre-surgery baseline values. The effect of AAV2-r-IL-10 was again transient, with complete reversal observed for a week, from 7 through 14 days after AAV2-r-IL-10. Thermal hyperalgesia then progressively returned with robust hyperalgesia recorded 20 days after AAV-r-IL-10. In contrast, rats receiving intrathecal AAV2-GFP exhibited stable ipsilateral thermal hyperalgesia throughout the timecourse tested. Neither AA2-r-IL-10 nor AAV-GFP altered thermal response thresholds in sham-operated rats. AAV-2-r-IL-10 attenuated established ipsilateral and mirror-image mechanical allodynia in these same animals, with allodynia again becoming fully re-expressed by 16–20 days after AAV2-r-IL-10. Rats receiving intrathecal AAV2-GFP exhibited marked bilateral mechanical allodynia throughout the timecourse tested, in agreement with prior studies [22‐24]. Again, neither vector altered response thresholds of sham-operated rats. AAV2 appears to predominantly infect meningeal cells surrounding the CSF space, as indicated by beta-galactosidase staining of spinal tissues from rats injected with AAV2-LacZ. This pattern of meningeal staining is in accordance with prior studies of intrathecal adenovirus administration [25].

IL-10 is only one of many endogenous anti-inflammatory cytokines. In addition to IL-10, the anti-inflammatory cytokine family also includes IL-4, IL-11, and IL-13 [28, 29]. Leukemia inhibitory factor, interferon-alpha, IL-6, and transforming growth factor-beta are categorized as either anti-inflammatory or pro-inflammatory cytokines, under various circumstances [29‐32]. Anti-inflammatory effects are also exerted by a variety of endogenous agents as well, such as IL-1 receptor antagonist, soluble and membrane-bound IL-1 decoy receptors, and soluble TNF decoy receptors [29]. Thus a number of anti-inflammatory substances exist which may potentially exhibit therapeutic effects for enhanced pain states.

IL-10 was chosen for the present study for several reasons. First, it is considered to be the most powerful anti-inflammatory cytokine, potently downregulating TNF, IL-1 and IL-6 production and release [19, 29, 33]. In addition, IL-10 can upregulate endogenous anti-cytokines and downregulate pro-inflammatory cytokine receptors [19, 29, 33]. Thus, it can counter-regulate production and function of pro-inflammatory cytokines at multiple levels. Second, simultaneous suppression of multiple pro-inflammatory cytokines, rather than targeting a single cytokine, has advantages for 2 reasons: (a) pro-inflammatory cytokines are redundant, such that blockade of a single pro-inflammatory cytokine results in its functions being taken over by other pro-inflammatory cytokines [34], and (b) TNF, IL-1 and IL-6 can vary greatly in their relative magnitude of production, dependent both upon the inciting stimulus and time. This has been observed in spinal cord under conditions of pain facilitation as well [35]. Third, acute administration of IL-10 protein has been documented in previous studies to suppress the development of spinally-mediated pain facilitation in diverse animal models, including intrathecal dynorphin, peri-sciatic snake venom phospholipase A2, and spinal cord excitotoxic injury [36‐40]. As IL-10 has a very short half-life (~2 hr) in rat cerebrospinal fluid (L. He, R. Chavez and K. Johnson, unpublished observations), IL-10 gene therapy may provide an efficient means of attaining neuropathic pain control across days. Lastly, evidence to date indicates that spinal cord neurons do not express receptors for IL-10 under either basal or inflammatory conditions [41]. Therefore, in spinal cord, IL-10 may selectively target glia, without disrupting neuronal function. Indeed, the major reported effect of IL-10 on neurons is enhancement of neuronal survival, an effect thought to be indirect via the inhibition of glially-derived neuroexcitotoxic products, including pro-inflammatory cytokines [42‐45].

Replication-defective adeno-associated viral vectors offer numerous advantages for use in human gene therapy. This vector rarely inserts into host DNA, thus avoiding insertional effects common to other gene therapy approaches [46]. In addition, AAV is less inflammatory than other gene delivery vectors, such as adenovirus [47]. For such reasons, adeno-associated viruses have recently attracted attention as vectors for human gene therapy [48].

However, the present report found the efficacy of AAV2 to be short-lived upon intrathecal administration. This was surprising, given the generally accepted long-term persistence of AAV transgene expression. When AAV is administered directly into spinal cord tissue or dorsal root ganglia, AAV-directed gene expression persists for at least 4–8 months [49, 50]. Since AAV injection into spinal cord or dorsal root ganglia requires traumatic surgery to expose the site, such approaches are not optimal for human application. Our intent is to explore therapies with potential use in humans, and therefore an acute intrathecal injection via percutaneous lumbar puncture route of administration was used. It is possible that AAVs of different serotypes and promoters may produce far more long-lasting effects following intrathecal administration, as serotype and promoter tropism of AAV can greatly affect what cell type(s) are infected and/or express the transgene [50‐53]. Indeed, our ongoing studies suggest that far longer IL-10 protein expression can be attained by altering the AAV serotype used for intrathecal gene therapy (T. Liu, R. Chavez, and K. Johnson, unpublished data). Thus, while the results with AAV2 reported in the present studies were transient, they do indicate the potential for intrathecal IL-10 gene therapy for pain.

The conclusions from the present experiments are clear. First, intrathecal IL-10 gene therapy shows promise for control of neuropathic pain. This is exciting as it demonstrates that targeting products of spinal cord glial activation can produce prolonged suppression of both mechanical allodynia and thermal hyperalgesia. Second, neuropathic pain was not only prevented but also reversed by intrathecal IL-10 gene therapy. This is important, as it implies that spinal cord glial activation contributes to both the initiation and maintenance of neuropathic pain. Third, intrathecally administered AAV2 targets primarily meningeal cells. This may be advantageous, as infection of meningeal cells avoids retrograde transport of AAV to distant sites, as has been observed in brainstem neurons following intra-spinal cord injections [25]. Lastly, it is anticipated that the duration and potency of AAV-IL-10 gene therapy can be improved by careful selection of the optimal AAV serotype.

A previously undetected point mutation, resulting in an amino acid change (F129S), has been identified in the rat IL-10 gene expressed in this study. This mutation lies outside of identified receptor binding regions. Its effect on IL-10 bioactivity in vitro, if any, is currently under investigation