In this section, the review will focus on two transgenes that appear particularly promising for translating AAV gene therapy for pain: prepro-ß-endorphin (ppßE) and interleukin-10 (IL-10). The therapeutic efficacies of these two genes have been investigated and validated in multiple vector systems including AAV by several independent groups including ours [61, 72–74]. The ppßE gene was originally developed by us for the use in pain gene therapy . A number of other analgesic genes have been reported in a variety of vector systems and might be alternative candidates for the clinical translation of AAV for pain, whereby the pharmacological assessment will follow similar principles as laid out here for ppßE and IL-10.
Dose–response relationship and therapeutic index: Transgene as a drug
Effects of the transgene products can be assessed by applying the principles established in pain pharmacology for conventional (non-gene therapy) drugs. Most analgesic drugs, including the peptide ß-endorphin (ßE, considered as a gene-encoded version here), show dose dependence in regards to both their analgesic- and toxic effect. Low doses of analgesic drugs thereby exert low (if any) pharmacological effect and dose escalation results in an analgesic effect and eventually, if doses exceed a certain threshold, the gradual emergence of toxicity. The drug level needed to produce the toxic effect relative to a therapeutic effect is the therapeutic index.
The emphasis on the dose–response relationship and the therapeutic index in pain pharmacology provides important guidance in regards to the development of AAV for pain. A certain degree of toxicity may be expected at high-therapeutic doses and may be not only acceptable but even the goal of dose-escalation in order to optimize the therapeutic benefit. The rationale for delivering the ppßE gene by AAV is the expectation of a favorable therapeutic index. The basis for its safe development will be the appropriate choice of the initial vector dose in a phase I/II trial.
Tissue-specific transgene expression: improving the therapeutic index
Conventional analgesic drugs, such as opioids, administered orally or intravenously expose all sites of the body to similar drug concentrations. In contrast, AAV gene therapy for pain by the IT or IG route aims at targeting anatomical sites known to mediate analgesic efficacy of opioids but not toxicity: the spinal compartment of the nervous system including the DRGs and the posterior horn of the spinal cord.
IT delivery improves the therapeutic index of conventional opioids. This principle is firmly established through animal studies [11, 12] and the use of IT pump-mediated delivery of opioids in patients with the most severe pain states . Vector systems that were initially used in gene therapy for pain, such as adenovirus [77, 78] or plasmids [74, 79] transduced only meningeal lining fibroblasts, which secreted the transgene product into the spinal CSF emulating an IT pump.
An important observation in the field of AAV for pain, however, has been the marked tropism of AAV for DRG neurons if administered IT or IG. This was a very favorable, while unexpected, finding and has subsequently been confirmed by several groups in different animal models [16, 18, 59–61, 80–82]. It has an important promising implication for the pharmacology of AAV for pain: the potential to not only match the therapeutic index of IT opioids but to improve further upon it. AAV-mediated transgene pharmacokinetics can be formalized as follows. If the transgene encodes a secreted protein such as ppßE and IL-10, the protein can reach the target cells by two routes: (i) through release into the CSF, i.e., emulating an IT pump, and (ii) in an autocrine/paracrine fashion within the immediate surrounding of the transduced cells. While the ratio of biological effects (ii)/(i) cannot be directly measured in animal models, it may be an important consideration when assessing the viability of AAV gene therapy for pain explicating why the therapeutic index of AAV will likely be improved not only over systemic opioids but also when compared with IT pump delivery of opioids.
Of note, a high degree of DRG-neuron selectivity has been the fundamental rationale for the HSV-based approach to pain gene therapy by the subcutaneous route of vector administration [4, 83]. IT AAV may be less selective for DRG neurons than subcutaneous HSV, but will likely improve the therapeutic index over IT delivery of conventional opioids (as discussed above). Furthermore, the IG route will very likely yield the most DRG neuron-selective transgene activity achievable with AAV potentially matching the DRG neuron-selectivity of subcutaneous HSV.
Specific consideration for ppßE
ßE, the secreted peptide product of ppßE, is a μ-opioid receptor agonist and thereby shares its pharmacological activity with the most commonly used conventional opioids such as morphine. The toxicological assessment of ppßE can therefore be guided by the robust clinical experience with exogenous opioids. In addition, ßE was tested in clinical trials.
The principal sites where opioid toxicity occurs at high doses are the CNS and the periphery. The most common adverse effects of opioids in the CNS are nausea and vomiting, mental status changes, pruritus, urinary retention, and respiratory depression [84–87], while the peripheral adverse effects include constipation, cardiovascular adverse events, derangement of hypothalamo-pituitary axis, and immune-modulation [87–91]. Both the CNS and peripheral adverse effects are dependent on the drug concentration at a given site and may be reversed by naloxone, a μ-, κ-, and δ-opioid receptor antagonist.
When IT administration of ß-endorphin was tested in humans, it resulted in profound analgesia similar to that attainable with exogenous opioids [92–97]. Dose escalation led to a toxicity profile that would be expected with an IT opioid drug, whereby doses of IT ßE of up to 7.5 mg resulted in miosis, drowsiness, and elevation plasma prolactin levels in some patients. No systemic adverse effects related to the treatment were found in those studies, reflecting the inability of ß-endorphin to cross the BBB, keeping it from escaping from the IT space to the systemic circulation.
In conclusion, the therapeutic and toxic effects of ppßE are expected to be vector dose-related and the specific toxicities can be predicted from the previous clinical experience with IT opioids and ßE. What remains to be defined in the preclinical large animals studies is the maximum tolerated dose (MTD) of the vector encoding ppßE for both the IT and the IG route. A phase I/II clinical would then test safety for vector doses starting with the lowest potentially efficacious dose and sequentially being escalated towards the MTD established in animals.
Specific consideration for IL-10
IL-10 is of interest for pain therapy because it may be the first/only available agent to target a mechanism of chronic pain first described about a decade ago, glial activation . Recombinant IL-10 administered IT has been found to be efficacious in animal models of chronic pain [99, 100]. The use of recombinant protein, however, has not been tested in humans due to its short half-life of IL-10 in the CSF, its inability to cross the BBB, and the impracticality of continuous IT delivery over a prolonged period of time . A pharmaceutical formulation of IL-10 needed for delivery by an IT pump, i.e. the stability at ambient temperature and the high concentration, has proved to be difficult. Therefore, gene therapy may provide the best (or only) strategy for testing targeted spinal delivery of IL-10 for pain in the clinical setting.
While toxicity of IL-10 delivery targeted to the PNS by either the IT or the IG route has not been addressed in large animal models, rodent data suggest good tolerability of IL-10 in the CSF [61, 74, 101, 102]. Furthermore, systemic toxicities of IL-10 have been thoroughly characterized in trials of recombinant IL-10 for Crohn disease, rheumatoid arthritis, Wegener granulomatosis, and psoriatic arthritis [103–107]. Safety studies showed that a single intravenous injection of up 100 μg/kg IL-10 in healthy humans led to transient, dose-related flu-like symptoms accompanied by neutrophilia, monocytosis, lymphopenia, and down-regulation of pro-inflammatory cytokines IL-1 and TNF-α . Long-term intravenous or subcutaneous administration of IL-10 in rats and NHP confirmed hematological effects of IL-10 and also showed its inhibitory effect on T-lymphocytes .
Establishing the MTD for AAV encoding IL-10 will be an important goal for both the IT and the IG route. Clinical testing would then start at a lower vector dose with subsequent dose escalation to establish the MTD in humans.
Transgene-specific immune response
Syngeneic transgenes will not induce serological or cellular immunity. Therefore, no immunity is to be expected in a clinical trial using human IL-10 as therapeutic gene and no special studies appear to be necessary towards assessing it.
The case of ppβE may require additional consideration, because the gene includes an artificial fusion site. Specifically, the pre-pro sequence of ppβE (the amino-terminal “pp” part of the protein) is taken from nerve growth factor (NGF); thus, this portion of the gene by itself is not of concern because it can be made syngeneic (by taking the pre-pro sequence from human NGF). The carboxy terminal portion of ppβE gene encodes the pharmacologically active βE peptide. By itself this portion of the gene is not of concern either, because it can also be made syngeneic (by using the βE sequence as it occurs in the precursor protein human pro-opiomelanocortin (POMC)). In the ppβE gene, these two syngeneic peptides are fused creating a potentially non-syngeneic epitope.
The rationale for using ppβE rather than POMC is to circumvent the requirement of POMC to be processed in acidified secretory vesicles found only in certain cell types and its reliance on the regulated secretory pathway. ppβE can be processed into βE by all cell types and secretion of βE occurs via the constitutive secretory pathway yielding continuous release of the pharmacologically active βE peptide, which is the desired property of this therapeutic gene. Because the processing of the ppβE peptide into its syngeneic components, the pre-pro-sequence and βE, occurs immediately following its ribosomal synthesis and co-translational transport into the ER, the fusion epitope may not even have a chance to become immunologically apparent. Therefore, to evaluate the possibility of cellular immune toxicity, pre-clinical toxicity studies of ppβE will have to assess the integrity of transduced tissues such as neuropathological evaluation of DRG.
A challenge for pre-clinical toxicology assessments is whether to perform animal testing with syngeneic transgenes, i.e., specifically adapted to the animal species employed, or to perform testing of the human-type vector slated to be used in a clinical trial, i.e. xenogeneic to the animal model. Xenogeneic transgenes are well-tolerated in rodents, which express even the bacterial EGFP gene long term. However, xenogeneic transgenes induced immunity in large animals . As a consequence, an AAV vector encoding a species-specific, i.e., syngeneic transgene could be used for preclinical toxicology studies of AAV for pain. The vector that is ultimately used in the clinic could also be given to animals, but the latter type of studies should be limited to effects observable within approximately one week because all later effects can be expected to be confounded by a xenogeneic immune response that is an artifact of the testing scenario and not applicable to the anticipated scenario of a clinical trial.