Arthritis Gene Therapy using the Interleukin-1 Receptor Antagonist

In its most basic form, arthritis gene therapy rests on a simple concept. Genes, or more pragmatically cDNAs, encoding anti-arthritic products are introduced into the joint by intra-articular injection [1]. Therapeutic gene products are produced within the joint in a sustained and potentially regulated fashion, thereby overcoming the major barrier to intra-articular therapy, namely the short dwell time of molecules within joints [2]. Although this concept has been around for over 25 years, its clinical implementation has been long and tortuous [3].

Two key questions for intra-articular gene therapy are which gene(s) to transfer to the joint and which gene delivery strategy to use? Osteoarthritis (OA), the form of arthritis studied in the paper by Nixon et al in this issue of Arthritis and Rheumatology [4], is a complex, polygenic condition. Although OA has high heritability, many genes make small contributions and there is no single, generally accepted genetic target, unlike the case with a classical Mendelian disorder. However, it is possible to make a decent case for targeting interleukin-1 (IL-1) by expressing its natural inhibitor, the interleukin-1 receptor antagonist (IL-1Ra). IL-1 is produced by cells within joints with OA, and it mediates many of the pathophysiologic processes occurring in and around such joints including cartilage loss, inflammation, osseous changes and pain. IL-1Ra is small, safe, native and has no agonist activity or other confounding properties. Its entire coding sequence can be comfortably packaged into practically any vector.

Data from a number of different animal models, stretching back over two decades, support the delivery of IL-1Ra cDNA to joints as a means of combatting OA (Table 1). Depending on the animal model, IL-1Ra inhibits the degenerative, osseous, inflammatory and symptomatic components of OA and a clinical trial delivering IL-1Ra cDNA to the knee joints of patients with OA is expected to start later this year (Clinicaltrials.gov Identifier NCT02790723). IL-1Ra cDNA has already been transferred to human joints in the setting of rheumatoid arthritis (RA) [5, 6].

Table 1.

IL-1Ra cDNA delivery in animal models of osteoarthritis

Animal Model	Vector	Reference
Canine ACL transection	Retrovirus, Ex vivo	[14]
Horse osteochondral defect	1st generation adenovirus, In vivo	[10]
Rabbit MCL-meniscus resection	1st generation adenovirus, In vivo	[15]
Rat ACL-meniscus resection	AAV, In vivo	[16]
Horse osteochondral defect	AAV, In vivo	[17]
Horse osteochondral defect	Helper-dependent adenovirus, In vivo	[4]

The novelty of the paper by Nixon et al. resides not so much with the concept or the transgene, but with the vector used for gene delivery. Reflecting the technology of the time, initial research into arthritis gene therapy explored ex vivo gene delivery using autologous cells transduced with retrovirus vectors. Although success was recorded in animal models and two small human trials in RA ensued [5, 6], this technology proved cumbersome and expensive, and ran the risk of insertional mutagenesis from retroviral integration. One solution to these limitations has been to use allogeneic cells from a universal donor and to irradiate the transduced cells prior to injection into the joint. This approach has led to the development of Invossa® which uses a line of chondrocytes transduced with retrovirus to express large amounts of transforming growth factor-β₁. This product has been approved in Korea for the treatment of OA and is about to start Phase III trials in the US [7].

In vivo gene delivery to joints is a simpler and more expeditious strategy, but it has been a challenge to identify clinically suitable vectors for this purpose. Adeno-associated virus (AAV) has emerged as a popular vector now that technology provides recombinant, self-complementing AAV containing a double stranded DNA genome that removes the barriers previously imposed by the single-stranded DNA genome of the wild-type virus. Two clinical trials using AAV to deliver genes to joints have been completed; one more is in progress and another is scheduled to begin later this year [1].

Nixon et al advance the field by evaluating a novel vector based upon adenovirus. This virus was the first to be tested experimentally as a vehicle for in vivo gene transfer to joints [8]. The vectors used in such experiments proved to be powerful vectors in joints, delivering genes very effectively to synoviocytes, but not chondrocytes [9]. However, these “first generation” adenovirus vectors continue to express low levels of adenoviral proteins, which are highly antigenic. This curtails transgene expression within about a month as the immune system destroys transduced cells. Nevertheless, such vectors have proved very useful experimentally to screen potential anti-arthritic genes and obtain proof of concept in animal models. Notably, a first generation adenovirus vector was used to demonstrate the anti-arthritic properties of IL-1Ra cDNA in the same equine model of OA as used here [10].

Building upon these observations, Nixon et al report the use of an improved adenoviral vector that lacks all viral coding sequences and thus does not provoke cell-mediated immune responses to virally transduced cells. The genome of adenovirus is approximately 36 kb in size. After removing all viral coding sequences, vectors contain only the noncoding termini of the viral genome and can deliver large DNA fragments of up to 36 kb into target cells. Because these viruses can no longer replicate, their production requires proteins delivered in trans by so-called helper virus; hence the name “helper-dependent adenovirus”. These vectors, also known as “high capacity”, “gutted”, “gutless” or “third-generation” adenoviruses, are very useful when it is necessary to deliver large or multiple transgenes and their regulatory elements. Because they provoke no cell-mediated immune response, extended periods of transgene expression can be achieved. The vector used in the present study places IL-1Ra transgene expression under the transcriptional control of a NF-κB response element. This ensures that the level of IL-1Ra production reflects the amount of inflammation in the joint. A similar promoter is being used in a clinical trial in the gene therapy of RA to control the level of interferon-β transgene expression after intra-articular delivery with AAV (Clinicaltrials.gov Identifier NCT02727764).

In the present work, helper-dependent adenovirus was used to deliver murine and equine IL-1Ra cDNA into the knee joints of mice and horses, respectively, with surgically-induced models of OA. In agreement with data from previous studies (Table 1) using retrovirus, first generation adenovirus and AAV in this context, the authors confirmed safety and specific anti-arthritic effects in both models. But there were interesting differences, depending on age, species, surgical model and, possibly, target joint.

The murine model involves transection of both cruciate ligaments of the knee, provoking a severe and rapidly progressing form of experimental OA. In 8-week old, skeletally immature mice, intra-articular injection of the vector preserved cartilage and reduced osteophyte formation but, surprisingly, had had no effect on synovitis. In 12-week old, skeletally mature mice a ten-fold higher dose of the vector also improved cartilage scores, but had no effect on osteophyte formation or synovitis; however, one measure of pain was reduced.

The equine model targeted the middle carpal joint and involves the creation of an osteochondral defect followed by exercising the horse on a treadmill. This induces a milder form of OA than the murine model, and one that progresses more slowly. Data from the equine model were more striking. Intra-articular injection of the vector reduced synovitis, effusion, osteophytes and lameness while increasing range of motion and preserving cartilage. That statistical significance was achieved with only 4 horses per group suggests a very high effect size.

Safety is an overriding concern when using gene therapy for non-lethal diseases such as OA. Several detailed studies, the present one included, find no evidence of persistent viral genomes outside the injected joint, and no major adverse events have been reported, thus providing reassurance on this issue [4, 11, 12].

Overall, the data of Nixon et al strengthen the case for using IL-1Ra as a transgene in the intra-articular gene therapy of OA. Whether high capacity adenovirus will emerge as the vector of choice in clinical development will be determined by a number of additional factors including cost and ease of manufacture. Another question is whether a robust, sustained anti-OA effect will require transduction of chondrocytes. Detailed in vitro studies by Grodzinsky and colleagues [13] have shown that particles larger than about 10 nm, depending on shape and charge, do not penetrate the extracellular matrix of normal cartilage. However, AAV, which is approximately 20 nm in size, is able to transduce chondrocytes in situ after injection into equine joints [12]. This may be enabled by the pumping action occurring in cartilage as the horse moves. Adenovirus, which is over 100 nm in size, is unable to penetrate the extracellular matrix of normal cartilage [9]. However, as OA progresses, proteoglycans are progressively lost, thereby allowing larger particles to gain access to chondrocytes.

Although vectors such as helper-dependent adenovirus and AAV do not express viral proteins in transduced cells, the virions themselves are antigenic. This may hinder gene transfer in individuals who have pre-existing neutralizing antibody to the serotypes commonly used for vector development, adenovirus5 and AAV2. Even when novel serotypes are used they will provoke a primary immune response which could prevent re-dosing. Intra-articular gene delivery has the advantage of injection into a defined cavity where simple lavage or local, transient immunosuppression may be sufficient to overcome immunologic barriers to gene transfer.

Another consideration of relevance to the entire field is the applicability of the animal models used for pre-clinical development of novel therapies for OA. Nearly all models are acute and follow surgical injury to the joint. They may be pertinent to the early treatment of post-traumatic OA, but their relevance to established disease and idiopathic OA is unknown. Nevertheless, the paper by Nixon et al adds to a growing body of evidence that permits optimism concerning the development of genetic medicines for OA which are more effective than existing therapies and possibly disease-modifying.