Magy Immunol/Hun Immunol 2004;3(1):5-28.
Gene therapy as a treatment for rheumatoid arthritis
JAMES M. WOODS, PhD
Department of Microbiology and Immunology, Chicago College of Osteopathic Medicine, Midwestern University, Downers Grove, IL 60515.
Phone: (1-630) 515-6173, fax: (1-630) 515-7245, e-mail: firstname.lastname@example.org
A clear understanding of the pathogenic events and/or environmental conditions that lead to the development of rheumatoid arthritis has not been accomplished. In recent years, some of the most capable therapies have targeted individual proteins, such as proinflammatory cytokines, which contribute to persistent inflammation. The success of these therapies in some patients underscores the importance of having a solid pathophysiologic knowledge of the mechanisms at play in the diseased joint. Targeting the joint therapeutically with proteins or other agents has presented many challenges in the treatment of rheumatoid arthritis. To circumvent these obstacles, the idea of providing transgenes to cells of the synovial lining was born. This use of gene therapy, as a delivery vehicle rather than replacement of a genetic deficit, has had many successes in preclinical animal studies. Preliminary results of the first Phase I clinical trial in humans suggests that an ex vivo approach can be safe and enable transgene expression.
This review provides a consolidated overview of many of the successful gene therapy strategies undertaken for the treatment of animal models of arthritis. The focus is on: 1. joint targeting strategies, including discussion on the local and systemic approaches as well as the contralateral joint; 2. the applicability of viral vectors, including comparison of adenoviral, retroviral, adeno-associated, and herpes simplex viruses; 3. timing and dosage of treatment; and 4. targets and candidate proteins that have been examined, including targeting proinflammatory cytokines or the use of anti-inflammatory cytokines.
rheumatoid arthritis, gene therapy, cytokine, animal model, inflammation
Érkezett: 2003. szeptember 30. Elfogadva: 2003. október 15.
Persistent inflammation is a characteristic of the joints of many patients suffering from rheumatoid arthritis (RA), a chronic autoimmune disease affecting the synovial tissue1. While significant advances have been made in our understanding of the mechanisms at play at the cellular and protein levels throughout the disease course, no studies have elucidated the elements that cause RA. One way to capitalize on mechanistic knowledge that already exists about factors which regulate inflammation, is to alter the cytokine balance in a manner which reduces inflammation2. This can be achieved by delivery of biologic agents including monoclonal antibodies, recombinant cytokines, cytokine inhibitors, or soluble receptors3. A caveat of using proteins in the treatment of any disease is that they are efficiently broken down during digestion which makes oral delivery problematic. To circumvent this barrier, and other problems associated with targeting the joint in vivo, the concept of using gene therapy was proposed4. Many studies have validated the concept of using transgenes to treat joint diseases through the use of animal models of arthritis. The approaches most commonly taken have aimed at inhibiting the proinflammatory cytokines interleukin (IL)-1b and tumor necrosis factor (TNF)-a, or have introduced anti-inflammatory cytokines that possess immuno-regulating properties, such as IL-4, IL-10, or IL-13. In addition to preclinical studies in animals, the first U.S. Phase I clinical trial for RA patients has been completed, and the preliminary results suggest that the method is feasible for delivery of transgene products in human joints57. Thus, continued advances in this field may lead to the development of more safe and cost-effective protein therapies that can effectively reduce inflammation and possibly repair joint tissue.
Approaches to targeting the joint
Protein and other molecular movement from synovial capillaries to synovial tissue (ST) progresses by the slow rate of passive diffusion, and is further hindered with increased molecular size8, 9. This leaves typical methods of drug delivery, such as oral and intravenous routes, at a significant disadvantage when it comes to targeting the RA ST. When this limitation is side stepped by direct intra-articular (i.a.) injection into the joint space, efficient joint clearance presents another problem, resulting in a short molecule half-life. In addition, if vascular perfusion in the joint is relied upon following traditional delivery of a biologic agent, there is a significant likelihood that non-target organs will be exposed to high levels of the agent, resulting in undesired side effects. With these obstacles in mind, dr. Christopher Evans proposed the use of gene therapy in the treatment of RA as a novel delivery mechanism4. In addition to the gratifying concept of having the diseased synovium produce its own therapeutic agents, this approach could significantly reduce the exorbitant costs associated with producing and delivering biologic agents, because the gene can conceivably be expressed into its gene products millions of times10.
When devising a strategy of how to obtain transgene expression in the RA joint, two approaches, a local and systemic approach have been conceived11. Each approach can be performed with either a direct in vivo injection, or by removing cells from the body, infecting them ex vivo, and then re-implanting the cells. Here, local and systemic transgene administration is discussed, as well as the contralateral effect, or the observation that treating one joint reduces the disease course in the non-treated contralateral joint.
Local gene therapy
Many animal studies, using a variety of transgenes, have suggested benefits in models of arthritis when using a local gene therapy approach1247. In this approach, the transgene is usually introduced by a local injection to the joint, typically by i.a. or peri-articular (p.a.) injection. The goal is to genetically modify cells within the joint to manufacture a therapeutic agent that will inhibit the pathogenesis of RA. The synovium is, more often than not, the target of this approach, because it has a large surface area and is immediately adjacent to the synovial fluid (SF) with no membrane blocking its ability to sequester vectors from the SF47. A chief benefit of this method is that the afflicted joints can be injected directly, which should allow for the highest production of transgene products to be made in close proximity to where they need to act, thus exposing irrelevant organs to much lower concentrations48.
The approach of using a local and direct i.a. injection in an animal model of arthritis is shown at the top of Figure 1. The background shows a hematoxylin and eosin stain of a tissue section taken from the hind ankle of a Lewis rat with adjuvant-induced arthritis on day 18, a time in which the erosive pannus can be seen invading the bone. The arrow demonstrates that an ideal i.a. injection would introduce the transgene into the space which houses the SF, allowing for optimal uptake by the surrounding synovial lining. Using the direct injection approach, the concern of toxicity to the liver and kidneys, which is often a factor with other treatments of RA, is anticipated to be lowered. Offsetting this advantage is the fact that RA is a systemic disease which often affects multiple joints. Thus, treatment through local direct injections would likely require multiple injections and may not be a satisfactory method for treating the systemic component of the disease. Since this method is relatively quick and more easily accomplished in comparison with the ex vivo methodology described below, it has often been used to screen the efficacy of transgene products in animal models of RA. Some of the gene products that have shown success using a direct local injection include those which inhibit angiogenesis36, 37, those which aim to induce synoviocyte apoptosis3335, and those which target pro-inflammatory cytokines3847 or introduce anti-inflammatory cytokines2332. In general, soluble transgene products have been more commonly studied than gene products which remain intracellular. When producing a soluble transgene product, lower infection efficiency may be tolerable if the cells infected are producing a large amount of the soluble mediator. Conversely, when the gene product is retained intracellularly, higher infection efficiency would be required for the gene products to have their effects on many cells. In the latter case, when the intracellular gene product may induce apoptosis, a lower efficiency may be allowable if there is a considerable amount of bystander effect, which is the ability of an infected cell have effects on surrounding cells.
In addition to the direct in vivo strategy described above, an alternative ex vivo approach has also been investigated. As shown in the bottom of Figure 1, in this approach, synovial cells can be removed from the joint during surgery, infected outside the body, expanded in culture, screened for undesired agents and/or non-infected cells, and finally re-injected into the joint. Following re-introduction to the joint, synoviocytes re-colonize into the lining while expressing their new transgene. From a safety perspective, the ex vivo methods were the natural choice for the first studies in human trials, since the genetically modified synoviocytes could be tested extensively prior to being put back into the joint. In addition, there is a certain level of control regarding the number of modified cells, and the type of cell that will be re-injected. Another benefit to this method is that free virus is never directly introduced into the patient, thereby lowering the threat of an adverse immune response to the vector used. Should a transgene disrupt the growth properties of the cells collected for infection, there is the possibility that a change in cell phenotype and/or growth properties will be recognized before re-introduction of the cells. It should be noted, that even though the ex vivo methodologies are clearly better suited for treatment of human patients (with the current vectors available), the time and costs are considerably higher when compared to a direct in vivo approach. Therefore, none of the existing vectors is ideal for the treatment of a large volume of patients49.
Systemic gene therapy
The goal of the systemic gene therapy approach is to introduce transgenes that will produce gene products that enter the circulation to modify the disease process. Alternatively, cells introduced systemically may be designed to target the joint and have a local effect after a systemic delivery. The advantages to a systemic approach include the ability to affect the component of RA that is systemic and to have effects on numerous organs simultaneously. Rather than introducing transgenes whose products may be beneficial locally in the synovium, such as anti-inflammatory cytokines, a transgene may be chosen whose product will target the major secondary lymphoid organs, such as the spleen and the lymph nodes. The disadvantages to this method include the strong possibility that non-target organs will be exposed to high concentrations of transgene products which may have non-desirable or even toxic effects. Nonetheless, preclinical studies in animal models using a systemic approach have been accomplished using an assortment of transgenes with a variety of levels of success22, 23, 28, 5052. In addition, the use of cells which target a specific organ or are encapsulated for protection from elimination by the immune system have also demonstrated success21, 5359.
The contralateral joint
A number of years ago, the observation was made that local treatment of one joint of a rabbit with antigen-induced arthritis conferred protection to the non-injected contralateral joint44. This study examined two soluble receptor fusion proteins which bind IL-1 and TNF. The most impressive effects were conferred when both transgenes were expressed44. However, no transgene products were detected in the non-injected joint, suggesting that the contralateral effect could not be explained by soluble receptor transport in the circulation. Follow up investigations with viral IL-10 (vIL-10) and marker transgenes also examined this contralateral effect26. Using the gene for the marker green fluorescent protein (GFP), adenovirally infected cells were suggested to traffic from the joint in which they were introduced to the contralateral joint. Lavage fluids, taken from injected and non-injected rabbit ankles and analyzed by flow cytometry, showed 20% GFP positive cells in injected joints compared with 3% GFP positive cells in the contralateral joint. Subsequent fluorescent microscopy showed multiple cell types as GFP positive in the injected joint while only cells of a single morphology appeared positive in the contralateral joint. Preliminary inspection hinted that these cells may be antigen presenting cells (APCs) such as macrophages or immature dendritic cells. Also noticed in this study was that cell trafficking may depend upon the amount of inflammation initially present in both joints at the time one is injected, as well as the quantity of transduced cells in the joint injected26. Subsequent studies used murine models of delayed-type hypersensitivity because of the presence of a contralateral effect in combination with the availability of murine reagents60. One theory proposed to explain the contralateral effect postulates that both synovial lining cells and APCs are transduced by the viral agents used6. Antigen presentation, which takes place after transduction, will occur in the presence of a high concentration of the transgene product, which in many of these studies is a cytokine (like vIL-10 or IL-4) which can influence the maturity of the T cells produced. Such an occurrence is likely to influence the production of a T helper 2 (Th2) subset of cells which may act locally and after trafficking to the contralateral joint. In addition to trafficking by these newly produced T cells, the APCs originally transduced by the viral agents may also make their way to the contralateral joint where they could influence T cell production6. In support of this theory, dendritic APCs which have been modified genetically to produce IL-4 and administered systemically, can prevent or reduce murine collagen-induced arthritis (CIA)61, 62. Dendritic cells resided in the liver, spleen, and lymph nodes and nearly abolished the disease61. Other cells, including fibroblasts or T cells which were genetically modified to produce IL-4, were not able to modify the disease62.
Vectors used in delivering genes to cells of the joint
Vectors which assist in the cellular uptake and expression of a transgene can be of a viral or non-viral origin63. Each vector has particular strengths and weaknesses dependent on factors which include the target tissue. In essence, vectors can be employed to deliver a variety of beneficial products, including antisense RNA, ribozymes, or proteins. One of the biggest challenges in treating RA patients with gene therapy lies in the development of vectors which can maintain long term production of transgene products in the joint while concurrently having irrelevant side effects. The gene therapy concept was intended to treat these issues38. While non-viral methods are easy to generate and are relatively safe with low immunogenicity, their efficacy is often unsatisfactory in vivo49. Discussion of these non-viral vectors will not be included here. Viral vectors, which have demonstrated a much higher potency in transgene delivery to the joint (and also have their own serious limitations), will be the center of attention here. This will include discussion of adenovirus, retrovirus, herpes simplex virus (HSV), and adeno-associated virus (AAV).
Adenoviral gene therapy
Adenoviruses have been selected for use in a high number of pre-clinical animal studies of arthritis based on several important benefits that they possess. Adenoviruses are prepared, amplified, and purified at high titer with relative ease. Further, they infect dividing or non-dividing cells of a wide variety of types, and within the joint are capable of delivering transgenes to synovium as well as cartilage, ligaments, and menisci64. Regardless of these advantages, however, serious limitations obstruct the prospects of using the currently available adenoviruses in RA patients. These include transient transgene expression, safety concerns, and an immune system reaction to cells transduced by adenoviruses. These shortcomings are likely associated with each other, since the expression of viral proteins by host cells transduced with adenovirus triggers an immune reaction65, in this way partially underlying the transient expression. The short expression time may also be partially related to the fact that adenoviruses deliver their transgenes in an episomal fashion and lose transgene DNA upon cell division. The obstacle of transient expression is not likely to be suitably fixed by multiple injections, since one administration of an adenovirus can initiate a sizeable immune response66. In addition to the detrimental effects of having host cells expressing viral proteins, more inflammation may be contributed by the antigenic nature of non-infectious viral particles, which may constitute the majority of a given viral preparation61. Further, dendritic and other APCs are readily infected by adenovirus, permitting easy presentation to the immune system61. With regards to safety of adenoviruses, this issue may also be linked with the expression of viral proteins by host cells. In the liver, a low level of viral (E1 deleted adenovirus) gene expression can result in an immune response which initiates hepatocyte destruction and subsequent hepatitis65. Similarly, hepatitis has been noted when adenovirus was used to deliver viral IL-10 in a systemic manner as a treatment for mouse CIA20. In humans, additional safety concerns have been raised following the death of a patient with a mild enzyme deficiency who received a large dose of recombinant adenovirus67. Finally, a relatively minor obstacle in the use of adenoviruses for the treatment of RA, is that the majority of individuals have developed humoral immunity to type 5 adenoviruses68. Thus, one study has shown that 70% of RA synovial fluids carry antibodies which can reduce transduction of synoviocytes by type five adenoviruses. If the other major caveats related to the use of adenoviruses can be solved, this problem is likely to be sidestepped by switching to a different serotype68.
Crippling adenoviruses and/or preparing gutless or extremely attenuated adenoviruses in an attempt to reduce some of the disadvantages associated with adenoviral gene therapy has also been accomplished69. One success along these lines has been to produce a gutted adenovirus missing all viral coding sequences which allowed an approximate 3.5-fold increase in ability to hold DNA of a non-viral origin70. Other studies have investigated adding various motifs to the viral fiber knob to allow new receptors to be used for adenoviral binding. For example, adding a Arg-Gly-Asp (RGD) or polylysine motif allows binding of a integrin or heparin sulfate moiety, respectively, increasing cell attachment which normally operates via coxsackie-adenovirus receptors71, 72. These alterations in vitro led to superior expression of marker transgenes following infection of monocyte-differentiated macrophages and normal fibroblasts72. In a rat model of arthritis in vivo, the RGD and polylysine additions increased marker expression by 150% and 60%, respectively72. Similarly in mice, the RGD motif significantly enhanced transduction efficiency in arthritic joints71. In the latter study, when the RGD-modified adenovirus secreted IL-1 receptor antagonist (IL-1ra), a significant inhibition of mouse CIA was noted in comparison with a non-modified adenovirus producing IL-1ra.
Retroviral gene therapy
With regards to clinical trials performed in humans, most have utilized Moloney murine leukemia virus-derived retroviral vectors73. The main advantages of retroviruses are their ability to stably insert a transgene into dividing cells without disturbing cell growth, and a lack of an immune reaction against vector particles or viral proteins74, 75. Using marker transgenes and high titer retroviral gene therapy introduced intra-articularly, expression is noted to be higher in chronically inflamed knee joints of rabbits when compared with naive knees76. Further, comparing secretable markers quantitatively using in vivo and ex vivo techniques, it was noted that in vivo injections were at least comparable to ex vivo methods in inflamed knees76.
A major hindrance in retroviral use is that their infectivity is generally limited to dividing cells, and synoviocytes tend to proliferate slowly73. One resourceful approach to using the benefits of both retroviruses and adenoviruses involves combining the advantageous aspects of each through assembly of retroviral/adenoviral chimeras77. Thus, Bilbao, Curiel, and colleagues developed an in situ strategy to generate infectious retroviral particles subsequent to adenoviral-mediated gene transfer of the retroviral vector and packaging functions. In this technique, adenoviruses are initially employed to infect non-proliferating cells and to transiently turn them into cells which produce infectious retroviruses77. Next, the locally produced retrovirus infects surrounding cells and achieves stable transduction. In the event that neighboring cells are not dividing, and thus not susceptible to infection by the retrovirus, this strategy could include the production of a growth factor of an appropriate nature. When optimized, this sort of strategy could ultimately eliminate the need for an ex vivo infection77.
AAV gene therapy
Over the last dozen years, much progress has been made in regards to using recombinant AAVs for long term and efficient gene transfer in a wide variety of tissues78. In general, a main advantage of using AAVs as a treatment for RA is that AAVs induce a low inflammatory response, if at all, and they are not cytotoxic. Their expression can last for long periods of time, and they can be produced in high titer with the ability to transduce non-dividing cells. One disadvantage for AAVs is their ability to incorporate only a relatively small gene insert. For example, recombinant AAVs have only 145 bp of viral DNA and their insert size is limited to between 1 and 4.7 kb for successful packaging79. Another disadvantage is that the time taken between infection and expression can be on the order of weeks. While this lag time has been demonstrated in other organs, synovial detection of marker gene expression has been shown to peak with inflammation in only three to seven days80. A further limiting feature is that preparation of a high titer stock of recombinant AAV can be a particularly challenging task.
AAVs have their origin in human parvovirus and the mammalian versions can usually only grow in replicating cells. The AAV name was derived from the realization that productive infection can take place in non-dividing cells that are co-infected with either adenovirus or herpes virus81, 82. It should be noted, however, that this is not a hard rule83. AAVs do not tend to induce production of neutralizing antibodies in humans, where they are ubiquitously found, and they are not associated with any disease84. Thus, AAVs are generally considered innocuous to humans85. Highlights of AAV transduction will be briefly described here, while the interested reader is referred to a more complete review on this topic78. Heparin sulfate proteoglycans on the surface of cells are utilized in AAV adherence and are internalized with the help of the aVb5 integrin and fibroblast growth factor receptor-186, 87. After reaching the nucleus, single stranded viral DNA undergoes second strand production. This is believed to be the rate limiting step for wild type AAV and likely requires host cell enzymes which are not normally active. Wild type AAVs have been demonstrated to stably integrate into the host genome into chromosome 1988, 89, while recombinant AAVs integrate into the host cell genome infrequently and in a non-predictable method90, 91. Interestingly, the episomal DNA of recombinant AAVs appears to exist in a circular form and is responsible for the long term persistence90, 92. As mentioned previously, the time from AAV infection to expression can take weeks, unlike the short 24 hour time taken by adenoviral transduction. DNA damage or helper virus infection have each been demonstrated to aid second-strand synthesis of wild type AAV. In vivo, cellular transduction by AAV can be enhanced approximately 900-fold when rodents are g-irradiated locally prior or subsequent to AAV administration94, 95.
One potentially beneficial observation made with recombinant AAV use in inflamed synovium is that transgene expression is markedly correlated with disease severity80. Using IL-1b or lipopolysaccharide (LPS) to induce acute inflammation, the use of AAV with a reporter gene demonstrated transgene expression levels which increased and decreased with inflammation. After allowing inflammation to decrease over 30 days, less than 8% of cells were positive for the marker protein. Re-challenge with LPS could induce detectable marker protein levels in greater than 90% of cells80. In normal joints, similarly, only 5% of cells expressed the marker protein after 30 days, while greater than 90% of cells were marker protein positive following LPS challenge. These results suggest that transgene expression regulated by the disease-state is feasible, possibly only in the presence of inflammation. While this property may be very useful to exploit during the inflammation associated with later stages of RA, early stages of RA or remission may not be suitable times to utilize AAVs80. A similar study was accomplished comparing normal versus transgenic mice expressing human TNF96. While healthy mice exhibited little evidence of rAAV transduction in joint cells, the TNF transgenic mice exhibited 10-fold higher transduction efficiency which correlated with joint destruction. Synoviocytes, articular chondrocytes, and meniscal cells of diseased mice all appeared capable of transduction in animals expressing TNF-a96.
Investigators studying the applicability of using AAVs as a treatment for RA, have been particularly interested in studies examining AAV expression within diseased synovium78. The results from several studies performed with recombinant AAVs in animal models of arthritis have suggested the feasibility of this approach. These have been accomplished using several different transgenes including soluble TNF receptor47, IL-1ra46, or IL-429, 97.
HSV gene therapy
Early on in the field of treating RA with gene therapy, an assessment of potentially useful vectors suggested that HSV and adenoviruses held the most advantages when trying to deliver transgenes to the synovium in vivo98. In direct comparison, HSV-based vectors were noted to achieve similar transduction levels in synovium of rabbits at significantly lower titers. This could reasonably be expected to correlate with a reduced in vivo immune response, because of the lower viral load98. When expression of marker genes was compared, HSV-based vectors were noted to last at least as well as expression by adenovirus.
As with other viral vectors, HSVs can be generally compared in their applicability to the treatment of RA. Among their relative advantages, HSVs can be prepared at high titer, they can carry a large amount of non-viral DNA, and they are highly infectious. The main disadvantage that was associated with first generation HSVs is their cytotoxicity99. This toxicity is not likely the result of its virion, but rather may be initiated by viral gene products, since UV-irradiated virus is non-toxic, and disrupting immediate early gene expression can also reduce toxicity100104. Studies on the combination of several immediate early (IE) genes suggest their involvement in HSV toxicity. Removal of certain IE genes appears to result in a trade off, which reduces the viral titer, but still demonstrates some success in vivo99. Three IE genes that have been targeted for deletion because they are cytotoxic in vitro include infected cell protein (ICP)4, ICP22, and ICP27105. Another viral gene which destabilizes host cell mRNA, termed UL41, is another logical target to disable99. A recombinant HSV which lacks functional UL41, ICP4, ICP22, and ICP27 genes has shown greatly reduced cytotoxicity and increased transgene persistence in the rabbit knee99.
The HSV genome is circularized after entering the host cell nucleus and, as mentioned, can hold a relatively large capacity of non-viral DNA. This feature is accommodating from the perspective that it may allow for insertion of transgene promoters that are both regulated and sustained, and may be able to coordinate expression of multiple transgenes. Further, HSVs viruses have evolved to develop stealth-like mechanisms which allow them to evade antigenic detection by the host immune system. One HSV IE protein, ICP47, is both necessary and sufficient to block major histocompatibility complex class I protein transport and inhibit lysis of infected cells by CD8+ cytotoxic T cells106. This occurs by retaining major histocompatibility complex class I complexes in the endoplasmic reticulum within three hours of HSV insertion. These types of stealth mechanisms may further be aided by addition of similar or complementary proteins, which may be possible since HSVs can retain large quantities of non-viral DNA.
Generalizations about comparing viral vectors
Based on the above sections and the papers referenced therein, some generalizations can be made when comparing the four viral vectors reviewed above. However, some of these over-simplifications may not apply when considering the treatment of other tissues. Relatively speaking, retrovirus may be the hardest virus to prepare in a high titer. Although high titer AAV preparations can be accomplished, these can often be technically challenging. With regards to the time period in which each of these viruses will maintain expression in the joint, the shortest expression is typically by adenovirus and HSVs. In comparison, the integration into the genome by retroviruses can confer long-term expression, while AAV expression also appears to be long-term based on animal studies of arthritis. Along with the advantage of long-term expression, however, comes the added risk of insertional mutagenesis. This is always a concern with retroviruses and is also possible with AAVs. Since transgenes introduced by HSV and adenoviruses remain episomal, they do not carry this additional safety risk. With regards to other disadvantages, including which of the viruses may induce an immune response, the retroviruses and AAVs do not express viral proteins, thus lowering the possibility of an immune reaction. Adenoviruses and HSVs, however, are known to induce an immune response which is partially mediated by viral produced proteins, and in the case of HSV, the viral proteins may be cytotoxic. When being compared on the basis of fitting large inserts of foreign DNA to add a transgene and its appropriate regulators, HSVs can be considered very accommodating. In contrast, AAVs are very limited in their capacity to accommodate additional DNA. Relative to the extremes of AAV and HSV, both adenoviruses and retroviruses are adequate for fitting many transgenes. Finally, with regards to transducing cells, the most limited of the viruses is probably the retroviruses, because of its inability to transduce non-dividing cells.
Timing and dosage of treatment
Many of the successful preclinical gene therapy trials that have been performed in animal models of arthritis have been organized into a digestible format in Tables 15. The tables are divided based upon the transgene utilized, and they attempt to accurately include the year and reference of the study, the viral vector (or other means) used for transgene delivery, the animal model, the route of administration, and the major effects of administering the transgene that were noted in the study. While a complete listing of each and every study is not possible, due to the extensive number of manuscripts on the topic, these tables allow for an appreciation of the wide variety of studies that have been performed, as well as for some generalizations to be made. First off, with regards to timing of treatment, it is noted that the majority of manuscripts which are able to ameliorate arthritis in an animal model used a prophylactic treatment, where the transgene is anticipated to deliver its products prior to or during disease onset. The successful use of preventative therapy is probably due to the likelihood that it is easier to interrupt the mechanisms that contribute to inflammation and bone damage than it is to reverse the established disease. In studies investigating IL-4 and IL-13, my colleagues and I have employed both a preventative and a therapeutic treatment30, 31. Using an adenoviral vector in conjunction with a direct i.a. injection, we made several observations relevant to the timing and dosage of virus. We noted that introduction of adenovirus producing IL-4 during rat AIA onset, eight days after initiation of disease with adjuvant, resulted in mean concentrations of IL-4 that were more than double the levels found when the same dose of virus was introduced to the ankle (18 days after initiation) at maximal disease30. One possible explanation for this discrepancy may be related to the fact that the paw volume injected at maximal disease is significantly larger than the paw injected on day 8, before inflammation is present. When treated prophylactically on day 8 before signs of inflammation were present, we demonstrated that the IL-4-treated ankles had significantly reduced clinical signs of ankle inflammation and X-ray scores in comparison with the ankles which received a control adenovirus or PBS30. X-ray scores were based on the sum scores of bone erosions, joint space narrowing, and soft tissue swelling. When treating the maximal disease with an adenovirus which produces IL-4, 18 days after initiation of disease with adjuvant, we also were able to demonstrate reduced clinical indicators of ankle inflammation. However, when looking at X-ray scores, we noted that we could only achieve a significant reduction when comparing the IL-4-treated ankles with the control adenovirus group, but not the PBS group. The PBS group represents animals that received adjuvant followed by injections of PBS into their ankles, in the absence of any virus, and therefore they are probably the best indicators of how we altered inflammation and destruction induced by adjuvant. Thus, when trying to intervene with IL-4 gene therapy to treat extensive disease which exists, the IL-4 may have acted to prevent additional inflammation and bone destruction due to the adenovirus, but not due to adjuvant alone. This may be related to the fact that the concentration of IL-4 achieved in the prophylactically treated ankles was at least twice as high as that achieved in ankles which received injections at maximal disease. This emphasizes the importance of timing of the injection, because administering the same dose at different times relative to disease may result in different doses of transgene product achieved. In turn, the concentrations of transgene products achieved in a particular microenvironment could dramatically alter the outcome of the study.
The method of injection and the dosage of virus introduced are other important factors which will play a role in determining whether arthritis-related symptoms are successfully reduced. The injection method should be determined based on the properties of the transgene product. For example, if altering the balance between various lymphocytes is the goal, then an i.v. injection with systemic distribution may be better suited than a local injection. On the other hand, if synovial ablation is the desired endpoint, an i.a. approach would likely be called for. Interestingly, a relatively low dose of a type five adenovirus injected i.a. increased expression of the cytotoxic lymphocyte antigen (CTLA)4-IgG fusion protein transgene and suppressed mouse CIA107. Introducing the same dose of virus via i.v., i.m., or s.c. injections, however, did not ameliorate CIA nor alter serum levels of CTLA4IgG. The CTLA4IgG gene product is an effective blocker of CD28 interaction with co-stimulatory B7-1 and B7-2 in vitro and in vivo, which results in non-responsive T lymphocytes108110. To optimize levels of CTLA4IgG for inhibition of CIA without compromising antibody response to foreign antigens, a dose-response study was performed. It demonstrated that mouse CIA development could be inhibited, concurrent with an antibody response against foreign antigens, at a dose of 1×105 PFU when delivered i.a.107. In rats, using a type five adenovirus with no transgene, my colleagues and I demonstrated that doses delivered i.a. up to 1×108 PFU did not induce inflammation30. In contrast, when ankles were inflamed from AIA, we demonstrated that doses of 1×107 PFU and higher resulted in additional inflammation above and beyond that induced by adjuvant. Using a low dose of adenovirus in rat AIA that would not induce inflammation beyond that induced by adjuvant (5×106 PFU), we were not able to suppress clinical signs of AIA, but rather exacerbated inflammation. While we expected that high doses of adenovirus would result in more inflammation, we did not expect to find that low doses of adenovirus producing IL-4, which did not induce inflammation due to the virus itself, would be pro-inflammatory. When Lubberts, van den Berg and colleagues employed IL-4 in a mouse model of arthritis, they demonstrated an exacerbation of inflammation25, 27, while we and others have shown amelioration24, 28, 30. This observation may be dose-related, and/or be associated to the effects of IL-4 on angiogenesis. At low doses, IL-4 has been shown to possess pro-angiogenic properties while being anti-angiogenic when higher concentrations are achieved111. Consistent with these properties, when we increased our dose from 5×106 PFU to 1×108 PFU, we demonstrated multiple anti-inflammatory effects that could have been partially due to inhibition of angiogenesis30. Together, these studies underscore the importance of the amount of transgene product made and the time during the disease at which that product is produced. Deciding when and how much of a viral vector to administer may depend upon factors inherent to the transgene product, the delivery vector, and the animal model being used.
Cytokines therapeutic candidates and targets
In efforts to move the field of gene therapy forward, from the treatment of preclinical animal models of arthritis to humans with RA, the first phase I clinical trial was completed in 20007, 112. In addition, a phase I study is underway by dr. Blake Roessler at the University of Michigan, another in Dusseldorf, Germany is being conducted by dr. Peter Wehling, and the first phase II study is undergoing regulatory review7. The majority of studies investigating the use of gene therapy for RA continue to be performed using animal models of arthritis. In addition to the method of delivery, the vector used, the timing of delivery, and the dosage, the transgene selected is of obvious importance. In accomplishing the studies performed in Tables 15., the use of a wide variety of transgenes has been examined. An imbalance of cytokines in the RA joint has been well documented by many studies over the years, and these cytokines likely drive a great deal of the i.a. pathology. The studies listed in Table 1. aimed to abate the effects of IL-1b, while those in Table 2. targeted TNF-a. In contrast, the studies outlined in Table 3. aimed to up regulate production of the anti-inflammatory cytokines IL-4, IL-10, and/or IL-13.
IL-1 and TNF-a
It has long been recognized that IL-1 and TNF-a are crucial mediators in RA137. Both cytokines are capable of initiating in vitro cartilage destruction, while IL-1 may be more potent in vivo138. When combined, these proinflammatory cytokines display considerable synergism when examining arthritogenic actions in vivo139. Overexpression of IL-1a, IL-1b, or TNF-a in animals leads to full signs of arthritis140142. An important observation made in these studies is that administration of anti-IL-1 receptor can abolish arthritis induction in mice overexpressing TNF-a, suggesting a primary role for IL-1142. TNF-a, on the other hand, is the most rapidly produced proinflammatory cytokine which may quickly act to coordinate the in vivo cytokine response to injuries143. With regards to animal models of arthritis and gene therapy, targeting IL-1 and TNF-a has been successful in many studies, some of which are included in Tables 1. and 2. It should be recognized that each of the studies included in the tables have made important contributions towards our understanding of how arthritis may best be treated by gene therapy. Many of the studies upregulated transgene production of the IL-1 receptor antagonist (IL-1ra) and achieved differing measures of success with reduction of leukocytosis and synovitis, and achieving a chondroprotective affect. A study which utilized rAAV to deliver IL-1ra conferred long term protection when delivered locally in either a preventative or a therapeutic manner46. Another noteworthy point from these studies is that IL-1ra, when expressed locally, was estimated to be approximately four orders of magnitude more therapeutically efficient than systemic administration of the recombinant protein42. Studies delivering transgenes expressing soluble receptors of IL-1 and TNF-a have shown success in addition to using receptor antagonists.
A cytokine produced by the Th2 subset of cells that has the ability to regulate cell differentiation of those same cells is IL-4144. One cell type known to contribute pro-inflammatory cytokines in the RA joint are activated monocyte/macrophages. Monocytes stimulated in vitro produce proinflammatory cytokines including IL-1a, IL-1b, TNF-a, IL-6, IL-8, granulocyte-colony stimulating factor, and macrophage inflammatory protein (MIP)-1b, each of which can be down regulated by IL-4145. Further, IL-4 increases the expression of IL-1ra in activated monocytes, which also contributes to down regulating inflammatory responses146. IL-4 enhances monocyte apoptosis, while acting on RA synoviocytes to inhibit their proliferation and production of pro-inflammatory prostaglandin E2 (PGE2) and granulocyte/macrophage-colony stimulating factor147, 148. IL-4 reduces bone resorption by acting on osteoclasts and in a ex vivo RA model of bone resorption inhibits production of proinflammatory cytokines149. These traits, in addition to the lack of IL-4 in the RA synovium150, 151, have implicated this cytokine as a target for up-regulation in the arthritic joint.
As a preliminary step before proceeding to in vivo studies, my colleagues and I, determined the effects of adenovirally produced IL-4 on cytokines secreted by RA synovial tissue explants ex vivo152. Conditioned media was collected from explants infected with adenovirus producing IL-4 or a b-galactosidase transgene, and subjected to ELISA analysis. We demonstrated significant reductions in the potent pro-inflammatory mediators IL-1b and TNF-a, in addition to significant decreases in IL-8, PGE2, epithelial neutrophil activating peptide-78, growth related gene product-a, and monocyte chemoattractant protein (MCP)-1 at various time points152. When we analyzed the effects of IL-4 in an animal model of arthritis, treating rat AIA with i.a. injections of adenovirus prophylactically or therapeutically, our results suggest that IL-4 could significantly suppress inflammation, pro-inflammatory cytokine production, angiogenesis, and bone damage30. Table 3. shows that many other preclinical studies examining IL-4 introduced in a variety of manners and by a variety of vectors have also achieved various levels of success in ameliorating arthritis. For example, when delivered to the mouse CIA model by adenovirus, IL-4 prevents cartilage erosion and chondrocyte death while simultaneously enhancing articular cartilage proteoglycan synthesis25. Also noted in this study were lower levels of IL-1b, TNF-a, MIP-2, and stromelysin-325. When AAV was used as the delivery vector in the same mouse model, protection from articular cartilage damage was again noted29. With regards to transgene expression, the latter study demonstrated expression for at least seven months, having stronger expression in arthritic versus non-arthritic mice, and noted no inflammation induced by the rAAV29. When delivered in a systemic manner by immature dendritic cells infected with an adenovirus carrying the IL-4 transgene, mouse CIA could be almost completely abolished61.
IL-13 is a cytokine which shares many biological properties with IL-4 and my colleagues and I found relatively low quantities of IL-13 in the RA synovium153. While a study by Isomaki and colleagues did consistently demonstrate IL-13 presence in RA synovium, they also noted that exogenously added IL-13 could reduce proinflammatory cytokine production by synovial fluid mononuclear cells154. Thus, both studies seem to suggest that IL-13 is present at sub-optimal concentrations to suppress disease in the RA joint153, 154. IL-13 acts on monocytes in vitro in multiple ways, deactivating them as well as altering their morphology, phenotype, function, and cytokine production profile145. Similar to IL-4, IL-13 can profoundly inhibit activated monocyte production of proinflammatory cytokines145. In addition, IL-13 suppresses IL-1a-activated bone-resorbing activity through reducing PGE2 production and cyclooxygenase-2 mRNA in long bone cultures155. In short, for reasons similar to those outlined above for IL-4, IL-13 is another prime candidate for over-expression in the RA synovium. For these reasons, my colleagues and I performed an ex vivo study with RA synovial tissue explants infected with adenovirus, similar to the study described above with IL-4. Production of human IL-13 by adenovirus, in comparison with a control virus producing an irrelevant protein, significantly reduced the mean levels of proinflammatory mediators secreted by the explants156. Further, our study suggested that IL-13 produced by virus was more potent than the same amount of recombinant protein added exogenously156.
We next moved to studying the therapeutic potential of an IL-13 transgene in rat AIA, and first demonstrated that levels of IL-13 are normally low in the rat ankle31. Using an adenovirus with a single i.a. injection, we demonstrated that IL-13 could reduce clinical measurements of inflammation when injected before disease onset. Radiological scores based on joint space narrowing and bone erosions suggested that the IL-13 transgene could significantly lower the amount of bone destruction31. A histological look at the joints demonstrated that IL-13 production significantly reduced polymorphonuclear cell infiltration and the quantity of synovial blood vessels. MCP-1 levels were also significantly suppressed by IL-13. Similarly, when administering the IL-13 transgene after maximal inflammation had ensued, our data demonstrated significant improvements in articular index scores and paw volumes. In addition, TNF-a levels, radiological scores, histologic analysis of infiltrating monocytes and lymphocytes, and the number of blood vessels were all suppressed in ankles receiving IL-13 produced by adenovirus31.
Another successful approach, taken by Bessis and co-workers, utilized xenogenic Chinese hamster ovary fibroblasts in possession of the IL-13 cDNA to secrete systemic IL-135759. In the mouse, incidence and severity of CIA were reduced when assessed by clinical and histological parameters. In addition, IL-13-treated mice had reduced splenic levels of TNF-a mRNA59. Also in CIA, articular disease severity was significantly reduced when a hydrogel-based hollow tube was used to encapsulate Chinese hamster ovary fibroblasts, thus allowing systemic administration of IL-13 while protecting the cells from the immune system57. Administration in this manner was shown to be well tolerated and useful for longterm treatment when implanted into the peritoneal cavity57. In short, each of these studies demonstrated that administration of an IL-13 transgene before the presence of inflammation was beneficial in ameliorating arthritis5759.
Targets other than cytokines
Animal studies of arthritis that appeared therapeutically modulated following gene therapy approaches which targeted intracellular signaling mediators have been organized into Table 4. Similarly, Table 5 shows many of the studies which targeted other mechanisms, including apoptosis and angiogenesis. Nuclear factor-kB (NF-kB) is a mediator involved in intracellular kinase signaling that is common to both IL-1 and TNF-a receptors. In addition, NF-kB has been shown to play a significant role in signaling in both human RA and rat CIA, thus making it a logical target for disruption13. Synthetic double-stranded DNA with high affinity for NF-kB has shown efficacy as a decoy in rat CIA, suppressing severity of paw swelling, improving histologic and radiographic measurements, and reducing IL-1 and TNF-a production in the synovium13. In conjunction with other studies which have targeted NF-kB signaling16, 18, these studies suggest that manipulating our understanding of intracellular kinase signaling pathways represents an attractive means in the treatment of RA.
Another strategy to inhibit inflammation and the subsequent bone damage focuses on reducing blood supply by altering the balance of angiogenic mediators present in the synovium. Successful preclinical studies have been accomplished and some of the targets are included in Table 5. Prevention of mouse arthritis has been demonstrated via administration of a lentiviral vector producing endostatin, a retroviral vector producing angiostatin, and an adenovirus producing a urokinase plasminogen activator receptor antagonist36, 37, 131. In each of these studies, a reduction in the presence of blood flow was shown in addition to amelioration of arthritis. A study targeting apoptosis of synovial blood vessels has also shown promise, by suppressing clinical symptoms of arthritis157. When expressed by phage introduced i.v., RGD-containing peptides bypass normal synovium and show preference for inflamed synovium, via binding avb3 and avb5 integrins, which are crucial in angiogenesis158161. Using a novel approach, a pro-apoptotic heptapeptide dimer, D(KLAKLAK)2, was linked to a chimeric peptide containing a RGD-4C motif (RGD-containing cyclic peptide) and injected i.v. into mice with CIA. This targeted apoptotic chimeric peptide reduced clinical parameters of arthritis in addition to increasing apoptosis of synovial blood vessels157. My colleagues and I have also investigated the effects of IL-4 on angiogenesis within the synovium, while simultaneously looking at alterations in inflammation. We anticipated that IL-4 overexpression may inhibit synovial blood vessel numbers, since previous studies suggested that high doses of IL-4 could act directly on endothelial cells to inhibit angiogenesis111. We introduced an adenovirus producing IL-4 into the rat AIA model before onset of arthritis, and then prepared stained tissue sections from the hind ankles to compare the number of synovial blood vessels. These methods demonstrated that the ankles receiving IL-4 had significantly lower numbers of blood vessels in comparison with ankles treated with PBS or a control adenovirus30. In these studies, as noted earlier, the levels of IL-4 achieved when the adenovirus was introduced before arthritis onset were higher than the levels obtained when the virus was injected after arthritis onset. We did not note a significant reduction in synovial blood vessel numbers when treating rat ankles therapeutically at maximal inflammation, which may be related to the lower dose achieved in these ankles30. In line with this, lower doses of IL-4 appeared to have a pro-angiogenic effect on endothelial cells, in contrast to their anti-angiogenic outcome when used at high doses111. While direct effects of IL-13 on angiogenesis were not noted to occur in the literature, we monitored the effects of an adenovirus producing IL-13 on synovial angiogenesis, since IL-13 shares many properties with IL-4. Our studies demonstrated that adenoviral production of IL-13 in the rat AIA model could significantly reduce the number of blood vessels in the synovium whether it was introduced before or after arthritis onset31.
A gene therapy strategy that has ongoing clinical trials in the U.S. targets induction of apoptosis in proliferating synoviocytes7. The method employs the use of a viral enzyme, herpes simplex virus thymidine kinase (HSV-tk), which aids nucleotide synthesis while viral DNA is replicating. When HSV-tk is successfully transferred to proliferating synoviocytes, it phosphorylates the prodrug ganciclovir, which can be systemically administered. The phosphorylated drug leads to halting DNA polymerization, which in turn kills the cell35. A substantial bystander effect occurs with this technique, since phosphorylated ganciclovir can move to adjacent cells, resulting in destruction of many more cells than the number that were originally infected161. This non-surgical method of synovectomy has been validated through preclinical studies in animal models of arthritis34, 35.
While many RA patients may present with similar macroscopic and histopathological features of inflammation, those features may be rooted in a markedly heterogeneous profile of proinflammatory cytokines, when compared patient to patient. Indeed, some studies suggest that a wide variation is present with regards to cytokine quantity and profile between the joints of individuals162. As a result, successful treatment of RA patients in the future via gene therapy may partially depend upon knowledge of the proinflammatory cytokine profile at the initiation of treatment. Research using gene therapy as a treatment for RA has clearly shown much progress by way of preclinical studies in animal models of arthritis. The tables in this paper demonstrate that a wide variety of approaches taken with numerous transgenes have demonstrated some success in ameliorating arthritis. The challenge that now exists is to translate those promising results into new treatments for use in the clinic.
The author would like to thank the Arthritis Foundation for financial support provided from an Arthritis Investigator Award.
The author would like to thank dr. Alisa E. Koch for her support and acknowledge her important role in all of the gene therapy studies that we have performed. In addition, I would like to thank all of my colleagues who have contributed to our gene therapy publications, including: M. Asif Amin, Ken-Ichi Arai, Joe C. Berry, Phillip L. Campbell, Matthew A. Connors, John A. Damergis Jr., G. Kenneth Haines III, Margaret M. Halloran, Shige Hosaka, Kenneth J. Katschke Jr., Pawan Kumar, Hirokazu Kurata, Jeffrey H. Ruth, Michihide Tokuhira, Michael V. Volin, and Drew C. Woodruff.
GÉNÁTVITEL RHEUMATOID ARTHRITISBEN
Egyre több ismeretet szerzünk a rheumatoid arthritis patogeneziséről, a synovitist kísérő főbb mechanizmusokról. Az utóbbi időben teret nyert a biológiai terápia, amelynek során eddig elsősorban proteinekkel, például citokinekkel szemben fejlesztettek ki specifikus terápiát. Ezen módszerekkel korántsem értek el teljes javulást, és számos kérdés maradt nyitott a rheumatoid arthritis kezelését illetően. Mivel a proteinek alkalmazása nem vezetett eredményre, felmerült a genetikai megközelítés, bizonyos genetikai elemeknek a synovialis membránba való bevitele. Itt tehát nem génterápiáról, azaz hiányzó gének beviteléről, hanem szabályozó gének transzferjéről van szó. Az arthritises állatmodellekben végzett preklinikai vizsgálatok igen ígéretes eredményeket adtak. Ezt követően humán I. fázisú vizsgálatot végeztek, amelynek során igazolódott, hogy hasonló géntranszfer emberek esetében is elvégezhető.
A szerző áttekinti az arthritises állatmodellekben géntranszferrel nyert terápiás tapasztalatokat. A következő területeket veszi górcső alá: 1. a bevitel stratégiái, lokális és helyi stratégiák, hatások az azonos és az ellenoldali ízületben; 2. virális vektorok alkalmazhatósága, adenovirális, retrovirális, herpes simplex és adenoasszociált vektorok összevetése; 3. a kezelés időtartama és a dozirozás kérdései; 4. a lehetséges célpontok, így a pro- és az antiinflammatorikus citokinek áttekintése.
rheumatoid arthritis, génátvitel, citokin, állatmodell, gyulladás