In-Vivo Generation of Antigen-Specic Tregs with AAV8 Suppresses Collagen-Induced Arthritis in DBA1 Mice through IL-10 Production

Available therapeutics for autoimmune disorders focused on mitigating symptoms, rather than treating the cause of the disorder. A novel approach using adeno-associated virus (AAV) could restore tolerance to the autoimmune targets and provide a permanent treatment for autoimmune diseases. Here, we evaluated the ability of collagen II T-cell epitopes packaged in adeno-associated virus serotype 8 (AAV-8) vectors to reduce pathogenic cellular and humoral responses against collagen and to mitigate the disease in the collagen-induced arthritis mouse model. The cytokines and immune cells involved in the immune suppression were also investigated. Mice treated with AAV-8 containing collagen II T-cell epitopes demonstrated a signicant reduction in the arthritis symptoms, pathogenic collagen specic antibody and T cell responses. The AAV-8 mediated immune suppression was mediated by increased interleukin-10 expression and regulatory T cells expansion. Altogether, this study strengthens the notion that AAV vectors are promising candidates for the treatment of autoimmune diseases. difference in their anti-CII IgG and 28 ng/ml Taken together these ndings indicate that both the cellular and humoral response to CII can be suppressed with AAV8.CII prophylaxis in DBA1 mice. These results also suggest that the AAV8 based immune regulation is transgene specic, as the same AAV8.GFP did not demonstrate a reduction in either T-cell or humoral responses.


Introduction
Rheumatoid arthritis (RA) is a chronic autoimmune disorder with a prevalence in western adult populations of approximately 1%, with a common presentation of small joint in ammation and tissue destruction 1,2 . Despite this relatively high prevalence, RA etiology is not completely understood 2,3 . Due to this incomplete understanding, diagnosis focuses on individual symptom presentation, while treatment aim to minimize chronic disease symptoms such as pain, in ammation, and joint degradation [4][5][6] . Current treatments include pain management, broadly acting non-steroid anti-in ammatory drugs, corticosteroids, and anti-proliferative medications along with speci c biological therapies 7 . These treatments however are only able to stall, and not reverse disease progression. As a result, new avenues of prevention and treatment are being explored using viral vectors for immune modulation.
Minimally immunogenic AAV serotypes such as AAV serotype 8 (AAV8) have been shown to generate robust antigen-speci c immunosuppression even when challenged with adenovirus serotype 5 (Ad5) vectors expressing the identical genes [8][9][10] . These ndings led to the testing of AAV8 based therapies in the experimental autoimmune encephalitis (EAE) mice model. In this animal model, administration of AAV8 encoding myelin oligodendrocyte glycoprotein was able to reverse paralysis in mice 8 . In the EAE model, pathogenicity is mediated by auto-reactive T-cells. However, to date there are no reports of successful therapies with AAV vectors in treating what are believed to be antibody mediated autoimmune disorders.
The mechanism of AAV8 induced gene tolerance has been previously explored with the liver being implicated in the ability of AAV8 to induce gene speci c tolerance in mouse models 9,[11][12][13] . Within these models, the immune tolerance is thought to be the result of antigen speci c peripherally induced CD4 + CD25 + FoxP3 + T regulatory cells (Tregs) 14 . These Tregs are generated in the liver in part due to the manner in which the secreted antigen is taken up and presented by kuffer cells. This antigen presentation, along with an increased concentration of immunoregulatory cytokines such as interleukin (IL)-10, 13, and transforming growth factor b 10,14-16 is thought to drive the expansion of hepatic Tregs. These Tregs may only suppress local immune responses in the liver, but may also migrate to exert their regulatory functions at other peripheral sites making AAV8 induced gene tolerance an option for the maintenance or restoration of peripheral tolerance 8,14,[17][18][19] The collagen induced arthritis (CIA) model is the most common mouse model of antibody mediated autoimmune arthritis and is an indispensable tool in the study of the pathogenic mechanisms of the disease and testing of therapeutics. This is in large part due to similarities in disorder presentation and tissue destruction, along with its close linkage to MHC alleles 20,21 . Both T and B cell responses to the collagen II (CII) protein are noted in the CIA model and the immunodominant T-cell and B-cell determinants have also been identi ed 21,22 . Early works in determining CII T-cell epitopes also noted that CII peptide tolerization in susceptible mice can be achieved by adoptive transfer of Tregs modi ed to present the immunodominant murine epitope of CII (aa259-270) suppressing the disease in recipients 2,21,22 .
Here, we assessed the ability of an AAV8 vectors encoding CII T-cell epitopes to inhibit the symptoms associated with CIA as well as the pathogenic humoral and cellular response against collagen.

Material And Methods
Viral construct preparation: Previously characterized CII T-cell epitopes were synthesized in a sequence derived from the native murine collagen II (Col2A1) gene cDNA sequence by Genescript (Piscataway, NJ).
The resulting construct was cloned via Gibson assembly (New England Biolabs, Rowley, MA) into either pAAV-MCS or pacAd5 CMV K-N (all from Cell Biolabs, San Diego, CA) to generate the pAAV-MCS: CII or pacAd5 CMV K-N: CII constructs.
The virus was released by freeze-thaw and subsequently ampli ed in HEK293Ad cells and puri ed as previously described 25 .
Mice: Female DBA1/J mice (8 weeks of age) were purchased from The Jackson Laboratory (Bar Harbor, MI). Mice were acclimatized for at least a week prior to any experiment. Three mice per group were injected by way of the tail vein with 10 11 vector genomes (VG) of AAV8 or 10 10 Ad5 viral particles (VP).
Blood and serum samples were collected via cardiac puncture following overdose of iso urane. In vivo, Treg depletion was performed by intraperitoneal injection of 0.75mg of PC61 (BioX-cell, West Lebanon, NH) at 2 or 7 days post-AAV administration, while blockage of IL-10 receptor (IL-10R) was achieved with 0.75mg of 1B1.3A (anti-IL-10R) (BioX-cell) 2 days pre-and post AAV8.CII administration. Induction of CIA and analysis of the progressive in ammatory phenotypic was performed as previously described 26 . All animal protocols were approved by the Institutional Animal Care and Use Committee of Université Laval. All experiments were performed according to the guidelines of the Canadian Council on Animal Care. Throughout the manuscript, the methods of the performed experiments were reported according to the ARRIVE guidelines.
Liver Leukocyte and Splenocyte Isolation: Liver Leukocytes were isolated by Ficoll. Brie y, the liver homogenate was treated with ACK lysis buffer (Thermo sher) to remove erythrocytes and the cell suspension was subjected to one round of Ficoll (Sigma Aldrich, Nadick, MA) separation. Nonparenchymal cells were recovered and subsequently centrifuged and washed with RPMI (Gibco) supplemented with 10% Fetal Bovine Serum (FBS) (Wisent). Splenocytes were isolated from spleen homogenate and treated with ACK lysis buffer to remove erythrocytes and the resultant cell suspension was suspended in RPMI supplemented with 10% FBS, 1X Penicillin/Streptomycin, 1X Glutamax.
Interferon-g Enzyme-Linked Immunosorbent Spot Assay: The interferon (IFN)-g enzyme-linked immunosorbent spot (ELISPOT) assay was performed according to the manufacturer's instructions (BD Biosciences, San Jose, CA) 14 . Brie y, cells were stimulated with 1 mg/ml chicken CII protein (Chondrex, Redmond, WA) for 48 hours and spots were visualized with an AEC substrate kit (BD biosciences) and counted with the AID ELISPOT reader system and ISPOT software (Autoimmun Diagnostica GMBH, Strabberg, Germany) Cell Culture and Cytokine Assays: Plating and culture of liver non-parenchymal cells and splenocytes were performed as previously described 14 . Cells were stimulated with a nal concentration of 1 mg/ml chicken CII protein (Chondrex). Interleukin (IL)-10 was detected by an Enzyme-Linked Immunosorbent Assay (ELISA) kit (MabTech) as per the manufacturer's instructions. IFN-g was detected by an ELISA kit (R&D Systems, Minneapolis, MN) following the manufacturer's instructions.
Proteome Pro ler Cytokine Assay: Analysis of cytokines in puri ed liver non-parenchymal cell culture supernatant was assessed using the Proteome Pro ler Cytokine Array Kit, Panel A (R&D Systems) as per manufacturer's instructions. Spot intensity was assessed using ImageJ 27 .
Statistics: Statistical analysis was signi cance assessed by Unpaired T-test or ANOVA using GraphPad Prism version 8.0.0, (GraphPad Software, San Diego, CA). A P < 0.05 was considered signi cant.

AAV8.CII injection mitigates the CII speci c immune response in DBA1 Mice:
To date, the ability of AAV to suppress antibody-mediated autoimmunity had not been tested. Here, we sought to evaluate the therapeutic e cacy of AAV8 in the mouse CIA model where pathogenicity is mediated by auto-reactive antibodies against collagen 21 . However, due to the limited packaging size of AAV, the native CII cDNA sequence could not be used. Instead, a targeted cDNA construct was designed to contain all known CII T-cell epitopes. We anticipated that the resulting AAV8.CII would generated regulatory T cells speci c to collagen II capable of suppressing both cellular and humoral response toward this epitope.
First, we evaluated the ability of AAV8.CII to prevent collagen speci c humoral and cellular responses when challenged with Ad5.CII encoding for the same CII epitopes. Ad5.CII has been shown to trigger immunity to CII mimicking an auto-immune status against this antigen. Mice rst received a tail vein injection with 10 11 VG of AAV8.CII or AAV8.GFP, the later used as a non-speci c antigen control. Mice were then challenged 14 days later with 10 10 Ad5.CII VP ( Figure 1A) as previously described 14 . T cell and antibody responses were measured 10 days and 28 days post Ad5.CII challenge respectively ( Figure 1A).
When animals were treated with AAV8.CII prior to Ad5.CII, only background levels of T-cell responses to CII were detected by IFNg-ELIPOST (~25 spot forming units (SFU) /10 6 ). This is in contrast to when mice received AAV8.GFP or PBS control prior to Ad5.CII challenges, with no difference of CII speci c T cells (200 and 180 SFU/10 6 cells respectively) being noted in these animals ( Figure 1B). The humoral response against CII demonstrated a similar pattern of reduction in animals receiving AAV8.CII prophylaxis, where they exhibited background levels of anti-CII IgG (19 ng/ml)). Mice receiving AAV8.GFP or PBS prior to Ad5.CII challenge showed no difference in their anti-CII IgG concentration (30 and 28 ng/ml respectively) ( Figure 1C). Taken together these ndings indicate that both the cellular and humoral response to CII can be suppressed with AAV8.CII prophylaxis in DBA1 mice. These results also suggest that the AAV8 based immune regulation is transgene speci c, as the same AAV8.GFP did not demonstrate a reduction in either T-cell or humoral responses.
Treg are a critical mediator in AAV8.CII mediated autoimmune suppression: Hepatic Treg have been previously implicated in AAV8-mediated immune suppression 18 . In order to assess their involvement in the suppression of CII immune responses, the frequency of hepatic Tregs was measured by ow cytometry in control and AAV8.CII treated mice. Live Treg were de ned by the coexpression of CD45, CD3, CD4, CD25 and Foxp3+, and the absence of staining by a LIVE/DEAD dye (Supplementary Figure 1). Compared with control mice, a signi cant increase in Treg average frequency from 6 % to 12 % of CD4+ T-cells was noted in mice receiving AAV8.CII prophylaxis.
Depletion of Tregs was performed prior to AAV8.CII treatment in mice to assess the role of this expanded population on T and B cell responses. For Treg depletion, mice received an intraperitoneal injection of 0.75mg of PC61, an antibody against CD25, 7 days post AAV8.CII injection (Supplementary gure 2). Mice were subsequently injected with Ad5.CII 7 days later (Figure 2A). Mice only injected with Ad5.CII were used as positive response controls. The frequency of CII-speci c T cells was analysed by IFNg ELISPOT 10 days after the Ad5.CII injection. The magnitude of the humoral response against CII was analysed by CII ELISA 28 days following the Ad5 injection ( Figures 2B and 2C). Treg depletion in AAV8.CII pre-treated mice led to signi cant increase in CII-speci c T cells as well as CII-speci c IgG after Ad5.CII injection (p=0.001 and 0.001 respectively). Indeed, CII-speci c T cells increased from 109 SFU/10 6 (SEM 13.47) in mock treated to 185 SFU/10 6 cells (SEM 17.27) in depleted mice ( Figure 2B).
Similarly, the amount of CII speci c IgG nearly doubled from 90 ng/ml to 150 ng/ml on average between mock treated and Treg depleted mice ( Figure 2C). Of note, an increase in both CII speci c cellular and humoral response was also observed in PC61 treated mice even in absence of Ad5 injections (Supplementary Figure 3).
Overall, a signi cant decrease in Treg cells was noted in the mice receiving PC61, and this decrease in Tregs was concomitant with an increase in CII-speci c cellular and humoral response. Taken together, these data suggest that Tregs are involved in the AAV8.CII mediated suppression of the CII immune response.
IL-10 is modulated and important to AAV8.CII mediated suppression of CII speci c immunity.
To better understand the soluble mediators involved in the AAV8.CII Treg dependent immune modulation.
The cytokines pro le of CII stimulated liver non-parenchymal cells from control and AAV8.CII treated mice were evaluated by immunoblot. To do so, mice were saline or AAV8.CII treated 14 days prior to Ad5 injection and euthanized 10 days later. Non-parenchymal cells were isolated from liver homogenate by Ficoll and a total of 10 6 cells were stimulated with 0.1ug/ml CII for 4 days prior to supernatant collection ( Figure 3A). The expression of 22 cytokines was then pro led using a cytokine proteome pro ler. Mice receiving AAV8.CII prophylaxis exhibited elevated IL-2 and immunoregulatory IL-10 in comparison to mice receiving only Ad5.CII ( Figure 3B).
To further explore the origin of IL-10, the above experiment was repeated with the additional of a group in which Tregs were depleted using PC61. The level of IL-10 in the supernatant of CII stimulated hepatic non-parenchymal cells was measured by ELISA. In culture from hepatic non-parenchymal cells originated from AAV8.CII but not Treg depleted or mock treated mice, IL-10 production was elevated with the OD increasing from background levels to an OD of 0.35. In mice that received PC61 post AAV8.CII administration, no increase in IL-10 above background levels was observed following ex-vivo stimulation ( Figure 3C). These results indicate that the detected IL-10 production in the response to CII is tied to Treg cells.
Finally, we sought to evaluate the importance of IL-10 in AAV8.CII mediated inhibition of CII speci c cellular response. To do so, mice were treated as in gure 3C and CII speci c IFNg production was measured via ELISPOT. A group of mice that received Ad5.CII alone was used as control ( Figure 3D). In mice receiving Ad5.CII alone, approximately 150 SFU/10 6 CII-speci c T cells were detected, while in AAV8.CII prophylactically treated mice, the CII-speci c T cells response was present only at background levels (20 SFU/10 6 cells). Interestingly, AAV8.CII prophylactically treated mice injected with the IL-10R blocking mAb demonstrated CII-speci c T cells response at an average of 125 SFU/10 6 cells. Together these ndings suggest that AAV8.CII suppression of CII speci c immunity is mediated at least in large part by IL-10, and that this IL-10 is primarily derived from Treg cells in response to CII.
AAV8 mediated hepatic gene expression mitigates the progression of collagen-induced arthritis: Based on the above AAV8.CII suppression of CII speci c humoral and cellular responses, we explored the possibility of AAV8.CII inhibition on the severity of CIA in DBA1 mice. Mice received either AAV8.CII or saline prophylaxis treatment 14 days before the induction of the disorder and were followed for 15 days after LPS induced disease onset ( Figure 4A). Disease progression was graded based on joint in ammation as previously described 26 , while the IgG response against CII was measured by ELISA posteuthanasia as described above. An approximate 35% reduction in paw in ammation was noted in the mice receiving AAV8.CII prophylaxis (Supplementary Figure 4). In addition, in AAV8.CII treated mice, there was a signi cant decrease in anti-CII IgG from 60 ng/ml in mock treated mice to 20 ng/ml in mice receiving AAV8.CII ( Figure 4B). Together, these data suggests that prophylactic administration of AAV8.CII can slow arthritis progression in the CIA model and decrease autoantibody production.
Both Treg cells and IL-10 are involved in the suppression of arthritis with AAV8.CII prophylaxis.
Next, we examine whether Treg and IL-10 were involved in the AAV8.CII-mediated reduction of the arthritis phenotype. First, the frequency of Treg in hepatic non-parenchymal cells was monitored by FACS at the time of euthanasia in CIA induced mice in the presence or absence of AAV8.CII. The frequency of hepatic Tregs increased from 7.3 to 10.1% of CD4 T cells in mice receiving AAV8.CII when compared to saline controls ( Figure 4C). This expansion of Treg cells is consistent with what was observed in the Ad5.CII challenge model where a 5% increase in hepatic Tregs was noted prior to Ad5.CII, and suggests that mechanisms of immunoregulation in the CII speci c response are shared between the Ad5.CII challenge, CIA models.
To con rm the suspected role of Treg in AAV8.CII arthritis mitigation, Treg depletion was performed in AAV8.CII treated mice prior to arthritis induction using PC61 ( Figure 4D). Disease progression and level of CII speci c IgG was monitored as above. The depletion of Tregs in AAV8.CII treated mice abrogated the reduction in disease progression and resulted in greater peripheral in ammation than arthritis controls ( Figure 4E). The elimination of Tregs also restored the levels of CII speci c IgG to the same level as mice with CIA induction alone ( Figure 4F).
Finally, the impact of Treg depletion on IFNg and IL-10 production in the CIA model was evaluated. To do this, the hepatic non-parenchymal cells from CIA induced mice were isolated with and without Treg depletion and stimulated ex-vivo with CII as described above. Mice that received AAV8.CII prophylaxis had an approximate 50% reduction in the OD of pro-in ammatory IFNg and a concomitant 2.5 fold increase in IL-10 level when compared to arthritis controls. Treg depletion with PC61 restored the CII speci c IFNg response and eliminated the increased IL-10 production normally observed with AAV8.CII prophylaxis ( Figures 4G-H). This increase in in ammation severity, CII speci c antibody, increase in pro-in ammatory cytokine production and decreased immunoregulatory cytokine expression strongly suggests that peripherally induced Tregs are one of the primary mediators of the immunoregulatory response in CIA mice receiving AAV8.CII prophylaxis.

Discussion
Rheumatoid arthritis is a multifaceted disorder whose treatment consists primarily of symptom management and of preventive measures to minimize exacerbation of the clinical syndrome. Antiin ammatory, immunosuppressive, and immunomodulating drug regimens have shown promises in slowing the clinical progression of RA. These treatments however, are often non-speci c, provide mainly temporary relief, and none are capable of inducing complete disease remission 4,6 .
Biological therapeutics targeting in ammatory cytokines along with the administration of IL-10 have previously been shown to reduce disease severity and reduce the duration of the disease in the CIA model. However, due to their prohibitive cost and broad spectrum action, they suppress the immune system broadly therefore rendering treated individuals susceptible to opportunistic infections and secondary complications [28][29][30][31] . In contrast, AAV-based therapies could provide tolerance targeted to the epitopes in its transgene. This is supported by the data obtained with AAV8.GFP and the absence of an immunoregulatory phenotype during the Ad5.CII challenge. This suggests that AAV expressing epitopes recognized in autoimmune disorders may be able to speci cally suppress pathogenic responses in humans. This speci c immunoregulation would eliminate the general immunosuppression seen in existing biological therapies and may not require repeated administration as AAV can stably express antigens in vivo for over 10 years.
It is worth noting that AAVs can only package relatively small transgenes of up to ~4 kb and this fact has been a major drawback in its clinical use. Numerous human proteins involved in autoimmunity are indeed larger than 4 kb necessitating a tailored approach to epitope expression. Here we demonstrated that selectively packaging T cell epitopes from the protein of interest is su cient to diminish both cellular and humoral responses against the transgene. By using epitopes rather than whole proteins, tolerance could theoretically be induced against any antigen, independent of its size.
The crucial role played by central and peripherally induced Tregs in immune regulation has been previously reported in animal models and in human autoimmune diseases. Lower numbers or functional impairment of Tregs is reported in multiple autoimmune disorders including RA, multiple sclerosis and lupus 8,14,32,33 . The current data also show that Tregs play a central role in regulating CIA with AAV8.CII treatment leading to expanded Treg populations, reduced IFNg and increased IL-10 in mice. This central role of Tregs was further de ned with depletion of Tregs removing any bene t in AAV8.CII treated mice and restored tissue damage and IFNg autoimmune responses against collagen.
The present study shows that the delivery of speci c autoimmune targets with AAV8 is able to reduce pathogenic autoimmune responses in mice. Importantly, prophylactic administration of the AAV8.CII vector was able to mitigate the severity of CIA. Due to the speed of disease induction, this model did not allow for evaluation of AAV as a therapeutic agent. Therefore, further exploration in slower acting models of autoimmunity must be explored in order to determine AAV therapeutic e cacy in established autoimmune disorders. This work warrants further investigation in order to advance AAV8 based therapeutic of autoimmune syndormes to the clinic.   Elevated IL-10 is essential for tolerance to CII. AAV8 mediated tolerance is antigen speci c: (A) Mice (n=3/group) were injected intravenously with 1011 VG of AAV8.CII or AAV8.GFP. If applicable, mice were injected intraperitoneally with 0.75 mg of 1B1.3A anti-IL-10R antibody 2 days prior and 2 days post intravenous injection of AAV. At day 14 mice were injected intravenously with 1010 VP Ad5.CII. 10 days after Ad5.CII injection liver non-parenchymal cells were cultured with anti-CD3 and anti-CD28 in the presence of CII; 96-hours later culture supernatants were collected, pooled and analyzed for cytokine production. (B) Dot blot analysis of cytokine production from cultured liver non-parenchymal cells stimulated with CII. Pixel density was measured with ImageJ software. (C) Quanti cation of IL-10 production in cell culture supernatants was analyzed by ELISA. (D) 10 days after Ad5.CII injection, liver non-parenchymal cells were stimulated with CII and IFN production was measured by ELISPOT. For each group, the mean +/-SEM is shown (B-D).