AAV mediated repression of Neat1 lncRNA combined with F8 gene augmentation mitigates pathological mediators of joint disease in haemophilia

Abstract Haemophilic arthropathy (HA), a common comorbidity in haemophilic patients leads to joint pain, deformity and reduced quality of life. We have recently demonstrated that a long non‐coding RNA, Neat1 as a primary regulator of matrix metalloproteinase (MMP) 3 and MMP13 activity, and its induction in the target joint has a deteriorating effect on articular cartilage. In the present study, we administered an Adeno‐associated virus (AAV) 5 vector carrying an short hairpin (sh)RNA to Neat1 via intra‐articular injection alone or in conjunction with systemic administration of a capsid‐modified AAV8 (K31Q) vector carrying F8 gene (F8‐BDD‐V3) to study its impact on HA. AAV8K31Q‐F8 vector administration at low dose, led to an increase in FVIII activity (16%–28%) in treated mice. We further observed a significant knockdown of Neat1 (~40 fold vs. untreated injured joint, p = 0.005) in joint tissue of treated mice and a downregulation of chondrodegenerative enzymes, MMP3, MMP13 and the inflammatory mediator‐ cPLA2, in mice receiving combination therapy. These data demonstrate that AAV mediated Neat1 knockdown in combination with F8 gene augmentation can potentially impact mediators of haemophilic joint disease.

[16] Neat1 (nuclear enriched abundant transcript 1) lncRNA has a crucial role in maintaining the integrity of the nuclear paraspeckle substructure. 17Neat1 is known to be upregulated (~1.5 fold) in joint tissues of both RA and OA patients. 10,13,15Neat1 has a significant impact on inflammation, chondrocyte apoptosis, and the degradation of cartilage. 13,15Neat1 lncRNA is known to alter the expression of cytosolic phospholipase A2 (cPLA2), an enzyme encoded by Pla2g4a by sequestering miR-543. 139][20] For example, Neat1 lncRNA has been implicated in the pathogenesis of OA and has been postulated as a potential therapeutic target. 20Whilst these studies have provided valuable insights into the crucial role of Neat1 lncRNA in other joint diseases, the role of targeting Neat1 lncRNA in the context of HA remains to be explored.In the present study, we studied the impact of combination gene therapy in haemophilia A mice, by Adeno associated virus (AAV) mediated knockdown of Neat1 lncRNA within the articular cartilage, and liver directed F8 gene therapy in HA.

| Design and development of Neat1 specific short hairpin (sh)RNA vector
The shRNA sequence (48 bp) for targeting mouse Neat1 was obtained from Wang et al., 21 and cloned into an AAV backbone containing U6 promoter as illustrated in Figure S1A.The presence of the insert was validated by restriction digestion and Sanger sequencing.The shRNA had previously been validated to target mouse Neat1 using AAV vector, so no scrambled shRNA was used as a control in this study. 21The expression of Neat1 shRNA was confirmed by transfection of the plasmid construct in murine fibroblast cells (NIH3T3) using Fugene HD (Promega, Wisconsin, USA).After 48 h of transfection, RNA was isolated using Trizol (Invitrogen, Massachusetts, USA) and cDNA was prepared from 1 μg of RNA using Quantitect reverse transcription kit (Qiagen, Hilden, Germany).The levels of Neat1 was measured by quantitative (q)PCR using commercially available mouse Neat1 specific primers (Qiagen, Geneglobe ID-SBM0868233).Gapdh (Qiagen, Geneglobe ID-SBM1220562) was used as reference gene for normalisation of qPCR data.

| Vector production
The Neat1 shRNA and the F8-BDD-V3 constructs were packaged into AAV5WT and AAV8 mutant capsid (AAV8K31Q), respectively as previously described. 22,23AAV vectors were produced by triple transfection of capsid, transgene, and helper plasmid in a 1:1:1 ratio in a producer cell line (AAV293) using polyethyleneimine (Polysciences Inc, Pennsylvania, USA).After 72 h of transfection, cells were harvested and lysed by freeze-thaw cycles in liquid nitrogen/ dry-ice ethanol bath.The cell lysate was subjected to benzonase digestion (Santacruz Biotechnology, Santacruz, California, USA) followed by iodixanol gradient ultracentrifugation.The virus was then purified by column chromatography (Cytiva, Massachusetts, USA) and concentrated using Amicon filter (Merck Millipore, Massachusetts, USA).The viral titres were quantified by qPCR using ITR specific primers with AAV2-RSS (ATCC, Virginia, USA) as a standard.

| Repression of Neat1 lncRNA in combination with factor 8 (F8) gene augmentation in mice model of HA
Haemophilia A mice (B6;129S-F8 tm1Kaz /J) and C57BL/6J mice (Jackson Laboratory, Maine, USA) were housed in animal facility with food and water provided ad libitum.The animal experiments were approved by Institutional Animal Ethics Committee, IIT Kanpur.Haemophilia A mice were administered with AAV8K31Q-F8-BDD-V3 at a low dose of 1 × 10 11 vgs/mouse and AAV5WT-U6-Neat1 shRNA at a dose of 2.5 × 10 11 vgs/joint.Groups of animals received PBS (Mock group, n = 9), or AAV8K31Q-F8-BDD-V3 by tail vein injection (F8 treated group, n = 11), or AAV5WT-U6-Neat1 shRNA (Neat1 shRNA group, n = 7) by intra-articular administration of 10 μL volume of the vector into their right knee joints, or the combination of AAV8K31Q-F8-BDD-V3 and AAV5WT-U6-Neat1 shRNA vectors via systemic injection and intra-articular injection in right knee joint, respectively (n = 9).We then performed joint injuries to recapitulate HA in the murine model of haemophilia A as described earlier. 24Briefly, a week after vector administration, the first joint injury (Day0) was performed in the right knee of the mice using a 31G needle, whereas the contralateral left joint that was uninjured served as the control. 25The joint injury was repeated on Day14 and Day30 to recapitulate the pathophysiology of multiple bleeding episodes.Blood samples were collected, 52 days post-vector administration for FVIII specific clotting assays.Joint tissue from the experimental animals was harvested for RNA isolation, immunohistochemistry and histopathology (Figure 1).

| RNA isolation
The vector administered injured (right knee) and uninjured joints (contralateral left knee) (n = 4 mice) from each group were pooled for RNA extraction, respectively.The joint capsule was isolated by a resection of the femur distal end and below the tibia proximal end.This tissue was then immediately transferred to RNA protect reagent (Qiagen).The surrounding tissue consisting of muscle and synovium was excised and cartilage tissue was extracted from femoral condyles and tibial plateau.Total RNA was isolated using Trizol reagent (Invitrogen, Massachusetts, USA), as per manufacturer's protocol.cDNA was synthesized using Quantitect reverse transcription kit (Qiagen) for quantification of Neat1 across different groups of mice.

| Factor VIII assays
Blood samples from experimental mice were collected in 3.2% sodium citrate and plasma was isolated for the clotting assay.FVIII specific clotting activity was measured using the STA-Deficient FVIII plasma (Diagnostica Stago, Asnières-sur-Seine, France) as per the manufacturer's protocol. 26

| Histological assessment
The control and injured joint tissues were harvested after 52 days of vector administration followed by multiple bleeding episodes.The joints were fixed in 4% paraformaldehyde (PFA) for 48 h, then decalcified in 14% ethylene diamine tetra acetic acid (EDTA) for 10-14 days.
Further, decalcified tissues were dehydrated in sucrose gradient and cryosections (Leica CM1520, Leica Biosystems, Wetzlar, Germany) of ~10 μm thickness were obtained using polyfreeze (Sigma Aldrich, Missouri, USA).The joint sections were stained by haematoxylin and eosin staining and analysed for joint characteristics, as described previously. 27We also performed a Prussian blue staining as per standard protocol 28 to document the iron deposits in the joint space.

| Immunohistochemistry
Joints were collected from experimental mice that had three bleeding episodes (Day0, Day14 and Day30) and at Day45 after the first injury.The harvested joint tissues were processed and fixed in 4% paraformaldehyde for 48 h at 4°C.Further, decalcification of the joint tissues was carried out in 14% EDTA for 10-14 days.The joint tissues were then dehydrated in sucrose gradient overnight and blocks were prepared using polyfreeze (Sigma Aldrich).Cryosectioning was performed to obtain joint sections of ~10 μm thickness.Prior to staining, sections were fixed in 4% PFA and washed in phosphate buffered saline containing 0.1% Tween 20 (1X PBST).For for 1 h at room temperature.This was followed by staining with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma Aldrich) and mounting was performed using Fluorsave (Merck Millipore).Fluorescent images were acquired using a confocal imaging system (LSM780NLO, Carl Zeiss GmbH, Wein, Austria) for visualisation of protein expression.For quantification of Alexafluor 568 or Cy3 signal, the fluorescence intensity was measured using ImageJ software as described previously.Briefly, a region of interest (ROI) was defined within the articular cartilage from the surface to the tide mark for all fluorescence images.The integrated density of Alexafluor 568 or Cy3 within the ROI was determined and used for statistical analysis.The exposure settings were kept uniform among various groups. 29

| Data analysis
Data are represented as mean ± SD.GraphPad Prism 8.0.2 was used to create all of the graphs, and Biorad CFX Manager 3.1 was used to analyse the gene expression data from quantitative PCR.An unpaired Student's t-test was performed for statistical analysis.p ≤ 0.05 was considered as statistically significant.

| Design of shRNA vector targeted against Neat1 and its validation in vitro
Our earlier data had demonstrated that Neat1 was substantially over expressed in the hemarthritic joints in a murine model of haemophilia A. 8 To investigate if downregulation of Neat1 is beneficial, we first cloned a previously reported shRNA sequence (48 bp) targeting mouse Neat1 21 in an Adeno-associated virus (AAV) backbone containing a ubiquitous (U6) promoter (241 bp) (Figure S1A).The shRNA against Neat1 used in this study was previously validated to be specifically targeting mouse Neat1 and hence, we did not use a scrambled shRNA as control for validation study. 21To further validate this construct we confirmed the expression of the target sequence and its effect on Neat1 by transient transfection of the cloned shRNA vector into murine fibroblast (NIH3T3) cells (Figure S1A).After 48 h, the relative levels of Neat1 was ~1.95 fold (p = 0.0001) lower in shRNA transfected cells, when compared to mock transfected cells (Figure S1B).Similar level of Neat1 knockdown by transient transfection of shRNA has been demonstrated in mouse cell lines in earlier studies. 30,31The level of shRNA mediated knockdown of Neat1 depends on the transfection efficiency and permissivity of cell line. 32,33However, we observed a significant level of knockdown in vitro consistently (n = 3 biological replicates).This is comparable to the data obtained earlier with this target shRNA sequence in mouse corneal endothelium (~75% Neat1 inhibition), 21 suggesting that the AAV based Neat1 shRNA is functional.

| Neat1 shRNA and factor 8 (F8) gene transfer in Haemophilia A mice
In our next set of experiments, we wished to study the impact of targeted inhibition of Neat1 by Neat1 shRNA, on mediators of joint damage administered either alone or in combination with liver directed factor 8 gene therapy.We thus packaged the Neat1 shRNA construct into an AAV serotype 5 vector.Previous reports have demonstrated that this serotype is best suited for transgene expression into the articular cartilage. 34,35Similarly, a B-domain deleted human F8 gene (F8-BDD-V3, a kind gift from Dr Amit Nathwani, UCL) was packaged into an AAV serotype 8 optimized for high liver directed expression (AAV8K31Q), as described earlier. 22,23oups of haemophilia A mice were pre-treated with AAV5WT-U6-Neat1 shRNA vector (2.5 × 10 11 vgs/joint, intra-articular) or AAV8K31Q -F8-BDD-V3 vector (1 × 10 11 vgs/mouse, intravenous) either alone or in combination, as detailed in Figure 1.A week later, joint injury was performed to mimic hemarthrosis in these animals as described earlier 6 and the impact on molecular mediators was evaluated after multiple bleeding episodes (Day0, Day14 and Day30).
Upon administration with AAV5WT-U6-Neat1 shRNA in the articular cartilage, a significant knockdown of Neat1 was observed in vector-treated mice in comparison to the mock group (Figure 2).Joint tissues from mice receiving AAV8K31Q-F8-BDD-V3 had slightly reduced expression of Neat1, although not significant (p = 0.09) compared to the mock group.This could be due to rescue of bleeding phenotype as a result of increased F8 levels 36  We administered only a low dose of F8 vectors (1 × 10 11 vgs/ mice), to study the impact of lncRNA therapy.Our data revealed that the FVIII activity was ~2-3.3 fold higher in mice administered with AAV8K31Q-F8-BDD-V3 vectors (16.5%) either alone or in combination with AAV5WT-U6-Neat1 shRNA vectors (28%) when compared to phosphate buffered saline (PBS) injected mice (8.3%) (Figure 3).The FVIII activity level in mock group was observed to be ~8.3% which could be attributed to 10-fold higher factor V and factor VII levels in mouse plasma compared to human plasma that results in apparent FVIII activity in haemophilia A murine model used in this study.This phenomenon has been observed in previous studies where one stage FVIII clotting assay was used to measure FVIII activity in haemophilia A mice. 7,[37][38][39]

| Regulation of Neat1 target proteins in vector treated mice
To further assess the impact of Neat1 inhibition on molecular mediators of joint disease, we studied the expression of target proteins of Neat1 such as cPLA2, MMP3 and MMP13 13,40,41 in the joints by immunohistochemistry. Our data revealed an increased expression of cPLA2 and MMP3 in the injured joints of untreated mice compared to control joint tissue, as demonstrated earlier 8 (Figure 4A,B,F,G).
Interestingly, the mice that received Neat1 shRNA vector alone (Figure 4D,I) or in combination with F8 vector (Figure 4E,J) showed notable reduction in these Neat1 target proteins (Figure 4) compared to the injured joints of PBS injected mice (Figure 4B,G).However, the expression level of the target proteins cPLA2 and MMP13 were similar in the injured joints of F8 administered mice and mock group (4C, 4 M).The expression of MMP13 was significantly reduced in joint tissue of mice treated with both F8 and Neat1 shRNA vectors (Figure 4O), when compared to untreated injured joints (Figure 4L).
Further semi-quantitative analysis of the immunohistochemical images revealed significant downregulation of the target proteins cPLA2, MMP3 and MMP13 in mice receiving Neat1 shRNA vector and F8 therapy (Figure 4P,Q,R).Upon immunostaining of joint tissue from haemostatically normal mice (C57BL/6J) for cPLA2, MMP3 and MMP13, a similar level of target protein expression was observed in both uninjured and injured joint tissue sections (Figure S6) indicating that repeated injuries in haemostatically normal mice do not lead to arthropathy phenotype as observed earlier. 7on gross examination, the joint capsules of F8 and Neat1 shRNA vector treated groups demonstrated substantially less bleeding when compared to the untreated injured group (Figure S2).In addition, the mice treated with combination of F8 and shRNA vectors, had reduced synovial hyperplasia and haemorrhage (Figure S3 and S4) and lesser iron deposition (Figure S5).

F I G U R E 2 AAV mediated expression of shRNA represses
Neat1 after intra articular gene transfer.Joint tissue was harvested 52 days post vector administration and pooled RNA was isolated from control and injured joints (n = 4 mice from each group) using Trizol reagent, respectively.RNA was converted to cDNA for gene expression analysis using quantiTect RT kit.Relative quantification of Neat1 (n = 3 replicates) was performed by quantitative PCR.Gapdh levels were used for normalisation of Neat1 expression.An unpaired Student's t-test was performed for statistical analysis.ns, non significant.

| DISCUSS ION
Although FVIII administration can rescue bleeding phenotype in patients with haemophilia, 42,43 there have been multiple reports of breakthrough joint bleeds in patients receiving replacement therapy. 44This indicates that FVIII expression alone would not be sufficient to prevent development of HA and a thorough understanding of molecular mediators is necessary to understand HA progression.
To the best of our knowledge the present study is the first to investigate the effect of lncRNA modulation in the context of molecular mediators of haemophilic joint disease.
LncRNAs, through their interactions with microRNAs, regulate chromatin remodelling and histone modification, and they are involved in a wide variety of biological processes, such as gene transcription, splicing, translation, cell cycle, cell structure maintenance, and apoptosis, among many more. 45The dysregulated expression of lncRNA has been linked to several inflammatory diseases. 46,47LncRNAs play essential role in the development of bone and cartilage, and the abnormal expression of these ln-cRNAs in OA cartilage promotes the breakdown of the cartilage extracellular matrix. 48 Semi-quantitative analysis of immunofluorescence images for cPLA2 (P), MMP3 (Q), MMP13 (R) (n = 3 animals; 3-4 joint sections per group) using Image J software.An unpaired Student's t-test was performed for statistical analysis.ns, non significant. of the miR-129 promoter.As a result, miR-129 is upregulated in rheumatoid arthritis, and this results in a decrease in ERK1/2 phosphorylation and an increase in fibroblast like synoviocytes (FLS) proliferation. 10Wang et al. discovered that the expression of Neat1 and osteopontin (OPN) in osteoarthritis synovial cells were elevated.Following Neat1 knockout, the expression of MMP13, IL-6 and IL-8 in synovial cells decreased, cell proliferation was inhibited, and OPN protein levels decreased. 20Previously, our group reported the upregulation of Neat1 in mouse model of chronic HA and its potential role in the regulation of MMPs. 8Since HA is a joint disease that shares phenotypic manifestations noted in OA and RA, 49 we reasoned that downregulation of Neat1 may impact joint disease in haemophilia.
Neat1 knockdown by short interference (si)RNA or shRNA has been applied in multiple studies for knockdown of matrix metalloproteases. 13,20For example, Xiao et al. have demonstrated that siRNA mediated knockdown of Neat1 resulted in decreased expression of MMP3, MMP9 and MMP13 in OA patient derived chondrocytes. 13In our work, we utilized AAV5 vectors targeted expression of Neat1 specific shRNA to regulate the expression of disease modulators, cPLA2 and MMP3 and MMP13.Since, lncRNA based therapy is local and targeted to the joints, we aimed to also reconstitute F8 at low levels and study its impact on markers of cartilage degradation.
One of the recent studies has shown that systemic administration of recombinant FVIII (100 IU/kg) in haemophilia A mice significantly reduced the synovitis. 36Similarly, the participants from the high dose cohort (6 × 10 13 vgs/kg) in the AAV5-hFVIII-SQ clinical trial showed complete resolution of bleeding in the target joints 48 months after vector administration. 502][53] Nonetheless, such strategies require careful consideration of the choice of AAV serotype based on the target tissue, 54 immune response to the AAV vectors and their dosage optimisation. 55,56r study has certain limitations.We injected animals with AAV F8/Neat1 shRNA vectors prior to initiation of joint injury and documented its effect on molecular mediators.Whilst this is advantageous to directly measure the impact of F8 gene augmentation and Neat1 inhibition, a further validation with on-demand administration of vectors, at different stages of arthropathy development may be required to comprehensively dissect the impact of this intervention at the molecular level in the joints.Since many of these lncRNAs and associated pathways are conserved, this knowledge can benefit a variety of other inflammatory joint diseases such as OA and Juvenile Idiopathic Arthritis.

F I G U R E 1
Study design to evaluate the impact of AAV based inhibition of lncRNA Neat1 in combination with F8 gene augmentation in a murine model of haemophilic arthropathy.Haemophilia A mice were divided into 4 groups: Mock (n = 9), F8 (n = 11) treated, Neat1shRNA treated (n = 7), and F8 and Neat1shRNA treated (n = 9).AAV5WT-U6-Neat1shRNA was administered intra-articularly in the right knee at a dose of 2.5 × 10 11 vgs/joint and AAV8K31Q-F8-BDD-V3 was injected systemically via tail vein at a dose of 1 × 10 11 vgs/mouse.Mock treated mice received PBS injection.After 7 days of vector administration (Day 0), joint injury was performed in the right knee whereas the left knee was uninjured and served as control.Repeated joint bleeding episodes were induced on Day 14 and Day 30 to mimic chronic hemarthropathy like condition.On Day 45, blood was collected to measure FVIII activity level and joint tissues were harvested for assessment of Neat1 expression.Immunohistochemical analysis was performed to check the expression of target proteins in the articular cartilage and histopathological assessment was done to examine the extent of joint damage.Image created using Biore nder.com.immunostaining, sections were incubated with primary antibodies to cPLA2 (1:100, sc-454) (Santacruz Biotechnology), MMP3 (1:50, sc-21,732) and MMP13 (1:100, ab39012) (Abcam, Cambridge, United Kingdom) for 24 h at 4°C.Next, tissue sections were washed in 1X PBST and incubated with blocking buffer containing normal goat serum (Abcam) overnight at 4°C.Further, secondary antibody [Alexafluor 568 (1:500, A-11004) (Invitrogen) or goat anti-rabbit Cy3 (1:200, 111-165-008) (Jackson Immuno Research, West Grove, Pennsylvania, USA)] was added to the tissue sections and incubated thereby resolving the dysregulation of crucial mediators like Neat1.Mice administered with only Neat1 shRNA vector demonstrated a 16 fold downregulation in Neat1 expression whilst mice receiving combinatorial therapy showed a 40 fold decrease in Neat1 expression.The combination of F8 reconstitution and shRNA-mediated Neat1 knockdown led to a remarkable decrease in Neat1 levels (~40 fold) in the combinationtreated group when compared to the mock group.This enhanced repression of Neat1 is possibly due to synergistic effect of FVIII expression and shRNA mediated Neat1 knockdown in combination treated mice as FVIII reconstitution may directly regulate the molecular mediators and decrease the severity of joint damage. 24,36These data further validated the effectiveness of the AAV5 based shRNA vectors to inhibit Neat1, in vivo.

F I G U R E 3
Liver directed F8 gene transfer in haemophilia A mice.Blood samples were collected via retro-orbital sinus in 3.2% sodium citrate (10:1 vol/vol).Blood samples were then centrifuged at 2000 rpm for 15 min to extract plasma.Factor VIII specific activity in the plasma was measured using clotting based assay (Diagnostica Stago) as per manufacturer's protocol to determine phenotypic rescue in treated mice.FVIII expression in mock (n = 9), F8 treated (n = 11), Neat1 shRNA treated (n = 7), and combinatorial therapy group (n = 8) is shown (Mean ± SD).An unpaired Student's t-test was performed for statistical analysis.
Chen et al. revealed that the RA-associated Neat1 lncRNA could bind to miR-204 and increase methylation F I G U R E 4 AAV mediated Neat1 knockdown regulates the target proteins in joint tissue of treated mice.Cryosections (~10 μm) from joint tissue (n = 3 animals per group) were obtained (Leica CM1520, Leica Biosystems, Germany) from experimental haemophilia A mice.Primary antibodies against cPLA2 (A-E), MMP 3 (F-J), and MMP13 (K-O) were used to stain these sections.Our results indicate that these molecular regulators are overexpressed in injured joints (denoted by white arrows) and are repressed in Neat1 shRNA treated or combination therapy group.Images were obtained by confocal microscopy (LSM780NLO, Carl Zeiss GmbH).Scale bar is 50 μm.Representative images are shown.
(p = 0.01), and cartilage degenerative enzymes MMP3 (p = 0.002) and MMP13 (p < 0.0001) in combinatorial gene transfer group compared to untreated injured joints.This demonstrates the synergistic effect of F8 gene augmentation and Neat1 inhibition by possibly strengthening the baseline coagulation function and anti-inflammatory/ chondroprotective function in mitigating the arthropathic phenotype.The use of AAV vectors for combinatorial gene transfer and targeting multiple disease pathways has been demonstrated in other