Intra-Articular Adeno-Associated Virus-Mediated Proteoglycan 4 Gene Therapy for Preventing Posttraumatic Osteoarthritis

Abstract

Lubricin, a glycoprotein encoded by the proteoglycan 4 (PRG4) gene, is an essential boundary lubricant that reduces friction between articular cartilage surfaces. The loss of lubricin subsequent to joint injury plays a role in the pathogenesis of posttraumatic osteoarthritis. In this study, we describe the development and evaluation of an adeno-associated virus (AAV)-based PRG4 gene therapy intended to restore lubricin in injured joints. The green fluorescent protein (GFP) gene was inserted the PRG4 gene to facilitate tracing the distribution of the transgene product (AAV-PRG4-GFP) in vivo. Transduction efficiency of AAV-PRG4-GFP was evaluated in joint cells, and the conditioned medium containing secreted PRG4-GFP was used for shear loading/friction and viability tests. In vivo transduction of joint tissues following intra-articular injection of AAV-PRG4-GFP was confirmed in the mouse stifle joint in a surgical model of destabilization of the medial meniscus (DMM), and chondroprotective activity was tested in a rabbit anterior cruciate ligament transection (ACLT) model. In vitro studies showed that PRG4-GFP has lubricin-like cartilage-binding and antifriction properties. Significant cytoprotective effects were seen when cartilage was soaked in PRG4-GFP before cyclic shear loading (n = 3). Polymerase chain reaction and confocal microscopy confirmed the presence of PRG4-GFP DNA and protein, respectively, in a mouse DMM (n = 3 per group). In the rabbit ACLT model, AAV-PRG4-GFP gene therapy enhanced lubricin expression (p = 0.001 vs. AAV-GFP: n = 7–14) and protected the cartilage from degeneration (p = 0.014 vs. AAV-GFP: n = 9–10) when treatments were administered immediately postoperation, but efficacy was lost when treatment was delayed for 2 weeks. AAV-PRG4-GFP gene therapy protected cartilage from degeneration in a rabbit ACLT model; however, data from the ACLT model suggest that early intervention is essential for efficacy.

INTRODUCTION

Up to 60% of patients, who undergo anterior cruciate ligament (ACL) replacement surgery, develop symptomatic knee osteoarthritis (OA) within 15 years.^1–6 The need to provide treatment options is urgent, as refinements in ACL repair techniques over the last few decades have not significantly reduced the risk for OA.⁴ Articular cartilage is vulnerable to the loss of joint lubrication, which is needed to minimize friction between joint surfaces. Lubricant failure in the aftermath of ACL injuries has been well documented, and it is clear that lubricant loss plays a major pathogenic role in OA, particularly in the ACL-deficient knee, where shear stresses are exaggerated by joint instability.⁷

Synovial fluid (SF) contains a number of substances that lubricate joint surfaces, including hyaluronic acid, phospholipids, and lubricin. Lubricin is encoded by the proteoglycan 4 (PRG4) gene, which is expressed by synovial cells and chondrocytes, and is an essential boundary lubricant that reduces friction between articular cartilage surfaces.⁸ It is a heavily glycosylated protein with a mucin-like domain consisting of 76 repeats of a threonine-rich amino acid sequence. In addition to its role as a cartilage lubricant, PRG4 impairs cell-matrix adhesion and elicits anti-inflammatory and antifibrotic cellular responses upon docking with CD44 and toll-like receptors (TLRs).^9,10 These activities may contribute independently to the therapeutic effects of PRG4 supplementation.

Ruan et al. has reported that intra-articular genetic overexpression of PRG4 was chondroprotective in age-related OA in the mouse.¹¹ In tissue and SF analyses from ACL transection (ACLT) models, lubricin concentration and expression were significantly decreased by elevated tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β).¹² The lost lubricin may not be readily replaced because its biosynthesis is inhibited by high levels of the proinflammatory cytokines that accompany trauma and inflammation.

Lubricin supplementation via intra-articular injection of the protein mitigates cartilage degeneration in ACLT and destabilization of the medial meniscus (DMM) models, indicating that its loss contributes to OA progression.^13,14 However, multiple injections are required for efficacy due to lubricin's relatively short half-life in vivo. This limitation stands as a barrier to implementation of lubricin therapy in humans at risk for OA who may require long-term supplementation. Intra-articular gene therapy offers an appealing solution to this problem, as a single injection of viruses carrying the PRG4 gene can provide a steady supply of the protein for many months.

The chondroprotective effects of PRG4 gene therapy in combination with interleukin-1 receptor antagonist (IL-1RA) gene therapy were recently demonstrated by Stone et al., who used helper-dependent adenovirus (HDV) as a vector in mouse models of posttraumatic osteoarthritis (PTOA).¹⁵ Adeno-associated virus (AAV) vectors offer some of the same advantages as HDV: Cells infected with AAV remain healthy, and inflammatory responses to the virus are minimal. Moreover, in contrast to retroviral systems where oncogenic mutagenesis is a risk, AAV seldom inserts into the host's chromosomes and transgene expression is a function of multiplicity of infection.^16,17 Moreover, cells in mouse joints transduced with AAV-green fluorescent protein (GFP) continued to express the transgene for more than a year after intra-articular injection of the virus.¹⁸

Evaluating in vivo transgene expression can be greatly facilitated by including sequences encoding fluorescence markers together with the gene of interest in bicistronic clones. Our recombinant PRG4-GFP clone produces a PRG4/GFP fusion protein designed to facilitate mapping its tissue distribution in joints.¹⁹ We hypothesized that the PRG4/GFP fusion protein is fully functional as a cartilage lubricant and that in vivo transduction with AAV-PRG4-GFP would delay PTOA progression in a rabbit ACLT model.

MATERIALS AND METHODS

Molecular cloning of lubricin-GFP fusion protein in AAV

The domain structures of lubricin are described in Fig. 1A. While all variants contain the mucin-like and cartilage-binding hemopexin (HP) domains essential for boundary lubrication, Lub C lacks domains 4 and 5, rendering it somewhat smaller than other variants and facilitating cloning and packaging in the AAV vector. Although the functions of these N-terminal domains are incompletely understood, their omission does not appear to disrupt boundary lubrication.²⁰ The threonine-rich amino acid sequence (KEPAPTT), which serves as a glycosylation site, is repeated 76 times in the mucin domain. In the PRG4-GFP fusion gene, the enhanced GFP (EGFP) coding sequence replaces a part of the original mucin domain sequence to avoid disrupting N- or C-terminal domains that function in lubricin dimerization and extracellular matrix binding. The sizes of Lub C-EGFP and Lub C are 3,036 and 3,951 bp, respectively.

Figure 1.

Molecular cloning and characterization of a lubricin (Lub)-enhanced EGFP (PRG4-GFP) fusion protein. (A) Domain structures of Lub A, B, C, and Lub C-EGFP. SM, somatomedin B; HB, heparin binding; HP, hemopexin. (B) Cloning in AAV. Human lubricin cDNA was cloned into AAV vector by multiple fragment cloning methods. (C) Transduction efficiency of AAV-PRG4-GFP in chondrocytes (CC) (D), synoviocytes (SC) (E), and meniscus cells (MC) (F) (n = 4–11). (G) PRG4 contents in conditioned medium harvested from transduced SC (1 × = 2 × 10⁵ v.g, n = 2). (H) PRG4 immunoblot in × 10 CCM harvested from transduced SC. (I) GFP expression on the cartilage surface (arrow) implanted with CCM. Scale bars = 100 μm. Data are expressed as mean ± 95% CI. AAV, adeno-associated virus; CCM, concentrated conditioned medium; cDNA, complementary DNA; CI, confidence interval; GFP, green fluorescent protein; PRG4, proteoglycan 4; v.g, viral genomes. Color images are available online.

The human lubricin complementary DNA was cloned by multiple fragment cloning methods (Fig. 1B). Lub C-EGFP (PRG4-GFP) gene was inserted into the AAV vector (G0347 pFBAAVCMVmscBgHpA; Viral Vector Core Facility, University of Iowa, Iowa City, IA) between the enzyme sites of Xho I and Not I. For AAV-GFP as a control of AAV-PRG4-GFP, EGFP DNA was amplified by polymerase chain reaction (PCR) method and was inserted into AAV, specifically between BamH I and Apa I enzyme sites. The final constructs of AAV-GFP and AAV-PRG4-GFP were confirmed by sequencing.

Osteochondral plug preparation and cell isolation

Fresh bovine stifle joints from young adult cattle (15–24 months old) were received from a local abattoir (Bud's Custom Meats, Riverside, IA). Osteochondral plugs (12 mm diameter and 10 mm height) were prepared using a trephine drill (Salvin Dental Specialties, Charlotte, NC) from the femoral condyles. The plugs were washed twice in Hanks Balanced Salt Solution (HBSS) and cultured in a 1:1 mixture of Dulbecco's modified Eagle medium (DMEM) and Ham's F-12 medium (F-12) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Waltham, MA), 100 U/mL penicillin, 100 μg/mL streptomycin, and 2.5 μg/mL Fungizone. All plugs were cultured in a 37°C incubator with 5% O₂/3% CO₂.

After 2 days of preculture, the osteochondral plugs were injured by 7 J/cm² blunt impact using a drop tower device with a 4 mm diameter flat-ended platen, which caused superficial chondrocyte death and mild fissuring of the superficial zone.^21,22

Separate from plug collection, bovine chondrocytes, synoviocytes, and meniscus cells were isolated from freshly harvested stifle joints using 2.5 mg/mL collagenase (Sigma-Aldrich, St. Louis, MO) and 2.5 mg/mL pronase (Sigma-Aldrich) digestion. The isolated cells were subcultured and used for assessing AAV transduction efficiency.

Transduction efficiency test

Bovine chondrocytes, synoviocytes, or meniscus cells (1.5 × 10⁵) were seeded in a six-well plate and infected with or without 2 × 10⁵ viral genomes (v.g) of AAV-PRG4-GFP. For increasing transduction efficiency, 1 μL/mL Hoechst 33342 (Sigma-Aldrich) and 2% FBS were added in DMEM for the first 6 h. The medium was then replaced with serum-free DMEM with 1 μg/mL ITS Premix (BD Biosciences, San Jose, CA). After 2 days, the PRG4-GFP conditioned medium was collected for further studies, and the cells were fluorescently imaged with an Olympus Fluoview 1000 Confocal Laser Scanning Microscope (Olympus America, Center Valley, PA) to confirm transfection/protein expression. Transduction efficiency was calculated by the number of positive cells (green) and negative cells (no color).

PRG4-GFP secretion in conditioned medium

After 2 days of infection, the conditioned medium which contained the secreted PRG4-GFP was collected from transduced synoviocytes. PRG4 content in the conditioned medium was measured using a sandwich ELISA kit (LS-F35212: Lifespan Biosciences, Seattle, WA) at 450 nm in a UV-VIS plate reader (SpectraMax^® Plus, Molecular Diagnostics, San Jose, CA). The conditioned medium was concentrated up to 10-fold using Amicon^® Ultra 50K centrifugal Filters (MilliporeSigma, Burlington, MA) under 1,600 g at 4°C for 10 min. The concentrated conditioned medium (CCM) was mixed with 1% protease inhibitor cocktail (MilliporeSigma) and stored at −20°C. Protein concentration in CCM was determined by a bicinchoninic acid method using bovine serum albumin (BSA), and 2.5 mg of the CCM were loaded into a 4–20% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel. The blot was probed with a rabbit polyclonal antibody (ab28484; Abcam, Cambridge, MA) raised against PRG4.

The CCM was layered on the surface of an osteochondral explant to confirm the binding of PRG4-GFP protein. After incubating for 2 days, the explant was washed with HBSS and cryosectioned for confocal imaging.

Shear loading/friction test

Fresh bovine osteochondral plugs were anchored along the bone surface to the base plate with polycaprolactone. The centers of the plugs were subjected to 7 J/cm² impact injury and treated with various lubricants, HBSS, 0.3 μM BSA in HBSS (equivalent concentration with PRG4-GFP), 20% (v/v) bovine SF, or CCM (50, 100, and 200 μg/mL), for 30 min. Next, the plugs were moved into a custom-designed shear loading/friction testing device (Fig. 2A). A stepper motor actuator (NEMA 17 Stepper, Ultra Motion, Cutchogue, NY) connected with a 10-pound (lb) load cell (Honeywell, Columbus, OH) created shear loading on the flat surface of the osteochondral plugs. A constant compressive load was also applied on the specimen via a 1 kg stainless steel bar (Fig. 2B).

Figure 2.

Friction test. (A) The scheme of friction device. Shear load and normal load were applied to specimens via a stepper motor actuator and a 1 kg mass, respectively. (B) Osteochondral plugs (12 mm diameter and 10 mm height) impacted with 7 J/cm² using a drop tower were securely fixed with screws in a custom-designed mold. (C) Validation of the friction device using glass, Teflon™, ice, and cartilage (n = 4). (D) Effect of CCM on reducing friction coefficients (n = 4–10). Data are expressed as mean ± 95% CI. Color images are available online.

The specimen was securely fixed in a custom-designed mold and contacted with a 1 × 1 cm flat steel bar. The entire device was located in a 37°C incubator with 5% O₂/3% CO₂ and operated using LabVIEW software (National Instruments, Austin, TX). For testing the frictional coefficient, the preload was applied for 10 min, then the bar/platen was driven back and forth across the impact site over a distance of 3 mm at sliding velocity of 1.0 mm/s for three cycles. For shear loading, the preload was applied for 30 min after impact with an 11 mm diameter stainless steel ball traveling 5 mm at 1.0 mm/s for 1 h.

Friction coefficients were calculated by the kinetic equation, F = μN (F; friction force, μ; friction coefficient, and N; compression force). For validation of the device, the flat highly polished stainless steel platen was slid across glass, Teflon™ (polytetrafluoroethylene [PTFE]), ice, and articular cartilage to measure the friction coefficients (μ).^23–25

Confocal examination

The bovine osteochondral plugs and mouse knee joints were stained with 1 μg/mL calcein AM (Thermo Fisher Scientific; green) and 1 μg/mL ethidium homodimer (Thermo Fisher Scientific; red) and imaged with an Olympus Fluoview 1000 Confocal Laser Scanning Microscope. The sites were scanned from cartilage surface down to ∼200 μm in depth in 50 μm intervals. All confocal images were stacked in the Z plane by ImageJ software (NIH, Bethesda, MD) for representation purposes, and cell viability was calculated using Quantitative Cell Image Process (QCIP™), a custom automated MATLAB^® cell counting program.²⁶ For osteochondral plugs, the values at each time period (days 1, 3, and 7) were normalized to those of day 0.

Ethical statement

In conducting research using animals, the investigator(s) adheres to the laws of the United States and regulations of the Department of Agriculture. As specified on the Animal Care and Use Review Office (ACURO) website, the following assurances were obtained and maintained before performing animal research: Institutional Animal Care and Use Committee (IACUC) approval (#1403053 and 7021965), ACURO approval (dated July 23, 2014), The University of Iowa has an Animal Welfare Assurance (A3021-01, D16-00009), is registered as a research facility with the United State's Department of Agriculture (42-R-0004), and is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (000833).

Animals were individually housed in standard cages with 12/12 light/dark cycle, 18.9°C–20°C temperature, 30–50% humidity, and free access to water and diet. Randomization was used to allocate experimental groups for surgery and treatment. Primary experimental outcome was histological examination, and secondary outcomes were confocal examination, southern blot, and Equilibrium Partitioning of an Ionic Contrast Agent-Microcomputed Tomography (EPIC-μCT) examination.

Mouse DMM model

Twelve young adult male C57BL/6J mice (8 weeks old) were obtained from The Jackson Laboratory (Bar Harbor, ME) and allowed to acclimate for 1 week. Under anesthesia with isoflurane, right hind limbs were mounted on a triangular cradle with stifle joints positioned in ∼90° of flexion. The DMM model was created by cutting of the medial meniscotibial ligament.²⁷ Five microliter of 0.5 × 10¹⁰ v.g AAV-GFP (n = 6) or AAV-PRG4-GFP (n = 6) was injected into the intra-articular space using a Hamilton^® syringe fitted with a 30 gauge needle (Hamilton, Reno, NV). After 2 and 4 weeks, stifles from both sides were isolated for confocal examination, DNA extraction (n = 3 per group), and histological examination (n = 3 per group), respectively.

DNA extraction

DNA PCR was performed to confirm that intra-articular tissue cells were transduced with AAV-PRG4-GFP. Total DNA was isolated from the whole mouse stifle joints. Primers were designed across the EGFP and the lubricin gene (forward prime; 5′-GACTGCGACGCCCAATGTAA-3′, reversed primer; 5′-CACCTTGATGCCGTTCTTCT-3′) from Integrated DNA Technologies (Coralville, IA). PCR product was 768 bp. Platinum DNA Taq polymerase (Thermo Fisher Scientific) was used for the PCR reaction. According to the manufacturer's protocol, 34 cycles were performed to amplify the DNA template, and the reaction products were analyzed by agarose electrophoresis.

Rabbit ACLT model

Eighty male adult New Zealand White rabbits (NZW; 12–15 months old with an average weight of 4.5 kg) obtained from Covance were acclimated for 1 week and randomly divided into four groups; Sham control with AAV-GFP, Sham control with AAV-PRG4-GFP, ACLT with AAV-GFP, and ACLT with AAV-PRG4-GFP. Surgeries were performed essentially as described in our previous article.²⁸ In brief, all surgeries were conducted in a surgical suite with use of inhalation anesthesia and aseptic technique.

The ACL was exposed through a medial parapatellar arthrotomy with lateral patellar dislocation. The mid-substance of the ACL was lifted up with a surgical hook to expose the entire ligament. The ligament was transected with use of a number-11 scalpel. For the sham group, arthrotomy and ACL exposure were performed in an identical manner as for the transection groups, without ligament transection. After joint capsule was closed, half of the animals were treated with 200 μL of 5 × 10¹² v.g. AAV-GFP, or AAV-PRG4-GFP immediately following ACLT or Sham surgery, and the other half were treated 2 weeks after surgery. After 8 weeks posttreatment, both knee joints were harvested for EPIC-μCT and histological examinations.

Histological examination

The femorotibial joints of mice and rabbits were harvested at 4 and 8 weeks, respectively, following injection of AAV-GFP or AAV-PRG4-GFP for histological examination. The joints were dissected, fixed in 10% buffered neutral formalin (BNF) solution, and decalcified in 5% buffered formic acid. The specimens were embedded in paraffin, and 5-μm-thick sagittal sections were cut. The sections were then stained with Weigert's iron hematoxylin/Fast Green/Safranin-O. A modified Mankin's histological grading system was adapted to evaluate rabbit cartilage OA (Table 1).²⁹ In a blinded manner, three graders independently scored according to four categories; structure (0–6 points), cellularity (0–3 points), PG depletion (0–4 points), and tide mark (0–1 point).

Table 1.

A modified Mankin's histological grading system for cartilage osteoarthritis

I. Structure	II. Cellularity
0. Normal	0. Normal
1. Surface irregularities	1. Pyknosis, hypercellularity
2. Pannus and surface irregularities	2. Clusters
3. Clefts to the transitional zone	3. Hypocellularity
4. Clefts to the radial zone
5. Clefts to the calcified zone
6. Complete lack of cartilage

III. Proteoglycan Depletion	IV. Tide mark
0. Normal	0. Intact
1. Slight destruction	1. Destroyed
2. Moderate destruction
3. Severe destruction
4. No staining or denudation

The sections of mouse knee joints were stained with a 1:250 dilution of antilubricin and anti-GFP antibodies (Abcam) with a peroxidase-linked goat anti-rabbit IgG secondary antibody. VECTASTAIN^® ABC reagent and 3,3′-diaminobenzidine (DAB) (Vector Laboratories, Burlingame, CA) were used as a detection system according to the manufacturer's instruction. The stained slides were imaged using an Olympus VS110 Virtual microscopy system (Olympus America).

Rabbit knee joints were prepared for lubricin immunohistochemistry (IHC) staining. Hyaluronidase was used for antigen retrieval. The sections were incubated with mouse antilubricin antibody (1:1,000 MABT400: MilliporeSigma) at room temperature (RT) for 1 h, followed by ImmPRESS™ alkaline phosphatase-conjugated anti-mouse IgG (Vector Laboratories) for 30 min at RT. Vector Red (Vector Laboratories) was used to localize the antigen. Sections were counterstained with Meyer's hematoxylin.

The stained slides were imaged using an Olympus VS110 Virtual microscopy system. To quantify the lubricin coverage in the rabbit cartilage from lubricin IHC images, we followed a semiquantitative procedure using ImageJ Fiji software (Version 1.53c; NIH).³⁰ In brief, an image was deconvoluted with an RGB vector option, and area and length values were measured on the surface of cartilage (Supplementary Fig. S1). The area was divided by the cartilage surface length, and then normalized by the average value of Sham group.

EPIC-μCT examination

After fixation of rabbit joints, EPIC-μCT was performed for morphological and compositional changes, especially PG integrity, after AAV injections on an ACL ruptured rabbit knee model. The proximal end of each tibia was immersed in 2 mL of 30% Hexabrix 320 contrast agent (Covidien, Hazelwood, MO) and 70% ion-free phosphate buffer saline (PBS) at 37°C for 30 min for equilibration of the agent. We confirmed that there was no difference between fresh and BNF-fixed cartilage in the average X-ray attenuation levels, thickness, volume, and surface area. Proximal tibiae were scanned using a SkyScan 1176 μCT scanner (Bruker, Aartselaar, Belgium) according to the manufacturer's guideline for rabbit long bones.

Statistical analysis

Continuous normally distributed data, including transduction percentage, PRG4 content, friction coefficient, viability, and lubricin coverage, were analyzed using one-way analysis of variance (ANOVA) with Tukey post-hoc pairwise comparison. Nonparametric data of Mankin score were analyzed using Kruskal-Wallis ANOVA on ranks with Dunn-Bonferroni post-hoc pairwise comparisons. Data are expressed as mean ± 95% confidence interval (CI). The reliability of Mankin score was evaluated for interobserver (between observers) correlation coefficients using Kendall's τ-b test (>0.7: excellent, 0.501–0.7: good, 0.301–0.5: moderate, ≤0.3: low).³¹ SPSS Statistics software (Ver. 27; IBM, Armonk, NY) was used for all parametric and nonparametric analyses. Statistical significance was set at p < 0.05 with power over 0.8.

RESULTS

Transduction efficiency of AAV-PRG4-GFP

The transduction efficiency of AAV-PRG4-GFP was evaluated in monolayer culture of bovine chondrocytes, synoviocytes, and meniscus cells (Fig. 1C–F). Fluorescence microscopy revealed that transduction percentages in synoviocytes (49.4% ± 9.2%) and meniscus cells (43.8% ± 8.4%) were higher than in chondrocytes (5.3% ± 1.2%; p < 0.001) (Fig. 1C). Based on these data, synoviocytes were chosen for routine PRG4-GFP production. PRG4 contents in the conditioned medium from transduced synoviocytes were dose-dependently increased (Fig. 1G). The CCM from infected synoviocytes with AAV-PRG4-GFP was processed for immunoblot analysis (Fig. 1H). The PRG4-GFP fusion protein were strongly detected by the lubricin antibody at ∼250 kDa in AAV-PRG4-GFP, while there was no detection in both AAV-GFP and control groups. Finally, CCM from synoviocytes transduced with PRG4-GFP were layered on the surface of an osteochondral explant to confirm the binding of PRG4-GFP protein (Fig. 1I). The secreted protein was successfully bound on the cartilage surface with a strong GFP signal.

Effect of PRG4-GFP on cartilage friction

The lubricating function of attached PRG4-GFP was evaluated using the friction test. A custom-designed friction device (Fig. 2A, B) was validated using a highly polished stainless steel platen on glass, Teflon, ice, and articular cartilage (Fig. 2C). Compared with glass and Teflon, the friction coefficient of articular cartilage was significantly lower (0.015 ± 0.003), which was consistent with the literature showing coefficients of 0.0005–0.04 for articular joint.^32,33

Injured osteochondral plugs were treated with 50, 100, and 200 μg/mL PRG4-GFP or 20% bovine SF as a positive control (Fig. 2D). The friction coefficient of cartilage injured with a 7 J/cm² impact (0.073 ± 0.009) was ∼5-times higher than that of intact cartilage (0.015 μ ± 0.003; p < 0.001). This high friction value was reduced in CCM-treated groups. In particular, frictional coefficients for CCM-treated impacted cartilage ranged from 0.011 ± 0.007 for the 100 μg/mL dose (p < 0.001 vs. Impact) to 0.015 ± 0.005 for the 200 μg/mL dose (p < 0.001 vs. Impact). These values were similar to normal and SF-treated cartilage and were significantly lower than untreated impacted cartilage (p < 0.001 vs. impact for both doses).

Cytoprotective effect of PRG4-GFP on chondrocytes

Figure 3A shows representative images on the surface of cartilage at day 0 and 7 postimpact and postshearing. The cartilage plugs cultured in HBSS and 0.3 μM BSA were severely damaged with ∼20% to 40% initial death of chondrocytes after an impact-injury. This cell death dramatically increased at day 7 (40.2% ± 1.1% viability in HBSS and 57.8% ± 9.2% viability in BSA). In contrast, the number of dead cells was dramatically reduced in both 20% SF and 100 μg/mL CCM. Viability was significantly improved in both SF and CCM versus HBSS and BSA controls (Fig. 3B). The percentage of viable cells in SF and CCM was 78.6% and 81.7% at day 7 (p < 0.001 vs. HBSS and p < 0.003 vs. BSA).

Figure 3.

Cytoprotective effect of PRG4-GFP on chondrocytes during ex vivo culture of osteochondral plugs with cyclic shear loading. (A) Confocal images showed live cells (green) and dead cells (red) on the surface of plugs at day 0 and 7 postimpact. HBSS, Hanks Balanced Salt Solution; BSA, 0.3 μM bovine serum albumin; SF, 20% (v/v) bovine synovial fluid; CCM: 100 μg/mL CCM from AAV-PRG4-GFP-infected medium with synoviocytes. Scale bars = 250 μm. (B) Cell viability was quantified using QCIP™ and then normalized by an initial viability (day 0) (n = 3). Data are expressed as mean ± 95% CI. Color images are available online.

Feasibility test of AAV-PRG4-GFP in the mouse DMM model

AAV-GFP or AAV-PRG4-GFP was injected into the DMM mouse knee joints to verify the potential use in an in vivo environment. Green fluorescence was detected in joint cells, including articular cartilage (Fig. 4A), synovium (Fig. 4B), and infrapatellar fat pad (Fig. 4C). In contrast to the contralateral joint (Fig. 4D), all three tissues showed positive expression with green fluorescence in the AAV-PRG4-GFP injection group. DNA isolated from the whole stifle joint confirmed PRG4-GFP transduction (Fig. 4F). The transgene was detected in all three AAV-PRG4-GFP-injected mice (Fig. 4E).

Figure 4.

Feasibility test of AAV-PRG4-GFP in DMM model. (A–F) Six mice had AAV-GFP (n = 3) or AAV-PRG4-GFP (n = 3) injection in the right knee joint. At 2 weeks, confocal images showed apple green fluorescence emitted from the articular cartilage (A), synovium (B), and infrapatellar fat pad (C) in the right knee joint in AAV-PRG4-GFP injection. (D) A noninfected contralateral (left) joint. Southern blots of PRG4-GFP expression in AAV-PRG4-GFP injection (E) and AAV-GFP (F). Scale bars = 100 μm. DMM, destabilization of the medial meniscus. Color images are available online.

Efficacy of AAV-PRG4-GFP in the rabbit ACLT model

There was no issue for animal health problem and adverse events. The rabbits treated with AAVs immediately following ACLT surgery were evaluated. EPIC-μCT images indicated that AAV-PRG4-GFP inhibited cartilage damage compared to AAV-GFP injection (Fig. 5A, C). In load-bearing cartilage regions, higher PG distribution was observed in AAV-PRG4-GFP injected knees (Fig. 5B, D). Three graders independently scored cartilage integrity using a Mankin scoring system (Table 1). Reliability of interobserver correlation coefficients ranged from excellent to moderate; Grader A and B: 0.711, Grader A and C: 0.460, and Grader B and C: 0.430. The data of two Sham controls (AAV-GFP and AAV-PRG4-GFP) were combined because of no differences in Mankin scores (p = 0.117) and lubricin coverage (p = 0.370) (Supplementary Fig. S2).

Figure 5.

Efficacy of AAV-PRG4-GFP in a rabbit anterior ACLT model. Joints were injected with AAV-GFP (named AAV-G) or AAV-PRG4-GFP (named AAV-P) immediately after ACLT or Sham surgery and euthanized at 8 weeks. (A–D) EPIC-μCT images (n = 6); (A) AAV-GFP injection, (B) AAV-PRG4-GFP injection, (C) AAV-GFP injection in load-bearing cartilage region, (D) AAV-PRG4-GFP injection in load-bearing cartilage region. Safranin-O/Fast Green/Weigert's iron hematoxylin staining of (E) sham control, (F) AAV-GFP injection, and (G) AAV-PRG4-GFP injection. Lubricin immunohistochemistry staining of (H, K) sham control, (I, L) AAV-GFP injection, (J, M) AAV-PRG4-GFP injection. (N) Mankin score (n = 9–10). (O) Quantified lubricin coverage (%) (n = 7–14). White and black scale bars = 1 mm, blue scale bars = 20 μm. Data are expressed as mean ± 95% CI. ACLT, anterior cruciate ligament transection; EPIC-μCT, Equilibrium Partitioning of an Ionic Contrast Agent-Microcomputed Tomography. Color images are available online.

The results from the Mankin scoring indicated that cartilage degeneration at 8 weeks postoperation (postop) was greater in the ACLT joints (AAV-GFP: p < 0.001 vs. Sham, AAV-PRG4-GFP: p = 0.041 vs. Sham) (Fig. 5E–G, N). When given immediately after ACLT surgery, AAV-PRG4-GFP therapy reduced the post-ACLT severity of PTOA (AAV-PRG4-GFP: 3.31 ± 1.18 vs. AAV-GFP: 5.42 ± 0.43, p = 0.014). In Safranin-O/Fast Green/Weigert's iron hematoxylin staining, structural damage, clustered cells, and discontinued PG distribution were apparent in the AAV-GFP group.

The extent of lubricin coverage of joint surfaces was notable between groups of AAV-GFP and AAV-PRG4-GFP (Fig. 5H–M). Lubricin IHC staining revealed that the percentage of cartilage surfaces and superficial chondrocytes coated with lubricin was higher in the AAV-PRG4-GFP group versus the AAV-GFP group (p = 0.001) (Fig. 5O).

DISCUSSION

AAV-PRG4-GFP gene therapy enhanced lubricin expression and was chondroprotective in a rabbit PTOA model, a result that comports with extensive published evidence supporting the efficacy of lubricin supplementation in mitigating PTOA in injured joints.^13,14 Including GFP in the AAV-PRG4-GFP construct facilitated assessment of transduction efficiency and enabled the fate of the secreted product to be easily followed in vivo. However, it was unclear if the physical, chemical, and biologic characteristics of native lubricin were retained in the PRG4/GFP fusion protein. In that regard, in vitro tests showed that the fusion protein decreased friction on cartilage surfaces and protected chondrocytes from shear-loading induced death, while in vivo tests revealed that a single intra-articular injection of AAV-PRG4-GFP forestalled the development of PTOA in the rabbit ACLT model. These results indicate that the PRG4/GFP fusion protein is functional as a lubricin surrogate.

In a recent report of gene therapy for PTOA, HDV was used as a vector for PRG4 and IL-1RA gene delivery 2 weeks postop in a mouse DMM model of mild-to-moderate PTOA, and a cruciate ligament transection (CLT) model, which leads to more severe PTOA.¹⁵ In that study, combination therapy with both genes significantly retarded PTOA progression in the CLT model, whereas PRG4 delivery showed similar protective outcomes only when administered before CLT surgery.¹¹ This is consistent with our finding in the rabbit ACLT model that the effect of PRG4-GFP injection was influenced by timing, in that the efficacy of immediate postop injection was lost when treatment was delayed until 2 weeks postop. Although this contrasts somewhat with our finding in the rabbit ACLT model that monotherapy with PRG4-GFP was sufficient to significantly inhibit disease progression, we cannot rule out the possibility that dual therapy with an anti-inflammatory gene would have enhanced this effect.

The friction coefficient of osteochondral plugs decreased significantly with the addition of PRG4-GFP to the culture medium. We measured the friction coefficient from three cycles. However, some researchers report that friction coefficients increase with repetitive motion.³⁴ Thus, more lengthy friction tests are needed to acquire a more comprehensive understanding of the effects of PRG4-GFP on friction.

Injection of AAV-PRG4-GFP in mouse stifle joints resulted in persistent expression of the transgene by superficial and transitional chondrocytes, as well as by synoviocytes, adipocytes, and meniscal cells. Unfortunately, we could not quantify the expression rate due to no counterstaining. At 4 weeks posttreatment, cartilage integrity and GFP and lubricin expression were examined histologically at load-bearing areas of cartilage (Supplementary Fig. S3). The cartilage in AAV-PRG4-GFP injected joints was well protected compared with AAV-GFP injected joints which showed proteoglycan loss (asterisks) on the cartilage surface of tibia plateau. The expression of GFP and lubricin was strongly positive in AAV-PRG4-GFP group compared with AAV-GFP.

These results demonstrate that AAV-PRG4-GFP gene therapy enhanced joint surface lubricin. While the findings suggested that the treatment was disease-modifying, the 4 weeks endpoint was too early to fully evaluate this potential. Moreover, our results may have been affected by the use of relatively young (8 weeks old) mice with immature cartilage and subchondral bone.^35,36

It is important to select an appropriate PTOA animal model to evaluate the effect of exogenous PRG4 supplementation. Changes inPRG4 concentrations in SF depend on injury type and/or species. In humans, PRG4 is significantly depleted following ACL injury,¹² but is elevated in tibial plateau fractures.³⁷ Studies in larger animal species indicate that some forms of joint trauma induce significant synoviocyte-driven increases in PRG4 levels in joint fluids.^38,39 In this study, we selected a rabbit ACLT model, which leads to decreased PRG4 levels in both SF and cartilage.⁴⁰ We are currently evaluating the effects of AAV-PRG4-GFP in a porcine model of intra-articular fracture, which is expected to provide a PRG4-rich environment.

There were no significant differences in Mankin scores between the AAV-GFP and AAV-PRG4-GFP groups when therapy was delayed until 2 weeks postop (Supplementary Fig. S4), indicating an early and limited window of opportunity for treatment. The finding that delayed treatment failed to forestall PTOA is consistent with previous observations of the influence of timing on the effectiveness of PRG4 gene therapy in animal models.¹¹ The mechanisms underlying the superior efficacy of immediate treatment are unclear. While it is possible that gene transfer was less efficient at 2 weeks postop, it is also possible that immediate treatment protected joints by reducing acute inflammatory responses to surgical insult, an opportunity that is lost by 2 weeks postop. This is consistent with the ability of PRG4 to suppress proinflammatory pathways in synoviocytes via interactions with CD44 and innate immune receptors (TLR2 and 4).^41,42 This hypothesis, and the influence of timing on transgene expression will be addressed in future work.

CONCLUSION

In conclusion, our results build on an evidence from prior studies indicating that PRG4 gene therapy holds promise as a preventative treatment for PTOA. Our findings also suggest that the timing of administration is important for efficacy, an issue that will be addressed in future work.

AUTHORs' CONTRIBUTIONS

All authors approved the final version of this article to be submitted for publication. Conception and design: H.Z. and J.A.M. Collection, analysis, and interpretation of the data: D.S., H.H.C., M.J.B., D.C.F., I.S., L.R.J.C., A.K.S., and J.A.M. Drafting of the article: D.S. and H.H.C. Critical revision of the article for important intellectual content: D.S., H.H.C., and J.A.M. Provision of study materials: H.Z. and M.J.B. Statistical expertise: D.S. and J.A.M. Obtaining of funding: J.A.M. Animal surgeries: D.S., D.C.F., and E.B.P. Final approval of the article: J.A.M.

AUTHOR DISCLOSURE

The authors declare that a patent application based on the lubricin gene therapy technology described herein has been filed with USPS. Currently, there are no financial/professional relationships with companies or manufacturers to commercialize this technology.

FUNDING INFORMATION

The U.S. Army Medical Research Acquisition Activity, 820 Chandler Street, Fort Detrick MD 21702-5014 is the awarding and administering acquisition office. This work was supported by the U.S. Army Medical Research Acquisition Activity, through the Peer Reviewed Orthopedic Research Program (PRORP) under Award No. W81XWH-14-1-0163 and through the Peer Reviewed Medical Research Program (PRMRP) under Award No. W81XWH-18-1-0658. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the U.S. Army Medical Research Acquisition Activity. The research reported in this publication was also supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) of the National Institutes of Health under Award No. P50AR055533.

SUPPLEMENTARY MATERIAL

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4