Minimally invasive, sustained-release relaxin-2 microparticles reverse arthrofibrosis

Abstract

Substantial advances in biotherapeutics are distinctly lacking for musculoskeletal diseases. Musculoskeletal diseases are biomechanically complex and localized, highlighting the need for novel therapies capable of addressing these issues. All frontline treatment options for arthrofibrosis, a debilitating musculoskeletal disease, fail to treat the disease etiology – the accumulation of fibrotic tissue within the joint space. For millions of patients each year, the lack of modern and effective treatment options necessitates surgery in an attempt to regain joint range of motion (ROM) and escape prolonged pain. Human relaxin-2 (RLX), an endogenous peptide hormone with antifibrotic and antifibrogenic activity, is a promising biotherapeutic candidate for musculoskeletal fibrosis. However, RLX has previously faltered through multiple clinical programs due to pharmacokinetic barriers. Here, we describe the design and in vitro characterization of a tailored drug delivery system for the sustained release of RLX. Drug-loaded, polymeric microparticles released RLX over a multi-week timeframe without altering peptide structure or bioactivity. In vivo, intraarticular administration of microparticles in rats resulted in prolonged, localized concentrations of RLX with reduced systemic drug exposure. Further, a single injection of RLX-loaded microparticles restored joint ROM and architecture in an atraumatic rat model of arthrofibrosis with clinically derived endpoints. Finally, confirmation of RLX receptor expression, RXFP1, in multiple human tissues relevant to arthrofibrosis suggests the clinical translational potential of RLX when administered in a sustained and targeted manner.

One Sentence Summary:

Intraarticular injection of relaxin-2-loaded polymeric microparticles restores joint range of motion and reverses arthrofibrosis in rats.

INTRODUCTION

Protein therapeutics are one of the largest and fastest growing classes of pharmaceuticals in the last 20 years(1). Approved protein therapeutics now include monoclonal antibodies, fusion proteins, peptide hormones, enzymes, and growth factors(1–3). These modern biotherapeutics treat cancer (e.g., ipilimumab, brentuximab, idarucizumab), blood disorders (e.g., coagulation factor IX), cardiac and vascular diseases (e.g., alirocumab, evolocumab), dermatological conditions (e.g., pembrolizumab, nivolumab), diabetes (e.g., albiglutide, dulaglutide) and autoimmune disorders (e.g. adalimumab, belimumab)(2). However, there is a distinct lack of biotherapeutics targeting musculoskeletal diseases. This treatment gap leaves patients to traditional therapeutic avenues, such as pain and symptom management (e.g., glucocorticoids, immunosuppressive agents), physical therapy, and surgery, which are non-curative and can be highly invasive(4–6). Unfortunately, the scarcity of comprehensive treatments does not correlate to prevalence; orthopaedic disorders are responsible for over 70 million doctors’ office visits each year in the U.S.(7). This discrepancy likely results from the biomechanical complexity and localized nature of diseased musculoskeletal tissues, which points towards the need for customized biotherapeutics and targeted delivery mechanisms for specific disease phenotypes.

Arthrofibrosis, also known as adhesive capsulitis or frozen joint, is a widely prevalent musculoskeletal condition that impacts greater than 4% of the adult population(4, 8–10). It often develops idiopathically and also arises from events such as injury, surgery, or post-operative immobilization, causing long-term pain as well as loss in active and passive range of motion (ROM)(4, 5, 11). Pathologically, arthrofibrosis manifests as excessive extracellular matrix (ECM) protein deposition inside and surrounding the joint space due to dysregulated Transforming Growth Factor (TGF)-β1 signaling and subsequent myofibroblast differentiation and proliferation(5, 12). Current treatments for frozen shoulder include physical therapy, non-steroidal anti-inflammatory drugs (NSAIDs), oral or intraarticular injections of corticosteroids to reduce inflammation and pain, as well as intraarticular sodium hyaluronate injections(11). With no targeted therapeutics available that treat the underlying disease etiology, the pain and decreased quality of life stemming from arthrofibrosis pushes millions of individuals every year to undergo invasive surgical operations in an attempt to alleviate their symptoms(8). Common surgical treatments are manipulation under anesthesia (MUA), where fibrous adhesions are physically torn by mobilizing the joint beyond the point of conscious comfort, and capsular release, where the tightened, fibrotic synovium is surgically opened to increase mobility(13). MUA and capsular release often exacerbate the condition and incite a new fibrotic response(11). Additionally, these invasive surgical procedures, as well as currently available therapeutics, only treat the symptoms of adhesive capsulitis; they fail to target the underlying pathology of the disease – fibroblast dysregulation and fibrotic tissue accumulation in the joint space(4, 5, 11).

To address these treatment gaps, we have identified human relaxin-2 (RLX) as a promising antifibrotic and antifibrogenic therapeutic for arthrofibrosis. RLX is a 6 kDa peptide hormone that is naturally upregulated during pregnancy. In preparation for childbirth, RLX stimulates ECM remodeling to prime the cervix, birth canal, and circulatory system(14, 15). Binding of RLX to its cognate G-protein-coupled receptor, RXFP1, results in increased intracellular cyclic AMP (cAMP), upregulation of matrix metalloproteinases (MMPs), downregulation of tissue inhibitors of MMPs, and degradation of collagen fibers. Importantly, RLX also inhibits TGF-β1-mediated myofibroblast differentiation and subsequent fibrotic ECM protein production(12, 14–18). In vitro, RLX treatment of primary human fibroblast-like synoviocytes (FLS) suppresses a TGF-β1-induced myofibroblast phenotype, and decreases total ECM proteins. In vivo, it reduces fibronectin content and the overall contractile phenotype of an arthrofibrotic joint when repeatedly delivered (8).

RLX’s promising preclinical profile led to three late-stage clinical trials for the induction of labor, and for the treatment of cutaneous systemic sclerosis and acute heart failure. However, these clinical programs ultimately failed to meet primary efficacy endpoints(18–23). We hypothesize that this failure, in part, is a consequence of improperly addressed pharmacokinetic limitations, given that RLX’s half-life is just 1.6 hours in humans after intravenous injection(18). In an attempt to overcome these pharmacokinetic barriers, all previous clinical trials utilized continuous systemic infusion, which ultimately proved ineffective (20, 22). Additionally, continuous infusion for the treatment of arthrofibrosis is not possible through traditional methods due to limitations of the tissue architecture.

Targeted and sustained drug delivery methods present an opportunity to both overcome RLX’s pharmacokinetic limitations and account for the complex biomechanical nature of diseased orthopaedic tissue. To alter RLX’s pharmacokinetics, multiple groups report the successful delivery of RLX via nanoparticles, either as a conjugated protein or as a gene therapy, for the treatment of liver(24) and lung(25) fibrosis, as well as tumor-associated fibrosis in pancreatic cancer(26). Polymeric encapsulation of RLX in a microparticle offers several advantages for treatment of arthrofibrosis, including i) an alternative to chemical conjugation that does not modify the native protein structure, thus maximizing bioactivity; ii) a suspension formulation suitable for administration into a confined space or location; and, iii) a larger particle size to prevent cellular internalization and reduce synovial clearance, allowing for prolonged release times and higher local RLX concentrations. Poly(lactide-co-glycolide) (PLGA) microparticles, in particular, are desirable due to their capability of releasing both lipophilic and hydrophilic molecules, tunable fabrication parameters, and long acting release properties that increase bioavailability of the payload(27, 28). To this end, we describe the optimized design and use of a locally delivered, single intraarticular injection of human RLX-loaded PLGA microparticles that afford multi-week sustained release of bioactive protein. Further, we report restoration of mobility and architecture in an arthrofibrotic joint in a rat model by maintaining therapeutic concentrations of RLX in the synovial space and periarticular tissue, while minimizing systemic drug exposure. This work demonstrates a superior treatment option for both the symptoms and underlying pathology of arthrofibrosis. Additionally, it provides further evidence for the power of tailored drug delivery methods to revitalize RLX’s clinical efficacy and expands preclinical understanding of biotherapeutics targeting musculoskeletal diseases.

RESULTS

PLGA microparticles encapsulate and release biologically active RLX over 4 weeks in vitro

We previously demonstrated the efficacy of RLX as a treatment for arthrofibrosis, with the caveat that the treatment required multiple intraarticular injections due to RLX being rapidly cleared from the joint(8). To overcome pharmacokinetic barriers observed in this past study and encountered in previous clinical programs for RLX (19–21), as well as to increase the translational potential for this therapy, we developed a sustained-release delivery system. This system eliminates the need for multiple injections, maximizes local drug concentration, and reduces off-target RLX accumulation. Specifically, we synthesized RLX-loaded PLGA microparticles (RLX MPs). We utilized a water/oil/water double emulsion method (29), which was further optimized over a large design space (fig. S1) due to the complexity of encapsulating a bioactive protein in a polymeric matrix. The RLX MPs exhibited an average hydrodynamic radius of 5.27±0.33 μm and a polydispersity of 0.36 as measured by dynamic light scattering (fig. S1). RLX was encapsulated with an efficiency of 94.6% at 139 ng/mg PLGA (fig. S1). In vitro RLX release and polymer degradation were evaluated longitudinally under conditions that mimic the physiological synovial space. RLX MPs demonstrated a quasilinear release (r²=0.99) over the first 28 days, with >99% of encapsulated RLX released by day 50, as determined by enzyme-linked immunosorbent assay (ELISA) (Fig. 1A). Analysis of released RLX at discrete timepoints in the linear release region showed no loss in RLX-RXFP1 binding or activation of cyclic adenosine monophosphate (cAMP) signaling (Fig. 1B). Additionally, released RLX downregulated elevated collagen-I in TGF-β1-stimulated FLS cells over 4 weeks, demonstrating the maintenance of its antifibrotic activity after release (fig. S1).

Fig. 1. In vitro characterization of RLX MPs.

(A) RLX release over time (DMEM, 37°C, 100rpm) (n=3). Arrows indicated time points selected for bioactivity analysis. (B) Bioactivity of RLX released from RLX MPs at day 10 (D10) and day 25 (D25) from (A). Values are normalized to maximum luminescence of HEK293T-CRE-fLuc-CMV-RXFP1 after treatment with released RLX (n=3). (C) SEM micrographs of RLX MP degradation. (a) day 0, (b) day 7, (c) day 14, (d), day 21, (e) day 35, and (f) day 42; scale bar = 10 μm. Circular dichroism spectra of (D) RLX before and after the chemical and physical processing associated with microparticle encapsulation and (E) after long-term incubation at physiological temperatures (37 °C). (F) Relative cell viability of human fibroblast-like synoviocytes (FLS) and chondrocytes (NHAC) treated with RLX MPs and vehicle MPs as determined by MTS assay (n=3). All plotted data presented as mean ± STDev.

Scanning electron microscopy (SEM) analysis revealed the creation of smooth, spherical particles at the time of formulation, which degraded via bulk erosion, as anticipated for PLGA (Fig. 1C)(30–32). Changes in surface morphology appeared by day 7, with pores distributed throughout the particles that increased in size by day 21. By day 34, RLX MPs lost a substantial amount of polymer and decreased in sphericity, with structural collapse occurring by day 42 (Fig. 1C). Circular dichroism (CD) of RLX before and after encapsulation further confirmed the maintenance of protein secondary structure during microparticle fabrication and degradation (Fig. 1D–E). The negative peaks present at 208 and 222 nm denoted maintenance of the alpha helical structure of RLX after extended incubation at physiological temperatures, as well as after exposure to the organic solvents and mechanical and thermal forces involved in the encapsulation process. Additionally, RLX MPs were not cytotoxic to FLS cells and human chondrocytes (NHAC) across a broad range of RLX MP treatment dosages (Fig. 1F). All in vitro analyses confirmed the successful synthesis of biocompatible RLX MPs that release stable and biologically active RLX over a therapeutically relevant window.

RLX MPs provide sustained, elevated, and localized concentrations of RLX

We next performed a pharmacokinetic study in healthy Sprague Dawley rats using RLX radiolabeled with ¹²⁵I (¹²⁵I-RLX). We determined the pharmacokinetic profile of ¹²⁵I-RLX MPs compared to ¹²⁵I-RLX by quantifying total gamma radiation content in any given tissue, organ, or fluid of interest after intraarticular injection. This method allows for assessment of the intact synovial space without mechanical or chemical disruption of the joint, thus overcoming the limitation of synovial fluid sampling by accounting for tissue-adhered, receptor-bound, encapsulated, and released RLX.

We performed a single intraarticular injection of ¹²⁵I-RLX or ¹²⁵I-RLX MPs with fluoroscopic guidance and assessed ¹²⁵I-RLX localization to the joint, as well as its biodistribution over 48 h for non-encapsulated ¹²⁵I-RLX and 672 h for ¹²⁵I-RLX MPs. ¹²⁵I-RLX and ¹²⁵I-RLX MPs were injected at 2 μg/kg and 10 μg/kg, respectively, to reflect the dose per injection of RLX used in the efficacy study. This method allows for comparison of single injection RLX MP pharmacokinetics to that of multi-injection dosing of RLX(8) for the same total dose. ¹²⁵I-RLX MPs provided prolonged joint localization of RLX compared to ¹²⁵I-RLX (Fig. 2A). In line with synovial concentrations, periarticular tissue has a similarly localized profile (fig. S2). Pharmacokinetic modelling(33) of the observed data revealed that RLX MPs increased RLX’s effective synovial half-life (t_1/2) >1000-fold from 0.305 h to 367.3 h, compared to non-encapsulated ¹²⁵I-RLX (Table 1). Correspondingly, ¹²⁵I-RLX MPs increased RLX’s effective synovial mean residence time (MRT) from 2.16 h to 529.9 h. This increase in t_1/2 and MRT resulted from a significantly decreased effective synovial clearance (CL) [2.63 (mL/kg)·hr for RLX and 0.01 (mL/kg)·hr for RLX MPs](P = 0.001). Model validity is supported by the lack of significant difference between synovial volume of distribution (V_D), synovial steady-state volume of distribution (V_ss), or plasma CL for ¹²⁵I-RLX and ¹²⁵I-RLX MPs (Table 1) (P_Vd = 0.33, P_Vss > 0.99, P_CL = 0.43). The significant increase in calculated plasma V_ss for ¹²⁵I-RLX MPs further confirmed localized and sustained in vivo release of RLX from the MPs to the synovial space and surrounding tissue (P < 0.00001). Though the predicted plasma V_ss for a single intraarticular injection of ¹²⁵I-RLX was slightly larger than the anticipated plasma volume of a healthy Sprague Dawley rat(34), we hypothesized this difference is due to the natural sequestration provided by the synovial capsule (Table 1). The prolonged localization of RLX in the joint and periarticular muscle demonstrated the capability of RLX MPs to localize bioactive RLX to a biomechanically complex target-tissue.

Fig. 2. RLX MPs provide prolonged RLX concentration at the joint and minimize systemic exposure.

Observed and predicted pharmacokinetic data of RLX concentrations in the (A) treated synovial capsule and (B) plasma of healthy female Sprague Dawley rats (n=4^†/timepoint; n=32/treatment). Single intraarticular injection of ¹²⁵I-RLX (2 μg/kg) was compared to single intraarticular injection of ¹²⁵I-RLX MPs (10 μg/kg). RLX concentration is calculated from total γ count, normalized to total tissue mass, and adjusted for radioactive decay. Observed data is displayed as mean and 95% CI. Pharmacokinetic prediction is displayed as a solid line denoting mean with colored bands denoting 95% CI of predicted pharmacokinetic behavior. Modelled multiple intraarticular (mIA) ¹²⁵I-RLX injections (5 × 2 μg/kg) (orange) compared to a single intraarticular (sIA) injection of ¹²⁵I-RLX MPs (1 × 10 μg/kg) (blue) in the (C) plasma, (D) kidney, (E) synovial capsule, and (F) periarticular tissue. mIA RLX 0–48 h reflects PK profile determined from observed data, with mIA RLX 48–240 h determined by repeat injection modelling. sIA RLX MPs data represents PK profile determined from the first 240 h of observed data. Limit of detection (LoD) determined as 2σ of background measurement = 72.01pg/g-tissue. [^†n=5 was used for 4 d, 10 d, 14 d, and 28 d RLX MP timepoints to account for potential animal loss; no loss occurred.]

Table 1.

Pharmacokinetic parameters of the synovial capsule and plasma in rats

	RLX		RLX MPs†		Adjusted P value
Parameter	Synovial capsule	Plasma	Synovial capsule	Plasma	Synovial capsule	Plasma
k10 (h−1)	1.28 (1.29, 1.28)	0.169 (0.141, 0.223)	1.89·10−3 (1.77·10−3, 2.44·10−3)	7.77·10−3 (6.71·10−3, 1.04·10−2)	4.10·10−5	1.90·10−5
k12 (h−1)	0.797 (0.943, 0.591)	0.329 (0.364, 0.252)	-	5.42·10−2 (6.45·10−2, 3.97·10−2)	-	5.4·10−4
k21 (h−1)	0.455 (0.500, 0.379)	0.298 (0.373, 0.202)	-	1.15·10−2 (1.27·10−2, 9.83·10−3)	-	5.88·10−4
t1/2-α(h)	0.305 (0.280, 0.346)	0.954 (0.852, 1.15)	367.3 (283.7, 391.6)	9.60 (8.37, 11.9)	<1·10−6	6.3·10−4
t1/2-β(h)	2.72 (2.66, 2.88)	10.0 (10.7, 9.27)	-	562.2 (675.9, 395.5)	-	4.54·10−4
Cmax (μg/mL)	0.970 (1.29, 0.675)	0.016 (0.019, 0.014)	1.83 (2.78, 0.894)	1.37·10−3 (1.85·10−3, 9.27·10−4)	>0.999	2.92·10−3
VD (mL/kg)	2.06 (1.55, .96)	121.9 (107.3, 142.5)	5.46 (3.59, 11.2)	7.27·104 (5.40·103, 1.08·104)	0.329	4.08·10−4
CL (mL/kg)·hr	2.63 (2.01, 3.78)	20.6 (15.1, 31.8)	0.010 (6.36·10−3, 0.027)	56.5 (36.2, 111.8)	1.33·10−3	0.434
VD2 (mL/kg)	3.62 (2.93, 4.62)	134.9 (104.6, 178.1)	-	3.44·104 (2.75·104, 4.35·104)	-	1·10−6
CL2 (mL/kg)·hr	1.64 (1.47, 1.75)	40.1 (39.1, 35.9)	-	394.4 (348.2, 428.2)	-	3.40·10−5
AUC0–48 (μg/g)·hr	0.800 (1.05, 0.554)	0.094 (0.127, 0.062)	0.116 (0.158, 0.077)	0.025 (0.032, 0.019)	2.12·10−3	0.022
AUC0–240 (μg/g)·hr	4.93 (6.50, 3.42)	0.491 (0.661, 0.324)	0.491 (0.678, 0.313	0.058 (0.076, 0.40)	1.14·10−3	1.77·10−3
AUC0-Inf (μg/g)·hr	4.93 (6.50, 3.42)	0.491 (0.661, 0.324)	0.970 (1.36, 0.569)	0.107 (0.148, 0.066)	0.013	0.011
MRT (h)	2.16 (2.23, 2.00)	12.5 (14.1, 10.1)	529.9 (564.9, 409.3)	737.8 (908.2, 486.1)	9.0·10−6	9.90·10−5
Vss (mL/kg)	5.68 (4.49, 7.58)	256.9 (212.0, 320.5)	5.46 (3.59, 11.19)	4.17·104 (3.29·104, 5.43·104)	>0.999	2.0·10−6

^†RLX MPs fit to mono-exponential decay.

RLX MPs yielded a significant reduction in maximum concentration (C_max) in all analyzed organs and plasma (tables S1 to S3). On average, the calculated non-target tissue C_max decreased by 13.9-fold with RLX MPs compared to RLX. By localizing RLX to the site of injection, and thus the target tissue, RLX MPs reduced total systemic drug exposure, as calculated by area under the curve (AUC) analysis of RLX concentration over time (Table 1 and table S4). To better understand the exposure to RLX of synovial tissues and clinically relevant organs in a therapeutically-relevant regimen, and to allow for a more robust comparison between a single injection of RLX MPs and multiple injections of RLX, we modelled the pharmacokinetic profile of five repeated doses of RLX (2 μg/kg each) over 240 hours, matching the total dose administered in the RLX MPs. Modelling tissue exposure to RLX after repeat injection was possible due to the rapid pharmacokinetics of RLX, with >99% clearance at the time of dose re-administration in all analyzed organs, tissues, and fluid (Fig. 2A–B, fig. S2)(8). This rapid clearance decreased the chance that residual RLX concentration would confound repeat injection modelling and maximized animal welfare by avoiding an unnecessary 5× increase in cohort size to account for repeated doses. We used the predicted concentration at 48 hr post-injection as the basal concentration present in the organ, tissue, or fluid prior to each additional modelled intraarticular injection. By recursively sampling the predicted RLX concentration at 48 hr post-modelled injection, we determined the RLX concentration in the subsequent 48 hr period (data file S1). Using AUC_0–240hr analysis to calculate total tissue RLX exposure over time, a single intraarticular injection of RLX MPs significantly reduced total drug exposure compared to modelled multiple intraarticular injections of RLX in the plasma (Fig. 2C), kidneys (Fig. 2D), and heart (fig. S3) by nearly 10-fold, from 1.24, 0.491, and 0.178 (μg/g)·hr to 0.138, 0.058, and 0.021 (μg/g)·hr, respectively (P_plasma < 0.002, P_kidney < 0.001, P_heart < 0.004) (table S4). AUC_0–240hr encompassed the complete timeframe for drug exposure in the modelled repeat injection treatment group but did not include the entirety of the RLX MP release profile. Analysis of AUC_0-Inf for RLX MPs demonstrated a significant decrease compared to AUC_0-Inf for multiple injections of RLX in all modelled major organs and plasma (table S3 and S4). RLX MPs mitigated the concentration spiking associated with a multiple dose regimen and further decreased the total tissue exposure in clinically meaningful organs.

While ¹²⁵I-RLX content in major organs and plasma (Fig. 2B,fig. S2) directly reflected tissue exposure to free-RLX for both RLX and RLX MP treatment groups, it did not fully reflect exposure to released RLX in the synovial space. Due to sampling the entire glenohumeral joint, as well as surrounding periarticular muscle, which included encapsulated ¹²⁵I-RLX in non-degraded MPs, the observed synovial RLX localization (Fig. 2A) did not solely reflect synovial capsule exposure to free-RLX. We calculated the synovial tissue exposure to released RLX as the ΔC_RLX over time, accounting for C_RLX decay using the t_1/2 for the synovial capsule and periarticular muscle after a single injection of RLX (Fig. 2E and 2F). Similar multiple intraarticular injection modelling and AUC_0-Inf analysis demonstrated that RLX MPs decreased the total synovial capsule exposure from 4.93 to 0.970 (μg/g)·hr, (P = 0.013) with no statistically significant decrease observed in periarticular tissue exposure (tables S3 and S4). The decrease in local tissue exposure was likely due to the rapid synovial clearance of RLX as it is released from the MPs. After injection of RLX MPs, the particles begin to degrade, and the released RLX concentration quickly equilibrated in the joint and periarticular muscle to a therapeutically relevant range at low ng/mL, which were maintained over the complete observed duration of in vivo release (Fig. 2E and 2F). Thus, RLX MPs provided prolonged in vivo localization and release of RLX at the target tissue, while limiting systemic exposure in major organs.

A single intraarticular injection of RLX MPs rescues shoulder ROM

To assess the in vivo efficacy of RLX MPs, we utilized a previously validated atraumatic rat model of shoulder arthrofibrosis(35, 36). In this experiment, the ability of RLX to rescue rat shoulder ROM over 8 weeks was determined via live, longitudinal biomechanical measurement and terminal histological assessment. We recorded torque measurements associated with maximum internal (100°) and external (60°) shoulder rotation for both forelimbs prior to immobilization, which served to individually standardize the subsequent biomechanical measurements. Detailed descriptions and diagrams of experimental timeline, contracture induction, measurement procedures and the mechanical testing apparatus are found in supplementary methods, Figure 3, and figure S4. After induction of shoulder arthrofibrosis in the left forelimb through scapular-humeral fixation – performed without disrupting the joint capsule – for 8 weeks, animals were randomly assigned to one of four treatment groups: multiple injections of RLX (5 × 2 μg/kg, 50 μL), multiple injections of saline (5 × 50 μL), a single injection of RLX MPs (1 × 10 μg/kg, 50 μL), or a single injection of vehicle MPs (1 × 50 μl) (Fig. 3A). We identified the optimal RLX MP dose via a pilot dose-range finding study with full-dose RLX MPs (10 μg/kg, 1×50 μL) or one-third (RLX^1/3 MPs, 3.33μg/kg, 1 × 50μL) of the total dose previously used in vivo(8). ROM improved more rapidly with 10 μg/kg RLX MPs compared to 3.33μg/kg RLX MPs, which led to selection of 10μg/kg for further study (fig. S5). For all comparisons between treatment groups, except multiple injections of RLX vs. a single injection of vehicle MPs (P = 0.014), there were no significant differences at day 0 between internal, external, or total ROM measurements (table S5). Shoulder ROM was then measured over 56 days, where the determined internal and external ROM values were the rotation angles at which baseline torque values were achieved (Fig. 3A).

Fig. 3. A single injection of RLX MPs restores joint ROM.

(A) Experimental timeline for use of atraumatic rat model of arthrofibrosis, RLX treatment and ROM assessment. Longitudinal assessment of (B) internal ROM and (C) external ROM of contracted joint after remobilization and intraarticular injection(s). (D) Corresponding measurements of the internal (top) and external (bottom) ROM of the healthy, contralateral joint. Arrowheads indicate injection timepoint with color corresponding to treatment group, where orange denotes multiple intraarticular (mIA) RLX injections (2 μg/kg per injection) or mIA saline control injections and blue denotes a single intraarticular (sIA) injection of RLX MPs (10 μg/kg) or vehicle MPs (unloaded particles, dosed at equivalent PLGA mass). −56 d timepoint displays baseline ROM measurements performed prior to the induction of contracture. Joint remobilization and commencement of ROM assessment occurred at 0 d, directly prior to the first injection. Data presented as mean and 95% CI of mIA RLX (n = 10), mIA saline (n = 10), sIA RLX MPs (n = 9), and sIA vehicle MPs (n = 11). All measurements performed in technical triplicate on each joint, at each timepoint. * indicates comparison between sIA RLX MPs and sIA vehicle MPs. # indicates comparison between mIA RLX and mIA saline. Statistical significance determined by two-tailed t-test. */^#P < 0.05, **/^##P < 0.005.

Longitudinal measurement of internal ROM for each treatment group showed that a single injection of RLX MPs restored internal ROM as effectively as multiple injections RLX (Fig. 3B). Internal ROM recovery after 8 weeks of multiple injections RLX agreed with our previously published results(8). A single dose of RLX MPs and multiple doses of RLX achieved statistically significant differences in internal ROM from their controls, vehicle MPs and saline, at 9 days (P=0.002) & 2 days (P=0.045) after remobilization, respectively (Fig. 3B). Both treatment groups maintained increased internal ROM from the onset of difference through the final measurement at 56 days. Final internal ROM measurements (Mean ± 95% CI) for each group were 92.87° ± 6.84° (RLX), 62.30° ± 20.11° (Saline), 102.04° ± 3.18° (RLX MPs), and 69.30° ± 10.89° (Vehicle MPs) (table S5). Multiple injections of RLX achieved a significantly increased external ROM compared to control between day 35–42 but did not maintain this difference throughout the treatment period(8). In comparison, a single injection of RLX MPs achieved a statistically significant difference in external ROM compared to the vehicle MPs control at day 7 (P=0.03) and again at day 28 (P=0.03), which was then maintained through the final measurement at 56 days (P=0.04) (Fig. 3C,table S5). Internal and external ROM of the right, contralateral shoulder was not significantly altered throughout the study, demonstrating that intraarticular injection of RLX or RLX MPs did not induce joint laxity distal to the contracted shoulder (Fig. 3D).

Single intraarticular injection of RLX MPs restores healthy joint architecture

Histological assessment of joint architecture at 14 and 56 days after intraarticular injection supported the observed ROM restoration after a single injection of RLX MPs. Figure 4A is a representation of the internal and external architecture of a healthy and arthrofibrotic human shoulder joint for reference. To observe differences between treatment groups in joint architecture, fibrotic deposition, synovial membrane thickness, and cartilage health after treatment, we performed H&E, Masson’s Trichrome and Safranin-O staining of coronal cross sections of the rat shoulder joints (Fig. 4B–D). Representative images of the complete joint (Fig. 4B, fig. S6) are depicted, with the axillary pouch (Fig. 4C, fig. S7) and humeral head (Fig. 4D, fig. S8) shown at 40X magnification to highlight architecture of interest. Non-contracted rat joints demonstrated healthy architecture with clear delineation between the humeral head and glenoid surface as well as a defined synovial membrane, and no excess fibrotic deposition or presence of fibrotic adhesions (Fig. 4B and 4C). In contrast, contracture was visibly observed in both the saline and vehicle MPs control groups 56 days after joint remobilization and treatment (Fig. 4B and 4C), which corresponded with the measured long-term ROM restriction (Fig. 3). These joints presented with markedly decreased synovial space, thickened synovial membrane, as well as fibrotic adhesions (Fig. 4C). RLX and RLX MPs (Fig. 4B and 4C) treatment groups both displayed restored delineation of the humeral head and glenoid surface, absence of fibrotic adhesions, and a return of the synovial membrane to a non-contracted phenotype. Safranin O staining showed unperturbed articular cartilage, independent of treatment condition, supporting previous findings of minimal chondro-toxicity of RLX after intraarticular injection (Fig. 4D)(8).

Fig. 4. RLX MPs restore healthy joint architecture with no off-target remodeling.

(A) Depiction of healthy (left) and fibrotic (right) human shoulder joint. Representative histological images of fixed and decalcified rat joints stained with (B) Hematoxylin and Eosin (H&E) (scale bar = 1 mm) (C) Masson’s Trichrome (scale bar = 200 μm), and (D) Safranin O (scale bar = 200 μm) at 56 days after joint remobilization and initial intraarticular injection. Representative locations of axillary pouch (dotted line box) and humeral head (dashed line box) (B) are shown as higher magnification images in C & D. Dotted outline in sIA vehicle MPs in (B) highlights a region of immune cell infiltration. Black arrows identify fibrotic adhesions in the axillary pouch in (C), and capped lines demonstrate the location for measurement of axillary pouch thickness in quantified in F. Treatment groups presented top to bottom: Healthy, multiple intraarticular (mIA) saline injections, single intraarticular (sIA) vehicle MPs injection, multiple intraarticular (mIA) RLX injections, and single intraarticular (sIA) RLX MPs injection. (E) Quantification of excess collagen content (% total area with >120% blue pixels compared to red pixels) in Masson’s Trichrome staining, indicative of fibrotic tissue present in the axillary pouch and synovial membrane (n=5). (F) Quantification of axillary pouch thickness (n=5). (G) Histological quantification of immune activation in analyzed H&E sections (n = 5). Statistical significance determined by ANOVA * P < 0.05, ** P < 0.005, *** P < 0.002, **** P < 0.0001. All plotted data presented as mean ± STDev.

Next, we quantified total fibrotic deposition present in the axillary pouch using a previously validated Masson’s Trichrome image processing method in a blinded manner (Fig. 4E)(37). At 56 days, there was a significant difference in the calculated area of fibrotic tissue between RLX and saline (P = 0.0002), and RLX MPs and Vehicle MPs (P < 0.0001). Additionally, there was no observed differences between RLX, RLX MPs, or the non-contracted contralateral control. Blinded measurements of synovial membrane thickness in the axillary pouch directly inferior to the humeral head (Fig. 4C, Fig. 4F) yielded a significant difference between both treatment groups and their respective controls as well as restoration to baseline values at 56 days after joint remobilization and treatment (P_mIA-RLX = 0.0052) (P_RLX-MPs < 0.0001). In agreement with the measured ROM recovery, quantified fibrosis and axillary pouch thickness were not fully restored to baseline values at 14 days post joint remobilization and treatment. However, both RLX treatment conditions numerically trend towards the contralateral control (fig. S9).

RLX MPs modulate foreign body response in vivo

Blinded histological scoring of inflammatory infiltrate in the shoulder joint revealed a distinct difference between treatments. Contracted joints treated with multiple injections of RLX or multiple injections of saline presented with negligible inflammation at 14- and 56-days post-treatment (Fig. 4G, fig. S9). In contrast, contracted joints treated with RLX MPs or vehicle MPs displayed an increase in observed inflammation, indicated by the presence of lymphocytes and macrophages, compared to non PLGA-based treatment groups 14 days after administration (P_{RLX MPs-RLX} < 0.0001) (P_{vehicle MPs-saline} < 0.0001) (fig. S9, fig. S10). Both RLX MPs and vehicle MPs stimulated a lymphohistiocytic and foreign body response (FBR) (Fig. 4B). The FBR was identified by the presence of multinucleated giant cells (fig. S8). At 56 days, the RLX MPs exhibited significantly decreased inflammation compared to the vehicle MPs control (P < 0.016), with only mild inflammation histologically observed (Fig. 4G). Additionally, the presence of multinucleated giant cells was noted in 100% of analyzed histological sections of the vehicle MPs control (n = 5) at both 14 and 56 days. By contrast, multinucleated giant cells were only observed in 60% of the RLX MPs group at 14 days (n = 5) and were entirely absent at 56 days (n = 5). There was no observation of non-localized inflammation in the contralateral joints at either timepoint (Fig. 4G,figs. S9–S10). Histological evidence suggests that RLX MPs demonstrated self-regulation of the FBR that is commonly observed in PLGA-based drug delivery systems, which is a potentially critical aspect of RLX as a sustained-release antifibrotic. Neutrophil infiltration was absent in all treatment groups, excluding the possibility of acute infection. This finding was supported by serum cytokine analysis 2- and 7-days after intraarticular injection of RLX MPs or vehicle MPs, which demonstrated no significant elevation of common inflammatory cytokines associated with acute infection or inflammation compared to a healthy control (fig. S11).

RLX MP administration does not alter RXFP1 expression in rat joints

The key to RLX biotherapeutic activity is the presence and tissue specificity of the RLX receptor, RXFP1, in the joint space. The antibody used to identify RXFP1 was validated by flow cytometry, western blot, and immunohistochemistry (IHC) for both specificity and cross-reactivity between rat and human RXFP1. Western blotting of rat uterine tissue yielded a single band at ~95kDa, agreeing with published literature values (fig. S14)(39–42). Previously explored in the male rat joint(38), we confirmed that RXFP1 is highly expressed in the target synovial tissue of female Sprague Dawley rats in both the healthy and contracted states, as well as in treated and untreated joints (Fig 5A,fig. S12). IHC analysis demonstrated that RXFP1 is expressed by rat chondrocytes in the humeral head, which raises concerns of off-target remodeling. However, Safranin O staining did not show gross changes in cartilage architecture with RLX treatment (Fig. 4D,figs. S6–8). IHC analysis for RXFP1 expression demonstrated that the receptor is present in the target tissue and that local administration of RLX does not impact healthy bone or cartilage.

Fig. 5. RXFP1 is expressed throughout target tissue in rats and human patients.

(A) Coronal cross section with RXFP1 IHC staining of a healthy rat joint with no treatment (top) and contracted with a single intraarticular injection of RLX MPs (sIA RLX MPs) (bottom), confirming RXFP1 expression before disease induction and after treatment. Corresponding H&E staining is also shown. Scale bars = 100 μm. (B) Representative H&E and RXFP1 IHC staining of shoulder joint tissue from human patients (n = 5) Donor information listed in table S6. Individual cell types shown at right, highlighted by dashed boxes in the full view images at left. (a) fibroblasts - synovial, (b) vascular endothelial cells, (c) articular chondrocytes, (d) labral chondrocytes, (e) fibroblasts - labrum, (f) osteocytes. Scale bars = 100 μm in full view and 20 μm in cell type specific images.

RXFP1 is selectively expressed in human synovial joints

We fixed donor tissue from the shoulder joints of human patients (n = 5) immediately after surgical harvest and subsequently sectioned and analyzed these tissues for expression of RLX’s cognate receptor, RXFP1, via immunohistochemistry (Fig. 5B,fig. S13), following Institutional Review Board (IRB) approval. Tissue was harvested from orthopaedic procedures following acute traumatic shoulder injury. Complete donor classification and staining characterizations are found in table S6. IHC analysis revealed that RXFP1 was expressed in all analyzed human shoulder joints (5/5 donors). Additionally, we documented cell type-specific expression of the receptor. Fibroblasts found in the joint capsule and labral lining stained positively for RXFP1 (3/5 eligible samples) (Fig. 5B, a&e). Chondrocytes, both articular and labral, did not express RXFP1 (0/3 eligible samples) (Fig. 5B, c&d). Vascular endothelial cells, which have previously been shown to positively express RXFP1 in human tissue(43), also stained positive in a cross-section of synovial vessels (4/5 eligible samples) (Fig. 5B, b). Other cells which stained positively for RXFP1 but are of less clinical interest due to lack of drug-tissue interaction and access after intraarticular administration, are osteocytes (2/3 eligible samples) and macrophages (5/5 eligible samples), which were found in the humeral head and bone marrow (Fig. 5B, f). Confirmation of RXFP1 expression in human tissues of interest sets the stage for rapid clinical translation of RLX MPs.

DISCUSSION

One of many fibrotic musculoskeletal diseases, arthrofibrosis affects millions of people every year, causing joint discomfort, restricting ROM, hindering daily activities, and severely impacting emotional and physical wellbeing(4, 5, 11). Traditional treatment options focus on pain management or attempt to surgically restore ROM without addressing the underlying disease biology - excessive and aberrant deposition of ECM proteins in and around the joint space. Human RLX-2 is an endogenously-occurring antifibrotic peptide hormone that stimulates ECM remodeling and mitigates TGF-β1-induced myofibroblast differentiation(12, 14, 16, 17). Its combined antifibrotic and antifibrogenic properties poise it as both a symptomatic and etiological treatment for arthrofibrosis. However, a major hurdle in the clinical translation of RLX is its short half-life(8, 19–21). To overcome this barrier, we fabricated RLX-loaded poly(lactide-co-glycolide) microparticles, which provide sustained and localized release of RLX into the synovial joint space after a single intraarticular injection. The antifibrotic efficacy of RLX MPs provides evidence to support the clinical re-evaluation of RLX, given a localized delivery method.

RLX MPs are morphologically consistent, smooth and spherical, and release bioactive human RLX over ~40 days in vitro. To assess the localized and sustained delivery of RLX MPs in vivo, we performed pharmacokinetic and biodistribution analyses using radiolabeled ¹²⁵I-RLX and ¹²⁵I-RLX MPs injected into the shoulder joint of healthy Sprague Dawley rats. Assessment of gamma radiation in joints and tissues of interest, as well as major organs, demonstrates that RLX MPs localize drug to the target tissue for at least 28 days, whereas intraarticular injection of non-encapsulated ¹²⁵I-RLX is >99% clears from the synovial space within 48 hr, and then disseminates to major organs. RLX MPs provide a sustained, localized, and therapeutically-relevant dose of RLX via a single intraarticular injection that is unmatched even by multiple injections of non-encapsulated RLX.

Toxicologically, RLX MPs limit non-target tissue exposure. In two of the three previous clinical programs exploring RLX as a therapeutic, it presented a favorable toxicological profile at 30 μg/kg/day for treatment of acute heart failure (AHF)(19, 44) and 75μg/kg/day for induction of labor,(21) when given via continuous IV infusion over 48 and 24 hr, respectively. However, in the phase 3 clinical trial of systemic sclerosis patients presenting with moderate to severe diffuse cutaneous scleroderma, severe adverse renal events – defined as grade 3 or 4 hypertension, renal crisis, or doubling of serum creatinine - were observed after the cessation of 25 μg/kg/day of RLX continuously infused over 24 weeks(20, 22). As we observed release from RLX MPs over multiple weeks both in vitro and in vivo, minimizing renal drug exposure is essential for a successful clinical program. To this end, a single injection of RLX MPs decreased maximum and average renal exposure by >25- and >12-fold, respectively, on a per hour basis compared to multiple intraarticular injections of RLX. Full toxicology studies of RLX MPs were not performed here, but available biodistribution data points towards a favorable renal toxicity profile with this delivery method.

Given RLX’s pleiotropic nature, minimization of systemic RLX concentrations is key to limiting its effects in other tissues. For example, RLX’s cardiac bioactivity has been extensively documented in preclinical and clinical studies (16, 17, 19, 45–47), including mouse models of myocardial infarction that demonstrate an average plasma exposure concentration, C_avg, of 20–40 ng/mL RLX is bioactive(46). In our study, a single injection of RLX MPs results in a maximum plasma exposure, C_max, of 1.34 ng/mL and average plasma exposure, C_avg, of 0.308 ng/mL/hr over the entire observed release, while modelled multiple injections of RLX over 240 hr demonstrates a C_max of 16 ng/mL and a C_avg of 2.05 ng/mL/hr. RLX is naturally upregulated during pregnancy, with peak circulating plasma RLX concentrations at 0.92–1.0 ng/mL during the first trimester(48, 49). While both RLX and RLX MPs temporarily reach serum concentrations higher than 1.0 ng/mL (16 ng/mL and 1.34ng/mL, respectively), RLX MPs result in an average subphysiologic serum concentration compared to the prolonged supraphysiologic systemic concentration with multiple RLX injections. This reduced systemic exposure to bioactive RLX points towards a minimized on-receptor off-target profile for RLX MPs in major organs and tissues.

With respect to target tissue, a single intraarticular injection of RLX MPs into the synovial space of a contracted rat shoulder restores internal ROM to baseline compared to unloaded particles and multiple injections of saline. When administered at the same overall dose, multiple injections of RLX and a single injection of RLX MPs shows equivalent improvements to internal ROM, demonstrating that a sustained concentration of synovial RLX is as efficacious as multiple, repeated doses, but with a lower systemic exposure profile. In an attempt to further reduce systemic RLX concentration while maintaining antifibrotic efficacy, a pilot dose-range finding study for RLX MPs was performed during the development process. A lower dose of RLX MPs at 3.33μg/kg is less effective than 10 μg/kg with respect to speed of ROM recovery. RLX^1/3 MPs achieve a significant difference in internal ROM at day 42 (P=0.002), however significance was not maintained compared to the basal recovery of the vehicle MPs control group. This loss of significance is likely due to the lower total RLX dose in RLX^1/3 MPs, resulting in fewer responders in the measured population. It is possible that with a longer measurement period (e.g., 12 or 16 weeks), the RLX^1/3 MPs treatment group would achieve the same ROM recovery as 10 μg/kg RLX MPs; however, the express purpose of RLX MPs as a treatment for arthrofibrosis is to accelerate ROM restoration, as prolonged recovery time increases patient discomfort and reduces clinical feasibility. Dose-range finding studies using RLX MP concentrations between 3.33–10 μg/kg RLX could be performed to further reduce systemic RLX concentration if desired. RLX MPs at 10 μg/kg is currently considered the optimal dosing for allometric scaling in a large animal study or clinical translation.

While there is statistically significant improvement in the external ROM with a single injection of RLX MPs, average ROM recovery does not fully return to baseline, which possess an interesting question about the ability of this model to induce contracture with reduced external ROM that is capable of full recovery without surgical intervention. The limited improvement in external ROM with multiple doses of RLX may have resulted from anterior capsular ligamental fibrosis. Due to the nature of the joint immobilization, some internal ROM mobility may be maintained throughout the contracture period, as opposed to external ROM, which is almost completely abrogated by restriction of humeral movement to the plane of the scapula. This orientation may result in more severe fibrosis and limited external ROM recovery as seen in this study. Additionally, as external rotation is a less naturally undertaken murine motion, it is possible that attempting to replicate human ROM recovery in the rat shoulder is non-optimal when investigating external ROM. A large animal pharmacology model that is more similar to human joints(50, 51) will both support future drug development as well as aid in parsing out potential limitations of RLX MPs with respect to direction-specific ROM recovery.

The multiple injection saline control show larger variation than the single injection vehicle MPs control throughout the study and at the final internal ROM measurement (final total ROM with 95%CI: 80.23° ± 24.07°, 90.88° ± 11.47°, respectively). This difference may be due to the dilating and lubricating effect of multiple saline injections. A leading therapy for shoulder arthrofibrosis is hydrodilatation of the synovial capsule by injection of saline, sodium hyaluronate, or corticosteroids(4, 5, 11). The hydrodilatatory effect of repeated injections of saline, when combined with the physical manipulation of ROM measurements, results in greater variation in intragroup ROM recovery for the saline control compared to the vehicle MPs control. The only time at which the ROM variation after multiple injections of saline is less than that of a single injection of vehicle MPs is during the course of injections (0, 2, 7, & 9 day). The trauma and inflammation associated with repeatedly piercing the synovial capsule may contribute to minimal ROM recovery during this period as well as to future variability, depending on the stimulation of an additional scarring response. The variability of outcomes with multiple injections further demonstrates the benefit of a single injection system.

Though both RLX and RLX MPs are delivered directly into the synovial space to treat intraarticular fibrosis, fibrous hyperplasia in periarticular soft-tissue also contributes to arthrofibrosis ROM limitations(52). Further, fibrosis and incorporation of the fatty tissue in contracture, such as the subacromial fat pad in the shoulder or the infrapatellar fat pad in the knee, is observed in some arthrofibrosis patients(5). We hypothesize that the efficacy of RLX MPs as a treatment for arthrofibrosis in part resulted from elevated periarticular RLX concentrations. Our pharmacokinetic study of RLX MPs demonstrates that RLX concentration is maintained in the joint, as well as in the periarticular soft-tissue, which may explain the increased ability of RLX MPs to restore external ROM compared to repeated RLX injections. Additionally, the elevated RLX concentrations in soft tissue proximal to the injection site points towards the possibility for RLX MPs to treat a multitude of musculoskeletal fibroses, such as those arising from prolonged immobilization or neuromuscular disease.

We selected our previously validated atraumatic rat shoulder contracture model to assess preclinical pharmacology due to its clinically relevant endpoints(8, 35, 36). Clinically, ROM assessments to measure shoulder mobility and diagnose arthrofibrosis include flexion/tension, abduction/adduction, and internal/external rotation. Our model specifically measures the internal and external rotation of the shoulder joint, as these movements are some of the most commonly impacted and assessed when diagnosing adhesive capsulitis in human patients, and importantly, are common endpoints for arthrofibrosis clinical trials(53–55). Additionally, the rat glenohumeral joint possesses anatomical features that closely resemble the human joint(56). Previous clinical programs with RLX utilized endpoints which were not directly mirrored in preclinical research(18, 19, 57, 58). Just as RLX MPs are rationally designed to overcome barriers to efficacy observed in previous clinical programs, the atraumatic rat shoulder contracture model is a preclinical pharmacology model with outcomes that closely reflect clinical assessment. An overarching goal of this work is to demonstrate that rational design of drug delivery systems and proper selection of preclinical models can maximize the chance of clinical efficacy, even with a compound that has multiple previous clinical failures.

An inherent limitation of this efficacy model is the physical-therapy-like motion of the assessment. Manipulation of the shoulder during ROM measurement acts as a form of physical therapy that likely contributes to improvements in ROM, which is supported by the gradual increase in ROM observed in the saline and vehicle MPs control groups. However, these controls also demonstrate how physical therapy alone stimulates unfavorably slow recovery. Furthermore, ROM manipulation is performed on all treatment groups in this study, and thus the ability of a single injection of RLX MPs to improve ROM as effectively as multiple injections of free RLX, compared to the controls, demonstrates a superior and more translatable therapeutic. Additionally, most clinical treatment strategies are combined with physical therapy in order to minimize recovery time and achieve lasting restoration of ROM(11); therefore, we propose that the injection of RLX MPs will be concurrent with physical therapy when translated to the clinic. An additional and overarching limitation of this work is the use of a single sex cohort, which was selected due to physical limitations of the custom radial-torque sensor(8). The sensor is not easily adapted between the large variation in body mass, humeral length, and thoracic thickness that exists between male and female Sprague Dawley rats. The pharmacokinetic study was limited to a single sex to properly match cohorts in the efficacy study. Future validation of RLX MPs efficacy in males is essential for clinical translation.

Histological assessment of RLX MPs further supports their use as an antifibrotic treatment for arthrofibrosis. A single injection of RLX MPs and multiple injections of RLX reduce the presence of fibrotic adhesions on the humeral head and maintained the glenohumeral synovial joint space in a contracted shoulder compared to saline and vehicle MPs controls. Masson’s Trichrome staining show a reduction in axillary pouch fibroses with RLX MPs. Detailed histological assessment also reveal possible activation of inflammatory events in the presence of RLX MPs and vehicle MPs. A foreign-body response (FBR), observed in H&E staining of shoulders treated with a single injection of vehicle MPs or RLX MPs was initially identified due to a localization of macrophages and lymphocytes to the axillary pouch and joint-associated soft tissue, with confirmation of macrophage presence in these groups by IHC. As the observed FBR was not present in the repeated doses of RLX or saline groups, we attributed this response to the PLGA microparticles. Our theory of an excipient-driven FBR is supported by pharmacological and toxicological review of multiple FDA-approved PLGA drug delivery systems(59–62). Most poignantly, a FBR is reported to occur after intraarticular administration of a corticosteroid-loaded microparticle, ZILRETTA(63, 64). Different from many other PLGA drug delivery systems, the release of RLX from the microparticles mitigates the FBR, pointing towards potential anti-inflammatory effects of RLX that may aid its efficacy as an antifibrotic. This surprise finding is in line with recent reports that demonstrate the immunomodulatory action of RLX, specifically suppression of macrophage activation and infiltration(65). Despite reduction in inflammation with RLX MPs compared to vehicle MPs, inflammation is still elevated relative to the multiple injections of RLX and saline at 56 days. Two potential strategies to reduce this excipient-driven FBR are increasing RLX loading percentage in the RLX MPs, thereby decreasing total mass of delivered polymer, or by co-delivery of an anti-inflammatory, such as a corticosteroid. With corticosteroids being a common treatment for arthrofibrosis, co-administration may enhance the antifibrotic efficacy of the RLX MPs, as well as reduce the FBR. No observable neutrophil infiltration or gross inflammation is present in animals that received injections of PLGA microparticles, ruling out a FBR driven by acute infection. Finally, the absence of pro-inflammatory cytokines in the plasma at 2- and 7-days post injection further confirms a lack of PLGA microparticle-driven systemic inflammatory response. These results support the use of RLX PLGA microparticles as a safe and efficacious method to alleviate shoulder arthrofibrosis.

To support the translation of RLX MPs as a treatment for human arthrofibrosis patients, we performed immunohistochemical analyses for the RXFP1 receptor in relevant human shoulder tissues samples. Corresponding to our staining in the rat shoulder, RXFP1 is present in the human shoulder capsule and labrum, specifically in synovial and labral fibroblasts. However, RXFP1 is not expressed by articular or labral chondrocytes. These results agree with previously published staining of the synovial tissue and non-osteoarthritic cartilage in the human trapeziometacarpal joint(66). Not only does confirmation of human RXFP1 expression in connective tissue types most implicated in arthrofibrosis point towards antifibrotic efficacy in the clinic, but the corresponding lack of expression in chondrocytes suggests limited on-receptor off-target activity for RLX MPs after intraarticular administration. These findings together support our hypothesis that RLX will specifically act on the collagenous fibrotic tissue in the synovial joint space and the contracted capsule to reverse the arthrofibrotic phenotype and increase joint ROM.

In conclusion, we present a minimally invasive, sustained-release biotherapeutic for the treatment of arthrofibrosis. Clinically available treatments for arthrofibrosis require multiple injections of NSAIDs, corticosteroids or nerve-blockers into the synovial space in conjunction with extensive physical therapy or surgery(5, 11). RLX is superior to current therapies in that it signals for the breakdown of pre-established collagenous scar tissue as well as prevents development of further fibrosis, as opposed to only treating symptomatic pain. Further, sustained release of RLX from polymeric microparticles provides prolonged and locally elevated biotherapeutic concentration after a single injection. Combined, these factors reduce the need for painful and invasive procedures and minimize possible disease recurrence. Additionally, this work demonstrates the potential for modern biotherapeutics to treat musculoskeletal diseases with an emphasis on the need for properly tailored drug delivery systems and clinically relevant in vivo pharmacology models. RLX MPs are a prime candidate for the treatment of arthrofibrosis with the potential to expand the repertoire of available therapies for fibrotic musculoskeletal diseases.

MATERIALS AND METHODS

Study design

The study aimed to validate a sustained-release intraarticular delivery system of human relaxin-2 (RLX) to treat shoulder contracture after a single injection. In vitro characterization of the formulated microparticles showed their degradation and subsequent release of bioactive RLX as well as determined the bioactivity of RLX after encapsulation and release from the microparticle. ELISA, an endogenous receptor binding assay using the cognate RLX receptor, RXFP1, circular dichroism, as well as SEM micrographs were used to interrogate particle and RLX behavior during formulation and degradation. The in vivo pharmacokinetic profile of RLX MPs was determined using a radioisotope labelled RLX to measure synovial residence time and biodistribution after release from the sustained-release system. In vivo efficacy was evaluated using a previously established atraumatic rat model of arthrofibrosis. RLX MP efficacy was assessed by live ROM measurement on both a fibrotic and healthy shoulder, functioning as the internal control. RLX MP efficacy was also determined via histopathological analyses of joint architecture. Finally, the translational potential of RLX MPs was assessed via analysis of receptor expression in human musculoskeletal tissues of interest.

Timepoint and treatment group sizes as well as end points were selected based on previous experience with the respective animal model. Animals were randomly assigned to treatment and timepoint groups and given a numerical identifier. Experimental group sizes (n) are stated in all figure legends. Efficacy study exclusion criteria and RXFP1 antibody characterization are described in detail in the Supplementary Materials and depicted in figure S15, with 5 animals excluded across all treatment groups and timepoints. ROM measurements were not blinded. The numerical identifier associated with a specific animal was input into the custom software running the mechanical testing apparatus to measure ROM without any intervention by the researchers. Histopathological scoring and quantification were blinded. All experiments involving animals and biological specimens were approved by either the Institutional Animal Care and Use Committee (IACUC) at Beth Israel Deaconess Medical Center or the Institutional Biosafety Committee (IBC) at Boston University. Female, Sprague Dawley rats were used for all in vivo experimentation i) in order to provide replication and verification of previous work and ii) due to geometric limitations of the radial torque sensor. All analysis and collection of human tissue was approved by the Institutional Review Board (IRB) at Beth Israel Deaconess Medical Center and performed under protocol #2019P000951.

Statistical methods

Statistical analysis was performed using GraphPad Prism 9.0 or MATLAB. Numerical data are represented as mean ± standard deviation (in vitro microparticle characterization, histological quantification) or mean ± 95% confidence interval (in vivo). All histology images are representative. Paired or unpaired Student’s t-tests were used to compare two groups. One-way or two-way analysis of variance (ANOVA) tests were performed to compare more than two groups. Bonferroni-Dunn, Holm-Sídak or Tukey corrections were performed, where appropriate, to adjust for multiple group comparisons to protect against Type I errors(67). Adjusted P values < 0.05 were reported as statistically significant. Individual subject-level data are reported in data file S1.