Multi-arm Avidin nano-construct for intra-cartilage delivery of small molecule drugs

Abstract

Targeted drug delivery to joint tissues like cartilage remains a challenge that has prevented clinical translation of promising osteoarthritis (OA) drugs. Local intra-articular (IA) injections of drugs suffer from rapid clearance from the joint space and slow diffusive transport through the dense, avascular cartilage matrix comprised of negatively charged glycosaminoglycans (GAGs). Here we apply drug carriers that leverage electrostatic interactions with the tissue’s high negative fixed charge density (FCD) for delivering small molecule drugs to cartilage cell and matrix sites. We demonstrate that a multi-arm cationic nano-construct of Avidin (mAv) with 28 sites for covalent drug conjugation can rapidly penetrate through the full thickness of cartilage in high concentration and have long intra-cartilage residence time in both healthy and arthritic cartilage via weak-reversible binding with negatively charged aggrecans. mAv’s intra-cartilage mean uptake was found to be 112× and 33× the equilibration bath concentration in healthy and arthritic (50% GAG depleted) cartilage, respectively. mAv was conjugated with Dexamethasone (mAv-Dex), a broad-spectrum glucocorticoid, using a combination of hydrolysable ester linkers derived from succinic anhydride (SA), 3,3-dimethylglutaric anhydride (GA) and phthalic anhydride (PA) in 2:1:1 M ratio that enabled 50% drug release within 38.5 h followed by sustained release in therapeutic doses over 2 weeks. A single 10 μM low dose of controlled release mAv-Dex (2:1:1) effectively suppressed IL-1α-induced GAG loss, cell death and inflammatory response significantly better than unmodified Dex over 2 weeks in cartilage explant culture models of OA. With this multi-arm design, < 1 μM Avidin was needed – a concentration which has been shown to be safe, preventing further GAG loss and cytotoxicity. A charge-based cartilage homing drug delivery platform like this can elicit disease modifying effects as well as facilitate long-term symptomatic pain and inflammation relief by enhancing tissue specificity and prolonging intra-cartilage residence time of OA drugs. This nano-construct thus has high translational potential for enabling intra-cartilage delivery of a broad array of small molecule OA drugs and their combinations to chondrocytes, enabling OA treatment with a single injection of low drug doses and eliminating toxicity issues associated with multiple high dose injections.

1. Introduction

Osteoarthritis (OA) is a common chronic inflammatory disease of the whole joint affecting knees, hips, fingers, and low spinal regions, and is one of the most disabling diseases in developed countries with an estimated social cost between 1 and 2.5% of gross domestic product [1]. Traumatic joint injuries lead to development of post-traumatic OA (PTOA) within 10 years of injury in a majority of cases [2,3]. Following a joint injury, there is an immediate increase in synovial fluid levels of inflammatory cytokines (e.g. interleukin-1 (IL-1), interleukin-6 (IL-6), and tumor necrosis factor alpha (TNFα)) which diffuse into cartilage and rapidly initiate proteolysis and matrix loss [4–7]. While several drugs have known potential to inhibit OA, none have yet translated to clinical practice as they suffer from poor cartilage targeting and off-target side effects [8,9].

Currently, intra-articular (IA) injections to affected joints are the primary route for directly delivering pain and inflammation relievers [10,11], however, a majority of drug is rapidly cleared from the joint space, therefore requiring multiple injections of high drug doses which causes systemic toxicity [9]. This problem is further aggravated by the complex architecture of avascular cartilage containing a dense meshwork of collagen, interspersed with a high density of negatively charged aggrecan-associated glycosaminoglycans (GAGs) which hinder drug penetration [8–12]. Breakdown of articular cartilage triggers production of various inflammatory cytokines which disturb chondrocyte homeostasis, causing these cells to produce more matrix degrading agents than matrix generating agents. It is therefore critical for drugs to reach their cell and matrix targets sites, requiring novel strategies to enhance localization of OA drugs to target chondrocytes [8,13]. Currently there are no delivery systems that can (i) locally and safely target cartilage and enable drugs to penetrate through the full depth of tissue to reach chondrocytes and matrix targets, (ii) bind within cartilage to prevent their diffusion back out to the synovial fluid, and (iii) provide sustained drug release over several days to weeks. None of the drug delivery systems proposed thus far (e.g. drug-encapsulating micelles, liposomes, polymeric particles, aggregating hydrogels, etc.) are able to penetrate the cartilage or bind with its cell/matrix targets [14–19].

The high negative fixed charge density (FCD) of cartilage resulting from the high density of negatively charged GAG chains provides a unique opportunity to use electrostatic interactions for enhancing transport, uptake and retention of cationic drug carriers [8,20–24]. Using short length cationic peptide carrier motifs, it was shown recently that there exists an optimal net positive charge to deliver a drug of given size to a tissue of known FCD that will result in rapid penetration through the full thickness of cartilage before a majority of it is cleared from the joint space, providing the highest intra-cartilage uptake and long-term retention [25]. Optimal net positive charge on the carrier is chosen to enable weak and reversible binding with the intra-tissue negatively charged groups such that the drug and its carrier can penetrate through the full tissue thickness and not get stuck in the tissue’s superficial zones. Despite weak binding, the high negative FCD of aggrecan associated GAGs inside cartilage greatly increases the residence time of optimally charged cationic drug carriers [8,25]. Similarly, the cationic glycoprotein, Avidin, due to its optimal net size (< 10 nm hydrodynamic diameter) and charge (between + 6 and + 20) [20] was shown to penetrate through full thickness of rabbit cartilage following IA injection [26], resulting in a high intra-cartilage uptake ratio of 180 (implying 180× higher concentration of Avidin inside cartilage than surrounding fluid at equilibration). Further, Avidin was found to be present through the full thickness of cartilage two weeks following its IA administration in a rabbit anterior cruciate ligament transection (ACLT) model of PTOA [27]. Avidin was covalently conjugated with 4 mol of Dexamethasone (Av-Dex) using its four biotin binding sites [28] and administered in a single low dose IA injection one week following ACLT in a rabbit model [27]. Av-Dex suppressed injury induced joint inflammation, synovitis, incidence of osteophyte formation and restored trabecular properties at 3 weeks significantly greater than free Dex [27]. However, to deliver a single low dose of 0.5 mg Dex, a high dose of 20 mg Avidin was required due to low drug loading content of the conjugate design, which likely resulted in enhanced GAG loss from cartilage. This high dose of Avidin (that translates to about 200 μM assuming 1 mL synovial fluid in inflamed rabbit knee joint) can reduce intra-tissue osmotic swelling pressures owing to its cationic charge leading to decreased water content and potential loss of proteoglycans [12]. Avidin doses < 100 μM have been shown to not affect GAG loss, chondrocyte viability or biosynthesis rates of proteins and GAGs in bovine cartilage explants [12]. Therefore, in order to effectively use the cartilage homing property of Avidin for delivering OA drugs and enable its clinical translation, it is necessary to design a carrier system with increased drug loading content.

Here, we design a charge-based intra-cartilage drug delivery nano-construct – multi-arm Avidin (mAv), containing 28 sites for covalent conjugation of drugs (Fig. 1) compared to 4 sites in previous designs [27,28]. We use Dexamethasone (Dex) as an example small molecule drug and conjugate it to mAv (mAv-Dex) by using a combination of hydrolysable ester linkers derived from succinic anhydride (SA), 3,3-dimethylglutaric anhydride (GA) and phthalic anhydride (PA) that enable sustained controlled release of Dex in therapeutic doses over several days. Using in-vitro cytokine-challenged bovine cartilage models of OA, we show that a single low dose of mAv-Dex can effectively suppress cytokine-induced GAG loss, cell death and inflammatory response significantly better than free (unmodified) Dex over 2 weeks. With this multi-arm design, < 1 μM Avidin was needed – a dose that did not cause any GAG loss or cytotoxicity. This nano-construct thus has high translational potential for enabling intra-cartilage delivery of a broad array of small molecule OA drugs and their combinations to chondrocytes without the previously contemplated side effects.

Fig. 1.

Schematic of charge-based intra-cartilage drug delivery of the nano-construct multi-arm Avidin conjugated with a small molecule drug, Dexamethasone (mAv-Dex) by using hydrolysable ester linkers derived from succinic, glutaric and phthalic anhydrides (SA, GA, PA) in 2:1:1 M ratio enabling tunable (and long) drug release half-lives. Following its intra-articular administration, mAv-Dex can rapidly penetrate through the full thickness of negatively charged cartilage in high concentrations due to electrostatic interactions thereby creating an intra-cartilage drug depot. The optimal net positive charge of mAv enables its rapid and high intra-cartilage uptake and long-term retention via weak-reversible binding with negatively charged aggrecans. Therapeutic doses of drug (Dex) is then released via hydrolysis from mAv over several days, which can reach the chondrocytes to bind with its glucocorticoid receptors triggering downstream signaling pathways and suppressing OA associated catabolic activity. This intra-cartilage depot drug delivery platform can be used to deliver a variety of drugs or combination of drugs and enable OA treatment with only a one shot injection of low drug doses thereby eliminating toxicity issues associated with multiple injections of high drug doses that are currently needed to maintain sustained drug doses within the joint.

2. Material and methods

2.1. Materials

10 kDa 8-arm polyethylene glycol (PEG) amine hydrochloride salt was purchased from Advanced Biochemicals (Lawrenceville, GA). N-Hydroxysuccinimido (NHS)-biotin, 1-ethyl-3-(3 dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (NHSS), Avidin and Avidin-Texas Red conjugated, 4′-hydroxybenzene-2-carboxylic acid (HABA), 3.5 kDa molecular weight cut-off (MWCO), 7.0 kDa MWCO SnakeSkin dialysis tubing was purchased from Thermo Fisher Scientific (Waltham, MA). Proteinase-K was purchased from Roche Diagnostics (Risch-Rotkreuz, Switzerland). Dulbecco’s Modification of Eagle’s Medium (DMEM) was from Cellgro (Manassas, VA). HEPES, non-essential amino acids (NEAA), penicillin-streptomycin Antibiotic-Antimycotic (PSA) and trypsin-EDTA phenol red were purchased from Gibco (Carlsbad, CA). Ascorbic acid and L-proline were from Fisher Bioreagents (Pittsburgh, PA). Propidium iodide (PI) was obtained from Thermofisher Acros Organics (Geel, Belgium). Human recombinant IL-1α was from PeproTech (Rocky Hill, NJ). Antibodies for type II collagen immunohistochemistry was acquired from the Developmental Studies Hybridoma Bank (University of Iowa), while the Vectastain Elite ABC kit was from Vector Laboratories (Burlingame, CA). Dex, SA, GA, PA, dimethyl sulfoxide‑d₆ (DMSO‑d₆) containing 0.03% (v/v) tetramethylsilane, fluorescein diacetate (FDA), fluorescein isothiocyanate isomer I (FITC), dimethylaminopyridine (DMAP), resazurin sodium salt, Griess reagent, histology reagents and other salts were purchased from Sigma-Aldrich (St. Louis, MO).

2.2. Chemical synthesis

2.2.1. Biotinylation of 8-arm PEG

10 kDa PEG was biotinylated by reaction with NHS-biotin. Briefly, 10 mg (0.001 mmol, 1.0 equiv.) of PEG was dissolved in 500 μL of nanopure water and 1.7 mg (0.005 mmol, 5.0 equiv.) of NHS-biotin was dissolved in 500 μL of DMSO. NHS-biotin solution was then added dropwise to the PEG solution (5:1 M ratio) and reacted for 2 h under gentle rotation at room temperature using click chemistry between the NHS group in biotin and amine groups in PEG. Excess NHS-biotin was removed from the PEG-biotin conjugate solution using dialysis (7.0 kDa MWCO) for 24 h against phosphate buffer saline (PBS). Extent of biotinylation was confirmed using the HABA dye assay [29].

2.2.2. Synthesis of dexamethasone hemisuccinate (Dex-SA), glutarate (Dex-GA) and phthalate (Dex-PA)

Three carboxylated derivatives of Dex were prepared by reacting 36.0 mg Dex (0.092 mmol, 1.0 equiv.) with 46 mg of SA, 52.0 mg of GA or 67.0 mg of PA (0.458 mmol, 5.0 equiv.) in presence of 2 mg DMAP (0.015 mmol, 0.2 equiv.) as a catalyzer in 1 mL of pyridine. The reaction for Dex-SA was conducted in a round bottom flask purged with nitrogen gas for 24 h at room temperature. For Dex-GA and Dex-PA, the reaction time was 48 h at 37 °C. Following completion of the reaction, pyridine was evaporated with constant purging of nitrogen gas, and 4 mL of the cold solution containing 25 mL water and 10 mL concentrated hydrochloride acid (HCl) was added to the flask to precipitate Dex-SA, Dex-GA and Dex-PA out. A white precipitate was observed, which was stirred for 10 min and then centrifuged at 10,000 g for 5 min for 5 cycles. In each cycle, the supernatant was replaced with fresh cold solution of HCl. The final products of Dex-SA, Dex-GA, and Dex-PA were lyophilized, weighed and stored at −20 °C for future use. Their structures were confirmed using Proton Nuclear Magnetic Resonance (¹H NMR). The carboxyl groups incorporated to the Dex were verified with thin layer chromatography (TLC).

2.2.3. Conjugation of Dex-SA, Dex-GA and Dex-PA to PEG-biotin

Dex-SA was conjugated to PEG-biotin using EDC/NHS chemistry. Briefly, 5.0 mg of Dex-SA, Dex-GA or Dex-PA (0.010 mmol, 100.0 equiv.) was dissolved initially in 120 μL of DMSO and added 600 μL of 2-morpholinoethanesulfonic acid (MES) dropwise. Then, 19.2 mg of EDC (0.104 mmol, 104.0 equiv.) and 21.7 mg NHSS (0.092 mmol, 92.0 equiv.) were added to Dex-SA, Dex-GA and Dex-PA solution, and all of them were purged with nitrogen to activate the reaction for 30 min. Subsequently, 1.0 mg of PEG-biotin (0.100 μmol, 1.0 equiv.) was added to each of the solutions and reacted for 2 h at room temperature, purged with nitrogen gas. Upon completion of the reaction, the final product was dialyzed using 7.0 kDa MWCO membrane to remove the excessive reagents under 4 °C for 24 h. The pure product was then lyophilized and stored at −20 °C for future purposes. The formation of these three chemical compounds was then confirmed using ¹H NMR.

2.2.4. Loading of Dex-PEG-biotin on Avidin to synthesize mAv-Dex

The three Dex-PEG-biotin products from section 2.2.3 were reacted with Avidin in nanopure water for 30 min under gentle shaking at room temperature in 1:1, 2:1, 3:1, 4:1, 5:1, 6:1 and 8:1 M ratios to determine the ratio at which all four biotin binding sites of Avidin are occupied by Dex-PEG-biotin to form 1:4 mAv, which was confirmed using the HABA dye assay [25]. For subsequent drug release and biological studies, we have used 1:4 mAv configuration as it has higher number of sites for drug loading than 1:2 mAv and it has been referenced simply as mAv in following text. In one formulation, mAv-Dex-SA, mAv-Dex-GA, and mAv-Dex-PA were physically mixed together in 2:1:1 M ratio of Dex, abbreviated as controlled release mAv-Dex (2:1:1). We compared its bioactivity with mAv-Dex-SA, which is referred as fast release mAv-Dex throughout the manuscript. The products were then lyophilized and stored at −20 °C until further use.

2.3. Analysis

2.3.1. Degree of biotinylation using HABA colorimetric assay

The extent of biotinylation of PEG and the loading of PEG-biotin on Avidin were determined by using the HABA colorimetric assay [29]. Changes in absorbance of the HABA-Avidin complex at 500 nm due to competitive displacement by the biotinylated PEG was used to estimate the degree of biotinylation. HABA dye was dissolved in 10 mL of nanopure water (2.42 mg/mL) and filtered using 0.2 μm filter. Excessive HABA dye was added to Avidin solution to a final concentration of 0.82 mg/mL (initial absorbance of 1.2). 20 μL of graded concentrations of PEG-biotin or Dex-PEG-biotin were added to 180 μL of HABA-Avidin complex (1:1 through 8:1 M ratio of PEG-biotin to HABA-Avidin) that competitively displaced HABA from the biotin binding sites of Avidin thereby reducing the absorbance value. 100% PEGylation of Avidin was achieved when the change in absorbance achieved a plateau.

2.3.2. Gel electrophoresis

Conjugation of PEG-biotin to Avidin was confirmed by using native polyacrylamide gel electrophoresis (PAGE) in 7.5% separating gel. In brief, 12 μL of protein samples (~7.5 μg protein) in DI water were mixed with 4 μL of 2× Native Tris-Glycine sample loading buffer without heating. Since the isoelectric point of Avidin is 10.5 and the protein mobility depends on both the charge and molecular weight in the native PAGE gel, the electrode polarity had to be reversed (anode was inserted at the top of gel and cathode was inserted at the bottom of gel). Electrophoresis was performed for approximate 4 h in 1× solution of non‑sodium dodecyl sulfate tris-base running buffer at 200 V, 40 mA and 4 °C.

Native gel was stained using iodine solution and Coomassie Brilliant Blue R-250. Gel was fixed and then washed with deionized (DI) water for 20 min. Gel was then incubated in 5% barium chloride solution for 15 min followed by 3 washes in DI water. Subsequently, the gel was stained with potassium iodide and iodine solution for 5 min to identify free or conjugated PEG. Following this, the gel was stained with Coomassie Brilliant Blue R-250 for Avidin, and de-stained three times in 100 mL of 10% acetic acid solution for 1 h.

2.3.3. Ultra performance liquid chromatography (UPLC) and zeta potential

PEGylation of Avidin in 1:2 and 1:4 mAv was further confirmed by using H-Class Acquity UPLC (Waters Corp, Milford, MA) equipped with an Acquity UPLC BEH200 Size Exclusion Column (200 Å, 1.7 μm column, 4.6 × 300 mm) with 20 mM ammonium bicarbonate buffer as the mobile phase at 0.2 mL/min. Avidin was detected at 280 nm. Zeta potential of Avidin and mAv was measured in nanopure water at 0.45 mg/mL concentration using a Zetasizer Nano-ZS90.

2.3.4. Proton nuclear magnetic resonance (¹H NMR)

Modification of Dex and conjugation of Dex-SA, Dex-GA, Dex-PA to PEG-biotin were verified using 500 MHz ¹H NMR (Varian Inova. Agilent Technologies). 1–2 mg of solutes to be tested were dissolved in 700 μL DMSO‑d₆. In addition, ¹H NMR spectra of Dex-SA, Dex-GA, Dex-PA reacted with PEG-biotin using EDC/NHS chemistry were also confirmed. The obtained NMR data was analyzed using MestRe Nova software.

2.3.5. Dex loading content

Dex-PEG-biotin was hydrolyzed using 0.1 N hydrochloric acid overnight and neutralized against 0.1 N sodium hydroxide. The amount of Dex released was quantified by HPLC (Agilent Technologies 1260 infinity II) equipped with a Variable Wavelength Detector using a Poroshell 120 EC-C18 4.6 × 150 mm column. A gradient of solvent A (0.1% trifluoroacetic acid (TFA) in water) and solvent B (0.1% TFA in acetonitrile) was used. The concentration of solvent B was increased linearly from 5% to 65% over 15 min. Column temperature of 30 °C and a flow rate of 1.0 mL/min were used. Dex was detected at 254 nm. Drug loading content (DLC) was calculated as:

$$\text{DLC}=\frac{TotalDexencapsulated(\mathrm{g})}{TotalDexencapsulated(\mathrm{g})+MassofAvidin(\mathrm{g})}$$ (1)

2.3.6. In-vitro Dex release rates

Dex release rates from Dex-PEG-biotin were estimated in PBS at pH 7.4, 37 °C using dialysis tubing (7.0 kDa MWCO) with continuous shaking under sink conditions: Dex concentration was kept 10× lower than the saturation solubility of Dex in PBS. At different time intervals, 200 μL of release media was used to estimate the Dex concentration by HPLC, which was replaced by equal amount of fresh release media.

2.4. Transport studies

2.4.1. Equilibrium uptake of mAv in cartilage

Cartilage explants were harvested from the femoropatellar groove of 2–3-week-old bovine knees (Research 87, Boylston, MA) with a 3 mm diameter biopsy punch. The cylindrical plugs were then sliced to obtain the superficial 1 mm layer of cartilage, and frozen until use. Dual labeled mAv was synthesized by conjugating Texas Red labeled Avidin with FITC labeled 8-arm PEG. Cartilage disks were equilibrated in 300 μL of 8.5 μM of labeled Avidin, 1:2 mAv or 1:4 mAv in presence of protease inhibitors for 24 h in a 96-well plate at 37 °C on gentle shaking to prevent formation of stagnant layers. To minimize evaporation, the empty wells in the plate were filled with DI water and the plate was wrapped in parafilm. At the end of the experiment, the surfaces of each disk were quickly blotted with Kimwipes and the wet weight was measured. The equilibrium bath fluorescence was measured using a plate reader (Synergy H1, Biotek). The final concentration was calculated based on a linear calibration correlating fluorescence to concentration of labeled Avidin. The moles of solutes absorbed into the cartilage were calculated using the difference between the initial and equilibrium concentration of the bath. The concentration of solutes inside was calculated by normalizing the number of moles inside cartilage to the wet weight of the tissue. The uptake ratio (R_U) was defined as the ratio of the concentration of solutes inside the tissue (C_tissue) to that of the solute in the equilibrium bath (C_bath).

$$R_U=\frac{C_{tissue}}{C_{bath}}$$ (2)

To measure uptake of labeled Avidin and mAv in partially-degraded tissue modeling osteoarthritic cartilage, cartilage explants were incubated in 0.10 mg/mL trypsin-EDTA phenol red solution in PBS for 5 h, which induced about 50% GAG depletion [25]. GAG content in cartilage was measured using the dimethyl-methylene blue (DMMB) assay [30]. The explants were then rinsed multiple times and incubated in PBS containing protease inhibitors for 1 h to wash out trypsin before using for uptake experiments.

2.4.2. Confocal imaging to estimate depth of penetration into cartilage

A previously described transport setup was used to study 1-dimensional diffusion of solutes in cartilage [20]. Briefly, 6 mm half disk cartilage explants were mounted in the mid-section of the chamber. The upstream compartment was filled with either 9.5 μM of labeled Avidin or dual labeled 1:2 and 1:4 mAv. The transport chamber was placed in a petri-dish containing water to minimize evaporation and placed on a shaker inside an incubator at 37 °C. After 24 h of adsorption, a 100 μm slice was cut from the center of the explant and imaged using a confocal microscope (Zeiss LSM 700). Texas Red was excited using 555 nm laser line and FITC was separately excited using 488 nm laser line. Z-stack multilayer image of both channels (Red and Green) were obtained to visualize distribution of mAv conjugates. The maximum intensity of each channel was projected to the Z-axis.

2.4.3. Intra-cartilage diffusion kinetics

A custom designed transport chamber made out of transparent poly (methyl methacrylate) was used to measure non-equilibrium one-dimensional diffusion of solutes in cartilage as described previously [25]. The interior walls of the transport chamber were equilibrated in 0.5% nonfat-dried bovine milk solution in PBS for 15 min to lessen nonspecific binding of solutes to inner walls of the transport chamber. Subsequently, the compartments of chamber were rinsed with DI water. 3 μM solution of labeled Avidin or mAv was added to the upstream chamber. A 6 mm diameter cartilage disk of 400–600 μm thickness was placed between the upstream and downstream compartments. Concentration of fluorescently labeled solutes in the downstream chamber was measured over time by exciting the downstream solution using a 480 nm laser line and detecting the emission. Non-equilibrium diffusion curves were thereby obtained by plotting normalized downstream concentration (C_D) to upstream concentration (C_U) over time. Effective Diffusivity (D_EFF), which is the diffusivity of solutes in cartilage while binding interactions exist within the tissue was estimated as follows [20,31]:

$${\tau }_{\text{Lag}}=\frac{{\text{L}}^2}{6{\text{D}}_{\text{EFF}}}$$ (3)

where L corresponds to cartilage thickness, and τ_Lag is the time it takes to reach a steady state flux. τ_Lag was estimated using the time-axis intercept of the linear slope of the normalized concentration versus time [20].

2.5. In-vitro cartilage explant culture model of OA

3 mm diameter cartilage explants harvested from calf knee joints were equilibrated separately in serum free culture media containing 96.2% low-glucose DMEM, 1.0% HEPES, 1.0% NEAA, 1.0% PSA, 0.4% proline and 0.4% ascorbic acid for 48 h at 37 °C, 5% CO₂ for 2 days prior to treatment. Tissue explants for all treatment conditions were matched for depth and location to prevent any bias. To test the biological effectiveness of mAv-Dex, equilibrated cartilage explants were treated with or without IL-1α (2 ng/mL) for 16 days in combination with (i) a single dose of 100 nM free Dex, (ii) a continuous dose of 100 nM free Dex, (iii) a single dose of 10 μM free Dex (iv) a single dose of fast release mAv-Dex (10 μM Dex) or (v) a single dose of controlled release mAv-Dex (2:1:1) (10 μM Dex). To evaluate the effect of mAv carrier alone on cartilage health, a high one-time dose of 10 μM mAv, which is 10× higher than that used to deliver 10 μM Dex in condition (v) above, was also tested in the absence of IL-1α. Note that mAv here refers to 1:4 mAv configuration. Media was changed every 2 days and IL-1α was replenished at each medium change. Single dose treated explants were subjected to the drug and its carrier for only the first 2 days; in the following media changes, the media did not contain the drug, thereby simulating a single intra-articular injection in-vivo [28,32]. In the continuous dosing condition, Dex was replenished throughout the culture duration. IL-1α concentration was chosen as it represents a moderately aggressive cytokine treatment [5]. Previous work has shown that a sustained (continuous) dosing of 10–100 nM Dex throughout the culture duration is effective in suppressing cytokine induced catabolism [5,6,28]; we, therefore, compared single and continuous dosing with 100 nM free Dex. Single dose of mAv-Dex with effective Dex concentration of 10 μM was chosen for intra-cartilage drug depot delivery to provide a sustained drug dose of at least 10 nM Dex throughout the culture duration (calculations are explained in Section 3.4) and compared with the equivalent concentration of free Dex.

2.6. GAG loss and nitrite release from cartilage

After 16 days of culture, cartilage explants were weighed and digested in proteinase K. The cumulative GAGs released to the media and residual GAGs in the digested explants were measured using the DMMB assay [30]. Nitrite content was measured using the Griess assay as an indicator of nitric oxide (NO) release from tissues. Equal volumes of Griess reagent and culture media collected every two days were mixed and incubated at room temperature for 15 min, and absorbance at 540 nm was measured using a plate reader. Sodium nitrite was used as a standard.

2.7. Chondrocyte viability and metabolism

Chondrocyte viability was analyzed by staining cartilage explants with FDA (4.0 mg/mL) and PI (10.0 mg/mL) for live (green) and dead (red) cells, respectively. Slices were washed with PBS and imaged at 4× magnification (Nikon Eclipse Ts2R). The live and dead images were overlaid using ImageJ. At the end of the culture, tissue explants were incubated with media containing 1× resazurin sodium salt (alamarBlue assay) for 3 h in dark at 37 °C and 5% CO₂. Cell metabolic activity was estimated by measuring fluorescence at 530 nm excitation and 590 nm emission wavelengths [32].

2.8. Histology and immunohistochemistry

Cartilage explants were fixed in 4% formalin, embedded in 0.75% agarose for ease of handling, dehydrated in a graded series of ethanol and xylenes, and embedded in paraffin. Transverse sections (to obtain full thickness cartilage slices with superficial and deep zones) were taken at 5 μm thickness. These sections were stained with 0.5% Safranin O, 0.02% Fast Green and Weigert’s iron hematoxylin for GAG detection. Adjacent sections underwent antigen retrieval using 0.1% hyaluronidase, 0.1% pronase in PBS for 30 min at 37 °C, then were immunostained for type II collagen using 1 μg/mL mouse IgG1 (clone II-II6B3) or normal mouse IgG1 as isotype control. Antibody detection was performed using a VectaStain Elite ABC kit with 3,3-diaminobenzidine staining. Stained sections were imaged using a Zeiss Axioplan2 equipped with an AxioCam HRc camera.

2.9. Statistical analysis

Data is presented as Mean ± Standard Deviation. For all studies, n = 6–8 explants per condition and experiments were repeated using explants from at least 3 animals. A general linear mixed effects model was used with animal as a random variable. For comparisons between different treatment conditions, Tukey’s Honestly Significant Difference test was used. P < .05 was considered statistically significant.

3. Results

3.1. Synthesis and characterization of mAv

The 1:5 M ratio of PEG to NHS-biotin was optimal for synthesizing biotinylated 8-arm PEG using NHS ester reaction chemistry, resulting in every molecule of PEG being conjugated with one molecule of biotin as confirmed by matrix-assisted laser-desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Supplementary Fig. S1); PEG MW was estimated as 10,620 Da which increased to 10,902 Da following biotinylation showing that on an average 1.15 biotin per PEG were present. In addition to the mass spectrometry data, degree of biotinylation was also confirmed by using the HABA dye assay where the addition of biotinylated PEG to the HABA-Avidin complex displaced the HABA dye and reduced the absorbance value. Using the Beer-Lambert Law [29], an average of 1.28 ± 0.02 biotin per PEG molecule was estimated, which is consistent with the mass spectrometry data.

PEGylation in mAv was confirmed using the HABA dye assay, gel electrophoresis and UPLC. Fig. 2A shows the reduction in absorbance value with increasing molar ratio of biotinylated PEG to HABA-Avidin from 1:1 to 4:1, following which a plateau is achieved meaning that a majority of biotin sites on Avidin were occupied by PEG-biotin indicating the formation of 1:4 mAv. Fig. 2B shows native PAGE gel in reverse polarity used to confirm PEGylation in 1:2 and 1:4 mAv containing two or four 8-arm PEGs, respectively. PEG is stained as yellow with iodine and protein is stained as blue with Coomassie Brilliant Blue R-250. In PEG-staining (left), bands only appear in the PEG and mAv channels. However, in protein-staining (right), bands only appear in the Avidin and mAv channels. Therefore, bands at the same position in the mAv channels with both PEG-staining and protein-staining verified the formation of mAv.

Fig. 2.

Characterization of multi-arm Avidin (mAv) synthesis. A. (Top) Schematic of HABA dye assay: HABA binds with Avidin resulting in a high absorbance value but is competitively displaced by biotin or biotinylated PEG reducing the absorbance value. (Bottom) Titration curve of biotinylated PEG with HABA-Avidin mixture. Absorbance value dropped with increasing biotinylated PEG:HABA-Avidin molar ratio, and a plateau was achieved after 4:1 M ratio confirming that all four binding sites of Avidin (Av) were occupied by PEGs to form 1:4 mAv configuration B. Native PAGE gel (7.5%) of Av, PEG, 1:2 mAv and 1:4 mAv under reverse polarity stained with (left) iodine for PEGs and with (right) Coomassie Brilliant Blue R-250 for protein. C. UPLC analysis of (left) native Av with peak ‘a’ at 6.29 min and (right) of 1:4 mAv formulation containing a majority of mAv with 4 PEGs (peak ‘b’ at 4.38 min) and a secondary population of mAv with 2 PEGs (peak ‘c’ at 5.33 min). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

UPLC also confirmed that a majority of the population in 1:4 mAv had 4 PEGs conjugated to Avidin (peak ‘b’ at 4.38 min) followed by a secondary population of mAv with 2 PEGs (peak ‘c’ at 5.33 min, Fig. 2C). No peak for native Avidin (6.29 min) was found in 1:4 mAv confirming that Avidin was successfully PEGylated. UPLC of 1:2 mAv also confirmed that the majority of Avidin conjugated with two PEGs (Supplementary Fig. S2, peak ‘c’ at 5.35 min) followed by configurations containing three PEGs (peak ‘b’) and one PEG (peak ‘d’). Due to this heterogeneity, the physical and transport properties determined, represent the collective behavior of different populations in each formulation.

PEGylation did not reduce mAv’s zeta potential (ζ) suggesting minimum shielding of cationic charge; its net size was within the 10 nm limit (Table 1). As such, its intra-cartilage transport diffusivities (estimated using setup in Fig. 3A) were not affected either (Table 1). The hydrodynamic diameter of Avidin and mAv was estimated from their molecular weights using the Stokes-Einstein equation [33]. The net size of mAv was within the 10 nm limit enabling it to penetrate through the full thickness of cartilage similar to unmodified Avidin as further explained in section 3.2.

Table 1.

Zeta potential (ζ), net size (diameter) and intra-cartilage effective diffusivities (D_EFF) of Avidin, 1:2 mAv and 1:4 mAv. Data shown as Mean ± SD.

Formulation	Avidin	1:2 mAv	1:4 mAv
ζ (mV)	18.3 ± 0.5	20.3 ± 0.3	25.3 ± 0.7
Diameter (nm)	~7.0	~7.6	~8.1
DEFF (cm2/s)	8.4 ± 2.0 × 10−8	2.7 ± 0.5 × 10−8	3.5 ± 1.0 × 10−8

Fig. 3.

Transport properties of mAv nano-constructs in cartilage. A. Non-equilibrium diffusion curve of 1:4 mAv in cartilage showing normalized downstream concentration (C_D) to upstream concentration (C_U) versus time. τ_Lag was used to estimate effective diffusivity, D_EFF, in presence of binding interactions within the cartilage. L refers to cartilage thickness. B. Intra-cartilage equilibrium uptake of Avidin (Av), 1:2 mAv and 1:4 mAv after 24 h in healthy and 50% GAG depleted arthritic cartilage (* vs respective condition for Av, # vs respective healthy condition, p < .05). C. % retention of Av, 1:2 and 1:4 mAv inside healthy cartilage following desorption in 1× PBS and 10× PBS over 7 days D. Confocal microscopy images demonstrating full thickness intra-cartilage penetration of dual labeled 1:4 mAv from superficial zone (SZ) to deep zone (DZ) of healthy and 50% GAG depleted cartilage. Red channel shows Texas Red labeled Avidin and green channel shows presence of FITC labeled PEGs in mAv. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.2. Transport properties of mAv in cartilage

Non-equilibrium transport properties of mAv in cartilage were estimated from diffusion curves (Fig. 3A) using a setup described previously [25]. Effective diffusivity (D_EFF) of mAv decreased 2.5–3× compared to unmodified Avidin, implying slower diffusion rates as a result of PEGylation. There was no significant difference in diffusivities of 1:2 and 1:4 mAv (Table 1).

Equilibrium intra-cartilage uptake (R_U) of 1:2 mAv and 1:4 mAv was 1.5× and 1.3× lower compared to Avidin in healthy cartilage, respectively (Fig. 3B), however, mAv still maintained high uptake in cartilage (mean R_U of 96 and 112 for 1:2 mAv and 1:4 mAv, respectively). Furthermore, following 50% GAG depletion to simulate a mid-stage OA condition, as expected, uptake of Avidin, 1:2 mAv and 1:4 mAv dropped by 13.2×, 8.5× and 4.8×, respectively, compared to healthy condition due to the loss of negatively charged GAG binding sites (Fig. 3B). It should, however, be noted that these uptake values are still very high. For example, a mean R_U of 33 for 1:4 mAv implies 33× higher concentration inside cartilage than in the surrounding equilibration bath. 1:4 mAv carriers can, therefore, also be used for targeting early to mid-stage OA cartilage. Additionally, a majority of mAv was retained inside healthy cartilage over at least a 7-day period similar to unmodified Avidin when desorbed in 1× PBS. Desorption in saline bath with high salt concentration (10× PBS) resulted in complete desorption for all species within 24 h highlighting the dominant role of charge interactions (Fig. 3C).

1:4 mAv penetrated through the full thickness of cartilage within 24 h (Fig. 3D), similar to Avidin in both healthy and OA cartilage explants [20]. The green channel shows presence of PEG-FITC while the red channel shows Avidin-Texas Red distribution in cartilage.

3.3. Synthesis and characterization of PEG-Dex-SA, PEG-Dex-GA, PEG-Dex-PA

Incorporation of carboxyl group in compounds 2, 3 and 4 (Fig. 4A) was verified with the appearance of a yellow spot (stained by bromocresol green) on TLC plate and their yields were estimated as ~92%, ~72% and ~94%, respectively, using HPLC quantitative analysis. PEG-Dex-SA (5), PEG-Dex-GA (6) and PEG-Dex-PA (7) were synthesized by coupling compounds 2, 3 or 4 with 8-arm PEG-amine through EDC/NHS reaction (Fig. 4A). The chemical structure of compounds 1–7 were confirmed by ¹H NMR as shown in Supplementary Fig. S3.

Fig. 4.

Reaction scheme, release profile and cytotoxicity of PEG-Dex compounds. A. Schematic of Dex conjugation with 8-arm PEG-amine using cross linkers (R: SA, GA and PA) to form ester linkers with varying carbon spacer lengths (X: X₁, X₂ and X₃). B. Corresponding Dex release rates at 37 °C, pH 7.4 in PBS. Controlled release PEG-Dex (2:1:1) represents Dex conjugated using combination of ester linkers synthesized from SA, GA and PA in 2:1:1 M ratio of Dex. C. Mechanism of fast hydrolysis of PEG-Dex-SA (5) in PBS (pH 7.4). Carbonyl in amide bond, an electron withdrawing group (EWG), withdraws electrons from methylene and ester bonds (inductive effect) thereby decreasing ester bond’s electron density and making it more electrophilic (δ⁺) and reactive to nucleophilic attack from hydroxide ion (⁻OH) causing faster hydrolysis of the ester bond. Repulsion between hydrophilic PEG and hydrophobic Dex further strains the ester linker making it unstable. When X₁ is replaced by X₂ or X₃, the carbon spacer length is increased that weakens the inductive effect of carbonyl and donates more electrons to stabilize the ester bond. This also reduces the repulsive effects between PEG and Dex. D. Cytotoxicity test: % cell viability in cartilage explants following 48 h treatment with PEG-Dex-SA, PEG-Dex-GA, and PEG-Dex-PA.

The amount of Dex conjugated to PEG was determined by analytical reverse-phase HPLC. As shown in Table 2, compounds 5, 6 and 7 had 6.6 ± 0.5, 1.6 ± 0.4, 3.3 ± 0.5 molecules of Dex on one molecule 8-arm PEG, respectively. After conjugating to Avidin in 1:4 mAv configuration, the DLC for mAv-Dex-SA, mAv-Dex-GA, and mAv-Dex-PA were calculated as 15.7 ± 1.0%, 3.8 ± 0.9%, and 7.8 ± 0.1%, respectively.

Table 2.

Hydrolysis half-lives of ester linkers between 8-arm PEG and carboxylated derivatives of Dex. Molar ratio of Dex conjugated with PEG for each configuration and the corresponding drug loading content (DLC). Data shown as Mean ± SD.

Ester linker	PEG-Dex-SA	PEG-Dex-GA	PEG-Dex-PA	PEG-Dex (2:1:1)
Half-life (h)	6.8 ± 0.2	79 ± 1.8	86 ± 2.3	38.5 ± 1.5
Dex: PEG (molar ratio)	6.6 ± 0.5:1	1.6 ± 0.4:1	3.3 ± 0.5:1	4.6 ± 0.5:1
DLC (%)	15.70 ± 1.0	3.81 ± 0.9	7.85 ± 0.1	11.1 ± 0.8

About 70% of Dex was released from PEG-Dex-SA in PBS within the first 24 h resulting in a short release half-life (t_1/2) of about 6.8 h (Fig. 4B). We addressed this by increasing the carbon spacer length between the ester and adjacent amide by replacing SA with GA or PA to form carboxylic acid derivatives of Dex, which significantly increased the mean release half-lives to 79 h and 86 h for PEG-Dex-GA and PEG-Dex-PA, respectively. These spacers weaken the adjacent carbonyl groups’ inductive effect (which reduces the ester’s electron density) by donating multiple electrons and stabilizing the ester bond (Fig. 4C) [34]. These linkers provide a simple aqueous based way of effectively controlling release rates depending on drugs, disease severity and tissue type. Based on the drug release profiles, compounds 5, 6 and 7 were physically mixed at a 2:1:1 M ratio by Dex content (PEG-Dex 2:1:1) to develop a controlled release formulation that combined the effects of burst-release from PEG-Dex-SA to reach therapeutic levels within a short period of time and sustained drug release over two weeks from PEG-Dex-GA and PEG-Dex-PA. The resulting Dex release half-life was measured as t_1/2 = 38.5 ± 1.5 h. In vivo release half-lives of linkers are expected to be altered due to the presence of proteases, other enzymes and binding with other synovial fluid proteins. Prior work, however, has shown that Dex release via ester hydrolysis did not change in synovial fluid and was similar to that in PBS [28].

Finally, before conjugating PEG-Dex-SA to Avidin to form fast release mAv-Dex or a combination of PEG-Dex-SA, −GA and -PA in 2:1:1 M ratio to Avidin to form controlled release mAv-Dex (2:1:1) and testing its bioactivity using cytokine challenged in-vitro cartilage explant culture models, cartilage explants were treated with compounds 5, 6 and 7 (with high equivalent concentration of 100 μM Dex, which is 10× higher than that used in subsequent bioactivity experiments) for 48 h. Treated explants showed similar cell viability as the untreated control condition (Fig. 4D), confirming no cytotoxic effects.

3.4. Bioactivity of single dose of mAv-Dex in IL-1α treated cartilage explant culture OA model

We first compared the effectiveness of single versus continuous (sustained) doses of 100 nM free Dex in suppressing IL-1α induced catabolic activity in cartilage. Treatment with IL-1α significantly increased GAG loss over the 16-day culture period compared to control (p < .0001) (Fig. 5A). A single dose of 100 nM Dex was not effective and a continuous dose of Dex was needed to effectively suppress IL-1α induced GAG loss throughout the culture duration to levels similar to the healthy condition, highlighting the importance of maintaining a sustained drug concentration throughout the culture period.

Fig. 5.

Effectiveness of single dose mAv-Dex treatment in suppressing IL-1α induced GAG loss and chondrocyte death. % cumulative GAG loss over 16 days: IL-1α treated cartilage explants treated with A. a single or continuous (C) dose of 100 nM free Dex, B. single doses of 10 μM free Dex, mAv-Dex or mAv-Dex (2:1:1). A concentration of 10 μM mAv alone in absence of IL-1α was also tested and compared with the control condition (* vs control, # vs IL-1α, $ vs single dose Dex condition, ^ vs mAv-Dex, p < .05. Statistical markers are colour coordinated with the curves. All data enclosed within statistical markers are significant.) C. Chondrocyte viability after 16 days of culture. Green indicates viable cells and red indicates non-viable cells. Scale bar = 200 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To evaluate the effect of single dose of mAv-Dex that creates an intra-cartilage drug depot, a concentration of 10 μM Dex was used to provide sustained therapeutic doses of at least 10 nM over the entire culture duration and the biological effects were compared with that of free Dex. The dosing calculations are explained here: Previous work has shown that a sustained (continuous) dosing of 10–100 nM Dex throughout the culture duration is effective in suppressing cytokine induced catabolism [5,6], which corresponds to a minimum of 0.0001–0.001 nmol inside cartilage, respectively. An initial dose of 10 μΜ mAv-Dex in the media equilibrates within 24 h resulting in 112× higher concentration inside cartilage than the outside media bath owing to charge interactions (Fig. 3B). Therefore, based on mass balance between the explant and surrounding media, the equilibrium Dex concentration in media is estimated as 2.1 μM, implying an intra-cartilage concentration of ~236 μM (112 X 2.1 μM), which is equivalent to 2.36 nmol of Dex inside the cartilage disk of 10 μL volume before the first media change at 48 h. Dex release half-life from mAv-Dex in 2:1:1 formulation is 38.5 h (by fitting a single-phase exponential decay model), which means 10 half-lives over 16-day culture period. Based on this, [(1/2)^10] of the initial administered dose will be present in cartilage after 16 days which equals to 0.0023 nmol (0.1% of 2.36 nmol). Therefore, a one-time dose of 10 μM mAv-Dex was used in our experiments as it can maintain therapeutic Dex dosage of at least 10 nM over 2 weeks to suppress cartilage degradation.

As shown in Fig. 5B, a single dose of 10 μM Dex significantly suppressed IL-1α induced GAG loss throughout the duration of culture (p < .0001 compared to IL-1α condition) but remained significantly elevated compared to the control condition (p < .0014 starting at Day 4), highlighting its short-term therapeutic benefit. Both fast and controlled release mAv-Dex suppressed IL-1α induced GAG loss throughout the culture duration significantly greater than single Dex dose (from Day 4; p < .0037), bringing levels down to that of the untreated control. Controlled release mAv-Dex (2:1:1) had a longer lasting therapeutic response (throughout the culture duration) than the fast release mAv-Dex whose GAG loss levels became statistically different from control condition by Day 8 (p < .04). mAv carrier in 10× higher concentrations than used in mAv-Dex (2:1:1) was associated with no GAG loss. In fact, GAG loss levels measured were lower than that for the control condition, a phenomenon that warrants more probing. We confirmed that the presence of positively charged Avidin did not hinder the activity of cationic DMMB assay as the total GAG measured (i.e. GAG lost to media over the entire culture duration + residual GAG in the cartilage explant) was similar in the untreated control and mAv treated conditions i.e. we were effectively able to measure the total GAG content using the DMMB assay.

Similar trends were observed in Day 16 chondrocyte viability images (Fig. 5C); treatment with IL-1α enhanced cell death which was rescued effectively by the continuous dose but not with a single dose of 100 nM Dex. On the other hand, single dose of both fast and controlled release 10 μM mAv-Dex rescued cell viability similar to continuous dosing condition while a single dose of 10 μM free Dex was not as effective. mAv treatment did not cause any cell death. Some cell death shown in the superficial zone was observed in control explants, along with the location of harvesting from the joint. Additionally, excision of tissues from the joint using punches can also lead to cell death at the cut surfaces [35].

Nitrites are a reactive oxygen species (ROS) triggered by an upregulation in inflammatory activity such as the presence of IL-1α in an OA environment [36]. As expected, IL-1α treated explants produced 14.4× and 7.7× higher amounts of nitrite compared to control at 2 and 8 days of culture (p < .0001), respectively (Fig. 6A). Treatment with all Dex containing conditions significantly lowered nitrite synthesis at Day 2 (p < .047 vs IL-1). By Day 8, continuous Dex dose resulted in 3.5× lower nitrite levels compared to single dose of 100 nM Dex. A single dose of mAv-Dex (both controlled and fast release formulations) suppressed nitrite synthesis levels by 1.5–2× compared to a single dose of 10 μM free Dex, bringing levels close to control levels at Day 8. Lastly, there was no difference between mAv and untreated control conditions confirming no associated toxicity/inflammatory effects of mAv. IL-1α reduced cell metabolism rates, however, were not rescued by either Dex or mAv-Dex (Fig. 6B) suggesting the need to incorporate a pro-anabolic drug for a comprehensive and a more effective OA treatment [4,5]. mAv treatment did not suppress cell metabolism levels.

Fig. 6.

Effect of mAv-Dex on nitrite release and cell metabolism in IL-1α treated cartilage. A. Nitrite released to the medium per mg cartilage tissue on Day 2 and Day 8 and B. Chondrocyte metabolism measured as relative fluorescence units (RFU) using alamarBlue assay on Day 16. (* vs untreated, # vs IL-1α, $ vs single dose Dex of same concentration, p < .05. Statistical markers are colour coordinated with the bars.)

Finally, GAG and collagen content through the explant tissue depth was assessed by histology and immunohistochemistry (Fig. 7). IL-1α treated explants showed visibly low levels of Safranin-O staining compared to controls, indicating significant GAG loss. This IL-1-stimulated GAG loss was mostly rescued by a continuous 100 nM Dex dose, particularly below the superficial layer where chondrocyte viability was maintained (Fig. 5); however, staining was marginally affected by a single 100 nM dose. At 10 μM, a single dose free Dex protected GAG content to a greater extent than at 100 nM. The fast release formulation of 10 μM mAv-Dex did not clearly alter GAG staining over 10 μM free Dex. This may be partly explained by the limited linear range of GAG staining by Safranin-O [37], which can mask more subtle differences in total GAG levels; in this regard, Safranin-O staining can better reflect the localization of GAG loss. In contrast to the fast release formulation, the controlled release mAv-Dex formulation displayed GAG staining that more closely approximated control disks compared to the other IL-1-challenged groups. Furthermore, mAv alone did not have any detrimental effects on tissue Safranin-O staining. Total type II collagen levels were not visibly affected by IL-1α treatment over the 16 days test period (Fig. 7).

Fig. 7.

Histological and immunohistochemical analysis of cartilage. Cartilage explants stained with Safranin-O and Fast Green (for GAG) or immunostained for collagen type II after a 16-day culture period. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4. Discussion

Here we designed a cartilage penetrating and binding mAv nano-construct that can be conjugated with a variety of small molecule drugs or their combinations using hydrolysable ester linkers with tunable drug release rates. The nano-construct incorporates four 8-arm PEGs, providing 28 sites for drug conjugation per mAv compared to only 4 sites in previous designs [28]. As a result, to deliver a one-time dose of 10 μM Dex to cartilage explants, < 1 μM Avidin was needed which is within safe limits. Previous research has shown that an Avidin concentration < 100 μM does not cause any GAG loss or affect chondrocyte viability or biosynthesis rates of proteins and GAGs in bovine cartilage explants [12]. mAv was designed to keep its net size within the previously determined 10 nm size limit for unhindered diffusion into native cartilage and its net charge was shown to be not shielded by addition of multi-arm PEGs (Table 1). As a result, mAv exhibited similar intra-cartilage transport properties as native Avidin; it penetrated through the full thickness of cartilage within 24 h, resulting in a high intra-cartilage uptake (mean Ru of 112 for 1:4 mAv) and long-term retention as the majority of it remained bound within the cartilage over a 7-day desorption period (the duration of experiment conducted) in 1× PBS (Fig. 3). mAv also showed high uptake (mean Ru of 33) in 50% GAG depleted cartilage explants, implying that it can be used for delivering drugs to mid- to late-stage arthritic cartilage. While some binding of mAv with negatively charged groups of synovial fluid can be expected [38,39], our previous work has confirmed that the high negative FCD of cartilage enables rapid diffusion of IA injected native Avidin into rat and rabbit cartilage [12,26,27] resulting in high uptake and long-term retention despite the presence of synovial fluid and dynamic compression-induced convective flow in animal knee joints. Since mAv’s net size and charge properties are similar to that of native Avidin, similar intra-joint kinetics are expected.

First, we conjugated Dex as an example OA drug with 8-arm PEG using hydrolysable ester linkers derived from SA (Fig. 4A). However, the fast-release ester linker in PEG-Dex-SA had a half-life of 6.8 ± 0.2 h, which burst-released a majority of Dex prior to the first media change at 48 h of culture (Fig. 4B). When conjugated with Avidin (fast release mAv-Dex), this resulted in significantly improved therapeutic benefit compared to free Dex throughout 16 days of culture. However, compared to continuous dose of 100 nM Dex, the fast release mAv-Dex resulted in higher GAG loss starting at Day 8 (Fig. 5B). Thus, in order to maintain therapeutic doses of Dex throughout the culture duration, we used GA or PA to synthesize carboxylated derivatives of Dex that increased the half-life of the ester linkers in PEG-Dex-GA and PEG-Dex-PA to 79 ± 1.8 h and 86 ± 2.3 h, respectively. The hydrolysis of an ester bond begins when hydroxide ions of water attack the electrophilic carbon in the ester bond, breaking the p-π conjugation of ester bond creating a tetrahedral intermediate [40]. Since the adjacent carbonyl group from the amide bond tends to compete with and withdraw electrons from the ester bond (inductive effect) resulting in a decrease in the ester bond’s electron density [34], the carbon in ester bond becomes more electrophilic and reactive to nucleophilic attack from hydroxide ion causing a faster release rate (as in the case of compound 5, PEG-DEX-SA in Fig. 4C). Furthermore, hydrophobic Dex conjugated to hydrophilic polymer PEG can produce repulsive forces, thereby accelerating the separation of Dex from PEG. As for compound 6, the GA cross-linker will increase the carbon length between the ester and amide bonds thus weakening the inductive effect of the carbonyl group. The two methyl groups in the GA branched chain can also donate electrons, thereby stabilizing the ester bond. Similarly, the phenyl group in PA can donate numerous electrons to the ester bond (compound 7). Our previous research has shown that a combination of fast and slow releasing linkers can yield a more effective treatment over a long period of time than a complex containing only one of the linkers [28]. Thus, we used ester linkers derived from SA, GA and PA in a molar ratio of 2:1:1 to formulate a controlled release mAv-Dex that enabled 50% release of Dex in 38.5 h followed by a sustained release of the remaining drug over the next two weeks.

As a result, single 10 μM dose of controlled release mAv-Dex (2:1:1) completely suppressed IL-1α induced GAG loss bringing levels down to that of untreated control throughout the 16-day culture period (Figs. 5B and 7), mimicking the continuous 100 nM Dex condition (Figs. 5A and 7). Its therapeutic effect was significantly better than that of the fast release mAv-Dex where ester linker was derived from SA only and had a short half-life of 6.8 h. Similar trends were observed in Day 16 chondrocyte viability images (Fig. 5C) where single dose of both fast and controlled release 10 μM mAv-Dex rescued cell viability similar to continuous 100 nM free Dex while a single dose of 10 μM free Dex was not effective. Treatment with mAv alone did not have any detrimental biological effect even at 10× higher concentrations than used in mAv-Dex (2:1:1).

We chose Dex as an example drug as it is a broad spectrum glucocorticoid (GC) with anti-catabolic properties and has intra-cellular receptors inside cartilage and in synovium and can elicit disease modifying effects [5,6,41] as well as suppress OA induced pain and inflammation [42,43]. Effectiveness of GCs, however, has been a subject of controversy showing both chondroprotective and deleterious effects following IA injection [5,28,44,45]. Their effectiveness depends largely on frequency, dosage and the duration of treatment [41,46–48]. GCs like Dex, triamcinolone acetonide (TCA) and prednisone are used in high doses and multiple times (up to 100 mM [9] compared to one time 10 μM dose used in the present study) for OA pain relief due to their short intra-joint residence time causing bone resorption, cell apoptosis and systemic toxicity [49]. For example, a recent human clinical trial (NCT01230424) concluded that IA injections of 40 mg TCA every 3 months for 2 years in patients with symptomatic knee OA resulted in greater cartilage loss compared to saline [50], emphasizing the critical need for targeted, low dose sustained therapy. A recent study also showed that multiple intra-articular injections of Dex (2.5 mg every 3 days or 0.5 mg/kg) were needed to significantly reduce inflammation and protect cartilage in a bone drill rabbit model of PTOA; this high multi-dose regimen, however, resulted in systemic toxicity in vital organs [45]. Nevertheless, a recent clinical trial (NCT01692756) concluded that early intervention using two doses of 40 mg TCA within 10 days of ACL rupture was able to prevent injury induced early chondral changes [51]. While more follow up studies are needed to conclude meaningful clinical difference in overall outcome, there remains continued interest in using GCs as disease modifying agents for OA/PTOA treatment [52] and not just for symptomatic relief. It is, however, imperative to develop strategies in parallel that can target cartilage and deliver therapeutic low drug doses over several days to weeks to avoid toxicity associated with multiple injections of high drug doses.

Recent research has focused on developing IA drug delivery systems, including drug-encapsulating microparticles [17,53,54], polymeric micelles [14,55], liposomes [15], aggregating hydrogels [16,18,56,57] and peptides [25]. The majority of carrier systems cannot penetrate the dense, negatively charged cartilage; consequently, while some have shown promise for suppressing pain and inflammation originating from the synovium and surrounding joint capsule, they are not effective at stimulating disease-modifying responses in chondrocytes. For example, recently intra-joint sustained release formulations of TCA encapsulated within micron sized PLGA particles (Flexion Therapeutics, Burlington, VT, USA) were approved by the FDA for OA pain and inflammation, but such systems naturally use high drug doses to induce biological response (40–60 mg of drug). Flexion’s microsphere based TCA delivery (FX006, 32 mg drug dose administered) showed prolonged synovial fluid joint residency (until Week 12), diminished peak plasma levels and thus reduced systemic exposure compared to free TCA following a single IA injection in patients with knee OA in a Phase II open label study [17]. Their Phase III, multicenter, double-blinded, 24-week study concluded that a single IA injection of FX006 provided significant improvement in average-daily-pain (ADP)-intensity scores compared to saline (placebo) but no significant improvements in OA pain were observed when compared to the free drug [58]. In a Phase III post-hoc study, where efficacy of FX006 was evaluated in a subgroup of participants with unilateral knee OA only (as bilateral knee pain has emerged as a cofounding factor in clinical trials evaluating the effect of single IA injection), significant improvement in WOMAC or ADP scores were reported for FX006 compared with both saline and free drug over a period of 5–6 months [59]. While these are promising data for longer lasting pain relief with a single IA injection, delivery systems like mAv can target cells inside cartilage to elicit long-term disease modifying effect to restore joint function. Joint inflammation has a complex etiology that involves not only the synovium but also subchondral bone and articular cartilage [38,60]. Therefore, effective treatments will likely need to deliver OA drugs into cartilage as well as to surrounding tissues. We predict that gradual release of drug from intra-cartilage drug depot enabled by mAv delivery can significantly inhibit IL-1α signaling, for example, in both cartilage and nearby synovium and other tissues inside the joint, and thus can provide long-term pain and inflammation relief along with restoring joint function. Future studies will investigate this.

While a single dose of mAv-Dex effectively suppressed cytokine-induced catabolic activity, it was unable to rescue anabolic activity within cartilage explants (Fig. 6B). Dex is known as a potential OA prophylactic agent that inhibits production of matrix metalloproteinases (MMPs), NO and inflammatory cytokines in OA joints [61,62]. When combined with a pro-anabolic factor like insulin-like growth factor 1 (IGF-1), it can effectively also reverse cytokine-induced inhibition of GAG synthesis [5]. Thus, a combination drug therapy comprising of an anti-catabolic agent like Dex that can induce broad spectrum inhibition of cytokine-related degradation, maintain cell viability along with a pro-anabolic factor like IGF-1 [63,64] or Kartogenin [65,66] that can stimulate biosynthesis of viable chondrocytes and replenish cartilage matrix components like aggrecans and collagens is a better approach than monotherapy with one drug. Multiple drugs can be conjugated using similar chemistry to mAv and their release rates can be modulated by varying the ratio of ester linkers derived from SA, GA and PA or designing new types of linkers based on the mechanism explained earlier.

Thus, by using charge interactions, our mAv nano-construct offers multiple advantages of (i) converting cartilage from a barrier to drug entry into a reservoir of drugs that can prevent rapid exit from synovial joint, (ii) accumulating higher drug dose at the cell and matrix target sites and (iii) presenting multiple sites for covalent conjugation of more than one drug for combination therapy. In summary, a charge-based cartilage homing drug delivery platform like this can, potentially, elicit disease modifying effects as well as facilitate long-term symptomatic pain and inflammation relief by enhancing tissue specificity and prolonging intra-cartilage residence time of OA drugs.

The mechanism of mAv targeting of cartilage and drug delivery is summarized as follows (Fig. 1): Following its intra-articular administration, mAv-Dex can rapidly penetrate through the full thickness of negatively charged cartilage in high concentrations due to electrostatic interactions thereby creating an intra-cartilage drug depot. The optimal net positive charge of mAv enables its rapid and high intra-cartilage uptake (112×) and long-term retention via weak-reversible binding with negatively charged aggrecans. Therapeutic doses of drug (Dex) is then released via hydrolysis from controlled release mAv-Dex (2:1:1) over 2 weeks, which can diffuse into chondrocytes to bind with its glucocorticoid receptors triggering downstream signaling pathways and suppressing OA associated catabolic activity. This intra-cartilage depot drug delivery platform can be used to deliver a variety of drugs or combination of drugs and enable OA treatment with only a single injection of low drug doses thereby eliminating toxicity issues associated with multiple injections of high drug doses that are currently needed to maintain sustained drug doses within the joint.

5. Conclusion

Our cationic multi-arm Avidin (mAv) nano-construct can enable intra-cartilage delivery of a broad array of small molecule OA drugs and their combinations to chondrocytes. Broad applicability is critical because OA has a range of causal factors and one class of OA drugs may not be effective for all clinical cases. Drug release rates can be modulated by using a combination of ester linkers with different rates of hydrolysis based on type of drug, its target sites and the state of the disease. This charge-based intra-cartilage drug delivery platform also offers a unique opportunity to re-examine drugs that failed OA clinical trials due to lack of tissue targeting and the resulting systemic side effects as well as repurpose FDA approved anti-rheumatic drugs for OA therapy.