Chronic inflammatory diseases (CIDs), such as rheumatoid arthritis (RA), osteoarthritis (OA) and asthma, have a substantial socio-economic impact, high prevalence and severely limit patients' quality of life. [1] Although the clinical symptoms of the diverse CIDs vary, the cellular processes in chronically inflamed tissues are similar. [2] In particular, RA can lead to severe disability and premature mortality. [3,4] In RA, the inflammation is polyarticular, primarily affecting the small diarthrodial joints of the hands and feet. However, any joint can be affected and severe non-joint systemic symptoms may occur. [5] Although non-steroidal anti-inflammatory drugs (NSAIDs), corticosteroids, disease-modifying anti-rheumatic drugs, and biologicals are used for the long-term treatment of RA [6], their application does not usually lead to complete remission.

Glucocorticoids (GCs) serve a central role in the therapeutic management of CIDs, especially in RA. [7,8] Nevertheless, the severe side effects induced by their long-term administration (e.g., osteoporosis) limit their clinical application and require careful dosing and often additional supportive medication. [9,10] The current recommendations for GC therapy generally favour their low-dose and short-term application. Since the ubiquitous GC receptors are present in nearly every cell in the body, site-specific GC delivery should present therapeutic benefits.

To increase bioavailability and reduce adverse effects, several targeted nanomedicines enhancing GC accumulation at the site of inflammation have recently been proposed for RA since macromolecules exhibit increased permeability of the inflamed synovium blood capillaries. [[11], [12], [13]] Among the available nanomedicine approaches, water-soluble polymer-drug conjugates offer various advantages, such as improved pharmacokinetics of the carried drug as well as easy handling and storage since the final products can be stored in solid form and prepared for parenteral administration by simple dissolution in a physiological solution. Additionally, the use of finely tailored pH-sensitive, redox-sensitive or enzyme-biodegradable linkers between the polymeric carrier and the active compound protects against premature drug escape from the polymer during blood circulation and enables drug release and activation upon exposure to acidified pH, a reductive environment or enzymes in target tissues and cells (e.g., inflammatory joints and cancer). [14,15]

Copolymer drug conjugates based on the biocompatible N-2-(hydroxypropyl)methacrylamide (HPMA) copolymers have repeatedly been proven to serve as efficient and safe drug carriers capable of the passive targeting to inflamed joints. [11,[16], [17], [18], [19]] Not only is the drug protected during blood circulation, but the HPMA copolymer nanomedicines are unable to penetrate certain tissues where exposure to GC causes severe side effects, i.e. brain (impaired hypothalamus and neural activity) [20], eyes (glaucoma) [21] and bone tissue (osteoblast and osteoclast dysfunction) [22], thereby preventing the negative effects of GCs.

We have previously optimised the synthesis of HPMA-based copolymer conjugates with the anti-inflammatory drug dexamethasone (DEX), which is attached via a biodegradable pH-sensitive spacer to obtain well-defined copolymer systems with enhanced biodistribution, increased joint accumulation and rapid pH-sensitive drug release in inflamed joints. The conjugates have proven to be significantly more efficient than free DEX in a model of single-joint acute arthritis (antigen-induced arthritis, AIA) in mice. [19]

Here, we have designed and synthesised properly defined tailored HPMA copolymer-DEX conjugates and studied their biological efficacy using collagen II-induced arthritis (CIA) in mice - a model of dissipated chronic arthritis that is more clinically relevant to RA than AIA. To minimise the exposure to DEX, the therapeutic activity of DEX-bearing conjugates was determined using various dosage schemes in CIA, where the drug was applied daily or every second or third day. Moreover, the impact of the nanomedicine structure, either size-switchable between glomerularly non-excretable/excretable, or permanently glomerularly non-excretable long-circulating systems, on therapy efficacy was studied in detail. Additionally, the influence of the administration of various DEX nanoformulations and dosing schemes on bone structure and the density of the inflamed joints was compared to that of free DEX.

Dichloromethane (DCM), tert-butanol (t-BuOH), dimethylformamide (DMF), dimethyl sulfoxide (DMSO), diethyl ether, dimethylacetamide (DMA), ethyl acetate (EtAc), methanol (MeOH), acetone, 2,2′-azobis(isobutyronitrile) (AIBN), 2,2′-azobis(4-methoxy-2,4-dimethylvaleronitrile) (V70), acetic acid (CH3COOH), dexamethasone (DEX), trifluoroacetic acid (TFA), N-(3 dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), 2,4,6-trinitrobenzene-1-sulphonic acid (TNBSA), 4-(dimethylamino)pyridine (DMAP), sodium borohydride (NaBH4), N-ethylmaleimide (NEMI), sodium cyanoborohydride (NaBH3CN) and porcine liver esterase were obtained from Merck KGaA (Darmstadt, Germany). Milli-Q water (H2O) was used for all experiments and obtained from the appliance Millipore Merck (Darmstadt, Germany) (resistivity 18.2 MΩ·cm, 25 °C, organic carbon ≤5 ppb). The 4-(2-oxopropyl) benzoic acid (OPB) was purchased from Rieke Metals LLC (Lincoln, Nebraska). Bovine native collagen II (CII), complete Freund's adjuvant (CFA) and incomplete Freund's adjuvant (IFA) were purchased from MD bioSciences (Zurich, Switzerland). The dexamethasone (2 mg/mL of 80 excipient QSP) used in in vivo testing was obtained from Virbac S.A. (Carros, France).

The two monomers N-(terc-butoxycarbonyl)-N´-(6-methacrylamidohexanoyl)hydrazine (Ma-Ah-NHNH-Boc)) and N-(2-hydroxypropyl) methacrylamide (HPMA) and the chain transfer agent (CTA)-S-2-cyano-2-propyl-S´-ethyl trithiocarbonate (sCPsE-TTc) were synthesised according to the literature [23,24]. The linear copolymer precursors, P1 and P2, were prepared by the controlled radical reversible addition-fragmentation chain transfer (RAFT) copolymerisation of HPMA and Ma-AH-NHNH-Boc at a molar ratio of 92:8. The sCPsE-TTc was used as a CTA and V70 was used as an initiator. A mixture comprised of 85% t-BuOH and 15% DMA was used as a solvent for the copolymerisation. The molar ratio of monomers, CTA and V70 was 250:1:0.5 for P1 and 580:1:0.5 for P2. The copolymerisation conditions were as previously reported. [19,25]

The trithiocarbonate (TTc) ω-end groups were removed using 2,2`-azobisisobutyronitrile (AIBN), as follows: [26] P1 (600 mg, 2.5 mmol of TTc) and AIBN (120 mg, 0.79 mmol) were dissolved in 7.2 mL of DMA (10 vol% solution). The reaction mixture was inserted into an ampule, bubbled with argon for 10 min and sealed. The reaction was carried out for 4 h at 80 °C and the copolymer was then isolated by precipitation into a 2:1 acetone-diethyl ether mixture, filtered and dried in a vacuum. To complete the removal of the chain end functional groups originating from CTA, their reduction using NaBH4 and subsequent modification by NEMI was performed. Copolymer precursor P1 (500 mg) was dissolved in dried methanol (10 vol% solution) and NaBH4 (5.5 mg, 0.15 mmol) was added in parts within 10 min. The mixture was stirred for another 10 min. As a next step, NEMI (20 mg, 0.16 mmol) was added to the solution in two parts within 15 min. The excess of reduction agent was then quenched via the addition of acetic acid (15 μL). Then, the solution was diluted to a 3 vol% solution with MeOH and the low molecular weight impurities were removed on an LH-20 filled column in MeOH. [27] The copolymer fraction was collected, concentrated under reduced pressure and the copolymer was then separated by precipitation into a 2:1 acetone-diethyl ether mixture, filtered and dried in a vacuum. The deprotection of the hydrazide groups was performed in distilled water at 100 °C after the removal of the TTc end groups, as recently described. [25] The P1 or P2 (450 mg) were dissolved in 4.5 mL of distilled water, placed in an ampule, bubbled with argon for 10 min, sealed, and heated for 1 h to 100 °C. The final copolymer (P1 or P2) was then separated by freeze-drying.

The dexamethasone 4-(2-oxopropyl) benzoate (DEX-d) (Fig. 1A) was synthesised as previously reported. [19] Briefly, DEX-d was synthesised by an esterification reaction of the hydroxyl functional group of the DEX (C21-OH) and the carboxyl functional group of the 4-(2-oxopropyl) benzoic acid (OPB). OPB (45.4 mg, 0.25 mmol) together with EDC (73.3 mg, 0.38 mmol) were dissolved in a mixture of 3.0 mL DMF and 1.0 mL DCM and maintained at −18 °C for 20 min. Then, a solution of DEX (100 mg, 0.25 mmol) and DMAP (31.1 mg, 0.25 mmol) in 3.0 mL DMF was added, and the reaction mixture was maintained at 4 °C for 24 h. The reaction course was monitored by TLC (EtAc:DCM, 2:1). The reaction mixture was then purified on a column filled with silica gel 60 (EtAc:DCM, 8:1 with a gradient lowering the ratio of the solvents, monitored by TLC). DEX-d was obtained after the evaporation of the solvent. The yield was 89 mg (89%). HPLC showed 99% purity (peak maximum at 3.7 min). The structure was confirmed via 1H NMR spectroscopy (Fig. 3). 1H NMR (600 MHz, DMSO, δ, ppm): 7.94 (d, 2H, Ar), 7.36 (d, 2H, Ar), 7.31 (d, 1H, Ar), 6.24 (d, 1H, Ar), 6.02 (s, 1H, Ar), 5.45 (d, 1H, OH), 5.28 (d, 1H, CH2), 5.23 (s, 1H, OH), 5.06 (d, 1H, CH2), 4.18 (m, 1H, CH), 3.91 (s, 2H, CH2), 2.90 (m, 1H, CH), 2.64 (m, 1H, CH2), 2.41 (m, 1H, CH2), 2.31 (m, 1H, CH2), 2.23 (m, 1H, CH), 2.17 (s, 3H, CH3), 2.14 (m, 1H, CH), 1.77 (m, 1H, CH2), 1.66 (m, 2H, CH2), 1.50 (s, 3H, CH3), 1.36 (m, 1H, CH2), 1.09 (m, 1H, CH2), 0.93 (s, 3H, CH3), and 0.82 (d, 3H, CH3).

DEX-d was bound to the copolymer precursor P1 or P2 via a pH-sensitive hydrazone bond. Briefly, copolymer precursor P1 (50 mg) and DEX-d (3.5 mg) were dissolved in the MeOH (10 vol% solution, 535 uL). The solution was stirred for 24 h at 10 °C. The unbound DEX-d was removed on the LH-20 column in MeOH. The copolymer fraction was collected, concentrated by solvent evaporation under a vacuum and P1-DEX-d was precipitated to EtAc and dried under a vacuum. The amount of DEX-d attached to the copolymer was determined by HPLC analysis after total hydrolysis of hydrazone bonds at pH 2 and confirmed by 1H NMR analysis.

The release rate of free DEX, as well as the free and/or ester derivative, from the polymer conjugate P1 was investigated via the incubation of the conjugate in model buffer environments with pH 5.0, 6.5 and 7.4 (0.1 M phosphate buffer) or plasma at 37 °C. The amount of free drug and ester derivative was determined via HPLC analysis after their extraction into DCM. Analysis was performed on an HPLC instrument equipped with a reverse phase column. All drug release data are expressed as the amounts of the free drug or its derivative relative to the total drug content in the conjugates. All experiments were performed in duplicate. The in vitro drug release kinetics of DEX-d and DEX were also evaluated in the presence of porcine liver esterase. First, the activity of the enzyme for a model reaction with a 4-methylumbelliferyl acetate substrate was explored using the spectroscopic determination of the 4-methylumbelliferon product of the enzymatic reaction (λ = 350 nm, ε350 = 12,200 L·mol−1·cm−1) as follows: the two solutions of the substrate (0.025 M solution in DMSO) and esterase (0.3 mg/mL in 0.05 PBS buffer of pH 5.0) were prepared and maintained at 37 °C. The solutions—970 μL of 0.05 M phosphate buffer of pH 5.0, 20 μL of the substrate and 10 μL of the solution of the enzyme—were inserted into a 1-cm cuvette. The reference value of the absorbance was set by adding 980 μL of 0.05 M phosphate buffer of pH 5.0 and 20 μL of the substrate. Cuvettes were thermostated and the measurement was performed at 37 °C. The specific activity of the enzyme was 6.8 mmol.min−1 mg−1.

The number-average molecular weight (Mn), weight-average molecular weight (Mw) and dispersity (Đ) of the copolymer precursors and conjugates were measured using size-exclusion chromatography (SEC) on an HPLC Shimadzu system equipped with an SPD-M20A photodiode array detector (Shimadzu, Japan), an OptilabrEX differential refractometer and a multi-angle light scattering DAWN HELEOS II (Wyatt Technology, USA) detector using a mixture of 0.15 M sodium acetate buffer at pH 6.5 (20 vol%) and methanol (80 vol%) as the mobile phase. A TSKgel SuperSW3000 column was used for P1 and a TSKgel 4000SWxl was used for P2. ASTRA software and a refractive index increment dn/dc of 0.167 mL/g were used for calculations.

The content of hydrazide groups was determined by UV/VIS spectrophotometry after derivatisation using TNBSA as previously described [25].

The DEX-d structure was determined via 1H NMR spectroscopy. 1H NMR spectra were acquired using a Bruker Avance III 600 spectrometer operating at 600.2 MHz. The width of the 1H NMR 90° pulse was 18 μs, with a relaxation delay of 10 s and an acquisition time of 2.73 s. Samples were dissolved in deuterated DMSOd6. The chemical shifts are relative to tetramethylsilane using a solvent signal (DMSO, δ = 2.50 ppm from TMS in 1H NMR spectra). All samples were measured at 295 K.

The total content of DEX-d conjugated to the polymer was determined by HPLC and 1H NMR spectroscopy. Concerning the HPLC, a sample of the conjugate was dissolved in the mobile phase of water:acetonitrile 95:5 with 0.1% of TFA (pH 2) and incubated for 1 h at 37 °C. The amount of DEX-d was determined using an HPLC device equipped with a Chromolith reverse phase column. To determine the molar content of DEX-d on P1 or P2 via NMR, a signal at 6.02 (s, 1H, Ar) was used.

The hydrodynamic diameter (Dh) of the copolymer precursors and conjugates was measured by dynamic light scattering (DLS) using a Nano-ZS instrument (ZEN3600, Malvern) in a phosphate buffer (pH 7.4, 0.1 M with 0.05 M NaCl). The concentration of the copolymer was 1.5 mg/mL. The intensity of the scattered light was detected at θ = 173° using a laser with a wavelength λ of 632.8 nm. The DTS (Nano) program was used for the dynamic light scattering data evaluation. The values are equivalent to the mean of at least five independent measurements. The values were not extrapolated to the zero concentration.

DBA/1 (n = 80) mice were purchased from Janvier (Le Genest-St-Isle, France) and allowed to acclimatise for at least 1 week. All mice were 8 weeks of age at the beginning of the experiment. They were provided with standard rodent chow and water ad libitum. Mice with different treatments were randomly assigned in each cage to prevent a cage effect. The animal experiment followed European and French animal experimentation regulations. The project was approved by the local Ethics Committee (CEEA) 34, Université Paris Descartes, on 5 September 2017 and registered by the French Ministry of Research (number 20123–2019021516441070).

An emulsion of the 2 mg/mL bovine native collagen II (CII) was formed by dissolving it overnight at 4 °C in 10 mM acetic acid and combining it with an equal volume of CFA. DBA/1 mice were injected at the base of the tail intradermally with a total of 100 μL of emulsion containing 200 μg CII emulsified in CFA. On day 21, an injection with CII in IFA was repeated as a booster. [28] The CII solution and the emulsion with CFA or IFA were always freshly prepared.

Mice were monitored for indication of polyarticular arthritis in their paws using a blind procedure by two examiners (S.S. and F.L.) and a clinical score based on disease severity was given for each mouse. The clinical assessment was performed three times per week in the case of mice without treatment and every day after the beginning of the therapeutic experiments. The date of disease onset was recorded and the clinical severity of each joint or group of joints (fingers, wrists, toes, ankles and tarsi) was graded as follows: 0 (normal, no swelling), 1 (minute swelling of the joint or the first phalange of the fingers or toes), 2 (mild swelling), 3 (severe swelling) or 4 (necrosis). The inflammation scores (IS) of each joint (graded 0–4) were summed to give a total arthritic score theoretically ranging from 0 to 40. The severity of CIA was expressed as the mean score per group observed on a given day, the mean of the score reached by each mouse during the experiment and the mean of the maximal arthritic score reached by each mouse.

Mice with a total severity score higher than 3 on the main joints (ankle, tarsus or wrist) were divided into ten groups (n = 8, Table 1). Mice were injected intraperitoneally with 250 μL of 0.5 mg/kg DEX as a control or with DEX eq. for the DEX-d copolymer conjugates either daily (6 × 0.5 mg/kg), every second day (3 × 0.5 mg/kg) or every third day (2 × 0.5 mg/kg). The free DEX and tested polymer compounds were dissolved in phosphate buffer. Each day, untreated mice were given phosphate buffer to account for the stress induced by repeated injections. At the end of the experiment, the mice of each group were sacrificed and the hind legs were withdrawn for further analysis.

To analyse the effect of the different treatments on bone quality, we performed micro-CT scans of hind legs, which were measured using a SkyScan 1272 (Bruker, micro-CT) ex vivo scanner with the following scanning parameters: pixel size 6 μm, source voltage 70 kV, source current 142 μA, Al 0.25 mm filter, rotation step 0.2°, frame averaging (2), rotation 180° and scanning time 2 h. Flatfield correction was updated prior to each image acquisition. Projection images were reconstructed using NRecon software (Bruker) and visualisations were acquired using a DataViewer and CTVox (Bruker). The image data of the specimens were semi-quantitatively evaluated according to the three parameters considered to be signs of inflammation: porosity of cortical bone, reactive new bone formation and osteoporosis. Each parameter was graded as follows: 0 (without any pathological finding), 1 (slight changes) and 2 (significant changes). The maximum total score was set as 6, which indicates significant bone damage (as seen in Fig. 2).

Animal experiment values are presented as the average of eight measurements. The standard deviation of two-way analysis of variance (ANOVA) was determined, followed by a post hoc test (Bonferroni's test) for multiple comparisons using Prism software (GraphPad, USA). A value of p ≤ 0.0001 was considered statistically significant (****) and a value of p > 0.05 was non-significant. The values between these two extreme values were labelled as *** for p ≤ 0.001, ** for p ≤ 0.01 and * for p ≤ 0.05.