Injectable acylhydrazone-linked RAFT polymer hydrogels for sustained protein release and cell encapsulation

Abstract

A new class of temperature responsive polymer, termed PADO, is synthesized by reversible addition-fragmentation chain-transfer (RAFT) polymerization. Synthesized from copolymerization of diacetone acrylamide (DAAM), di(ethylene glycol) ethyl ether acrylate (DEGA), and oligo(ethylene glycol) methyl ether acrylate (OEGA), PADO polymer phase separates at temperature above its LCST (36°C to 42°C) due to enhanced hydrophobic interactions between the short ethylene glycol side chains. Solution of PADO polymers exhibited injectable shear-thinning properties and reached sol-gel transition rapidly (< 5 min) at 37°C. When the ketone moieties on DAAM were linked by adipic acid dihydrazdie (ADH), PADO polymers formed crosslinked and injectable acylhydrazone hydrogels, which were hydrolytically degradable at a mild acidic environment owing to the pH sensitive acylhydrazone bonds. The pH-responsive degradation kinetics could be controlled by tuning polymer contents and ketone/hydrazide ratio. Importantly, the injectable PADO hydrogels were highly cytocompatible and could be easily formulated for pH-responsive sustained protein delivery.

Graphical Abstract

A novel class of PADO polymers containing diacetone acrylamide (DAAM), di(ethylene glycol) ethyl ether acrylate (DEGA), and oligo(ethylene glycol) methyl ether acrylate (OEGA) are polymerized via RAFT polymerization. Via acylhydrazone chemistry, the PADO polymers are crosslinked into pH/thermo-sensitive injectable hydrogel for tunable and sustained protein delivery.

1. Introduction

Injectable protein delivery formulations are highly sought-after in many therapeutic applications. For treating cronic diseases such as osteoarthiritis (OA), the concentration of the delivered therapeutics should be maintained within a therapeutic window to reduce undesired side effects and to prevent the needs for repeated dosing.^[1] The ideal delivery carrier should shield the protein therapeutics from unwanted degradation and/or denaturation.^[2,3] To this end, numerous synthetic hydrogels have been designed for sustained delivery of environmentally sensitive macromolecules (e.g., proteins, cytokines, growth factros) and even cells.^[4,5] Synthetic materials offers many benefits, including well-deinfed molecualr structures and properties, high purity and batch-to-batch consistency, as well as high adaptability for modification. Furthermore, the degradation kinetics of synthetic carriers can be readily tuned to maximize therapeutic performance without eliciting cytotoxicity. For example, hydrolytically degradable poly(lactic-co-glycolic acid) (PLGA) has been widely used for sustained delivery of proteins and small molecular weight therapeutics.^[6,7]

When stimuli-responsive functional groups are incorporated into injectable carriers,^[8] the release kinectics of the therapeutic payloads can be further regulated by physical and chemical environmental cues such as temperature, pH, redox potential, magnetic field, etc.^[9] In particular, thermoresponsive polymers and hydrogels exhibiting a lower critical solution temperature (LCST) slightly below the body temperature (37°C) are among the most widely studied biomaterials as they exhibit rapid physical gelation upon injection into the body. The polymers collapsed at temperature above their LCST due to increased inter/intra-molecular hydrophobic interactions. Poly(N-isopropylacrylamide) (PNIPAAM) is considered the gold standard of thermo-responsive polymers in controlled release applications due to its relatively stable LCST over fluctuations of pH and concentration.^[10,11] However, PNIPAAM-based hydrogels often exhibit slow temperature response and low mechanical strength,^[12,13] necessitating the copolymerization of other monomers. Another viable option for fabricating thermosensitive polymer hydrogels include the family of poly(oligo(ethylene glycol) (meth)acrylates) (POEGA).^[14] The LCST of POEGA polymers can be tailored by different poly(ethylene glycol) (PEG) side chain lengths and the hydrophobicity/hydrophilicity of the end groups. With a narrower hysteresis profile compared to PNIPAAM and excellent biocompatibility conferred by PEG side chains, ^[11] POEGA is a suitable class of polymers for formulating injectable thermosensitive hydrogels. For example, Xu et al. recently synthesized a series of POEGMA-based polymers with a range of LSCT and charge to screen suitable injectable formulations for delivering proteins.^[15]

In addition to thermosensitivity, it is highly desirable that the injectable hydrogels are also sensitive to other stimuli to better facilitate therapeutic targeting and delivery. Schiff base dynamic covalent chemistry is being adapted for biomaterials fabrication due to its physiological reaction conditions along with self-healing and pH responsive properties.^[16] Schiff base reactions produce imines (aldehyde/ketone-amine reaction), hydrazones (aldehyde-hydrazide reaction), acylhydrazones (ketone-hydrazide reaction), or oximes (aldehyde/ketone-oxyamine reaction).^[17] The bonding stability of acylhydrazone falls between imines and oximes, making it an ideal chemistry for degradation-mediated controlled release. Furthermore, the dissociation of acylhydrazone bonding is accelerated in acidic conditions but substaintially slower in alkalic environments. Hydrogels containing acylhydrazone bonds have been utilized for controlled release of small molecule drugs,^[18,19] as well as for promoting wound dressing^[20] and bone repair.^[21] For example, Chen and coworkers designed gelatin/dextran-based acylhydrazone hydrogels for sequential delivery of chlorhexidine acetate (CHA) and basic fibroblast growth factor (bFGF) to promote wound healing.^[22] In this example, bFGF was encapsulated in PLGA microspheres that were embedded in the CHA-loaded acylhydrazone hydrogels. While the microspheres-loaded hydrogels provided a mechanism for sustained protein delivery, no pH responsiveness was reported. Using acylhydrazone and disulfide chemistries, Xu et al. reported a redox and pH dual-responsive hyaluronic acid-based system for protein and cell delivery.^[23] The release of bovine serum albumin (BSA) was accelerated at acidic condition and in the presence of dithiothreitol. However, the protein release rate was relatively fast, with about 80% and 50% BSA released at pH 6.0, and pH7.4 in 5 days.

An injectable hydrogel formulation that provides slow and extended protein release is highly desirable for treating chronic diseases, such as OA. Hydrogels exhibiting LCST and pH responsiveness are particularly useful in this endeavor. To this end, we utilized reversible addition-fragmentation chain transfer (RAFT) polymerization to synthesize a series of stimuli-responsive polymers, named Poly(diacetone acrylamide-s-di(ethylene glycol) ethyl ether acrylate-s-oligo(ethylene glycol) methyl ether acrylate) or PADO. By adjusting the hydrophobic and hydrophilic monomers, PADO polymers exhibit tunable LCST below/around body temperature. Through forming acylhydrazone dynamic covalent bonds with adipic acid dihydrazide (ADH), the PADO polymers were further formulated into injectable and thermo/pH dual-responsive hydrogels. The acylhydrazone bonds formed in the PADO hydrogels were highly sensitive to mild pH changes, leading to stable hydrogels at neutral pH but acid-labile gels at a mild acidic environment. In this work, we thoroughly characterized crosslinking kinetics, degradation, in vitro cytotoxicity, and in vivo compatibility of the injectable acylhydrazone PADO hydrogels. We further used model proteins to demonstrate the potential of PADO hydrogels for tunable, stimuli-responsive, and extended release of macromolecular proteins.

2. Result and Discussion

2.1. Polymer Synthesis and injectable hydrogel formulation

Using RAFT polymerization, we designed PADO polymer containing ketone moiety (from DAAM) for acylhydrazone crosslinking, hence exhibiting pH-responsiveness (Figure 1A). Furthermore, poly(ethylene glycol) acrylates (PEGAs) with different ethylene glycol chain lengths were copolymerized to afford temperature responsiveness of the resulting PADO polymers and hydrogels. Since the polyacrylic backbone of PEGA was hydrophobic, the LCST of PEGAs increased with longer hydrophilic ethylene glycol side chains. The LCSTs of DEGA and OEGA were reported as 13°C and 92°C, respectively.^[14] Balancing the ratio of DEGA and OEGA would give rise to PADO polymers with tunable LCST. To this end, we synthesized 3 PADO polymers with low (L), medium (M), or high (H) ketone content (Figure 1B). Using GPC, we characterized the number-average molecular weight (Mn) of PADO polymers, ranging from was controlled at around 20 kDa and measured via GPC to with dispersity (D) of 1.14–1.28 (Figure 1C). The number-average molecular weight (Mn) of PADO polymer was controlled at around 20 kDa (Figure 1C), while the dispersity (D) was between 1.14–1.28 (Figure 1C). We estimated the compositions of the polymerized units (Figure 1C) based on ¹H NMR spectra (Figure 1B) according to the characteristic peaks of dimethyl group (peak a) from DAAM, methyl group (peak e) from OEGA, and ethyl group (peak f) from DEGA. The content of DEGA decreased with increasing hydrophobic DAAM, affording tunable LCST as characterized by a temperature-dependent turbidity assay (Figure 1D). Of note, the temperature at which solution turbidity started to increase rapidly (i.e., decreased light transmission) scaled inversely with the ketone content on PADO (i.e., 36, 40, and 42 °C for PADO_H, PADO_M, and PADO_L, respectively). These results indicated that the LCST of PADO polymers were successfully controlled at around physiological temperature.

Figure 1. Poly(diacetone acrylamidei-s-di(ethylene glycol) ethyl ether acrylate-s-oligo(ethylene glycol) methyl ether acrylate) (PADO) synthesis and characterization.

(A) Reaction scheme. (B) ¹H NMR spectra. (C) Polymer compositions, number-average molecular weights (Mn), and dispersities (D). (D) Polymer turbidity as a funciton of temperature and keton content on PADO.

As shown in Figure 1D, PADO polymers precipitated from solution when temperature was above the polymer LCST. In the presence of crosslinker ADH, however, PADO polymers were crosslinked into hydrogels via the formation of acylhydrazone linkages (Figure 2A). Depending on the polymer weight content and ketone concentration, the precursor mixture was either a viscous solution or a soft hydrogel with self-healing and injectable property. Upon heating, the polymers containing ADH condensed into hydrogels, likely due to the aggregation of DEGA units. In the absence of ADH, the copolymers alone precipitated from the solution due to phase separation (Figure 1D, Figure S1). We reasoned that the dynamic covalent acylhydrazone bonds forming between the PADO polymers prevented their thermally induced phase separation and gave rise to reversible gelation above polymer LCST (Figure 2B). Therefore, PADO precursor mixture exhibited a reversible sol-gel transition between room temperature and 37°C. The PADO polymer hydrogels were injectable and showed rapid temperature induced gelation, as demonstrated by injection through a 23G needle. While gelation in the presence of ADH at room temperature took hours to form gels, the injected polymer rapidly crosslinked into standalone hydrogel within 5 minutes of incubation at 37°C (Figure 2C and Supporting movie).

Figure 2. PADO hydrogel formulation via acylhydrazone chemistry and DEGA thermo responsive units.

(A) Reaction scheme. (B) Temperature-dependent reversible gelation for 10% PADO_M. (C) Injectability of 15% PADO_M. (D-G) SEM micrographs of PADO_M at 10% (D), 15% (E), 20% (F), and PADO_L at 20% (G). Scale bar = 10 μm.

To evaluate the effect of polymer compositions (e.g., acylhydrazone concentration and polymer weight percent) on PADO hydrogel properties, the hydrogels were snap-frozen in liquid nitrogen and lyophilized for scanning electron microscopy (SEM) imaging (Figure 2D-G, Figure S2). It is worth noting that the porous structures revealed in the SEM micrographs were not a direct indication of hydrogel mesh size as they were created during sample processing and/or altered by the vacuum environment during imaging. Nonetheless, SEM micrographs could still provide valuable morphological information by comparing the porous structures and pore sizes from different hydrogel formulations. With increasing polymer weight percent, the dried hydrogel polymer pore size reduced from ca. 5 μm in PADO_M-10% to ca. 2 μm in PADO_M-15% and PADO_M-20%. Moreover, dried polymer pore size decreased with increasing ketone content, exemplified by PADO_L (7 μm) and PADO_M (2 μm) at 20%, and PADO_M (5 μm) and PADO_H (3 μm, Figure S2) at 10%. While the pore sizes of these dried polymers cannot be used to predict protein release kinetics, they may be used to qualitatively correlate protein released from different formulations.^[24]

2.2. Crosslinking, shear-thinning, and temperature responsiveness of PADO hydrogels

Adjusting PADO polymer weight percent and ketone content affected the crosslinking density of the resulting hydrogels. Using ADH as the crosslinker, we first evaluated PADO hydrogel crosslinking at different PADO polymer contents using an oscillatory rheometer. As shown in Figure 3A, shear moduli (G’) of PADO hydrogels scaled with both polymer weight content (wt%) and ketone concentration (L, M, or H). For example, at the same polymer content (e.g., 20 wt%), increasing ketone content on PADO resulted in a wide range of hydrogel crosslinking density, as demonstrated by increasing G’ from 1 kPa to 85 kPa for PADO_L and PADO_H, respectively. On the other hand, hydrogels crosslinked by 8 wt% to 20 wt% of PADO_H exhibited G’ of 1.7 kPa to 85 kPa. These results were as expected as a higher ketone content in the polymer would provide more reactive moieties for acylhydrazone bond formation, resulting in stiffer hydrogels. In addition to tuning polymer weight percent, varying hydrazide/ketone (Hz/K) ratio in the formulations also altered PADO hydrogel properties (Figure S3). Specifically, hydrogel shear moduli peaked (G’ = 4.3 kPa) at a stoichiometric ratio (i.e., Hz/K=1) and decreased with either component in excess, a result consistent with that reported by Liu et al.^[25] The imbalance between hydrazide and ketone led to partial dissociation of acylhydrazone bonds and decreased the shear moduli of hydrogels. These moduli results agreed well with the hydrogel microstructure as revealed in the SEM micrographs (Figure 2C-F, Figure S2), wherein hydrogels with smaller pores exhibited stronger mechanical properties.

Figure 3. PADO hydrogel rheological properties.

(A) Shear modili (G’) of PADO hydrogels at 37°C. (B) Strain step measurement of PADO_M−15%. The oscillation strain alternated between 1% and 300% for every 2 minutes at 37°C. (C, D) Temprature responsiveness of PADO_M hydrogels. The temperature was ramped up at 1 °C/min (C) or set at 10°C for 1 minute and then raised to 37°C within 1 minute (D).

To demonstrate the shear-thinning properties of PADO hydrogels, we subjected PADO_M hydrogels (15 wt%) to low (1%) and high (300%) strain cycles at 37ºC using a oscillatory rheometer (Figure 3B). During the low strain measurements, the PADO hydrogels exhibited typical elastic gel properties, with shear moduli (G’) consistently higher than the loss moduli (G”). However, when the strain was increased to 300%, G’’ (~ 90 Pa) rapidly surpassed G’ (~50 Pa), indicating that the network structure of the PADO hydrogel was deformed by the high strain. When the strain was lowered to 1% again, G’ quickly recovered back to ~330 Pa with G’’ dropping to ~10 Pa. This indicated the excellent shear-thinning property of the PADO hydrogels as its network structure was fully restored after experiencing large oscillatory shear strain. Similar behavior was also displayed in other PADO hydrogels (Figure S4). The shear-thinning property was endowed by the dynamic covalent acylhydrazone bonds, which re-connected through reversible metathesis after rupture due to the exchange between different acylhydrazone moieties.^[26]

In addition to shear-thinning property, we also evaluated the temperature responsiveness of PADO hydrogels. Similar to the PADO polymers, ADH-crosslinked PADO acylhydrazone hydrogels also showed high degree of temperature responsiveness. However, different from PADO polymers (Figure 1D), PADO-ADH acylhydrazone hydrogels exhibited slower response rate to an increasing temperature, either with a slower temperature ramping at 1 °C/min (Figure 3C) or with a rapid step increase from 10 °C to 37 °C in 1 min (Figure 3D). Nonetheless, with the addition of ADH, hydrogel precursor gelled or further stiffened above its LCST, which can be characterized via a temperature scan. For example, the precursor of PADO_M-10% and −15% started with G’~100 Pa at low temperature. The onset of G’ increasing began at ~38°C and ~21°C for 10% and 15% PADO_M, respectively. We marked these values as their LCSTs. At 20 % PADO_M, however, the precursor solution turned into soft gel even at low temperature (15°C) owing to its higher polymer weight percent and thus higher ketone concentration. As temperature was ramped up slowly to 45°C, 20% PADO_M hydrogels gradually stiffened, with G’ increased from 4 kPa to 45 kPa (Figure 3C). The temperature responsiveness of PADO hydrogels was also evaluated through a step increase in temperature (from 10 °C to 37 °C in 1 min, Figure 3D). The LCST of hydrogels decreased with increasing polymer weight percent, which is commonly observed in materials with LCST properties.^[27]

Additional testing revealed rapid gelation of PADO hydrogels induced by temperature change. In particular, 10% PADO_M hydrogels showed higher LCST (at 40°C), thus it took about 5 minutes to gel at 37°C, but the gelation time was reduced to ~2 minutes for 15% PADO_M. For the pre-gelled 20% PADO_M, moduli hardening started immediately after temperature raised to 37°C. Due to the geometry limitation of the rheometer and the nature of the hydrogels, continuing moduli hardening was observed in 10% and 15% PADO_M even after sol-gel transition. This rheological behavior may be attributed to the moisture loss in these hydrogels at 37°C. The LCST and response time for other hydrogel formulations are included in Table 1 and Figure S5.

Table 1.

Summary of PADO hydrogel LCST and gelation time.

Precursor	Polymer weight %	Hz/K	LCST [°C] a)	Gelation time [min] b)
PADOL	20%	2.0	26	4.5
PADOM	10%	2.0	40	5.0
PADOM	15%	2.0	21	2.3
PADOM	15%	1.0	25c)	0.0
PADOM	15%	0.5	33	0.0
PADOM	20%	2.0	18c)	0.0
PADOH	8%	2.0	27	0.0
PADOH	10%	2.0	27	0.0

^a)The temperature was ramped at 1°C/min between 15–45°C;

^b)The response time of hydrogel after set to 37°C.

^c)Precursor gelled and resulted in blunt LCST transition. The temperature was set at 10°C for 1 minute and then equilibrated at 37°C within 1 minute. Results were summarized from Figure 3C, D and Figure S5.

2.3. Acid-catalyzed hydrolysis of PADO hydrogels

To affirm that acid-labile acylhydrazone bonds^[28] afforded hydrolytic degradation of PADO hydrogels, we evaluated the degradation process by measuring hydrogel mass loss at different pH values (Figure 4). When the hydrogel was immersed in buffer solution at above its LCST, PADO polymers collapsed, leading to de-swelling of the hydrogel. Therefore, a parallel set of hydrogel mass loss incubated at pH 7.4 PBS was measured on dry weight basis for comparison (Table S1). The dry mass loss was 10–18%, compared to ~24% wet mass loss on day 1 at pH 7.4. These results indicated that moderate hydrogel degradation along with hydrogel de-swelling took place simultaneously at pH 7.4. After day 1, the mass loss increased slowly to reach ~40% at pH 7.4 and was relatively insensitive for hydrogels with different weight contents (Figure 4A).

Figure 4. PADO hydrogel pH responsiveness.

Mass loss % of PADO_M hydrogels at 10%, 15%, and 20% incubated at (A) pH 7.4 and (B) pH 6.0 PBS at 37°C. (C) Mass loss % of PADO_M 15% hydrogel incubated at different pH at 37°C for an extended period.

While hydrogel formulation did not significantly alter PADO gel degradation at pH7.4, the hydrolytic degradation profiles were highly depended upon polymer weight percent at pH6.0 (Figure 4B). Specifically, hydrogels disintegrated faster in lower polymer weight percent formula. By day 4, PADO_M-10% has lost 84% of its initial weight while PADO_M-20% lost only 38% at pH6.0. Compared PADO_M-15% mass loss at pH7.4, 6.0, and 5.0 (Figure 4C), the influence of acylhydrazone dissociation on PADO hydrogel degradation was even more profound. It took about 10 and 20 days to fully degraded PADO_M-15% hydrogels at pH5.0 and 6.0, respectively. In contrast, the mass loss at pH7.4 was limited at around or below 40% after 20 days. Even with a subtle change from pH7.4 to 6.8 (Figure S6), the mass loss was 23% higher than at pH7.4 on day 7, indicating that acylhydrazone dissociation had been accelerated at a milder acidic condition. The results of these highly sensitive macroscopic hydrogel degradation studies will inform the selection of formulations for controlled protein delivery.

2.4. In vitro protein release from PADO hydrogels

Bovine serum albumin (BSA) was used as a model protein to demonstrate the potential of PADO hydrogels as sustained protein delivery carriers. The protein encapsulation efficiency and in vitro release profiles were regulated by the ketone content on PADO, polymer weight percent, the hydrazide/ketone ratio, and the pH of the release medium (Figure 5, Table S2). As shown in the SEM micrographs (Figure 2D-G), both ketone content and polymer weight percent affected the pore size of PADO hydrogels. A more porous microstructure such as PADO_L-20% and PADO_M-10% might lead to faster protein diffusion compared to the less porous PADO_M-15% and PADO_M-20% at pH7.4. Indeed, protein release was completed from PADO_L-20% hydrogels within 2 days at pH7.4, whereas the release from PADO_M-20% was only ~35% (Figure S7A). Since hydrogels crosslinked by PADO_M provided tunable gelation rate and pH-responsive degradation profiles, the following experiments were conducted using PADO_M.

Figure 5. Sustained protein release from PADO hydrogels.

(A) Effect of PADO_M wt% on BSA release at pH 7.4. (B) Effect of hydrazide/ketone (Hz/K) ratio on BSA release at pH7.4. (C) Effect of pH values on BSA release. (D) Effect of polymer wt% on BSA release at pH6.0.

We evaluated the initial burst release of BSA from PADO_M hydrogels crosslinked at 10%, 15%, and 20%. As shown in Figure 5A, substantial protein release was only observed in the first few hours upon placing the PADO hydrogels into pH7.4 buffer. However, PADOM hydrogels crosslinked at 10% exhibited high burst release at ~80%, compared to ~40% for 15%, and ~35% for 20% PADO_M hydrogels (Figure 5A). This result is as expected as PADO hydrogel with smaller pore size (at higher polymer weight percent or higher ketone content) would reduce the initial burst release. Since all PADO_M hydrogels regardless of wt% did not degrade more than 40% at pH7.4 in the first 7 days (Figure 4A), we explored other parameters to control protein release. As shown in Figure 5B, formulating PADO_M hydrogels at different Hz/K ratio (0.5, 1, 2) enabled sustained protein release at pH 7.4 for at least 35 days. Of note, for hydrogels crosslinked at Hz/K = 1.0 and 2.0, the protein release leveled off from day 2 to day 14 but gradually increased afterward, likely a result from slow degradation of PADO hydrogels over a long period of time at neutral condition.

At an acidic condition (e.g., pH 6.0, 6.8), the accelerated acylhydrazone bond dissociation would cause rapid protein release. Of note, the PADO hydrogels were highly sensitive to pH change and even a small drop in pH (e.g., from 7.4 to 6.8) could result in vastly different release profiles (Figure 5C). It is also possible to control the release at pH6.0 by adjusting polymer wt%. As shown in Figure 6D, increasing PADO_M wt% from 10% to 20% extended the release period from ~5 days to over 21 days. This finding correlated well with the hydrolytic degradation profile obtained from hydrogel mass loss at pH 6.0 (Figure 4B), where the degradation rate was inversely correlated to polymer wt%. Interestingly, changing Hz/K ratio did not affect protein release at pH 6.0, where BSA finished its release within 7 days across all 3 formulations (Figure S7B). This indicated that PADO hydrogels degraded somewhat similar at acidic condition when the dissociation of acylhydrazone was accelerated regardless of Hz/K ratio. When comparing hydrogel degradation kinetics (Figure 4) with protein release profiles (Figure 5), it is worth noting that the rate of protein release was faster than that of hydrogel degradation. For example, at pH6.0 BSA was fully released from 15% PADO_M hydrogels before day 7 when the hydrogel only lost ~82% of its original mass at the same pH. This discrepancy was ascribed to the faster progress of microscale pore increase (due to rapid degradation at an acidic pH) compared to the slower macroscale mass loss in the bulk hydrogel degradation.

Figure 6. Cytocompatibility of PADO hydrogels.

(A) Cytokine secretion from THP-1 cells cultured in the absence of presence of PADO_H−10% hydrogels. (B) Live/dead staining of C28/I2 chondrocytes encapsulated in PADO_H−10% hydrogels. (C, D) H&E staining images for no hydrogel control (C) and intramuscularly injected PADO_M−15% hydrogel (D). Injection site: mice hindlimb.

In addition to BSA (MW ~ 66.5 kDa, pl ~ 4.5), lysozyme (LYS) and glucose oxidase (GOX) were also encapsulated into PADO hydrogels to study the impact of protein molecular weight and net charge on release profiles (Figure S7C). For the positively charged and smaller protein like lysozyme (MW ~14 kDa, pl ~11), the initial burst release was higher ~80%. After the initial burst phase, the lysozyme release profile was relatively pH independent and gradually reached completion on D7. This behavior indicated that the pore size of PADO was too large to confine lysozyme inside the hydrogel. For the negatively charged and larger protein like GOX (MW ~ 160 kDa, pl ~4.2), its release profile was more similar to BSA. GOX only released ~33% at pH 7.4 during 7 days incubation. However, at pH6.0 GOX experienced a fast release period and reached 60% release on day 1 and revealed a degradation-related release profile after that (Figure S7D). These findings help us define the suitable protein size range for the sustained release using PADO hydrogels as smaller proteins like lysozyme may not have the prolonged release profiles.

The BSA release profile of various Hz/K formulations at neutral condition and formulations at acidic conditions exhibited a period of near-zero order release kinetics (Figure S8, Figure S9, and Table S3) with a high correlation variance (r² > 0.91). Lysozyme and GOX also somewhat followed the near-zero order release kinetics after the initial release stage at acidic condition (Figure S10 and Table S3). This confirms that the protein release from PADO hydrogels followed an erosion-control release model.^[30] Hence, the injectable and pH-responsive PADO hydrogels are promising for local and sustained therapeutics delivery to treat chronic disease.

2.5. In vitro and in vivo cytocompatibility evaluation

To demonstrate the therapeutic potential of the new PADO hydrogels as sustained protein release carriers, we tested the cytocompatibility of the hydrogels using THP-1 cells, a human monocyte cell line; C28/I2 cells, a human chondrocyte cell line; MH7A cells, a human synovial cell line; and NIH3T3, a mouse embryonic fibroblast cell line. Cells were treated with hydrogels for 3 days, followed by viability assay. As shown in Figure S11, there was no significant difference between the hydrogel treated cells and untreated cells for all 4 cell lines, indicating that the PADO hydrogels were highly cytocompatible in vitro. Moreover, we investigated cytokine secretion from THP-1 cells as an assurance that the PADO hydrogels would not unintentionally bias monocyte polarization. Indeed, there was limited to no significant difference in the secretion of 80 cytokines/growth factors from THP-1 cells cultured in the absence or presence of PADO hydrogels (Figure 6A and Figure S12). Among the 80 cytokines, only macrophage inflammatory protein-1β (MIP-1β), a chemoattractant for monocytes,^[29] was slightly upregulated (39% with gel versus 24% without gel). All other proteins secreted from THP-1 cells were comparable with or without PADO hydrogels.

Another potential application for PADO hydrogel is to use it as a carrier for cell delivery. Since our main interest for the PADO hydrogel is in treating OA, we performed in situ encapsulation of chondrocytes and evaluated the cytocompatibility of the hydrogels via live/dead staining of the encapsulated cells. As shown in Figure 6B, we demonstrated high cytocompatibility of PADO hydrogels as the majority of the encapsulated C28/I2 chondrocytes remained alive over 7 days of in vitro culture. Since the current PADO hydrogel did not provide any bioactive site for cell adhesion and signaling, C28/I2 cells could remain in the dedifferentiation state within the hydrogels.

To test the biocompatibility of PADO_M hydrogels in vivo, we injected 15% PADO_M hydrogels in mice intramuscularly at the femurs. After the injection, the mice weights were tracked over 7 days (Figure S12) and we found no significant changes in the animal weights. Muscle tissues at the injected sites were explanted for histology analysis. Gross observation of the tissue explants 7 days after hydrogel injection did not show sign of gel remnants, indicating that the hydrogels had degraded completely in vivo. The hematoxylin and eosin (H&E) staining results also revealed no residual hydrogels and no substantial difference in the degree of cell infiltration (Figure 6C), suggestive of minimal inflammation induced by the injected PADO hydrogels. The injected PADO hydrogel has the potential to tempararily fill the joint defects while releasing proteins under the slightly acidic microenvironment of osteoarthritis to facilitate joint repair.^[31,32] While further testing and analyses are required to gain a full picture of the in vivo biocompatibility and therapeutic potential of the new PADO hydrogels, the current study nonetheless established a new class of injectable and pH-responsive hydrogel system for controllable protein release.

3. Conclusion

We have designed a new class of injectable hydrogel with dual pH and temperature responses using inexpensive acrylic-based polymers, including DAAM, DEGA, and OEGA. While DEGA was responsible for the temperature sensitivity, the crosslinking between DAAM and the crosslinker ADH controlled the pH responsiveness of the hydrogel. The PADO hydrogel properties such as LCST, relation time at 37°C, and stiffness were governed by the ketone content, polymer weight percent, and Hz/K ratio. The formulations with higher ketone content and polymer weight percent resulted in lower LCST, faster gelation rate, and stiffer hydrogels. The PADO hydrogel was demonstrated to be cytocompatible and has the potential as a protein and cell delivery carrier for sustained release. Depending on the formulations, the in vitro protein release were modulated between 2 and 14 days at pH6.0, at least 21 days at pH 6.8 with the prolonged release of at least 36 days at pH 7.4. The results indicated this hydrogel platform has great potential for slow release of protein therapeutics to treat chronic diseases.

4. Experimental Section

Materials

2-(2-Carboxyethylsulfanylthiocarbonylsulfanyl)propionic acid (CPA, 95%, Sigma-Aldrich), adipic dihydrazide (ADH, 98%, Spectrum Chemical), methanol (ACS reagent, 99.8%, Sigma-Aldrich), and bovine serum albumin (BSA, GE healthcare) were used as received. 4,4′-Azobis(4-cyanovaleric acid) (ACVA, 98%, Sigma-Aldrich) was recrystallized from methanol. Di(ethylene glycol) ethyl ether acrylate (DEGA, technical grade, Sigma-Aldrich), poly(ethylene glycol) methyl ether acrylate (OEGA, Mn ~ 480, Sigma-Aldrich), and diacetone acrylamide (DAAM, 99%, Sigma-Aldrich) were passed through the inhibitor remover column prior to conducting the polymerization.

Polymer synthesis and characterization

For an exemplary PADO_M synthesis, CPA (75 mg, 295 μmol), DAAM (1.5 g, 8.86 mmol, 30 equiv.), DEGA (5.56 g, 29.55 mmol, 100 equiv.), OEGA (2.84 g, 5.91 mmol, 20 equiv.), ACVA (16.6 mg, 59 μmol, 0.2 equiv.) and methanol (16 g) were charged in a reaction vial equipped with a stir bar. The reaction was sealed and purged under nitrogen for 15 min before conducting the reaction at 60°C for 8 h. The product was dialyzed in regenerated cellulose (MWCO = 3.5k) bag against pure water for 3 d and subsequently lyophilized with yield ≥ 90%. The feed ratio of [CTA]:[DAAM]:[DEGA]:[OEGA]:[ACVA] was set at 1:15:105:30:0.2 for PADO_L and 1:60:70:20:0.2 for PADO_H, respectively. The reaction conversion and final composition was examined by ¹H NMR (Bruker, 500 MHz) in deuterated DMSO-d₆. The number-average molecular weight of polymers and the corresponding dispersity were characterized by Waters Breeze HPLC system from the Nanoscale Characterization Facility at Indiana University-Bloomington. The system was run at 35°C in tetrahydrofuran with a flow rate of 1 mL/min. The system was equipped with a dual UV/Vis detector, Styragel HR2, HR4, and HR5 columns and calibrated by polystyrene standards. Samples were prepared at 5 mg/mL and passed through 0.45 μm PTFE filter before testing.

Polymer turbidity test

PADO polymers were dissolved in pH 7.4 PBS at 10 mg/mL and transferred to a 96-well plate. The polymer response to temperature was recorded by its absorbance at 600 nm using a microplate reader. The temperature was ramped between 30 and 46°C with a 2°C interval. The temperature was equilibrated for 5 min before each acquisition. The absorbance was converted into transmittance% (T%) according to $T\%={10}^{(2-{OD}_{600})}$ using Beer’s law.

Scanning electron microscopy (SEM)

For SEM specimen preparation, the predetermined amount of PADO and ADH were mixed in pure water overnight at Hz/K = 2. 50 μL precursor was deposited on the glass slide and incubated in a humidified chamber at 37°C for 30 min to afford hydrogels. The glass slide was quickly transferred onto the surface of liquid nitrogen to quench hydrogel microstructure. The hydrogel was then lyophilized. Prior to SEM imaging, the specimen was spotter-coated with a thin layer gold to enhance its conductivity. The hydrogel microstructure was acquired by JSM-7800F operated at 5 kV in vacuum.

Dynamic shear rheology

PADO and ADH were mixed in pH 7.4 phosphate buffer solution (PBS) overnight at Hz/K = 2 if not otherwise specified. To obtain shear moduli, the precursor was deposited into the mold with a diameter of 8 mm and incubated at 37°C for 30 min to afford hydrogel. The hydrogel was incubated in pH 7.4 PBS at 37°C for 1 h before the measurement. The shear moduli of hydrogels (n = 3) was obtained from Anton Paar MCR102 with 8 mm parallel plates at 37°C through amplitude sweep between strain of 0.1 and 5% at 1 Hz under 0.25 N normal force. For the strain step measurement, the strain alternated between 1% and 300% for 3 cycles with each step of 2 min using the same 1 Hz under 0.25 N normal force at 37°C.

For temperature scan, 100 μL precursor (n = 3) was deposited between 8 mm parallel plates with a gap size of 1 mm. The precursor was surrounded by mineral oil to prevent water evaporation. The acquisition was conducted at 1 Hz with a strain of 1% and temperature ramping between 15 and 45°C at 1°C/min. The response temperature was reported as the onset of G’ increase. The similar procedure was applied to temperature response measurement with a different temperature profile. The precursor was equilibrated at 10°C for 1 min followed by a rapid temperature jump to and equilibrated at 37°C. The response time was also reported as the onset of G’ increase.

Hydrogel mass loss test

PADO hydrogel precursor was deposited and weighed onto the 3.5 cm petri dish to obtain the initial wet weight (w_w0, n = 3). The precursor was incubated at 37°C for 30 mins to afford the hydrogel prior to the addition of warm PBS. To obtain hydrogel wet weight at time t (w_wt), PBS was decanted, and the hydrogel was blotted dry and weighed. Fresh warm PBS was added to the petri dish after weighing. The hydrogel was continued to be incubated at 37°C. The wet mass loss was calculated based on $wetmassloss\%=\left(1-w_{wt}/w_{w0}\right)\times 100\%$.

A parallel experiment measuring hydrogel dry weight was also recorded. PADO hydrogel precursor was deposited into the pre-weighed 2 mL microcentrifuge tube (n = 3). The precursor was incubated at 37°C for 30 mins and then quenched in liquid nitrogen. The hydrogel was lyophilized to obtain the initial dry weight (w_d0). The hydrogel was then immersed in pH 7.4 PBS at 37°C overnight. The PBS was then decanted. Hydrogel was quenched in liquid nitrogen and lyophilized to obtain dry weight (w_d1) on day 1. The dry mass loss was calculated based on $drymassloss\%=\left(1-w_{d1}/w_{d0}\right)\times 100\%$.

Protein encapsulation and release

PADO, ADH, and the protein (BSA, lysozyme, or GOX) were mixed in pH 7.4 PBS overnight at Hz/K = 2 if not otherwise specified. The protein concentration was 40 mg/mL. 50 μL precursor was deposited into an open-cut 1 mL syringe and incubated at 37°C for 30 min. The hydrogel was quickly rinsed with warm pH 8.0 PBS 3 times before incubating in 1.5 mL PBS at different pH at 37°C. At designated time point, 50 μL solution was aliquoted followed by replenished equal volume of PBS. Each condition has 3 replicates. To determine the amount of total encapsulated protein, another pre-washed hydrogel was incubated in 1.5 mL pH 5.0 PBS at 4°C overnight to reach a full degradation. The protein encapsulation efficiency was calculated as the amount of total encapsulated protein over protein in the precursor. The protein concentration was quantified by Bradford assay (Bio-Rad) according to manufacturer’s protocol. Briefly, 250 μL Bradford reagent was added to 5 μL sample. The absorbance was read at 595 nm using the corresponding protein as the standard.

Cytocompatibility evaluation

The cytocompatibility of PADO hydrogel was evaluated by THP-1, C28/I2, MH7A and NIH3T3 cell lines. THP-1 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco) containing 10% Fetal bovine serum (FBS, Corning) and 1% Penicillin-Streptomycin (P/S, Gibco) at 37°C with 5% CO₂. Before the treatment, THP-1 cells were centrifuged and re-suspended at 30k cell/mL in serum-containing media. Each well contained 1 mL cell solution. C28/I2 cells were cultured in high glucose Dulbecco’s modified eagle medium (DMEM) containing 10% FBS and 1% antibiotic-antimycotic (Gibco) at 37°C with 5% CO₂. C28/I2 cells were trypsinized by 0.05% trypsin-EDTA (Gibco) and seeded in 24-well plate at 5k cell/well in serum-containing media for 24 h prior to the treatment. MH7A cells were cultured in RPMI 1640 medium containing 10% FBS and 1% P/S at 37°C with 5% CO₂. MH7A cells were trypsinized by 0.25% trypsin-EDTA and seeded in 24-well plate at 8k cell/well in serum-containing media for 24 h prior to the treatment. NIH3T3 cells were cultured in high glucose DMEM containing 10% FBS and 1% P/S at 37°C with 5% CO₂. NIH3T3 cells were trypsinized by 0.05% trypsin-EDTA and seeded in 24-well plate at 5k cell/well in serum-containing media for 24 h prior to the treatment. 50 μL PADO_H 10% precursor was deposited in cell insert and incubated at 37°C for 30 mins before transferred into the wells (n = 3). Cells were incubated with hydrogels for 3 d.

After the treatment, the cell insert was removed before evaluating cell viability by (3-(4,5-diemthylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay for THP-1 and NIH3T3 cell lines, and alamar blue (Bio-Rad) assay for C28/I2 and MH7A cell lines. NIH3T3 cells were treated with 0.5 mg/mL MTT. 80 μL 5 mg/mL MTT was directly added to each well for THP-1 cells. Cells were further incubated for 4 h. The formazan crystal was dissolved in 400 μL/well DMSO. The absorbance was read at 590 nm. C28/I2 cells and MH7A cells were incubated with alamar blue reagent (0.1x diluted in serum-free medium) for 4 h before reading at 560/590 nm. The cell viability was normalized by the control group, which was not exposed to hydrogels.

Cell encapsulation

20 M/mL C28/I2 cell solution was mixed with cold PADO_H-10% precursor solution at 1:9 ratio. 25 μL cell-laden precursor solution was casted in 24-well plate and incubated at 37°C. After 30 min, regular warm cell culture media was added to the well plate. The cell viability was evaluated by staining hydrogels with calcein AM/ethidium homodimer 1 live/dead staining kit prior to confocal imaging (Olympus Fluoview FV100 laser scanning microscope). Images were z-staged from 10 μm slices over 100 μm in total.

Animal study

All animal studies were approved by the Indiana University Purdue University Indianapolis School of Science Institutional Animal Care and Use Committee (Approval number: SC303R). A total of three C57BL/6J mice were used. Under sterile conditions, 50 μl of PADO_M−15% precursor solutions were injected into the region of quadriceps muscle of the right hindlimb of each mouse. Body weights of the mice were measured daily. A week later, mice were sacrificed and the quadriceps muscles with the surrounding connective tissues were collected from both right and left hindlimbs. Muscle tissues from the left hindlimbs served as the controls. Muscle tissues were fixed in formalin for histology process. Tissue specimens were embedded in paraffin, sectioned, and stained by hematoxylin and eosin (Indiana University pathology lab).

Statistical analysis

Data were presented as mean ± SEM. One-way ANOVA was used to determine the statistical significance between groups when p < 0.05.