Restoring Carboxylates on Highly Modified Alginates Improves Gelation, Tissue Retention and Systemic Capture

Abstract

Alginate hydrogels are gaining traction for use in drug delivery, regenerative medicine, and as tissue engineered scaffolds due to their physiological gelation conditions, high tissue biocompatibility, and wide chemical versatility. Traditionally, alginate is decorated at the carboxyl group to carry drug payloads, peptides, or proteins. While low degrees of substitution do not cause noticeable mechanical changes, high degrees of substitution can cause significant losses to alginate properties including complete loss of calcium cross-linking. While most modifications used to decorate alginate deplete the carboxyl groups, we propose that alginate modifications that replenish the carboxyl groups could overcome the loss in gel integrity and mechanics. In this report, we demonstrate that restoring carboxyl groups during functionalization maintains calcium cross-links as well as hydrogel shear-thinning and self-healing properties. In addition, we demonstrate that alginate hydrogels modified to a high degree with azide modifications that restore the carboxyl groups have improved tissue retention at intramuscular injection sites and capture blood-circulating cyclooctynes better than alginate hydrogels modified with azide modifications that deplete the carboxyl groups. Taken together, alginate modifications that restore carboxyl groups could significantly improve alginate hydrogel mechanics for clinical applications.

1. Introduction

Hydrogel biomaterials offer utility in biomedical applications due to their stability, tunable mechanics, and degradation profiles, as well as biocompatibility with surrounding tissues. Clinical and preclinical applications of hydrogel biomaterials include tissue engineered constructs [1-3], depots for drug [4-8] and cell delivery [9,10], and cellular scaffolds used in the study of biological processes [11-13].

Among hydrogels, alginate is rapidly gaining attention because it gels under neutral, physiological conditions, exhibits good biocompatibility at tissue sites, and has wide chemical versatility. Alginate gelation utilizes ionic cross-links between the carboxyl groups on alginate and divalent cations. The calcium cross-linked hydrogels are injectable and self-healing [14-16], enabling facile injection into tissues [5,17]. In vivo, calcium cross-linked alginate hydrogels elicits low levels of foreign body responses, very low toxicity and limited immunogenicity [1,18-20]. In addition, alginate is generally recognized as safe (GRAS) by the FDA [21], which has motivated preclinical testing to deliver drugs [22-24], biologicals [25-27], viruses [28,29] and cells [30,31] and clinically as a dietary supplement [32], material for wound dressings [33,34], sealant agent [35], and as an injectable implant [36-38].

Chemical functionalization of alginate is rapidly gaining attention to improve or modulate the mechanical and biological properties of this important material. Chemical functionalization with small molecules regulates immune and foreign body response [39-41], functionalization with peptides mediates cellular and tissue responses [42-44], and modification with reactive chemical groups enables new modes of drug delivery [24,45-49]. Alginate polymers conjugated to bioorthogonal “click” chemical motifs enhance cross-linking [15,50-54] and expedite polymer modification [55-57]. More recently, alginates modified with click motifs have been used as targetable drug depots [18,58-60], capable of repeatedly capturing and releasing drugs. For many of these applications, a central limiting factor has been low achievable degree of substitution, limiting modification to 5-10% of all sugars within the polymer. Expanding alginate’s usage to more applications will require increasing modification density and using multiple modifications. Robust and predictable methods for chemical functionalized alginate must be developed to meet this need.

Although alginate is straightforward to chemically modify, modification carries undesired complications. Most frequently, alginate polymers are modified through carbodiimide coupling between the carboxyl group on alginate and nucleophiles such as alcohols or amines. The popularity of this modification strategy is due to the wide array of available nucleophiles, the solubility of all reagents in aqueous conditions and lack of harsh acids and bases, which minimizes polymer degradation. Unfortunately, this chemical modification inhibits gelation. Indeed, alginate hydrogels modified to a high degree of substitution (DS) suffer from poor or nonexistent calcium cross-linking. We and others have demonstrated that poor calcium cross-linking directly caused by high DS causes loss of gel stiffness, migration of implants away from desired injection sites, and enhanced calcium leaching, leading to an increased foreign body response to the gel [50,61].

One approach to loss of calcium cross-linking at high DS is to use alternative cross-linkers, including click cross-linking [50,61], but this approach has several drawbacks. Introduction of new chemical cross-linkers creates additional complication to the regulatory pathway for clinical use and the new crosslinks may have unexpected physiological toxicity. Additionally, these cross-linking strategies rely on formation of covalent bonds and thereby eliminate two main advantages of using alginate in the first place - alginate’s ability to self-heal after injection and its shear-thinning characteristic.

To overcome the challenge of achieving high DS modification without sacrificing calcium cross-linking, we report a new strategy that uses modifications that restore carboxyl groups on the alginate. Since the carboxyl groups in alginate monomers are required for both calcium cross-linking and for carbodiimide coupling, we hypothesized that the depletion of the carboxyl groups during EDC coupling was responsible for the observed loss of calcium cross-linking and that further increasing the DS on alginate will worsen this effect. We propose that the modifications that are coupled to alginate carry their own carboxyl groups so that carboxyl groups are replaced at every modified spot (Fig. 1). In this report we demonstrate that alginate gels conjugated to azide modifications that restore the carboxyl groups (restorative modifications) provide much improved calcium cross-linking and gelation properties as compared azide modifications that deplete the carboxyl groups (depletive modifications). We also show that, in vivo, alginates with restored carboxyl groups have improved retention and drug capture at injected sites. Taken together, restorative modifications are a promising approach to highly modified calcium cross-linked alginate hydrogels.

Figure 1. Overview of depletive and restorative alginate modifications.

Alginate can be modified with chemical groups (blue) through the carboxyl groups (red), which depletes the carboxyl groups, and destroys calcium cross-linking. In contrast, if the modifications restore carboxyl groups, alginate gels maintain calcium cross-linking and gelation.

2. Materials and Methods

Reagents

Nutrition grade Alginate (Protanal LF 20/40, average MW = 300 kDa, >60% guluronic acid) was purchased from Dupont Nutrition and Health. Calcium sulfate dihydrate (C3771), MES (M3671), and N,N-Diisopropylethylamine (DIPEA, D125806) were purchased from Sigma-Aldrich. Azide-PEG4-amine (1868) was purchased from Lumiprobe corporation. DBCO-sulfo-amine (1227) and DBCO-Cy7 (1047) were purchased from Click Chemistry Tools. 1-Ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) was purchased from Oakwood Chemical (024810). Azidoacetic acid (35109), N-Boc-L-lysine (02708) were purchased from Chem-Impex. N-hydroxysuccinimide (NHS) was purchased from Chem-Impex International (00182) and Alfa Aesar (A10312). 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (A299) was purchased from AK Scientific. N,N-Dimethylformamide (DMF) was purchased from Acros Organics (227056). Trifluoroacetic Acid (O4901-500), Methanol (A433P-4), Acetonitrile (A998-4) and Toluene (T290-4) were purchased from Fisher Chemical.

2.1. Azide-lysine synthesis

(S)-2-ammonio-6-(2-azidoacetamido)hexanoate (S4) was prepared utilizing a modified literature protocol [62].To a flame dried 250 mL round bottom flask was added NHS (3.037 g, 26.39 mmol, 1.3 equiv.) and stir bar. Septa was added with an Argon balloon. 101 mL of anhydrous DMF (0.2 M) was added and reaction was set to stirring. Septa was removed and EDC (5.06 g, 26.4 mmol, 1.3 equiv.) was added. Septa was replaced and the reaction was set to stirring at room temperature. Then 2-azidoacetic acid (S1) (1.52 mL, 20.3 mmol 1 equiv.) was injected into reaction. Reaction was stirred for 2 hours upon which septa was removed and N-Boc-L-lysine (S2) (5.0 g, 20.3 mmol, 1 equiv.) and DIPEA (12.7 mL, 73.1 mmol, 3.6 equiv.) were added in that order. Septa was replaced and reaction was stirred for an additional 17 hours. Over the course of 1 hour, the reaction became an amber color and remained this color during the completion of the reaction. The crude reaction mixture was concentrated on a rotary evaporator followed by high vacuum to remove all DMF and DIPEA leaving an amber oil. Toluene was added and then evaporated by rotary evaporator and high vacuum to help remove any trace DMF and DIPEA. This crude oil was then taken forward to the next step without purification. To the crude N6-(2-azidoacetyl)-N2-(tert-butoxycarbonyl)-L-lysine (S3) (6.69 g, 20.3 mmol, 1 equiv.), 44 mL of anhydrous dichloromethane and 22 mL of trifluoroacetic acid (2:1 ratio, 0.3 M reaction molarity) was added at room temperature. Reaction was allowed to stir for 4 h, upon which complete conversion was observed on LCMS. Reaction was then concentrated to volume on a rotary evaporator. Then 100 mL of acetonitrile was added and this was concentrated. Then 100 mL of toluene was added and this was also concentrated. This crude oil was then triturated twice with a small volume of methanol followed by addition of copious amounts of diethyl ether providing (S)-2-ammonio-6-(2-azidoacetamido)hexanoate (S4) (3.56 g, 15.5 mmol, 77%) as a light pink/brown solid with trace impurities. Isolated solid was triturated one time further obtaining (S)-2-ammonio-6-(2-azidoacetamido)hexanoate (S4) (2.10 g, 9.16 mmol, 45%) pure as a light tan solid.

2.2. Azide conjugation of alginate

Nutrition grade, high guluronic acid content, high molecular weight (MW) alginate (PROTANAL LF 20/40) purchased from Dupont Nutrition and Health was first dissolved in DI water and charcoal was added (0.5 g of charcoal per 1 g of alginate), filtered, and then MES buffer was added to reach the desired 0.5% weight per volume (w/v) concentration (1 g of alginate in 200 mL, 20 μM, 1 eq.) (100 mM MES, 300 mM NaCl, pH: 6.5). To begin the coupling reaction, EDC (40 mM, 2000 eq.) and NHS (20 mM, 1000 eq.) was added to the alginate solution and allowed to stir for 5 minutes. Either azide-lysine (10 mM, 500 eq.) or azide-amine (10 mM, 500 eq.) was added to the solution, and both solutions stirred overnight, covered, at room temperature. The solution was dialyzed against 4 L of water with successively lower salt content, changing solution 2-3 times per day for 4-5 days. Dialyzed solutions were frozen and lyophilized under high vacuum. For additional coupling, the samples were dissolved in 0.5% w/v in 1X MES buffer solution (1 g in 200 mL) and the coupling steps were repeated as described. Each round of coupling results in incremental increases of DS so that batches could be closely compared in DS. Four rounds of coupling resulted in high DS alginate for both depletive and restorative alginates.

2.3. Azide quantitation

The number of azides conjugated to the alginate was quantified by looking at the decrease in DBCO absorbance upon incubation of alginate azides with a known excess amount of DBCO [50]. 0.1% w/v (2.5 mg in 2.5 ml) alginate solutions were made in phosphate buffer solution (PBS). 80 μM solutions of DBCO-amine (Click Chemistry Tools-1227) were made and three different amounts of alginate was added (400 pmol, 800 pmol and 1.6 nmol of alginate) to separate 80 μM solutions of DBCO-amine. A DBCO negative control solution was also tested along with a positive control consisting of DBCO reacted with sodium azide (1 μmol, 2500 eq.). Spectrophotometry was performed using a UV/Vis spectrophotometer (Thermo Scientific Nanodrop 2000c) using a cuvette with a 1 cm pathlength. Absorbance changes were observed at 308 nm wavelength. This process was repeated with replicates of the 0.1% alginate (alg) solutions and the absorbance values were averaged. Decrease in absorbance in alginate samples compared to DBCO negative control indicated the quantity of azides that reacted with DBCO.

$$\begin{gathered} DS=\frac{\#azides}{\#strands}=\frac{molazide}{molalgstrands}=\frac{molazide}{massalg∕MWalg}=\frac{molazide\ast MWalg}{massmodifiedalg-massazides} \\ =\frac{molesazide\ast MWunmodifiedalginate}{massofmodifiedalginate-molesazide\ast MWazidealone} \end{gathered}$$

2.4. Gel Formation

Alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). 10:1 ratio of the 2% w/v alginate solution was mixed with a calcium sulfate solution (final concentration of 18.2 mM) in a two-syringe system while minimizing air. The solutions were mixed by pushing the syringe barrels back and forth 10 times. This system was used to create gel sizes ranging from 400 μL to 1 mL.

2.5. Rheology Testing

Alginate was cross-linked as described in section 2.4, with 1 mL gels mixed and injected into a 24-well plate. The gels were allowed at least 24 hours to settle and then collected for rheology testing. Rheology test of prepared gel was performed on a Molecular Compact Rheometer (MCR302, Anton Paar, Graz, Austria). A cone plate geometry of a 50 mm diameter and 1 degree (CP50-1/TG, Anton Paar, Graz, Austria) was used in this experiment. After installing the geometry, zero force and zero gap with a truncation gap of 0.1 mm was set. Before analysis began, to calibrate the system, adjust measurement system inertia and adjust system tests were performed. After calibration was done, 1 mL of gel samples (gently mashed with a lab spatula) was placed on the preinstalled temperature-controlled Peltier plate at 21 °C. The geometry cone was set at the truncation height and excess of samples were trimmed. Before each test, cone geometry and Peltier plate were cleaned with distilled water and absolute ethanol and instrument was calibrated to ensure the consistency.

To determine the rheological characteristics of the gels, the following simultaneous tests were performed using the method of Chen et al. [63] with slight modification.

Time sweep:

Two minutes of time sweep was performed at 0.2% strain and a frequency of 10 Hz and a total 150 data points at a constant rate were collected before and between each test performed.

Amplitude sweep:

Amplitude sweep test of the gel samples was performed on a logarithmic ramp ranging from 0.01 to 500% at 10 Hz with 3 seconds of conditioning and sampling time, and 10 data points per decade were recorded.

Frequency sweep:

Frequency sweep on a logarithmic ramp ranging from 0.01-100 Hz at 0.2% strain was performed with 3 seconds of sample conditioning and sampling time, and 10 data points per decade were recorded.

Cyclic amplitude sweep:

A continuous cyclic amplitude sweep test was performed in five intervals (1-5). Intervals 1, 3 and 5 were set at 0.2% strain, 10 Hz for 2 minutes, whereas intervals 2 and 4 set at 500% strain, 10 Hz for one minute.

Shear rate ramp:

Two continuous shear rate ramps from 0 to 50 s⁻¹ and 50 to 0 s⁻¹ for 2.5 minutes each were performed to study the continuous flow behavior of the gel. Total of 74 data points with 20 data points per decade was recorded.

2.6. Degradation and Swelling

Swelling:

Alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). Alginate solutions were cross-linked with calcium sulfate slurry (0.41 g CaSO₄/ml H2O, ratio of 40 μl CaSO₄ slurry to 1 ml alginate) and mixed in two syringes. 100 μL of gel was injected into 2 mL microcentrifuge tubes and allowed to further gel overnight. 500 μL of 0.1 M CaCl₂ solution was added to each tube. Samples were weighed to find the swelling ratio using the formula swelling ratio = (wet weight – dry weight) / dry weight. Samples were repeatedly weighed over the course of 24 hours. High DS depleted alginate was not measured due to the lack of gelation.

Degradation:

Alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). Alginate solutions were cross-linked with calcium sulfate slurry (0.41 g CaSO₄/ml H2O, ratio of 40 μl CaSO₄ slurry to 1 ml alginate) and mixed in two syringes. 100 μL of gel was injected into 2 mL microcentrifuge tubes and allowed to further gel overnight. 500 μL of 0.1 M CaCl₂ solution was added to each tube. At specific time points, the samples were weighed (for wet weight) and freeze dried (for dry weight). High DS depleted alginate was not measured due to the lack of gelation.

2.7. Cytotoxicity of modified alginates in vitro.

Sterilization of 2% alginate solutions were prepared by placing lyophilized alginate under UV treatment for 15 minutes. Alginate was then dissolved in sterile PBS (2% w/v, 20 mg in 1 mL). Additionally, a restorative alginate solution was cross-linked with calcium sulfate slurry (0.41 g CaSO₄/ml H2O, ratio of 40 μl CaSO₄ slurry to 1 ml alginate) and mixed in two syringes. Neonatal human dermal fibroblasts (HDFn) were cultured for this experiment using DMEM Media with 10% FBS and 1x Penicillin-Streptomycin and L-Glutamine. A sterile 12-well plate was seeded with 5714.28 cells/cm² (20,000 cells/well). HDFn were incubated at 37 °C overnight to allow adherence to the bottom of the 12-well plate. Following overnight incubation, the cell media was aspirated. Stock solutions of media with restored alginate polymers or depleted alginate polymers were created in 15 mL centrifuge tubes with 1080 μL of media and 20 μL of alginate polymer solution. A volume of 1.2 mL of media was added to negative and positive controls, and restored alginate cross-linked gel samples. The restored alginate gels were injected on top of media at a volume of 20 μL. The HDFn were incubated at 37 °C for 24 hours. Following a 24 hr incubation period, the cells were stained with 2 drops of Invitrogen NucGreen™ Dead 488 ReadyProbes™ Reagent (SYTOX™ Green) and NucBlue™ Fixed Cell ReadyProbes™ Reagent (DAPI), incubating for 15-20 min. The stained cells were imaged using Echo Revolve, using wavelengths DAPI and AF488.

2.8. SEM Imaging

Alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). Alginate solutions were cross-linked with calcium sulfate and mixed in two syringes. 100 μL of gels were stored and allowed to further gel overnight. The gels were then freeze dried and provided for SEM imaging. Dry gels were coated with 70 nm AuPd (Au: 60%, Pd: 40%) for 10 min at 7 nm min⁻¹ and analyzed on Hitachi S-3200N Variable pressure SEM. Magnifications were chosen to highlight pore sizes and distinct structures.

2.9. Gel retention of intramuscular implanted alginate gels

All animal work was done in compliance with institutional ethical use protocols, including the NIH Guide for Care and Use of Laboratory Animals. Unmodified and modified alginate were prepared as described in section 2.4 with the addition of 21.8 μM DBCO-Cy7. 12-week-old CD1 mice (Charles River, 022) were injected intramuscularly in the left limb with 50 μL of azide reactive fluorophore mixed with unmodified alginate, depleted azide-alginate, or restored azide-alginate (n=4). Cy7 fluorescence was monitored over two weeks using an IVIS imager to obtain a fluorescence signal. ICG/ICG excitation and emission filters were used for all IVIS images presented and no image math in the Living Image software was performed. For all IVIS images, only radiance efficiency values were used to normalize the data over variable exposure times. Regions of Interest (ROIs) were used to sum the fluorescent signal associated with the injected calf, the injected ankle, and the contralateral calf.

2.10. Systemic capture of fluorophore in intramuscular implanted alginate gels

All animal work was done in compliance with institutional ethical use protocols, including the NIH Guide for Care and Use of Laboratory Animals. Unmodified and modified alginate were prepared as described in section 2.4. 12-week-old CD1 mice (Charles River, 022) were injected intramuscularly in the left limb with 50 μL of unmodified alginate, depleted azide-alginate, or restored azide-alginate (n=3). 5 g/L stock of DBCO-Cy7 in water was prepared and diluted 100x and sterile filtered. 1 week after injection, 100 μL of the sterile 50 mg/L DBCO-Cy7 solution was injected retro-orbitally. Cy7 fluorescence was measured after 1 week using an IVIS imager to obtain a fluorescence signal. ICG/ICG excitation and emission filters were used for all IVIS images presented and no image math in the Living Image software was performed. For all IVIS images, only radiance efficiency values were used to normalize the data over variable exposure times. ROIs were used to sum the fluorescent signal associated with the left calf.

2.11. Statistical Analysis

Figures show mean ± SEM. For experiments described in figure 3, two-way ANOVA followed by Tukey’s multiple comparison test was performed and compares all groups. Statistical significance represented as †,* p<.05 ††,** p<.01, †††,***p < 0.001, and ††††,****p < 0.0001. In experiments described in figure 4, one-way ANOVA followed by Tukey’s multiple comparison test of all groups was performed. Statistical significance represented as **p < 0.01, and ****p < 0.0001. For cytotoxicity test, statistical significance represented as ****p<.0001 by ordinary one-way ANOVA followed by Tukey’s multiple comparison test. N=3. For swelling and degradation statistical significance is determined by multiple unpaired t tests assuming gaussian distribution. ns = p>0.05.

Figure 3. Calcium cross-linked alginate gels with restorative azide modification reduce off-target gel migration.

(A) Representative images (same BLI scale) and (B) quantitation of fluorescence signal after intramuscular injection of alginate hydrogels over 2 weeks with either carboxyl-depletive or -restorative modifications. Gels were labeled with an azide-reactive fluorophore with unconjugated fluorophore service as negative control. (C) Representative locations for quantified off-target migration. Quantitation of fluorescence signal in non-injected contralateral limb (D) and ankle (E) over 2 weeks. Region of interest (ROI) values were quantified as total radiance efficiency in equivalent sized regions. Figures show mean ± SEM. For statistical testing * compare statistical significance between modified alginate and unconjugated fluorophore and † compare depletive to restorative modified alginates. Statistical significance represented as ††,** p<.01, †††,***p < 0.001, and ††††,****p < 0.0001 by two-way ANOVA followed by Tukey’s multiple comparison test. Full images of all mice can be found in Supplementary Figure S6.

Figure 4. Restorative azide modification improves on-target capture of circulating DBCO fluorophores.

A) Representative images (same BLI scale) and B) quantitation of fluorescence at intramuscular sites from i.m.-injected alginate hydrogels cross-linked with calcium. DBCO-Cy7 was administered i.v. one week following gel injection. One week following i.v. administration, mice were imaged to compare the capture of the fluorescent signal at the injected site. ROI values were quantified as total radiance efficiency in equivalent sized regions. Samples show mean ± SEM. Statistical significance represented as **p < 0.01, and ****p < 0.0001 by one-way ANOVA followed by Tukey’s multiple comparison test of all groups. N=3.

3. Results

3.1. Synthesis of alginate polymers modified with depletive and restorative azide modifications.

We and others have previously found that alginate polymers modified with azide modifications lose the ability to cross-link in the presence of calcium [24,50]. In these previous alginate modifications, azide-PEG4-amine (azide-amine) was conjugated to alginate through carbodiimide coupling, replacing the carboxyl group on the alginate with an amide. Because these modifications deplete the available carboxyl groups, we name these modifications “depletive”. We hypothesized that modifications that contain carboxyl groups would restore calcium cross-linking and termed such modifications “restorative”.

To prepare alginate strands with restored carboxyl groups, we hypothesized we could conjugate the carboxyl-containing azide-lysine S4. Azide-lysine was prepared utilizing a modified three-step literature protocol [62] starting from 2-azidoacetic acid (Supp. Scheme S1). Synthesis of compound S4 began with NHS derivatization of 2-azidoacetic acid S1 with NHS. The azide-NHS was coupled to α-BOC protected lysine S2 to give α-BOC-γ-azido-lysine S3, which was deprotected to give compound the final azide-lysine S4 (Supp. Fig. S1).

Preparation of alginate modified with depletive and restorative azide modifications was achieved through successive rounds of carbodiimide coupling. Each round utilized EDC (2000 eq.), NHS (1000 eq.), and azide (500 eq.) followed by extensive dialysis to remove reactants and lyophilization. Table 1 shows azide quantitation of alginates for each type of modification after a single and multiple rounds of coupling. The initial coupling of alginate to the carboxyl-depletive modification azide-amine (Supp. Scheme S2) yielded a degree of substitution of 6.4% (87 azides/strand) while coupling to the carboxyl-restorative modification azide-lysine (S4) yielded a degree of substitution of 3.8% (51 azides/strand). These low DS materials were submitted to successive rounds of coupling, which increased the DS for both depletive (13.2%; 179 azides/strand) and restorative (10.7%; 145 azides/strand) modifications.

Table 1: Select information for modified alginate polymers used in this study.

High DS was achieved through successive carbodiimide couplings using restorative and depletive modifications. Degree of substitution is defined as the number of azides per alginate strand and as % of COOH modified.

Modification Type	Alginate Type	Azide Material	Degree of Substitution (azides/strand)	Degree of Substitution (% COOH)	DS Classification
Restorative	LF 20/40	Azide-Lysine	51	3.8%	Low
Restorative	LF 20/40	Azide-Lysine	145	10.7%	High
Depletive	LF 20/40	Azide-Amine	87	6.4%	Low
Depletive	LF 20/40	Azide-Amine	179	13.2%	High

3.2. Modification of alginate with carboxyl-restorative groups efficiently maintains calcium cross-linking

We next assessed whether carboxyl restoration also restores calcium cross-linked gel mechanics and injectability. 2% (w/v) Alginate solutions were cross-linked with a final calcium concentration of 18.2 mM). Calcium sulfate was chosen because its low solubility enables slow dissolution, increasing the time to gelation and improving injectability [64,65]. At low DS, both restorative and depletive gel modifications permitted gel formation, likely because at low substitution levels enough carboxyl groups are left for calcium cross-linking. At high DS, depletive modifications completely eradicated alginate gelation. In sharp contrast, alginate gels modified with restorative modifications maintained gel integrity (Fig. 2A).

Figure 2. Restorative modifications at high degrees of substitution maintain alginate gel mechanics.

A) Photos of calcium cross-linked alginate gels using depletive or restorative modifications at low and high DS azide. B) Representative strain sweep of calcium cross-linked alginate gels with restorative modifications (linearity limit of γ_L=24%) showing viscoelastic behavior. N=3 for each gel as well as carboxyl-depleted and unmodified hydrogel can be found in Supplementary Figure S2. ±5 % tolerance range of deviation was used to select the plateau value of the linearity limit. C) Representative frequency sweeps [1-100 Hz] within the LVE region showing the gel-like behavior and structural stability of calcium cross-linked gels modified with high DS restored azide-alginate and high DS depleted azide-alginate. Storage modulus (G’) and loss modulus (G”) measurements are shown. N=3 for each gel and low DS gel mechanics can be found in Supplementary Figure S3. D) Cyclic strain time sweep rheology showing self-healing behavior of calcium cross-linked alginate gels with restorative modifications demonstrated by the recovery after repeated deformation of high strain [500%] followed by low strain [0.2%] (frequency = 10Hz). N=3 for each gel type as well as unmodified and carboxyl-depleted gels are shown in Supplementary Figure S5.

Alginate hydrogels formed through vigorous mixing of alginate and calcium sulfate were subjected to rheological testing. Rheological testing confirmed that while low DS alginate with either modification form viscoelastic gels, only the carboxyl-restoring modifications permitted calcium cross-linking at high DS. Because the advantage of restorative modifications was most marked at high DS, subsequent experiments compared high DS carboxyl-depleted and -restored alginate gels.

Viscoelastic behaviors and injectability are crucial characteristics to hydrogel function. Restored alginate gels were submitted to strain sweep rheological testing. Alginate with restorative modifications showed gel-like behavior (G’ > G") with a relatively strong three-dimensional network as indicated by increase in G" at high shear strains (Fig. 2B). The linearity limiting value of the linear viscoelastic (LVE) range for the sample with restorative modifications was 26% as compared to 5% for unmodified gels (Fig. 2B, Supp. Fig. S2), indicating the modifications induced a relatively high degree of cross-linking and strengthened the structural organization of the sample. Frequency sweep tests were performed within the LVE region (Fig. 2C, Supp. Fig. S3). The figures showed that at low frequency range, G’>G" and a flat slope are found in the sample with restorative modifications, whereas the sample using depletive modifications showed the opposite results. The results confirmed that the sample with restorative modifications is in a gel-like state and is relatively more structurally stable when at rest in comparison to the sample using depletive modifications (Fig. 2C). A rotational test was performed to characterize injectability. The increased shear rate induced a decrease in viscosity in the alginate gels with restorative modification (shear-thinning behavior) similarly to unmodified alginate (Supp. Fig. S4). The reversibility of the alginate-calcium bonds confers self-healing behavior to calcium cross-linked alginate gels[66,67]. To test whether restorative modifications also conferred this property, self-healing behavior was assessed through cyclic strain time sweep rheology experiments [63]. Restorative gel samples showed preserved self-healing behavior (Fig. 2D, Supp. Fig. S5).

The impact of these modification on alginate hydrogel swelling and degradation was next studied. There were no significant changes in swelling and degradation (Supp. Fig. S6-7). However, we found interesting changes to the microstructure of the gels when observed with SEM (Supp. Fig. S8). Increase in DS using restorative modification seemed to increase porosity. Low DS using depletive modification had a large impact on the structure.

Additionally, we tested the cytotoxicity of modified alginate materials. Uncross-linked polymeric solutions or calcium cross-linked gels were incubated with neonatal human dermal fibroblasts and found no difference in percent cell viability as measured by live/dead staining after 24 hours (Supp. Fig. S9). Taken together these results demonstrate that in sharp contrast to modification that deplete alginate carboxyl groups, alginate polymers conjugated to restorative modifications maintain calcium cross-linking, cytocompatibility, injectability, self-healing and other desired gel properties.

3.3. Alginate conjugated to restorative modifications demonstrates improved gel retention and systemic capture in vivo.

Hydrogel function requires gels to remain at injected sites. Previously, we found that poorly cross-linking hydrogels lose their migrate to other parts of the body [50]. We tested whether alginate carrying depletive and restorative modifications are retained at the site on injection sites.

We tested in vivo gel retention of high DS carboxyl-depleted and carboxyl-restored alginate gels after intramuscular (i.m.) injection into the hind limb. Azide-modified alginate gels were incubated with Cyanine7-dibenzocyclooctyne (Cy7-DBCO), which covalently reacts with azide modifications to label the alginate. As a control of fast tissue clearance, Cy7-DBCO was incubated with unmodified alginate, which lacks the azide modification. As expected, the unmodified alginate control quickly shed fluorescence signal owing to clearance of the unconjugated dye from the gel (Fig. 3A-B). Both high DS azide-alginates maintained signal at the injected site and were statistically significant compared to unconjugated fluorophore after the first day and through the entire two-week period.

To quantify the off-target accumulation of alginate at undesired sites, we measured the fluorescence in the ankle of the injected limb and in the non-injected contralateral limb (Fig. 3C). We observed no significant off-target accumulation with the restored alginate as compared to unconjugated fluorophore. In contrast, we found depleted alginate to be significantly different from unconjugated probe and restored alginate after the first day (ankle) and the third day (contralateral limb) and through the entire two-week period (Fig. 3D-E). Interestingly, a small amount of off-target fluorescence appeared to resist clearance, even in the unconjugated fluorophore samples. The reason for this residual fluorescence is unclear and will be the subject of further studies. Taken together, this data demonstrates that loss of hydrogel mechanical strength leads to migration of material from the injection site to undesired locations throughout the body.

In previous reports, we demonstrated that azide-modified alginate captures circulating DBCO molecules from the blood [18,50] for applications in ultra-specific drug delivery. Therefore, we tested whether restored alginate could also capture systemically circulating fluorescent DBCO molecules at intramuscular sites. Alginate gels with restorative and depletive modifications were injected intramuscularly into the hind limb of outbred CD1 mice. One week after i.m. injections, DBCO-fluorophore was administered intravenously (i.v.), and fluorescence in muscle was measured over time. Unmodified alginate, which lacks azide modifications and therefore cannot capture Cy7-DBCO, was used as a negative control. One week after i.v. injection, although both gels captured systemically circulating DBCO-Cy7, the restored gels showed significantly higher fluorescence capture (Fig. 4A-B).

Taken together, our data demonstrates that the improvements in gelation observed in vitro directly translate to in vivo function, highlighting the importance of stabilizing hydrogel mechanics for in vivo studies. Restorative modifications enable high levels of modification without sacrificing alginate gelation.

4. Discussion

This paper reports that restoration of carboxyl groups during alginate modification preserves the calcium cross-linking of highly modified alginate gels. Carboxyl restoration enables increasing the degree of substitution while maintaining mechanical properties. In addition, carboxyl restoration improves in vivo retention of highly modified alginate gels at injection sites. Finally, carboxyl restoration improved click-specific targeting of small molecules to locally injected alginate hydrogels.

We have previously demonstrated increasing alginate degree of substitution through repeat couplings. However, we and others[24,50] have shown that alginate gels with high DS fail to cross-link, forcing us to resort to alternative cross-linking strategies [50]. In this work we show that these alternative cross-linker strategies are not necessary and that calcium cross-linking can be recovered by restoring the carboxyl groups.

The two modifications, azide-lysine and azide-amine presented different yields in EDC/NHS-mediated coupling to alginate. As has been previously reported, the overall yields of these reactions are quite low, with 500 equivalents of azide yielding only 50-80 coupled azides per strand. We hypothesize that the poor yields derive from hydrolysis of alginate-NHS esters as well as steric hindrance preventing nucleophilic attack on the NHS carbonyl when attached to large alginate polymers. We observed slightly higher yields in EDC/NHS-mediated conjugation for the depletive modification as compared to the restorative. The reason for this difference is not entirely clear, but may have to do with the nucleophilicity of the amines on the respective molecules. The difference in coupling yields led to a difference in DS for the highly modified materials (179 for depletive, 145 for restorative) after four couplings. While we do not believe that the small difference in DS between restorative and depleted gels impacts the conclusions of this work (depletive alginates with a DS=130 also failed to gel, Supp. Fig. S11), future efforts will focus on varying reactant concentration and reaction time to improve yields as well as bring these two molecule's coupling into alignment. An interesting possibility is that restorative modification of alginate could allow modification of alginate far more than that possible through depletive modification. Modifications that restore carboxyl groups create new sites for EDC/NHS reactivity, allowing for further coupling.

Despite the significant improvements in gelation efficiency, there was a still a decrease noted in the mechanical properties of high DS restored gels as compared to low DS gels with the same modification. This decrease in storage modulus could be because the restored carboxyl groups are not unable to fully participate in the “egg-box” model of calcium cross-linking [68-70]. However, the egg-box model has recently been questioned [71] and much further work will need to be done to elucidate the full mechanism. Future work could focus on introduction of covalent cross-links [50,61] or incorporation of nanomaterials [72-74] to further improve mechanical properties and controlled drug delivery capability.

Interesting changes in gel microstructure were noted by SEM as DS increased. This was surprising, but follows a similar trend noted in the mechanical properties and could be related to the interaction of the modified alginate polymers with calcium ions. Further investigation is required to better understand the mechanisms involved.

Hydrogel retention at introduced sites is central to their use as tissue engineering scaffolds and as cell and drug delivery platforms. It was initially surprising that alginate hydrogels incorporating both restorative and depletive modifications were well retained over two weeks, especially in light of previous published results in which poorly cross-linked alginate rapidly migrated away from the injection site [50]. Thus, despite the complete loss of gelation behavior in vitro, some hydrogel coherence must have remained. Calcium cross-linked alginate gels containing depletive modification showed significantly increased presence at off-target sites, indicating that the loss of cross-linking does lead to polymer shedding over time. Lack of precise control over hydrogel localization and off-target accumulation can severely impede clinical translation so the restorative modifications would be highly desirable in scaffold and drug delivery platforms that require high DS.

Alginate hydrogels with restorative carboxyl groups improved small molecule capture at intramuscular sites. Despite, the lack of gelation and the migration of polymer to off-target sites, depleted gel still managed to capture DBCO-fluorophore from the circulation. Although, surprising, the result has been observed in previous studies [44]. We hypothesize that the pressures of the intramuscular environment confine the polymer to the location. The amount of DBCO-fluorophore injected is still a small percentage of the theoretical number of available azides and we only expect 1-10% of the total molecule reaching the gel at any point [18,57,58], but the increased chance of an interaction between an azide and a circulating DBCO molecule in the restorative gels increases the accumulation.

5. Conclusions

Taken together, alginate modifications that restore carboxyl groups on the polymer backbone represent a significant improvement in hydrogel stability for in vitro and in vivo applications. This simple change to alginate functionalization makes possible mechanically robust, highly modified alginate gels with superior calcium cross-linking, mechanical properties, in vivo retention, and depot capture. Future studies could focus on further increasing azide degree of substitution, testing of alternative modifications and preclinical testing in disease models.

Statement of Significant.

Chemical modification of hydrogels provides a powerful tool to regulate cellular adhesion, immune response, and biocompatibility with local tissues. Alginate, due to its biocompatibility and easy chemical modification, is being explored for tissue engineering and drug delivery. Unfortunately, modifying alginate to a high degree of substitution consumes carboxyl group, which are necessary for ionic gelation, leading to poor hydrogel crosslinking. We introduce alginate modifications that restore the alginate’s carboxyl groups. We demonstrate that modifications that reintroduce carboxyl groups restore gelation and improve gel mechanics and tissue retention. In addition to contributing to a basic science understanding of hydrogel properties, we anticipate our approach will be useful to create tissue engineered scaffolds and drug delivery platforms.

0. Abbreviation

Alg

alginate

Azide-amine

azide-PEG4-amine

Azide-lysine

(S)-2-ammonio-6-(2-azidoacetamido)hexanoate

Cy7

Cyanine7

DBCO

dibenzocyclooctyne

DS

degree of substitution

EDC

N-Ethyl-N'-(3-dimethylaminopropyl)carbodiimide

G’

storage modulus

G”

loss modulus

i.v.

intravenous

i.m.

intramuscular

LVE

linear viscoelastic

MES

2-(N-morpholino)ethanesulfonic acid

NHS

N-hydroxysuccinimide

NMR

nuclear magnetic resonance

PBS

phosphate buffer saline

ROI

region of interest

w/v

weight per volume