Click Cross-linking Improves Retention and Targeting of Refillable Alginate Depots

Abstract

Injectable alginate hydrogels have demonstrated utility in tissue engineering and drug delivery applications due in part to their mild gelation conditions, low host responses and chemical versatility. Recently, the potential of these gels has expanded with the introduction of refillable hydrogel depots alginate gels chemically decorated with click chemistry groups to efficiently capture prodrug refills from the blood. Unfortunately, high degrees of click group substitution on alginate lead to poor viscoelastic properties and loss of ionic cross-linking. In this work, we introduce tetrabicyclononyne (tBCN) agents that covalently cross-link azide-modified alginate hydrogels for tissue engineering and drug delivery application in vivo. Adjusting cross-linker concentration allowed tuning the hydrogel mechanical properties for tissue-specific mechanical strength. The bioorthogonal and specific click reaction creates stable hydrogels with improved in vivo properties, including improved retention at injected sites. Azide-alginate hydrogels cross-linked with tBCN elicited minimal inflammation and maintained structural integrity over several months and efficiently captured therapeutics drug surrogates from the circulation. Taken together, azide-alginate hydrogels cross-linked with tBCN convey the benefits of alginate hydrogels for use in tissue engineering and drug delivery applications of refillable drug delivery depots.

Graphical Abstract

1. Introduction

Injectable hydrogels have demonstrated wide utility in clinical applications including in chemotherapeutic drug delivery[1–4], tissue regeneration after injury[5] and as scaffolds for tissue engineered constructs[6,7]. The particular advantages of hydrogels include chemical and biological stability, tunable mechanical and degradation properties and high water content, which combine to make them biocompatible and bioinstructive. This synchronization between hydrogels and surrounding tissues allows for improved cellular integration and efficient drug diffusion.

Particular attention has focused on injectable alginate hydrogels due to their neutral gelation conditions, high biocompatibility and high chemical versatility. Alginate is a natural polymer composed of units of guluronic and mannuronic acid residues. When injected in tissues, natural alginate demonstrates low immunogenicity[8–10], slow degradation rates[11] and controlled drug release[12] kinetics. Due to these qualities, alginate is generally recognized as safe by the FDA, and is currently undergoing clinical testing as a dietary supplement[13], as a material for wound dressings[14,15] and as an implant[16,17]. Alginate hydrogels have been used to deliver therapeutic levels of small molecule drugs[18,19], growth factors[20], and viral vectors[21] for sustained periods of time. In addition, the carboxylic acid groups on the alginate backbone can be readily chemically modified for conjugation of click groups[8,22–24], integrin ligands[25,26], drug conjugates[27] or proteins[28].

Strain promoted alkyne-azide cycloadditions (SPAAC) is rapidly gaining recognition as a selective and robust chemical reaction for bioconjugation[29–35]. One version of SPAAC involves a concerted [3+2] cycloaddition between azide and dibenzocyclooctyne (DBCO) groups[36–39]. This reaction efficiently proceeds in a complex biological milieu and the two reactants largely avoid reacting with the biological molecules, leading to very little off-target reactivity. Click chemistry has been used to couple materials together in vitro[40–42] and has demonstrated utility for selectively conjugating hydrogels[40–42], nanoparticles[43], and antibodies in vivo[44].

We and others have described in prior work that SPAAC chemistry can be used to noninvasively drug-eluting depots with chemotherapeutic payloads, which improve efficacy and tolerability in tumor models[8,22]. Refillable drug-eluting depots enable for locally-injected depots to be repeatedly refilled without systemic side-effects, leading to persistence of drug presentation far in excess of that possible with preloaded drug depots. Compared to preloaded drug depots, refillable drug depots eliminate burst release, extend drug release kinetics, provide for complete temporal control over drug presentation and allow for the drug or dose to be changed in response to disease progression. In the previously published system, calcium-crosslinked alginate gels decorated with azide molecules efficiently and repeatedly capture DBCO-conjugated prodrugged therapeutics from the blood. The prodrugged therapeutics prevented side effects while circulating in the body and, once captured, the prodrugs underwent slow, sustained hydrolysis, releasing drugs to the local environment[8,22,45].

Many of the potential applications of calcium cross-linked alginate hydrogels, including refillable depots, require high degrees of chemical modification. Unfortunately, because the same carboxylic groups are used in chemical modification as well as in binding divalent calcium during cross-linking, highly modified alginate hydrogels frequently lose their ability to be cross-linked by calcium. Loss of calcium cross-linking is deleterious because poorly cross-linked alginate gels can migrate away from the injection site and can be taken up and removed by phagocytic cells. Breakdown of poorly cross-linked gels can also lead to calcium leaching, which may increase foreign body response to the gel[46].

In this paper, we report improved methods to create hydrogels from highly modified alginate strands by using multi-arm cyclooctyne cross-linkers. We demonstrate that these click cross-linked gels elicited only minor inflammatory host responses. We also show that in comparison to calcium cross linked gels, click cross-linked gels are better retained in tissues, do not move away from the injection site and demonstrate improved click-mediated capture of small molecules from the circulation. Taken together, we suggest that click cross-linked gels constitute an improvement toward clinical utility of refillable drug depots.

2. Materials and Methods

2.1. Materials

Ultrapure Alginate (Pronova UP MVG-Sodium Alginate, average MW = 250 kDa, 67% guluronic acid) was purchased from NovaMatrix. Nutrition grade Alginate (Protanal LF 20/40, average MW = 300 kDa, >60% guluronic acid) was purchased from FMC. Calcium sulfate dihydrate (C3771), MES (M3671), PAMAM dendrimer gen 0 (412368), dimethyl sulfoxide (276855), and methanol (34860) were purchased from Sigma-Aldrich. Azide-PEG4-amine was purchased either from Sigma (17748) or from Lumiprobe corporation (1868). DBCO-Cy7 (1047) and DBCO-sulfo-amine (1227) were purchased from Click Chemistry Tools. 1-Ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC, 024810) was purchased from Oakwood Chemical. Sulfo-N-hydroxysuccinimide (sNHS, 00182) was purchased from Chem-Impex International. BCN-NHS (BLA 54) was purchased from Berry & Associates, Inc.

2.2. Synthesis of tBCN Molecule

See Supp. Fig. 1 for schematic. BCN-NHS ester was dissolved in 800 μL of 50:50 DMSO and methanol at a 25.5 mM concentration. 14.2 μL of triethylamine and 52.8 μL of PAMAM dendrimer was added and vortexed. The reaction was mixed for 5 minutes before analysis on a high performance liquid chromatography (HPLC) analytical system to confirm the reaction (Supp. Fig. S2). The reaction was directly purified by preparative HPLC. The collected material was evaporated on a rotary evaporator to remove methanol and then frozen and lyophilized to remove water. The compound was analyzed by high resolution mass spectrometry (HR-MS) to confirm molecular weight (Supp. Fig. S3). Proton and carbon nuclear magnetic resonance (NMR) was performed to confirm the expected molecule (Supp. Fig. S4 & S5). The mass collected was resuspended in DMSO at a 10 mM concentration. 50–200 μL aliquots were made to prevent excessive freeze-thaw cycles of the stock.

2.3. Conjugation of azide amine to alginate

See Supp. Fig. 6 for schematic. Medical grade, high guluronic acid content, high molecular weight (MW) alginate (MVG) was purchased from NovaMatrix (Sandvika, Norway). 1 g of alginate (4 μmol, 1 eq.) was dissolved overnight in 200 mL MES buffer (100 mM MES, 300 mM NaCl, pH: 6.5). Azide-PEG4-Amine (8 mmol, 2000 eq., Lumiprobe-1868) was added to the solution and the mixture was stirred for an additional 1 hour at room temperature. A mixture of 1-Ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) (8 mmol, 2000 eq.) and sulfo-N-hydroxysuccinimide (sNHS) (4 mmol, 1000 eq.) was added in three equal doses eight hours apart and the solution was stirred for an additional eight hours. The solution was dialyzed against 4 L of water with successively lower salt content, changing solution 2–3 times per day. Dialyzed solutions were frozen and lyophilized under high vacuum. For in vitro experiments and experimental coupling procedures, high guluronic acid content, high molecular weight (MW) alginate (PROTANAL LF 20/40) purchased from FMC was used and the alginate was initially dissolved in water at a 1% w/v solution (1 g in 100 mL) and then charcoal filtered (0.5g of charcoal per 1g alginate) prior to adding two equal volume of 2X MES buffer (200 mM MES, 600 mM NaCl, pH: 6.5). For double coupled material, the samples were dialyzed, frozen, and lyophilized after the first coupling and then resuspended at 0.5% w/v in 1X MES buffer solution (1 g in 200 mL) and the coupling steps were repeated as described. Adjustments and substitutions were made to increase the azide degree of substitution and are listed in Supp. Table 1.

2.4. Quantification of azides on alginate

Quantification of azide groups on alginate was performed by looking at the decrease in DBCO absorbance upon incubation of alginate-azides with an excess amount of DBCO. Alginate samples were dissolved in PBS at 0.5% w/v (5 mg into 1mL). Sulfo-DBCO-amine (700 nanomoles, Click Chemistry Tools 1227) was dissolved in 750 μL of PBS. Alginate solution (50 μL) was added to DBCO-amine solution, mixed and incubated overnight. Samples were analyzed on a Nanodrop UV/Vis spectrometer (Thermo Scientific Nanodrop 2000c; 1 mm gap) at wavelength 308 nm. Azide concentrations were determined from the decrease in absorbance at 308 of a DBCO-amine with alginate sample (extinction coefficient at 308 = 11,800 M⁻¹ cm⁻¹) compared to a DBCO solution without alginate (Supp. Fig. S7). Alginate degree of substitutions were calculated using the mass and molecular weight of azide-alginate taking into account MW change due to azide modification (Supp. Fig S7).

$$DS=\frac{mol\hspace{0.17em}azide*MW\hspace{0.17em}a\lg }{mass\hspace{0.17em}aza\lg -mol \hspace{0.17em}azide*MW\hspace{0.17em}azide}$$

where mol azide = moles of azide in solution calculated from UV/Vis; MW of alg = molecular weight of unmodified alginate; mass azalg = mass of azide-alginate in solution; MW azide =molecular weight of the azide contribution to the azide-alginate.

2.5. Gel formulation

For in vitro experiments, PROTANAL LF20/40 alginate batches were used with mechanical strengths confirmed for MVG UP alginate. In vivo experiments used MVG UP alginate and sterile solutions. Alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). 800 μL of alginate solution was pulled up in a 3 mL Luer lock syringe. 80 μL of cross-link solution was then added to a separate 1 mL luer lock syringe and the two syringes were connected while being careful to avoid introducing air to mixture. The solutions were mixed by pushing syringe barrels back and forth 10 times. For certain applications smaller volumes were mixed but with the same 10:1 ratio of 2% w/v alginate solution to cross-link solution (minimum of 440 μL).

2.6. Rheology testing of alginate gels

To determine gelation kinetics, 800 μL of 2% w/v azide-modified alginate was gelled as described in section 2.5 with 80 μL of tBCN solution in DMSO. The sample was immediately dispersed between a cone (Anton Paar measuring Cone Plate CP50–1/TG, 500 mm diameter, 1 degree angle with TruGap™) and a heated plate surface (37°C) with a 0.1 mm gap and a time sweep was run at a 1% shear strain (oscillating) at 10 rad/s angular frequency for 2 hours at 1 minute points. Mineral oil was applied surrounding the sample to prevent sample drying out. Gelation time was defined as time point storage modulus exceeds the loss modulus.

For determining final gel stiffness, gels were formulated as described in section 2.5 with tBCN in DMSO or calcium sulfate in water and immediately placed on a parallel plate (Anton Paar Measuring Parallel Plate PP25/TG, 25 mm diameter with TruGap™) at RT with a 0.5 mm gap. Mineral oil was applied around the sample to prevent sample drying out. Samples were left to gel for at least 6 hours to fully gel and then a 0.5% shear strain (oscillating) was applied with an angular frequency logarithmic ramp (0.1–100 rad/s) and 1 rad/s was measured for storage modulus.

2.7. Histology of 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. 12-week-old CD1 mice (Charles River, 022) were injected intramuscularly in the left limb with 50 μL of tBCN cross-linked gels, calcium cross-linked gels, or PBS (n=3). These mice were sacrificed four weeks after intramuscular (i.m.) injection, and the injected muscle and main organs (heart, liver, kidney, lung, and spleen) were excised, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E) by NC State College of Veterinary Medicine’s histology core to evaluate inflammatory host response to the hydrogel. The sections were sent to a certified, veterinary pathologist who was blinded to the groups for evaluations.

2.8. In vivo stability of intramuscular gel injections

Azide-modified alginate was dissolved in PBS (2% w/v, 20 mg in 1 mL). DBCO-Cy7 (3.94 μL of 4.06 mM) was added to the alginate solution and incubated overnight with stirring. Gels were prepared as previously described in section 2.5 with either tBCN (0.90 mM) in DMSO or calcium sulfate (17.47 mM) in water. After mixing, the gels were allowed to cross-link for an additional 30 minutes prior to injecting 50 μL intramuscularly in the left limb of 12 week-old CD1 mice. For the PBS control group, a PBS solution with equivalent DBCO-Cy7 (14.5 μM) was injected intramuscularly instead of a preloaded gel. Cy7 fluorescence was monitored over 2 weeks using an IVIS imager to obtain a fluorescence signal. ICG/ICG excitation and emission filters were used for all IVIS images presented with no image math in the Living Image software 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 left calf.

2.9. Targeting of intramuscular depot with DBCO-Cy7

800 μL of 2% w/v solutions of alginate (20 mg/mL) gels were made as described in section 2.5. Calcium cross-linked gels were made with 17.47 mM calcium sulfate in water. tBCN cross-linked gels were made with 0.9 mM tBCN in DMSO. 30 minutes after mixing, 50 μL of the gel or PBS control was injected intramuscularly in the left limb of 12 week-old CD1 mice. 5 g/L stock of DBCO-sulfo-Cy7 in water was prepared and diluted 100x and sterile filtered. 1 day after injection, 100 μL of the sterile 50 mg/L DBCO-Cy7 solution was injected retro-orbitally. Cy7 fluorescence was monitored over 2 weeks using an IVIS imager to obtain a fluorescence signal. ICG/ICG excitation and emission filters were used for all IVIS images presented with no image math in the Living Image software 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.

2.10. Repeated targeting of intramuscular depot with DBCO-Cy7

800 μL of 2% w/v solutions of alginate (20 mg/mL) gels were made as described in section 2.5. Calcium cross-linked gels were made with 17.47 mM calcium sulfate in water. tBCN cross-linked gels were made with 0.9 mM tBCN in DMSO. 30 minutes after gel mixing, 50 μL of gels were injected intramuscularly in the left limb of 12 week-old CD1 mice. Calcium (17.47 mM) cross-linked gels of azide-modified and unmodified alginate and PBS solutions were also performed as controls. 5 g/L stock of DBCO-sulfo-Cy7 in water was prepared and diluted 100x and sterile filtered. 1 day after i.m. injection, 100 μL of the sterile 50 mg/L DBCO-Cy7 solution was injected retro-orbitally. Cy7 fluorescence was monitored over 2 weeks using an IVIS imager. DBCO-Cy7 injections were repeated immediately following 2 week imaging for up to four refills. Capture per refill was graphed in Figure 6 and was determined by taking the fluorescence in the limb 1 week after the DBCO-Cy7 injection and subtracting the signal before the DBCO-Cy7 injection. Quantification without subtraction can be found in Supp. Fig. S8. ICG/ICG excitation and emission filters were used for all IVIS images presented with no image math in the Living Image software 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.

Figure 6: Repeated capture of circulating DBCO fluorophores by azide-alginate gels over multiple weeks.

TBCN and calcium cross-linked gels were injected intramuscularly in mice. DBCO-Cy7 was injected i.v. every two weeks and mice were imaged seven days aftereveryi.v. injection. The i.v. administration was repeated four times over eight weeks. Samples show mean ± SEM. **p < 0.01, ***p <0.001 by multiple t-tests for each refill. N=3.

2.11. Statistical methods

For experiments described in Figure 4, multiple t-tests were performed with multiple comparisons correction performed using the Holm-Sidak method, with alpha = 0.05. Computations assume that all rows are sampled from populations with the same scatter (SD). Number of t-tests = 6. For experiments described in Figure 5, statistical differences were determined using unpaired t-tests. For experiments described in Figure 6, multiple t-tests performed with multiple comparisons correction performed using the Holm-Sidak method, with alpha = 0.05. Computations assume that all rows are sampled from populations with the same scatter (SD). Number of t-tests = 4. Samples show mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p<0.0001.

Figure 4: Gels cross-linked with tBCN are better retained as compared to calcium cross-linked gels.

(A) Representative images and (B) quantitation of fluorescence signal after intramuscular injection of either free fluorescent probe,or fluorescently -labeled calcium and tBCN cross-linked azide-alginate gels monitored over 2 weeks. Red box shows region for higher magnification. Figures show mean ± SEM. Statistical significance represented as *p < 0.05 and **p < 0.01 between az-alg+tBCN and az-alg+Ca groups by multiple t-tests withHolm-Sidak correction for multiple comparisons. All samples n=3.Full images of all mice can be found in Supp. Fig. S10.

Figure 5: Cross-linking with tBCN improves on-target capture of circulating DBCO fluorophores.

Representative images (A) and quantitation (B) of fluorophore capture by intramuscularly-injected alginate hydrogels. Azide-alginate cross-linked with either tBCN or calcium were compared to control gels lacking azide groups and a PBS injection. DBCO-Cy7 was administered i.v. 24 hours following depot implantation. One week following i.v. administration, m ice were imaged to assaycapture and retention of fluorescent signal at the target site. Samples showmean SEM. Statistical significance represented as**p < 0.01,and****p < 0.0001between groups by multipleunpaired t-tests with Holm-Sidak correction for multiple comparisons.. N=6. Individual mouse images in high resolution can be found in Supp.Fig. S11.

3. Results

3.1. Synthesis of alginate strands with high azide degree of substitution

The refillable depot technology uses azide groups to capture prodrug refills. Consequently, translation of refillable drug-eluting depots relies on achieving a high number of azide groups on injectable hydrogel. Thus, we set out to maximize the conjugation efficiency of azide-amine and maximize the degree of substitution (DS) of azide motifs on alginate strands.

A number of changes to previously published conditions for carbodiimide coupling of azide-amine to LF 20/40 alginate strands[8] were tested with azide DS quantified through reaction with DBCO and UV/Vis spectroscopy[47]. The results of these efforts are summarized in Supp. Table 1. Both highly basic (pH=8.5) and highly acidic (pH=5.5) reaction conditions suppressed the coupling yields, while small changes in pH did not substantially change the reaction efficiency. Increasing the equivalents of azide-amine did improve azide conjugation, but only marginally. Although, changing the coupling agent to DMTMM from EDC has been reported by others to increase coupling efficiency[48,49], no improvement in coupling yield was observed with this reagent in our experiments. Increasing the equivalent of coupling agent (EDC) in this reaction led to wholesale cross-linking of the reaction mixture to create a gelled, unworkable solution.

Since doubling the reagent concentrations led to a cross-linked, unworkable solution, we wondered whether repeating the coupling protocol a second time could succeed where doubling reagent concentration failed. “Double coupling” is commonly used in solid-state synthesis when reactants are sterically hindered or hidden due to secondary structure formation[50,51]. A double coupling protocol was tested in which the azide-alginate material from a conjugation reaction is purified and submitted to a second round of carbodiimide coupling. We found that this protocol increased azide conjugation significantly. With double coupling, it was observed that azide DS increased from ~150 azides per strand to more than 300 azides per strand. Since in vivo work requires highly purified alginate, a subset of these reactions was tested with GMP-grade ultra-pure (UP MVG) alginate materials. In most conditions, azide coupling yielded similar DS with UP-MVG as with LF20/40. However, double coupling with 40mM azide-amine yielded lower azide DS with UP-MVG than LF20/40 alginate (DS=300 vs. 400, Supp. Table 1). To reduce variability in moving from LF 20/40 to UP-MVG materials, azide-alginate with a DS of ~200, resulting from double coupling with 10 mM azide-amine, was selected for further study.

3.2. TBCN efficiently cross-links azide-alginate gels.

In previous publications, we and others have reported the utility of azide-modified, calcium cross- linked alginate gels. However, all attempts at using calcium to cross-link alginate with a high azide degree of substitution resulted either very soft gels with poor mechanical properties or a failure to gel entirely, even with very high calcium content (110 mM). Since the same carboxylic acid groups are used both for calcium cross-linking and carbodiimide coupling, we suspect that alginate strand with high degree of substitution removes too many carboxylic groups, hindering calcium cross-linking.

Given click chemistry’s widespread use for creating robust hydrogel cross-links, we reasoned that a small fraction of click groups intended for capture of drug refills could be adopted for cross-linking of the alginate hydrogel. In addition, it was hoped that removing calcium as a cross-linker would avoid any potential host response associated with calcium[24,46], calcium leaching[52] and inconsistent mechanical properties[53]. A tetrameric cross-linker bearing four cyclooctyne groups was synthesized and tested for its ability to cross-link azide-modified alginate. Bicyclononyne[54] (BCN) was chosen due to its robust click chemistry kinetics, small size and hydrophilicity. The four-armed, generation-0 PAMAM dendrimer was reacted with bicyclononyne N-hydroxysuccinimde to yield a tetramer of BCN (termed tBCN) in a single step and purified by preparative HPLC (Supp. Fig. S2).

TBCN cross-linking of alginate strands with a high azide degree of substitution was tested through a rheological time sweep. When mixed with 0.90 mM tBCN, a 2% solution of azide-alginate quickly formed hydrogel (Fig. 2A), which continued to strengthen over the several hour. As shown in Fig. 2B, varying the concentrations of tBCN allowed tuning the gelation time. In addition to 0.90 mM, 0.45 mM and 0.22 mM concentrations of tBCN efficiently cross-linked a 2% w/v solution of alginate, with gelation times varying from 5 minutes (0.9 mM tBCN) to 30 minutes (0.22 mM). Lower tBCN concentration (0.11 mM) did not produce gels. If fully reacted under these conditions, the 0.90 mM and 0.45 mM tBCN conditions correspond to the utilization of 50 (25%) and 25 (12.5%) azide molecules per strand of alginate, respectively. Thus, in either case, more than 150 azide molecules per strand remain for capturing prodrugs.

Figure 2: TBCN efficiently crosslinks azide-alginate.

(A) One-hour time sweep of tBCN cross-linking of a 2% solution of azide-alginate. (B) Gelation times for tBCN cross-linking of azide-alginate. Gelation time was defined as time at which storage modulus exceededloss modulus. (C) Storage moduli of alginate gels formed with calcium (4.4mM) or tBCN(0.9 and 0.45mM)at a 1 rad/s frequency sweep. Graph shows individual values, mean (n=3) and standard deviation.

The stiffness of tBCN cross-linked azide-alginate gels was measured and compared to those of calcium cross-linked gels (Fig. 2C). Unmodified alginate (2% w/v) cross-linked with calcium sulfate (4.4 mM) showed a storage modulus of approximately 250 Pa. Cross-linking azide-alginate with 0.9 mM tBCN led to gels with similar stiffnesses, while attempts to cross-link azide-alginate with calcium failed. Due to their favorable gelation kinetics and mechanical properties, 2% azide-alginate gels cross-linked with 0.9 mM tBCN were selected for further study.

3.3. TBCN cross-linked azide-alginate gels elicit mild host responses and are retained at injection site

Biocompatibility, lack of toxicity and lack of immunogenicity of tissue-injected gels is crucial in almost all medical applications. Thus, we interrogated the safety and biocompatibility of tBCN cross-linked azide-alginate gels over multiple weeks. Azide-alginate gels (2% w/v) cross-linked with tBCN (0.9 mM) or calcium (17.5 mM) were injected intramuscularly into 12-week old, outbred CD1 mice. No ulceration was observed in any mice over the time period and there were no noted health issues. Four weeks after intramuscular injection, the injected muscles and main organs heart, lung, spleen, kidney and liver were removed, fixed in paraformaldehyde, sectioned and stained with hematoxylin and eosin (Fig. 3). These sections were evaluated and scored by a veterinary histopathologist blinded to sample identity (Supp. Table 2).

Figure 3: Biocompatibility of tBCN cross-linked azide-alginate gels.

Representative images of H&E stained sections from injection site and five major organs four weeks after intramuscular injection ofcalcium andtBCN cross-linked azide-alginate hydrogels and PBS-injectedcontrols. Scale bar = 400 μm.

No differences were observed in the central organs of any of the mice analyzed, suggesting that tBCN cross-linked azide-alginate, whether calcium or tBCN cross-linked, had minimal systemic toxicity after i.m. injection. Analysis of the injection site showed a small fibrous capsule(1–2 cells thick) formed around the tBCN cross-linked azide-alginate gels, but no other differences as compared to the PBS group. Interestingly, azide-alginate cross-linked with calcium could not be observed in the muscle sections, further confirming the instability of gels formed by calcium cross-linking.

Intramuscularly injected gels must be retained at the tissue sites to serve active function. Thus, we set out to determine the persistence and retention of tBCN and calcium cross-linked azide-alginate gels at intramuscular sites. Azide-alginate gels (2% w/v, 50 μL) cross-linked with tBCN (0.9 mM) or calcium (17.5 mM) were covalently labeled with the Cy7 fluorophore and injected intramuscularly into outbred CD1 mice. Gel retention and stability were monitored by in vivo fluorescence imaging on an IVIS imager. As shown in Figure 4, when injected intramuscularly, signal from the unconjugated fluorophore quickly disappeared from the muscle (Fl in PBS) with minimal residual fluorescent signal remaining after 24 hours. Fluorescence signal from the calcium cross-linked azide-alginate gels (Fl-az-alg +Ca²⁺) showed significant decrease at the target site within 24 hours. At the same time, the gel signal appeared to migrate down the leg to the knee and ankle area (see enlarged image). In sharp contrast, gels formed through tBCN cross-linking of azide-alginate (Fl-az-alg +tBCN) were stable and well retained at the injected muscle over the two weeks, with no change in fluorescence signal in the limb, suggesting minimal migration from the targeted injection site.

With the observation that calcium-crosslinked gel migrates within the leg, we assessed the migration of fluorescence to the non-injected, opposite limb. Significantly more accumulation of fluorescence was observed in the non-injected limb for mice injected with calcium cross-linked gels as compared to tBCN cross-linked gels or unconjugated fluorophore (Supp. Fig S9).

3.4. TBCN gels efficiently capture blood-circulating fluorophores

A central attribute of refillable, drug-eluting depots is the noninvasive capture of small molecules from systemic circulation. To assess whether tBCN cross-linked gels retain this function, gels were assayed for their ability to capture fluorescent drug surrogates. Azide-alginate gels (2% w/v, 50 μL) cross-linked with tBCN (0.9 mM) or calcium (17.5 mM) were injected intramuscularly into the left hindlimb of outbred CD1 mice. I.m. injections of PBS and calcium cross-linked unmodified alginate gels were used as negative controls. One day after i.m. gel injection, DBCO-Cy7 (100 μL, 50 mg/L ) was injected retro-orbitally. The capture of intravenously (i.v.)-administered DBCO-Cy7 was followed over one week through live animal fluorescence imaging on the IVIS imager. As shown in Figure 5, neither PBS nor the unmodified alginate gels captured circulating fluorophores from the blood. TBCN cross linked azide-alginate gels (az-alg+tBCN) efficiently captured circulating material. In comparison, calcium-crosslinked gels captured significantly less material.

Taken together, our data demonstrate that click cross-linking of alginate strands with high azide degree of substitution produces more mechanically stable gels which are better retained at target sites and demonstrate less migration within the limb or to the non-injected limb.

3.5. Systemic capture by click cross-linked gels is maintained over multiple weeks and refill events.

It was next tested whether the superior retention and refilling capabilities of tBCN cross-linked gels as compared to calcium cross-linked gels could be maintained over multiple refills and over many weeks. Azide-alginate gels (2% w/v, 50 μL) cross-linked with tBCN (0.9 mM) or calcium (17.5 mM) were injected intramuscularly into the left hindlimb of outbred CD1 mice. PBS injections and calcium cross-linked unmodified alginate gels were used as negative controls. DBCO-Cy7 (100 μL, 50 mg/L) was injected retro-orbitally and the capture of i.v.-administered DBCO-Cy7 was followed over one week through live animal fluorescence imaging on the IVIS imager (Fig. 6). This capture test was repeated every two weeks for a total of four refill events. Both gels efficiently captured circulating fluorophores over the eight week experiment. Although tBCN cross-linked gels outperformed calcium cross-linked gels after the first refill, this improvement was not statistically significant for subsequent captures. Taken together, our data demonstrate that a high density of active azide groups can be achieved without sacrificing tissue retention or regioselective control over tissue capture. The stability and repeatability of capture as well as the higher density of azides at the target site bodes well for future application of these gels in medical applications.

4. Discussion

This paper reports the synthesis, mechanical properties and in vivo function of click cross-linked alginate gels modified with azide groups to a high degree of substitution. We report that calcium fails to cross-link alginate strands with high azide DS. To alleviate this problem, four-arm BCN cross-linkers (tBCN) were synthesized and shown to efficiently cross-link azide-modified alginate strands to form stable, mechanically robust hydrogels. We demonstrated that tBCN cross-linked azide-alginate gels demonstrated mild to no host response when injected intramuscularly in mice. In addition, tBCN cross linked gels were better retained and more stable at intramuscular sites as compared to calcium cross linked azide-alginate gels. Finally, the tBCN cross-linked gels improved capture of systemically administered small molecule drug surrogates at intramuscular sites and this in vivo targeting was sustained over multiple weeks and multiple rounds of systemic capture. Taken together, click cross linking of azide-modified alginate represents a significant improvement in hydrogel and refillable depot stability for applications at intramuscular sites.

The number of azides conjugated to refillable depots limit the number and dose of refills. Because of this limitation, we set out to increase the density of azides on alginate strands. Previously published protocols for conjugation of azide groups onto alginate strands used EDC/NHS coupling and a 2,000-fold excess of both the azide and the carbodiimide reagents[8,24]. Some protocols increased EDC equivalents even farther, up to 4,000 equivalents[55,56], but necessitated potentially destructive sodium hydroxide incubations because of inter strand cross-linking at these high EDC concentrations. Despite such high reagent equivalents, these protocols still yielded DS of 50–100 azides per alginate. To overcome the poor coupling yields that result even with excess reagents and reactants, we tested a panel of alternative conditions, including alternative pH, reactant concentrations, and coupling reagents, but none of these changes yielded significant improvement to azide DS or conditions led to undesired cross-linking of alginate in the reaction. However, performing “double coupling” of the alginate supplied significantly higher azide DS, including as high as 400 azides per strand.

Successful increases in the azide DS on alginate led to a new challenge loss of calcium-mediated cross-linking. In agreement with previous studies[55], high azide DS alginates presented marked decreases in viscosity and almost completely lost the ability to gel in the presence of calcium. The storage modulus of calcium cross-linked hydrogels made from azide-alginate was 2,000-fold lower than calcium cross-linked gels made from unmodified alginate. In addition, calcium cross-linked high-azide gels showed significant loss and unexplained migration within days of injection in vivo. The poor cross linking and poor site retention are of particular concern because they translate to significant loss of target sites as well as capture of drugs at undesired, off-target locations. Although calcium cross-linked high azide alginates still captured circulating cyclooctynes, they raise concern of higher off-target accumulation at undesired sites such as the non-targeted limb.

A four-arm bicyclononyne (tBCN) cross-linker structured on the G-0 PAMAM efficiently cross-linked azide-alginates to form stable, mechanically robust hydrogels. Although PAMAM molecules have been reported to elicit in vitro and in vivo toxicity[57], this toxicity is likely due to the high positive charge of the G3-G6 compounds[58]. The small, end-capped G0 cross-linker used in this study likely does not share this toxicity, as demonstrated by minimal toxicity after four weeks in vivo. Rheology of cross-linked gels showed that tBCN cross-linking could restore gel strength lost to calcium cross-linking upon high azide substitution, while using only a fraction (<25%) of the azides. The use of excess target groups simultaneously for cross-linking and systemic capture simplifies gel formulation and provides tunability. A similar approach was recently reported by Webber and colleagues for systemic cross-linking and capture through host-guest chemistry. The covalent bonds formed through our cross-link strategy is likely more longer lasting, with in vivo capture still efficient at day 45, at which point host-guest cross linked gels were completely eliminated (Supp. Fig S8) [59]. Future studies will focus on increasing azide degree of substitution even higher, further decreasing the fraction of azides needed for cross-linking.

TBCN cross-linked azide-alginate gels demonstrated low host immune responses when injected intramuscularly, with little foreign body response and minimal inflammation. The controlled kinetics of BCN-azide conjugation (0.1–0.3 M⁻¹s⁻¹)[54] provides an opportunity for in situ gelation, decreasing injection pressures and potential tissue damage to the surrounding area during injection. Intramuscular injections were chosen due to their proven injectability, vascularity and the established use of injectable alginate gels in intramuscular preclinical applications[60–63]. Specifically, in the intramuscular space, short-range gel migration before completion of cross-link might allow for increased surface area and more exposure to blood vessels as well as more interaction to the targeted tissue for therapeutic delivery. In the case of tBCN-crosslinked gels, the gels are only partially gelled at the time of injection. It is not clear whether the gels at this time point are fully self-healing[64] or demonstrate some irreversible fragmentation and future studies will focus on this important point. As is typical with many biomaterials, a small fibrotic capsule was formed at the periphery of the hydrogel. It is worth noting that in reported side-by-side comparisons, the fibrotic responses to click-crosslinked gels are significantly lower than to calcium cross-linked alginate gels[7,8,65]. Although we were not able to identify any literature for the immunogenicity of azide-conjugated or tBCN-conjugated polymers, additional research is needed into this important point. Nevertheless, click-mediated cross-linking of alginate may assuage some worries related to calcium-related toxicities that have been reported for calcium cross-linked alginate gels[46] as long-term depots. As demonstrated in our capture experiment, the fibrotic capsule does not isolate the hydrogel from capturing circulating molecules from the surrounding vasculature an essential feature to the refillable depot platform.

The fact that tBCN cross-linked azide-alginate gels were better retained and more stable at intramuscular sites as compared to calcium cross-linked azide-alginate gels is significant for their use as a drug delivery platform. The fluorescence signal from calcium cross-linked gels dropped significantly over the first week and migrated to the animal’s knee and ankle as well as other parts of the body. Although we do not propose a mechanism for the clear migration of the gel from the thigh to the ankle and knee, the migration of this material throughout the animal is consistent both with breakdown of the hydrogel and potential phagocytosis by phagocytic macrophages[46,66,67]. In vivo targeting of azide-modified hydrogels was sustained over multiple weeks and multiple rounds of systemic targeting. While only the non-injected limb was measured for off-target accumulation in this study, it was particularly concerning that circulating molecules appeared to accumulate at the non-target limb to a much higher degree in calcium cross-linked azide gels than tBCN gels, although other tissues, including key organs, were not assessed. The migration of azide-modified alginate throughout the mouse would lead to delivery of drug to vital organs and increased off-target toxicity.

We observed a substantial difference in terms of capture efficiency between the first refill and subsequent refills, something we have observed previously[46], but not at other times[22]. One possibility is that fibrous capsule formation limits accessibility to some extent, though not fully. In this case, better methods to prevent fibrosis will improve refilling. Another possibility is that a small amount of tissue damage from the injection elicits angiogenesis or inflammation, which increases circulating molecule accessibility to the injected gels. In this case, increasing vascularization around the gel may improve refilling. A final possibility is that after injection certain azides are more accessible than others, and these azides are used up in the first refill, leaving less accessible azides for subsequent injections. In this case, increasing azide degree of substitution further may improve refilling.

The present manuscript raises a number of important questions for future studies. First, although only roughly 25% of azides are used by the tBCN crosslinker, it is not known whether other azides are rendered inaccessible due to the spatial arrangement in the cross-linked gel. Second, although DBCO Cy7 was conjugated to the gels for in vivo stability studies, we did not evaluate whether this conjugation changes the mechanical properties of the click-crosslinked gels in vitro. Thus, the in vivo gel retention of Cy7-conjugated click gels may not be representative of the behavior of click gels without DBCO conjugation Third, the present study did not incorporate or study drug release from tBCN cross-linked gels. Although we presume that the same release mechanisms used in previous studies[8,33] can be directly applied to tBCN cross-linked gels without modification, this will need to be studied further. Fourth, the current study did not address the drug distribution within the gel or gel distribution within the injected tissue, which will be the subject of future studies. Finally, capture efficiency, especially after multiple rounds of capture, will need further study. Previous research suggests that roughly 4–10% of the systemically administered dose is captured at locally-injected gels[8,22,45,59]. The efficiency of capture appears to be governed more by the pharmacokinetics and local availability of small molecule as opposed to the number of azide groups or kinetics of click reaction. However, much additional work remains to be done to understand the factors influencing capture rates.

5. Conclusions

Taken together, our system represents an improvement in hydrogel and refillable depot stability for applications in the body. Refillable drug depots made possible by capturing drug refills from circulation by azide groups constitute a less-invasive way to deliver drugs locally for repeatable dosing. Click mediated cross-linking overcomes the complications discovered with calcium cross-linking of azide-conjugated gels and improves their biocompatibility properties. Future studies could investigate yet higher azide substitutions and increasingly potent disease responses.

Figure 1: Using click chemistry to cross link refillable depots.

Left to Middle: Azide-alginate strands are combined and injected into target tissues and cross-link in situ. Middle to Right: Intravascular administration of cyclooctyne-conjugated therapeutics allows selective capture and display of drug at gel site.

Statement of Significance.

Ionically cross-linked, injectable alginate biomaterials hold promise in many different clinical settings. However, adding new chemical functionality to alginate can disrupt their ionic cross-linking, limiting their utility. We have developed a “click” cross-linking strategy to improve the mechanical properties and tissue function of modified alginate biomaterials and enable them to capture small molecule drugs from the blood. We show that click cross-linked materials remain in place better than ionically cross-linked materials and efficiently capture payloads from the blood. Development of click cross-linking for refillable depots represents a crucial step toward clinical application of this promising drug delivery platform.

Abbreviations:

Az-Alg+tBCN

azide-modified alginate cross-linked with tBCN

Az-Alg+Ca²⁺

azide-modified alginate cross-linked with calcium sulfate

Alg

alginate

Az

azide

BCN

bicyclononyne

COOH

carboxyl group

Ctrl-alg+Ca²⁺

alginate cross-linked with calcium sulfate

Cy7

Cyanine7

DBCO

dibenzocyclooctyne

DMSO

dimethyl sulfoxide

DMTMM

(4-(4,6-dimethoxy-1,3,5-triazin-2-yl)- 4-methyl-morpholinium chloride)

DS

degree of substitution

EDC

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

Fl-az-alg

Cyanine7 fluorescence labeled azide-modified alginate

H&E

hematoxylin and eosin

HPLC

high performance liquid chromatography

HR-MS

high resolution mass spectrometry

i.v.

intravenous

i.m.

intramuscular

LCMS

liquid chromatography–mass spectrometry

MES

2-(N-morpholino)ethanesulfonic acid

NHS

N- hydroxysuccinimide

NMR

nuclear magnetic resonance

PAMAM

polyamidoamine

PBS

phosphate buffer saline

RO

retroorbital

ROI

region of interest

sNHS

sulfo-N- hydroxysuccinimide

SPAAC

strain-promoted alkyne-azide cycloadditions

tBCN

tetrabicyclononyne

UP

UltraPure Alginate

w/v

weight per volume