Dexime: A Selectively Enzyme‐Degradable Hydrogel for Protein Therapeutic Release Authors: Quinton E. A. Sirianni, Nitzan Letko Khait, Adam Forman, Edward A. Armstrong, Karim Fouad, Molly S. Shoichet Published: 2026 ABSTRACT ABSTRACTHydrogels are of great interest for the delivery of protein therapeutics as they obviate the denaturing shear forces and/or organic solvents common to other delivery systems. Dextrin, a polysaccharide derived from starch, has been largely ignored in the design of hydrogels, yet is biodegradable and can be modified to allow in situ gelation after injection by click‐crosslinking. Herein, dextrin is modified with either ketone or aldehyde moieties and then click‐crosslinked with poly(ethylene glycol)‐tetraoxyamine to form dextrin‐oxime (dexime) hydrogels. The rheological and mechanical properties are tunable by changing formulation parameters such as total dextrin weight percentage (wt.%) and pH. The hydrogels are injectable through a 30G needle, selectively degradable via α‐amylase, yet stable and minimally swelling in the presence of other enzymes, including hyaluronidase and chondroitinase ABC (ChASE). The presence of low amounts of α‐amylase in many fluids and tissues of the body allow for the slow degradation of dexime with minimal swelling. Dexime is particularly useful for local delivery of ChASE: an enzyme that degrades the aggregate scar formed after injury in the central nervous system. Dexime is cytocompatible and releases active ChASE both in vitro and in vivo as evidenced by chondroitin sulphate proteoglycan degradation in the spinal cord. Introduction Hydrogels have been widely investigated for 3D cell culture [1], bioprinting and tissue engineering [2], and controlled therapeutic delivery [3]. For the latter, hydrogels can be shear thinning and/or gel in situ [4, 5], allowing for targeted and minimally invasive injection at the site of interest. Local protein delivery, by injection of a biocompatible hydrogel to the central nervous system, is particularly useful as the blood‐brain‐barrier limits the diffusion of therapeutics therein [6, 7]. Dextrin, a short homopolymer composed of D‐glucose monomers primarily linked via α(1→4) glycosidic linkages, is produced via the incomplete hydrolysis of starch or glycogen and can be degraded in vivo by amylases [8]. Despite its biodegradability and commercial availability, there are only a few dextrin hydrogels reported in the literature. For example, dextrin has been crosslinked by free radical graft polymerization using (meth)acrylic acid derivatives [9, 10, 11, 12, 13], poly(lactic acid) [14], or vinyl‐modified dextrin [15, 16]; however, in situ free radical polymerizations are difficult to control and the free radicals themselves can be cytotoxic [17, 18]. Alternatively, the glucose rings of dextrin have been oxidized with sodium periodate to produce aldehydes, which have been crosslinked with adipic acid dihydrazide [19] or used to crosslink other polysaccharides, such as chitosan [20]. While effective for crosslinking, these hydrogels have been shown to degrade via hydrolysis of the crosslinks, which can lead to undesirable swelling and uncontrolled degradation. This motivated us to design a cytocompatible, hydrolytically stable, and specifically α‐amylase‐degradable in situ gelling crosslinked dextrin hydrogel. Although α‐amylase is primarily produced by salivary glands and pancreas, it is also found in low amounts in the blood and urine [21], and many tissues [22, 23, 24]. Thus, hydrogel degradation by α‐amylase activity is expected to occur at a slow rate in vivo in most tissues. This design avoids the issues of more rapid hydrolysis degradation, such as hydrogel swelling [25], while providing an enzyme‐degradation pathway that ensures a bioresorbable material. We report the synthesis of oxime‐crosslinked dextrin hydrogels, which we call dexime. The oxime chemistry results in stable, crosslinked hydrogels, as shown with other polymers including hyaluronic acid [26, 27, 28], poly(ethylene glycol) [29, 30, 31, 32], and polyacrylamides [33]. In oxime click‐crosslinking, oxyamine and aldehyde/ketone react with each other in a bioorthogonal manner [34]. The reaction rate of oxime formation is tunable by adjusting parameters such as pH, where acid catalyzes the reaction, and the nature of the carbonyl, where aldehydes react faster than ketones. Importantly, and unlike the similar and more commonly used imine/hydrazone click‐crosslinking chemistry, which uses amine/hydrazine nucleophiles instead of oxyamines, oxime crosslinks are much more hydrolytically stable [35]. Thus, oxime crosslinking limits rapid hydrogel degradation via hydrolysis and the consequential swelling that typically occurs. Dextrin was modified via a two‐step synthetic procedure to introduce either ketone or aldehyde moieties, which were each crosslinked with poly(ethylene glycol)‐tetraoxyamine (PEGOA4). Dexime hydrogels were thoroughly characterized by rheology, mechanical testing, injection force, enzymatic degradation, swelling, cytocompatibility, and utility as a vehicle for protein release in vivo. We were particularly interested in using dexime to deliver proteins of therapeutic interest that are often difficult to release — i.e., enzymes like hyaluronidase (HAase) and chondroitinase ABC (ChASE), which degrade glycosaminoglycans (GAGs) in the extracellular matrix (ECM) and bioresorbable hydrogels common in biomedical engineering such as hyaluronan, heparin, and chondroitin sulfate [36]. Methyl cellulose and agarose hydrogels have been studied [37, 38, 39] for ChASE delivery, yet these materials are limited by their inherent stability and slow bioresorption, which may lead to long‐term issues including inflammation and fibrosis [40]. HAase degrades hyaluronic acid in the ECM and has been used in numerous medical applications including cancer [41], primary immunodeficiency [41], and cosmetic surgery [42, 43]. ChASE degrades chondroitin sulfate of chondroitin sulfate proteoglycans (CSPGs). It is used to treat spinal disc herniation clinically [44, 45, 46] and has been shown to degrade the aggregate scar tissue that forms after traumatic spinal cord injury (SCI) and stroke experimentally [47, 48, 49, 50, 51, 52, 53]. ChASE requires local delivery for a defined duration to be efficacious at the site of injury, making controlled local delivery a compelling strategy [54]. ChASE is thermally unstable at body temperature [55] and has a short half‐life (0.7 d), which was overcome with ChASE37, having 37 point mutations and a half‐life of 4.4 d [56]. Importantly, local, sustained delivery of ChASE37 showed tissue benefit in the stroke‐injured rat brain when delivered from a crosslinked methyl cellulose hydrogel [50]. Notwithstanding these important results, the crosslinked methyl cellulose is, at best, very slowly bioresorbable, making injection into tissue undesirable. We describe the chemical and physical characterization of our novel dexime hydrogel, its degradation by α‐amylase, release of active ChASE37 over 7 d in vitro, and its efficacy in vivo with the degradation of chondroitin sulfate proteoglycans (CSPGs) of the perineuronal nets (PNNs) in the rat spinal cord (Figure 1). Study design: (a) Dexime, a novel and versatile hydrogel, was synthesized from dextrin‐ketone or dextrin‐aldehyde crosslinked with poly(ethylene glycol)‐tetraoxyamine (PEGOA4) to yield an oxime‐crosslinked hydrogel. The dexime hydrogel can (b) be selectively degraded by α‐amylase, (c) release bioactive ChASE37 in vitro, and (d) degrade CSPG in vivo, in the rodent spinal cord. Results and Discussion Synthesis of Modified Dextrin Dextrin (Mn = 2.3 kDa, determined by mass spectrometry; Figure S1) was modified by a two‐step synthesis, first to dextrin‐carboxylate (Dxt‐COONa) and then to either dextrin‐ketone or dextrin‐aldehyde. Specifically, dextrin‐hydroxyl groups reacted with a molar excess of bromoacetic acid via a Williamson ether synthesis, resulting in carboxymethyl groups (Scheme 1), building on a previously reported modification of methyl cellulose hydroxyl groups [37, 57]. Synthetic pathway to produce carboxylate‐modified dextrin via a Williamson ether synthesis, with carbons numbered for Dxt‐COONa. The modified polymer has carboxylated glucose repeat units statistically (randomly) distributed among the polymer chain (denoted with “stat”). While all free alcohols could be substituted with sodium carboxymethyl groups, substitution along the polymer chains was found to be primarily on C2′ when investigated with NMR spectroscopy (as described in the Supporting Information). The resulting Dxt‐COONa was a white, cottony solid that was soluble in water at room temperature, unlike unmodified dextrin, which only dissolves completely in water under alkaline conditions at room temperature. Dextrin and Dxt‐COONa were characterized by 1H NMR spectroscopy (Figure 2a; Figure S2, respectively): modification is apparent as the anomeric proton peak of the dextrin glucose repeat unit (∼5.4 ppm) splits from one peak into two and broadens. The carboxylate groups on the repeat units withdraw electrons from the anomeric proton and cause a downfield shift. Additionally, the other dextrin proton peaks broaden to yield a large multiplet from ∼3.4–4.4 ppm in Dxt‐COONa. To quantify the degree of carboxylate substitution, Dxt‐COONa was acidified with DCl, and the 1H NMR spectrum of Dxt‐COOD (Figure 2b) showed that 30%–34% of polymer repeat units were modified with carboxyl groups, determined by integrating the carboxy methylene peaks at ∼4.3 ppm against the anomeric proton peaks at ∼5.3–5.7 ppm. Further NMR characterization showed the regioselectivity of the carboxyl group of Dxt‐COONa to be primarily located on C2′ and the reducing end C1′ of the modified polymer (characterization details are provided in the Supporting Information and additional NMR spectra of dextrin and Dxt‐COOD, including 1H‐13C{1H} heteronuclear single quantum coherence (HSQC, Figure S3 for dextrin, Figure S4 for Dxt‐COOD), 1H total correlation spectroscopy (TOCSY, Figure S5 for Dxt‐COOD), and 1H nuclear Overhauser effect spectroscopy (NOESY, Figure S6 for Dxt‐COOD). 1H NMR spectra (400 MHz; D2O) of (a) dextrin, (b) Dxt‐COOD, (c) Dxt‐Ket, and (d) Dxt‐Ald. Changes of note in the 1H NMR spectra during the synthesis of the intermediate Dxt‐COOD include a splitting of the anomeric proton peak (red square) and the presence of a new peak representing the methylene protons of the carboxyl moieties (green box) in the presence of acid. Integrating both of these peaks allowed carboxyl substitution to be determined. Further modification to Dxt‐Ket produces another new peak representing the methyl protons of the ketone moieties (blue oval). Integration of this peak against the anomeric proton peaks allowed ketone substitution to be calculated. In contrast, further modification to Dxt‐Ald produces four new peaks representing the methylene protons on the α and β carbons next to the aldehyde moieties (tan ovals). Integration of these peaks against the anomeric proton peaks allowed aldehyde substitution to be determined. FTIR spectra of (e) dextrin, (f) Dxt‐COONa, (g) Dxt‐Ket, and (h) Dxt‐Ald. Changes of note in the FTIR spectra are in the 1500–1800 cm−1 region. The conversion of dextrin to Dxt‐COONa produces a carboxylate C═O stretch at 1590 cm−1. Further conversion to either ketone or aldehyde produces multiple peaks in this region resulting from ketone/aldehyde/carboxylic acid C═O stretching (at 1730 cm−1), amide C═O stretching (at 1640–1660 cm−1), and amide N─H bending (at 1560–1580 cm−1). Initially, regioselective substitution of the carboxymethyl group on C6′ was expected as this is the only primary hydroxyl group of the glucose ring, which should be more reactive in an SN2‐type reaction than the secondary hydroxyls on C2′ and C3′. Previously, vinyl acrylate substitution onto dextrin resulted in a regioselective reaction primarily at C2′ [15]. We postulate the regioselectivity is due to the alkaline conditions in which the reaction was conducted. The C2 hydroxyl group of aldohexose carbohydrates has been reported to have higher acidity than C3–C6 hydroxyls due to its proximity to the electron withdrawing anomeric hemiacetal, which translates to higher reactivity under alkaline conditions [58]. By the same reasoning, the C1 hydroxyl of the reducing end of the dextrin polymer chains should also be highly acidic and thus allow for higher reactivity under alkaline conditions. Dextrin‐ketone (Dxt‐Ket) was synthesized by the amide coupling of a ketal to Dxt‐COONa, followed by deprotection (Scheme 2a). Dxt‐COONa was activated with 4‐(4,6‐dimethoxy‐1,3,5‐triazin‐2‐yl)‐4‐methyl‐morpholinium chloride (DMTMM), after which 2,2‐dimethoxy‐1‐propanamine (1) was added to yield dextrin‐ketal, which was then converted to dextrin‐ketone under acidic conditions. The resulting dextrin‐ketone polymer was a white solid that dissolved freely in water and was characterized by both 1H NMR and FTIR spectroscopy. The 1H NMR spectra of Dxt‐Ket (Figure 2c) has a new broad singlet peak at ∼2.3 ppm corresponding to the methyl protons of the ketone moieties and by integrating with the anomeric proton peaks at ∼5.3–5.8 ppm, we determined that 17%–20% of dextrin polymer repeat units of Dxt‐Ket had a ketone group. Synthesis of (a) Dxt‐Ket and (b) Dxt‐Ald from Dxt‐COONa via an amide coupling reaction, followed by acidic deprotection of the ketal/acetal. Carboxyl and carbonyl modified glucose repeat units are statistically (randomly) distributed among the polymer chain (denoted with “stat”). Dextrin‐aldehyde (Dxt‐Ald) was synthesized similarly to Dxt‐Ket. The acetal amine 4,4‐dimethoxy‐1‐butanamine (2) was coupled to DMTMM‐activated Dxt‐COONa to provide dextrin‐acetal, which was subsequently deprotected with acid to yield Dxt‐Ald (Scheme 2b). Dxt‐Ald was a white solid that dissolved in water with heating and was characterized by both 1H NMR and FTIR spectroscopy. The 1H NMR spectrum of Dxt‐Ald (Figure 2d) possesses three new peaks upfield of 3 ppm in comparison to its precursor. These new peaks correspond to the protons of the α and β carbons next to the aldehyde moieties. While the β protons manifest as one peak at ∼2.0 ppm, the α protons are shown as three separate peaks at ∼2.6, ∼1.9, and ∼1.3 ppm. This is due to the neighboring aldehyde group, which can assume one of three forms in solution (free aldehyde, hydrate, or hemiacetal) [59, 60]. Aldehyde substitution was consistently 14% based on integrating these peaks against the anomeric proton peaks at ∼5.3–5.8 ppm. Fourier Transform Infrared (FTIR) spectroscopy was additionally used to characterize the synthesis of the dextrin derivatives and confirm successful syntheses. Comparing the spectra of dextrin (Figure 2e) and Dxt‐COONa (Figure 2f), the latter possesses a new peak at 1590 cm−1, corresponding to the C═O stretch of the sodium carboxylate. The FTIR spectrum of Dxt‐Ket (Figure 2g) has three small peaks at: 1730 cm−1 ascribed to C═O stretch of ketone/carboxylic acid groups; 1660 cm−1 to the C═O stretch of amide groups produced during the coupling reaction; and 1560 cm−1, which arises from the N─H bend of the amide moieties. The FTIR spectrum of Dxt‐Ald (Figure 2h) reveals three small new peaks at: 1730 cm−1 arising from the aldehyde/carboxylic acid C═O stretch; 1640 cm−1 from the C═O stretch of the amides formed during coupling of the acetals; and 1580 cm−1 from the N─H bend of the amides. Dexime Hydrogel Formulation and Characterization To produce dexime hydrogel, either Dxt‐Ket or Dxt‐Ald was reacted with PEGOA4 under mild acidic conditions to yield a crosslinked hydrogel (Figure 3a). PEGOA4 was synthesized based on a previously described procedure [26, 27] and characterized by 1H NMR spectroscopy (Figure S7). A simple inversion test showed that a gel formed within 60 min when incubated at 37°C (Figure 3b). Evaluation of the rheological and mechanical properties of dexime formulated from modified dextrin and PEGOA4. (a) Dexime was synthesized by reacting Dxt‐Ket or Dxt‐Ald (red) and PEGOA4 (blue) in a mildly acidic solution to produce an oxime‐crosslinked hydrogel. (b) The liquid‐like dexime gels within an hour at 37°C after mixing Dxt‐Ket and PEGOA4, as visualized with the addition of bromophenol blue. Rheological graphs of the reaction of PEGOA4 with (c) 2.25 wt.% Dxt‐Ket and (d) 2.50 wt.% Dxt‐Ket formulations at 37°C and pH 5.8 (n = 3, mean + standard deviation plotted). (e) Compression testing of the Dxt‐Ket crosslinked dexime formulations. Significance was determined using Welch's unpaired t‐test (n = 4, mean ± standard deviation plotted, *** p < 0.001). Rheological graphs of the reaction of PEGOA4 with 2.50 wt.% Dxt‐Ald at 37°C at either (f) pH 5.8 or (g) pH 6.8 (n = 3, mean + standard deviation plotted). (h) Compression testing of the Dxt‐Ald crosslinked dexime formulations. Significance was determined using an unpaired t‐test (n = 4, mean ± standard deviation plotted, ** p < 0.01). Injection forces of the most promising formulation of crosslinked (i) Dxt‐Ket (2.50 wt.%, pH 5.8) and (j) Dxt‐Ald (2.50 wt.%, pH 6.8) were measured at room temperature to assess the ease of injection through a 30G needle (n = 3, mean ± standard deviation plotted). Both gels had injection forces under the maximum force limit of 30 N (dashed line) for up to 5 h. Rheometry was used to further validate the gelation time and viscoelastic behavior of different dexime formulations. 2.25 wt.% Dxt‐Ket (22.5 mg mL−1, 20.1 mм ketone) was combined with PEGOA4 (18.1 mg mL−1, 14.1 mм oxyamine) to achieve up to 70 mol% oxime bonds. The 2.25 wt.% formulation was kept at pH 5.8 and 37°C over 3 h while monitoring its rheological properties. Under these conditions, the formulation showed a gelation time, defined as the crossover of the storage (G′) and loss moduli (G′′) of the hydrogel, at 63 ± 10 min, with a final G′ of 200 ± 120 Pa (Figure 3c). A more concentrated formulation of 2.50 wt.% Dxt‐Ket (25.0 mg mL−1, 22.4 mм ketone) with PEGOA4 (20.1 mg mL−1, 15.7 mм oxyamine), under the same pH and temperature conditions, gelled faster, at 37 ± 5 min, and had a stiffer final G′ of 720 ± 130 Pa (Figure 3d). Young's moduli of these formulations were measured by compression: the 2.25 wt.% Dxt‐Ket dexime had a modulus of 740 ± 50 Pa while that of the 2.50 wt.% formulation was significantly stiffer at 2500 ± 200 Pa (*** p < 0.001, Figure 3e). At pH 6.8, a more biologically relevant value, the crosslinking reaction slowed substantially, with gelation occurring for the 2.50 wt.% formulation at 147 ± 12 min (Figure S8). As this gelation time is longer than desired for in vivo surgery, the higher pH formulation of Dxt‐Ket was not explored further. Rheometry and compression testing were similarly used to characterize formulations with Dxt‐Ald instead of Dxt‐Ket. As aldehydes are more reactive than ketones, faster gelation was anticipated. A 2.50 wt.% Dxt‐Ald (25.0 mg mL−1, 18.3 mм aldehydes) was crosslinked with PEGOA4 (16.5 mg mL−1, 12.8 mм oxyamine) to achieve up to 70 mol% oxime bonds. The formulation was kept at pH 5.8 and 37°C for 2 h while monitoring rheological properties: gelation occurred within 2–3 min and G′ plateaued at 1700 ± 200 Pa (Figure 3f). At pH 6.8, which is closer to physiological pH, gelation of the Dxt‐Ald formulation occurred at 4 ± 1 min (Figure 3g); however, G′ only reached 600 ± 70 Pa, about one‐third of that observed at pH 5.8 [34]. Nevertheless, the ability of the Dxt‐Ald dexime formulations to set quickly at a pH closer to physiological is beneficial for its use as an injectable therapeutic delivery system. Young's moduli of these formulations were also measured by compression: the pH 5.8 Dxt‐Ald crosslinked dexime had a modulus of 2100 ± 200 Pa whereas that at pH 6.8 was significantly softer at 1200 ± 400 Pa (** p < 0.01, Figure 3h). The rheology results reveal different levels of crosslinking density for each of the formulations tested, as crosslinking density is proportional to plateau G′ [61]. An increase in crosslinking density was expected for 2.50 wt.% vs. 2.25 wt.% Dxt‐Ket formulation due to the additional polymer and crosslinker. Interestingly, 2.50 wt.% Dxt‐Ald has a higher plateau G′ and hence higher crosslinking density than 2.50 wt.% Dxt‐Ket (both crosslinked at pH 5.8) despite the same concentration of modified dextrin and lower concentration of crosslinker being used for the Dxt‐Ald. For Dxt‐Ald formulations crosslinked at pH 6.8 vs. pH 5.8, the resulting hydrogel had a lower crosslinking density despite having the same dextrin and crosslinker concentrations, indicating the impact of rate of gelation (which is slower at higher pH) [34] on hydrogel properties [62]. Overall, the rheometric and compression results show that the gelation and mechanical properties of dexime are readily tunable by changing formulation variables such as dextrin weight percentage, the type of crosslinking moiety (ketone vs. aldehyde), and pH. Moving forward, one dexime formulation for each of Dxt‐Ket and Dxt‐Ald crosslinked hydrogels were chosen based on ease of use: 2.50 wt.% Dxt‐Ket at pH 5.8 gelled faster than the 2.25 wt.% formulation while 2.50 wt.% Dxt‐Ald at pH 6.8 gelled quickly and at a more physiologically relevant pH. Both of these formulations were soft gels with storage moduli between 100–1000 Pa, which is similar to methyl cellulose‐based gels that have previously been injected into the subarachnoid space [38, 63, 64], indicating that dexime, based on mechanical properties, should be well‐tolerated in vivo. We next measured the injection force of the dexime formulations through a fine 30G needle at room temperature using a force gauge, as described previously [27]. For Dxt‐Ket crosslinked dexime, injection force increased from the initial 1.0 ± 0.2 N to 21.0 ± 1.3 N by 90 min where it plateaued (Figure 3i). For Dxt‐Ald crosslinked dexime, injection force increased from the initial 1.2 ± 0.1 N to 11.0 ± 5.6 N by 10 min where it plateaued (Figure 3j). The force of injection increases with time as the dexime crosslinking reaction progresses, as reflected in the rheometry data. Importantly, the dexime formulations were both easily injected within a reasonable timeframe and below the 30 N force limit for injections performed by hand [65]. Dexime Degrades Selectively and is Minimal Swelling One of the motivations for pursuing dexime as a novel biodegradable carrier of protein therapeutics is the unsuitability of existing biodegradable hydrogels to deliver glycosaminoglycan‐degrading enzymes of therapeutic importance, such as HAase and ChASE37. These enzymes degrade the hydrogel carrier itself, as evidenced when a hyaluronan‐based hydrogel, hyaluronan‐oxime (HAO) was exposed to a low concentration of ChASE37 (Figure S9). Dexime was designed to deliver this class of enzymes without being degraded by them while at the same time retaining specific enzymatic degradation by amylase. Modification of dextrin was carefully performed to leave several unmodified repeat units as α‐amylase utilizes the C2 and C3 hydroxyl groups of these unmodified repeat units to successfully bind and hydrolyze the polysaccharide chains of the dexime hydrogel [66]. We wanted to avoid hydrolytic degradation because this often results in swelling, as has been observed with other dextrin‐based hydrogels [16], which is undesirable in a confined volume, such as the subarachnoid space around the spinal cord. To better understand the mechanism of degradation, a comparison of hydrolytic vs. enzymatic action on dexime was conducted using the 2.50 wt.% formulations exposed to either artificial cerebrospinal fluid (aCSF) or one of 3 enzymes dissolved in aCSF: α‐amylase, HAase, or ChASE37. Degradation was investigated using α‐amylase at 100, 1, 0.1, and the more clinically relevant 0.01 U mL−1, which is similar to 0.015 U mL−1 reported in the CSF of healthy individuals [67]. To determine enzyme specificity, a similarly high (100 U mL−1) concentration of HAase was utilized. For ChASE37, a concentration of 0.3 U mL−1 was used due to the difficulties in preparing higher quantities of the enzyme for testing higher concentrations. It should be noted that 0.3 U of ChASE37 is the enzyme loading shown previously to be efficacious in vivo [63, 64] and a concentration of 0.1 U mL−1 degraded a hyaluronan‐based hydrogel (Figure S9). Dxt‐Ket dexime degraded with 1 U mL−1 α‐amylase within ∼2 months. Biological levels of α‐amylase (0.01 U mL−1) showed negligible degradation over the same period (Figure 4a), although this would likely be accelerated in vivo due to the continuous enzyme turnover and mechanical forces exerted by cerebrospinal fluid flow [68]. Notably, these lower levels of α‐amylase resulted in a much lower initial swelling of the hydrogel of 0%–8 ± 5%, correlated to α‐amylase concentrations (Figure 4a, dashed box on the right). Thus, injection of dexime in a confined space, such as in the subarachnoid space or even into tissue lesions in the spinal cord, should not cause any tissue damage due to hydrogel swelling. Furthermore, the slow biodegradation of dexime in the presence of more systemic levels of α‐amylase allows the hydrogel to serve its intended function of spatiotemporal delivery of protein therapeutics prior to degradation. Evaluation of hydrolytic stability, swelling, and specific enzymatic degradation of different dexime formulations in vitro. (a) To evaluate the degradation and swelling behavior of dexime at more clinically relevant concentrations of α‐amylase, hydrogels formulated with 2.50 wt.% of Dxt‐Ket were subjected to various levels of α‐amylase in aCSF, or aCSF alone over the course of 2 months. Significant differences between the groups were determined using two‐way ANOVA with Tukey's post hoc (n = 4; mean ± standard deviation plotted; *** p < 0.001, **** p < 0.0001, ns = not significant). To evaluate the specific degradation of dexime, hydrogels were prepared using the 2.50 wt.% formulations of either (b) Dxt‐Ket or (c) Dxt‐Ald and subjected to one of: α‐amylase, HAase, or ChASE37 in aCSF vs. aCSF alone over the course of 1 month. Significant differences comparing the amylase group to the others were determined using two‐way ANOVA with Tukey's post hoc (n = 4; mean ± standard deviation plotted; * p < 0.05, ** p < 0.01, **** p < 0.0001, ns = not significant). For figures a–c, the dashed boxes on the right are magnified to show the first week of measurements. In the accelerated degradation study with 100 U mL−1 of α‐amylase, Dxt‐Ket dexime swelled immediately, peaking after 1 d of exposure, and then degraded fully by 2 weeks. In contrast, neither HAase nor ChASE37 affected dexime hydrogel mass over 28 d (Figure 4b), demonstrating the specificity of α‐amylase degradation and the suitability of dexime for the delivery of glycosaminoglycan‐degrading enzymes like HAase and ChASE37. With dexime formulated from Dxt‐Ald, a similar degradation profile occurred in the presence of high α‐amylase concentrations, showing immediate swelling that peaked at 2 d followed by full degradation by 7 d. None of the other conditions tested demonstrated any significant swelling or loss of mass, emphasizing the specificity of α‐amylase degradation of dexime (Figure 4c). These data suggest that dexime would be suitable for the delivery of either ChASE37 or HAase. Dexime is Cytocompatible Dexime was evaluated in vitro with the microglial SIM‐A9 cell line, using nitrite production as a proxy for an immunogenic response. Cells were plated either directly on the dexime 2.50 wt.% formulations prepared from either Dxt‐Ket or Dxt‐Ald or cultured in wells with dexime‐conditioned media (CM). Both dexime formulations exhibited similar nitrite values to the no treatment control, according to the Griess assay (and normalized to cell number), demonstrating low immunogenicity (Figure 5a,b). As expected, cells in the lipopolysaccharide (LPS) positive control group secreted significantly more nitrite, as we and others have previously reported [50, 69]. In vitro immunogenicity and cytocompatibility assays of dexime formulations. As a proxy for immunogenicity, nitrite levels of SIM‐A9 cells (a) plated directly on top of dexime or (b) exposed to dexime conditioned medium (CM) were measured. For cytocompatibility, the (c) cytotoxicity based on lactate dehydrogenase (LDH) production and (d) viability, based on metabolic activity (by Presto Blue) of NPCs cultured in dexime coated Transwells. Significant differences were calculated using an ordinary one‐way ANOVA with Tukey's post hoc (n = 3 biological replicates; mean ± standard deviation plotted; *** p < 0.001, **** p < 0.0001, ns = not significant). To evaluate cytocompatibility, human neural progenitor cells (NPCs) were cultured with either Dxt‐Ket or Dxt‐Ald crosslinked hydrogels placed in Transwell inserts. After 24 h of culture, the lactate dehydrogenase (LDH) assay was used to measure cytotoxicity (Figure 5c) while the PrestoBlue metabolic assay was used as a proxy for viability (Figure 5d). Both dexime formulations were cytocompatible, with neither increase in LDH secretion nor reduction in metabolic activity compared to the no gel control, and significantly different from the cytotoxic control. While both Dxt‐Ket and Dxt‐Ald formulations showed similar cytocompatibility, we advanced Dxt‐Ald‐based gels for further investigation in vivo as this formulation gels faster at near physiological pH and is easier to handle. Dexime Releases Bioactive ChASE37 To evaluate the capacity of dexime to deliver an active enzyme therapeutic over time, the release of ChASE37 was investigated first in vitro and then in vivo. ChASE37 was recombinantly expressed in E. coli, purified, and characterized for molar mass and activity (Figure S10). ChASE37 was combined with the Dxt‐Ald 2.50 wt.%, pH 6.8 dexime formulation and its release was followed into artificial cerebrospinal fluid (aCSF) at 37°C for 7 d using a custom enzyme‐linked immunosorbent assay (ELISA) [37, 63], and for activity, using chondroitin sulfate A (CSA) degradation [37, 56]. Greater than 90% of the loaded ChASE37 was released from dexime over a 7 d period, with the most (80%) of this release occurring within the first 48 h (Figure 6a). Importantly, while ChASE and its derivatives lose specific activity over time due to their thermal instability, ChASE37 remained bioactive after 7 d of release at physiological temperature (Figure 6b, raw data in Table S1). Both the release and bioactivity results are consistent with previous observations for ChASE37 delivered from crosslinked methyl cellulose [56]. In vitro and in vivo release studies of ChASE37 from dexime. For the in vitro study, (a) the release profile of the enzyme from dexime was measured with a custom ELISA and (b) the released enzyme's specific activity was measured based on chondroitin sulfate‐A degradation (n = 3, mean ± standard deviation plotted). For the in vivo study, (c) dexime alone or a combination of dexime and ChASE37 (ChASE37‐Dex) was injected into the dorsal sub‐arachnoid space of the rat spinal cord and sacrificed 7 d later. (d) WFA staining of the ventral horn of a spinal cord cross section. The ventral horn refers to gray matter, which is delineated from the white matter with a dashed line. Gray matter contains cell bodies, with motor neurons being the largest cells, appearing as black empty spaces (identified by white arrows). These cells are surrounded by a green glowing halo from the WFA staining, which demarcates the PNNs that contain CSPGs. In the control sample that received only dexime (left), the glowing PNN halo is clearly visible around the motor neuron cell bodies. In contrast, after ChASE37‐Dex application (right), the PNN halo is absent and only the motor neuron cell bodies remain due to ChASE37 activity. Scale bars = 50 µm, 20× objective. (e) ChASE37‐Dex shows a significant decrease in CSPGs in PNNs. Significance was determined using an unpaired t‐test (n = 7 biological replicates; mean ± standard deviation plotted; **** p < 0.0001). (f) Iba‐1 (microglia) staining and (g) quantification performed on cross sections of the spinal cord where dexime or ChASE37‐Dex was injected. White arrows in both the samples indicate hydrogel that was visible on the dorsal side of the cord, indicating that it remained in place over the time of the experiment and covered the dorsal half of the cord. (h) GFAP (astrocytes) staining and (i) quantification performed on cross sections of the spinal cord where dexime or ChASE37‐Dex was injected. Scale bars = 500 µm, 20× objective. In both the Iba‐1 and GFAP histology, no significant differences were observed between dexime and ChASE37‐Dex. Significant differences were determined using an unpaired t‐test (n = 7 biological replicates; mean ± standard deviation plotted; ns = not significant). Next, Dxt‐Ald crosslinked dexime was evaluated in vivo for biocompatibility and bioactive ChASE37 delivery. Our dexime/enzyme system, denoted as ChASE37‐Dex, was injected into the dorsal subarachnoid space of healthy Lewis rats at thoracic level 9 (Tx 9) (Figure 6c). The rats were maintained for 7 d before being sacrificed for tissue analysis. Immunohistochemistry was used to analyze spinal cord tissue at the site of injection, using Wisteria floribunda agglutinin (WFA) staining to detect CSPGs of the PNNs (Figure 6d). In the dexime control (without ChASE37), WFA staining showed PNNs surrounding the motor neuron cell bodies in the ventral horn (indicated with white arrows). In contrast, the ventral horn of the rat treated with ChASE37‐Dex demonstrated a lack of PNNs around the motor neuron cell bodies. This indicates that the released ChASE37 remained bioactive and diffused from the dorsal to the ventral side of the cord. Quantification of WFA staining revealed a significant difference between ChASE37‐Dex and the dexime control (Figure 6e), thereby providing proof‐of‐concept evidence for bioactive release of ChASE37‐Dex in vivo for the first time while at the same time setting the stage for future investigations in a SCI model. To assess biocompatibility, we stained the spinal cord at the site of injection for ionized calcium‐binding adapter molecule 1 (Iba‐1) to detect microglia (Figure 6f) and found no significant differences between dexime and ChASE37‐Dex (Figure 6g). Interestingly, the injected dexime is visible in the Iba‐1 staining (indicated with white arrows) and remained at the injection site for 7 d, despite CSF flow and animal movement. The capacity of dexime to remain where it is injected is promising as it gives the user spatiotemporal control for therapeutic delivery to a specific location within the body. Similarly, we stained the spinal cord tissue for glial fibrillary acidic protein (GFAP) to detect astrocytes (Figure 6h) and found no significant differences between dexime and ChASE37‐Dex (Figure 6i). Animal health that was monitored post‐surgery included piloerection, isolation within the cage, vocalization, porphyrin staining, reduced grooming, altered gait, and reduced mobility. We did not observe any changes in any of these signs post‐subdural injection of either ChASE37‐Dex or Dexime alone. As expected, weights of animals in both groups dropped post‐surgery and after 3 days began to increase again (Table S2). Overall, the rats remained healthy without any adverse behavior after treatment. Conclusions For the first time, we described the synthesis of dexime: an enzyme‐degradable, dextrin‐oxime hydrogel engineered for the spatiotemporal delivery of protein therapeutics, especially those that degrade GAGs. Dextrin was modified with either ketone or aldehyde moieties and crosslinked with poly(ethylene glycol)‐tetraoxyamine to yield dexime. The rheological and mechanical properties of dexime were tunable by changing pH, ketone or aldehyde functional groups, and the weight percentage of dextrin‐ketone/aldehyde itself. Formulations were easily injectable through a fine 30G syringe. Dexime was hydrolytically and enzymatically stable and minimally swelling when exposed to ChASE37 and HAase, while being specifically degradable by α‐amylase, an endogenous enzyme that hydrolyses α(1→4) D‐glucose polysaccharides. Dexime was cytocompatible in vitro, biocompatible in vivo and released bioactive ChASE37 both in vitro and in vivo. This novel hydrogel is promising for local and sustained delivery of enzymes as it is injectable, minimally swelling, and selectively biodegradable. Methods General Materials All materials were used as received unless otherwise noted. Dextrin (commercial grade Type II powder from corn), α‐amylase (from A. oryzae), hyaluronidase (Type IV‐S, from bovine testes), 2,2′‐azino‐bis(3‐ethylbenzothiazoline‐6‐sulfonic acid) (ABTS) liquid substrate solution, sodium dodecyl sulfate, CSA, LPS, bromophenol blue, bromoacetic acid, diisopropylethylamine, bovine serum albumin (protease‐free grade; BSA), MgCl2, Triton X‐100, Accutase, chlorohexidine digluconate, heparin sodium salt, and WFA were obtained from Sigma–Aldrich (US). 2,2‐Dimethoxy‐1‐propanamine and 4,4‐dimethoxy‐1‐butanamine were obtained from 1PlusChem (US). Tween 20, glycine, Tris base, Coomassie brilliant blue G‐250, 2‐(N‐morpholino)ethanesulfonic acid (MES), MES sodium salt, Na2HPO4, NaH2PO4•H2O, CaCl2•2H2O, HCl, NaOH, KCl, and NaCl were obtained from BioShop (CA). 4‐Arm poly(ethylene glycol) amine hydrochloride salt (Mn = 5000 Da; PEG[NH3Cl]4) was purchased from JenKem (US). (Boc‐aminooxy)acetic acid was obtained from AK Scientific (US). N,N'‐diisopropylcarbodiimide (DIC) and DMTMM were obtained from TCI (JP). Regenerated cellulose dialysis tubing was obtained from SpectraPor (US). EDTA‐free Complete Mini Protease Inhibitor Tablets were obtained from Roche (CH). Surgical staples and heated water blankets (TP‐700/TP‐R 22G) were acquired from Braintree (US). 5‐0 vicryl sutures were acquired from Ethicon (GB). Systane ophthalmic ointment was acquired from Alcon (CH). Dulbecco's phosphate buffered saline (PBS) was obtained from Wisent Inc (CA). Novex Wedgewell 4%–20% tris‐glycine mini gels, AF488 Streptavidin for WFA (S32354), and Fluoromount G were acquired from Invitrogen (US). Ni‐NTA 96‐well coated plates were obtained from Qiagen (DE). UV‐star 96‐well plates were obtained from Greiner Bio‐One (AT). ChASE37 was expressed in Escherichia coli BL21(DE3) cells (ClearColi from Biosearch Technologies, US) and purified as previously reported [56]. SIM‐A9 microglial cells were obtained from ATCC (US). Griess and LDH assay reagents were obtained from Promega (US). PageRuler Plus prestained protein ladder (10–250 kDa), PrestoBlue assay reagents, DMEM:F12 and DMEM:F12:Glutamax media, horse serum, fetal bovine serum, β‐mercaptoethanol, B27, BSA‐FV, penicillin‐streptomycin, optimal cutting temperature compound (OCT), 2‐methylbutane, and SuperFrost Plus microscope slides were obtained from Thermo Fisher Scientific (US). Sodium pentobarbital (Euthanyl) and buprenorphine were acquired from Western Drug Distribution Center Ltd (CA). N2 was obtained from Wisent Bioproducts (CA). Basic FGF was obtained from R&D Systems (US). NPCs were differentiated from iPSCs as previously reported [50]. Artificial cerebrospinal fluid (aCSF) was formulated in water using the following solute concentrations: 149 mм NaCl, 3 mм KCl, 0.8 mм MgCl2, 1.4 mм CaCl2•2H2O, 1.5 mм Na2HPO4, and 0.2 mм NaH2PO4•H2O. Antibodies used are listed in Table 1: TABLE 1Information on antibodies used in this work.NameSourceCatalog NumberAnti‐FLAG rabbit (HRP)Abcam (GB)AB2493Anti‐GFAP rabbitAbcam (GB)AB278054Anti‐Iba‐1 rabbitFujifilm Wako (JP)019‐19741Anti‐rabbit goat (AF555)Invitrogen (US)A21428Anti‐rabbit goat (AF647)Invitrogen (US)A21244John Wiley & Sons, Ltd. Information on antibodies used in this work. General Instrumentation NMR spectra were acquired with either a Bruker Ascend 400 MHz spectrometer, Agilent DD2 700 MHz spectrometer, or Agilent DD2 500 MHz spectrometer. The residual HOD solvent signal (4.79 ppm) was used as a reference peak within the spectra whenever appropriate. FTIR spectra were acquired with a Perkin Elmer Spectrum Two with attenuated total reflectance (ATR) sampling. Matrix‐assisted laser desorption ionization (MALDI) mass spectra were acquired with a Bruker Autoflex Speed. Samples were mixed with a 2,5‐dihydroxybenzoic acid matrix in water. Electrospray ionization (ESI) mass spectra were acquired with an Agilent 6538 UHD. Samples were diluted in a 1:1 volumetric solution of MeOH and 0.1% v/v aqueous formic acid. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) was run using a BioRad Powerpac 1000 and imaged using a Life Technologies E‐Gel imager. Rheology measurements were made with a TA Instruments Discovery HR‐2. Compression measurements were made with a Biomomentum Mach‐1 Mechanical Tester with a Newport ESP300 Universal Motion Controller. Injection forces were measured using a Mark 10 Series 5 Model M5‐50 Advanced Digital Force Gauge. UV–vis spectra of culture plates were acquired with a Tecan Infinite M200 Pro plate reader. Histology slides were prepared using a Leica Biosystems 3050S cryostat. Immunohistology images were acquired using a Leica TCS SP8 confocal microscope and analyzed using ImageJ (v. 1.54f) software. Dextrin Characterization 1H NMR (400 MHz, D2O): δ = 5.42 (br s, 1H; H1), 3.93–4.07 (m, 1H; H3), 3.76–3.93 (m, 3H; H5, 2×H6), 3.57–3.76 (m, 2H; H2, H4) ppm. FTIR (ATR): ν = 3290 (O─H stretch), 2930 (C─H stretch) cm−1. MS (MALDI): Mn = 2.3 kDa, Mw = 3.0 kDa, Đ = 1.3. Synthesis of Dxt‐COONa (32% Carboxyl sub) Dextrin (4.0 g, 25 mmol of repeat unit, 1.0 equiv) was dissolved in 400 mL of 1.5 м NaOH solution. The solution was cooled to 4°C and bromoacetic acid (34 g, 240 mmol, 9.6 equiv) was added before sealing the flask and allowing it to stir. After 20 h, the reaction was brought to room temperature and NaH2PO4•H2O (5.5 g to make a 0.1 м solution) was added, followed by the addition of 6 м HCl until the solution was neutralized. The solution was dialyzed in 1 kDa MWCO RC tubing against water for purification. The dialyzed solution was then lyophilized to afford 3.4 g of a white solid (yield = 73%). 1H NMR (400 MHz, D2O): δ = 5.67 (br s, 0.32H; H1′), 5.48 (br s, 0.68H; H1), 3.38–4.33 (m, 6.17H; H2, H3, H4, H5, 2×H6, CH 2COONa) ppm. FTIR (ATR): ν = 3290 (O─H stretch), 2930 (C─H stretch), 1590 (C─O stretch) cm−1. Dxt‐COOD Characterization (32% Carboxyl sub) 1H NMR (400 MHz, D2O/DCl): δ = 5.68 (br s, 0.32H; H1′), 5.42 (br s, 0.68H; H1), 4.44 (br s, 0.64H; CH 2COOD), 3.38–4.30 (m, 6.02H; H2, H3, H4, H5, 2×H6) ppm. Dxt‐COONa Substitution To determine the carboxyl substitution of Dxt‐COONa, a sample was dissolved in D2O at 10 mg mL−1 and then acidified with a small amount of DCl (1 µL of 12 м DCl in 500 µL total) to obtain a spectrum of Dxt‐COOD. When in an acidic environment, it was found that the methylene protons of the carboxyl α carbon would separate from the rest of the large multiplet at ∼3.3–4.3 ppm and shift downfield as a broad singlet at ∼4.4 ppm. By normalizing the anomeric proton peaks (∼5.3–5.8 ppm) to 1H and integrating the α carbon methylene protons, carboxyl substitution could be determined using Equation (1): (1) CarboxylSubstitution=i2H100% where i is integration of the methylene protons on the carboxyl α carbon and 2H is number of protons on the carboxyl α carbon. Synthesis of Dxt‐Ket (20% Ketone sub) Dxt‐COONa (1.0 g, 5.3 mmol of repeat unit, 1.0 equiv) was dissolved in 100 mL of MES buffer (0.1 m, pH 6.6). DMTMM (2.4 g, 8.7 mmol, 1.6 equiv) was added and the reaction was sealed and stirred at room temperature for 30 min before 2,2‐dimethoxy‐1‐propanamine was added (0.51 g, 4.3 mmol, 0.81 equiv). The reaction was sealed and stirred for 48 h before dialyzing the reaction mixture in 1 kDa MWCO RC tubing against 0.2 м HCl for 2 h. The reaction was further dialyzed against water, then lyophilized to afford 0.72 g of a white solid (yield = 71%). 1H NMR (400 MHz, D2O): δ = 5.53–5.80 (m, 0.32H; H1′), 5.41 (br s, 0.68H; H1), 3.35–4.55 (m, 7.61H; H2, H3, H4, H5, 2×H6, CH 2CONH), 2.26 (br s, 0.60H; COCH 3) ppm. FTIR (ATR): ν = 3330 (O─H stretch), 2930 (C─H stretch), 1730 (C═O stretch), 1660 (C═O stretch), 1560 (N─H bend) cm−1. Dxt‐Ket Substitution Ketone substitution of Dxt‐Ket was measured using its 1H NMR spectrum. By normalizing the anomeric proton peaks (∼5.3–5.8 ppm) to 1H and integrating the area under the peak representing the ketone methyl group (∼2.3 ppm), substitution could be determined according to Equation 2: (2) KetoneSubstitution=i3H100% where i is ketone methyl group integration and 3H is number of protons on the ketone methyl group. Synthesis of Dxt‐Ald (14% Aldehyde sub) Dxt‐COONa (0.50 g, 2.7 mmol, 1.0 equiv) was dissolved in 50 mL of MES buffer (0.1 m, pH 6.6). DMTMM (1.8 g, 6.5 mmol, 2.4 equiv) was added and the reaction was sealed and stirred at room temperature for 30 min before 4,4‐dimethoxy‐1‐butanamine was added (0.36 g, 2.7 mmol, 1.0 equiv). The reaction was sealed and stirred for 48 h before dialyzing the reaction mixture in 1 kDa MWCO RC tubing against 0.2 м HCl for 48 h. The reaction was further dialyzed against water, then lyophilized to afford 0.43 g of a white solid (yield = 85%). 1H NMR (400 MHz, D2O): δ = 9.71 (br s, 0.03H; COH), 5.67 (br s, 0.32H; H1′), 5.43 (br s, 0.68H; H1), 3.16–4.53 (m, 6.35H; H2, H3, H4, H5, 2×H6, CH 2CONH, CH 2CH2CH2COH), 2.62 (br s, 0.07H; CH2CH2CH 2COH), 1.79–2.24 (m, 0.35H; CH2CH 2CH 2COH), 1.65 (br s, 0.14H; CH2CH2CH 2COH) ppm. FTIR (ATR): ν = 3290 (O─H stretch), 2930 (C─H stretch), 1730 (C═O stretch), 1640 (C═O stretch), 1580 (N─H bend) cm−1. Dxt‐Ald Substitution Aldehyde substitution of Dxt‐Ald was measured using its 1H NMR spectrum. By normalizing the anomeric proton peaks (∼5.3–5.8 ppm) to 1H and integrating the area under the peaks representing the methylene peaks of the α and β carbons of the aldehydes (∼2.6, ∼1.8–2.2, ∼1.3 ppm) substitution could be determined according to Equation 3: (3) AldehydeSubstitution=ia+iβ4H100% where ia is combined integrations of α protons, i β is integration of β protons, and 4H is number of methylene protons on the aldehyde α and β carbons. Synthesis of PEGOA4 PEGOA4 was prepared based on a previously reported procedure [26]. Boc‐aminooxyacetic acid (0.69 g, 3.6 mmol, 12 equiv) was dissolved under nitrogen in 40 mL of anhydrous CH2Cl2. The reaction flask was cooled to 0°C before DIC (1.1 mL, 7.0 mmol, 23 equiv) was added, and the reaction was stirred for 1 h. PEG(NH3Cl)4 (1.5 g, 0.30 mmol, 1.0 equiv) and diisopropylethylamine (2.5 mL, 14 mmol, 47 equiv) were added and the reaction was warmed up to room temperature before being stirred under nitrogen for 72 h. Solvent was removed under vacuum and 40 mL of water was added before filtering the mixture to remove the diisopropyl urea byproduct. The filtrate was dialyzed against 0.2 м HCl for 48 h using 1 kDa MWCO dialysis tubing, followed by 0.1 м NaHCO3 and then water. The final solution was lyophilized to yield 0.83 g of a brittle off‐white solid (yield = 54%). 1H NMR (400 MHz, D2O): δ = 4.22 (s, 1H; NHCOCH 2ONH2), 3.72 (br s, 110H; CH 2CH 2O), 3.47–3.55 (m, 4H; CH 2NHCO, CCH 2O) ppm. Dexime Formulations Dexime was prepared by dissolving modified dextrin in water, followed by the addition of an appropriate buffer solution to adjust to the desired pH. For all in vitro work, an MES buffer at a final working concentration of 0.1 м was used. For in vivo work, a phosphate buffer at a final working concentration of 0.1 м was used. To initiate gelation, PEGOA4 dissolved in water was added to the dextrin/buffer solution. Rheology Rheology of the hydrogels was performed using a 20 mm steel flat plate geometry with a water reservoir and a 100 µm geometry gap. Hydrogel formulations were prepared immediately prior before measurements were conducted, and a solvent trap was employed with mineral oil to prevent loss of water from the sample during measurements. Measurements were conducted with a plate oscillation at a strain of 2% and a frequency of 1 Hz at a temperature of 37°C. Changes in storage (G′) and loss (G′′) moduli were monitored over time, with gelation time defined as the intercept where G′ becomes greater than G′′. Rheology results are plotted as an average of the replicates moduli + standard deviation. To correct plots with negative values before gelation (caused by instrument sensitivity issues), 0.1 Pa was added to all data points in order to normalize the values and allow for their visualization on a logarithmic plot (plots with this correction are explicitly stated as normalized). Unconfined Mechanical Testing for Compression Moduli Hydrogel formulations were prepared as batches and then aliquoted 4 × 150 µL into a 96‐well plate mold with a temporary bottom. The hydrogels were allowed to set at 37°C (1.5 h for Dxt‐Ket, 45 min for Dxt‐Ald) before adding 150 µL of aCSF on top of each. The hydrogels were placed in a Petri dish containing water and the dish was sealed with parafilm before storing the hydrogels at 37°C overnight to cure. Hydrogels were removed from the mold by removing the temporary bottom and using a fine gauge needle to help break adhesion between the hydrogel and the mold wall. Each hydrogel was placed under the mechanical tester compression head with minimal aCSF added to promote adhesion to the instrument and avoid water loss from the hydrogel. After finding contact and compressing an additional 20% of the overall height of the hydrogel to ensure good contact, compression force was measured over a series of 5% × 2% strain loads being applied. Each load was applied over 1 s, with a 30 s relaxation period between each load application. The slopes of the curves were averaged and used to determine the Young's modulus of each hydrogel via Equation 4: (4) E=101.9716Ngf−1mL0A where E is Young's modulus (in Pa), m is Curve slope (in gf mm−1), L0 is Overall height of the hydrogel (in mm), and A is Surface area of the hydrogel (in m2, defined as 0.00004 m2 for the 96 well mold) Injection Force Hydrogel formulations were prepared at room temperature and drawn into a syringe with a 30G needle attached. The syringe was placed in a holder and the force to cause a drop to form on the tip of the needle was measured using a force gauge at set timepoints. At each timepoint, 4–6 technical replicates were taken. The first measurement at each timepoint was discarded due to additional force sometimes required to clear the needle of cured gel. Injection force results are plotted as an average of separate injection force experiments on a formulation ± standard deviation (n = 3). Degradation and Swelling Assays Hydrogel formulations were prepared as batches and then aliquoted 4 × 100 µL into pre‐weighed Eppendorf tubes for each degradation condition being assessed. The hydrogels were cured overnight at 37°C. Separately, enzyme solutions for the various conditions to be evaluated were prepared in aCSF, aliquoted for each timepoint, and frozen at −80°C for future use. After curing, the hydrogels were weighed to obtain their initial mass. A 900 µL aliquot of the appropriate enzyme solution (or aCSF only for the control) was applied to each hydrogel and the samples were incubated at 37°C on an orbital shaker. At set timepoints, the hydrogels were removed from the incubator, their solutions poured out, and the were blotted dried with KimWipes before weighing them. Hydrogels were then treated with fresh aliquots of the appropriate enzyme solution and placed back in the incubator until the next timepoint. The mass of the hydrogel was calculated relative to the initial mass over time. Immunogenicity SIM‐A9 microglial cells were cultured in DMEM:F12 media containing 5% heat‐inactivated horse serum, 10% heat‐inactivated fetal bovine serum, and 1% penicillin‐streptomycin. Media was replaced every 2–3 d. For immunogenicity assays using direct contact, wells were coated with hydrogel formulation and allowed to gel, followed by cell seeding (1 × 105 cells per well of a 96‐well plate). For indirect assays, hydrogels (100 µL) were allowed to set in a 96‐well plate and then covered with complete media overnight. Simultaneously, cells were plated in 96‐well plates and allowed to adhere overnight. The following day, media was replaced with a mixture of fresh media and dexime CM at a 1:1 ratio. For both assays, media was collected 24 h later and nitrite secretion was evaluated using the Griess assay according to the manufacturer's instructions, and viability was measured with 10% PrestoBlue in complete media. For controls, cells were either cultured in the absence of dexime or CM, or in the presence of 100 ng mL−1 LPS to induce a pro‐inflammatory phenotype. Cytocompatibility Human NPCs were differentiated from the Zan11 induced pluripotent cell line, kindly donated by the Zandstra lab as previously reported [50]. NPCs were plated on Matrigel and cultured in DMEM:F12:Glutamax media, supplemented with 0.2% β‐mercaptoethanol, 1% N2, 2% B27, and 0.05% BSA‐FV (7.5% solution). Media was replaced every other day and supplemented with 20 ng mL−1 basic FGF. Cells were dissociated with Accutase, seeded at 400 000 cells per well in the bottom of 24‐well Transwell plates, and allowed to adhere overnight. Dexime was prepared from sterile filtered components combined under aseptic conditions. The inserts for the Transwells were coated with 100 µL of dexime hydrogel or left uncoated. Cells were incubated for 24 h before the media was collected and analyzed for cytotoxicity using the LDH assay and viability using the PrestoBlue assay, both according to the manufacturer's protocol. As a cytotoxic positive control, cells were treated with 2% Triton X‐100 for 15 min before media was collected and analyzed. Release and Activity of ChASE37 From Dexime Dxt‐Ald dexime was prepared containing 0.1 µg µL−1 of ChASE37. The crosslinker PEGOA4 was added and dexime solution was aliquoted 3 × 100 µL into Eppendorf tubes. The hydrogels were allowed to set at 37°C for 1.5 h. Aliquots of release media (aCSF containing 0.1% w/v BSA and protease inhibitor tablets, 900 µL) were added on top of the hydrogels and they were incubated at 37°C on an oscillator for set time points. At each time point, hydrogels were removed from the incubator, release media was collected and replaced with fresh media. Release media samples for each time point were stored at −80°C for later analysis. Release Sample Enzyme Quantification Enzyme release samples were quantified using a previously developed custom HIS/FLAG tag ELISA protocol [37, 63]. Samples were thawed and diluted appropriately with release buffer before 200 µL of each sample was placed in a Ni‐NTA coated 96‐well plate well. After incubating for 2 h at 37°C with gentle shaking, the plate was washed (using PBS with 0.05 vol% Tween 20) and 200 µL of anti‐FLAG HRP conjugated rabbit antibody (diluted 1:5000 in PBS containing 0.2% BSA) was added to each well before incubating at 37°C with gentle shaking for an additional 2 h. The plate was washed again and 100 µL of ABTS solution was added to each well before developing the plate for 10 min with gentle shaking at room temperature. The wells were read for absorbance at 405 nm (with a 650 nm correction). Protein concentration was calculated using a ChASE37 standard curve developed on the same ELISA plate. Release Sample Activity Quantification Activity of each release sample was determined by monitoring the development of the CSA digestion product at 232 nm, as previously described [37, 56]. A 50 µL aliquot of each release sample was placed in a well of a UV‐star plate. PBS was also aliquoted into wells to serve as a control. A 50 µL aliquot of CSA in PBS (18 mg mL−1) was subsequently added to each well and the plate was assayed immediately afterward. After shaking the plate for 5 s, the absorbance of each well was recorded every 20 s for 20 min. Specific activity of each sample was determined using the slope of the absorbance over time plot of each sample, and the sample concentration (determined previously via ELISA) using Equation 5: (5) SpecificActivityUmg−1=m260smin−1c0.32cm5.1mM−1cm−1 where m is slope of sample absorbance over time (in s−1) and c is concentration of sample (in mg mL−1). Intrathecal Injection in a Healthy Rat Dexime was prepared from sterile filtered components combined under aseptic conditions. The components were combined with ChASE37 (at a loading of 0.05 U µL−1) if preparing ChASE37‐Dex, or with no ChASE37 if preparing dexime only. After combining the components, dexime was drawn into a 50 µL Hamilton syringe, with a 22G needle bent to a 45° angle. The solution was allowed to set in the syringe for ∼20 min immediately prior to performing the injection. All animal procedures were approved by the Animal Care and Use Committee at the University of Alberta. Female Lewis rats (Charles River, CA), 180–220 g, were anesthetized with isoflurane (4% induction, 3% maintenance) in a 50:50 oxygen:air mix. The hair over the thoracic vertebrae was removed and the rat was placed in a stereotactic frame to prevent movement. Ophthalmic ointment was applied to the eyes, and the skin was wiped with 10% chlorhexidine digluconate. A partial laminectomy was made at T9 to expose the spinal cord. The syringe needle was placed between the dura and spinal cord before injecting either ChASE37‐Dex (10 µL; 0.5 U; 7 rats), or dexime alone (10 µL; 7 rats), in the dorsal subarachnoid space. The muscle was then sutured and the skin closed with 9 mm staples. Buprenorphine (0.03 mg kg−1) was injected before the skin incision was made and 8 h post‐surgery. Rats were kept warm on heated water blankets until they were fully awake, at which time they were transferred back to their home cages. Perfusion and Tissue Collection After 7 d post injection, animals were euthanized with a lethal dose of sodium pentobarbital (0.03 mg kg−1). Rats underwent transcardial perfusion with PBS containing 0.02 g L−1 heparin, followed by 4% paraformaldehyde in a 5% sucrose solution, and post fixed in the same solution overnight at 4°C. The tissue was then cryoprotected with 30% sucrose in 0.1 м phosphate buffer at 4°C for 7 d. The thoracic spinal cord at the level of the injection site was then extracted. The tissue segments were embedded in OCT, mounted on filter paper, frozen in 2‐methylbutane at −50°C to −60°C, and stored at −80°C processed. Twenty‐five µm sections covering the injection site were collected as cross‐sections using a Leica 3050S cryostat at −20°C. All tissue sections were mounted on SuperFrost Plus microscope slides and stored at −20°C until histological analyses. Histology Slides were taken from −20°C and incubated at 37°C for 1 h. The slides were washed in Tris‐buffered saline (TBS) 1 time for 10 min and then in TBS with 0.3% Triton X‐ 100 (TBS‐T) 2 times for 10 min at room temperature. The slides were then blocked for 1 h at room temperature with TBS‐T containing 10% normal goat serum (NGS). The blocking solution was removed, and the slides were incubated with primary antibodies in TBS‐T with 2% NGS. One slide was stained for CSPGs and astrocytes, 1:200 WFA and 1:200 GFAP rabbit antibody, respectively, and another for microglia, 1:1000 Iba‐1 rabbit antibody, respectively. The next day the slides were washed 3 times for 10 min in TBS followed by a 2 h room temperature incubation with secondary antibodies in TBS‐T with 2% NGS. The secondary antibodies used were 1:500 AF488 Streptavidin for WFA, 1:500 AF555 goat anti‐rabbit for GFAP, and 1:500 AF647 goat anti‐rabbit for Iba‐1. The slides were then washed 3 × 10 min in TBS‐T and 1 × 10 min in TBS and mounted with a coverslip with Fluoromount and stored at 4°C until imaged. Image Analysis Images of sections were taken on a confocal microscope and analyzed using Image J. Images were taken with the 20× objective. Image analysis was blinded to which group was being analyzed. The numbers of either WFA positive motor neurons were counted in the ventral horns of 3 sections per animal. The optical density of GFAP and Iba‐1 staining was taken from the dorsal spinal cord from the level of the lateral funiculus on both, in the region where the dextrin gel was visible. Background optical density, from a region with no GFAP or Iba‐1 staining, was subtracted from the optical density value. Statistical Analyses Statistics were calculated using GraphPad Prism (v. 10.1.1). Tests performed are noted where they were used. Significance is represented using the symbols: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. A p > 0.05 is denoted as ns (not significant). Conflicts of Interest The dextrin‐oxime hydrogel (dexime) described herein has been submitted for patent protection. The authors declare no other competing interests. Supporting information Supporting File: adma73262‐sup‐0001‐SuppMat.docx.