Local delivery of decorin via hyaluronic acid microrods improves cardiac performance and ventricular remodeling after myocardial infarction

Heart failure (HF) is a global public health burden and associated with significant morbidity and mortality. HF can result as a complication following myocardial infarction (MI), with cardiac fibrosis forming in the myocardium as a response to injury. The dense, avascular scar tissue that develops in the myocardium after injury following MI creates an inhospitable microenvironment that hinders cellular function, survival, and recruitment, thus severely limiting tissue regeneration. We have previously demonstrated the ability of hyaluronic acid (HA) polymer microrods to modulate fibroblast phenotype using discrete biophysical cues and to improve cardiac outcomes after implantation in rodent models of ischemia-reperfusion MI injury. Here, we developed a dual-pronged biochemical and biophysical therapeutic strategy leveraging bioactive microrods to more robustly attenuate cardiac fibrosis after acute myocardial injury. Incorporation of the anti-fibrotic proteoglycan decorin within microrods led to sustained release of decorin over one month in vitro and after implantation, resulted in marked improvement in cardiac function and ventricular remodeling, along with decreased fibrosis and cardiomyocyte hypertrophy. Together, this body of work aims to contribute important knowledge to help develop rationally designed engineered biomaterials that may be used to successfully treat cardiovascular diseases.


Introduction
Heart failure (HF) affects approximately 6 million Americans, and the prevalence is projected to increase 46% from 2012 to 2030. 1 The prognosis for HF is poor, with an estimated mortality rate of ≈ 50% within 5 years of the diagnosis. 1 Myocardial infarction (MI) from coronary artery disease is the leading cause of HF and despite advances in the management of MI, subsequent pathologic remodeling of ischemic myocardium with brotic scar tissue and aneurysmal degeneration leads to HF and death. 2 Therefore, therapies that can prevent scar tissue formation, increase cell survival, and promote contractile tissue regeneration need to be identi ed to treat this growing patient population in the wake of increasing rates of obesity and heart disease. 1 To date, cellular therapies have had limited success in promoting long-term cardiovascular repair 3,4 and no delivery system for an effective pharmacotherapy exists. Moreover, current treatment of brosis involves systemic inhibition of cytokines and chemokines which leads to many adverse side effects for patients. 5,6 In recent years, there has been growing interest in targeting pathways that lead to altered cardiovascular cell phenotypes and microenvironments after injury to reduce maladaptive repair and promote functional recovery. As most of these cell types are mechanosensitive and rely on micro-and nanoscale cues from the extracellular matrix (ECM) to dictate homeostatic function, it is possible to harness these interactions through biomaterials with similar size scale biophysical cues to elicit more native cell phenotypes, thereby mitigating cardiovascular disease progression and enhancing regenerative potential. 7 Fibroblasts represent the largest percentage of cells in the heart and coordinate numerous functions including ECM turnover, cell-cell signaling, and cytokine and growth factor secretion. 8 After MI, broblasts transform into a highly contractile, activated myo broblast phenotype, which functions to stabilize the injury site by increasing ECM deposition to preserve the integrity of the myocardial wall and maintain the pressure generating ability of the heart. 9 While this compensatory process is initially bene cial, problems arise when pro-brotic in ammatory signals such as TGF-β1 persist, leading to continual deposition of stiff scar tissue which ultimately impairs contractility and perfusion within in the heart. 9 We have previously demonstrated the ability to modulate broblast morphology and function using polymeric microstructural cues to achieve less brotic phenotypes, which could have tremendous implications in heart failure therapy. [10][11][12][13] Recent work showed that in vitro treatment with polymeric microrods (15 x 15 x 100 µm) decreased broblast proliferation and that microrod injections in preclinical rodent models of heart failure cause reductions in scar tissue and improvements in cardiac function by in uencing the cardiac microenvironment. 12,13 Bene ts of using hydrogel microstructure strategies with mechanobiological mechanisms of action as therapeutic approaches include being injectable, cell-free, and highly tunable in terms of geometry, stiffness, and material. Further, as their therapeutic effects are restricted based on proximity, concerns related to systemic side effects are avoided. Microrod hydrogels also allows for combination therapies with the ability to load and release various therapeutic factors from the microstructures. 14,15 The ability to devise more potent, multi-faceted therapies can have tremendous implications on myocardial regeneration after MI by addressing multiple pathological processes.
Reperfusion strategies are a key component in the management of acute MI but can lead to ischemia reperfusion injury (IRI), which is widely characterized by oxidative stress, in ammation, intracellular Ca 2+ overload, brosis, and endothelial dysfunction. [16][17][18] Although the exact mechanisms of IRI remain unknown, targeting oxidative damage and subsequent brotic response that occurs after myocardial injury and reperfusion is critical to cardiac recovery after MI. [19][20][21][22][23][24][25][26] Several naturally occurring biological macromolecules within the body possess unique characteristics that may enhance intrinsic wound healing functions after injury. Small leucine rich proteoglycans (SLRPs) are ubiquitous ECM components involved in structural organization and are known regulators of collagen bril assembly. 27 Decorin, a class I SLRP, has been shown to have both anti brotic and antioxidant properties. It has been reported to sequester the pro brotic cytokine TGF-β with high a nity and modulate collagen brillogenesis. [28][29][30][31][32][33][34] Preclinical studies have also demonstrated a therapeutic role of decorin in mitigating brosis in various in vivo models of ischemic injury [35][36][37] and in vitro brosis models. 38-40 Further, it has been reported that there is a protective role of decorin after traumatic injury in vivo which is linked to oxidative stress response. 35,41 In cell studies with high glucose and oxygen/glucose-deprived environments, protective effects of decorin center on involvement in apoptosis and oxidative stress pathways. 42,43 Therefore, there may exist an important role for decorin in early response therapies for cardiovascular injury.

Synthesis of hyaluronic acid methacrylate
Hyaluronic acid methacrylate (HAMA) was synthesized based on a protocol adapted from Bencherif et al. 44 These methods have been previously reported by our lab. 13 Brie y, one gram of sodium hyaluronate (100 kDa) was dissolved at 3.76 mg/mL in a 1:1 solution of deionized (DI) water:DMF (266 mL). After solubilizing, a 73-fold molar excess of GM (24.86 mL, 0.1822 mol) and 26.5-fold molar excess of TEA (9.235 mL, 0.066 mol) with respect to the primary hydroxyl/hydroxymethyl functional group on hyaluronic acid was added to the mixture. The reaction was left to stir for 24 hours at ambient temperature while protected from light. HA and HAMA products were recovered via precipitation in an excess of isopropanol. Brie y, 35 mL of isopropanol was added to 15 mL of reaction solution and then the precipitate was isolated by centrifugation at 1275 × g for 5 min. This process was repeated until all of the reaction solution had been precipitated. The recovered precipitate was subsequently dissolved in 90 mL of DI water. The resulting solution was then dialyzed (3500 Dalton MWCO) against DI water (10 times volume of the solution) for 48 hours with three changes of water. The product was then lyophilized for 3 − 4 days at − 40°C and 65 mTorr and the resulting white powder was then stored at − 20°C until further use. 1 H NMR spectroscopy (Bruker Avance III HD 400 NMR) was used to determine the degree of methacrylation. Methacrylate peaks are observed at ~ 6.5, ~ 5.6, and ~ 1.85 ppm. Degree of methacrylation was calculated based on the ratio of the relative peak integration of the methacrylate peak at 1.85 ppm and HA's acetamide peak which occurs at 1.9 ppm and was determined to be approximately 50.1 ± 6.9% substitution (data not shown).

Hyaluronic acid microrod fabrication
HAMA was dissolved at 75 mg/mL in DI water containing 0.5% w/v of the photoinitiator 2-hydroxy-4′-(2hydroxyethoxy)-2-methylpropiophenone. The solution was stirred for 2 hours at ambient temperature to facilitate solubilization and protected from light. After, the solution was centrifuged at 15000 × g for 5 minutes to remove any impurities. Subsequently, a 15 µm layer of precursor solution was deposited onto an oxygen plasma-or piranha-treated 3" silicon wafer. The wafer was then patterned using a Karl Suss MJB3 or Quintel Q4000 mask aligner by exposing the wafer through a photomask (15 µm × 100 µm features) to a 365 nm UV light source. Crosslinked microrods were then gently removed from the wafer using a cell scraper and collected into DI water, where any uncrosslinked HAMA would fully dissolve. Microrods were then passed through a 150 µm mesh lter to remove any aggregates and then concentrated by centrifugation. The microrods were subsequently sterilized with 70% ethanol for 30 minutes and then resuspended in saline prior to use. Microrod concentration was determined using a hemocytometer by counting the number of microrods that appeared in the nine gridded boxes after pipetting 10 µL of microrod solution into the hemocytometer. The number of counted microrods was then divided by 9, and then multiplied by 10 4 and the dilution factor to get the microrod concentration in the stock solution. The microfabrication process is illustrated in Fig. 1.

Decorin loading of microrods
To passively load microrods with decorin, 750,000 microrods were concentrated in 1 mL of deionized (DI) water and then 333 µL of decorin (600 µg/mL) was added. The microrods were passively loaded via incubation with inversion over four days at 4°C. After incubation, the microrods were centrifuged at 15,000 × g for 10 minutes and the supernatant was removed. The microrods were then recentrifuged at 15,000 × g for 10 minutes once more to concentrate them and remove any residual solution. The microrods were then resuspended in saline to the desired concentration. Decorin loading was assessed via NanoOrange Protein Quantitation Kit (Thermo Fisher, Waltham, MA). Values were validated with an ELISA for human decorin (Abcam, Cambridge, UK) (data not shown). Decorin loading was determined by calculating the amount of free decorin in the supernatant post-centrifugation, and then subtracting that value from the initial amount of decorin added (200 µg), and then dividing that value by 750,000 to get the amount of decorin (µg) per microrod.

Collagen turbidity assay
First, a 1 mg/mL stock solution of collagen type I was prepared by dilution of high concentration collagen type I with PBS and was subsequently kept on ice along with a 96-well plate. Varying decorin concentrations ranging from 5 − 80 µg/mL were then prepared by diluting 600 µg/mL stock decorin solution with PBS. Then, a 1:1 ratio of 1 mg/mL collagen type I and the decorin stocks were added to each well in triplicate resulting in nal decorin doses of 2.5, 5, 10, 20, and 40 µg/mL. Absorbance measurements at 405 nm were then taken every 5 min over the span of 2 hours at ambient temperature using a plate reader. The assay also included blank controls of water and PBS as well as a control for the highest concentration of decorin (40 µg/mL).

Cell culture & qCPR
NIH 3T3 mouse broblasts (ATCC, Manassas, Virginia) were cultured in Dulbecco's modi ed Eagle's medium with 10% fetal calf serum and 1% penicillin/streptomycin. Cells were seeded into 24-well plates and then media was added that either contained only 10 ng/mL TGF-β1 or 10 ng/mL TGF-β1 plus 10 µg/mL decorin. Genetic material was harvested and puri ed using the RNeasy Mini Kit (Qiagen, Hilden, Germany). RNA was converted into cDNA using the iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA) and a Viia7 qPCR machine (Life Technologies, Carlsbad, CA) was used to measure relative expression levels of gene targets compared to the housekeeping gene 60S ribosomal protein L19 (RPL19). Expression levels of all genes were evaluated using the Fast SYBR Green Mastermix (Life Technologies, Grand Island, NY) and custom DNA primers (Integrated DNA Technologies, Coralville, IA) in triplicate for three biological replicates (Table S1).

Infarct model and microrod injections
The animal protocol for MI induction was approved by the Committee for Animal Research of the University of California, San Francisco and was performed in accordance with the recommendations of the American Association for Accreditation of Laboratory Animal Care. The ischemia reperfusion (I/R) MI model used in this study has been extensively tested and successfully used in our labs. 12,13,45,46 To produce the MI model, male Sprague-Dawley rats (200-225g) underwent occlusion of the left anterior descending coronary artery for 30 minutes followed by reperfusion while under general anesthesia which was achieved by inhalation of 2% L/min iso urane. 45 The chest was then sutured closed, and the animal was allowed to recover. After 3-4 days, the rats were randomized to saline-injected, microrod-injected, decorin microrod-injected, or free decorin-injected treatment groups (Table 1), and were given one intramuscular injection into the myocardial wall via ultrasound guided transthoracic injection using a 29gauge syringe under blinded conditions. 12,13,45,47 Each injection consisted of 50 µL of sterile 0.9% sodium chloride solution (n = 5), 50 µL of 12 µg of decorin in 0.9% sodium chloride solution (n = 4), 50 µL of 50,000 microrods in 0.9% sodium chloride solution (n = 7), or 50 µL of 50,000 decorin microrods in 0.9% sodium chloride solution (n = 10). Successful injection to the center of the infarct region was con rmed by a local increase in ultrasound signal and brief thickening of the left ventricle (LV) wall near the tip of the syringe. Transthoracic echocardiography was performed with a 15-MHz linear array transducer system (Vevo 3100LT, FUJIFILM VisualSonics, Ontario, Canada) on all animals while under general anesthesia (2% L/min iso urane). Echocardiography was performed prior to injection on day 3 or 4 post-MI and eight weeks post-injection using standard methods that have been performed reproducibly in our lab. 12,13,45,46 To determine left ventricular end diastolic volume (LVEDV), left ventricular end systolic volume (LVESV), and ejection fraction (EF) at 3-4 days and eight weeks, the left ventricular endocardium was outlined in both the end-systolic and end-diastolic phase and the single plane area length algorithmic method was applied. Two-dimensional images were obtained in both parasternal long-and short-axis views at the papillary muscle level. Stroke volume (SV) was calculated by SV = LVEDV-LVESV, while change in EF was calculated by ΔEF = EF 8 weeks -EF 3 − 4 days . All image and subsequent analyses were performed in a blinded manner. In situations where the ventricular endocardium was not clearly identi able in the 3-4 day or eight-week image, the animal was excluded from echocardiographic analyses. Cases where ejection fraction was above 45% at 3-4 days were excluded because they indicated an insu cient infarct model.

Histology
Sacri ce was performed after eight weeks after performing endpoint echocardiography. The animal was maintained at 5% L/min iso urane for ve minutes, followed by bilateral thoracotomy and injection of potassium chloride into the right atrium to arrest the heart in diastole. The heart was then extracted and frozen in OCT (Sakura Finetech USA, Inc., Torrance, CA) on 2-methylbutane (Sigma Aldrich, St. Louis, MO) on dry ice and sectioned for histology, immuno uorescent imaging, and image analysis. Tissue blocks were cryo-sectioned at a thickness of 10 µm starting at the apex of the LV, collecting 10 serial sections every 350 µm until a total of 100 sections were collected. Sections were stained with H&E and Sirius red using standard protocols. Brie y for Sirius red staining, frozen sections were brought to room temperature, soaked in xylene for a total of 15 min., rehydrated using an ethanol series (100%, 95%, and 80%) for 1 min. each, and then soaked in water. Sections were then incubated in 0.01% Fast Green FCF solution in saturated picric aqueous solution for 1 hr. followed by a 1 hr. incubation in 0.04% Fast Green FCF/0.1% Sirius red in saturated picric acid solution. Next, sections were dipped in acidi ed water, 100% alcohol, and then xylene prior to mounting. For H&E staining, slides were soaked in tap water for 5 min. to remove OCT from the sections. They were then stained in Harris's Hematoxylin for 5 min., followed by dipping in water for 1 min., and then dipping in Differentiating Solution for 1 second for a total of 3 times.
Next, slides were immersed in Bluing Solution (0.1% sodium bicarbonate, pH ~ 8.0) for 30 seconds, followed by dipping in water for 1 min. Slides were then soaked in 70% ethanol for 1 min. and then in Eosin Y for 45 sec. Sections were then dehydrated using an ethanol series (85%, 95%, 100%) for 1 min. each before soaking in xylene and mounting. For immuno uorescent stains, tissue sections were air dried and blocked with 10% serum, followed by incubation with primary and secondary antibodies using standard protocols. In brief, samples were xed in ice cold acetone for 5 min., blocked with 10% serum for 20 min., and incubated overnight with primary antibodies at 4°C in a solution of 0.05% Tween-20, 10% serum, and 1% BSA in PBS (anti-sarcomeric alpha actin 1:500, anti-alpha smooth muscle actin 1:500). After washing, secondary antibody and WGA (1:100) was added for 45 minutes at room temperature. Hoechst (1:500) was added for 5 min at room temperature to visualize nuclei.

Imaging and quanti cation
For whole heart histology, images were taken using a Nikon 6D optical microscope (NIKON Instruments, Inc., Melville, NY) using 4-40x objectives (H&E and Sirius red). Subsequent quanti cations were performed using custom scripts and ImageJ.
For collagen analysis, ve alternating sections of each heart were selected from throughout the coronal plane of the heart and stained with Sirius red to assess the quantity and density of collagen in the injured hearts. These sections were imaged under bright eld as well as under cross-polarized light to visualize the collagen bers using a 4x objective. Representative images for bright eld and polarized shading corrections were taken prior to the start of each imaging session. Infarct area and intensity was quanti ed in the LV, which included the septum. Intensity measurements in the regions of interest were determined using ImageJ.
For wall thickness analysis, ve measurements across the LV free wall were recorded for each tissue section where the LV free wall was distinguishable. The average length across all analyzed sections in each heart sample was reported.
For cardiomyocyte cross-sectional area analysis, three sections (apex, middle, and end of the heart) per heart (n = 4 per group) were selected and stained with anti-sarcomeric alpha actinin, WGA, and Hoechst. Four high-magni cation (40x objective) images of the border zone and remote regions in each section, for each animal were averaged to obtain the cross-sectional area, cells per area, eccentricity, major axis length, and minor axis length. Speci cally, WGA-stained sections aided in the determination of number of cells per area and were quanti ed in Python (3.9.7) using OpenCV (4.5.5.64) and scikit-image (0.18. 3) packages to conduct watershed thresholding of cell bodies and subsequent area measurements. 48, 49 Images were inverted and then binarized using an adaptive mean threshold before applying the watershed segmentation algorithm. The area of segmented cells, cell count per area, eccentricity, major axis length, and minor axis length were then computed. Segmented images were then manually checked to ensure good quality segmentation, with those showing a high error rate excluded (e.g., high fractionation of cells into multiple segments, doublet or multiplet cells recognized as one cell body, membrane stains segmented as cells). For each image, the average cell area, number of cells per µm 2 , eccentricity, major axis, and minor axis are reported.
For vascular density analysis, three sections (apex, middle, and end of the heart) per heart (n = 4 per group) were selected and stained with anti-alpha smooth muscle actin and Hoechst. Four highmagni cation (40x objective) images of the remote, infarct, and border zone regions in each section, for each animal were averaged to obtain the number of arterioles per eld. Only vessels that stained positively for alpha smooth muscle actin and possessed a visible lumen were included in the analysis.

Statistical analysis
All data are presented as the mean ± standard deviation unless otherwise indicated. In vitro analysis was performed using a student's t-tests to identify statistical differences between two groups. One-way analysis of variance (ANOVA), followed by the Tukey's multiple comparisons test was used to identify differences between three or more groups, unless otherwise stated. Statistical comparisons made between groups to assess differences in EF, LV end systolic volume (ESV), and LV end diastolic volume (EDV) at eight weeks post-MI were made using a Two-way repeated measures ANOVA followed by the Tukey's multiple comparisons test.

Decorin treatment reduces myo broblast-like gene expression and inhibits collagen brillogenesis
To con rm anti-brotic biologic effects of decorin, NIH 3T3 broblasts were stimulated with 10 ng/mL TGF-β1 and cultured in the presence or absence of 10 µg/mL decorin. Relative gene expression results from RT-qPCR using primers speci ed in Table 1 indicate stark reductions in ACTA2 (p < 0.0001), COL1A2 (p < 0.01), COL3A1 (p < 0.001), and MMP2 (p < 0.05) expression in decorin-treated samples compared to control ( Figure S1A-C). These results demonstrate decorin's ability to sequester TGF-β1 and prevent activation of its downstream signaling pathways that result in myo broblast-like phenotypes in broblasts.
Turbidity was used to assess the degree of collagen brillogenesis. A broad range of decorin concentrations (2.5-40 µg/mL) were used to assess dose-dependent effects of decorin on collagen ber formation. It is evident that bril formation kinetics are dependent on decorin concentration, and while all doses of decorin tested signi cantly inhibited brillogenesis, the highest concentration of 40 µg/mL yielded the greatest impact ( Figure S2, p < 0.01).

Sustained release of decorin is achieved from microrods
Next, we examined the loading and in vitro release dynamics of decorin from microrods. After 4-days of passive loading, decorin microrods exhibited a loading e ciency of 98%. Release kinetics were assessed in PBS solution using aliquots of 50,000 microrods containing ~ 12.5 µg of decorin. Within the rst 10 hours, decorin exhibited burst release from the microrods with more than 5.7 µg being released into solution. Following this burst release, decorin release exhibits a plateau in release from hours 24 to 168 (Fig. 2). Overall, the total hourly amount of eluted decorin decreased with time indicating concentration dependent ( rst-order) release kinetics from the decorin microrods. However, decorin elution was readily detectable and quanti able over 31 days. Morphological characteristics including size and shape of the microrods remained comparable before and after loading (data not shown).

Effect of decorin microrods in ischemia-reperfusion myocardial infarction in vivo models
Decorin was loaded into microrods to achieve a more potent anti-brotic strategy that leveraged biophysical regulation of microstructures and the ability of decorin to sequester TGF-β1 in the post-infarct environment. Microrods were fabricated as previously described and loading according the scheme in Fig. 1. 13 From three independent studies, decorin loading was calculated to be approximately 10.7 µg of decorin per 50,000 microrods as measured via protein quantitation (data not shown). A rodent model of ischemia-reperfusion myocardial infarction was utilized to generate a cardiac brosis model. All treatments-saline, microrods, decorin microrods, or free decorin were delivered into the infarct via ultrasound-guided, intramyocardial injection. For subsequent animal studies, we chose to use the higher end of observed loading (~ 12 µg per 50,000 microrods) as the injection dose for the free decorin treatment group that served as a positive control (Table 1).
Microrod and decorin microrod treatment were shown to result in signi cant improvements in cardiac function (Fig. 3). When comparing baseline EF to EF 56 days post-MI in Fig. 3A, while saline-treated animals experienced a decline in function (p = 0.0707), microrod-treated animals did not experience a decline, and decorin microrod-treated animals exhibited a signi cant increase in EF (p < 0.001). Change in EF was assessed by taking the difference in EF measured at Day 3-4 post-MI and at Day 56 post-MI (Fig. 3B). Rats treated with microrods (2.12% ± 4.35%) or decorin microrods (5.21% ± 4.29%) exhibited better change in EF after 56 days than rats treated with saline (-4.18% ± 2.78%) or free decorin (-3.42% ± 1.86%). Since decorin microrods performed better than microrods and free decorin, this suggests that the ability to locally retain decorin at the infarct site through the use of the microstructures was crucial to improving the e cacy of the microrod strategy. Rats treated with decorin microrods demonstrated a signi cantly better change in EF at 56 days post-MI compared to both saline and free decorin (Fig. 3B, p < 0.001 and p < 0.01, respectively). Ventricular remodeling was also improved in rats treated with decorin microrods as evidenced by signi cant reduction in ESV (Fig. 3C, p < 0.05, p < 0.05, p < 0.001, respectively) and importantly EDV compared to rats treated with saline, microrods, and free decorin (Fig. 3D, p = 0.0505, p < 0.01, p < 0.01, respectively). Both microrods and decorin microrods led to a better change in stroke volume than the saline and free decorin treatments after 56 days post-MI. Microrod-treated animals had signi cantly higher changes in stroke volume compared to saline and free decorin groups (p < 0.05 and p < 0.01, respectively) while decorin microrods showcased a trend towards increased change in stroke volume when compared to both vehicle and free decorin (Fig. 3E, p = 0.1772, p = 0.0595, respectively).
Histological evaluation of each treatment group was performed using H&E and Sirius red to assess LV wall thickness and degree of brosis (Fig. 4A). Wall thickness measurements were performed on all sections throughout the coronal plane of the heart where the LV cavity was distinguishable. Animals treated with decorin microrods (2076 ± 399 µm) exhibited trends in having greater LV wall thickness compared to those treated with saline (1711 ± 350 µm) and free decorin (1480 ± 195 µm) as seen in sections throughout each heart under cross-polarized imaging. Rats with no MI (p < 0.05) and rats treated with microrods (p < 0.05) and decorin microrods (p < 0.05) all had signi cantly reduced intensity of collagen staining in the LV compared to those treated with saline (Fig. 4C). While not statistically signi cant, treatment with free decorin (p = 0.0726) also decreased average collagen intensity in the LV compared to treatment with saline.
We also investigated the impact of decorin microrods on cardiomyocyte and endothelial cell behaviors.
Given that microrods were observed to be retained in cardiac tissue for at least 8 weeks ( Figure S3), we hypothesized that there could be additional long-term bene ts bestowed by the presence of these structures. A crucial compensatory mechanism after myocardial injury is hypertrophic growth of cardiomyocytes. 50 Hypertrophy was assessed by immuno uorescence staining for sarcomeric alpha actinin, cell membrane, and nuclei in the border zone and remote zone (Fig. 5A). Three sections from throughout the heart were assessed (n = 4 animals per group). While no differences in cardiomyocyte area were observed between groups in the border zone, reduced cardiomyocyte area in the remote zone was identi ed in decorin microrod (p < 0.05) and free decorin-treated animals (p < 0.01) compared to saline-treated animals ( Fig. 5B-C). Accordingly, we observed an increase in cardiomyocytes per area in the remote zone for decorin microrod (p < 0.05) and free decorin (p < 0.05) groups compared to the saline group ( Fig. 5D-E). Further morphometric analysis of cardiomyocytes indicated the observed smaller areas in decorin microrod and free decorin groups were due to proportional reduction in dimensions of the cells as no change in eccentricity was found ( Figure S4). Studies were performed to identify if an increase in vascular density was also responsible for the observed improvements in cardiac function brought about by the decorin microrod treatment. Arteriole number was assessed by immuno uorescence staining for alpha smooth muscle actin and nuclei in the infarct, border zone, and remote zone (Fig. 6A). Three

Discussion
In this work, we investigated the capacity for microrods to locally deliver decorin to the infarct and modulate the post-infarct environment to dampen pathophysiological responses that occur post-MI. Our results demonstrate that decorin-loaded microrods release decorin for up to a month and can signi cantly improve cardiac function and attenuate deleterious ventricular dilatation, cardiac brosis, and hypertrophy in chronic models of ischemia-reperfusion MI.
An exciting area in therapeutic biomaterials research focuses on leveraging the mechanosensing machinery of cells to elicit speci c phenotypes and behaviors. 7,51,52 The ability to reliably dictate cellular responses can be a powerful tool in addressing pathological responses that result from cardiovascular diseases, such as heart failure. 7 While injections of bulk polymers may provide the myocardium with mechanical support to improve cardiac function and delay LV dilatation, biophysical regulation via microand nanostructure cues can lead to reprogramming of resident cells to create microenvironments that are more amenable to tissue repair and regeneration. 51 Our lab has previously shown the utility of high aspect ratio microstructures made from polypropylene and polyethylene glycol (PEG) to discourage myo broblast transition. 11,12 However, developing therapeutic strategies should aim to employ materials that have biocompatibility, exhibit biodegradation, and possess tunable properties for facile translation to the clinic. As such, next generation microrods were fabricated from HA, a naturally occurring polysaccharide that is implicated in wound healing responses. [53][54][55] Advantages of employing bioactive and biodegradable materials such as HA as opposed to bioinert materials include the potential for longacting therapeutic bene ts -as HA degrades, the released oligosaccharide degradants can stimulate angiogenic processes. 55,56 HA microrods were shown to outperform other materials in attenuating brotic response in vitro and in improving cardiac function in in vivo models of ischemia-reperfusion MI. 13 These observed bene ts are likely due to additional biochemical effects bestowed by the HA material itself.
Not only do polymeric microrods have the capacity lend mechanical strength to the myocardial wall and modulate broblast behaviors through mechanical regulation, but they can be formulated to release bioactive factors. We have successfully utilized PEG microrods to locally deliver E-domain peptide and β-NGF to injured tissue in myocardial infarction and tibial fracture models. 14,15 To enhance e cacy of our microrod platform for applications in post-MI therapy, we investigated the ability for microrods to locally deliver decorin, a SLRP with both anti brotic and antioxidant attributes. The core protein of decorin has two binding sites for TGF-β1 and binds collagen through its leucine rich repeat region. 57,58 Several studies have implicated decorin in playing an important role in post-infarct remodeling and others have demonstrated how decorin treatment can mitigate adverse outcomes in various models of cardiovascular disease. 35,37,[59][60][61] The results from our described in vivo experiments support these observations about the ability of decorin to improve cardiovascular outcomes after injury. While microrods improved change in ejection fraction after 56 days post-MI, decorin microrods caused the greatest improvement in cardiac function compared to all other groups, including saline and free decorin. The fact that decorin microrods performed better than microrods and free decorin individually, points to the importance of the decorin being retained in the infarct region by the microrods to achieve the optimal therapeutic effects from the combination strategy (Fig. 3). The improvements in cardiac function observed in both microrod and decorin microrod groups may likely be due to reduced LV wall stress due to the mechanical support provided by the polymer microrod injection. This is in line with what was observed in other studies where LV function improved in response to reduced LV volume and wall stress due to increased LV wall thickness resulting from biopolymer injection. 62 Similarly, in our study decorin microrods showed a trend towards increased average LV wall thickness compared to saline and free decorin groups (Fig. 4B). Based on in vitro release data (Fig. 2) and the fact that HA microrods remain visually intact after 8 weeks within the myocardium ( Figure S3), it is expected that decorin is primarily released passively from the microrods as opposed to from microrod degradation.
The presence of HA microrods after 8 weeks is not surprising as it has been shown that methacrylated HA-based hydrogels, such as our microrods, degrade enzymatically as the hydrolytic degradation sites are sterically hindered. 63 Hyaluronidases are present at low amounts in the heart both before and after injury 64 , explaining the slow degradation of these microrods. However, our group and others have shown that when exposed to higher concentrations of hyaluronidases, these HA hydrogels will readily degrade. 13,65,66 The presence and stability of HA microrods in the myocardium may pose bene ts to longterm myocardial remodeling. Previous work has shown that similar HA hydrogels improved maintenance of wall thickness, compared to more rapidly (hydrolytically) degradable HA hydrogels with similar mechanical properties otherwise. Further, these stable enzymatic HA hydrogels elicited less in ammatory cues from the surrounding cells compared to their degradable counterparts. 63 The immunomodulatory role that these HA microrods have within the infarct microenvironment still requires further examination, but similar HA hydrogels have been reported to elicit minimal in ammation and favorable cytocompatibility. 67 The long-term presence of HA microrods may play a bene cial role within the myocardium remodeling, as the extended stability of similar HA gels has led to prolonged reduction in LV volume compared to degradable systems. 63 We have also previously seen that HA microrods do not affect the contractility of cardiomyocytes. 13 The long-term effects of HA microrods and similar HA hydrogel systems in the myocardium have not been well explored and requires further longitudinal studies to understand the myriad of cell-material and cell-cell interactions that HA systems affect.
While it was apparent that bulk cardiac metrics are enhanced in both microrod groups compared to vehicle and free decorin groups, the incorporation of decorin was hypothesized to also bestow favorable changes at the cellular level to achieve a microenvironment that is conducive to wound healing. Similar to ndings from Faust et al. in studies utilizing decorin gene transfer, both decorin microrod and free decorin groups had decreased brosis (Fig. 4C) and cardiomyocyte hypertrophy compared to saline (Fig. 5). 36 Interestingly however, only rats treated with decorin microrods exhibited improved cardiac function and ventricular remodeling outcomes (Fig. 3). Given the differences in therapeutic outcomes regarding ejection fraction and LV volumes after 56 days post-MI between the two decorin groups, it points to the presence of microrods as playing a role in the improved treatment e cacy. This additional therapeutic bene t bestowed by the microrods may be attributed to bulking of the ventricular wall as well as improved vascularization given the given the differences in vascular density between both microrod groups and the free decorin group (Fig. 6). As HA degrades, oligosaccharides are released which are known to promote angiogenic endothelial processes including proliferation, migration, and tube formation. 55,56 However, because the injection occurred at the center of the infarct where analysis of capillary density could be subject to high variability due to confounding variables such as tissue distortion and compression, we opted instead to investigate arteriole presence as a surrogate marker of downstream vascularization. 68 Thus, the presence and subsequent solubilization of microrods in the infarct zone may promote vessel formation. However, further investigations will need to be performed to provide more insight into this phenomenon.
While our results are promising, we recognize that there may exist some limitations to the work described here. The animal studies shown utilized male Sprague-Dawley rats to ensure that consistent models of HF were achieved. Of note, it has been documented that there may exist differences in post-MI left ventricular remodeling based that is in uenced by sex. Prior studies have demonstrated favorable remodeling processes in females compared to males after myocardial injury that may be due to the ability to retain advantageous myocardial properties, such as reduced myocyte apoptosis and hypertrophy. 69-73 These sex-related differences in remodeling responses post-MI may also explain why in the prior investigation using all female rats, the HA microrod group appeared to have a more pronounced improvement in cardiac performance compared to what was observed in the microrod groups in this study. 13 To better account for differences in therapeutic response based on sex, future studies involving both male and female rats are necessary. Additionally, while power analysis shows that we were able to achieve signi cant power in the current study, future work will bene t from larger sample sizes. By identifying the relevant pathways and physiological responses that are affected by these microstructures, it will be possible to optimize this therapeutic strategy to achieve more holistic myocardial repair after injury.

Conclusion
In this study, we have demonstrated the capacity for polymeric microstructures made of hyaluronic acid to achieve sustained delivery of the anti-brotic agent decorin which directly translated to therapeutic cardiac outcomes in preclinical models of I/R MI. Further, we have showcased the ability for a dual, biochemical and biophysical therapeutic strategy to synergistically improve cardiac and ventricular remodeling outcomes while attenuating collagen brosis and cardiomyocyte hypertrophy experienced following myocardial injury. Therefore, the use of decorin microrods represents a promising and novel translational strategy for cardiac treatment after MI. Applications of this biophysical platform in additional clinically relevant models including bone healing, cirrhosis, and implantable devices represents exciting prospects to further advance our ability to facilitate wound healing and tissue repair.  Microrod fabrication scheme and loading with decorin. HA was rst modi ed with a photosensitive handle to generate hyaluronic acid methacrylate (HAMA) polymer that is amenable to crosslinking via exposure to UV light in the presence of the photoinitiator 2-hydroxy-4′-(2-hydroxyethoxy)-2methylpropiophenone. HAMA was deposited onto an oxygen plasma-or piranha-treated wafer and exposed to UV light through a photomask patterned with xed rectangular geometries of 15 μm width by 100 μm length. After exposure, the polymer lm was then developed in water to isolate the crosslinked Page 23/29 microrods. Microrods were loaded with decorin via incubation with inversion at 4°C over 4 days and then subsequently puri ed via centrifugation.

Figure 2
Decorin release pro le from microrods. Hourly release rate (ng) of decorin from 75 mg/mL microrods over a 31-day period shown (n = 3). It is apparent that decorin microrods exhibit burst release of decorin within rst 10 hours and subsequently plateau, following rst-order release behavior. The data is presented as mean +SD. Treatment with decorin microrods improves cardiac outcomes and reduces maladaptive ventricular remodeling. Echocardiography was used to compare ejection fraction (EF) at 3-4 days after infarct and at 56 days after infarct in rats that had no MI performed (n = 5) and rats with MI that were treated with saline (n = 5), microrods (n = 7), decorin microrods (n = 10), and free decorin (n = 4). (A) The average ejection fraction (EF) at 3-4 days post-MI and EF at 56 days post-MI for each group is plotted. While saline animals show a trend for decreased EF, decorin microrod-treated animals show signi cant improvement in EF over the 8-week time period. (B) Rats treated with decorin microrods and microrods had a signi cantly higher change in EF compared to saline-treated animals. Rats treated with decorin microrods also had a signi cantly higher change in EF compared to those treated with free decorin.
Echocardiography was used to evaluate (C) end systolic volume and (D) end diastolic volume at 3-4 days after infarct and at 56 days after infarct in all experimental groups. End systolic volume and end diastolic volume were signi cantly reduced in rats treated with decorin microrods compared to rats treated with saline, microrods, or free decorin. (E) Both microrod-and decorin microrod-treated animals exhibited improved change in stroke volume compared to saline and free decorin treatments. The data are presented as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. distinguishable. Rats treated with decorin microrods showed a trend towards increased LV wall thickness compared to the free decorin group. (C) Tissue sections were stained with Sirius red and visualized under cross-polarized light. Intensity of collagen staining in the LV (including the septum) was measured.