Enzyme‐Responsive Nanoparticles for Dexamethasone Targeted Delivery to Treat Inflammation in Diabetes

Diabetes is a global epidemic accompanied by impaired wound healing and increased risk of persistent infections and resistance to standard treatments. Therefore, there is an immense need to develop novel methods to specifically target therapeutics to affected tissues and improve treatment efficacy. This study aims to use enzyme‐responsive nanoparticles for the targeted delivery of an anti‐inflammatory drug, dexamethasone, to treat inflammation in diabetes. These nanoparticles are assembled from fluorescently‐labeled, dexamethasone‐loaded peptide‐polymer amphiphiles. The nanoparticles are injected in vivo, adjacent to labeled collagen membranes sub‐periosteally implanted on the calvaria of diabetic rats. Following their implantation, collagen membrane resorption is linked to inflammation, especially in hyperglycemic individuals. The nanoparticles show strong and prolonged accumulation in inflamed tissue after undergoing a morphological switch into microscale aggregates. Significantly higher remaining collagen membrane area and less inflammatory cell infiltration are observed in responsive nanoparticles‐treated rats, compared to control groups injected with free dexamethasone and non‐responsive nanoparticles. These factors indicate improved therapeutic efficacy in inflammation reduction. These results demonstrate the potential use of enzyme‐responsive nanoparticles as targeted delivery vehicles for the treatment of diabetic and other inflammatory wounds.


Introduction
Diabetes is one of the most prevalent epidemics in human history, [1] having a major impact on patient health, with a high risk for severe complications, [2] such as increased susceptibility for delayed wound healing, [2] and deterioration of inflammatory and infection-related pathologies. [3]A common manifestation of most diabetic complications is an exaggerated and prolonged inflammatory response, including excessive pro-inflammatory cytokine production together with increased numbers of immune cells. [4]Increased levels of inflammation, as evidenced either by the cellular make-up and/or the molecular repertoire, have been widely described [5] in diabetic human patients [6] and in animal models. [7]urthermore, many diabetes drugs possess anti-inflammatory properties. [8]Indeed, inflammation is the natural response of the body to injury or infection.However, chronic and exaggerated inflammation, involving constant or repetitive activation of the immune system, may lead to tissue damage and organ dysfunction. [9]Current systemic medications usually require repeated dosing due to limited drug circulation half-life and poor tissue penetration, [10] leading to patient noncompliance and enhanced risk for dose-dependent off-target effects.
Diabetes is a risk factor for delayed wound healing, with an increased incidence of persistent infections showing elevated resistance to standard treatments. [11]Therefore, to improve treatment efficacy, novel methods to target therapeutics to the inflamed tissues in diabetic patients are required. [11,12]These might be further applied to the development of novel treatment strategies for diabetes complications.As part of the exaggerated, dysregulated inflammatory state, many of the diabetic complications are associated with increased expression and production of matrix metalloproteinases (MMPs). [13]Recently, the use of enzymedirected assembly of particle therapeutics for minimally invasive delivery of drugs to inflamed tissues has been reported. [14]hese materials are assemblies of peptide-polymer amphiphiles (PPAs) prepared via graft-through ring-opening metathesis polymerization (ROMP). [15]14a,b,16] When the PPAs experience a solvent switch from organic to aqueous solutions, the larger cone angles from the hydrophilic peptide brush give higher surface curvature structures, driving the formation of micellar nanoparticles (20 nm in diameter), which provide the opportunity for intravenous (IV) or local administration.17d-e] We envision that the application of enzyme-responsive nanoparticles for targeted and sustained drug delivery in diabetes models bears the potential to improve treatment efficacy by targeting anti-inflammatory drugs in response to the inflammation itself.
Dexamethasone (DEX) is a common glucocorticoid medication used for many years for the treatment of a wide variety of diseases and conditions, mainly for its glucocorticoid effects as an anti-inflammatory and immunosuppressant agent. [18,19]Recently, the National Institutes of Health (NIH), and others, [20] recommended the use of corticosteroids such as DEX to treat certain COVID-19 patients. [20,21]However several severe side effects have been reported, such as hyperglycemia, psychiatric effects, avascular necrosis, or increased risk of opportunistic infections, especially during the use of systemic corticosteroids. [21,22]Thus, in order to utilize the therapeutic effect of DEX, and at the same time minimize its systemic side effects, there is a great need to develop novel delivery methods to target the DEX to the diseased tissues.
Therefore, we hypothesized that by incorporating DEX into the PPA scaffold and subsequently the hydrophobic core of the micellar nanoparticles formation via cleavable covalent ester bonds, the anti-inflammatory effect of the incorporated DEX would be improved compared to the drug alone, by promoting targeted and sustained drug release following a single dose in a diabetic rat model.
Herein, we aimed to study the delivery, accumulation, and therapeutic effect of fluorescently-labeled DEX-incorporated enzyme-responsive nanoparticles in collagen membranes implanted in the sub-periosteal calvaria of diabetic rats.In this model, the degree of collagen membrane resorption following implantation is directly correlated to the level of inflammation in the surrounding tissues. [23]Thus, it provides a quantitative approach to histologically evaluate the residual collagen contents within the implanted membrane, correlated to the degree of inflammation, and monitor an eventual change in the efficacy of inflammation reduction.The responsive nanoparticles described in this work display strong accumulation and retention around the membranes at both 7-and 14-days post-injection and show enhanced anti-inflammatory effects compared to the no-injection, free-DEX, and non-responsive nanoparticles control groups.We expect that these studies will facilitate the development of novel efficient targeted drug delivery methods to inflamed tissues, thus improving treatment efficacy and reducing negative side effects.
14a,c] In addition to the responsive L-and non-responsive D-MMP peptide substrate, a negative control peptide was designed in which the MMP recognition sequence PLGLAG was replaced with GS-GSGS.The resulting control peptide maintained good water solubility and had the same net charge of −1 at pH 7 as the MMP peptide substrate.To confirm MMP responsiveness, monomers functionalized with L-, D-MMP peptides and control peptide (NorMMP L , NorMMP D , and NorMMP C , respectively) were incubated with MMP-9 at 37 °C for 24 h in Dulbecco's phosphatebuffered saline (DPBS) buffer.The resulting mixtures were analyzed by high-pressure liquid chromatography (HPLC) with a UV detector to monitor peptide cleavage.For reference, the cleaved peptide sequence LAGGWGERDGS was prepared and analyzed.As shown in Figure S6 (Supporting Information), NorMMP L was completely cleaved by the enzyme, giving rise to an elution peak similar to that observed for the cleaved sequence LAG-GWGERDGS.ESI-MS analysis was further employed to confirm the mass.In comparison, both NorMMP D and NorMMP C remained intact post-MMP-9 treatment, demonstrating the desired non-responsiveness of the control versions.Next, a test copolymerization of NorDEX with NorMMP L was performed.Chain extension was completed within 4 h as confirmed by the molecular weight (MW) increase identified using size exclusion chromatography coupled with multi-angle light scattering (SEC-MALS) and the disappearance of the NorMMP L olefin signal by NMR (Figure S7, Supporting Information).The same monomer-to-catalyst ratio and reagent concentrations were used for PPA preparation to maintain similar polymerization kinetics.

Synthesis and In Vitro Analysis of DEX-Incorporated Nanoparticles
Cy5.5-labeled DEX-incorporated peptide-polymer amphiphiles PPA L (including NorMMP L ), PPA D (including NorMMP D ), and PPA C (including NorMMP C ) were synthesized by sequentially polymerizing the DEX, peptide, and dye functionalized norbornene monomers (NorDEX, NorMMP, and NorCy5.5, respectively) via graft-through ROMP (Figure 1A).The reaction was terminated by adding ethyl vinyl ether (EVE).Aliquots post NorDEX and NorMMP incorporation were removed and terminated with EVE before SEC-MALS analysis to determine polymer MW (Figure 1B).As shown in Figure 1C, the measured MWs were in agreement with the theoretical values, confirming a complete monomer-to-polymer conversion.The resulting PPAs were then dissolved in DMSO and subsequently solvent was switched into DPBS buffer.The solutions were concentrated by spin filtration to give 300 μM relative to PPAs by measuring Cy5.5 UV absorbance (Figure S8, Supporting Information).Transmission electron microscopy (TEM) analysis confirmed that all PPAs assembled into micellar nanoparticles with an average diameter of 15 nm (Figure 1D,F,H).
Upon treatment with the Zn-metalloproteinase thermolysin, which serves as a robust alternative to MMP with higher in vitro activity, the nanoparticles containing an L-MMP peptide shell (NP L ) underwent cleavage-induced aggregation (Figure 1E), while the ones with D-MMP peptide or control peptide (NP D and NP C , respectively) maintained their nanoparticle morphology as expected (Figure 1G,I).

DEX-Incorporated Nanoparticles in a Rat Diabetes Model
Given the promising in vitro results, all three types of nanoparticles were applied in a rat diabetes model.The study was comprised of 51 Sprague-Dawley rats, weighing 240-260 g.23a,b] The successful hyperglycemia induction was confirmed by a blood glucose level >180 mg/dL at 7 days after streptozotocin injection (Figure S9A, Supporting Information).The biotin-labeled collagen membrane was then implanted under the periosteum covering the parietal bone of the calvaria anterior to the ears line (Figure S9B), soft tissues were then repositioned and sutured.Following, rats were divided into five groups, each receiving a single dose of: 1) NP L , 2) NP D , 3) NP C , 4) free DEX, or 5) no injection through a subcutaneous injection posterior to the ears line (n = 12 for each treatment group and n = 3 for the no injection control group) (Figure 2A,B).For each treatment group, 260 μg of DEX (either free or conjugated to the polymer) was used to provide a 1 mg mL −1 dosage similar to previously tested values. [25]EX-incorporated nanoparticles were tested for their ability to respond to inflammation and elicit an anti-inflammatory function in vivo.Due to the difference in enzyme responsiveness, we expected these nanoparticles to show different retention times in the diseased tissue with the order: NP L > NP D ≈ NP C to elicit different levels of anti-inflammatory effects.23b] Accordingly, previous reports have shown that streptozotocin-induced hyperglycemia in rats results in enhanced degradation of collagen membranes, [23a,e] which has been linked to increased blood vessel formation and macrophage infiltration.We hypothesized that if DEX-incorporated nanoparticles possess an anti-inflammatory effect in vivo, they should slow-down collagen membrane degradation, leading to a higher level of remaining collagen within the membranes, together with less cellular infiltration and blood vessels compared to the control groups.Biotin labeling of the collagen membranes prior to implantation enables their subsequent staining with horseradish peroxidase-conjugated streptavidin, and provides a quantitative histological approach to evaluate the residual collagen amount of the membranes. [24] At 7 days post-injection, trace fluorescence was detected in the rats receiving free DEX (Figure 2C,D).In comparison, strong fluorescence was observed around the implanted membranes instead of at the site of injection in rats receiving NP L treatment (Figure 2E,F).This finding confirms the targeted accumulation of the material at the site of inflammation.At both 7-and 14days post-injection, the in vivo fluorescence intensity in NP Ltreated rats was significantly higher than in the NP D and NP C control groups (Figure 2G, Figure S10).The order of fluorescence intensity NP L > NP D ≈ NP C is in accordance with the level of MMP responsiveness of the incorporated peptides, all significantly higher than the free DEX.Inflamed tissue typically has enhanced permeability due to leaky vasculature.Without the morphology change, the non-responsive nanoparticles enter and exit the tissue through passive diffusion.As shown in Figure 2G, the fluorescence accumulation in the inflamed tissue at days 7 and 14 post NP D and NP C injection was significantly lower than the level observed for NP L , suggesting gradual clearance of the nonresponsive NPs.
Tissue staining and histological analysis showed a higher collagen content within the membranes implanted in the NP L -treated rats compared to the free DEX and no injection groups (Figure 3; Figure S11, Supporting Information).The residual membrane area post NP L treatment was significantly larger than in the notreatment and free DEX groups, displaying 72.78% ± 1.20% of residual collagen content, compared with 23.70% ± 3.62% and 43.46% ± 3.44%, respectively (Figure 3B-D).No significant difference in residual membrane area was detected between the free DEX-, NP D -, and NP C -treated groups (Figure 3E).This trend correlates well with the degree of material accumulation in the inflamed tissue as analyzed by fluorescence imaging.Furthermore, the DEX-related modulation of the inflammatory process was assessed via the examination of macrophage and endothelial cell infiltration into the collagen membrane (Figure 4).Foci of infiltrated macrophages were observed within the colla-gen membrane in the control, while in the DEX-injected rats, the macrophages foci were found at the membrane periphery (Figure 4A,B).In contrast, the NP L -treated membranes were virtually free of infiltrated macrophages foci (Figure 4C).At a higher magnification of the collagen membrane edge, the pristine state of the NP L -treated membrane was in sharp contrast to the considerable infiltration of macrophages seen both in the control and the DEX-injected rats (Figure 4A-C, insets).Endothelial cells massively penetrated the collagen membrane from the periphery inwards in the control and were apparently confined to the membrane borders both in the DEX-and NP L -treated rats (Figure 4D-F).Nevertheless, at higher magnification, the edge of the membrane revealed a lower amount and a reduced penetration of endothelial cells to the membrane in NP L -treated rats as compared to DEX-injection (Figure 4E,F, insights).Overall, macrophage and endothelial cells infiltration were visibly reduced following NP L -injection as compared to DEX alone and control, suggesting a higher effect of NP L in reducing the inflammatory response at the implanted membrane.These results suggest that DEX therapeutic efficacy in reducing inflammation can be significantly improved by incorporation into the enzyme-responsive nanoparticles.

Conclusions
We demonstrated the versatility and therapeutic potential of the enzyme-directed assembly of particle therapeutics approach for delivery in wound inflicted in a diabetic model.Specifically, we showed that enzyme-responsive nanoparticles can respond to local inflammation for targeted and prolonged retention up to 14 days post-injection in diabetic-diseased tissues.Most importantly, our results suggest that incorporation of DEX into the enzyme-responsive nanoparticles improves its therapeutic efficacy in reducing inflammation, noteworthy within a single injec- tion, potentially by increasing local drug concentration and prolonging therapeutic effect.
The current in vivo study in rats points out the potential of the presented delivery system to reduce drug doses compared to systemic administration of the drug alone.Thus, it is an important proof of principle, and may serve as the basis for further studies aimed to translate the system to an in vivo scenario in humans.Ongoing experiments are currently underway to fully determine the pharmacokinetic parameters and metabolic pathways post-administration.We believe that utilization of enzymedirected assembly of particle therapeutics as a generalizable platform for drug delivery could potentially achieve desired therapeutic effects, including anti-inflammation, anti-infection, and po-tentially tissue repair and regeneration with reduced doses and side effects.
Micellar Nanoparticle Formulation: PPAs (50% wt.% with and 50% wt.% without Cy5.5)were dissolved at 3 mg mL −1 in DMSO.1× DPBS (without Ca 2+ and Mg 2+ ) was added via a syringe pump at a speed of 100 μL h −1 until reaching 30% DPBS in DMSO (v/v).The solution was left stirring overnight and transferred into SnakeSkin Dialysis Tubing (10K MWCO) to dialyze against DPBS for 48 h with three buffer changes.The resulting solution was filtered through a 0.22 μm PES membrane filter to remove bacteria and any large aggregates.The polymer concentration of the filtered solution was confirmed by UV absorbance from Cy5.5 (Figure S8, Supporting Information), which showed less than 5 wt.% material loss by filtration.The nanoparticle solution was concentrated by spin centrifugation to produce 300 μm regarding PPA and 150 μm regarding Cy5.5 dye.
Peptide and Nanoparticle Cleavage: Stock solution of NorMMP (510 μm) was prepared in DPBS buffer.Then, 1.5 μL MMP-9 stock (38.46μM) and 105 μL of NorMMP stock were mixed, giving a final concentration of 0.5 μM MMP-9 and 500 μM NorMMP.The solution was incubated at 37 °C for 24 h to promote peptide cleavage.To prepare the HPLC sample, 50 μL of the reaction mixture was added to 150 μL of DPBS buffer and 40 μL of the resulting solution was injected into HPLC for analysis.For NorMMP L , NorMMP D , and a sequence (LAGGWGERDGS) equivalent to that produced via enzyme-catalyzed cleave of NorMMP L , a gradient of 20-60% Buffer B over 30 min was used.For NorMMP C , a gradient of 20-40% Buffer B over 30 min was used.The same HPLC samples were then analyzed by ESI-MS to determine the change in mass post-enzyme treatment.
Micellar nanoparticles (100 μM, with respect to the polymer) were treated with thermolysin (1 μM), an MMP alternative with higher in vitro activity, or DPBS for 24 h at 37 °C in 1× DPBS.The resulting nanoparticle solutions were analyzed by TEM to examine the change in morphology.For the TEM analysis, 5 μL of sample was applied to a 400-mesh carbon grid (Ted Pella, Inc.) that was glow discharged for 15 s.Five microliters of 2 wt.% uranyl acetate solution was then applied and wicked away after 30 s for staining.
Type 1 Diabetes Induction in Rats: Three-to 6-month-old female Sprague Dawley rats (weight 240-260 g) were used in all assays.Diabetes was induced by selective destruction of the insulin-producing -cells of the pancreas by a single injection of streptozotocin (STZ, 65 mg kg −1 ).Seven days after the STZ injection, blood glucose concentrations measured via the tail were higher than 180 mg dL −1 , confirming hyperglycemia.All experiments were conducted in accordance with the animal experiment guidelines after approval of the Tel Aviv University Animal Care Committee (research number 01-19-082, authorization number 034_b15917_36).
In Vivo Collagen Membrane Implantation: Resorbable collagen membranes were cut using a biopsy punch into 5-mm diameter discs and labeled with biotin at 3 mg mL −1 for 1 h, to allow future measurement of residual collagen within the membrane.Following blood glucose measurements, rats were anesthetized with ketamine and xylazine, and the dorsal part of the skin covering the scalp was shaved and aseptically prepared for surgery.A lateral incision was made in the scalp, a partial-thickness flap, not involving the periosteum, was elevated, a pouch was then created underneath the periosteum and a single membrane disc was placed on top of the parietal bone of the calvaria, anterior to the ears line.Soft tissues were repositioned and sutured with resorbable sutures.Nanoparticles and control solutions were injected 24 h later, posterior to the ears line (200 μL injected subcutaneously).
In Vivo Nanoparticles Imaging: In vivo nanoparticles imaging was performed at different time points following the nanoparticles administration.Animals were anesthetized with a parenterally administered cocktail of ketamine (37 mg kg −1 body weight) and xylazine (0.5 mg kg −1 body weight) and images of the administrated fluorescently labeled nanoparticles were acquired with an optical imager (Maestro TM, CRI Inc, Woburn, MA) using deep red filters (excitation range 671-705 nm, emission of 750 nm long pass).Animals were positioned to capture images of the upper half of the body from dorsal view.The nanoparticles signal intensity was analyzed using the ImageJ software (version 1.53t, National Institutes of Health, Bethesda MD) after defining the specific regions of interest.The nanoparticle signal intensity expressed as counts per second per mm 2 was normalized to the adjacent background intensity.
Tissue Harvesting and Staining: Two weeks after injection, the rats were sacrificed by a carbon dioxide overdose, and the skull, together with the over lining periosteum and the membrane, were harvested.Harvested tissues from ten rats per group were fixed using buffered 4% paraformaldehyde, processed for decalcified tissue histology with EDTA 10%, and embedded in paraffin.Sections were cut in a coronal plane from the middle area of the collagen membrane and were stained with Horseradish peroxidase-conjugated streptavidin for measurements of membrane thickness and residual collagen using a monoclonal mouse anti-rat cluster of differentiation 68 (CD68) at 1:500 for macrophage cells and a monoclonal mouse anti-human/rat transglutaminase II (TGII) at 1:100 for endothelial cells.Horseradish peroxidase-conjugated anti-mouse secondary antibodies were used.Images were captured using an automated scanning optical microscope (Aperio VERSA 8, Leica) by a CMOS camera at magnification x20.
Residual Collagen Membrane Quantification: The residual collagen content of each membrane was quantified according to the histological staining of its biotin labeling using the ImageJ plugin Color Segmentation (EPFL Biomedical Imaging Group, available at http://bigwww.epfl.ch/sage/soft/colorsegmentation).This ImageJ plugin allows to segment a color image by pixel clustering.The cluster was manually defined by the user through the friendly interface, and K-means was used as the algorithm of clustering.Briefly, for each stained section, three regions of interest were drawn from the outer borders of the collagen membrane (as in Figure 3B-D), the color segmentation algorithm was run separately for each region, and the resulting residual collagen content was expressed as the percentage of collagen stained pixels, each section score being the average over the 3 regions.

Figure 1 .
Figure 1.Synthesis and enzyme responsiveness of DEX-incorporated nanoparticles with L-, D-MMP peptide and control peptide shell.A) Synthetic scheme showing PPAs and nanoparticles preparation.b -refers to block copolymer.B) SEC traces of NorDEX 20 first block and PPAs.Note: due to Cy5.5 interference with light scattering, the MWs of polymers formed post-NorMMP incorporation (i.e., NorDEX 20 -b-NorMMP 5 ) were reported as the MWs of PPAs.C) Tablesummarizingthe MWs and dispersity of PPAs.Theoretical MW was calculated by M n,theo = ∑ DP monomer × MW monomer .Measured MWs and dispersity were determined by SEC-MALS with a dn/dc of 0.179 mL g −1 .D-I) TEM images of nanoparticles before (top panel) and after (bottom panel) incubation with thermolysin at 37 °C for 24 h.The sizes of NP L , NP D , and NP C were measured to be 15 ± 2, 15 ± 2, and 14 ± 2 nm, respectively.The formation of microscale aggregates was observed only for NP L .Scale bars: 100 nm.

Figure 2 .
Figure 2. In quantification of nanoparticle accumulation in the inflamed tissue.A) Timeline of the study.B) Illustration of the collagen membrane implantation surgery and the nanoparticle injection site.C-F) In vivo fluorescence images of (C,D) DEX and (E,F) NP L -injected rats at day 7 post-injection.The fluorescence images (C,E) were superposed on the black-and-white image to measure the fluorescence intensity in the area of the implanted collagen membrane (labeled with a black circle) (D,F).Scale bars: 2 cm.G) Quantification of fluorescence accumulation in the inflamed tissue at days 7 and 14 post-injection.Values indicate the mean fluorescence ± SEM (control n = 3, DEX n = 8, NP C n = 10, NP D n = 11, NP L n = 10, one-way ANOVA with Tukey HSD Test for post hoc pairwise comparison, significance meaning: *p < 0.05, ***p < 0.001).

Figure 3 .
Figure 3. Residual collagen content of the implanted membranes at day 14 post-injection.A) Illustration of a histological section comprising the stained collagen membrane in between the lining periosteum and the calvaria bone.B-D) Residual collagen content in stained membranes implanted in B) control, C) DEX-injected, and D) NP L -injected rats.Scale bars: 500 μm.E) Residual collagen membrane area expressed as percentage of the original pre-implanted membrane (Values indicate the mean ± SEM (control n = 3, DEX n = 7, NP C n = 8, NP D n = 9, NP L n = 9, one-way ANOVA with Tukey HSD Test for post hoc pairwise comparison, significance meaning: ***p < 0.001, ****p < 0.0001).

Figure 4 .
Figure 4. Macrophage and endothelial cells infiltration at the borders of the collagen membrane at day 14 post-injection.A-C) CD68 + macrophage immuno-staining in A) control, B) DEX-injected, and C) NP L -injected rats.D-F) TG-II+ endothelial cells immuno-staining in D) control, E) DEX-injected, and F) NP L -injected rats.Notable foci of macrophage and endothelial cells are highlighted with arrows, marking the limit of the infiltration edge.The collagen membrane is delimited with dashed line in each picture.Scale bars: pictures: 500 μm, insets: 150 μm.