Self-Assembled Peptide Amphiphile Nanofibers for Controlled Therapeutic Delivery to the Atherosclerotic Niche

Abstract

Atherosclerotic plaque remains the leading contributor to cardiovascular disease and requires invasive surgical procedures for its removal. Nanomedicine offers a minimally invasive approach to alleviate plaque burden by targeted therapeutic delivery. However, nanocarriers are limited without the ability to sense and respond to the diseased microenvironment. In this study, targeted self-assembled peptide amphiphile (PA) nanofibers were developed that cleave in response to biochemical cues expressed in atherosclerotic lesions—reactive oxygen species (ROS) and intracellular glutathione—to deliver a liver X receptor agonist (LXR) to enhance macrophage cholesterol efflux. The PAs released LXR in response to physiological levels of ROS and reducing agents and could be co-assembled with plaque-targeting PAs to form nanofibers. The resulting LXR PA nanofibers promoted cholesterol efflux from macrophages in vitro as well as LXR alone and with lower cytotoxicity. Further, the ApoA1-LXR PA nanofibers targeted plaque within an atherosclerotic mouse model in vivo and activated ATP-binding cassette A1 (ABCA1) expression as well as LXR alone with reduced liver toxicity. Taken together, these results demonstrate the potential of self-assembled PA nanofibers for controlled therapeutic delivery to the atherosclerotic niche.

Graphical Abstract

Peptide amphiphiles containing peptides that target plaque and linkages that release a therapeutic (LXR) in response to the atherosclerotic niche can self-assemble to form nanofibers in aqueous solution. These nanofibers promote macrophage cholesterol efflux in vitro (bottom left, scale bar: 25 μm) and localize to atherosclerotic plaque in mice upon intravenous injection (bottom right, scale bar: 100 μm).

1. Introduction

Cardiovascular disease is the leading cause of death worldwide.^[1][2] Atherosclerosis is a major contributor and develops from a diminished lipid metabolism coupled with a prolonged inflammatory response.^[3][4] Atherosclerotic plaque forms when low density lipoproteins (LDLs) in blood diffuse through endothelial junctions and accumulate in vessel intima.^[5] In an attempt to clear the LDL, endothelial cells and macrophages secrete reactive oxygen species (ROS), which promote oxidation of LDLs (oxLDLs). Macrophages phagocytose the oxLDLs, yet cannot catabolize the excessive levels of free cholesterol, thus transforming into foam cells that eventually undergo necrosis. The necrotic debris are not efficiently cleared from the plaque, causing additional inflammation and endothelial dysfunction.^[5] The resulting plaque in blood vessels restricts oxygen delivery to tissues and can lead to ischemia, heart attack, or stroke.

Current treatments for advanced atherosclerosis involve invasive surgical procedures including endarterectomy, bypass, and angioplasty—all of which risk vessel restenosis.^[6][7] Nanomedicine is a minimally invasive approach to reduce plaque burden through administering nanoparticles that target atherosclerotic lesions and deliver therapeutics.^[8] Peptide amphiphiles (PAs) have shown promise for cardiovascular disease treatment.^{[9][10][11][12][13]}

These PAs are composed of palmitic acid conjugated to peptides with β-sheet hydrogen bonding and can self-assemble into high-aspect ratio nanofibers in aqueous solution. Biofunctionality can be incorporated into the PA by adding peptides that participate in surface receptor binding and intracellular signaling.^[14] PAs also enable multiplexing of biological signals through the self-assembly of several different types of PAs into a single nanofiber. Furthermore, PAs are bioresorbable, lack immunogenicity, and palmitic acid has been shown to reduce atherosclerosis severity.^[14][15] We previously demonstrated that PA nanofibers containing collagen-binding peptides allow superior binding to injured vasculature in comparison to PA nanospheres.^[10] Additionally, we developed PA nanofibers containing an apolipoprotein A1-mimetic peptide (ApoA1 PAs) that target plaques in atherosclerotic mice.^[11][16]

Despite these benefits, PA applications in drug delivery are limited without the ability to sense and respond to the diseased microenvironment, which would enable controlled, therapeutic delivery. In this study, we aimed to develop PA nanofibers that could target plaque and release therapeutics in response to biochemical cues expressed in the atherosclerotic niche. As a model therapeutic, we chose liver X receptor agonist GW3965 (LXR), which restores lipid homeostasis by stimulating cholesterol efflux from macrophage foam cells through increased expression of ATP-binding cassette A1 (ABCA1) cholesterol transporters, yet promotes liver steatosis when administered systemically.^[17][18][19] This adverse side effect may be avoided by incorporating LXR agonists into nanoparticles that target atheroma.^[20][21][11]

We hypothesized that PA nanofibers containing linkages sensitive to ROS and intracellular glutathione could tether LXR and be co-assembled with ApoA1 PAs to form injectable ApoA1-LXR PA nanofibers that allow targeted, controlled therapeutic delivery to atherosclerotic plaque. To evaluate this hypothesis, we synthesized PAs with ROS- and glutathione-cleavable linkages using prolines and succinimidyl 3-(2-pyridyldithio)propionate (SPDP) or 4-nitrophenyl 2-(2-pyridyldithio)ethyl carbonate (NDEC), respectively, and characterized their ability to release LXR under physiological conditions. We then determined processing conditions that allowed for ApoA1-LXR PA nanofiber formation and characterized their physical and biological properties in vitro. Using an LDL receptor knockout (LDLR KO) mouse model of atherosclerosis, we evaluated the targeting and therapeutic potential of ApoA1-LXR PAs in vivo.

2. Results

2.1. Design of Atheroma Niche-Responsive PAs

Schematic 1 illustrates our nanofiber approach for targeted drug delivery to the atherosclerotic niche. The nanofiber is formed from the co-assembly of three PAs. The first PA—ApoA1 PA— contains 4F (red sequence in Schematic 1B), an apolipoprotein A1-mimetic peptide that binds to oxidized LDLs and promotes targeting to atherosclerotic plaque.^[11][16][22] The second PA contains LXR, which is tethered by linkages that cleave in response to reactive oxygen species (ROS-LXR PAs) or intracellular glutathione (SPDP-LXR or NDEC-LXR PAs).^[23][24][25] The third PA is a diluent PA (E2 PA), which aids nanofiber formation.^[12][16]

Schematic 1.

Design of peptide amphiphile (PA) nanofibers for targeted, controlled therapeutic delivery to the atherosclerotic niche. a) Molecular graphic of nanofiber formation from the self-assembly of three PAs: a diluent PA (Backbone), a PA containing a plaque-targeting peptide (ApoA1 PA), and a PA containing liver X receptor agonist GW3965 (LXR) attached by a cleavable linkage. b) Corresponding chemical structures for the PAs. The diluent PA contains an alkyl tail followed by a V2A2 peptide sequence that promotes β-sheet hydrogen bonding, and a charged peptide sequence (E2) that aids solubility. Shown in red is 4F (DWFAKDYFKKAFVEEFAK), an ApoA1-derived peptide that targets plaque. The light blue box represents the cleavable linkage, comprised of either prolines cleavable by reactive oxygen species (ROS), or intracellular glutathione-cleavable disulfide linkages derived from SPDP or NDEC. The SPDP and NDEC linkages contained PEG, attached using cysteine.

Given the nuclear target of LXR and 4F peptide binding to ABCA1, we sought linkages that could be cleaved within macrophages.^[26] Proline linkages are known for their cell-penetrating properties and are disrupted by intracellular ROS. We observed superior delivery of therapeutics from PAs with proline linkages (hereafter denoted as ROS linkages) to macrophages in comparison to protease-cleavable linkages.^[12][27] Our SPDP- and NDEC- derived linkages contain disulfide bonds that can be cleaved by intracellular glutathione. NDEC linkages offer the additional benefit of acting as self-immolative linkers due to the release of the original therapeutic without any vestiges of NDEC.^[25] PEG spacers were included in the cleavable LXR PA design to improve solubility. LXR was incorporated onto PAs with ROS, SPDP, or NDEC linkages using solid phase peptide synthesis techniques.

Once we confirmed successful incorporation of ROS-, SPDP-, and NDEC-LXR linkages into PAs by LC-MS (Figure S1 and S2, Supporting Information), we assessed their cleavage in response to ROS at levels found within the atherosclerotic niche, using hydrogen peroxide (H₂O₂ , 250 μM).^[28] Intracellular levels of glutathione are estimated to be near 10 mM.^[29] However, given the potential for glutathione to undergo side reactions, we chose to use tris(2-carboxyethyl)phosphine (TCEP, 10 mM) or dithiothreitol (DTT, 10 mM) as the reducing agent.^[30] The ROS-LXR PAs released approximately 40% of the LXR by 24 hours (Figure 1, Figure S3a, b), and were completely cleaved after 1 week of treatment with H₂O₂. The disulfide containing linkers were more efficient, where approximately 87% of the SPDP-LXR PA and nearly all (97%) of the NDEC-LXR PA, were cleaved within 24 hours of treatment with 10 mM of TCEP (Figure 1, Figure S3c–f). Figure S3f shows the NDEC-LXR PA cleavage peaks of LXR release after reduction of the disulfide bond (7.12 min) and subsequent release of the original K-LXR component (6.32 min) after the thiyl radical attack on the carbonyl carbon. Taken together, these results demonstrate the potential for therapeutic release from PAs in the atheroma niche using ROS- and glutathione-responsive linkages.

Figure 1:

PA cleavage over time for NDEC-, SPDP-, and ROS-LXR PAs. Samples were tested at 24 h, 4 day, and 7 day time points against controls. NDEC-LXR PAs were treated with 10 mM TCEP and showed complete cleavage by day 4 of treatment. SPDP-LXR PAs were treated with 10 mM TCEP or 10 mM DTT, ROS-LXR PAs were treated with 250 μM H₂O₂ + 50 μM CuSO₄.

2.2. Parameters for Developing Targeted, Atheroma Niche-Responsive PA Nanofibers (ApoA1-LXR PAs)

After confirming the cleavage potential of the ROS-, SPDP-, and NDEC-LXR PAs, we determined parameters that enabled their co-assembly with ApoA1 and diluent PAs (Figure S4, Supporting Information) to form nanofibers. Specifically, we varied the PA molar ratio and the processing conditions. Our prior work indicated that at least 40 mol% of ApoA1 PA was required for targeting.^[11][16] Given this restriction, we found that 10 mol% was the maximum amount of LXR PA that could be incorporated (Figure 2).

Figure 2.

Representative TEM images depicting processing conditions that enabled nanofiber formation from ApoA1-LXR PA co-assemblies. The co-assemblies contained 40 mol% ApoA1 PA and 50 mol% diluent E2 PA.

Aging and annealing PAs can shift their self-assembly towards more thermodynamically favored states to produce nanofibers by allowing them more time to self-assemble and increasing their thermal energy, respectively.^[31][32] We found that the ROS-LXR PA co-assemblies formed nanofibers after 24 hours of aging at 4 °C. In contrast, the SPDP- and NDEC-LXR PA co-assemblies required an annealing process, which consisted of placing the PAs in a water bath at 80 °C for 30 minutes before gradually reducing the temperature to room temperature (RT, 23 °C) overnight. Nanofiber self-assembly was not disrupted by serum proteins, as shown by cryogenic transmission electron microscopy (cryoTEM, Figure S5, Supporting Information), suggesting that the nanofiber structure would remain intact upon intravenous injection.

2.3. Characterization of ApoA1-LXR PAs

We used small-angle X-ray scattering (SAXS) to characterize the geometry of the ApoA1-LXR PA co-assemblies. The Guinier low q region provides insight into whether the nanofibers exist as cylindrical or lamellar structures based on whether the slope is −1 or −2, respectively.^[33] The E2 nanofibers showed a slope of −1.2, indicating mostly cylindrical or thin ribbon-like structures (Figure 3a, Figure S6, Supporting Information). In contrast, the increasing slopes of ApoA1 (−2.6), ROS-LXR (−2.8), and SPDP-LXR (−1.8) PAs suggest more flattened lamellar structures. The lower slope of NDEC-LXR PAs (−0.8) implies the presence of both cylindrical and micelle structures. The diameter of the nanofibers increased with co-assembly (Table S1, Supporting Information). For example, the diameter for the E2 PA was 8.7 nm, while the ApoA1 PA co-assembly (40% ApoA1 PA, 60% E2 PA) was 10 nm. Accordingly, the ROS-LXR, SPDP-LXR, and NDEC-LXR PAs (co-assembled with 40% ApoA1 PA, 50% E2 PA, and 10% of the respective LXR PA) diameters were 14.7, 12.6, and 14.5 nm, respectively. Nanofiber length decreased with co-assembly. E2 PA nanofibers had an estimated length of 1330 nm, which was reduced upon co-assembly with ApoA1, ROS-LXR, SPDP-LXR, and NDEC-LXR PAs to 1070, 742, 410, and 839 nm, respectively.

Figure 3.

Characterization of co-assembled PA nanofibers using a) SAXS, b) circular dichroism spectroscopy, and c) zeta potential. Nile Red Assay results for d) ROS-LXR, e) SPDP-LXR, and f) NDEC-LXR PA co-assemblies. CAC indicates critical aggregation concentration, a.u. denotes arbitrary units. NDEC-LXR, SPDP-LXR, and ROS-LXR refer to PA co-assemblies of 40% ApoA1 PA, 50% E2 PA, and 10% NDEC-LXR, SPDP-LXR, or ROS-LXR PA, respectively.

Due to the α-helical character of the ApoA1 PA being critical for targeting plaque lesions, we characterized the PA secondary structure using circular dichroism spectroscopy. The α-helical character was preserved after its co-assembly with the LXR PAs, as shown by the negative ellipticities near 222 and 208 nm (Figure 3b).^[26][34] The ApoA1-LXR PAs possessed negative zeta potentials, as expected (Figure 3c). ApoA1 PA co-assembly with the ROS-LXR PA increased the average zeta potential from −31.8 ± 1.5 mV to −16.1 ± 1.12 mV. The NDEC- and SPDP-LXR PA co-assemblies had similar zeta potentials of −19.8 ± 2.6 and −22.7 ± 1.8 mV, respectively.

We next employed Nile Red assays to find the critical aggregation concentration (CAC) for the PAs (Figure 3d–f, Figure S7, Supporting Information). The process of co-assembling the PAs increased their CAC from 7.5 μM in E2 PAs to 45 μM in ApoA1 PA co-assemblies. Co-assembly with ROS-LXR PAs further increased the CAC to 63 μM. Interestingly, annealing the PAs countered this increasing effect and improved the CAC values over ApoA1 PA co-assemblies to 24 and 20 μM for SPDP- and NDEC-LXR PA co-assemblies, respectively. These values are at least 20-fold lower than the injection concentration (1 mM), indicating that the nanofiber structure is retained after dilution in the bloodstream.^[10]

2.4. Evaluating ApoA1-LXR PA Biological Effects In Vitro

To assess the potential for the ApoA1-LXR PA co-assemblies to promote cholesterol efflux, we loaded J774.2 macrophages with fluorescent cholesterol and treated them with 1–64 μM LXR-equivalent PA concentrations. As seen in Figure 4a, the ROS-, SPDP-, and NDEC-LXR PA co-assemblies promoted cholesterol efflux in a similar manner as LXR. Surprisingly, the ApoA1 PA (at the epitope equivalent of the 64 μM LXR PA co-assemblies) also promoted cholesterol efflux. In contrast, the E2 PA showed no significant increase in cholesterol efflux versus the controls. Further, the LXR PA co-assemblies significantly reduced the cytotoxic effect of LXR as revealed through the methylthiazolyldiphenyl-tetrazolium bromide (MTT) assay (Figure 4b). Among the ApoA1-LXR PAs, the ROS-LXR PA co-assemblies showed the greatest potential for promoting cholesterol efflux and improving cell viability versus LXR. This is also interesting as our cleavage data suggested the ROS cleavage was slower compared to the other linkers over 24 hours.

Figure 4.

Characterization of PA effects on a) cholesterol efflux and b) cell viability using an MTT assay on J774.2 macrophages in vitro. For (A), ^ p<0.05, *p<0.05 vs. control, #p<0.05 vs. DMSO. The ROS-, SPDP-, and NDEC-LXR indicate PA co-assemblies of 40% ApoA1, 50% E2, and 10% of the respective LXR PA. The ApoA1 PA co-assembly (40% ApoA1, 60% E2) and E2 PA are at the epitope equivalent to the 64 μM LXR PA co-assemblies. For b), *p<0.05 vs. LXR, ^p<0.05 vs. E2 PA, #p<0.05 vs. 1 μM LXR. Control refers to untreated cells. Data were analyzed by ANOVA followed by a post hoc Student’s t test, n≥3.

We sought to better understand the interaction between the PAs and cholesterol-laden macrophages by conducting confocal microscopy using fluorescently-labelled PAs (Figure S8, Supporting Information). Surprisingly, we observed the ApoA1-LXR PAs appearing to aid in the removal of cholesterol from the macrophages (Figure 5a, Videos S1–3, Supporting Information). Consistent with our assay results, we found that macrophages treated with E2 PAs retained the majority of their cholesterol (Video S4, Supporting Information). Additionally, macrophages treated with LXR appeared to be undergoing cell death (Figure S9, Videos S5,6, Supporting Information). We found no significant difference in the Manders coefficients between the ApoA1 and SPDP-, NDEC-, and ROS-LXR PAs and cholesterol, indicating their colocalization (Figure 5b). In contrast, the E2 PAs showed significant differences between their Manders coefficients, implying that not all of the cholesterol pixels were colocalizing with the PA, which is consistent with the imaging and cholesterol efflux results.

Figure 5.

PA interactions with cholesterol-laden macrophages observed through a) confocal microscopy and b) Manders colocalization coefficient analysis. Macrophages were treated with 32 μM LXR or PA epitope equivalents for 24 hours. Scale bar: 25 μm. M1 coefficient indicates the amount of PA pixels that colocalized with cholesterol pixels, M2 coefficient indicates the reverse association, *p<0.05. At least 26 cells were analyzed for each PA’s interaction with cholesterol. Data were analyzed with a two-factor ANOVA followed by a post hoc Tukey-Kramer HSD test.

Macrophages residing in plaque lesions are activated to an inflammatory phenotype due to excessive oxLDL uptake and the constant presence of ROS. This causes stress to the mitochondria, decreasing their ATP production and diminishing ABCA1-mediated cholesterol efflux.^[35][36] Therefore, we evaluated whether the ApoA1-LXR PA co-assemblies could promote cholesterol efflux in an inflammatory microenvironment in vitro. As shown in Figure 6a–e, the SPDP-, ROS-, and NDEC-LXR PA co-assemblies improved cholesterol efflux in macrophages stimulated with lipopolysaccharide and IFN-γ by 10–25% versus LXR alone. Further, the SPDP-LXR PA co-assemblies significantly outperformed ApoA1 PA (p=0.0422, Figure 6f).

Figure 6.

Characterization of macrophage cholesterol efflux under inflammatory conditions after treatment with a) LXR, b) PEG-ApoA1, c) SPDP-LXR, d) ROS-LXR, and e) NDEC-LXR PA co-assemblies. f) Quantification of percent increase in cholesterol efflux versus LXR treatment. *p<0.05. Macrophages were stimulated with 10 mg/mL lipopolysaccharide and 100 ng/mL interferon gamma. Data was analyzed by nonparametric testing using Wilcoxon multiple comparisons for each pair, n≥4.

2.5. ApoA1-LXR PA Targeting to Plaque In Vivo

Based on the ability of the ApoA1-LXR PA co-assemblies to promote cholesterol efflux and reduce LXR cytotoxicity in vitro, we extended their evaluation in vivo using LDLR KO mice. We have previously shown the LDLR KO mice can develop atherosclerosis after 14 weeks of a high-fat diet.^[11] In addition, we demonstrated that ApoA1 PA co-assemblies can target plaque within 24 hours after injection.^{[11] [16]} We wanted to confirm that this targeting was maintained despite two major changes in the PAs: the addition of a PEG spacer before the 4F targeting peptide in the ApoA1 PA, and co-assembly with the LXR-containing PAs. Although incorporating fluorescent labels for each PA component in the co-assembly would most accurately indicate the degree of targeting, the presence of multiple fluorophores would impede nanofiber self-assembly.^[37] Hence, we co-assembled the LXR PAs with fluorescently labeled ApoA1 and scrambled ApoA1 PAs to visualize the targeting (Supplemental Figures S8b, S10a–c, Supporting Information). The fluorescently-labeled PAs were able to self-assemble into nanofibers (Figure S11). As demonstrated in Figure 7, all ApoA1-LXR PA co-assemblies (ROS-, SPDP-, and NDEC-LXR) localized to atherosclerotic plaque in the aortic root as well as the original ApoA1 PA co-assembly by 24 hours after intravenous injection. No targeting was seen in mice injected with phosphate buffered saline (PBS) or PA co-assemblies containing the scrambled ApoA1 targeting sequence. Fluorescence examination of the mice organs indicated that the PAs were mostly metabolized by the kidney and liver (Figure S12, Supporting Information).

Figure 7.

Evaluation of ApoA1-LXR PA co-assembled nanofiber targeting of atherosclerotic plaque in LDLR KO mice 24 hours after intravenous injection. a) Representative images of mice aortic roots showing localization of fluorescently labeled PAs (red) to plaque. Lamina autofluorescence is shown in green. Scr indicates scrambled ApoA1 PA co-assemblies. Scale bar: 50 μm. b) Corresponding quantification of fluorescent pixel density in aortic roots. *p<0.05. Data was analyzed by an ANOVA test followed by a post hoc Tukey-Kramer HSD test, n≥4.

2.6. Potential of ApoA1-LXR PAs for Atherosclerosis Treatment

Given the potential for the ApoA1-LXR PAs to target plaque lesions, we evaluated whether they could deliver LXR to reduce atherosclerosis progression. After eight weeks of twice weekly injections, we found no significant decrease in the amount of atherosclerosis between mice treated with PBS, LXR, or PAs (Figure 8). We did, however, observe a significant increase in ABCA1 expression relative to the number of CD68+ macrophages in the PA and LXR treatments versus the PBS control (Figure 9). Among the PA treatments, the NDEC-LXR PA co-assembly showed the greatest ABCA1 activation based on its significant increase in comparison to the PBS control (p=0.0220, Figure 9b). We also analyzed the blood of mice and found no significant differences in lipid levels, glucose, or kidney, muscle, and liver function among the treatments (Figure S13, Supporting Information). Liver toxicity was observed in the PBS and LXR treated mice based on increased RIP3 expression, an indicator of necroptosis (Figure 10). The ROS-LXR and SPDP-LXR PA treatments caused significantly less liver toxicity than LXR (p=0.0081and p=0.0131, respectively), while the NDEC-LXR PA treatment had a decreasing trend (p=0.0656) versus LXR. Thus, incorporating LXR into the ApoA1 PA co-assemblies reduced LXR toxicity while maintaining its potential to activate ABCA1 for cholesterol efflux.

Figure 8.

Evaluation of ApoA1-LXR PA co-assembled nanofibers on atherosclerosis progression in LDLR KO mice after eight weeks of treatment. a) Representative images of atherosclerotic plaque in mouse aortic roots shown by Oil Red O staining. Scale bar: 100 μm. b) Corresponding quantification of atherosclerosis. Data was analyzed by an ANOVA test followed by a post hoc Tukey-Kramer HSD test, n≥6.

Figure 9.

Effect of ApoA1-LXR PA co-assembled nanofibers on ABCA1 expression and resident macrophage number (CD68) in atherosclerotic plaque of LDLR KO mice after eight weeks of treatment. a) Representative images of ABCA1 and CD68 expression in aortic roots shown by immunofluorescence staining. Scale bar: 50 μm. b) Corresponding quantification of fluorescence pixel intensity. *p<0.05, #p<0.05 vs. LXR, ^p<0.05 vs. NDEC-LXR. Data was analyzed by a two factor ANOVA with post hoc Dunnett’s test against the PBS control, n≥6.

Figure 10.

Effect of ApoA1-LXR PA co-assembled nanofibers on liver toxicity. a) Representative images of necroptosis occurring in liver tissue after eight weeks of treatment based on RIP3 expression. Scale bar: 50 μm. b) Corresponding quantification of fluorescence pixel intensity. *p<0.05, #p<0.05 vs. LXR. Data was analyzed by nonparametric testing using Wilcoxon multiple comparisons for each pair, n≥6.

3. Discussion

The increasing prevalence of cardiovascular disease necessitates new strategies for managing atherosclerotic plaque burden.^[2][6] An emerging therapeutic approach is to instruct niche cells in atheroma to self-repair by restoring lipid metabolism and resolving inflammation.^[38][39][40] Nanomedicine technologies are ideal for this approach due to their potential for local, targeted therapeutic delivery. Significant advancements have been made over the past decade in developing nanoparticles that can target atherosclerotic plaque.^[9][13][41] More recent studies explored the potential of delivering therapeutics to the plaque site. ^{[20][21][42][43][44][45][46]}

For example, Flores and colleagues utilized PEG-functionalized carbon nanotubes to deliver a small-molecule inhibitor of CD47-SIRPα signaling in macrophages, which decreased inflammatory gene expression and increased phagocytosis of apoptotic cells in apolipoprotein E-deficient (ApoE−/−) mice.^[45] While promising, this technology is limited by relying on diffusion for drug release, which may not occur quickly enough to impart a therapeutic effect. To overcome this limitation, Wang et al. developed ROS-sensitive β-cyclodextrin nanoparticles and showed their feasibility in delivering rapamycin to slow atherosclerosis progression and improve plaque stability in ApoE−/− mice.^[46] Nevertheless, most technologies for atherosclerosis nanomedicine still rely on passive strategies to target atherosclerotic lesions through the enhanced vascular permeability and retention in inflamed vessel endothelium, which can require a long circulation time to accumulate therapeutic levels of nanoparticles in atheroma.^[47][48] Moreover, delivering therapeutics with systemic toxicity is limited by this approach as the nanoparticles pass through the large fenestrae in the liver and spleen.^[48] Therefore, we sought to develop nanocarriers capable of both targeting atherosclerotic plaque and controlling drug delivery to atherosclerotic plaque. To the best of our knowledge, this study is the first example of a nanocarrier that is capable of targeted, controlled delivery of a lipid metabolism-restoring therapeutic (LXR) to the atherosclerotic niche.

We developed PAs that release LXR in response to physiological levels of ROS and reducing agents. Specifically, glutathione-sensitive linkages SPDP and NDEC allowed at least 1.6-fold higher release of LXR in comparison to ROS-responsive proline linkages after 24 hours of treatment. Under annealing and aging processing conditions, these niche-responsive LXR PAs could be co-assembled with plaque-targeting ApoA1 PAs. The resulting ApoA1-LXR PAs promoted cholesterol efflux from macrophages in vitro as well as LXR alone, with superior efflux seen against LXR under inflammatory conditions. The ApoA1-LXR PAs also avoided cytotoxic effects of LXR. An unexpected finding was the ApoA1-LXR PAs appearing to assist with cholesterol removal from macrophages, visualized by confocal microscopy. This mirrors the role of HDL as a cholesterol-carrier.^[49] These phenomena might be explained by the ability of ApoA1 expressed on HDL to bind to ABCA1 on the macrophage cell membrane where it is directly loaded with free cholesterol.^[50] Although a synthetic mimic of ApoA1, the 4F peptide present on the ApoA1 PAs can participate in ApoA1-like function, including cholesterol efflux.^[26] Accordingly, we observed that ApoA1 PA co-assemblies promoted cholesterol efflux and aided cholesterol removal from macrophages. Co-assembling LXR PAs with ApoA1 PAs enhanced macrophage cholesterol efflux under inflammatory conditions in comparison to LXR alone, specifically for SPDP-LXR PA co-assemblies.

Co-assembled PA nanofibers possess structural diversity, with a locally solid-like behavior in the interior and liquid-like dynamics dominating on the surface.^[31] Further, through super-resolution microscopy, we observed PA molecules undergoing dynamic exchange between fibers and demonstrated that fluorescently labelled molecules can ‘insert/co-assemble’ within fibers of a different fluorescent label.^[51] This structural diversity enhances the ability of PAs to adapt to different environments and may assist their binding to bioactive targets. In support of this theory, we found that all ApoA1-LXR PA co-assemblies showed colocalization to cholesterol in vitro and successfully targeted the plaque within an atherosclerotic mouse model in vivo despite the differences in their physical characteristics.

The ApoA1-LXR PAs targeted plaque within an atherosclerotic mouse model in vivo and activated ABCA1 expression as well as LXR alone with significantly reduced liver toxicity. LDLR KO mice are highly sensitive to oxLDL.^[52] Accordingly, we suspect that the liver toxicity seen from PBS treatment was related to the effects of the high-fat diet. Thus, LXR treatments likely reduced liver toxicity based on their ability to increase cholesterol metabolism through increased ABCA1 expression, which would decrease the amount of oxLDL available to cause liver damage.

The lack of therapeutic effect on atherosclerosis progression may be explained by the very low dosage of LXR administered—32-fold lower than used in our previous study that saw significant reduction of plaque versus PBS treatment (100 μM vs. 3.2 mM).^[11] This lower dosage was due to 10 mol% being the maximum amount of LXR PA that could be co-assembled to form nanofibers with ApoA1 PAs, which were administered at 1 mM. A potential solution to this limitation in our PA design is to utilize a prodrug approach. Here, the hydrophobic component of the PAs would consist of the LXR therapeutic, in place of palmitic acid, tethered to the cleavable linkage and the 4F sequence attached to the epitope region. This would allow the LXR PA and ApoA1 PA to exist as a single PA that could self-assemble into nanofibers, greatly increasing the drug loading capacity. A similar concept was demonstrated by Wang et al. for cancer therapy using pro-drug PAs nanofibers.^[53] Additionally, future PA designs would benefit from developing a means to assess drug localization to the plaque.

Given the limited therapeutic effect in vivo, it would be premature to recommend which linkage, ROS-, SPDP-, or NDEC-LXR, is best suited for atherosclerosis treatment. Still, PAs containing these cleavable linkages can serve as versatile platforms for several new up-and-coming therapeutic realms for atherosclerosis. For instance, the idea of restoring normal immune resolution to plaque macrophages could be performed by delivering peptides of pro-resolving proteins and mediators, for example, IL-10, annexin A1, lipoxins, and resolvins.^[5] Another example is macrophage immunometabolism therapy, which seeks to repair metabolic pathways and epigenetics that are altered as macrophages uptake oxLDL.^[36] As well, blocking mechanosensitive transcription factors in macrophages that are activated during the hardening of plaque is gaining recognition as a means to increase their response to therapeutics.^[54] Our targeted, atheroma niche-responsive PA technology could aid in the translation of these emergent treatment strategies by enabling dual therapeutic delivery through the co-assembly of PAs with ROS- or GSH-cleavable linkages.

4. Conclusion

In this study, we developed PAs that released LXR in response to ROS and reducing conditions found in the atherosclerotic microenvironment and promoted cholesterol efflux in vitro. These PAs also targeted plaque, activated ABCA1, and reduced LXR liver toxicity in vivo. These results support the use of self-assembled PA nanofibers for controlled therapeutic delivery to the atherosclerotic niche as a minimally invasive approach to manage cardiovascular disease.

5. Experimental Section

PA Synthesis:

PAs were synthesized using solid-phase peptide synthesis as previously described.^[12] PEG linkages were incorporated at n=6 (ChemPep Inc.) for ApoA1 PAs and n=4 (Quanta BioDesign) for ROS-, SPDP-, and NDEC-LXR PAs. NDEC linkages were synthesized according to previously reported methods, and validated for structure using proton NMR (Figure S1, Supporting Information).^[25]

The niche-responsive PA products were initially synthesized into two parts, where C₁₆-VVAAEE-PEG₄-C and X-K(LXR) were synthesized separately, X = SPDP (Sigma) or NDEC, and LXR = GW3965.HCl (Aurum Pharmatech). These HPLC purified materials were then reacted in solution to construct the desired final PA structures, C₁₆-VVAAEE-PEG₄-C*-(X)-K(LXR). After this reaction was complete as determined by ESI-MS, the reaction mixture was directly injected and purified by a reverse phase HPLC. A mobile phase of acetonitrile and water was used, both containing 0.1% NH₄OH. Pure fractions were identified for C₁₆-VVAAEE-PEG₄-SPDP-K(LXR) (1999 m/z) and C₁₆-VVAAEE-PEG₄-NDEC-K(LXR) (2015 m/z) by ESI-MS and combined. Excess acetonitrile was removed by rotary evaporation and the remaining water freeze dried to yield the final products. The purity was confirmed by LC-MS. The ROS-LXR PA was created by first synthesizing the C₁₆-VVAAEE-PEG₄-K-PPPPP-K(Mtt) using automated instrumentation and completed by adding the LXR through manual coupling after deprotection of the Mtt group on resin. The ApoA1 PA C₁₆-VVAAEE-PEG₆-DWFKAFYDKVAEKFKEAF was synthesized using automated instrumentation. Additional details for the PA synthesis can be found in the Supporting Information.

Cleavage of Niche-Responsive LXR PAs:

To cleave the prolines in the ROS-LXR PA, we generated hydroxyl radicals by adding 250 μM H₂O₂ and 50 μM CuSO₄ (Fisher Chemical) to the PAs in phosphate buffered saline (PBS).^[1] The disulfide bonds in the NDEC-LXR PAs were reduced by treatment with 10 mM of TCEP (Sigma-Aldrich) in PBS. The disulfide bonds in the SPDP-LXR PAs were reduced by treatment with 10 mM of TCEP or 10 mM of dithiothreitol (DTT, Sigma-Aldrich) in PBS. All LXR PAs were used at 1 mM concentrations and the controls were treated with PBS alone. The cleavage assays were conducted in an incubator at 37 °C and 5% CO₂ for 24 hours and stored at −20 °C until further analysis. A ThermoFisher Q Exactive HF-X (ThermoFisher) mass spectrometer coupled with a Waters Acquity H-class liquid chromatograph system was used to evaluate PA cleavage.^[12] Samples were diluted to 0.1 mM with HPLC grade water and 0.1% NH₄OH was added. The samples were held at 10 °C before introducing 3 μL to a heated electrospray source at a flow rate of 0.6 mL/minute. The electrospray conditions were: spray voltage 4.7 kV, sheath gas (nitrogen) 45 arb, auxiliary gas (nitrogen) 30 arb, sweep gas (nitrogen) 0 arb, capillary temperature 350 °C, capillary voltage 40 V, and tube lens voltage 100 V. The mass range was 150–2000 m/z with 120,000 resolution. Separations were conducted on a Waters Acquity UPLC BEH C18 column (2.1 × 50 mm, 1.7 μm particle size) held at 40 °C. LC conditions were set at 100% water with 0.1% formic acid (A) ramped linearly over 9.8 minutes to 95% acetonitrile with 0.1% formic acid (B) and held until 10.2 minutes. At 10.21 minutes the gradient was returned to 100% A and allowed to re-equilibrate until 12 minutes. All data was analyzed using Xcalibur software (ThermoFisher). Peaks for the LXR PAs in the control and treatment conditions were compared using the automated peak integration tool from FreeStyle^™ software (ThermoFisher). The data was plotted in GraphPad Prism version 9.1.1 and fitted using either hyperbola (NDEC- and SPDP-LXR PAs) or third order polynomial (ROS-LXR PA) least squares fit.

PA Co-Assembly:

Lyophilized PAs were combined in varying molar ratios and co-assembled in hexafluoroisopropanol (HFIP, Sigma) at 2 mg/mL using probe sonication (Q700 Sonicator, Qsonica, 10% amplitude, 110 V, 20 kHz) in a water bath. ApoA1, ROS-, SPDP-, and NDEC-LXR PAs containing PEG were sonicated for five minutes of 10-second on/off pulse cycles. ApoA1 and scrambled ApoA1 PAs without PEG were sonicated for 15 minutes of 10-second on/off pulse cycles. The resultant solution was flash frozen in liquid nitrogen and evaporated under high vacuum for 2–4 hours to remove the HFIP. The PAs were then reconstituted in distilled water and pH-adjusted to 7.5–8.0 using a 200 mM solution of NaOH. Probe sonication was conducted for 1 minute of 10-second on/off cycles. The solution was again flash frozen in liquid nitrogen and dried on a lyophilizer (Labconco Freezone 1L Freeze Dryer System). PAs were stored at −20 °C until use.

To visualize PA colocalization with cholesterol, 5 mol% of the E2 PAs in the E2, ApoA1, ROS-LXR, and SPDP-LXR co-assemblies was replaced with E2 PAs containing tetramethylrhodamine (TAMRA). NDEC-LXR PA co-assemblies were fluorescently labeled by replacing 4.3 mol% of the ApoA1 PA with ApoA1 PAs containing AlexaFluor 546 (AF546). For in vivo targeting studies, PAs were co-assembled in the following molar percentages: Scrambled ApoA1 PAs (4.3% Scrambled ApoA1 PA with AF546, 35.7% Scrambled ApoA1 PA, 60% E2 PA); ApoA1 PAs—without PEG—(4.3% ApoA1 PA with AF546, 35.7% ApoA1 PA, 60% E2 PA); ROS-LXR PAs (4.3% ApoA1 PA with PEG and AF546, 35.7% ApoA1 PA with PEG, 50% E2 PA, 10% ROS-LXR PA); SPDP-LXR PAs (4.3% ApoA1 PA with PEG and AF546, 35.7% ApoA1 PA with PEG, 50% E2 PA, 10% SPDP-LXR PA); NDEC-LXR PAs (4.3% ApoA1 PA with PEG and AF546, 35.7% ApoA1 PA with PEG, 50% E2 PA, 10% NDEC-LXR PA).

TEM:

TEM was performed as previously described.^[12] For cryoTEM, we used 300-mesh copper grids with lacey carbon film (Electron Microscopy Sciences, Hatfield, PA, USA) that were glow discharged for 30 seconds in a PELCO easiGlow system (Ted Pella, Inc., Redding, CA, USA). PAs were prepared at 3 mM in PBS and diluted to 1 mM with 15% FBS immediately prior to blotting; 7 μL of each sample was transferred to the copper grids and plunge-frozen using a Vitrobot Mark IV (FEI) vitrification robot. Samples were blotted at room temperature (RT) with 95–100% humidity and plunge frozen into liquid ethane. Samples were then transferred into a liquid nitrogen bath and placed in a Gatan 626 cryo-holder through a cryo-transfer stage. CryoTEM was performed using a liquid nitrogen-cooled JEOL 1230 TEM at 100 kV accelerating voltage. Images were acquired using a Gatan 831 CCD camera.

SAXS:

SAXS experiments were performed at beamline 5-ID-D of the DuPont-Northwestern-Dow Collaborative Access Team (DND-CAT) Synchrotron Research Center at the Advanced Photon Source, Argonne National Laboratory. PAs were prepared at 5 mM in PBS and irradiated for five frames at 5 seconds per sample. Data was collected with an X-ray energy of 17 keV (l = 0.83 Å). The sample to detector distance was 201.25 mm. The scattering intensity was recorded in the interval 0.002390 < q < 4.4578 Å⁻¹. The wave vector q is defined as = (4π/λ) sin(θ/2), where θ is the scattering angle. Azimuthal integration (Fit2D) was used to average 2D scattering images to produce 1D profiles of intensity versus q. Samples were oscillated with a syringe pump during exposure to prevent beam damage. Background scattering patterns were obtained from samples containing PBS. Following background subtraction and reduction, the data were modeled to a poly-disperse core-shell cylinder using the NCNR Analysis macro in IgorPro software.

Circular Dichroism Spectroscopy:

A Chirascan^™-plus Circular Dichroism Spectrometer (Applied Photophysics) was used to analyze the secondary structure of all PAs as previously described.^[12] PA samples were analyzed at 37 °C from 185 to 260 nm with 0.3 nm step size and analysis time of 1.25 seconds per data point. Spectrum data was averaged from at least two scans and the mean molar ellipticity was determined per residue, with the amino acids averaged for PA co-assemblies.

Zeta Potential:

A Zetasizer Nano ZS (Malvern) was used to characterize the PA zeta potential. PAs at 0.25–0.5 mM in PBS were passed through a 0.2 μm filter (Acrodisc® Syringe Filters with Supor® Membrane, Pall). Three measurements were taken at 37 °C with 10 scans per sample.

Nile Red Assay:

Nile Red is used as an indicator of critical aggregation concentration (CAC) in amphipathic molecules because it experiences a blue shift in fluorescence as the hydrophobic character in the solvent increases. There is a corresponding increase in fluorescence intensity from reduced twisted intramolecular charge transfer.^[55] The CAC for PA nanofibers was determined by diluting PAs from 1 mM to 100 nM in PBS and adding Nile Red (2.5 μM final concentration).^[10] ROS-LXR PA dilutions were aged for 24 hours at 4 °C. Prior to diluting, SPDP- and NDEC-LXR PAs were annealed at 1 mM in the Nile Red solution for 30 minutes at 80 °C and brought to RT overnight. All PA samples were allowed to normalize in the Nile Red solution for 2 hours before aliquoting in triplicate on a 96-well plate. The samples were excited at 550 nm and the fluorescence read from 580 to 450 nm. The data was plotted as the maximum fluorescence intensity versus the log of the concentration. CAC was determined from the intersection of the baseline and tangent line to the rising curve.^[56][57]

In Vitro Cholesterol Efflux Assay:

To simulate cholesterol efflux conditions in vitro, we utilized J774.2 macrophages (Sigma-Aldrich), cultured as previously described.^[12] The macrophages were first plated at 2.63 × 10⁴ cells/cm² on 24-well plates and allowed to adhere overnight. The second day, macrophages were loaded with complete media (DMEM, 4.5 g/L glucose, 2 mM L-glutamine, 2.5% v/v fetal bovine serum, 100 U/mL penicillin, 100 μg/mL streptomycin, 2 mg/mL bovine serum albumin) containing 10 μM of fluorescent NBD cholesterol (22-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-23,24-bisnor-5-cholen-3β-ol, Thermo Fisher Scientific) and 2 μg/mL of Acyl-CoA:cholesterol acyltransferase (ACTA) inhibitor (Sigma-Aldrich), which prevents cholesterol processing by esterification as well as the uptake/accumulation of cholesteryl esters. On the third day, the cholesterol-loading media was removed and the LXR treatments were added at 1, 16, 32, and 64 μM LXR or LXR PA epitope equivalents overnight in media free of serum of albumin. LXR was prepared from 1, 16, 32, or 64 mM stock solutions in dimethyl sulfoxide (DMSO) diluted to the final concentrations in media. Finally, on the fourth day, cholesterol efflux was stimulated by replacing the media with DMEM media that did not contain serum or albumin and adding high-density lipoproteins (HDL, from human plasma, Fisher Scientific) at 50 μg/mL. After 4 hours, the media was removed, the cells were rinsed with PBS and then lysed with RIPA buffer for 10 minutes at 4 °C. The lysates were scraped and centrifuged at 16,000 × g for 15 minutes at 4 °C. The supernatant was collected in 96-well black-walled, clear-bottom plates and the fluorescence read in triplicate on a Cytation 3 Multi-Mode Reader with 473 nm excitation and 530 nm emission. Percent cholesterol efflux was calculated using the following formula:

$$\%\mathit{\text{ cholesterol efflux}}=\frac{F.I{.}_{t0}-F.I{.}_{tx}}{F.I{.}_{t0}}\times 100\%$$

where F.I._t0 is the average fluorescence of cell lysates at time t=0 across triplicate wells of the control, and F.I._tx is the average fluorescence of cell lysates at time t=4 hours across triplicate wells of the treatment groups.

MTT Assay:

PA and LXR effects on macrophage viability were assessed by a methylthiazolyldiphenyl-tetrazolium bromide (MTT, Sigma-Aldrich) assay. Macrophages were plated at 2.63 × 10⁴ cells/cm² on 96-well black-walled, clear-bottom plates and followed procedures for cholesterol loading and LXR treatments as described for the cholesterol efflux assay. After overnight treatment with LXR or PAs, the media was removed and a 0.4 mg/mL MTT solution prepared in complete media was added to each well. After 4 hours of incubation at 37 °C, the solution was removed and 100 μL of DMSO was added per well, incubated for 10–15 minutes at RT on a plate shaker at low speed, and read by an Epoch plate reader at 560 nm absorbance with background subtraction at 670 nm. All values were normalized to the untreated control values.

PA Colocalization with Cholesterol:

Macrophages were seeded onto 6-channel μ-Slide VI 0.4 (ibidi) at 2.63 × 10⁴ cells/cm² and followed the same cholesterol-loading and treatment procedures for 32 μM LXR or LXR PA epitope equivalent as described in the cholesterol efflux assay. All PAs contained a red fluorescence signal from either TAMRA or AlexaFluor 555-conjugation. After overnight treatment, the cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences) in PBS, permeabilized with 0.125% Triton-X in PBS for 10 minutes, rinsed with PBS, and blocked with 2% w/v BSA for 1 hour at RT. The samples were then incubated with 5 μg/mL 4’,6-diamidino-2-phenylindole (DAPI, Thermo Fisher Scientific) and phalloidin (1:1000 dilution, CruzFluor^™ 647 conjugate, Santa Cruz Biotechnology) in PBS for 1 hour at RT, rinsed, and stored in PBS at 4 °C until analysis.

Microscopy was performed using Zeiss 880 confocal microscopy at 63× magnification as previously described.^[12] Video animations of the confocal images were compiled using Imaris software version 9.5.1 (Oxford Instruments). To examine colocalization of PAs to cholesterol, we performed Manders coefficient calculations using the Coloc 2 plugin from Fiji. The nuclei, shown by DAPI staining, were subtracted from the cytoskeleton, shown by phalloidin staining. The resulting area was combined with the cholesterol area for colocalization analysis.

Cholesterol Efflux Under Inflammatory Conditions In Vitro:

Macrophages were plated on 24-well plates at 2.11 × 10⁵ cells/cm² and loaded with cholesterol as described above. The cells were then treated with 100 ng/mL of interferon gamma (IFN-γ, mouse carrier-free protein, Biolegend) and 32 μM LXR or LXR PA epitope equivalents overnight in complete media. The cells were rinsed with PBS before adding lipopolysaccharide (derived from Escherichia coli 0111:B4, Sigma-Aldrich) at 10 μg/mL in complete media without serum or albumin for 24 hours before flow cytometry analysis. Flow cytometry for induced nitric oxide synthase (iNOS) was performed as previously described with an Attune NxT Acoustic Focusing Cytometer (Thermo Fisher).^[12] Cholesterol and iNOS antibody (eBioscience^™) fluorescence were read using the BL1 and RL1 channels, respectively.

In Vivo Studies:

Low-density lipoprotein receptor knockout (LDLR KO) male and female mice were obtained at four weeks of age (Jackson laboratories, B6.129S7-Ldlr/J stock #002207) and fed a high-fat diet (Envigo, TD.88137) for 14 weeks to develop atherosclerosis. For the targeting studies, PAs were co-assembled with fluorescently labeled PAs according to the processing conditions above and injected into the tail vein of mice at 2 mg/mL in PBS (approximately 1 mM), 8 mg/kg with a 29 G syringe. Prior to injection, mice were anesthetized with inhaled isoflurane (Pivetal). After 24 hours, the mice were sacrificed by inhaled isoflurane followed by cardiac puncture. Mice were then perfused with PBS and 2% paraformaldehyde (in PBS). The heart, lung, liver, kidney, and spleen were removed for fluorescence imaging analysis using an Ami HT Optical Imaging System. The aortic root was cryosectioned at 5 μm based on a previous technique using a Thermo Scientific NX70.^[11][58] Images of PA targeting to plaque in the aortic root were taken on a Zeiss AxioVision Imager.A2 fluorescence microscope using a 5× objective and 561 nm excitation and 600/50 nm emission filters. The fluorescence pixel intensity was quantified and normalized to the plaque area using Fiji software.

Oil Red O (Sigma) staining was performed on the aortic roots following the manufacturer’s protocol. The percent atherosclerosis was calculated as the ratio of plaque area, stained by Oil Red O, to the total aortic root area using Fiji software. Immunofluorescence for ABCA1 and CD68 was performed by first blocking nonspecific binding on the slides using Tris-buffered saline with Tween 20 (TBST, 0.1% Triton) and 10% horse serum for 1 hour at RT. Then, the slides were incubated with anti-CD68 (Abcam, ab125212, 1:500) and anti-ABCA1 (Abcam, ab18180, 1:1000) at 4 °C overnight. The next day, the slides were washed twice with PBS for 5 minutes each wash and incubated with secondary antibodies AlexaFluor 647 (Life Technologies, A32733, 1:1000) and AlexaFluor 555 (Life Technologies, A21425, 1:1000) for 1 hour at RT. The slides were again washed twice with PBS, dipped in distilled water, then mounted on coverslips using ProLong Gold antifade mountant with DAPI (Fisher Scientific). Images were also taken on a Nikon A1R microscope using 2.5× (Oil Red O) or 5× (immunofluorescence) objectives.

Therapeutic studies were performed on mice after 14 weeks of the high-fat diet over an 8-week period. PA or PBS injections were administered twice per week through the tail vein under isoflurane anesthesia. Mice were then sacrificed, perfused, and the aortic root sectioned following the same procedures as in the targeting study. The blood was analyzed by the Animal Histopathology and Laboratory Medicine Core at the University of North Carolina-Chapel Hill for clinical toxicology panels. The liver was sectioned at 7 μm intervals and analyzed for apoptosis using an anti-RIP3 antibody (1:100 dilution; Novus Biologicals NBP1–77299), added to slides diluted in IHC Tek (IHC-Tek diluent, pH 7.4, IHC World) and incubated overnight at 4 °C. The next day, slides were washed 3 times with PBS, then stained with goat anti-rabbit AF555 secondary antibody (1:3000 dilution; Fisher Scientific), for 1 hour in the dark at RT. The slides were then washed 3 times with PBS, quickly rinsed with DI water, and mounted with coverslips using ProLong Gold antifade mountant with DAPI. Finally, the slides were left for 24 hours in the dark to allow the mountant to cure.

Statistical Analysis:

All data were analyzed using JMP^® Pro 15 statistical software (SAS). One way ANOVA with post hoc Tukey-Kramer honestly significant difference (HSD) test or Student’s t-tests were used, as well as two factor ANOVA with post hoc Dunnett’s test. For data that did not follow a normal distribution based on the Welch’s test for unequal variances, nonparametric testing was performed using Wilcoxon multiple comparisons for each pair. A p-value of less than 0.05 indicated statistical significance. All error bars indicate standard error of the mean.