Selective MMP-9 Inhibitor (R)-ND-336 Alone or in Combination with Linezolid Accelerates Wound Healing in Infected Diabetic Mice

Abstract

Diabetic foot ulcers (DFUs) are a common complication of diabetes that are recalcitrant to healing due to persistent inflammation. The majority of DFUs have bacterial biofilms, with Staphylococcus epidermidis as a predominant bacterium, requiring infection control with antibiotics before treatment of the wound. Matrix metalloproteinases (MMPs) play roles in the pathology and repair of DFUs. However, defining the roles of the 24 human MMPs has been challenging due to the presence of three forms for each MMP, of which only one is catalytically competent, and the lack of convenient methods to distinguish among the three forms of MMPs. Using an affinity resin that binds only to the active forms of MMPs, with identification and quantification by mass spectrometry, we found that infected wounds in mice had increased levels of active MMP-9 compared to uninfected ones, paralleling infected human DFUs. MMP-9 activity prevents diabetic wounds from healing. We evaluated the efficacy of the selective small-molecule MMP-9 inhibitor, (R)-ND-336, in the infected diabetic mouse model of wound healing and showed that (R)-ND-336 alone or in combination with the antibiotic linezolid improves wound healing by inhibiting the detrimental MMP-9, mitigating macrophage infiltration to diminish inflammation, and increasing angiogenesis to restore the normal wound healing process. An advantage of this strategy is the ability to administer (R)-ND-336 concurrently with an antibiotic.

Diabetic foot ulcers (DFUs) are a complication of diabetes that affects 26.1 million individuals globally,¹ with an incidence of 19–34%.¹ The inability of these chronic wounds to heal resulted in 108 000 lower-limb amputations in the United States in 2014.² Survival rates are 47.7% at 5 years and 24.1% at 10 years after amputation.³ Treatment of DFUs includes wound debridement, off-loading, and infection control with antibiotics. While skin substitutes such as Apligraf and Dermagraft are available, they are contraindicated for infected DFUs, and only Apligraf was found to offer modest improvement over the standard of care.⁴ There is only one FDA-approved drug, becaplermin, which contains platelet-derived growth factor that stimulates angiogenesis to accelerate wound healing. However, becaplermin is not the first-line therapy in treatment of DFUs due to its modest efficacy and increased risk of cancer and death.⁵

Infection is common in DFUs, with osteomyelitis present in 50–60% of hospitalized patients and in 10–20% of ambulatory patients.⁶ Most of the isolates (68%) in DFUs are Gram-positive bacteria, with Staphylococci accounting for 645 of 1817 isolates.⁷S. epidermidis is more frequently found in patients with ischemic foot ulcers, accounting for 55% of staphylococci isolated from patients with DFUs.⁸ The majority (60%) of DFUs are characterized by the presence of bacterial biofilms,⁹ which are resistant to conventional antibiotics.¹⁰S. epidermidis is one of the predominant bacteria in infected DFUs.¹¹

Matrix metalloproteinases (MMPs) play roles in both the pathology of DFUs and in their repair.¹² For example, MMP-1, MMP-2, MMP-9, MMP-13, and MMP-14 are involved in keratinocyte migration; MMP-3 affects wound contraction and delayed healing; MMP-7 is required for re-epithelialization; MMP-8 and MMP-13 promote wound healing; MMP-10 overexpression results in normal wound healing; and MMP-12 regulates angiogenesis.¹² However, defining the roles of the 24 human MMPs has been challenging due to the existence of three forms for each MMP: inactive zymogens or proMMPs, active MMPs arising from removal of the prodomain, and tissue inhibitor of matrix metalloproteinase (TIMP)-complexed MMPs. Of these, the proMMPs and TIMP-MMPs are catalytically incompetent, thus, they do not play a role in pathology or repair. Only the active unregulated MMPs are catalytically competent, thereby playing roles in pathology or repair. Most methods for MMP profiling do not distinguish among the three forms of MMPs. In addition, these methods require screening for a specific MMP with antibodies, fluorescent substrates, or a library of selective MMP-directed probes, rather than “fishing” out and identifying the MMP(s) that plays critical roles in repair and disease.

We devised an affinity resin that couples covalently a broad-spectrum MMP inhibitor with Sepharose resin¹³ and binds exclusively to the active forms of MMPs, which are subsequently identified and quantified by mass spectrometry (MS).¹⁴ We identified only active MMP-8 and MMP-9 in wounds of diabetic (db/db) mice.¹⁵ Selective inhibition of MMP-9 or MMP-9 gene knockout accelerated wound healing, indicating that MMP-9 was detrimental.^15,16 Selective inhibition of MMP-8 delayed wound healing,¹⁵ while exogenously added recombinant MMP-8 accelerated wound healing in diabetic mice,¹⁶ demonstrating that MMP-8 plays a repair function. The best strategy for treatment of DFUs would appear to be a selective inhibitor of MMP-9 that suppresses the root cause of why DFUs do not heal—the MMP-9 activity—and spares the activity of MMP-8 to allow the body’s natural repair mechanisms to regenerate the damaged tissue. We recently reported on the discovery of (R)-ND-336, a selective small-molecule inhibitor of MMP-9.¹⁷ (R)-ND-336 inhibits MMP-9 as a slow-binding inhibitor with a K_i value of 19 ± 3 nM and a long residence time bound to MMP-9 of 300 ± 1 min.¹⁷ In contrast, (R)-ND-336 poorly inhibits MMP-8 as a noncompetitive inhibitor with K_i of 8590 ± 230 nM and binds to MMP-8 with a short residence time of seconds. The exquisite selectivity of (R)-ND-336 resulted in superior efficacy compared to becaplermin (platelet-derived growth factor) in uninfected db/db mice.¹⁷ We showed that (R)-ND-336 decreases the levels of reactive oxygen species and nuclear factor kappa beta (NF-κβ)¹⁷ that induce dermal inflammation.¹⁸ Furthermore, we found that the more severe and infected human DFUs have 34× higher active MMP-9 levels than nondiabetic controls.¹⁷ The questions that become pertinent are whether increased levels of active MMP-9 are found in infected db/db mouse wounds and if (R)-ND-336 has efficacy in an infected animal model that parallels the human condition with commonly infected DFUs.

Results and Discussion

We investigated S. aureus ATCC 29213, S. epidermidis ATCC 12228, and S. epidermidis ATCC 35984 for their ability to form biofilms by staining with crystal violet. S. epidermidis ATCC 12228 was nonadherent (OD ≤ OD_control),¹⁹S. aureus ATCC 29213 was weakly adherent (OD_control < OD ≤ 2× OD_control),¹⁹ and S. epidermidis ATCC 35984 was strongly adherent (2× OD_control < OD ≤ 4x OD_control)¹⁹. Therefore, we selected S. epidermidis ATCC 35984.

Healing was delayed in S. epidermidis infected compared to uninfected diabetic wounds (Figure 1A–C). Wounds were larger on day 3 than when initially created (day 0) in both groups. However, uninfected diabetic wounds started healing on day 7, while infected wounds did not start healing until day 10 (P = 0.025). Re-epithelialization was increased in the uninfected wounds (Figure 1D) compared to the infected ones (Figure 1E).

Figure 1.

Infected mouse model of diabetic wound healing. Wounds (8 mm full-thickness) were infected on day 1 and compared to uninfected ones. (A) Wound area measurements relative to day 0, mean ± SEM; n = 6 for uninfected and n = 9 for infected on day 3, n = 9 for uninfected and n = 19 for infected on day 7, n = 4 for uninfected and n = 5 for uninfected on day 10, n = 8 for uninfected and n = 11 for infected on day 14; *P < 0.05 by Mann–Whitney U two-tail test. (B) Representative images of the wounds. (C) Scanning electron microscopy of S. epidermidis ATCC 35984 shows biofilm formation, scale bar 2 μm. Representative H&E of (D) uninfected wound and (E) infected wound on day 14, scale bar 50 μm; re-epithelialization shown by the white line. Levels of (F) IL-6 and (G) VEGF in uninfected and infected mouse wounds measured by ELISA, mean ± SD, n = 4 mice per group per day, *P < 0.05, **P < 0.01 by Student’s t test with two tails.

As chronic wounds are characterized by prolonged inflammation and decreased angiogenesis, we measured interleukin (IL)-6, a pro-inflammatory factor that plays a role in wound healing,^20,21 and vascular endothelial growth factor (VEGF), which mediates angiogenesis in wound healing,²² using ELISA (Figure 1F,G). IL-6 levels were significantly increased on days 3 (P = 0.013) and 7 (P = 0.031) in infected wounds compared to uninfected ones, while VEGF decreased on days 7 (P = 0.039) and 14 (P = 0.042).

While an infected db/db mouse model has been reported,²³Pseudomonas aeruginosa, a Gram-negative bacterium, was used to infect the wounds. Gram-positive bacteria are found in DFUs three times more frequently than Gram-negative bacteria, with P. aeruginosa accounting for 3.2% of isolates.⁷ As such, our current mouse model with S. epidermidis is more clinically relevant. We found increased inflammation and decreased angiogenesis in S. epidermidis infected diabetic mouse wounds.

As we had observed increased levels of active MMP-9 in human DFUs, especially in the more severe and infected wounds,¹⁷ we analyzed the mouse wounds with our affinity resin. The affinity resin (Figure 2A) consists of a broad-spectrum MMP inhibitor (red) attached covalently to Sepharose resin (blue) via a linker (black) and binds only to active forms of MMPs (or the related ADAMs),¹⁴ as the active site is accessible (Figure 2B, open gray enzyme with active site exposed), but not to pro-MMPs (gray enzyme with pro-domain in yellow covering the active site) or TIMP-complexed MMPs (gray enzyme with TIMP in green covering the active site). The captured active MMPs are trypsin digested and the resulting peptides are identified by mass spectrometry. This method allows not only for identification of active MMPs, but also for their quantification. Using the affinity resin analysis, we identified only active MMP-8 and MMP-9 in wounds of infected diabetic mice, the exact same active MMPs identified in human DFUs.¹⁷ No statistical differences in levels of active MMP-8 were observed in infected wounds compared to uninfected ones (Figure 2C). However, concentrations of active MMP-9 (Figure 2D) increased 20% on day 7 in infected mouse wounds compared to uninfected controls (P = 0.045); in human DFUs we found 5 times higher levels of active MMP-9 in the more severe and infected human DFUs relative to uninfected ones.¹⁷ Thus, this model is clinically relevant.

Figure 2.

Upregulation of active MMP-9 in infected diabetic mouse wounds. (A) Structure of the affinity resin; a broad-spectrum MMP inhibitor is covalently attached to Sepharose resin. (B) Method for the identification and quantification of active MMPs. Wound homogenate is incubated with the affinity resin; only active MMPs bind to the affinity resin. Cysteines in active MMPs are reduced with dithiothreitol (DTT), then alkylated with iodoacetamide (IAA), followed by trypsin digestion. The resulting peptides are identified by MS using a peptide database engine; active MMPs are quantified by MS/MS. (C,D) Analysis of infected diabetic wounds with the affinity resin-identified active MMP-8 and MMP-9. (C) No difference in the levels of active MMP-8 was observed between infected and uninfected wounds, while (D) active MMP-9 was upregulated in infected wounds. Mean ± SD, n = 4 mice per group per day, *P < 0.05 by Student’s t test with two tails.

The presence of active MMP-8 and MMP-9 in wounds of infected diabetic mice was confirmed by in situ zymography. We used dye-quenched (DQ)-collagen for MMP-8 activity and DQ-gelatin for MMP-9 activity. Of the two collagenases (MMP-8 and MMP-13) in mice, only active MMP-8 is observed in diabetic mouse wounds.¹⁵ Similarly, of the two gelatinases (MMP-2 and MMP-9), only active MMP-9 is present in diabetic mice.¹⁵ However, other proteases can digest DQ-collagen and DQ-gelatin.²⁴ There was no difference in MMP-8 activity (green) in infected compared to uninfected wounds (Figure 3A), while slightly higher MMP-9 activity (green) was discernible in infected wounds (Figure 3B).

Figure 3.

Infected mouse model of diabetic wound healing shows upregulation of active MMP-9 and macrophages. Representative in situ zymography of day 7 wounds shows (A) no difference in MMP-8 activity (green) in infected wounds relative to uninfected ones with increased macrophage staining with F4/80 (red) and (B) measurable increase in MMP-9 activity (green) in infected wounds with increased macrophages (red); DAPI for DNA nuclear staining in blue; scale bars 50 μm.

Macrophages play important roles in wound healing.²⁵ After injury, neutrophils arrive to the wound and produce reactive oxygen species (ROS) to mitigate or prevent infection, then monocytes are converted into macrophages, which phagocytose bacteria to clean the wound. Macrophages are also known to produce and release MMP-9,²⁶ which is needed for macrophage migration during the inflammation stage of wound healing. We used F4/80 as a marker for wound macrophages and found that infected mouse wounds have increased macrophage recruitment (Figure 3, red); the merged images indicated colocalization of MMP-8 and MMP-9 activity with macrophages.

We had previously reported that the selective small-molecule MMP-9 inhibitor (R)-ND-336 accelerated wound healing in uninfected diabetic mice.¹⁷ We wanted to ascertain whether (R)-ND-336 was efficacious in an infected diabetic mouse model of wound healing, a more clinically relevant animal model, and whether if used in combination with an antibiotic, it improved wound healing further.

We investigated the activity of antibiotics against S. epidermidis ATCC 35984. The minimal-inhibitory concentration (MIC) values were determined to be 1 μg/mL for linezolid, 4 μg/mL for vancomycin, >256 μg/mL for gentamicin, and 0.125 μg/mL for rifampin. Since linezolid is effective for treatment of Gram-positive infections in DFU patients,²⁷ we selected this antibiotic for evaluation in infected diabetic mice.

Before conducting the combination study, we investigated the potential for drug–drug interactions of (R)-ND-336. Alteration in the activities of CYP450 enzymes, which play important roles in the metabolism of drugs, via inhibition or induction is the major cause of drug–drug interactions when two or more drugs are coadministered. Inhibition of CYP450 enzyme(s) responsible for the metabolism of one drug can cause an increase in the plasma concentration of a second drug, leading potentially to serious adverse drug reactions.²⁸ Predicting in vivo drug–drug interactions from in vitro studies may prevent adverse drug reactions in the clinic. We investigated the potential of (R)-ND-336 for drug–drug interactions (Figure 4). (R)-ND-336 inhibits CYP1A2 (Figure 4A), CYP2C8 (Figure 4C), CYP2C9 (Figure 4D), and CYP2C19 (Figure 4E) with IC₅₀ values of 7.9 μM, 39.0 μM, 3.1 μM, and 3.5 μM, respectively, and it does not inhibit CYP3A4/A5 (Figure 4B) and CYP2D6 (Figure 4F). These concentrations are unlikely to be achieved in systemic circulation after topical administration of (R)-ND-336.

Figure 4.

Inhibition of CYP450 metabolizing enzymes by (R)-ND-336. Recombinant human CYPs were incubated with (R)-ND-336, marker substrate, and NADPH. Decreases in metabolite formation compared to the control (in the absence of (R)-ND-336) were used to calculate IC₅₀ values. (A) Phenacetin O-deethylation for CYP1A2, (B) testosterone 6β-hydroxylation for CYP3A4/5, (C) paclitaxel 6α-hydroxylation for CYP2C8, (D) diclofenac 4′-hydroxylation for CYP2C9, (E) S-mephenytoin for CYP2C19, and (F) bufaralol 1′-hydroxylation for CYP2D6.

We then investigated whether (R)-ND-336 has any antibacterial activity. Up to a concentration of 128 μg/mL we observed no such activity. To determine the appropriate doses to use, we conducted a dose–response study with topical linezolid and (R)-ND-336 (Figure 5A,B) each at three dose levels. The wound area in the linezolid-treated mice was not statistically different (P > 0.05) than that of the vehicle-treated animals (Figure 5C,D), while all three dose levels of (R)-ND-336 showed statistically significant improvement (P < 0.05) compared to the vehicle control. On the other hand, treatment with (R)-ND-336 did not affect bacterial count (Figure 5E, red), while all three dose levels of linezolid reduced bacterial load by 0.79-, 0.77-, and 1.85-log₁₀ in the 0.5, 2.5, and 5 μg/wound/day groups, respectively (Figure 5E, blue). The dose of 5 μg/wound/day of linezolid was chosen as it resulted in almost 2-log₁₀ reduction in bacteria. Since all three dose levels of (R)-ND-336 resulted in similar wound areas on day 10, we selected the middle dose level of 10 μg/wound/day of (R)-ND-336 for further studies.

Figure 5.

Dose response of linezolid and (R)-ND-336. Full-thickness excisional wounds (8 mm) were created on day 0 and infected with S. epidermidis ATCC 35984 on day 1. Mice were randomized to groups on day 7 and treated with vehicle (water) or 0.5, 2.5, or 5 μg of linezolid per wound per day or 5, 10, or 25 μg of (R)-ND-336 per wound per day starting on day 7 once a day until day 9 (total three doses). (A) Structure of (R)-ND-336. (B) Structure of linezolid. (C) Wound area relative to day 0; mean ± SEM; n = 35 mice on day 7, n = 5 mice per group on day 10; *P < 0.05, **P < 0.01 by Mann–Whitney U two-tail test. (D) Representative images of the wounds on day 10. (E) Bacterial counts in the wounds on day 10, mean ± SD, n = 4 mice per group; **P < 0.01 by Student’s t test with two tails.

We evaluated the efficacy of linezolid, (R)-ND-336, and (R)-ND-336 + linezolid, compared to vehicle in wounds of infected diabetic mice. Wound areas and representative images of the wounds are shown in Figure 6A,B. Linezolid did not accelerate wound healing, with wound areas similar to those of the vehicle control (P = 0.67, 0.97, and 0.47 on days 10, 14, and 21, respectively). (R)-ND-336-treated wounds were smaller than vehicle-treated ones at all times (P = 0.033, 0.036, and <0.049, on days 10, 14, and 21, respectively). The combined (R)-ND-336 + linezolid showed further acceleration of wound healing compared to vehicle at all times (P = 0.00002, 0.037, and 0.008 on days 1, 14, and 21, respectively). The combined treatment was also statistically significant compared to both linezolid alone (P = 0.0043, 0.0056, and 0.00017 on days 10, 14, and 21, respectively) and to (R)-ND-336 alone only on day 21 (P = 0.035). H&E analysis showed complete re-epithelialization for the infected wounds treated with (R)-ND-336 alone or the combined therapy, whereas only partial re-epithelialization was observed in vehicle- and linezolid-treated wounds (Figure 6C).

Figure 6.

(R)-ND-336 alone or in combination with linezolid accelerates wound healing in infected diabetic mice. Wounds (8 mm full-thickness) were created on day 0 and infected with S. epidermidis ATCC 35984 on day 1. Mice (n = 12 per group) were randomized on day 7 and the wounds were treated topically with vehicle (water), 5 μg linezolid, 10 μg (R)-ND-336, or combined 5 μg linezolid +10 μg (R)-ND-336 once a day starting on day 7 until day 21. (A) Wound area measurements relative to day 0, mean ± SEM; n = 11 mice for vehicle, n = 12 mice for linezolid, (R)-ND-336 and (R)-ND-336 + linezolid on day 10; n = 8 mice per vehicle and linezolid, n = 9 mice for (R)-ND-336 and (R)-ND-336 + linezolid on day 14; n = 5 mice for vehicle, n = 6 mice for linezolid, (R)-ND-336 and (R)-ND-336 + linezolid on day 21; * P < 0.05 and **P < 0.01 by Mann–Whitney U two-tail test. (B) Representative images of the wounds. (C) Representative H&E of the wounds on day 21, re-epithelialization shown by the white line, scale bar 50 μm.

The (R)-ND-336-treated group showed a 0.69-log₁₀ reduction in CFU/wound compared to vehicle (Figure 7A), however it was not statistically significant (P = 0.18). In contrast, treatment with linezolid or (R)-ND-336 + linezolid resulted in 1.94- (P = 0.04) and 2.01-log₁₀ (P = 0.03) bacteria reduction, respectively. Measurement of VEGF, a marker for angiogenesis, on day 10 wounds revealed no differences in linezolid-treated wounds compared to vehicle (P = 0.47, Figure 7B). In contrast, VEGF increased in the (R)-ND-336-treated wounds (P = 0.043) and in the (R)-ND-336 + linezolid group (P = 0.038, Figure 7B). On the other hand, levels of IL-6, a marker for inflammation, on day 10 were similar across the groups (P > 0.05, Figure 7C), as inflammation in the wound healing cycle occurs early (on day 3, Figure 1F).

Figure 7.

(R)-ND-336 alone or in combination with linezolid increases VEGF and decreases macrophages in infected diabetic mouse wounds. Wounds were treated topically with vehicle (water), 5 μg linezolid, 10 μg (R)-ND-336, or combined 5 μg linezolid +10 μg (R)-ND-336 once a day starting on day 7 until day 21. (A) Bacterial counts on day 10 show 1.9- and 2.0-log₁₀ reduction in linezolid and combined groups, respectively; mean ± SD, n = 4 mice per group, *P < 0.05 by Student’s t test with two tails. (B) Increased VEGF levels by ELISA are observed in the (R)-ND-336 and combined groups on day 10, mean ± SD, n = 4 mice per group, *P < 0.05 by Student’s t test with two tails. (C) IL-6 levels by ELISA are similar (P > 0.05 by Student’s t test with two tails) across the groups on day 10, mean ± SD, n = 4 mice per group. (D) Representative in situ zymography of day 14 wounds with DQ-Gelatin (green) shows inhibition of MMP-9 activity (green) in (R)-ND-336 and combined groups, decreased or no macrophages (F4/80, red) in linezolid, (R)-ND-336, and combined groups relative to vehicle control. (E) Representative in situ zymography of day 14 wounds with DQ-Collagen (green) shows no difference in MMP-8 activity (green) in all groups, decreased or no macrophages (red); DAPI for DNA nuclear staining in blue; scale bars 50 μm.

In-situ zymography with DQ-gelatin showed that treatment with (R)-ND-336 or (R)-ND-336 + linezolid suppressed MMP-9 activity (Figure 7D, no green fluorescence). F4/80, a marker for macrophage infiltration occurring during inflammation, decreased in linezolid-treated wounds (Figure 7D,E, decreased red compared to vehicle), and was absent in (R)-ND-336 or (R)-ND-336 + linezolid treated wounds (Figure 7D,E, absence of red). Inhibition of MMP-8 activity was not observed in any treatment (Figure 7E, green fluorescence).

Current therapies for the treatment of infected DFUs require first the use of antibiotics to clear the infection prior to further wound treatment, such as becaplermin and/or skin substitutes. An advantage of our strategy is the ability to coadminister (R)-ND-336 with an antibiotic. We showed in an infected diabetic mouse model of delayed wound healing that (R)-ND-336 alone or in combination with linezolid improves wound healing by mitigating macrophage infiltration and increasing the angiogenesis marker VEGF. Thus, (R)-ND-336 inhibits the detrimental MMP-9, decreases inflammation, and increases angiogenesis that restores the normal wound-repair process, demonstrating that (R)-ND-336 converts a chronic wound to an acute wound.

Materials and Methods

Animals

The incidence of DFUs in humans is similar between males and females (6.0% and 5.9%, respectively).²⁹ Therefore, we chose to evaluate the efficacy of the selective MMP-9 inhibitor (R)-ND-336 only in female mice. Female db/db mice (BKS.Cg-Dock7^m+/+Lepr^db/J, 8-weeks old, 40.6 ± 3.8 g) were purchased from Jackson Laboratory. Animals were fed 5001 Laboratory Rodent Diet (LabDiet) and water ad libitum. Mice were housed in cages containing ground corn cob (Bed ‘O Cobs, The Andersons, Inc.) and Alpha Dri (Shepherd Specialty Papers) bedding at 72 ± 2 °F and exposed to 12 h light/12 h dark. The studies were conducted with approval and oversight of the University of Notre Dame Institutional Animal Care and Use Committee.

Bacterial Strains and Electron Microscopy

S. aureus ATCC 29213, S. epidermidis ATCC 12228, and S. epidermidis ATCC 35984 (a biofilm-producing strain)³⁰ were purchased from ATCC. Bacteria were incubated for 8 h at 37 °C under rotation to a concentration of approximately 10⁵ CFU/mL. The culture was inoculated into tryptic soy broth at a 1:100 dilution and seeded on a cover glass at 37 °C under static conditions. After growing overnight, planktonic bacteria were removed by rinsing the glass three times with sterile 0.1 M sodium cacodylate and fixed with 2% (v/v) paraformaldehyde (Electron Microscopy Sciences) for 1 h at room temperature. Following fixation, the samples were washed with 0.1 M sodium cacodylate and stained with 1% (v/v) osmium tetroxide for 1 h. The cover glass underwent a series of ethanol dehydration (50%, 70%, 80%, 95%, and 100%) and critical drying point with CO₂. Samples were mounted onto scanning electron microscope stubs and sputter coated with iridium/carbon. Images were acquired using a Magellan 400 scanning electron microscope (TSS Microscopy).

Test Compounds

(R)-ND-336 was synthesized as previously described,¹⁷ with a purity of >95%. Linezolid was purchased from AmplaChem Inc. Vancomycin, gentamicin, and rifampin were purchased from Sigma-Aldrich. The compounds were dissolved in water, filter-sterilized on an Acrodisc syringe filter (VWR), and stored at 4 °C. The dosing solutions were freshly prepared every day.

Infected Wound Healing Animal Model

The dorsal area of the mice was shaved, swabbed with povidone and alcohol, and a single 8 mm diameter wound was made on day 0 with a biopsy punch (Miltex), while the mice were under isofluorane anesthesia. One day later (day 1), the wound (n = 32 mice) was infected with 50 μL of 10⁶ CFU/mL of S. epidermidis ATCC 35984. Semiocclusive Tegaderm dressing (3 M Company) was used to cover the wound. A control group (n = 12 mice) was not infected. On days 0, 3, 7, 10, and 14, the wounds were photographed with a Nikon D5300 camera (Nikon, Inc.) on a tripod; the distance was fixed and a ruler was included in the photographic frame. Wound area was determined by NIH ImageJ software (version 1.52). Wound area was calculated as percent change relative to day 0. On days 3, 7, and 14, n = 4 mice per group were sacrificed, the wounds were harvested, immediately frozen in liquid nitrogen, and stored at −80 °C for cytokine measurements. On days 3 and 7, the wounds were harvested, frozen, and stored at −80 °C for analysis with our affinity resin.

Affinity Resin

The affinity resin was synthesized as previously described³¹ in 14 synthetic steps. The wound samples (100 mg, n = 4 infected mice and 4 uninfected mice) on days 3 and 7 were homogenized in 1 mL of cold 25 mM Tris·HCl pH 7.5, 100 mM NaCl, 1% v/v Nonidet P-40 in a Bullet Blender (Next Advance). An aliquot (10 mg) of tissue homogenate was incubated with 100 μL of affinity resin for 2 h at 4 °C and processed as described previously.³² The peptide mixture was analyzed by electrospray ionization in a Thermo Q Exactive Orbitrap mass spectrometer (Thermo Fisher Scientific) and Uniprot database to identify active MMPs. Quantification of the identified active MMPs was determined on a 6500 QTrap mass spectrometer (AB Sciex) using three peptides per MMP and three Q1 → Q3 transitions per peptide. The peptides used for quantifying MMPs were custom-synthesized and purchased from Genscript. Calibration curves in control mouse wound samples containing known amounts of the synthetic peptides relative to internal standard (yeast enolase) were used to quantify the MMPs. The peptides and transitions are shown in Table 1.

Table 1. Identification and Quantification of Active MMPs in Mouse Wounds^a.

protein	peptide sequence	Q1 precursor ion m/z	Q3 product ion m/z
MMP-8	QYWALSGYDLQQGYPR	972.97 [M + 2H]2+	1196.57 [M + H]+y10
			1139.55 [M + H]+ y9
			976.48 [M + H]+y8
	DISNYGFPR	534.76 [M + 2H]2+	840.40 [M + H]+y7
			753.37 [M + H]+y6
			639.32 [M + H]+ y5
	TYFFINNQ[C]WR	774.86 [M + 2H]2+	1137.53 [M + H]+ y8
			990.46 [M + H]+ y7
			877.37 [M + H]+y6
MMP-9	DMIDDAFAR	527.24 [M + 2H]2+	807.40 [M + H]+ y7
			694.32 [M + H]+y6
			579.29 [M + H]+ y5
	AFAVWGEVAPLTFTR	832.94 [M + 2H]2+	1090.59 [M + H]+y10
			904.53 [M + H]+ y8
			734.42 [M + H]+ y6
	GSPLQGPFLTAR	622.34 [M + 2H]2+	889.49 [M + H]+ y8
			761.43 [M + H]+y7
			704.41 [M + H]+ y6
Internal Standard	TAGIQIVADDLTVTNPK	878.48 [M+2H]2+	1172.62 [M + H]+y11
			1002.51 [M + H]+ y9
			772.46 [M + H]+ y7
	VNQIGTLSESIK	644.86 [M + 2H]2+	1075.60 [M + H]+y10
			777.44 [M + H]+ y7
			676.39 [M + H]+ y6
	NVNDVIAPAFVK	643.86 [M + 2H]2+	844.53 [M + H]+ y8
			632.38 [M + H]+y6
			561.34 [M + H]+ y5

^aQuantification was done using three peptides per MMP. Precursor m/z ions were selected in the first quadrupole (Q1), then fragmented in Q2. The specified fragment was monitored in Q3. The fragments used as “quantifier” are represented in bold and were assigned using conventional peptide fragmentation nomenclature. [C] = alkylated cysteine or carbamidomethyl cysteine.

Interleukin (IL)-6 and Vascular Endothelial Growth Factor (VEGF) Quantification

Wounds were homogenized in cold lysis buffer containing EDTA-free protease inhibitor cocktail (Pierce) using a Bullet Blender. The homogenate was centrifuged (20000g for 20 min at 4 °C) and protein content was determined by BCA. Samples were normalized and stored at −80 °C until analysis. The levels of IL-6 and VEGF in the wound lysates were quantified by ELISA following the manufacturer’s protocols (Abcam).

Histology and In-Situ Zymography

The harvested wounds were embedded in optimum cutting temperature compound (Tissue-Tek) and cryosectioned at 12-μm thickness for H&E staining and at 8-μm thickness for in situ zymography and F4/80 staining. Analysis for re-epithelialization was conducted³³ using a Nikon Eclipse 90i Fluorescent Microscope (Nikon Instruments Inc.). In-situ zymography with DQ-gelatin or DQ-collagen (Molecular Probes, Inc.) was conducted as reported previously.¹⁶ F4/80 immunofluorescence staining was conducted following a published procedure with minor modifications;³⁴ the cryosectioned wound tissues were blocked with PBS containing 5% BSA/0.1% Tween-20, and incubated 1 h with antimouse F4/80 (BioLegend). The embedded tissues were counterstained with DAPI (Invitrogen), imaged by confocal microscopy, and analyzed by ImageJ.

Dose–Response Study with (R)-ND-336 and Linezolid

A study to determine the doses of (R)-ND-336 and linezolid was conducted. Mice (n = 5 mice per group, 7 groups, total 35 mice) were given full-thickness excisional wounds (day 0) and were infected the following day (day 1). On day 7, the mice were randomized to vehicle (water), 10 μg/mL linezolid, 50 μg/mL linezolid, 100 μg/mL linezolid, 100 μg/mL (R)-ND-336, 200 μg/mL (R)-ND-336, and 500 μg/mL (R)-ND-336. Wounds were treated with 50 μL of solution per wound per day (equivalent to 0.5, 2.5, and 5 μg/wound/day of linezolid and 5, 10, and 25 μg/wound/day of (R)-ND-336) once a day until day 10. Wounds were photographed on days 0, 7, and 10. On day 10, the mice were sacrificed and the wounds were harvested for bacterial counts.

Minimal-Inhibitory Concentration (MIC)

The MICs of linezolid, vancomycin, gentamicin, rifampin, and (R)-ND-336 against S. epidermidis ATCC 35984 were determined in cation adjusted Mueller-Hinton II broth (Becton, Dickinson and Co.) using the microdilution technique.³⁵ The MICs were recorded as the lowest concentration of compound that produced no visible bacterial growth.

Bacterial Count

Wounds were harvested and excised, and placed in separate sterilized Eppendorf tubes containing 1 mL of PBS. Samples were homogenized with a Bullet Blender and the homogenates were serially diluted and plated on tryptic soy agar (Becton, Dickinson and Co.) and incubated overnight at 37 °C. CFUs were determined by the standard colony counting method.³⁶

In Vitro CYP Inhibition

The following marker reactions were used for CYP450-mediated metabolism: phenacetin O-deethylation for CYP1A2, testosterone 6β-hydroxylation for CYP3A4/5, paclitaxel 6α-hydroxylation for CYP2C8, diclofenac 4′-hydroxylation for CYP2C9, S-mephenytoin 4′-hydroxylation for CYP2C19, and bufaralol 1′-hydroxylation for CYP2D6. Phenacetin and bufuralol were purchased from Sigma-Aldrich; S-mephenytoin 4′-hydroxylation, paclitaxel, and diclofenac were purchased from Cayman Chemical Company.

(R)-ND-336 at concentrations of 0.01, 0.05, 0.1, 0.5, 1, 5, 10, 25, and 50 μM was incubated at 37 °C for 15 min with recombinant human CYP enzymes (40 pmol protein/mL, XenoTech, LLC) with the marker substrate (at concentrations approximately equal to K_m; 40 μM for S-mephenytoin; 20 μM for phenacetin and testosterone; 10 μM for paclitaxel, diclofenac, and bufuralol) in potassium phosphate buffer (50 mM, pH 7.4) containing 3 mM MgCl₂ to a final incubation volume of 200 μL. The reaction was started by the addition of 1 mM NADPH (XenoTech, LLC). The reaction was terminated by the addition of 200 μL of cold acetonitrile containing an internal standard. Precipitated protein was removed by centrifugation (15 000g for 20 min at 4 °C). The supernatant was analyzed by UPLC with UV detection at 245 nm or by MS with multiple reaction monitoring. The chromatographic conditions consisted of a Kinetex 2.6 μm C18 2.1 mm × 100 mm column (Phenomenex) at 0.4 mL/min with 2 min 10% acetonitrile/90% water containing 0.1% formic acid, 8 min linear gradient to 90% acetonitrile/10% water containing 0.1% formic acid. A decrease in metabolite formation compared to the control (in the absence of (R)-ND-336) was used to calculate the IC₅₀ value.

Combined (R)-ND-336 and Linezolid Study

Mice (n = 12 per group, 4 groups, total 48 mice) were given full-thickness excisional wounds on day 0, infected on day 1, and randomized on day 7: vehicle (water), 100 μg/mL linezolid, 200 μg/mL (R)-ND-336, and 100 μg/mL linezolid + 200 μg/mL (R)-ND-336. Starting on day 7, wounds were treated topically with 50 μL per wound per day (equivalent to 5 μg/wound/day of linezolid, or 10 μg/wound/day of (R)-ND-336, or combined treatment) for the next 14 days. On day 10, n = 4 mice/group were sacrificed, and the wounds were harvested and homogenized for bacterial counts and VEGF and IL-6 levels. On day 14, n = 4 mice/group were sacrificed, and the wounds were harvested and embedded in optimum cutting temperature compound, and cryosectioned for in situ zymography and F4/80 staining for macrophages. The remaining mice were sacrificed on day 21, and the wounds were harvested for H&E analysis.

Statistical Analysis

Wound areas were analyzed for statistical significance using Mann–Whitney U two-tail test (GraphPad Prism 5). VEGF, IL-6, bacterial counts, and active MMP-8 and MMP-9 levels were analyzed by the Student t-test (Excel) using a two-tail distribution and unequal variance; the data was normally distributed. P < 0.05 was considered statistically significant.

Author Contributions

Z.P. determined MICs, developed the infected model, measured IL-6 and VEGF, conducted the affinity resin with proteomic analysis, determined the appropriate doses of (R)-ND-336 and linezolid, and performed the (R)-ND-336, linezolid, and combination study. T.T.N. and B.A. performed H&E analysis, in situ zymography, DAPI, and macrophage staining for the (R)-ND-336, linezolid, and combination study. W.S. and T.T.N. performed CYP inhibition assays. W.R.W. and V.A.S. conducted the in-life portion of the animal studies. D.H. and M.L. synthesized the affinity resin under the direction of S.M. M.C. designed and directed the studies, and wrote the manuscript with contributions from the authors.

This study was supported by Department of Defense Therapeutic Development Award W81XWH1910493 and by the American Diabetes Association Pathway to Stop Diabetes Grant 1-15-ACN-06.

The authors declare the following competing financial interest(s): U.S. Patent 9,604,957, U.S. Patent 9,951,035, U.S. Patent 10,253,103, European Patent 3107905, Japan Patent 6,5,21,995, and China Patent ZL201580023723.6 have been issued for (R)-ND-336.