Efficacy of a Multimechanistic Monoclonal Antibody Combination against Staphylococcus aureus Surgical Site Infections in Mice

Surgical site infections (SSIs) are commonly caused by Staphylococcus aureus. We report that a combination of three monoclonal antibodies (MEDI6389) that neutralize S. aureus alpha-toxin, clumping factor A, and four leukocidins (LukSF, LukED, HlgAB, and HlgCB) plus vancomycin had enhanced efficacy compared with control antibody plus vancomycin in two mouse models of S. aureus SSI.

ABSTRACT

Surgical site infections (SSIs) are commonly caused by Staphylococcus aureus. We report that a combination of three monoclonal antibodies (MEDI6389) that neutralize S. aureus alpha-toxin, clumping factor A, and four leukocidins (LukSF, LukED, HlgAB, and HlgCB) plus vancomycin had enhanced efficacy compared with control antibody plus vancomycin in two mouse models of S. aureus SSI. Therefore, monoclonal antibody-based neutralization of multiple S. aureus virulence factors may provide an adjunctive perioperative approach to combat S. aureus SSIs.

INTRODUCTION

Despite advances in aseptic surgery, decolonization efforts, and perioperative antibiotics (1–3), Staphylococcus aureus remains the most common cause of surgical site infections (SSIs), with half complicated by methicillin-resistant S. aureus (MRSA) (4). Vancomycin (Vanc) is recommended for perioperative coverage in MRSA-colonized patients or in populations with a high prevalence of MRSA (5). Because antibiotic stewardship programs aim to reduce antibiotic usage to prevent other infections (e.g., Clostridium difficile) and select for resistant bacteria, there is an unmet need for nonantibiotic coverage against SSIs (6).

S. aureus produces virulence factors involved in SSI and biofilm formation, which inhibit antibiotic penetration and immune clearance (7–9). Notably, inhibition of alpha-toxin (AT, also known as hla) and/or clumping factor A (ClfA) by antibody-based vaccines or monoclonal antibodies (MAbs) was efficacious in preclinical models of S. aureus skin/wound and implant-associated infections (10–12). The combined anti-AT and anti-ClfA MAbs were significantly more efficacious than the individual MAbs in mouse models of S. aureus bacteremia and hematogenous implant infection (12–14). Additionally, S. aureus produces leukocidins, including Panton-Valentine leukocidin (encoded by LukSF-PV genes), LukAB, LukED, and gamma-hemolysin (comprised of HlgAB and HlgCB), which promote inflammation and tissue damage and might contribute to SSIs (15, 16). Herein, we evaluated MEDI6389, a combination of three human MAbs (anti-AT [MEDI4893*], anti-ClfA [SAR114], and anti-LukF/LukD/HlgB cross-reactive MAb [SAN481]), which neutralize AT, ClfA, LukSF, LukED, HlgAB, and HlgCB, as adjunctive perioperative coverage in two models of SSI.

An S. aureus skin SSI model (17–19) was modified, whereby diabetic 10-week-old TallyHo/JngJ mice (blood glucose, >300 mg/dl) were anesthetized (2% isoflurane) and placed on a 37°C pad, and their backs were shaved and prepped with three betadine and 70% alcohol scrubs. A midline longitudinal 1-cm full-thickness skin incision was made (11-blade scalpel) (Fig. 1A, I). A 4-0 braided-silk suture with a C-3 needle was passed through the undersurface of the skin distal to the incision (Fig. 1A, II), tied twice, and cut, leaving a 1-cm tail (Fig. 1A, III). The suture was placed into the surgical site, and an inoculum of MRSA (strain USA300 LAC::lux [20]; 1 × 10⁵ CFU in 5 μL phosphate-buffered saline [PBS]) was pipetted onto the suture (Fig. 1A, IV) before closure with a wound clip (Fig. 1A, V and VI).

FIG 1.

Enhanced efficacy of MEDI6389 in a skin SSI model. The mouse model of skin SSI with MRSA strain USA300 LAC::lux was performed in diabetic TallyHo/JngJ mice (n = 10/group). (A) Surgical procedures: full-thickness incision (I), placement of a silk suture on the undersurface of the skin (II), 1-cm suture tail (III), inoculum of MRSA pipetted onto the suture in the surgical site (IV), closure with a wound clip (V, VI). (B) Timeline of preoperative and postoperative administration in the experimental groups: c-IgG (isotype control MAb), c-IgG/Vanc, and MEDI6389/Vanc. (C) Representative in vivo BLI signals on a color scale overlaid on a gray-scale photograph of the mice. (D) Mean total flux (photons/s) ± SEM (logarithmic scale). *, P < 0.05 (2-way analysis of variance [ANOVA]). LOD, limit of detection. (E) Ex vivo CFU (horizontal bars, geometric mean) isolated from skin biopsies performed on euthanized mice on day 14 (logarithmic scale). *, P < 0.05, Kruskal-Wallis test with 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli to correct for multiple comparisons (29).

To determine the perioperative efficacy of adjunctive MEDI6389 plus Vanc, three experimental groups (n = 10 mice/group) were evaluated with preoperative and postoperative dosing (beginning on day 6 to account for the human MAb half-life in mice [see Fig. S1 in the supplemental material]): (i) preoperative c-IgG (isotype control MAb, anti-HIV gp120 [R347] [21]) (15 mg/kg i.p. [intraperitoneal] on day −1) and no postoperative administration on day 6 (none); (ii) c-IgG/Vanc: preoperative (c-IgG plus Vanc [110 mg/kg i.v., human exposure equivalent dose (22)] at −2 h) and postoperative (c-IgG on day 6 plus Vanc [110 mg/kg subcutaneous (s.c.) twice daily on days 6 to 14]); and (iii) MEDI6389/Vanc preoperative (MEDI6389 [5 mg/kg of each MAb i.p. on day −1] plus Vanc [110 mg/kg i.v. at −2 h]) and postoperative (MEDI6389 [once on day 6] plus Vanc [110 mg/kg s.c. twice daily on days 6 to14]) (Fig. 1B). MEDI6389/Vanc had decreased in vivo bioluminescence imaging (BLI) signals throughout the 14-day experiment compared with c-IgG or c-IgG/Vanc (P < 0.05) (Fig. 1C and D). On day 14, a 10-mm punch biopsy was performed, skin tissue was homogenized, and ex vivo CFUs were enumerated, as previously described (23). MEDI6389/Vanc had significantly decreased ex vivo CFUs compared with c-IgG or c-IgG/Vanc (P < 0.05) (Fig. 1E).

Next, perioperative efficacy of adjunctive MEDI6389 plus Vanc was evaluated in an orthopedic SSI model (24–26). Briefly, 8-week-old C57BL/6 anesthetized mice (2% isoflurane) underwent a medial parapatellar arthrotomy, and an orthopedic-grade titanium Kirschner wire (0.6 mm diameter, 9 mm length) was surgically placed into the intramedullary femoral canal with the distal end protruding into the knee joint. An inoculum of MRSA (strain SAP231 derived from parental strain NRS384 [27]; 1 × 10³ CFU in 2 μl PBS) was pipetted onto the distal implant before reducing the patellar complex to midline and closure with absorbable sutures. The same experimental groups as in Fig. 1B were used except that preoperative Vanc was given at −1 h and the experimental duration was 10 days (Fig. 2A). MEDI6389/Vanc had decreased in vivo BLI signals compared with c-IgG or c-IgG/Vanc (P < 0.05) (Fig. 2B and C). On day 10, bone/joint tissue was homogenized, implants were sonicated, and ex vivo CFUs were enumerated, as previously described (26). MEDI6389/Vanc significantly decreased ex vivo CFUs from the bone/joint tissue (P < 0.001) and implants (P < 0.01) compared with c-IgG; there were no significant differences compared with c-IgG/Vanc (Fig. 2D and E). However, MEDI6389/Vanc had a significant decrease in the percentage of mice that had detectable ex vivo CFUs compared with c-IgG/Vanc (P < 0.001) (Fig. 2F).

FIG 2.

Enhanced efficacy of MEDI6389 in an orthopedic SSI model. The mouse model of orthopedic SSI with MRSA strain SAP231 was performed in C57BL/6 mice (n = 10/group). (A) Timeline of preoperative and postoperative administration in the experimental groups: c-IgG (isotype control MAb), c-IgG/Vanc, and MEDI6389/Vanc. (B) Representative in vivo BLI signals on a color scale overlaid on a gray-scale photograph of the mice. (C) Mean maximum flux (photons/s/cm²/steradian) ± SEM (logarithmic scale). LOD, limit of detection. *, P < 0.05 (2-way ANOVA). (D and E) Ex vivo CFU (horizontal bars, geometric mean) isolated from joint/bone tissue (D) and implants (E) from euthanized mice on day 10 (logarithmic scale). †, P < 0.01; ‡, P < 0.001, Kruskal-Wallis test with 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutieli to correct for multiple comparisons (29). n.s., not significant. (F) Percentage of mice with (filled) or without (open) ex vivo CFUs detected from tissue (D) or implant (E) specimens. ‡, P < 0.0001, Fisher’s exact test.

The study had some limitations. First, given the proof-of-concept efficacy for perioperative MEDI6389, the individual MAbs were not evaluated separately, but the anti-AT and anti-ClfA MAb combination had enhanced efficacy compared with the individual MAbs through complementary mechanisms (toxin neutralization, opsonophagocytic killing, and inhibition of biofilm formation and agglutination) (12–14). Second, additional efficacy of SAN481 was not observed in the orthopedic SSI model (see Fig. S2 in the supplemental material), which was likely due to the low sensitivity of mice to leukocidins (relative to humans and rabbits) (15, 16) or to limited leukocidin production or virulence in the SSI models. Importantly, SAN481 was not detrimental and might be beneficial against human SSIs. Third, the timing of preoperative Vanc was different in the skin (−2 h) and orthopedic (−1 h) SSI models because the −1-h-preoperative Vanc alone in the skin SSI model resulted in low initial in vivo BLI signals and CFUs in some mice (see Fig. S3 in the supplemental material), which prohibited studying the additional efficacy of MEDI6389. Because the half-life of Vanc is 0.97 h in mice (28), the beginning Vanc levels in the skin SSI model were likely half of those in the orthopedic SSI model. Finally, two different USA300 MRSA strains were used in the skin and orthopedic SSI models that produced lower (USA300 LAC::lux [20]) and higher (SAP231 [27]) bioluminescent signals, respectively, for optimal in vivo BLI. Nonetheless, MEDI6389 demonstrated enhanced efficacy against both strains. The data in this study suggest that neutralization of S. aureus virulence factors may provide adjunctive perioperative coverage against SSIs.