A minimal-passaged SF8300 clinical strain representative of the epidemic clone USA300-0114 was used to induce lethal septic shock in the mechanically ventilated rabbit model. SF8300 was grown in 10 mL of tryptic soy broth (TSB) media (Sigma-Aldrich) overnight, then diluted 1:50 into 50 mL of TSB media in a 250 mL Erlenmeyer flask. Bacteria were grown at 37°C with shaking at 120 RPM and harvested at an OD_{600 nm} of 0.4. Cells were washed and resuspended in sterile phosphate-buffered saline (PBS) to a concentration of 1 × 10¹⁰ colony-forming units (CFU)/mL, aliquoted into individual cryovials, and immediately stored at -80°C (19). Frozen stocks were titered in triplicate on three separate occasions before use in any experiment. Based on this value, the inoculum containing 9.4 x 10⁸ CFUs (in 2.2 mL saline solution) was prepared and again tittered in triplicate to document the numbers of bacterial CFUs used for intravenous challenge of rabbits.

Anti-Hla mAb (37), anti-Luk mAb that cross-neutralizes LukF-PV (PVL), LukD (leukocidin ED), HlgB (γ-hemolysin) (16), and anti-ClfA mAb (38) were generated as described previously. An irrelevant isotype-matched control IgG (c-IgG) directed against the gp120 protein of HIV was used as control (37).

The rabbit model was reviewed and approved by the University of California San Francisco Institutional Animal Care and Use Committee. Experiments were conducted in a facility certified by the Association for Assessment and Accreditation of Laboratory Animal Care International. New Zealand White rabbits (Western Oregon Rabbit Co.), 3.4 - 4.0 kg, 16 - 20 weeks of age, both male or female rabbits, were sedated by administration of buprenorphine (0.01-0.05 mg/kg) subcutaneously, and then, after 30 minutes, injected intramuscularly with a solution containing 36 mg/kg ketamine and 5.2 mg/kg xylazine. Ophthalmic ointment was placed on each eye. Once the desired depth of anesthesia was achieved as assessed by lack of pedal reflex, a 3.0 mm cuffed endotracheal tube was introduced orally into the trachea. The endotracheal tube was connected to a neonatal ventilatory circuit (Fisher & Paykel), connected to a heated humidifier (Fisher & Paykel) and then a ventilator (ADS2000, Engler) with the following settings: peak inspiratory pressure (PIP) of 12-15 cm H₂O, positive end-expiration pressure (PEEP) of 5 cmH₂O, respiratory rate of 30 breaths/min, flow rate of 4 L/min, fraction of inspired oxygen (FiO₂) of 0.35, and 2.0-2.5% isoflurane to maintain general anesthesia. The ventilator settings remain constant for the duration of the study, except when FiO₂ is increased to 1.0 when performing suctioning for removal of mucus that built up at the distal end of the endotracheal tube. The respiratory rate may be adjusted to maintain PaO₂ between 140 and 200 mmHg and PaCO₂ between 35 and 45 mmHg. The left or right marginal ear vein was catheterized with a 22G x 1” IV catheter for continuous infusion of Normosol-R with 5% dextrose (4 mL/kg/h) using an infusion pump (Hospira) for fluid maintenance.

The right carotid artery was catheterized with either 4 French pediatric PiCCO catheter (Getinge) for the natural history study or 18G arterial catheter (Argon Medical Devices) for the efficacy study and positioned approximate 2-3 cm from the common aortic arch, whereas the right internal jugular vein was catheterized with a 4 French dual-lumen pediatric central venous catheter (Medline) and positioned 1-2 cm from the right atrium using a common cutdown approach as detailed here. The surgical site was shaved, scrubbed, and prepped aseptically. The surgical site was draped out with sterile towels. The rabbit was placed in a supine position and depth of anesthesia was assessed by pinching between the toes. If rabbit was responsive, isoflurane was increased to effect. A 2-3 cm incision was made to access the right carotid artery/internal jugular vein using aseptic technique. Blunt and sharp dissection was used to identify and isolate the carotid artery from the Vagus nerve as well as the internal jugular vein. A distal 3-0 silk ligature was placed around the carotid artery/jugular vein to prevent back bleeding. A loose proximal ligature was placed around the carotid artery/jugular vein. A small (approximately 1 mm) cut was made at an angle across the vessel and the catheter was passed through the cut proximally and advanced about 5 cm so that the catheter tip was positioned approximately 2-3 cm from the common aortic arch or 1-2 cm from the right atrium. The proximal ligature was tied to secure the catheter in place. The vessels were repositioned and the wound was closed in at least 2 layers. The catheters were connected to pressure transducers with integrated flush device (Getinge) and the patient monitor (MP90, Philips Healthcare) for continuous blood pressure monitoring. To maintain patency, the catheters were continuously flushed with 3 ml/h normal saline containing 1U/ml heparin, and manually flushed every 1 h with three quick flushes of normal saline containing 1U/ml heparin.

Once the catheters were in place, the transpulmonary thermodilution method was used to measure cardiac output every 2 h by injecting a bolus of 2-mL cold normal saline through the central venous catheter. A thermistor located at the tip of the PiCCO catheter recorded the very brief drop in the blood temperature for calculation of a thermodilution curve and cardiac output (CO) on the patient monitor (MP90, Philips Healthcare). For each timepoint, CO was measured two times about 3-4 min apart, and then averaged. If the two CO measurements differed >10% from one another, a third CO measurement was performed, and the three measurements averaged. At 6 h after surgery, rabbits were challenged intravenously via the IV catheter placed in the marginal ear vein with a 2.2 ml saline solution containing 9.4 x 10⁸ CFUs of USA300.

Two-dimensional echocardiography was used to assess global left ventricular hypokinesia in the natural history study rabbits. Images were obtained with a cardiac ultrasound machine (iE33, Philips Healthcare) equipped with S12-4 sector array pediatric transducer. We measured the left ventricular and auricular dimensions as well as left ventricular ejection fraction (LVEF) using the parasternal long axis view of the heart via Teicholz method.

Arterial blood samples were collected at pre-infection baseline (within 1 h of intravenous challenge with USA300), and then every 2 h for the first 24 h post infection (hpi) and every 4 h until the end of the study. Blood gas analysis was determined using RapidPoint 500 (Siemens), including partial pressure of oxygen (PaO₂) and of carbon dioxide (PaCO₂), bicarbonate [HCO₃ ^-]), base excess, sodium (Na⁺), potassium (K⁺), chloride (Cl^-), calcium (Ca⁺⁺), glucose, and lactate. The Element HT5 Veterinary Hematology Analyzer^© (Heska, Loveland, CO) was used to determine complete blood count with five-part differential. VetScan VS2^© (Abaxis, Union City, CA) was used for analysis of alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatinine and blood urea nitrogen.

Heart, lung, liver, kidney, and spleen were weighed and processed for bacterial count. For each organ, 3 to 4 pieces (0.05-0.10 grams each) were removed from different parts of the organ, totaling 0.1-0.2 grams, and then homogenized for 30-90 seconds using a Tissue-Tearor (BioSpec, Bartlesville, OK). Each homogenate (100 µL) was then serially diluted in 900 µL of normal saline and 100 µL of each dilution plated onto 5% blood sheep agar and incubated for 16-24 h at 37°C for CFU count.

Our hypothesis is that survival of rabbits pretreated with c-IgG is shorter than survival of those pretreated with anti-Hla/Luk/ClfA mAb combination in a rabbit model of USA300-induced septic shock. We calculated that a sample size of 9 animals per experimental group would provide a power of 80% to detect a hazard ratio of 0.25, with a two-sided type I error 0.05 by means of a log-rank test (Schoenfeld method) using STATA version 17.0 (StataCorp). The sample size was increased to 10 animals per experimental group to account for potential exclusion of animals in this model for technical problems related to intubation, surgery for placement of catheter in the carotid artery, or occlusion of endobronchial tube from mucus buildup during mechanical ventilation (35). Survival curves were generated using the Kaplan-Meier method, and significance was assessed by means of the log rank (Mantel-Cox) test. Nonparametric statistics, median [25^th percentile – 75^th percentile], were computed for all variables. Bacterial count and lung weight/body weight (LW/BW) ratio were compared using a nonparametric two-sided Mann-Whitney U test. Two-sided Fisher’s exact test was used to determine whether rabbits pretreated with c-IgG or anti-Hla/Luk/ClfA mAb combination differ in the proportions of categories for the various blood tests (e.g., hypocapnia). Nonparametric one-way analysis of variance (ANOVA) with Kruskal-Wallis test followed by Dunn’s multiple comparison test were used to evaluate effect of pretreatment with c-IgG or anti-Hla/Luk/ClfA mAb combination on blood tests conducted at pre-infection baseline or terminal endpoint.

We performed a natural history study to characterize the pathogenesis of USA300-induced septic shock in anesthetized rabbits that were intubated for mechanical ventilation with a lung-protective low-tidal volume ( Figure 1A ), as previously described (35). Anesthetized rabbits were instrumented with the PiCCO catheter and central venous catheter for advanced hemodynamic monitoring, including intermittent measurement of cardiac output by transpulmonary thermodilution method ( Figures 1A, B ). Rabbits were infused with a total of 60-65 mL/kg of a balanced crystalloid solution over the course of 6 h preceding S. aureus challenge for (i) fluid maintenance at 4 mL/kg/h, (ii) continuous flushing the arterial and central venous catheters at 3 mL/h each ( Figure 1A ), and (iii) fluid resuscitation at 30 mL/kg to counteract the hypotension associated with overstimulation of Vagus nerve during carotid artery catheterization (35), thereby providing adequate volume repletion that may be crucial for the development of a hyperdynamic state of septic shock after S. aureus challenge.

Mortality rate was 100% (6/6) for USA300-challenged rabbits, with acute circulatory failure and death occurring between 15 and 27 h post infection ( Figures 2A–F ), whereas the two saline-challenged rabbits survived to the end of the study period at 36 h post challenge ( Figures 2G, H ). The hyperdynamic state of USA300-induced septic shock, which was characterized by elevated cardiac output, increased stroke volume, and decreased systemic vascular resistance, occurred early at 2-6 h post infection in five of the rabbits ( Figures 2A, B, D–F ) or later at 8-18 h post infection in one rabbit ( Figure 2C ), thereby reproducing the clinical hyperdynamic state with increased heart rate, elevated cardiac output, increased stroke volume, and decreased systemic vascular resistance in adequately volume-resuscitated patients with septic shock (26–29). Then the ensuing hypodynamic state, which was characterized by decreased heart rate, decreased MAP, increased central venous pressure, decreased cardiac output, decreased stroke volume, and elevated systemic vascular resistance, occurred in the 4-6 h period before the rabbits succumbed to USA300-induced septic shock ( Figures 2A–F ), mimicking the terminal phase in non-survivor patients with hypodynamic shock and death (39). MAP, central venous pressure, cardiac output, stroke volume and systemic vascular resistance remained within normal limits for the two saline-challenged rabbits, consistent with our previous findings that our lung-protective low-tidal volume mechanical ventilation protocol and instrumentation did not negatively impact rabbits without infectious challenge (35).

Transthoracic echocardiography was used in the rabbit natural history study to characterize further potential septic shock-induced acute myocardial dysfunction, including global left ventricular hypokinesia, which occurred in 60% of mechanically ventilated patients with septic shock (40). Representative M-mode echocardiographic tracings for Rabbit #1 (see Figure 2A for corresponding changes in hemodynamic parameters) obtained at pre-infection baseline ( Figure 3A ), at the early hypodynamic state when MAP decreased 20% from pre-infection baseline value ( Figure 3B ), and at the late hypodynamic state when MAP decreased >30% from pre-infection baseline value ( Figure 3C ) illustrated the development of progressive global left ventricular hypokinesia. For all six rabbits with USA300-induced septic shock, left ventricular ejection fraction (LVEF) was 70% (range from 64% to 75%) at pre-infection baseline, and decreased progressively to 47% (28% to 61%) during the early hypodynamic state, and then to 35% (25% - ;46%) during late hypodynamic state (test for linear trend P<0.001; Figure 3D ).

Rabbits were randomized for intravenous injection with either 30 mg/kg each of the anti-Hla/Luk/ClfA mAb combination or irrelevant isotype-matched c-IgG. After 16-20 h, rabbits were challenged intravenously with the community-associated MRSA USA300, which is known to produce the virulence determinants targeted by our mAb combination: α-hemolysin is neutralized by anti-Hla mAb; F components of the bicomponent toxin LukF-PV (PVL), LukD (leukocidin ED), HlgB (γ-hemolysin) are cross-neutralized by anti-Luk mAb (41); and clumping factor A is neutralized by anti-ClfA mAb. Overall survival rates were 11% (1/9) for rabbits pretreated with c-IgG compared to 80% (8/10) for those pretreated with anti-Hla/Luk/ClfA mAb combination (P<0.001 by log-rank test; Figure 4A ).

S. aureus quantified in heart and kidney samples showed significant log₁₀CFU reduction (P<0.001 and P=0.043 by nonparametric Mann-Whitney test), whereas those in lung, spleen, kidney, and liver samples showed non-significant reduction in rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to those pretreated with c-IgG ( Figure 4B ).

Although cardiac output and central venous pressure were measured in the natural history study, these advanced hemodynamic parameters were not measured for the efficacy study of the anti-Hla/Luk/ClfA mAb combination because they are labor intensive, and the more expensive thermodilution and central venous catheters that we used are more prone to thrombotic occlusion due to their narrower tip design, thereby requiring removal and exchange with new catheters in 20-30% of rabbits. Nonetheless, for the efficacy study, we catheterized the right carotid artery for continuous blood pressure monitoring with a similarly sized arterial catheter in which the narrow tip was cut off with a sharp surgical blade, creating a 90° blunt tip which helped reduce the frequency of thrombotic catheter occlusion to <5% of rabbits. Mean arterial pressure (MAP) was recorded every 15 min, providing a high-resolution view of the individual rabbit’s hemodynamics ( Figures 5A, B ). MAP declined sharply in all (9/9) c-IgG-pretreated rabbits in the 2-4 h preceding death ( Figure 5A ), corresponding to the hypodynamic state of septic shock described above ( Figures 2 , 3 ), whereas MAP remained within normal limits for 80% (8/10) of anti-Hla/Luk/ClfA mAb combination-pretreated rabbits that survived to the end of the study at 36 h post infection and declined sharply in only the remaining 20% (2/10) of rabbits that died ( Figure 5B ). The one survivor pretreated with c-IgG developed severe hypotension at 25 h post infection when MAP decreased to 14 mmHg, but then increased rapidly and remained from 25 to 32 mmHg until the animal was euthanized at 36 h post infection, the end of the study period ( Figure 5A ). USA300-induced acute circulatory failure, defined as >70% decreased in MAP from pre-infection baseline, occurred in 20% (2/10) of rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to 100% (9/9) rabbits pretreated with c-IgG compared to (two-sided Fisher’s exact test, P < 0.001; see also Figure 5C ).

Despite USA300 infecting the lungs of all rabbits ( Figure 4B ), the lung weight to body weight (LW/BW) ratio in g/kg, a quantitative measure of acute lung injury, was 4.2 [3.8 – 4.9] for rabbits pretreated with anti-Hla/Luk/ClfA mAb combination and 3.9 [3.7 - 4.5] for those pretreated with c-IgG (Mann Whitney test, P=0.50; Figure 6A ), which are similar to the LW/BW ratio of 3.9 for uninfected control rabbits that were managed using the same aforementioned lung-protective ventilatory strategies (35). Accordingly, hypoxemic respiratory failure was not a consistent feature of the rabbit model because the ratio of partial pressure arterial oxygen and fraction of inspired oxygen (PaO₂/FiO₂) decreased to 100 mmHg - 300 mmHg, thresholds for mild to moderate in ARDS severity according to the Berlin definition (42), in only 20% (2/10) and 11% (1/9) of rabbits pretreated with anti-Hla/Luk/ClfA mAb combination or c-IgG, respectively (two-sided Fisher’s exact test, P = 1.00; Figure 6B ).

To determine whether temporal changes in circulating white blood cell populations are associated with survival of rabbits pretreated with anti-Hla/Luk/ClfA mAb combination or c-IgG, serial blood samples were collected from each rabbit at pre-infection baseline, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 28, 32, 36 h post infection or within 1 h of the time when rabbit succumbed to infection for complete blood count with differential. Peripheral white blood cells, including neutrophils and monocytes, declined very rapidly within the first 6 h post infection for all rabbits, but then increased among the 80% (8/10) and 11% (1/9) of rabbits pretreated with anti-Hla/Luk/ClfA mAb combination or c-IgG, respectively, that survived to the end of the study at 36 h post infection (two-sided Fisher’s exact test, P = 0.006; Figures 7A–F ). Among the 8 non-survivors pretreated with c-IgG, those with the steepest decline in white blood cells, neutrophils and monocytes also developed earliest acute circulatory failure and death ( Figures 7A, C, E ). Lymphocyte counts also declined, but less rapidly compared to the decline of neutrophils and monocytes, in all rabbits from both experimental groups ( Figures 7G, H ). Compared to rabbits pretreated with c-IgG, those pretreated with anti-Hla/Luk/ClfA mAb combination had significantly greater numbers of white blood cells (4.9 [3.7 - 6.4] vs. 1.4 [0.7 - 2.9] * 10³/µL, P<0.001), neutrophils (2.5 [1.7 - 3.4] vs. 0.7 [0.2 - 1.5] * 10³/µL, P<0.001), monocytes (0.49 [0.29 - 0.56] vs. 0.05 [0.01 - 0.14] * 10³/µL, P=0.002), and lymphocytes (1.2 [0.9 - 1.4] vs. 0.4 [0.3 - 0.8] * 10³/µL, P=0.002) at the terminal endpoint ( Figures 7A–H ).

Lactate at the terminal endpoint was 1.3 [0.8 - 6.4] mmol/L for rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to 13.9 [9.0 - 15.7] mmol/L for those pretreated with c-IgG (multiplicity-adjusted P<0.01; Figure 8D ). Lactate increased in all (10/10) non-survivors but remained similar to pre-infection baseline for 9/9 survivors from both experimental groups ( Figure 8D ), consistent with hyperlactatemia being one of the strongest prognostic biomarker of survival in patients with septic shock (43, 44).

Despite no differences at pre-infection baseline, PaCO₂ at the terminal endpoint was 35 [29 - 44] mmHg for rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to 19 [11 - 27] mmHg for those pretreated with c-IgG (multiplicity-adjusted P<0.05; Figure 8A ). Severe hypocapnia, defined as PaCO₂ < 25 mmHg, occurred in 10% (1/10) of rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to 78% (7/9) of those pretreated with c-IgG (two-sided Fisher’s exact test, P = 0.005). This is consistent with findings in patients with septic shock in which severe hypocapnia was associated with death (45, 46).

pH showed broad changes, ranging from marked acidosis to alkalosis, in rabbits from both experimental groups, and did not correlate with survival or death ( Figure 8B ). In contrast, bicarbonate decreased to the lowest level in all (10/10) non-survivors compared 0/9 survivors, although differences between the two experimental groups at the terminal endpoint were not statistically significant ( Figure 8C ).

Glucose also displayed a broad range at the terminal endpoint for rabbits in both experimental groups, with hypoglycemia (<100 mg/dL) occurring in 37% (7/19) and hyperglycemia (>200 mg/dL) in 16% (3/19) of rabbits ( Figure 8E ). Potassium levels at the terminal endpoint was 3.0 [2.7 - 4.8] mmol/L for rabbits pretreated with anti-Hla/Luk/ClfA mAb combination compared to 7.4 [4.5 - 8.3] mmol/L for those pretreated with c-IgG (multiplicity-adjusted P<0.01; Figure 8F ). Extreme hyperkalemia (>6.5 mEq/L) occurred in 70% (7/10) of rabbits from both experimental groups that succumbed to USA300-induced septic shock and none (0/9) of the survivors ( Figure 8F ).

Acute myocardial injury biomarkers, including cardiac troponin I, CK-MB, and myoglobin, increased significantly from pre-infection baseline to the terminal endpoint for rabbits pretreated with c-IgG ( Figures 9A–C ). In contrast, cardiac troponin I, but not CK-MB or myoglobin, increased significantly from pre-infection baseline to the terminal endpoint for rabbits pretreated with anti-Hla/Luk/ClfA mAb combination ( Figures 9A–C ). When comparing rabbits pretreated with c-IgG or anti-Hla/Luk/ClfA mAb combination at the terminal endpoint, significant differences were noted with cardiac troponin I (multiplicity-adjusted P<0.01), but not CK-MB or myoglobin ( Figures 9A–C ). The aspartate aminotransferase to alanine aminotransferase (AST/ALT) ratio increased to greater than 5 in 44% (4/9) rabbits pretreated with c-IgG compared to 0% (0/10) for those pretreated with anti-Hla/Luk/ClfA mAb combination ( Figure 9D ). In the four c-IgG-pretreated rabbits with AST/ALT >5, ALT was only mildly elevated (suggestive of minimal liver necrosis) whereas AST was markedly elevated (suggestive of myocardial necrosis) (47), which is consistent also with their markedly elevated levels of cardiac troponin I, CK-MB, and myoglobin ( Figures 9A–C ). Creatinine and blood urea nitrogen levels were mildly elevated in only 22% (2/9) and 10% (1/10) of rabbits pretreated with c-IgG and anti-Hla/Luk/ClfA mAb combination (two-sided Fisher’s exact test, P = 0.58; Figures 9E–F ), suggesting minimal impact of USA300-induced septic shock on kidney function.