Development of Novel Immunoprophylactic Agents against Multidrug-Resistant Gram-Negative Bacterial Infections

ABSTRACT

The widespread emergence of antibiotic resistance, including multidrug resistance in Gram-negative (G–) bacterial pathogens, poses a critical challenge to the current antimicrobial armamentarium. Antibody-drug conjugates (ADCs), primarily used in anticancer therapy, offer a promising treatment alternative due to their ability to deliver a therapeutic molecule while simultaneously activating the host immune response. The Cloudbreak platform is being used to develop ADCs to treat infectious diseases, composed of a therapeutic targeting moiety (TM) attached via a noncleavable linker to an effector moiety (EM) to treat infectious diseases. In this proof-of-concept study, 21 novel dimeric peptidic molecules (TMs) were evaluated for activity against a screening panel of G– pathogens. The activities of the TMs were not impacted by existing drug resistance. Potent TMs were conjugated to the Fc fragment of human IgG1 (EM), resulting in 4 novel ADCs. These ADCs were evaluated for immunoprophylactic efficacy in a neutropenic mouse model of deep thigh infection. In colistin-sensitive infections, 3 of the 4 ADCs offered protection similar to that of therapeutically dosed colistin, while CTC-171 offered enhanced protection. The efficacy of these ADCs was unchanged in colistin-resistant infections. Together, these results indicate that the ADCs used here are capable of potent binding to G– pathogens regardless of lipopolysaccharide (LPS) modifications that otherwise lead to antibiotic resistance and support further exploration of ADCs in the treatment of infections caused by drug-resistant G– bacteria.

INTRODUCTION

Antimicrobial resistance presents a great challenge in the care of hospitalized and critically ill patients (1). The treatment landscape has become considerably more challenging for Gram-negative (G–) infections, primarily due to significant increases in the frequency of pathogens resistant to existing antimicrobial agents. Multidrug resistance (MDR) (resistance to three or more antimicrobial classes) is prevalent among bacterial pathogens, including Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Escherichia coli (KAPE). Data demonstrate that in 2018, 9% of P. aeruginosa and E. coli, 12% of K. pneumoniae, and 39% of A. baumannii hospital-onset infections were MDR (2). This antibiotic resistance results in a significant reduction in antibiotic options for clinicians (3, 4). Many MDR G– pathogens produce beta-lactamases, including extended spectrum beta-lactamases (ESBL) and carbapenemases, which render them resistant to extended-spectrum cephalosporins and carbapenems (5). In the United States, the rates of carbapenem resistance among A. baumannii clinical strains range from 33% to 58% (6). Alarmingly, a new resistance mechanism, MCR-1, was found globally, which confers resistance against colistin (COL), the last-line agent against MDR P. aeruginosa and A. baumannii (7). Moreover, the mcr-1 gene is carried on a plasmid and can be transferred to other types of bacteria, spreading its resistance. Thus, there is an urgent need for novel agents that leverage immunological mechanisms to prevent the infection, as opposed to relying solely on a drug’s antibacterial action.

Antibody-drug conjugates (ADCs) are a class of therapeutic agents mainly applied to cancer therapy (8). ADCs employ a cytotoxic small-molecule “payload” attached via a cleavable or a noncleavable linker to an antibody with high affinity to a surface-exposed antigen on the host cell. To date, 10 ADCs have been approved by the FDA, and more than 150 are in preclinical and clinical development (9). Despite recent advances in cancer immunotherapy (10), the use of ADCs to treat infectious diseases remains largely unexplored.

The innovative Cloudbreak platform developed by Cidara Therapeutics, Inc. (San Diego, CA), is being developed to meet this need. Cloudbreak differs from current ADC design by employing an antimicrobial therapeutic targeting moiety (TM) attached via a noncleavable linker to an effector moiety (EM) that engages the host immune system. Similar to anticancer bispecific agents (11, 12), the Fc-conjugates in this work bind a conserved target broadly found on G– pathogens through the TM while simultaneously engaging multiple arms of the immune system via the EM, specifically, the Fc domain of human IgG1 (hIgG1).

The TMs employed in these Cloudbreak ADCs are novel peptidic molecules related to the polymyxins that tightly bind to the lipid A segment of lipopolysaccharide (LPS), causing disruption of the outer membrane (OM) of susceptible G– bacteria (13). Pathogens can alter the makeup of lipid A through their ability to esterify one of the 1′- or 4′-phosphate groups with ethanolamine (14), leading to reduced affinity of polymyxins to LPS. We hypothesized that two polymyxin molecules linked together could bind to LPS in a cooperative manner (15–17). Therefore, in situations where monomeric polymyxin binding is reduced by an LPS mutation, a larger dimeric molecule could enjoy additional hydrophobic and/or phosphate interactions on an adjacent molecule to restore potent binding (Fig. 1).

FIG 1.

Hypothetical binding of polymyxins (PMX) to wild type (wt) or resistant (mcr-1) lipid A. In the mcr-1 mutation, one of the phosphate groups (P) is modified to give a phosphoethanolamine (PE), reducing the number of binding sites for monomeric PMX. However, due to entropic effects, dimeric PMX benefits from additional interactions through an adjacent lipid A molecule.

Fusion of TMs to an Fc further enhances killing by complement-dependent cytotoxicity (CDC), antibody (Ab)-dependent cell-mediated cytotoxicity (ADCC), and Ab-dependent cell phagocytosis (ADCP) (Cidara Therapeutics, unpublished.) In addition, Fc-mediated recycling through the neonatal receptor (FcRn) on host cells improves circulating conjugate exposure and facilitates efficient tissue distribution, conferring pharmacokinetic advantages to the conjugates over the TM alone. An ancillary benefit of these conjugates is that their size limits kidney exposure, where cationic peptides, as a class, have the potential to cause nephrotoxicity.

RESULTS

Synthesis of TMs.

Novel polymyxin dimers were prepared from polymyxin B or colistin (polymyxin E) by enzymatic hydrolysis of the Dab-Thr-Dab-octanoyl tripeptide (18), and reacylation with a variety of orthogonally protected tripeptides gave polymyxin analogs terminating in an amino group upon deprotection. Acylation via the free amine with a bivalent central linker (depicted in blue in Fig. 1) gave a series of novel polymyxin dimers. Incorporation of a terminal alkyne moiety into the central linker allowed for the subsequent coupling of the TM to an azido-modified Fc protein using a copper(I)-catalyzed click reaction (19).

In vitro susceptibility screening of TMs.

A total of 21 novel TMs were evaluated for in vitro activity against a screening panel of KAPE pathogens. This screening panel was composed of 12 clinical isolates per species with various susceptibilities to colistin and carbapenems (see Table S1 in the supplemental material). Four TMs were selected for conjugation based on broad KAPE activity (Table 1) and having sufficient material after the initial synthesis. Other broadly active TMs that were not conjugated included H-0013 and H-0017 (Table S1).

TABLE 1.

TMs selected for conjugation inhibit growth across KAPE G– strains^a

TM	Geometric mean MIC (μg/ml) of select TMs of:				Corresponding ADC
	K. pneumoniae (μg/ml) (range)	A. baumannii (μg/ml) (range)	P. aeruginosa (μg/ml) (range)	E. coli (μg/ml) (range)
Colistin	0.25 (0.125–64)	8 (0.3–128)	0.25 (0.125–2)	0.25 (0.125–0.5)
H-0009	1.70 (1–4)	0.67 (0.5–2)	1.38 (1–4)	1.17 (1–2)	CTC-172
H-0010	1.45 (0.5–4)	0.59 (0.5–2)	1.05 (0.5–2)	1.05 (1–2)	CTC-173
H-0027	0.72 (0.3–2)	0.52 (0.3–2)	1.12 (0.5–4)	0.58 (0.5–1)	CTC-174
H-0028	2.00 (0.5–8)	1.12 (0.5–4)	2.00 (1–8)	0.85 (0.5–2)	CTC-171

^aTwelve clinical strains of K. pneumoniae, A. baumannii, P. aeruginosa, and E. coli were evaluated for MICs using the CLSI broth microdilution method.

The A. baumannii strains evaluated were of interest due to the high degree of variation in susceptibility to colistin, ranging from highly susceptible (0.5 μg/ml) to highly resistant (128 μg/ml) (Table 2). The MICs of the novel TMs did not correlate with colistin susceptibility, displaying no more than 2-fold differences observed between the MIC in AB 383 and AB 457, despite having over 100-fold differences in colistin MICs (Table 2). This indicates that the TM design which aimed to optimize binding to LPS in both COL-sensitive (COL-S) and COL-resistance (COL-R) strains was successful and was not significantly affected by LPS modifications.

TABLE 2.

TMs inhibit growth of COL-S and COL-R strains of A. baumannii^a

Strain	A. baumannii MICs (μg/ml) for:
	Colistin	H-0009	H-0010	H-0027	H-0028
AB 383 (45321)	128	1	0.5	0.5	2
AB 317 (45380)	32	1	0.5	0.5	4
AB 328 (45388)	16	0.5	0.5	0.5	2
AB 368 (45429)	8	0.5	0.5	0.3	0.5
AB 482 (45358)	1	0.5	0.5	0.3	0.5
AB 457 (45328)	0.5	0.5	0.5	0.5	1

^aMICs to select TMs displayed for 6 A. baumannii strains with varied COL susceptibility.

Preparation and characterization of novel ADCs.

Selected TMs from those described above were conjugated to an azide-derivatized Fc protein utilizing a copper(I)-catalyzed azide-alkyne Huisgen cycloaddition (click reaction). The coupling was generally high yielding and required a 1-fold excess of the TM coupling partner for quantitative derivatization.

Lysine residues on hIgG1 Fc protein were acylated using activated azido-PEG4-NHS ester and purified by protein A affinity chromatography and buffer exchange dialysis. The acylation reaction provided the azido-PEG4-Fc as a defined mixture of species with an average DAR (drug-antibody ratio) of approximately 6.0, ranging from DAR 2 to DAR 9 in a statistical manner, where the DAR 6 species predominated.

TMs terminating in an alkynyl group were coupled to the azido-PEG4-Fc in phosphate-buffered saline (PBS) by treatment with a Cu(I) catalyst to join the TM and EM via a 1,2,3-triazole linkage (Fig. 2). Affinity purification by protein A chromatography followed by size exclusion chromatography (SEC) gave the final TM-Fc conjugates after dialysis buffer exchange into PBS. The final compounds were characterized by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF) to determine the average DAR of the product, which ranged from 6.2 to 6.5.

FIG 2.

Preparation of TM-Fc conjugates (ADCs). hIgG1 Fc was acylated via lysine residues with a PEG4 linker terminating in an azido group. Cu(I)-catalyzed cycloaddition with an alkyne-derivatized TM (click reaction) joins the Fc and TM via a 1,2,3-triazole linkage to provide the final conjugate.

Pharmacokinetic (PK) evaluation of the novel ADCs.

Following purification of the conjugates, the pharmacokinetics of the resultant ADCs (CTC-171, CTC-172, CDC-173, and CTC-174) were evaluated. Naive immunosuppressed mice were administered a single dose of 20 mg/kg of the indicated ADC via intraperitoneal (IP) injection. The maximum concentration of drug in serum (C_max) of the ADCs was observed 2 h following injection (Fig. 3) and was quantified to be 5.6 to 13.3 μg/ml (Table 3). The elimination half-life of the ADCs ranged from 9.9 to 22.1 h (Table 3), generally longer than that of colistin, which has been reported as anywhere from 1.2 to 14.4 h depending on the model, dose, and administration route (20–23). The area under the concentration-time curve from 0 to 24 h (AUC_0–24) varied from 49.2 to 141.0 μg · h/ml between compounds (Table 3). The amount of TM-EM that remained conjugated was evaluated by measuring both the TM and EM plasma concentrations individually, with the percentage of the TM/EM ratio indicating the percentage of intact ADC relative to total EM. From 2 h to 24 h postdose the compounds were found to degrade with 3.7 to 18.2% of intact ADC remaining compared to the 18.2 to 50.6% of intact ADC observed at 2 h following injection (Table 3).

FIG 3.

PK parameters for novel ADCs in mice. At the selected intervals following IP administration with the indicated ADC at 20 mg/kg, whole blood was obtained from naive neutropenic mice. LPS ELISA was used to quantify the concentration of conjugated ADC. The concentration of ADC is displayed as the plasma concentration (μg/ml) over time.

TABLE 3.

ADC pharmacokinetics demonstrate delayed clearance and degradation of conjugates^a

Compound	PK summary novel ADCs
	AB 482 MIC (μg/ml)	AB 317 MIC (μg/ml)	Cmax (SD)	T1/2 (h) (SD)	AUC0–24 (μg · h/ml) (SD)	2-h TM/EM (%) (SD)	24-h TM/EM (%) (SD)	24-h AUC/MIC AB 482	24-h AUC/MIC AB 317
CTC-171	8	8	5.8 (2.1)	9.9	97.0 (23.4)	22.8 (15.0)	3.7 (1.3)	12.1	12.1
CTC-172	16	16	5.6 (1.4)	21.5	49.2 (4.2)	18.2 (0.9)	5.4 (1.0)	3.1	3.1
CTC-173	16	8	7.3 (0.9)	22.1	90.6 (9.6)	50.6 (9.4)	18.2 (3.3)	5.7	11.3
CTC-174	16	16	13.3 (3.2)	21.7	141.0 (9.7)	39.4 (9.5)	17.0 (1.6)	8.8	8.8

^aNaive neutropenic mice were administered the 20 mg/kg of the indicated ADC. Whole blood was isolated and LPS ELISA was used to calculate the C_max, T1/2, and AUC_0–24. The percentage of TM-EM conjugate relative to total EM calculated at 2 and 24 h postinjection indicates the percentage of intact ADC at the indicated time.

Novel ADCs act as an effective immunoprophylactic.

Given the prolonged circulation of the ADCs due to Fc-mediated recycling, we aimed to evaluate the efficacy of these ADCs in an immunoprophylaxis model. The neutropenic murine thigh infection model (24) has been well established and addresses the concern whether the immunomodulation effect of novel compounds can be achieved in an immunocompromised host.

A. baumannii strains with high and low COL MICs were evaluated in vivo for growth and sensitivity to COL in a neutropenic thigh infection model in CD-1 mice. Mice were treated with 2.5 mg/kg of COL or 200 μl of PBS twice daily. After 48 h, bacterial burdens were enumerated. AB 317 (COL-R) and AB 482 (COL-S) were found to grow approximately 1 log without causing significant mortality while maintaining the observed in vitro colistin response phenotype (not shown). Therefore, AB 317 and AB 482 were selected as representative strains for the assessment of ADC efficacy in vivo.

To evaluate the immunoprophylactic efficacy of the ADCs, the indicated mice were administered 20 mg/kg of ADC 12 h prior to infection and again 1 h following infection with COL-S AB 482. Colistin-treated mice were used as a comparator control for restriction of bacterial burdens. In all the infections, the novel ADCs were observed to offer some protection compared to the vehicle-treated control mice (Fig. 4A and B; Fig. S1). To directly compare the efficacy between the ADCs, the infection data were normalized to the vehicle control for each infection (Fig. 4C). CTC-172, CTC-173, and CTC-174 partially reduced bacterial burdens and were found to perform similarly to mice therapeutically dosed with colistin. Impressively, CTC-171 immunoprophylaxis reduced bacterial burdens of AB 482-infected mice more than treatment with colistin (mean difference from vehicle control, 90.17% for CTC-171 versus 66.42% for colistin). The efficacy of the ADCs was also evaluated in a COL-R infection. Similar to their performance in COL-S AB 482 infections, in COL-R AB 317 infections, immunoprophylactic treatment with CTC-171, CTC-172, and CTC-173 allowed for the partial reduction of bacterial burdens at 48 h, whereas colistin was ineffective (Fig. 4D). Again, immunoprophylaxis with CTC-171 led to enhanced reduction of bacterial burdens compared to the other ADCs (Fig. 4D). Importantly, the novel ADCs offered similar protection in COL-S AB 482 and COL-R AB 317 infections (Fig. S2), indicating that like the individual TMs, the performance of the TM-EM conjugates is similar in colistin-susceptible and colistin-resistant infections, and they display comparable MICs to the TMs on a molar basis (Table S2).

FIG 4.

ADCs show immunoprophylactic efficacy in COL-S and COL-R models of infection. Independent experiments examining the efficacy of (A) CTC-171 and (B) CTC-173, where “control 1h” is the bacterial burden in the thigh at 1 h postinfection and burdens of all other groups are at 48 h postinfection. All burdens are shown as CFU for each thigh. (C) Data from separate infections evaluating efficacy of CTC-171-CTC-174 in COL-S AB 482 normalized as a percentage of vehicle control for each independent experiment. (D) Data from separate infections evaluating efficacy of CTC-171, CTC-172, and CTC-173 in COL-R AB 317 normalized as a percentage of the vehicle control for each independent experiment. (E) Weight of each mouse as a percentage of day of infection demonstrates no significant weight loss in ADC-treated groups relative to vehicle or colistin controls. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. n = 3 to 5 for each experiment.

Throughout the course of the infections, all infected mice were evaluated twice daily for signs of significant morbidity. As part of this evaluation, weight loss was monitored as an indication of acute toxicity. The weights of mice treated with CTC-171, CTC-173, and CTC-174 did not differ significantly from either the vehicle or colistin-treated mice, whereas CTC-172 treated mice lost less weight compared to the vehicle or colistin-treated mice (Fig. 4E). This indicated that the ADCs were well tolerated and did not induce adverse clinical observations over the course of treatment that could be a sign of acute toxicity in mice.

Efficacy of ADCs associated with drug exposure.

Finally, we sought to determine whether the observed efficacy of the ADCs was associated with the drug exposure. As a measure of efficacy, we calculated the average percentage of the burden reduction from both COL-S and COL-R infections since differences were not observed in the efficacy of the ADCs between infections (Fig. S2). The AUC_0–24 of each compound, as a measure of drug exposure, was divided by the in vitro MIC values for AB 482 and AB 317 to obtain the AUC/MIC (Table 4). These averaged values were plotted and fitted to a sigmoidal curve (Fig. 5; R² = 0.25). While there are not enough data points to determine further significance of this fit, these data suggest that enhanced bacterial clearance is likely associated with elevated drug exposure and reinforces the need to improve the stability of the ADCs and reduce their rate of clearance.

TABLE 4.

Immunoprophylactic efficacy linked with drug exposure^a

Compound	24-h AUC/MIC	Burden reduction (%) (SD)
CTC-171	12.1	87.3 (8.9)
CTC-172	3.1	32.0 (50.5)
CTC-173	8.5	40.9 (30.4)
CTC-174	8.8	56.4 (34.9)

^aAverage AB 482 and AB 317 24-h AUC/MIC values from Table 3 and percentage bacterial burden reduction calculated from bacterial burdens of AB 482- and AB 317-infected mice relative to PBS-treated control for each infection.

FIG 5.

Immunoprophylactic efficacy linked with drug exposure. Sigmoidal modeling demonstrates the relationship between the average AUC_0–24/MIC and average percentage burden reduction of the tested ADCs. R squared = 0.25.

DISCUSSION

At present, treatment options, particularly for hospital-acquired infections, rely on antibacterial action and will inevitably become ineffective due to emerging antibiotic resistance. Thus, there is an urgent need for novel agents that leverage immunological mechanisms to prevent or treat infections. The appeal of ADCs to meet this need involves not only their direct antimicrobial activity, allowing for the tight binding of LPS and the disruption of OM of KAPE G– pathogens, but also their potent engagement of the immune system which enhances killing by CDC, ADCC, and ADCP mechanisms. As a proof-of-concept study, we demonstrated the development of novel ADCs and their potential usefulness as an immunoprophylactic agent against A. baumannii infections in a mouse model.

Novel dimeric peptides serving as TMs were designed and optimized to bind LPS of both sensitive and COL-resistant (COL-R) KAPE pathogens. Following an initial screen, four TMs were conjugated to EM. The hIgG1 Fc was chosen as the EM over other Fc isotypes due to its potent, balanced effector functions, immune cell activation (via Fc-gamma receptor binding), and long plasma half-life (25). A secondary benefit of this conjugation is that hIgG1 Fc cross-reacts with the murine FcRn receptor (26) and certain murine Fc-gamma receptors and is able to recruit murine complement (26, 27). These properties suggest that mouse PK, tissue distribution, and efficacy profiles for the resultant ADCs will have predictive translational value for humans.

Initial PK studies of the novel ADCs demonstrated that the C_max (5.6 to 13.3 μg/ml) was reached 2 h following IP injection, with the AUC_0–24 ranging from 49.2 to 141.0 μg · h/ml (Table 3). These studies also demonstrated significant degradation of the ADCs following administration, with less than 20% of intact molecules present after 24 h (Table 3). Although it is known that ADCs can undergo deconjugation as a result of chemical or enzymatic activities (28), current studies are focused on peptide stabilizing strategies such as d-amino acid incorporation into the linker to stabilize the conjugates.

The neutropenic murine thigh infection model of COL-sensitive and -resistant A. baumannii strains was used to assess the efficacy of the four ADC molecules. Compared to the vehicle-treated mice, all conjugates showed antibacterial activity at 20 mg/kg (Fig. 4D and E). Strikingly, CTC-171 immunoprophylaxis showed enhanced reduction compared to the other ADCs and resulted in bacterial burdens lower than those observed in colistin-treated mice (Fig. 4A). H-0028, the TM associated with CTC-171 showed low MIC values across the KAPE pathogens, suggesting that it may be an effective immunoprophylaxis if implemented in other infection models such as those of K. pneumoniae or P. aeruginosa.

This study used bacterial burdens as the primary readout for efficacy of ADCs; however, clinically, bacterial burdens are not the only primary concern. In the AB 482 infection in which CTC-173 was evaluated, 40% of the vehicle control mice succumbed to infection after approximately 36 h; however, all mice receiving either colistin or CTC-173 survived until the endpoint of the infection (not shown). While premature, the survival of ADC-treated mice is encouraging and suggests that these ADCs should be advanced to pneumonia and septicemia studies, where their reduction of morbidity and mortality can be better evaluated and the mechanisms by which ADCs promote survival can be defined. If successful in other models, clinical implementation of these novel ADCs as a viable prophylactic prior to scheduled procedures or at the time of hospital admission will be supported.

In summary, this proof-of-concept study demonstrated that novel TMs comprising dimeric peptides effectively inhibit growth of G– KAPE pathogens regardless of their antibiotic sensitivity. Furthermore, the ADCs produced by conjugation of these TMs to hIgG1 Fc have been successfully studied in mice, where it has been demonstrated that ADCs have half-lives of ≥20 h and are effective immunoprophylactic agents in a neutropenic thigh model of A. baumannii infection.

MATERIALS AND METHODS

TM synthesis.

Dimeric polymyxin TMs containing a centrally linked alkynyl group were prepared by Cidara Therapeutics (San Diego, CA, USA) or WuXi AppTec (Wuhan, Hubei, People’s Republic of China) from either commercially available polymyxin B or colistin. Final compounds were purified by reversed phase high-performance liquid chromatography (HPLC) (Gemini C₁₈ 5-μm column, 0.1% trifluoroacetic acid [TFA], acetonitrile/H₂O gradient elution) and lyophilized to give a solid with purity of >79% (220 nm).

hIgG1 Fc protein.

The hIgG1 Fc protein (29) was prepared by Genscript Biotech Corporation (Piscataway, NJ, USA) and was stored frozen in pH 7.4 PBS at −80°C and thawed at 4°C before use. The mass as measured by electrospray ionization (ESI) was 53,486 Da. Deglycosylation of an analytical sample gave a mass of 50,603 Da (theoretical mass, 50,597 Da).

Preparation of azido-PEG4-Fc.

The Fc protein was acylated via solvent-exposed lysine residues (confirmed by protein digestion and MS analysis, data not shown) with the N-hydroxysuccinimide (NHS) ester, NHS-OC(CH₂CH₂O)₄CH₂CH₂N₃ (NHS-PEG4-N₃), to obtain the intermediate conjugated protein with an average of 6.0 PEG4-azide moieties per Fc (DAR, ∼6.0) as determined by MALDI-TOF analysis of a known derivative.

Conjugation of TMs and azido-PEG4-Fc.

The azido-PEG4 Fc (1.0 eq; DAR, ∼6.0) in PBS buffer was conjugated with TM (10 to 12 eq) dissolved in PBS using the Cu(I)-catalyzed Huisgen 1,3-dipolar cycloaddition with several modifications (click reaction) over a 2-h period. Affinity purification by protein A column chromatography followed by size exclusion (SEC) chromatography gave, after buffer exchange into PBS, the desired conjugate.

Antibiotics and bacterial strains.

Colistin sulfate (PHR1605-1G; Sigma) was purchased from Sigma-Aldrich (St. Louis, MO, USA), and solution was prepared for administration to mice in 1× PBS. Strains of K. pneumoniae, A. baumannii, P. aeruginosa, and E. coli were streaked on LB agar overnight and resuspended in PBS and diluted in Mueller-Hinton Broth (MHB) prior to experiments. Twelve clinical isolates of each species were selected from the bacterial repository in the Kreiswirth lab at Center for Discovery and Innovation (CDI). Control strains include K. pneumoniae ATCC 700603, P. aeruginosa ATCC 27853, and E. coli ATCC 25922. MICs were determined for TMs and ADCs using the CLSI broth microdilution methodology (30).

Animals.

Female 6-week-old CD-1 mice weighing 22 to 28 g were obtained from Charles River Labs and were housed in the Animal Biosafety Level-2 Research Animal Facility at the Center for Discovery and Innovation, Hackensack Meridian Health (HMH). All experimental procedures were performed in accordance with National Research Council guidelines and approved by the HMH Institutional Animal Care and Use Committee (IACUC).

Neutropenic murine thigh infection model.

A well-established murine neutropenic thigh model was used for this study (24). The A. baumannii strain used for infection was grown in Mueller-Hinton broth at 37°C with shaking overnight. The culture was centrifuged, supernatant was aspirated, and the bacteria were gently washed once in sterile 1× PBS and suspended in 10 ml. The optical density was checked at 600 nm, and the OD was adjusted to get 2.0 × 10⁸ CFU/ml. The inoculum count was verified by dilution plating on LB plates which were incubated at 37°C for 24 h. Mice were rendered neutropenic by receiving 150 mg/kg and 100 mg/kg of cyclophosphamide via intraperitoneal (IP) injection on day −4 and day −1, respectively, prior to infection. An immunoprophylactic dose of ADC at 20 mg/kg was given IP 12 h prior to infection (day –1). In order to infect the mice, the mice were manually restrained and infected with 2.0 × 10⁸ CFU/ml of bacterial suspension in a volume of 0.05 ml via intramuscular (IM) injection. This resulted in 1 × 10⁷ to 4 × 10⁷ CFU/thigh as evaluated 1 h postinfection. Mice which received ADC immunoprophylaxis were administered a second 20-mg/kg dose 1 h postinfection. Treatment was also initiated for the vehicle (200 μl PBS, IP, twice daily) and colistin (2.5 mg/kg, subcutaneous (SC), twice daily) groups at 1 h postinfection.

Following infection, mice were observed twice daily for mortality and morbidity. This observation included monitoring for weight loss as a possible sign of acute toxicity. No abnormal clinical signs or morbidity was observed in ADC-treated mice, nor in vehicle- or colistin-treated mice.

To enumerate bacterial burdens at the appropriate intervals, mice were humanely euthanized by CO₂ asphyxiation. Thighs were removed, and muscle tissue was dissociated and diluted on LB agar.

PK of ADCs.

Single-dose PK studies were performed in neutropenic uninfected mice. The PK parameters for each ADC were evaluated in female CD-1 mice (4 animals/group) after 20 mg/kg IP administration. Whole-blood samples were collected via tail vein at 0.5, 2, 4, and 8 h or via cardiac puncture at 24 h postinjection. Plasma ADC concentrations at each time point were measured by sandwich enzyme-linked immunosorbent assay (ELISA).

Drug exposure was also evaluated in infected mice which received ADC immunoprophylaxis. For these studies, whole-blood samples were collected via tail vein within 30 min following infection prior to the second administration of ADC and via cardiac puncture at the time of sacrifice.

PK ELISA.

TM ELISA was performed with a TaKaRa peptide coating kit (MK100; TaKaRa Bio). The plates were coated with LPS from E. coli O127:B8 (L4516; Sigma) in coating buffer as per the TaKaRa peptide coating kit manual. Plates were incubated overnight at 4°C. Following 5 washes in washing buffer (300 μl PBS with 0.05% Tween 20), wells were blocked with blocking buffer (5% nonfat dry milk [9999S; Cell Signaling] in PBS with 0.05% Tween 20) for 2 h at room temperature (RT). The plates were washed again as described above, and then serial dilutions of the plasma samples (1:50, 1:150) in sample buffer (2.5% nonfat dry milk in PBS with 0.025% Tween 20) were loaded and incubated at RT for 2 h. An ADC standard curve ranging from 0.0004572 to 1 μg/ml, in duplicate, was run on each plate. Following the 2-h incubation, plates were washed 5 times in 300 μl PBS with 0.05% Tween 20. ADC bound to LPS on the plates was then probed with a horseradish peroxidase (HRP) conjugated anti-human IgG Fc F(ab′)2 (709-036-098; Jackson ImmunoResearch) diluted 1:2,000 in PBS for 1 h at RT. Plates were then washed 8 times in 300 μl PBS with 0.05% Tween 20 and developed with 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (34021; Thermo Fisher) for 2 min. The reaction was stopped with 1N H₂SO₄. Absorbance was read at 450 nm. EM ELISA was run using an anti-hIgG Fc capture antibody (109-005-098; Jackson ImmunoResearch) at 1 μg/ml in carbonate buffer overnight at 4°C. The samples dilutions (1:200, 1:600) and reference ADC curve were plated as described above and detected using an anti-human IgG Fc F(ab′)2 (709-036-098; Jackson ImmunoResearch) diluted 1:2,000 in PBS for 1 h at RT.

Statistical analysis and data visualization.

GraphPad Prism v8.4.3 was used to visualize data and determine statistical significance when appropriate. Ordinary one-way analysis of variance (ANOVA) was used to compare multiple groups with Tukey’s multiple-comparison test evaluating the means between groups. Statistical significance is indicated as *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001. TM and EM values in plasma were interpolated in test samples using GraphPad Prism’s nonlinear regression analysis (Sigmoidal, 4PL analysis) of the ADC standard curves.