A multifunctional bispecific antibody protects against Pseudomonas aeruginosa

See allHide authors and affiliations

Science Translational Medicine  12 Nov 2014:
Vol. 6, Issue 262, pp. 262ra155
DOI: 10.1126/scitranslmed.3009655


Widespread drug resistance due to empiric use of broad-spectrum antibiotics has stimulated development of bacteria-specific strategies for prophylaxis and therapy based on modern monoclonal antibody (mAb) technologies. However, single-mechanism mAb approaches have not provided adequate protective activity in the clinic. We constructed multifunctional bispecific antibodies, each conferring three mechanisms of action against the bacterial pathogen Pseudomonas aeruginosa by targeting the serotype-independent type III secretion system (injectisome) virulence factor PcrV and persistence factor Psl exopolysaccharide. A new bispecific antibody platform, BiS4, exhibited superior synergistic protection against P. aeruginosa–induced murine pneumonia compared to parent mAb combinations or other available bispecific antibody structures. BiS4αPa was protective in several mouse infection models against disparate P. aeruginosa strains and unexpectedly further synergized with multiple antibiotic classes even against drug-resistant clinical isolates. In addition to resulting in a multimechanistic clinical candidate (MEDI3902) for the prevention or treatment of P. aeruginosa infections, these antibody studies suggest that multifunctional antibody approaches may be a promising platform for targeting other antibiotic-resistant bacterial pathogens.


Antibody therapy with animal serum targeting bacterial toxins or capsular polysaccharides predates the discovery and development of small-molecule antibiotics (1, 2). Broad-spectrum antibiotic chemotherapy eventually displaced serum therapy and became the cornerstone of modern medicine given its relative safety in comparison to serum derived from nonhuman sources and because it enabled more convenient empiric therapy. However, increasing drug resistance to virtually all antibiotic classes, particularly within the designated ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) pathogens (3), threatens modern medicine as we know it (4, 5). This severe situation and paucity of new antibiotic classes in development coupled with the belief that empiric broad-spectrum antibiotic therapy has led to bacterial cross-resistance are stimulating renewed interest in pathogen-specific strategies including monoclonal antibody (mAb) technology for the most problematic microorganisms. Although mAbs offer considerable potential, the principal challenge for single-target mAb-based approaches is obtaining adequate protective activity against a broad range of strains and disease states. P. aeruginosa is one of the most recalcitrant ESKAPE pathogens and a leading cause of acute pneumonia in the hospital environment and of chronic lung infections in cystic fibrosis patients. Its intrinsic drug resistance, owing to its comparatively large genome and regulatory capacity, makes P. aeruginosa a challenging target for a single mAb approach. We previously reported the identification of protective mAbs that mediate serotype-independent opsonophagocytic killing of P. aeruginosa and inhibit adherence to cultured epithelial cells (6). The most protective mAb in multiple infection models was elucidated to target a determinant associated with exopolysaccharide Psl, an abundantly expressed extracellular sugar polymer implicated in immune evasion and biofilm formation (68). In addition, we recently identified a new highly active mAb against the clinically validated target PcrV, which strongly inhibits P. aeruginosa type III secretion (T3S) transport of multiple virulence factors (911). Given the important role of Psl and T3S expression in the establishment of acute and persistent P. aeruginosa infections (1215), we reasoned that a combination of the anti-Psl and anti-PcrV mAb mechanisms of action (MOAs) could enhance strain and disease coverage against P. aeruginosa. Although it is possible to co-administer antibody combinations, it is more practicable to develop a single-molecule clinical candidate. Bispecific antibody technology was first described with the derivation of hybrid hybridomas (16); however, this technology often resulted in mispaired byproducts and poor overall product yield. With the increased implementation of molecular biological methods and the development of single-chain variable fragments (scFvs) (17), construction of bispecific antibodies by fusion of scFv domains to free termini of mAb heavy chain or light chain sequences became feasible (18). Further developments in bispecific antibody technology were forged by engineering heterodimerization motifs into constant sequences, resulting in more than 50 bispecific platforms described to date (19). Whereas bispecific antibodies against viral targets have been reported (20), bispecific antibodies targeting bacterial antigens are limited (21, 22). Herein, we describe enhanced anti–P. aeruginosa activity afforded by a multimechanistic bivalent, bispecific antibody configuration targeting Psl and PcrV designated BiS4αPa. The potent serotype-independent activity of BiS4αPa observed against diverse strain types, including multidrug-resistant (MDR) strains, in multiple animal infection models in both prophylactic and therapeutic regimens, and the surprisingly potent in vivo synergy in adjunctive therapy with multiple antibiotic classes, supports this molecule as a promising clinical candidate (designated MEDI3902) for the prophylaxis or adjunctive treatment of P. aeruginosa infections. The successful application of this approach suggests a more broadly applicable strategy to combat other serious antibiotic-resistant bacterial pathogens with multifunctional antibodies.


Anti-pseudomonal MOAs are functional in bispecific mAb formats

In efforts to identify serotype-independent protective mAbs against P. aeruginosa, we had previously identified unique progenitor mAbs targeting the Psl exopolysaccharide and the PcrV component of the T3S transport system (6, 11). Given the different roles of Psl and T3S in P. aeruginosa infection (1215), and the protective activities afforded by the individual anti-Psl and anti-PcrV parent mAbs in multiple P. aeruginosa infection models (6, 11), we reasoned that combining both specificities and activities could significantly enhance protection and broaden P. aeruginosa strain coverage. A broad geographical survey of 269 recent P. aeruginosa clinical isolates showed that the vast majority of strains are capable of expressing Psl (89.8 to 91.2%) and PcrV (87.7 to 90.2%), whereas 97.3 to 100% of isolates expressed either or both targets (table S1). The relative prevalence of cytotoxic exoU and invasive exoS strain types was also consistent with previous reports (table S1) (2325). Because the length of the surface-expressed T3S system needle is thought to extend 80 to 120 nm from the surface of the bacterium (2628), it appeared plausible that a combination of anti-Psl and anti-PcrV binding units separated by suitable interparatopic distances on a single molecule, coupled with adequate flexibility and geometry, could allow simultaneous binding to both Psl and PcrV surface targets. We therefore constructed bispecific antibodies possessing anti-PcrV and anti-Psl specificities with varying intramolecular distances between binding units using the anti-PcrV mAb, V2L2-MD [human immunoglobulin G1 (IgG1)] (11), as the bispecific antibody scaffold (Fig. 1A). Two previously described bispecific antibody formats (18, 29) were initially selected for this study, in which the anti-Psl scFv is genetically linked to the heavy chain N terminus (BiS2αPa) or the heavy chain C terminus (BiS3αPa), resulting in proximal and distal interparatopic distances, respectively (Fig. 1, B and C). In addition, we devised and constructed a unique bispecific configuration with an intermediate interparatopic distance between antigen binding sites, designated BiS4αPa, by genetically inserting the anti-Psl scFv in the upper hinge region of the anti-PcrV mAb coding scaffold (Fig. 1D).

Fig. 1. P. aeruginosa bispecific antibodies compared to IgG1.

(A) Parental IgG1. (B to D) Bispecific constructs using anti-PcrV mAb V2L2-MD (11) as a scaffold. scFvs of anti-Psl mAb Psl0096 (derived from Cam-003) (6) were linked via a 10–amino acid linker (GGGGSx2) to the V2L2-MD heavy chain N terminus (B), heavy chain C terminus (C), or between C220 and D221 of the upper hinge region flanked by 10–amino acid linkers (GGGGSGGGGS) (D). Fab light chain is light blue, and heavy chain is dark blue; CDR1-3H, red; CDR1-3L, salmon; scFv variable light (VL) chain is light orange and variable heavy (VH) chain is dark orange; hinge and Fc are green. Linkers between VH and VL in scFv and between scFv and IgG sequences are gray. Carbohydrates at the N297 Fc glycosylation site are rendered in the stick conformation. Illustrations depict the complementarity determining regions (CDRs) of the scFvs pointing outward toward antigen binding sites. Subunit images were rendered in PyMOL (Delano Scientific) and assembled in PowerPoint.

All bispecific constructs were assessed for their in vitro potency compared to each respective parental mAb. Opsonophagocytic killing activity (Fig. 2A), cell attachment inhibition (Fig. 2B), and inhibition of cytotoxicity (Fig. 2C) were measured for anti-Psl and anti-PcrV mAbs. Whereas all bispecific constructs exhibited strong anti-Psl opsonophagocytic killing activity, a modest reduction in opsonophagocytic killing activity was observed for the BiS2αPa and BiS4αPa bispecific antibodies, while the BiS3αPa construct exhibited the greatest reduction in opsonophagocytic killing activity relative to the parent anti-Psl mAb or mixture of anti-Psl and anti-PcrV mAbs (Fig. 2A). In contrast, BiS3αPa and BiS4αPa exhibited enhancement of anti-Psl–mediated inhibition of P. aeruginosa attachment to cultured epithelial cells in comparison to BiS2αPa, the parent anti-Psl mAb, or mAb mixture (Fig. 2B).

Fig. 2. Bispecific antibody anti-Psl and anti-PcrV activity against P. aeruginosa.

(A) Opsonophagocytic killing activity of antibody constructs using the P. aeruginosa luminescent reporter strain PAO1.lux. Samples were run in duplicate and are representative of three independent experiments. (B) Anti–cell attachment activity of antibody constructs using PAO1.lux reporter strain. Error bars indicate the SEM of four samples and are representative of four independent experiments. (C) P. aeruginosa epithelial cell cytotoxicity assay using cytotoxic P. aeruginosa strain 6077 (ExoU+). Samples were run in duplicate and are representative of four independent experiments. Statistical comparisons were conducted by calculating the area under the curve (AUC) comparing bispecific constructs to anti-Psl mAb (A and B) and anti-PcrV mAb (C).

The anti-PcrV component of the bispecific constructs was next evaluated for inhibition of P. aeruginosa T3S-mediated acute cytotoxicity, which is mediated by strains that express the ExoU phospholipase (30, 31). The constructs with the greatest interparatopic spacing, BiS3αPa and BiS4αPa, displayed significantly enhanced anti-cytotoxic activity at lower antibody concentrations (<1 nM) compared to the parent anti-PcrV mAb (P = 0.0001 for both BiS3αPa and BiS4αPa), whereas no difference in activity was observed between the anti-PcrV mAb, the mAb mixture, and BiS2αPa in the in vitro cytotoxic inhibition assay (Fig. 2C). This finding suggests that the BiS2 platform containing the shortest interparatopic distance between anti-Psl and anti-PcrV binding units may not be adequately spaced to confer the benefit of anti-Psl binding avidity or simultaneous engagement with both targets. Together, these in vitro data for the three distinct MOAs suggested a collective superiority for BiS4αPa in comparison to the other bispecific constructs, parent mAbs, or mAb combination.

BiS4αPa exhibits enhanced protective activity in vivo

We next evaluated each bispecific antibody in comparison to individual parental anti-Psl and anti-PcrV mAbs and the mAb mixture for protection against lethality in a murine lung infection model caused by cytotoxic (ExoU+) strain 6206, which is representative of the most highly pathogenic P. aeruginosa challenge strains we have tested (6, 11) (Table 1). Prophylactic administration of the single-parent anti-Psl mAb provided no protection against strain 6206, whereas anti-PcrV, BiS2αPa, and BiS3αPa mAbs protected 40 to 80% of the challenged animals at 5 mg of antibody per kilogram of body weight (mpk) and provided minimal protection (10 to 20%) at the 1 mpk dose (Table 1). In contrast, we consistently observed enhanced activity in mice dosed prophylactically with both the mAb mixture (1 mpk for each mAb) and BiS4αPa (1 mpk total) (78 and 88%, respectively). Whereas isobologram analysis was not possible given the lack of activity for the anti-Psl mAb at all tested doses for strain 6206, the enhanced activity for the mAb combination and BiS4αPa was synergistic given the marked increase in potency over the anti-PcrV parent mAb in the absence of observable anti-Psl activity. Further titration of BiS4αPa (0.5 mpk) and the mAb mixture (0.5 mpk for each mAb) indicated enhanced efficacy for BiS4αPa (P < 0.0001), even though the larger–molecular weight BiS4αPa construct (200.3 kD versus 150 kD for an IgG1, respectively) and its anti-Psl and anti-PcrV binding units are at a molar and weight disadvantage in these studies. The enhanced BiS4αPa activity compared with the mAb mixture was also confirmed in studies with another P. aeruginosa cytotoxic (ExoU+) clinical isolate, MDR strain 6077 (Table 1), which is more representative of a broader range of clinical isolates used as challenge strains in the murine pneumonia model. Activity against strain 6077 was also confirmed to be synergistic in comparison to the individual parental mAbs by isobologram analysis [isobole (I) = 0.43] (fig. S1A). In addition to prophylactic activity, therapeutic administration (1 hour after infection) of BiS4αPa also resulted in potent protective activity against challenge strains 6206 and 6077 in the murine lethal pneumonia model (fig. S2, A and B).

Table 1. Protection against lethal P. aeruginosa pneumonia.

Summary of lethal pneumonia survival studies completed for the indicated mAbs against P. aeruginosa strains 6206 and 6077. Mice were treated prophylactically with antibody 24 hours before infection. No survival was observed in mice treated with the isotype control IgG against either strain. Differences in survival for BiS4αPa versus the anti-Psl/anti-PcrV mAb mixture at 0.5 mpk for 6206 and 0.02 mpk for 6077 were evaluated by the log-rank test (*P < 0.0001). 6206: Results were compiled from 15 independent experiments; n is indicated in parentheses. 6077: Results were compiled from 11 independent experiments; n is indicated in parentheses. For the anti-Psl and anti-PcrV mAb mixture, the indicated mpk is for each mAb. Parentheses indicate total number of animals analyzed at each antibody dosage. nd, not determined.

View this table:

Given the consistently enhanced in vivo activity of BiS4αPa, we focused our remaining analyses on this molecule as a potential candidate for clinical development. We first evaluated the impact of BiS4αPa on bacterial burden in lung and distal organ tissues using the lethal pneumonia mouse model. All doses of BiS4αPa significantly reduced P. aeruginosa lung colony-forming units (CFUs) in comparison to mice dosed prophylactically with control IgG (P < 0.05) (fig. S3, A and B). BiS4αPa also significantly reduced bacterial dissemination to the spleen and kidneys in comparison to control IgG (P < 0.05) (fig. S3, A and B). Furthermore, histopathological evaluation of lungs from infected mice dosed prophylactically with BiS4αPa alone 24 hours before infection or treated 4 hours after infection revealed marked reductions in pathological markers of pneumonia and cellular damage in comparison to mice treated with control IgG (Fig. 3, A and B).

Fig. 3. BiS4αPa-mediated reduction in pneumonia lung pathology in mice.

(A and B) Mice received IgG control or BiS4αPa prophylactically (n = 3 for each, respectively) (A) or as a single-agent treatment at 4 hours after infection with P. aeruginosa strain 6077 (n = 3 for each, respectively) (B). Mice receiving control IgG presented with hemorrhagic interstitial pneumonia involving most of the examined lung section, along with marked bronchial epithelial damage and severe edema surrounding the bronchioles and blood vessels. In addition, control IgG–treated animals also showed the presence of bacterial colonies (inset) in three of three mice in the treatment group and in one of three mice in the prophylaxis group. In contrast, mice receiving BiS4αPa alone in prophylaxis or treatment exhibited minimal inflammatory infiltrate in the alveolar spaces with no detectable bacteria, primarily showing bronchial pneumonia with little or no alveolar hemorrhage and edematous changes surrounding the larger blood vessels. The pathologist was blinded to the treatment groups.

The protective activity of BiS4αPa was also evaluated against a collection of recent clinical isolates of P. aeruginosa. BiS4αPa exhibited high-level in vitro anti-Psl–mediated opsonophagocytic killing activity (fig. S4, A to D) and anti-PcrV–mediated anti-cytotoxic activity (fig. S5, A to I) against all strains tested. Furthermore, prophylactic or therapeutic administration of BiS4αPa prevented lethality from acute pneumonia (Fig. 4, A to D) against isolates known to be resistant to all anti-pseudomonal standard-of-care antibiotics with the exception of colistin [minimal inhibitory concentrations (MICs) presented in table S2].

Fig. 4. BiS4αPa protective activity against MDR P. aeruginosa lethal pneumonia.

(A to D) BiS4αPa was screened against MDR isolates ARC3928 (A and B) and ARC3502 (C and D) under both prophylactic (A and C) (T = 24 hours before infection) and therapeutic regimens [T = 1 hour after infection (B) and T = 0.15 hour after infection (D)]. Results are represented as Kaplan-Meier survival curves; significant differences were calculated by the log-rank test by comparing each concentration of BiS4αPa to the control IgG. (A to D) Data are representative of three independent experiments. (A) Control IgG, n = 8; BiS4αPa at 1.0, 0.3, and 0.1 mpk, n = 8; BiS4αPa at 0.03 mpk, n = 7. (B) Control IgG, n = 8; BiS4αPa, n = 10 for all groups. (C) Control IgG, n = 8; BiS4αPa, n = 8 for all groups. (D) Control IgG, n = 8; BiS4αPa, n = 8 for all groups.

Multiple BiS4αPa MOAs contribute to synergistic protection

Given the superior in vivo activity of BiS4αPa in comparison to the parent mAb mixture, we next investigated the potential mechanism by which BiS4αPa mediates enhanced anti-cytotoxic activity in comparison to the anti-PcrV mAb or parent mAb mixture. We hypothesized that the high-avidity lower-affinity binding of the BiS4αPa anti-Psl module (6) to the abundant surface Psl exopolysaccharide effectively increases the local antibody concentration and residence time around the bacterium without impeding the higher-affinity engagement of the anti-PcrV component (11) of the bispecific antibody to its target, thereby allowing cooperative target engagement while increasing anti-PcrV activity around the bacterium. To test this hypothesis, we constructed several BiS4αPa antibody variants that lacked one or more functional activities contained within parental BiS4αPa. First, BiS4αPcrV was constructed with a negative control scFv in place of anti-Psl binding unit. The in vitro anti-cytotoxic activity of BiS4αPcrV was significantly reduced in comparison to BiS4αPa (P < 0.001) but was not different when compared to the parental anti-PcrV mAb (P = 0.434) (Fig. 5A). A significant difference was also observed in a direct comparison of BiS4αPa versus anti-PcrV (P < 0.001). The increased anti-cytotoxic activity of BiS4αPa in comparison to mAbs lacking anti-Psl specificity became apparent at mAb concentrations lower than 1 nM in the in vitro dose-response curve. Consistent with these results, no difference in activity was observed between the anti-cytotoxic activities of BiS4αPa, BiS4αPcrV, and the parental anti-PcrV mAb when evaluated against a Psl-deficient strain of P. aeruginosa (6206ΔpslA) (Fig. 5B). These in vitro data suggested that the enhanced anti-cytotoxic activity observed with BiS4αPa is associated with Psl engagement by BiS4αPa. These data were further confirmed in the acute pneumonia mouse model induced by strain 6206ΔpslA, in which no difference in activity was observed in mice prophylactically treated with either BiS4αPa or the parent anti-PcrV mAb at molar equivalent doses (Fig. 5C).

Fig. 5. Contributions of BiS4αPa MOAs to enhanced activity.

(A) P. aeruginosa epithelial cell cytotoxicity assay with wild-type P. aeruginosa. Error bars represent the SEM for six samples and are representative of six independent experiments. (B) Cytotoxicity assay with Psl-deficient P. aeruginosa. Samples were run in duplicate and are representative of three independent experiments. (C) Acute pneumonia mouse model. Mice were treated with molar equivalent doses of BiS4αPa and anti-PcrV mAb (V2L2-MD) 24 hours before infection with 6206ΔpslA. No statistical difference was observed by the log-rank test between treatment groups (n = 10). Representative data from three independent experiments. (D) Opsonophagocytic killing assay. Error bars represent the SEM for three samples and are representative of three independent experiments. (E) Anti–cell attachment activity with luminescent P. aeruginosa. Error bars represent the SEM for three samples and are representative of two independent experiments. (F) Acute pneumonia model. Mice were treated with molar equivalent doses of the indicated antibodies followed by infection with wild-type strain 6206. Results compiled from six independent experiments. Control IgG, n = 48; anti-Psl, n = 16; BiS4αPsl, n = 24; anti-PcrV, n = 16; BiS4αPcrV, n = 16; BiS4αPa-N297Q, n = 16; BiS4αPa, n = 48. (A, B, D, and E) Statistical comparisons were conducted by calculating the AUC.

In addition to enhancing the anti-cytotoxic activity mediated by anti-PcrV, the anti-Psl component of BiS4αPa mediates opsonophagocytic killing in the presence of effector cells (Fig. 2A) and inhibits P. aeruginosa binding to epithelial cells (Fig. 2B). We next evaluated the relative contribution of anti-Psl opsonophagocytic killing and inhibition of cell attachment to the overall activity of BiS4αPa by constructing variants that lacked anti-cytotoxic activity as well as a clone deficient for effector function necessary for opsonophagocytic killing activity. BiS4αPsl, lacking anti-cytotoxic activity, was constructed by replacing the anti-PcrV binding unit sequence of BiS4αPa with an irrelevant negative control antibody sequence. In addition, we constructed BiS4αPa Fc domain variant, with an N297Q point mutation previously shown to diminish FcγR engagement and C1q binding (32), thereby diminishing effector cell–mediated opsonophagocytic killing activity without affecting Psl and PcrV target binding. No difference in the opsonophagocytic killing (Fig. 5D) and adherence inhibitory activity (Fig. 5E) was observed between BiS4αPsl and BiS4αPa. In contrast, BiS4αPa-N297Q exhibited significantly impaired anti-Psl opsonophagocytic killing activity in comparison to BiS4αPa (P < 0.001) (Fig. 5D), without affecting the enhanced in vitro inhibition of Psl-mediated adherence (Fig. 5E) or the anti-PcrV–mediated anti-cytotoxic activity (Fig. 5A). The activity of BiS4αPa, associated variants, and parent mAbs was then compared in the pneumonia model induced by wild-type P. aeruginosa strain 6206. A significant reduction in activity was observed with all variant constructs in comparison to BiS4αPa (Fig. 5F). No difference in activity was observed between BiS4αPcrV and the anti-PcrV mAb. Although BiS4αPa-N297Q exhibited intermediate activity, reduced in comparison to the complete BiS4αPa (P = 0.01) construct, its activity was significantly greater than that of the BiS4αPcrV (P < 0.0001) or parental anti-PcrV mAbs (P < 0.0001) (Fig. 5F). Together, these results indicate that Psl targeting potentiates BiS4αPa anti-cytotoxic activity and that anti-Psl opsonophagocytic killing and possibly anti–cell attachment activity also contribute to efficacy in the murine pneumonia model. In addition, these data further confirm that targeting both Psl and PcrV with BiS4αPa results in significantly enhanced efficacy in comparison to the single-parent mAbs with the potential of broadening coverage against P. aeruginosa strains that might not express one of the two targets.

BiS4αPa is protective in multiple infection models

Given that P. aeruginosa infections are particularly problematic in immunosuppressed patients and because BiS4αPa functions at least in part by effector cell–mediated opsonophagocytic killing, we further evaluated BiS4αPa activity in a cyclophosphamide-induced immunocompromised pneumonia model in which total white blood cell and neutrophil counts are markedly reduced (33). BiS4αPa provided concentration-dependent protection in both the prophylactic and post-infection treatment models in these immunosuppressed mice (fig. S6, A and B).

P. aeruginosa also remains a significant cause of bacteremia and mortality in the critically ill and in serious infections of burn wounds (34, 35). We therefore evaluated BiS4αPa in murine thermal injury and bacteremia infection models. In the thermal injury model, attachment and colonization of injured murine tissues with P. aeruginosa result in the formation of a biofilm (36) followed by systemic bacterial dissemination as a consequence of immune cell anergy (37). Prophylaxis or treatment with BiS4αPa provided significant protection at all antibody concentrations evaluated for both cytotoxic (ExoU+) (fig. S6, C and D) and invasive (ExoS+) (fig. S6, E and F) clinical isolates; the ExoS+ strain is highly lethal in the thermal injury model as indicated by the low LD100 (absolute lethal dose) challenge dose. BiS4αPa administered prophylactically or therapeutically also demonstrated potent activity against bacteremia induced by intravenous challenge with multiple strains of P. aeruginosa (fig. S6, G to J).

BiS4αPa promotes synergistic protection with antibiotic therapy

Given the steady increase in global MDR infection rates, we tested whether BiS4αPa antibody–mediated therapy could complement the antibacterial activity of multiple antibiotics used to treat P. aeruginosa. We first selected marginal subtherapeutic antibiotic doses of ciprofloxacin (CIP) and meropenem (MEM) (two antibiotics with disparate MOAs) to simulate the insufficient drug exposure for antibiotic-resistant strains in the lethal pneumonia model. These subtherapeutic antibiotic doses were then used to treat mice in which subprotective prophylactic doses of BiS4αPa had been administered 24 hours before infection to evaluate the potential for synergistic activity. Animals receiving control IgG or single-agent treatments all succumbed to infection, whereas mice receiving the subtherapeutic combinations of antibiotic plus BiS4αPa survived challenge (Fig. 6, A and B). Isobologram analyses confirmed synergistic activity when combining BiS4αPa with either CIP (I = 0.21) or MEM (I = 0.45) (fig. S1, B and C). We next evaluated if this activity could be recapitulated in a post-infection treatment setting. Subtherapeutic doses of individual drugs and the BiS4αPa/antibiotic combinations were adjunctively delivered to mice 4 hours after infection. Enhanced protection was observed in mice receiving the BiS4αPa/antibiotic combinations in comparison to little or no protection in animals treated with control IgG or individual antibiotics (Fig. 6, C and D). We further evaluated the impact of BiS4αPa on P. aeruginosa bacterial burden and lung pathology in mice receiving CIP with BiS4αPa treatment at 4 hours after infection. CIP-treated groups demonstrated significant reductions in bacterial burden in comparison to non-CIP–treated groups in all tissues evaluated with or without co-administration of the control mAb or subtherapeutic dose of BiS4αPa at 4 hours after infection (Fig. 6E). The combination of CIP and BiS4αPa provided only marginal additional CFU reduction over CIP treatment alone in the lung. However, only the combination of BiS4αPa and CIP prevented death in this model, illustrating that P. aeruginosa lethality is not solely dependent on bacterial burden and that antibody functional activities can further enhance antibiotic therapy by disarming pathogenic mechanisms and reducing pathology while at the same time promoting bacterial clearance (Fig. 6E). This protection was evident in further histopathological evaluation of lungs from infected animals. Control IgG– and CIP-treated animals presented with severe bronchial epithelial damage, edema, and severe hemorrhagic interstitial pneumonia, whereas these pathological characteristics were absent in mice treated with BiS4αPa and CIP (Fig. 6F).

Fig. 6. Synergistic activity of BiS4αPa and antibiotics against P. aeruginosa pneumonia.

(A and B) Previously determined subprotective doses of BiS4αPa were administered to mice 24 hours before intranasal infection with P. aeruginosa strain 6206. Subtherapeutic doses of CIP (A) or MEM (B) were administered 1 hour after infection. (C and D) Mice were treated with subprotective doses of both BiS4αPa and CIP (C) or MEM (D) 4 hours after intranasal infection with P. aeruginosa strain 6206. (F) Histological evaluation of lungs from mice treated with IgG control (n = 3) or BiS4αPa (n = 3) in combination with subtherapeutic doses of CIP 4 hours after infection. (A to D) Results are represented as Kaplan-Meier survival curves; differences in survival were calculated by the log-rank test for the BiS4αPa and antibiotic combination versus antibiotic treatment alone. (A to D) n = 6 for all antibody treatment groups. Representative data from two (A and B), five (C), and four (D) independent experiments. (E) Organ burden analysis of animals treated with BiS4αPa and CIP 4 hours after infection. Statistical comparisons were performed by one-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison test. Results are presented as a scatter dot plot and are marked by the SEM and mean. Control IgG, Control IgG + CIP, CIP alone and BiS4αPa + CIP, n = 10; BiS4αPa alone, n = 9. Data are representative of three independent experiments.

The synergy observed with subtherapeutic doses of BiS4αPa and multiple antibiotic classes strongly suggested the potential for enhancing marginally active antibiotics against drug-resistant strains of P. aeruginosa. We further confirmed this potential by testing BiS4αPa combined with a dose regimen of the aminoglycoside tobramycin (TOB) approximating human drug exposure versus both a TOB-susceptible and TOB-resistant strain of P. aeruginosa in the murine acute pneumonia model. As expected, the human equivalent multidose regimen exposure of TOB protected against susceptible P. aeruginosa strain 6206 (MIC: 0.5 μg/ml) (Fig. 7A) but failed to protect against TOB-resistant P. aeruginosa strain 6077, which has an MIC eightfold above the Clinical and Laboratory Standards Institute (CLSI)–established MIC break point (1 μg/ml) for human TOB therapy (Fig. 7B). In contrast, even at a low single dose of 0.1 mg/kg, which is subtherapeutic under the conditions of this experiment, BiS4αPa was synergistically protective in adjunctive therapy with TOB against this resistant strain (Fig. 7C). We next evaluated the impact of BiS4αPa on strain 6077 bacterial burden in mice receiving control mAb, TOB, or BiS4αPa individually and BiS4αPa and TOB in combination. Whereas control IgG– and TOB-treated mice yielded similar bacterial burdens, mice receiving the subtherapeutic dose of BiS4αPa in combination with TOB exhibited significantly lower bacterial burden in evaluated tissues (P = 0.0006) (Fig. 7D).

Fig. 7. Synergistic activity of BiS4αPa and TOB against resistant P. aeruginosa.

(A to C) Efficacy of TOB against (A) TOBS strain 6206 and (B) TOBR strain 6077. (C) Mice were treated with subprotective doses of both BiS4αPa (T = 24 hours before infection) and TOB (T = 1 hour) after intranasal infection with P. aeruginosa strain 6077. (A to C) Results are represented as Kaplan-Meier survival curves. Significant differences were calculated by the log-rank test comparing (A and B) TOB-treated mice to the diluent control, and (C) TOB treatment alone to BiS4αPa + TOB treatment. (A to C) Data are representative of three independent experiments. (A) n = 6, (B) n = 5, and (C) n = 8 for all treatment groups. (D) Organ burden analysis in 6077-infected animals treated with BiS4αPa + TOB. Results are presented as a scatter dot plot and are marked by the SEM and mean. Differences in CFU were determined by the Mann-Whitney U test for pairwise comparison of TOB versus BiS4αPa + TOB–treated groups. Control IgG, n = 7; TOB, BiS4αPa, and BiS4αPa + TOB, n = 8. Data are representative of two independent experiments.


In addition to the treatment of primary community or trauma-associated infections, much of modern medical practice is only possible with the use of antibiotics to prevent or treat opportunistic infections associated with high-risk surgeries. However, empiric use of broad-spectrum antibiotics has led to the progressive selection of cross-resistance to most antibiotic classes. With insufficient potential for new antibiotic classes on the horizon, the antibiotic resistance crisis has brought more focus on strategies for preserving the valuable antibiotics we have left or will have in the future. This dire situation has also stimulated work on pathogen-specific biological strategies using modern antibody-based technologies as one possible approach to preserve and augment antibiotic options. mAbs offer considerable potential for preventing or managing bacterial infections including enhanced specificity and safety, lack of drug-drug interactions, long half-lives, and complementary MOAs that can enhance antibiotic activities without driving antibiotic resistance or disruption of the beneficial microbiota. However, in many circumstances, mAbs aimed at single targets may offer only marginal effectiveness against diverse strain types across different disease manifestations. Among the drug-resistant bacterial ESKAPE pathogens, P. aeruginosa may present the greatest challenge because of its large genome coding capacity and complex regulatory networks, allowing the bacterium to phenotypically adapt to environmental pressures and to form persistent biofilm communities. Multiple P. aeruginosa proteins and carbohydrates have been explored individually as antibody targets in the laboratory, few of which are currently under clinical evaluation (9, 38).

For challenging drug-resistant bacterial infections, multidrug therapies with multiple MOAs are increasingly required to reduce the potential for increased drug resistance. Here, we reasoned that a combination of complementary mAbs with different MOAs against the TS3 system and Psl exopolysaccharide, targets that are expressed by most P. aeruginosa primary clinical isolates (table S1), could provide enhanced activity against acute infection and prevent the establishment of persistent infections. Indeed, the anti-cytotoxic activity provided by anti-PcrV to blunt the pathogenesis and invasive progression of infection coupled with anti-Psl–mediated opsonophagocytic killing, clearance, and anti-adherence activities augmented activity against P. aeruginosa–induced lethal pneumonia in mice. In addition, we hypothesized that in a bispecific mAb format, the high-avidity binding activity against the abundant Psl target could enhance activity against the low-abundance PcrV target, thereby resulting in a higher local mAb concentration “cloud” around the bacterium, and thus effectively increasing bacterial cell surface–associated anti-PcrV activity, which would not be observed with an anti-PcrV mAb alone. This hypothesis was supported by the enhanced anti-cytotoxic activity for the two bispecific constructs in vitro, particularly at low concentrations where the control PcrV mAb alone was ineffective. Of the three bispecific mAb configurations tested, the BiS4 configuration was the only construct exhibiting enhanced cytotoxic activity without a substantial deficiency in the anti-Psl–mediated opsonophagocytic killing activity. The novelty of the BiS4αPa bispecific format lies in the placement of the anti-Psl scFv, which is inserted into the upper hinge region rather than appended to the N or C termini of the antibody heavy chain as previously described in other bispecific configurations. This configuration offers an intermediate distance between paratopes and appears to be most optimal for dual target engagement in the applications examined herein. Although the BiS4 format worked best for this target combination, these studies also demonstrate that having a sufficient range of multispecific formats may be necessary for success in other antibacterial applications. These studies may also shed light on what types of antibacterial mAb target combinations might result in synergistic outcomes.

Because antibodies generally do not directly kill bacteria, it may seem most reasonable to consider these agents for preventing disease before bacterial burden is high and the infection is established. It would also seem that antibody prophylaxis would be most practical in severely ill patients at highest risk for infections causing high morbidity and mortality, and for infections that are extremely expensive to treat. P. aeruginosa infections in intensive care settings certainly fall into this class (39). However, the data provided herein for BiS4αPa also suggest a potential beyond prophylaxis, particularly for the treatment of acute drug-resistant P. aeruginosa infection for which few desirable antibiotic options remain. Although BiS4αPa showed considerable potential in monotherapy with multiple clinical strain types in multiple animal models, also striking was the unexpected synergy observed in combination with multiple antibiotic classes, even with strains that are highly resistant to the partner antibiotic. These data indicate that BiS4αPa has the potential to complement the bactericidal or bacteriostatic activities of antibiotics, perhaps by limiting the tissue damage caused by the cytotoxic pathogenic mechanisms of the bacteria while also promoting bacterial clearance, thereby minimizing the potential for a concomitant destructive hyper-inflammatory innate host response. These complementary and synergistic protective activities coupled with the long antibody half-life and unlikely drug-drug interactions between antibody and antibiotics offer the attractive option of combination therapy with antibiotics. The data provided herein demonstrate considerable potential for the prevention and treatment of acute P. aeruginosa lung infections. However, we have also demonstrated protective post-infection treatment potential in the thermal injury mouse model, a comparatively less acute infection model in which biofilms are reported to form (36). Given that the Psl exopolysaccharide target has been implicated in biofilm formation and maintenance, it seems reasonable to consider that antibodies targeting Psl with attendant antibody-dependent effector cell engagement may have the potential to aid in containing established biofilm communities. However, additional studies would be required to assess anti-biofilm activities in an in vivo setting of infection (40, 41).

If the full potential for antibody-based strategies to augment our antibacterial options is to be realized, multifunctional approaches must be pursued. However, mAb combinations and recombinant polyclonal approaches present formidable and arduous clinical development challenges including the potential requirement to develop multiple bioprocesses and to produce, purify, and clinically assess each individual mAb. Notwithstanding the potential advantages for single-molecule multifunctional bispecific mAbs versus antibody mixtures, the potential for undesirable biophysical properties or more immunogenic structures may be greater for non-germline, engineered multispecific antibody structures. Although not addressed in the study here, biophysical formulation stability must also be monitored in the decision schema for selection of a multispecific clinical candidate. However, immunogenicity liabilities that could affect mAb candidate pharmacokinetics or activities in humans cannot be adequately assessed in preclinical studies and may only be finally assessed in human clinical trials. Our derivation of BiS4αPa portends one possible multifunctional bispecific strategy to ameliorate some of the challenges associated with mAb combinations. Additionally, these findings support a significant benefit of synergistically augmenting antibiotic bactericidal activities through inhibiting cytotoxicity, bacterial adherence, and promoting effector cell clearance mediated by advanced biological therapeutics. While providing supporting data for BiS4αPa as a human clinical candidate (designated hereafter as MEDI3902) to prevent or possibly treat acute P. aeruginosa infections, these results also suggest a promising new bispecific antibody platform, enabling multimechanistic strategies to augment and preserve our dwindling antibiotic options for serious drug-resistant bacterial infections.


Study design

We constructed multimechanistic bispecific antibodies with specificities against P. aeruginosa Psl exopolysaccharide and PcrV, a protein required for type III secretion injectisome activity, to evaluate the potential of enhanced strain and disease coverage against P. aeruginosa. All antibody constructs were first tested in in vitro functional activity screens followed by validation and comparison of efficacy in a P. aeruginosa acute pneumonia mouse model. One new multimechanistic bispecific antibody, BiS4αPa, exhibiting synergistic protection against lethal challenge in the pneumonia model, was selected as the lead clinical candidate and further evaluated against lethal infection in P. aeruginosa mouse models of thermal injury and bacteremia. In addition, we evaluated BiS4αPa activity when combined adjunctively with antibiotics. Experts in statistics and experimental design were consulted to validate the design of all in vivo experiments before execution. All in vivo studies were performed in accordance with federal, state, and institutional guidelines and were approved by the MedImmune Institutional Animal Care and Use Committee in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited facility. Animals were monitored closely for survival up to 120 hours after challenge. Guidelines for humane endpoints were strictly followed for all in vivo experiments; moribund animals were immediately euthanized by CO2 asphyxiation and recorded as a nonsurvivor. Sample sizes in all animal studies for each model were estimated using log-rank test with 5% type I error rate and 80% power. The hypothesized effect size for each comparison was derived from historical data or pilot study data. Sample sizes were calculated using nQuery Advisor software. All animals were randomly assigned to treatment groups using a randomization tool implemented in MS Excel. Animal allocations were not blinded to study scientists for containment and personnel considerations. Pathologists that evaluated tissue sections were blinded to study groups.

Synergistic effects between different antibodies and antibodies in combination with antibiotics were evaluated using a typical isobologram approach as described (42, 43). All samples were included in all experiments. All in vitro and in vivo experiments were repeated at least three times unless otherwise stated within the figure legends.

Bispecific antibody construction

Antibody sequences were cloned into cytomegalovirus promoter-driven expression vectors. Anti-PcrV and control specificities in the Fab position of the molecules were transferred from existing expression plasmids using restriction digestion and ligation. scFv sequences were generated by gene synthesis (Eurofins MWG Operon), amplified by polymerase chain reaction using primers that generated a 15-nucleotide overlap to the target insertion site, and inserted using InFusion (BD Clontech). Anti-Psl scFv sequences were generated in a VH-VL orientation with a 20–amino acid linker (GGGGSx4). Negative control scFv sequences were generated similarly but in a VL-VH orientation. All scFv sequences were stabilized by engineering a VH-VL interdomain disulfide bond generated by mutating VH44 and VL100 (Kabat numbering) to Cys. BiS2 was constructed by fusing scFv sequences to the N terminus of the heavy chain, separated by a 10–amino acid linker (GGGGSx2). Similarly, BiS3 was constructed by appending the linker-scFv to the C terminus of the heavy chain. BiS4 constructs were generated by flanking the scFv with 10–amino acid linkers (GGGGSx2) and inserting into the upper hinge region between C220 and D221 (EU numbering). All proteins were expressed by transient transfection in 293 cells, purified by protein A affinity chromatography, and polished using size exclusion chromatography. The integrity of the molecules was verified using mass spectrometry, both intact mass and peptide mapping, to ensure proper formation of engineered and endogenous disulfide bonds.

Opsonophagocytosis killing assay

Assays were performed as described (6, 44). Briefly, assays were performed in 96-well plates using 0.025 ml of each component, luminescent P. aeruginosa strains, diluted baby rabbit serum, differentiated HL-60 cells, and mAb. Data were acquired using an Envision Multilabel plate reader (Perkin Elmer) and plotted as percent killing compared to a control lacking antibody.

Cell attachment assay

Assays were performed as described (6). Briefly, antibodies were added to confluent A549 cells grown in opaque 96-well plates (Nunc Nunclon Delta). Log-phase luminescent PAO1 was added at a multiplicity of infection (MOI) of 10. After incubation at 37°C for 1 hour, cells were washed followed by addition of LB + 0.5% glucose. Bacteria were quantified after a brief incubation at 37°C.

Cytotoxicity assay

Assays were performed as described (11). Briefly, antibodies were added to A549 cells seeded in white 96-well plates (Nunc Nunclon Delta) in Dulbecco’s modified Eagle’s medium plus 10% fetal bovine serum. Log-phase P. aeruginosa clinical isolates capable of expressing ExoU were added at an MOI of 10 and incubated for 2 hours at 37°/5% CO2 followed by measurement of lactate dehydrogenase released from lysed cells.

Calculation of MICs

MICs were performed using the materials, standards, and methods set forth by the CLSI (45).

P. aeruginosa acute pneumonia mouse model

The P. aeruginosa acute pneumonia was performed as described (6, 44). Antibodies or phosphate-buffered saline was intraperitoneally administered 24 hours before or 1 to 4 hours after infection. Antibiotics were administered subcutaneously 1 or 4 hours after infection. For acute pneumonia organ burden experiments, mice were infected followed by harvesting of lungs, spleens, and kidneys 24 hours after infection for determination of CFU.

P. aeruginosa immunocompromised pneumonia mouse model

Mice were rendered immunocompromised by intraperitoneal delivery of cyclophosphamide monohydrate (CyM) (150 and 100 mg/kg) suspended in 0.9% saline on days −4 and −1, respectively, followed by P aeruginosa intranasal challenge on day 0. Depletion of total white blood cells was confirmed by counting blood cells at day 0 from CyM-treated mice compared to nontreated controls using the Sysmex XT-2000i Automated Hematology Analyzer.

P. aeruginosa thermal injury and bacteremia mouse models

The thermal injury and bacteremia models were performed as described with modifications (6, 44). For thermal injury, 11-week-old non-Swiss albino–1 mice were shaved dorsally and anesthetized with isoflurane before exposure to a custom-made aluminum platform. This platform was designed to induce a 12 to 15% total body surface area thermal injury as determined using the Meeh equation (46). Animals were exposed to a constant temperature of 92°C for 5 s followed by intraperitoneal injection of 0.5 ml of saline for hydration. P. aeruginosa was then injected subcutaneously under the thermal wound. Buprenorphine hydrochloride was administered to mice twice daily for the duration of the experiment to mitigate pain and distress from the thermal injury. Animals were monitored closely for 5 days after infection. In the bacteremia model, 7-week-old BALB/c mice were treated with control IgG or BiS4αPa by intraperitoneal administration 24 hours before intravenous challenge with P. aeruginosa through the tail vein.


Infected mice were euthanized; lungs were removed, inflated, and fixed in 10% neutral buffered formalin (VWR). All lung samples underwent routine histological processing and paraffin embedding; 4-μm histological sections were stained with Gills hematoxylin (Mercedes Medical) and eosin (Surgipath). All stained sections were analyzed using a Nikon 80i microscope with 10× and 40× objectives and reviewed by a pathologist who was blinded to the treatment groups.

Statistical analyses

The log-rank test was used to compare Kaplan-Meier survival curves between different treatment groups generated in GraphPad Prism version 5. ANOVA analysis with Bonferroni correction was used for comparison of antibody-treated groups in lung burden studies. The lung burden CFU in each group was normally distributed after log transformation using D’Agostino-Pearson normality test. The sample variances were also similar in each group. GraphPad Prism version 5.0 software was used for construction of figures and for ANOVA. The AUC for each antibody activity-response curve was calculated using the linear trapezoidal rule on the means at different concentrations in the log scale. AUC calculation and statistical comparisons between different antibodies were performed using PK package in R software.


Table S1. P. aeruginosa strain survey for Psl, PcrV, exoU, and exoS.

Table S2. Determination of MICs.

Fig. S1. Isobologram analysis for synergistic activity.

Fig. S2. BiS4αPa treatment protects mice from lethal pneumonia.

Fig. S3. BiS4αPa reduces P. aeruginosa organ burden.

Fig. S4. BiS4αPa opsonophagocytic killing activity against P. aeruginosa clinical isolates.

Fig. S5. BiS4αPa anti-cytotoxic activity against P. aeruginosa clinical isolates.

Fig. S6. BiS4αPa protection in P. aeruginosa lethal challenge models.


  1. Acknowledgments: We thank J. Lin and L. Shirinian for expressing and purifying antibody constructs. We thank M. Jones, S. Oldham, A. Alfaro, and G. Wilson for excellent technical assistance with animal experiments. In addition, we thank J. Goldberg (Emory University) and V. Fowler (Duke University) for P. aeruginosa clinical isolates. Funding: Research was supported by MedImmune, LLC. Author contributions: A.D., C.K.S., G.J.R., and B.R.S. contributed to experimental design; A.D., P.W., and C.M.G. conducted key in vitro assays and clinical isolate strain surveys for antibody activities; A.D., A.E.K., M.M.C., J.B., and J.H. conducted key multiple in vivo infection model work and independent in vivo replicate experiments; N.D., B.B., Cuihua Gao, R.F., Changshou Gao, and G.J.R. contributed to the design, construction, and production of bispecific antibodies; W.Z. and X.-Q.Y. performed statistical and isobologram analyses; V.D. performed blinded histological analyses; A.D., J.A.S., G.J.R., and C.K.S. prepared the manuscript. Competing interests: This work was funded by MedImmune, LLC, a wholly owned subsidiary of AstraZeneca Pharmaceuticals. All authors were employed by MedImmune, LLC when work was executed and may currently hold AstraZeneca stock or stock options. Patents describing the activity of the lead antibody (BiS4αPa or MEDI3902) in this work have been filed by MedImmune: PCT/US2012/041538, PCT/US2012/063639, PCT/US2012/063722, PCT/US2013/068609, PCT/US2014/037839, and national applications thereof.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article