A High-Affinity Native Human Antibody Disrupts Biofilm from Staphylococcus aureus Bacteria and Potentiates Antibiotic Efficacy in a Mouse Implant Infection Model

Abstract

Many serious bacterial infections are difficult to treat due to biofilm formation, which provides physical protection and induces a sessile phenotype refractory to antibiotic treatment compared to the planktonic state. A key structural component of biofilm is extracellular DNA, which is held in place by secreted bacterial proteins from the DNABII family: integration host factor (IHF) and histone-like (HU) proteins. A native human monoclonal antibody, TRL1068, has been discovered using single B-lymphocyte screening technology. It has low-picomolar affinity against DNABII homologs from important Gram-positive and Gram-negative bacterial pathogens. The disruption of established biofilm was observed in vitro at an antibody concentration of 1.2 μg/ml over 12 h. The effect of TRL1068 in vivo was evaluated in a murine tissue cage infection model in which a biofilm is formed by infection with methicillin-resistant Staphylococcus aureus (MRSA; ATCC 43300). Treatment of the established biofilm by combination therapy of TRL1068 (15 mg/kg of body weight, intraperitoneal [i.p.] administration) with daptomycin (50 mg/kg, i.p.) significantly reduced adherent bacterial count compared to that after daptomycin treatment alone, accompanied by significant reduction in planktonic bacterial numbers. The quantification of TRL1068 in sample matrices showed substantial penetration of TRL1068 from serum into the cage interior. TRL1068 is a clinical candidate for combination treatment with standard-of-care antibiotics to overcome the drug-refractory state associated with biofilm formation, with potential utility for a broad spectrum of difficult-to-treat bacterial infections.

INTRODUCTION

The understanding of bacterial physiology has fundamentally changed since the discovery of biofilms in the bacterial life cycle (1–3). Biofilms provide an anchor and physical protection for bacterial cells and the physiology and genetic programming of the bacteria shift from the planktonic (free-floating) to a sessile (adherent) state. This shift can result in a substantial reduction of antibiotic sensitivity in the biofilm (4). As much as 65 to 80% of clinically significant bacterial infections resistant to antibiotics are associated with biofilm (5, 6), including those of implants and catheters, infective endocarditis, lung infections associated with cystic fibrosis and chronic obstructive pulmonary disease (COPD), persistent infections of the ears and urinary tract, osteomyelitis, and surgery-associated nosocomial infections. Accordingly, a promising approach to treatment is to disrupt biofilms so that the freed bacteria become sensitive to available antibiotics as well as more fully subject to immune control (7).

Biofilms are not simply random assemblies of bacterial and host components. Rather, the polymers in a biofilm form a multinode scaffolding with a semirigid, three-dimensional web-like architecture (8) which serves to exclude host immune cells while allowing the diffusion of nutrients and waste. Comparative genomic studies have identified tens of proteins associated with the adherent state but not the planktonic state (9). Among the bacterial proteins identified as part of the biofilm matrix, DNA binding proteins are of particular interest in light of the considerable amounts of extracellular DNA (eDNA) present in the biofilm (10) and the observation that cell lysis and DNA release are critical for both early biofilm formation and mature biofilm structure in vitro (11, 12). Further, DNase treatment has been shown to disrupt biofilms taken from chronic sinusitis patients (13). The DNABII family includes integration host factor (IHF) and histone-like DNA-binding (HU) proteins and has conserved homologs in a wide variety of bacterial species (14). They share structural features and the key activity of inducing bends in DNA (15).

The use of a polyclonal rabbit serum against IHF has been shown to extract the protein from an established Haemophilus influenzae biofilm in vitro; the loss of this key scaffolding protein resulted in the dissolution of the biofilm (15). Moreover, the antibodies to IHF were active even when separated from the biofilm by a membrane (16), presumably by shifting the IHF binding equilibrium in the biofilm from the DNA-bound state toward the free state. This suggests that high affinity is a desirable property for a therapeutic antibody.

To generate a clinically useful therapeutic monoclonal antibody (MAb), we used our previously described single B-lymphocyte screening platform (17, 18) to clone high-affinity native human MAbs that bind to DNABII homologs from diverse bacterial species. In this paper, we describe the biochemical and biological properties of one of these MAbs, TRL1068. We show that TRL1068 disrupts established biofilm and restores antibiotic sensitivity to the bacteria released from the biofilm.

MATERIALS AND METHODS

IHF and HU proteins.

Integration host factor (IHF) and histone-like DNA-binding proteins (HU) from Staphylococcus aureus, Pseudomonas aeruginosa, Klebsiella pneumoniae, Haemophilus influenzae, and Acinetobacter baumannii were produced by transient transfection in HEK 293 Freestyle cells (Thermo Fisher Scientific, Waltham, MA). Genes encoding the IHF/HU proteins (GenBank accession numbers are listed in Table 1) were synthesized by GeneArt (Thermo Fisher Scientific, Waltham, MA) and cloned as 6× His-tagged fusion proteins into the pTT5 expression vector (licensed from Canadian National Research Council). HEK 293 cells were transfected using linear polyethyleneimine (PEI; Polysciences, Warrington, PA) (19), and purification from the supernatant was performed using His60 beads (Clontech Laboratories, Mountain View, CA) according to the manufacturer's recommendations. For animal studies, large-scale production of TRL1068 was performed by Blue Sky BioServices (Worcester, MA).

TABLE 1.

Estimated affinities from ELISA binding curves of TRL1068 to IHF alpha and HU protein and peptide homologs from diverse bacterial species

Target strain (GenBank accession no.)	Gram stain result	Estimated affinity (pM)
		Protein	Peptide
Staphylococcus aureus (WP_033859538)	+	15	49
Streptococcus pneumoniae (WP_001284638)	+	Not done	46
Pseudomonas aeruginosa (WP_031638906)	−	11	39
Acinetobacter baumannii (WP_032027224)	−	10	42
Klebsiella pneumoniae (WP_004143152)	−	11	35
Borrelia burgdorferi (WP_002657662)	−	Not done	39
Haemophilus influenzae (WP_005657421)	−	5,400	470

IHF and HU peptides.

For peptide mapping, alanine scanning, and specificity determination, different sets of peptides were synthesized by Mimotopes Pty. Ltd. (Victoria, Australia) and supplied as lyophilized powders. All of the peptides were capped at the N terminus with biotin-SGSG and amidated at the C terminus, with the exception of the C-terminal peptide (peptide mapping set), which had a free carboxylic acid. All of the peptides were resuspended in dimethyl sulfoxide (DMSO) prior to assay.

Single B-lymphocyte MAb discovery technology.

Leukopaks were obtained from a total of 11 anonymized donors under informed consent approved by Stanford's Institutional Review Board (Stanford Blood Center, Stanford, CA). Peripheral blood mononuclear cells (PBMCs) were prepared by standard methods, and individual memory B cells were assayed following stimulation to proliferate and differentiate into plasma cells (17). A portion of the culture was allowed to secrete IgG, and the footprints were screened at the single-cell level using a multiparameter assay that allows the concurrent measurement of binding to different IHF/HU proteins conjugated to distinguishable fluorescent beads using sodium cyanoborohydride (20). The lack of binding to beads coated with bovine serum albumin (BSA) was used as a specificity counterscreen.

After identifying a human B-cell secreting a MAb meeting the selection criteria, the encoding mRNAs for heavy and light chains were amplified by single-cell reverse transcription-PCR (RT-PCR) from sibling cells and subcloned into the previously described pTT5 vector. Following the same protocol as that for the expression of the IHF/HU proteins, recombinant antibodies were produced in HEK 293 Freestyle cells (Thermo Fisher Scientific, Waltham, MA) by transient transfection and purified using protein A (MAb Select Sure; GE Healthcare).

ELISA for full-length IHF and HU proteins.

Binding was evaluated against purified IHF/HU proteins from S. aureus, P. aeruginosa, K. pneumoniae, H. influenzae, and A. baumannii in an enzyme-linked immunosorbent assay (ELISA) at pH 7.4. Additionally, the specificity of TRL1068 binding was evaluated against a panel of two unrelated bacterial proteins, one viral protein, and seven mammalian proteins with different tags. The IHF/HU or other proteins were diluted in phosphate-buffered saline (PBS) and passively adsorbed on a Microlon high binding plate (Grenier/Thermo Fisher Scientific, Waltham, MA) overnight at 4°C. TRL1068 was diluted in blocking buffer (PBS–3% bovine serum albumin [BSA]) and added to the plate in serial dilutions. Horseradish peroxidase (HRP)-conjugated anti-human IgG was used as a detection antibody in conjunction with TMB (tetramethylbenzidine) substrate, recording the optical density at 450 nm. Affinity calculations were performed using Prism (GraphPad Software, La Jolla, CA). The binding of TRL1068 to IHF/HU proteins from S. aureus and P. aeruginosa also was evaluated at pH ranging from 2.5 to 8.5.

ELISA for peptide mapping, alanine scanning, and specificity determination.

Biotinylated peptides were screened by an ELISA similar to that for the full-length proteins. Pierce streptavidin-coated high-sensitivity plates (Thermo Fisher Scientific, Waltham, MA) were used according to the manufacturer's recommendations. Peptides were diluted in DMSO (15 to 25 mg/ml for epitope mapping, 10 mM for alanine scanning, and 5 mM for specificity assessment) and further diluted 1/1,000 in 0.1% BSA (single-point determination for epitope mapping, serial dilutions for others). TRL1068 was diluted in blocking buffer, and the procedure was identical to that for the ELISA of full-length proteins.

Thermal stability assay.

The melting temperature of TRL1068 was determined by diluting TRL1068 in duplicate to 2 μg/ml into PBS containing 1× SYPRO Orange (Thermo Fisher Scientific, Waltham, MA) in a final volume of 20 μl. The samples were heated in a StepOne Plus instrument (Thermo Fisher Scientific, Waltham, MA) for 2 min at 25°C, followed by a 1.58°C/minute ramp rate to 99°C and then 2 min at 99°C. Fluorescence intensity was monitored over the course of the incubation using the filters for Rox dye. Melting temperatures were determined by curve fitting of the fluorescence data in Microsoft Excel (Redmond, WA) and are an average of two replicate measurements for each assay with the assay repeated in three independent experiments.

ELISA for quantification of TRL1068.

TRL1068 was quantified by ELISA following capture from the sample matrix (serum or cage fluid) via binding to immobilized affinity-purified goat anti-human Fcγ-specific antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). The capture of TRL1068 was detected using an affinity-purified goat anti-human kappa light-chain peroxidase-conjugated antibody (Sigma-Aldrich Corp., St. Louis, MO). Optical density readings were used to calculate TRL1068 concentrations from a 4-parameter logistic regression fit curve for the standards using SoftMax Pro (ver. 5.4.5; Molecular Devices, Sunnyvale, CA).

In vitro biofilm-disrupting assay and SEM.

The biofilm-disrupting activity of TRL1068 was evaluated using the minimum biofilm eradication concentration (MBEC) assay performed by Innovotech, Inc. (Edmonton, Alberta, Canada). The MBEC device consists of a plastic lid with 96 conical plastic pegs arranged to fit into a 96-well microplate, with multiple parameters optimized to grow a reproducible biofilm, including plastic composition, rate of motion of the inoculated reactor, incubation temperature and time, inoculum size, atmospheric gas composition, composition of the growth medium, and frequency of medium exchange. Innovotech has successfully cultured over 65 different bacterial species in this manner in the past (21). The S. aureus (ATCC 29213) and P. aeruginosa (ATCC 27853) strains for the present study had been optimized previously by Innovotech for this assay. A first subculture from a cryogenic stock was streaked onto tryptic soy agar plates and incubated at 37°C ± 1°C for 24 h. From the first subculture, a second subculture again was streaked onto tryptic soy agar plates and incubated at 37°C ± 1°C for 24 h. The second subculture then was used within 48 h to prepare an inoculum that matched a 0.5 McFarland standard (1.5 × 10⁸ cells per ml) in 3 ml sterile water in a glass test tube. The inoculum was diluted 1:250 in tryptic soy broth adjusted to pH 7.2 for biofilm growth and inverted 3 to 5 times to achieve uniform mixing of the organism. The cell density was confirmed by serially diluting and spot plating triplicate samples onto tryptic soy agar plates. From this standardized culture, 150 μl was transferred into each of the wells of the 96-well base plate of the MBEC device. The lid with the conical pegs was placed onto the base plate containing the specific organism(s). The device then was placed on an orbital shaker set at 110 rpm and incubated at 37°C ± 1°C for 24 h for biofilm growth. The lid then was transferred to a sterile 96-well microtiter rinse plate (200 μl per well of 0.9% sterile saline) for 1 to 2 min. The conical pegs were transferred to a challenge plate and then treated with vehicle, IgG1 isotype control MAb (1.2 μg/ml), or TRL1068 (1.2 μg/ml) for 12 h. Adherent bacteria were visualized by scanning electron microscopy (SEM). For visualization, the pegs were removed from the lid and fixed (5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.5) at 4°C for 24 h and then air dried for 96 h. The samples were attached to labeled aluminum stubs and sputter coated with a conductive layer (Platinum-Gold) before imaging in a Hitachi S3700N. The SEM images ranged from magnifications of ×30 to ×5,000 (representative ×500 and ×2,500 magnification images are presented here).

In vivo mouse infectious implant model.

TRL1068 biofilm-disrupting activity on methicillin-resistant S. aureus ATCC 43300 was evaluated using a murine tissue cage infection model as previously described (22, 23). The in vivo studies were performed at the University Hospital of Basel, Basel, Switzerland, and were conducted according to the regulations of Swiss veterinary law. Sterile cylindrical Teflon-coated tissue cages (8.5 by 1 by 30 mm; volume, 1.9 ml) with 130 regularly spaced holes (Angst+Pfister AG, Zurich, Switzerland) were subcutaneously implanted into the backs of 12- to 15-week-old female C57BL/6 mice. Two weeks later, the sterility of the cages was assessed by aspirating tissue cage fluid (TCF) from the lumen of the tissue cages. The sterile cages then were infected by injection of 700 CFU of MRSA into the lumen of the tissue cage (day 0). To prepare the inoculum, one bead of the cryovial bead preservation system (Microbank; Pro-Lab Diagnostics, Richmond Hill, Ontario, Canada) was incubated at 37°C without shaking in Trypticase soy broth (TSB) (Becton Dickinson and Company, Allschwil, Switzerland). After overnight incubation, the bacteria were washed twice with 0.9% saline (Bichsel, Interlaken, Switzerland) and diluted to provide the needed inoculum. The infection was confirmed by TCF aspiration 1 day after infection. The planktonic bacterial load was assessed by plating TCF in appropriate dilutions. All treatments were started at 24 h postinoculation (day 1).

Four in vivo studies were performed that varied in frequency of treatment. The following treatments were used: (i) TRL1068 and/or daptomycin (DAP) were administered once (single-dose study), followed by the termination of mice on day 3; (ii) TRL1068 and/or DAP were administered daily for 3 days (3-day treatment study), followed by termination of mice on day 4; (iii) TRL1068 or IgG1 isotype control antibody was administered twice (days 1 and 3) and DAP was administered daily for 5 days (5-day treatment study), with termination of mice on day 6; and (iv) TRL1068 was administered on day −1 prior to infection, with the study format otherwise the same as that for treatment iii. In all studies, either saline or an IgG1 isotype control antibody was used as a negative control, DAP alone was the active control (50 mg/kg of body weight; Cubicin, Novartis Pharma Schweiz AG, Switzerland), and the treatment arm was comprised of DAP (50 mg/kg) plus TRL1068 (15 mg/kg). All treatments were administered intraperitoneally (i.p.), except for the single-dose study for which an additional treatment group received 190 μg TRL1068 injected directly into the cage. TCF samples were taken at prespecified time points to quantify bacterial burden (CFU per milliliter) and to quantify the concentration of TRL1068 in the TCF. All mice were euthanized at the end of the treatment period and the cages explanted; mouse serum also was collected. The explanted cages were washed twice with PBS and sonicated in saline at 130 W for 3 min to release residual adherent bacteria, which were quantified as CFU/milliliter. Concentrations of the IgG isotype control and TRL1068 in the TCF samples and in mouse serum were determined by ELISA as described above.

For statistical analysis, CFU/milliliter values were transformed to log₁₀ CFU/milliliter. Treatment effect was evaluated as the change in log₁₀ CFU/milliliter from day 1 and by log₁₀ CFU/milliliter for cage-adherent residual bacteria. One-way analysis of variance (ANOVA) followed by Tukey's HSD (honest significant difference) post hoc test was used to determine statistical significance of differences among the treatment groups. Statistical analysis was performed using IBM SPSS Statistics, version 20 (IBM, Armonk, NY).

RESULTS

TRL1068 discovery.

We used our previously described MAb discovery platform that utilizes digital microscopy to examine the secreted IgG footprint (∼100 fg per cell over a 5-h period) from each of millions of individual memory B cells (17). Following stimulation with a mixture of mitogens and cytokines, ∼70% of resting memory B cells begin a course of proliferation and antibody secretion that lasts 10 days. The miniaturization to the single-cell level allows two rounds of assay of the secreted antibody within this window of viability. In the first round, pools of ∼200 memory B cells in each well of a 96-well microplate are surveyed. In the second round, the contents of positive wells are distributed at high dilution and reassayed. The miniaturization allows rare clones to be identified by providing high throughput. Using multiple selection criteria in the primary assay typically reduces the hit rate to <50 clones per million B cells. Following the B-cell dilution step, the mRNA encoding the antibody of interest is recovered by single-cell cDNA cloning.

From sequence analysis of clinically relevant bacterial species representing broad sequence diversity across the published sequences, five IHF/HU homologs were chosen as representatives of the natural diversity: S. aureus, P. aeruginosa, A. baumannii, K. pneumoniae, and H. influenzae. As it was difficult to express the proteins in bacteria, a mammalian cell expression system was used to produce enough protein for the MAb discovery and characterization studies.

Three of the recombinant IHF/HU proteins (S. aureus, K. pneumoniae, and H. influenzae) were conjugated to distinguishable fluorescent beads and used in a multiplexing format as probes of binding specificity. A fourth bead type was coated with BSA as a specificity counterscreen. The frequency of memory B cells secreting a MAb meeting all of the screening criteria was <4 per 100,000 across 11 blood donors surveyed. A total of 9.6 million B cells were interrogated. The V domains of candidate MAbs were cloned by single-cell RT-PCR into an IgG1 expression vector and expressed transiently in HEK 293 cells. Twenty MAbs from 9 different donors were confirmed as positive for binding to the IHF/HU antigens. Based on affinity and breadth of cross-reactivity to IHF/HU homologs, TRL1068 was chosen for further characterization.

TRL1068 specificity and stability.

To study the breadth of specificity, TRL1068 binding was measured by ELISA against the set of five DNABII proteins described above and against a panel of unrelated proteins. High affinity (in the low-picomolar range) to many IHF/HU full-length proteins was observed by ELISA (Fig. 1 and Table 1). A rough estimate of the epitope was determined by measuring binding to a set of 26 linear peptides comprising overlapping 15-mers spanning the entire S. aureus HU protein, with offsets of 3 residues (Fig. 2A). Peptides 19 and 20 of this set showed strong binding to TRL1068 across the sequence AARKGRNPQTGKEIDIPA. Figure 2E shows a model from the crystal structure of S. aureus HU bound to DNA (24) with the sequences of peptides 19 and 20 shown in blue on each chain of the homodimer (green) wrapped around the DNA. The epitope is contained within an anti-parallel beta-sheet capped with a beta-turn (Fig. 2D).

FIG 1.

ELISA binding of TRL1068 to IHF alpha/HU homologs from diverse bacterial species. Two unrelated bacterial, one viral, and seven mammalian proteins (including BSA) with different tags also were tested for binding to TRL1068. No binding was detected up to 6 nM for the 10 unrelated proteins.

FIG 2.

Epitope mapping. (A) Full-length S. aureus HU sequence. A series of overlapping 15-mers, offset by 3, were assessed for binding by TRL1068 by semiquantitative ELISA, with the positive peptides denoted in red. (B) High conservation of the epitope in Gram-positive and Gram-negative bacterial species. Yellow denotes full identity across all species, blue denotes partial identity, and green denotes conservative substitutions. (C) Alanine substitution for each amino acid in turn (parental sequence shown below peptide number) was tested for binding to TRL1068 in a semiquantitative ELISA. Results are shown as fold change over the level of the parent peptide. (D) Key residues (in red) from the peptide binding experiments. Dotted lines indicate hydrogen bonds in the crystal structure of HU from S. aureus bound to DNA (24). (E) Location of the epitope mapped on the crystal structure. The HU dimer is shown in green with the peptide 19-20 region in blue.

To better identify the specific residues which are necessary for antibody binding in the sequence common to peptides 19 and 20, an alanine scan was performed on the peptide KGRNPQTGKEID. Each amino acid in turn was replaced with an alanine for a total of 13 peptides, including the parent sequence. Reduced binding from the replacement of any amino acid with the small uncharged alanine was interpreted as evidence of important antibody-antigen interactions. Figure 2C shows the alanine scan ELISA results as the fold change over the level for the parent peptide. Lower values indicate the loss of binding when that amino acid is replaced with alanine. All but two amino acids contribute to binding with the key residues highlighted in red in Fig. 2D. Not all of these necessarily make contact with the antibody. For example, in the case of glycine, which has no side chain, reduced binding following replacement with the larger alanine suggests a tight fit with the antibody. Similarly, proline is essential for the beta turn, and replacement with an alanine could affect the structure of the epitope, leading to loss of binding. Based on the peptide-mapping and alanine-scanning assays, the binding epitope for TRL1068 was identified as the 11-mer GRNPQTGKEID. This sequence is highly conserved within the HU (subunits alpha and beta) and IHF (subunits alpha and beta) proteins of many Gram-positive and Gram-negative pathogens.

Based on the strong binding of peptide 19 from the first peptide series, a third set of peptides was designed to assess specificity against a broader set of DNABII proteins. Using the aligned protein sequences of alpha IHF and HU from various organisms, the peptide corresponding to peptide 19 (AARKGRNPQTGKEID) was synthesized for each of the following species, including both Gram-positive and Gram-negative organisms: S. aureus, P. aeruginosa, K. pneumoniae, H. influenzae, A. baumannii, S. pneumoniae, and B. burgdorferi. Figure 2B shows the alignments of these sequences. Binding of TRL1068 to the peptides was tested by ELISA. As shown in Table 1, the affinity (midpoint of ELISA binding curve) was similar for the whole protein (when available) and the corresponding peptide. Thus, the binding affinity to peptides for which the whole protein was not available likely provides a reasonable estimate of cross-reactivity. TRL1068 bound to most examples of both Gram-positive and Gram-negative species.

A BLAST search (http://www.ncbi.nlm.nih.gov/GenBank/) was done to further determine the breadth of distribution of the binding epitope for TRL1068. The input peptide sequences included those from K. pneumoniae (NQRPGRNPKTGEDIP), P. aeruginosa (RQRPGRNPKTGEEIP), A. baumannii (RERPGRNPKTGEEIP), S. aureus (AARKGRNPQTGKEID), S. pneumoniae (AERKGRNPQTGKEIT), H. influenzae (SSRPGRNPKTGDVVP), and B. burgdorferi (KGRLNARNPQTGEYVK). The BLAST search showed that variants of the binding epitope for TRL1068 are found in the HU/IHF proteins across hundreds of bacterial species within the classes Betaproteobacteria and Gammaproteobacteria of the Proteobacteria phylum (Gram negative), classes Coccus and Bacilli of the Firmicutes phylum (Gram positive), and class Spirochaetes of the Spirochaetes phylum. The peptide sequences for H. influenzae, S. aureus, S. pneumoniae, and B. burgdorferi were found to be unique to their respective bacterial genera. However, the peptide sequences for K. pneumoniae, P. aeruginosa, and A. baumannii could be found in bacterial species within all orders and families of the Betaproteobacteria and Gammaproteobacteria classes. None of the peptide sequences showed homology with any human protein sequence.

Additional peptides (15-mers) corresponding to variants of the TRL1068 epitope were screened by semiquantitative ELISA to determine the breadth of TRL1068 binding. The peptides included variants from HU alpha (K. pneumoniae, H. influenzae, A. baumannii, E. aerogenes, N. gonorrhoeae, P. aeruginosa, and Y. pestis), HU beta (P. aeruginosa, B. pseudomallei, E. faecium, E. faecali, N. gonorrhoeae, and Y. pestis), IHF alpha (B. pseudomallei and N. gonorrhoeae), and IHF beta (K. pneumoniae, E. aerogenes, P. aeruginosa, A. baumannii, H. influenzae, and N. gonorrhoeae). TRL1068 bound to nearly all of the peptides screened, with the exception of the HU alpha subunit peptide for P. aeruginosa and IHF beta subunit for K. pneumoniae and E. aerogenes. Changes in pH at the site of a bacterial infection may affect the binding and stability of antibodies. Therefore, the pH dependence of TRL1068 binding to the S. aureus and P. aeruginosa proteins was measured from pH 2.5 to 8.5. There was negligible change in binding at pH 4.5 to 8.5 (data not shown).

Thermal stability of TRL1068 was assessed by the method of Niesen et al. (25), whereby a modified quantitative RT-PCR thermocycler is used to monitor the thermal unfolding of a protein in the presence of the fluorescent dye (SYPRO Orange) that preferentially binds to the exposed hydrophobic regions of the protein as it melts. The melting temperature for TRL1068 was 67.3°C ± 0.4°C, as derived from averages of two measurements per experiment repeated after 3 days (data not shown).

Biofilm disruption in vitro.

The MBEC assay is a high-throughput screening assay used to determine the efficacy of therapeutic agents against biofilms (4, 21). Conical plastic pegs are arrayed in a 96-well format, allowing serial dilutions and replicates to be processed in parallel for several experimental conditions. Pegs were treated with vehicle, IgG1 isotype control antibody at 1.2 μg/ml, or TRL1068 at 1.2 μg/ml for 12 h, and adherent bacteria were visualized by SEM. There were marked reductions in biofilm observed for both S. aureus and P. aeruginosa. The biofilms on the untreated control pegs tended to be substantially thicker and more extensive than the TRL1068-treated pegs. Results similar to those for the untreated pegs were obtained with an isotype-matched control antibody. Representative images are shown in Fig. 3. No further reduction in the biofilm was achieved when TRL1068 was provided at 6 μg/ml; efficacy became more variable and generally was lower at 0.24 μg/ml. The mechanism of action of TRL1068 involves structural alterations in the biofilm which likely have a cooperativity component that accounts for the weak dose-response relationship.

FIG 3.

In vitro biofilm disruption. Bacterial biofilms were formed on conical plastic pegs in 96-well format (MBEC assay by Innovotech, Inc.). Pegs were treated with vehicle (left) or TRL1068 (1.2 μg/ml; right) for 12 h. Adherent bacteria were visualized by scanning electron microscopy. Marked reductions in biofilm were observed for both S. aureus and P. aeruginosa.

Mouse infectious implant model.

The effect of TRL1068 in combination with daptomycin (DAP) on established methicillin-resistant S. aureus (MRSA) biofilms was tested in a previously described mouse infectious implant model (23). MRSA ATCC 43300 bacteria were injected into the lumen of the implanted tissue cage on day 0, resulting in the formation of a mature bacterial biofilm coating the interior surface within 24 h based on prior studies. At various times pre- or postinfection, mice were treated with (i) saline, (ii) DAP alone (50 mg/kg, i.p.), (iii) DAP (50 mg/kg, i.p.) plus TRL1068 (15 mg/kg, i.p.), (iv) DAP (50 mg/kg, i.p.) plus TRL1068 (190 μg injected directly into the cage), or (v) IgG1 isotype control MAb (15 mg/kg, i.p.). The DAP dose was chosen to obtain an area under the concentration-time curve for 0 to 24 h (AUC_0–24) similar to that in humans after a dose of 6 mg/kg (23, 26). Samples were taken from the tissue cage fluid (TCF) on various days and from serum at sacrifice to measure planktonic bacteria and antibody concentrations. The level of adherent bacteria within the cage, explanted at the end of the experiment, also was quantified.

TRL1068 levels.

In the first study, a single dose of TRL1068 was given; the mean concentration of the MAb in the TCF on day 2 postinfection (24 h after MAb administration) was higher in the group for which the MAb was directly injected than for the group for which it was delivered i.p. (41.8 ± 18.2 and 20.6 ± 9.7 μg/ml, respectively). However, by day 3 postinfection the TCF levels for the two groups were similar (19.5 ± 3.0 μg/ml direct injection and 21.2 ± 9.1 μg/ml i.p.). The serum levels also were similar at sacrifice on day 3 (15.4 ± 6.4 μg/ml direct injection and 20.1 ± 9.7 μg/ml i.p.). Based on the substantial equilibration of TRL1068 between serum and the cage interior, subsequent treatment with TRL1068 was by the i.p. route only.

In the 3-day treatment study (MAb given on days 1, 2, and 3 postinfection), the mean (± standard deviations) TRL1068 concentration in the TCF rose from 14.2 ± 15.2 μg/ml on day 2 to 84.5 ± 40.1 μg/ml on day 4. The level in serum at sacrifice was 156.3 ± 23.0 μg/ml.

In the 5-day treatment study (MAb given on days 1 and 3 postinfection), the three treatment groups consisted of an IgG1 isotype control in place of the saline previously used, thereby providing a control for possible nonspecific IgG effects, DAP alone, and DAP plus TRL1068. DAP was dosed daily for 5 days. This schedule reflects the faster clearance time for DAP compared to that for the MAb, for which serum levels stayed substantially above the concentration that was effective in vitro for disrupting established biofilms. TRL1068 mean concentration in the TCF was similar on both sampling days, 41 ± 11 μg/ml on day 4 and 45 ± 20 μg/ml on day 6, and the serum concentration was 55 ± 31 μg/ml on day 6. The isotype control concentration was determined to be 15 ± 5, 8 ± 3, and 27 ± 12 μg/ml on days 4 and 6 and in serum, respectively.

Single-dose study.

On days 2 and 3, TCF was aspirated and CFU of planktonic bacteria were determined. On day 2 postinfection, following a single dose of DAP plus TRL1086 on day 1, planktonic bacteria in the cage were reduced by more than 2 logs for the two DAP-plus-TRL1068 treatment groups compared with those for the saline group (Fig. 4A). More importantly, the addition of TRL1068 to DAP was more effective than DAP alone (which showed only a 1-log reduction). The two TRL1068 groups were not significantly different from each other (P = 0.8), and in the combined treatment groups, the planktonic bacterial count was lowered by 1 log versus that with DAP treatment alone (P < 0.01).

FIG 4.

In vivo efficacy in a murine tissue cage infection model. Four in vivo studies were conducted with different durations of treatment. (A and B) In the one-day treatment study, saline was used as a negative control. Two routes of administration were used for TRL1068: direct injection into the tissue cage and i.p.; as there were no significant differences between the two, the pooled data from both are plotted in blue. (A) 1-Log reduction after a single dose of DAP plus TRL1068 on day 2 (left) (P < 0.01) but rebound on day 3 (right), attributed to insufficient DAP to eradicate the continuously released bacteria. (B) 1-Log reduction (P < 0.02) in adherent bacteria inside the cage released by sonication after the explantation of the cage. (C and D) In the three-day treatment study, saline was used as a negative control. Panels C and D show a greater reduction in both planktonic and adherent bacteria than in the 1-day study; however, the results did not reach statistical significance. (E and F) In the five-day treatment study, an IgG1 isotype control antibody was used as a negative control (indistinguishable from saline controls used previously). (E) 1.5-Log reduction (P < 0.01) and 2-log reduction (P < 0.001) in planktonic bacteria observed for day 4 and day 6, respectively. (F) 3-Log reduction in adherent bacteria on day 6 (P < 0.01). The 5-day treatment study was repeated but with the additional administration of TRL1068 24 h prior to infection. At day 6, both planktonic (G) and adherent (H) bacteria within the cage were eradicated.

On day 3, the DAP-only group showed growth, while both TRL1068 groups again showed a reduction in the bacterial growth. The rebound of bacterial growth on day 3 was attributed to rapid clearance of DAP in the mouse. In the cage, DAP concentration likely was reduced to below the minimum bactericidal concentration within 24 h after a single i.p. injection (23). Without continuous resupply with DAP, bacteria released from the biofilm at later times would not be killed.

Figure 4B shows the adherent bacterial CFU counts from the cages released by sonication and represents the residual bacteria embedded in the biofilm. The DAP and saline groups were not significantly different from each other, but the combined TRL1068 treatment groups significantly reduced adherent bacteria by 1 log (P < 0.02).

Multiple-dose studies.

The rebound of bacteria on day 3 in the single-dose study suggested that longer treatment durations are beneficial. In the next study we evaluated the effects of TRL1068 in combination with DAP with daily i.p. dosing of the two drugs for 3 days. Otherwise, the study design was analogous to that of single-dose studies, with TCF samples taken for analysis at days 2 and 4 and the explanted cage analyzed for adherent bacteria at day 4.

Compared to treatment with DAP alone, the administration of DAP plus TRL1068 reduced planktonic bacterial CFU in the TCF on all assessment days (>2-log reduction on day 4) and reduced residual adherent bacteria on the cages at sacrifice (∼2-log reduction), as shown in Fig. 4C and D. Compared with the previous 1-day treatment study, which showed high statistical significance, greater variability in the initial infection rates was observed in this experiment, which may be why the differences from the DAP-only group did not reach significance (Fig. 5A). This variability was in spite of the confirmation of the initial inoculum by CFU plating (all mice received the same amount of bacteria).

FIG 5.

Meta-analysis of planktonic bacterial reduction in the tissue cage fluid. (A) Dispersion, in CFU/ml, on day 1 prior to treatment for three separate experiments. (B) Consolidation of day 4 data from the 3-day and 5-day treatment studies to yield a larger number of mice per group, compensating for the higher variability in the 3-day study. This meta-analysis shows highly significant reduction in planktonic bacteria with TRL1068 plus DAP versus DAP alone.

In the next experiment, DAP was dosed for 5 days (days 1 to 5), and TRL1068 was dosed on days 1 and 3 postinfection. In this study, the initial inoculum rates had low variability, as in the first experiment (Fig. 5A), and high statistical significance was achieved (Fig. 4E and F). On days 4 and 6, DAP plus TRL1068 versus DAP alone showed significant reductions in planktonic bacteria in the TCF of >1.5 log (P = 0.008) and >2 log (P < 0.001), respectively. On day 6, the explanted cages were analyzed for residual adherent bacteria. DAP plus TRL1068 versus DAP alone showed a 3-log reduction in adherent bacteria (P = 0.002). In two of the four mice treated with the combination of DAP and TRL1068, there were no detectable adherent bacteria. This study also included an IgG1 isotype control group, which had negligible efficacy (no difference from the saline control was seen; P = 0.8). DAP plus TRL1068 versus the isotype control showed significant reductions in both planktonic and adherent bacteria (P < 0.001).

Pretreatment.

Since the adherent bacteria were eradicated completely in two of the four mice from the 5-day treatment study, we examined the efficacy of including a pretreatment dose at day −1 (24 h prior to infection) with the rest of the experimental conditions being the same, including similar consistency in initial infection burden. The reduction of both planktonic and adherent bacteria in this experiment was complete at day 6 (Fig. 4G and H).

Day 4 meta-analysis.

Unusual variability in the initial bacterial CFU (Fig. 5A) was observed in the 3-day treatment experiment, but no clear cause has been identified. The TRL1068 levels in the cage for both the 3-day and 5-day treatment studies were far above the levels associated with efficacy in the in vitro biofilm study. For that reason, the TRL1068 day 4 postinfection changes in log₁₀ CFU/milliliter from the second and third experiments were pooled for a meta-analysis resulting in n = 7 for the saline/isotype control group, n = 9 for the DAP alone group, and n = 9 for the DAP plus TRL1068 group. For this combined data set (Fig. 5B), the reduction in planktonic bacteria on day 4 for DAP plus TRL1068 showed a 1.5-log reduction in planktonic bacteria in the TCF compared to that of DAP alone (P = 0.025).

DISCUSSION

Biofilm is a major contributor to antibiotic treatment failures in a variety of clinical settings (5, 6). In particular, the number of implanted medical devices is steadily increasing. Infections of these devices are a significant problem, however, and are caused primarily by biofilm-forming staphylococci that are difficult to treat due to the decreased susceptibility to both antibiotics and host defense mechanisms (27). The rate of infection ranges from 2% to 40% depending on the type of surgical implant (28). These infections can occur perioperatively, by direct bacterial contamination during surgery or wound healing, or via hematogenous lymphatic infection (29). More than half are caused by staphylococci, and the rise of methicillin-resistant S. aureus (MRSA) has created additional therapeutic challenges.

Thus, preventing or disrupting biofilms represents an important clinical goal. Although polyclonal rabbit serum raised against IHF from E. coli had been characterized in this regard (14), it was not clear whether a single monoclonal antibody could achieve equivalent activity. As reported here, the single MAb TRL1068 displays low-picomolar affinity to IHF/HU homologs from diverse bacterial species, including both Gram-positive and Gram-negative species. TRL1068 has additional favorable properties: high affinity maintained at pH ranging from 4 to 8.5, substantial equilibration into the cage, suggesting it will have good tissue penetration, and high melting temperature, which generally correlates with ease of expression at commercially relevant levels.

TRL1068 was shown to bind to a broad spectrum of naturally occurring peptide variants of the TRL1068 binding epitope found on the HU alpha and beta and IHF alpha and beta subunits from both Gram-positive and Gram-negative bacteria. The TRL1068 binding epitope was mapped to a short stretch of amino acids (GRNPQTGKEID) found within the DNA binding domain of IHF/HU proteins. A BLAST search showed that the binding epitope is found in hundreds of bacterial species, both pathogenic and nonpathogenic, within the classes Betaproteobacteria and Gammaproteobacteria of the Proteobacteria phylum (Gram-negative), classes Coccus and Bacilli of the Firmicutes phylum (Gram-positive), and class Spirochaetes of the Spirochaetes phylum. According to the crystal structure of the S. aureus HU protein, the epitope is part of a beta-sheet beta-turn motif, defining a significant conformational character for the actual interaction between the MAb and antigen in its most functionally relevant region, namely, the site of interaction with DNA. These results suggest broad therapeutic potential for TRL1068.

TRL1068 degraded established biofilm in vitro over the course of 12 h at a concentration of 1.2 μg/ml using the 96-well microplate format MBEC assay, which provided multiple replicates. Based on a prior TRL1068 pharmacokinetic study in rats (unpublished data), a dose of 15 mg/kg was selected for the mouse study in order to maintain serum levels substantially above 1.2 μg/ml for at least 7 days. The directly measured mouse in vivo studies showed that the serum levels of TRL1068 averaged >30 μg/ml during the study period.

To understand the pathogenesis and treatment tolerance of implant-associated infection, animal models that closely resemble human disease are needed. The murine tissue cage infection model, first established in 1982 by Zimmerli et al. (30) and modified by Kristian et al. (31), has proven to be a suitable tool to study the antiadhesive and anti-infective efficacy of different biomaterial coatings and to assess the pharmacokinetics, efficacy, and cytotoxicity of antimicrobial compounds (22). The model is particularly useful for evaluating the activity of TRL1068, since it distinguishes between planktonic and adherent bacteria. The effect of TRL1068 in disrupting biofilm can be directly observed as a reduction of adherent bacteria. The effect on planktonic bacteria is a secondary effect of the mechanism of action as the released bacteria regain sensitivity to antibiotic treatment (22). The model involves sterile implantation of a perforated Teflon cage subcutaneously in a mouse. After healing, bacteria are injected directly into the lumen of the cage, where they form a localized biofilm. In this model, the tissue cage fluid can be aspirated repeatedly without the need to sacrifice the animal (22).

Using this model, we first tested TRL1068 in combination with daptomycin (DAP) as a single dose given 24 h after the infection. Reduction in the bacterial burden in the tissue cage fluid was observed initially, but the bacteria began to grow back after a day. This rebound growth was attributed to the DAP concentration falling below the minimum bactericidal concentration within 24 h after a single i.p. injection (23). Without redosing, there is insufficient DAP present to eradicate the continuously released bacteria as they became planktonic following biofilm dissipation. Moreover, there is a lag period of up to 24 h for the bacteria to revert to the planktonic state.

Accordingly, we conducted two additional experiments in which DAP was given daily for 3 or 5 days. The day 4 cohorts in the two experiments were treated identically, and a meta-analysis of the combined planktonic bacterial levels (n = 7 to 9 mice/group) provided statistically significant results despite variability in the level of planktonic bacteria at the initiation of treatment. In all of the experiments, either a non-IHF binding isotype-matched MAb or saline was used as a control (growth control). There was no observed difference between the use of saline and the isotype control. Compared to the growth control, DAP alone was only moderately effective, reflecting the biofilm-associated drug refractory state. When combined with TRL1068, the efficacy of DAP was greatly potentiated (P < 0.01). This result was consistent across all time points for which smaller numbers of mice were tested.

At study termination, the explanted cages were rinsed to remove planktonic bacteria and adherent bacteria were dislodged by sonication. Again, the DAP-alone groups showed a moderate reduction in the biofilm-embedded bacteria compared to that of the growth controls. The combination of DAP plus TRL1068 showed significantly lower levels of adherent bacteria (P < 0.001). Finally, administering a dose of TRL1068 at 24 h prior to infection, followed by the same 5-day treatment regimen as before, resulted in the complete eradication of both the planktonic and adherent bacteria at day 6 (P < 0.05).

The properties described here support the development of TRL1068 for the treatment of implant-associated infections. The breadth of binding to homologs of the target protein across extensive phylogenetic distances, combined with in vitro data showing the disruption of biofilms of both S. aureus and P. aeruginosa, suggest that TRL1068 will have utility for a broad spectrum of biofilm-associated infections in combination with standard-of-care antibiotics.

Funding Statement

All of the studies reported here were funded primarily by Trellis Bioscience LLC. Partial funding for the work shown in Fig. 2 and 3 was provided by a grant from the U.S. National Institute of Allergy and Infectious Disease (R41 AI120425). The in vivo experiments at University Hospital Basel were funded through a contract with Trellis Bioscience, LLC. The UHB investigators have no other financial interest in Trellis. All other authors are paid employees or consultants of Trellis Bioscience.