Bactericidal/Permeability-Increasing Protein Preeminently Mediates Clearance of Pseudomonas aeruginosa In Vivo via CD18-Dependent Phagocytosis

Chronic Pseudomonas aeruginosa infection mysteriously occurs in the airways of patients with cystic fibrosis (CF), bronchiectasis (BE), and chronic obstructive pulmonary disease (COPD) in the absence of neutrophil dysfunction or neutropenia and is strongly associated with autoimmunity to bactericidal permeability-increasing protein (BPI). Here, we define a critical role for BPI in in vivo immunity against P. aeruginosa. Wild type and BPI-deficient (Bpi-/-) mice were infected with P. aeruginosa, and bacterial clearance, cell infiltrates, cytokine production, and in vivo phagocytosis were quantified. Bpi-/- mice exhibited a decreased ability to clear P. aeruginosa in vivo in concert with increased neutrophil counts and cytokine release. Bpi-/- neutrophils displayed decreased phagocytosis that was corrected by exogenous BPI in vitro. Exogenous BPI also enhanced clearance of P. aeruginosa in Bpi-/- mice in vivo by increasing P. aeruginosa uptake by neutrophils in a CD18-dependent manner. These data indicate that BPI plays an essential role in innate immunity against P. aeruginosa through its opsonic activity and suggest that perturbations in BPI levels or function may contribute to chronic lung infection with P. aeruginosa.


INTRODUCTION
Polymorphonuclear cells or neutrophils (PMNs) are an essential component of innate immunity against Pseudomonas aeruginosa infection (1,2). Neutropenic mice are more susceptible to P. aeruginosa-related mortality (2), while humans deficient in neutrophil-mediated antimicrobial mechanisms (typically altered phagocytic function) have markedly increased susceptibility to P. aeruginosa infection (3,4). Thus, immunity against P. aeruginosa is dependent on neutrophilmediated phagocytosis. Chronic P. aeruginosa infection occurs in the airways of patients with cystic fibrosis (CF), non-CF bronchiectasis (BE), and chronic obstructive pulmonary disease (COPD) in the absence of neutrophil dysfunction or neutropenia (5)(6)(7)(8)(9)(10). The underlying mechanism(s) of chronic P. aeruginosa airway infection in the context of intact neutrophil function is unknown. We previously reported a strong association between autoreactivity to bactericidal/permeabilityincreasing protein (BPI) and chronic P. aeruginosa infection in the serum of CF and BE patients (5,6). Moreover, levels of autoantibodies to BPI inversely correlate with patient lung function in CF, BE, and COPD (5)(6)(7)(8)(9).
In an infected airway, neutrophils mediate killing of P. aeruginosa through phagocytosis, neutrophil extracellular trap (NET) formation, and neutrophil degranulation to release bactericidal proteases (11)(12)(13). BPI is an antimicrobial agent within neutrophil azurophilic granules that mediates membrane injury of gram-negative bacteria (GNB) following its high affinity binding to lipopolysaccharides (LPS) (14)(15)(16). This bactericidal activity of BPI has been recognized in vitro and is transduced by the N-terminal BPI domain (15,(17)(18)(19)(20)(21)(22). Mouse BPI shows 53% amino acid identity and 71% similarity to human BPI; overexpression of mouse BPI in human embryonic kidney 293 cells resulted in antibacterial activity against Escherichia coli, comparable with that obtained with human BPI (23). The ability of BPI to mediate E. coli opsonization has been reported in vitro and attributed to the ability of the BPI C-terminal domain to engage myeloid cells through an undetermined pathway (17). The function of BPI in innate immunity against GNB in vivo has not been studied.
We demonstrated that BPI is required for clearance of P. aeruginosa in vivo, in both lung and peritoneal infection models. The absence of BPI resulted in increased inflammation, characterized by enhanced neutrophil recruitment and inflammatory cytokine production in both the lung and the peritoneum. In the absence of BPI, altered P. aeruginosa clearance was due to impaired phagocytosis of P. aeruginosa by BPI-deficient neutrophils and not due to an impairment in degranulation or production of reactive oxygen species (ROS). Exogenous human BPI decreased bacterial load and reduced inflammation in vivo by facilitating P. aeruginosa uptake into Bpi-/-neutrophils, in a CD18-dependent manner. Together, these findings provide the first in vivo evidence that BPI is essential for eliminating P. aeruginosa infection. Moreover, this activity of BPI is transduced through enhanced phagocytosis of P. aeruginosa, rather than by direct bacterial cytotoxicity.

Mice
Age-matched male and female (6-10 weeks old) mouse littermates on the C57BL/6 background were used for all experiments. Heterozygous Bpi+/-(C57BL/6NJ-Bpi em1(IMPC)J / Mmjax) mice were obtained from The Jackson Laboratory (MMRRC Stock: 42125-JAX, Bar Harbor, ME) and bred in our facility. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Research Council (24). All experiments were conducted according to protocols approved by Dartmouth College's Institutional Animal Care and Use Committee (IACUC Protocol 00002007). Tail DNA genotyping was performed on pups between 8-14 days of age (Supplemental Figure 1) following the specific strain protocol provided by The Jackson Lab (MMRRC Stock: 42125-JAX). No developmental phenotype was observed with homozygous Bpi-/-mice.
Bacterial Culture and Phagocytosis Assays P. aeruginosa strains PAO1 and PA14 (wt and DpopB) were obtained from Dr. Brent Berwin. Bacteria were cultured overnight in lysogeny broth (LB) at 37°C and sub-cultured for 3hrs in LB (1:30) at 37°C. Phagocytosis assay of live P. aeruginosa bacteria was performed as previously described (27,28). Neutrophils (250,000 in 250ul RPMI, with or without 30mins 37°C pre-incubation with human or mouse BPI) were incubated with P. aeruginosa (PAO1 or PA14) for 45mins at 37°C at 10 or 25 multiplicities of infection (MOI) and then treated with 250ul of gentamicin (200mg/ml) for 20mins at 37°C to kill extracellular bacteria before being pelleted and washed twice in 500ul RPMI. Neutrophils were resuspended in lysis buffer (0.1% Triton X-100 in PBS) and lysed contents (10ul) plated on 1.5% LB agar plate and incubated overnight at 37°C. Bacterial colonies (colonyforming unit, CFU) were counted on the following day, representing the number of live bacteria phagocytosed by the neutrophils. To confirm phagocytosis, neutrophils were treated with GFP-expressing P. aeruginosa (PAO1) as above. After incubation with gentamicin and washing step, 200ul of lysis buffer was added to the tubes, mixed, and transferred to a 96-well Flat Clear Bottom Black Polystyrene Microplates (Fisher Scientific). Fluorescence was measured with BioTek Synergy plate reader (485/528 nm).

Intraperitoneal and Oropharyngeal P. aeruginosa In Vivo Challenge and Cytokine Analysis
For in vivo intraperitoneal (i.p.) and oropharyngeal exposure sublethal infection, mice (wt or Bpi-/-) were anesthetized with isoflurane (to effect) and either injected with 3x10 6 CFU of subcultured PA14 or PBS i.p. (300ul) or by oropharyngeal administration (50ul). For those receiving BPI treatment, 10mg of human BPI in PBS (Athens Research) was injected i.p. 15mins following bacterial infection. Mice were euthanized 3hrs postinfection (hpi). Peritoneal and bronchoalveolar lavage (BAL) fluids were collected using 1.5ml and 0.8ml PBS, respectively and 10ul plated (1:10 dilution peritoneal lavage; no dilution BAL) on 1.5% LB agar plate. Plates were incubated overnight at 37°C and recovered CFUs determined by colony counts that were backcalculated to the original lavage volume. Excess peritoneal and BAL samples were spun down (3000rpm 5 mins) and cell-free lavages stored at −80°C for cytokine analysis. Lysates from cell pellets were made in citrate-phosphate buffer as mentioned above and stored for immunoblotting. Blood samples were collected from the inferior vena cava and serum extracted. TNFa, IL-1b, and IL-6 concentrations in the lavage and serum samples were determined with mouse ELISAs: TNFa MAX Deluxe Set (BioLegend), IL-1b Duoset, and IL-6 Duoset (R&D Systems). Murine hind limb long bone dissection and bone marrow isolation was performed as described previously (29). Briefly, mice were euthanized using CO 2 according to IACUC protocol and guidelines. Following femoral extraction, bone marrow cells were flushed out with cold PBS and a 22-gauge needle. Cells from the perfused bone marrow were passed through a 70-mm filter cell strainer. Erythrocytes were lysed using 1X BD Pharm Lyse (BD Bioscience). Staining and flow cytometry were performed as mentioned above.

In Vivo Phagocytosis Blocking via CD18 Receptor
Blocking of phagocytosis was achieved by i.p. injection of anti-CD18 M18/2 antibody or a rat IgG2b isotype control (50mg, Bio X Cell) 15 mins after mice were infected with 3x10 6 CFU of P. aeruginosa PA14 in 200ul PBS. In Bpi-/-mice, BPI (10mg) was administered i.p. 15 mins following the antibody injection. After 1.5hpi, i.p. lavage was performed (due to unequal immune cellrecruitment at 3hpi, shorter experiment was performed here). Cell-free lavage was plated onto LB agar plates for P. aeruginosa colony count. Peritoneal immune cells were counted to ensure comparable cell recruitment into the peritoneal cavity.

Lung Histology
Mice (wt and Bpi-/-) infected via oropharyngeal route (noninvasive) with P. aeruginosa (PA14, 3x10 6 in 50ul PBS, 3hpi) were sacrificed and lungs harvested and fixed in 10% formalin (Fisher Scientific) for at least 24hrs. Paraffin sections were generated and stained with Hematoxylin and eosin (H&E) by the Pathology Research Resource at Dartmouth-Hitchcock Medical Center. The slides were imaged using optical microscope (Olympus BX50), at 40x magnification.

Quantification of Reactive Oxygen Species (ROS)
Thioglycollate-induced neutrophils from wt and Bpi-/-mice were stained with DCFDA dye (Cellular ROS Assay Kit, Abcam) for 30mins at 37°C in the dark, washed, and resuspended in 1X supplemented buffer (supplied with the kit) with 10% FBS. Stained neutrophils (100,000 cells/well) were left untreated or were treated with ROS inducers: tert-butyl hydrogen peroxide (TBHP, 100uM) or glucose oxidase (GO, 2U/ml) for 4hrs, 37°C. Fluorescence was measured at 485/535 nm (BioTek, Synergy HT) nm in end point mode, with sensitivity 50 and bottom read.

Lung Tissue Homogenization
Mice were infected oropharyngeally with P. aeruginosa PA14wt (3x10 6 CFU in 50ul PBS) or just PBS for 3hrs. Euthanized mouse lungs were perfused with 5 mL of ice-cold PBS injected through the right ventricle of the heart. Perfused lungs were removed and placed in RPMI on ice until processing. Lungs were minced and placed in digestion buffer of 1 mg/mL collagenase type I + 60 U DNase type I (Sigma-Aldrich) in DPBS Ca2+ Mg2+ (HyClone). Lungs were digested for 30 min at 37°C while shaking, and passed through a 70-mm filter cell strainer. Digested contents of the lungs were plated on LB agar plates to determine recovery CFU. Cells were washed (in RPMI, 1500rpm, 4°C) and residual red blood cells were lysed using 1X BD Pharm Lyse lysis buffer (BD Biosciences), for 2 min, and counted. Cells were washed and subsequently blocked using mouse Fc Block (1:200, 10 min, 4°C). Staining for flow cytometry was performed as mentioned above.

Murine BPI (mBPI) Production and Affinity Purification of mBPI Peptide-Specific Antisera
Murine BPI plasmid was obtained from Origene (Cat. MR219864). Plasmid was transfected into TOP10 comp cells and DNA was purified using Qiagen Maxi Prep kit (Qiagen) according to manufacturer's instruction. DNA was transfected into Expi293 cells using Expi293 Expression System Kit (Thermo Fisher Cat. A14635) according to manufacturer's instruction. Mouse BPI was purified using Anti-FLAG Affinity Gel (Bimake, Cat. B23101) according to manufacturer's instruction. Protein concentration was measured by the Pierce BCA Protein Assay Kit (ThermoFisher Scientific). Murine BPI was detected by immunoblotting using affinity purified rabbit anti-BPI raised against the murine-specific amino acid (256-269) antibody (Pocono Rabbit Farm & Laboratory (PRF&L), Canadensis, PA). P. aeruginosa Killing Assay by BPI BPI (5mg/ml or 90nM) was incubated with 10 8 CFU of PAO1 (in 100ul PBS). After 30mins at 37°C incubation, the samples were diluted 1:10000 in ice-cold LB before plated (10ml) onto LB agar plates. Plates were incubated overnight at 37°C before CFUs were counted. CFU data was normalized to no-BPI treatment (100% survival).

Statistical Analyses
Data were analyzed using GraphPad Prism 6 software (GraphPad Software, Inc., California, USA). Student's paired and unpaired t-tests with Welch's correction were applied to determine the significant difference between two data sets. Oneway ANOVA with multiple comparisons were applied for more than two data sets. P <0.05 was considered significant.

Study Approval
All experiments were conducted according to protocols approved by Dartmouth College's Institutional Animal Care and Use Committee (IACUC Protocol 00002007).

BPI Is Required for Efficient P. aeruginosa Clearance and Control of Lung Inflammation
We investigated the role of BPI in immunity against P. aeruginosa, using BPI-deficient (Bpi-/-) mice (Supplemental Figure 1). BPI deficiency was confirmed in the Bpi-/-neutrophil lysates, which showed no reactivity when immunoblotted with a mouse-specific BPI (mBPI) antibody, compared to neutrophils from wild type (wt) mice, which demonstrated a~64kD band, corresponding to that seen with recombinant mBPI ( Figure 1A and Supplemental Figures 2A, B). Immunofluorescence staining of thioglycollaterecruited peritoneal neutrophils confirmed cytoplasmic residence of BPI in wt and absence of BPI in Bpi-/-neutrophils ( Figure 1B).
The role of BPI in the clearance of P. aeruginosa in vivo has not been reported. We therefore asked if BPI is required for the clearance of P. aeruginosa (PA14) following acute (3hr) infection. Mice, wt and Bpi-/-, were infected with P. aeruginosa via intraperitoneal or oropharyngeal routes. Colony forming units (CFU) were assayed in the lavage (peritoneal or bronchoalveolar, respectively) or lung homogenate ( Figure 1C). Higher numbers of colony forming units (CFUs) were recovered from both the peritoneal fluid and the bronchoalveolar lavage (BAL) samples of Bpi-/-mice, compared to wt littermates ( Figures 1D, E). Bpi-/-mice also demonstrated higher CFUs in the whole lung homogenates following acute lung infection with P. aeruginosa ( Figure 1F). While no substantive differences in lung pathology were observed between the two genotypes at baseline (Supplemental Figure 3), the inflammatory response to P. aeruginosa infection in Bpi-/-mice was associated with increased cellularity and inflammation, compared to wt mice through H&E staining ( Figure 1G). These findings reveal that BPI is required for in vivo clearance of P. aeruginosa and suggest that lack of P. aeruginosa clearance in the absence of BPI leads to excessive lung tissue inflammation.

BPI Deficiency Enhances Neutrophil Recruitment and Cytokine Production
To investigate whether reduced clearance of P. aeruginosa in Bpi-/-mice was due to an impaired inflammatory response, we evaluated neutrophil recruitment to the infection site. On the contrary, significantly higher neutrophil numbers were detected in both the BAL and the peritoneum of Bpi-/-mice compared to wt controls following infection (Figures 2A, B). The enhanced neutrophil response in BPI-deficient mice was accompanied by a decrease in the percentage of neutrophils in the bone marrow of Bpi-/-mice ( Figure 2C), suggesting greater infection-triggered neutrophil recruitment in the absence of BPI.
The increased neutrophil recruitment in Bpi-/-mice was accompanied by significantly higher concentrations of inflammatory cytokines (TNFa, IL-1b, and IL-6) in both the peritoneal and the BAL fluid samples, compared to the wt mice (up to 3.6 fold increase in peritoneal lavage and 37.4 fold increase in the BAL) ( Figures 2D, E). No differences in the peritoneal and the BAL fluid cytokine levels were observed between wt and Bpi-/-mice challenged with PBS, which were overall significantly lower in the absence of PA14 infection (Supplemental Figures 4A, B and Figures 2D, E). Peritoneal infection led to significantly higher levels of serum TNFa, IL-1b, and IL-6 in Bpi-/-, compared to wt mice (Supplemental Figure 5A). With lung infection, a difference in serum cytokines was only seen with IL-1b (Supplemental Figure 5B). Therefore, these data demonstrate that absence of anti-microbial and anti-inflammatory functions mediated by BPI resulted in increased airway, peritoneal and systemic inflammation. Hence, not only is BPI essential for the clearance of P. aeruginosa in vivo, intact BPI is required for the control of the inflammatory cellular and cytokine responses during acute P. aeruginosa infection.
Compromised P. aeruginosa Clearance In Vivo in the Absence of BPI Is Due to Impaired Neutrophil Phagocytosis To identify the mechanism(s) responsible for defective bacterial clearance in vivo in the absence of BPI, we investigated different   Figures 2A, B, e.g. 90% in the peritoneum and 97.3% in the BAL). Degranulation was evaluated by quantifying release of myeloperoxidase (MPO), a bactericidal protein residing in neutrophil azurophilic granules alongside BPI (30). MPO levels in the peritoneal lavage fluid were higher in Bpi-/-compared to wt mice ( Figure 3A), suggesting that degranulation was not compromised in Bpi-/-neutrophils. Thioglycollate-elicited peritoneal neutrophils were investigated in vitro for differences in reactive oxygen species (ROS) production or phagocytosis ability ( Figure 3B). In agreement with their intact ability to undergo degranulation, Bpi-/-neutrophils produced equal levels of neutrophil extracellular traps (NETs) and ROS as the wt neutrophils upon stimulation with P. aeruginosa or known ROS stimuli (TBHP−tert-Butyl hydroperoxide or GO−glucose oxidase) ( Figure 3C and Supplemental Figure 6).
Together, these data demonstrate that compromised P. aeruginosa clearance in BPI-deficient mice in vivo is not due to impaired degranulation or ROS production by Bpi-/-neutrophils. While cytocidal activity of BPI toward GNB has been reported in vitro (14-16, 31, 32), its role in neutrophil-mediated phagocytosis of GNB is less well understood (17,19,20). To investigate this, thioglycollate-elicited Bpi-/-and wt neutrophils were assayed for the number of ingested CFUs of P. aeruginosa (PAO1 or PA14) in vitro following gentamicin treatment to kill extracellular bacteria. Cell lysates showed significantly fewer P. aeruginosa CFUs were recovered from BPI-deficient neutrophils ( Figures 3D, E). This effect was independent of the P. aeruginosa strain or multiplicity of infection (MOI). To confirm that CFU counts accurately represent relative phagocytic ability of the neutrophils, levels of GFP-expressing P. aeruginosa in neutrophil lysates were quantified by fluorescence, following gentamicin (D, E) Inflammatory cytokines TNF, IL-1b, and IL-6 (pg/ml) were measured in (D) peritoneal fluid (n=5) and (E) BAL (n=3,4) of mice infected with PA14 (3x10 6 CFU, 3hpi). All data were analyzed by unpaired t-test with Welch's correction, except for data in (C) which were analyzed by paired Student t-test, **p < 0.01, *p < 0.05; Error bars represent mean ± SEM. protection assay. Lower levels of GFP fluorescence intensity was detected in Bpi-/-neutrophil extracts relative to wt controls (Supplemental Figure 7). Addition of exogenous BPI (mouse or human) to Bpi-/-neutrophils prior to in vitro culture with P. aeruginosa increased intracellular CFU recovery in a concentration-dependent manner ( Figures 3F, G). Moreover, the addition of exogenous BPI neutralized LPS-induced TNFa release by Bpi-/-neutrophils (Supplemental Figure 8). Together, these data suggest that BPI is required for efficient phagocytosis of P. aeruginosa by neutrophils in vitro and that it mediates antiinflammatory properties by reducing LPS-induced TNFa release.

Addition of BPI Enhances P. aeruginosa Clearance and Lowers Inflammation in Bpi-/-Mice In Vivo
To investigate if treatment with BPI can rescue the defect in bacterial clearance seen in Bpi-/-mice in vivo ( Figures 1D-F), we administered human BPI (hBPI, 0.2-10 mg) 15 min following intraperitoneal P. aeruginosa infection of Bpi-/-mice ( Figure 4A and Supplemental Figure 9). Addition of hBPI significantly reduced P. aeruginosa CFU counts in the peritoneal lavage of Bpi-/-mice, compared to untreated controls ( Figure 4B). CFU counts from Bpi-/-mice treated with hBPI were comparable to those of wt mice ( Figure 4B). Moreover, treatment of Bpi-/-mice with hBPI resulted in a significant reduction in TNFa, IL-1b, and IL-6 levels in the cell-free peritoneal fluid, compared to infected Bpi-/-mice without hBPI treatment ( Figure 4C). Full length human BPI (~64kD) was detected in the lysates of peritoneal immune cells (3h post infection) by immunoblotting, but not in the cells of untreated Bpi-/-mice ( Figure 4D and Supplemental Figure 2C). Immunofluorescence staining confirmed that hBPI was internalized by the infiltrating immune cells, localizing in the cytoplasm ( Figure 4E). Exogenous hBPI enabled uptake of P. aeruginosa by neutrophils [>89% of CD45+ cells in the peritoneum ( Figure 2B)] as demonstrated by co-localization of hBPI with td-Tomato expressing P. aeruginosa ( Figure 4E, top  -test; (F, G) Intracellular CFU counts from wt and Bpi-/-mPMNs treated with P. aeruginosa PAO1 (MOI 25, n=3) with increasing concentrations of (F) native human BPI (hBPI, 0-600 nM) or (G) recombinant mouse BPI (mBPI, 0-600 nM). Statistical significance was determined by one-way ANOVA with multiple comparisons for (C), and unpaired t-test with Welch's correction for others. ****p < 0.0001, ***p < 0.001, *p < 0.05, ns, not significant; Error bars represent mean ± SEM. Theprungsirikul et al. BPI Mediates P. aeruginosa Clearance In Vivo panels). No td-Tomato expressing P. aeruginosa was detected in the peritoneal cells of Bpi-/-mice without hBPI supplementation ( Figure 4E, middle panels). Td-Tomato P. aeruginosa were observed in the peritoneal cells of wt mice, suggesting that the endogenous BPI also mediates P. aeruginosa cellular uptake in vivo ( Figure 4E, bottom panels). These findings reveal that administration of exogenous BPI can rescue the ability to clear P. aeruginosa in vivo and suggest that a prominent function of BPI in vivo is to mediate phagocytosis of P. aeruginosa.
BPI Mediates Clearance of P. aeruginosa In Vivo by Phagocytosis, Dependent on CD18 While prior studies of BPI function, largely limited to in vitro systems, have stressed its direct cytocidal activity (15,19,20), our data suggest an important role for BPI-mediated phagocytosis in the clearance of GNB in vivo (Figure 4). To determine whether the exogenous BPI mediates P. aeruginosa clearance in vivo via phagocytosis or direct killing, we blocked the ability of neutrophils to phagocytes P. aeruginosa prior to treatment with hBPI by neutralizing CD18, a b2 integrin required for the uptake of P. aeruginosa by phagocytes (33). Anti-CD18 blocking antibody (clone M18/2) or isotype control was given 15 minutes following P. aeruginosa (PA14) infection via intraperitoneal route; Bpi-/mice received hBPI 15mins after antibody treatment, and peritoneal lavage was performed to determine PA14 CFUs ( Figure 5A). Anti-CD18 treatment completely abolished the enhanced PA14 clearance mediated by administration of exogenous hBPI ( Figure 5B). We validated the ability of anti-CD18 antibody to effectively block phagocytosis by treating PA14 infected wt mice and demonstrating higher CFU counts in the lavage, compared to isotype treated controls ( Figure 5B). Peritoneal immune cell count in both Bpi-/-and wt mice was comparable between all treatments ( Figure 5C), suggesting that the effect seen on peritoneal CFUs ( Figure 5B) was not due to disparities in immune cell recruitment. Pseudomonas strains with T3SS can manipulate host cells leading to evasion of phagocytosis and clearance (34). Infecting mice with P. aeruginosa strain deficient in Type III secretion system (T3SS) translocon (PA14DpopB) showed comparable recovered CFU to mice infected with P. aeruginosa strain sufficient in T3SS translocon (PA14wt) (Supplemental Figure 11). These two strains of P. aeruginosa are equally susceptible to BPI killing in vitro (Supplemental Figure 12). Thus, absence or presence of a functional T3SS exhibited no significant effect on Pseudomonas CFU in wt and Bpi-/-mice. Together, these data suggest that BPI mediates P. aeruginosa clearance in vivo mainly via bacterial phagocytosis and that this process is dependent on CD18.

DISCUSSION
The selective ability of P. aeruginosa to chronically infect the airways in patients with CF, BE and COPD in the absence of neutropenia is poorly understood (5)(6)(7)(8)(9)(10). We report that BPIdeficient mice exhibit decreased ability to limit P. aeruginosa infection in both the lung and peritoneum in vivo. Thus, in this model system, BPI plays a functionally non-redundant role in innate immunity against P. aeruginosa. This defect is characterized by increased neutrophil accumulation and production of inflammatory cytokines at sites of infection, propagated by increased recruitment of neutrophils from the bone marrow in Bpi-/-mice. The absence of BPI did not impair neutrophil degranulation or ROS production. Rather, in vitro and in vivo studies demonstrated that Bpi-/-neutrophils exhibited markedly decreased phagocytosis of P. aeruginosa. This defect was corrected in vivo by administering exogenous human BPI. Following phagocytosis, ingested P. aeruginosa in neutrophils co-localized with exogenous BPI, suggesting a specific and direct role of intact BPI in bacterial opsonization. Lastly, the inability of exogenous BPI to clear P. aeruginosa in the presence of CD18 blockade supports the pre-eminent role of opsonization by BPI as well as identifies its transduction via a CD18-dependent pathway. The anti-bacterial functions of BPI have mainly been studied in vitro in systems without cell interactions (14-17, 19, 20, 31, 32). We confirmed that BPI exhibits direct cytocidal activity towards P. aeruginosa in vitro (Supplemental Figure 10). These studies do not recapitulate the complexity and heterogeneity of the in vivo immune cell environment. Our study of BPI-deficient mice is novel, as it is the first study to demonstrate the function of BPI in P. aeruginosa phagocytosis and clearance in vivo. Prior in vitro studies have shown that BPI cytotoxic activities reside in the Nterminal domain which also confers LPS binding (19,20,31,32). In contrast, the C-terminal domain is thought to confer opsonic activity of Escherichia coli through an as-yet-undefined pathway in neutrophils and monocytes in in vitro systems (17,35). For the first time, we provide in vivo evidence of BPI-mediated phagocytosis, by demonstrating P. aeruginosa co-localization with intact human BPI (confirmed by immunofluorescence) in Bpi-/neutrophils in peritoneal infection. Immunoblotting of neutrophil lysates confirmed the absence of BPI proteolysis with neutrophil uptake, suggesting the requirement for intact BPI. Thus, our data strongly support the model by which the cationic N-terminal portion of intact BPI binds LPS on P. aeruginosa while the Cterminal domain mediates neutrophil phagocytosis of an intact BPI-P. aeruginosa complex. Moreover, CD18 blockade inhibits this activity of exogenous human BPI in vivo, thereby confirming that the functional importance of BPI in this model is largely opsonic in nature, rather than by transducing bacterial cytotoxicity.
The ability of CD18 neutralizing antibodies to inhibit BPIdependent phagocytosis in Bpi-/-neutrophils suggests a direct interaction. Recent study has shown that CD18 expression facilitates the uptake of both motile and nonmotile P. aeruginosa strains (33). This central role in P. aeruginosa uptake has been confirmed; CD18 is the only phagocytic cell surface receptor with genetic evidence to support that it is critical for uptake of P. aeruginosa (36, 37). CD18 (b2 integrin) forms heterodimers with CD11a, CD11b, CD11c and CD11d (38). Thus, two models are possible: i) BPI alters P. aeruginosa membrane structure to allow for CD18 binding, or ii) BPI directly engages with CD18 for binding and phagocytosis of P. aeruginosa. Therefore, our findings of the key importance of BPI in mediating P. aeruginosa phagocytosis provide a strong rationale to investigate the CD18-BPI interactions in chronic infectious diseases such as CF, non-CF BE, and COPD, where reduced phagocytosis has been reported (39)(40)(41)(42)(43). In CF and non-CF BE, decreased phagocytosis by both neutrophils and macrophages has been identified and correlated with reduced expression of phagocytic receptors (39)(40)(41). Whether and how this might influence BPI mediated-phagocytosis warrants investigation. In COPD patients, who frequently suffer from airway P. aeruginosa infection (10,44), macrophages demonstrate significantly reduced phagocytic capacity of bacteria and apoptotic cells (42,43,45). Since BPI is also expressed by human macrophages albeit at lower levels than neutrophils (14,46), BPI is likely to play an important role in macrophage-mediated phagocytosis. Together, our data suggest BPI as a potential therapy to enhance immune cell-mediated phagocytosis in diseases such as CF, BE, and COPD.
The observation of a BPI-specific defect in clearing P. aeruginosa is notable given its ubiquitous and chronic nature in CF, BE, and COPD. P. aeruginosa infection in these diseases is strongly linked with autoantibodies to BPI (5)(6)(7)(8)(9)(10)47). Moreover, autoantibodies to BPI correlate with worsening lung function in patients with these diseases (5)(6)(7). While autoantibodies to BPI are found in other disease states, high-avidity autoantibodies are restricted to patients with chronic lung infection with P. aeruginosa (6,48). Given our data that BPI is required for P. aeruginosa immunity in vivo, the serologic findings in CF, BE, or COPD suggest that autoantibodies against BPI might shape a protective niche that enables chronic infection by P. aeruginosa in the clinical setting. The unique properties of this model not only derive from the specific association of lung infection by P. aeruginosa with BPI autoimmunity but also our demonstration of the importance of BPI in clearing this organism in vivo. Interestingly, no relationship was seen between circulating BPI levels and antibodies to P. aeruginosa in a cohort of bacteremic patients (48). This evidence suggests that patients who are susceptible to P. aeruginosa infections are not defective in producing or releasing BPI to combat infections. This further supports the notion that the presence of autoantibodies to BPI restricts its function during infection. In addition to bacterial clearance, the therapeutic potential of BPI is also to modulate inflammation, as phagocytosis in its nature is an antiinflammatory process (49). Excessive inflammation leads to lung injury in CF (50) particularly by neutrophils. Our data strongly implicate BPI to be a potent anti-inflammatory molecule, as seen with the levels of cytokines in vivo and lung pathology. This is particularly important in the light of possible diminished BPI function in patients with anti-BPI autoantibodies. Studies which confirm a functional role of BPI autoantibodies as well as address their specific association with P. aeruginosa lung infection would potentially lead to novel immunomodulatory strategies.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study protocols were reviewed, approved, and conducted according to Dartmouth College's Institutional Animal Care and Use Committee (IACUC Protocol 00002007).

AUTHOR CONTRIBUTIONS
JT, WR, and SS-G conceived the project and designed the experiments. JT and RW performed the experiments. AB produced and purified essential reagents. JT and SS-G performed data analyses and interpretation. JT, WR, and SS-G wrote the manuscript. JT prepared the figures. WR supervised all aspects of the study. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by the Cystic Fibrosis Foundation (CFF) [grant number: RIGBY 20G0], and the BioMT Institute for Biomolecular Targeting core lab (grant: P20GM113132). Further support came from the Translational Research Core at the Dartmouth Geisel School of Medicine, the Shared Resource facility, Irradiation, the Pre-clinical Imaging and Microscopy Resource (IPIMSR) at the Norris Cotton Cancer Center (NCI Cancer Center Support Grant 5P30CA023108-37 and NIH shared instrument grant S10OD21616, for the use of microscopy, pathology and flow cytometry instruments.