Sap Transporter Mediated Import and Subsequent Degradation of Antimicrobial Peptides in Haemophilus

Antimicrobial peptides (AMPs) contribute to host innate immune defense and are a critical component to control bacterial infection. Nontypeable Haemophilus influenzae (NTHI) is a commensal inhabitant of the human nasopharyngeal mucosa, yet is commonly associated with opportunistic infections of the upper and lower respiratory tracts. An important aspect of NTHI virulence is the ability to avert bactericidal effects of host-derived antimicrobial peptides (AMPs). The Sap (sensitivity to antimicrobial peptides) ABC transporter equips NTHI to resist AMPs, although the mechanism of this resistance has remained undefined. We previously determined that the periplasmic binding protein SapA bound AMPs and was required for NTHI virulence in vivo. We now demonstrate, by antibody-mediated neutralization of AMP in vivo, that SapA functions to directly counter AMP lethality during NTHI infection. We hypothesized that SapA would deliver AMPs to the Sap inner membrane complex for transport into the bacterial cytoplasm. We observed that AMPs localize to the bacterial cytoplasm of the parental NTHI strain and were susceptible to cytoplasmic peptidase activity. In striking contrast, AMPs accumulated in the periplasm of bacteria lacking a functional Sap permease complex. These data support a mechanism of Sap mediated import of AMPs, a novel strategy to reduce periplasmic and inner membrane accumulation of these host defense peptides.

Nontypeable Haemophilus influenzae (NTHI) is a commensal of the human nasopharnyx, yet causes opportunistic diseases such as conjunctivitis, sinusitis, exacerbations of chronic obstructive pulmonary disease, complications of cystic fibrosis and chronic and acute otitis media [20][21][22][23][24][25]. During the transition from a commensal to pathogen, NTHI must acquire nutrients and defend against host innate immune defense strategies including increased production of AMPs in response to infection. NTHI outer membrane remodeling provides a first line of defense against cationic AMPs. Lysenko and colleagues demonstrated that the presence of phosphorylcholine, a phase variable modification of NTHI lipooligosaccharide, alters outer membrane hydrophobicity that confers resistance to the cathelicidin LL-37 [26]. Additionally, HtrB is required for hexaacylation of NTHI lipid A, thus mutants lacking htrB are unable to fully acylate their lipid A rendering NTHI susceptible to AMP mediated killing [27]. NTHI lack the other described resistance mechanisms such as AMP efflux or exoprotease activity. The sap (sensitivity to antimicrobial peptides) operon encodes an inner membrane ABC-transporter, previously shown to play a crucial role in defense against AMPs [28][29][30][31][32][33][34][35][36]. Previously, we demonstrated that NTHI SapA, the periplasmic substrate binding protein of the Sap transporter, binds AMPs [37]. NTHI strains deficient in either SapA or the SapD ATPase are susceptible to killing by recombinant chinchilla b-defensin-1, an orthologue of human bdefensin-3, sharing 77% amino acid identity [37,38]. Moreover, SapA and SapD are required for virulence in a mammalian host [37,38]. Expression of the sap operon is up-regulated in vivo during NTHI-induced otitis media and in response to AMP exposure in vitro [37,38]. The mechanism by which the Sap transporter complex confers AMP resistance remains unknown.
Here, we demonstrated that NTHI lacking the ligand binding protein SapA were directly susceptible to AMP exposure in the mammalian host, as neutralization of cBD-1 in vivo reversed attenuation and clearance of the sapA mutant strain. Further, we describe a novel mechanism of AMP import into the bacterial cytoplasm, which is dependent upon Sap transporter function. The Sap-permease was required for cytoplasmic localization and in a Sap-permease deletion strain, AMPs accumulated in the bacterial periplasm. Since AMPs were susceptible to intracytoplasmic peptidase activity, we hypothesize that AMP import, coupled with cytoplasmic degradation, decreases AMP accumulation in the bacterial periplasm and protects the bacterium from subsequent perturbation of the bacterial cytoplasmic membrane, a counter strategy to evade innate immune defense and ultimately benefit the bacterium nutritionally.

Results
SapA is required for NTHI to directly counter host defensin microbicidal activity in vivo Previously, we demonstrated that the SapA periplasmic binding protein is required for NTHI to persist and cause disease in the middle ear of a mammalian model of otitis media [38]. In this coinfection model of middle ear disease (equal amounts of the wild type NTHI strain 86-028NP and the isogenic sapA mutant strain were introduced directly into the middle ear) we observed significant attenuation of the sapA mutant strain, which was unable to compete for survival and cleared from the middle ear. We subsequently demonstrated that recombinant SapA binds the AMP molecule, chinchilla b-defensin 1 (cBD-1) in vitro [37], and further, binds numerous AMPs, including the human b-defensin 3 (hBD-3) and cathelicidin (LL37) peptides [39]. These data support a mechanism whereby SapA confers protection from AMP lethality in vivo during disease progression. To directly assess this hypothesis, we sought to neutralize activity of cBD-1 in vivo and monitor the consequence of this treatment on survival of the SapA-deficient strain during a co-infection model of otitis media. We predicted that SapA was required by NTHI to counter the defensin bactericidal activity in vivo, and that neutralization of cBD-1 would reverse attenuation and clearance of the SapAdeficient strain in this environment.
Altered expression of AMPs can impact the ability of bacteria to colonize a host [11,40]. We previously demonstrated that bdefensin expression controls NTHI bacterial colonization of the nasopharyngeal mucosal surface in a chinchilla model of NTHImediated disease [41]. Neutralization of available native cBD-1 via passive inhalation of affinity-purified antibody assessed the direct contribution of AMPs in mediating NTHI colonization levels.
Here, we further developed and utilized this methodology to assess the direct contribution of AMPs in mediating attenuation of sapA mutant infection in the middle ear. We delivered affinity-purified anti-recombinant cBD-1 [(r)cBD-1] polyclonal antibody (or preimmune serum as a negative control) to the chinchilla middle ear cavity (n = 5 per cohort) via the superior bullae, to inhibit the activity of native cBD-1. Twenty minutes after pre-treatment, we challenged chinchillas transbullarly with NTHI in a co-infection model, then measured the relative concentration of wild type and sapA mutant present in middle ear effusions over a 14 day period. A second dose of neutralizing antibody (or pre-immune serum) was delivered 24 hours after bacterial challenge.
We demonstrated that animals receiving anti-(r)cBD-1 neutralizing antibody remained colonized with the sapA mutant strain ( Figure 1, panel B), which was sufficiently able to compete (in many cases out compete) with the wild type strain as shown in competitive index calculations (Figure1, panel C). In contrast, the SapA-deficient strain is attenuated in animals receiving preimmune serum (Figure 1, panel A). Consistent with our previous findings, we demonstrated that the sapA mutant was unable to persist over the course of infection (Figure 1, Panel A), and was significantly impaired (p,0.01) in its ability to infect the middle ear compared to that of the wild type strain 14 days after infection. Additionally, as previously described, anti-(r)cBD-1 antibody or the control pre-immune serum was not bactericidal against NTHI at concentrations used for AMP neutralization [41]. These data demonstrate that SapA functions to counter AMP lethality during NTHI infection in vivo. Since SapA binds and delivers periplasmic substrate to the inner membrane Sap transporter complex, we were interested to determine whether AMP molecules are transported to the cytoplasm of intact cells and whether this transport was dependent upon Sap permease activity.

Co-fractionation of AMPs in cytoplasmic extracts of NTHI
We predicted that AMP substrates, bound by the SapA periplasmic binding protein, would be transported into the bacterial cytoplasm. Thus, we sought to obtain cytoplasmenriched fractions of NTHI by bacterial fractionation and monitor the presence of AMPs in these fractions following exposure to sublethal concentrations of hBD3 or LL37. Traditional methods previously used to fractionate Escherichia coli were not successful with the prototypic NTHI clinical isolate strain, 86-028NP. Therefore, we utilized membrane destabilization and differential centrifugation (see Materials and Methods) and obtained both periplasm and cytoplasm-enriched fractions from lysed NTHI cells. The protein profiles of these two fractions were unique when visualized by two dimensional gel electrophoresis ( Figure 2A) and SDS-PAGE analysis ( Figure 2B). We confirmed fractionation by immunodetection of the periplasmic enzyme b-lactamase and the cytoplasmic protein SapD ( Figure 2C). b-lactamase was localized to periplasm-enriched fractions and the cytoplasmic protein SapD was only detected in the cytoplasm-enriched fraction (not present in the periplasm). Importantly, we demonstrated that hBD-3 and LL37 co-fractionated with cytoplasm-enriched fractions of NTHI, demonstrating that AMPs, even at sublethal concentrations,

Author Summary
The opportunistic pathogen Haemophilus influenzae is a normal inhabitant of the human nasopharynx, and is commonly implicated in respiratory tract infections, particularly of the middle ear (otitis media), sinuses, and lung (pneumonia, chronic obstructive pulmonary disease and cystic fibrosis). We have identified a multifunctional bacterial uptake system that is required for critical mechanisms of bacterial survival in the host. This Sap transporter system recognizes and transports host immune defense molecules and is involved in uptake of an iron-containing nutrient (heme) that is host-limited, yet required for bacterial growth and survival. We propose that bacteria utilize this, and likely other similar transport systems, for numerous functions that are important for bacterial survival in the host, including host immune evasion and metabolism. Our findings significantly advance our understanding of how single bacterial protein systems co-operate and coordinate multiple functions to equip bacteria to survive and cause disease in the hostile host environment. Our long-range goal is to block this uptake system thereby starving the bacterium of essential nutrients and also promoting clearance by the host immune response. Removal of this important bacterial survival mechanism will thwart the ability for Haemophilus to survive as a pathogen and thus decrease the incidence of disease development.
gained access to the bacterial cytoplasm of intact NTHI ( Figure 2D).

AMP structural classes display different kinetics of transport into the bacterial cytoplasm
We hypothesized that AMP substrates bound by SapA are delivered to the inner membrane Sap transporter complex for energy dependent transport into the bacterial cytoplasm. Indeed, we demonstrated that AMPs localized to the bacterial cytoplasm of intact cells within 30 minutes of exposure to sublethal AMP concentrations. Thus, we sought to characterize the kinetics of localization to the cytoplasm. NTHI were incubated with AMPs for 30 minutes (pulse period), cells were washed to removed unbound AMPs and then incubated in buffer alone (without AMPs) for 0, 2, or 4 hours (chase period). Incubation of AMP-exposed NTHI in buffer, during this chase period, did not affect NTHI viability during the time course (data not shown). At each time point, cytoplasm-enriched fractions were monitored for the presence of AMP molecules by immunoblot ( Figure 3). We observed maximal import of hBD3 at the earliest time point tested ( Figure 3A and 3C). Interestingly we observed a decrease in the detection of hBD3 over the remainder of the chase period (from 0 to 4 hours, Figure 3A and 3C). Import of hBD-3 was dependent upon SapA production as hBD-3 did not localize to the cytoplasm in a SapA-deficient strain (data not shown). Although similarly detected in the bacterial cytoplasm following the 30 minute pulse period (0 hr), LL-37 peptide appeared to accumulate during the first 2 hours of the chase period and decrease only slightly when incubated an additional 2 hours (4 hr, Figure 3A and 3C). These data suggest differential kinetics of AMP accumulation within NTHI, likely related to AMP structure and charge. We were unable to follow the fate of LL37 beyond the 4 hour chase period due to loss of NTHI viability.

AMPs are degraded by cytoplasmic peptidase activity
The absence of known MTR in NTHI [15], combined with our observation of decreasing AMP levels in the cytoplasm, led us to hypothesize that the AMPs are degraded in the cytoplasm. To determine susceptibility of AMPs to peptidase activity, we incubated cytoplasm-enriched fractions of NTHI with hBD-3 or LL37 for 0, 2, 5, or 13 hours in the presence or absence of a protease inhibitor cocktail mix (2/+ inhibitor , Figure 4). At each time point, samples were collected and monitored for AMP by immunoblot ( Figure 4). Similar to our observations in intact cells, we observed nearly complete loss of hBD-3 detection within 5 hours whereas LL37 loss was not observed to the same extent. After overnight incubation (T 13 ) however, LL-37 detection was dramatically reduced. In both cases, AMPs were susceptible to cytoplasmic fraction peptidase activity that was blocked in the presence of protease inhibitors ( Figure 4, + inhibitor). Collectively, these data suggest that AMP molecules are susceptible to cytoplasmic protease activity, and are likely beneficial to NTHI for metabolic purposes.

Transport of antimicrobial peptides to the cytoplasm of NTHI requires the Sap permease
The Sap transporter belongs to a family of ABC transporters that recognize and transport substrates that are typically small and cationic Figure 1. SapA is required for NTHI to directly counter host defensin lethality in vivo. Chinchilla middle ears were pre-treated with (r)cBD-1 antiserum (B) or pre-immune serum (A) then co-infected with a mixture of wild type NTHI strain 86-028NP and the isogenic sapA mutant to determine the consequence of neutralization of native cBD-1 on persistence of the sapA mutant in vivo. A second dose of serum was delivered one day after bacterial challenge. (A) The sapA mutant strain (black triangles) was unable to persist in the middle ear, and showed a significant decrease in colonization relative to that of the wild type strain (gray circles) 14 days after infection (p,0.01). (B) Neutralization of cBD-1 restored the ability of the sapA mutant strain to persist in the middle ear, at levels equal to or greater than that of the wild type strain. (C) Competitive index ratios (wild type to sapA mutant strain) were determined for each cohort receiving either the pre-immune (gray squares) or (r)cBD-1 (black squares) neutralizing antiserum. The black line indicates a competitive index of 1 (equal wild type to sapA mutant ratio). Black and gray horizontal lines (panels A and B) represent the geometric mean value of each cohort (n = 5 animals per cohort). Statistical significance was determined by non-parametric Mann-Whitney U test of geometric means, significance at p#0.05. doi:10.1371/journal.ppat.1002360.g001 in nature [29]. We previously demonstrated that AMPs can displace additional substrates previously bound by SapA, and that mutations in Sap transporter proteins alter NTHI susceptibility to AMP exposure [37][38][39]. We hypothesized that the Sap transporter may mediate the transfer of AMPs across the cytoplasmic membrane, as a mechanism of AMP transport into the cell. We therefore generated a deletion mutant lacking the permease components of the Sap transporter (SapB and SapC) and determined the subcellular localization of LL-37 and hBD3. NTHI (parent and sapBC permease mutant) were exposed to sub-lethal concentrations of either LL-37 or hBD3 for 30 minutes and processed to assess AMP localization by transmission electron microscopy, to visualize AMP localization in intact cells. We observed that both LL-37 ( Figure 5, top panels) and hBD3 ( Figure 5, bottom panels) localized to the periplasm and the cytoplasm of the parental cells. Accumulation of AMPs was not observed in either the bacterial outer or inner membranes under these conditions. In contrast, neither LL-37 nor hBD3 were observed in the cytoplasm of the sapBC permease-deficient cells, yet a striking accumulation of AMPs was observed in the periplasm and cytoplasmic membrane. These data suggest that subcellular localization of AMP molecules to the bacterial cytoplasm is dependent upon Sap-mediated transport. Consistent with this observation, sapBC permease-deficient cells are more sensitive to AMP exposure (data not shown). This targeted transport of AMPs, to the cytoplasm for degradation, may serve to reduce periplasmic accumulation of AMPs, and thus protect the cytoplasmic membrane from AMP association and lethality. Thus, Sap-mediated transport of AMP to the cytoplasm for degradation serves as a newly described subversion mechanism that would provide a nutritional benefit to the bacterium thus coupling innate immune resistance to metabolic activity.

Discussion
The ability of NTHI to persist in both the commensal and pathogenic lifestyles requires mechanisms to avert the host innate immune response. The epithelium in the upper airway produces AMPs that assist in the control of NTHI [41][42][43]. This led us to investigate the mechanisms used by NTHI to evade killing by AMPs. Modifications of the outer membrane primarily serve as a first line of defense against AMP lethality including addition of phosphorylcholine to lipooligosaccharide to repel positively charged AMPs [26] and lipid A acylation to increase NTHI hydrophobicity, thus altering membrane permeability [27]. Although effective as a first line defense strategy, AMP accumulation and disruption of the outer membrane, periplasmic localization and permeabilization of the cytoplasmic membrane is microbicidal. Many bacteria possess additional mechanisms to detect and resist AMP insult shown to be regulated through twocomponent regulatory signaling [44,45], release of extracellular peptidases [16], secreted proteins to bind AMPs [19], and AMP efflux pumps [15,46,47]. NTHI however, lacks two-component regulatory systems that sense AMP exposure, and additional mechanisms of AMP resistance have not been identified.
We previously demonstrated that the Sap transporter conferred NTHI resistance to AMPs. The periplasmic binding protein, SapA, binds AMPs [37], contributes to AMP resistance [38] and is also upregulated in response to AMP exposure during otitis media. This, taken with our data presented here, suggests that resistance to AMP lethality, conferred via Sap mediated transport of AMPs, is important for NTHI pathogenesis. Since we observed marked attenuation of Sap transporter mutants in vivo [37,38], we predicted that neutralization of AMPs would rescue this attenuated phenotype, supporting a critical role for Sap-dependent AMP import in NTHI pathogenesis. In support of this, in vivo neutralization of the hBD3 orthologue, chinchilla b-defensin-1, which is highly expressed in the chinchilla upper respiratory tract [48], results in increased colonization by NTHI [41]. These data suggest that AMPs play an important role in limiting infection in a mammalian host. Indeed, alterations in AMP production can affect the ability of other microorganisms to colonize a host [11,40,49]. The ability of NTHI to exploit mechanisms to resist AMP lethality may equip transition to that of opportunistic pathogen and ability to cause disease. We demonstrated here that neutralization of b-defensin activity in the middle ear of a mammalian model of otitis media rescues the attenuated phenotype of the SapA-deficient strain, such that the mutant strain is no longer cleared from this environment (Figure 1). This is the first direct evidence that SapA contributes to bacterial virulence by conferring protection from AMP lethality in vivo. Having established this, we monitored the consequence of SapA-AMP interaction in intact bacterial cells. We describe a novel  mechanism of inner membrane transport of AMP molecules, via the Sap ABC transporter, to the bacterial cytoplasm resulting in AMP degradation ( Figure 6). This direct mechanism of AMP influx serves to decrease both periplasmic and inner membrane accumulation of AMPs and protect NTHI from AMP bactericidal effects. Current models predict that AMP-induced microbicidal activity results from transmembrane pore formation subsequent to AMP accumulation at the cell surface [1]. However, we observed by TEM analysis, that AMPs localized to the bacterial cytoplasm, even when NTHI were exposed to sublethal concentrations that do not appear to accumulate in the bacterial membranes, suggesting a mechanism whereby bacteria can directly import AMP molecules. Since accumulation of AMPs in the bacterial cytoplasm would likely be detrimental, we hypothesized that AMPs are subsequently targeted for cytoplasmic degradation. Indeed, we observed that hBD3, localized to the bacterial cytoplasm, was susceptible to cytoplasmic degradation, as the ability to detect hBD3 decreased over time ( Figure 3A, C). Although we did not observe an appreciable decrease in LL-37 detection following uptake into intact cells, we demonstrated that LL37, like hBD-3, was susceptible to peptidase activity in cytoplasm-enriched fractions (Figure 4). These data suggest differential kinetics of uptake and degradation, likely dependent upon AMP structure and charge. In support of differential kinetic uptake of AMP molecules, we previously demonstrated that LL-37 and hBD3 differ in their abilities to bind and displace the ironcontaining compound heme, also shown to bind SapA [39]. Consistent with these studies, cytoplasmic accumulation of LL-37 appeared delayed, relative to hBD3 ( Figure 3C). Although we have not, as of yet, identified the binding sites for these SapA ligands, our data indicate that hBD3, which is highly cationic, may preferentially bind SapA based upon charge (unpublished). Additionally, it has been shown that the cathelicidin LL-37 contains proline-rich sequences that resist degradation by serine proteases [50] and NTHI lack PgtE and OmpT homologues, shown to mediate degradation of alpha-helical peptides [51][52][53][54], which may explain our observation that LL-37 appears more resistant to proteolysis within the bacterial cytoplasm. NTHI encodes homologues of bacterial cytoplasmic proteases, ClpX and Lon proteases, shown to contribute to AMP resistance in other microorganisms [55,56]. The contribution of these proteases to AMP degradation in NTHI is currently unknown, but remains the focus of future work.
An alternative hypothesis is that an MTR drug efflux pump functions to export AMPs that have accumulated in the bacterial inner membrane and in the cytoplasm. It was recently shown that the MtrC periplasmic membrane fusion protein conferred resistance to LL-37 and b-defensins in Haemophilus ducreyi [47]. Interestingly, the H. ducreyi Sap transporter does not mediate bdefensin resistance [36,47]. Although NTHI strain 86-028NP contains week homologs of the MTR system, our evidence suggests that this system is not functional or does not function to confer resistance to defensin or cathelicidin peptides, a well described function of the NTHI Sap transporter [37][38][39]. Further, we demonstrated that in vivo neutralization of native cBD-1 was sufficient to rescue competitive growth of the sapA mutant, suggesting that SapA was sufficient to mediate resistance to this defensin molecule in the host. Finally, a functional MTR system would be expected to export AMP molecules thus avoiding periplasmic, membrane and cytoplasmic accumulation of AMPs. We observed, in the absence of a functional Sap permease complex, AMP accumulation ( Figure 5) and increased sensitivity to AMP-mediated killing (data not shown). We would expect, based upon these published reports that a functional MTR system would not result in this phenotype.
Development of antibiotic resistant strains secondary to conventional antibiotic use has prompted interest in the use of AMPs as therapeutic alternatives to treat bacterial infections [57]. The use of AMPs in therapeutic regimens may be hindered due to bacterial subversion mechanisms of AMP lethality [54]. Since the Sap transporter functions to protect against the accumulation of AMPs in the bacterial periplasm, and subsequent interaction with the cytoplasmic membrane, a mechanism to block transporter function may provide novel therapeutic supplements for NTHI infections. Delivery of small molecule inhibitors to block substrate binding and transport would confound an important AMP resistance mechanism in NTHI, rendering NTHI susceptible to innate immune attack while preserving the normal flora that are often disrupted by conventional antibiotic use.
The described novel mechanism of AMP import and subsequent degradation expands our understanding of host-pathogen interactions, particularly those that mediate resistance to key components of the host innate immune response. Future studies to identify AMP specific intracellular peptidases, AMP degradation products, and metabolic consequences of targeted peptide degradation, are needed to better understand bacterial survival strategies in the hostile host environment. Further, a better understanding of bacterial mechanisms used to transition from commensal to that of an opportunistic pathogen will better equip the design of therapeutics to combat disease.

Ethics statement
All animal experiments were carried out in strict accordance with the accredited conditions in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee (Welfare Assurance Number A3544-01) at The Research Institute at Nationwide Children's Hospital, AR08-00027.
All experimental procedures were performed under xylazine and ketamine anesthesia, and all efforts were made to minimize suffering.

Animals
Healthy adult chinchillas (Chinchilla lanigera), purchased from Rauscher's Chinchilla Ranch (LaRue, OH), were used for these studies, after allowing them to acclimate to the vivarium for 7 to 10 days. Chinchillas were anesthetized with xylazine (2 mg/kg, Fort Dodge Animal Health, Fort Dodge, IA) and ketamine (10 mg/kg), Phoenix Scientific Inc., St. Joseph, MO), and nasopharyngeal lavage fluids were obtained by passive inhalation of 500 ml of pyrogen-free sterile saline into one naris with recovery of lavage fluid from the contralateral naris as liquid was exhaled. Middle ear fluids were recovered via epitympanic tap through the superior bullae, and directly obtained from the inferior bullae behind the tympanic membrane. All animal experiments were performed using accredited conditions for animal welfare approved by the Institutional Animal Care and Use Committee (Welfare Assurance Number A3544-01) at The Research Institute at Nationwide Children's Hospital, AR08-00027.

Bacterial strains and culture conditions
Nontypeable Haemophilus influenzae strain 86-028NP is a minimally passaged clinical isolate obtained at Nationwide Children's Hospital in Columbus, Ohio. This prototypic wild type strain has been sequenced and extensively characterized in chinchilla models of otitis media [58,59]. The parental NTHI strain 86-028NP:: rpsL is a streptomycin resistant strain constructed as previously described [60]. The sapBC permease deletion mutant was constructed as an unmarked, non-polar deletion mutation of the sapB and sapC genes by recombineering strategy as previously described [39]. Bacterial strains were grown on chocolate II agar (Becton Dickinson, Sparks, MD) or in brain heart infusion broth supplemented with 2 mg heme/mL and 1 mg NAD/mL (sBHI). Bacteria were cultured from overnight growth on chocolate II agar, resuspended in sBHI to OD 490 = 0.65 (equivalent to 1610 8 CFU/ml), diluted 1:5 into fresh sBHI medium and grown to mid-log phase (3 hours) at 37uC, 5% CO 2 , static.

Antibody-induced neutralization of native cBD-1 in vivo
In vivo neutralization of chinchilla b-defensin 1 (cBD-1) was performed as previously described with the following modifications [41]. We prepared and purified (r)cBD-1 based upon previously published methods [41,48]. A HiTrap protein G HP column (GE healthcare, Pittsburgh, PA) was used to affinity purify total IgG from rabbit anti-(r)cBD-1 and the cognate pre-immune serum. One milliliter of serum was dialyzed [3.5 kDa molecular weight cutoff (MWCO), EMD Chemicals Inc., San Diego, CA] at 4uC against 20 mM sodium phosphate buffer, pH 7.0 and applied to the affinity column. Immunoglobulins were eluted from the affinity matrix with 0.1 M glycine-HCl, pH 2.7 in one milliliter fractions into eppendorf tubes that contained 200 ml of 1.0 M Tris-HCl, pH 9.0, to neutralize the acidic elution conditions. Samples that contained the greatest amount of protein (fractions 1 and 2) were pooled and dialyzed overnight at 4uC against sterile saline. Protein concentrations of the anti-(r)cBD-1 and the pre-immune serum were determined using Coomassie Plus Protein Assay Reagent (Thermo Scientific, Rockford, IL). We confirmed that anti-(r)cBD-1 antiserum, and not the pre-immune antiserum, bound purified (r)cBD-1 by immunoblot (data not shown). Five adult chinchillas were administered 90 mg anti-cBD-1 or pre-immune rabbit immunoglobulin, in a total volume of 200 ml. Antisera were administered by passive inhalation of droplets of the solution delivered to the nares or direct transbullar inoculation of the middle ears of anesthetized chinchillas. Animals were then placed in a prone position for 20 minutes prior to intranasal challenge with approximately 5.0610 7 cfu NTHI wild type strain 86-028NP mixed with 5.0610 7 cfu sapA mutant (co infection) in a 100 ml volume or transbullarly with 2.5610 3 cfu NTHI wild type strain 86-028NP mixed with 2.5610 3 cfu sapA mutant in a 200 ml volume. Nasopharyngeal lavage and epitympanic taps were performed 1, 2, 4, 7, 10, 14 days after NTHI co-challenge, and bacterial counts were determined by dilution plating of bacteria on chocolate agar.

NTHI fractionation
Parental NTHI strain 86-028NP::rpsL was grown to mid-log phase in brain heart infusion broth supplemented with 2 mg heme/ml and 2 mg NAD/ml (sBHI). Cells were pelleted by centrifugation and resuspended in 1X phosphate buffered saline containing 2 mg Polymyxin B sulfate (Sigma-Aldrich, St. Louis, MO)/ml, 0.05% glycerol, to bind lipooligosacharide and generate spheroplasts, thereby releasing periplasmic contents to the supernatant. Following incubation for 1 hour at 37uC on a rotating platform (100 rpm), spheroplasts were pelleted by centrifugation and resuspended in 1 ml 10 mM HEPES pH 7.4 (Sigma-Aldrich, St. Louis, MO), 0.05% glycerol and subsequently lysed by freeze-thaw method (10 cycles, 278uC to 37uC). Lysed spheroplasts were then incubated at room temperature, on an orbital rocker, with 1 ml Benzonase Nuclease (Novagen, Darmstadt, Germany) and 5 ml 1 M MgCl 2 (Ambicon, Alamo, CA)) to a final concentration of 5 mM. Bacterial membranes were pelleted by ultracentrifugation (40,000 rpm for 1 hour, 4uC) and cytoplasmic proteins were removed in the supernatant. Cytoplasm and periplasm-enriched fractions were observed by two dimensional gel electrophoresis (Bio-Rad, Hercules, CA) of enriched fractions, stained with SYPRO Ruby Protein Gel Stain (Invitrogen, Carlsbad, CA) and imaged at 535 nm (Kodak Image Station 2000, New York, NY). To ensure successful NTHI fractionation, cytoplasm and periplasm-enriched fractions were analyzed by silver stain (Pierce, Rockford, IL) following SDS-PAGE separation. In addition, NTHI were exposed to a sub-lethal concentration of 15 mg ampicillin/mL to induce expression and trafficking of blactamase to the NTHI periplasm. Periplasm and cytoplasmenriched fractions were separated by SDS-PAGE and transferred to nitrocellulose (Bio-Rad, Hercules, CA). Membranes were then blocked in 3% nonfat dried milk and probed for the periplasmic protein, b-lactamase, by incubating with mouse anti-b-lactamase for 45 minutes or the cytoplasmic protein, SapD, by incubating with rabbit anti-SapD overnight. Membranes were then washed, incubated with goat anti-mouse IgG (H+L) HRP conjugate (Invitrogen, Carlsbad, CA) for 30 minutes or goat anti-rabbit IgG (H+L) HRP conjugate (Invitrogen, Carlsbad, CA) for 1 hour,