Discovery and Contribution of Nontypeable Haemophilus influenzae NTHI1441 to Human Respiratory Epithelial Cell Invasion

Nontypeable Haemophilus influenzae (NTHi) is the primary cause of bacterially induced acute exacerbations of chronic obstructive pulmonary disease (COPD). NTHi adheres to and invades host respiratory epithelial cells as a means to persist in the lower airways of adults with COPD. Therefore, we mined the genomes of NTHi strains isolated from the airways of adults with COPD to identify novel proteins to investigate their role in adherence and invasion of human respiratory epithelial cells.

ABSTRACT

Nontypeable Haemophilus influenzae (NTHi) is the primary cause of bacterially induced acute exacerbations of chronic obstructive pulmonary disease (COPD). NTHi adheres to and invades host respiratory epithelial cells as a means to persist in the lower airways of adults with COPD. Therefore, we mined the genomes of NTHi strains isolated from the airways of adults with COPD to identify novel proteins to investigate their role in adherence and invasion of human respiratory epithelial cells. An isogenic knockout mutant of the open reading frame NTHI1441 showed a 76.6% ± 5.5% reduction in invasion of human bronchial and alveolar epithelial cells at 1, 3, and 6 h postinfection. Decreased invasion of the NTHI1441 mutant was independent of either intracellular survival or adherence to cells. NTHI1441 is conserved among NTHi genomes. Results of whole-bacterial-cell enzyme-linked immunosorbent assay (ELISA) and flow cytometry experiments identified that NTHI1441 has epitopes expressed on the bacterial cell surface. Adults with COPD develop increased serum IgG against NTHI1441 after experiencing an exacerbation with NTHi. This study reveals NTHI1441 as a novel NTHi virulence factor expressed during infection of the COPD lower airways that contributes to invasion of host respiratory epithelial cells. The role in host cell invasion, conservation among strains, and expression of surface-exposed epitopes suggest that NTHI1441 is a potential target for preventative and therapeutic interventions for disease caused by NTHi.

INTRODUCTION

Nontypeable Haemophilus influenzae (NTHi) is a Gram-negative bacterium that colonizes the nasopharynx in its exclusive host, humans (1). NTHi is a pathobiont, and nasopharyngeal colonization by this organism precedes middle ear infection in children and infection of the lower airways of adults with chronic obstructive pulmonary disease (COPD) (1–4). NTHi is a primary cause of otitis media and is the leading cause of bacterially induced acute exacerbations of COPD (5–7). Antibiotics are used to treat both of these acute disease states. However, antibiotic treatment does not prevent subsequent infections, nor does it eradicate chronic lower airway infection in COPD. Consequently, continued use causes antibiotic resistance in NTHi (8, 9). There is currently no vaccine against NTHi licensed in the United States, despite the major burden of disease in adults with COPD and children. There is a crucial need to understand the complex biology of NTHi infection of secondary sites of the middle ear and COPD lower airways in order to identify targets of preventative therapeutics, such as vaccines and novel drugs (1, 10).

NTHi persists in the lower airways of adults with COPD for months to years (4, 11). NTHi uses several virulence mechanisms to establish and maintain COPD lower airway persistence. One such persistence virulence mechanism includes attachment to and invasion of host respiratory epithelial cells (2, 12, 13). Attachment allows NTHi to co-opt host cell endocytic pathways to subsequently invade and persist intracellularly (13–16). Intracellular survival protects bacteria from direct recognition from innate and humoral immune responses as well as antibiotic treatment. NTHi utilizes a suite of proteins with surface-exposed epitopes that interact with host cells to confer attachment and invasion (1, 2). Deletion of individual proteins does not completely ablate the capacity of NTHi to adhere to and invade host cells (1, 2, 17–19). The redundancy in proteins conferring adherent and invasive phenotypes supports this as a critical mechanism used by NTHi to colonize and persist in its human host. Additionally, NTHi surface-exposed proteins are genetically diverse, undergo genetic variation during COPD lower airway persistence, and are subject to phase variation (4, 20–22). These factors dictate that preventative therapies must target multiple conserved and invariant proteins to prevent NTHi infection of privileged sites of the middle ear and COPD lower airways.

We mined the genomes of NTHi strains that persisted in the lower airways of adults with COPD for novel proteins with ideal vaccine antigen characteristics, including (i) extracellular exposure on the bacterial cell surface, (ii) probable antigenicity, and (iii) absence of mutations incurred during persistence in the COPD airways. We further investigated top candidates for their role in adherence to and invasion of host respiratory epithelial cells. Proteins with surface-exposed epitopes have the capacity to interact with host cells and coordinate adherence to and invasion of host cells. Surface-exposed, conserved, and antigenic NTHi proteins are accessible to host immune responses that may block adherence and invasion and clear NTHi from sites of infection. Such proteins make ideal targets for preventative and therapeutic intervention strategies to prevent or eliminate infections by NTHi.

We identified the NTHI1441 open reading frame (ORF) as a conserved and invariant gene among persistent NTHi strains that is involved in invasion of host respiratory epithelial cells. We further showed that the NTHI1441 protein expresses extracellular epitopes on the bacterial cell surface and that adults with COPD develop increased serum IgG against NTHI1441 after experiencing an exacerbation with a strain of NTHi. The conservation, surface-exposed epitopes, and contribution of this previously undescribed NTHi protein to human respiratory epithelial cell invasion support the idea that NTHI1441 is involved in host infection. Furthermore, this work suggests that NTHI1441 is a candidate therapeutic target to prevent and treat NTHi infections.

RESULTS

Genome mining.

Bioinformatics programs were used to predict the subcellular localization and antigenicity and calculate sequence similarity of the translated annotated open reading frames (ORFs) of three NTHi strains, 67P38H1, 6P24H1, and 11P6H. These strains were isolated from the airways of adults with COPD and genome sequences were determined previously (4). The strains contained 1,714, 1,809, and 1,685 annotated open reading frames, respectively. We selected translated ORFs, proteins, that are predicted by PSORTb 3.0 to localize to the extracellular, outer membrane, periplasmic, and multiple unknown-location subcellular fractions (23). Proteins were selected from these subcellular fractions because proteins in these locations have the potential capacity to express epitopes on the bacterial cell surface. Proteins with surface-exposed epitopes are capable of interacting with host cell components to coordinate adherence and invasion and are accessible to host immune responses, an ideal characteristic of preventative therapeutic targets such as vaccine antigens (1, 2, 9, 10). To reduce the number of candidates missed by the limitations of protein localization predictions we also included proteins predicted to localize to periplasmic and multiple unknown locations. Proteins with predicted periplasmic, or intracellular, location have been shown to be expressed on the cell surface in NTHi and other bacterial pathogens (24–27). There were 143, 145, and 383 proteins from th 67P38H1, 6P24H1, and 11P6H genomes, respectively, predicted to localize to the selected sites. The 11P6H genome contained 282 proteins that have multiple unknown locations, compared to 35 in 67P38H1 and 38 in 6P24H2.

From the proteins predicted to localize to these sites, we selected those that are predicted to be probable antigens using VaxiJen, a protein antigen prediction program (28, 29). Antigenic proteins, such as those used in subunit vaccines, induce the production of host humoral immune responses that recognize the protein, block its virulence function, and target the associated pathogen for clearance. Of the surface-exposed proteins assessed, 68, 61, and 70 probable antigens were predicted in the respective 67P38H1, 6P24H1, and 11P6H genomes.

The annotated nucleotide ORFs of probable surface-exposed and antigenic proteins from each of the three strains were grouped based on sequence homology using Sequencher (GeneCodes 5.0.1). In total, we curated 41 proteins that are predicted to locate to subcellular fractions with potential surface exposure, are probable antigens, are prevalent in all three strains examined, and have a mean nucleotide sequence similarity of ≥85% (see Table S1 in the supplemental material). Of the mined protein candidates, previously described NTHi proteins involved in adherence to and invasion of host cells and vaccine candidate antigens were identified along with novel proteins (1, 2, 9). Other proteins previously shown to confer adherence to and invasion of host cells were not identified in our final set due to low percent similarity among the three strains and/or due to their antigenicity scores being below the set threshold (2, 17–19).

Characterization of top candidate sequences.

We expanded our initial analyses described above, which were based on three genomes, to a set of genomes of 101 strains of NTHi that persisted in the airways of adults with COPD for a mean of 313 days (range, 27 to 1,422 days) (4). We identified three top candidate ORFs that were predicted to encode two hypothetical proteins, NTHI1441 and NTHI1101, and the opacity-like porin opa. These were selected as top candidates because they localize to sites with potential bacterial surface accessibility, have high probable antigen scores, and have yet to be investigated.

The top candidates are present and conserved in all 101 persistent strains (Table 1) (4). A nucleotide BLAST search of the candidate ORFs revealed the prevalence and conservation of the candidates in all publicly available NTHi genomes (data not shown). The 101 persistent NTHi strains include the genome of the isolate when first acquired by the patient and the isolate immediately prior to clearance by the patient for each of the strains. We therefore analyzed whether the ORF of the candidates in each of the 101 NTHi strains developed mutations during persistence in the COPD airways. NTHI1101 incurred three nonsynonymous single nucleotide polymorphisms in 1 of 101 strains during persistence; no mutations were observed in NTHI1441 or opa during persistence. We conclude that the top candidates NTHI1441 and opa remain stable during persistence in the COPD airways. Positive selection in amino acids of surface-exposed proteins drives antigenic variation that can facilitate immune evasion, enabling persistence in the airways (4, 30). Positive selection was detected in two amino acids codons of the NTHI1101 in the genomes of 101 persistent strains (Table 1). Two of the three mutations that the NTHI1101 sequence incurred during persistence (noted above) included the amino acid codons experiencing positive selection.

TABLE 1.

Genetic analysis of top candidates in 101 pairs of persistent pair NTHi from the airways of persons with COPD

Genea	Annotationa	Locationb	Antigen scorec	No. of haplotypes	MHPD (%)d	Mutations during persistenced	Positive selectiond
NTHI1441	Hypothetical	Multiple	2.3	7	89.2 ± 8.1	None	None
NTHI1101	Hypothetical	Extracellular	0.7	22	98.4 ± 0.8	1 strain: 3 SNPs	A88V and M122L
opa	Opacity-like porin	Outer membrane	1.0	12	98.2 ± 1.2	None	None

^aGenes and annotations based on the strain 86-028NP genome (64, 68).

^bSubcellular protein location determined by PSORTb v3.

^cAntigen scores calculated by VaxiJen v2.0.

^dDetermined using methods described by Pettigrew et al. (4). SNPs, single nucleotide polymorphisms.

Adherence and invasion by top candidate isogenic knockout strains.

Adherence to and invasion of host cells are virulence mechanisms important for NTHi survival in the host (1, 2, 16). Therefore, we generated isogenic knockout mutants of the three top candidate ORFs in strain 86-028NP and assessed the mutants and parent strain in an adherence-and-invasion assay using human bronchial epithelial, NCI-H292, cells (15). A 71.3% ± 6.9% (mean ± standard deviation) decrease in invasion of NCI-H292 cells was observed in the isogenic knockout mutant of the top candidate NTHI1441 (∼1 × 10⁶ CFU/ml) compared to the parent strain (∼4 × 10⁶ CFU/ml). Deletion of NTHI1101 and opa ORFs had a marginal effect on invasion of host cells compared to the parent strain. Deletion of the top candidate open reading frames did not significantly affect bacterial cell adherence to host cells (Fig. 1).

FIG 1.

Adherence and invasion of top candidate isogenic knockout mutants 3 h after infection of NCI-H292 human bronchial epithelial cells. Data are presented as the percent adherence and invasion or invasion normalized to that of the parent strain for each group. Bars show means and error bars represent SDs from three independently performed experiments. A Kruskal-Wallis test with Dunn’s multiple-comparison test was used to calculate differences between adherence and invasion samples. ns, nonsignificant. *, P value < 0.05.

Characterization of NTHI1441 sequences in the 101 persistent NTHi strains.

Prior to this study, NTHI1441 was annotated as a conserved hypothetical gene in the 86-028NP genome, and its function is unknown. Thus, we characterized the NTHI1441 amino acid sequences among the set of 101 persistent NTHi strains isolated from the airways of adults with COPD. Sequences range from 78 to 107 amino acids, with 87% of strains being of 95 amino acids in length. The 86-028NP NTHI1441 sequence is within the haplogroup of the most common haplotype, including 75% of the persistent NTHi strains (Fig. 2A). The NTHI1441 sequence has a 24-amino-acid signal peptide identified by SignalP 4.0 (31). Within the NTHI1441 sequences we identified amino acid repeat regions with 11 conserved amino acids, AEGKCGEGKCG, spaced by six amino acids, A(D/S)K(P/A)K(A/S), between each repeat. There were 87% of strains with haplotypes containing three repeats, followed by 12% of strains with two repeats. One strain contained two of the conserved repeats with an additional repeat of a 12-amino-acid duplication located in between its two repeats (Fig. 2B). The NTHI1441 repeat domain is not present in other NTHi proteins. The NTHI1441 haplotypes have a mean haplotype pairwise diversity (MHPD) of 89.2% ± 8.1% with a range of 71 to 100%, based on a Gonnet similarity scoring matrix (Fig. 2C and D). Seven different NTHI1441 amino acid sequence haplotypes are present among the 101 persistent COPD strains of NTHi. The most common NTHI1441 haplotype was found in 76 of the 101 persistent strains; other haplogroups included 12, 6, 3, and 2 strains (Fig. 2E). Overall, we conclude that NTHI1441 is highly conserved among strains of NTHi.

FIG 2.

NTHI1441 amino acid sequence haplotypes in the 101 persistent NTHi strains isolated from the airways of adults with COPD. (A) Phylogram and amino acid sequence alignment of the 86-028NP reference strain and the seven NTHI1441 haplotypes observed in persistent NTHi strains, generated in MacVector version 16.0.8. Individual haplotype sequences are represented with the name of a persistent strain in the haplogroup. The percentage of strains within the haplogroup among the 101 persistent strains is located between the strain name and amino acid sequence. Numbers following the amino acid sequences indicate the length of the mature peptide of each sequence haplotype. The dotted underline in the 56P1H1 sequence indicates the unique amino acid repeat region of the haplogroup. (B) Diagrammatic representation of the 11-amino-acid sequence repeats of the NTHI1441 haplotypes. Percentages show the total number of strains among the persistent strains with each repeat pattern. The dotted line references the secondary repeat pattern observed in the haplogroup underlined in panel A. (C) Percent similarity matrix of the seven haplotypes calculated with a Gonnet scoring matrix in MacVector. (D) Mean haplotype pairwise diversity (MHPD) box-and-whisker plot showing individual pairwise percent similarities from panel C. (E) Distribution of the number of strains of each haplotype among the 101 persistent NTHi strains. The key shows the number of persistent strains contained within the haplogroup.

A protein-protein BLASTp search of the NTHI1441 sequence of strain 86-028NP identified NTHI1441 as a member of the COG3767 superfamily of uncharacterized low-complexity proteins with unknown function (32). Homologues of NTHI1441 identified by the search include those with high percent identities in Actinobacillus ureae (98.95%), Haemophilus haemolyticus (96.84%), Haemophilus parahaemolyticus (86.32%), and Haemophilus parainfluenzae (71.58 to 83.16%). Other bacteria of the genera Aggregatibacter, Mannheimia, Glaesserella, and Pasteurella had homologues with lower percent identity, between ∼60 and 80%. Query of the full-length and mature peptide NTHI1441 sequence produced low-confidence and low-percent-identity matches to structural homologues when analyzed by the Protein Homology/analogy Recognition Engine V 2.0 (Phyre2) (33).

NTHI1441 transcription in a four-gene operon.

ORFs are located up- and downstream in close proximity to the NTHI1441 ORF in the strain 86-028NP genome (GenBank accession number CP000057.2). We used Softberry-FgenesB to predict operon contents of the NTHI1441 ORF locus, including 1,957 and 3,191 bases of upstream and downstream sequence. FgenesB identified that NTHI1441 is located in a putative four-gene operon including the ORFs NTHI1440, NTHI1443, and NTHI1444 (Fig. 3A) (34). Endpoint reverse transcriptase PCR with RNA isolated from strain 86-028NP was used to determine the contents of the predicted NTHI1441 operon. Primer sets were designed to span portions of the ORFs and intergenic regions between the predicted NTHI1441 operon ORFs in addition to ORFs immediately upstream and downstream outside the predicted operon (Fig. 3A). Reactions with 86-028NP RNA and primer sets spanning predicted NTHI1441 operon ORF intergenic regions using primer sets B, C, and D generated amplicons of expected sizes (Fig. 3B). Reactions with RNA and primer sets spanning intergenic regions between the NTHI1441 operon and surrounding genes not predicted to be in the operon, primer sets A and E, did not produce amplicons (Fig. 3B). The reverse transcriptase PCRs support the idea that NTHI1441 is transcribed as an operon in an RNA transcript that includes ORFs NTHI1440, NTHI1441, NTHI1443, and NTHI1444. The nucleotide sequence of the NTHI1441 operon is present and conserved in all publicly available NTHi genomes, including the 101 persistent strains (data not shown).

FIG 3.

NTHI1441 expression in an operon. (A) Diagram of NTHI1441 operon contents from the 86-028NP genome locus. Genes predicted to be in an operon are gray with the NTHI1441 open reading frame outlined in black. Amino acid (aa) lengths are listed below the open reading frames. Reverse transcriptase PCR primers sets designed to bridge intergenic regions of open reading frames to assess the contents of the NTHI1441 transcript are shown above the operon diagram. (B) Ethidium bromide-stained agarose gel of reverse transcriptase PCRs. The endpoint reverse transcriptase PCR was performed with either 1 μg of RNA, 100 ng of DNA, or a water template. Reaction mixtures are indicated in the table below the respective lanes. A double-stranded DNA marker, with bases indicated below each marker, is located in the first and last lanes.

In the 86-028NP genome the NTHI1440, NTHI1443, and NTHI1444 ORFs are annotated as hypothetical conserved proteins. A protein-protein BLASTp analysis of the protein sequences identified NTHI1440 as a predicted YphA inner membrane protein with homology to the DoxX-COG2259 superfamily of unknown function. Low-confidence and low-percent-identity matches were found for the NTHI1440 protein sequence by the Phyre2 prediction (33). The NTHI1443 protein was identified as an AP2Ec-COG3220 superfamily of uncharacterized conserved proteins. NTHI1444 is a DUF2063-COG3219 superfamily protein with unknown function. However, the DUF2063 superfamily pfam09386 annotation identifies proteins of the family to have N-terminal structure homology to a putative DNA binding domain-transcriptional regulator in Neisseria gonorrhoeae. A Phyre2 structure prediction analysis supported the homology of NTHI1444 protein sequence to the solved crystal structure of the Neisseria gonorrhoeae FA1090 ngo_1945 protein product (33, 35). Phyre2 prediction of the NTHI1443 protein revealed homology to the solved crystal structure of the Haemophilus somnus hs_1138 protein product, which is annotated as a metal binding protein (PDB code 3BWW) (33).

Effects of NTHI1441 deletion on whole-cell lysates and growth.

The presence of NTHI1441 in an operon with a putative transcriptional regulatory gene suggests that deletion of the NTHI1441 ORF may affect expression of other NTHi genes. Sequencing of the NTHI1441 operon locus in the parent and NTHI1441 deletion (Δ1441) strain revealed no mutations in the surrounding operon sequence as a result of NTHI1441 deletion. We further determined the effects of NTHI1441 deletion in gross protein expression and growth. Deletion of NTHI1441 showed no differences in protein expression in whole-cell lysates of the strains grown in broth culture compared to the parent strain (Fig. 4A). The presence of the NTHI1441 protein band was not visible in Coomassie-stained whole-cell lysates, but polyclonal rabbit serum raised against a recombinantly expressed and purified NTHI1441 protein did however detect an ∼14 kDa protein band in the parent and complemented strain, which was absent in the NTHI1441 knockout strain (Fig. 4B). Deletion of NTHI1441 had no effect on growth of strains in laboratory medium (Fig. 4C).

FIG 4.

Effects of deletion and complementation of NTHI1441 on whole-cell lysate protein expression and growth. (A) Coomassie-stained whole-bacterial-cell lysates of the 86-028NP parent and Δ1441 and C′Δ1441 strains. Molecular weights are listed on the sides of the gel. (B) Detection of the NTHI1441 protein band, indicated by the white arrow, in an immunoblot of whole-cell lysates by polyclonal rabbit anti-NTHI1441 serum in the parent and complemented strain. (C) Twenty-four-hour growth curves of strains in supplemented brain heart infusion broth.

Effect of NTHI1441 deletion on host cell entry and intracellular survival.

Invasion of host cells is a dynamic process of both entry and intracellular survival. NTHi enters host cells by adhering to cells and co-opting endocytic pathways and resides within lysosomal compartments. NTHi resists phagolysosomal maturation by secretion of IgA protease and maintains a population of intracellular bacterial communities within the lysosomal compartment (7, 13, 15). We observed a deficit in the Δ1441 knockout mutant to invade cells 3 h postinfection (h.p.i.) (Fig. 1), which could be a result of a deficiency to enter cells, survive intracellularly, or both. We therefore assessed the adherence to and invasion of NCI-H292 cells at earlier and later time points to determine the effect of NTHI1441 ORF deletion on entry of host cells. The Δ1441 strain showed a reduced H292 cell invasion of 74.6% ± 3.7% (mean ± standard deviation of means at 1, 3, and 6 h.p.i.) over all three time points tested compared to that of the parent strain (Fig. 5A to C). When persisting in the lower airways of adults with COPD, NTHi colonizes both the bronchioles and alveoli. Thus, we also determined the adherence to and invasion of the A549 human alveolar epithelial cell line by strains at 1, 3, and 6 h.p.i. The Δ1441 strain showed a 78.6% ± 6.2% reduction in invasion of the A549 cell line (Fig. 5D to F). The complemented, C′Δ1441, and constitutively expressing complemented, C′Δ1441CE, strains had recovered host cell invasion at the 1-h.p.i. time point. At the later time points of 3 and 6 h, the complemented strains had greater mean invasion than that of the Δ1441 strain yet did not fully recover invasion to the parent strain levels (Fig. 5B, C, E, and F). We conclude that the deficiency in host cell invasion of the Δ1441 strain is at least in part due to decreased entry into host cells.

FIG 5.

Adherence and invasion by the parent strain, NTHI1441 mutant, and complemented strains in NCI-H292 and A549 cells. (A to C) Adherence to and invasion of NCI-H292 cells at 1 h, 3 h, and 6 h postinfection (h.p.i). (D to F) Adherence to and invasion of A549 cells at 1 h, 3 h, and 6 h.p.i. Percent adherence and invasion or invasion of NTHI1441 knockout mutant and complemented strains is normalized to that of the parent strain for each group. Bars represent means and error bars represent the SDs from at least three independently performed experiments. A Kruskal-Wallis test with Dunn’s multiple-comparison test was used to identify statistical differences. *, P value < 0.05; **, P value < 0.01; ****, P value < 0.0001.

The deficiency in the host cell invasion of the Δ1441 knockout could be due to intracellular survival in addition to decreased entry into host cells. Therefore, we evaluated the contribution of NTHI1441 to intracellular survival at 1, 3, and 6 h following infection of H292 cells with an intracellular survival assay (7, 13, 15). After 3 h of infection of H292 cells, medium was replaced with gentamicin-containing medium to kill extracellular bacteria and thus block additional entry. CFU of intracellular bacteria of each strain after incubation for 0, 1, 3, and 6 h were determined. While there were fewer intracellular bacteria of the Δ1441 knockout mutant and complemented strains than those of the parent at the 0-h time point, we observed no difference in intracellular survival among the parent, knockout mutant, and complemented strains (Fig. 6). We conclude that the deficiency of the Δ1441 strain to invade cells is due to decreased host cell entry (Fig. 5).

FIG 6.

Intracellular survival of the parent strain, NTHI1441 knockout mutant, and complemented strains in H292 cells. (A) Intracellular CFU per milliliter of strains at 0, 1, 3, and 6 h after treatment of infected NCI-H292 monolayers with medium containing gentamicin. (B) Intracellular survival over time. The y axis shows percentage of intracellular bacteria remaining from the 0-h time point. Bars represent the means of observations from three independently performed experiments and error bars represent the SDs. A Kruskal-Wallis test with Dunn’s multiple-comparison test was used to calculate differences among percentage of intracellular bacteria strains at the time points tested.

NTHI1441 expression of epitopes on the bacterial cell surface.

NTHI1441 was predicted to localize to multiple subcellular locations, including the periplasm, outer membrane, and extracellular sites. We observed that deletion of the NTHI1441 ORF in the Δ1441 strain resulted in decreased invasion of host cells (Fig. 5). Proteins with surface-exposed epitopes have potential to interact with host cell factors to confer adherence to and invasion of cells. We hypothesize that NTHI1441 has surface-exposed epitopes, which contributes to invasion of host cells. We tested for the presence of epitopes expressed on the bacterial cell surface with a whole-bacterial-cell enzyme-linked immunosorbent assay (wcELISA). Using polyclonal rabbit serum raised to a recombinantly expressed and purified NTHI1441 protein and adsorbed with the Δ1441 strain to remove background antibodies, we assessed the ability of the serum to detect parent versus NTHI1441 isogenic knockout and complemented strains. Significantly more NTHI1441 was detected by antiserum in the parent and complemented strains than in the Δ1441 strain (Fig. 7A). A control assay with an antibody to an intracellular antigen showed that cells remained intact under the conditions of the assay (Fig. S2).

FIG 7.

Expression of NTHI1441 epitopes on the bacterial cell surface. (A) Whole-cell ELISA with bacterial strains and dilutions of adsorbed polyclonal rabbit NTHI1441 antiserum. (B) Immunoblot of whole bacterial cell lysates with a dual monoclonal N- plus C-terminal six-histidine tag primary antibody. (C) Whole-cell ELISA with strains and monoclonal six-histidine-tagged antibody used for panel B. (D) Flow cytometry of whole bacterial cells of constitutively expressing non-tagged (C′Δ1441CE) and C-terminally six-histidine tagged (C-His-C′Δ1441) NTHI1441 assayed with FITC-labeled anti-six-His antibodies. (E) Mean fluorescence intensity (MFI) of flow cytometry with strains as noted. Antibodies are noted in the key to the right. Bars represent the means and error bars represent SDs from three independently performed experiments. A one-way analysis of variance (ANOVA) with Dunnett’s multiple comparison was used to identify statistical differences from the parent strain OD₄₅₀ readings of each serum dilution of ELISA. This analysis was also used to determine differences among the strain MFIs of each antibody group. **, P value < 0.01; ***, P value < 0.001; ****, P value < 0.0001.

To further characterize the NTHI1441 surface-exposed epitopes, we generated strains constitutively expressing either an amino- or carboxy-terminal six-histidine-tagged NTHI1441 (Fig. 7B). Tagging the termini or NTHI1441 permitted the use of monoclonal antibodies to determine whether the termini were expressed on the bacterial cell surface. Mouse dual monoclonal amino- and carboxy-terminal six-histidine antibodies were used to determine the termini of NTHI1441 expressed on the bacterial cell surface. Low expression of NTHI1441 was observed in whole-cell lysates (Fig. 4A). We therefore used strains constitutively expressing tagged and nontagged NTHI1441 to increase epitope abundance for detection by the monoclonal antibodies in wcELISA and flow cytometry. Both the amino- and carboxy-terminally tagged forms of NTHI1441 were detected on the bacterial cell surface, in contrast to strains without histidine-tagged proteins, in wcELISA (Fig. 7C).

Flow cytometry was used to detect surface exposure of the six-histidine-tagged NTHI1441, with gating on whole bacterial cell events. The histogram of the strain expressing NTHI1441 with the carboxy-terminal six-histidine tag showed a positive shift to the right in fluorescein isothiocyanate (FITC) fluorescence intensity compared to the strain constitutively expressing nontagged NTHI1441 when incubated with the anti-six-histidine monoclonal antibodies (Fig. 7D). The strain expressing NTHI1441 with the carboxy-terminal six-histidine tag also showed significantly greater mean fluorescence intensity when incubated with anti-six-histidine antibodies than did nontagged strains (Fig. 7E). No differences in mean fluorescence intensity of strains incubated with an isotype control or antibody diluent were observed (Fig. 7E).

In total, we conclude that NTHI1441 expresses epitopes on the bacterial cell surface. The observation that antibody to histidine tags detected the tagged amino and carboxy termini in wcELISA supports the idea that both the carboxy and amino termini of NTHI1441 are expressed on the bacterial cell surface. Results of flow cytometry suggest that the amino terminally tagged NTHI1441 protein may be intracellular. However, we interpret this observation as a difference in the sensitivity of between wcELISA and flow cytometry. We hypothesize that the absence of detection of amino terminally tagged NTHI1441 by flow cytometry is a result of lower expression of the tagged protein than of the carboxy-terminally targeted protein or a function of lower affinity and avidity of the monoclonal antibody to the N-terminal six-histidine tag (Fig. 7B and C).

Antibody response to NTHI1441 following exacerbations in adults with COPD.

To evaluate the development of antibodies to NTHI1441 following infection of the human respiratory tract, we assessed 22 paired pre-and postexacerbation serum samples from adults with COPD for the development of antibody to purified recombinant NTHI1441. Serum collected approximately 1 month after an exacerbation due to NTHi was compared to serum collected from the same individual approximately 1 month prior to the exacerbation. The cutoff for a significant change between pre- and postexacerbation serum IgG levels was determined as described previously (36, 37). Briefly, we used 10 control pairs of serum samples collected approximately 2 months apart (the same interval of time between the exacerbation samples) from adults with COPD who were clinically stable and had negative sputum cultures for NTHi. Serum samples were subjected to ELISA with purified recombinant NTHI1441. The percent change in values for optical density at 450 nm (OD₄₅₀) between serum samples was calculated. Control pairs of serum samples were found to have a 7.4% ± 6.2% change when tested with recombinant NTHI1441. A change in OD₄₅₀ of 12.5% represented the upper limit of the 99% confidence interval of the control samples. Thus, any increase of serum IgG in the exacerbation serum pairs of >12.5% was considered statistically significant. A statistically significant increase in 12 of the 22 pre- and postexacerbation serum pairs was observed (Fig. 8). We conclude that approximately half of COPD patients who experienced exacerbations caused by NTHi developed antibody responses to NTHI1441. This observation further supports the conclusion that NTHi expresses NTHI1441 during human lower airway infection.

FIG 8.

Results of ELISA showing percent changes in paired pre- and postexacerbation serum IgG to recombinant NTHI1441 from 22 patients with acute exacerbations of COPD caused by NTHi. The dotted horizontal line at 12.53% indicates the 99% confidence interval of a significant increase of serum IgG to NTHI1441 determined from 10 pairs of negative-control serum samples.

DISCUSSION

The NTHI1441 ORF, previously annotated as a conserved hypothetical protein, was identified in the present study through mining of annotated NTHi genomes. We show that the NTHI1441 ORF contributes to entry of NTHi into human respiratory bronchial and alveolar epithelial cells, with no observed effects on adherence or intracellular survival. We further show that the NTHI1441 protein expresses epitopes on the bacterial cell surface, including the amino and carboxy termini when tagged and constitutively expressed. After experiencing an exacerbation with NTHi, about half of adults with COPD generate serum IgG against recombinant NTHI1441 protein, indicating that NTHI1441 is expressed during infection and is a target a humoral immune response in adults with COPD. Together, these results support the idea that NTHI1441 is a conserved NTHi virulence factor involved in invasion of host cells during infection of the airways in persons with COPD. These properties suggest that NTHI1441 has potential as a target of preventative and therapeutic interventions to eliminate NTHi from the airways and thus reduce morbidity caused by NTHi infections in the setting of COPD.

Mining the annotated genomes of three NTHi strains isolated from the lower airways of adults with COPD curated 41 candidate proteins with (i) predicted bacterial cell surface subcellular location, (ii) antigenicity, and (iii) conservation among strains. NTHi proteins with surface-exposed epitopes interact with host cells and extracellular matrix proteins and are targets of the host immune response. We therefore mined for proteins predicted to localize to subcellular sites with accessibility to the bacterial cell surface, which included extracellular, outer membrane, and periplasmic sites and multiples of these locations. A growing body of evidence supports the idea that bacterial proteins with periplasmic or even intracellular predicted localizations have epitopes expressed on the bacterial cell surface (25–27, 38). These proteins have pleotropic functions and moonlight on the bacterial cell surface. The mechanisms of how these proteins traffic to and express epitopes on the bacterial surface have yet to be fully characterized. As a result, a limitation of bacterial protein subcellular location prediction programs is that they are unable to identify all cellular locations of proteins accurately. Proteins with predicted periplasmic subcellular localization were therefore included. Included among the initial 41 candidates identified were proteins with previously described roles in adherence to and invasion of host cells and candidate vaccine antigens of NTHi, such as PilA, OapA, P4, P5, and P6. Other NTHi proteins with known adherence and invasion function, antigenicity, and surface-exposed epitopes, like Hap, HMW1/2, and PE, were not among the candidates identified, because they did not meet either the antigenicity criteria or the percent similarity cutoffs (Table S1).

The top novel candidate identified by genome mining, NTHI1441, is a conserved and invariant gene during persistent infection of the COPD airways (Table 1 and Fig. 2). The NTHI1441 protein haplotypes contain 11 amino acid conserved repeats in the carboxy-terminal region of the sequence. Haplotypes contain either two or three repeats, with a majority of persistent NTHi strains containing 3 repeats (Fig. 2A and B). Differences in sequence repeats among the NTHI1441 haplotypes account in large measure for the percent similarity differences between haplotypes, but the amino acid sequence itself is highly conserved. Most of the persistent NTHi strains, 75%, are of one haplotype sequence, which has three repeat domains and has high percent similarity to the other haplotypes (Fig. 2C to E).

Repeat domains of proteins allow for arrangement of spatial and functional groups, which confer protein structure and/or interactions with target molecules (39). Two cysteine residues are located four residues apart in the repeat domains of NTHI1441 (Fig. 2B). Cysteine residues in proteins can form disulfide bonds under oxidative conditions, which confer protein structure and function (40–42). The cysteines present in the repeat domains of NTHI1441 may undergo intra- or interprotein disulfide bond formation, conferring structure and function to the mature NTHI1441 peptide. Additionally, cysteine residues in proteins can confer binding to substrates such as metals, although such cysteines are spaced by two amino acids (42, 43). Interestingly, the NTHI1441 native and recombinant protein migrates to approximately twice the predicted molecular weight of the mature protein in spite of subjecting the protein to denaturants and reducing agents (data not shown), suggesting that NTHI1441 forms a homodimer via bonds between cysteines in the repeat domain (Fig. 4B and Fig. 7B). Future investigations will test our hypotheses that the NTHI1441 repeat domains are involved in protein structure, multimerization, and function in the observed NTHI1441-dependent phenotypes.

NTHI1441 is expressed during growth in laboratory medium in a four-gene operon including the open reading frames NTHI1440, NTHI1443, and NTHI1444 (Fig. 3B and Fig. 4B). Inquiry of the proteins in the NTHI1441 operon components identified no specific function or role for the operon. NTHI1440 was annotated as an uncharacterized inner membrane protein, while NTHI1443 and NTHI1444 had homology to cytoplasmic metal binding and transcriptional regulator proteins, respectively. Though homologues of two of the NTHI1441 operon were identified, this did not provide insights into the function or role of the operon to investigate further. The presence of a putative transcriptional regulator is interesting, as this regulator may alter expression of the operon and or other NTHi genes. However, deletion of the NTHI1441 open reading frame did not result in differences in protein expression or growth in strains compared to the parent (Fig. 4A and C). Discovery of the function and conditions under which the NTHI1441 operon is expressed in future studies will help to further clarify the contribution of NTHI1441 and other operon components to infection of the COPD airways.

NTHi invades host cells by a “zipper” mechanism of membrane cytoskeletal rearrangements resulting in endocytosis. NTHi lacks secretion systems and effectors used by other bacterial pathogens in “trigger” mechanisms of receptor-independent induced host cell uptake of the bacterium (16, 44). NTHi binds to host platelet-activating growth factor receptor and β-glucan receptors, which results in receptor-mediated invasion of host cells (45–48). We hypothesize that attachment to host cell receptors, which induces endocytosis of the bacterium, is a crucial prerequisite to entry into host cells. Interestingly, we observed no difference in the ability of the NTHI1441 knockout to adhere to host cells compared to the parent strain (Fig. 5 and 6). The redundancies in several NTHi proteins to confer adherence to host cells likely compensates for the marginal effect of NTHI1441 attachment to its cognate host receptor. Therefore, the effect of NTHI1441-dependent invasion may be a result of receptor-mediated endocytosis. Alternatively, the role of NTHI1441 in invasion of host cells may be indirect or epistatic, modifying other NTHi genes or proteins involved in invasion of host cells. A zinc uptake operon knockout mutant of Moraxella catarrhalis showed a similar phenotype of reduced invasion of A549 cells that was independent of adherence to host cells. However, this was observed at one time point and intracellular survival was not assessed (49). We speculate that the NTHI1441 repeat domains discussed above play a role in the invasive phenotype, perhaps through protein folding and interacting with host cell receptors that mediate endocytosis of the bacterial cells.

We generated a complemented strain and a constitutively expressing complemented NTHI1441 knockout strain and showed that they restored or partially restored the phenotype in each assay. However, expression from the pSPEC1 plasmid and the constitutively expressing pSPEC1 derivative plasmid may not completely mimic NTHI1441 expression levels and function compared to those of the parent strain, accounting for the incomplete restoration to parent strain function in selected assays. For example, in the whole-cell ELISAs, the complemented strains expressed significantly greater NTHI1441 on the bacterial cell surface than did the parent strain (Fig. 7A), potentially altering accessibility of other NTHi adhesins and invasins to their cognate host cell receptors that induce bacterial uptake. In addition, increased NTHI1441 production may cause the protein to aggregate via the cysteine residues in the repeat domains, altering the functionality of the protein during invasion of host cells. Complementation restored invasion at 1 h postinfection (Fig. 5A and D) but did not completely restore invasion to the parent strain levels at 3 and 6 h postinfection. This result is likely due to altered expression of NTHI 1441 in the complemented strains as noted above. Alternatively, NTHI1441 may play a more important role in the early interactions with host cells than at later time points.

To determine the extent to which the NTHI1441 termini are expressed on the bacterial cell surface, we engineered strains constitutively expressing NTHI1441 with amino (N)- and carboxy (C)-terminal six-histidine tags (Fig. 7B and C). Whole-cell ELISA detected both the N- and C-terminally six-his-tagged NTHI1441 (Fig. 7C), whereas flow cytometry detected only the C-terminally six-histidine tagged construct on the bacterial cell surface (Fig. 7D and E). The inability to detect the N-terminally six-histidine-tagged NTHI1441 could be a result of decreased sensitivity of flow cytometry compared to that of wcELISA. Additional explanations for lack of detection of the N-terminally six-histidine-tagged construct is the lower expression of the N-terminally tagged NTHI1441 protein (Fig. 7B) and/or lower affinity and avidity of the monoclonal antibody to the N-terminal six-histidine tag than the C-terminal tag. NTHI1441 C-terminal domain bacterial surface exposure suggests that the repeat regions are accessible on the bacterial cell surface and interact with host cells. These hypotheses will be tested by future investigations. A limitation of this approach to consider is that the six-histidine tag may alter the native conformation of the protein on the cell surface. With this limitation in mind, these experiments provide evidence that both the N-terminal and C-terminal regions of NTHI1441 are expressed on the bacterial cell surface (Fig. 7C to E).

To assess whether adults with COPD made antibodies to NTHI1441 following infection by NTHi and to assess the expression of NTHI1441 in vivo, we assayed pre- and postexacerbation serum from adults with COPD who experienced an infection with NTHi in ELISA using purified NTHI1441. To optimize detection of human antibodies to NTHI1441, the recombinant C-terminally six-histidine-tagged NTHI1441 protein was purified from an NTHi background to aid in abundant expression and proper folding of the protein in the native NTHi strain. In previous purifications we identified that the NTHi ZnuA protein bound to the affinity resin used to purify recombinant NTHI1441, and we therefore generated a knockout of the znuA open reading frame in the expressing strain (data not shown).

We observed a significant increase in serum IgG against recombinant NTHI1441 in 12 of the 22 serum pairs from pre- to postinfection serum. These results indicate that NTHI1441 is expressed during infection of the human respiratory tract in vivo. In addition, the results show that NTHI1441 is a target of a humoral antibody response following infection in adults with COPD. It is somewhat surprising that despite the low expression of the protein in the nutrient-rich culture medium (Fig. 4), more than half of adults with COPD generated antibodies to NTHI1441 following infection. Other NTHi surface proteins that generate robust antibody responses are often highly expressed. For example, the major outer membrane protein P2 is one such protein that comprises over half of the total outer membrane proteins of NTHi (50) and is a major target of antibodies generated following NTHi infection (51–55). One possible explanation is the potent antigenicity of the mature NTHI1441 peptide predicted by VaxiJen (Table 1). The antigenicity score of NTHI1441 is quite high (2.0) compared to those of other NTHi proteins that are known to be immunogenic (P2, 0.72; P4, 0.82; P5, 0.73; and PilA, 0.74) (Table S1) (28, 56, 57). An alternative explanation is that NTHI1441 could be expressed at higher levels in the environment of the human respiratory tract than in nutrient-rich laboratory media. Because it is an exclusively human pathogen, it is especially important to assess host immune responses to potential vaccine candidates in humans (58, 59). Thus, a strength of this study is that the ELISA with patient serum provides true in vivo data about human immune responses during infection (Fig. 8).

In this study, we identified the NTHI1441 ORF through genome mining and characterized its translated amino acid sequence among 101 persistent NTHi strains from the airways of adults with COPD. The NTHI1441 ORF was conserved among strains and did not change during persistence for months to years in the airways of persons with COPD. An isogenic knockout of NTHI1441 has decreased invasion of human respiratory epithelial cells that is independent of adherence to cells or intracellular survival. The NTHI1441 protein is expressed during infection of the airways of persons with COPD and is the target of serum antibodies following infection in adults with COPD. We hypothesize that the surface-exposed epitopes of NTHI1441 interact with host cell receptors to stimulate invasion and uptake of NTHi by the host cells. Future investigations will test this hypothesis and determine the direct or indirect role of NTHI1441, and associated operon components, in invasion of host cells. Additionally, we aim to characterize the role of the NTHI1441 repeat domains in multimerization and function of the protein in the observed phenotypes. In total, we conclude the NTHI1441 is a newly identified NTHi invasin of host respiratory epithelial cells and is a potential target for preventative and therapeutic interventions.

MATERIALS AND METHODS

Prospective study of COPD.

Strains of NTHi and serum were isolated from patients with COPD as a part of a 20-year prospective study in Buffalo, NY, as described previously (60). The institutional review boards of the University at Buffalo and the Veterans Affairs Western New York Healthcare System approved this study; study participants provided written informed consent before enrollment. In brief, patients were seen monthly and at the time of suspected exacerbations. Serum and expectorated sputum samples were collected at each visit and were subjected to bacterial culture.

Genome mining and sequence analyses.

The genomes of NTHi strains 67P38H1, 6P24H1, and 11P6H were sequenced and annotated previously (4). PSORTb 3.0 was used to predict the subcellular localization of the translated open reading frames of each annotated genome (23). Proteins predicted to localize to the extracellular, outer membrane, periplasmic, and multiple-location subcellular fractions were evaluated for their probability of being antigens by VaxiJen v2.0 (28). Probable antigens were identified using a bacterial antigen threshold set to ≥0.55 to optimize the accuracy, sensitivity, and specificity of the prediction (28, 29). The respective nucleotide sequences of proteins identified as probable antigens were used to determine sequence similarity of the candidates among each of the three genomes. Nucleotide sequences were grouped and aligned and percent similarity was determined using Sequencher (Gene Codes 5.0.1-Build 8764) with the automatic assembly function, 80% minimum similarity, and 20-bp minimum match parameters. Open reading frames present in all three genomes analyzed with an average percent similarity of >85% were identified as candidates (Table S1). The three top candidate (NTHI1441, NTHI1101, and opa) open reading frame sequences from the 101 persistent NTHi strain pairs (202 isolates; first acquisition isolate and final clearance isolate for each strain) were extracted using a local BLAST (blast-2.2.29) database made with genomes of persistent strains (4, 32). The 86-028NP gene sequence of top candidates was used as the query to call sequences from the genomes. Top candidate sequences from the persistent NTHi strains were analyzed for presence, conservation, and haplotype distribution as described previously by Pettigrew et al. (4) (Table 2 and Fig. 2).

TABLE 2.

Strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
NTHi strains
67P38H1	Strain annotated genome used for genome mining	4
6P24H2	Strain annotated genome used for genome mining	4
11P6H	Strain annotated genome used for genome mining	4
86-028NP	Parent strain	64
Δ1441 mutant	86-028NP NTHI1441::apha3 Kanr	This study
Δ1101 mutant	86-028NP NTHI1101::apha3 Kanr	This study
Δopa mutant	86-028NP opa::apha3 Kanr	This study
C′Δ1441	Δ1441 strain complemented with p1441spec	This study
C′Δ1441CE	Δ1441 strain complemented with p1441CEspec. Constitutively expresses NTHI1441.	This study
N-His-C′Δ1441	Δ1441 complemented with pN6×His1441CEspec	This study
C-His-C′Δ1441	Δ1441 complemented with pC6×His1441CEspec	This study
C-His-C′Δ1441ΔZnuA	C-His-C′Δ1441 znuA::cat1 Cmr. Purification of C-terminally 6×His-tagged NTHI1441.	This study
86-028NP GFP	86-028NP pGM1.1. Flow cytometry gating.	67
3P14H1	Representative strain of NTHI1441 haplotype	4
1P16H5	Representative strain of NTHI1441 haplotype	4
19P68H7	Representative strain of NTHI1441 haplotype	4
47P38H1	Representative strain of NTHI1441 haplotype	4
72P12H1	Representative strain of NTHI1441 haplotype	4
56P1H1	Representative strain of NTHI1441 haplotype	4
96P6H3	Representative strain of NTHI1441 haplotype	4
E. coli strains
TOP10	TOP10 OneShot chemically competent cells	Invitrogen
TOP10-NTHI1441	Top10 with pET100-1441	This study
BL21 Star DE3	BL21 Star DE3 OneShot chemically competent cells	Invitrogen
BL21*DE3-1441	BL21 Star DE3 with pET100-1441	This study
Plasmids
pUCK18	Kanamycin resistance cassette	69
pET-100 Topo	Topoisomerase cloning expression plasmid	Invitrogen
pET100-1441	NTHI1441 mature peptide sequence cloned into pET-100 Topo	This study
pSPEC1	E. coli-to-NTHi shuttle vector for complementation	70
p1441spec	pSPEC1 with NTHI 1441 ORF with 68 bp upstream	This study
pKM1.1	Backbone for constitutive gene expression in NTHi	65
p1441CEspec	pKM1.1 mCherry::NTHI1441 ORF	This study
pN6×His1441CE	p1441CEspec with 6-histidine tag cloned after signal peptide sequence	This study
pC6×His1441CE	p1441CEspec with 6-histidine tag cloned before stop codon sequence	This study
pACYC184	Chloramphenicol acetyltransferase (cat1) gene for cloning	New England BioLabs

Bacterial growth.

Bacterial strains used in this study are listed in Table 2. NTHi and Escherichia coli strains were cultured using standard techniques described previously (4, 61, 62). NTHi was grown on chocolate agar plates (Remel) at 35°C and 5% CO₂ or in brain heart infusion broth supplemented with 10 μg/ml of β-nicotinamide adenine dinucleotide (β-NAD) (Sigma) and 10 μg/ml of hemin (Sigma) at 37°C shaking at 225 rpm with or without antibiotics (sBHI). E. coli used for cloning, protein expression, and purification was grown on Luria-Bertani (LB) agar plates or in LB (Sigma) broth with the appropriate antibiotics.

Reverse transcriptase PCR.

RNA was isolated from strain 86-028NP grown in sBHI to an OD₆₀₀ of ∼1.00 with the QIAshredder and RNEasy kits (Qiagen) and then treated with RQ1 DNase (Promega) as per the manufacturer’s instructions. DNA was purified from overnight broth cultures with the Promega Wizard genomic DNA kit as per the manufacturer’s suggested protocol (Promega). The nucleic acid concentrations were determined by measuring absorbances at 260 and 280 nm using NanoDrop 2000c (Thermo Scientific). Reverse transcriptase primer sequences are listed in Table S2. The endpoint reverse transcriptase PCR (RT-PCR) was performed with a OneStep RT-PCR kit with either 1 μg of RNA, 100 ng of DNA, or a water template with the manufacturer’s suggested mixture and PCR conditions (Qiagen). A reaction mixture was made and run without reverse transcriptase to ensure the absence of DNA contamination of the RNA template samples. The ethidium bromide-stained 2% agarose gel of reverse transcriptase PCRs included Novagen 50- to 2,000-bp markers (Millipore-Sigma) (Fig. 3).

Recombinant DNA for cloning and protein expression.

Isogenic knockout mutants of open reading frames were generated by allelic exchange as described previously (4, 36). Plasmids were purified from overnight broth cultures using the QIAprep Spin Miniprep kit as per the manufacturer’s protocol (Qiagen). Recombinant fragments containing antibiotic resistance cassettes flanked by ∼1 kb of upstream and downstream sequences of open reading frames were generated by sequence overlap extension PCR using Q5 polymerase high-fidelity master mix (New England BioLabs). Cells of 86-028NP were made competent by starvation and transformed with 150 ng of the transforming fragment (63). Antibiotic-resistant transformants of open reading frame knockouts were confirmed by sequencing of the transformed gene locus.

The isogenic NTHI1441 knockout (Δ1441) strain was complemented in trans with the p1441spec plasmid. The p1441spec plasmid was generated by ligating the 86-028NP NTHI1441 open reading frame and 68 bases of upstream sequence (bases to the stop codon of NTHI1440) into the multiple-cloning site of the E. coli-to-NTHi shuttle vector pSPEC1 (64). The NTHI1441 open reading frame and upstream sequence were amplified from 86-028NP DNA with primers containing 5′ BamHI and 3′ EcoRI restriction enzyme sites. The pSPEC1 plasmid and amplified NTHI1441 cloning fragment were digested with BamHI and EcoRI FastDigest restriction enzymes (Invitrogen). The digested plasmid was gel purified with the Monarch DNA gel extraction kit (New England BioLabs), while the digested PCR fragment was purified with a QIAquick PCR purification kit (Qiagen) using both the manufacturers’ instructions. The PCR fragment was ligated into the plasmid at a 1:3 vector-to-insert ratio with T4 DNA ligase (Invitrogen), after which the reaction was used to transform OneShot TOP10 chemically competent E. coli cells (Invitrogen). Plasmid from antibiotic-resistant transformants selected was purified and sequenced to ensure the proper insertion sequence. The Δ1441 strain was made electrocompetent and transformed with 1 μg of plasmid, and an antibiotic-resistant clone was selected as the C′Δ1441 strain, as described previously (36).

The complemented mutant constitutively expressing NTHI1441, C′Δ1441CE, was generated by electroporation of the p1441CEspec plasmid. The p1441CEspec plasmid was generated by replacement of the mCherry sequence of the pKM1.1 plasmid with the NTHI1441 open reading frame by BamHI and EcoRI restriction enzyme digestion, ligation, transformation, and electroporation, as described above with primers listed in Table S2 (65). The complemented strains constitutively expressing amino- and carboxy-terminally six-histidine-tagged NTHI1441 were made by electroporation of the pN6×His1441CE and pN6×His1441CE plasmids into the Δ1441 knockout strain. The plasmids expressing six-histidine-tagged NTHI1441 were made by site-directed mutagenesis of p1441CEspec with primers listed in Table S2 and the Q5 site-directed mutagenesis kit per the manufacturer’s suggested protocol (New England BioLabs). Expression of the six-histidine-tagged NTHI1441 proteins in the complemented Δ1441 knockout strains N-His-C′Δ1441 and C-His-C′Δ1441 was confirmed by immunblotting with a mouse dual monoclonal (4E3D10H2/E3) anti-N- and anti-C-terminal six-His antibody (Invitrogen) (Fig. 7B).

The NTHI1441 open reading frame sequence, without the signal peptide, was cloned into the pET-100 Directional TOPO vector and transformed into OneShot TOP10 cells (Invitrogen). Plasmid pET100-1441 was purified from an overnight culture of an antibiotic-resistant clone of TOP10-NTHI1441 and transformed into BL21 Star (DE3) OneShot to make strain BL21*DE3-1441. This strain was used for expression and purification of recombinant NTHI1441 from E. coli. The znuA open reading frame of C-His-C′Δ1441 was replaced with the chloramphenicol resistance cassette from the pACYC184 plasmid, using methods described above, to generate strain C-His-C′Δ1441ΔZnuA. The C-His-C′Δ1441ΔZnuA strain was used for expression and purification of the C-terminally six-histidine-tagged NTHI1441 protein from NTHi. Primers used for recombinant DNA techniques and sequencing are listed in Table S2.

Expression and purification of recombinant NTHI1441 protein.

Recombinant NTHI1441 was expressed and purified from E. coli for generation of polyclonal NTHI1441 antiserum. The E. coli strain BL21*DE3-1441 from an overnight plate containing LB and 200 μg/ml of carbenicillin (Sigma) was seeded into LB broth with 200 μg/ml of carbenicillin and grown overnight. The overnight culture was used to seed TB broth (MBio) with 200 μg/ml of carbenicillin and grown at 37°C and 225 rpm to an OD₆₀₀ of 0.9. The culture was cooled on ice before addition of isopropyl β-d-1-thiogalactopyranoside (Sigma) to a final concentration of 1 mM. Induction was carried out at 16°C with shaking at 175 rpm for 24 h. Cells were collected by centrifugation at 3,900 × g and 4°C for 15 min, the supernatant was discarded, and the pellet was frozen at –20°C.

Cells were thawed on ice and then incubated with five pellet volumes of 1-mg/ml lysozyme from chicken egg white (Sigma) in Tris-EDTA buffer (pH 8.0; 20 mM Tris-HCl and 2 mM EDTA; Fisher Chemical) for 30 min at room temperature with nutation. Three volumes of lysis buffer (pH 7.8; phosphate-buffered saline [PBS] with 1% [vol/vol] Triton X-100 [J. T. Baker-Avantor] and 20 mM β-mercaptoethanol [Bio-Rad]) were added to the lysozyme-treated pellet before sonication three times on ice. Each sonication was carried out for 30 s at a 50% duty cycle, output of 10, with a medium sonication horn of a Bronson Sonifier. Lysate was centrifuged at 23,700 × g and 4°C for 20 min. The supernatant was collected and mixed 1:1 with wash buffer (pH 7.8; PBS with 20 mM imidazole; Sigma). A 2-ml volume of Talon metal affinity resin (Clonetech) slurry, 1 ml of resin volume, was equilibrated with five resin volumes of PBS (pH 7.8) at room temperature for 10 min. The lysate with wash buffer was added to the equilibrated resin and allowed to bind for 1 h at room temperature with nutation. After binding, the sample was centrifuged at 1,200 × g and 4°C for 5 min. The supernatant was decanted and the resin was washed three times with 50 resin volumes of wash buffer. The recombinant NTHI1441 protein was eluted from the resin three times with two resin volumes of elution buffer (PBS with 200 mM imidazole). The eluates were combined and concentrated with an Amicon ultracentrifugal filter unit with a 10-kDa nominal molecular weight limit (NMWL) (Millipore-Sigma) before being dialyzed three times overnight against 1,000 volumes of PBS (pH 7.8) at 4°C in SnakeSkin pleated 3,500-molecular-weight-cutoff (MWCO) dialysis tubing (Bio-Rad). The recombinant NTHI1441 protein concentration was determined with the Pierce bicinchoninic acid (BCA) protein assay (Thermo Scientific).

A C-terminally six-histidine-tagged recombinant NTHI1441 protein was expressed and purified from strain C-His-C′Δ1441ΔZnuA for use in ELISA with pre- and postexacerbation serum. Cells from an overnight broth culture of C-His-C′Δ1441ΔZnuA in sBHI with 15 μg/ml of kanamycin, 200 μg/ml of spectinomycin, and 2 μg/ml of chloramphenicol (Sigma) were collected by centrifugation at 3,900 × g and 20°C for 15 min. The supernatant was discarded and the pellet was frozen at –20°C. Purification of the C-terminally six-histidine-tagged recombinant NTHI1441 protein from thawed cells was done as described for the E. coli purification, with slight modifications. The recombinant NTHI1441 protein was eluted from the resin five times with three resin volumes of elution buffer. The eluates were combined and filtered through with a 30-kDa-NMWL Amicon ultracentrifugal filter unit (Millipore-Sigma) to remove a higher-molecular-weight contaminating protein. Flowthrough was then concentrated with a 3-kDa-NMWL Amicon ultracentrifugal filter unit (Millipore-Sigma) before being dialyzed three times overnight against 1,000 volumes of PBS (pH 7.6) at 4°C. The recombinant NTHI1441 protein concentration was determined with the Pierce BCA protein assay (Thermo Scientific). A Pierce Endotoxin Chromogenic Quant kit (Thermo Scientific) was used to quantify the amount of endotoxin contamination in the purified protein. The purified C-terminally six-histidine-tagged recombinant NTHI1441 protein purified from NTHi was determined to have 7.1 × 10⁻⁵ endotoxin units per microgram of protein.

Production of NTHI1441 polyclonal rabbit antiserum.

Purified NTHI1441 protein from E. coli was sent to Covance (Denver, PA) for antibody production in New Zealand White rabbits. Specific-pathogen-free rabbits were immunized with 250 μg of purified protein with Freund’s adjuvant subcutaneously on day 0, followed by three subcutaneous boosts with 125 μg with Freund’s adjuvant on days 21, 42, and 63. Final bleed serum was collected on day 73.

Adsorption of NTHI1441 antiserum.

Serum was adsorbed to the Δ1441 strain to remove cross-reacting antibodies to bacterial surface-exposed NTHi proteins. Serum adsorption was performed as described previously, with slight modifications (66). An overnight broth culture of Δ1441 grown in sBHI with 15 μg/ml of kanamycin was separated into 50-ml aliquots, and cells were pelleted by centrifugation at 3,900 × g and 4°C for 20 min. Cell pellets were washed three times with 50 ml of ice-cold PBS-calcium-magnesium (PCM) buffer (PBS with 1.25 mM CaCl₂ and 0.50 mM MgCl₂). The polyclonal rabbit NTHI1441 antiserum and preimmunized rabbit serum were heat inactivated at 56°C for 30 min and diluted 1:100 in 3 ml of PCM buffer. Serum was adsorbed by resuspending washed cell pellets with the diluted serum and incubating them on ice for 30 min. Cells were then pelleted by centrifugation, as described above, and the supernatant was adsorbed to four additional cell pellets. After the final adsorption, the adsorbed serum supernatant was filter sterilized through a Whatman Puradisc 25-mm sterile 0.2 μM polyehtersulfone membrane ES filter (GE Healthcare) and kept at 4°C until use. Three independently adsorbed batches of the polyclonal rabbit NTHI1441 antiserum and preimmunized rabbit serum were made for each replicate of whole-cell ELISA.

Whole-cell ELISA.

Whole-cell ELISA was performed as described previously (66). Strains were grown to mid-log phase in sBHI broth, harvested by centrifugation at 1,300 × g and 20°C for 10 min, and washed three time with 1 volume of PBS (pH 7.4). The final bacterial pellet was resuspended to an OD₆₀₀ of 0.20 in PBS (pH 7.4), and 100 μl (approximately 5 × 10⁷ CFU) was pipetted into duplicate wells of an Immulon-4HBX flat-bottom ultrahigh-binding polysterene microtiter 96-well plate (Thermo Scientific). No-coat wells containing 100 μl of PBS (pH 7.4) were used to determine background binding of each serum dilution tested. After overnight incubation at 4°C, samples were aspirated from wells and washed three times with 300 μl of whole-cell ELISA plate wash buffer (pH 7.0; wcPWB) (3.8 mM NaH₂PO₄, 12.7 mM Na₂HPO₄, 126.6 mM NaCl, and 0.05% Tween 20; Fisher Chemical). Wells were blocked for 1 h with 300 μl of wcPWB with 3% nonfat dry milk at room temperature. Wells were then washed three times with wcPWB, and 100 μl of serum or antibody in PWB with 1% nonfat dry milk was added to bacterium-containing or no-coat wells. Plates were incubated at 37°C and 5% CO₂ for 2 h, after which wells were washed three times with wcPWB. To measure bound serum IgG, horseradish peroxidase-conjugated goat anti-rabbit IgG (Kirkegaard & Perry Laboratories antibodies; SeraCare), diluted 1:3,000 in assay diluent (wcPWB containing 3% heat-inactivated goat serum), was added to the wells. To measure bound mouse dual monoclonal (4E3D10H2/E3) anti-N- and C-terminal six-histidine tag antibodies (Invitrogen), horseradish peroxidase-conjugated goat anti-mouse IgG (Kirkegaard & Perry Laboratories antibodies; SeraCare) diluted 1:3,000 in assay diluent was used. After a 1-h incubation at room temperature, wells were washed three times with wcPWB and color was developed with 100 μl of ELISA developer (0.1 mg of 3,3′,5,5′-tetramethylbenzidine-dimethyl sulfoxide [Bio-Rad], 0.02% hydrogen peroxide per ml in 0.1 M C₂H₃NaO₂ adjusted to pH 4.5 with citric acid [Fisher Scientific]). After a 15-min incubation at room temperature, the reaction was stopped by the addition of 4 N H₂SO₄ (Fisher Scientific). The OD₄₅₀ was read by an iMark microplate reader (Bio-Rad) (Fig. 7A and C).

ELISA.

Wells of an Immulon-4HBX flat-bottom ultrahigh-binding polystyrene microtiter 96-well plate were coated overnight at room temperature with 1 μg of C-terminally six-histidine-tagged recombinant NTHI1441 protein purified from NTHi in 100 μl of coating buffer (pH 9.6; 0.1 M Na₂CO₃ and 0.1 M NaHCO₃). The ELISA protocol was carried out as described for the whole-cell ELISA, with slight modifications. The plate wash buffer for the ELISA contained 0.15% Tween 20 (PWB). All buffers described for the whole-cell ELISA protocol containing wcPWB instead used PWB. After overnight incubation, wells were washed once with PWB. After blocking, wells were washed twice with PWB. To measure bound human serum IgG, horseradish peroxidase-conjugated goat anti-human IgG (λ) chain (Kirkegaard & Perry Laboratories antibodies; SeraCare), diluted 1:3,000 in assay diluent, was used (Fig. 8).

Coomassie-stained whole-cell lysates and immunoblotting.

Strains were grown in sBHI with or without the appropriate antibiotic for the strain to an OD of 1.0. A 10-ml volume of each culture was centrifuged at 3,900 × g and 20°C for 10 min to collect cells. The supernatant was discarded and cells were resuspended in 250 μl of H₂O to which 250 μl of 2× Laemmli sample buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol, 0.004% bromphenol blue, and 0.125 M Tris HCl; Bio-Rad) was added and boiled for 10 min. A 7-μl volume of Precision Plus Protein dual-color molecular weight marker (Bio-Rad), Spectra multicolor low-range protein ladder (Thermo Scientific), and lysates were resolved on Novex 10 to 20% tricine protein gels (Invitrogen). For staining, gels were stained in Coomassie stain (25% methanol, 10% glacial acetic acid, and 0.05% [wt/vol] Coomassie brilliant blue R-250 in H₂O) and destained in 10% methanol and 7.5% glacial acetic acid in H₂O (Fig. 4A). Immunoblotting of whole-cell lysates resolved on gels was performed as described previously (4). NTHI1441 polyclonal rabbit antiserum and mouse dual monoclonal (4E3D10H2/E3) anti-N- and C-terminal six-His antibodies were used at a 1:1,000 dilution. Both peroxidase-labeled goat anti-mouse and rabbit IgG secondary antibodies were used at a dilution of 1:1,000 (Kirkegaard & Perry Laboratories antibodies; SeraCare).

Growth curves.

Growth curves of strains were analyzed using the Bioscreen C automated growth curve analysis system (Oy Growth Curves AB Ltd., Helsinki, Finland). Bacterial strains from an overnight broth culture were diluted 1:1,000 in fresh sBHI, and 200 μl was seeded into plate wells. Bacteria were grown at 37°C in atmosphere for 24 h with low shaking speed and medium amplitude.

Adherence and invasion.

Bacterial strain adherence to and invasion of human respiratory epithelial cells were determined using a gentamicin protection assay as described previously (15). In brief, confluent monolayers of cells were inoculated with bacteria grown to mid-log phase and diluted in tissue culture medium to a multiplicity of infection of ∼2 (CFU of bacteria to cells). Infected monolayers were centrifuged for 5 min at 175 × g to settle bacteria onto cells. After the desired hours postinfection, medium was collected from infected monolayers to enumerate planktonic non-cell-associated bacteria. Monolayers were then washed, and fresh medium or fresh medium containing 50 μg/ml of gentamicin was added to infected monolayers and incubated for 1 h. Wells replenished with fresh medium were used to enumerate adherent and invading bacteria, while those receiving gentamicin-containing medium were used to enumerate intracellular bacteria. Monolayers were washed and harvested as described previously. In the screen of top candidates, the percent adherence and invasion or percent invasion was calculated by dividing the CFU per milliliter of either adherent and invading or invading bacteria by the total bacteria per well and multiplying the resulting value by 100. The total bacteria per well was the summation of the adherent and invading and planktonic non-cell-associated bacteria. The adherence and invasion of Δ1441 knockout mutant and complemented strains were normalized by dividing percentages by the percentage of the 86-028NP parent strain (Fig. 1 and 5).

Intracellular survival.

Intracellular survival of strains was determined using a modification of the adherence-and-invasion assay described above. Duplicate wells containing monolayers of NCI-H292 monolayers were infected with strains as before. Three hours postinfection, monolayers were washed three times with 1 volume of PBS. After the washing, medium containing 50 μg/ml of gentamicin was added to each well and incubated for 1 h at 37°C in 5% CO₂. After this step, monolayers were washed, treated, and harvested as described above to enumerate starting intracellular CFU per milliliter of bacteria, representing the 0-h time point. Medium containing 50 μg/ml of gentamicin was added to the remaining infected monolayers for the desired times. At each time point cells were washed, treated, and harvested to enumerate intracellular CFU per milliliter of each bacterial strain (Fig. 6).

Flow cytometry.

Strains were grown to mid-log phase and then centrifuged at 1,300 × g and 20°C for 10 min. Cell pellets were washed once with 1 volume of PBS–1% (wt/vol) bovine serum albumin (Sigma), pH 7.4 (PBS-BSA), and then resuspended to an OD₆₀₀ of 0.20. One milliliter (∼5 × 10⁸ CFU) was pelleted by centrifugation at 5,000 × g and 20°C for 5 min, and pellets were resuspended in 100 μl of either mouse dual monoclonal anti-six-His 4E3D10H2/E3 (Invitrogen) or isotype control mouse IgG MOPC-21 (BioLegend). Primary antibodies were diluted in PBS-BSA to a final concentration of 0.2 μg/μl. Samples were incubated standing at room temperature for 1 h. After incubation, cells were centrifuged at 5,000 × g and 20°C for 5 min and washed 3 times with 5 volumes of PBS-BSA. Washed cells were resuspended in 100 μl of 200-μg/μl FITC-conjugated goat anti-mouse IgG polyclonal serum (SeraCare) secondary antibody diluted in PBS containing 1% (vol/vol) heat-inactivated goat serum (pH 7.4). Samples were incubated standing at room temperature in the dark for 1 h. Cells were centrifuged as before and washed twice with 5 volumes of PBS (pH 7.4). The final cell pellets were resuspended in 1 ml of PBS, and FITC intensity was measured by using a BD LSRFortessa flow cytometer. Gating for cell events was set using green fluorescent protein (GFP)-expressing strain 86-028NP (Table 1) (67). Background FITC autofluorescence of each strain was measured with samples incubated in the respective primary and secondary antibody buffers without antibodies. FlowJo was used to analyze the flow cytometry data (Fig. 7D and E).

Statistical analyses.

Statistical analyses were performed in GraphPad Prism (8.1.0) and are described in the figure legends. Comparisons with P values of <0.05 were regarded as statistically significant.