Chronic Infection by Mucoid Pseudomonas aeruginosa Associated with Dysregulation in T-Cell Immunity to Outer Membrane Porin F

Abstract

Rationale: Pseudomonas aeruginosa (PA) is an environmental pathogen that commonly infects individuals with cystic fibrosis (CF) and non-CF bronchiectasis, impacting morbidity and mortality. To understand the pathobiology of interactions between the bacterium and host adaptive immunity and to inform rational vaccine design, it is important to understand the adaptive immune correlates of disease.

Objectives: To characterize T-cell immunity to the PA antigen outer membrane porin F (OprF) by analyzing immunodominant epitopes in relation to infection status.

Methods: Patients with non-CF bronchiectasis were stratified by frequency of PA isolation. T-cell IFN-γ immunity to OprF and its immunodominant epitopes was characterized. Patterns of human leukocyte antigen (HLA) restriction of immunodominant epitopes were defined using HLA class II transgenic mice. Immunity was characterized with respect to cytokine and chemokine secretion, antibody response, and T-cell activation transcripts.

Measurements and Main Results: Patients were stratified according to whether PA was never, sometimes (<50%), or frequently (≥50%) isolated from sputum. Patients with frequent PA sputum-positive isolates were more likely to be infected by mucoid PA, and they showed a narrow T-cell epitope response and a relative reduction in Th1 polarizing transcription factors but enhanced immunity with respect to antibody production, innate cytokines, and chemokines.

Conclusions: We have defined the immunodominant, HLA-restricted T-cell epitopes of OprF. Our observation that chronic infection is associated with a response of narrowed specificity, despite strong innate and antibody immunity, may help to explain susceptibility in these individuals and pave the way for better vaccine design to achieve protective immunity.

At a Glance Commentary

Scientific Knowledge on the Subject

Non–cystic fibrosis (CF) bronchiectasis is a chronic, progressive lung disease associated with impaired lung function, sputum production, and recurrent infection. Chronic infection with the gram-negative bacteria Pseudomonas aeruginosa (PA) is associated with higher morbidity and mortality. It is unclear why, among susceptible patient groups, some individuals develop chronic PA infections and others do not. Chronic infection with PA is associated with transcriptional changes in bacterial gene expression, such as those associated with the development of a mucoid phenotype.

What This Study Adds to the Field

In patients with non-CF bronchiectasis, we looked at immunologic differences associated with differential susceptibility to chronic infection with PA. Patients from whom PA was frequently cultured from sputum tended to be infected with mucoid isolates and showed altered T-cell immunity to an immunodominant PA antigen: T-cell immunity was much more narrowly targeted on the PA protein outer membrane porin F, despite strong innate and antibody immunity. These findings help to explain susceptibility to chronic infection and should prove valuable for vaccine design.

Pseudomonas aeruginosa (PA) is an environmental pathogen, able to exploit predisposing host conditions to cause chronic infection (1, 2). Chronic respiratory PA colonization of patients with non–cystic fibrosis (CF) bronchiectasis impacts on morbidity and mortality (3, 4). Features of host–pathogen interplay during PA infection pose novel and important challenges. Chronic infection can be long term, with the pathogen locating in hypoxic masses in the airway lumen and an associated morphotypic shift to a biofilm state involving generation of a protective, alginate, exopolysaccharide coat (5, 6). This state can persist in the face of high titers of neutralizing antibody (6). The biofilm state is an adaptation presumed to protect the pathogen from host immunity (7, 8), but little is known about adaptive immunity to PA during long-term infection.

As with many other large, variable bacterial genomes, there is a paucity of data available on detailed mapping of functional, protective antigen targets; often there are clear data on serodominant targets, but ability to activate antibody can differ from the ability to elicit protective immunity (9). Various PA antigens have been considered in mouse, rat, and nonhuman primate, as well as trials in human CF or non-CF vaccinees (10–16). In a phase III vaccine trial of patients with noncolonized CF, a flagellin vaccine roughly halved the number of PA infections in the recipient cohort (10). Studies have considered the lipopolysaccharide O antigens, flagellin, and the outer membrane porins F and I (OprF and I). OprF is a major outer membrane protein that is conserved across clinical isolates (17–19). In a comparison of aerobic and anaerobic PA biofilms, OprF is up-regulated 39-fold during anaerobic (relative to aerobic) biofilm growth (6). Studies have considered the immunogenicity and protective efficacy of OprF (20–25), commonly focusing on OprF antibodies.

Non-CF bronchiectasis offers a highly relevant clinical group in whom to map the immune events predisposing to chronic PA colonization. A disease of chronic lung inflammation, damage, and infection, it is defined by a pathologic endpoint: irreversible, abnormal, bronchial dilatation with chronic airway inflammation, reached through diverse etiologies (26, 27). The pathobiology suggests interplay between immunogenetic susceptibility, immune dysregulation, lung damage, and chronic bacterial infection (27–30). Damaged epithelium impairs mucus removal, allowing mucus accumulation that facilitates bacterial infection, promoting chronic colonization associated with chronic cough and sputum production, recurrent chest infections, and airflow obstruction. Chronic infections can encompass Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, and PA as well as nontuberculous mycobacteria, and Aspergillus fumigatus (26, 27). Genetic associations suggest involvement of genes controlling adaptive immunity and innate immunity (28–30). We here analyzed adaptive immunity to PA in a cohort of 57 patients with bronchiectasis stratified according to the frequency with which PA was cultured from monthly, longitudinal sputum samples.

Methods

Patient Cohort and Clinical and Microbiologic Investigations

A total of 57 unrelated clinically stable subjects (mean age, 58 ± 29 yr [2 SD]; 28% male) with a diagnosis of non-CF bronchiectasis were recruited from the Royal Brompton Hospital (Table 1). Patients were clinically diagnosed as per the British Thoracic Society bronchiectasis guidelines (28, 31). Sputum samples were collected monthly for 6 months and sent for microscopy and culture by standard microbiology. Peripheral blood samples were collected. Patients were stratified into three groups: (1) never (53 ± 31 yr [2 SD]; 29% male), (2) less than 50% (66 ± 17 yr [2 SD]; 8% male), and (3) more than 50% (61 ± 26 yr [2 SD]; 43% male) culture positive for PA in sputum samples collected. The study was approved under ethics code 10/H0801/53 with full, informed, patient consent.

Table 1.

Study Subjects

Underlying Cause of Non–Cystic Fibrosis Bronchiectasis	Patient ID	Age (yr)	Sex (M/F)	HLA-Cw		HLA-DR		Percentage of Sputum Samples Culture Positive for Pseudomonas aeruginosa (0%, <50%, ≥50%)	Other Pathogens Isolated in Sputum
PCD	B01	57	F	6	7	11	15	0%	Haemophilus influenzae, Streptococcus pneumoniae
CVID	B03	66	F	7	8	1	17	0%	H. influenzae, S. pneumoniae, Moraxellacatarrhalis
Postinfective	B10	71	F	10	16	4	7	0%	URT flora, Staphylococcus aureus
Idiopathic	B09	62	F	7	7	15	17	0%	—
Postinfective	B18	76	M	6	10	13	14	0%	—
Idiopathic	B19	40	M	1	4	4	12	0%	—
Idiopathic	B25	75	M	4	7	1	1	0%	—
Idiopathic	B26	77	F	12	14	4	10	0%	URT flora
Idiopathic	B28	64	F	5	7	17	4	0%	URT flora
Postinfective	B29	31	F	4	15	14	15	0%	—
Idiopathic	B33	42	M	5	7	4	13	0%	URT flora
PCD	B35	40	F	7	7	1	15	0%	S. aureus
PCD	B36	59	F	9	7	11	13	0%	URT flora, Stenotrophomonasmaltophilia
PCD	B38	22	F	6	12	17	11	0%	H. influenzae, S. pneumoniae, M. catarrhalis, S. aureus
Postinfective	B39	51	F	4	5	1	14	0%	URT flora, S. aureus
Youngs	B45	59	M	1	4	4	15	0%	H. influenzae
Postinfective	B46	29	F	8	10	4	7	0%	S. pneumoniae
ABPA	B27	56	F	10	6	17	13	0%	URT flora, Aspergillus fumigatus
ABPA	B55	60	M	10	10	7	15	0%	URT flora
CVID	B56	41	F	1	5	1	11	0%	URT flora, H. influenzae
Postinfective	B54	59	F	6	10	7	13	0%	URT flora, H. influenzae
ABPA	B63	27	F	7	7	4	11	0%	URT flora, S. aureus
Idiopathic	B77	64	M	7	12	15	17	0%	URT flora, M. catarrhalis
PCD	B79	33	F	2	12	4	13	0%	URT flora, M. catarrhalis, H. influenzae
Idiopathic	B83	55	M	5	7	7	15	0%	URT flora
Postinfective	B73	60	M	7	7	4	15	0%	URT flora, S. pneumoniae, coliform
Youngs	B64	62	F	5	7	15	17	0%	URT flora, Candida albicans
PCD	B72	29	F	7	16	11	15	0%	URT flora, S. aureus, S pneumoniae
Idiopathic	B74	61	F	7	7	4	15	0%	URT flora, S. aureus
Postinfective	B75	64	F	7	6	1	15	0%	URT flora, H. influenzae
Postinfective	B62	53	F	7	7	15	15	0%	H. influenzae
ABPA	B12	67	M	4	7	1	14	<50%	S. aureus
								3/7 nonmucoid
Postinfective	B17	72	F	1	7	1	15	<50%	URT flora
								1/6 nonmucoid
Idiopathic	B31	72	F	7	14	1	17	<50%	URT flora, Proteus vulgaris
								2/6 mucoid
Idiopathic	B43	61	F	1	12	1	12	<50%	URT flora
								3/8 mucoid
Postinfective	B48	81	F	10	15	4	9	<50%	URT flora
								1/6 nonmucoid
Idiopathic	B50	64	F	7	5	7	15	<50%	URT flora, M. catarrhalis, S. maltophilia
								1/7 nonmucoid
Idiopathic	B52	52	F	7	7	1	7	<50%	URT flora, coliform, S. maltophilia, S. aureus
								1/9 nonmucoid
Idiopathic	B76	67	F	7	8	1	17	<50%	URT flora, β-hemolytic Streptococcus group C
								1/9 nonmucoid
Postinfective	B78	72	F	1	1	8	14	<50%	URT flora, C. albicans
								2/7 mucoid
Postinfective	B81	63	F	7	9	17	7	<50%	URT flora, H. influenzae, S. aureus
								2/7 nonmucoid
PCD	B68	53	F	4	12	15	4	<50%	URT flora
								3/8 nonmucoid
Postinfective	B71	68	F	2	14	1	4	<50%	S. aureus
								1/7 mucoid
Idiopathic	B02	78	F	2	7	4	17	≥50%	M. catarrhalis
								4/8 mucoid
Postinfective	B13	70	M	6	12	11	14	≥50%	—
								7/7 mucoid
Postinfective	B14	65	M	14	16	7	9	≥50%	—
								7/7 mucoid
ABPA	B16	64	M	7	10	1	7	≥50%	S. aureus, H. influenzae
								5/7 mucoid
Postinfective	B21	62	F	9	7	7	15	≥50%	—
								7/7 mucoid
PCD	B23	30	M	7	7	4	15	≥50%	—
								1/1 mucoid
ABPA	B40	64	M	4	7	1	13	≥50%	C. albicans
								5/7 mucoid
Postinfective	B47	72	F	7	8	4	14	≥50%	—
								4/4 nonmucoid
Idiopathic	B57	76	F	4	12	4	15	≥50%	—
								4/7 mucoid
ABPA	B82	60	M	7	12	13	17	≥50%	—
								6/7 mucoid
Postinfective	B67	66	F	17	4	1	7	≥50%	URT flora, A. fumigatus
								4/7 nonmucoid
ABPA	B70	55	F	5	7	4	15	≥50%	—
								7/7 mucoid
Idiopathic	B61	57	F	7	12	1	15	≥50%	URT flora
								5/7 mucoid
PCD	B85	39	F	7	10	4	15	≥50%
								6/6 nonmucoid

Definition of abbreviations: ABPA = allergic bronchopulmonary aspergillosis; CVID = common variable immunodeficiency; HLA = human leukocyte antigen; ID = identification; PCD = primary ciliary dyskinesia; URT = upper respiratory tract.

Sputum samples were collected monthly for 6 months and analyzed by microscopy and culture using standard microbiology techniques. Patients were divided into three groups defined by the percentage of sputum cultures collected that were positive for P. aeruginosa (0%, <50%, and ≥50%). The column also shows the number of sputum samples positive for mucoid isolates over the total number of sputum samples collected and analyzed. Where no mucoid P. aeruginosa was isolated, the number of sputum samples positive for nonmucoid isolates is given over the total number of sputum samples collected.

Production and Purification of OprF Protein and Peptide Panel

OprF protein was produced using the recombinant vector pSUMO-OprF as described previously, with minor modifications (25, 32). The identity and purity of OprF protein was confirmed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and mass spectrometry. An OprF (PA1777) peptide library comprising 20-mer peptides overlapping by 10 amino acids (Table 2) was synthesized (GL Biochem Ltd., Shanghai, China).

Table 2.

OprF Peptide Panel

Peptide Number	Peptide Name	Peptide Sequence
1	PA1777 [1–20]	MKLKNTLGVVIGSLVAASAM
2	PA1777 [11–30]	IGSLVAASAMNAFAQGQNSV
3	PA1777 [21–40]	NAFAQGQNSVEIEAFGKRYF
4	PA1777 [31–50]	EIEAFGKRYFTDSVRNMKNA
5	PA1777 [41–60]	TDSVRNMKNADLYGGSIGYF
6	PA1777 [51–70]	DLYGGSIGYFLTDDVELALS
7	PA1777 [61–80]	LTDDVELALSYGEYHDVRGT
8	PA1777 [71–90]	YGEYHDVRGTYETGNKKVHG
9	PA1777 [81–100]	YETGNKKVHGNLTSLDAIYH
10	PA1777 [91–110]	NLTSLDAIYHFGTPGVGLRP
11	PA1777 [101–120]	FGTPGVGLRPYVSAGLAHQN
12	PA1777 [111–130]	YVSAGLAHQNITNINSDSQG
13	PA1777 [121–140]	ITNINSDSQGRQQMTMANIG
14	PA1777 [131–150]	RQQMTMANIGAGLKYYFTEN
15	PA1777 [141–160]	AGLKYYFTENFFAKASLDGQ
16	PA1777 [151–170]	FFAKASLDGQYGLEKRDNGH
17	PA1777 [161–180]	YGLEKRDNGHQGEWMAGLGV
18	PA1777 [171–190]	QGEWMAGLGVGFNFGGSKAA
19	PA1777 [181–200]	GFNFGGSKAAPAPEPVADVC
20	PA1777 [191–210]	PAPEPVADVCSDSDNDGVCD
21	PA1777 [201–220]	SDSDNDGVCDNVDKCPDTPA
22	PA1777 [211–230]	NVDKCPDTPANVTVDANGCP
23	PA1777 [221–240]	NVTVDANGCPAVAEVVRVQL
24	PA1777 [231–250]	AVAEVVRVQLDVKFDFDKSK
25	PA1777 [241–260]	DVKFDFDKSKVKENSYADIK
26	PA1777 [251–270]	VKENSYADIKNLADFMKQYP
27	PA1777 [261–280]	NLADFMKQYPSTSTTVEGHT
28	PA1777 [271–290]	STSTTVEGHTDSVGTDAYNQ
29	PA1777 [281–300]	DSVGTDAYNQKLSERRANAV
30	PA1777 [291–310]	KLSERRANAVRDVLVNEYGV
31	PA1777 [301–320]	RDVLVNEYGVEGGRVNAVGY
32	PA1777 [311–330]	EGGRVNAVGYGESRPVADNA
33	PA1777 [321–340]	GESRPVADNATAEGRAINRR
34	PA1777 [331–350]	TAEGRAINRRVEAEVEAEAK

Definition of abbreviation: OprF = outer membrane porin F.

An OprF peptide library comprising 20-mer peptides overlapping by 10 amino acids was synthesized from the sequence of OprF, PA1777 (GL Biochem Ltd., Shanghai, China).

HLA Class II Transgenic Mice and Mapping of HLA-restricted T-Cell Responses

Human leukocyte antigen (HLA)-DR1(DRA*0101/DRB1*0101)Aβ°, HLA-DR4(DRA*0101/DRB1*0401)Aβ°, and HLA-DR15(DRA*0101/DRB1*1501)Aβ° transgenic mice described previously (33–36) aged 8–15 weeks were used. U.K. Home Office regulations for animal welfare were strictly observed. HLA-DR1, -DR4, and -DR15 transgenic mice were footpad immunized with 25 μg OprF emulsified in Complete Freund’s adjuvant. Popliteal draining lymph nodes were harvested at Day 10 and stimulated with 25 μg/ml protein or individual peptides and cellular responses detected by murine IFN-γ enzyme-linked immunospot assay (ELISPOT) (Gen-Probe Diaclone SAS, Besançon, France).

Human T-Cell Responses to OprF Protein and Peptide Panel: Quantification of Immunologic Effector Proteins and OprF IgG-Specific ELISA

Peripheral blood mononuclear cells (PBMCs) were stimulated with 50 μg/ml protein or individual peptides and cellular responses detected using the ELISpot^PRO Human IFN-γ Kit (Mabtech, Nacka Strand, Sweden) (Table 3). Supernatants from PBMCs cultured with OprF protein in human IFN-γ ELISPOT assays were used to determine cytokine and chemokine levels (Table 3) using the 30-plex Luminex assay (Invitrogen, Life Technologies Ltd., Paisley, UK) on the Bio-plex 200 system (Bio-Rad Laboratories Ltd., Hemel Hempstead, UK). OprF IgG antibody titers were determined in patient serum samples by ELISA (Table 3).

Table 3.

Subjects Used in T-Cell Functional Analysis: Enzyme-linked Immunospot Assay

Group Classification	Luminex Protein	Age (±2 SD)	Sex (F/M)	Number
Luminex
Never culture positive for PA	IL-6/IL-8/MIP-1α	52 ± 33	11/2	13
	All other proteins	51 ± 30	18/6	24
>50% cultures positive for PA	IL-6/IL-8/MIP-1α	64 ± 21	7/5	12
	All other proteins	55 ± 28	4/4	8
ELISpot
Never sputum culture positive for PA		53 ± 31	22/9	31
<50% of sputum cultures positive for PA		66 ± 17	11/1	12
>50% of sputum cultures positive for PA		61 ± 26	8/6	14
ELISA
Never sputum culture positive for PA		54 ± 31	33/12	45
<50% of sputum cultures positive for PA		64 ± 23	14/2	16
>50% of sputum cultures positive for PA		61 ± 22	10/10	20
PCR
Never culture positive for PA		47 ± 32	6/1	7
>50% of sputum cultures positive for PA		58 ± 26	2/2	4

Definition of abbreviation: ELISpot = enzyme-linked immunospot assay; MIP = macrophage inflammatory protein; PA = Pseudomonas aeruginosa; PCR = polymerase chain reaction.

Real-Time Polymerase Chain Reaction Analysis of Patient PBMC

PBMCS were stimulated using phorbol myristate acetate/ionomycin as a generic form of lymphoid activation to obtain a snapshot of activated transcription. RNA was prepared (Absolutely RNA microprep kit; Agilent Technologies, Wokingham, UK) and cDNA synthesized using SuperScript III reverse transcriptase (Invitrogen, Life Technologies). Real-time polymerase chain reaction reactions were performed using glyceraldehyde phosphate dehydrogenase (PPH00150F), T-box transcription factor 21 (TBX21) (PPH00396A), retinoic acid receptor–related orphan receptor C (RORC) (PPH05877A), trans-acting T cell–specific transcription factor (Gata3) (PPH02143A), S1P receptor 1 (S1pr1) (PPH01350F) (RT² quantitative polymerase chain reaction primer assays; Qiagen, Manchester, UK), and glyceraldehyde phosphate dehydrogenase (Hs02758991_g1), forkhead box P3 (FoxP3) (Hs01085834_m1), and IL-10 (Hs00961622_m1) (Applied Biosystems, Paisley, UK) primers. Polymerase chain reactions were run in triplicate and Ct values obtained using a MX3000P real-time polymerase chain reaction machine (Stratagene, Stockport, Cheshire, UK). Variance RNA amount between samples was controlled by normalizing to glyceraldehyde phosphate dehydrogenase and relative levels of gene expression between samples were determined using the standard curve method (Table 3).

Peptide/HLA-DR Binding Assays

HLA-DR heterodimers were purified and relative binding of the OprF peptide panel to the HLA-DR molecules HLA-DRB1*01:01, 03:01, 04:01, 07:01, 09:01, 11:01, 13:01, and 15:01 was measured by competitive ELISA, as previously described (33–36). Data were expressed as relative affinity: ratio of the half maximal inhibitory concentration of the peptide to the half maximal inhibitory concentration of the reference peptide.

Results

Strong T-Cell Responses to PA Protein and OprF in Patients with Non-CF Bronchiectasis

We initially compared the T-cell responses with the major PA immunogen, OprF, in three groups of patients with bronchiectasis, stratified according to whether PA was never, sometimes (<50%), or frequently (≥50%) cultured from monthly sputum samples collected over 6 months (Tables 1 and 3). Patients showed a strong IFN-γ response to the protein antigen. Individuals that were never sputum culture positive for PA or sputum culture positive less than 50% of the time had mean frequencies of 158 (±377) and 202 (±484) spot-forming cells/10⁶, respectively, suggestive of robust immune memory devoted to the host PA response (Figure 1A). However, patients with evidence of chronic PA infection that were sputum culture positive for PA greater than or equal to 50% of the time showed reduced responsiveness to OprF, with a mean response of 119 (±262) spot-forming cells (Figure 1A). Patients with chronic PA infection were more likely to be infected by mucoid isolates (chi-square test, 5.42; P = 0.0199, Pearson). In the group that were less than 50% sputum sample culture positive for PA, 4 of 12 (25%) individuals grew mucoid isolates and in the greater than or equal to 50% group, 11 of 14 (79%) individuals grew mucoid isolates (see Figure E1A in the online supplement). Figure 1B shows the patient ELISPOT response to OprF in terms of infection by mucoid or nonmucoid PA isolates.

Figure 1.

T-cell and serum antibody responses to Pseudomonas aeruginosa (PA) protein outer membrane porin F (OprF) in patients with non–cystic fibrosis bronchiectasis. (A) The magnitude of the T-cell response to OprF and control anti-CD3 in the three bronchiectasis groups classified according to the PA infection status were determined (never [0%], n = 31; sometimes [<50%], n = 12; and frequent [≥50%], n = 14 culture positive for PA). In the group that were less than 50% sputum sample culture positive for PA, 4 of 12 (25%) individuals grew mucoid isolates, and in the greater than or equal to 50% group, 11 of 14 (79%) individuals grew mucoid isolates (chi-square test, 5.42; P = 0.0199, Pearson). (B) The magnitude of the T-cell response to OprF and control anti-CD3 was measured and plotted according to the PA isolate culture status (never [0%], n = 31; nonmucoid, n = 11; or mucoid, n = 15). (C) The IgG OprF-specific antibody titers in the three bronchiectasis groups classified according to the PA infection status were determined (never [0%], n = 45; sometimes [<50%], n = 16; and frequent [≥50%], n = 21). (D) The IgG OprF-specific antibody titer was measured and plotted according to the PA isolate culture status (never [0%], n = 45; nonmucoid, n = 17; or mucoid, n = 19). Data are presented ±SEM. Statistical significance was determined using the Kruskal-Wallis test followed by a post hoc Dunn test. **P < 0.01, ***P < 0.001. PBMC = peripheral blood mononuclear cells; sfu = spot-forming unit.

Strong Antibody Responses to PA Protein, OprF in Patients with Non-CF Bronchiectasis, and Patients with Evidence of Chronic PA Infection

Next, we measured serum antibody responses to OprF in patients with bronchiectasis (Tables 1 and 3), stratified according to whether PA was never, sometimes (<50%), or frequently (≥50%) cultured from monthly sputum samples (Figure 1C) and according to the type of PA isolate (no PA, nonmucoid PA, or mucoid PA) (Figure 1D). Individuals that were sputum culture positive for mucoid and nonmucoid PA made significantly stronger antibody responses to OprF than those that were consistently sputum culture negative for PA (mean, ± 2 SD for mucoid PA [1,692 ± 3,059; P < 0.001] and nonmucoid PA [947 ± 2,355; P < 0.01]) compared with culture negative for PA (325 ± 457). Individuals with non-CF bronchiectasis with evidence of chronic PA infection that were frequently culture positive for PA (≥50%) had significantly higher serum antibody responses to OprF (mean, ± 2 SD for >50% for PA [2,082 ± 3,131; P < 0.001], <50% PA [414 ± 330; P < 0.01], and no PA [325 ± 457]).

OprF Sequence Contains Strong HLA-DR–Binding Epitopes

Because OprF is immunogenic and protective in experimental Pseudomonas (20–25) and seroreactive in patients with non-CF bronchiectasis with chronic PA infection (37), we further investigated the candidacy of OprF as a CD4 T-cell antigen. A synthetic peptide panel of 20 mers overlapping by 10 amino acids was generated, covering the full coding sequence (Table 2) and binding affinities determined for the HLA-DR alleles, HLA-DR1, -DR3, -DR4, -DR7, -DR9, -DR11, -DR13, and -DR1501 (Table 4). Of the 34 peptides in the panel, more than 60% were able to bind one or more of the HLA alleles tested with moderate to high relative affinity, showing that OprF contains peptides that can bind across many HLA class II alleles.

Table 4.

OprF Peptide Relative Binding Affinity to HLA-DR Molecules

OprF Peptide	DR1	DR3	DR4	DR7	DR9	DR11	DR13	DR1501
p1	1	>1,060	3	2	10	50	>309	4
p2	2	>1,060	3	4	3	11	>309	3
p3	13	>1,060	350	9	23	756	2	94
p4	1,134	224	10	964	420	75	3	173
p5	15	>1,060	249	1,420	122	1,500	>309	3
p6	36	21	11	10	52	354	>309	49
p7	1,793	>1,060	>2,968	534	37	>4,297	>309	81
p8	1,120	>1,060	>2,968	>5,056	>1,642	537	>309	>1,119
p9	15	533	2,000	1,412	>1,642	1,333	>309	1,771
p10	18	447	753	91	42	129	>309	8
p11	112	>1,060	67	67	39	548	200	0.5
p12	8,571	>1,060	>2,968	>5,056	>1,642	>4,297	>309	>1,119
p13	11,429	26	1,333	355	345	1,500	>309	417
p14	11	23	283	49	13	707	2	67
p15	8	>1,060	5	>5,056	13	22	>309	20
p16	2,673	>1,060	>2,968	9,412	574	149	>309	1,458
p17	4,286	>1,060	>2,968	1,765	548	4,167	>309	>1,119
p18	66	>1,060	460	127	57	420	>309	123
p19	212	>1,060	4,333	241	177	8,333	>309	1,250
p20	28,571	>1,060	6,000	11,765	>1,642	>4,297	>309	>1,119
p21	20,000	>1,060	>2,968	>5,056	>1,642	6,667	>309	>1,119
p22	3,714	683	>2,968	7,059	>1,642	>4,297	>309	>1,119
p23	94	>1,060	>2,968	>5,056	>1,642	>4,297	>309	>1,119
p24	>11,262	>1,060	>2,968	>5,056	>1,642	>4,297	>309	>1,119
p25	2,857	13	1,467	1,412	237	2,083	>309	833
p26	822	>1,060	258	1,065	1,200	335	>309	70
p27	71	>1,060	103	80	85	866	169	4
p28	>11,262	>1,060	>2,968	>5,056	>1,642	>4,297	>309	>1,119
p29	1,225	>1,060	>2,968	1,412	600	342	>309	1,042
p30	949	>1,060	467	142	63	1,667	>309	38
p31	90	>1,060	161	290	64	130	>309	33
p32	35	>1,060	>2,968	183	31	391	>309	15
p33	14,286	>1,060	2,667	>5,056	3,600	2,917	>309	>1,119
p34	>11,262	>1,060	>2,968	>5,056	>1,642	7,500	>309	>1,119

Definition of abbreviations: HLA = human leukocyte antigen; OprF = outer membrane porin F.

Results are expressed as a relative binding ratio obtained by dividing the half maximal inhibitory concentration of each peptide by that of a reference peptide that binds strongly to the HLA molecule tested. Lower numbers correspond to a higher binding affinity. Numbers in bold (ratio of 20 or less) = high-affinity binding; numbers in italics (ratio 20–100) = moderate binding affinity. Each peptide:MHC combination was evaluated in two independent experiments.

Narrowed OprF T-Cell Epitope Responses in Patients with Non-CF Bronchiectasis with Chronic PA Infection

Patient T-cell responses to OprF were epitope mapped by screening responses to an overlapping peptide panel spanning the full-length sequence (Tables 1 and 3). Responses in patients with non-CF bronchiectasis from whose sputum PA was never cultured show broad immune responses to multiple T-cell epitopes within OprF (Figure 2A). The spread of epitope responses seemed somewhat similar in the patients who were sometimes culture positive (<50%) for PA. However, in the individuals with evidence of chronic PA infection, the response focused on a significantly smaller number of epitopes (mean, ±2 SD; 9 ± 13 in the never–sputum PA culture–positive group, compared with 3 ± 10 in the >50% sputum PA culture–positive group; P < 0.01). For clarity, this relationship is displayed in Figure 2B, with patients stratified by mucoid morphotype of PA isolates (see Figure E1B). Representative examples of patient T-cell responses to the peptide panel are shown for the never–PA culture–positive group (Figure 2C), sometimes–culture-positive group (<50% of sputum samples) (Figure 2D), and frequently culture-positive group (≥50% of sputum samples) (Figure 2E).

Figure 2.

T-cell epitope responses to outer membrane porin F (OprF) peptide panel in patients with non–cystic fibrosis bronchiectasis with evidence of chronic Pseudomonas aeruginosa (PA) infection. Epitope mapping of patient T-cell responses to OprF were determined by screening responses to an overlapping peptide panel of OprF for (A) PA infection status (never [0%], n = 31; sometimes [<50%], n = 12; and frequent [≥50%], n = 14 culture positive) or (B) PA isolate (never PA positive, n = 31; nonmucoid, n = 11; or mucoid, n = 15). Representative examples of patient T-cell responses to the peptide panel are shown for the never (0%) (C), sometimes (<50%) (D), and frequently (≥50%) (E) culture positive. Two SD above the mean of the media-only control is shown as a horizontal dotted line. (F) The overlapping relationships of epitopes identified in bronchiectasis patients with PA isolates classified as never PA culture positive (n = 31) and nonmucoid (n = 11) and mucoid (n = 15) positive cultures visualized as a Venn diagram. Data are presented ±SEM. Statistical significance was determined using the Kruskal-Wallis test followed by a post hoc Dunn test. *P <  0.05, **P < 0.01. PBMC = peripheral blood mononuclear cell; sfc = spot-forming cell.

The T-cell response to OprF in patients with non-CF bronchiectasis with evidence of chronic PA infection is focused on fewer epitopes within the antigen. Because it was theoretically possible that breadth of epitope recognition was simply a correlate of overall magnitude of response to the protein, we checked to see if such a correlation could be determined, but found none (see Figure E2). The notion that breadth of epitope recognition is not a simple correlate of magnitude of response to the whole antigen is further supported by analysis of those specific epitopes most commonly recognized by T cells from the patients in each group (Figure 2F); whereas some epitopes were recognized by some donors from each of the groups, others were more commonly recognized by individuals from one group or another.

CD4 T-Cell Epitope Responses to OprF Peptide Panel in HLA Cass II Transgenic Mice

Because of the difficulties inherent in any attempt to impute patterns of peptide-HLA class II restriction in human populations where most individuals are heterozygotes across the HLA region and, furthermore, each antigen-presenting cell expresses multiple, different HLA class II heterodimers, we compared patterns of CD4 OprF epitope recognition from patients with the pattern of responses in HLA transgenic mice (Figure 3). For this, we used a panel of mice that carry a knockout for endogenous mouse H2-Aβ and used human HLA class II heterodimers, HLA-DR1, -DR4, and -DR15. As can be seen in Figure 3, screening of OprF epitopes in the context of individual HLA-DR molecules allows for unequivocal attribution of peptides presented in this context: the response of HLA-DR1 transgenics focuses on p4; of DR4 transgenics on p6, p14, and p15; and of DR15 on p11. Although the complexities of human T-cell repertoire selection and immunogenetics make it challenging to juxtapose findings in HLA transgenics and in humans, comparison of the response patterns offers some insight as to peptide-HLA complexes involved in the PA response.

Figure 3.

CD4 T-cell epitope responses to outer membrane porin F (OprF) peptide panel in HLA class II transgenic mice. Draining lymph nodes from transgenic mice for DRB1*0101, n = 15 (A), DRB1*0401, n = 9 (B), and DRB1*1501, n = 10 (C) were harvested 10 days postimmunization with OprF protein in complete Freund’s adjuvant. Recall peptide responses were determined using the overlapping peptide panel for OprF. Responses were considered positive if the response was greater than 2 SD above the mean of the negative control (see horizontal dotted line). SEB = staphylococcal enterotoxin B; SFC = spot-forming cell.

We next looked at to what extent p11, the sole immunodominant epitope presented to CD4 T cells by HLA-DR15 in transgenic mice, features in the T-cell response of patients who carry one or two HLA-DR15 alleles. It is one of the more common epitopes in HLA-DR15⁺ patients, and 7 of 14 (50%) individuals in the group from which PA was never cultured respond to p11. However, for individuals with evidence of PA infection in the sputum, this epitope was recognized in only one of nine (11%) of HLA-DR15⁺ patients (Fisher exact probability test, two tailed; P = 0.0858). The number/percentage of individuals carrying the HLA-DR1501 allele did not significantly differ between groups (n = 14 of 31 [44%] for never PA culture positive vs. n = 6 of 14 [43%] for ≥50% culture positive). Thus, the altered disease state to chronic infection may be associated not just with a quantitative diminution in T-cell response but in a qualitative shift in the recognition of immunodominant epitopes.

Enhanced Cytokine and Chemokine Responses in Patients with Non-CF Bronchiectasis with Evidence of Chronic PA Infection

Despite a substantially reduced antigen- and peptide-specific IFN-γ response to PA OprF, the patients with evidence of chronic PA infection showed an enhanced response with respect to several innate cytokines and chemokines when compared with responses in patients from whom PA was never isolated (Tables 3 and 5 and Figure 4). This increased response encompassed significantly raised levels of IL-12, IL-6, macrophage inflammatory protein-1α (CCL3), and macrophage inflammatory protein-1β, collectively suggesting increased chemotaxis for neutrophils, monocytes, and natural killer cells. IL-8 was also increased, although this difference did not reach significance. A small, but significant increase in secreted IL-4 may indicate a Th2-skewed program in these individuals.

Table 5.

Cytokine and Chemokine Responses in Non–Cystic Fibrosis Bronchiectasis

Immunologic Protein	Never Culture Positive (pg/ml)*	Culture Positive > 50% (pg/ml)†	P Value
Cytokines
GM-CSF	136 ± 25	240 ± 50
G-CSF	348 ± 77	779 ± 258
TNF-α	517 ± 96	1,027 ± 319
IFN-α	3.9 ± 2	10.7 ± 6
IL-1β	1,671 ± 175	2,315 ± 276
IL-2	0.2 ± 0.05	0.2 ± 0.1
IL-4	1.2 ± 0.4	3.6 ± 0.9	<0.01
IL-5	0.5 ± 0.1	0.8 ± 0.3
IL-6	5339.1 ± 1,508	12,772 ± 3,401	<0.05
IL-7	15.1 ± 2	21.6 ± 5.8
IL-8	35,230 ± 17,683	95,584 ± 34,319
IL-10	88.3 ± 21	159 ± 68
IL-12	12.4 ± 2	24.9 ± 5	<0.05
IL-13	1.1 ± 0.3	2.2 ± 0.7
IL-15	52.8 ± 10	60.6 ± 28
IL-17	0.1 ± 0.05	0.3 ± 0.2
IL-1RA	0	0
IL-2R	53.8 ± 11	83.5 ± 25
Chemokines
MIP-1α	7,338 ± 2,465	18,963 ± 4,145	<0.05
MIP-1β	1,215 ± 230	4,038 ± 1,040	<0.01
RANTES	161 ± 36	202 ± 46
EOTAXIN	0.05 ± 0.02	0.05 ± 0.02
IP-10	1.9 ± 0.7	2.7 ± 2
MCP-1	0	0
MIG	2.5 ± 1	2.3 ± 2
Growth factors
VEGF	72.1 ± 13	78.3 ± 32
EGF	1.3 ± 0.3	0.6 ± 0.3
HGF	3.1 ± 1	3.0 ± 2
FGF	6.6 ± 1	7.8 ± 2

Definition of abbreviations: EGF = epidermal growth factor; FGF = fibroblast growth factor; G-CSF = granulocyte colony–stimulating factor; GM-CSF = granulocyte-macrophage colony–stimulating factor; HGF = hepatocyte growth factor; IP = IFN-γ–induced protein; MCP = monocyte chemoattractant protein; MIG = monokine induced by IFN‐γ; MIP = macrophage inflammatory protein; RANTES = regulated on activation, normal T-cell expressed and secreted; TNF = tumor necrosis factor; VEGF = vascular endothelial growth factor.

Data are presented ± SEM. Statistical significance was determined using the Mann-Whitney U test.

^*Never (0%) (n = 13 for IL-6, MIP-1α, and IL-8; n = 24 for MIP-1β and IL-12) culture positive for Pseudomonas aeruginosa.

^†Frequently (>50%) (n = 12 for IL-6, MIP-1α, and IL-8; n = 8 for MIP-1β and IL-12) culture positive for P. aeruginosa.

Figure 4.

Cytokine and chemokine responses to outer membrane porin F protein in non–cystic fibrosis bronchiectasis with evidence of chronic Pseudomonas aeruginosa infection. Proinflammatory cytokine concentrations for (A) IL-6, (B) macrophage inflammatory protein (MIP)-1α, (C) MIP-1β, (D) IL-12, and (E) IL-8 were determined from the supernatant of peripheral blood mononuclear cells stimulated with outer membrane porin F. Concentrations were measured using the 30-plex Luminex assay in patients with bronchiectasis classified as never (0%; n = 13 for IL-6, MIP-1α, and IL-8; n = 24 for MIP-1β and IL-12) or frequently (≥50%; n = 12 for IL-6, MIP-1α, and IL-8; n = 8 for MIP-1β and IL-12) culture positive for P. aeruginosa. Data are presented ±SEM. Statistical significance was determined using the Mann-Whitney U test. *P < 0.05, **P < 0.01.

Lymphocyte Transcriptional Changes in Patients with Non-CF Bronchiectasis with Evidence of Chronic PA Infection

PBMC responses to OprF suggested an immune profile in chronically infected individuals whereby antigen-specific IFN-γ T-cell effector responses are narrowed in epitope specificity, but the innate/inflammatory and antibody response is enhanced. We therefore investigated whether there was evidence for any intrinsic difference between patient groups in terms of transcripts associated with T-cell adaptive immunity programs (Table 3 and Figure 5). We looked at the hallmark transcription factors of Th1, Th17, and Th2 cells, that is, TBX21, RORC, and Gata3. We also looked at S1pr1 transcription, a molecule implicated in T-cell trafficking between tissues and lymph nodes. Transcripts for TBX21 were significantly reduced in PBMC samples from patients that were frequently (≥50%) sputum culture positive for PA compared with those that were never sputum culture positive for PA (Figure 5A). This is suggestive of a relative reduction in Th1 programs in patients with non-CF bronchiectasis with evidence of chronic PA infection. No change was seen with respect to transcription of Gata3 the Th2 transcription factor (Figure 5D), and RORC the Th17 transcription factor is reduced, but this did not achieve significance (Figure 5C). We found no evidence that these changes may have been mediated through a systemic bias to a more regulated phenotype: transcription of FoxP3 and IL-10 was also reduced, although this did not reach statistical significance (Figures 5E and 5F). Transcription of S1pr1 was significantly reduced in PBMC of the chronic PA carriers relative to PBMC from individuals who were never culture positive for PA (Figure 5B).

Figure 5.

Lymphocyte transcriptional changes in patients with non–cystic fibrosis bronchiectasis with evidence of chronic Pseudomonas aeruginosa infection. The expression of (A) T-box transcription factor 21 (TBX21), (B) S1P receptor 1 (S1pr1), (C) retinoic acid receptor–related orphan receptor cT (RORcT), (D) trans-acting T cell–specific transcription factor (Gata3), (E) forkhead box P3 (FoxP3), and (F) IL-10 were determined by quantitative polymerase chain reaction. Peripheral blood mononuclear cells were stimulated with phorbol myristate acetate/ionomycin, from which RNA was isolated and converted to cDNA. Patients with broncheiectasis were classified as never (0%; n = 7) or frequently (>50%; n = 4) culture positive for P. aeruginosa. Data are presented ±SEM. Statistical significance was determined using the Mann-Whitney U test. *P < 0.05.

Discussion

OprF has received attention as an immunogen that confers protection in murine challenge studies and trialed for immunogenicity as a PA vaccine in humans (11–13, 15, 16, 22–25). It is an abundantly expressed outer membrane protein of PA, initially characterized with respect to anchoring to the peptidoglycan layer (38) and with a structural role in adhesion to eukaryotic cells (39). It is up-regulated during biofilm formation under anaerobic conditions and the airway mucus of some patients with CF contains cleaved OprF protein. OprF antibody titer is regarded as a potential biomarker of biofilm infection (6). OprF binds IFN-γ, leading to stimulation of the quorum-sensing network (40). Oprf deletion mutants define a role of OprF in virulence, partly through modulation of quorum-sensing (41).

In light of this evidence for OprF immunogenicity and hyperexpression during biofilm formation, it might be predicted that patients undergoing such infection would show enhanced OprF adaptive immunity: it has previously been shown by others and confirmed by us that chronic infection is associated with raised antibody titers (6, 37). Studies in a mouse infection model allowing comparison of PA that was either alginate wild-type, hyperexpressing, or null showed that alginate overexpression was associated with poor bacterial clearance and exacerbated lung pathology, yet with heightened proinflammatory cytokines, that is, raised IFN-γ and IL-12 but lower IL-10 in lung homogenates (7). However, in other studies, mucoid or biofilm phenotypes are associated with reduced innate immunity (8, 42).

The present study shows that patients with chronic PA infection have reduced T-cell immunity to OprF, although with enhanced innate immunity with respect to several innate cytokines and chemokines. OprF has intrinsic ability to stimulate release of diverse innate cytokines from dendritic cells (43); although this clearly would not have contributed to the differences in epitope focus we observed in peptide-specific CD4 responses, it may be that in chronically PA-infected patients showing up-regulated OprF, this up-regulation may stimulate increased systemic IL-12 and tumor necrosis factor-α. From our study we cannot be certain whether the altered T-cell response pattern we observed is cause or effect of a heightened susceptibility to PA infection. It is likely that polymorphisms in pathways underpinning both innate and adaptive immunity to PA contribute to susceptibility (27–30). Certainly the enhanced innate inflammatory response contributes to the progressive lung damage, deterioration in lung function, and associated increased morbidity and mortality seen in patients with non-CF bronchiectasis with chronic PA infection (4).

A key finding here was the narrowed repertoire of T-cell OprF epitopes during chronic PA infection. This raises the question whether functional clearance of PA by adaptive immunity benefits most from a highly focused or a more diffuse response. The issue of epitope focus versus breadth has been considered most extensively in relation to viral infections, including HIV, cytomegalovirus, and hepatitis C virus. Simian immunodeficiency virus and HIV studies have argued variously that either a very broad or very focused repertoire response can be correlated with better control of infection and improved outcome (44–46). In the case of PA, curated sequences for OprF are largely conserved and indeed, the epitopes we have characterized in the present study are highly conserved across diverse Pseudomonas family members, including Pseudomonas alcaligenes, Pseudomonas indica, Pseudomonas citronellolis, and Pseudomonas stutzeri. Thus, in principle, a T-cell response targeted at any epitope should be able to support a protective response. This suggests that the correlation of multiple epitopes with effective host immunity may lie not in the breadth of epitope responses per se but in the cumulative frequency of responding cells.

This study showed a strong correlation between chronic PA infection and mucoid isolates, making it difficult to establish whether the driving correlate of altered adaptive immunity in chronic infection is the altered morphotype/transcriptome itself, or some other parameter of chronic infection (e.g., the altered inflammatory environment). Changes in the bacterial transcriptome with shift to a mucoid morphotype are considerable, with likely impacts on innate recognition, phagocytosis and uptake, and accessibility to proteolytic enzymes for antigen processing. It is noteworthy, however, that the change in T-cell epitope recognition in chronically infected patients is not just a matter of “missing responses.” For example, a response to p22 is seen most commonly in the chronic infection group (data not shown).

At the same time as analyzing antigen-specific changes at the level of epitope recognition, we analyzed systemic changes in adaptive immunity programs at the level of T-cell transcription. Individuals from whom PA was never isolated in sputum show heightened PBMC transcription of the Th1-controlling transcription factor, Tbet, and the S1P1 receptor, implicated in both central nervous system and lung lymphocyte trafficking (47). A relative reduction in Tbet transcription in chronic infection may indicate an intrinsic defect in (or, at least, program bias away from) Th1 polarization. Related observations have focused on the corollary of this: a switch to Th2 immunity (48). Tbet was first defined as a hallmark transcription factor for differentiation of Th1 cells, and since then has often been taken as a biomarker of Th1 immunity (49); in Tbet knockout mice the key phenotype relates to Th1 defects (50, 51). However, alterations in Tbet transcription may encompass programmatic changes in other cell types: CD8 cells, natural killer cells, innate lymphoid cells, dendritic cells, and B cells (52–56). The lipid second messenger, S1P and its receptor, S1PR1, control lymphoid organ egress, trafficking, and T-cell subset determination (57).

Our finding that patients with chronic PA infection have an intrinsic reduction in S1PR1 lymphocyte expression may help explain their abnormal immunity to infection, suggesting the potential for therapy via modulation of this pathway. With relatively large numbers of patients with multiple sclerosis receiving fingolimod (FTY720, S1P analog) as a new therapeutic of choice, there is a case for increased clinical vigilance for possible breakdown of lung surveillance of respiratory infections (58). The reduction in RORC transcription in patients with chronic PA did not achieve significance, but the findings seem compatible with a trend to reduced systemic Th17 immunity. The importance of Th17 immunity is highlighted by a study from Chan and colleagues (59) showing enhanced antigen-specific Th17 responses in draining lymph nodes of explanted lungs of patients with CF, suggesting that this subset may be important for lung immunity. Patients chronically sputum PA culture positive show a trend to reduced systemic FoxP3 and IL-10 transcription, compatible with the notion that defects in immune regulation may contribute to an inflammatory environment that facilitates PA lung infection (60).

It has become apparent that chronic immune stimulation, commonly by tumor antigens or viral antigens, can lead to an “exhausted” T-cell phenotype, characterized by raised expression of programmed cell death-1 (61). Because the example of chronic PA infection encompasses protracted immune exposure to enhanced antigen expression and yet results in lowered rather than heightened T-cell immunity, it is interesting to speculate that this may constitute a bacterial-driven example of exhaustion. Alternatively, other lung coinfections could be contributing to an altered immune environment in the lung. The lung is a complex immunologic environment, with interplay between innate and adaptive receptors and signals from diverse microbial species (27). This encompasses not just coinfections identified in sputum by standard microbiology, but also the microbiota species that are far from immunologically silent. A fully integrated analysis of the interplay between immune and microbial repertoires is both necessary and timely. We have previously reported observations on the immunogenetics of susceptibility to idiopathic bronchiectasis in terms of HLA-C and DR (28, 29). Determining within this patient group the genetic factors driving susceptibility to chronic PA infection is an important goal, but a much larger cohort needs to be enrolled into a study dedicated to specifically address this question.

Although several PA vaccines have been trialed including those using OprF, OprI, and flagellin (10–16, 20–25), studies such as this one raise an additional possibility of exploiting the potency and flexibility of incorporating a large number of stimulatory T- and B-cell epitopes into epitope string vaccines. In other contexts, such as Plasmodium falciparum vaccination, these can be successfully administered in a “prime-boost” regime through incorporation into an adenoviral vector followed by DNA boost (62). The appeal of the approach to a patient group such as this would be the ability to tailor a polyvalent construct to cover several different pathogens, including PA.