Pseudomonas aeruginosa Lipid A Structural Variants Induce Altered Immune Responses

Abstract

Pseudomonas aeruginosa causes chronic lung infection in cystic fibrosis (CF), resulting in structural lung damage and progressive pulmonary decline. P. aeruginosa in the CF lung undergoes numerous changes, adapting to host-specific airway pressures while establishing chronic infection. P. aeruginosa undergoes lipid A structural modification during CF chronic infection that is not seen in any other disease state. Lipid A, the membrane anchor of LPS (i.e., endotoxin), comprises the majority of the outer membrane of Gram-negative bacteria and is a potent Toll-like receptor 4 (TLR4) agonist. The structure of P. aeruginosa lipid A is intimately linked with its recognition by TLR4 and subsequent immune response. Prior work has identified P. aeruginosa strains with altered lipid A structures that arise during chronic CF lung infection; however, the impact of the P. aeruginosa lipid A structure on airway disease has not been investigated. Here, we show that P. aeruginosa lipid A lacks PagL-mediated deacylation during human airway infection using a direct-from-sample mass spectrometry approach on human BAL fluid. This structure triggers increased proinflammatory cytokine production by primary human macrophages. Furthermore, alterations in lipid A 2-hydroxylation impact cytokine response in a site-specific manner, independent of CF transmembrane conductance regulator function. It is interesting that there is a CF-specific reduction in IL-8 secretion within the epithelial-cell compartment that only occurs in CF bronchial epithelial cells when infected with CF-adapted P. aeruginosa that lacks PagL-mediated lipid A deacylation. Taken together, we show that P. aeruginosa alters its lipid A structure during acute lung infection and that this lipid A structure induces stronger signaling through TLR4.

Keywords: LPS; Pseudomonas; TLR4; airway adaptation; cystic fibrosis.

Figure 1.

Hexa-acylated lipid A for Pseudomonas aeruginosa in human airway infection. (A) P. aeruginosa canonical lipid A structure. Lipid A dioxygenase enzymes, LpxO1 and LpxO2, add hydroxyl residues onto secondary acyl chains in a site-specific manner, indicated in blue and red, respectively. PagL is a lipid A deacylase that removes a 3OH-C10 acyl chain (gray) from P. aeruginosa lipid A within the Gram-negative outer membrane. (B) Matrix-assisted laser desorption/ionization time-of-flight mass spectra of P. aeruginosa (PAK strain) grown to logarithmic phase in lysogeny broth (LB) or synthetic CF media (SCFM). BAL fluid underwent microbial culture in the clinical microbiology laboratory to confirm growth of organisms. This human BAL sample grew more than 100,000 cfu/ml P. aeruginosa. Peaks at a mass-to-charge ratio (m/z) of 1,462 are representative of lipid A that undergoes PagL-mediated deacylation in (A). (C–E) Gene expression analysis was performed using qRT-PCR of PAK grown in LB, in SCFM, or during acute lung infection of wild-type (WT) BALB/c or Scnn1b-transgenic mice; (C) pagL, (D) lpxO1, and (E) lpxO2. The ribosomal gene, rpsL, was used as a housekeeping gene to evaluate relative expression. Primer gene sequences are listed in Table E3. Unpaired t tests were used to compare growth conditions for each gene tested.

Figure 2.

Loss of PagL-mediated lipid A deacylation induces a cystic fibrosis (CF) transmembrane conductance regulator (CFTR)-independent proinflammatory cytokine response in myeloid cells. LPS was purified from two CF clinical strains (SE4 and SE9), where SE9 has an acquired loss-of-function mutation in pagL (273C→A producing a premature stop codon: Y91X) (see Figure E12). (A) THP-1 NF-κB reporter cell lines were then used to evaluate differences in TLR4 signaling after LPS stimulation with increased concentrations of LPS. (B and C) Next, THP-1 macrophages (that lack the NF-κB reporter machinery) were exposed to 100 ng/ml LPS for (B) 6 and (C) 24 hours before supernatant collection. Secreted IL-8 was measured by ELISA, and an unpaired t test was performed to determine significant differences between WT P. aeruginosa lipid A (Pa) and LPS with lipid A that lacked PagL-mediated deacylation (ΔpagL). Escherichia coli LPS, which has a highly stimulatory lipid A structure, was used as a positive control. (D and E) Last, primary monocyte-derived macrophages (MDMs) were obtained from subjects (D) without CF (non-CF) and (E) subjects with CF and were exposed to WT and mutant LPS. (F) A subset of CF MDMs were exposed to elexacaftor/tezacaftor/ivacaftor (ETI) before and during LPS exposure. Cell supernatant was collected 6 hours post-LPS exposure, and IL-6 levels were measured by ELISA. A paired t test was performed to determine differences between groups. Unpaired t tests in (B) and (C) and paired t tests in (D) through (F) were used to compare groups. OD = optical density.

Figure 3.

IL-6 secretion after LPS stimulation of MDMs. (A and B) Primary MDMs were obtained from subjects (A) without or (B) with CF and cultured ex vivo. These macrophages were exposed to purified P. aeruginosa LPS from defined lipid A mutant strains (engineered in P. aeruginosa PAK strain) (Figure E12) (28). Secreted IL-6 was measured by ELISA at 6 hours post-LPS stimulation. (C) MDMs from subjects CFBR-EM-348, -267, -658, and -398 were treated with ETI (at concentrations of 6, 3, and 30 nM, respectively) at the time of plating and during LPS exposure. A paired t test was conducted to assess for differences between treatment samples.

Figure 4.

P. aeruginosa LpxO mutant strains induce altered lipid A gene transcription in murine lung infection. (A) Scnn1b-transgenic (Tg) mice underwent pulmonary infection through the nasal route with P. aeruginosa (PAK strain) that lacked LpxO1- and/or LpxO2-mediated lipid A 2-hydroxylation; a survival curve was generated. Eight mice were included in each group, and a log-rank test was used to determine differences in survival among groups. (B) Lung P. aeruginosa burdens were determined at 7 days postinfection. (C) A similar murine lung infection was performed in Scnn1b-Tg mice, and 8 hours postinfection, lungs were harvested, and RNA was extracted from whole lungs. qRT-PCR was performed for lpxO1 and lpxO2. (D) In the same cohort of animals described in (C)—and using a separate cohort of wild-type BALB/c mice—qRT-PCR was performed on lptD. A P. aeruginosa 30S ribosomal protein gene, rpsL, was used as the housekeeping gene to determine relative changes in gene expression. Fold change was determined to be relative to gene expression in PAK-infected lungs. An unpaired t test was performed to determine differences between groups; P values are shown.

Figure 5.

P. aeruginosa PagL mutants promote CF-specific decrease in IL-8 secretion in human bronchial epithelial cells. WT (16HBE) and CF (IB3–1) human bronchial epithelial cells were infected with various P. aeruginosa PagL mutant strains at multiplicity of infection of 10 for 2 hours before gentamicin protection. Supernatants were collected at 24 hours postinfection and assayed for differences in IL-8 secretion by ELISA. An unpaired t test was used to compare groups.

Figure 6.

P. aeruginosa lipid A structural variation is associated with genetic divergence in mutS or mutL. (A) Proportion of P. aeruginosa strains isolated in CF with WT lipid A structure (n = 144) or lipid A structural variant (n = 35) that have the hypermutator genotype. Furthermore, subjects with CF from whom at least one P. aeruginosa lipid A variant strain was isolated (n = 12) were included. (B) A large-scale BLAST score ratio (LS-BSR) analysis was performed on genes commonly mutated in hypermutator P. aeruginosa strains, as well as three lipid A genes (lpxO1, lpxO2, and pagL) that can be mutated in CF chronic lung infection. This cohort (previously reported) includes 179 P. aeruginosa strains obtained from subjects with CF (Patients 1–22), as well as strains from non-CF sources. P. aeruginosa gene sequences were obtained using the PAO1 complete sequenced genome (National Center for Biotechnology Information accession number: NC_002516.2) An LS-BSR of 1 indicates full sequence similarity with consequence sequence, and an LS-BSR of 0 indicates no sequence similarity. One asterisk after patient numbers indicates at least one isolate from that patient with loss of function in LpxO1, two asterisks indicate at least one isolate from that patient with loss of function in LpxO2, and three asterisks indicate at least one isolate from that patient with loss of function in PagL.