A Pseudomonas aeruginosa hepta‐acylated lipid A variant associated with cystic fibrosis selectively activates human neutrophils

Short abstract

Lipid A produced by P. aeruginosa only in the late stages of CF stimulates IL‐8 production and survival, but not granule exocytosis by human neutrophils.

Abstract

Pseudomonas aeruginosa (PA) infection in cystic fibrosis (CF) lung disease causes airway neutrophilia and hyperinflammation without effective bacterial clearance. We evaluated the immunostimulatory activities of lipid A, the membrane anchor of LPS, isolated from mutants of PA that synthesize structural variants, present in the airways of patients with CF, to determine if they correlate with disease severity and progression. In a subset of patients with a severe late stage of CF disease, a unique hepta‐acylated lipid A, hepta‐1855, is synthesized. In primary human cell cultures, we found that hepta‐1855 functioned as a potent TLR4 agonist by priming neutrophil respiratory burst and stimulating strong IL‐8 from monocytes and neutrophils. hepta‐1855 also had a potent survival effect on neutrophils. However, it was less efficient in stimulating neutrophil granule exocytosis and also less potent in triggering proinflammatory TNF‐α response from monocytes. In PA isolates that do not synthesize hepta‐1855, a distinct CF‐specific adaptation favors synthesis of a penta‐1447 and hexa‐1685 LPS mixture. We found that penta‐1447 lacked immunostimulatory activity but interfered with inflammatory IL‐8 synthesis in response to hexa‐1685. Together, these observations suggest a potential contribution of hepta‐1855 to maintenance of the inflammatory burden in late‐stage CF by recruiting neutrophils via IL‐8 and promoting their survival, an effect presumably amplified by the absence of penta‐1447. Moreover, the relative inefficiency of hepta‐1855 in triggering neutrophil degranulation may partly explain the persistence of PA in CF disease, despite extensive airway neutrophilia.

Abbreviations

7‐AAD

= 7‐aminoactinomycin D

APC

= allophycocyanin

BALF

= bronchoalveolar lavage fluid

CD

= cluster of differentiation

CF

= cystic fibrosis

FCC

= ferricytochrome C

fMLF

= N‐formyl‐met‐leu‐phe

geo MFI

= geometric mean fluorescence intensity

HEK 293

= human embryonic kidney epithelial cell line 293

LB

= lysogenic broth

MD‐2

= myeloid differentiation factor‐2

m/z

= mass‐to‐charge ratio

PA

= Pseudomonas aeruginosa

PagL

= PhoP/PhoQ‐activated lipase

PagP

= PhoP/PhoQ‐activated palmitoyl transferase

ROS

= reactive oxygen species

SEAP

= secreted alkaline phosphatase

TOF

= time‐of‐flight

Introduction

The progression of CF disease is marked by massive neutrophil infiltration that is ineffective with respect to clearance of bacterial pathogens, such as PA. PA, isolated from the airways of patients with CF, expresses unique LPS structural variants [1], but none of these variants have been tested for their effects on neutrophil function. Hence, this study was undertaken to provide the first such characterization, and it reveals a surprisingly broad range of response patterns to the CF‐specific lipid A variants.

Neutrophils are the most abundant immune cells in human peripheral blood and are rapidly recruited to sites of infection to attack invading pathogens [2, 3]. Exposure to LPS, a cell‐wall component of Gram‐negative bacteria, activates several functional responses needed for immune defense, including phagocytosis, oxidant release, cytokine‐chemokine production, and prolonged survival [4, 5, 6–7]. However, activated neutrophils can also contribute to inflammatory disorders, as occurs during CF progression when neutrophils accumulate in lung tissue but fail to clear the bacteria that drew them there [8, 9–10].

CF is a genetic disorder resulting from mutation in the gene that encodes the CF transmembrane conductance regulator, an anion channel. In CF‐associated obstructive lung disease, clogging of the airway by thick, sticky mucus favors persistence of opportunistic pathogens. The ensuing inflammation further impairs lung function, often resulting in premature death. Innate immune cells, predominantly neutrophils and alveolar macrophages, drive this strong inflammation but paradoxically fail to clear pathogens, such as PA, which infects almost 80% of patients with CF by early adulthood [10, 11, 12–13]. In the airways of these patients, PA is known to acquire multiple distinct characteristics, as it establishes a chronic infection, of which one of the most prominent involves structural modifications to LPS and its TLR4 stimulatory core component, lipid A [13, 14–15].

Lipid A, the bacterial membrane anchor of LPS, consists of a phosphorylated disaccharide head group with up to 7 fatty acid side‐chains. The positions and numbers of the acyl chains are highly variable among Gram‐negative bacteria, as determined by the activities of several biosynthetic enzymes [16, 17–18]. PA has at least 3 modes of lipid A biosynthesis that reflect its ability to thrive in a wide range of environments ( Fig. 1 ). First, PA, isolated from environmental samples or from non‐CF patients with acute or chronic infections, synthesizes a penta‐acylated lipid A, which is designated here as penta‐1419, reflecting its acyl chain number, followed by m/z in mass spectrometry. Second, in CF patients, beginning as early as 3 yr of age, PA converts to synthesis of a mixture of lipid A isoforms, designated penta‐1447 and hexa‐1685, in which penta‐1447 is typically predominant. Finally, in a subset of patients with late‐stage CF disease, characterized by severe pulmonary dysfunction, PA is observed to synthesize an unusual lipid A with 7 acyl side‐chains, hepta‐1855 [15, 19, 20]. The consistent recurrence of these patterns of synthesis strongly suggests that each lipid A variant helps PA adapt successfully to the respective environments from which it can be isolated.

Figure 1.

Lipid A variants synthesized by PA. All lipid A variants are named by the number of acyl chains and m/z of each, and arrows with text show precursor‐product relationships in biosynthesis pathways and the enzymes involved. PagL, a 3‐O‐deacylase, removes the 3 position 3‐OH C10 fatty acid, and PagP, an acyltransferase, adds a palmitate (C16 fatty acid) at the 2 position. Chemical structures are shown for the lipid A portion of (A) PA LPS isolated from environmental samples and acute and chronic infections other than in CF disease. (B and C) Variants present in all early and two‐thirds of late CF stage‐specific PA isolates, as a result of expression of PagP and PagL. (D) The hepta‐acylated variant expressed uniquely by one‐third of late CF stage‐specific PA isolates that have lost PagL activity. (E) Hexa‐1616, the biosynthetic precursor of hepta‐1855 and penta‐1447. (F) Summary of disease associations.

Many pathogens also alter the number and distribution of the fatty acid chains to subvert recognition by TLR4/MD‐2 and the subsequent immune‐defensive response [21, 22]. In general, hexa‐acylated LPS, produced by enteric bacteria, such as Escherichia coli and Salmonella, induces robust TLR4 signaling, whereas hypoacylated (tetra or penta) forms show weak or antagonistic activity [21, 23, 24]. PA lipid A acylation variants are similarly associated with differential recognition by human TLR4. For example, compared with penta‐1447_, hexa‐1685 induces a robust TNF‐α response in the human monocytic cell line THP‐1, higher IL‐8 response in endothelial cells, and strong NF‐κβ activation in a reporter assay [1, 20, 25]. However, the response of human neutrophils, key players in CF‐related lung inflammation, to each of these variants is unknown. In the case of the novel hepta‐1855 variant, no studies of any kind have been performed.

The transition of PA in CF lung disease from acute to chronic phases is characterized by reduced expression of many of its virulence factors, suggesting that hepta‐1855 may be similarly attenuated. On the other hand, establishment of chronic infection with PA is also known to be associated with intense inflammation in the late stages of disease, which might indicate that hepta‐1855 is highly immunostimulatory. Studies of hepta‐acylated lipid A variants from other Gram‐negative bacteria show unpredictable activities; some are weak [26], and others are strong [27] stimulators of TLR4. In the first tests of its activity, we found the late stage‐specific hepta‐1855 PA lipid A variant to be a strong stimulator of both neutrophils and monocytes with high TLR4 agonist activity in a reporter cell system. These findings may provide a rational basis for the association of the unique hepta‐acylated form of PA LPS with disease severity during late stages of CF. In addition, we found that penta‐1447 is likely a partial agonist/antagonist of TLR4, which has important implications for its role in establishment of chronic infection early in the CF disease process.

MATERIALS AND METHODS

Bacterial strains and growth conditions

PA was grown in LB ( Table 1 ), supplemented with 1 mM or 8 µM MgCl₂ to suppress or favor lipid A acylation modifications regulated by the 2‐component system, PhoP/Q [1, 19, 25, 28].

Table 1.

Bacterial strains and growth conditions used in preparation of PA LPS variants

LPS acyl variant	PAK strain	MgCl2 concentration	Reference
penta‐1419	ΔhtrB1	1 mM	[28]
penta‐1447	Wild‐type	1 mM	[25]
hexa‐ 1685	Wild‐type	8 μM	[25]
hepta‐1855	ΔPagL	8 μM	[19]
hexa‐ 1616	ΔPagL	1 mM	[19]

LPS acyl variants are identified by the number of acyl chains and observed m/z from mass spectrometry analysis. PA strain K (PAK), with the indicated mutations, was cultured in LB supplemented with high (1 mM) or low (8 µM) magnesium before LPS extraction. htrB1, 2‐hydroxy lauryl transferase.

LPS and lipid A isolation

PA LPS was extracted by the hot phenol/water method [29]. Freeze‐dried bacterial pellets were resuspended in endotoxin‐free water at a concentration of 10 mg/ml. A volume of 12.5 ml 90% phenol (Thermo Fisher Scientific, Waltham, MA, USA) was added, and the resultant mixture was vortexed and incubated for 60 min in a hybridization oven at 65°C. The mixture was cooled on ice and centrifuged at 12,096 g at room temperature for 30 min. The aqueous phase was collected, and an equal volume of endotoxin‐free water was added to the organic phase. The extraction was repeated, and aqueous phases were combined and dialyzed against Milli‐Q‐purified water to remove residual phenol and then freeze dried. The resultant pellet was resuspended at a concentration of 10 mg/ml in endotoxin‐free water and treated with DNase (Qiagen, Valencia, CA, USA) at 100 μg/ml and RNase A (Qiagen) at 25 μg/ml and incubated at 37°C for 1 h in a water bath. Proteinase K (Qiagen) was added to a final concentration of 100 μg/ml and incubated for 1 h in a 37°C water bath [30]. The solution was then extracted with an equal volume of water‐saturated phenol. The aqueous phase was collected and dialyzed against Milli‐Q‐purified water and freeze dried as above. The LPS was further purified by the addition of chloroform/methanol 2:1 (vol:vol) to remove membrane phospholipids [31] and further purified by an additional water‐saturated phenol extraction and 75% ethanol precipitation to remove contaminating lipoproteins [32]. No protein contamination was observed by the Bradford protein assay (Thermo Fisher Scientific). For mass structural analysis, 1 mg purified LPS was converted to lipid A by mild‐acid hydrolysis with 1% SDS (Sigma‐Aldrich, St. Louis, MO, USA) at pH 4.5, as described previously [33].

TLRgrade LPS from E. coli serotype O55:B5 (product number ALX‐581‐013‐L001; Enzo Life Sciences, Farmingdale, NY, USA) was used as a positive control for the experiments (see all figures except Fig. 3), in which Salmonella minnesota R595 (product number ALX‐581‐008‐L001; Enzo Life Sciences) was used to prime superoxide release. Tests of these LPS preparations confirmed that they were equally active in that assay (data not shown).

Figure 3.

Similar priming of neutrophil respiratory burst by CF‐specific hexa‐1685 and hepta‐1855 PA lipid A. released (nM) by human neutrophils in response to fMLF stimulation after priming with control S. minnesota LPS (SmLPS; 100 ng/ml) or (A) 100 ng/ml or (B) 1000 ng/ml of the indicated PA lipid A, as quantified by FCC reduction. Data show means ± sem from 4 to 11 experiments. UT, Untreated. ****P ≤ 0.0001, **P ≤ 0.01, *P ≤ 0.05 when compared with fMLF only; and ns, P > 0.05 when compared with hexa‐1685 PA lipid A.

MALDI TOF mass spectrometry

Lipid A, isolated by small‐scale lipid A isolation procedures, was analyzed on an Autoflex Speed MALDI TOF mass spectrometer (Bruker, Billerica, MA, USA). Data were acquired in reflectron negative and positive modes with a Smart Beam laser with 1 kHz repetition rate, and up to 500 shots were accumulated for each spectrum. Instrument calibration and all other tuning parameters were optimized using Agilent tuning mix (Agilent Technologies, Santa Clara, CA, USA). Data were acquired and processed using flexControl and flexAnalysis version 3.3 (Bruker).

Assay for TLR4 activity

HEK‐Blue TLR4 cells (InvivoGen, San Diego, CA, USA) are HEK 293, stably transfected with plasmids expressing human TLR4, MD‐2, and CD14 genes and a SEAP reporter gene, under the control of a minimal promoter with multiple NF‐κβ and AP‐1 binding sites. Cells were cultured in DMEM medium (Thermo Fisher Scientific) containing 1 mM sodium pyruvate, 50 units/ml penicillin, 50 mg/ml streptomycin, and 10% heat‐inactivated FBS (Thermo Fisher Scientific). Cells were plated at 5 × 10⁴ cells/well of a 96‐well plate and stimulated with agonists or medium as vehicle control for 24 h at 37°C. For all assays, 100 ng/ml of lipid A used is a physiologically relevant concentration, whereas further higher doses are for determining plateau. Cell‐free supernatants were analyzed for SEAP activity using the QUANTI‐Blue colorimetric enzyme assay, as directed by the manufacturer (InvivoGen).

Isolation and purification of human neutrophils

Neutrophils were isolated from venous blood of healthy donors using plasma‐Percoll gradients, as described elsewhere [34]. Isolated cells were >90–95% neutrophils, as evaluated by microscopy, and are referred to as “pure” neutrophils in this study. Trypan blue exclusion indicated that >97% of cells were viable. Human donor recruitment, blood draws, and the use of the materials were in accordance with the guidelines approved by the Institutional Review Board of the University of Louisville. In some experiments, pure neutrophil populations were further enriched to obtain highly pure cells (>99%) by negative magnetic selection using the EasyEights EasySep Magnet and human neutrophil enrichment kit (Stemcell Technologies, Vancouver, BC, Canada). Cell purity was assessed by simultaneously staining with FITC‐conjugated anti‐CD66b (clone G10F5; BioLegend, San Diego, CA, USA) and APC‐conjugated anti‐CD16 (clone CB16; eBioscience, San Diego, CA, USA) antibodies and determining the percentage of CD66b⁺CD16⁺ cells using BD LSR II flow cytometer (BD Biosciences, San Jose, CA, USA). Both pure (>90–95%) and highly pure (>99%) neutrophils were cultured in complete RPMI‐1640 medium (Thermo Fisher Scientific) containing 2 mM l‐glutamine, 50 units/ml penicillin, 50 mg/ml streptomycin, and 1 mM sodium pyruvate in a total volume of 200 μl in 96‐well plates for all overnight experiments. As a source of soluble CD14, LPS‐binding protein, and growth factors [35, 36–37], the medium was also supplemented with 5% heat‐inactivated human AB serum (Sigma‐Aldrich), an amount determined to be optimal for IL‐8 responses to control LPS in pilot experiments (data not shown).

Isolation of human monocytes

PBMCs were isolated using Histopaque ‐1077 (Sigma‐Aldrich) gradients or plasma‐Percoll gradients from the whole blood of normal, healthy donors. Isolated PBMCs were resuspended in complete RPMI‐1640 medium (Thermo Fisher Scientific) containing 2 mM l‐glutamine, 50 units/ml penicillin, 50 mg/ml streptomycin, 1 mM sodium pyruvate, and 10% heat‐inactivated human AB serum (Sigma‐Aldrich) and 50 mM 2‐ME (both added freshly). Cells were plated at 5 × 10⁵ cells/well of a 96‐well plate and incubated for 2 h at 37°C. Nonadherent cells were washed by gently pipetting with Ca²⁺Mg²⁺ HBSS (Thermo Fisher Scientific) twice and RPMI medium once at room temperature. The remaining adherent monocytes were cultured in fresh complete RPMI medium in a total volume of 200 μl for all experiments, generally for 20 h.

Cytokine production

Adherent monocytes at 0.5 × 10⁵ (estimated) cells/well and pure (>90–95%) or highly pure (>99%) neutrophils at 0.1 × 10⁶ cells/well of a 96‐well plate were cultured with TLR4 agonists or medium as vehicle control for 20 h. IL‐8 or TNF‐α or both were measured in cell‐free supernatants by ELISA (human IL‐8 and TNF‐α ELISA Ready‐SET‐Go! kit; eBioscience), according to the manufacturer's protocol.

Assay for partial agonist and antagonist activity

EC75, the concentration of an agonist that gives 75% maximal response, was calculated for each of the CF hexa‐1685, CF hepta‐1855 lipid A, and E. coli LPS in HEK‐Blue TLR4 cells for their SEAP‐inducing activity and in primary cells for their cytokine response. EC75 is commonly used as the concentration of target agonists for testing inhibitory activity of partial agonists [38]. The log EC50 value for each agonist‐induced response was calculated, as described in Kolb et al. [39], by generating a 4‐parameter logistic curve [log (agonist) vs. response − variable slope] by nonlinear regression, using GraphPad Prism software (GraphPad Software, La Jolla, CA, USA) and the following equation

EC50 and Hill slope values were used to calculate EC75 further using the following equation

Partial agonist and antagonistic activity of the penta‐1447 lipid A was determined in HEK‐Blue TLR4 cells and in primary cells. In brief, both cell types were cultured with EC75 values of the agonists, either alone or in the presence of increasing concentrations of penta‐1447 lipid A or with penta‐1447 alone. For IL‐8 and TNF‐α inhibition assays in monocytes, penta‐1447 lipid A was used for 4–5 different individuals, whereas penta‐1447 LPS was used for 1–2 donors.

Neutrophil survival assay

Pure (>90–95%) neutrophils were suspended in complete RPMI‐1640 medium, supplemented with 5% heat‐inactivated human AB serum. Cells were seeded in 96‐well plates at 5 × 10⁵ cells/well with TLR4 agonists, GM‐CSF‐positive control (CYT‐416; ProSpec‐Tany Technogene, East Brunswick, NJ, USA), or medium as vehicle control for 24 h at 37°C. Following incubation, the cells were washed with Ca²⁺Mg²⁺ HBSS (Thermo Fisher Scientific) and stained with APC‐conjugated Annexin V (BD Biosciences) and 7‐AAD (Thermo Fisher Scientific), according to the manufacturers’ protocol. Cell viability was assessed by the proportion of cells that were double negative for Annexin V and 7‐AAD, using a BD LSR II flow cytometer and FlowJo software.

Priming of the respiratory burst activity

Superoxide released by fMLF exposure in the presence or absence of TLR4 priming was quantified, as described elsewhere [40, 41]. In brief, pure neutrophils (>90–95%) were resuspended at 4 × 10⁶ cells/ml/replicate Eppendorf (polypropylene) tube (1.5 ml) in Krebs‐Ringer phosphate buffer containing 0.2% dextrose (“Krebs +”) and 5% heat‐inactivated human serum and preincubated for 5 min in a water bath at 37°C with gentle shaking. TLR4 agonists or vehicle control (Krebs+ with serum) were added for 1 h, followed by addition of FCC (Sigma‐Aldrich). One set of replicates was then stimulated with 300 nM fMLF (Sigma‐Aldrich) for 5 min; another set was not further manipulated to determine any response to TLR4 priming agents alone. All tubes were centrifuged for 10 min at 600 g at 4°C, and the cell‐free supernatants tested for OD values at 550 nm and the amount of Cytochrome c reduction were calculated in nanomoles.

Neutrophil granule exocytosis

Pure neutrophils, 4 × 10⁶ cells/ml/polypropylene tube (1.5 ml), were stimulated with TLR4 agonists and diluted in Krebs+ buffer with 5% heat‐inactivated human serum or with vehicle control (Krebs+ with serum) alone for 1 h in a water bath at 37°C with gentle shaking. Following incubation, cells were washed and stained at 4°C for 1 h with antibodies specific for the granule‐specific markers CD35, CD66b, or CD63. Exocytosis of secretory vesicles, specific granules, and azurophilic granules was determined by measuring the increase in plasma membrane expression (by geo MFI) of CD35 (PE‐conjugated anti‐human CD35 antibody; clone E11; BioLegend), CD66b (FITC‐conjugated anti‐human CD66b antibody; clone G10F5; BioLegend), or (by geo MFI) CD63 (FITC‐conjugated anti‐human CD63 antibody; clone AHN16.1/46‐4‐5; Ancell, Bayport, MN, USA), respectively. A BD LSR II flow cytometer was used, and the data were analyzed with FlowJo software (Tree Star, Ashland, OR, USA).

Statistical analysis

GraphPad Prism software was used for statistical analysis by tests (see each figure legend). Tests used included nonlinear regression, log (agonist) versus response‐variable slope (4 parameters) analysis, comparison of fits on log EC50 using the Extra Sum‐of Squares F‐test, and 2‐way ANOVA with Sidak's or Dunnett's multiple comparisons.

RESULTS

TLR4 stimulation by fatty acyl variants of PA lipid A

To determine whether human TLR4 responds differentially to fatty acyl variants of PA lipid A, we first tested individual lipid A preparations over a wide dose range (0.0003–1000 ng/ml) using HEK‐Blue TLR4 reporter cells ( Fig. 2 ). The positive control hexa‐acylated E. coli LPS, a potent TLR4 agonist, showed robust activity as expected. CF‐specific hexa‐1685 PA lipid A, although less potent than E. coli LPS, reached a similar dose plateau, as reported previously [20, 25]. The severe CF disease‐associated hepta‐1855 PA lipid A stimulated TLR4 with a potency and efficacy similar to the proinflammatory hexa‐1685 variant. A precursor of both penta‐1447 and hepta‐1855 isoforms, hexa‐1616, which is a minor component, was markedly less active, even at a very high concentration (1000 ng/ml). Both penta‐acylated PA lipid A variants had similarly weak TLR4 activity, as shown by the shift in the dose curve to the right; only the non‐CF penta‐1419 reached the maximum plateau at higher concentrations. Taken together, these data indicate that lipid A fatty acyl variants of PA activate human TLR4 differentially in a reporter cell assay and demonstrate for the first time that CF hepta‐1855 is highly active as a TLR4 agonist.

Figure 2.

Differential stimulation of human TLR4 by PA lipid A variants. HEK‐Blue TLR4 cells were cultured for 24 h with increasing doses of the indicated PA lipid A or E. coli LPS as control. The colors used to identify agonists are the same in all figures. TLR4‐stimulating activity was measured in a SEAP reporter assay and normalized to E. coli LPS activity (E. coli = 100%). Data points show means ± sem from at least 3 independent experiments.

PA lipid A variants prime ROS production by human neutrophils

Extracellular release of ROS is a potential mechanism for neutrophil‐mediated inflammatory damage in airways of patients with CF [42, 43], and TLR4 agonists can prime or potentiate neutrophil respiratory bursts that are triggered by activators, such as the bacterial peptide fMLF [37]. Therefore, we determined whether PA lipid A variants with varying TLR4‐stimulating potencies and efficacies differentially prime oxidative burst in human neutrophils.

We quantified superoxide release from peripheral blood neutrophils primed with control LPS at 100 ng/ml ( Fig. 3A and B ) or different PA lipid A isoforms for 1 h at a physiologically relevant dose of 100 ng/ml (Fig. 3A) or a higher dose of 1000 ng/ml (Fig. 3B). Neither LPS nor any of the PA lipid A variants induced any ROS release on their own (Fig. 3A and B). The precursor hexa‐1616 lipid A and both penta‐acylated variants failed to prime significant ROS release at a dose of 100 ng/ml (Fig. 3A), and only hexa‐1616 showed some priming effect at a higher dose of 1000 ng/ml (Fig. 3B). hexa‐1685 lipid A primed neutrophil respiratory burst at either dose. The late CF stage‐specific hepta‐1855 variant primed neutrophils as strong as that of the hexa‐1685 lipid A at both concentrations tested (Fig. 3A and B), in agreement with the TLR4 reporter cell assay (Fig. 2). Both isoforms were similarly weak in priming at concentrations lower than 100 ng/ml (data not shown). Thus, the robust TLR4 activation by hepta‐1855 in reporter cells was also seen in primary neutrophils at the level of an inflammatory function, superoxide release.

PA lipid A variants induce neutrophil granule exocytosis differentially

LPS is well known to induce neutrophil degranulation [44, 45–46]. Neutrophils contain 4 granule subtypes that are enriched in antimicrobial proteins and proteases. Degranulation or granule exocytosis involves incorporation of the granule membrane‐specific proteins onto the cell membrane and release of their luminal contents in the extracellular milieu. Once released outside of the cells, those granule contents not only help to kill extracellular pathogens but can also contribute to tissue destruction [47, 48].

To determine whether the PA lipid A variants induce granule exocytosis, human neutrophils were treated (1 h) with positive control LPS at 1000 ng/ml or different PA lipid A isoforms at 100 and 1000 ng/ml concentrations. At both concentrations tested, the CF hexa‐1685 variant induced the secretory vesicle and specific granule exocytosis measured as a plasma membrane increase of CD35 ( Fig. 4A ) and CD66b (Fig. 4B), respectively. Both hepta‐1855 and hexa‐1616 failed to increase either granule marker at 100 ng/ml, whereas at 1000 ng/ml, both showed minimum induction of secretory vesicle exocytosis (CD35), however, a stronger stimulation of specific granule exocytosis (CD66b; Fig. 4A and B). Both hexa‐1685 and hepta‐1855 variants induced azurophilic granule exocytosis (CD63) only at a higher dose of 1000 ng/ml (data not shown). penta‐1447 had no activity in these assays at any dose. Therefore, these data suggest that only hexa‐1685 PA lipid A, among all of the variants, is a potent inducer of neutrophil granule exocytosis.

Figure 4.

Differential induction of neutrophil granule exocytosis by PA lipid A variants. Human neutrophils were treated with the indicated PA lipid A or E. coli LPS (EcLPS; 1 μg/ml) as control and tested for neutrophil granule exocytosis by flow cytometry to measure increases in surface expression of (A) the secretory vesicle marker CD35 or (B) the specific granule marker CD66b. PA lipid A of 100 ng/ml and 1000 ng/ml are represented by left and right sides of the triangles, respectively. Bars show mean increases in geo MFI over untreated (=baseline) ± sem from 4 to 8 individuals. ****P ≤ 0.0001, ***P ≤ 0.001, **P ≤ 0.01 when compared with untreated.

Effect of PA lipid A variants on neutrophil survival

Several studies showed that blood neutrophils from patients with CF undergo delayed apoptosis when cultured ex vivo [49, 50]. As TLR4 agonists, such as LPS, are known to prevent neutrophil apoptosis, we next characterized the survival effect of PA lipid A variants using pure neutrophil populations (purity 90–95%).

Human neutrophils were cultured for 24 h in medium containing human serum with positive control GM‐CSF at 0.1 ng/ml or LPS at 1000 ng/ml or with PA lipid A isoforms at both 100 and 1000 ng/ml ( Fig. 5B ) and tested for cell viability. Approximately 55% of the untreated neutrophils remained viable after 24 h in independent assays (example shown in Fig. 5A; untreated, upper right). At both concentrations tested, hepta‐1855 and hexa‐1685 PA lipid A increased the percentage of viable cells similarly, as well as that by GM‐CSF, a very potent survival factor of neutrophils (Fig. 5B). The survival effects of both isoforms were also indistinguishable when tested at concentrations lower than 100 ng/ml (data not shown). hexa‐1616 increased neutrophil viability significantly at either concentration. No penta variants showed a substantial effect. Therefore, PA lipid A variants improve neutrophil survival differentially, with hepta‐1855, hexa‐1685, and the precursor hexa‐1616 showing substantial activity.

Figure 5.

hexa‐1685 and hepta‐1855 PA lipid A have similar survival effects on neutrophils. Purified human neutrophils (>90%) were incubated 24 h in the absence (untreated) or presence of control GM‐CSF (0.1 ng/ml), E. coli LPS (1 μg/ml), or PA lipid A variants. Cell viability before and after culture was analyzed by Annexin V‐APC/7‐AAD staining and flow cytometry. (A) Dot plots from a representative experiment showing the percent of cells that were viable (negative for Annexin‐V and 7‐AAD) at 0 or 24 h after the indicated treatments. (B) Percentage of viable neutrophils after culture with the indicated PA lipid A at 100 and 1000 ng/ml (by left and right sides of triangle, respectively) or controls. Bars show percent means ± sem from 3 to 6 individuals. PMN, Polymorphonuclear neutrophils. ****P ≤ 0.0001, ***P ≤ 0.001, **P ≤ 0.01 when compared with untreated.

PA lipid A differentially induces IL‐8 production from human neutrophils

IL‐8, a major chemoattractant of neutrophils, is elevated in BALF or sputum samples from patients with CF and negatively correlates with pulmonary function [51, 52, 53, 54–55]. Respiratory infections in patients with CF have higher levels of IL‐8 in their BALF compared with uninfected patients [56, 57]. The massive numbers of neutrophils infiltrating the CF airway can serve as a potential source of this chemokine [54]. Moreover, LPS is a known IL‐8 stimulator in neutrophils. Therefore, we compared the PA lipid A variants as stimulators of IL‐8 production from purified blood neutrophils (90–95% pure). penta‐1447 showed very little IL‐8‐inducing activity, even at nonphysiologically high concentrations, such as 10 μg/ml ( Fig. 6A ). In contrast, both the CF hexa‐1685 and ‐1616, as well as hepta‐1855, induced IL‐8 to the same dose plateau as E. coli LPS (Fig. 6A and C).

Figure 6.

Strong IL‐8 induction in human neutrophils by CF‐specific hexa‐1685 and hepta‐1855 PA lipid A variants. ELISA measurement of IL‐8 released from (A and C) pure neutrophils (>90%) or (B and D) highly pure neutrophils (>99%) after 20 h stimulation with the indicated dose range of PA lipid A or E. coli LPS. (E) Log EC50 of hexa‐1685 and hepta‐1855 PA lipid A in stimulating IL‐8 production by pure neutrophils, calculated as described in Materials and Methods. Data shown are the means ± sem from (A–D) 3 to 9 or (E) 3 donors. ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05; and ns, P > 0.05 when comparing (A and B) hexa‐1616 with penta‐1447 or (C and D) hexa‐1685 with hepta‐1855 at the indicated dose or (E) over the full curves.

As even low levels of monocyte contamination in neutrophil preparations can contribute to IL‐8 production [58], we next compared the PA lipid A variants using highly purified (>99% pure) neutrophil populations (Fig. 6B and D). The amount of IL‐8 released from these highly purified preparations was approximately one‐tenth of that from the less‐pure cultures, but the relative patterns of IL‐8 induction were nearly identical. As shown in Fig. 6B, hexa‐1616, although less potent, reached the same dose plateau as E. coli LPS, whereas penta‐1447 showed minimal IL‐8‐inducing activity. CF hexa‐1685 and hepta‐1855 lipid A variants were similarly potent in stimulating IL‐8 (Fig. 6D). These 2 lipid A isoforms were further compared for their potencies, triggering IL‐8 responses by pure neutrophils in more extensive dose curves. We calculated the log EC50 values, where EC50 represents the concentration of ligand that triggers one‐half of the maximum response. The difference between their log EC50 values was statistically significant, with hepta‐1855 having a lower value (more potent) than hexa‐1685 (Fig. 6E).

Characterizing the effect of CF‐specific PA lipid A variants on human monocytes

In addition to neutrophils, mononuclear phagocytes, namely monocytes or monocyte‐derived cells and macrophages, are implicated in lung inflammation [59, 60]. Our finding that the presence of a small proportion of mononuclear cells can greatly amplify the IL‐8 response of neutrophils to PA lipid A variants (Fig. 6) prompted us to determine whether those variants directly induce monocytes for differential IL‐8 responses. Interestingly, the IL‐8 response pattern was distinct from that found in neutrophils (Fig. 6). penta‐1447 reached the dose plateau as E. coli LPS, although with lower potency ( Fig. 7A ). CF hexa‐1616 was more potent than CF penta‐1447 in reaching the same dose plateau. The hexa‐1685 variant was the most potent of the PA lipid A variants, followed by hepta‐1855 (Fig. 7C). To determine if this monocyte‐specific pattern of response was true for other cytokines, we also compared PA lipid A variants for their induction of TNF‐α, another proinflammatory marker in CF lung disease [53, 57]. penta‐1447 showed significantly lower TNF‐α‐inducing activity throughout the dose range, never reaching the dose plateau as E. coli LPS (Fig. 7B). CF hexa‐1685 was the most potent of the lipid A variants in stimulating maximal TNF‐α, followed by hepta‐1855 and hexa‐1616, respectively (Fig. 7B and D).

Figure 7.

CF‐specific PA lipid A variants differentially modulate IL‐8 versus TNF‐α responses by human monocytes. (A and C) IL‐8 and (B and D) TNF‐α released from adherent monocytes after 20 h culture with the indicated dose range of PA lipid A variants or E. coli LPS, as quantified by ELISA. (E) Log EC50 of CF hexa‐1685 and hepta‐1855 in IL‐8, or (F) TNF‐α induction was calculated as described in Materials and Methods. Data shown are means ± sem from (A–D) 3 to 7 or (E and F) 3 individuals. ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05; and ns, P > 0.05 when comparing (A and B) hexa‐1616 with penta‐1447 or (C and D) hexa‐1685 with hepta‐1855 lipid A at the indicated dose or (E and F) over the full curves.

As our data also indicated a differential response of monocytes to hexa‐1685 versus hepta‐1855 at lower concentration, we further compared their potencies by calculating log EC50 values for induction of IL‐8 and TNF‐α (Fig. 7E and F). Although hepta‐1855 showed a similar potency as hexa‐1685 for induction of IL‐8, it was significantly less potent for TNF‐α. Taken together, these data indicate that PA lipid A variants modulate IL‐8 versus TNF‐α responses of monocytes differentially, with patterns that were, moreover, distinct from those seen in neutrophil populations.

Inhibitory activity of penta‐1447 PA lipid A

We next focused on characterizing CF penta‐1447, which we hypothesized is a partial TLR4 agonist that could interfere with the activities of other TLR4 agonists. To test our hypothesis, we first calculated the EC75 value of each of E. coli LPS, hexa‐1685, or hepta‐1855 PA lipid A for TLR4‐stimulating activity in HEK‐Blue TLR4 cells. The EC75 corresponds to the concentration of a ligand that leads to a 75% maximal response and is pharmacologically a more suitable target to evaluate the inhibitory effect of a candidate antagonist by remaining below the dose plateau. Reporter cells were treated with the EC75 doses of E. coli LPS, hexa‐1685, or hepta‐1855, alone or in the presence of increasing concentrations of penta‐1447. In parallel, a dose curve of penta‐1447 was performed, and as seen before (Fig. 2), penta‐1447 failed to trigger maximal activity on its own. When added to the EC75 doses of each of the other agonists, increasing amounts of penta‐1447 gradually reduced their activities to the penta‐1447 submaximal dose plateau ( Fig. 8 ), a characteristic pattern of partial agonists.

Figure 8.

CF‐specific penta‐1447 PA lipid A is a partial TLR4 agonist with inhibitory activity. HEK‐Blue TLR4 cells were stimulated for 24 h with the EC75 dose of E. coli LPS (1 ng/ml), CF hexa‐1685 (16 ng/ml), or CF hepta‐1855 (6 ng/ml) PA lipid A alone or in combination with increasing doses of penta‐1447. Yellow bars show the activity of penta‐1447 alone. TLR4 stimulation was measured by SEAP reporter assay, with normalization to the E. coli LPS dose plateau (E. coli = 100%). Bars show means ± sem from 3 independent experiments. ****P ≤ 0.0001, ***P ≤ 0.001 when compared with the corresponding agonist alone (1:0).

To determine if penta‐1447 also functions as a partial agonist/antagonist in primary cell cultures, human neutrophils were stimulated with EC75 doses of CF hexa‐1685, hepta‐1855, or E. coli LPS alone (1:0 wt:wt) or in the presence of increasing ratios of penta‐1447 (1:1, 1:2, 1:5, and 1:10). penta‐1447 significantly inhibited the IL‐8 response to hexa‐1685 ( Fig. 9A ). Although mixtures of penta‐1447 with hepta‐1855 are not physiologic, as they are not synthesized at the same time, we tested its effect on hepta‐1855 and found inhibition (Fig. 9B) beginning at a 1:2 ratio and increasing at greater ratios. Unlike the reporter cell assay, penta‐1447 did not show any inhibitory effect on E. coli LPS in neutrophil cultures (Fig. 9C).

Figure 9.

CF penta‐1447 PA lipid A efficiently inhibits IL‐8 production by other PA agonists in neutrophils but not in monocytes. (A–C) Pure neutrophils were stimulated for 20 h with the EC75 doses of CF hexa‐1685 (81 ng/ml), CF hepta‐1855 (36 ng/ml), or E. coli LPS (309 ng/ml), alone or with increasing mass ratios of CF penta‐1447. (D–F) Human monocytes were stimulated with the EC75 doses of CF hexa‐1685 (11 ng/ml), CF hepta‐1855 (25 ng/ml), or E. coli LPS (0.1 ng/ml), alone or with increasing mass ratios of CF penta‐1447. After culture, IL‐8 was quantified by ELISA. Each symbol represents 1 donor. Data are mean IL‐8 responses from (A–C) 4 or (D–F) 7 individuals. ****P ≤ 0.0001, ***P ≤ 0.001, *P ≤ 0.05; and ns, P > 0.05 when compared with the corresponding agonist alone (1:0) group.

To investigate the inhibitory effect of penta‐1447 in primary monocyte cultures, we treated the cells with EC75 doses of CF hexa‐1685, hepta‐1855, or E. coli LPS for inducing IL‐8 and TNF‐α in the presence of increasing amounts of penta‐1447 ( Figs. 9D–F and 10A–C ). Unlike neutrophils, there was generally no inhibition of IL‐8 responses detectable in monocyte cultures other than at the highest ratios of penta‐1447:hepta‐1855 (Fig. 9E). TNF‐α responses by monocytes to CF hexa‐1685 (Fig. 10A) or to E. coli LPS (Fig. 10C) were not inhibited by penta‐1447, a pattern similar to that of IL‐8 responses. However, significant inhibition of hepta‐1855 was seen when the amount of penta‐1447 was 5:1 or 10:1 of the agonist's EC75 (Fig. 10B). To determine if penta‐1447 had an inhibitory effect at receptor‐saturating concentrations, we stimulated adherent monocytes with >4× the EC75; 100 ng/ml CF hexa‐1685_, hepta‐1855, or E. coli LPS alone; or in the presence of 1000 ng/ml penta‐1447. Under these conditions, we found significant inhibition by penta‐1447 of TNF‐α responses to either PA lipid A variant (Fig. 10D). Together, these data suggest that penta‐1447 lipid A functions as a partial agonist that lacks strong proinflammatory activity on its own but can moderate the activity of other TLR4 agonists, especially that of hexa‐1685 lipid A, its coexpressed isomer.

Figure 10.

CF penta‐1447 PA lipid A is a weak inhibitor of TNF‐α responses induced by other PA lipid A agonists in monocytes. (A–C) TNF‐α responses by human monocytes stimulated with the EC75 doses of CF hexa‐1685 (11 ng/ml), CF hepta‐1855 (25 ng/ml), or E. coli LPS (0.7 ng/ml), alone or with increasing mass ratios of CF penta‐1447. (D) TNF‐α responses by monocytes stimulated with 100 ng/ml E. coli LPS, CF hexa‐1685, or CF hepta‐1855 PA lipid A, alone or with 1000 ng/ml CF penta‐1447. Each symbol represents 1 donor. Black symbols show responses to CF penta‐1447 plus the indicated agonist; colored symbols show agonist alone. Bars indicate the mean values from (A–C) 8 or (D) 4 individuals. ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05; and ns, P > 0.05 when compared with each agonist alone (1:0).

DISCUSSION

The distinct inflammatory activities associated with different lipid A fatty acyl isoforms synthesized by PA only in CF disease are important to understand. We manipulated the genetics and growth conditions of PA to isolate the major lipid A isoforms associated with CF and compared their immunostimulatory activities. Two major conclusions can be drawn from these comparisons: PA lipid A hepta‐1855 is an effective stimulator of TLR4 and of most but not all neutrophil responses we tested, and penta‐1447 is a weak/partial agonist of TLR4 that can inhibit responses to the other major lipid A isoform, hexa‐1685, which is present when penta‐1447 is.

One‐third of PA isolates from patients with late‐stage CF contains the hepta‐1855 lipid A variant instead of a mixture of penta‐1447 plus hexa‐1685 lipid A (Fig. 1) [15, 19]. Microbiological studies of PA with hepta‐1855 lipid A reveal that it confers resistance to β‐lactam antibiotics [unpublished results], perhaps as a selective adaptation to the frequency of antibiotic treatments needed to manage CF disease [61]. Whether conversion of PA lipid A from mixed penta‐1447/hexa‐1685 to the hepta‐acylated isoform increases or decreases immunostimulatory activity had been an open question given the range of immune activities observed for hepta‐acylated lipid A from other pathogens [26, 27, 62, 63]. Here, we report that relative to the well‐characterized hexa‐1685, hepta‐1855 lipid A had high activity in all neutrophil assays that we performed except for granule exocytosis. Hepta‐1855 stimulated TLR4 tested with the HEK 293 TLR4 reporter cell assay (Fig. 2), induced priming of extracellular oxidative bursts (Fig. 3), prolonged survival in ex vivo culture (Fig. 5), and triggered IL‐8 secretion from primary neutrophils (Fig. 6C) and monocytes (Fig. 7C) and TNF‐α secretion from monocytes at the physiologically relevant dose of 100 ng/ml (Fig. 7D). At the same dose, however, hepta‐1855 was markedly less efficient than hexa‐1685 in inducing secretory vesicle and specific granule exocytosis, as determined by surface display of CD35 and CD66b, respectively (Fig. 4A and B).

The relative inability of hepta‐1855 to induce neutrophil granule exocytosis could be related to a bacterial strategy of immune evasion. On the other hand, the strong IL‐8 induction and inhibition of neutrophil apoptosis by hepta‐1855 suggest that it could potentiate the influx and persistence of neutrophils, thus contributing to the characteristic CF airway neutrophilia. What is less clear is why PA is selected to retain a lipid A structure that is a robust activator of several neutrophil responses but selectively week at inducing degranulation. In the CF airway, despite limited oxygen availability, PA can grow to densities as high as 10⁹ cfu/ml, most likely through anaerobic respiration, using alternative electron acceptors, such as nitrites and nitrates [13, 64, 65–66]. As superoxide released by neutrophils favors production of nitrites and nitrates, hepta‐1855‐induced airway neutrophilia may help PA thrive by favoring extracellular ROS generation if release of antimicrobial peptides and proteases by degranulation is weak [65, 66].

Others have shown that priming of the respiratory burst response can be achieved as a result of granule exocytosis through delivery of components of the NADPH oxidase complex, such as gp91^phox and p22^phox, from granule membranes to the plasma membrane [44, 45, 67]. Here, a physiologic dose of hepta‐1855 had a priming effect without inducing degranulation, which indicates that a distinct mechanism is responsible. For example, signaling by hepta‐1855 may induce partial phosphorylation and translocation of cytosolic components, such as p47^phox, to the cell surface in sufficient quantities to facilitate oxidase activation [45, 67].

We also found that penta‐1447, the dominant isoform in all other infected CF patients, has the characteristics of a partial agonist, as it has a low‐dose plateau and can antagonize the activity of stronger TLR4 agonists. penta‐1447 was a weak TLR4 agonist in the HEK reporter cell assay (Fig. 2), induced limited cytokine responses from neutrophils (Fig. 6A and B) and monocytes (Fig. 7B), and inhibited responses to hexa‐1685 and hepta‐1855 (Fig. 9A and B). Weak agonistic activity of penta‐1447 is consistent with lipid A hypoacylation as an immunoevasive strategy of PA and other Gram‐negative bacteria [68, 69]. Moreover, the fact that penta‐1447 was antagonistic to hexa‐1685 in its induction of IL‐8 responses by neutrophils (Fig. 9A) suggests a potential role of this isoform in moderating hexa‐1685‐driven inflammation in CF airways. The inhibitory activity of penta‐1447 is consistent with other published observations. Bäckhed et al. [70] reported that a commercially available penta‐acylated LPS from PA antagonizes hexa‐acylated E. coli LPS in bladder epithelial cell cultures. Hypoacylated LPS from other pathogens, such as Shigella flexneri and Porphyromonas gingivalis, has also been shown to have antagonistic effects on TLR4 stimulation [68, 69].

Interestingly, IL‐8 responses by monocytes were not inhibited by penta‐1447 (Fig. 9D–F), which may be consistent with the reported amplifying autocrine loop that exists for IL‐8 in monocytes but not in neutrophils [71]. The fact that penta‐1447 was ineffective in inhibition of responses to the most potent TLR4 agonist used in this study—control E. coli LPS—is an indication that its inhibitory activity is moderate. As a partial agonist/antagonist, the relative abundance of penta‐1447 in a lipid A mixture may therefore be particularly critical in terms of understanding the progression of CF disease severity. For example, the presence of high amounts of antagonistic penta‐1447 might mitigate hexa‐1685‐driven inflammation in early stages, whereas its absence from the subset of late PA isolates expressing hepta‐1855 might enable more severe inflammation. These observations indicate that continued efforts to monitor the PA lipid A profile in patients can make invaluable contributions to our understanding of CF disease progression.

AUTHORSHIP

S.S. designed and performed all of the experiments, analyzed and interpreted the data, and wrote the manuscript. L.E.H. provided PA lipid A compounds. R.K.E. contributed PA lipid A compounds and provided expertise in Pseudomonas biology. S.M.U. provided the neutrophil expertise. T.C.M. provided the TLR4 expertise. R.K.E., S.M.U., and T.C.M. analyzed and interpreted the data and wrote and edited the manuscript.

ACKNOWLEDGMENTS

This work was supported by a grant from the U.S. National Institutes of Health (K99/R00 HL087924; to S.M.U.) and the Barnstable‐Brown Foundation (to T.C.M.). The authors thank Terri Manning for neutrophil isolation and for her expert technical help.