Apolipoprotein E Signals via TLR4 to Induce CXCL5 Secretion by Asthmatic Airway Epithelial Cells

Abstract

The primary function of APOE (apolipoprotein E) is to mediate the transport of cholesterol- and lipid-containing lipoprotein particles into cells by receptor-mediated endocytosis. APOE also has pro- and antiinflammatory effects, which are both context and concentration dependent. For example, Apoe^−/− mice exhibit enhanced airway remodeling and hyperreactivity in experimental asthma, whereas increased APOE levels in lung epithelial lining fluid induce IL-1β secretion from human asthmatic alveolar macrophages. However, APOE-mediated airway epithelial cell inflammatory responses and signaling pathways have not been defined. Here, RNA sequencing of human asthmatic bronchial brushing cells stimulated with APOE identified increased expression of mRNA transcripts encoding multiple proinflammatory genes, including CXCL5 (C-X-C motif chemokine ligand 5), an epithelial-derived chemokine that promotes neutrophil activation and chemotaxis. We subsequently characterized the APOE signaling pathway that induces CXCL5 secretion by human asthmatic small airway epithelial cells (SAECs). Neutralizing antibodies directed against TLR4 (Toll-like receptor 4), but not TLR2, attenuated APOE-mediated CXCL5 secretion by human asthmatic SAECs. Inhibition of TAK1 (transforming growth factor-β–activated kinase 1), IκKβ (inhibitor of nuclear factor κ B kinase subunit β), TPL2 (tumor progression locus 2), and JNK (c-Jun N-terminal kinase), but not p38 MAPK (mitogen-activated protein kinase) or MEK1/2 (MAPK kinase 1/2), attenuated APOE-mediated CXCL5 secretion. The roles of TAK1, IκKβ, TPL2, and JNK in APOE-mediated CXCL5 secretion were verified by RNA interference. Furthermore, RNA interference showed that after APOE stimulation, both NF-κB p65 and TPL2 were downstream of TAK1 and IκKβ, whereas JNK was downstream of TPL2. In summary, elevated levels of APOE in the airway may activate a TLR4/TAK1/IκKβ/NF-κB/TPL2/JNK signaling pathway that induces CXCL5 secretion by human asthmatic SAECs. These findings identify new roles for TLR4 and TPL2 in APOE-mediated proinflammatory responses in asthma.

Clinical Relevance

APOE (apolipoprotein E) is secreted by alveolar macrophages to function as a concentration-dependent pulmonary danger signal that augments pulmonary inflammatory responses. This study demonstrates that increases in APOE induce CXCL5 (C-X-C motif chemokine ligand 5) secretion by airway epithelial cells via a TLR4 (Toll-like receptor 4)/TAK1(transforming growth factor-β–activated kinase 1)/IκKβ (inhibitor of nuclear factor κ B kinase subunit β)/NF-κB p65/TPL2 (tumor progression locus 2)/JNK (c-Jun N-terminal kinase) signaling pathway. This defines new roles for TLR4 and TPL2 in APOE-mediated signaling and identifies a mechanism by which APOE may promote neutrophilic airway inflammation, which is relevant for the pathobiology of viral exacerbations of allergic asthma.

APOE (apolipoprotein E) is a 34-kD protein with two main structural motifs: an amino-terminal low-density lipoprotein (LDL) receptor binding domain and a carboxy-terminal lipid-binding domain that mediates interactions with multiple lipoprotein particles, including chylomicron remnants, very-low-density lipoprotein (VLDL), intermediate-density lipoprotein, and subclasses of high-density lipoprotein (1). The tertiary structure of the amino-terminal domain of APOE consists of a four-helix bundle arranged in an antiparallel fashion, and the carboxy-terminal domain is comprised of three amphipathic α-helices (2, 3). The amino- and carboxy-terminal domains are linked by a nonstructural hinge region (4). Humans have three APOE isoforms, E2, E3, and E4, which are encoded by the APOE ε2, ε3, and ε4 alleles on chromosome 19 (5), with ε3 being the most prevalent allele (6). The three protein isoforms differ by single base substitutions at residues 112 and 158 (1), which modify the tertiary structure of APOE and its binding affinity for the LDL receptor (2). Hence, although E3 and E4 possess similar binding affinity for the LDL receptor, the affinity of E2 is greatly reduced (7). The majority of APOE is synthesized by hepatocytes; however, APOE is also produced by macrophages as well as cells in the central nervous system (8–10). APOE plays an important role in lipoprotein metabolism by mediating the cellular uptake of lipoprotein particles, such as chylomicron remnants and VLDL, via binding to the LDL receptor, in a process known as receptor-mediated endocytosis (11, 12). In contrast, APOE-containing high-density lipoprotein particles can facilitate reverse cholesterol transport out of cells (13). In addition to its important role in lipid homeostasis, the APOE ε4 allele is a major genetic risk factor for Alzheimer’s disease and other neurodegenerative disorders (14). This predisposition is attributed mainly to accelerated accumulation of amyloid-β oligomers and plaques in the brain, which ultimately leads to neuronal cell death (15).

APOE is also expressed by multiple cell types in the respiratory system, including alveolar macrophages, type 1 and 2 pneumocytes, and pulmonary artery smooth muscle cells (16, 17). Experimental studies in murine model systems have demonstrated important roles of APOE in the lung (16). Apoe^−/− mice have impaired alveologenesis in utero and abnormal lung development later in life, manifesting as airspace enlargement and loss of gas exchange surface area, as well as enhanced airway hyperreactivity and altered elastic properties. In addition to its role in normal lung homeostasis, APOE has both protective and antiinflammatory properties in the setting of lung disease. For example, Apoe^−/− mice are susceptible to the development of neutrophilic lung inflammation, acute lung injury, and emphysema upon pulmonary exposure to cigarette smoke and nanoparticles, as well as granulomatous lung inflammation after being fed a high-fat diet (16). Apoe^−/− mice maintained on a high-fat diet also manifest pulmonary artery smooth muscle proliferation and pulmonary hypertension, which can be rescued by supplementation with APOE and adiponectin. Moreover, APOE gene expression was reduced in lung tissue from individuals with primary pulmonary hypertension (18). In a murine model of house dust mite (HDM)-induced asthma, both Apoe^−/− and Ldlr^−/− mice exhibited enhanced airway hyperreactivity and goblet cell hyperplasia (19).

Conversely, APOE also has proinflammatory effects in the lung and other organs. We recently reported that APOE activated the NLRP3 (NACHT, LRR and PYD domains-containing protein 3) inflammasome and induced IL-1β secretion from asthmatic BAL fluid (BALF) macrophages in a concentration-dependent fashion when APOE levels were increased above those found in lung epithelial lining fluid (ELF) from healthy volunteers (17). Furthermore, HDM-derived cysteine and serine proteases induced APOE secretion by asthmatic BALF macrophages via protease-activated receptor 2. This suggests that APOE can function as a concentration-dependent pulmonary danger signal that augments pulmonary inflammatory responses. APOE also induced phosphorylation of NF-κB, p38 MAPK (mitogen-activated protein kinase), and JNK (c-Jun N-terminal kinase) in murine peritoneal macrophages via MyD88 (myeloid differentiation primary response protein 88) (20). Similarly, APOE activated IL-6 secretion by subretinal mononuclear phagocytes via a TLR2 (Toll-like receptor 2)-CD14–dependent pathway (21). In rat hippocampal neurons, APOE induced the phosphorylation of ERK (extracellular signal–regulated kinase) with resultant activation of CREB (cAMP-response element binding protein) in a MEK (MAPK kinase)-dependent fashion (22). In a rat model of polymicrobial sepsis, mortality was increased in animals that received APOE via a mechanism involving enhanced presentation of lipid antigens (23).

APOE-mediated airway epithelial cell inflammatory responses have not been defined. Here, we sought to determine whether APOE modulates antiinflammatory or proinflammatory responses in asthmatic airway epithelial cells. We show that APOE induces both proinflammatory gene and cytokine secretion signatures in human bronchial brushing cells (HBBCs) from individuals with asthma. In particular, we demonstrate that APOE, at concentrations similar to those found in murine lung ELF during viral exacerbations of HDM-induced airways disease (17), promotes secretion of CXCL5 (C-X-C motif chemokine ligand 5) by human asthmatic small airway epithelial cells (SAECs) (24). Moreover, the signaling pathways by which APOE regulates cellular inflammatory responses are incompletely defined. Therefore, we characterized a signaling pathway by which APOE induces CXCL5 secretion by human asthmatic SAECs. We show that APOE-mediated CXCL5 secretion occurs via a TLR4-dependent pathway that activates TAK1 (transforming growth factor-β–activated kinase 1; MAP3K7) and the IκK complex. IκKβ (inhibitor of nuclear factor κ B kinase subunit β) then phosphorylates both NF-κB p65 and TPL2 (tumor progression locus 2; MAP3K8), with TPL2 mediating the downstream phosphorylation of JNK. Thus, we conclude that elevated levels of APOE in the airway activate a TLR4/TAK1/IκKβ/NF-κB p65/TPL2/JNK signaling pathway that induces CXCL5 secretion by airway epithelial cells. This defines new roles for TLR4 and TPL2 in APOE-mediated signaling and identifies a mechanism by which APOE may promote neutrophilic airway inflammation, which is relevant for the pathobiology of viral exacerbations of allergic asthma (17).

Methods

Research Participants

Research volunteers with asthma (n = 8) provided informed consent to participate in this study (National Heart, Lung, and Blood Institute research protocol 99-H-0076). Table E1 in the data supplement provides the demographic, anthropometric, and clinical characteristics of the participants. Eligibility was determined by detailed screening, including medical history, physical examination, laboratory investigations, and pulmonary function testing. Asthma was confirmed by a compatible medical history in conjunction with the presence of reversible airflow obstruction as evidenced by a significant response to inhaled short-acting β2-agonist by pulmonary function testing or bronchial hyperresponsiveness as demonstrated by a methacholine bronchoprovocation study.

RNA Sequencing

Total RNA was isolated using the Direct-Zol RNA MiniPrep kit (#R2052; Zymo Research) and quantified with a NanoDrop One Microvolume UV-Vis spectrophotometer (#ND-ONE-W; ThermoFisher Scientific).

Sequencing libraries were constructed from 100–500 ng of total RNA using the TruSeq Stranded Total RNA Library Prep (#20020596; Illumina) and Ribo-Zero rRNA Removal (#MRZH11124; Illumina) kits. The fragment size of the RNA sequencing (RNA-seq) libraries was verified using a 2100 Bioanalyzer instrument (#G2939BA; Agilent Technologies) and concentrations were determined using a Qubit 3 fluorometer (#Q33226; ThermoFisher Scientific). Libraries were loaded onto HiSeq 3000 (#SY-401–3001; Illumina) for 2 × 75 bp paired-end read sequencing. Fastq files were generated using bcl2fastq Conversion Software v2.20 (Illumina).

Signaling Pathway Inhibition

The TAK1 inhibitor Takinib (#SML2216) and the IκKβ inhibitor TPCA-1 (#T1452) were obtained from Sigma-Aldrich. The IκBα phosphorylation inhibitor BAY-11-7082 (#196871) and the TPL2 inhibitor CAS 871307-18-5 (#616373) were obtained from EMD Millipore. The MEK1/2 inhibitor PD98059 (#1213), the p38 MAPK inhibitor SB203580 (#1202), and the JNK inhibitor SP600125 (#1496) were obtained from Tocris Bioscience.

RNA Interference

Predesigned Stealth RNA interference (RNAi) siRNAs targeting MYD88, TLR2, TLR4, MAP3K7, IKBKB, MAP3K8, and MAPK8 (Table E2), as well as the siRNA negative control (#12935112) were obtained from Invitrogen. For reverse transfection, 7.5 μl of Lipofectamine RNAiMAX reagent (#13778; Invitrogen) and 2.5 pmol of siRNA were suspended in 250 μl of Opti-MEM reduced serum media (#31985062; ThermoFisher Scientific) and added into wells in 6-well plates for 20 minutes at room temperature. Human asthmatic SAECs, suspended in 2,250 μl of supplemented airway epithelial cell basal medium without antibiotics, were added at a density of 2.0 × 10⁵ cells per well and incubated for 96 hours, which resulted in a final siRNA concentration of 1 nM. Lipofectamine RNAiMAX reagent without siRNA was used as a vehicle control.

Statistical Analysis

Quantitative data are presented as percentages or mean ± SD or SEM. One-way ANOVA or Kruskal-Wallis tests were used to compared continuous measures. Wilcoxon matched-pairs rank tests were used to compared matched-paired continuous measures. The significance level was set at P < 0.05. Data analysis was performed with GraphPad Prism version 7.02.

Additional details are provided in the data supplement.

Results

APOE Induces Proinflammatory Gene Expression and Protein Secretion Signatures in Asthmatic HBBCs

To investigate the role of APOE in modulating HBBC function, cells from volunteers with asthma (n = ), which were comprised of 93.6% ± 1.6% epithelial cells, were divided into two equal batches and stimulated ex vivo with APOE (500 nM) or media as a control. As shown in Figure 1A, 21 of the 14,442 transcripts identified by RNA-seq demonstrated increased expression with a log₂ fold change > 1.0 (P and q values < 0.05) in APOE-treated asthmatic HBBCs. This APOE-induced gene signature was characterized by the increased expression of multiple proinflammatory genes, including IL36G, CXCL5, IL1B, CSF3, S100A8, CCL20, SAA1, SAA2, and IL1A. We next assessed whether the APOE-induced proinflammatory gene signature was accompanied by the increased secretion of the corresponding protein. As shown in Figures 1B–1D, the protein levels of CXCL5, CCL20, and G-CSF (granulocyte colony-stimulating factor; colony-stimulating factor 3) were increased in the media of APOE-treated cells, whereas IL-36γ, IL-1α, IL-1β, S100A8, SAA1 (serum amyloid A1), and SAA2 were below the limit of detection. A sensitivity analysis showed that the APOE-mediated increases in CXCL5 and G-CSF secretion remained statistically significant when the subject with the highest level of expression was removed from the analyses. These data show that APOE induces a proinflammatory gene expression signature in asthmatic HBBCs with associated increases in secretion of CXCL5, CCL20, and G-CSF.

Figure 1.

APOE (apolipoprotein E) induces proinflammatory gene expression and protein secretion signatures in asthmatic human bronchial brushing cells (HBBCs). (A) HBBCs from asthmatic volunteers (n = 8) were divided into two equal batches and stimulated ex vivo with APOE (500 nM) or media for 24 hours. Next-generation sequencing of mRNA transcripts was performed to identify genes that were upregulated by APOE, which was defined as a log₂ fold change (log2FC) > 1.0 with P and q values < 0.05. A volcano plot is presented, showing mRNA transcripts that were significantly increased in red, and mRNA transcripts that did not reach statistical significance in black. (B–D) Secretion of CXCL5 (C-X-C motif chemokine ligand 5) (B), CCL20 (C), and G-CSF (granulocyte colony–stimulating factor) (D) into the media was quantified by ELISA and compared between APOE (500 nM)- and media-treated cells (n = 8, *P < 0.01 vs. media, Wilcoxon matched-pairs signed rank test).

APOE Induces Two Concentration-Dependent Proinflammatory Cytokine Secretion Programs in Human Asthmatic SAECs

We then investigated whether the APOE-mediated cytokine secretion profile in human asthmatic SAECs exhibited concentration-dependent effects. SAECs were stimulated with APOE at concentrations that ranged from 500 nM, which is similar to the level present in plasma from healthy individuals and individuals with asthma (25–27), to 0.25 nM, which is similar to the level in lung ELF from healthy individuals (28). In addition, SAECs were stimulated with APOE (50 nM), which approximates the level (32 nM) that was previously shown to be present in ELF from a murine model of neutrophilic airway inflammation induced by HDM and the double-stranded RNA surrogate polyinosinic:polycytidylic acid, which mimics a concurrent viral-induced asthma exacerbation (17). As shown in Figures 2A–2H, APOE (500 nM) increased the secretion of CXCL5, G-CSF, SAA1, and IL-1β by human asthmatic SAECs, whereas IL-36γ, S100A8, CCL20, and IL-1α were not affected. In addition, APOE (50 nM) induced significant increases in CXCL5 secretion. Conversely, APOE (0.25 and 0.5 nM) reduced the secretion of IL-36γ, S100A8, and CCL20, whereas CXCL5, G-CSF, SAA1, IL-1α, and IL-1β were not affected. To assess whether the LPS present in the APOE preparation (0.0276 endotoxin units per microgram of protein) had an effect on cytokine secretion, SAECs were also stimulated with 0.047 ng/ml of LPS, which is comparable to the LPS content of APOE (500 nM). LPS alone did not modify the secretion of any cytokine, which shows that the APOE-mediated effects were not a consequence of LPS contamination. Taken collectively, these data show that APOE-mediated proinflammatory cytokine secretion by human asthmatic SAECs is concentration dependent and can be divided into two distinct programs: 1) upregulated secretion of CXCL5, G-CSF, SAA1, and IL-1β by high concentrations of APOE; and 2) reduced secretion of IL-36γ, S100A8, and CCL20 by low concentrations of APOE. Furthermore, 50 nM APOE, which is similar to the concentration present in ELF from a murine viral exacerbation model of allergic airways disease (17), induced increases in CXCL5 secretion. This is consistent with the conclusion that APOE can achieve physiologically relevant levels in the airway to mediate this effect.

Figure 2.

APOE induces two concentration-dependent proinflammatory protein secretion programs in human asthmatic small airway epithelial cells (SAECs). Human asthmatic SAECs were stimulated with APOE (0.25, 0.5, 5, 50, and 500 nM) or LPS (0.047 ng/ml) for 24 hours, with media as control. (A–H) Quantification of CXCL5 (A), G-CSF (B), SAA1 (serum amyloid A1) (C), IL-1β ( D), IL-36γ (E), S100A8 (F), CCL20 (G), and IL-1α (H) in the media was performed by ELISA (n = 12, *P < 0.05 vs. media and **P < 0.01 vs. media, one-way ANOVA with Dunnett’s multiple comparisons test [IL-36γ, CXCL5, G-CSF, S100A8, SAA1, and IL-1β], or Kruskal-Wallis with Dunn’s multiple comparisons test [CCL20 and IL-1α]). Pooled data from four independent experiments are shown, except for LPS (n = 9), where three independent experiments were performed.

APOE-induced Secretion of CXCL5 by Human Asthmatic SAECs Is Mediated via TLR4, TAK1, IκKβ, TPL2, and JNK

We next delineated the APOE-mediated signaling pathway that induces CXCL5 secretion by human asthmatic SAECs. We elected to focus primarily on the neutrophil chemotactic factor CXCL5 because it showed the largest increase in secretion by APOE-treated human asthmatic SAECs (Figure 2). Furthermore, a previous transcriptomic analysis of induced sputum from ex-smokers with severe asthma showed increased expression of CXCL5, as well as increased expression of genes involved in TLR signaling (29). Our RNA-seq analysis of APOE-treated asthmatic HBBCs similarly identified increased expression (as defined by P and q values < 0.05) of mRNA transcripts encoding TLR2, as well as multiple genes involved in TLR signaling (Figure 3A). In addition, APOE was reported to induce the phosphorylation of NF-κB, p38 MAPK, and JNK in a MyD88-dependent manner in murine peritoneal macrophages (20). We therefore assessed whether TLR2 or TLR4 mediates APOE signaling via Myd88 in human asthmatic SAECs. Although TLR2 mRNA transcripts were increased in our RNA-seq analysis of asthmatic HBBCs, APOE treatment did not modify the protein levels of TLR2, TLR4, or MyD88 in asthmatic SAECs (Figures 3B and E1). Therefore, we next considered whether APOE might instead signal via TLR2 or TLR4. SAECs were preincubated with anti-TLR2 and anti-TLR4 neutralizing antibodies for 1 hour, and APOE (500 nM) was added for 24 hours before the media were collected for quantification of CXCL5 (30–32). As shown in Figure 3C, neutralization of TLR4, but not TLR2, attenuated APOE-induced CXCL5 secretion. RNAi was also used to knock down the expression of TLR2, TLR4, and MyD88. As shown in Figures 3D and E1, RNAi-induced reductions in the protein expression of TLR4 and Myd88, but not TLR2, attenuated APOE-mediated CXCL5 secretion (Figure 3E). To demonstrate that this effect was not specific to asthmatic airway epithelial cells, we showed that neutralizing TLR4 antibodies similarly attenuated APOE-mediated CXCL5 secretion by normal human small airway cells (Figure E2). Collectively, these results demonstrate that APOE signals via TLR4 in human airway epithelial cells to induce CXCL5 secretion.

Figure 3.

APOE-induced secretion of CXCL5 by human asthmatic SAECs is mediated via TLR4 (Toll-like receptor 4). (A) Heat map showing expression of genes involved in TLR signaling from HBBCs stimulated with APOE (500 nM) as compared with media control. (B) Human asthmatic SAECs were stimulated with APOE at 0.5, 5, 50, or 500 nM for 24 hours, with media as control, and cellular proteins were analyzed by IB for the indicated targets. The Western blot shown is representative of three separate experiments. Densitometry is shown in Figure E1. (C) Human asthmatic SAECs were preincubated with a TLR2-neutralizing antibody, a TLR4-neutralizing antibody, or an IgG2a isotype control antibody (all at 10 μg/ml) for 1 hour. APOE (500 nM) was added for 24 hours, with media as the control. Quantification of CXCL5 in the media was performed by ELISA (n = 9, *P < 0.01 vs. APOE + media, one-way ANOVA with Dunnett’s multiple comparisons test). Pooled data from three independent experiments are shown. (D) Human asthmatic SAECs were transfected for 96 hours with siRNAs targeting TLR2 (siTLR2), TLR4 (siTLR4), and MYD88 (siMYD88) or negative control siRNA (siControl), all at a concentration of 1 nM, using Lipofectamine RNAiMAX reagent (Invitrogen). Two siRNAs (#1 and #2) were used for each target. The Western blot shown is representative of three separate experiments. The Western blot for TLR4 shows two regions from one Western blot. (E) Human asthmatic SAECs were transfected for 96 hours with siRNAs targeting TLR2, TLR4, and MYD88 or negative control siRNA, all at a concentration of 1 nM. APOE (500 nM) was added for 24 hours. Lipofectamine RNAiMAX reagent without siRNA was used as the vehicle control. CXCL5 secretion into the media was quantified by ELISA (n = 9, **P < 0.01 vs. APOE + vehicle, one-way ANOVA with Dunnett’s multiple comparisons test). Pooled data from three independent experiments are shown. (F) Balb/c mice received intranasal administration of PBS (control), IgG2a (10 μg), or a neutralizing anti-TLR4 antibody (10 μg) every morning, and PBS, APOE (17 μg), or LPS (170 pg, as a control for the amount of LPS present in APOE) every afternoon for 5 days. The mice were rested for 2 days, an additional boost was administered on day 8, and BAL was performed on day 9. (G and H) BAL fluid (BALF) LIX (LPS-induced CXC chemokine) (G) and BALF neutrophils (H). n = 12 mice per group, ***P < 0.001 and ****P < 0.0001, one-way ANOVA with Sidak’s multiple comparison test. NS = not significant.

Additional experiments were performed to confirm that APOE induces CXCL5 expression via TLR4 in vivo (Figure 3F) (17). As shown in Figure 3G, intranasal administration of an anti-TLR4 neutralizing antibody to wild-type mice inhibited APOE-mediated increases in BALF levels of LIX (LPS-induced CXC chemokine), a murine chemokine that is most closely related to human CXCL5, whereas a control IgG2a antibody did not (30, 33–36). Similarly, the anti-TLR4 neutralizing antibody suppressed APOE-mediated increases in BALF neutrophils (Figure 3H), which is consistent with the function of LIX as a neutrophilic chemotactic factor. Although APOE also induced increases in BALF macrophages and lymphocytes, these were not attenuated by the anti-TLR4 neutralizing antibody (Figure E3). Intranasal administration of the amount of LPS present in the APOE preparation did not increase BALF LIX or BALF inflammatory cell numbers, which shows that the APOE-mediated effects were not a consequence of LPS contamination. Collectively, these results show that intranasal administration of APOE to the murine lung induces increases in BALF LIX and BALF neutrophils via TLR4.

Having shown that APOE-induced CXCL5 secretion by human asthmatic SAECs is mediated by TLR4 and Myd88, we next sought to characterize the downstream signaling pathways. In particular, we considered whether the APOE-mediated signaling pathway that induced CXCL5 secretion involved TAK1, the IκK complex, NF-κB, JNK, p38 MAPK, TPL2, and MEK/ERK. Human asthmatic SAECs were preincubated for 1 hour with the following inhibitors before stimulation with APOE (500 nM) for 24 hours: the TAK1 inhibitor Takinib, the IκK inhibitor BAY-11-7082, the IκKβ (IκK-2) inhibitor TPCA-1, the JNK inhibitor SP600125, the p38 MAPK inhibitor SB203580, the TPL2 inhibitor CAS 871307-18-5 (4-(3-chloro-4-fluorophenylamino)-6-(pyridin-3-yl-methylamino)-3-cyano-(1, 7)-naphthyridine), and the MEK1/2 inhibitor PD98059. As shown in Figures 4A–4G, inhibition of TAK1, IκKβ, TPL2, and JNK, but not p38 MAPK or MEK1/2, attenuated APOE-mediated CXCL5 secretion by human asthmatic SAECs. The attenuation of APOE-mediated CXCL5 secretion by inhibitors of TAK1, IκKβ, TPL2, and JNK was confirmed by Western blotting (Figures 4H and E4).

Figure 4.

Inhibition of TAK1 (transforming growth factor-β–activated kinase 1; MAP3K7), IκK, TPL2 (tumor progression locus 2), or JNK (c-Jun N-terminal kinase) attenuates APOE-induced secretion of CXCL5 by human asthmatic SAECs. (A–G) Human asthmatic SAECs were preincubated with the TAK1 inhibitor Takinib (100 nM) (A), the IκKβ (IκK-2) inhibitor TPCA-1 (1 μM) (B), the IκK inhibitor BAY-11-7082 (10 μM) (C), the TPL2 inhibitor CAS 871307-18-5 (1.56 μM) (D), the JNK inhibitor SP600125 (10 μM) (E), the p38 MAPK (mitogen-activated protein kinase) inhibitor SB203580 (25 μM) (F), and the MEK1/2 (MAPK kinase 1/2) inhibitor PD98059 (25 μM) (G) for 1 hour. DMSO (0.1%) was used as the vehicle control. Cells were then stimulated with APOE (500 nM) for 24 hours and secretion of CXCL5 in the media was quantified by ELISA (n = 9, *P < 0.01 vs. APOE + vehicle, one-way ANOVA with Dunnett’s multiple comparisons test [Takinib, BAY-11-7082, CAS 871307-18-5, SP600125, and SB203580] or Kruskal-Wallis with Dunn’s multiple comparison tests [TPCA-1 and PD98059]). Pooled data from three independent experiments are shown. (H) Human asthmatic SAECs were preincubated with the TAK1 inhibitor Takinib (100 nM), the IκKβ (IκK-2) inhibitor TPCA-1 (1 μM), the IκK inhibitor BAY-11-7082 (10 μM), the TPL2 inhibitor CAS 871307-18-5 (1.56 μM), and the JNK inhibitor SP600125 (10 μM) for 1 hour. DMSO (0.1%) was used as the vehicle control. Cells were then stimulated with APOE (500 nM) for 24 hours and secretion of CXCL5 in the media was quantified by IB for the indicated targets. The Western blot shown is representative of three separate experiments. Densitometry for the three experiments is shown in Figure E3.

RNAi experiments were then performed to validate the roles of TAK1, IκKβ, TPL2, and JNK in the APOE-induced CXCL5 secretion signal transduction pathway in human asthmatic SAECs. As shown in Figures 5A–5D and E5, siRNAs targeting MAP3K7, IKBKB, MAP3K8, and MAPK8 attenuated the expression of TAK1, IκKβ, TPL2, and JNK, respectively. Furthermore, as shown in Figure 5E, the siRNA-mediated knockdown of MAP3K7, IKBKB, MAP3K8, and MAPK8 genes attenuated CXCL5 secretion. Taken together, these inhibitor and RNAi experiments show that TAK1, IκKβ, TPL2, and JNK signaling mediates APOE-induced CXCL5 secretion by human asthmatic SAECs. Furthermore, inhibition of APOE-induced CXCL5 secretion by the IκKβ (IκK-2) inhibitor TPCA-1, and siRNAs targeting IKBKB are consistent with the involvement of the canonical NF-κB pathway (37). In contrast, neither p38 MAPK, which was previously shown to be downstream of TAK1 (38), nor MEK1/2, which was previously shown to be downstream of TPL2 (39), participated in the APOE-mediated signaling pathway that induced CXCL5 secretion from human asthmatic SAECs.

Figure 5.

siRNA-mediated knockdown of MAP3K7, IKBKB, MAP3K8, and MAPK8 attenuates APOE-induced secretion of CXCL5 by human asthmatic SAECs. (A–D) Human asthmatic SAECs were transfected for 96 hours with siRNAs targeting MAP3K7 (siMAP3K7) (A), IKBKB (siIKBKB) (B), MAP3K8 (siMAP3K8) (C), and MAPK8 (siMAPK8) (D) or negative control siRNA (siControl), all at a concentration of 1 nM, using Lipofectamine RNAiMAX reagent (Invitrogen). Two siRNAs (#1 and #2) were used for each target. The Western blot shown is representative of three separate experiments. Densitometry for the three experiments is shown in Figure E4. The Western blot for JNK shows two regions from the same membrane. (E) Human asthmatic SAECs were transfected for 96 hours with siRNAs targeting MAP3K7, IKBKB, MAP3K8, and MAPK8 or negative control siRNA, all at a concentration of 1 nM. APOE (500 nM) was added for 24 hours. Lipofectamine RNAiMAX reagent without siRNA was used as the vehicle control. CXCL5 secretion into the media was quantified by ELISA (n = 9, *P < 0.05 vs. APOE + vehicle; **P < 0.01 vs. APOE + vehicle, one-way ANOVA with Dunnett’s multiple comparisons test). Pooled data from three independent experiments are shown.

Because IL-8 (CXCL8) plays a key role in neutrophil chemotaxis in humans, we also assessed whether APOE upregulated its secretion by asthmatic SAECs. As shown in Figure E6, APOE induced concentration-dependent increases in IL-8 that could be attenuated by both an anti-TLR4 neutralizing antibody and siRNA-mediated knockdown of TLR4 and MyD88, but not TLR2. In addition, pathway inhibition using either small-molecule inhibitors or RNAi showed that TAK1, IκKβ, TPL2, and JNK signaling also mediated APOE-induced IL-8 secretion by human asthmatic SAECs. This effect was not specific to asthmatic airway epithelial cells, as neutralizing TLR4 antibodies similarly inhibited APOE-mediated IL-8 secretion by normal human small airway cells (Figure E7).

APOE-mediated Activation of TAK1 and IκKβ Induces Downstream Signaling via NF-κB p65 and TPL2/JNK in Human Asthmatic SAECs

Western blots were performed to further characterize the APOE-mediated signaling pathway in human asthmatic SAECs. As shown in Figures 6 and E8, APOE induced the phosphorylation of IκKβ, NF-κB p65, TPL2, and JNK after 30 seconds of APOE stimulation. Next, human asthmatic SAECs were transfected with siRNAs targeting MAP3K7, IKBKB, MAP3K8, and MAPK8 for 96 hours before stimulation with APOE (500 nM) for 5 minutes. As shown in Figures 7 and E9, the siRNA-mediated knockdown of MAP3K7 attenuated the APOE-induced phosphorylation of IκKβ, NF-κB p65, TPL2, and JNK, whereas the knockdown of IKBKB attenuated the APOE-induced phosphorylation of NF-κB p65, TPL2, and JNK. In contrast, the siRNA-mediated knockdown of MAP3K8 attenuated the APOE-induced phosphorylation of JNK, but not NF-κB p65. Finally, the siRNA knockdown of MAPK8 did not affect the APOE-induced phosphorylation of TPL2. Collectively, these data support the conclusion that APOE activates TAK1 in human asthmatic SAECs, which in turn induces the downstream phosphorylation of IκKβ. Subsequently, IκKβ phosphorylates both NF-κB p65 and TPL2, which allows TPL2 to mediate the downstream phosphorylation of JNK (Figure E10). Furthermore, in contrast to prior studies using murine embryonic fibroblasts, phosphorylation of NF-κB p65 appeared to be independent of TPL2 signaling in human SAECs (39).

Figure 6.

APOE induces the phosphorylation of IκKβ, NF-κB p65, TPL2, and JNK in human asthmatic SAECs. (A–D) After a 6-hour serum deprivation, human asthmatic SAECs were treated with APOE (500 nM) at the times indicated and cellular proteins were analyzed by IB for IκKβ (A), NF-κβ p65 (B), TPL2 (C), and JNK (D). Except for TPL2, membranes were initially reacted with antibodies directed against the phosphorylated protein, stripped, and then reacted against the nonphosphorylated protein. Because an antibody that detected human TPL2 by Western blotting could not be identified, β-actin was used as a control for equivalency of protein loading for experiments using the anti-phospho-TPL2 antibody. The blots shown are representative of three separate experiments. Densitometry for the three experiments is shown in Figure E7.

Figure 7.

APOE-induced activation of TAK1 and IκKβ induces downstream signaling via NF-κB p65 and TPL2/JNK in human asthmatic SAECs. (A–D) Human asthmatic SAECs were transfected for 96 hours with siRNAs targeting MAP3K7 (siMAP3K7) (A), IKBKB (siIKBKB) (B), MAP3K8 (siMAP3K8) (C), and MAPK8 (siMAPK8) (D) or negative control siRNA (siControl), all at a concentration of 1 nM, using Lipofectamine RNAiMAX reagent (Invitrogen). Two siRNAs (#1 and #2) were used for each target. After a 6-hour serum deprivation, APOE (500 nM) was added for 5 minutes and cellular proteins were analyzed by IB. Except for TPL2, membranes were initially reacted with antibodies directed against the phosphorylated protein, stripped, and then reacted against the nonphosphorylated protein. The blots shown are representative of three separate experiments. Densitometry for the three experiments is shown in Figure E8.

Discussion

The major function of APOE is to facilitate the receptor-mediated endocytosis and transport of lipoprotein particles, such as VLDL and chylomicrons, and their cargo of cholesterol and lipids into cells. In the lung, the lipid transport function of APOE is important for both normal alveologenesis and the synthesis and secretion of pulmonary surfactant, which is essential for reducing surface tension within alveoli. In addition to its lipid transport capabilities, APOE has context-dependent effects that can either suppress or amplify inflammation and adaptive immune responses (16). For example, APOE can directly bind and neutralize LPS, which suppresses neutrophilic recruitment to the lung and the subsequent development of acute lung injury and septic shock. APOE mimetic peptides can also be transported into the cytoplasm, where they bind SET, which liberates protein phosphatase 2A (PP2A) and attenuates phosphorylation of p38 MAPK and Akt signaling pathways (40). In addition, APOE was shown to suppress IL-1β signaling in vascular smooth muscle cells via an interaction with LRP-1 (LDL receptor–related protein 1) that inhibited IRAK-1 (IL-1 receptor–associated kinase-1) phosphorylation (41). In contrast to LPS models of septic shock, administration of APOE to rats with polymicrobial sepsis increases mortality via a mechanism where APOE facilitates the presentation of endogenous lipid antigens (23). This promotes natural killer T-cell activation with augmented T-helper cell type 1 cytokine production and liver injury. Uptake of APOE-bound exogenous lipid antigens, such as those from Mycobacterium tuberculosis, also facilitates antigen uptake and presentation by dendritic cells to promote adaptive immune responses (42).

Similarly, APOE has context-dependent effects in asthma. After sensitization and challenge with HDM, which is a common aeroallergen that mediates asthma, Apoe-deficient mice displayed a phenotype of enhanced mucous cell metaplasia and airway hyperreactivity, whereas inflammation was not affected (19). In contrast, in a murine viral exacerbation model of HDM-induced asthma, APOE levels in lung ELF were increased and acted synergistically with HDM to prime and activate the NLRP3 inflammasome, with resultant secretion of mature IL-1β (17). This suggests that when levels are increased in lung ELF, APOE acts synergistically with HDM to induce IL-1β production and secretion, which may promote neutrophilic inflammation in the setting of viral exacerbations of HDM-induced asthma (17). Collectively, these findings demonstrate that APOE has both proinflammatory and antiinflammatory functions that are context and concentration dependent.

Here, we first sought to determine whether APOE augments or, alternatively, attenuates inflammatory responses by human asthmatic airway epithelial cells, and to assess whether these effects are concentration dependent. We used RNA-seq to show that 500 nM APOE, which is a concentration similar to that found in the blood of healthy individuals and individuals with asthma, induces the expression of multiple genes encoding proinflammatory proteins, including CXCL5, by asthmatic HBBCs. In addition, APOE at a concentration of 50 nM, which is physiologically relevant and similar to the concentration of APOE found in ELF (32 nM) from a murine viral exacerbation model of HDM-induced airways disease (17), stimulated the secretion of CXCL5 by human asthmatic SAECs. CXCL5 is a chemokine produced by lung bronchial and alveolar epithelial cells that recruits neutrophils to the lung via an interaction with its receptor, CXCR2 (C-X-C chemokine receptor 2) (24, 43). CXCL5 may also promote neutrophilic airway inflammation in asthma, as increased CXCL5 expression was previously found in sputum samples from cigarette-smoking patients with severe asthma and neutrophilic inflammation (29). However, we found that APOE at concentrations similar to the levels observed in lung ELF from healthy individuals suppressed the secretion of IL-36γ, S100A8, and CCL20. Thus, APOE has concentration-dependent effects, such that elevated levels promote and normal levels suppress the secretion of proinflammatory cytokines by human asthmatic SAECs.

An RNA-seq analysis of APOE-stimulated HBBCs also identified the increased expression of mRNA transcripts encoding TLR2, as well as TPL2 and NF-κB family members. Therefore, we considered whether TLR signaling might mediate the APOE-induced increases in CXCL5 secretion by human asthmatic SAECs. In support of this hypothesis, TLR2 and CD14 have previously been shown to mediate APOE-induced IL-6 secretion by subretinal mononuclear phagocytes (21). Another apolipoprotein, apolipoprotein A-I, has also been reported to signal via TLR2 and TLR4 to activate p38 MAPK, JNK, and NF-κB pathways in murine peritoneal macrophages (20). Furthermore, APOE-mediated activation of p38 MAPK, JNK, and NF-κB signal pathways in Myd88-deficient murine peritoneal macrophages was shown to be impaired (20). In contrast to these proinflammatory effects, APOE was found to suppress TLR4- and TLR3-mediated phosphorylation of JNK in RAW 264.7 macrophages (44). Here, we used neutralizing antibodies directed against TLR2 and TLR4 to show that APOE signals via TLR4, but not TLR2, in human asthmatic SAECs to induce CXCL5 secretion. Furthermore, we demonstrated that APOE administration stimulates increases in BALF LIX and BALF neutrophils via TLR4 in the murine lung.

Limited information exists regarding APOE-activated proinflammatory signaling pathways. APOE has previously been shown to mediate the MyD88-dependent phosphorylation of p38 MAPK and JNK, as well as to induce IκBα degradation in murine peritoneal macrophages (20). APOE has also been reported to phosphorylate and activate CREB via a calcium-dependent Rap-1/B-Raf/MEK/ERK pathway in hippocampal neurons (22). Here, we characterized the signaling pathway of APOE-induced CXCL5 secretion by human asthmatic SAECs, and identified the involvement of TLR4 and TPL2 in the APOE signaling pathway. We showed that APOE signals via TLR4 to activate TAK1 and IκKβ. IκKβ then phosphorylates both NF-κB p65 and TPL2, which in turn phosphorylates JNK, with the resultant induction of CXCL5 secretion by human asthmatic SAECs. However, the proinflammatory signaling pathway downstream of JNK that induces CXCL5 expression is not known.

TLRs are pattern recognition receptors that bind pathogen-associated molecular patterns, such as LPS in the case of TLR4, as well as endogenous ligands, such as those released from necrotic cells during sterile inflammation (45, 46). The signaling pathway downstream of TLR4 is complex. Upon ligand binding, Myd88 is recruited to TLR4 and forms a Myddosome complex containing IRAK4, IRAK1, IRAK2, and the E2 ubiquitin ligase TRAF6 (tumor necrosis factor receptor–associated factor 6) (47, 48). TRAF6 and the E2 UBC13 (ubiquitin-conjugating enzyme 13) then polyubiquitinate TRAF6, which recruits the TAK1 complex, comprised of the MAP3K TAK1, TAB2 (TAK1-binding protein), and TAB3 (49, 50). TAK1 phosphorylates MAP2Ks, such as MKK3/MKK6 (MAPK kinase 3/6) and MKK4/MKK7, which in turn phosphorylate MAPKs, such as p38 MAPK and JNK1/JNK2, respectively. TAK1 can also activate the IκK complex, which contains IκKα, IκKβ, and IκKγ (NEMO [NF-κB essential modulator]), which mediates the phosphorylation and degradation of IκBα (inhibitor of NF-κB), thereby liberating canonical NF-κB (p50 and p65 subunits) to translocate to the nucleus.

In unstimulated cells, TPL2 resides in a complex with NF-κB1 (p105) and ABIN2 (A20 binding inhibitor of NF-κB 2) (45, 51, 52). Activation of IκKβ in the IκK complex mediates the phosphorylation and activation of p105, which liberates TPL2 to phosphorylate the MAP2Ks MKK1 and MKK2, which in turn phosphorylate the MAPKs ERK1 and ERK2 (53, 54). TPL2 can also phosphorylate and activate MKK3/6, with the resultant phosphorylation of p38 MAPK (55, 56). In addition, tumor necrosis factor-α signaling in murine embryonic fibroblasts has been shown to activate MKK4 upstream of JNK, as well as MSK1 (ribosomal protein S6 kinase α-5) upstream of NF-κB p65 (39). Here, we show that TPL2 is downstream of IκKβ in APOE-stimulated human asthmatic SAECs and phosphorylates JNK. In addition, we show that the NF-κB pathway is not downstream of TPL2, and the MEK/ERK pathway is not involved in APOE-mediated CXCL5 secretion by human asthmatic SAECs. Of relevance, TPL2 signaling has been shown to have a protective effect in asthma, as Map3k8-deficient mice have exaggerated type 2 inflammation, as well as enhanced production of antigen-specific IgE (57, 58). Interestingly, the ability of TPL2 to negatively regulate asthmatic airway inflammation in mice was shown to be independent of its catalytic activity, and instead was a consequence of its adaptor function, which stabilized and prevented the degradation of ABIN2 (52). In contrast, our data show that the catalytic function of TPL2 may mediate proinflammatory effects in the airway when local concentrations of APOE are increased above basal levels, which may occur during viral exacerbations of HDM-induced asthma, via the phosphorylation of JNK.

In summary, the data presented in this study define a pathway by which APOE induces proinflammatory signaling and CXCL5 secretion in human asthmatic SAECs. We show that APOE signals via TLR4 to activate TAK1 and IκKβ, which are upstream of NF-κB p65 and TPL2, whereas TPL2 is upstream of JNK. To the best of our knowledge, this is the first time that TLR4 and TPL2 have been identified as participating in APOE signaling pathways. Furthermore, our data are consistent with the conclusion that when concentrations of APOE are elevated above the basal levels normally found in ELF, APOE may promote airway inflammation by inducing the secretion of CXCL5, which is an epithelial cell–derived chemokine with chemotactic activity toward neutrophils.