Mammalian target of rapamycin regulates neutrophil extracellular trap formation via induction of hypoxia-inducible factor 1 α

Neutrophils (polymorphonuclear leukocytes, PMNs) are key effector cells in infection, inflammation, and tissue injury.¹  Formation of neutrophil extracellular traps (NETs), first identified with human PMNs, is a function of neutrophils.²  NETs are complex lattices of decondensed chromatin that trap and kill bacteria, fungi, and some parasites by exposing them to high concentrations of NET-associated microbicidal factors.^3,4  Rapidly evolving studies indicate that NETs effect extracellular microbial killing while limiting the spread of pathogens in vivo.^3,5 

The intracellular signaling pathways that regulate NET formation by PMNs remain largely unknown. There is evidence that generation of reactive oxygen species (ROS) is a key event.^4,6  Nevertheless, we showed in primary human PMNs that NET formation requires signaling events and regulatory pathways in addition to ROS generation.⁴  Consistent with our results, subsequent studies in human HL-60 myeloid leukocytes and genetically altered mice indicate that activity of peptidylarginine deiminase 4, an enzyme responsible for chromatin decondensation, is also required.^5,7  Recent observations further suggest that NET formation requires enzymatic activity of neutrophil elastase (NE) and myeloperoxidase to initiate degradation of core histones that lead to chromatin decondensation before plasma membrane rupture.⁸  Furthermore, ROS generation and NET formation can be dissociated under some conditions.^9,10  Thus, molecular regulation of NET formation is complex and may involve multiple signaling pathways and effector events, depending on the neutrophil agonists and inflammatory context.

The mammalian target of rapamycin (mTOR) is a highly conserved PI3K-like serine/threonine kinase with functional homologs found in all studied eukaryotic organisms.¹¹  mTOR integrates nutrient, energy, oxygen sensing, and mitogenic input signals.¹²  We found that mTOR also responds to inflammatory signals and mediates a previously unrecognized pathway of posttranscriptional gene regulation in human PMNs.¹³  These results identified a new mechanism by which mTOR can regulate innate, as well as, adaptive, immune responses. Immunoregulatory activities of mTOR are now increasingly recognized.¹⁴ 

Recent observations indicate that hypoxia inducible factor 1α (HIF-1α), the regulated subunit of the transcription factor HIF-1, is a novel “command and control” protein in activities of myeloid leukocytes. In a murine model, conditional deletion of HIF-1α resulted in decreased glycolytic capacity with dramatically reduced cellular ATP levels, leading to impaired myeloid leukocyte aggregation, motility, extravasation, and microbial killing.¹⁵  In additional murine studies, bacterial infection induced expression of HIF-1α in macrophages even in the absence of hypoxia,¹⁶  a key environmental regulator of HIF-1α accumulation in tumor cells and other cell types.¹⁷  Bacterial infection also triggered HIF-1α–dependent synthesis of antimicrobial products, granule proteases, nitric oxide, and cytokines, and deletion of HIF-1α in mice decreased bactericidal capacity and resulted in systemic spread of pathogens.¹⁶  HIF-1α has thus been termed a “master regulator” of innate host defense responses.¹⁸  Nevertheless, the activities of HIF-1α in human neutrophils remain largely uncharacterized, and nothing is known about its roles in the regulation of NET formation by human PMNs.

Here, we report that mTOR regulates NET formation by human PMNs through induction of HIF-1α protein expression and show that inhibition of mTOR and HIF-1α signaling in these immune effector cells leads to decreased bacterial killing. Our findings identify a previously unrecognized pathway to NET formation with key nodal points of biologic control and potential for therapeutic manipulation. The importance of these observations may extend to NET-mediated vascular injury and inflammatory tissue damage.¹⁹ 

Lipopolysaccharide (LPS; Escherichia coli serotype 0111:134), poly-L-Lysine, Cytochalasin B, Cytochalasin D, glucose oxidase, paraformaldehyde, and N-(Methoxysuccinyl)-Ala-Ala-Pro-Val 4 nitroanilide were purchased from Sigma-Aldrich. Additional reagents included 2-methoxyestradiol (Sigma-Aldrich), rapamycin, diphenylene iodonium (DPI; Calbiochem), torin 1 (a generous gift from the Whitehead Institute for Biomedical Research and Dana-Farber Cancer Institute), TOPRO-3, DNA standard (Invitrogen), Syto Green and Sytox Orange (Molecular Probes), DNase (Promega), Anti-CD15⁺ impregnated microbeads (Miltenyi Biotec), medium 199 (Lonza Biologics), micrococcal DNase (Worthington Biochemical), and platelet-activating factor (PAF; Avanti Lipids).

PMNs were isolated from acid-citrate dextrose anticoagulated venous blood of healthy adults under protocols approved by the University of Utah Institutional Review Board. This study was conducted in accordance with the Declaration of Helsinki. PMN suspensions (> 96% pure) were prepared by positive immunoselection with the use of anti-CD15–coated microbeads and an auto-MACS (Miltenyi Biotec) and were resuspended at 2 × 10⁶ cells/mL concentration in serum-free M-199 media at 37°C.

HL-60 cells were obtained from ATCC and differentiated into surrogate PMNs with the use of 1.3% DMSO treatment over 5 days as previously described.¹³  Surrogate PMNs were then resuspended at a concentration of 2 × 10⁶ cells/mL in serum-free M-199 media at 37°C.

We transduced undifferentiated HL-60 leukocytes via lentiviral infection with pLVTHM-GFP plasmids²⁰  modified to constitutively express a short hairpin loop inhibitory-RNA (shRNA) against HIF-1α mRNA transcripts (www.genelink.com/siRNA/shrnaconstruct.asp; sequence: sense, 5′-caccggttaggtaatctggtaaggacgaatccttaccagattacctaacc-3′; antisense, 5′-aaaaggttaggtaatctggtaaggattcgtccttaccagattacctaacc-3′). As a control, PLVTHM-GFP was modified with a scrambled shRNA. Twenty-four hours after lentiviral transduction, the undifferentiated HL-60 cells were underwent FACS on the basis of green fluorescent protein (GFP) expression. High GFP-expressing cells were then differentiated as described in the previous section and resuspended at a concentration of 2 × 10⁶ cells/mL in serum-free M-199 media at 37°C in preparation for various assays described below.

The expression and sequence of mRNA coding for HIF-1α in adult human PMNs were determined with paired-end next-generation RNA sequencing, as previously described with the use of the Illumina GAIIx sequencer with the assistance of the University of Utah Bioinformatics core,²¹  and quantitative PCR (qPCR) with the following primer sets: HIF-1α sense (5′-agagccgcttcatttcttaga-3′) and antisense (5′-tatccaaatcaccagcatcca-3′) and GAPDH sense (5′-gaacatcatccctgcctctactg-3′) and antisense (5′-agcttgacaaagtggtcgttgag-3′). The estimated secondary structure energy of the HIF-1α 5′–untranslated region (UTR) was determined with the Vienna RNA Websuite.²² 

We assessed HIF-1α protein expression in human PMNs and SC Control and HIF-1αKD surrogate PMNs with the use of the In-Cell Western technique with a primary antibody against human HIF-1α protein (Novus Biologicals) and an 800-nm secondary antibody (Li-Cor Biosciences) on an Odyssey Infrared scanner (Li-Cor Biosciences). Sapphire700 dye (Li-Cor Biosciences) uptake allowed for normalization to cell number in these assays. In selected experiments, HIF-1α protein expression was also quantified with quantitative immunocytochemistry and ImageJ Version 1.46 (National Institutes of Health) image analysis software.

To show mTOR-dependent, translational regulation of HIF-1α protein expression, we pretreated human PMNs with either rapamycin (200nM), torin 1 (250nM), or cycloheximide (500nM) for 1 hour before a 1-hour incubation with LPS (100 ng/mL) or PAF (10nM) in 5% CO₂/95% air at 37°C. The cells in each well were then fixed (2% p-formaldehyde) and permeablized (0.1% Triton X-100). Imaging and fluorescence quantitation were performed with the Odyssey Imaging system and Version 3.0 software (Li-Cor Biosciences). We used the same technique to show inhibition of HIF-1α protein expression in HIF-1αKD surrogate PMNs.

Primary or surrogate PMNs (2 × 10⁶ cells/mL) were incubated with control buffer or stimulated with LPS (0.1-100 ng/mL) for 1 hour at 37°C in 5% CO₂/95% air on glass coverslips coated with poly-L-lysine and imaged with an Olympus FV300 1 × 81 confocal microscope and FluoView Version 2.1c software as previously described.⁴  NETs were detected with a mixture of cell permeable (Syto Green; Molecular Probes) and impermeable (Sytox Orange; Molecular Probes) DNA fluorescent dyes. For selected experiments, primary or surrogate PMNs were preincubated with rapamycin (200nM), FK-506 (200nM), 2 methoxyestradiol (2-ME2; 2μM), vinblastine (10μM), or paclitaxol (10μM) for 1 hour before imaging.

We determined supernatant total histone H3 content as follows. After live cell imaging of control and LPS-stimulated primary or surrogate PMNs (2 × 10⁶/mL; 100 ng/mL), the cells were incubated with new media containing DNase (40 U/mL) for 15 minutes at room temperature to break down and release NETs formed in response to LPS stimulation. The supernatant was gently removed and centrifuged at 420g for 5 minutes. The cell-free supernatant was then mixed 3:1 with 4× Laemelli buffer before Western blot analysis. We used a polyclonal primary antibody against human histone H3 protein (Cell Signaling Technology) and infrared secondary antibodies (Li-Cor Biosciences). Imaging and densitometry were performed on the Odyssey infrared imaging system (Li-Cor Biosciences). Comparison of histone H3 quantitation with other surrogates for NET formation⁴  is shown in supplemental Figure 2 (available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

We determined bacterial killing efficiency of primary and surrogate PMNs as previously described.⁴  Leukocytes were incubated for 30 minutes at 37°C in 5% CO₂/95% air alone or with the phagocytosis inhibitors Cytochalasin B and D (10μM). The leukocytes were then stimulated with LPS (100 ng/mL), placed in poly-L-lysine–coated wells of a 24-well tissue culture plate, and incubated at 37°C for 1 hour to induce cellular activation and formation of NETs. To inhibit NET-mediated bacterial killing, we incubated selected wells with DNase (40 U/mL) for 15 minutes before adding bacteria. E coli (1 colony forming unit/100 PMNs) were added to the PMNs, followed by continued incubation for 2 hours. The PMNs were then permeabilized with 0.1% Triton X-100 for 10 minutes, and each well was scraped to free all cells. Serial dilutions were performed, and bacterial cultures were grown on 5% sheep blood agar plates (Hardy Diagnostics). After a 24-hour incubation, bacterial counts were determined. Total, phagocytotic, and NET-mediated bacterial killing were determined as described.⁴ 

Statistical analyses were performed with Stata 11.0 (Stata Corporation) and Prism 5.02 (GraphPad software). Descriptive statistics are reported as the mean ± SEM. For parametric results that compared 2 groups, we used an unpaired, single-tailed Student t test, whereas the 1-way ANOVA with Tukey Multiple Comparisons posthoc testing was used for comparison of 3 or more groups. For nonparametric results that compared 2 groups we used the 2-sample Wilcoxon rank sum test, and for 3 or more groups we used the Kruskal-Wallis equality-of-populations rank test with 2-sample Wilcoxon rank sum post hoc testing. A P value of < .05 was considered statistically significant.

We first determined whether specific pharmacologic knockdown of mTOR activity with rapamycin²³  altered NET formation by stimulated PMNs assessed by direct microscopic observation as previously described.⁴  We found that rapamycin, a selective blocker of mTOR activity,²⁴  inhibits assembly of NETs by PMNs stimulated with LPS, whereas FK-506 did not (Figure 1A). FK-506, like rapamycin, binds to the regulatory molecule FKBP12, but in contrast, does not inhibit mTOR activity,²⁵  and thus served as a control. In a complementary strategy that used genetic manipulation, we also attempted to silence mTOR in differentiated HL-60 surrogate PMNs with the use of siRNA. The cells with mTOR knocked down, however, repeatedly failed to divide and grow, consistent with known activities of mTOR¹² ; therefore, experiments that examined NET formation under these conditions could not be accomplished. NETs are 3-dimensional structures that, in principal, must be visually assayed. Nevertheless, markers such as NE activity and extracellular DNA associated with NETs are used as surrogate assays for NET formation.^4,6  We used histone release in this fashion and found that treatment of PMNs with rapamycin inhibited release of histone H3 by LPS-stimulated PMNs, whereas the control agent FK-506 did not (Figure 1B). These are the first observations that link the mTOR signaling pathway to NET formation.

We have previously shown that mTOR exerts specialized translational control and regulates synthesis of key proteins in activated human PMNs.¹³  Furthermore, mTOR regulates specialized translation in other cell types.^11,12  We considered HIF-1α a candidate for specialized posttranscriptional regulation by mTOR.

We first probed HIF-1α messenger RNA (mRNA) transcript structure and expression in control and stimulated human PMNs with the use of paired-end next-generation RNA sequencing (RNA-seq) and qPCR.²¹  mRNA for full-length HIF-1α was present in freshly isolated, unstimulated PMNs (Figure 2A top). Consistent with this, HIF-1α protein was also basally present under normoxic conditions (Figures 2D-E and 3A). Analysis of HIF-1α transcript expression after stimulation with LPS showed a less than 1.5-fold increase in multiple experiments (Figure 2A top, C). In contrast, there was a 2- to 4-fold increase in HIF-1α protein expression (Figure 2D-E), even though the increase in HIF-1α mRNA was minimal, a pattern indicating a major element of posttranscriptional control.²⁶  For comparison, we found a 5- to 8-fold increase in interleukin 8 (IL-8) mRNA expression in response to stimulation of PMNs under the same conditions (Figure 2A bottom). In previous experiments, we have shown that IL-8 protein synthesis is regulated by transcription in activated human PMNs.^13,27 

We examined the sequence of the 5′-UTR of the HIF-1α transcript identified by RNA-seq in PMNs (Figure 2A) and found that it exhibited substantial complexity with a base pair length of 339 and estimated secondary structure energy of −182.4 kcal/mol (Figure 2B). These features are characteristic of transcripts whose translation is controlled by mTOR in nonimmune cells²⁸  and in human PMNs.¹³  Consistent with this, treatment of PMNs with rapamycin inhibited increased HIF-1α protein expression in stimulated PMNs, as did cycloheximide (Figure 2D-E). In addition, a second selective inhibitor of mTOR complex 1 activity, torin 1, had a similar effect (Figure 2E). These results indicated that HIF-1α is downstream in a pathway regulated by mTOR in human neutrophils, a previously unrecognized signaling cascade in this cell type.

Conditions that induced NET formation by PMNs, including LPS stimulation, also induced HIF-1α protein expression (Figures 2–3A). Furthermore, PAF, an inflammatory agonist that activates mTOR in human PMNs¹³  and induces NET formation,⁴  stimulated expression of HIF-1α protein in an mTOR-dependent manner (Figure 2F). In addition, we found enhanced HIF-1α protein expression in PMNs in areas of active NET formation by immunocytochemical analysis (Figure 3A).

We next treated human PMNs with cobalt chloride, a hypoxia mimetic that increases cellular HIF-1α protein expression,²⁹  and found it induced NET formation (supplemental Figure 5). To further examine this relation, we treated PMNs with 2-ME2, a chemotherapeutic inhibitor of HIF-1α signaling,³⁰  and assayed NET formation both qualitatively and quantitatively. We found that 2-ME2 blocked NET formation by LPS-stimulated PMNs (Figure 3B-C). Antimicrotubule activities of 2-ME2 have also been reported,³⁰  so we examined the effect of paclitaxol and vinblastine, known antimicrotubule agents, on NET formation. Neither drug inhibited NET deployment by stimulated PMNs (Figure 3B-C). Furthermore, 2-ME2 inhibited vascular endothelial growth factor transcript expression, a key downstream mRNA with expression dependent on the HIF-1 transcription factor, whereas vinblastine did not (supplemental Figure 4). We also found that inhibition of endogenous ROS with DPI failed to inhibit vascular endothelial growth factor transcript expression, suggesting that ROS generation was not required for HIF-1α's activity as a subunit of the HIF-1 transcription factor (supplemental Figure 3). These experiments indicated that, in response to LPS, NETosis and NET formation³¹  involve signaling to HIF-1α.

We next sought to confirm the contribution of HIF-1α to NET formation by genetic inhibition. Because primary PMNs are not readily tractable to genetic manipulation, we used differentiated HL-60 leukocytes as surrogates for PMNs as previously described.¹³  As previously reported, we found that these cells form NETs in response to LPS stimulation (supplemental Figure 1).⁷  With the use of lentiviral infection to transduce undifferentiated HL-60 leukocytes,³²  we inhibited HIF-1α protein expression with the use of shRNA constructs that target HIF-1α mRNA (HIF-1αKD). Constructs encoding a scrambled shRNA sequence served as controls (SC Control). Both lentiviral constructs constitutively expressed GFP as a marker of transduction. The HL-60 leukocytes were then differentiated into surrogate PMNs.⁷  We achieved a substantial reduction in HIF-1α protein expression in the HIF-1αKD cells compared with the SC Control cells under control and LPS-stimulated conditions (Figure 4A-B). We then assessed the ability of HIF-1αKD and SC Control surrogate PMNs to form NETs with the use of confocal microscopy and histone H3 quantitation. NET formation in HIF-1αKD cells was dramatically decreased after a 2-hour incubation with LPS compared with the SC Control cells assayed in parallel (Figure 4C-D). This result was consistent with pharmacologic knockdown of HIF-1α in primary human PMNs (Figure 3B-C). These results showed for the first time that HIF-1α regulates NET formation.

We next determined whether inhibition of mTOR signaling and HIF-1α activity decreased NET-mediated extracellular bacterial killing. We first performed bacterial killing assays as previously described⁴  by incubating a pathogenic strain of E coli bacteria with LPS-stimulated primary human PMNs pretreated for 1 hour with either rapamycin or FK-506. Total and extracellular NET-mediated bacterial killings were significantly decreased in rapamycin-pretreated PMNs compared with FK-506–pretreated and control PMNs (Figure 5A). A trend toward decreased phagocytotic bacterial killing was also noted (Figure 5A). This result was consistent with inhibition of NET formation by rapamycin (Figure 1) and suggested that mTOR also regulated cellular pathways required for phagocytotic as well as NET-mediated, extracellular microbial killing.³³ 

In parallel experiments, we pretreated human PMNs with 2-ME2 for 1 hour before LPS stimulation and determined phagocytotic and extracellular NET-mediated bacterial killing as described above. We found that pharmacologic inhibition of HIF-1α activity significantly decreased NET-mediated extracellular bacterial killing by 35% while not affecting phagocytotic bacterial killing (Figure 5B).

Finally, we determined whether genetic inhibition of HIF-1α protein expression affected extracellular NET-mediated bacterial killing. We incubated LPS-stimulated HIF-1αKD and SC Control cells with E coli bacteria as described above and determined phagocytotic and extracellular NET-mediated bacterial killing. We found significantly decreased NET-mediated extracellular bacterial killing in HIF-1αKD cells compared with the SC Control cells, with no difference detected in phagocytotic bacterial killing between the 2 groups of cells (Figure 5C). Thus, pharmacologic inhibition of HIF-1α in primary PMNs and genetic knockdown of HIF-1α in surrogate neutrophils inhibit NET-mediated bacterial killing.

Although the importance of NET formation in protection from microbial invasion is well established,^10,34,35  the pathologic power of dysregulated NET formation is now increasingly recognized. Dysregulated NET formation, occurring in response to pathologic stimuli, may contribute to sepsis,¹⁹  small vessel vasculitis,³⁶  and systemic inflammatory response syndrome.^19,37  Recent studies also document the importance of dysregulated NET formation in vascular injury associated with systemic lupus erythematosus.^38,39  Yet the regulatory pathways governing NET formation remain incompletely characterized. Others have highlighted the importance of downstream effector molecules such as ROS, peptidylarginine deiminase 4, and the granule enzymes NE and myeloperoxidase for NET formation,^5,6,8  but knowledge of the upstream regulatory molecules governing NET formation remains elusive.

Here, we link for the first time the mTOR signaling pathway to NET formation by human PMNs and characterize mTOR regulation of HIF-1α expression in primary human PMNs. Our findings also show for the first time that HIF-1α regulates NET formation and indicate that HIF-1α is downstream in a pathway regulated by mTOR in human neutrophils, a previously unrecognized signaling cascade in this critical immune effector cell. Earlier reports indicate that mTOR regulates HIF-1α protein accumulation and stability in cancer cells and cell lines,^29,40-42  but this mechanism has not been previously demonstrated in primary human inflammatory cells. In other cell types, HIF-1α protein accumulation has additional complex regulatory elements, including alterations in proteolytic degradation.^43,44  Thus, it may also be controlled at multiple checkpoints in PMNs, depending on the physiologic context.⁴⁵ 

Our studies found that regulation of NET formation by mTOR operates at least in part through translational control of HIF-1α protein expression. Pharmacologic and genetic “knockdown” of mTOR and HIF-1α expression in primary human and surrogate PMNs result not only in decreased NET formation, assessed both qualitatively and quantitatively, but also in decreased NET-mediated antibacterial activity. Thus, 2 “master” molecular regulators, which are nodal points of cellular control that respond to a variety of inflammatory and metabolic signals, influence the deployment of NETs by human PMNs.

Our studies do not, however, determine the mechanism through which HIF-1α regulates NET formation. Although HIF-1α is widely recognized as the regulated subunit of the HIF-1 transcription factor which governs the cellular response to hypoxia, the rapidity of the PMN response to stimulation with NET formation suggests that other, more direct mechanisms may be at play. Of interest is the described role of HIF-1α in the regulation of ROS generation.⁴⁶  Initial experiments in human PMNs found that treatment with exogenous ROS restores NET formation to PMNs pretreated with DPI, a pharmacologic inhibitor of NADPH oxidase.^4,6 

Our observations have translational relevance. Agents known to increase HIF-1α activity enhanced neutrophil bacterial killing in a previous study.⁴⁴  The antimicrobial activity was extracellular and was eliminated by DNase treatment. This combination of characteristics is unique to NET-mediated antibacterial killing.^2,4,6  Although no increase in NET formation was detected in that report,⁴⁴  our findings suggest that HIF-1α is a point of molecular regulation of NET formation that may be positively manipulated by pharmacologic interventions in syndromes of microbial invasion or in syndromes of innate immune deficiency, such as the complex neutrophil dysfunction exhibited by human neonates.⁴  In parallel, identification of mTOR and HIF-1α as regulators of NET formation suggests that one or both may be molecular points of intervention in sepsis, lupus erythematosus, and vasculitis, where NET deployment appears to be injurious to the host rather than protective.^19,36,38 

The authors thank Jenny Pearce, Alex Greer, and Diana Lim for their help with preparation of the manuscript and figures and Andrew Weyrich and Guy Zimmerman for advice and collaboration in the studies reported and for suggestions and editorial help during drafting of this manuscript.

This work was supported by the National Institutes of Health (grant K08 HD049699, C.C.Y.), The Children's Health Research Center of the University of Utah, and the Primary Children's Medical Center Foundation (award to C.C.Y.).

National Institutes of Health

Contribution: A.M.M. performed, directed, and interpreted experiments and wrote major portions of the paper; M.J.C., E.A.E, J.W.G., and J.W.R. performed experiments, provided key experimental approaches, and/or interpreted data; M.T.R. provided statistical expertise and helped interpret data; and C.C.Y. provided overall direction for the project, reviewed and analyzed all experiments, wrote sections of the paper, and edited the entire paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Christian C. Yost, Pediatrics/Neonatology, University of Utah, Williams Building, 295 Chipeta Way, Salt Lake City, UT 84108; e-mail: christian.yost@u2m2.utah.edu.