Myeloid Cell–Derived HIF-1α Promotes Control of Leishmania major

Hypoxia inducible factor-1α (HIF-1α) promotes the adaption of cells to low O₂ microenvironments. Low O₂ conditions result in HIF-1α stabilization by impairing the enzymatic activity of prolyl hydroxylase domain enzymes (PHDs) and subsequent hydroxylation of two critical proline residues that target HIF-1α for proteasomal degradation (reviewed in Refs. 1, 2). However, under conditions of ample O₂, HIF-1α accumulation can be induced by a plethora of other nonhypoxic signals (reviewed in Refs. 3–5). Accordingly, HIF-1α accumulation was found in mouse and human tissues infected with various pathogens (6).

In myeloid cells, exposure to inflammatory and pathogen-associated molecular pattern molecules, such as LPS, results in HIF-1α accumulation even under normoxic conditions, which requires transcriptional and posttranslational signaling events (7–10). This inflammatory HIF-1α stabilization in myeloid cells is necessary for their proinflammatory, antibacterial, and antifungal activity (11–15). Moreover, inflammatory HIF-1α expression in myeloid cells provides the metabolic basis for innate immune memory termed “trained immunity” (16).

In parasitic infections, the role of HIF-1α is less clear. In bovine macrophages infected with Theileria annulata, HIF-1α is required to transform these cells into aggressive leukemic-like cells (17). Toxoplasma gondii and Leishmania donovani induce HIF-1α stabilization in host cells to promote their survival within the intracellular niche (18, 19). In a mouse model of visceral leishmaniasis elicited by L. donovani, HIF-1α signaling in CD11c⁺ dendritic cells (DCs) was linked to an impaired DC function and CD8⁺ T cell response and supported parasite survival (20). Leishmania amazonensis infection of mice led to accumulation of HIF-1α in cutaneous lesions, but the functional role of HIF-1α in this model of nonhealing progressive cutaneous leishmaniasis has not been investigated (21, 22). There is circumstantial evidence that HIF-1α signaling might affect microRNA-210 expression in self-healing human leishmaniasis caused by L. major (23). However, it is unknown whether HIF-1α expression is necessary for the control of self-healing L. major infections in vivo. Given that control of cutaneous L. major infection requires type 2 NO synthase (NOS2)-dependent production of leishmanicidal NO (24–27) and that HIF-1α promotes the expression of NOS2 in macrophages (12, 28–30), we hypothesized that HIF-1α might support macrophage-driven antileishmanial control. Therefore, we tested the contribution of myeloid cell–specific HIF-1α to antileishmanial defense in vitro and in vivo.

LPS (Escherichia coli O111:B4) was purchased from Sigma-Aldrich (Taufkirchen, Germany). Recombinant murine IFN-γ and TNF were purchased from eBioscience (Frankfurt, Germany). 2-(1-Chloro-4-hydroxyisoquinoline-3-carboxamido) acetate (ICA) was synthesized as previously described (31) and kindly provided by Prof. Nicolai Burzlaff (Department of Chemistry and Pharmacy, Friedrich-Alexander-Universität Erlangen-Nürnberg).

Promastigotes of L. major strain MHOM/IL/81/FEBNI were propagated as described earlier (32, 33) in vitro in RPMI 1640 (10% FCS) on Novy-MacNeal-Nicolle blood agar slants for a maximum of five passages.

For flow cytometry analysis, fluorochrome-labeled Abs against CD11b (M1/70; BD Biosciences), NOS2 (6/iNOS/NOS Type II; BD Transduction Laboratories, Heidelberg, Germany), and the respective isotype controls were used. For Western blot analysis, primary rabbit Abs were used against the following mouse target proteins: actin (catalog [cat.] no. A2066; Sigma-Aldrich, Taufkirchen, Germany), HIF-1α (cat. no. 10006421; Cayman Chemical, Ann Arbor, MI), and HSP90 (cat. no. sc-7947, Santa Cruz Biotechnology, Heidelberg, Germany). HRP-conjugated polyclonal swine anti-rabbit IgG (P0399; Dako, Hamburg, Germany) was used as secondary Ab.

All animal experiments were carried out according to protocols approved by the Animal Welfare Committee of the local governmental authorities (Regierung von Mittelfranken, Ansbach and Regierung von Unterfranken Würzburg, Germany). Hif1α^flox/flox mice (34) were crossed with lysozyme M-driven Cre (LysM^Cre)-transgenic mice (35), both on a C57BL/6 background (36), as described earlier (37), to obtain mice deficient for Hif1α in their myeloid cell compartment (Hif1α^Δmyel). At the age of 8–9 wk, female C57BL/6 Hif1α^Δmyel mice and littermate controls were injected in the hind footpads with 3 × 10⁶ stationary-phase L. major promastigotes (of low in vitro passage [≤5]) in 50 μl of PBS to induce infection.

Paraffin sections (2–4 μm) were dewaxed in xylene and rehydrated in a series of ethanol washes. The following primary Abs were used for immunodetection: polyclonal rabbit anti–HIF-1α (1:10,000; cat no. 10006421; Cayman Chemical) and polyclonal rabbit anti-mouse CD68 (1:500; cat. no. ab125212; Abcam, Cambridge, U.K.). Biotinylated secondary goat anti-rabbit (cat. no. BA1000; Vector Laboratories, Peterborough, U.K.) was used at a dilution of 1:1000. Abs were detected by a catalyzed signal amplification system (Dako) and counterstained. Use of human tissue in this study was approved by the Yale Human Investigation Committee, as part of the Yale Dermatopathology Stored Specimen Repository.

In vitro restimulation of immune cells from popliteal lymph nodes of infected animals was performed as described earlier using soluble Leishmania Ag (26). Quantification of the number of parasites in the tissue was determined by limiting dilution analysis, as described earlier (26).

Mouse bone marrow–derived macrophages (BM-MΦs) were generated from C57BL/6 mice (Charles River Breeding Laboratories, Sulzfeld, Germany), Hif1α-deficient (Hif1α^Δmyel) mice, or control mice and cultured in Teflon bags (FT FEP 100 C; Dupont) purchased via American Durafilm (Holliston, MA), as described earlier (38, 39). On day 7–8, macrophages were harvested and transferred onto polystyrene plates. Infection was carried out by coculturing of BM-MΦs with L. major promastigotes at a 1:30 ratio for 4 h under normoxic conditions (37°C, 5% CO₂, ∼20% O₂), as described earlier (26). Postinfection, extracellular L. major promastigotes were washed using PBS and cultured in the absence or presence of TNF (20 ng/ml), TNF/IFN-γ (each at 20 ng/ml), or LPS/IFN-γ (10 and 20 ng/ml, respectively) for 18–24 or 72 h under normoxic or hypoxic conditions. Normoxic conditions were achieved in a regular humidified incubator (37°C, 5% CO₂, ∼20% O₂), whereas hypoxic conditions (37°C, 5% CO₂, 2.8% O₂) were generated using an adjustable hypoxic chamber suitable for cell culture experiments (Whitley H35 hypoxystation purchased via Meintrup DWS Laborgeräte, Herzlake, Germany). After 72 h, the cell cultures were analyzed microscopically using Diff-Quik staining (Medion Diagnostics, Düdingen, Switzerland) to determine the percentage of infected cells and the number of parasites per infected cell. The parasite load was calculated as the number of L. major in ∼100 BM-MΦs after 72 h/the number of L. major in ∼100 BM-MΦs after 4 h.

Extraction of proteins, preparation of cell lysates, and immunoblotting were performed as described earlier (9, 30). Signals were visualized by the Chemo Star Imager (Intas Science Imaging Instruments, Göttingen, Germany) or the ChemiDoc XRS+ System in combination with Image Lab Software (Bio-Rad, München, Germany). Images were processed with Adobe Photoshop CS6 software (Adobe Systems, San Jose, CA).

Total RNA was extracted from infected and stimulated cells and reverse transcribed as described earlier (9, 30). The following TaqMan assays were used for quantitative real-time PCR (Applied Biosystems, Darmstadt, Germany): Hprt1 (Mm00446968_m1), phosphoglycerate kinase 1 (Pgk1; Mm01225301_m1), Nos2 (Mm00440485_m1), Hif1a (Mm01283760_m1), and Tnfaip3 (Mm00437121_m1). Quantitative real-time PCR was carried out with an ABI Prism 7900 sequence detector (Applied Biosystems) and FastStart Universal Probe Master (Rox) (Roche Diagnostics, Mannheim, Germany). Data were analyzed using the ΔΔC_T method. The normalized ratio of target mRNA/control Hprt1 in unstimulated and/or uninfected cells was set to 1.

Nonsilencing small interfering RNA oligonucleotides (cat. no. 1027281) and small interfering RNA duplexes directed against Hif1α (cat. no. L-04065-01-0005) were purchased from QIAGEN (Hilden, Germany) and Thermo Fisher Scientific (Darmstadt, Germany), respectively. Small interfering RNA was transferred via electroporation, as described earlier (39, 40).

Nitrite accumulation in cell supernatants was determined by a Griess reaction, as described earlier (41, 42).

Results are expressed as mean + SEM and “n” represents the number of biological samples, analyzed murine tissue specimens, and/or quantified high power fields obtained from independent experiments. Statistical significance was calculated with Prism v6.0 software (GraphPad, La Jolla, CA). Normality distribution was tested with the Kolmogorov–Smirnov test. When comparing two groups, the unpaired two-tailed Student t test and Mann–Whitney U test were used for normally and nonnormally distributed data sets, respectively. When comparing more than two groups, one-way ANOVA, followed by the Tukey multiple-comparison test and the Kruskal–Wallis test, followed by the Dunn multiple-comparison test were used for normally and nonnormally distributed data sets, respectively. The p values < 0.05 were deemed statistically significant (unless indicated otherwise).

As the first step, lesions from patients with human cutaneous leishmaniasis were analyzed for the expression of HIF-1α protein. In accordance with earlier observations (43), HIF-1α expression was detectable to comparable extents in the epidermal layers of Leishmania-infected skin and healthy skin. In contrast, elevated HIF-1α protein levels were seen in the dermal infiltrates of Leishmania skin lesions (Fig. 1A), whereas the dermis of uninfected control specimens was HIF-1α negative (Fig. 1B). Next, we assessed whether HIF-1α also accumulates in a mouse model of cutaneous leishmaniasis. Infection of C57BL/6 mice with L. major results in the development of a transient cutaneous lesion (reviewed in Refs. 44–47) (Supplemental Fig. 1). We analyzed the lesional HIF-1α levels and macrophage infiltration when the lesion reached maximum size at 14 d postinfection (first time point). We detected robust lesional HIF-1α accumulation (Fig. 1C) in macrophage-rich dermal and s.c. infiltrates (Fig. 1D). Resolution of the disease (second time point) was paralleled by diminished levels of lesional HIF-1α (Fig. 1E) and reduced numbers of infiltrating macrophages (Fig. 1F). From these findings, we concluded that infection with L. major triggers lesional HIF-1α accumulation in vivo.

Given that several pathogens and pathogen-derived danger signals are able to induce HIF-1α accumulation in innate immune cells under conditions of ample O₂ (reviewed in Refs. 3–5), we analyzed whether infection of macrophages with L. major is sufficient to induce HIF-1α accumulation on its own. Infection of mouse macrophages with L. major was not able to induce Hif1α mRNA (Fig. 2A) or to mediate HIF-1α stabilization under normoxic conditions (Fig. 2B). Accordingly, infection of macrophages with L. major did not promote expression of the HIF-1α target gene Pgk1 (Fig. 2C). In contrast, stimulation of macrophages with LPS/IFN-γ resulted in robust induction of Hif1α mRNA and protein, as well as expression of a glycolytic HIF-1α target gene Pgk1 (Fig. 2A–C). Because inflammatory HIF-1α accumulation under normoxic conditions requires NF-κB activation (reviewed in Refs. 3–5), we tested expression of the NF-κB response gene Tnfaip3 (A20) upon infection of macrophages with L. major, which is known to impair NF-κB activation of macrophages (48–50). We observed that infection of macrophages with L. major did not induce the NF-κB response gene Tnfaip3, whereas stimulation with LPS/IFN-γ resulted in robust Tnfaip3 induction (Fig. 2D). We conclude that infection of macrophages with L. major alone is not sufficient to allow for HIF-1α accumulation, most likely as a result of insufficient NF-κB activation.

We hypothesized that signals from the tissue microenvironment, such as the O₂ levels found in L. major–infected tissue and/or inflammatory signals, might account for HIF-1α accumulation in macrophages in vivo. Our previous analyses revealed that O₂ levels in L. major–infected tissue corresponded to ∼2.8% O₂ when the L. major skin lesions reach maximum size (26). IFN-γ–dependent (25, 51), TNF-dependent (52), and TLR-dependent (53–55) signals play a key role in macrophage activation and the antileishmanial response. We tested whether these factors allow for HIF-1α accumulation in L. major–infected macrophages. Stimulation of L. major–infected macrophages with TNF/IFN-γ or LPS/IFN-γ under conditions of ample O₂ promoted robust HIF-1α accumulation.

Incubation of L. major–infected macrophages under the reduced O₂ tensions found in infected tissues (2.8% O₂) resulted in HIF-1α accumulation but did not augment the induction of the NF-κB response gene Tnfaip3 (Fig. 2E, Supplemental Fig. 2). In the presence of exogenous inflammatory signals, low O₂ environments (2.8% O₂) enhanced HIF-1α accumulation (Fig. 2E). Moreover, in the context of LPS/IFN-γ costimulation, low O₂ levels (2.8% O₂) significantly boosted the expression of the NF-κB-response gene Tnfaip3 (Supplemental Fig. 2). From these results, we conclude that inflammatory signals from the tissue microenvironment, rather than L. major itself, drive HIF-1α accumulation in macrophages upon L. major infection in vivo. Moreover, our data demonstrate that reduced atmospheric O₂ levels (2.8% O₂) cannot induce significant proinflammatory responses in macrophages on their own but require exogenous inflammatory signals.

Given that the antileishmanial activity of cytokine-activated macrophages depends on NOS2 (24, 25) and that NOS2 expression can be regulated by HIF-1α in myeloid cells (12, 28–30), we tested whether inflammatory HIF-1α contributes to the antileishmanial defense of macrophages.

First, we analyzed whether pharmacological stabilization of HIF-1α under normoxic conditions and without exogenous macrophage activation by LPS/IFN-γ was sufficient to promote clearance of intracellular L. major. To this end, we treated L. major–infected macrophages with the lipophilic α-ketoglutarate analog PHD inhibitor ICA (31). ICA stabilized HIF-1α (Fig. 3A) and promoted upregulation of the metabolic HIF-1α–dependent target gene Pgk1 in comparison with vehicle treatment (Fig. 3B). However, ICA treatment on its own did not upregulate expression of the NF-κB response gene Tnfaip3 (Fig. 3C) or induce the production of NO (Fig. 3D). Accordingly, ICA treatment alone did not enhance the antileishmanial activity of macrophages, whereas LPS/IFN-γ was able to do so (Fig. 3E). Interestingly, despite remarkable HIF-1α accumulation due to ICA alone, ICA per se did not amplify Tnfaip3 induction or NO production; subsequently, it did not reduce L. major load in macrophages (Fig. 3). From these data, we conclude that HIF-1α accumulation in the absence of exogenous inflammatory macrophage-activating signals does not promote leishmanicidal activity of macrophages.

Next, we examined whether LPS/IFN-γ–induced inflammatory HIF-1α accumulation was required for the killing of intracellular Leishmania in macrophages. For that purpose, we silenced HIF-1α in macrophages by using RNA interference, as described earlier (39, 40). Hif1α-specific small interfering RNA effectively abolished HIF-1α protein accumulation in LPS/INF-γ–treated macrophages (Supplemental Fig. 3A). Blunting of HIF-1α impaired LPS/INF-γ–induced NO release and abolished the leishmanicidal activity of macrophages (Supplemental Fig. 3B).

As an alternative approach to address the importance of inflammatory HIF-1α in the antileishmanial response, we used macrophages from Hif1α^Δmyel mice. In line with earlier observations (56), immunoblotting demonstrated a truncated HIF-1α protein in macrophages generated from Hif1α^Δmyel mice (Fig. 4A). This most likely reflects deletion of the loxP-flanked exons (Fig. 4A), which encode for the dimerization and transactivation domain of HIF-1α (34). In macrophages derived from Hif1α-deficient mice (Hif1α^Δmyel), LPS/INF-γ–induced upregulation of the HIF-1α target gene Pgk1 (Fig. 4B) was abolished. In line with earlier observations (30), inflammatory HIF-1α signaling in macrophages did not affect the induction of the NF-κB response gene Tnfaip3 (Fig. 4C). However, LPS/INF-γ–induced NO release was diminished (Fig. 4D), which was paralleled by significant impairment of antileishmanial macrophage activity (Fig. 4E).

From these data, we conclude that HIF-1α stabilization in the absence of a proinflammatory macrophage-activating signal is insufficient to allow for production of leishmanicidal NO. However, in the context of classical macrophage activation, inflammatory HIF-1α is required for high-level NO production and for induction of a robust leishmanicidal activity.

Based on the findings described above, we hypothesized that HIF-1α in myeloid cells might also contribute to the control of L. major infection in vivo. Therefore, we infected Hif1α^Δmyel mice and littermate controls with L. major. They showed comparable development of cutaneous lesions during the first 2 wk postinfection; however, the lesion sizes in the control animals decreased steadily thereafter, whereas Hif1α^Δmyel mice showed a nonresolving course of infection (Fig. 5A). Restimulation of draining lymph node cells with soluble Leishmania Ag demonstrated that deficiency of HIF-1α in myeloid cells did not affect the Leishmania-specific IFN-γ T cell responses (Fig. 5B). In contrast, expression of NOS2 protein by CD11b⁺ myeloid cells from skin lesions of Hif1α^Δmyel mice was diminished compared with littermate controls (Fig. 5C). This was paralleled by an increase in the parasite load in the infected cutaneous lesion compared with littermate controls (Fig. 5D). From these data, we conclude that HIF-1α in myeloid cells plays an important role in controlling cutaneous L. major infection in vivo.

In this study, we demonstrated that HIF-1α is present in the macrophage-rich cutaneous Leishmania lesions of mice and humans. We further show that HIF-1α expression by myeloid cells contributes to the control of L. major parasites in vitro and in vivo. Our results extend earlier findings in mice infected with L. amazonensis (a causative agent of chronic nonhealing cutaneous leishmaniasis) or L. donovani (which causes visceral leishmaniasis), in which an accumulation of HIF-1α in the skin lesions (57) or parasitized spleens was noted (20).

Importantly, although the function of HIF-1α was not analyzed in L. amazonensis–infected mice (21, 22, 57), it appeared to be rather detrimental for the host immune response in L. donovani–infected mice (20). This contrasts with our observations obtained from self-healing cutaneous leishmaniasis of L. major–infected mice (this study). However, similar to our findings earlier reports established a critical role of myeloid cell–derived HIF1α in the defense against various bacterial pathogens including Streptococcus pyogenes (11, 12), Pseudomonas aeruginosa (13), Mycobacterium avium (14), and M. tuberculosis (58).

Infection of macrophages with L. major alone was not sufficient to promote HIF-1α accumulation in vitro; it required additional exogenous inflammatory signals, such as costimulation with LPS/IFN-γ or TNF/IFN-γ. In line with this, Hammami et al. (20) observed that in vivo HIF-1α accumulation in DCs was not directly elicited by L. donovani parasites but was secondary to IRF5-mediated signaling in CD11c⁺ cells. In other studies, infection of macrophages with L. donovani or L. amazonensis per se resulted in HIF-1α accumulation under normoxic conditions, and L. donovani or L. amazonensis was required for pathogen survival within the host cell (19, 59). Given that Leishmania species differ greatly in their metabolism (60, 61) and that HIF-1α plays an important role in reprogramming the metabolism of macrophages (reviewed in Refs. 62, 63), it is tempting to speculate that certain Leishmania species, such as L. donovani or L. amazonensis, require host cell HIF-1α to fuel their metabolic demands. In contrast, L. major’s metabolic requirements might even be met in the absence of HIF-1α signaling in macrophages. However, further studies are required to understand the divergent role of HIF-1α in macrophages infected with different Leishmania species.

Although L. major does not induce HIF-1α on its own, our data suggest that the context of HIF-1α induction is critical for its biological effect. Despite the fact that inflammatory HIF-1α accumulation promotes the production of NO and the leishmanicidal activity of macrophages, pharmacological HIF-1α stabilization with the PHD inhibitor ICA in the absence of inflammatory macrophage activation was not sufficient to induce NO production and did not exert antileishmanial activity in macrophages. In contrast to our findings, pharmacological HIF-1α stabilization helped to contain Streptococcus pyogenes–infected and M. tuberculosis–infected phagocytes (58, 64). However, unlike S. pyogenes and M. tuberculosis, L. major is a silent intruder that is able to shut down inflammatory responses, including NF-κB activation (48–50). Hence, in the absence of inflammatory signaling, canonical hypoxia-induced HIF-1α stabilization did not significantly enhance the expression of inflammatory genes, including Nos2 (30). Nevertheless, for instance, in the context of mycobacterial cord factor– or LPS-mediated activation of mononuclear phagocytes, HIF-1α promoted the expression of NOS2 and other inflammatory genes (30, 56). Consistent with this finding, our results demonstrate that LPS/IFN-γ costimulation required HIF-1α to establish high levels of NO production and leishmanicidal activity. This further supports the idea that the context of HIF-1α activation critically affects its function.

In a previous study, we demonstrated that, at the time point of maximal skin lesion sizes, O₂ tension in the infected cutaneous tissues dropped to levels that impaired O₂-dependent NO production and leishmanicidal activity of LPS/INF-γ–costimulated macrophages (26). Normalization of tissue O₂ levels was paralleled by cure of the disease (26). After 14 days postinfection, when the size of the leishmanial lesions reached their maximum size and skin O₂ levels reached their lowest point, tissue O₂ tension of the leishmanial lesions started to increase steadily (26). At day 29 postinfection, tissue O₂ levels allowed for substantial leishmanicidal activity of activated macrophages resulting in a reduced cutaneous parasite burden (26). In the current study, we analyzed the effect of HIF-1α in myeloid cells in vivo around day 29 postinfection. The cutaneous burden of L. major in control animals was of the same order of magnitude as the number of L. major parasites in the skin lesions obtained in the previous study (32). However, we noted increased parasite load at the site of infection and decreased NOS2 expression in lesional myeloid cells from Hif1α^Δmyel mice. Moreover, the antileishmanial activity of Hif1α-deficient macrophages under normoxic conditions was impaired in vitro and was associated with impaired NO production. In summary, our data suggest that, in cutaneous L. major infection, HIF-1α plays a key role in arming the macrophages with the antileishmanial effector enzyme NOS2, whereas a sufficient supply of molecular O₂ as a cosubstrate of NOS2 is required for the production of high levels of NO (65–67).

In addition to the indirect antimicrobial actions of HIF-1α that are exerted via its targets NOS2 [this study and (58)] and cathelicidin (12), HIF-1α expression in myeloid cells might have immunoregulatory functions in infectious diseases. For instance, in mouse models of Histoplasma capsulatum infection, HIF-1α suppressed the production of IL-10 and, thereby, helped to fight the infection (68). Moreover, infection of mice with Helicobacter pylori resulted in HIF-1α accumulation in myeloid cells, which was not required for H. pylori elimination by phagocytes but was required for termination of tissue inflammation. The mechanism by which myeloid cell–derived HIF-1α contributes to resolution in this model remains to be determined (69). Although our data suggest that HIF-1α deficiency in myeloid cells results in impaired leishmanicidal activity in vivo, largely as a result of impaired induction of Nos2 in lesional macrophages, further studies are needed to investigate whether other Hif1α-regulated targets, such as vascular endothelial growth factor A (70), contribute to the control of Leishmania parasites in the skin.

We thank Kirstin Castiglione, Vincent Klump, and Monika Nowottny for technical assistance.

The authors have no financial conflicts of interest.