Hypoxia Induces Macrophage tnfa Expression via Cyclooxygenase and Prostaglandin E2 in vivo

Abstract

Macrophage phenotypes are poorly characterized in disease systems in vivo. Appropriate macrophage activation requires complex coordination of local microenvironmental cues and cytokine signaling. If the molecular mechanisms underpinning macrophage activation were better understood, macrophages could be pharmacologically tuned during disease situations. Here, using zebrafish tnfa:GFP transgenic lines as in vivo readouts, we show that physiological hypoxia and stabilization of Hif-1α promotes macrophage tnfa expression. We demonstrate a new mechanism of Hif-1α-induced macrophage tnfa expression via a cyclooxygenase/prostaglandin E2 axis. These findings uncover a macrophage HIF/COX/TNF axis that links microenvironmental cues to macrophage phenotype, with important implications during inflammation, infection, and cancer, where hypoxia is a common microenvironmental feature and where cyclooxygenase and TNF are major mechanistic players.

Introduction

Innate immune activation during homeostasis and disease are tightly regulated, in part by the coordination of microenvironmental cues and cytokine signaling. Macrophages are important innate immune cells in disease and their activation status, commonly termed polarization, is especially important during the response to injury and infection. Mammalian macrophages have been classified into two broad polarization states: M1 (or classically activated) and M2 (alternatively activated) (1, 2). M1 macrophages are highly antimicrobial and can phagocytose and efficiently kill bacteria. M2 macrophages are central players in tissue healing and restoration of homeostasis post injury/infection (3, 4). M1/M2 polarization states have largely been determined in vitro by addition of exogenous factors, e.g., cytokines, to the growth media, however the complex regulation of the diverse macrophage phenotypes observed in vivo remains poorly understood. It is only in whole organism in vivo models, where tissues remain intact, that microenvironmental cues of macrophage activation can be fully investigated.

Leukocytes, including macrophages and neutrophils, are exquisitely sensitive to tissue hypoxia, a signature of many disease microenvironments (e.g., in inflammatory diseases, infections, and cancers) due to lack of local blood supply and a high turnover of local oxygen by pathogens (5, 6). Tissue hypoxia can polarize macrophages, via stabilization of the Hypoxia Inducible Factor (HIF)-1α transcription factor (7). HIF-1α is a master transcriptional regulator of the cellular response to hypoxia (8) and has been shown to have activating effects on macrophages in vitro, increasing their pro-inflammatory cytokine profile and bactericidal capabilities, yet the molecular mechanisms behind these observations are not well-understood in vivo (6).

Eicosanoids are lipid signaling molecules and important pro-inflammatory mediators during homeostasis and disease (9, 10). For example, tissue hypoxia increases cyclooxygenase (COX)/prostaglandin E2 (PGE2) production in epithelial cells in cancer models (11–13). Recent evidence indicates that hypoxia-induced tumor necrosis factor alpha (TNF) expression in osteoblasts (bone producing cells) is mediated through cyclooxygenase enzymes by an, as yet, unknown mechanism (14). This HIF/COX/TNF axis has not, to date, been observed in macrophages, and could be a potential novel pathway to macrophage activation in diseases where hypoxia is a microenvironmental factor.

Here, we investigated the regulation of expression of the important, primarily pro-inflammatory, cytokine tnfa by Hif-1α stabilization in vivo using optically transparent zebrafish larvae. We demonstrate that genetic, pharmacological, and hypoxia-mediated Hif-1α stabilization upregulated macrophage tnfa. We utilize well-characterized, non-invasive, zebrafish models of inflammation and infection to show that Hif-1α-mediated tnfa acts via an alternative, cyclooxygenase dependent mechanism, that differs from better understood DAMP/PAMP mediated pathways that are cyclooxygenase independent (15, 16). We show that the Hif-1α-mediated tnfa acts via active PGE2. Furthermore, this new macrophage HIF/COX/TNF axis is found in primary human macrophages. These findings have important implications in inflammation, infection, and cancer biology where macrophage phenotypes are influenced by microenvironmental hypoxia, and where HIF, COX, and TNF are major mechanistic players.

Materials and Methods

Zebrafish

Zebrafish were raised and maintained on a 14:10 h light/dark cycle at 28°C, according to standard protocols (17), in UK Home Office approved facilities at The Bateson Center aquaria at the University of Sheffield. Strains used were Nacre (wildtype), TgBAC(tnfa:GFP)pd1028 (Source: Bagnat lab, Duke University, USA), Tg(tnfa:eGFP-F)ump5Tg:Tg(mpeg1:mCherry-F)ump2Tg (Source: Lutfalla Lab, Montpellier University, France), Tg(mpeg1:mCherryCAAX)sh378 (Source: University of Sheffield), Tg(lyz:Ds-RED2)nz50 (Source: Hall Lab, University of Auckland, New Zealand), TgBAC(il-1β:eGFP)sh445 (Source: University of Sheffield), and Tg(phd3:EGFP)i144 (Source: University of Sheffield) (18–22).

Tailfin Injury

Inflammation was induced in zebrafish embryos by tail transection at 2 days post-fertilization (dpf). Larvae were anesthetized and the tailfin was transected using a sterile scalpel blade (23). Larvae were imaged by confocal microscopy at 16 h post-wounding (hpw) on a Leica TCS-SPE confocal on an inverted Leica DMi8 base and imaged using 20× or 40× objective lenses.

Mycobacterium marinum

Mycobacterium marinum (Mm) M (ATCC #BAA-535), containing a psMT3-mCherry or psMT3 mCrimson vector were used (24). Injection inoculum was prepared from an overnight liquid culture in the log-phase of growth resuspended in 2% polyvinylpyrrolidone40 (PVP40) solution (CalBiochem, CAS 9003-39-8) (25). One hundred or one hundred and fifty colony forming units (CFU) were injected into the caudal vein at 28–30 h post-fertilization (hpf) (26).

Confocal Microscopy of Transgenic Larvae

Larvae were mounted in 0.8–1% low melting point agarose (Sigma-Aldrich, A9414) and imaged on a Leica TCS-SPE confocal on an inverted Leica DMi8 base and imaged using 20× or 40× objective lenses.

For quantification purposes acquisition settings and area of imaging were kept the same across groups. Corrected total cell fluorescence was calculated for each cell using Image J by drawing around the cell and correcting for background levels of staining (27, 28).

RNA Injections

Embryos were injected with dominant hif-1αb (ZFIN: hif1ab) variant RNA at the one cell stage as previously described (23). hif-1α variants used were dominant active (DA) that is a constitutively stable Hif-1α and dominant negative (DN) hif-1α that cannot signal due to lack of transactivation domain. Phenol red (PR) (Sigma-Aldrich, P0290) was used as a vehicle control.

Bacterial Pixel Count

Mm mCherry infected zebrafish larvae were imaged at 4 days post-fertilization (dpi) on an inverted Leica DMi8 with a 2.5× objective lens. Brightfield and fluorescent images were captured using a Hammamatsu OrcaV4 camera. Bacterial burden was assessed using dedicated pixel counting software as previously described (29).

Pathway Inhibitors

Unless otherwise stated, embryos were treated from 4 h pre-Mm infection to 24 h post-infection (hpi) by addition to the embryo water in six well plates and dimethyl sulfoxide (DMSO, Sigma-Aldrich, D8418), was used as a negative solvent control. The pan hydroxylase inhibitor, DMOG (dimethyloxaloylglycine, Enzo Life Sciences, BML-EI347-0010), was used at a 100 μM concentration by incubation in E3 embryo media (23). The selective PHD inhibitors JNJ-402041935 (Cayman Chemicals, 14316) and FG-4592 (Selleckchem, FG-34595) were used at 100 and 5 μM, respectively (30, 31). The selective inhibitor of COX-1, SC-560 (Cayman Chemical, 70340), was used at 30 μM and the selective inhibitor of COX-2, NS-398 (Cayman Chemical, 70590), was used at 15 μM by incubation in E3 embryo media (32). The 15-LOX inhibitor PD146176 (Tocris, 2850) was microinjected (1 nl of 40 nM) at 1 h post-infection (hpi) (33). The leukotriene B4 receptor 1 (BLTR1) inhibitor, U-75302 (Abcam, ab141736), was used at 30 mM by incubation in E3 embryo media (33) together with the BLTR2 inhibitor, LY255283 (Abcam, ab144472), which was used at 1 mM (34). Exogenous prostaglandin E2 (Cayman Chemical, 14010) and 15-keto-prostaglandin E2 (Cayman Chemical, 14720) were added by incubation in E3 at a concentration of 1 μM (35). The EP4 antagonist AH23848 (Cayman Chemical, 19023) was used at 1 μM (36).

Hypoxia Hood

Embryos were incubated in 5% oxygen (with 5% carbon dioxide) in a hypoxia hood (Baker-Ruskinn Sci-tive UM-027) from 32 hpi for 6 h and were imaged at 2 dpf. Embryos from the same clutch kept in incubated normoxic room air were used as controls (37).

Human Cells

Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Paque PLUS (GE Healthcare, GE17-1440-03) density centrifugation of whole blood from healthy donors (National Research Ethics Service reference 07/Q2305/7). PBMC were seeded in RMPI 1640 media (Gibco, 11875-093) containing 2 mmol/L L-glutamine (Lonza, BE17-605F) and 10% newborn calf serum (Gibco, 26010074). Cells were cultured in 5% CO₂ at 37°C for 14 days in RPMI 1640 supplemented with 2 mmol/L L-glutamine and 10% heat-inactivated fetal bovine serum (FBS) (PAN Biotech, P30-3306) (38).

RPMI 1640 + 25 mM HEPES (Gibco, 11550496) containing 2 mmol/L L-glutamine and 10% heat-inactivated FBS and drugs were pre-equilibrated in normoxia or hypoxia (0.8% O₂, 5% CO₂, 70% humidity at 37°C, in a Baker-Ruskinn Sci-Tive UM-027) for 24 h prior to the experiment (39). Cells were treated in duplicate or triplicate wells with 100 μg/ml LPS (InvivoGen, tlrl-b5lps), DMSO (Sigma-Aldrich, D8418), or 15 mM COX-2 inhibitor (NS398) (Cayman Chemical, 70590) and incubated for 18 h. For normoxic controls, cells were treated identically within a class-2 tissue culture hood and transferred to a normoxic tissue-culture incubator.

Cell supernatants were collected following 18 h incubation and were assayed in triplicate using a human TNF ELISA MAX (Biolegend, 430204).

Statistical Analysis

All data were analyzed (Prism 7.0, GraphPad Software) using unpaired, two-tailed t-tests for comparisons between two groups and ANOVA (with Bonferonni post-test adjustment) for other data. P-values shown are: *P < 0.05, **P < 0.01, and ***P < 0.001.

Results

Hypoxia and Hif-1α Stabilization Upregulate tnfa:GFP Expression in vivo

Tumor necrosis factor (TNF) is a central regulator of macrophage pro-inflammatory phenotypes during homeostasis, inflammation, and infection. Here, we use two transgenic zebrafish lines that express GFP under the control of a tnfa promoter region; TgBAC(tnfa:GFP)pd1028 and Tg(tnfa:eGFP-F)ump5Tg (18, 20). The TgBAC(tnfa:GFP)pd1028 is a BAC (bacterial artificial chromosome) transgenic line that contains 50 kb of promoter region (18). The Tg(tnfa:eGFP-F)ump5Tg line has a smaller promoter region (3.8 kb) than the BAC line (20) and has previously been demonstrated to be upregulated in macrophages after tailfin injury in a zebrafish model (20). Here, we demonstrate that injury-induced upregulation of macrophage tnfa:GFP is also found in the BAC transgenic line at 16 h post-wound (hpw), shown in double transgenic TgBAC(tnfa:GFP)pd1028; Tg(mpeg1:mCherryCAAX)sh378 larvae (Figure 1A).

Figure 1

TNF has been implicated in many bacterial infections, however the cell types involved have been difficult to observe in vivo. To investigate this we used a well-characterized zebrafish infection model, the M. marinum (Mm) model of tuberculosis infection (40). We used the tnfa BAC promoter GFP line to establish the transcriptional regulation of tnfa during pre- and early- larval Mm granuloma formation stages. tnfa:GFP expression was induced in larvae following Mm infection before granuloma onset (at 1 day post-infection, dpi), and during granuloma formation (at 4 dpi) (Figure 1B). tnfa:GFP expression was predominantly found in macrophages, shown by co-localization of fluorescence with mpeg-1:mCherry expressing macrophages (Figure 1C). Macrophage expression was confirmed using the other tnfa promoter driven line, Tg(tnfa:eGFP-F)ump5Tg; Tg(mpeg1:mCherry-F)ump2Tg (Figure S1). Our data demonstrate that injury and Mm induced tnfa expression occurs in macrophages as part of an early response.

Hypoxia and Hif-1α stabilization are key features of diseased tissue microenvironments that can trigger macrophage pro-inflammatory responses. We first tested whether Hif-1α promotes tnfa in vivo, by stabilizing Hif-1α genetically in the tnfa:GFP BAC transgenic line (18, 20). Dominant active (DA) Hif-1α RNA (23) resulted in upregulation of macrophage tnfa:GFP expression (Figures 1D–F) indicating a shift of macrophage phenotype toward a pro-inflammatory response. Dominant negative (DN) Hif-1α caused no change in tnfa:GFP expression (Figures 1E,F).

To ensure that endogenous levels of Hif-1α could trigger a tnfa response, Hif-α was stabilized in embryos pharmacologically. Treatment with the well-described hypoxia mimetic DMOG (dimethyloxaloylglycine, a pan-hydroxylase inhibitor that stabilizes endogenous Hif-1α via inactivation of regulatory prolyl hydroxylase enzymes) alongside newer hydroxylase inhibitors with reportedly greater prolyl hydroxylase selectively than DMOG, JNJ-42041935, and FG4592 were tested in the Tg(phd3:GFP)i144 hypoxia reporter zebrafish. phd3 is a Hif-1α transcriptional target that is highly upregulated after hypoxia/Hif-1α stabilization (22). Treatment with all the hydroxylase inhibitors increased phd3:GFP levels compared to DMSO controls demonstrating stabilization of the Hif-α pathway (Figure S2) with FG4592 treatment leading to the highest phd3:GFP levels (23, 30, 31, 37). FG4592 (Figure 1G), DMOG (Figure S3A), and JNJ-42041935 (Figure S3B) treatment upregulated tnfa:GFP expression compared to a DMSO solvent control, phenocopying the DA Hif-1α response. tnfa:GFP induced by hydroxylase inhibition could be blocked using dominant negative (DN) Hif-1α, both in the case of FG4592 (Figures 1H,I) and DMOG (Figure S4), demonstrating that hydroxylase inhibitors are acting via Hif-1α to induce tnfa:GFP.

Finally, we subjected the tnfa:GFP line to physiological hypoxia (5% oxygen for 6 h at 32 hpf) and looked for GFP expression at 48 hpf. This level of hypoxia was sufficient to turn on phd3:GFP expression in Tg(phd3:GFP)i144 hypoxia reporter zebrafish (Figure S5) (37). Hypoxia induced tnfa:GFP expression, compared to normoxic controls at 48 hpf, to a similar level as that observed with genetic or pharmacological Hif-1α stabilization (Figure 1J).

Together, these data indicate that tnfa expression is induced in macrophages in response to hypoxia and Hif-1α, a response that is targetable by pharmacological agents and has the potential to manipulate macrophage responses during disease.

Hif-1α Dependent tnfa:GFP Transcription Requires Cyclooxygenase

Eicosanoids are lipid signaling molecules and can be produced by macrophages as part of a pro-inflammatory response (9, 10). We tested whether cyclooxygenase inhibition affected Hif-1α-mediated macrophage tnfa:GFP expression, using Cox-1 and Cox-2 inhibitors (SC560 and NS398, respectively) (32). First, we showed that neither SC560 nor NS398 caused increased Hif-α transcriptional activity, using the Tg(phd3:GFP)i144 hypoxia reporter zebrafish (Figure S6). In the absence of Hif-1α stabilization tnfa:GFP was not altered by SC560 or NS398 compared to DMSO treated negative controls (Figures 2A,B). Strikingly, the DA Hif-1α-induced tnfa:GFP expression was reduced to basal levels by both SC560 and NS398 (Figures 2A,B). These findings were replicated using the hydroxylase inhibitors FG4592 (Figures 2C,D) and DMOG with NS398 (Figure S7), (N.B. DMOG with SC560 caused toxicity and developmental delay therefore data is not shown for this group). The production of an unrelated Hif-1α target gene phd-3 by hydroxylase inhibitors, shown in the Tg(phd3:GFP)i144 transgenic line, was not altered by SC560 or NS398 treatment, demonstrating that Hif-1α transcriptional potential is not affected by cyclooxygenase inhibition (Figure S8).

Figure 2

We tested whether HIF-1α stabilization induced TNF in human monocyte derived macrophages (hMDMs). We found that human monocyte derived macrophages produced higher levels of TNF protein in hypoxia (0.8% oxygen) than those in normoxia, measured by an anti-human TNF ELISA (Figure 2E). Furthermore, treatment with the COX-2 inhibitor, NS398, reduced this hypoxia-induced TNF to equivalent levels found in normoxia (Figure 2E). This was replicated when HIF-1α was stabilized in hMDMs using the hypoxia mimetic FG4592 (Figure S9). These data indicate that a HIF/COX/TNF axis is found in zebrafish and human macrophages, and could be a druggable target in disease situations (Figure 2F).

We have previously shown that DA Hif-1α induces the expression of another important pro-inflammatory cytokine, il-1β, shown by an il-1β:GFP transgenic line (Figure S10A) (37). Cox-1 inhibition by SC560 did not abrogate Hif-1α-induced il-1β:GFP, while Cox-2 inhibition with NS398 caused a small decrease that did not reach basal levels (Figures S10B,C). These data indicate that Hif-1α-induced tnfa expression is caused by a product of the cyclooxygenase arm of the arachidonic acid pathway via a novel macrophage HIF/COX/TNF axis, while Hif-1α-induced il-1β does not fully act via this pathway suggestive of complex regulation of inflammatory cytokines by Hif-1α stabilization.

We next tested whether injury- and infection-induced tnfa are Hif-1α and cyclooxygenase dependent processes. Induction of tnfa:GFP expression after injury was not blocked by dominant negative Hif-1α (Figures 3A,B), and this was also the case for Mm infection-induced tnfa:GFP expression (Figures 3C,D). Similarly, neither genetic nor pharmacological Hif-1α stabilization were additive to the tnfa:GFP expression caused by either injury (Figures 3E,F) or Mm infection (Figures 3G,H; (Figures S11A,B). These data indicate that the early tnfa response of injury and Mm infection do not require active Hif-1α. Macrophage tnfa:GFP expression induced after injury was not abrogated by cyclooxygenase inhibition using NS398 (Figures 3E,F). Similarly, cyclooxygenase inhibition using either SC560 or NS398 did not alter the expression of Mm-induced tnfa:GFP (Figures 3G,H). SC560 and NS398 mediated inhibition of Hif-1α-induced tnfa expression did not diminish the host-protective effect of DA Hif-1α (Figures S12A–C) nor DMOG treated larvae (Figures S12D–F), suggesting that the presence of Mm induced tnfa is sufficient for protection. Our data indicate that Hif-1α-induced macrophage tnfa expression via cyclooxygenase differs from DAMP- (injury) or PAMP- (Mm) induced tnfa mechanisms.

Figure 3

Blocking Cyclooxygenase Independent Arachidonic Acid Pathways Do Not Abrogate Hif-1α Upregulation of tnfa:GFP

To investigate whether the effect of the cyclooxygenase inhibitors on tnfa was specific to the prostaglandin path of arachidonic acid signaling, we targeted the lipoxin and leukotriene producing arms using the 15-Lipoxygenase inhibitor PD146176 and leukotriene B4 receptor antagonists (Figure 4A) (33, 34). The 15-Lipoxygenase inhibitor PD146176 did not block the tnfa:GFP induced by DA Hif-1α (Figures 4B,C). Furthermore, PD147176 increased tnfa:GFP expression on its own in the absence of infection, although not to the same extent as DA Hif-α (Figures 4B,C). PD146176 also did not affect tnfa:GFP expression after Mm infection (Figures 4B,C), nor did it block the protective effect of DA Hif-1α. Treatment with PD146176 was sufficient to decrease Mm burden, but not to the same extent as DA Hif-1α (Figure 4D). Leukotriene B4 inhibition, using the BLTR1/2 antagonists U75302 and LY255283, did not increase tnfa:GFP levels and also did not block DA Hif-1α-induced tnfa:GFP (Figures 4E,F). These data indicate that blocking components of the lipoxygenase dependent arms of the arachidonic acid pathway does not block the Hif-1α effect on tnfa expression and do not replicate the effects observed by blocking the cyclooxygenase dependent arm.

Figure 4

Hif-1α-Induced tnfa:GFP Requires Active Prostaglandin E2

A key family of immune regulators downstream of arachidonic acid and cyclooxygenases are prostaglandins. The best characterized of these as a regulator of macrophage function is prostaglandin E2 (PGE2) (41). We tested whether PGE2 was a mediator in the HIF/COX/TNF pathway by addition of exogenous PGE2 to DA Hif-1α larvae (35). Exogenous PGE2 had no effect on tnfa:GFP expression in the absence of infection (Figures 5A,B). However, PGE2 was able to rescue the decrease in tnfa:GFP expression after inhibition of either Cox-1 (Figure S13) or Cox-2 (Figures 5A,B) in DA Hif-1α larvae. Furthermore, addition of PGE2 alone was sufficient to increase the DA Hif-1α-induced tnfa:GFP expression (Figures 4A,B). These tnfa:GFP activating effects were not observed by addition of exogenous 15-keto prostaglandin E2, an immunologically inactive degradation product of PGE2 (Figures 5C,D) (35, 42, 43). These data indicate that Hif-1α-induced tnfa:GFP requires active prostaglandin E2.

Figure 5

PGE2 signals via prostanoid receptors. In human and zebrafish there are four different receptors, EP1-4 (44). Of these, EP4 is the best characterized in immune cells and is expressed on human macrophages. When EP4 is blocked, macrophage pro-inflammatory cytokine release is decreased (45). We therefore tested a previously published EP4 antagonist AH23848 (36). We first demonstrated that AH23848 did not activate Hif-1α transcription using the Tg(phd3:GFP) hypoxia reporter line (Figure S6). The increased level of phd3:GFP induced by hydroxylase inhibitors was also not altered by AH23484 treatment, demonstrating that Hif-1α transcriptional potential is not affected by EP4 inhibition (Figure S8). Addition of AH23848 abrogated dominant active Hif-1α-induced tnfa:GFP, indicating that Hif-1α stabilization acts through PGE2 signaling via EP4 to upregulate Tnfa expression (Figures 5E,F).

Discussion

Control of macrophage function during homeostasis and disease are critical for maintaining healthy tissues and must integrate changes in the local microenvironment with cytokine signaling. Understanding of the signaling pathways that link microenvironment with macrophage phenotypic outcomes may identify novel therapeutic avenues for control of macrophages during disease.

Here, we show that a disease relevant microenvironmental cue, hypoxia signaling via Hif-1α, upregulates macrophage tnfa expression in a cyclooxygenase dependent. TNF upregulation by hypoxia has been previously demonstrated in a range of mammalian cells, both directly, through HIF responsive elements (HREs) in its promotor region, and indirectly (46–49). Here, we addressed hypoxia-induced tnfa levels using promoter driven GFP expression in two zebrafish transgenic lines as a proxy for endogenous tnfa expression (18, 20). Changes in endogenous tnfa expression on a wholebody-level after Hif-1α stabilization were not observable by qPCR (data not shown), presumably because the changes in transcription at the macrophage level are too small to be detected using these methods due to macrophages representing a small proportion of the zebrafish larvae, a challenge that was previously reported in wholebody detection of another pro-inflammatory cytokine, il-1β (37). These promoter driven tnfa transgenic lines have therefore opened up opportunities to study tnfa induction in situ in the in vivo zebrafish. We observed increased tnfa expression with stabilized Hif-1α, demonstrating that hypoxia signaling alone can lead to a change of macrophage phenotype, consistent with our previous observations with il-1β expression, suggestive of a pro-inflammatory response (37). Our findings are also in line with HIF-1α stabilization being important for the production of TNF and other pro-inflammatory molecules, such as nitric oxide (NO), leading to host-protective effects against group A Streptococcus in in vivo murine models (50). Not only could tnfa activation be achieved by genetic stabilization of Hif-1α, but also using hypoxia mimetics (hydroxylase inhibitors) and physiological hypoxia, that stabilize endogenous Hif-1α. These findings indicate that the Hif-1α-Tnfa axis is targetable by pharmaceuticals and could be druggable during disease.

A similar tnfa induction was induced by infection and inflammation in our in vivo zebrafish models. It has been widely demonstrated that Tnfa is required for control of early Mm infection, with perturbation of Tnfa signaling leading to high infectious burdens (51). While Hif-1α-induced tnfa was downregulated by cyclooxygenase inhibition, it was notable that the high level of tnfa expression after inflammation or infection was not affected by cyclooxygenase inhibition. These findings are consistent with the novel HIF/COX/TNF macrophage axis being independent of infection/inflammation-induced macrophage Tnfa. Mm-induced Tnfa is widely reported as being toll-like receptor (TLR) mediated, and our data indicate that Mm-induced levels of Tnfa are sufficient for control of bacteria in the presence of stabilized Hif-1α (28, 52). In the absence of the TLR adapter protein MyD88, bacterial induction of Tnfa is lost, demonstrating the central role of TLR pathways in post-infection Tnfa induction (28, 52–54). The best characterized exogenous TLR signal, LPS via TLR4, can induce Tnfa production via activation of the central pro-inflammatory regulator NFκB (54, 55). As cyclooxygenase inhibition did not alter either inflammation or infection driven tnfa it suggests that these complex damage/infection driven mechanisms of Tnfa induction are not dependent on the HIF/COX/TNF pathway. Our data demonstrate that Hif-1α stabilization can stimulate a macrophage tnfa producing response in the absence of a TLR stimulus, which may have important implications in sterile diseased tissue and infers an alternative macrophage tuning mechanism.

Hypoxic regulation of TNF via COX has previously been demonstrated in mammalian osteoblasts, however, little is known about this interaction (14). Here, we demonstrate that Hif-1α upregulation of macrophage tnfa is likely to be via the production of PGE2 signaling via the EP4 receptor. PGE2 is an inflammatory mediator that is known to affect macrophage polarization states during infection, and synergises with cytokines to amplify pro-inflammatory responses (41, 56). The degradation metabolite 15-keto-PGE2 did not rescue the loss of tnfa expression after cyclooxygenase inhibition, consistent with previous reports that 15-keto-PGE2 is unable to bind the prostaglandin E2 EP receptors, thus demonstrating a requirement for active PGE2 (57).

Here, we demonstrate that a macrophage HIF/COX/TNF pathway is found in human monocyte derived macrophages, indicating that this pathway may have conserved roles in macrophage homeostasis and disease, and opening up possibilities for pharmaceutical intervention to promote proinflammatory macrophage phenotypes in human disease. Our findings in human MDMs indicate that the HIF/COX/TNF pathway may, at least in part, be cell-autonomous to macrophages, however, this was not tested in vivo where it is likely that contributing signals from other celltypes, for example endothelial cells, may play important roles in macrophage phenotype outcomes (58).

Although the effect of the HIF/COX/TNF axis in early Mm infection was negligible, this pathway is likely to have important roles in other disease situations. Mycobacterial infection, along with many other bacterial pathogens, stimulates early TNF production, however some pathogens remain undetected by macrophages during early disease. An example of this is in Lyme disease, caused by Lyme borreliosis, where early low TNF production is associated with worse disease outcomes (59). Multiple infectious/inflammatory disease pathologies have hypoxia as a feature of the tissue microenvironment, concurrent with the presence of macrophages. For example, Hif-1α mediated TNF activation may be pertinent in later TB infection situations where granulomas have stabilized Hif-1α due to necrotic and hypoxic centers (60, 61). Hypoxia is a key hallmark of cancers with high levels of HIF-1α widely found in those that produce large tumors where the center is hypoxic (62). These tumors can also produce cyclooxygenase/PGE2, where these important immunomodulatory signals are required for cellular adaption to the tumor microenvironment (63). Macrophages play central roles in cancer-inflammation that could be exploited utilizing pharmaceutical control of this novel HIF/COX/TNF mechanism. Further investigation of the HIF/COX/TNF axis in models where hypoxia is a key hallmark of disease pathology is required to uncover the full therapeutic relevance of this potentially important novel macrophage pathway.

In conclusion, we have identified a novel mechanism for macrophage tnfa production via Hif-1α and cyclooxygenase that is found in zebrafish and humans. Importantly, this axis links a commonly found microenvironmental cue, hypoxia, to a macrophage cytokine, TNF. We provide strong evidence to show that this response acts via the cyclooxygenase/PGE2 arm of the arachidonic pathway. Due to the central roles of these modulators in disease microenvironments we anticipate that this HIF/COX/TNF pathway may have important implications in conditions such as inflammation, infection and cancer.

References

• 1.

  BenoitMDesnuesBMegeJ-L. Infections macrophage polarization in bacterial macrophage polarization in bacterial infections. J Immunol Ref J Immunol. (2008) 181:3733–9. 10.4049/jimmunol.181.6.3733

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 2.

  MillsCDKincaidKAltJMHeilmanMJHillAM. M-1/M-2 macrophages and the Th1/Th2 paradigm. J Immunol. (2000) 164:6166–73. 10.4049/jimmunol.164.12.6166

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3.

  WiegertjesGFWentzelASSpainkHPElksPMFinkIR. Polarization of immune responses in fish: the ‘macrophages first' point of view. Mol Immunol. (2016) 69:146–56. 10.1016/j.molimm.2015.09.026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 4.

  MantovaniABiswasSKGaldieroMRSicaALocatiM. Macrophage plasticity and polarization in tissue repair and remodelling. J Pathol. (2013) 229:176–85. 10.1002/path.4133

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5.

  PeyssonnauxCCejudo-MartinPDoedensAZinkernagelASJohnsonRSNizetV. Cutting edge: essential role of hypoxia inducible factor-1alpha in development of lipopolysaccharide-induced sepsis. J Immunol. (2007) 178:7516–9. 10.4049/jimmunol.178.12.7516

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6.

  NizetVJohnsonRS. Interdependence of hypoxic and innate immune responses. Nat Rev Immunol. (2009) 9:609–17. 10.1038/nri2607

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 7.

  WangNLiangHZenK. Molecular mechanisms that influence the macrophage M1-M2 polarization balance. Front Immunol. (2014) 5:614. 10.3389/fimmu.2014.00614

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 8.

  CramerTYamanishiYClausenBEFörsterIPawlinskiRMackmanNet al. HIF-1α is essential for myeloid cell-mediated inflammation. Cell. (2003) 112:645–57. 10.1016/S0092-8674(03)00154-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9.

  NorrisPCReichartDDumlaoDSGlassCKDennisEA. Specificity of eicosanoid production depends on the TLR-4-stimulated macrophage phenotype. J Leukoc Biol. (2011) 90:563–74. 10.1189/jlb.0311153

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10.

  GuptaSMauryaMRStephensDLDennisEASubramaniamS. An integrated model of eicosanoid metabolism and signaling based on lipidomics flux analysis. Biophys J. (2009) 96:4542–51. 10.1016/j.bpj.2009.03.011

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 11.

  KaidiAQualtroughDWilliamsACParaskevaC. Direct transcriptional up-regulation of cyclooxygenase-2 by hypoxia-inducible factor (HIF)-1 promotes colorectal tumor cell survival and enhances HIF-1 transcriptional activity during hypoxia. Cancer Res. (2006) 66:6683–91. 10.1158/0008-5472.CAN-06-0425

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12.

  LeeJJNatsuizakaMOhashiSWongGSTakaokaMMichayliraCZet al. Hypoxia activates the cyclooxygenase-2-prostaglandin E synthase axis. Carcinogenesis. (2010) 31:427–34. 10.1093/carcin/bgp326

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13.

  ZhaoLWuYXuZWangHZhaoZLiYet al. Involvement of COX-2/PGE2signalling in hypoxia-induced angiogenic response in endothelial cells. J Cell Mol Med. (2012) 16:1840–55. 10.1111/j.1582-4934.2011.01479.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 14.

  XingYWangRChenDMaoJShiRWuZet al. COX2 is involved in hypoxia-induced TNF-α expression in osteoblast. Sci Rep. (2015) 5:10020. 10.1038/srep10020

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15.

  MeijerAH. Protection and pathology in TB: learning from the zebrafish model. Semin Immunopathol. (2016) 38:261–73. 10.1007/s00281-015-0522-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16.

  LoynesCAMartinJSRobertsonATrushellDMInghamPWWhyteMKet al. Pivotal advance: pharmacological manipulation of inflammation resolution during spontaneously resolving tissue neutrophilia in the zebrafish. J Leukoc Biol. (2010) 87:203–12. 10.1189/jlb.0409255

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17.

  Nusslein-VolhardC DR. Zebrafish: A Practical Approach. 1st ed. Oxford: Oxford University Press (2002).

  • Google Scholar

• 18.

  MarjoramLAlversADeerhakeMEBagwellJMankiewiczJCocchiaroJLet al. Epigenetic control of intestinal barrier function and inflammation in zebrafish. Proc Natl Acad Sci USA. (2015) 112:2770–5. 10.1073/pnas.1424089112

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 19.

  BojarczukAMillerKAHothamRLewisAOgryzkoNVKamuyangoAAet al. Cryptococcus neoformans intracellular proliferation and capsule size determines early macrophage control of infection. Sci Rep. (2016) 6:21489. 10.1038/srep21489

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20.

  Nguyen-ChiMLaplace-BuilheBTravnickovaJLuz-CrawfordPTejedorGPhanQTet al. Identification of polarized macrophage subsets in zebrafish. Elife. (2015) 4:e07288. 10.7554/eLife.07288

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21.

  HallCFlores MVStormTCrosierKCrosierP. The zebrafish lysozyme C promoter drives myeloid-specific expression in transgenic fish. BMC Dev Biol. (2007) 7:42. 10.1186/1471-213X-7-42

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22.

  SanthakumarKJudsonECElksPMMcKeeSElworthySvan RooijenEet al. A zebrafish model to study and therapeutically manipulate hypoxia signaling in tumorigenesis. Cancer Res. (2012) 72:4017–27. 10.1158/0008-5472.CAN-11-31480008-5472.CAN-11-3148

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23.

  ElksPMvan EedenFJDixonGWangXReyes-AldasoroCCInghamPWet al. Activation of hypoxia-inducible factor-1α (Hif-1 α) delays inflammation resolution by reducing neutrophil apoptosis and reverse migration in a zebrafish inflammation model. Blood. (2011) 118:712–22. 10.1182/blood-2010-12-324186

  • CrossRef
  • Google Scholar

• 24.

  van der SarAMSpainkHPZakrzewskaABitterWMeijerAH. Specificity of the zebrafish host transcriptome response to acute and chronic mycobacterial infection and the role of innate and adaptive immune components. Mol Immunol. (2009) 46:2317–32. 10.1016/j.molimm.2009.03.024

  • CrossRef
  • Google Scholar

• 25.

  CuiCBenardELKanwalZStockhammerOWvan der VaartMZakrzewskaAet al. Infectious disease modeling and innate immune function in zebrafish embryos. Methods Cell Biol. (2011) 105:273–308. 10.1016/B978-0-12-381320-6.00012-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26.

  BenardELvan der SarAMEllettFLieschkeGJSpainkHPMeijerAH. Infection of zebrafish embryos with intracellular bacterial pathogens. J Vis Exp. (2012) 15:3781. 10.3791/37813781

  • CrossRef
  • Google Scholar

• 27.

  ElksPMBrizeeSvan der VaartMWalmsleySRvan EedenFJRenshawSAet al. PLoS Pathogens: hypoxia inducible factor signaling modulates susceptibility to mycobacterial infection via a nitric oxide dependent mechanism. PLoS Pathog. (2013) 9:e1003789. 10.1371/journal.ppat.1003789

  • CrossRef
  • Google Scholar

• 28.

  ElksPMVan Der VaartMVan HensbergenVSchutzEReddMJMurayamaEet al. Mycobacteria counteract a TLR-mediated nitrosative defense mechanism in a zebrafish infection model. PLoS ONE. (2014) 9:e100928. 10.1371/journal.pone.0100928

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29.

  StoopEJSchipperTHuberSKNezhinskyAEVerbeekFJGurchaSSet al. Zebrafish embryo screen for mycobacterial genes involved in the initiation of granuloma formation reveals a newly identified ESX-1 component. Dis Model Mech. (2011) 4:526–36. 10.1242/dmm.006676

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 30.

  BarrettTDPalominoHLBrondstetterTIKanelakisKCWuXHaugPVet al. Pharmacological characterization of 1-(5-Chloro-6-(trifluoromethoxy)-1H-benzoimidazol-2-yl)-1H-pyrazole-4-carboxylic Acid (JNJ-42041935), a potent and selective hypoxia-inducible factor prolyl hydroxylase inhibitor. Mol Pharmacol. (2011) 79:910–20. 10.1124/mol.110.070508

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31.

  WuKZhouKWangYZhouYTianNWuYet al. Stabilization of HIF-1α by FG-4592 promotes functional recovery and neural protection in experimental spinal cord injury. Brain Res. (2016) 1632:19–26. 10.1016/j.brainres.2015.12.017

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32.

  FengYRenshawSMartinP. Live imaging of tumor initiation in zebrafish larvae reveals a trophic role for leukocyte-derived PGE2. Curr Biol. (2012) 22:1253–9. 10.1016/j.cub.2012.05.010

  • CrossRef
  • Google Scholar

• 33.

  TobinDMVaryJCJr.RayJPWalshGSDunstanSJBangNDet al. The lta4h locus modulates susceptibility to mycobacterial infection in zebrafish and humans. Cell. (2010) 140:717–30. 10.1016/j.cell.2010.02.013S0092-8674(10)00128-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 34.

  KojoKItoYEshimaKNishizawaNOhkuboHYokomizoTet al. BLT 1 signalling protects the liver against acetaminophen hepatotoxicity by preventing excessive accumulation of hepatic neutrophils. Sci Rep. (2016) 6:29650. 10.1038/srep29650

  • CrossRef
  • Google Scholar

• 35.

  EvansRJPlineKLoynesCANeedsSAldrovandiMTiefenbachJet al. 15-keto-prostaglandin E2 activates host peroxisome proliferator-activated receptor gamma (PPAR-γ) to promote Cryptococcus neoformans growth during infection. PLoS Pathog. (2019) 15:e1007597. 10.1371/journal.ppat.1007597

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36.

  LoynesCALeeJARobertsonALSteelMJGEllettFFengYet al. PGE2 production at sites of tissue injury promotes an anti-inflammatory neutrophil phenotype and determines the outcome of inflammation resolution in vivo. Sci Adv. (2018) 4:eaar8320. 10.1126/sciadv.aar8320

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37.

  OgryzkoNVLewisAWilsonHLMeijerAHRenshawSAElksPM. Hif-1α-induced expression of Il-1β protects against mycobacterial infection in zebrafish. J Immunol. (2019) 202:494–502. 10.4049/jimmunol.1801139

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38.

  DockrellDHLeeMLynchDHReadRC. Immune-mediated phagocytosis and killing of Streptococcus pneumoniae are associated with direct and bystander macrophage apoptosis. J Infect Dis. (2001) 184:713–22. 10.1086/323084

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39.

  GhezziPDinarelloCABianchiMRosandichMERepineJEWhiteCW. Hypoxia increases production of interleukin-1 and tumor necrosis factor by human mononuclear cells. Cytokine. (1991) 3:189–94. 10.1016/1043-4666(91)90015-6

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40.

  TobinDMRocaFJOhSFMcFarlandRVickeryTWRayJPet al. Host genotype-specific therapies can optimize the inflammatory response to mycobacterial infections. Cell. (2012) 148:434–46. 10.1016/j.cell.2011.12.023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41.

  HariziH. The immunobiology of prostanoid receptor signaling in connecting innate and adaptive immunity. Biomed Res Int. (2013) 2013:683405. 10.1155/2013/683405

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42.

  ChenIJHeeSWLiaoCHLinSYSuLShunCTet al. Targeting the 15-keto-PGE2-PTGR2 axis modulates systemic inflammation and survival in experimental sepsis. Free Radic Biol Med. (2018) 115:113–26. 10.1016/j.freeradbiomed.2017.11.016

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43.

  TaiHHEnsorCMTongMZhouHYanF. Prostaglandin catabolizing enzymes. Prostaglandins Other Lipid Mediat. (2002) 68–69:483–93. 10.1016/S0090-6980(02)00050-3

  • CrossRef
  • Google Scholar

• 44.

  KalinskiP. Regulation of immune responses by prostaglandin E 2. J Immunol. (2012) 188:21–8. 10.4049/jimmunol.1101029

  • CrossRef
  • Google Scholar

• 45.

  IkegamiRSugimotoYSegiEKatsuyamaMKarahashiHAmanoFet al. The expression of prostaglandin E receptors EP 2 and EP 4 and their different regulation by lipopolysaccharide in C3H/HeN peritoneal macrophages. J Immunol. (2001) 166:4689–96. 10.4049/jimmunol.166.7.4689

  • CrossRef
  • Google Scholar

• 46.

  GhoshSPaulASenE. Tumor necrosis factor alpha-induced hypoxia-inducible factor 1α-β-catenin axis regulates major histocompatibility complex class I gene activation through chromatin remodeling. Mol Cell Biol. (2013) 33:2718–31. 10.1128/MCB.01254-12

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47.

  CharbonneauMHarperKGrondinFPelmusMMcDonaldPPDuboisCM. Hypoxia-inducible factor mediates hypoxic and tumor necrosis factor alpha-induced increases in tumor necrosis factor-alpha converting enzyme/ADAM17 expression by synovial cells. J Biol Chem. (2007) 282:33714–24. 10.1074/jbc.M704041200

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48.

  ImtiyazHZSimonMC. Hypoxia-inducible factors as essential regulators of inflammation. Curr Top Microbiol Immunol. (2010) 345:105–20. 10.1007/82-2010-74

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49.

  SandauKBZhouJKietzmannTBrüneB. Regulation of the hypoxia-inducible factor 1α by the inflammatory mediators nitric oxide and tumor necrosis factor-α in contrast to desferroxamine and phenylarsine oxide. J Biol Chem. (2001) 276:39805–11. 10.1074/jbc.M107689200

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50.

  PeyssonnauxCDattaVCramerTDoedensATheodorakisEAGalloRLet al. HIF-1alpha expression regulates the bactericidal capacity of phagocytes. J Clin Invest. (2005) 115:1806–15. 10.1172/JCI23865

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51.

  ClayHVolkmanHERamakrishnanL. Tumor necrosis factor signaling mediates resistance to mycobacteria by inhibiting bacterial growth and macrophage death. Immunity. (2008) 29:283–94. 10.1016/j.immuni.2008.06.011S1074-7613(08)00327-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52.

  BasuJShinD-MJoE-K. Mycobacterial signaling through toll-like receptors. Front Cell Infect Microbiol. (2012) 2:145. 10.3389/fcimb.2012.00145

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53.

  van der VaartMvan SoestJJSpainkHPMeijerAH. Functional analysis of a zebrafish myd88 mutant identifies key transcriptional components of the innate immune system. Dis Model Mech. (2013) 6:841–54. 10.1242/dmm.010843

  • CrossRef
  • Google Scholar

• 54.

  KissnerTLCisneyEDUlrichRGFernandezSSaikhKU. Staphylococcal enterotoxin A induction of pro-inflammatory cytokines and lethality in mice is primarily dependent on MyD88. Immunology. (2010) 130:516–26. 10.1111/j.1365-2567.2010.03249.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55.

  ZuckermanSHEvansGFGuthrieL. Transcriptional and post-transcriptional mechanisms involved in the differential expression of LPS-induced IL-1 and TNF mRNA. Immunology. (1991) 73(4): 460–5.

  • Pubmed Abstract
  • Google Scholar

• 56.

  ManciniADDi BattistaJD. The cardinal role of the phospholipase A(2)/cyclooxygenase-2/prostaglandin E synthase/prostaglandin E(2) (PCPP) axis in inflammostasis. Inflamm Res. (2011) 60:1083–92. 10.1007/s00011-011-0385-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57.

  CogginsKGLatourANguyenMSAudolyLCoffmanTMKollerBH. Metabolism of PGE2by prostaglandin dehydrogenase is essential for remodeling the ductus arteriosus. Nat Med. (2002) 8:91–2. 10.1038/nm0202-91

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58.

  HeHXuJWarrenCMDuanDLiXWuLet al. Endothelial cells provide an instructive niche for the differentiation and functional polarization of M2-like macrophages. Blood. (2012) 120:3152–62. 10.1182/blood-2012-04-422758

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59.

  WidheMGrusellMEkerfeltCVrethemMForsbergPErnerudhJ. Cytokines in lyme borreliosis: lack of early tumour necrosis factor-α and transforming growth factor-β1 responses are associated with chronic neuroborreliosis. Immunology. (2002) 107:46–55. 10.1046/j.1365-2567.2002.01500.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60.

  Duque-CorreaMAKuhlAARodriguezPCZedlerUSchommer-LeitnerSRaoMet al. Macrophage arginase-1 controls bacterial growth and pathology in hypoxic tuberculosis granulomas. Proc Natl Acad Sci USA. (2014) 111:E4024–32. 10.1073/pnas.1408839111

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61.

  QuallsJEMurrayPJ. Immunometabolism within the tuberculosis granuloma: amino acids, hypoxia, and cellular respiration. Semin Immunopathol. (2016) 38:139–52. 10.1007/s00281-015-0534-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 62.

  AjdukovicJ. HIF-1–a big chapter in the cancer tale. Exp Oncol. (2016) 38:9–12. 10.31768/2312-8852.2016.38(1):9-12

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 63.

  GreenhoughASmarttHJMMooreAERobertsHRWilliamsACParaskevaCet al. The COX-2/PGE2 pathway: key roles in the hallmarks of cancer and adaptation to the tumour microenvironment. Carcinogenesis. (2009) 30:377–86. 10.1093/carcin/bgp014

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64.

  LewisAElksPM. Hif-1α stabilisation polarises macrophages via cyclooxygenase/prostaglandin E2 in vivo. bioRxiv [Preprint]. (2019). 10.1101/536862

  • CrossRef
  • Google Scholar