Hypoxia regulates the production and activity of glucose transporter-1 and indoleamine 2,3-dioxygenase in monocyte-derived endothelial-like cells: possible relevance to infantile haemangioma pathogenesis
Lowy Cancer Research Centre and Prince of Wales Clinical School, University of New South Wales, Sydney, NSW 2052, Australia
Background Infantile haemangioma (IH) may present as a precursor area of pallor prior to the initial proliferative phase, which implies that the early lesion may be hypoxic.
Objectives To examine the effect of hypoxia on the expression and activity of two key molecular markers of IH, glucose transporter-1 (GLUT1) and indoleamine 2,3-dioxygenase (IDO).
Methods IH endothelial cells express both haematopoietic and endothelial cell markers. CD14+ monocyte-derived endothelial-like cells have been employed in the study of IH and is the cell type used in this study.
Results GLUT1 transcript, protein and activity levels were strongly induced by hypoxia and remained elevated following 2 days of normoxic recovery. IDO transcript levels were not affected by hypoxia, although IDO protein level was reduced fivefold and IDO activity > 100-fold following 2 days of hypoxia. The protein and activity levels returned to normal following 2 days of normoxic recovery.
Conclusions The findings link the tissue hypoxia that precedes lesion development and the expression and/or activity of two key IH proteins. The early hypoxic insult may contribute to the elevated GLUT1 levels in IH lesions, while the very low IDO activity during the hypoxic phase may promote activation of immune cells in the lesion, which release cytokines that trigger IDO expression and activity and entry into the proliferative phase. Interestingly, IH lesion development shares some common features with ischaemia-reperfusion injury.
Infantile haemangioma (IH) is a common, benign tumour of infancy composed of endothelial cells which have lost their ability to organize in lumenized structures.1 IH is not usually present at birth but then proliferates rapidly for weeks to months, sometimes resulting in severe disfigurement with painful ulceration, and occasionally can induce life-threatening complications when vital structures are involved. The rapid growth phase is followed by a transitional ‘plateau’ phase and then a variable period of involution, over years. There is an increased incidence of IH in caucasians2 and in low birth weight, premature infants.3,4 Pre-eclampsia and placenta praevia are also associated with a statistically significant increase in the incidence of IH.4
While there are various hypotheses that attempt to account for all of the features of IH, the pathogenesis of the disease remains unknown. One study recently reported that expression of the proangiogenic growth factor receptor vascular endothelial growth factor (VEGF) receptor-1 is suppressed in IH tissue, and mutations in the genes encoding tumour endothelial marker-8 and VEGF receptor-2 were found in a subset of individuals with IH.5 In addition to these genetic factors, which may provide the basis for potential antiangiogenic treatment options, the origin of the endothelial cells that comprise the vascular lobules characteristic of IH is a subject of intense interest and studies have demonstrated the presence of endothelial progenitor cells (expressing surface CD133 or CD34) in IH,6 suggesting that they may contribute to the early vascular development in proliferative lesions. Consistent with this, an increase in circulating endothelial progenitor cells has been identified in patients with IH.7 Also, a recent study suggests that CD133+ cells derived from proliferative IH are particularly efficient in inducing IH-like lesions in nude mice.8 Further, differentiated CD31+ endothelial cells re-isolated from these lesions retained the capacity to generate blood vessels in secondary recipients. However, the proportion of endothelial progenitor cells in IH lesions is very small. It remains to be seen whether the vasculogenic capacity of endothelial progenitor cell populations can be reproduced by equivalent cells of primary origin that have not undergone ex vivo culture and expansion.
IH endothelial cells express both haematopoietic and endothelial cell markers. Early lesions consist predominantly of endothelial-like cells expressing both myeloid (CD83, CD32, CD14 and CD15) and endothelial [CD31, CD34, von Willebrand factor (VWF)] cell markers.9,10 So called ‘angiogenic monocytes’, which share both endothelial and monocytic features, are an important proangiogenic population that contributes to blood vessel growth and repair in both physiological and pathological settings.11–15 The presence of myeloid markers in addition to endothelial markers points to the involvement of a haematopoietic progenitor cell in IH. Notably, monocytes have been detected in proliferating IH tissue10 and peripheral blood CD14+ mononuclear cells can be induced to differentiate into endothelial-like cells when exposed to angiogenic growth factors.16,17 Monocytes isolated from peripheral blood and expanded with angiogenic growth factors developed an endothelial-like phenotype that expressed glucose transporter (GLUT)-1 while retaining the expression of FcγR11.18 These findings have led to the hypothesis that circulating monocytes may be the precursor cell for IH endothelial cells.19 For these reasons, we have employed CD14+ monocyte-derived endothelial-like cells in this study.
A ‘precursor’ area of pallor with telangiectases is often observed prior to the proliferative IH lesion.20 The initial description of a promontory mark of IH as ‘an anaemic naevus’ was suggested by several authors to reflect tissue ischaemia.21,22 Others have reported an association of IH with areas of ischaemic changes and infarction in the placenta.23 In much the same way, hypoxia is an initial stimulus for the placental tumour chorangioma (with a similar histology to IH).24 The blanching of the area subjects the associated tissue to a hypoxic insult, which may be the trigger for lesion development.20 Indeed, hypoxia appears to be an important factor in the recruitment of circulating progenitor cells to IH lesions.25,26 It has been suggested that some areas of the fetal dermis may not be adequately vascularized during the transition to postnatal life (which is accompanied by pronounced hypoxia), leading to recruitment of stem/progenitor cells.20
Cells adapt to low oxygen levels by changing their transcription and translation of certain genes.27,28 The transcriptional effects are mediated largely by a family of hypoxia inducible factors (HIFs), which are stabilized by hypoxia and activate genes by binding to hypoxia response elements in their promoters, introns and/or enhancers.28 Proliferating IH lesions contain higher HIF-1α protein levels than involuting lesions25 and the levels of VEGF, stromal-derived factor 1 and insulin-like growth factor-2, all downstream effectors of HIF-1α, are significantly increased in the proliferating lesion.20,25 HIF-1α dysregulation has also been implicated in the pathophysiology of other tumours.29
The plasma membrane glucose transporter, GLUT1, is a marker of IH.30 It is expressed in the proliferating phase and during involution. GLUT1 is not present in the vasculature of normal skin or solid tumours, although it is highly expressed in microvascular endothelia at sites of blood–tissue barriers, such as in the central nervous system and the placenta.31 The cytosolic haem enzyme that converts tryptophan to kynurenine, indoleamine 2,3-dioxygenase (IDO), is expressed in proliferating IH but not in the involuting lesion.32 IDO is highly expressed in the placenta and is thought to prevent rejection of the allogeneic fetus by catabolizing tryptophan, the least abundant essential amino acid that is crucial for the activation of T cells.33 It has been postulated that the immune suppressive action of IDO might ‘protect’ haemangioma cells from immune surveillance during the proliferating phase.34
In this study, we examine the association between hypoxia and expression and activity of GLUT1 and IDO in monocyte-derived endothelial-like cells.
Materials and methods
Isolation of CD14 monocytes
Peripheral blood mononuclear cells (PBMC) were isolated from white cell concentrates of healthy donors of the New South Wales Red Cross Blood Transfusion Service (Sydney, NSW, Australia) by density gradient centrifugation using Ficoll-Plaque-Plus (Amersham Pharmacia Biotech, Little Chalfont, U.K.). CD14+ monocytes were separated from PBMC using MACS purification (Miltenyi Biotec, Bergisch Gladbach, Germany). In short, cells were incubated with MACS anti-CD14 MicroBeads (Miltenyi Biotec) for 15 min at 4 °C (80 μL per 107 cells), washed with phosphate-buffered saline (PBS) pH 7·2 containing 0·5% bovine serum albumin (BSA) and 20 mmol L−1 ethylenediamine tetraacetic acid (EDTA), and separated using either the VarioMACS or autoMACS separation systems (Miltenyi Biotec).
Endothelial-like cells were generated from CD14+ human monocytes as previously described.11,16,17,35 Briefly, the CD14+ cells were resuspended in Endothelial Growth Medium 2 (Lonza Bioscience, Basel, Switzerland) with 20% (v/v) heat-inactivated human serum (HS) and plated on fibronectin-coated six-well Primaria plates (BD Biosciences, Franklin Lakes, NJ, U.S.A.). Human recombinant interferon (IFN)-γ (R&D Systems, Minneapolis, MN, U.S.A.) was added to induce the expression of IDO. The cultured cells were subjected to hypoxia by sealing in a chamber with two sachets of AnaeroGen (Oxoid, Cambridge, U.K.) to convert the O2 into CO2. An O2 concentration of < 1% was achieved.
Cells were harvested in 0·02% (w/v) EDTA in PBS, then blocked with PBS containing 5% (v/v) fetal calf serum (FCS), 0·5% (w/v) BSA, 2 mmol L−1 NaN3. Subsequently, cells were washed with PBS containing 1% FCS, 2 mmol L−1 NaN3, then incubated with antibodies either fluorescently conjugated (CD14, CD31, CD32, CD34; Invitrogen, San Diego, CA, U.S.A.) or nonconjugated [VWF (Dako, Glostrup, Denmark), GLUT1 (R&D Systems)] followed by compatible fluorescently labelled secondary antibodies (Invitrogen). Cells were fixed with 1% (w/v) paraformaldehyde. Flow cytometry was carried out on a FACSVantage SE Cell Sorter (BD Biosciences) and analysed using FCS Express (De Novo Software, Los Angeles, CA, U.S.A.).
Measurement of glucose transport activity
Glucose transport activity was determined by the rate of 2-deoxyglucose (2-DG) uptake. Cells were incubated for 5 min at 37 °C in PBS containing 0·5 mmol L−1 2-DG (Sigma, St Louis, MO, U.S.A.) and [3H]-2-DG (1 μCi mL−1, specific activity 370 GBq mmol−1; Perkin Elmer, Waltham, MA, U.S.A.). The cells were then washed repeatedly with cold PBS and solubilized in 0·2 mol L−1 NaOH for 1 h at room temperature. Uptake of [3H]-2-DG was determined by counting an aliquot of cell lysate using a Tricarb scintillation counter (Perkin Elmer). 2-DG uptake was expressed as nmol mg−1 of protein.
Real-time polymerase chain reaction
RNA was harvested using TriSure (Bioline, London, U.K.) according to the manufacturer’s instructions and quantified with a Nanodrop Spectrometer (Thermo Scientific, Wilmington, DE, U.S.A.). Reverse transcription was carried out using a Superscript III kit (Invitrogen) according to the manufacturer’s instructions. Real-time polymerase chain reaction was performed on a Rotor-Gene-3000 (Corbett Research, Sydney, NSW, Australia) using SyBrGreen (Bioline) and the following primers: GLUT1 forward, AAGGTGATCGAGGAGTTCTACA, GLUT1 reverse, ATGCCCCCAACAGAAAGATG; IDO forward, GATGTCCGTAAGGTCTTGCCA; IDO reverse, TGCAGTCTCCATCACGAAATG.
Determination of kynurenine and tryptophan
IDO activity was assessed by measuring the extent to which l-tryptophan was converted into l-kynurenine in the culture medium.36 For this, culture medium was collected and an aliquot was deproteinized with 20% trichloroacetic acid (w/v) and the sample centrifuged at 13 000 r.p.m. for 10 min at 4 °C. The resulting supernatants were subjected to high-performance liquid chromatography (HPLC) analysis [Agilent 1200 HPLC system equipped with a Hypersil 3 micron ODS C18 (1) column; Phenomenex, Torrance, CA, U.S.A.] eluted with 100 mmol L−1 chloroacetic acid/9% acetonitrile pH 2·4 at 0·6 mL min−1. Tryptophan and kynurenine were detected by ultraviolet absorbance at 280 and 364 nm, respectively. For studies examining IDO, media were routinely supplemented with 200 μmol L−1 tryptophan.
Indoleamine 2,3-dioxygenase protein expression
At the end of the experiments, cells were washed with PBS and lysed in 1× cell lysis buffer (Cell Signaling Technology, Danvers, MA, U.S.A.). Protein concentrations were determined by the bicinchoninic acid protein assay using BSA as standard (Sigma). The samples were diluted with 3× sodium dodecyl sulphate (SDS) sample buffer containing 0·5 mol L−1 Tris pH 6·8, 30% (v/v) glycerol, 6% SDS, 150 mmol L−1 dithiothreitol and 0·03% (w/v) of bromphenol blue. Equal amounts of cell protein were separated using 4–12% precast Bis–Tris gel (Invitrogen) and transferred to polyvinylidene difluoride membranes. Membranes were blocked with Tris-buffered saline containing 0·1% (v/v) Tween 20 (TBS-T) and 5% (w/v) defatted powder milk (blocking buffer) and probed with mouse monoclonal antihuman IDO antibody at 4 °C (1 : 10 000)36 and mouse anti-α-tubulin (1 : 1000; Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.). The blots were washed three times with TBS-T and incubated with goat antimouse monoclonal secondary antibody linked to horseradish peroxidase (1 : 2500). Membranes were washed three times with TBS-T and developed using ECL detection reagents (Amersham Pharmacia Biotech).
Data are presented as mean ± SEM. Unpaired Student’s t-tests were performed using GraphPad Prism (San Diego, CA, U.S.A.) to determine statistical significance. Statistical significance was defined as P < 0·05.
Derivation of endothelial-like cells from CD14+ monocytes
Endothelial-like cells were derived from CD14+ human monocytes using anti-CD14 antibody-coated microbeads (Fig. 1a).11,16,17,35 CD14+ cells were isolated from PBMC. The median CD14+ cell purity from 10 separate preparations was 99·4% (range 53·4–99·8%). When cultured on fibronectin-coated plates in a proprietary endothelial cell growth medium containing 20% HS, the cells developed an endothelial-cell like morphology. They changed from the rounded shape of monocytes to an elongated, spindle shape more typical of endothelial cells (Fig. 1b). The surface expression of the vascular endothelial cell markers VWF and CD31, as well as CD34 (which is also a haematopoietic progenitor cell antigen) increased during culture (Fig. 1c). These results are consistent with previous reports of monocyte-derived endothelial-like cells.11,16,17,35
The cells were cultured for 7 days, exposed to normoxia (20% O2) or hypoxia (< 1% O2) for 2 days, followed by normoxic recovery for a further 2 days. Cells were harvested at days 7, 9 and 11 and examined for expression and activity of GLUT1 and IDO (Fig. 1d). The hypoxia challenge did not change the pH of the culture medium, as reported by the indicator dye, and there was minimal cell death (< 10% of cells) during the course of the experiment.
Hypoxia triggers increase in glucose transporter-1 message and transport activity in monocyte-derived endothelial-like cells
GLUT1 surface expression was enhanced by hypoxia treatment (14-fold, P < 0·01) and expression remained elevated (sevenfold, P < 0·01) following recovery for 2 days under normoxic conditions (Fig. 2a). To test whether the hypoxia-mediated increase in GLUT1 protein was due to release of the protein from preformed stores or from new protein synthesis, GLUT1 mRNA levels were measured. The trend in GLUT1 message levels correlated with GLUT1 protein levels (Fig. 2b), indicating that new synthesis of the transporter was triggered by hypoxia. To determine if the new GLUT1 protein was functional, rate of uptake of [3H]-2-DG in the differentiated monocytes was measured. Glucose transport activity correlated with both GLUT1 message and protein levels (Fig. 2c). There was an eightfold (P < 0·01) increase in transport activity following hypoxia treatment and activity was still elevated after 2 days of recovery from hypoxia (1·5-fold, P < 0·01).
Hypoxia inhibits indoleamine 2,3-dioxygenase activity in monocyte-derived endothelial-like cells
IDO expression is usually very low under basal conditions, but strongly induced by immune cytokines such as IFN-γ.37 In accordance with this observation, the expression of IDO mRNA remained low during differentiation of the monocytes, although there was a trend of increased IDO transcript following the hypoxia challenge (Fig. 3a). IDO protein and activity were not detectable, however, indicating that the modest changes in IDO mRNA expression upon hypoxia are not likely to be functionally significant (data not shown).
Treatment with IFN-γ triggered an approximately 1000-fold increase in IDO message level irrespective of whether the cells were cultured under normoxic or hypoxic conditions (Fig. 3b). Despite similar IDO transcript levels, hypoxic treatment resulted in an approximately fivefold reduction in cellular IDO protein (Fig. 3c, d). Cellular IDO enzyme activity was measured as the extent of conversion of tryptophan into kynurenine. Hypoxia treatment reduced IDO activity > 100-fold compared with normoxic controls (Fig. 3e). This was expected considering that IDO protein levels were approximately 20% of normal and O2 is an essential cofactor for IDO-catalysed oxidative conversion of tryptophan to kynurenine. Hypoxic recovery of the cells for 2 days led to restoration of IDO protein and activity levels to 65–80% of that exhibited by normoxic controls (Fig. 3d, e). Similar trends were observed using either 10 or 100 U mL−1 of IFN-γ (data not shown).
The precursor IH lesion may present as a promontory localized area suggesting vasoconstriction. This observation implies that the early IH lesion is hypoxic, which may be the trigger for the proliferating phase. This has led us to examine the effect of hypoxia on the expression and/or activity of two key molecular markers of IH, GLUT1 and IDO, in monocyte-derived endothelial-like cells.
CD14+ monocytes isolated from blood developed an endothelial cell-like morphology and expressed characteristic endothelial proteins (VWF, CD31 and CD34) when cultured for 7 days with angiogenic growth factors. When these cells were subjected to hypoxia, GLUT1 transcript, protein and activity levels increased in the cells and remained elevated 2 days after normoxic recovery. In contrast, IDO transcript levels were not affected by hypoxia, although IDO protein level was reduced fivefold following 2 days of hypoxia and IDO activity > 100-fold. Both IDO protein and activity levels returned to normal following 2 days of normoxic recovery.
The GLUT1 gene is a HIF-1α target,28 so increased expression of this transporter was anticipated in the cells. Our finding that GLUT1 expression and activity remained elevated following 2 days of normoxic recovery suggests that the early hypoxic insult is a contributing factor to the elevated GLUT1 levels in IH lesions. The high GLUT1 level also has implications for the proliferative capacity of IH lesion cells. There is a parallel between GLUT1 expression in IH lesions and increased expression of this transporter in solid tumours. Tumours reorganize the metabolic steps used by normal tissues for glucose utilization and ATP production. Normal tissues rely heavily on oxidative phosphorylation for ATP, while tumour cells employ a route that includes a much greater dependency on glycolysis.38 Tumour cells enhance their capacity to scavenge glucose from the surroundings, which supports the glycolytic phenotype. This is achieved through increased expression of glucose transporters, in particular GLUT1, GLUT3 and GLUT4.39 Hypoxia-triggered expression of GLUT1 in IH lesion cells is predicted to provide a metabolic and, therefore, proliferative advantage to the lesion cells. It is noteworthy that myeloid cells have been shown to promote the growth of new vessels under low oxygen conditions.10
It will be of interest to analyse other metabolic features of IH lesion cells. Tumours, for instance, harness hexokinase (HK) II to entrap and channel glucose towards glycolysis.40 Normal tissue employs HKIV, which is an approximately 250-fold less efficient enzyme than HKII.41 HKII expression is silenced in normal tissue due to methylation of the promoter42 and triggered in tumour cells by hypoxia.43 Importantly, while HKIV is cytosolic, HKII is bound to mitochondria via interaction with the voltage-dependent anion channel.40 This affords HKII preferential access to mitochondrial ATP, rapidly and quantitatively to phosphorylate and trap the incoming glucose. The identity of the HK isoform that IH lesion cells employ will be informative.
The finding that IDO protein but not transcript levels were reduced by hypoxia implies an effect on either translation of the IDO message or degradation of IDO protein. In addition to the transcriptional effects mediated by HIF transcription factors, hypoxic stress is associated with global attenuation of protein synthesis, primarily via the modification of eukaryotic translation initiation factors.27 The drop in IDO protein levels is likely to be a consequence of blocked protein synthesis during the hypoxic challenge and normal or enhanced protein degradation. Further studies are required to understand this balance. Hypoxia is known to mediate the recruitment of immune cells to the tissue microenvironment.44,45 During the hypoxic phase of IH lesion development, invading immune cells may release cytokines that, upon reoxygenation, trigger IDO expression and activity. Interestingly, prolactin at concentrations equal to serum levels in late pregnancy increases the sensitivity of monocytes to IFN-γ-mediated IDO induction.46 Only minute amounts of IFN-γ may be required, therefore, effectively to induce the expression of IDO in IH lesions. IDO activity in IH lesions is thought to shield the cells from immune surveillance mechanisms, through depletion of tryptophan which is important for T-cell survival.32,47,48 The induction of IDO activity following reoxygenation could subsequently allow the IH lesion to avoid detection by the normal immune response during the proliferative phase.
Our observations link two important features of IH lesions: the tissue hypoxia that precedes lesion development and the expression and/or activity of two key IH proteins. It is likely that hypoxia regulates the expression and activity of other IH lesion factors. Interestingly, IH lesion development shares some common features with ischaemia-reperfusion injury. This is where restoration of normal circulation to tissue deprived of oxygen and nutrients results in inflammation and oxidative damage. The damage to the precursor IH lesion as a result of the hypoxia may be the precipitating event that leads to the abnormal angiogenic response, rather than restoration of normal tissue function. A potential limitation of this study is the cell type employed. Of particular interest would be similar studies performed in haemangioma-derived stem cells, as recent evidence has identified these as the likely cellular origin of IH.8
What’s already known about this topic?
• A ‘precursor’ area of pallor with telangiectases may be observed prior to the proliferative infantile haemangioma lesion.
• The blanching of the area seen in some lesions subjects the associated tissue to a hypoxic insult, which may be the trigger for lesion development.
• Glucose transporter-1 and indoleamine 2,3-dioxygenase are two key molecular markers of infantile haemangioma lesions.
What does this study add?
• The findings link the tissue hypoxia that precedes lesion development and the expression and/or activity of two key infantile haemangioma proteins.
• Infantile haemangioma lesion development shares some common features with ischaemia-reperfusion injury.
The authors acknowledge flow cytometrist Leonie Gaudry for technical assistance.