Modulation of Angiogenic and Inflammatory Response in = Glioblastoma by Hypoxia

Abstract

Glioblastoma are rapidly = proliferating=20 brain tumors in which hypoxia is readily recognizable, as = indicated by=20 focal or extensive necrosis and vascular proliferation, two = independent=20 diagnostic criteria for glioblastoma. Gene expression profiling of = glioblastoma revealed a gene expression signature associated with=20 hypoxia-regulated genes. The correlated gene set emerging from=20 unsupervised analysis comprised known hypoxia-inducible genes = involved in=20 angiogenesis and inflammation such as VEGF and = BIRC3,=20 respectively. The relationship between hypoxia-modulated = angiogenic genes=20 and inflammatory genes was associated with outcome in our cohort = of=20 glioblastoma patients treated within prospective clinical trials = of=20 combined chemoradiotherapy. The hypoxia regulation of several new = genes=20 comprised in this cluster including ZNF395, TNFAIP3, and = TREM1=20 was experimentally confirmed in glioma cell lines and primary = monocytes=20 exposed to hypoxia in vitro. Interestingly, the cluster = seems to=20 characterize differential response of tumor cells, stromal cells = and the=20 macrophage/microglia compartment to hypoxic conditions. Most genes = classically associated with the inflammatory compartment are part = of the=20 NF-kappaB signaling pathway including TNFAIP3 and = BIRC3=20 that have been shown to be involved in resistance to = chemotherapy.

Our results associate = hypoxia-driven=20 tumor response with inflammation in glioblastoma, hence = underlining the=20 importance of tumor-host interaction involving the inflammatory=20 compartment.

Introduction

Maintenance of oxygen = homeostasis is=20 critical for cell survival. Hypoxia is a common condition in = cancer tissue=20 due to rapid tumor growth, accompanied by inadequate angiogenesis = with=20 formation of structurally aberrant, leaky blood vessels with poor = blood=20 flow and formation of edema. In fact, such aberrant vascular = proliferation=20 characterized by glomeruloid and garland-like patterns are a = hallmark of=20 glioblastoma [1], the most malignant primary = brain tumor.=20 Cancer cells undergo adaptive changes and are selected for genetic = alterations that allow them to survive and proliferate in a = hypoxic=20 environment. Hypoxia-regulated genes, mediating adaptive = physiologic=20 changes, include genes regulating the glycolytic pathway and = blood-vessel=20 formation, and genes encoding chemotactic molecules such as CCL2, = IL8 and=20 VEGF [2]. In cancer, such changes are = associated with recruitment of macrophages along a = hypoxia-mediated=20 chemotactic gradient. Macrophages recruited to hypoxic sites exert = a=20 tumor-promoting effect through the expression of genes with = mitogenic,=20 angiogenic, and migration/invasion stimulating properties, such as = VEGF, EGF, or HGF [3], [4]. The relative contribution of=20 hypoxia-induced genes expressed by tumor cells or macrophages to = tumor=20 progression is unknown. Tumor hypoxia is associated with = aggressive tumor=20 behavior and worse outcome in many cancers and its role in driving = tumor=20 malignancy and resistance to therapy is receiving increased=20 attention.

The key mediator of the molecular = responses to=20 hypoxia is the hypoxia-inducible factor-1 (HIF-1), a heterodimeric = transcription factor consisting of an =CE=B1 and a =CE=B2 = subunits. In the presence=20 of oxygen the HIF-1=CE=B1 subunit is hydroxylated, and upon = ubiquitination is=20 targeted to the proteasome for degradation. Under hypoxic = conditions,=20 however, HIF-1=CE=B1 hydroxylation and degradation no longer = occur, since the=20 hydroxylation reaction requires oxygen. Stabilized, = non-hydroxylated=20 HIF-1=CE=B1 translocates to the nucleus and binds to the = hypoxia-response=20 element (HRE) thereby activating expression of numerous = hypoxia-responsive=20 genes [5]. Hypoxia-inducible pathways = regulate=20 several biological processes, including angiogenesis, cell = proliferation,=20 metabolism, survival and apoptosis, immortalization, and = migration.=20 Besides hypoxia, HIF-1 can also be activated by growth factor = receptors=20 and oncogenic signaling pathways.

Using gene expression profiling of = human gliomas,=20 we have previously identified angiogenesis/response to hypoxia as = one of=20 the most discriminating features between malignancy grades of = astrocytic=20 glioma [6]. Accordingly, a = hypoxia-induced gene=20 expression signature is a feature of glioblastoma expression = profiles [7], [8] which has been reported to = classify=20 gliomas into prognostic groups in some datasets [6], [9].

The aim of the present study = was to=20 investigate biological and potential clinical implications of a=20 hypoxia-related gene signature emerging from our glioblastoma gene = expression data set [7]. All patients were enrolled in = a=20 prospective clinical trial of combined chemoradiotherapy for newly = diagnosed glioblastoma [10].

Results

Hypoxia-inducible genes related = to=20 angiogenesis and inflammation

Unsupervised analysis of = our gene=20 expression data-set derived from 80 glioblastoma and 4 non-tumoral = brain=20 tissues identified stable gene clusters that were associated with = known=20 biological processes, including a cluster characterized by = hypoxia-induced=20 genes as shown in Figure=20 1

,=20 Table=20 1

.=20 The hypoxia cluster drew our attention, when we found that the = second=20 principle component (PC) of this cluster was strongly associated = with=20 survival (p 3D"

=3D

0.0015; HR, 1.73; 95% CI, 1.23 to 2.43; multivariate = analysis=20 including age, treatment, MGMT-methylation status [a = predictive=20 factor for benefit from temozolomide treatment [11]]), while the 1^st = PC, as well=20 as the mean expression of the cluster, had no prognostic value in = our=20 cohort of patients. Subsequent investigation of the loadings = (coefficients=20 of the genes) in the PCs of the hypoxia cluster revealed that the=20 2^nd PC was mainly defined by two groups of genes: the = first=20 characterized by high negative coefficients and comprising = IL8, TREM1,=20 SERPINE1, BIRC3, and TNFAIP3; the second = characterized by=20 high positive coefficients and consisting of genes such as = VEGF, ADM,=20 ZNF395, and KISS1R (Fig.=20 2A

).=20 The first group is enriched for inflammation-related genes, while = the=20 second group consists more of universal hypoxia-regulated genes, = including=20 pro-angiogenic factors, such as VEGF and ADM, in = accordance with the gene dendrogram of the hypoxia cluster (Fig.=20 1

,=20 Table=20 1

).=20 Next we tried to identify biological features, characterized by = gene=20 expression, that might be correlated with the 2^nd PC. = In our=20 data-set we identified a list of genes enriched in = inflammatory/innate=20 immune response genes (e.g. S100A8, S100A9, = ITGB2,=20 TLR1) that were anti-correlated with the 2^nd PC, = and a=20 list of positively correlated genes comprising signal = transduction-related=20 genes. Samples with lower 2^nd PC have higher expression = of=20 inflammatory genes and vice versa. In accordance, our previously = published=20 immune response-related cluster (G24) displayed a = similar=20 correlation (=E2=88=920.54, Pearson correlation of mean = expression) with the=20 2^nd PC. The top anti-correlated genes of the = 2^nd PC=20 (cut-off at =E2=88=920.5 correlation) that passed the initial = variation filter,=20 were almost exclusively clustered in G24 (33/46 = genes, Table S1).

Figure=20 1

The = hypoxia gene=20 = cluster.

Table=20 1 ¹Probesets of=20 the Hypoxia Gene=20 Cluster.

Figure=20 2

Components of the=20 hypoxia gene=20 cluster.

To investigate if the two directions = that account=20 for the largest variance in the hypoxia cluster were consistent in = different glioblastoma cohorts, we analyzed three external = datasets [8], [9], [12] including only samples labeled = as=20 glioblastoma (WHO grade IV) from initial surgery (no other prior = therapy).=20 We performed PC analysis using as descriptive variables 52 = probe-sets from=20 our hypoxia cluster that were common to all datasets (Table=20 1

).=20 The loading plots of all datasets displayed a similar influence of = the=20 genes to the first two PCs: the group of genes containing = IL8,=20 TREM1, SERPINE1, BIRC3, and TNFAIP3 were = separated from=20 the group containing VEGF, ADM, ZNF395, and = KISS1R in=20 the 2^nd PC (Fig.=20 2A

).=20 In particular, most coefficients of genes in the first group are = negative=20 and most coefficients of genes in the second group are positive in = all the=20 four datasets. Hence, there is a reproducible pattern in which a=20 consistent component of tumor variability is described by the = differential=20 expression of these two groups of genes.

We observed that the loading vectors = for the four=20 datasets were highly correlated, both for the 1^st and = the=20 2^nd PC, while there is a dramatic drop of the = correlations for=20 the 3^rd PC loadings. The pair-wise correlations between = the=20 2^nd PC loadings of the four datasets range from 0.76 to = 0.88=20 (Fig.=20 2B

).=20 To examine the association of the 2^nd PC with survival, = a=20 combined analysis of the four cohorts was performed using a Cox=20 proportional hazards model with stratification by study. This = model is in=20 agreement with a potential prognostic value of the 2^nd = PC=20 (n 3D"

=3D

242, p

=3D

0.010, HR, 1.09, 95% CI, 1.02 to 1.16) (Fig.=20 3

).=20 However, if we consider a Cox model combining just the three = external=20 studies, the 2^nd PC does not reach statistical = significance at=20 the conventional 5% significance level (HR 3D"

=3D

1.06 95% CI: 0.98, 1.14, p-value: 0.15). It is of note = that the=20 patients in these datasets are more heterogeneous since, unlike = our study,=20 they have not been collected prospectively and were not treated=20 uniformly.

Figure=20 3

Meta-analysis using=20 four gene expression data-sets of=20 = glioblastoma.

The inflammation-related gene set = (negative=20 coefficients in the linear combination defining the 2^nd = PC)=20 contains CXCL3, TNFAIP3 and BIRC3, three = NF-=CE=BAB=20 down-stream target genes [13], [14] known to be expressed by = macrophages [15], [16]. Other members of this = group=20 include ADFP and SLC2A3 (GLUT3). Both = genes,=20 together with IL8 and SERPINE1, have been = reported to be=20 highly expressed in macrophage foam cells [17], [18], [19], [20], in accordance with the = presence of=20 phagocytic cells in necrotic tissues. Thus, the re-grouping of = these genes=20 as observed in the 2^nd PC loading is consistent with = their=20 reported expression by macrophages.

In contrast, the genes in the second = group=20 (positive coefficients in the linear combination defining the=20 2^nd PC) may be expressed by any cell type and reflect = essential=20 adaptive responses to hypoxia. The genes in this group comprise = angiogenic=20 factors such as VEGF, ADM, PDK-1, encoding an = enzyme of=20 the glycolytic pathway active in hypoxic tumors [21], and survival factors, = such as=20 NTN-1 (Neuritin) [22], which is also expressed by=20 microendothelial cells in perinecrotic areas [23].

Taken together, the hypoxia = cluster=20 seems to capture the hypoxia-induced genes in the tumor as a = whole, while=20 the loading plot of the 2nd PC provides some information = reflecting a more=20 specific patho-physiological context associated with the presence = of=20 tumor-infiltrating monocytes/macrophages. These cells may be = attracted by=20 the tumors' hypoxic areas and necrosis, and may subsequently = contribute to=20 the inflammatory signature (enriched on the negative side of the = loading=20 plot) within the cluster of hypoxia-inducible genes. Thus the two = PCs of=20 this cluster seem to mirror differential response of glioblastoma = to=20 hypoxic conditions: a basic angiogenic response of the tumor cell=20 compartment on one hand, and a more specific inflammatory response = of=20 macrophages /microglia on the other.

Upregulation of = inflammation-related=20 genes in microglia/macrophages isolated from glioblastoma = tissue

To investigate = differential gene=20 expression in the distinct tumor compartments, we performed gene=20 expression profiling of paired samples of glioblastoma tissue and = the=20 respective glioma-infiltrating microglia/macrophage (GIM) cell = fraction.=20 The GIM cell fraction was isolated by a modified Percoll-gradient = that=20 minimizes artificial microglia/macrophage activation and enriches=20 CD11b+/CD11c+/CD45+ cells as previously described [24]. In accordance, gene = expression=20 profiles exhibited at least two-fold increased RNA expression of = the=20 microglia/macrophage markers CD11b, CD11c, and CD45 as compared to = the=20 level of expression in the tumor tissue (Table=20 2

).=20 Similarly, the GIM fraction also expressed at least four-fold = increased=20 RNA levels of Fc receptor genes II and III (CD32 and CD16) and = CD14. The=20 genes with negative coefficients in the 2^nd PC = (inflammation=20 related, hypoxia-regulated genes) exerted an increased expression = in the=20 GIMs as compared to the respective tumor tissue, suggesting = enrichment of=20 the cells expressing these genes in the GIM fraction. Of the genes = belonging to the second group (positive coefficients in the = 2^nd=20 PC), most displayed a lower level of expression in the GIMs as = compared to=20 the unsorted tumor bulk (Table=20 2

).=20 However, these gene expression levels are merely indicative of = expression=20 by microglia/macrophages compared to the tumor bulk, since the = isolation=20 procedure may have introduced some artifacts.

Table=20 2

Differential=20 Expression of Microglial Markers and Selected Genes from the = 2nd=20 PC.

Novel hypoxia-induced genes in = the=20 expression profiles of glioblastoma patients

More than two-thirds of the = genes in=20 this hypoxia cluster have previously been reported to be = hypoxia-induced=20 (31 of 45 genes) (references, Table S2), from which we deduced = that the=20 remaining genes that have never been reported as such might also = be=20 regulated by hypoxia. We chose to examine four genes for = hypoxia-induction=20 and cell-type specific expression experiments: triggering = receptors=20 expressed by myeloid cells-1 (TREM-1), tumor necrosis = factor=20 (TNF)-induced protein 3 (TNFAIP3), zinc-finger protein = 395=20 (ZNF395) and kisspeptin receptor (KISS1R).

TREM1 and TNFAIP3 = (also known=20 as A20) are both NF-=CE=BAB target genes [13], [25]. While TREM1 is = implicated in=20 innate immune response as an amplifier of pro-inflammatory = mediators such=20 as IL-8, MCP1 and TNF-=CE=B1 [26], TNFAIP3 is part = of the=20 negative feedback regulatory mechanism of NF-=CE=BAB and inhibitor = of apoptosis=20 [13]. Both genes are known to be = expressed=20 in macrophages.

ZNF395 is a transcription = factor that=20 binds to the promoter of the Huntington Disease (HD) gene = [27]. It's expression has been = associated=20 with worse prognosis in Ewing's sarcoma family of tumors (ESFT) [28].

KISS1R (also = GPR54), is a=20 G-coupled protein receptor for the KISS1 peptide, which has been = shown to=20 be an inhibitor of tumor metastasis across a range of cancers = (reviewed in=20 [29]. KISS1R is known as a = regulator of=20 endocrine function and involved in the = hypothalamic-pituitary-gonadal axis=20 of the reproductive system [30].

To investigate these genes' = hypoxic=20 responsiveness, freshly isolated monocytes and four glioblastoma=20 cell-lines, LN229, LN319, LNZ308 and U87, were cultured under = hypoxia=20 (1%O₂) and the expression of the genes was measured = using=20 real-time quantitative RT-PCR (qRT-PCR). Expression of = TREM1 was=20 increased in freshly isolated monocytes after 18 hours of hypoxia = as=20 compared to normoxic culture conditions (Fig.=20 4A

).=20 However, neither under normoxia, nor hypoxia TREM1 = expression was=20 detectable in the glioblastoma cell-lines (data not shown), thus=20 supporting the hypothesis that TREM1 is expressed by = inflammatory=20 cells. In contrast, ZNF395 and KISS1R expression = and=20 their hypoxia induction were observed in all four glioblastoma = cell-lines,=20 while TNFAIP3 was hypoxia-inducible in LN229 and LNZ308. = A=20 time-course revealed that all three genes were induced to a = significant=20 level after 8 hours of hypoxia. The levels of expression were = either=20 maintained or increased after 12 hours (Fig.=20 4B=E2=80=93D

).=20 In addition, all genes chosen for the hypoxia experiments, except=20 KISS1R were also hypoxia-inducible in monocytes. In = accordance,=20 assessment of the promoter sequences of TREM-1, TNFAIP3, = ZNF395=20 and KISS1R revealed presence of the consensus HRE A/(G)CGTG. In addition, the=20 TREM-1 gene promoter comprises a binding sites for the = AP-1=20 transcription factor (TGAGTC/G),=20 KISS1R' contains SP-1 binding sites (GGCGGG), while the = ZNF395 and=20 TNFAIP3 promoters contain both binding sites. AP-1 and = SP-1=20 transcription factors are known to be necessary to form a = functional=20 hypoxia transcriptional enhancer complex together with = co-activators p300=20 and CREB Binding Protein (CBP). The genes thus appear to be = potentially=20 HIF-1-regulated.

Figure=20 4

Hypoxia = induction=20 of TREM1, ZNF395 and=20 = KISS1R.

Expression patterns of = TREM1,=20 ZNF395 and KISS1R in glioblastoma

RNA expression = patterns for=20 TREM1 and ZNF395 were analyzed in situ = on=20 glioblastoma sections. TREM1 and ZNF395 = expression were=20 both mostly expressed in proximity of multilayered blood vessels = of=20 glioblastoma (Fig.=20 5

).=20 KISS1R immunoreactivity was detected in endothelial cells on blood = vessels, as described previously [31] and hypoxic areas adjacent to = necrosis,=20 particularly prominent in pseudopalisading cells lining necrosis. = For=20 comparison, expression of KDR/VEGFR-2, an angiogenic marker = expressed by=20 endothelial cells; CD11b, a monocyte/macrophage marker; and CD45, = a=20 pan-leukocyte marker were determined by immunohistochemistry on = sequential=20 frozen sections of a glioblastoma used for the in situ=20 hybridization (Fig=20 5

Figure=20 5

Expression patterns=20 of hypoxia inducible genes in=20 = glioblastoma.

Discussion

Investigating a = hypoxia-related gene=20 expression signature in glioblastoma we observed two classes of=20 hypoxia-induced genes. The first class, reflecting a common = response to=20 hypoxia, includes genes involved in regulating angiogenesis and = glycolytic=20 metabolism. The second group comprises genes more commonly = expressed in=20 the macrophage/microglia compartment, in accordance with the gene=20 expression profiles of paired samples of glioblastoma and the = respective=20 GIM fraction. This second class of genes might reflect the = presence of the=20 tumor infiltrating macrophages attracted to hypoxic sites in the = tumors. A=20 similar expression pattern was confirmed in independent, published = glioblastoma data-sets. Most interestingly, the 2^nd PC = of the=20 hypoxia cluster was associated with inferior outcome in our = dataset of=20 prospectively and homogenously treated glioblastoma patients. A = weak trend=20 in the same direction was observed when combining 3 independent = external=20 glioblastoma datasets. The weaker signal may be expected given the = fact of=20 various treatments, in particular for the chemotherapy component. = Our=20 patient cohort was treated homogenously adding an alkylating = agent,=20 temozolomide, concomitant and adjuvant to standard radiation, a = regimen=20 demonstrated to have improved efficiency.

The importance of the inflammatory = response or=20 the involvement of tumor-associated macrophages in tumor promotion = have=20 been widely described, but not in the perspective of = hypoxia-induced tumor=20 resistance. Although acute hypoxia can induce cell death, exposure = to=20 chronic or repeated hypoxia can initiate adaptive changes and = select for=20 genetic alterations in tumors that allow survival and = proliferation in a=20 hypoxic environment. Our analysis of expression profiles from four = independent glioblastoma datasets suggests that hypoxia may induce = TREM1, TNFAIP3 and BIRC3 in the inflammatory=20 compartment. BIRC3 and TNFAIP3 are known anti-apoptotic factors = involved=20 in the NF-=CE=BAB pathway. Interestingly, both have been = implicated in=20 conferring resistance to apoptosis induced by anticancer drugs. = BIRC3 is=20 highly expressed in cisplatin-resistant prostate cancer cell lines = [32], and in doxorubicine- and=20 busulfan-resistant leukaemia cell lines [33], [34] while TNFAIP3 is = overexpressed in=20 tamoxifen-resistant breast carcinoma cell lines [35] and part of a signature = for=20 resistance to alkylating agents in glioblastoma cells [36]. Hence, resistance to = hypoxia may also=20 trigger resistance to anticancer treatment.

Evidence from clinical and = experimental studies=20 have indicated that macrophages can promote tumor progression and=20 metastasis [37]. Macrophages are recruited = to the=20 tumor microenvironment by growth factors and chemokines, such as = CSF-1,=20 TGF-=CE=B2, and MCP-1 produced by the tumor cells themselves or = cells of the=20 microenvironment, similar to those expressed upon challenge by = pathogens=20 or wounding. Once recruited, tumor-educated macrophages take on=20 non-immunological functions, in particular the production of = factors that=20 promote progression of tumors to a more malignant state through = paracrine=20 cues, including angiogenic factors (VEGF, TNF-=CE=B1, ANG1), = proteases (MMPs),=20 growth factors (FGF, PDGF, TGF-=CE=B2), and motility factors (EGF, = HGF) [3]. Similar to amplifying the = immune=20 response, TREM1 may have a role in amplifying these processes in = tumors.=20 During innate immune response, engagement of TREM1 stimulates = phagocytes=20 to secrete pro-inflammatory cytokines and chemokines, such as = IL-8, MCP-1,=20 TNF-=CE=B1 and IL-1 [26]. Interestingly, some of = these=20 tumor-promoting factors are also expressed by glioblastoma cells=20 themselves [38], [39], [40]. TREM1 has been shown to be=20 upregulated only in infectious inflammatory responses, but not in=20 non-microbial inflammatory processes [41]. However, in tumors or more = likely in=20 tumor-associated macrophages according to our results, = TREM1 may=20 also be induced by hypoxia. High expression of TREM1 in = tumor-associated=20 macrophages was reported to correlate with cancer recurrence and = poor=20 survival in patients with non-small cell lung cancer [42]. TREM1 expression in = macrophages=20 is also regulated by NF-=CE=BAB at the transcriptional level [25] again emphasizing the = contribution of=20 NF-=CE=BAB pathway activation in bridging inflammation and tumor = promotion and=20 progression. It is known from a number of studies that hypoxia = activates=20 NF-=CE=BAB, and NF-=CE=BAB has been described to promote = malignancy in=20 inflammation-associated cancers by means of its cytoprotective and = anti-apoptotic properties [43]. In these cancers, the = activation=20 state of NF-=CE=BAB is controlled by pro-inflammatory mediators = produced by=20 neighboring inflammatory cells. In glioblastoma, NF-=CE=BAB could = thus be=20 activated in a similar way, in addition to constitutive NF-=CE=BAB = activity=20 reported from various glioma cell lines and primary cultures from = tumor=20 tissue [44]. Of interest is also = that=20 S100A8 and S100A9 transcripts were found = correlated with=20 the inflammatory component in the 2^nd PC. S100A8 and = S100A9 are=20 chemoattractant proteins that can promote the recruitment of=20 immune-suppressive immune cells into tumor niches [45], and their detection here = may=20 therefore be consistent with this notion. In accordance, = HIF-1=CE=B1-mediated=20 recruitment of bone marrow-derived myeloid cells modulating tumor=20 angiogenesis has been shown in mouse models of glioblastoma [4].

Little is known on the potential role = of the=20 other two hypoxia inducible genes, ZNF395 and = KISS1R, in=20 cancer. Several studies have described a prognostic value of=20 KISS1 and KISS1R expression in tumors across a = variety=20 of cancer types and a role in metastasis and trophoblast invasion = [46], both invasive processes. = Here we=20 showed that KISS1R is expressed in hypoxic areas and endothelial = cells of=20 tumor blood vessels, pointing to a putative role in angiogenesis. = In=20 accordance, recent evidence proposes a role of kisspeptins in the=20 cardiovascular system, acting as vasoconstrictor with discrete=20 localization of the KISS1R to artherosclerosis-prone vessels [31].

Our results have indicated that = hypoxia may=20 induce genes specific to the inflammatory tumor compartment (e.g.=20 macrophages). The relative increase of the hypoxia-inducible and=20 inflammatory-associated genes compared to angiogenesis-related = genes was=20 associated with better outcome in our data-set. This is in line = with a=20 better outcome associated with the inflammatory response gene = cluster=20 (G24) as published previously [7] that we show here to be = correlated with=20 the inflammatory component of the hypoxia cluster. This = observation may=20 seem paradoxical at first since tumor-infiltrating inflammatory = cells have=20 been associated with a more malignant phenotype. Likely it is the=20 preponderance of the type of inflammation present =E2=80=93 be it = either=20 pro-inflammatory consisting of effector anti-tumor responses or=20 anti-inflammatory such as with immune suppressive = microglia/macrophages or=20 T regulatory cells that either limit or contribute to tumor = malignancy.=20 Alternatively, the inflammatory component may be actually related = to its=20 potential to promote tumor vascularization [3], [4] and thereby may improve delivery = of the=20 alkylating agent during chemotherapy though increased perfusion. = We=20 previously reported that increased expression of a gene signature = for=20 blood vessel markers (G7) was also associated = with better=20 outcome in our patients. This hypothesis with potentially relevant = therapeutic implications needs to be validated prospectively in = patients=20 treated uniformly with the new standard of care.

Taken together, the = investigation of the=20 different components of our hypoxia cluster (loadings of the=20 2^nd PC) and their association with outcome suggest that = the=20 effect of hypoxia is more complex than through sole induction of=20 tumor-promoting growth factors or angiogenesis alone. The = hypoxia-mediated=20 tumor-host interaction, including the inflammatory compartment may = have a=20 profound influence on response to classic therapeutic agents as = suggested=20 by our data. Moreover, this hypoxia-modulated interaction may be = of=20 clinical relevance for response to anti-angiogenic = therapy-mediated=20 induction of hypoxia that has been shown recently to invoke = important=20 secondary effects triggering massive tumor invasion [47].

Materials and = Methods

Ethics Statement

Collection of samples = used in this=20 study was approved by the Institutional Review Board of the = Faculty of=20 Biology and Medicine of the University of Lausanne.

Tumor Samples, Gene Expression = Profiles,=20 and Patient Characteristics

The microarray data is = deposited=20 in the Gene Expression Omnibus (GEO) database at http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0lMPT2k1nwX9UHDYZwfFSMUCeiDY54ooiyRRcOQ4n&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&http://www.ncbi.nlm.nih.gov/geo/=20 (accession-number GSE7696) and described in accordance = with MIAME=20 guidelines. In brief, gene expression profiles were established = from 80=20 frozen glioblastoma (grade IV astrocytoma), comprising 70 tumors = from=20 initial surgery, and 10 samples resected at recurrence, and from = four=20 non-neoplastic brain tissue samples. All glioblastoma patients = were=20 treated within two prospective clinical trials that led to = establishing=20 combined chemoradiotherapy as the current and worldwide accepted = standard=20 of care [10]. All patients provided = written informed=20 consent for molecular studies of their tumor. Data from external = datasets=20 (Freije, Phillips [8], [9]) were downloaded from GEO, = while the=20 Sun data-set [12] was downloaded from the caArray = database,=20 publicly accessible at http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0Z5cE4hde0LZWRJV2PJ-tK1eHytbTbif0_zqVC5pR&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&https://caarraydb.nci.nih.gov/caarray/publicExperimentDeta= ilAction.do?expId=3D101589758985234=20 (Identifier : gov.nih.nci.ncicb.caarray: = Experiment:1015897589852334:1 ).=20 Gene expression data of paired samples of glioblastoma tissue and=20 respective glioma-infiltrating microglia/macrophage (GIM) cell = fraction=20 are deposited in GEO (accession-number GSE16119).

Data Analysis

All statistical = analyses were=20 carried out with R, a free software environment available at http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0zsFShv1SFtSa4tcrjGLqSAlAjTX0kY6xBjs6yAiB&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&http://www.r-project.org/=20 or Coupled Two Way Clustering (CTWC) [48], publicly available at the = CTWC-Server:=20 http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0PMoPJQ1RXMSkPycS9p77bnCIoofic4WZiKBvBkbY&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&http://ctwc.weizmann.ac.il.=20 The expression intensities for all probe-sets from Affymetrix = CEL-files=20 from our and external data sets [7], [8], [9], [12] were estimated using robust = multi-array=20 average (RMA) with probe-level quantile normalization followed by = median=20 polish summarization as implemented in the BioConductor open = source=20 software (available at http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0V_JuKQ3gLKWzKREKNQlWHr3B7IIr7qCX2dvv-Qdf&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&http://www.bioconductor.org/).=20 Cluster analysis using our data set was performed with CTWC using = the=20 expression matrix of 84 samples (80 glioblastoma, 4 non-tumoral = brain=20 samples) and 3,860 probe-sets passing a non-specific filter based = on=20 standard deviation (>0.75 sd) [7]. All results of CTWC analysis = can be=20 viewed at: http://www.pubmedcentral.nih.gov/redirect3.cgi?&&= ;auth=3D0Bi6DV7InxCXgwu3_d-9VqshOfA2QvBfaljVhYAx-&reftype=3Dextlink&a= mp;article-id=3D2694268&issue-id=3D178938&journal-id=3D440&FR= OM=3DArticle%7CBody&TO=3DExternal%7CLink%7CURI&rendering-type=3Dn= ormal&&http://bcf.isb-sib.ch/projects/cancer/glio/=20 . Probe-sets comprised in stable gene clusters emerging from CTWC = served=20 as input for supervised analyses. The genes belonging to the = hypoxia=20 cluster were fixed according to the stable hypoxia cluster = emerging from=20 CTWC using our dataset (cluster G84) [7] (Fig.=20 1

,=20 Table=20 1

).=20 We performed a principal component analysis on the matrix using = primary=20 glioblastoma as samples and gene expression values for probe-sets = in each=20 mutually exclusive stable cluster as variables. We used the scores = of the=20 first and second components to associate samples to survival data. = To=20 visualize the contribution of genes to the principal components in = Figure=20 2A

,=20 we plotted the loadings i.e. the coefficients of each gene in the = linear=20 combination that defines the principal component. The data was = scaled and=20 centered independently for each data-set. Univariate or = multivariate Cox=20 proportional-hazards models were computed using the mean, the = first, and=20 the second principal component (PC) of each mutually exclusive = stable=20 cluster. Cox proportional-hazards models were used to examine the=20 association of the hypoxia cluster with survival. For all four = data sets,=20 including our own, survival analysis and PC analysis was performed = using=20 data from patients at initial surgery (no previous = treatment).

Isolation of = Microglia/Macrophages from=20 Human Glioblastoma Tissue

GIM cell fraction was = isolated by=20 a modified Percoll-gradient as described previously [24]. Briefly, after resection, = freshly=20 isolated glioblastoma tissue was mechanically dissociated through = a=20 stainless steel sieve. The dissociated material was centrifuged, = washed,=20 and layered onto a Percoll gradient. After centrifugation, the = cell layer=20 was removed, washed, layered on top of a second gradient and = centrifuged.=20 Microglia/macrophages were collected from the interphase between = the=20 1.065-g/ml and 1.03-g/ml layers.

Glioblastoma Cell Lines and = Human=20 Monocytes

Human glioblastoma = cell lines were=20 grown in high glucose Dulbecco's Modified Eagle Medium (DMEM) = (Invitrogen)=20 supplemented with 100 units/ml penicillin, 100 =C2=B5g/ml = streptomycin and 5%=20 fetal calf serum (FCS, Bioconcept). Human peripheral blood = lymphocytes and=20 monocytes were purified from dextran-sedimented leukocytes. The = monocytic=20 and lymphocytic fractions were obtained using the Nycoprep = protocol. Cells=20 were maintained under normoxic condition in a humidified = incubator, 21%=20 O₂/5% CO₂ at 37=C2=B0C.

Culture under Hypoxic = Condition

Monocytes = (1=C3=9710⁶) were=20 plated in 3.5 cm tissue culture dishes, incubated under normoxic = or=20 hypoxic conditions for 18 hours. For hypoxic conditions, cells = were=20 incubated in an Oxoid Gas anaerobic system chamber (Unipath Ltd.,=20 Hampshire, UK) using 1% O₂, 94% N₂, 5%=20 CO₂. Glioblastoma cells lines were plated in 10 cm = tissue=20 culture dishes. When the cultures reached 80% confluence, fresh = culture=20 medium was added and cells were incubated under normoxic or = hypoxic=20 conditions as indicated.

RNA extraction & = qRT-PCR

Total RNA was = extracted using the=20 Qiagen RNeasy total RNA extraction kit (Qiagen) and reverse = transcribed=20 using random hexamers and Superscript RT II (Invitrogen). = Quantitative=20 real time PCR was performed in triplicates using the QuantiTect = SYBR Green=20 PCR-kit (Qiagen) on a LightCycler 2.0 Instrument (Roche Applied = Science).=20 Primer sequences were as follows: TREM-1 (5=E2=80=B2-GCAGATAATAAGGGACGGAGA-3=E2=80=B2; 5=E2=80=B2=E2=80=93CCACTTGGACTGGATGG=E2=80=933=E2=80=B2), = ZNF395 (5=E2=80=B2=E2=80=93CTGCATGGAAGTCAAAGGAA=E2=80=933=E2=80=B2= ; 5=E2=80=B2=E2=80=93AACCCAATGTCTGAGGGAAC=E2=80=933=E2=80=B2= ), POLR2A=20 (RefSeq ID NM_000937) (5=E2=80=B2=E2=80=93CTGCCAACACAGCCATCTAC=E2=80=933=E2=80=B2= ; 5=E2=80=B2=E2=80=93TCACCCATTCCTGATCCTCT=E2=80=933=E2=80=B2= ) and TNFAIP3=20 as previously published [36]. Quantitect Primers (Qiagen) = were used=20 for KISS1R. The following PCR conditions were used: 95=C2=B0C, 15 = minutes; 45=20 cycles at 94=C2=B0C, 30 seconds; 50=C2=B0C, 30 seconds; 72=C2=B0C, = 20 seconds. The=20 measured transcript abundance was normalized to the level of = POLR2A (RNA=20 Polymerase II) for all samples.

In situ Hybridization = and=20 Immunohistochemistry

Fragments of the = TREM-1 and=20 ZNF395 cDNAs were obtained by PCR using specific primers=20 (TREM-1: forward 5=E2=80=B2-TTGTCTCAGAACTCCGAGCTGC-3=E2=80=B2, reverse=20 5=E2=80=B2-GAGACATCGGCAGTTGACTTGG-3=E2=80=B2;=20 ZNF395: forward 5=E2=80=B2-TTTTGGTTCTCCCCAAACTG-3, reverse 5=E2=80=B2-GGTGGAAGAGCAGACAGAGG-3=E2=80=B2). PCR=20 products were purified and cloned into pBluescript-KS-M13+.=20 Digoxigenin-11-UTP (DIG)-labeled riboprobes were synthesized from=20 =C3=97baI or PstI linearized plasmids by in = vitro=20 transcription reaction with T7 or Sp6 polymerase and with = incorporation of=20 DIG, and used for in situ hybridization as described = (protocol=20 Roche Applied Science). Briefly, freshly cut frozen sections were = fixed=20 with 4% paraformaldehyde and hybridized with 0.4 ng/=C2=B5l = riboprobes in=20 hybridization buffer (50% formamide, 4=C3=97 SSC, 10% dextran = sulfate, 1=C3=97=20 Denhardts solution, 2 mM EDTA, and 500 =C2=B5g/ml sheared salmon = sperm DNA) at=20 45=C2=B0C overnight, followed by increasingly stringent washes of = SSC to 0.02=C3=97,=20 immunodetection using anti-DIG antibody (1 3D[ratio]=20

200 dilution), and stained by a combination of BCIP=20 (5-Bromo-4-Chloro-3=E2=80=B2-Indolyphosphate p-Toluidine Salt) and = NBT (Nitro-Blue=20 Tetrazolium Chloride) (Roche Applied Science). Slides were = subsequently=20 counterstained with methyl green.

Immunohistochemistry for = CD11b=20 (Pharmingen, dilution 1 3D[ratio]=20

100), CD45 (DAKO, dilution 1 3D[ratio]=20

100) and KDR (Santa Cruz, dilution 1 3D[ratio]=20

500) were performed on 5 =C2=B5m frozen sections = according to standard=20 procedures using 5 min cold acetone fixation and 1 hour incubation = with=20 the primary antibody. Expresssion of KISS1R was evaluated on = paraffin=20 embedded glioblastoma samples (NOVUS Biologicals, GPR54 antibody, = dilution=20 1 3D[ratio]=20

500) according to standard procedures using high = temperature=20 epitope retrieval (citrate buffer pH 6.0, pressure cooker 20=20 min).

Top

Supporting Information

Table S1

(0.06 MB DOC)

Click=20 here for additional data file.^{=20
(55K)}

Table S2

(0.12 MB DOC)

Click=20 here for additional data file.^{=20
(119K)}

Acknowledgments

We thank our colleagues from = Neurosurgery for=20 providing fresh tumor tissue and Robert Janzer for helpful=20 discussions.

Footnotes

Competing Interests: The = authors=20 have declared that no competing interests exist.

Funding: This project = was=20 supported by the NCCR Molecular Oncology (to MH and MD), the Swiss = National Science Foundation (grant 3100AO-108266/1 to MH) and the = National=20 Institute of Health (RO1 CA120813 to AH). The funders had no = role in=20 study design, data collection and analysis, decision to publish, = or=20 preparation of the manuscript.