Hypoxia-Inducible Factor-Alpha Activation in Lymphoma and Relationship to the Thioredoxin Family

Abstract

Hypoxia-Inducible-Factors (HIFs) activate oncogenic pathways, while thioredoxins, including thioredoxin-1 (Trx1) and thioredoxin reductases-1 and -2 (TrxR1 and TrxR2), promote HIF-α stabilization. Elevated levels of thioredoxin or HIF have been associated with poor outcomes in solid tumors and each represent potential therapeutic targets. In lymphoma cell line immunoblotting studies, we found that Raji and SUDHL4 cells exhibited normoxic HIF-2α protein-stabilization. Furthermore, five lymphoma cell-lines showed increased TrxR1 expression, while only Namalwa, HF1, and SUDHL4 had Trx1 and TrxR2 activation. Utilizing tissue microarrays from diffuse large B-cell (DLBCL) and follicular lymphoma (FL) patient specimens (n=82), we found different frequency of HIF expression in FL versus DLBCL as well as differing HIF-1 versus HIF-2 expression within each histologic subgroup. Forty-four percent of DLBCL versus 11% of FL cases had moderate-to-high expression of both HIF-1α and HIF-2α (p=0.0017), while 56% of DLBCL and 32% of FL samples had at least low HIF-1α and HIF-2α expression (p=0.042). Survival analysis among newly-diagnosed DLBCL cases showed 44% 2-year event-free survival (EFS) and 67% overall survival (OS) with high HIF-1α and HIF-2α expression compared with 67% and 76%, respectively, (p=0.10 and p=0.64, respectively) without high expression. These data demonstrate that HIF and the thioredoxins are activated in lymphoma and that HIF expression may influence prognosis.

Introduction

Hypoxia-Inducible-Factors (HIFs) are a family of heterodimeric transcription factors that regulate expression of more than 150 genes involved in the response to hypoxia.¹ HIF is important in cancer, as it regulates various oncogenic genes as well as genes involved in cell survival, proliferation, and migration.^1,2 In tumors, the combination of tissue hypoxia and cell growth leads to increased HIF activity, which drives tumor angiogenesis, fosters motility, and inhibits apoptosis.² Several solid tumors demonstrate constitutive HIF activation^3,4 and elevated HIF levels in solid tumors have been linked to poor prognosis.³ A role for HIF in lymphoma has not been well established.

The thioredoxin family are proteins that contain a Cys-Gly-Pro-Cys dithiol active site in the reduced form catalyzing reduction of disulfides.⁵ Increased expression of thioredoxin-1 (Trx1) has been linked to cancer progression/prognosis,^6,7 including diffuse large B-cell lymphoma (DLBCL).⁸ Trx1 over-expression has also been found to trigger HIF-α stabilization.⁹ Furthermore, thioredoxin reductases-1 and -2 (TrxR1 and TrxR2) are important effectors in the thioredoxin system and they are also over-expressed in many tumors.^6,10 However, little is known about the thioredoxin family in lymphoma.

Material and methods

Western blotting for HIF and thioredoxin

Whole cell lysate protein was electrophoresed in SDS-PAGE gel (12%-Trx, 10%-HIF) and transferred to nitrocellulose membrane (Invitrogen, Carlsbad, CA). Membranes were probed with primary antibodies (Trx1- Santa Cruz SC-18215; TrxR1- Santa Cruz SC-18220; TrxR2-Santa Cruz SC-46278; HIF-1α- BD Transduction Laboratories-BD610958; HIF-2α- Novus Biologicals-NB100-122; β-actin- Abcam-AB6276-100; GAPDH- Chemicon-MAB374) at 1:200, 1:1000, 1:1500, or 1:5000 at 4°C. They were incubated with horseradish peroxidaseconjugated bovine anti-goat IgG secondary antibody at 1:1000. Immune complexes were visualized by enhanced chemiluminescence (Western Blotting Detection Reagents, Amersham Biosciences) using high-performance chemiluminiscence film.

Tissue microarrays

Paraffin-embedded sections of lymphoma tissue were retrieved through the Northwestern University Pathology Core archives following Institutional Review Board approval. Northwestern University Pathology Standard Operating Procedure (SOP) for lymph node procurement: Under clinical Pathology practices, surgical resections and biopsies were harvested, fixed and processed using automated instruments (Leica TP1050). In addition, following collection in the operating room, all lymph nodes biopsies were either immediately placed in fixative or were brought fresh to the Pathology Core center where they were processed immediately. Tissue microarrays (TMA) were comprised of DLBCL and follicular lymphoma (FL) specimens (45 FL specimens- 32 untreated; 37 DLBCL specimens- 30 untreated). All DLBCL patients had received initial treatment with rituximab CD20 antibody in combination with anthracycline-based chemotherapy, while all FL patients had received frontline rituximab/chemotherapy.

Immunohistochemical stains were performed on 5μm tissue sections using heat-induced antigen retrieval. Sections were deparaffinized, dehydrated, and stained with monoclonal antibody (anti-HIF-1α dilution 1:2000, Novus- NB100-105; anti-HIF-2α dilution 1:500, Novus- NB100-132). HIF expression was assessed with semiquantitative immunohistochemical analysis. Briefly, ten random fields at 40× magnification were examined in each specimen, and the percentage of positively-stained lymphoma cells were estimated for HIF-1α (nuclear) and HIF-2α (cytoplasmic). Level of HIF expression was classified as absent (0%), low (1–10%), moderate (11–49%), or high (50–100%).¹¹ Relevant HIF expression cut-off points were determined to be approximate medians of underlying empirical mass functions. We report differences between groups having HIF-1α and HIF-2α expression levels between <10% and ≥10%, concomitant HIF-1α and HIF-2α expression <50% and ≥50%, and cases having only HIF-1α levels between <50% and ≥50%. SAS^(c) version 9.1 (Cary, NC) was used for data analysis. Differences in HIF staining were compared using Fisher's exact test. Survival analyses were performed using Kaplan and Meier curves, which were compared using log-rank testing.

Results/Discussion

Normoxic HIF stabilization and thioredoxin expression in lymphoma cell lines

We assessed HIF-1α and HIF-2α protein levels in multiple lymphoma cell lines (HF1 follicular lymphoma, SUDHL4 large B-cell lymphoma, and the Burkitt's lymphoma/leukemia cell lines Namalwa, Raji, Ramos, and Daudi). SUDHL4 and Raji cells demonstrated evidence of normoxic HIF-2α stabilization that was not seen in normal lymphocytes (Figure 1A). A small amount of normoxic HIF-1α stabilization was also seen in these cell lines. Normally, HIF-1α levels are minimally expressed during normoxia, because the constitutively expressed protein is rapidly degraded by the ubiquitin-proteosome system. Hydroxylation of conserved proline residues in the alpha subunit facilitates interaction with the von Hippel Lindau (VHL) protein, the E3 ubiquitin ligase for HIF-α.^12,13 During hypoxia, HIF-α stabilization occurs because HIF prolyl hydroxylation becomes inhibited, allowing the protein to increase. The mechanism underlying the increase in HIF-1α and/or HIF-2α stabilization observed in lymphoma is not known. Contributing mechanisms could include an increase in transcription/translation of the alpha-subunits, tissue hypoxia within the lymph tissue, and/or genetic modifications leading to loss-of-function in the prolyl hydroxylase–VHL–proteosome degradation pathway.

Figure 1. Hypoxia inducible factor (HIF) and thioredoxin activation in lymphoma cell lines and HIF in tissue microarray patient cases.

A. Immunoblots of HIF-1α and HIF-2α. Left: SUDHL4 cells (replicate lanes). Right: Normal lymphocytes (N.L.) and Raji cells. Normoxic HIF-2α is evident in SUDHL4 and Raji cells, but not in normal lymphocytes. One set of lanes included cells cultured under static conditions (− Agitation), while a second set was performed with agitation (+ Agitation) to eliminate possible contribution of hypoxia artifact in the tissue culture flask. + indicates DFO treatment, a hypoxia mimetic, as the positive control. B. Activation of the thioredoxin family in multiple lymphoma cell lines. C. TMA of diffuse large B-cell lymphoma for HIF-1α expression. D. Representative high magnification core section from a follicular lymphoma patient showing increased HIF-1α. E. Representative high magnification TMA section from a diffuse large B-cell lymphoma case with increased HIF-2α. F. Representative high magnification TMA section from a follicular lymphoma patient showing increased HIF-2α expression.

The thioredoxin components have never been studied in lymphoma. To explore this, we measured expression of Trx1, TrxR1, and TrxR2 in lymphoma cell lines. Figure 1B shows varying expession of thioredoxin system components. Namalwa, HF-1, and SUDHL4 showed the strongest expression of Trx1, while TrxR1 activation was seen in all lymphoma cell lines. TrxR2 activation was seen primarily with Namalwa, HF1, and SUDHL4, while minimal or no TrxR2 activity was seen in Ramos and Daudi lines.

Trx1 over-expression enhances HIF expression (under normoxic conditions) and HIF binding to DNA in the nucleus, while it increases HIF-dependent gene transcription.^5,14 TrxR1 deficiency leads to a significant attenuation of tumor progression in vivo,¹⁰ indicating its importance in cancer. Many cancers over-express Trx1, TrxR1, and TrxR2,^6–8,10,14 including DLBCL, where gene array data from Tome and colleagues showed that patients with the worst prognosis had decreased expression of antioxidant enzymes with increased thioredoxin-system function.⁸ It is possible that up-regulation of the thioredoxin system in lymphoma promotes HIF activation, which may potentially drive lymphomagenesis and clinical outcome.

HIF-1α and HIF-2α TMAs

An important role for HIF has emerged in cancer biology, including a relationship between elevated HIF-α expression with high tumor grade in solid tumors.^10,11 Stewart and colleagues showed that HIF was expressed in lymphoma, although level(s) of HIF expression and survival analysis were not reported.¹⁵ We examined 82 lymphoma nodal samples for HIF-1α and HIF-2α protein expression (Figure 1 C–F). Fifty-four percent of DLBCL cases showed moderate-to-high (>10%) HIF-1α expression compared with 20% of FL cases (p=0.001). A valid question is whether “physiological hypoxia” arising from the specimen collection be responsible for HIF stabilization? If physiologic hypoxia accounted for abnormal HIF stabalisation, there should be correlation between HIF-1 and HIF-2 levels. Instead, we found differing levels of HIF expression among DLBCL and FL and differing stabalisation of HIF-1 and HIF-2 within DLBCL and FL patient groups (Table 1). Forty-four percent of DLBCL versus 11% of FL biopsies had moderate-to-high expression of both HIF-1α and HIF-2α (p=0.0017), while 56% of DLBCL and 32% of FL samples had at least low expression of both HIF-1α and HIF-2α (p=0.042). On the other hand, varying levels of HIF-1 versus HIF-2 expression were seen within the FL and the DLBCL patient groups. Within the FL TMA, 30% of cases had moderate-to-high HIF-2 stabalisation with concurrent absent expression of HIF-1 in the same cases; while 17% of FL cases had at least low HIF-1 expression with absent HIF-2 expression. Furthermore, within the DLBCL TMA, 13% of cases had detectable HIF-2 expression with coexistent absent HIF-1 expression, and 10% of DLBCL cases had at least low HIF-1 expression with absent HIF-2 expression. Finally, 27% of FL and 25% of DLBCL samples exhibited no detectable stabilization of either HIF-1α or HIF-2α. There were no appreciable HIF-expression differences between untreated and relapsed patients, although the number of relapsed cases was small (data not shown).

Table 1.

Lymphoma tissue microarray (TMA) analysis for hypoxia inducible factor (HIF) protein expression.

	Follicular lymphoma (n=45)	Diffuse large B-cell lymphoma (n=37)
HIF-1α expression			P value+
Absent (0%)	24/45 (53%)	13/37 (35%)	0.12
Low (1–10%)	12/45 (27%)	4/37 (11%)	0.10
Moderate (11–49%)	5/45 (11%)	10/37 (27%)	0.09
High (50–100%)	4/45 (9%)	10/37 (27%)	0.04
HIF-2α expression *
Absent (0%)	19/44 (43%)	12/35 (34%)	0.49
Low (1–10%)	3/44 (7%)	5/35 (14%)	0.46
Moderate (11–49%)	5/44 (11%)	9/35 (26%)	0.14
High (50–100%)	17/44 (39%)	9/35 (26%)	0.24

^*One follicular lymphoma and two large B-cell specimens contained inadequate tissue for HIF-2α evaluation.

⁺p values for comparison of follicular lymphoma HIF expression with diffuse large B-cell lymphoma.

Event-free survival (EFS) and overall survival (OS) were calculated based on HIF expression. Analysis of histologic subgroups showed inferior survival trends with high HIF-1α and HIF-2α expression, although the differences were not statistically significant. Among newly-diagnosed DLBCL cases (median followup 26 months), 2-year EFS was 44% and OS was 67% for patients with high expression of both HIF-1α and HIF-2α versus EFS of 67% and OS of 76% for DLBCL patients without high HIF-1α and HIF-2α expression (EFS comparison p=0.10, OS comparison p=0.64). Among newly-diagnosed FL cases, 2-year EFS and OS were 71% and 90%, respectively, for patients with high expression of HIF-1α and HIF-2α versus EFS of 90% and OS of 95% for patients without high HIF-1α and HIF-2α expression (EFS p=0.10, OS p=0.62).

Our data indicate that HIF-α and the thioredoxins are activated/expressed in lymphoma cell lines and among primary patient lymph node samples. HIF and the thioredoxins represent potential novel therapeutic targets for the treatment of cancer.^2,16–18 This may be accomplished through direct inhibition of HIF-1α and HIF-2α and/or Trx1 and TrxR1/2 using direct small molecule inhibitors;^18,19 with indirect inhibitors of HIF through downstream signaling pathways such as with histone deacetylase inhibitors;²⁰ and possibly through hypoxia-regulated genes such as vascular-endothelial-growth-factor (VEGF). Among TMA specimens, we found significantly higher expression of HIF-α among DLBCL cases compared with FL. It is possible that this observation is related to the more aggressive nature of DLBCL compared with FL. Prospective analysis of the prognostic importance of HIF-α and the thioredoxins in larger lymphoma treated patient cohorts is warranted.

The involvement of HIF in lymphoma is likely to differ from its involvement in solid tumours, possibly with respect to angiogenesis. Likewise, the involvement of Trx in lymphoma, and its potential relationship to HIF, will require further study to dissect. Our observation that a significant fraction of patients with lymphoma show enhanced stabilisation of HIF-1 and/or HIF-2 suggests that HIF-dependent genes may confer a proliferative advantage, a survival advantage, or an enhanced resistance to therapy. Based on the very large number of genes regulated by HIF, it seems likely that more than one gene target is responsible for these effects. Dissecting the responsible candidates will require extensive future study in the laboratory. Further studies are necessary to determine whether HIF-1α and/or HIF-2α promote lymphoma growth or progression, and to define the molecular interactions between HIF and thioredoxin that may mediate this association.