Hereditary leiomyomatosis and renal cell carcinoma (HLRCC) is caused by mutations in the Krebs cycle enzyme fumarate hydratase (FH). It has been proposed that “pseudohypoxic” stabilization of hypoxia-inducible factor-α (HIF-α) by fumarate accumulation contributes to tumorigenesis in HLRCC. We hypothesized that an additional direct consequence of FH deficiency is the establishment of a biosynthetic milieu. To investigate this hypothesis, we isolated primary mouse embryonic fibroblast (MEF) lines from Fh1-deficient mice. As predicted, these MEFs upregulated Hif-1α and HIF target genes directly as a result of FH deficiency. In addition, detailed metabolic assessment of these MEFs confirmed their dependence on glycolysis, and an elevated rate of lactate efflux, associated with the upregulation of glycolytic enzymes known to be associated with tumorigenesis. Correspondingly, Fh1-deficient benign murine renal cysts and an advanced human HLRCC-related renal cell carcinoma manifested a prominent and progressive increase in the expression of HIF-α target genes and in genes known to be relevant to tumorigenesis and metastasis. In accord with our hypothesis, in a variety of different FH-deficient tissues, including a novel murine model of Fh1-deficient smooth muscle, we show a striking and progressive upregulation of a tumorigenic metabolic profile, as manifested by increased PKM2 and LDHA protein. Based on the models assessed herein, we infer that that FH deficiency compels cells to adopt an early, reversible, and progressive protumorigenic metabolic milieu that is reminiscent of that driving the Warburg effect. Targets identified in these novel and diverse FH-deficient models represent excellent potential candidates for further mechanistic investigation and therapeutic metabolic manipulation in tumors. Cancer Res; 70(22); 9153–65. ©2010 AACR.
The hereditary leiomyomatosis and renal cell carcinoma (HLRCC) syndrome is an inherited condition in which affected individuals are at risk of developing predominantly benign cutaneous and uterine leiomyomas, and aggressive type II papillary and collecting duct renal cell carcinomas (RCC; ref. 1). These RCCs are relatively resistant to systemic therapies (2–4). Although patients with HLRCC carry heterozygous germline mutations in the gene encoding the Krebs cycle enzyme fumarate hydratase (FH E.C. 18.104.22.168.; refs. 3, 4), the mechanisms of tumorigenesis and tissue specificity remain to be defined. Elucidation of these mechanisms would support the development of targeted therapies for HLRCC and might inform the biology of a broadening array of tumor syndromes and somatic (acquired) tumors associated with mutations in Krebs cycle enzymes (5). These include mutations in succinate dehydrogenase (SDH) subunits B, C, and D that cause paragangliomas and pheochromocytomas (6); mutations in the SDH-interacting protein SDH5 (SDHAF2, required for flavination of SDHA; ref. 7) that cause paragangliomas; and mutations affecting isocitrate dehydrogenase 1 and 2 (8) converting the Krebs cycle intermediate isocitrate to α-ketoglutarate that have been described in brain tumors.
Several possible mechanisms have been proposed to contribute to the tumor formation that results from FH mutations (5). An important observation is that elevated fumarate levels have been noted in HLRCC tumors (9). Because fumarate inhibits 2-oxoglutarate–dependent enzymes, one potential pathogenic pathway is the inhibition of hypoxia-inducible factor (HIF) hydroxylases (Ki values of 50–80 μmol/L; ref. 10), resulting in “pseudohypoxic” HIF stabilization and HIF target gene upregulation (9, 11, 12). HIF stabilization induces expression of a broad range of target genes, including those that stimulate growth and angiogenesis that may contribute to tumorigenesis (13). Furthermore, it has been observed that HIF stabilization in HLRCC may also result from glucose-dependent reactive oxygen species generation (12, 13). In a similar manner, fumarate may also inhibit other 2-oxoglutarate–dependent dioxygenases, including JMJD6, which catalyzes lysyl hydroxylation of U2AF65, an RNA splicing protein (14); the JmjC family of histone demethylases (15); and extracellular matrix processing enzymes (collagen prolyl and lysyl hydroxylases). Recently, evidence of the contribution of bioenergetics, as exemplified by the Warburg effect, to tumorigenesis in FH-deficient cells has been adduced from cell lines including an immortalized line from a patient with aggressive HLRCC-associated recurring kidney cancer (16, 17).
We hypothesized that as well as activating a pseudohypoxic response (18), FH deficiency activates a biosynthetic milieu, i.e., an environment that is both permissive of and contributory to tumorigenesis. To define the progressive consequences of FH deficiency in detail, we developed a novel Fh1 (murine orthologue of FH)-deficient mouse embryonic fibroblast (MEF) line and characterized changes in the expression profile of these cells. These expression changes were compared with the transcriptome of Fh1-deficient murine renal cysts and an advanced renal cancer from an HLRCC patient. To test the generality of any biosynthetic changes that emerged from these systems, we assessed the effect of FH deficiency in an in vitro (pancreatic β-cells) and in vivo smooth muscle model of FH deficiency. What emerges is the importance of synergizing expression changes in a single pathway (glycolysis—HK2, PFKP, ENO2, PKM2) in HLRCC tumorigenesis.
Materials and Methods
Mouse husbandry and generation of MEFs
Procedures involving live animals were in accordance with the Home Office guidelines and licensing regulations (project license 30/2368). MEFs were isolated by standard techniques from littermate embryos as previously described (18). Fh1/Ksp-Cre conditional knockout mice have been previously described (19). Mice expressing Cre recombinase under the smooth muscle myosin heavy chain (smMHC) promoter were a kind gift from Dr. Michael Kotlikoff (20). All mice were backcrossed onto a C57BL/6 genetic background for at least five generations.
Gene expression arrays
mRNA was extracted from each sample with the GenElute mammalian total RNA miniprep kit (Sigma, RTN70) All subsequent steps in the microarray expression analysis were carried out by the Cancer Research UK GeneChip Microarray Service. Briefly, after quality control, the RNAs were reverse transcribed to double-stranded cDNAs and biotin-labeled complementary RNA (cRNA) was amplified by in vitro transcription with T7 RNA polymerase. cRNAs were then hybridized to either HGU133A (Human RCC) or MOE430_2 (mouse kidney tissue) Affymetrix Chips. Full protocols are available at The Paterson Institute Microarray Service Web site (http://bioinformatics.picr.man.ac.uk/mbcf/protocols.jsp).
Chips were quantile normalized and quantified using robust multiarray averaging. A linear model was fitted to the knockout and wild-type groups and differential genes selected by applying a 0.001 false discovery rate threshold to P values corrected using the Benjamini and Hochberg method. T statistics were moderated before transcript selection by empirical Bayes shrinkage of the standard errors. The analysis was carried out using the Limma and Affy packages from Bioconductor 2.2 within R 2.7.0.
Magnetic resonance spectroscopy metabolite analysis
Magnetic resonance spectroscopy (MRS) analysis was done as previously described (9). In brief, metabolites were extracted from cells and tissue using perchloric acid and freeze dried. The samples were resuspended in D2O and analyzed using 1H MR. Media from the cell study were also collected and analyzed by 1H MRS. Sodium 3-trimethylsilyl-2,2,3,3-tetradeuterpropionate (TSP) was added to the samples for chemical shift calibration and quantification.
Cell culture, viral production, and standard molecular and histologic methodology are described in the Supplementary Materials and Methods.
Fh1-deficient MEFs exhibit a direct upregulation of the HIF-1α pathway
We generated several MEF lines as a cellular model of FH deficiency. Karyotype analysis by multiplex-fluorescence in situ hybridization identified a variable degree of spontaneous polyploidy in all MEF lines analyzed at passage number 10, including the wild type. In particular, the Fh1-deficient fibroblasts showed a chromosome number ranging from 4n± to 6n±. However, neither structural rearrangements nor numerical clonal abnormalities were detected (Fig. 1A). These MEFs had a metabolomic profile consistent with FH deficiency (fumarate was undetectable in wild-type cells, but 8.9 fmol/cell in knockout; Table 1). To characterize these lines further, we measured metabolites in the overlying culture media after 16 hours incubation with these cells. There was some evidence of accelerated glycolysis in the Fh1-deficient MEFs as previously reported for other FH deficiency cells and as corroborated by markedly increased lactate levels (Table 1; ref. 12). These cells manifested HIF-1α protein stabilization (Fig. 1B, i) and an upregulation in HIF target genes: glucose transporter 1 (Glut1; 17.5 ± 1.9-fold), Hk2 (3.8 ± 0.4-fold), and lactate dehydrogenase A (Ldha; 8.7 ± 0.9-fold; Fig. 1B, ii). To confirm that the changes were a direct consequence of FH deficiency rather than a secondary genetic or epigenetic modification, we showed that viral introduction of an (human) FH protein normalized this HIF upregulation and gene expression profile in a time-dependent manner (Fig. 1C). Forty-eight hours postviral infection, HIF-1α protein was barely detectable and HIF target genes were significantly reduced (Egln3, 0.09 ± 0.04-fold; Hk2, 0.14 ± 0.01-fold; Slc2a1, 0.19 ± 0.01-fold; and Vegfa, 0.43 ± 0.02-fold). Fh1-deficient MEFs seem to be a model in which to study the early consequences of FH deficiency.
Fh1-deficient MEFs manifest a significant metabolic phenotype
Because our MEF lines strongly upregulated Hk2, Glut1, and Ldha and manifested augmented glycolytic flux, we assessed whether they exhibited a more elaborate metabolic phenotype at RNA expression level. As detailed in Fig. 2A, these MEF lines significantly upregulate the expression of the platelet isoform of phosphofructokinase (Pfkp), aldolase A (Aldoa), the M2 isoform of pyruvate kinase (Pkm2), Ldha (genes all involved with glycolysis), and NADP(+)-dependent malic enzyme 1 (cytosolic; Me1; also involved in anaplerosis). This metabolic profile in Fh1-deficient MEFs mimics exactly that observed in both murine renal cysts and HLRCC-related human RCCs where PKM2 and LDHA proteins are progressively upregulated and strengthens the validity of this model system (Fig. 2B). The upregulation of PKM2 and LDHA was ordered and strictly confined to cysts. In RCCs, these enzymes were very highly expressed and diffusely distributed (Fig. 2B). As noted previously (21), we observed that PKM2 was upregulated in malignant non-HLRCC–related RCCs (Fig. 2B).
Renal cysts from Fh1-deficient mice and HLRCC renal cancer show consistent and progressive changes in gene expression
Gene expression profiles from cysts derived from animals with renal Fh1 deficiency (Fh1fl/fl Ksp1.3/Cre), which manifest an appropriate metabolomic profile (46.8-fold increased cellular fumarate; P < 0.002) compared with control tissue (Fig. 3A), were compared with an HLRCC papillary cell carcinoma. The genes with the most profound increase in expression in both FH-deficient cysts and RCC, as detailed in the ontogeny analysis (Table 2), were those predominantly involved in intermediary metabolism and aerobic respiration. Specifically, there was a prominent and consistent upregulation of HIF target genes; e.g., hexokinase II (HK2), ceruloplasmin (CP), neuron-specific enolase (ENO2, a glycolytic enzyme predictive of prognosis in RCC; refs. 22, 23), cyclin-dependent kinase inhibitor 1a (CDKN1A), and prolyl hydroxylase 3 (PHD3/EGLN3) highlighted in gray in Table 2. Several other genes with increased expression in Fh1-deficient cysts and RCC are functionally related to tumorigenesis and tumor progression: lipocalin 2 (LCN2; strongly upregulated during inflammation and an inducer of tumorigenesis; ref. 24), group-specific component (GC; Vitamin D-binding protein), tribbles homologue 3 (TRIB3; a homologue of the Drosophila tribbles activated in hypoxia and that controls cell growth through ubiquitination-dependent protein degradation; refs. 25, 26) MYC and FOS (myelocytomatosis and Finkel-Biskis-Jinkins murine osteogenic sarcoma viral oncogenes—transcription factors regulating the expression of 15% of all genes including those involved in cellular proliferation; ref. 27) and CD44 (involved in cell-cell interactions and migration; refs. 28, 29). To validate these observations, we did quantitative reverse transcription-PCR analyses on Fh1-deficient renal cysts (Fig. 3B), confirming that the microarray observations, Lcn2, Hk2, Cp, c-Myc, and c-Fos were upregulated in cysts. Moreover, immunohistochemical analysis with CD44, C-Myc, and C-Fos showed that there were corresponding increases in numbers (and intensity) of cells stained which were distributed over the proliferating cyst surfaces (Fig. 3C). Expression analysis of human HLRCC renal cancer showed similar increases in HK2, CP, and CD44 throughout the RCC tumor mass. Despite the appropriate distribution of these proteins (CP and HK2—intracellular and within the core of the tumor, CD44—tumor surface), these molecules seemed to be spatially less organized than in cysts, consistent with the less differentiated state of the RCC (Fig. 3D). A further feature of these changes in cysts (benign) and renal cancer (malignant), notwithstanding the challenges of normalization in microarray studies, was the progressive increase in the expression of HIF target genes (HK2, CP, ENO2, CDKN1a, and EGLN3) and in genes known to be involved in tumorigenesis and metastasis [e.g., MYC, CD44, CDKN1a, heparan sulfate glucosamine 3-O-sulfotransferase 1 (HS3ST1) and transmembrane glycoprotein NMB (GPNMB)]. Of equal importance was a progressive downregulation of several genes where again the expression profiles are consistent with those seen in tumorigenesis (e.g., PVALB downregulation has been noted in inherited renal neoplasm syndromes; ref. 30) or of importance to the tumor metabolism [argininosuccinate synthetase 1 (ASS1) and phenylalanine hydroxylase (PAH) are involved in amino acid metabolism, dihydropyrimidinase (DPYS) in nucleic acid metabolism (31) and endoplasmic reticular glucose-6-phosphatase (G6PC) and phosphoenolpyruvate carboxykinase (PCK1) which encode gluconeogenic enzymes that have been related to cell proliferation and growth arrest; ref. 32].
FH deficiency promotes cellular proliferation and growth
We reasoned that the metabolic changes resulting from FH deficiency, at least in part, recapitulated the Warburg effect, which, as has been suggested, may promote tumorigenesis (33). Therefore, we hypothesized that this milieu might correlate and contribute to cellular proliferation in HLRCC. Using an Fh1fl/fl SMC22a-Cre strategy (19, 20) to eliminate Fh1 from smooth muscle (using four Fh1fl/fl SMC22a-Cre+/− and four Fh1fl/fl controls), we assessed whether Fh1 deficiency in mice would recapitulate smooth muscle cell (SMC) changes noted in HLRCC (i.e., leiomyoma). The life span of these animals was severely curtailed. Due to animal welfare considerations, this mouse line was promptly terminated, permitting only a limited set of experiments to be done. Nevertheless, using genomic PCR, we confirmed that Fh1 had been efficiently deleted in the SMC of the intestine (and the uterus; Fig. 4A). Moreover, consistent with successful Fh1 deletion of smooth muscle, metabolomic analysis of Fh1fl/fl SMC22a-Cre bowels displayed a predicted profile [i.e., 8.03-fold (P < 0.02) increased cellular fumarate compared with control tissue; Table 3]. Consistent with our hypothesis that metabolic changes lead to a promotion of cell growth/proliferation, we observed that mice carrying a conditional Fh1 deletion in smooth muscle manifest prominent (5- to 10-fold) intestinal wall thickening throughout the gastrointestinal tract (GIT) from stomach to colon (Fig. 4B). Immunohistochemical assessment of the Fh1-deficient GIT showed no differences in either cellular proliferation (assessed by phospho-histone H3; ref. 34), or mucosal differences (assessed by endomucin) when compared with wild-type GITs (Fig. 4C, i). In contrast, there was a clear expansion of the muscularis layer of the bowel as assessed by immunostaining for smooth muscle (SM actin; Fig. 4C, i). There was, however, no evidence of mammalian target of rapamycin (mTOR)/phospho-mTOR activation as noted in another prominent model of widespread GIT SMC hyperplasia [i.e., mice carrying a homozygous SMC-specific deletion of phosphatase and tensin homologue (Pten) alleles which developed widespread SMC hyperplasia and abdominal leiomyosarcomas; data not shown; ref. 35]. Expansion of the muscularis layer resulted from both significant cellular hyperplasia (7.44 × 10−4 ± 5.16 × 10−5 cells/μm2 in knockout versus 2.07 × 10−3 ± 6.09 × 10−5 cells/μm2 in wild type, respectively; P < 4.4 × 10−6) and cellular hypertrophy (area: 1.35 × 103 ± 96 μm2 in knockout versus 4.83 × 102 ± 14 nm2 in wild type, respectively; P < 5.0 × 10−5). Most strikingly, as seen with the renal progression to cancer, there was a compelling and progressive increase in PKM2 and LDHA in the smooth muscle leiomyoma (Fig. 4C, ii). Finally, to assess whether these metabolic consequences of FH deficiency were applicable to other cells, we knocked down Fh1 in a murine pancreatic β-cell line (Min6) which is histogenically unrelated to HLRCC tissues but is important in glucose metabolism (36). Successful Fh1 knockdown recapitulated the results noted above with statistically significant increases in Pdk1, Ldha, and Pkm2 mRNA (Supplementary Fig. S1).
We provide evidence to support our hypothesis that expression changes (e.g., those related to HIF-1α stabilization) and the metabolic profile (e.g., increased PKM2) is a primary consequence of Fh1 deficiency in MEF lines and are recapitulated in Fh1-deficient renal cysts and HLRCC type II papillary carcinomas contributing to tumorigenesis. Genetic manipulations in these MEFs confirm that this HIF-1α stabilization is related to the early pseudohypoxic consequences of FH deficiency and fumarate accumulation. Fh1-deficient renal cysts and HLRCC type II papillary carcinomas manifest consistent and progressive changes in a select set of HIF target genes, genes related to intermediary (especially glycolytic) metabolism and in genes related to cell proliferation. FH deficiency is associated with a metabolic phenotype reminiscent of the Warburg effect, correlating with and potentially contributing to tumorigenesis.
Although two cell lines (UOK 262 and A549) already exist for the study of HLRCC/FH deficiency, these are either models of advanced cancer with established cytogenetic changes [47,X,−X,+1,i(1)(q10), + 5,der(21)t(15;21)(q15;p11.2),+22; ref. 16] or are predicated on small interfering RNA (siRNA) modification of already aberrant cell lines (alveolar basal epithelial cells derived from human carcinoma; ref. 17) respectively. Although these existing lines may be representative of advanced tumors and recapitulate many of the features of HLRCC-RCC described above including HIF upregulation, upregulation of genes involved in glucose metabolism and features consistent with a highly invasive phenotype (12, 16), these lines do not provide a system to study the earliest events leading to tumorigenesis. To this end we derived primary Fh1-deficient MEFs. As predicted, these MEFs upregulated Hif-1α and its target genes and showed metabolomic changes consistent with FH deficiency (increased fumarate and aspartate levels with accelerated glycolysis). Re-expression of FH by viral transduction reversed Hif-1α and target gene expression confirming the established direct role of FH deficiency in HIF upregulation (11, 37).
The observation that Fh1-deficient cysts are monoclonal and have moderately increased proliferative indices suggests that they may be a good model of early preneoplastic HLRCC lesions (19). Supporting this inference, we observed that the expression profiles identified in cysts and in HLRCC-derived RCC are remarkably consistent both in the genes that are altered and with a transition to malignancy from benign cysts to malignant tumors. A common feature of these expression studies is the accumulation of HIF-1α and select HIF target genes (e.g., HK2, CP, ENO2, SOX9, and EGLN3) most likely driven by pseudohypoxia. Under these conditions HIF-1α stabilization is mediated by competitive inhibition (fumarate competing with 2-oxoglutarate) of non-haem Fe(II) and 2-oxoglutarate–dependent dioxygenases [e.g., prolyl hydroxylase domain proteins (PHD) 1–3; ref. 13]. As a corollary, accumulation of intracellular fumarate achieved either by biochemical inhibition of FH or siRNA-mediated FH knockdown results in HIF-1α stabilization and has been proposed to promote tumorigenesis (11, 19, 37). In keeping with this proposal, there is also an upregulation in genes known to enhance proliferation and metastasis (e.g., LCN2, ANLN, MYC, TRIB3, CDKN1A, and CD44). Thus, not only did these expression changes confirm the congruity between Fh1-deficient cysts and HLRCC cancer, but they also identify several biological pathways progressively activated in FH deficiency. These changes (e.g., upregulation of prometastatic CD44; refs. 28, 29) were confirmed immunohistologically in cysts and RCC.
It is well recognized that HIF-1α upregulation results in the increased transcription of genes involved in glucose transport (e.g., GLUT1), glucose metabolism and lactate formation (e.g., ENO2 and LDHA), lactate export from cells (e.g., monocarboxylate transporter 1, MCT1), and the diversion of glycolytic carbon units away from the mitochondria and toward fermentation by inhibition of the pyruvate dehydrogenase complex (e.g., PDK1; ref. 13). Thus, the immediate metabolic consequences of FH deficiency coupled with HIF-1α upregulation may be sufficient to facilitate the aerobic glycolysis characteristic of tumors (Warburg effect; refs. 38, 39). HIF-α, especially HIF-2α, is also known to cooperate with c-MYC (which is upregulated in these FH-deficient tumors; Fig. 1) and promotes cellular proliferation (40, 41). Because c-MYC upregulates transcription of PTB, hnRNPA1, and hnRNPA2, ensuring a high cellular PKM2 level, several mutually compatible pseudohypoxic pathways exist that may synergize into an oncogenic phenotype (42) potentially providing a mechanism that we could extrapolate to our diverse models of FH deficiency. Despite being a model of early FH deficiency, our MEFs exhibited elevations in genes with a metabolic signature typical of proliferating cancer cells (e.g., Pkm2, Ldha, and Me1; refs. 6, 33). Moreover, our immunohistochemical and gene expression analysis showed that this Pkm2 and Ldha upregulation was progressive from MEFs, through cysts to the HLRCC-RCCs.
As leiomyomatosis is an important but neglected aspect of HLRCC, we generated novel mice with Fh1-deficient smooth muscle (Fh1fl/fl SMC22a-Cre) to test the above hypothesis. Although these mice had compromised viability, the knockout animals exhibited profound and diffuse GIT enlargement, with a 5- to 10-fold increase in bowel diameter (Fig. 4B and C). Moreover, histologically, these mice manifested profound and uniform SMC hypertrophy limited to the muscularis. Immunohistochemical studies, concordant with the renal studies, showed striking PKM2 and LDHA upregulation that was progressive from normal tissue, through benign intestinal SMC thickening to HLRCC-related leiomyoma. Interestingly, this phenotype (i.e., upregulation of PDK1, PKM2, LDHA, and ME1) is reminiscent of the expression profile in leiomyoma (43).
It is tempting to infer that FH deficiency drives cells toward a common final tumorigenic phenotype when such a striking number of changes converge upon the glycolytic pathway (Supplementary Fig. S2; ref. 6). Upregulation of HK (38, 44), PFKP, ENO2 (22, 23), PKM2 (21, 33, 42), and LDHA (ref. 17; with increased Aldoa in MEFs) coordinates the increased glycolysis as confirmed by the metabolomic studies, whereas downregulation of PCK1 (32) and G6PC retards the antagonistic effects of gluconeogenesis. The upregulation of malic enzyme (ME1) also accentuates this glycolytic carbon cycling while increasing cytosolic NADPH (6). Hence, consequent to FH deficiency, glycolytic carbon flux is accelerated and glucose is channeled through the pentose phosphate pathway, increasing the NADPH/NADP+ ratio and encouraging incorporation of nutrients into the biomass (e.g., nucleotides, amino acids, and lipids) necessary for proliferating cells (6). Overlaid onto this metabolic profile are cross-talks with proliferogenic signaling pathways (including augmented HIF-1α expression), with the potential to induce cellular growth and proliferation (6) and allow for further changes [e.g., upregulation LCN2 (24), MYC (27, 40, 41), TRIB3 (25), CDKN1A (45), ANLN (46), HS3ST, ITPR3 (inositol 1,4,5-triphosphate receptor, type 3), GPNMB, and CD44 (28, 29)] leading to tumorigenesis. Possible evidence for a more general extrapolation of this hypothesis is underlined by the observation in Fig. 3 that PKM2 is also found in renal tumors, especially those with extrarenal malignant metastasis (21). Confirming the potential ubiquity of this common metabolic pathway, as shown in Supplementary Fig. 2, even histogenically unrelated cells such as Min6 (pancreatic β-cell line) recapitulate the biosynthetic profile when Fh1 is successfully knocked down.
Our primary contention is that primary FH deficiency advances the metabolic component of tumorigenesis but we also propose that FH deficiency may be acquired secondarily through selective advantage by other (non-HLRCC) tumors. To begin to investigate this question, we searched existing public databases (GEO) to see whether FH was downregulated in other tumors. We found that in diverse tumors, including N-methyl-N-nitrosourea (NMU)–induced mammary tumors in rats (47), normal-appearing colonic mucosa in early onset colorectal cancer (CRC; ref. 48), prostate cancer (49), and in lung cancer cell lines (50), FH mRNA was downregulated by 35% (P < 4 × 10−8), 41% (P < 5 × 10−3), 61% (P < 6 × 10−2), and 37% (P < 9 × 10−3), respectively. Caveats to these data include the reliability of expression studies without RNA/protein validation and without biochemical validation by measurement of increased fumarate. However, the results of these analyses are consistent with the hypothesis that secondary FH downregulation might contribute to tumorigenesis.
In conclusion, we suggest that the cellular response to FH deficiency is manifested in early, rapid, and consistent expression changes (e.g., HIF-1α target and glycolytic/biosynthetic gene upregulation). As proposed in other tumors, these metabolic and gene expression changes (e.g., HK2, ENO2, PKM2, LDHA, ME1 upregulation) seem to be sufficient to promote cellular growth and proliferation (6, 33). Subsequent expression changes putatively arising from genetic and epigenetic changes within preneoplastic cells might drive tumorigenesis. We propose that tumors arising in the context of FH deficiency represent an exemplar for the role of metabolism in supporting and driving cellular proliferation and highlight potential biological and therapeutic targets to be studied.
Disclosure of Potential Conflicts of Interest
P.H. Maxwell and C.W. Pugh are scientific cofounders of ReOX Ltd. The other authors disclosed no potential conflicts of interest.
We thank Stuart Pepper and Yvonne Hey (Cancer Research UK Microarray Facility, Manchester, United Kingdom) and Alison Martin (Clare Hall Laboratories, South Mimms, Cancer Research UK).
Grant Support: Cancer Research UK (C6079/A9485 and C10843/A12027), The Wellcome Trust (WT091112MA; P.J. Pollard), and a British Heart Foundation Centre for Research Excellence Award (H. Ashrafian). P.J. Pollard is a Beit Memorial Fellow.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received May 31, 2010.
- Revision received August 12, 2010.
- Accepted September 4, 2010.
- ©2010 American Association for Cancer Research.