NECAB3 Promotes Activation of Hypoxia-inducible factor-1 during Normoxia and Enhances Tumourigenicity of Cancer Cells

Unlike most cells, cancer cells activate hypoxia inducible factor-1 (HIF-1) to use glycolysis even at normal oxygen levels, or normoxia. Therefore, HIF-1 is an attractive target in cancer therapy. However, the regulation of HIF-1 during normoxia is not well characterised, although Mint3 was recently found to activate HIF-1 in cancer cells and macrophages by suppressing the HIF-1 inhibitor, factor inhibiting HIF-1 (FIH-1). In this study, we analysed Mint3-binding proteins to investigate the mechanism by which Mint3 regulates HIF-1. Yeast two-hybrid screening using Mint3 as bait identified N-terminal EF-hand calcium binding protein 3 (NECAB3) as a novel factor regulating HIF-1 activity via Mint3. NECAB3 bound to the phosphotyrosine-binding domain of Mint3, formed a ternary complex with Mint3 and FIH-1, and co-localised with Mint3 at the Golgi apparatus. Depletion of NECAB3 decreased the expression of HIF-1 target genes and reduced glycolysis in normoxic cancer cells. NECAB3 mutants that binds Mint3 but lacks an intact monooxygenase domain also inhibited HIF-1 activation. Inhibition of NECAB3 in cancer cells by either expressing shRNAs or generating a dominant negative mutant reduced tumourigenicity. Taken together, the data indicate that NECAB3 is a promising new target for cancer therapy.

PDK1, and LDHA 4,8 . However, the mechanisms that drive HIF-1 activation during normoxia are largely unclear, especially in terms of how HIF-1 is released from inhibition by prolyl hydroxylase or FIH-1.
Mint3 (also known as APBA3), a recently identified regulator of FIH-1, provides a clue 9 . Mint3 belongs to the X11 family of proteins, each of which contains a phosphotyrosine binding domain and two PDZ domains at the C-terminus 10,11 . Mint3 localises to the Golgi apparatus by interacting with membrane proteins such as amyloid precursor protein and furin 12,13 . The N-terminal region of Mint3 binds to and inhibits the enzymatic activity of FIH-1, resulting in HIF-1 activation even during normoxia without affecting HIF-1α levels 9 . Accordingly, Mint3 depletion in cancer cells reduces expression of HIF-1 target genes, glycolysis, and tumourigenicity 14,15 .
In this study, we surveyed Mint3-binding proteins by yeast two-hybrid screening, and identified N-terminal EF-hand calcium binding protein 3 (NECAB3) as a regulator of the MT1-MMP/Mint3 axis. Importantly, inhibition of NECAB3 suppresses normoxic glycolysis in cancer cells and reduces tumourigenicity.

NECAB3 binds Mint3.
Yeast two-hybrid screening using Mint3 as a bait identified eight potential Mint3binding proteins, including NECAB3 (Supplementary Table S1). We characterised NECAB3 further, as it has been reported to bind APBA2, which also belongs to the Mint family, and to localise it like Mint3 to the Golgi apparatus 26,27 . To confirm interaction between Mint3 and NECAB3, FLAG-tagged NECAB3 was co-expressed with Myc-tagged Mint3, FIH-1, or MT1-MMP in mammalian 293FT cells. Mint3 was specifically co-precipitated with NECAB3 using beads coated with anti-FLAG (Fig. 1A). Conversely, FLAG-tagged Mint3 also co-precipitated V5-tagged NECAB3 (Fig. 1B). Interaction between endogenous Mint3 and NECAB3 was also confirmed by immunoprecipitation using human fibrosarcoma HT1080 lysates (Fig. 1C). Subsequently, FLAG-tagged NECAB3 was stably expressed in HT1080 cells, and was found to aggregate in the perinuclear region with Mint3 and GM130, a Golgi marker protein (Fig. 1D). Of note, we found that the staining pattern for Mint3 was quite distinct from those of the endoplasmic reticulum marker calnexin and the endosome marker EEA1, with which FLAGtagged NECAB3 also partially co-localised (Fig. 1D). MT1-MMP, which co-localises with Mint3 at the Golgi apparatus 18 , also co-localised with FLAG-tagged NECAB3 (Fig. 1D). Taken together, the results indicate that NECAB3 binds and co-localises with Mint3 at the Golgi apparatus.
HIF-1 activation by Mint3 promotes glycolysis in cancer cells, even at normal oxygen levels 14 . Thus, glucose consumption and lactate production due to glycolysis were analysed in HT1080 cells from which NECAB3 had been knocked down. As expected, both glucose consumption and lactate production decreased significantly in NECAB3-depleted cells (Fig. 2F,G). Similar results were obtained in human squamous cell carcinoma A431 cells and human lung adenocarcinoma A549 cells (Fig. 2H,I). These results indicate that NECAB3 promotes glycolysis during normoxia in various cancer cells that exhibit the Warburg effect.
NECAB3 forms a ternary complex with Mint3 and FIH-1. To map Mint3-binding sites, truncation mutants of NECAB3 were expressed in 293FT cells. NECAB3 contains an EF-hand domain at the N-terminus  and an antibiotic biosynthesis monooxygenase domain at the C-terminus (Fig. 5A). Immunoprecipitation assays revealed that a fragment consisting of amino acids 181-190 is critical for binding Mint3, but not the EF-hand or the monooxygenase domains (Fig. 5B, C2 and C3). The homologous monooxygenase domain in Streptomyces synthesises several antibiotics via a conserved, catalytic histidine 28 . However, a mutated form of NECAB3, in which the catalytic His 324 in the monooxygenase domain is replaced with Ala, bound Mint3 (Fig. 5B, H/A). Hence, the data indicate that Mint3 binding depends neither on sequences in the monooxygenase domain nor on enzymatic activity.
Conversely, FLAG-tagged truncation mutants of Mint3 ( Fig. 5C) were expressed in 293FT cells and subjected to immunoprecipitation with V5-tagged NECAB3. The N-terminus of Mint3, which binds to FIH-1 9 , did not bind to NECAB3, in contrast to the C-terminus (Fig. 5D, N and C). As noted, the C-terminus of Mint3 consists of a phosphotyrosine-binding domain and two PDZ domains (Fig. 5C). These domains were individually deleted, and only deletion of the phosphotyrosine-binding domain abolished binding (Fig. 5D, dPTB). On the other hand, the phosphotyrosine-binding domain by itself bound to NECAB3 (Fig. 5D, PTB), suggesting that it is necessary and sufficient for binding.
Unlike NECAB3, FIH-1 binds to the N-terminus of Mint3 9 , and we examined whether NECAB3 binds to FIH-1 indirectly via Mint3. Indeed, NECAB3 was immunoprecipitated with FIH-1 only when Mint3 was expressed in 293FT cells (Fig. 5E), suggesting that NECAB3 formed a ternary complex with Mint3 and FIH-1. NECAB3 mutants that bind Mint3 but lack an intact monooxygenase domain are dominant negative. As NECAB3 depletion suppressed HIF-1 activity and glycolysis during normoxia in HT1080 cells ( Fig. 2E-G), we tested whether exogenous expression of NECAB3 or its mutants boosts HIF-1 activity. V5-tagged wild type or mutant NECAB3 was stably expressed in parental HT1080 cells (Fig. 6A), and real-time RT-PCR indicated that exogenous wild type NECAB3 did not stimulate expression of HIF-1 target genes (Fig. 6B, WT). Remarkably, NECAB3 mutants that bind Mint3 but lack an intact monooxygenase domain specifically suppressed expression of HIF-1 target genes (Fig. 6B, N1 and H/A). Accordingly, glucose consumption and lactate production diminished (Fig. 6C,D). In addition, these mutants did not further reduce glycolysis activity in NECAB3-depleted HT1080 cells (Supplementary Fig. S3). Thus, we hypothesised that these mutants are dominant negative against endogenous NECAB3. Indeed, immunoprecipitation experiments demonstrated that these mutants destabilised the interaction between Mint3 and FIH-1 (Fig. 6E) to a similar extent as NECAB3 depletion (Fig. 4A), and in a manner that also depended on Mint3 (Supplementary Fig. S4). These results suggest that NECAB3 mutants that bind Mint3 but lack an intact monooxygenase domain inhibit endogenous NECAB3. NECAB3 depletion reduces tumourigenicity. Next, we examined whether NECAB3 depletion reduces tumourigenicity. We first verified that NECAB3 depletion did not affect cell growth in vitro under normoxic and hypoxic conditions (Fig. 7A,B), as observed in Mint3 depletion 14,15 . Control (shLacZ) and NECAB3-depleted (shNECAB3) HT1080 cells were then grafted into immunodeficient mice. NECAB3 depletion decreased subcutaneous tumour growth to approximately 40-50% of control cells (Fig. 7C). Similarly, NECAB3 depletion decreased tumour growth from A431 and A549 cells (Fig. 7D,E) to a comparable extent as Mint3 depletion ( Supplementary  Fig. S5). Exogenous expression of NECAB3 dominant negative mutants also decreased tumour growth to approximately 50% (Fig. 7F, N1 and H  increased in five tumours (urothelial bladder cancer, chromophobe renal cell carcinoma, papillary kidney carcinoma, liver hepatocellular carcinoma, and prostate adenocarcinoma), but significantly decreased in four (oesophageal cancer, lung adenocarcinoma, lung squamous cell carcinoma, and phaeochromocytoma and paraganglioma) (Fig. 8A). Mint3 was modestly induced in some tumours (Fig. 8B), while MT1-MMP was elevated in many (Fig. 8C). Interestingly, NECAB3, Mint3, and MT1-MMP were induced in three tumours (urothelial bladder cancer, papillary kidney carcinoma, liver hepatocellular carcinoma), implying that NECAB3-mediated HIF-1 activation might play some role in these cancers.

Discussion
Cancer cells activate HIF-1 to promote glycolysis during normoxia through mechanisms that are not fully understood. Our results suggest that NECAB3, a novel Mint3-binding protein, activates HIF-1 to promote normoxic glycolysis and tumourigenicity. NECAB3 did not directly bind to FIH-1, an inhibitor of HIF-1, but formed a ternary complex with Mint3 and FIH-1. NECAB3 depletion prevented formation of the complex between Mint3 and FIH-1, suppressed expression of HIF-1 target genes and glycolysis during normoxia, and reduced tumourigenicity. Though exogenous expression of NECAB3 did not further enhance HIF-1 activity, NECAB3 mutants that binds Mint3 but lacks an intact monooxygenase domain had a dominant negative effect on endogenous NECAB3 (Fig. 6A-E). These results are summarised in Fig. 8D.
Notably, an intact monooxygenase domain is necessary to promote glycolysis during normoxia ( Supplementary Fig. S3A-C). Thus, NECAB3 monooxygenase activity is also a possible target to control MT1-MMP/Mint3-induced activation of HIF-1 in cancer cells. We anticipate that further studies will identify the direct target(s) of oxygenation by NECAB3. On the other hand, we previously demonstrated that mTOR complex 1 phosphorylates Mint3 and thereby stabilises the interaction between Mint3 and FIH-1 15 . However, NECAB3 depletion did not affect the phosphorylation state of Mint3 ( Supplementary Fig. S6), indicating that NECAB3 supports the interaction between Mint3 and FIH-1 independently of mTOR complex 1.
There are three known NECAB proteins. NECAB1 and 2 are expressed predominantly in the brain, whereas NECAB3 is expressed not only in the brain but also in the heart, muscle, and pancreas 26,29 . These proteins are thought to be involved in amyloid precursor protein metabolism 26,30 , but their role in cancer cells has not been investigated. Using RT-PCR, we found NECAB2 and NECAB3, but not NECAB1, to be expressed in cancer cells (Supplementary Fig. S7). However, knockdown of NECAB3 alone was sufficient to inhibit normoxic glycolysis and reduce tumourigenicity. Thus, NECAB3 seems to function distinctly from other NECABs, at least in cancer cells.
NECAB3 was previously identified as a binding partner of APBA2 (also known as X11L) 26 . Like Mint3, APBA2 is a member of the X11 protein family, and contains one PTB domain and two PDZ domains at the C-terminus. However, APBA2 binds NECAB3 via the N-terminal region (1-367 AA) 26 while Mint3 binds to NECAB3 via the phosphotyrosine-binding domain (Fig. 5D). In addition, APBA2 is predominantly expressed in the brain, while Mint3 is ubiquitously expressed 10 . Thus, it is unclear whether APBA2 controls NECAB3 function, but such regulation might be restricted to the brain, if at all.
HIF-1 is precisely and specifically regulated in cells, because it controls various cellular activities, including the expression of a wide array of genes. Thus, HIF-1α is regulated by various post-transcriptional modifications such as hydroxylation, acetylation, and sumoylation 7,31 , and cancer cells seem to co-opt these mechanisms to promote cell survival and proliferation. Mint3-dependent activation of HIF-1 is additionally regulated, so far as known, by MT1-MMP, by the AKT/mTOR signalling pathway, and, now, by NECAB3 14,15 . However, we identified seven other potential Mint3-binding proteins by yeast two-hybrid screens (Supplementary Table S1), and further characterisation of these candidates may illuminate the precise regulatory mechanism that drives HIF-1 activation at normal oxygen levels.
In summary, we demonstrated that NECAB3, a novel Mint3-binding protein, plays an essential role in normoxic HIF-1 activation by MT1-MMP/Mint3. Hence, further studies on the interaction between the MT1-MMP/ Mint3 axis and NECAB3 may uncover the molecular basis of the Warburg effect. Finally, targeting NECAB3, especially its monooxygenase activity, may prove useful as a cancer therapeutic strategy.

Methods
Yeast two-hybrid screens. Yeast two-hybrid screens against a randomly primed cDNA library from human placenta were performed by Hybrigenics (Paris, France), using Mint3 as a bait. Protein-protein interactions were individually assigned a statistical confidence score as described previously 32 . Cell culture. HT1080 human fibrosarcoma, A431 human epidermoid carcinoma, and A549 human lung carcinoma cells were purchased from the American Type Culture Collection (Manassas, VA, USA). 293FT cells, which are derived from HEK293 human embryonic kidney cells and express the simian virus large T antigen, were purchased from Life Technologies (Carlsbad, CA). Cells were cultured at 37 °C and humidified 5% CO 2 in DMEM (HT1080, A431, and A549) or high-glucose DMEM (293FT) containing 10% foetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin (Sigma-Aldrich, St Louis, MO, USA). For experiments in hypoxic conditions, cells were cultured with 1% O 2 and 5% CO 2 in a model 9200 incubator (Wakenyaku, Tokyo, Japan).
Cell proliferation. Cells (1 × 10 4 ) were seeded in a plastic dish, cultured at 37 °C in a humidified CO 2 incubator, and counted periodically using a Coulter counter (Beckman, Fullerton, CA).
Immunoprecipitation. Immunoprecipitation was performed as previously described 9,15 . Briefly, 293FT cells were co-transfected using Lipofectamine ™ 2000 (Life Technologies) with expression plasmids encoding a V5-tagged construct, a FLAG-tagged construct, and a Myc-tagged construct. Cells were lysed in lysis buffer 24 h after transfection, and centrifuged at 20,000 × g for 15 min at 4 °C. Supernatants were collected and incubated with beads conjugated to anti-FLAG M2 (Sigma-Aldrich). Beads were washed, and bound proteins were eluted with FLAG peptide, and analysed by immunoblotting. To detect interaction between FIH-1 and Mint3, HT1080 cells were lysed with lysis buffer, and centrifuged at 20,000 × g for 15 min at 4 °C. Supernatants were collected and incubated overnight at 4 °C with rabbit anti-FIH-1 polyclonal antibody (Novus Biologicals, Littleton, CO, USA) or control rabbit IgG (Sigma-Aldrich). Lysates were then incubated with protein G-Sepharose (GE Healthcare, Little Chalfont, UK) for 30 min at 4 °C. Beads were washed four times with lysis buffer, and proteins were eluted with Laemmli sample buffer, and analysed by immunoblotting.
Tumourigenicity. Experimental protocols were approved by the the Animal Care and Use Committees of The Institute of Medical Science, University of Tokyo (permit number: PA13-115), and all experiments were conducted according to the institutional ethical guidelines for animal experiments and the safety guidelines for gene manipulation experiments by The Institute of Medical Science, University of Tokyo. Briefly, 1 × 10 6 (HT1080 and A431) or 5 × 10 5 (A549) cells were injected subcutaneously into 6-week-old female BALB/c nude mice (CLEA Japan, Tokyo, Japan). Dimensions of the resulting tumours were measured with a calliper, and volumes were calculated according to V = (L × W 2 )/2, where V is the volume (mm 3 ), L is the largest diameter (mm), and W is the smallest diameter (mm).
Gene expression in human tumour and normal tissues. Expression of NECAB3, Mint3, and MT1-MMP in human tumour and normal tissues was compared using TCGA GDAC Firehose standard data version "2015_11_01 stddata Run". Data were downloaded from the Broad TCGA GDAC web site (http://gdac. broadinstitute.org/), and cancer types with less than three samples were excluded from analysis. mRNA levels were estimated by RSEM 34 .
Statistical analysis. Groups were compared pairwise using two-sided t-test or Mann-Whitney U-test.