Kinome-Wide Functional Genomics Screen Reveals a Novel Mechanism of TNFα-Induced Nuclear Accumulation of the HIF-1α Transcription Factor in Cancer Cells

Hypoxia-inducible factor-1 (HIF-1) and its most important subunit, HIF-1α, plays a central role in tumor progression by regulating genes involved in cancer cell survival, proliferation and metastasis. HIF-1α activity is associated with nuclear accumulation of the transcription factor and regulated by several mechanisms including modulation of protein stability and degradation. Among recent advances are the discoveries that inflammation-induced cytokines and growth factors affect protein accumulation of HIF-1α under normoxia conditions. TNFα, a major pro-inflammatory cytokine that promotes tumorigenesis is known as a stimulator of HIF-1α activity. To improve our understanding of TNFα-mediated regulation of HIF-1α nuclear accumulation we screened a kinase-specific siRNA library using a cell imaging–based HIF-1α-eGFP chimera reporter assay. Interestingly, this systematic analysis determined that depletion of kinases involved in conventional TNFα signaling (IKK/NFκB and JNK pathways) has no detrimental effect on HIF-1α accumulation. On the other hand, depletion of PRKAR2B, ADCK2, TRPM7, and TRIB2 significantly decreases the effect of TNFα on HIF-1α stability in osteosarcoma and prostate cancer cell lines. These newly discovered regulators conveyed their activity through a non-conventional RELB-depended NFκB signaling pathway and regulation of superoxide activity. Taken together our data allow us to conclude that TNFα uses a distinct and complex signaling mechanism to induce accumulation of HIF-1α in cancer cells. In summary, our results illuminate a novel mechanism through which cancer initiation and progression may be promoted by inflammatory cytokines, highlighting new potential avenues for fighting this disease.


Introduction
Inflammation is a primary defense process against various extracellular stimuli, such as viruses, pathogens, foods, and environmental pollutants. Several studies have shown that tumorigenesis in many cancers is closely associated with chronic inflammation. Abnormal cellular alterations that accompany chronic inflammation such as oxidative stress, gene mutation, epigenetic change, and inflammatory cytokine release are shared with carcinogenic processes, which form a critical cross-link between chronic inflammation and carcinogenesis. Almost 25% of cancers are reported to occur through chronic inflammationrelated processes [1,2]. The pro-inflammatory regulators such as TNFa and other cytokines and their receptor networks seem to play crucial functions in tumorigenesis [3].
Hypoxia-inducible factor-1 (HIF-1) and its most important subunit, HIF-1a, plays a central role in tumor progression by regulating genes involved in cancer cell survival, proliferation and metastasis [4]. HIF-1 is a major component of the oxygen sensing system that governs cellular responses to decreased oxygen availability. The hypoxia inducible transcription factor HIF-1 is a heterodimer composed of the helix-loop-helix-Per-Arnt-Sim (bHLH-PAS) proteins HIF-1a and the aryl hydrocarbon nuclear translocator (ARNT) also known as HIF-1b. Transactivation of HIF-1 transmits a hypoxic signal into a multitude of pathophysiological responses by regulation of numerous target genes [4,5].
In addition to hypoxia, more recent evidence suggest that HIF-1 can be accumulated and activated during normoxia by growth factors, cytokines and other factors associated with inflammation [5]. Several reports have indicated an important role of TNFa in regulation of HIF-1a stability and activity [6][7][8]. However, details of HIF-1 regulation by TNFa remain unclear.
Here, we describe signaling mechanisms that incite HIF-1a accumulation in response to TNFa. To improve our understanding of HIF-1 regulation by the cytokine, we screened a kinasespecific small interference RNA (siRNA) library using a HIF-1a-eGFP chimera reporter assay under TNFa treatment. This screen determined that depletion of ADCK2, PRKAR2B, TRIB2 and TRPM7 most significantly downregulates nuclear accumulation of HIF-1a in response to the treatment of osteosarcoma cells. Furthermore, our results suggest that this pathway is also present in prostate cancer cells. Surprisingly, the mechanism of regulation of TNFa-elicited HIF-1a accumulation was associated with a nonconventional NFkB signaling pathway and alleviation of superoxide activity. Taken together our data allow us to conclude that TNFa uses a distinct and complex signaling mechanism to induce accumulation of HIF-1a.

Results
TNFa is a major inflammatory cytokine reported to be a potent inducer of HIF-1a nuclear accumulation [5][6][7][8]. We examined several cancer cell lines for HIF-1a accumulation under TNFa treatment. In our experiments, TNFa produced a significant increase in nuclear accumulation of HIF-1a in several cancel cell lines (Fig. 1a). Similarly, TNFa induced nuclear buildup of a HIF-1a-eGFP chimera protein (Fig. S1a, Fig. 1b,c) in the HIF-1a_U2OS Redistribution assay based on an osteosarcoma cell line. The observed effect was concentration-and time-dependent (Fig. 1b,c). 24 hr incubation with TNFa at 10 ng/mL was selected for all screening experiments to provide an appropriate window to study up-and down-regulation of HIF-1a accumulation.
There are two receptors described for TNFa, namely TNF receptor 1 (TNFR1, p55 receptor) and TNF receptor 2 (TNFR2, p75 receptor). TNFR1 is ubiquitously expressed while TNFR2 is mainly expressed in immune cells [9]. Although both receptors bind TNFa, the main receptor mediating cellular effects in most cell types is TNFR1. In our experiments, knockdown of the TNFR1 effectively diminished TNFa-dependent nuclear accumulation of HIF-1a ( Fig S1b). TNFa is known to activate multiple pathways downstream of TNFR1 [9]. To explore the role of kinases in regulating HIF-1a accumulation under TNFa treatment, we depleted kinases in HIF-1a-U2OS cells using a siRNA kinase library targeting 788 kinases and then analyzed cellular accumulation of HIF-1a-eGFP after incubation with TNFa ( Fig. 2a). To minimize siRNA off-target activity we used ON-TARGETplus version of the human kinases collection of SMARTpool siRNA reagents [10]. The same kinase library was screened in a control cell line that expresses only eGFP to subtract possible non-specific effects. HIF-1a-eGFP screening data were subjected to Student t-test p-value analysis, Benjamini-Hochberg multiple comparisons correction [11], and performance ranking followed by comparison between two independent screening experiments with a 1.5 fold change threshold. Resulting data was further compared with data from the counter-screen with cells expressing eGFP only (1.2 fold change threshold for eGFP only, Fig S2) and overlapping hits dismissed. Among the 788 kinases screened by siRNA-mediated silencing, depletion of 77 genes increased HIF-1a-eGFP accumulation above 2 fold (Table S1) and depletion of another seven target genes decreased accumulation under TNFa treatment (Fig. 2a). The genes demonstrating a siRNA-mediated decrease of HIF-1a accumulation were of particular interest because these could potentially represent members of TNFa signaling pathways. These include PRKAR2B, ADCK2, TRPM7, RIOK2, TRIO, ADRA1B and TRIB2. To confirm that the decrease in TNFa-induced HIF-1a accumulation in siRNA-transfected cells was directly related to depletion of selected targets we repeated this experiment using newly synthesized siRNA pools (Fig. 2b). Only one out of seven selected hit candidates was not confirmed: siRNA targeting ADRA1B (data not shown), which was subsequently omitted from further analysis.
A High Content Analysis approach allows simultaneous acquisition of multiple data streams from the same set of samples. We utilized this approach to further analyze the screening data. Collected data (cell number per field) suggested that none of the selected siRNAs (PRKAR2B, ADCK2, TRPM7, RIOK2, TRIO, and TRIB2) produced any effect on cell viability (Fig S3). In addition to the control of protein stability, HIF-1a function can be regulated by processes that influence its subcellular localization, e.g. cytoplasmic vs. nuclear. Simultaneous measurement of HIF-1a accumulation in the nuclei and cytoplasm revealed that none of the siRNA targets selected for a decrease in nuclear accumulation produced an increase in cytoplasmic retention of HIF-1a (data not shown).
To examine if selected candidate hits can influence TNFamediated accumulation of HIF-1a in other cell types we determined HIF-1a nuclear buildup in prostate cancer cell lines. HIF-1a accumulation in LNCaP and DU145 cells was found to be sensitive to TNFa while PC3, a prostate cell line with significant invasive potential [12] , demonstrated no such sensitivity (Fig. 1a,). Depletion of PRKAR2B, ADCK2, TRPM7 and TRIB2 significantly decreased HIF-1a accumulation in LNCaP, an androgen-sensitive human prostate adenocarcinoma cell line with low invasive potential (Fig. 2c). Data similar to U2OS and LNCaP were also obtained from MCF10a, a non-tumorigenic mammary epithelial cell line (Fig S4a). In DU145 cells, a prostate cancer cell line with moderate invasive potential, only TRIB2 depletion was effective in abrogating TNFa-induced HIF-1a accumulation ( Fig S4b). Based on the above results PRKAR2B, ADCK2, TRPM7 and TRIB2 were selected for further analysis. siRNA pools targeting these genes produced concentration-dependent effects on TNFa-stimulated HIF-1a accumulation in U2OS osteosarcoma cells (Fig. 3a) In all of our experiments we employed strategies that are known to diminish siRNA off-target effects: application of chemically modified siRNA molecules and usage of a siRNA pooling strategy. To further confirm that the decrease in TNFa-induced HIF-1a accumulation in siRNA-transfected cells was directly related to ontarget effects of the siRNA, we repeated this experiment using newly synthesized siRNA pools and four separate siRNA duplexes (that comprise each pool) to deplete all four cellular targets. These experiments produced results similar to the screening data -for all four targets siRNA pools and four individual siRNA duplexes produced significant decrease in HIF-1a accumulation (.2 fold, Fig. 3b). Effectiveness of target gene knockdown for selected siRNA hits was determined using Q-PCR and Solaris probes. Target gene depletion correlates with the HIF-1a phenotypical assay -for all four targets siRNA pools used in the screening campaign and four individual siRNA duplexes demonstrated potent knockdown (.60%, Fig. 3c). TRPM7 is poorly expressed in U2OS cells and application of any RNAi reagent effectively eliminated expression of this target to an undetectable level (Fig. 3c).
TNFa is known to regulate expression of proteins within its own signaling cascades. We found that expression of ADCK2 and TRIB2 mRNA is regulated by TNFa in U2OS cancer cells ( Fig  S5). No statistically significant changes were detected for PRKAR2B and TRPM7.
Recent reports indicate that HIF-1a stability and activity may be regulated through oxidative stress-sensitive pathways [6,13,14]. Such pathways are also well known regulators of TNFa signaling [6,15]. To examine a possible role of oxidative stress mechanisms in TNFa-stimulated HIF-1a accumulation we investigated the effects of exogenous hydrogen peroxide. While hydrogen peroxide alone produced a significant increase in HIF-1a accumulation, an opposite effect was observed on TNFa-pretreated cells (Fig. 4a). This result suggests possible negative regulation of HIF-1a accumulation by superoxide, a main source of intracellular peroxide [16]. We hypothesized that newly discovered positive regulators of HIF-1a accumulation may control either superoxide production or a conversion to peroxide. In U2OS cells, TNFa robustly increased expression of MnSOD, one of the major superoxide/peroxide conversion enzymes ( Fig S6). However, depletion of the identified positive regulators of HIF-1a accumulation had no effect on expression of MnSOD ( Fig S6).
We observed that superoxide scavengers Tiron and TEMPOL are able to rescue HIF-1a accumulation in TNFa-treated cells transfected with siRNA against ADCK2, and TRIB2 when applied in a concentration that does not significantly affect control cells (Fig. 4b). Cells transfected with PRKAR2B and TRPM7 siRNA were unaffected by superoxide scavengers (Fig. 4b). Taken together our data suggest that TNFa mediates HIF-1a accumulation through a mechanism that mitigates superoxide production, and ADCK2, PRKAR2B and TRIB2 are positive regulators of this mechanism.
Multiple pathways and factors are reported as regulators of HIF-1a activity in other experimental systems and conditions, including conventional NFkB, JNK, STAT3, and proteasome activity [5]. In addition, TNFa is known to induce signaling through the conventional NFkB pathway [9,17]. Analysis of our results revealed that depletion of kinases that are necessary for these pathways produced no negative impact on TNFa-mediated HIF-1a accumulation. In our experiments, TNFa produced no significant effect on the STAT3-dependent pathway ( Fig S7a). We hypothesized that because TNFa induces HIF-1a accumulation, it may also inhibit proteasome activity. To this end we tested TNFa as a possible agonist in the U2OS_E6-AP: p53 degradation and U2OS_SCF-Skp2 E3: p27 degradation Redistribution assays. Proteasome activity inhibitor MG132 induced accumulation of eGFP chimeras in both assays while incubation with TNFa had no effect (Fig S7b,c).
Data from our screening experiments suggest that depletion of upstream regulators of the JNK pathway results in an increase of HIF-1a accumulation in response to TNFa treatment (Table S1). Depletion of the upstream regulators of the conventional NFkB pathway CHUK, IKBKB or IKBKE do not modulate HIF-1a accumulation in response to TNFa treatment ( Fig S8). Moreover, siRNA-mediated depletion revealed that NFkB proteins RELA and NFkB2 may act as negative regulators of such accumulation because their knockdown produced sharp increase in HIF-1a accumulation (Fig. 5a). In contrast, NFkB proteins RELB, cREL and NFkB1 appear to be necessary for TNFa-induced HIF-1a accumulation because depletion of corresponding genes produced strong negative effects on accumulation (Fig. 5a). Activation of NFkB proteins correlates with their intracellular translocation [17]. We found that in U2OS osteosarcoma cells, TNFa stimulates translocation of RELB and cREL between the nucleus and cytoplasm, with RELB being excluded from the nucleus and cREL accumulating in the nucleus. Similar to depletion of TNFR1, depletion of TRIB2 and TRPM7 prevented nuclear exclusion of RELB (Fig. 5b). cREL translocation was not affected by depletion of selected targets (data not shown). Furthermore, the effect of RELB depletion was attenuated by superoxide scavengers Tiron and TEMPOL (Fig. 5c). Taken together our results suggest that TNFa-mediated HIF-1a accumulation may be at least partially governed by a non-conventional NFkB signaling pathway activated by TRIB2 and TRPM7.

Discussion
TNFa is well known to evoke multiple signaling mechanisms where various kinases play irreplaceable roles. Recent advances in functional genomics and cell imaging techniques allowed us to perform systematic investigation of possible mechanisms of TNFamediated HIF-1a accumulation. To explore the role of kinases in regulating HIF-1 activity under treatment with TNF-a, we depleted kinases in U2OS osteosarcoma cells using a chemically modified siRNA kinase library targeting 778 kinases and then analyzed nuclear accumulation of HIF-1a-eGFP chimera constitutively expressed in these cells. In this assay, TNFa strongly increased HIF-1a-eGFP protein accumulation (Fig. 1b,c).
Under normal physiological conditions HIF-1a accumulation is heavily repressed by several regulatory pathways. Kinases and related proteins are well known to play an important role in such pathways. Thus, we expected that the majority of siRNA hit candidates would provide a release from the repression of HIF-1a accumulation. Indeed, among the 788 kinases screened by siRNAmediated silencing, depletion of 6 kinases significantly decreased HIF-1a accumulation and depletion of another 89 kinases increased HIF-1 activity in cells treated with TNFa (Fig. 2a,b and Table S1).
To some extent, our data recapitulates previous findings regarding negative regulators of HIF-1a activity [18]. It was reported that SMG-1 suppresses HIF-1 activity under hypoxic conditions and that siRNA-mediated depletion of the gene product significantly increases activity of a HIF-1a-sensitive reporter [18]. Results of our screening experiments indicate that depletion of SMG-1 specifically up-regulates TNFa-induced HIF-1a accumulation (data not shown).
Several pathways were found to be significantly over-represented in the group of 77 negative regulators: 16 (Table  S1). These data suggest that such pathways may oppose TNFa signaling and inhibit HIF-1a accumulation.
Depletion of several target genes inhibit TNFa-mediated HIF-1a accumulation (Fig. 2). These genes may represent one or more TNFa signaling mechanisms, and are of particular interest. In our screening campaign we identified six targets of this kind (Fig. 2b). Furthermore, our results suggest that four of these genes -ADCK2, PRKAR2B, TRIB2 and TRPM7 -seem to regulate HIF-1a accumulation in multiple cancer cell lines (Fig. 2, Fig S4). All four genes were previously described in connection with regulation of cancer cell proliferation and motility but existing data did not suggest their participation in TNFa signaling [19][20][21][22][23]. The functions of ADCK2 protein are not yet clear. It is not known if it has protein kinase activity and what type of substrate it would phosphorylate. Several reports established a connection of ADCK2 to cancer cell proliferation and motility [20]. PRKAR2B encodes the cAMP-dependent protein kinase type II-beta regulatory subunit. The cAMP-dependent protein kinase A (PKA) is a ubiquitous serine/threonine protein kinase. PKA is accepted as a major mediator of intracellular cAMP signals in eukaryotes. To date, a large number of cytoplasmic and a few nuclear PKA substrates have been reported [23]. Interestingly, depletion of PRKAA1 (the catalytic subunit of the PKA) also produced a decrease in HIF-1a accumulation although to a lower extent than PRKAR2B depletion (data not shown). Further studies are necessary to clarify the exact role of PKA in TNFa-stimulated HIF-1a accumulation. Although the molecular function of TRIB2 (Tribbles homolog 2) is still unclear, it has been identified as a potential driver of lung tumorigenesis and a myeloid oncogene [21,22]. TRPM7 is a ubiquitously expressed and constitutively active divalent cation channel. It provides a mechanism for Mg2+ entry and thus it is essential for cell survival and proliferation [19,24].
Key regulators of TNFa signaling pathways are reactive oxygen species (ROS; e.g., superoxide, hydrogen peroxide, and hydroxyl radical) [6,15]. ROS have been suggested to modulate TNFa signaling, providing both positive and negative regulation of the NFkB system downstream of TNFR1 depending on the experimental system and conditions [6,25]. Our results imply that hydrogen peroxide suppresses TNFa-mediated HIF-1a accumulation (Fig. 4a). These data suggest that the source of intracellular hydrogen peroxide, superoxide anion may inhibit TNFa-mediated HIF-1a accumulation as well. We hypothesized that the newly described regulators of the accumulation may elicit their effect through modulation of superoxide production. Indeed, alleviation of superoxide anion activity rescues HIF-1a accumulation on the background of depletion of ADCK2 and TRIB2 (Fig. 4b). All these results suggest that TNFa-induced HIF-1a accumulation may be regulated by a superoxide sensitive pathway and that the above three proteins may be involved in negative regulation of superoxide production (Fig. 6). It is well established that conventional and non-conventional NFkB signaling cascades are major mechanisms that convey effects of TNFa on intra-cellular physiology [26]. Our screening results suggest that depletion of positive regulators upstream of conventional NFkB -IKK-a (CHUK), IKK-b (IKKB) or IKK-e (IKBKE) -produced no negative impact on HIF-1a nuclear accumulation ( Fig S8).
Also, we found that depletion of RELA and NFKB2 results in a significant upregulation of HIF-1a accumulation while depletion of RELB, cREL and NFKB1 produces a decrease in HIF-1a accumulation. Such a decrease can be rescued by mitigation of superoxide anion activity (Fig. 5). Furthermore, depletion TRIB2 and TRPM7 was found to prevent intracellular translocation of RELB upon treatment with TNFa. These findings allowed us to speculate that TNFa may regulate HIF-1a accumulation through both conventional and non-conventional NFkB pathways. The actual amount of accumulated HIF-1a will then depend on a balance between different TNFa-induced NFkB pathways. The proposed model seems to be in line with the known complexity of inflammation-cancer relationships [3].
Taken together our results suggest that TNFa-induced HIF-1a buildup is regulated by a several pathways (Fig. 6). At least in part, TNFa may convey its effect through TNFR1 receptor signaling leading to a non-conventional NFkB-dependent mechanism that negatively regulates production of reactive oxygen species. This mechanism appears to be controlled by ADCK2 and TRIB2. TRPM7 appears to stimulate RELB translocation, but its depletion phenotype can not be rescued by alleviation of superoxide activity. Thus TRPM7 may represent an independent pathway of regulation of HIF-1a accumulation. Further studies are necessary to understand the greater complexity of TNFadependent stimulation of HIF-1a nuclear accumulation and its role in tumorigenesis and tumor progression.
HIF-1a_U2OS Redistribution assay was used in screening campaign to monitor HIF-1a nuclear accumulation: recombinant U2OS cells stably expressing human HIF-1a (NM_001530) fused to the C-terminus of enhanced green fluorescent protein (eGFP). U2OS cells are adherent epithelial cells derived from human osteosarcoma. Expression of eGFP-HIF-1a is controlled by a standard CMV promoter and continuous expression is maintained by addition of G418 to the culture medium according to manufacturer protocol. U2OS stable cell line that expresses eGFP only was used as a subtraction control in screening campaign.
In addition to HIF-1a_U2OS the effect of TNFa treatment was tested in three separate Thermo Fisher Scientific Redistribution Assays: the STAT3_U2OS Redistribution Assay, the E6-AP: p53 degradation Redistribution Assay (U2OS), and the SCF-Skp2 E3 Ligase: p27 degradation Redistribution Assay (U2OS). Assays were performed according to manufacturer protocols for all reference compounds. For TNFa-treatment, each assay cell line was treated for 24 hours at 37uC with TNFa at the following concentrations: 80 ng/mL, 40 ng/mL, 20 ng/mL, 10 ng/mL, 5 ng/mL, 2.5 ng/mL, 1.25 ng/mL, and 0.63 ng/mL. Each TNFa titration was performed in quadruplicate on 96-well assay plates, and each assay plate was performed in triplicate. After TNFa treatment, plates were fixed, stained with Hoechst 33258, and imaged as described below. TNFa was purchased from R&D Systems, hydrogen peroxide, Hoechst 33258, Tiron and 4hydroxy-TEMPO (TEMPOL) were purchased from Thermo Fisher Scientific.  Our results suggest that TNFR1 receptor may convey its effect through a complex mechanism that includes a nonconventional NFkB-dependent pathway. This mechanism may negatively regulate production of reactive oxygen species and appears to be controlled by TRIB2, ADCK2 and PRKAR2B.

Cell Imaging
Imaging of harvested cells was performed using the ArrayScanH VTI HCS Reader and CellInsight TM Personal Cell Imager (Thermo Fisher Scientific). eGFP fluorescence was analyzed using the Molecular Translocation BioApplication (Thermo Fisher Scientific). Immunofluorescence was analysed using the Compartmental Analysis BioApplication (Thermo Fisher Scientific). Images were acquired using a 106 objective. Images and data were collected for three fields per well.

Gene expression analysis
The SV 96 Total RNA Isolation System (Promega, Madison, WI, catalog #Z3505) was used for total RNA purification. The Nanodrop (Thermo Fisher Scientific) was used to determine average concentration of the RNA preps. After isolation, total RNA was frozen at 280uC for storage until further use. Total RNA was thawed once for the RT step, which 5 mL of RNA was used in all cDNA reactions. Verso cDNA Synthesis Kit (Thermo Scientific, catalog #AB-1453) was used for the cDNA synthesis step. cDNA reactions were set up according to the supplier's protocol in a total of 20 mL reactions. Random hexamers and oligo-dT primers in a ratio of 3 to 1 were used in cDNA reactions. Cycling conditions were 42uC for 30 minutes then an inactivation step of 95uC for 2 minutes. No reverse transcriptase enzyme and no template controls were used for each RT run and each gene assay to determine presence of contamination, which all came up negative. Expression of ADCK2 (NM_052853), PRKAR2B (NM_002736), RIOK2 (NM_018343), TRIB2 (NM_021643, TRIO (NM_007118), and TRPM7 (NM_017672) was determined by qRT-PCR in both TNFa treated and untreated cells and transfected cells. Corresponding Thermo Scientific Solaris Human qPCR Gene Expression Assays were purchased from Thermo Scientific -ADCK2 #AX-005304-00, PRKAR2B #AX-007673-00, RIOK2 #AX-005002-00, TRIB2 #AX-005391-00, TRIO #AX-005047-00, and TRPM7 #AX-005393-00. PPIB Solaris Human qPCR Gene Expression Assay (catalog #AX-004606-00) was used to determine PPIB (NM_000942) expression for sample input normalization and analysis of relative expression of the specific genes. All Solaris assays listed do not span an exon-exon boundary or map to any pseudogenes. Either the assay probes or primers cross a splice site. cDNA was diluted 3-fold, and 3 mL of the dilution was the cDNA input for the qPCR reactions. The Thermo Scientific Solaris qPCR Gene Expression Master Mix (Thermo Scientific, catalog #AB-4350) was used for the qPCR step for a 15 mL final reaction volume, and the Ct values used for further analysis were obtained using the Roche Light Cycler 480 (Roche). All qPCR reactions were set up according to the supplier's protocol (1 cycle at 95uC for 15 minutes, then 95uC for 15 seconds followed by 60uC for 1 minute for 40 cycles). In all transfection experiments, relative gene expression was normalized to ON-TARGETplus Non-Targeting Pool transfected siRNA controls (Thermo Scientific) for each plate separately. All treatments were tested in biological triplicates. Additional information is presented in Table S2.
Gene Ontology analysis L2L (University of Washington) and DAVID (NIAID) tools were used to analyze RNAi screening data.  Figure S2 Schematic description of selection of hit candidates for positive regulation of HIF-1a accumulation. Screening data were subjected to Student t-test p-value analysis followed by Benjamini-Hochberg multiple comparisons correction, and performance ranking (top 10% selected). This analysis was followed by comparison between two independent screening experiments. Resulting data were further compared with data from the counter-screen with cells expressing eGFP only. Finally, only hits demonstrating fold change above 1.5 fold were selected for further experiments. siRNA targeting ADCK2, RIOK2, PRKAR2B, TRIB2, TRIO and TRPM7 were transfected into the U2OS osteosarcoma cell line. Cells were harvested 72 hr after transfection. Cells were treated with TNFa (10 ng/mL) for 24 hr before harvesting. All data normalized to untreated cells transfected with control siRNA NTC1. All data (Median+/2MAD) normalized to cells transfected with control siRNA NTC1. (TIF) Figure S7 Effect of selected siRNAs on nuclear translocation of STAT3, degradation of p27, and degradation of p53 in U2OS osteosarcoma cells. Cells were incubated with TNFa (10 ng/mL) for 24 hr. Nuclear translocation of STAT3 (A), degradation of p27 (B), and degradation of p53 (C) were determined as described in Materials and Methods. All data (Median+/2MAD) normalized to cells transfected with control siRNA NTC1. (TIF) Figure S8 Effect of CHUK, IKBKB and IKBKE siRNAs on HIF1a accumulation in U2OS osteosarcoma cells. siRNA targeting CHUK, IKBKB and IKBKE were transfected into U2OS osteosarcoma cells. Cells were harvested 72 hr after transfection. Cells were treated with TNFa (10 ng/mL) for 24 hr before harvesting. All data normalized to untreated cells transfected with control siRNA NTC1. All data normalized to cells transfected with control siRNA NTC1. Data (Median+/ 2MAD) are representative of two independent experiments performed in triplicate. All data normalized to TNFa-treated cells. (TIF)