Canine OS cell lines, OSA8, OSA16, OSA32, and OSA36, were provided by Dr. Jaime Modiano (University of Minnesota, Minneapolis, MN, USA) (35–37). The canine D17 OS cell line and human U2OS cell line were purchased from American Type Cell Culture Collection (ATCC, Manassas, VA, USA). All canine OS cell lines were maintained in RPMI-1640 (Gibco Life Technologies, Grand Island, NY, USA) supplemented with 10% fetal bovine serum, non-essential amino acids, sodium pyruvate, penicillin, streptomycin, l-glutamine, and HEPES (4-(2-dydroxethyl)-1-piperazineethanesulfonic acid) at 37°C, supplemented with 5% CO₂ (media supplements from Gibco). Normal canine osteoblasts (catalog no. Cn406-05) (Cell Applications Inc., San Diego, CA, USA) were cultured in canine osteoblast medium (Cell Applications Inc., catalog no. Cn417-500). The human RMS RH30 cell line was established at St. Jude Children’s Research Hospital, and characteristics have been described (38). Unless otherwise mentioned, all the cell lines used were cultured in DMEM supplemented with 10% FBS. The human OS cell lines OS-1 and OS-17 and human Ewing sarcoma cell lines EW5, ES-1, ES-2, ES-7, and ES-8 were generously provided by the Pediatric Preclinical Testing Program. These lines were maintained by serial passaging in severe combined immune deficient mice as described previously by the Pediatric Preclinical Testing Program (39). Normal canine skeletal muscle tissue collections were approved by the OSU IACUC (protocol 2010A0015). Fresh frozen canine OS tumor samples were obtained from dogs presenting to The Ohio State University Veterinary Medical Center (OSU-VMC). Tumor sample collections were performed in accordance with established OSU-VMC hospital protocols and approved by the OSU IACUC (protocol 2010A0015).

For luciferase assays, cells were transfected in 12-well plates, and luciferase activity was monitored as previously described (40). pBabe-puro was used for expressing IκBα SR. pCL-Eco was used for expressing IKKα ΔNBD. pSIREN-RetroQ was used for expressing shHK2. pWZL-Neo-Myr-Flag-HK2 was obtained from Addgene (Cambridge, MA, USA) (Plasmid 20501) (41). Retroviruses were prepared by calcium phosphate-mediated gene transfection kit (Promega, Madison, WI, USA) as previously described (42). RH30 and U20S cells were infected by applying 4–5 cfu/cell viruses in a total volume of 4 mL full culture medium containing 4 µg/mL polybrene for 8 h. Cells were transfected with 10 nM of AllStars negative control siRNA or p65-targeting siRNA (5′-CTATGATGAGTTTCCCACCATGGTG-3′) according to standard protocols using Lipofectamine 2000.

Cell extracts were made by lysing PBS-washed cell pellets in radio-immunoprecipitation assay buffer supplemented with complete Protease Inhibitor Cocktail (Roche Diagnostics, Indianapolis, IN, USA). Canine cell extracts were made by resuspending the cells in lysis buffer consisting of 20 mM Tris–HCl pH 8.0, 137 mM NaCl, 10% glycerol, 1% IPEGAL CA-630, 10 mM ethylenediaminetetraacetic acid, 1 mg mL⁻¹ aprotinin, 1 mg mL⁻¹ leupeptin, 1 mg mL⁻¹ pepstatin A, 1 mM phenylmethylsulphonyl fluoride, 1 mM sodium orthovanadate, and 10 mM sodium fluoride (all from Sigma, St. Louis, MO, USA) for 1 h at 4°C. In addition, protein lysates were prepared from frozen canine normal muscle or OS tumor samples by pulverizing tissues in liquid nitrogen and resuspending them in the same lysis buffer with proteinase and phosphatase inhibitors. For each sample, 30 µg of protein was loaded on each gel. Proteins were transferred onto a PVDF membrane using a tank blotter (Bio-Rad, Hercules, CA, USA). The membranes were then blocked using 5% milk and 1× Tris-buffered saline plus Tween 20 (TBST) and incubated with primary antibody overnight at 4°C. Membranes were then washed with 1× TBST and incubated with the corresponding secondary antibody. Membranes were again washed with 1× TBST, incubated with SuperSignal West Pico Chemiluminescent Substrate according to the manufacturer’s protocol (Thermo Fisher Scientific, Waltham, MA, USA), and visualized by autoradiography. The antibodies used included anti-p65 (Millipore, Billerica, MA, USA; Cell Signaling, Danvers, MA, USA), anti-p52 (Cell Signaling), anti-p100 (Cell Signaling), anti-IκBα (Santa Cruz Biotechnology, Dallas, TX, USA), and anti-HK2 (Cell Signaling). Anti-GAPDH (Santa Cruz Biotechnology) and anti-α-Tubulin (Sigma) were used as loading controls.

Electrophoretic mobility shift assays were performed as previously described (40). For quantitative RNA expression, total RNA was isolated from cells by Trizol extractions (Invitrogen). Two micrograms of total RNA were reversed transcribed with MultiScribe™ MuLV reverse transcriptase (Applied Biosystems, Foster City, CA, USA). cDNA equivalent to 40 ng was used for quantitative polymerase chain reaction (qPCR) amplification (Applied Biosystems) with SYBR green PCR master mix (Applied Biosystems). Samples in which no reverse transcriptase was added (no RT) were included for each RNA sample. The relative levels of expression of genes were normalized according to those of human GAPDH (5′-GGGGAGCCAAAAGGGTCATC-3′, 5′-GGCATGGACTGTGGTCATGAG-3′). qPCR data were calculated using the comparative Ct method (Applied Biosystems). Primers to the coding region of human HK2, 5′-GAATGGGAAGTGGGGTGGAG-3′ and 5′-TCTCGTCCAGGCTGTTCTGC-3′ were used for qRT-PCR analysis. Where possible, intron spanning primers were used. All qPCR reactions were performed in triplicate, and three independent RNA samples were assayed for each time point.

Chromatin immunoprecipitation assays were performed and quantified as described previously (43, 44) with the following modifications: 1 × 10⁷ cells were used for each immunoprecipitation, and protein A agarose beads (Invitrogen) were used to immunoprecipitate the antibody:antigen complexes. The assays were performed using an antibody against p65 (Millipore). Rabbit IgG (Millipore) was used as a non-specific control. Primers spanning the NF-κB binding sites on the HK2 (site 1 TCTGGCCTCCATCAGTCTCT, site 2 CCCTCCAGTAACAGGGAAAA, site 2 GTGTAGGAGACGAGCGGTTC, site 2 AGAGGCAGATCCTGAGGACA, site 3 ATGGGGGTCATCTCTCTTCC, site 3 GCACACTGCAGCGCTTAAT) gene and IgH (GCCGATCAGAACCAGAACACCTGC and TGGTGGGGCTGGACAGAGTGTTTC) were used to detect enrichment. Real-time PCR reactions were performed in triplicate. Values of [Δ][Δ]Ct were calculated using the following formula based on the comparative Ct method: [Δ]Ct, template (antibody) − [Δ]Ct, template (IgG) = [Δ][Δ]Ct. Fold enrichments were determined using the formula: 2^{−[Δ][Δ]Ct} (experimental)/2^{−[Δ][Δ]Ct} (reference, IgH). SE from the mean was calculated from replicate [Δ][Δ]Ct values. The IgH locus was used to normalize the fold enrichments for the individual promoters. All ChIP assays shown are representative of at least three individual experiments.

The metabolic profile of RMS and OS cells was determined using Seahorse XF^e24 Extracellular Flux Bioanalyzer (Agilent) to measure basal and maximal oxygen consumption rate (OCR), an indicator of mitochondrial respiration, as suggested by the manufacturer. The Seahorse Bioanalyzer provides real-time measurements of OCR as a measure of OXPHOS. In brief, cells were incubated for 1 h in DMEM supplemented with 10 mM glucose, 2 mM l-glutamine, and 1 mM sodium pyruvate. For the MitoStress test, cells were analyzed under basal conditions and after injection of three inhibitors: 1 µM oligomycin, 0.3 µM carbonyl cyanide 4-trifluoromethoxy-phenylhydrazone (FCCP), and 0.5 µM rotenone. Oligomycin inhibits ATP synthase, which allows measurement of ATP-coupled oxygen consumption through OXPHOS. FCCP is an uncoupling agent that allows maximum electron transport, therefore a measure of maximum OXPHOS capacity. Rotenone is an inhibitor of complex I of the mitochondrial respiratory chain that allows calculation of the rate of respiration solely attributable to mitochondria. ATP-linked respiration was calculated by subtracting the lowest OCR value after administrating of oligomycin from the third basal OCR measurement. All chemicals were obtained from Sigma.

For growth analysis, cells were plated in triplicate into 24-well plates, and the cells were trypsinized and counted each day. Cell growth was also determined using time-lapse live cell imaging. Cells were seeded in triplicate in 96-well plates at a density of 3,000 cells/well and placed in the IncuCyte Zoom (Essen Biosciences, Ann Arbor, MI, USA) in 37°C and 5% CO₂ conditions. Three high-definition phase-contrast images were collected from each well every hour. IncuCyte Zoom 2015A software (Essen Bioscience) was used for analysis to apply a mask that determined the percent confluence based on the phase-contrast images. The percent confluence was normalized to the initial seeding confluence. For cell death assays, cells were seeded in 96-well plates as described earlier, and Caspase-Glo 3/7 assay (Promega) was performed according to the manufacturer’s protocol. Annexin V (BD Biosciences, Franklin Lakes, NJ, USA) and propidium iodide (Sigma) were used to stain cells, and flow cytometry was used for determining the percentage of apoptotic cells.

Total RNA was isolated using a standard Trizol extraction protocol and purified with a cleanup kit (Qiagen). Affymetrix GeneChip Human Transcriptome Array 2.0 (Affymetrix) arrays were used for microarray. Probe intensity (CEL) files were obtained, and Affymetrix Transcriptome Analysis Console software was used for array analysis. Differentially expressed genes were identified by the Robust Multichip Average method (45).

Gene set enrichment analysis v2.0 was used to determine the enrichment of functional categories in the entire linear gene expression dataset (46). GSEA was performed using MSigDB C2 CP: Canonical pathways gene set collection. Statistical significance was determined using 1,000 random permutations of each gene set to obtain a nominal p-value and false discovery rate. Heat map was generated using GenePattern (47). HeatmapViewer (v11) was used to display values in heat map format for the genes identified from GSEA analysis.

The microarray data have been deposited into the GEO database with the accession number GSE112119.

Quantitative data are presented as mean ± SEM. Statistical analysis was performed using GraphPad Prism software. Statistical comparisons were performed using unpaired two-tailed Student’s t tests, with a probability value of <0.05 taken to indicate significance.

Our previous research using the murine C2C12 myoblast cell line showed that alternative NF-κB signaling is activated during myogenesis to promote OXPHOS in myotubes (3, 48). Interestingly, RMS tumors are thought to derive from a skeletal muscle lineage that improperly differentiates and exhibits defects in mitochondrial gene expression, favoring a Warburg effect (49, 50). Given these observations, we asked whether NF-κB signaling pathways were altered in RMS. We first performed western blot analysis on tissue extracts from xenografted human RH30 RMS tumors compared with normal skeletal muscle tissue. While normal tissue exhibited low levels of nuclear p65, processing of p100 to p52 was readily visible, indicative of an activated alternative signaling pathway (Figure 1A; Figure S1 in Supplementary Material). By contrast, in RH30 tumors, nuclear p65 expression was markedly increased while p100 processing to p52 was impaired (Figure 1A). In addition to human samples, we examined NF-κB signaling pathways in canine OS tumors. Attempts to use various p65 commercially available antibodies were unsuccessful in recognizing the canine form of the NF-κB subunit. Hence, we probed for NF-κB by EMSA using both canine control and OS samples. Results showed increased DNA binding activity in tumor samples, which by supershift analysis contained the p50 subunit of NF-κB (Figures 1B,C). This is evidence that classical signaling is increased in canine OS tumors. Similar to human sarcomas, we also observed that alternative signaling was decreased in canine OS tumors (Figure 1D). Together, results suggest that during the development of sarcomas, an NF-κB signaling switch occurs that causes the inactivation of the alternative pathway and the activation of the NF-κB classical pathway.

Next, we asked whether NF-κB signaling pathways were coupled during sarcomagenesis. To examine this regulation, we established RH30 and U2OS cell lines stably expressing the IκBα super repressor inhibitor of NF-κB (RH30/U2OS-IκBα SR) (Figure 2A). EMSAs and NF-κB cell luciferase reporter assays were used to confirm the efficiency of IκBα SR on the inhibition of NF-κB activity in RH30 and U2OS cells treated with the NF-κB activating cytokine, tumor necrosis factor (TNF) (Figures 2B,C). We then analyzed whether the inhibition of classical NF-κB signals impacted the alternative pathway in sarcoma cells. Compared with the p100 to p52 processing observed in RH30/U2OS-vector cells, there was a modest, but noticeable increase in p100 to p52 processing in RH30/U2OS-IκBα SR cells (Figure 2D). This suggested that the activation of the classical pathway occurs in the context of diminished alternative signaling, potentially driving the shift in NF-κB pathways observed in sarcomas.

To determine whether the loss of alternative signaling in sarcomas was reversibly coupled to the classical pathway, we generated RH30 and U2OS cells stably expressing an inactive form of the IKKα subunit where the NEMO-binding domain (NBD) had been deleted (IKKα ΔNBD) (Figure 2E). This domain is normally required for IKKα to form a heterotrimeric complex with IKKβ and IKKγ (NEMO) to activate classical NF-κB signaling. Deletion of NBD predominantly forces IKKα dimerization and activation of the alternative pathway (51). When we examined stable RH30 and U2OS IKKα ΔNBD clones for effects on classical pathway activation, no difference in p65 DNA binding activity was detected by EMSA compared with vector control cells (Figure 2F). Similarly, no difference was observed by western blot when probing for the nuclear localization of p65 in RH30 and U2OS IKKα ΔNBD clones versus controls (Figure 2G). These findings suggested that activation of alternative signaling does not affect the reciprocal activation of the classical pathway. Thus, the shift in NF-κB pathway activity in sarcoma cells likely results from a dominant classical signaling pathway, which when activated causes a concomitant reduction in alternative pathway activity.

We previously demonstrated that alternative signaling of NF-κB is important in regulating oxidative metabolism in skeletal muscle cells (30, 33, 34). Since this same pathway is downregulated in sarcoma cells, we asked whether the forced inactivation of classical NF-κB signaling in sarcoma cells, which we showed in Figure 2 partially re-establishes alternative signaling, would also promote oxidative metabolism. To examine the bioenergetic profile of sarcoma cells, we used the Seahorse Bioscience XF^e24 extracellular flux analyzer to measure OCR with a Mito stress test. OCR was used as a measure of aerobic respiration comparing RH30/U2OS vector and RH30/U2OS-IκBα cells. RH30/U2OS-IκBα SR cells lacking NF-κB activity had a substantially increased OCR profile compared with vector control cells (Figure 3A). Treatment of these cells with the ATP synthase inhibitor oligomycin, which blocks a necessary step in OXPHOS for ATP production, also revealed a significant increase (p < 0.05) in ATP-linked respiration (Figure 3B). Knockdown of p65 using siRNA recapitulated the metabolic effect seen with IκBα SR expression in U2OS cells (p < 0.05), but not in RH30 cells (Figure S2A in Supplementary Material). When we activated the alternative pathway by expressing IKKα ΔNBD, no metabolic difference was observed in RH30 or U2OS cells (Figure S2B in Supplementary Material), reaffirming our earlier results that alternative signaling does not reverse the activation of the classical pathway. Thus, findings suggest that activation of classical NF-κB reduces mitochondrial oxidative metabolism in sarcoma cells.

To probe the potential mechanism by which NF-κB regulates metabolism in sarcoma cells, we performed global gene expression analysis comparing RMS cells lacking NF-κB activity versus vector control cells. Using GSEA, we found that a gene set for glycolysis was significantly enriched (nominal p < 0.05; FDR < 0.25) in the control cells compared with the cells lacking NF-κB activity (Figure 4A). Since glycolytic genes showed decreased expression with loss of NF-κB activity, we examined these genes by rVista for potential conserved NF-κB binding sites. Results revealed HK2 as a potential transcriptional target of NF-κB, containing multiple binding sites in the first intron (Figure 4B). Consistent with the notion that HK2 is regulated by NF-κB, we observed that HK2 expression was reduced in RH30/U2OS-IκBα SR compared with vector control cells (Figure 4C). To further examine whether HK2 is regulated by NF-κB, we performed knockdown of p65 and observed decreased expression of HK2 in RH30 and U2OS cells silenced for p65 (Figure S3 in Supplementary Material). This suggested that NF-κB is a regulator of HK2 in sarcoma cells.

To determine whether HK2 is a direct target gene of NF-κB, we next performed ChIP assays using primers against three conserved NF-κB binding sites within the first intron of the HK2 gene located at positions +2123, +2604 and +3717 (named sites 1, 2, and 3, respectively) (Figure 4B). Results showed that p65 was significantly bound to all three NF-κB binding sites in both RH30 and U20S cells, and importantly, that this binding was reduced in sarcoma cells expressing the IκBα SR transgene (Figure 4D). Such data are supportive that classical NF-κB is a direct transcriptional regulator of HK2.

Previous reports have shown that HK2 is an important regulator of tumor growth and the Warburg effect (52–54). To determine if a similar role existed in sarcoma cells, we silenced HK2 in RH30 and U2OS cells using a short hairpin RNA (shRNA) that was confirmed by qRT-PCR and western blotting analysis (Figure 5A). Proliferation assays showed that the growth rate of RH30/U2OS-HK2 shRNA cells were not significantly different compared with RH30/U2OS-control shRNA cells (Figures S4A,B in Supplementary Material), nor did we observe an increase in cell death in RH30/U2OS-HK2 shRNA cells compared with RH30/U2OS-control shRNA cells (Figures S4C,D in Supplementary Material). These data indicated that HK2 was not required for sarcoma tumor cell proliferation or survival.

We next utilized the Seahorse Bioscience XF^e24 extracellular flux analyzer to examine whether HK2 depletion affected the metabolic profile of sarcoma cells. Similar to what we observed in sarcoma cells devoid of NF-κB, RH30 and U2OS cells silenced for HK2 also exhibited an increase in ATP-linked mitochondrial oxygen consumption, suggestive of a shift away from the Warburg effect (Figures 5B,C). Results further showed that cells lacking HK2 expression displayed a higher oxidative metabolism. To determine whether HK2 regulation of metabolism in sarcoma cells was dependent on NF-κB, we induced HK2 expression in RH30 and U2OS IκBα SR cells (Figure 5D). Results showed that expression of HK2 rescued the metabolic shift seen in U2OS IκBα SR cells (p < 0.05), but not in RH30 cells (Figures 5E,F). Taken together, these data support the notion that HK2 function in sarcoma cells is dispensable for the proliferation and viability of these cells, but important as a regulator of metabolism.