SP1 and NFY Regulate the Expression of PNPT1, a Gene Encoding a Mitochondrial Protein Involved in Cancer

The Polyribonucleotide nucleotidyltransferase 1 gene (PNPT1) encodes polynucleotide phosphorylase (PNPase), a 3′-5′ exoribonuclease involved in mitochondrial RNA degradation and surveillance and RNA import into the mitochondrion. Here, we have characterized the PNPT1 promoter by in silico analysis, luciferase reporter assays, electrophoretic mobility shift assays (EMSA), chromatin immunoprecipitation (ChIP), siRNA-based mRNA silencing and RT-qPCR. We show that the Specificity protein 1 (SP1) transcription factor and Nuclear transcription factor Y (NFY) bind the PNPT1 promoter, and have a relevant role regulating the promoter activity, PNPT1 expression, and mitochondrial activity. We also found in Kaplan–Meier survival curves that a high expression of either PNPase, SP1 or NFY subunit A (NFYA) is associated with a poor prognosis in liver cancer. In summary, our results show the relevance of SP1 and NFY in PNPT1 expression, and point to SP1/NFY and PNPase as possible targets in anti-cancer therapy.


Introduction
Mitochondria play a major role in the biology of eukaryotic cells as ATP producers through oxidative phosphorylation. Nevertheless, these organelles also work as a hub for metabolic functions, such as nutrient catabolism for energy production, the generation of biosynthetic precursors, redox homeostasis, and the management of metabolic waste (reviewed by [1]), and are involved in other cellular processes in health and disease (reviewed by [2]). Since mitochondria are organelles with a dual biogenesis, the transport of nuclear-encoded proteins and cytosolic RNA are essential processes for mitochondrial function. Unlike the well-characterized mitochondrial protein import, (reviewed by [3]), RNA transport into the mitochondria is a still a poorly understood process (reviewed by [4]).
The human Polyribonucleotide Nucleotidyltransferase 1 (PNPT1) gene, whose cytogenetic localization is 2p16.1, encodes Polynucleotide Phosphorylase (PNPase) which is essential for cell survival and mitochondrial DNA maintenance [5,6]. PNPase participates in a wide array of cellular processes, including nucleus-encoded RNA import into mitochondria, the processing and decay of miRNA and mRNA [7], and the processing of polycistronic mitochondrial transcripts and tRNAs [8]. Although PNPase was discovered in the 1950s [9], new functions of this protein have been recently discovered [10,11]. Once synthesized, PNPase is guided to the mitochondrial intermembrane space (IMS) [6,12] by a mitochondrial targeting sequence which is cleaved off upon translocation into the IMS [13]. Within the IMS, PNPase binds and facilitates the mitochondrial import of RNase P protein, 5S rRNA, and RNase MRP RNAs [5]. In E. coli, PNPase also plays a role in RNA turnover element (NRSE)-like sequence (consensus 5′-TTCAGCACCACGGACAGCGCC-The GC-box was part of a putative SP1 transcription factor binding site [conse (G/A)(C/T)(C/T)CCGCCC (C/A)-3′] [37] (Figure 1A). Transcription starting sites shown (TSS) [38].

Analysis of Cis-Elements with a PNPT1-Promoter Construct
The contribution of individual DNA elements to PNPT1 promoter activity sessed with HeLa cells using a firefly luciferase reporter construct containing a PNPT1 promoter region (Figure 2), and with derived constructs holding point mu of the four predicted elements. Cells were transiently transfected and the luciferas ity in lysates was measured. Maximum promoter activity was assessed by dose-re analysis; cell viability after transfection was also evaluated (not shown).
The activity of the PNPT1 promoter was similarly affected by single muta ISRE, CCAAT (NFY) or SP1 motifs (Figure 2), which downregulated promoter act  The ISRE, SP1 and NFY binding sites were almost fully conserved in rabbit, cow and sheep genomes, which also displayed NRSE-like sequences similar to the reported consensus (5 -TTCAGCACCACGGACAGCGCC-3 ) [36] ( Figure 1B).

Analysis of Cis-Elements with a PNPT1-Promoter Construct
The contribution of individual DNA elements to PNPT1 promoter activity was assessed with HeLa cells using a firefly luciferase reporter construct containing a 396-bp PNPT1 promoter region (Figure 2), and with derived constructs holding point mutations of the four predicted elements. Cells were transiently transfected and the luciferase activity in lysates was measured. Maximum promoter activity was assessed by dose-response analysis; cell viability after transfection was also evaluated (not shown).
to residual levels, suggesting the promoter could be working with two complexes, one for ISRE and a second one for SP1/NFY motifs. The ove and NRSE elements raises the possibility that the mutation in the latter co the SP1 assembly rather than the binding of a specific trans-element to th Figure 2. Mutational analysis of the predicted motifs of the human PNPT1 pro were transiently transfected with a luciferase reporter plasmid containing 396 bp 5087 of NG_033012.1 NCBI Reference sequence) of the PNPT1 promoter (396bp-5′ upstream from the ATG (WT, wild-type), or derived plasmids containing dele tations of the indicated motifs. The activity of the reporter was assayed 40 h afte represent the relative activity with respect to the native version of the promote analysis: 1 way ANOVA /Tuckey: * p < 0.05; ** p < 0.01; n = 6 (3 assays on duplica

SP1 and NFYA Bind PNPT1 Promoter
To determine whether the predicted cis-elements of the PNPT1 prom by their putative transcription factors, EMSA assays with labeled double nucleotide probes representing the identified motifs were performed.
Since SP1 and NRSE cis-elements overlap within the PNPT1 prom used probes encompassed both motifs ( Figure 1). The three probes mixe clear extracts (basal conditions) bound to high molecular weight comple suggesting that the nuclear extracts contained trans-elements recognizi motifs.
The previously described and characterized ISRE motif [11], was no assays.
To further characterize the specificity of complex binding to the probes were competed with unlabeled, double-stranded wild-type and m cleotides. Mutation of SP1 and NFY motifs in the unlabeled oligonucle their capacity to complete the formation of complexes of the labeled wild nuclear factors ( Figure 3B,C), indicating the existence of nuclear proteins bind to these motifs. Besides, a mutation in the NRSE-like (NRSE) motif d formation of probe-protein complexes, suggesting that NRSE-specific tra not present in the complex, or that the NRSE motif is not really functiona The identity of trans-acting elements was confirmed with specific a the use of αSP1 but not αNRSE hampered probe-protein complex forma The activity of the PNPT1 promoter was similarly affected by single mutations in ISRE, CCAAT (NFY) or SP1 motifs (Figure 2), which downregulated promoter activity by~60%. Meanwhile, a point mutation in NRSE-like sites reduced the promoter activity by~25%. A double mutant construct containing mutations in both SP1 and NFY motifs showed an activity similar to that of the singly mutated counterparts, suggesting that these elements work in a coordinated fashion. The mutation of one of these elements could be affecting the activity of the other, and for this reason the activity of the double SP1+NFY motif mutants is similar to that of single (SP1 or NFY motif) mutants.
Interestingly, the mutating ISRE motif further reduced the activity of NFY mutants to residual levels, suggesting the promoter could be working with two transcriptional complexes, one for ISRE and a second one for SP1/NFY motifs. The overlapping of SP1 and NRSE elements raises the possibility that the mutation in the latter could be affecting the SP1 assembly rather than the binding of a specific trans-element to this motif.

SP1 and NFYA Bind PNPT1 Promoter
To determine whether the predicted cis-elements of the PNPT1 promoter were bound by their putative transcription factors, EMSA assays with labeled double-stranded oligonucleotide probes representing the identified motifs were performed.
Since SP1 and NRSE cis-elements overlap within the PNPT1 promoter, one of the used probes encompassed both motifs ( Figure 1). The three probes mixed with HeLa nuclear extracts (basal conditions) bound to high molecular weight complexes ( Figure 3A), suggesting that the nuclear extracts contained trans-elements recognizing the proposed motifs.
( Figure 3D), ChIP assays were performed using HeLa cells. Both αSP1 and αNFYA enriched the amount of the PNPT1 promoter present in the precipitated material compared to controls, which is indicative of SP1 and NFYA binding to the PNPT1 promoter within the genomic DNA of cultured HeLa cells. Furthermore, similar results that confirm these findings can be found in online databases ( Figure S1).
Collectively, these results strongly suggest that SP1 and NFYA are part of the transcriptional machinery driving the PNPT1 expression.  (C) the arrow points out the complex that is competed with the unlabeled oligonucleotide containing a native SP1 motif or with αSP1 antibodies but not with the native NRSE motif or the presence of αNRSE antibodies. Shown are representative EMSA. (D) SP1 or NFYA chromatin immunoprecipitation in HeLa cells. Immunoprecipitation was performed with antibodies to SP1 or NFYA and the corresponding unspecific control antibody (Cont IgG). Enrichment relative to controls was determined by real-time PCR and expressed as a percentage (fold) over control. Shown are representative ChIP assays performed in triplicate (mean ± SD). Two different pairs of oligonucleotides encompassing the promoter region yield similar results. Statistical analysis: t-test, *** p < 0.001 (n = 6 independent IP).
The previously described and characterized ISRE motif [11], was not included in our assays.
To further characterize the specificity of complex binding to the motifs, labeled probes were competed with unlabeled, double-stranded wild-type and mutated oligonucleotides. Mutation of SP1 and NFY motifs in the unlabeled oligonucleotides impaired their capacity to complete the formation of complexes of the labeled wild-type probe with nuclear factors ( Figure 3B,C), indicating the existence of nuclear proteins that specifically bind to these motifs. Besides, a mutation in the NRSE-like (NRSE) motif did not affect the formation of probe-protein complexes, suggesting that NRSE-specific trans-elements are not present in the complex, or that the NRSE motif is not really functional.
The identity of trans-acting elements was confirmed with specific antibodies. Thus, the use of αSP1 but not αNRSE hampered probe-protein complex formation, suggesting that SP1, and not NRSE, binds the corresponding motif in HeLa cells ( Figure 3B). Likewise, antibodies against NFY subunit A (αNFYA) produced higher molecular weight complexes (supershift), which were not observed using antibodies to IRF-2, and which can be considered as control antibodies in this case ( Figure 3C). All of the above indicates that SP1 and NFYA polypeptides bind their predicted elements within the PNPT1 promoter in vitro.
To investigate the in vivo relevance of SP1 and NFYA binding to the PNPT1 promoter ( Figure 3D), ChIP assays were performed using HeLa cells. Both αSP1 and αNFYA enriched the amount of the PNPT1 promoter present in the precipitated material compared to controls, which is indicative of SP1 and NFYA binding to the PNPT1 promoter within the genomic DNA of cultured HeLa cells. Furthermore, similar results that confirm these findings can be found in online databases ( Figure S1).
Collectively, these results strongly suggest that SP1 and NFYA are part of the transcriptional machinery driving the PNPT1 expression.

PNPT1 Promoter Regulation by SP1 and NFYA in HeLa Cells
To evaluate the impact of SP1 and NFYA transcription factors on PNPT1 expression, we performed transfections with SP1-and NFYA-specific siRNAs followed by RT-qPCR analysis ( Figure 4A). SP1 and NFYA silencers displayed significant and similar efficacies at 10 nM and 50 nM, respectively, in reducing the expression of their target genes. As expected, SP1 silencing runs in parallel with a reduction of PNPT1 expression. Interestingly, SP1 silencing also produced a slight rise of NFYA expression (and vice versa), which could not counteract the downregulation of SP1 on PNPT1 mRNA levels. On the contrary, siRNA-mediated NFYA downregulation did not reduce PNPT1 expression. This could be the consequence of the parallel rise of SP1 levels observed in NFYA-silenced HeLa cells ( Figure 4A), an insufficient reduction of NFYA activity, the presence of other transacting factors that could bind the same motif (i.e., CCAAT enhancer binding proteins or C/EBPs), or a posttranslational compensatory stabilization of PNPT1 mRNA not observed with the reduction in SP1 levels.
To analyze the relevance of SP1 and NFYA expression in the transcriptional activity of the PNPT1 promoter ( Figure 4B), we used SP1 (10 nM) and NFYA (50 nM) siRNAs in luciferase reporter assays with the wild-type 396 bp PNPT1 promoter construct. Both SP1 and NFYA silencers reduced luciferase activity, although the effect of the SP1 silencer was 3-fold higher than that of the NFYA counterpart. These results indicated that both SP1 and NF-Y factors control, at least in part, the PNPT1 promoter basal activity in HeLa cells.
To analyze the relevance of SP1 and NFYA expression in the transcriptional a of the PNPT1 promoter ( Figure 4B), we used SP1 (10 nM) and NFYA (50 nM) siRN luciferase reporter assays with the wild-type 396 bp PNPT1 promoter construct. Bo and NFYA silencers reduced luciferase activity, although the effect of the SP1 silenc 3-fold higher than that of the NFYA counterpart. These results indicated that bo and NF-Y factors control, at least in part, the PNPT1 promoter basal activity in HeL

PNPT1, SP1 and NFYA Overexpression Is Correlated with a Poor Prognosis in Liver Cancer Patients
There is growing evidence suggesting that mitochondrial activity is an essential factor in cancer progression, and not just as a source of pathogenic stress. In this sense, mitochondrial activity is associated with hepatocellular cancer prognosis. Interestingly, liver cancer patients with higher PNPase levels showed a poorer prognosis than patients with a lower expression. In addition, high SP1 and NFYA expression levels were evidenced as being linked to a worse outcome. On the contrary, the transacting factors which bind ISRE motif in the PNPT1 promoter did not relate to disease progression ( Figure 5).

PNPT1, SP1 and NFYA Overexpression Is Correlated with a Poor Prognosis in Liver Cancer Patients
There is growing evidence suggesting that mitochondrial activity is an essential factor in cancer progression, and not just as a source of pathogenic stress. In this sense, mitochondrial activity is associated with hepatocellular cancer prognosis. Interestingly, liver cancer patients with higher PNPase levels showed a poorer prognosis than patients with a lower expression. In addition, high SP1 and NFYA expression levels were evidenced as being linked to a worse outcome. On the contrary, the transacting factors which bind ISRE motif in the PNPT1 promoter did not relate to disease progression ( Figure 5).
These results suggested that the activity of the SP1/NFY transcriptional complex on the PNPT1 promoter could be relevant for tumor growth and for the survival of liver cancer cells; they also suggested that there must be a pathological context in which the activity of the PNPT1 promoter depends on both SP1 and NFY, rather than just SP1. Consequently, the regulation of the formation of the SP1/NFY complex could be the underlying cause for the rise in the PNPT1 expression associated with a poor outcome in liver cancer patients.

Discussion
Promoters are critical regulatory elements in physiological and pathological conditions. Although they have not been considered as good pharmacological targets, several promising approaches to produce chemical probes and drugs targeting trans-acting fac- These results suggested that the activity of the SP1/NFY transcriptional complex on the PNPT1 promoter could be relevant for tumor growth and for the survival of liver cancer cells; they also suggested that there must be a pathological context in which the activity of the PNPT1 promoter depends on both SP1 and NFY, rather than just SP1. Consequently, the regulation of the formation of the SP1/NFY complex could be the underlying cause for the rise in the PNPT1 expression associated with a poor outcome in liver cancer patients.

Discussion
Promoters are critical regulatory elements in physiological and pathological conditions. Although they have not been considered as good pharmacological targets, several promising approaches to produce chemical probes and drugs targeting trans-acting factors have been proposed [39].
Our analysis of the PNPT1 promoter identified the previously characterized ISRE motif and additional SP1 and NFY cis-acting sites. These elements are conserved among mammals, pointing out their relevance in the regulation of PNPT1 expression. The functionality of these motifs has been demonstrated by mutagenesis using luciferase reporter plasmids. Besides, EMSA assays showed in vitro binding of NFYA and SP1 transcription factors to the corresponding putative sites on PNPT1, and in vivo binding was confirmed by ChIP assays.
Both SP1 [40] and NFY [41] are involved in the expression of mitochondrial proteins, and they have been reported to work in association [42], and even to physically interact [43]. In this sense, it has been reported that SP1 and NFY could operate as a unique trans-acting complex when their binding sites are in close proximity [44], as we found in the PNPT1 promoter. Our results showed that the mutation of both NFY and SP1 binding elements did not reduce the promoter activity beyond that displayed by single mutants of NFY or SP1 elements. However, additional mutation on ISRE caused a further reduction in activity. Collectively, these results suggested that the PNPT1 promoter operates with two transcriptional complexes, one for the ISRE motif, and a second one for the NFY/SP1 sites.
The ISRE motif binds a transcriptional complex which is active in basal conditions and responds to interferon γ by increasing transcriptional activity. Our results strongly suggest that the second complex operates under the control of SP1 in cancerous cells growing in conventional unstimulated conditions. In these conditions, NFY binds the promoter, but do not critically control its activity. In addition, we found a reduction of PNPT1 levels secondary to SP1 silencing, reinforcing the idea that the transcription factor SP1 regulates PNPT1 expression. However, in our experimental conditions, we did not observe a decrease in PNPT1 mRNA levels when NFYA was silenced, in contrast to the recombinant promoter activity which was downregulated upon NFYA silencing. The apparent discrepancy between the effect of NFYA silencing on native and recombinant promoter activities could be a consequence of the relatively high number of promoter copies present in cells transfected with the reporter construct, which reduces the ratio of available transacting factors. Moreover, the inefficacy of NFYA silencing to reduce PNPT1 mRNA levels could also be due to a compensatory stabilization of PNPT1 mRNA rather than to an absence of NFYA activity on the promoter. This compensatory mechanism could be dependent on SP1 but not on NFY factors.
The relevance of the PNPT1 promoter regulation in vivo can be deduced from the analysis of Kaplan-Meier survival plots. We found that a high PNPase expression in the tumor cells of hepatocellular cancer patients is linked to a poor prognosis. In addition, we found that SP1 or NFYA levels were also associated with a poor outcome, whereas the levels of trans-acting elements of the ISRE motif did not show a significant relation with liver cancer progression. In this sense, dysregulation of mitochondrial biogenesis has been linked to a hepatocellular carcinoma.
SP1 is overexpressed in a number of cancer types and is considered as a poor prognosis factor [45], and NFY has been reported to be overexpressed in lung [46] and hepatocellular cancer [47]. Our results suggest that in the context of tumor cell cultures, SP1 shows a more relevant role than NFY in driving PNPT1 expression. However, in the context of tumors of hepatocellular cancer, both NFY and SP1 seem to be linked to PNPase overexpression and to the outcome of patients. This suggests that the tumor microenvironment of patients has a key influence on the gene expression pattern of tumor cells, which is not mimicked by cultured cells (Figure S2).
Further studies are necessary to determine whether, under liver cancer pathological conditions, the accumulation of PNPase is a consequence of SP1/NFY complex formation or just a matter of expression levels of the individual transcription factors. The characterization of transcriptional complexes could allow the identification of target-specific molecules and the development of novel pharmacological repertoires for the therapeutic regulation of the transcription of genes such as PNPT1, whose overexpression is associated with serious diseases like liver cancer.

Analysis of the Promoter Region of the PNPT1 Gene
The proximal promoter sequence of the PNPT1 gene was obtained from the NCBI website. Nucleotides 4692 to 5095 of NG_033012.1 NCBI reference sequence were analyzed with MatInspector (Genomatix, Munich, Germany) and Promo (Alggen, Barcelona, Spain) online applications.
Ninety-six-well plates were used to assess mitochondrial activity and determine the number of mitochondria per well in gene-silenced cells.
Mitochondrial activity was determined by adding resazurin solution (10% vol/vol) (Biotium, Fremont, CA, USA) to culture medium, and relative quantification of mitochondria was performed with MitoTracker Green FM at 150 nM in PBS (ThermoFisher Scientific, Waltham, MA, USA), following manufacturer's indications. Resazurin reduction and Mito-Tracker green fluorescence emission were quantified using a Victor X5 fluorimeter (Perkin Elmer, Melbourne, Australia).
Both procedures display a direct linearity between 1000 and 10,000 cells per well.

DNA Constructs
The 396 bp sequence upstream of the ATG of PNPT1 and derived mutants were synthesized and cloned into the KpnI and HindIII sites of the pUC57 plasmid (Genescript, Piscataway, NJ, USA).
Wild-type and mutated promoters were subcloned into KpnI/HindIII of pGL4-Luc luciferase reporter vector (Promega, Sydney, Australia). DH5-competent bacteria were transformed with pGL4-Luc constructs and grown in selective LB Agar ampicillin (100 g/mL) plates. Isolated colonies were grown in selective LB medium supplemented with amplicilin. PureYield™ Plasmid Miniprep and Midiprep kits (Promega) were used for the purification of plasmids. The fidelity of the promoter region of all clones was verified by sequencing. Sequences of mutants are shown in Table 1.

Purification of RNA, Retrotranscription and qPCR
Total RNA was isolated from approximately 10 6 HeLa cells using the miRNeasy Mini Kit (Qiagen) following the manufacturer's instructions. Total RNA quantity and quality (260/280 absorbance ratio) were assessed using NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA).
One microgram of RNA was retrotranscribed using a High-Capacity RNA-to-cDNA Kit (Applied Biosystems, Foster City, CA, USA) following manufacturer's recommendations and using a Veriti thermal cycler device (Applied Biosystems).

Transfection of Reporter Constructs and Luciferase Reporter Assay
Luciferase reporter assays were performed with HeLa cells. Transfections were carried out with FuGENE (Promega), following the manufacturer's protocol. The day before transfection, HeLa cells were plated onto 96-well tissue culture dishes at a density that allowed cells to reach 60-70% confluence by the time of transfection (6000-7000 cells per well). Fresh culture media without antibiotics was used. The maximal activity of each reporter construct determined transfecting growing amounts (dose-response: 10-200 ng per well).
Luciferase activities were determined using a Victor X5 luminometer (Perkin Elmer), 0.5-s measuring period. The activities were expressed as relative values of the maximal activity, considering the activity of the wild-type promoter construct (396bp-PNPT1-promluc) as 100% and the activity of an empty vector (without promoter) as 0%.
Co-transfections of 396bp-PNPT1-prom-luc and siRNA probes were performed in 96-well plates using Lipofectamine 2000 (Invitrogen). Luciferase activity was assayed as described above. Silencing efficacies were assessed in parallel assays using 12-well plates.

Electromobility Shift Assays
Single-strand oligonucleotides were synthesized (IDT, Leuven, Belgium) for each region of the promoter and annealed to generate the probes. Probes were labeled with biotin for detection. Unlabeled probes were used in competition assays ( Table 2). Preparation of the probes was carried out according to manufacturer's instructions. Briefly, just before the experiment, complementary single-strand oligonucleotides (with or without biotin) were dissolved in DNAase-free water to a final concentration of 20 M in annealing buffer (60 mM KCl, 6 mM HEPES pH 7.5 and 0.2 mM MgCl 2 ) in a total volume of 100 µL, incubated for 1 min at 95 • C and gradually cooled to room temperature. EMSA experiments were carried out using wild-type and mutated probes for the different putative transcription factor binding sites ( Table 2). EMSAs for PNPT1 promoter probes were performed using HeLa nuclear extract (HeLaScribe ® Nuclear Extract, Promega). EMSAs were carried out as follows: 6 µg of commercial HeLa nuclear extract were preincubated for 10 min at 4 • C in binding buffer, then incubated with an excess of non-labeled wild-type or mutated probes for 10 min at 4 • C, and finally during 30 min with 10 fmol of annealed biotin-labeled probes. The binding buffer was composed of 10 mM Tris-HCl pH 7.0, 50 mM KCl, 1 g of poly dI·dC, 5% glycerol, 1 mM EDTA and 1 mM DTT. For supershifts, 2 µg of antibody targeting NFYA (Santacruz, Heildeberg, Germany), SP1 (Santacruz), NRSF (Santacruz), or IRF-2 (Santacruz) were added and incubated for 30 additional min at 4 • C. Afterward, samples were loaded onto a 4% polyacrylamide (40:1, acrylamide:bisacrylamide), 0.5× tris-borate-EDTA (TBE) buffer gel and electrophoresed at 320 V, 4 • C in 0.5× TBE buffer. Electrophoretic blotting to a nitrocellulose membrane (Hybond™-N+, GE Healthcare) was carried out in semidry conditions (30 V) for 15 min in 0.5× TBE, followed by crosslinking under UV light (1250 J, 1 min). Membranes were blocked for 15 min in blocking buffer [5% sodium dodecyl sulfate (SDS) in phosphate saline buffer, pH 7.0] followed by incubation with horseradish peroxidase (HRP)-labeled streptavidin at 125 pg/mL for 15 min in the same buffer. Blots were washed 5 times with a 10-fold dilution of blocking buffer. Signals on the membranes were visualized by chemiluminescence detection (Las4000, GE Health Care, Grens, Switzerland).

Chromatin Immunoprecipitation
Chromatin immunoprecipitations (ChIPs) were performed in HeLa cells using the Pierce™ Magnetic ChIP Kit (ThermoScientific). Briefly, cells were crosslinked with 1% formaldehyde (Thermo Scientific) for 10 min and the kit manufacturer's instructions followed. Chromatin was sheared enzymatically followed by sonication (30 s at 15% amplitude on ice, using an S-250 digital sonifier cell disruptor, Branson Ultrasonics, Danbury, CT, USA), as instructed by the kit. Immunoprecipitations were performed using 2 µg of the followiong antibodies: mouse monoclonal anti-NFYA antibody (Santa Cruz Biotechnology, sc-17753), rabbit policlonal anti-SP1 antibodies (Santa Cruz Biotechnology, sc-14027), normal mouse IgG antibody (EMD Millipore, 12-371) or rabbit policlonal IgG antibodies provided in the kit. Real-time PCR over the ChIP products was performed in a LightCycler 480 ® thermocycler (Roche) using a commercial polymerase mix (PowerUp™ SYBR ® Green Master Mix, Thermo Scientific) in triplicate. Enrichment as a percentage of the input was calculated and those values were referenced to the unspecific enrichment observed in the intergenic region. The primers used were: ON-PNPT1-PROM-3F, 5 -CACCGCGGAAACGAAACTC-3 and ON-PNPT1-PROM-3R, 5 -CAGGCTGTGCCCTGATTGG-3 , to amplify the PNPT1 promoter; and ON-CN-1F 5 -CTCTGTGCTGATACCTGGAGTCT-3 and ON-CN-1R 5 -GGAAATGGGGGCCTATGTTTTGG-3 to amplify the intergenic region.

Kaplan-Meier Survival Analysis
Data were downloaded from the Human Protein Atlas (data available from v21.protein atlas.org searching the indicated genes; all links are shown in Appendix A).
Cancer samples of The Cancer Genome Atlas (TCGA) project had been analyzed by RNA-Sequencing (RNA-Seq). The mRNA expression levels were determined as FPKMs (number Fragments Per Kilobase of exon per Million reads). Based on the FPKM value of each gene, patients were classified into two expression groups and the correlation between expression level and patient survival was examined. The prognosis of each group of patients was examined by Kaplan-Meier survival estimators, and the survival outcomes of the two groups were compared by log-rank tests.

Statistical Analysis
Prism 5.0 software (GraphPad, San Diego, CA, USA) was used for all calculations. Gaussian distribution was determined with Shapiro-Wilk or Kolmogorov-Smirnov tests. Bars (and whiskers) indicate mean + SD. The statistical significance of two groups was determined using the unpaired t-test, and more than two groups were analyzed using one-way ANOVA followed by the post hoc Tukey test. Differences between two conditions were considered statistically significant at p < 0.05 (* p < 0.05; ** p < 0.01; *** p < 0.001).

Conclusions
In summary, we have characterized the promoter of the human PNPT1 gene and established the essential roles of SP1 and NFY transcription factors in PNPT1 expression. Our results suggest that SP1 and NFY binding sites work as a single cis-element within the PNPT1 promoter, and that in hepatocellular cancer overexpressing PNPase, the promoter activity of PNPT1 depends on SP1 and/or NFY.