Structurally differentiated cis-elements that interact with PU.1 are functionally distinguishable in acute promyelocytic leukemia
© Qian et al.; licensee BioMed Central Ltd. 2013
Received: 3 February 2013
Accepted: 18 March 2013
Published: 2 April 2013
Transcription factor PU.1, a member of the ETS family, is a master regulator of myeloid differentiation whose functional disruption is often associated with acute myeloid leukemia (AML). Although much has been learned about PU.1 over the past decades, relatively little is known about cis-elements that interact with this factor under physiological or pathological conditions, especially in the whole-genome scale. We aimed to define the cistrome of PU.1 in acute promyelocytic leukemia (APL) cells and characterize the cis-elements bound by PU.1.
Chromatin immunoprecipitation with specific antibody coupled with deep sequencing (ChIP-seq) was used to investigate the in vivo PU.1 binding sites at the whole-genome scale in APL-derived NB4 cells. The ChIP-quantitative (q)-PCR and luciferase reporter assays were used to validate the binding events and trans-activity, respectively. Various computational analyses, including motif mining, evolutionary conservation analysis and functional enrichment analysis, were performed to characterize the cis-elements that interacted with PU.1.
A total of 26,907 significantly enriched binding regions of PU.1 were identified under the false discovery rate 0.1% in NB4 cells. PU.1 bound to various types of genomic regions and acted as a promoter-enhancer dual binding transcription factor. Based on the sequence length and composition, two types of representative motifs were identified in PU.1 binding sites: a long and a short motif. The long motif, characterized by high sequence specificity and binding affinity, predominantly resided in the promoter-distal regions. In contrast, the short one, with strong evolutionary constraint, represented the primary PU.1 cis-elements in the promoter-proximal regions. Interestingly, the short one showed more preference to be correlated with the binding of other factors, especially PML/RARα. Moreover, genes targeted by both PU.1 and PML/RARα were significantly involved in categories associated with oncogenesis, hematopoiesis and the pathogenesis of acute myeloid leukemia.
Our results demonstrate that structurally differentiated cis-elements that interact with PU.1 are functionally distinguishable in APL, suggesting that the sequence diversity of cis-elements might be a critical mechanism by which cells interpret the genome, and contribute to distinct physiological and/or pathological function.
KeywordsPU.1 PML/RARα ChIP-seq Acute promyelocytic leukemia cis-element
Biological processes are orchestrated by precise temporal and spatial regulation of gene expression, requiring proper interactions between cis-regulatory elements and trans-acting factors (TFs). Disruption of such interactions often causes disease. Over the past decades, much has been learned about actions of TFs under various physiological or pathological conditions, whereas relatively little is known about the cis-elements that control TF-specific gene expression . Recent advances in genomic technologies, including chromatin immunoprecipitation coupled with deep sequencing (ChIP-seq), have allowed the genome-wide recognition of in vivo cis-trans interacting sites, thus facilitating the survey of structural and functional features of thousands of cis-elements simultaneously, and providing the opportunity to understand the mechanism of gene regulation in a more comprehensive manner.
In hematopoiesis, a number of master TFs which play major instructive roles for hematopoietic development or malignant transformation have been identified and intensively studied. PU.1, one of these master TFs, is exclusively expressed in hematopoietic cells and has been identified as a crucial transcription factor in normal hematopoiesis and in generation of myeloid leukemia through disruption of its function . In acute promyelocytic leukemia (APL), a subtype of AML with the typical promyelocytic leukemia-retinoic acid receptor α (PML/RARα) fusion protein in the disease cells, PU.1 is expressed at reduced levels and increased expression mediated by gene transfer of PU.1 is sufficient to induce neutrophil differentiation, similar to the effect of all-trans retinoid acid (ATRA) [3, 4]. With transgenic mouse models, the penetrance rate of APL development is significantly increased in offspring when PML/RARα mice are crossed with PU.1+/- mice . These observations collectively suggested the presence of crosstalk between PU.1 and PML/RARα in APL. Using ChIP combined with whole-genome promoter arrays, we previously investigated the early molecular effects of PML/RARα in hematopoietic progenitor cells and demonstrated that PML/RARα disrupts the PU.1 regulated genes and thus results in a blockage of the downstream PU.1 signaling . However, a question regarding whether cis-elements that interact with PU.1 contribute to the selective binding of PML/RARα in APL remains unknown.
It is interesting to note that myeloid specific genes such as granulocyte colony-stimulating factor (G-CSF) receptor , granulocyte-macrophage (GM)-CSF receptor α  and macrophage (M)-CSF receptor  have PU.1 binding sites in their promoter regions. However, as these genes are investigated at the single-gene level, it is challenging to gain a comprehensive understanding of the cis-elements that are essential for the regulation of myeloid-specific genes. For instance, PU.1 was first reported to bind to purine-rich 5′-GGAA/T-3′ sequences, similar to other members of the Ets family [10, 11]. However, later studies showed that the DNA binding specificity of PU.1 was quite different from that of the other members, in that some sites required a string of adenosine residues at the 5′ of the GA core [8, 9, 12, 13]. Additional data indicated that some of the PU.1 binding sites were biologically significant but lack of the adenosine string at the 5′ of GA core . Also, nucleotides (i.e., G and T) flanking the 3′ end of the GAGGAA sequence appear to be important for PU.1 binding and its transactivation activity . Given the reported variability of PU.1 binding sites, and our interest in this transcription factor in disease, we performed ChIP-seq experiments in the APL-derived NB4 cells using PU.1-specific antibody to identify the in vivo PU.1 binding sites at the whole-genome scale. This revealed a number of interesting features which are potentially important for regulating myeloid differentiation and leukemogenesis.
Identification and validation of in vivo binding regions of PU.1
ChIP-seq reads and peaks threshold at FDR=0.001
Total sequenced (millions)
Total, mapped (millions)
Total, uniquely mapped and Non-redundant (millions)
In peaks (millions)
Peak coverage (Mb)
Median width (bp)
Number of peaks
Minimum fold enrichment
Average fold enrichment
Median fold enrichment
Characterizing PU.1 as a promoter-enhancer dual binding TF
Next, we performed a correlation analysis between the PU.1 binding sites versus total gene number and nucleotide number on each chromosome, respectively. As a control, we conducted the same analysis with the binding sites of a classic promoter-binding factor RNA polymerase II (RNAPII), a typical enhancer-binding protein estrogen receptor (ER)  and some other factors including CTCF , STAT1 , FOXA1 , GATA1  and GATA2  (Figure 2C and Additional file 4: Table S2), whose genome-wide binding sites have been documented previously. Interestingly, the PU.1 binding sites were correlated with both the chromosomal gene number (r 2 =0.70) and nucleotide number (r 2 =0.72) (Figure 2C), which appeared to be distinguishable from most of the other tested factors except for STAT1, a known promoter-enhancer dual binding TF [31, 32]. The factors like RNAPII, CTCF, GATA1 and GATA2, were obviously correlated with the number of genes, whereas those like ER and FOXA1, known as two specific enhancer-binding proteins, were correlated with the number of nucleotides. The above observations suggest that PU.1 may act as a versatile factor able to interact with cis-elements not only in promoter regions but also in enhancer regions.
In addition, we investigated the PU.1 binding locations and numbers on their corresponding RefSeq genes (9,556), revealing that 33.3% (3,184/9,556) of the RefSeq genes harbored the binding sites on their promoter-proximal regions whereas 66.7% (6,372/9,556) contained the binding sites on the promoter-distal regions. These observations, together with data shown in Figure 2A, suggest that promoter-distal binding of PU.1 may play at least as equally important roles as the promoter-proximal binding in transcriptional regulation. Interestingly, more than half of the genes (1,598/3,184) with promoter-proximal binding of PU.1 appeared to contain additional binding sites in their promoter-distal regions, suggesting that PU.1 regulatory mechanisms can be far more complex than previously recognized, by involving multiple trans-cis interaction sites. This would allow for precise control of gene expression that is essential for myeloid differentiation. Indeed, auto-regulation of the PU.1-encoding gene SPI1 appears to require PU.1 binding at both the promoter  and enhancer (i.e., 17 kb upstream) . In this study, we found in addition to the sites reported previously two additional sites were identified 9.6 kb and 14.6 kb upstream of the gene (Figure 2D). An additional example is the integrin alpha M chain (CD11b) gene, ITGAM. This gene is known to be important for the adherence of neutrophils and monocytes during differentiation. It contains one PU.1 binding site at the promoter as previously reported . In our study we have identified another site 16 kb upstream of the gene, representing a potential enhancer region (Figure 2D). Such structures may typically represent PU.1-involved trans-cis interactions required for myeloid differentiation. In consistent with this notion, numerous other myeloid differentiation-required genes are also multi-targeted by PU.1, such as LMO2, BCL3, IL1B and IL12B (Figure 2D and Additional file 5: Figure S3).
Distinct features of the short and long motifs in the PU.1-bound regions
Next, we scanned the total binding regions of PU.1 using the above long and short motifs by MotifScan. As shown in Figure 3B, 37.1% of the binding regions contained one or more long motifs, and 46.3% of the binding regions contained only the short motif. The remaining 16.6% revealed neither long nor short motifs, which were likely due to undetected PU.1-binding motifs present in these regions, or due to the possibility that for these sites PU.1 does not directly bind to chromatin, but rather forms a complex through protein-protein interactions. Positional distribution analysis revealed that both motifs, especially the long one, appeared to reside near the center of the binding regions (Figure 3C). We then evaluated the binding affinities of the long and short motifs by comparing their enrichment levels. As shown in Figure 3D, the long motif exhibited a much higher mean tag density than the short one, particularly with respect to the tag density in the regions (-100 to +100) flanking the summit of peaks. Consistently, we found that the higher the enrichment levels, the more (less) the percentages of the long (short) motifs (Figure 3E). These observations suggest that the long motif exhibits higher binding affinity to PU.1 than the short one. Interestingly, sequence evolutionary conservation analysis showed that the binding sites with the short motif appeared to be much more conserved than the long motif-containing sites (Figure 3F). These results together suggest that the motif preference may correlate with the motif location in the genome, implicating that functional roles played by the two types of motifs can be different in general. Then, we examined the proportional distributions of the short or long motifs in promoter regions vs. non-promoter regions, respectively. As shown in Figure 3G, the percentage of short or long motifs (48.8% vs. 36.1%, outer circle) in non-promoter regions was equivalent to that for the total of PU.1 binding sites (46.3% vs. 37.1%, Figure 3B), whereas, that in the promoter regions with 63.3% short motifs and 14.5% long motifs (inner circle) was significantly different from that.
In sum, cis-elements that interact with PU.1 can be classified into short and long motif classes based on their sequence patterns. In promoter regions, PU.1-cis elements are predominantly represented by those in the class of short motif, which are highly conserved across species but with lower binding affinity to PU.1. In contrast, cis-elements in the class of long motif are relatively depleted from promoter regions whereas primarily present in non-promoter regions, representing PU.1 binding sites of high affinity but evolutionally less conserved.
Short motif-containing binding regions of PU.1 preferentially targeted by other factors including PML/RARα
Next, we wanted to know whether these PU.1-bound regions with distinct cis-elements could be co-bound or tethered differentially by other factors in addition to PML/RARα. For this analysis we took advantage of the published Encyclopedia of DNA elements (ENCODE) project data that includes the genomic regions bound by 119 human transcription factors involving 72 different cell types [36, 37]. We compiled three different PU.1-bound region sets, including the PU.1-specific with the long motif (Long), the PU.1-specific with the short motif (Short) and the PU.1&PR, and carried out overlapping analysis with 1328 ChIP-seq data sets downloaded from the UCSC ENCODE data center. As shown in Figure 4D and 4E, most of the ENCODE TFs showed significantly less preference to the Long set than the Short set and the PU.1&PR set (paired t-test p<1e-303 and p<1e-296, respectively). In particular, factors, such as MAX (21.3% vs. 37.7%) and MYC (17.8% vs. 34.4%) in NB4 cells, JUND in K562 cells (13.4% vs. 26.0%) and CTCF in HCPE (12.3% vs. 27.3%) and HBMEC (12.9% vs. 25.9%) cells showed the most distinct overlapping percentage between the Long and Short sets. In contrast, the Long set was significantly more covered by 9 ENCODE PU.1 data sets involving 4 cell types (GM12878, GM12891, HL60 and K562) than the Short and PU.1&PR sets (paired t-test p<1e-6 and p<0.0002, respectively; Figure 4F). The result suggested that the long-motif-containing sites kept more PU.1 binding stability among different cell types, consistent with their higher binding affinity to PU.1 revealed above, while the short-motif-containing sites could recruit more TFs to regulate the corresponding gene expression.
Furthermore, we wanted to know whether genes regulated by different PU.1 cis-elements or binding factors were functionally differentiated. Thus, we conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses respectively on the three gene sets (Long, Short and PU.1&PR). As shown in Figure 4G, the most striking GO terms revealed in the gene set targeted by both PU.1 and PML/RARα (PU.1&PR) were highlighted by genes involved in oncogenesis and hematopoiesis. The former was represented by JUNB, BCL2, MLLT10, FES, FOS and MCL1, and the latter was represented by RUNX1, CEBPA and STAT3. Similarly, the most significant KEGG pathways revealed in this gene set were represented by those involved in acute myeloid leukemia, including SPI1, RARA, KRAS and so on (Figure 4H and Additional file 8: Figure S5). These results provide additional evidence that genes targeted by both PU.1 and PML/RARα are indispensable for the normal hematopoiesis, and are crucial for leukemogenesis (Figure 4G and 4H). Besides, we also noticed that the genes belong to the Long set specifically evolved in several biological processes and functional pathways, such as developmental processes and adherens junction, respectively. The Short gene set, however, showed little particularly functional enrichment, implying PU.1 may involve more extensive biological functions rather than some specific ones through the different dynamic combinations with other factors.
PU.1 is a master TF in myeloid differentiation, regulating numerous myeloid genes involved in hematopoiesis. Understanding cis-regulatory elements that interact with PU.1 may provide insights into regulatory networks underlying myeloid differentiation and related diseases. In this study, we identified 26,907 in vivo PU.1 binding sites in APL-derived NB4 cells by ChIP-seq. Through various analyses, we found the PU.1 binding sites were primarily represented by two types of cis-elements. One, with high sequence specificity and binding affinity, resides primarily in the promoter-distal regions, and a second, with strong evolutionary constraint, represents the primary PU.1 cis-elements in the promoter-proximal regions; the latter tend to be bound by PU.1 in association with other factors such as PML/RARα, MYC and MAX. Our findings suggest that sequence diversity of TF binding cis-elements is a critical mechanism by which cells interpret the genome, and contribute to distinct physiological and/or pathological function.
Physical interaction between distinct chromosome regions separated by hundreds of kilobases is thought to be important in the regulation of gene expression . By analyzing the distribution of PU.1 binding sites in APL, we found that over 85% of the sites resided in chromatin regions away from promoters. Interestingly, after mapping the PU.1 binding sites to the known RefSeq genes, more than half of the promoter-targeted RefSeq genes (1,598/3,184) contained additional PU.1 binding site at non-promoter regions, suggesting the presence of complex networks of interconnected chromatin. Also, it is tempting to assume that there is a mode of action through the long-range regulation in addition to conventional models of the promoter regulation in PU.1-regulated gene expression. Such an assumption appears to be supported by recent findings in many cell types, in which long-range interactions are common for transcriptional regulation , and by the findings that interacting loci between transcription start sites (TSSs) and enhancers are strongly correlated with gene expression level . Another example of long-range regulation is provided by estrogen receptor alpha (ERα) induced looping , in which the vast majority of ERα binding sites reside in non-promoter regions . It has been reported that PU.1 mediated looping between promoter pIII and the distal element termed hypersensitive site 1 (HSS1) of transcriptional co-activator CIITA in B cell of mouse model , but whether looping is common or infrequent in the regulatory networks of PU.1 remains to be elucidated. Consistent with previous investigations at the single gene level, our study added additional 1,598 PU.1 target genes potentially regulated across long distances.
The availability of genome-wide TF binding sites benefits our investigation regarding the regulatory mechanisms underlying protein-DNA interaction and improves the accuracy for analyzing cis-regulatory elements. In this study, we found potential cis-elements of PU.1 target genes can be classified into the canonical short consensus (AG(A/G)GGAAG) and the extended long consensus ((A/G)AAAG(A/G)GGAAGTG). The shorter sequence appears to be highly conserved across species but shows less binding affinity to PU.1, whereas the extended one demonstrates high binding affinity but low evolutionary conservation. This observation is also supported by a microarray-based binding affinity study, in which the vast majority of the 104 DNA-binding proteins tested showed different binding affinities to different DNA consensuses . Interestingly, potential cis-elements in the promoter-proximal regions of PU.1 regulated genes are predominantly represented by the short motif elements (63.4%) whereas these regions are relatively depleted of the long motif elements (14.5%). In contrast, the long motif elements primarily reside in the promoter-distal regions (including enhancers). Although it remains to be elucidated why these long and evolutionally diverse elements are predominantly located in non-promoter regions, it might be of interests to speculate that these elements may play species-specific roles in precise regulation of gene expression required for myeloid.
Of note, PML/RARα targeted PU.1 binding regions are highly enriched with the canonical short consensus (69.2%) but depleted of the extended long consensus sequence (11.6%). An intensive overlapping analysis revealed that most of the over 100 tested transcription factors showed more preference to the short-motif-containing binding regions of PU.1. Although much remains to be elucidated, it is tempting to assume that binding of PU.1 to the low affinity canonical consensus may represent an important mechanism that controls physiological or pathological process through the differential dynamic combinations with other factors, such as MYC and MAX. The extended long consensus elements are mostly distributed in non-promoter regions, and represent high affinity binding sites of PU.1. We speculate that these sites cooperate with the binding sites in promoter regions and regulate hematopoietic specific genes through long-range regulation. In the presence of the pathological protein PML/RARα, these sites are devoid of targeting by PML/RARα to a large extent, probably needed to keep certain expression levels of the genes. Indeed, 80% down-regulation of PU.1 in hematopoietic cells causes the blockage of cell differentiation  and restoration of PU.1 expression induces neutrophil maturation . Obviously, much remains to be investigated about regulatory networks of PU.1 in myeloid differentiation or leukemogenesis. However, findings and a resource of thousands of potential cis-elements of PU.1 from this setting may facilitate such investigation in a more effective and efficient manner.
We here describe a genome-wide characterization of in vivo binding sites of PU.1 in APL-derived NB4 cells. Our results demonstrate that PU.1 can regulate target genes by binding to both the promoter-proximal and distal cis-elements. Moreover, we reveal that structurally differentiated cis-elements that interact with PU.1 are functionally distinguishable in acute promyelocytic leukemia, suggesting that sequence diversity of cis-elements that interact with trans-acting factors might be a critical mechanism by which cells interpret the genome, and contribute to distinct physiological and/or pathological function.
Materials and methods
NB4 cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS). HEK 293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, in a humidified atmosphere with 5% CO2 at 37°C.
Plasmid construction, cell transfection and luciferase reporter assays
Promoter and enhancer regions harboring PU.1 motifs were cloned into the pGL3-basic and pGL3-promoter vector (Promega, Madison, WI), respectively. The primers used for the plasmids constructs are listed in Additional file 9: Table S4. The renilla luciferase plasmid pRL-SV40 (Promega, Madison, WI) was used as control for transfection efficiency. The expression plasmid was pCMV4-PU.1. HEK 293T cells were transiently transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). Transfected cells were cultured for 48 hours and then assayed for luciferase activity using Dual-Luciferase Reporter Assay System reagents (Promega, Madison, WI).
ChIP-qPCR was performed using Power SYBR® Green PCR Master Mix (Applied Biosystems, Foster City, CA) and ABI Prism 7900HT detection system (Applied Biosystems, Foster City, CA). The fold enrichment of the tested binding regions over the input DNA was estimated as previously described . The primers used for ChIP-qPCR are listed in Additional file 10: Table S5.
ChIP-seq and data analysis
ChIP was performed using specific antibodies according to the Affymetrix protocol as described previously . ChIPed and Input DNA were sequenced with Illumina Genome Analyzer II. The 35 bp reads (or tags) were aligned (mapped) to the unmasked human reference genome (NCBI v36, hg18) using the Eland application (Illumina) allowing two mismatches. Only uniquely mapped reads were retained to further analyses. Next, MACS  algorithm was used to identify PU.1 binding regions. For the visualization of the enrichment level of transcription factor’s binding sites, we calculated the tag density of each ChIP-seq sample with 500 bp window, aligned it to the same coordinate and visualized them in a bar plot using the IGB (Affymetrix) program.
Peak mapping and annotation
We used the RefSeq Genes’ database from UCSC to map and annotate the peak regions. For each peak region, we first searched the nearest RefSeq gene in both directions unless no gene was found within 50 kb. For a peak region lying within a gene, we classified it to the proximal promoter (-2 kb upstream to 1kb downstream to TSS), gene body (1 kb downstream of the TSS to the transcription end site (TES)), upstream enhancer (between at most -50 kb and -2 kb upstream to the TSS), or downstream enhancer (from TES to at most 50 kb downstream). Otherwise, we marked these peaks as distal intergenic region (>50 kb from a RefSeq gene). To avoid multiple genomic region type annotation of one peak, we uniquely mapped the peaks to genomic region type following the priority rule: proximal promoter > gene body > upstream enhancer > downstream enhancer > distal intergenic region.
Sequence evolutionary conservation analysis
The enriched peaks were first uniquely mapped to certain genomic regions according to the above peak mapping and annotation criteria. Then, the regions were aligned at their summits from 5′ to 3′ in accordance with the orientation of the corresponding genes (if a peak belongs to the distal intergenic region, we arbitrarily assume that the peak is on the positive strand) and uniformly expanded to 3,000bp in each direction, and phastCons scores were retrieved from UCSC genome browser (http://genome.ucsc.edu) and averaged at each position.
Statistical analysis of the known motifs with MotifScan
The acquisition of motif PSFM
The motif PSFM can be from the result of de novo motif analysis on a set of DNA sequence data. In this study, the long motif PSFM of PU.1 was from the result of AMD , an automated motif discovery tool using stepwise refinement of gapped consensuses, analysis on the highly enriched PU.1 occupancy sites of ChIP-seq data set. It also could be from some motif information database directly and the TRANSFAC database was used in this study.
The calculation of similarity between a motif and a DNA sequenceGiven a motif with length l bp, the similarity (S) for a DNA sequence with equal length was calculated as follow:
Where Ij indicates the information content of the motif j-th column, P ij is the frequency of a particular letter i in the j-th column and is the frequency of a specific letter with the maximum value in the j-th column. Details described as follow:
P ij – The frequency of a particular letter i in the j-th column (i. e. if G occurred 3 out of 6 times in an alignment column, this would be 0.5). We define P ij log P ij = 0, when P ij = 0.
Q i – The expected frequency of a letter i. This is an optional argument, usage of which is left at the user’s discretion. By default, it is automatically assigned to 0.25 = 1/4 for a nucleic acid alphabet. This is for getting the information content without any assumption of prior distributions.
The computation of fold enrichment (F) and Z-score (Z)
For a specific motif, the sequence with a similarity higher than the threshold was marked as a matched sequence. Suppose that the total number of matched sequences was X in a data set with total effective sequence length L bp (repeat masked). Then the random variable X followed the binomial distribution with parameters n and p, written as X ∼ B(n, p), where n ≈ L, and p was the possibility that each bp contained a matched sequence. In this case, n was very large generally. Meanwhile, n*p and n*(1-p) were large enough. Thus, to simplify the calculation, we used the normal distribution X ∼ N(np, np(1 − p)) to excellently approximate the binomial distribution. Subsequently, we further simplified the 1–p as 1, since the p was small generally. Therefore, the expected value (E) and the variance (σ2) of X would be the same.
Next, the total number of matched sequences was counted in a background data set and a sample data set, respectively. Suppose that the total number of matched sequences is N s in a sample data set with total effective sequence length L s (bp), and N c in a background data set with total effective sequence L c (bp). We calculated the expected number of matched sequences in the sample data set as Followed the above simplified formula, the fold enrichment and Z-score were calculated as and , respectively.
Percentage calculation of the motif-containing peaks
For a specific motif, if a peak region contained at least one matched sequence, then we marked this peak as a motif-containing one. The percentage of certain motif-containing peaks was calculated as the number of motif-containing ones divided by all the number of peaks. The similarity was calculated as the same as in MotifScan algorithm. In this study, to enhance the accuracy and specificity, the similarity threshold for the “long” PU.1 motif was set at 0.8 and that for the “short” PU.1 motif and RARE half was set at 0.9. We first excluded all the long PU.1 motif-containing peaks when calculating the only short PU.1 motif-containing ones, as the short PU.1 motif was nearly covered by the long one.
Acute myeloid leukemia
Acute promyelocytic leukemia
Chromatin immunoprecipitation coupled with deep sequencing
Promyelocytic leukemia-retinoic acid receptor alpha
all trans retinoid acid
False discovery rate
RNA polymerase II
Retinoic acid response element
University of California Santa Cruz
Encyclopedia of DNA Elements
Kyoto encyclopedia of genes and genomes
Hypersensitive site 1
Position-specific frequency matrix
Transcription start site
Transcription end site
We would like to thank Prof. Mark D Minden at the Ontario Cancer Institute / Princess Margaret Hospital for the critical reading and professional editing of the manuscript. We acknowledge Prof. X. Shirley Liu at the Harvard School of Public Health for the helpful discussion. This work was supported in part by Ministry of Science and Technology Grants of China (2013CB966802, 2012AA02A211, 2012AA02A505, 2009CB825607 and 2011CB910202), the National Natural Science Foundation (90919059, 81170503, 31171257 and 30971623), Chinese Academy of Sciences (KSCX2-EW-Q-1-08), the Shanghai Commission of Science and Technology (11431922402), and SA-SIBS Scholarship Program.
- Wittkopp PJ, Kalay G: Cis-regulatory elements: molecular mechanisms and evolutionary processes underlying divergence. Nat Rev Genet. 2012, 13 (1): 59-69. 10.1038/nri3362.View ArticleGoogle Scholar
- Tenen DG: Disruption of differentiation in human cancer: AML shows the way. Nat Rev Cancer. 2003, 3 (2): 89-101.View ArticlePubMedGoogle Scholar
- Mueller BU, Pabst T, Fos J, Petkovic V, Fey MF, Asou N, Buergi U, Tenen DG: ATRA resolves the differentiation block in t(15;17) acute myeloid leukemia by restoring PU.1 expression. Blood. 2006, 107 (4): 3330-3338.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhu X, Zhang H, Qian M, Zhao X, Yang W, Wang P, Zhang J, Wang K: The significance of low PU.1 expression in patients with acute promyelocytic leukemia. J Hematol Oncol. 2012, 5: 22-10.1186/1756-8722-5-22.PubMed CentralView ArticlePubMedGoogle Scholar
- Walter MJ, Park JS, Ries RE, Lau SK, McLellan M, Jaeger S, Wilson RK, Mardis ER, Ley TJ: Reduced PU.1 expression causes myeloid progenitor expansion and increased leukemia penetrance in mice expressing PML-RARalpha. Proc Natl Acad Sci USA. 2005, 102: 12513-12518. 10.1073/pnas.0504247102.PubMed CentralView ArticlePubMedGoogle Scholar
- Wang K, Wang P, Shi J, Zhu X, He M, Jia X, Yang X, Qiu F, Jin W, Qian M: PML/RARalpha targets promoter regions containing PU.1 consensus and RARE half sites in acute promyelocytic leukemia. Cancer Cell. 2010, 17: 186-197. 10.1016/j.ccr.2009.12.045.View ArticlePubMedGoogle Scholar
- Smith LT, Hohaus S, Gonzalez DA, Dziennis SE, Tenen DG: PU.1 (Spi-1) and C/EBP alpha regulate the granulocyte colony-stimulating factor receptor promoter in myeloid cells. Blood. 1996, 88: 1234-1247.PubMedGoogle Scholar
- Hohaus S, Petrovick MS, Voso MT, Sun Z, Zhang DE, Tenen DG: PU.1 (Spi-1) and C/EBP alpha regulate expression of the granulocyte-macrophage colony-stimulating factor receptor alpha gene. Mol Cell Biol. 1995, 15: 5830-5845.PubMed CentralView ArticlePubMedGoogle Scholar
- Zhang DE, Hetherington CJ, Chen HM, Tenen DG: The macrophage transcription factor PU.1 directs tissue-specific expression of the macrophage colony-stimulating factor receptor. Mol Cell Biol. 1994, 14: 373-381.PubMed CentralView ArticlePubMedGoogle Scholar
- Klemsz MJ, McKercher SR, Celada A, Van Beveren C, Maki RA: The macrophage and B cell-specific transcription factor PU.1 is related to the ets oncogene. Cell. 1990, 61: 113-124. 10.1016/0092-8674(90)90219-5.View ArticlePubMedGoogle Scholar
- Nguyen VC, Ray D, Gross MS, de Tand MF, Frezal J, Moreau-Gachelin F: Localization of the human oncogene SPI1 on chromosome 11, region p11.22. Hum Genet. 1990, 84: 542-546.View ArticlePubMedGoogle Scholar
- Pahl HL, Scheibe RJ, Zhang DE, Chen HM, Galson DL, Maki RA, Tenen DG: The proto-oncogene PU.1 regulates expression of the myeloid-specific CD11b promoter. J Biol Chem. 1993, 268: 5014-5020.PubMedGoogle Scholar
- Galson DL, Hensold JO, Bishop TR, Schalling M, D’Andrea AD, Jones C, Auron PE, Housman DE: Mouse beta-globin DNA-binding protein B1 is identical to a proto-oncogene, the transcription factor Spi-1/PU.1, and is restricted in expression to hematopoietic cells and the testis. Mol Cell Biol. 1993, 13: 2929-2941.PubMed CentralView ArticlePubMedGoogle Scholar
- Gangenahalli GU, Gupta P, Saluja D, Verma YK, Kishore V, Chandra R, Sharma RK, Ravindranath T: Stem Cell Fate Specification: Role of Master Regulatory Switch Transcription Factor PU.1 in Differential Hematopoiesis. Stem Cells Dev. 2005, 14: 140-152. 10.1089/scd.2005.14.140.View ArticlePubMedGoogle Scholar
- Li SL, Schlegel W, Valente AJ, Clark RA: Critical flanking sequences of PU.1 binding sites in myeloid-specific promoters. J Biol Chem. 1999, 274: 32453-32460. 10.1074/jbc.274.45.32453.View ArticlePubMedGoogle Scholar
- Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, Liu XS: Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008, 9: R137-10.1186/gb-2008-9-9-r137.PubMed CentralView ArticlePubMedGoogle Scholar
- Muller S, Sideras P, Smith CI, Xanthopoulos KG: Cell specific expression of human Bruton’s agammaglobulinemia tyrosine kinase gene (Btk) is regulated by Sp1- and Spi-1/PU.1-family members. Oncogene. 1996, 13: 1955-1964.PubMedGoogle Scholar
- van Dijk TB, Baltus B, Caldenhoven E, Handa H, Raaijmakers JA, Lammers JW, Koenderman L, de Groot RP: Cloning and characterization of the human interleukin-3 (IL-3)/IL-5/ granulocyte-macrophage colony-stimulating factor receptor betac gene: regulation by Ets family members. Blood. 1998, 92: 3636-3646.PubMedGoogle Scholar
- Eklund EA, Kakar R: Recruitment of CREB-binding protein by PU.1, IFN-regulatory factor-1, and the IFN consensus sequence-binding protein is necessary for IFN-gamma-induced p67phox and gp91phox expression. J Immunol. 1999, 163: 6095-6105.PubMedGoogle Scholar
- Li SL, Valente AJ, Qiang M, Schlegel W, Gamez M, Clark RA: Multiple PU.1 sites cooperate in the regulation of p40(phox) transcription during granulocytic differentiation of myeloid cells. Blood. 2002, 99: 4578-4587. 10.1182/blood.V99.12.4578.View ArticlePubMedGoogle Scholar
- Chen H, Ray-Gallet D, Zhang P, Hetherington CJ, Gonzalez DA, Zhang DE, Moreau-Gachelin F, Tenen DG: PU.1 (Spi-1) autoregulates its expression in myeloid cells. Oncogene. 1995, 11: 1549-1560.PubMedGoogle Scholar
- Kuhn RM, Karolchik D, Zweig AS, Wang T, Smith KE, Rosenbloom KR, Rhead B, Raney BJ, Pohl A, Pheasant M: The UCSC Genome Browser Database: update 2009. Nucleic Acids Res. 2009, 37: D755-D761. 10.1093/nar/gkn875.PubMed CentralView ArticlePubMedGoogle Scholar
- Brown RP, Feder ME: Reverse transcriptional profiling: non-correspondence of transcript level variation and proximal promoter polymorphism. BMC Genomics. 2005, 6: 110-10.1186/1471-2164-6-110.PubMed CentralView ArticlePubMedGoogle Scholar
- Wray GA: The evolutionary significance of cis-regulatory mutations. Nat Rev Genet. 2007, 8: 206-216.View ArticlePubMedGoogle Scholar
- Carroll JS, Meyer CA, Song J, Li W, Geistlinger TR, Eeckhoute J, Brodsky AS, Keeton EK, Fertuck KC, Hall GF: Genome-wide analysis of estrogen receptor binding sites. Nat Genet. 2006, 38: 1289-1297. 10.1038/ng1901.View ArticlePubMedGoogle Scholar
- Kim TH, Abdullaev ZK, Smith AD, Ching KA, Loukinov DI, Green RD, Zhang MQ, Lobanenkov VV, Ren B: Analysis of the vertebrate insulator protein CTCF-binding sites in the human genome. Cell. 2007, 128: 1231-1245. 10.1016/j.cell.2006.12.048.PubMed CentralView ArticlePubMedGoogle Scholar
- Robertson G, Hirst M, Bainbridge M, Bilenky M, Zhao Y, Zeng T, Euskirchen G, Bernier B, Varhol R, Delaney A: Genome-wide profiles of STAT1 DNA association using chromatin immunoprecipitation and massively parallel sequencing. Nat Methods. 2007, 4: 651-657. 10.1038/nmeth1068.View ArticlePubMedGoogle Scholar
- Lupien M, Eeckhoute J, Meyer CA, Wang Q, Zhang Y, Li W, Carroll JS, Liu XS, Brown M: FoxA1 translates epigenetic signatures into enhancer-driven lineage-specific transcription. Cell. 2008, 132: 958-970. 10.1016/j.cell.2008.01.018.PubMed CentralView ArticlePubMedGoogle Scholar
- Yu M, Riva L, Xie H, Schindler Y, Moran TB, Cheng Y, Yu D, Hardison R, Weiss MJ, Orkin SH: Insights into GATA-1-mediated gene activation versus repression via genome-wide chromatin occupancy analysis. MolCell. 2009, 36: 682-695.Google Scholar
- Fujiwara T, O’Geen H, Keles S, Blahnik K, Linnemann AK, Kang YA, Choi K, Farnham PJ, Bresnick EH: Discovering hematopoietic mechanisms through genome-wide analysis of GATA factor chromatin occupancy. Mol Cell. 2009, 36: 667-681. 10.1016/j.molcel.2009.11.001.PubMed CentralView ArticlePubMedGoogle Scholar
- Robertson AG, Bilenky M, Tam A, Zhao Y, Zeng T, Thiessen N, Cezard T, Fejes AP, Wederell ED, Cullum R: Genome-wide relationship between histone H3 lysine 4 mono- and tri-methylation and transcription factor binding. Genome Res. 2008, 18: 1906-1917. 10.1101/gr.078519.108.PubMed CentralView ArticlePubMedGoogle Scholar
- Wormald S, Hilton DJ, Smyth GK, Speed TP: Proximal genomic localization of STAT1 binding and regulated transcriptional activity. BMC Genomics. 2006, 7: 254-10.1186/1471-2164-7-254.PubMed CentralView ArticlePubMedGoogle Scholar
- Okuno Y, Huang G, Rosenbauer F, Evans EK, Radomska HS, Iwasaki H, Akashi K, Moreau-Gachelin F, Li Y, Zhang P: Potential autoregulation of transcription factor PU.1 by an upstream regulatory element. Mol Cell Biol. 2005, 25: 2832-2845. 10.1128/MCB.25.7.2832-2845.2005.PubMed CentralView ArticlePubMedGoogle Scholar
- Shi J, Yang W, Chen M, Du Y, Zhang J, Wang K: AMD, an automated motif discovery tool using stepwise refinement of gapped consensuses. PLoS One. 2011, 6: e24576-10.1371/journal.pone.0024576.PubMed CentralView ArticlePubMedGoogle Scholar
- Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, Noble WS: MEME SUITE: tools for motif discovery and searching. Nucleic Acids Res. 2009, 37: W202-W208. 10.1093/nar/gkp335.PubMed CentralView ArticlePubMedGoogle Scholar
- Consortium EP, Bernstein BE, Birney E, Dunham I, Green ED, Gunter C, Snyder M: An integrated encyclopedia of DNA elements in the human genome. Nature. 2012, 489: 57-74. 10.1038/nature11247.View ArticleGoogle Scholar
- Wang J, Zhuang J, Iyer S, Lin X, Whitfield TW, Greven MC, Pierce BG, Dong X, Kundaje A, Cheng Y: Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Res. 2012, 22: 1798-1812. 10.1101/gr.139105.112.PubMed CentralView ArticlePubMedGoogle Scholar
- Dekker J: Gene regulation in the third dimension. Science. 2008, 319: 1793-1794. 10.1126/science.1152850.PubMed CentralView ArticlePubMedGoogle Scholar
- Sanyal A, Lajoie BR, Jain G, Dekker J: The long-range interaction landscape of gene promoters. Nature. 2012, 489: 109-113. 10.1038/nature11279.PubMed CentralView ArticlePubMedGoogle Scholar
- Carroll JS, Liu XS, Brodsky AS, Li W, Meyer CA, Szary AJ, Eeckhoute J, Shao W, Hestermann EV, Geistlinger TR: Chromosome-wide mapping of estrogen receptor binding reveals long-range regulation requiring the forkhead protein FoxA1. Cell. 2005, 122: 33-43. 10.1016/j.cell.2005.05.008.View ArticlePubMedGoogle Scholar
- Lin Z, Reierstad S, Huang CC, Bulun SE: Novel estrogen receptor-alpha binding sites and estradiol target genes identified by chromatin immunoprecipitation cloning in breast cancer. Cancer Res. 2007, 67: 5017-5024. 10.1158/0008-5472.CAN-06-3696.View ArticlePubMedGoogle Scholar
- Yoon H, Boss JM: PU.1 binds to a distal regulatory element that is necessary for B cell-specific expression of CIITA. J Immunol. 2010, 184: 5018-5028. 10.4049/jimmunol.1000079.PubMed CentralView ArticlePubMedGoogle Scholar
- Badis G, Berger MF, Philippakis AA, Talukder S, Gehrke AR, Jaeger SA, Chan ET, Metzler G, Vedenko A, Chen X: Diversity and complexity in DNA recognition by transcription factors. Science. 2009, 324: 1720-1723. 10.1126/science.1162327.PubMed CentralView ArticlePubMedGoogle Scholar
- Rosenbauer F, Wagner K, Kutok JL, Iwasaki H, Le Beau MM, Okuno Y, Akashi K, Fiering S, Tenen DG: Acute myeloid leukemia induced by graded reduction of a lineage-specific transcription factor, PU.1. Nat Genet. 2004, 36: 624-630. 10.1038/ng1361.View ArticlePubMedGoogle Scholar
- Cawley S, Bekiranov S, Ng HH, Kapranov P, Sekinger EA, Kampa D, Piccolboni A, Sementchenko V, Cheng J, Williams AJ: Unbiased mapping of transcription factor binding sites along human chromosomes 21 and 22 points to widespread regulation of noncoding RNAs. Cell. 2004, 116: 499-509. 10.1016/S0092-8674(04)00127-8.View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.