Molecular mechanisms underlying SHP-1 gene expression Hing Wo Tsui 1 , Kathleen Hasselblatt 2 , Alberto Martin 3 , Samuel Chi-ho Mok 2 and Florence Wing Ling Tsui 1,4 1 Division of Cellular & Molecular Biology, Toronto Western Research Institute, University Health Network, Toronto, Ontario, Canada; 2 Laboratory of Gynecologic Oncology, Department of Obstetric Gynecology and Reproductive Biology, Brigham and Women’s Hospital, Dana-Farber Harvard Center, Boston, Massachusetts, USA; 3 Department of Cell Biology, Albert Einstein College of Medicine, Bronx, New York, USA; 4 Department of Immunology, University of Toronto, Toronto, Ontario, Canada SHP-1, a protein-tyrosine phosphatase with two src- homology 2 domains, is expressed predominantly in hematopoietic and epithelial cells and has been implicated in numerous signaling pathways as a negative regulator. Two promoters direct the expression of human and murine SHP-1, and two types of transcripts (I) and (II) SHP-1,are initiated from each of these promoters. The cDNA sequences of (I)SHP-1 and (II)SHP-1 are identical except in the 5¢ untranslated region and in the first few coding nucleotides. In this report, we show that promoter usage is similar in mouse and human hematopoietic cells, but different in epithelial cells. In human epithelial cells, only (I)SHP-1 transcripts were expressed. In addition, 4b-phorbol 12-myristate 13-acetate up-regulates human (I)SHP-1 transcript expression in SKOV3 cells (an ovarian cancer cell line). Indirect evidence suggests that nuclear factor-jB might play a role in this induction. We also show that a 12-bp repeat in the distal SHP-1 promoter, which directs (I)SHP-1 expression, is of functional relevance as deletion of one copy of this E-box-containing 12-bp repeat resulted in a significant decrease in promoter activity. Elec- trophoretic mobility shift assays and supershift experiments showed that the upstream stimulatory factors USF1 and USF2 hetero-dimerize and interact with this 12 bp repeat. Our results suggest that USFs which have antiproliferative functions might regulate the expression of SHP-1, which itself is predominantly a negative growth regulator. Keywords: cis elements; distal promoter; NFjB; promoter usage; USFs. Phosphorylation of proteins serves to alter their activity, providing a simple and mostly reversible change in molecular function. The regulation of tyrosine phosphory- lation is important in the control both of normal cellular processes including cell growth, cell cycle regulation, and differentiation, and of pathological events such as malignant transformation. Protein tyrosine kinases and phosphatases are the key players in the regulation of protein tyrosine phosphorylation. Among the known protein tyrosine phosphatases, SHP-1 and SHP-2 are distinguished by the presence of two tandem src-homology 2 domains. Src- homology 2 domains interact with phospho-tyrosine resi- dues in many growth factor receptors and thus play an important role in directing the effects of tyrosine phos- phorylation [1]. We [2] and others [3] showed that motheaten mice have mutations in the SHP-1 gene. These mutant mice thus provide insight into the role of SHP-1. Motheaten mice die prematurely and have characteristics of both immuno- deficiency and autoimmunity [4]. From analyses of moth- eaten mice and other work in cell lines, SHP-1 functions predominantly as a negative regulator in hematopoietic signaling pathways [5]. SHP-1 is expressed predominantly in hematopoietic and epithelial cells [6]. It has recently been shown that localiza- tion of SHP-1 differs between hematopoietic and epithelial cells (i.e. cytoplasmic in hematopoitic cells vs. nuclear in epithelial cells) [7]. Two promoters direct the expression of human [8] and murine SHP-1 [9], and two types of transcripts are initiated from the promoters. Transcripts that contain the 5¢-most exon [termed (I)SHP-1] encode SHP-1 with the initial amino acid sequences being MLSRG as compared to the MVR sequence encoded by transcripts that contain the 3¢ exon 1 [termed (II)SHP-1]. As there are minor to no enzymatic differences between (I) and (II) isoforms [9], we favor the view that different forms have arisen because of a need to regulate SHP-1 transcription using distinct promoters. Very little is known regarding the functionality of the two promoters and their usage in different cell types. In this study, we assessed the generation of (I) and (II) SHP-1 transcripts in various human and mouse cell lines and carried out functional deletional analyses of the distal promoter of human SHP-1 in epithelial cell lines. MATERIALS AND METHODS RT-PCR Reverse transcription of total RNA from cell lines, prepared by the Trizol (BRL) method, was carried out as described previously [2]. The primers (I)SHP-1-90-5¢ (5¢-AA CAGCTGTGCCACTCGATTG-3¢) and SHP-1-1859-3¢ (5¢-CCACAGGTCTCAGTCTATCGGGT-3¢); (II)SHP- 1-74-5¢ (5¢-GTGCCTGCCCAGACAAACTG-3¢)and SHP-1-1859-3¢ were used in RT-PCR of (I)SHP-1 and Correspondence to F. W. L. Tsui, Toronto Western Hospital, Mc14-417, 399 Bathurst Street, Toronto, Ontario M5T 2S8, Canada. Fax: + 1 416 603 5745, Tel.: + 1 416 603 5904, E-mail: ftsui@uhnres.utoronto.ca Abbreviations: bGal, b-galactosidase; EMSA, electrophoretic mobility shift assay; HPRT, hypoxanthine guanine phosphoribosil transferase gene; PMA, 4b-phorbol 12-myristate 13-acetate. (Received 24 October 2001, revised 10 April 2002, accepted 7 May 2002) Eur. J. Biochem. 269, 3057–3064 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02986.x (II)SHP-1 transcripts, respectively. Ten-fold serial dilution of the RT products were used to amplify either (I)SHP-1 or (II)SHP-1 transcripts. The RT-PCR products were separ- ated by electrophoresis through 0.7% agarose gels, blotted to nitrocellulose and probed with a 32 P-labeled DNA fragment containing sequences encoding the phosphatase domain of SHP-1 and autoradiographed. The intensity of bandswasmeasuredbydensitometryonanimager(Bio- Rad Fluor-S TM multi-imager). Electrophoretic mobility shift assay (EMSA) Nuclear extracts were prepared from cell lines [10]. Protein concentrations of the nuclear extracts were determined using the Coomassie protein assay reagent (Pierce). Five lg of nuclear extracts were mixed with herring DNA (BMC) and labeled oligonucleotide with and without competitor DNA (500-fold excess) in a buffer containing 25 m M Tris/HCl pH 7.5, 50 m M KCl, 0.6 m M dithiothre- itol, 1 m M EDTA, 0.5 m M spermidine, 12% glycerol for 20 min at room temperature. For supershift experiments, the nuclear extracts were incubated with 2 lgofthe antibody for 30 min at room temperature before adding the labeled oligonucleotide. The reaction was subjected to electrophoresis on a 6% native polyacrylamide gel in Tris/ glycine buffer (25 m M Tris pH 7.7, 200 m M glycine, 1 m M EDTA). The gels were dried and exposed to phospho- screens and the images were visualized on a phosphoi- mager (Bio-Rad). Quantitation of bands was carried out using the QUANTITY-ONE software. Antibodies to the upstream stimulatory factors USF1, USF2, Max, NFjB p50 were from Santa Cruz Biotech. Inc.; antibody to NFjB p65 (RelA) was from Upstate Biotechnology. Generation of reporter constructs An 845-bp BamHI–PvuII DNA segment containing the (I)SHP-1 promoter and 83 bp of the 5¢ UTR [without the (I)SHP-1 AUG] was isolated from a SHP-1-containing cosmid clone (LL12NCOIN 143H6, a gift from P. Mary- nen, Leuven, Belgium) and inserted upstream of the luciferase reporter gene in the pGL2-Basic vector (Promega) (construct A). For generation of construct B, a 420-bp KpnI segment was removed from construct A, followed by re-circularization of the vector. For generation of construct C, a 610-bp SmaI segment was removed from construct A, followed by re-circularization of the vector. In the 12 bp repeat sequences, there is an SstI site in each of the 12-bp segments. By deleting the 12 bp SstI–SstIfragmentineither construct A or B and re-circularizing the constructs, one copy of the 12-bp repeat was removed from each of the two constructs to form constructs A)12 bp and B)12 bp. Analysis of promoter function SKOV3 cells were cotransfected with the (I)SHP-1 pro- moter–luciferase constructs and pSV 2 bGal (pCH110) by lipofection using Fugene 6 (Roche Molecular Biochem). Forty-eight hours after transfection, fractions of each cell extract were used for the b-galactosidase (bGal) [11] and luciferase [12] assays. The conditions used for the luciferase assay were within the linear range of the assay for the promoters tested in this study. Each construct was tested in three different transfection experiments, with triplicates for each experiment. RESULTS Differential expression of SHP-1 isoform transcripts in human vs. murine cell lines In mouse as well as in human, SHP-1 proteins are detected in both hematopoietic and epithelial cells. As the two SHP-1 protein isoforms only differ in the first few amino acids, it is difficult to distinguish the two protein isoforms. Thus, it is unclear whether the SHP-1 proteins are translated from the (I)SHP-1 or (II)SHP-1 transcripts or both. To assess whether both SHP-1 promoters are transcriptionally active in hematopoietic and epithelial cells, we used RT-PCR to specifically amplify either the (I)SHP-1 or (II)SHP-1 transcripts. Expression of (I)SHP-1 and (II)SHP-1 tran- scripts were assessed using the primers (I)SHP-1-90-5¢ and SHP-1-1859–3¢ vs. (II)SHP-1-74-5¢ and SHP-1-1859–3¢, respectively. We determined the relative abundance of the two isoform transcripts using a quantitative RT-PCR assay. We previously [13] showed that the (I)SHP-1 and (II)SHP-1 isoforms were amplified to a similar extent using isoform specific primers, as mentioned above. As explained in the legend to Fig. 1, a serial dilution of the initial RT reaction mixture was used to amplify type (I)SHP-1 and (II)SHP-1 cDNAs. Blots of the electro- phoresed RT-PCR products were probed with P 32 -labeled sequences of the SHP-1 phosphatase domain, and the intensity of the autoradiographed bands was measured by densitometry. We analysed six human hematopoietic cell lines (K562, Raji, HL60, BL-JC, CEM and U937) and six human epithelial cell lines (HeLa, CAOV3, SKOV3, MDA453, Calu 1 and HT1080) as well as eight mouse hematopoietic cell lines (BW5147, M1, NFS-5C1, J774, IC21, 70Z/3, J558L and A20) and four mouse epithelial cells (Y1, L cells, LA-4 and MMT060562) (Table 1). In both human [8] and mouse SHP-1 [9], alternative transcripts (both longer and shorter than the major transcripts) have been reported and except for one human splice variant [14], most of these variant transcripts contain premature stop codons ([9] and our unpublished results for the human variants) and therefore cannot be translated into functional phosphatases. Thus, in this study, we only quantified the major transcripts. Fig. 1 shows representative profiles of SHP-1 isoform transcripts expressed in both human and mouse cell lines and the relative abundance of these isoforms are summarized in Table 1. In most human (4/6) and mouse (6/8) hematopoietic cell lines, both SHP-1 isoform transcripts were detected. However, some cell lines expressed only one of the two isoforms (Table 1). Of the cell lines that expressed both isoforms, the ratio of (II)SHP-1 to (I)SHP-1 transcripts ranged from 0.3 : 1 to 63 : 1 (human) and 28 : 1 to 110 : 1 (mouse). Similarly, in mouse epithelial cell lines (3/4), both isoform transcripts were present, although the ratio of (II)SHP-1 to (I)SHP-1, which ranged from 1.3 : 1 to 2 : 1 is much lower than that found in hematopoietic cells. However, in human epithelial cell lines (5/6), only (I)SHP-1 transcripts were detected. As all the cell lines used for this study are transformed, we asked whether the SHP-1 promoter usage is similar in untransformed hematopoietic cells. For human, we used 3058 H. W. Tsui et al. (Eur. J. Biochem. 269) Ó FEBS 2002 tonsillar T cells grown in the presence of phytohemagglu- tinin (TON-phytohemagglutinin) and for mouse, we used splenic T cells as well as thymus. In all three cases, both (II)SHP-1 and (I)SHP-1 transcripts were detected, with the former isoform being the predominant species (Table 1). (I)SHP-1 transcripts were up-regulated by 4b-phorbol 12-myristate 13-acetate (PMA) in HL60 and SKOV3 cells As human epithelial cells expressed only (I)SHP-1 tran- scripts, these cells (such as SKOV3, an ovarian cancer cell line) are ideal for the study of the distal promoter function of SHP-1. We first wished to identify agent(s) that can modulate the expression of (I)SHP-1 transcripts. Nuclear run-on experiments showed that PMA treatment increased SHP-1 transcription in HL60 cells [15].However, it is unclear which promoter is responsible for the increase in SHP-1 transcription. We used RT-PCR to assess the relative abundance of (I)SHP-1 and (II)SHP-1 transcripts [nor- malized to hypoxanthine guanine phosphoribosil transferase (HPRT)] in untreated vs. PMA treated (48 and 72 h) HL60 cells. We observed that the relative levels of (I)SHP-1 and (II) SHP-1 transcripts were up-regulated 48-fold and fivefold, respectively, when the cells were treated with PMA (Fig. 2A). SHP-1 proteins were increased  5-fold in PMA treated HL60 cells (data not shown). Because HL60 is a hematopoietic cell line, we asked whether PMA induces a similar effect in epithelial cells. We treated SKOV3 cells, which expressed only (I)SHP-1 transcripts, with PMA and compared the relative abundance of this isoform transcript in treated vs. untreated cells after normalization with HPRT. We found a lower but significant increase in the relative level of (I)SHP-1 transcripts (twofold to fourfold) in PMA treated SKOV3 cells (Fig. 2B). Thus, in both hematopoietic and epithelial cells, PMA can up-regulate the expression of (I)SHP-1 transcript. Role of NFjB in the expression of (I)SHP-1 transcripts As we found that PMA up-regulates the expression of human (I)SHP-1 transcripts (Fig. 2), we were interested in identifying potential activator(s) of the human distal SHP-1 promoter. PMA is a known nonphysiological activator of NFjB. In the distal promoter of human SHP-1,thereisa putative NFjBsiteat)314 (GGGATTTTCC). We first asked whether NFjB proteins can bind to this putative NFjB consensus sequence. We carried out EMSAs using SKOV3 nuclear extracts and a double-stranded oligonucle- otide containing this consensus sequence as a probe. We detected two specific DNA–protein complexes (Fig. 3, lane 2), both of which can be super-shifted using anti-NFjBIg (p50) and anti-NFjB Ig (p65) (Fig. 3, lanes 3 and 4). If (I)SHP-1 transcription is increased because PMA activated NFjB, we would expect to find more NFjB binding to this NFjB site located in the distal SHP-1 promoter. We thus carried out EMSA using equal amounts of untreated and PMA-treated SKOV3 nuclear extracts. As expected, we found that nuclear extracts from PMA treated SKOV3 cells had a 4–5-fold higher NFjB activity than those from untreated cells (Fig. 3, compare lane 8 with lane 6). These data suggest that the up-regulation of (I)SHP-1 transcrip- tion by PMA is mediated via the NFjB site in the distal promoter of SHP-1. Fig. 1. Relative abundance of (I)SHP-1 and (II)SHP-1 transcripts in human vs. mouse cell lines. Raji is a Burkitt’s Lymphoma cell line (i.e. hematopoietic); HeLa and HT1080 are human epithelial cancer cell lines. BW5147 is a mouse T-cell line and L cell are a mouse epithelial cell line. RNA from the cell lines were reverse transcribed, and serial dilutions (shown below each lane) of the RT mixture were used in PCR for (I)SHP-1 with primer pair (I)SHP-1-90-5¢ and SHP-1-1859-3¢,or(II)SHP-1 with primer pair (II)SHP-1-74-5¢ and SHP-1-1859-3¢. The RT-PCR products were separated by electrophoresis, transferred to nitrocellulose and probed with 32 P-labeled sequences of the phosphatase domain for SHP-1. Arrows denote the SHP-1 transcripts which are translatable into proteins [9,13]. Densitometry was performed on this species of SHP-1 transcripts. Bottom panels: Schematics showing the generation of (I)SHP-1 vs. (II)SHP-1 transcripts from the SHP-1 gene. Ó FEBS 2002 Regulation of SHP-1 expression (Eur. J. Biochem. 269) 3059 Table 1. Relative abundance of (II) vs. (I)SH P-1 transcripts in human (A) and mouse (B) cell lines and untransformed cells. (II)SHP-1 (I)SHP-1 Ratio(II) : (I)SHP-1 Human cells Epithelial cell lines HeLa 0 10 CAOV3 0 36 Calu I 0 40 SKOV3 0 8 MDA453 0 60 HT1080 3 0 Hematopoietic cell lines K562 1 3 0.3 : 1 Raji 100 1.6 63 : 1 HL60 167 9 19 : 1 BL-JC 206 43 5 : 1 CEM 346 0 U937 971 0 Hematopoietic cells TON-photohaemagglutinin 193 69 3 : 1 Mouse cells Epithelial cells lines Y1 3 2.2 1.4 : 1 L cells 8 4 2 : 1 LA-4 7 5.6 1.3 : 1 MMT060562 15 0 Hematopoietic cell lines BW5147 110 1 110 : 1 M1 229 3 76 : 1 NFS-5C1 195 4.4 44 : 1 J774 666 24 28 : 1 IC21 977 24 41 : 1 70Z/3 1320 27 49 : 1 J558L 140 0 A20 520 0 Hematopoietic cells Splenic T cells 82 1.2 68 : 1 Thymus 629 11 60 : 1 Fig. 2. Up-regulation of SHP -1 expression in PMA treated HL60 (A) and SKOV3 (B) cells. Relative abundance of (I)SHP-1 and (II)SHP-1 transcripts in untreated or PMA treated cells was estimated by quantitative RT-PCR (as in Fig. 1). 3060 H. W. Tsui et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Functional deletional analyses of the distal promoter of human SHP-1 in epithelial cells To characterize further the distal promoter, we needed to obtain a genomic segment containing the distal promoter. From a human SHP-1-containing cosmid clone (LL12NCOIN 143H6), we isolated the 5¢ flanking region upstream of the first exon of (I)SHP-1. We sequenced the region 986 bp upstream of the transcription initiation site, and the sequence was identical to the published one [8]. To test the functionality of the human SHP-1 distal promoter, we generated three deletion constructs (A, B and C) which were adjoined to a luciferase reporter gene (Fig. 4). These constructs contained different amounts of 5¢ flanking DNA and lacked the (I)SHP-1 AUG. They were individually transfected into SKOV3 cells, and lysates were assayed for luciferase activities. A bGal construct was cotransfected with each deletion construct and bGal activities were used to normalize the efficiency of each transfection. The promo- terless vector, pGL2-basic (D), was included as a negative control. The pGL2-control vector with both the SV40 promoter and enhancer (E) was included as a positive control. As shown in Fig. 4, maximal (I)SHP-1 activity was observed with construct A [845 bp of the (I)SHP-1 5¢ flanking region including 83 bp of 5¢UTR]. Deletion construct B (425 bp 5¢ flanking region) and C (235 bp 5¢ flanking region) produced less luciferase activities in SKOV3 transfections (47% and 30% of construct A, respectively). Identification of an activator(s) that binds to a 12-bp repeat Located in both constructs A and B, about 190 bp upstream of the distal SHP-1 initiation site, is a 12-bp repeat. As direct repeats in promoter regions usually represent important regulatory elements, we asked whe- ther this 12-bp repeat contributes to (I)SHP-1 promoter activity. We generated two additional constructs: construct A)12 bp differs from construct A by 12 bp (one copy of the 12 bp repeat was deleted from construct A) and likewise construct B)12 bp differs from construct B by the same 12 bp. Transfection studies using both sets of constructs (A vs. A)12 bp and B vs. B)12 bp; Fig. 4) showed that deletion of one copy of the 12-bp sequences in both cases resulted in a significant decrease in the luciferase activity. Construct A)12 bp had 75% of construct A activity, and construct B)12 bp had only 16% of construct B activity (Fig. 4) suggesting that an activator(s) binds to this 12-bp repeat. The reasons for a much larger effect on construct B will be considered in the Discussion. Fig. 3. EMSA and supershift analyses. The NFjBsite(TGTTAGG GATTTCCTTA) from (I)SHP-1 promoter was used as a probe. Lanes: 1 and 5, no nuclear extracts present in the reaction mix; lanes 2 and6,twospecificcomplexes(AandB)formedwhenthereactionmix contains both nuclear extracts and labeled probe. The lowest shifted band is nonspecific, as it cannot be competed out with excess unlabeled oligonucleotide in the reaction mix (lane 7); Both complexes A and B were supershifted when either anti-NFjB Ig, p50 or p65 (Rel A) were included in the reaction mix; lane 8, more complexes A and B were formed when nuclear extracts from PMA treated SKOV3 cells were used. Fig. 4. Schematic of the (I)SHP-1 deletion constructs and luciferase activities of these constructs in SKOV3 cells. Constructs A, B and C contain various lengths of (I)SHP-1 promoter region. Construct D is promoterless and was used as a negative control. Construct E is a luciferase construct driven by the SV40 promoter and enhancer; it served as a positive control. Both copies of the 12-bp repeat are present in constructs A and B, while only one copy of the repeat is present in either construct A)12 bp or B)12 bp. K, KpnI; S, SstI. Ó FEBS 2002 Regulation of SHP-1 expression (Eur. J. Biochem. 269) 3061 USF1 and USF2 bind to the 12-bp repeat in the (I)SHP-1 promoter We were interested in identifying the nuclear factor(s) that bind to the 12-bp repeat and activates (I)SHP-1 expression. Within the 12-bp sequences, there is an E-box (GAG CTCCAGGTG). Using this 12-bp repeat as a probe for binding factors in nuclear extracts from SKOV3 (an ovarian cancer cell line) and MDA453 (a breast cancer cell line) (Fig. 5, lanes 2 and 7) for EMSA, we detected several shifted bands. One band (U) was completely inhibited with 500- fold excess of the same unlabeled probe (Fig. 5, lanes 1 and 6) but not by an excess amounts of mutated oligonucleotide (GAGCTCCAGGGA; Fig. 5, lane 5), indicating that the protein complex binds to the E-box sequences in the 12-bp repeat. Two other shifted bands (a doublet X, and Y) were only partially inhibited in the presence of 500-fold excess of unlabeled probe (Fig. 5, lanes 1 and 6), and were not detectably competed with excess mutated oligonucleotide (Fig. 5, lane 5). These findings indicate that proteins in the X and Y complexes also have specificity to the E-box within the 12-bp repeat sequences. c-Myc and Max proteins are known E-box binding proteins [16]. We therefore tested whether antibody to Max can supershift the protein complex. We first used a known Myc–Max consensus probe and Ramos (a Burkitt’s Lym- phoma cell line) nuclear extract to check whether the anti- Max Ig can be used for supershift experiments. We detected two protein complexes, one of which can be supershifted by the anti-Max Ig (Fig. 5, lane 12). However, the same anti- Max Ig failed to supershift the protein complexes formed using the wild-type 12-bp repeat oligonucleotide and nuclear extracts from both SKOV3 and MDA453 cells (Fig. 5, lanes 2 and 7). As Myc hetero-dimerizes with Max, the inability of anti-Max Ig to supershift the complex would imply that Myc, like Max, does not bind to the E-box sequences in the 12-bp repeat. USFs are also known E-box binding proteins [17]. We therefore asked whether the protein complexes formed, contain USF1 and/or USF2 using the 12-bp repeat oligonucleotide and nuclear extracts from SKOV3 and MDA453 cells. As shown in Fig. 5 (lanes 3, 4, 8 and 9), one of the protein complexes was supershifted using either anti- USF1 Ig or anti-USF2 Ig. Therefore, both USF1 and USF2 proteins form a stable complexes with the 12 bp repeat. We have not identified the proteins involved in the formation of complexes X and Y. DISCUSSION Differential usage of SHP1 promoters in mouse vs. human epithelial cell lines A previous report [8] showed that a few human hemato- poietic cell lines expressed only (II)SHP-1 transcripts. Contrary to their finding that HL60 cells expressed only (II)SHP-1 transcripts, we found that HL60 cells not only express (I)SHP-1 transcript, but also can be stimulated by PMA to express up to 48-fold more (I)SHP-1 mRNA. In addition, we found that most human (5/7) and mouse (8/10) hematopoietic cells, expressed both SHP-1 transcript isoforms, albeit with (II)SHP-1 transcripts being the predominant species. In mouse hematopoietic cell lines (II)SHP-1 transcripts were always much more abundant than (I)SHP-1 transcripts. However, the relative difference between (II)SHP-1 and (I)SHP-1 transcripts was less pronounced in human hematopoietic cell lines. In both human and mouse, similar ratios were found in untrans- formed vs. transformed hematopoietic cells. The relative abundance of the SHP-1 transcript isoforms in epithelial cells was different from that of hematopoietic cells. In mouse epithelial cell lines, both SHP-1 transcript isoforms are of similar abundance. However, no (II)SHP-1 transcripts were detected in most human epithelial cell lines. Thus, the control of SHP-1 promoters appears to be different in mouse vs. human epithelial cell lines. It is not clear whether this species difference is due to cis-elements or trans-activating factors that regulate the SHP-1 promoters. It has recently been shown that in human epithelial cells (such as HeLa, A549 and MCF-7), SHP-1 proteins were Fig. 5. EMSA and supershift analyses. Either the 12-bp repeat from (I)SHP-1 promoter (TTGAGCTCCAGGTGGAGCTCCAG GTG; E-box consensus sequences are in bold) or a Myc–Max consensus (TTAAGCA GACCAC GTGGTCTGCAACC) was used as a probe. Nuclear extracts from SKOV3 (an ovarian cancer cell line) or MDA453 (a breast cancer cell line) or Ramos (a Burkitt’s Lym- phoma cell line) were used. To show specificity of the shifted bands, a 500-fold excess of either cold 12-bp repeat oligonucleotide (lanes 2 and 6) or cold 12-bp repeat mutant oligonu- cleotide (TTGAGCTCCA GGGAGAG CTCCAGGGA; lane 5) was included in the reaction mix for EMSA. 3062 H. W. Tsui et al. (Eur. J. Biochem. 269) Ó FEBS 2002 localized in the nuclei [7]. As we showed that only (I)SHP-1 transcripts were expressed in human epithelial cells, it appears that SHP-1 proteins derived from human (I)SHP-1 transcript are localized in the nuclei and thus might have different signaling substrates compared to that of the cytoplasmic (II)SHP-1 proteins. In support of this notion, tyrosine-phosphorylated stat-5b and SHP-1 com- plex has been detected in the nuclei of growth hormone stimulated liver cells in culture [18]. Activators of the distal promoter of human SHP-1 Our deletional analyses of the distal promoter of SHP-1 (in an ovarian cancer cell line, SKOV3) showed less promoter activity with sequential deletion of the 5¢ flanking region. This suggests that the distal promoter of SHP-1 is regulated by multiple activators. Indeed, we found two motifs within the distal promoter that were important for promoter activity. One such motif was an E-box containing a 12-bp repeat. Deletion of one copy of the repeat resulted in significantly lower promoter activity (Fig. 4). The additional region I (420 bp) in construct A presumably contains redundant regulatory elements, thus masking the contribu- tion of the 12-bp repeats in the comparison of construct A vs. construct A)12 bp activities. It appears that the two tandem E-boxes separated by 6 bp are crucial for presum- ably high affinity binding of the activator(s) involved. EMSA and supershift experiments showed that USF1 and USF2 hetero-dimerize and interact with this 12-bp repeat. USFs are thought to have anti-proliferative functions as their over-expression inhibited growth of numerous cancer cell lines [19]. As SHP-1 is predominantly a negative regulator of growth, it is possible that USFs mediate their anti-proliferative functions via the regulation of SHP-1 expression. To confirm whether USF proteins bind to the 12 bp repeat in the (I)SHP-1 promoter, in vivo binding of USF proteins can be assessed by formaldehyde cross-linking followed by chromatin immunoprecipitation and PCR amplification of the (I)SHP-1 promoter. In addition, it will be of interest to assess whether cotransfection of USF dominant negative mutants and a (I)SHP-1 promoter– luciferase construct would down-regulate luciferase activity in SKOV3 cells. In our EMSA analyses, aside from the shifted band that contained USF1 and USF2, we observed other shifted bands (X and Y). As bands X and Y were only partially inhibited by 500-fold excess of unlabelled wild-type oligo- nucleotide, we propose that the proteins involved in these complexes have very low ÔonÕ rates, resulting in an ineffi- cient, albeit stable binding to the oligonucleotides. Our finding that an oligonucleotide bearing a mutated E-box competed less than the wild-type oligonucleotide suggests that the proteins involved in the X and Y complexes recognize sequences in the E-box. However, we have not identified the proteins involved in the formation of complexes X and Y. The second motif in the distal promoter of SHP-1 which might contribute to the regulation of SHP-1 expression is a NFjB site located 105 bp upstream of the E-box containing 12-bp repeat. EMSA and supershift experiments show that NFjB p50 and p65 bind this NFjB consensus sequence (GGGATTTTCC). It was previously shown that PMA treatment of HL60 cells increased SHP-1 transcription [15]. We found that (I)SHP-1 transcripts were upregulated by PMA in HL60 and SKOV3 cells. Furthermore, PMA- treated SKOV3 nuclear extracts showed more NFjB binding activity (fourfold to fivefold; Fig. 3) than those from untreated cells. Thus, it is likely that PMA activates NFjB proteins which in turn leads to higher expression of (I)SHP-1 transcripts. Confirmation of this result can be achieved by deleting the NFjB site in the promoter construct and assessing whether this will render transfected cells unresponsive to PMA. Our analyses of the deletion constructs transfected into SKOV3 cells indicated that deletion of the region I (the 5¢ 420 bp sequences, Fig. 4) from the promoter construct (construct B) resulted in a 54% reduction of luciferase activity. Interestingly, no consensus sequences for known nuclear factors are found in region I. This result indicates that there might be novel nuclear factors (activators) which contribute to (I)SHP-1 promoter activity. Contrary to our results (i.e. progressively less promoter activity with sequential deletion of the 5¢ flanking region), a recent deletional study of the same distal promoter in MCF7 cells (a breast cancer cell line), showed a dramatic drop of promoter activity to  15% using a deletion construct with 5¢ flanking sequences up to 60 bp upstream of the 12-bp repeat [20]. It is possible that the vast difference in the level of SHP-1 transcripts expressed in SKOV3 vs. MCF7 cells might account for the discrepancy in the results between the two deletional analyses. We found that MCF7 expressed at least 10-fold more SHP-1 transcripts and SHP- 1 proteins than SKOV3 cells (unpublished data). It has also been reported that SHP-1 was up-regulated in MCF7 cells, as in human breast cancers [21]. We showed previously that SKOV3 expressed SHP-1 levels similar to normal ovarian epithelial cells [22] and thus our deletional analysis in SKOV3 might reflect a more physiological (and not pathological) situation of SHP-1 expression. Although we favor the above explanation, we cannot rule out the possibility that the control of SHP-1 expression might differ in ovarian vs. mammary cells. ACKNOWLEDGEMENTS This work was funded by the National Cancer Institute of Canada. We thank Dr P. Marynen for the generous gift of the SHP-1 containing cosmid clone, and Dr M. Shulman for a critical review of the manuscript. REFERENCES 1. Koch, C.A., Anderson, D., Moran, M.F., Ellis, C. & Pawson, T. (1991) SH2 and SH3 domains: elements that control interactions of cytoplasmic signaling protein. Science 252, 668–672. 2. Tsui, H.W., Siminovitch, K.A., de Souza, L. & Tsui, F.W.L. (1993) Motheaten and viable motheaten mice have mutations in the haematopoietic cell phosphatase gene. Nat. Genet. 4, 124–129. 3. Shultz, L.D., Schweitzer, P.A., Rajan, T.V., Yi, T., Ihle, J.N., Matthews, R.J., Thomas, M.L. & Beier, D.R. (1993) Mutations at the murine motheaten locus are within the heamtopoietic cell protein-tyrosine phosphatase (Hcph) gene. Cell 73, 1445–1454. 4. Shultz, L.D., Coman, D.R., Bailey, C.L., Beamer, W.G. & Sidman, C.L. (1984) ÔViable motheatenÕ, a new allele at the moth- eaten locus. Am. J. Pathol. 116, 179–192. 5. Neel, B.G. (1997) Role of phosphatases in lymphocyte activation. Curr. Opin. Immunol. 9, 405–420. Ó FEBS 2002 Regulation of SHP-1 expression (Eur. J. Biochem. 269) 3063 6. Plutzky, J., Neel, B.G. & Rosenberg, R.D. (1992) Isolation of a src homology 2-containing tyrosine phosphatase. Proc. Natl Acad. Sci. USA 89, 123–1127. 7. Cragg, G. & Kellie, S. (2001) A functional nuclear localization sequence in the C-terminal domain of SHP-1. J. Biol. Chem. 276, 23719–23725. 8. Banville, D., Stocco, R. & Shen, S.H. (1995) Human protein tyrosine phosphatase 1C (PTPN6) gene structure: alternate pro- moter usage and exon skipping generate multiple transcripts. Genomics 27, 165–173. 9. Martin, A., Tsui, H.W., Shulman, M.J., Isenman, D. & Tsui, F.W.L. (1999) Murine SHP-1 splice variants with altered src homology 2 (SH2) domains: implications for the SH2-mediated intramolecular regulation of SHP-1. J. Biol. Chem. 274, 21725–21734. 10. Dignam, J.D., Lebovitz, T.M. & Roeder, R.G. (1983) Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res. 11, 1475–1489. 11. Rosenthal, N. (1987) Identification of regulatory elements of cloned genes with functional assays. In Methods in Enzy- mology Vol. 152 (Berger, S.L. & Kimmel, A.R., eds) pp. 704–720. 12. Alam, J. & Cook, J.L. (1990) Reporter genes: application to the study of mammalian gene transcription. Anal. Biochem. 188, 245–254. 13. Martin, A., Tsui, H.W. & Tsui, F.W.L. (1999) SHP-1 variant proteins are absent in motheaten mice despite presence of splice variant transcripts with open reading frames. Mol. Immunol. 36, 1029–1041. 14. Jin, Y.J., Yu, C.L. & Burakoff, S.J. (1999) Human 70-kDa SHP-1L differs from 68-kDa SHP-1 in its C-terminal structure and catalytic activity. J. Biol. Chem. 274, 28301–28307. 15. Uchida, T., Matozaki, T., Matsuda, K., Suzuki, T., Matozaki, S., Nakano, O., Eada, K., Konda, Y., Sakamoto, C. & Kasuga, M. (1993) Phorbol ester stimulates the activity of a protein tyrosine phosphatase containing SH2 domains (PTP1C) in HL-60 leuke- mia cells by increasing gene expression. J. Biol. Chem. 268, 11845– 11850. 16. Cole, M.D. & McMahon, S.B. (1999) The myc oncoprotein: a critical evaluation of transactivation and target gene regulation. Oncogene 18, 2916–2924. 17. Gregor, P.D., Sawadogo, M. & Roeder, R.G. (1990) The adeno- virus major late transcription factor USF is a member of the helix- loop-helix group of regulatory proteins and binds to DNA as a dimer. Genes Dev. 4, 1730–1740. 18. Ram, P.A. & Waxman, D.J. (1997) Interaction of growth hor- mone-activated STATs with SH2-containing phosphotyrosine phosphatase SHP-1 and nuclear JAK2 tyrosine kinase. J. Biol. Chem. 272, 17694–17702. 19. Luo, X. & Sawadogo, M. (1996) Antiproliferative properties of the USF family of transcription factors. Proc. Natl Acad. Sci. USA 93, 1308–1313. 20. Xu, Y., Banville, D., Zhao, H F., Zhao, X. & Shen, S H. (2001) Transcriptional activity of the SHP-1 gene in MCF7 cells is dif- ferentially regulated by binding of NF-Y factor to two distinct CCAAT-elements. Gene 269, 141–153. 21. Yip, S.S., Crew, A.J., Gee, J.M.W., Hui, R., Blamey, R.W., Robertson, J.F.R., Nicholson, R.I., Sutherland, R.L. & Daly, R.J. (2000) Up-regulation of the protein tyrosine phosphtase SHP-1 in human breast cancer and correlation with grb2 expression. Int. J. Cancer. 88, 363–368. 22. Mok, S.C., Kwok, T.T., Berkowitz, R.S., Barrett, A.J. & Tsui, F.W.L. (1995) Overexpression of the protein tyrosine phospha- tase, nonreceptor type 6 (PTPN6), in human epithelial ovarian cancer. Gynecol. Oncol. 57, 299–303. 3064 H. W. Tsui et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . to specifically amplify either the (I )SHP-1 or (II )SHP-1 transcripts. Expression of (I )SHP-1 and (II )SHP-1 tran- scripts were assessed using the primers (I )SHP-1- 90-5¢. activity of the SHP-1 gene in MCF7 cells is dif- ferentially regulated by binding of NF -Y factor to two distinct CCAAT-elements. Gene 269, 141–153. 21. Yip,