UXT interacts with the transcriptional repressor protein EVI1 and suppresses cell transformation

Authors


C. Bartholomew, Glasgow Caledonian University, Department of Biological & Biomedical Sciences, City Campus, Cowcaddens Road, Glasgow G4 OBA, UK
Fax: +44 (0)141 331 3208
Tel: +44 (0)141 331 3213
E-mail: c.bartholomew@gcal.ac.uk

Abstract

The EVI1 transcriptional repressor is critical to the normal development of a variety of tissues and participates in the progression of acute myeloid leukaemias. The repressor domain (Rp) was used to screen an adult human kidney yeast two-hybrid library and a novel binding partner designated ubiquitously expressed transcript (UXT) was isolated. Enforced expression of UXT in Evi1-expressing Rat1 fibroblasts suppresses cell transformation and UXT may therefore be a negative regulator of Evi1 biological activity. The Rp-binding site for UXT was determined and non-UXT-binding Evi1 mutants (Evi1Δ706–707) were developed which retain the ability to bind the corepressor mCtBP2. Evi1Δ706–707 transforms Rat1 fibroblasts, showing that the interaction is not essential for Evi1-mediated cell transformation. However, Evi1Δ706–707 produces an increased proportion of large colonies relative to wild-type, showing that endogenous UXT has an inhibitory effect on Evi1 biological activity. Exogenous UXT still suppresses Evi1Δ706–707-mediated cell transformation, indicating that it inhibits cell proliferation and/or survival by both Evi1-dependent and Evi1-independent mechanisms. These observations are consistent with the growth-suppressive function attributed to UXT in human prostate cancer. Our results show that UXT suppresses cell transformation and might mediate this function by interaction and inhibition of the biological activity of cell proliferation and survival stimulatory factors like Evi1.

Abbreviations
CtBP

C-terminal binding protein

DBD

DNA-binding domain

GST

glutathione S-transferase

Rp

repressor domain

SD

synthetic dropout medium

UXT

ubiquitously expressed transcript

EVI1 is a member of the PR domain family of proteins [1], which include PRDM1 (PRDI-BF1) [2], PRDM2 (RIZ) [3] and PRDM16 (MEL1) [4], and share structural similarities with N-terminal PR domains, functional similarities with transcriptional repressor activities, have multiple cys2/his2 zinc-finger motifs, and roles in cell differentiation and tumorigenesis. The complex phenotype observed in Evi1 knockout mice suggests that it has a role in a range of biological processes including general cell proliferation, vascularization and cell-specific developmental signalling [5]. The pleiotropic phenotype reflects Evi1's functions as a transcription factor [6] as well as its ability to interfere with several signalling pathways through interactions with Smad [7,8] and JNK [9] proteins.

The Evi1 transcriptional repressor protein is essential for normal development [5] and when inappropriately expressed participates in the progression of a subset of leukaemias and myelodysplasias [10]. In vitro studies have shown that a number of biological properties can be attributed to the Evi1 protein including: (a) deregulation of cell proliferation [11]; (b) inhibition of transforming growth factor-β, bone morphogenic protein and activin signalling [7,8]; and (c) inhibition of stress-induced apoptosis [9]. These activities require the DNA-binding domains ZF1/ZF2 [12,13] and a 200-amino acid transcriptional repressor domain (Rp) [6].

Rp is critical to the biological activity of Evi1. We, and others, have shown that Rp binds the C-terminal binding protein (CtBP) family of corepressor proteins that interact via two conserved PLDLS-like motifs [14,15]. Critically, the ability of Evi1 to repress transcription, deregulate cell growth [14] and inhibit transforming growth factor-β, bone morphogenic protein and activin signalling [7,8,15], all rely on the recruitment of CtBP.

Several lines of evidence suggest that Rp is involved in additional protein interactions that may also contribute to Evi1's broad spectrum of biological activities. C-terminal deletion mutants of Rp, which retain CtBP-binding sites, create more effective repressors [6], suggesting that this activity is regulated. Naturally occurring splice variants in both human and murine Evi1 exist that insert nine amino acids (FP/QLPDQRTW) into Rp [16,17], which may either create or disrupt novel or pre-existing interactions, respectively. Inspection of aligned human and murine Rp primary amino-acid sequences with the corresponding region of the related MEL1 protein [4] reveal significant stretches of conservation in addition to the CtBP-binding sites, suggesting additional common activities might have been conserved during evolution for these two proteins.

To date, several Evi1-binding proteins have been identified using a candidate protein approach. It remains possible that Evi1 interacts with as yet unknown cellular proteins responsible for regulating biological activity. To investigate this, we screened a yeast two-hybrid library to identify new Rp-binding partners.

Results

Isolation of a new Evi1 Rp-domain binding protein

Rp was subcloned from pGBT9Rp [14] into the kanamycin-selectable yeast vector pKGI [18] to create pKGIRp (see Experimental procedures). As expected, pKGIRp produces a GAL4 DNA-binding domain (DBD)/Rp protein which interacts with mCtBP2 in yeast AH109 cells, confirming its suitability for use in screening a yeast two-hybrid library (Fig. 1A–C; pKGIRp, pGAD10mCtBP2).

Figure 1.

 Interaction of Rp and A1 in yeast cells. (A,D) Yeast AH109 cells (Clontech) were transformed with the combination of plasmids shown. The growth of single yeast colonies containing these plasmids are shown in (B) and (E) on SD lacking l-histidine and adenine and (C) on SD plus l-histidine and adenine.

In total, 1 × 106 independent clones from a human adult kidney yeast two-hybrid library (Clontech) were screened with pKGIRp in AH109 cells. Initially, 37 potentially interacting clones were identified of which only 17 grew under more stringent conditions. Target plasmid DNAs recovered from these clones were introduced into AH109 cells with pKGIRp and growth was reassessed. Six recombinant plasmids contained genes encoding putative Rp-interacting proteins (Table 1, secondary screen). Sequencing of the plasmid DNA inserts revealed that five were HuCtBP2 (A6, A7, A10, A14 and A37). The remaining clone, A1, contained a novel gene. This gene was isolated repeatedly from additional independent library screens (data not shown) and together with HuCtBP2 represents the only Rp-interacting proteins identified in the human adult kidney library.

Table 1.   Isolation of Rp-interacting proteins in yeast. The number of colonies obtained (clone A1 to A37), their growth on various selective media (SM), and their production of α- and β-galactosidase from both the primary and secondary library screening are shown. ND, not done.
AssayPrimary screenSecondary screen
Growth on SM -Trp/-LeuNDA1, A6, A7, A10, A11, A14, A15, A17
A20, A24, A25, A26, A29, A32, A33
Growth on SM -Trp/-Leu/-HisClones A1 to A37ND
Growth on SM -Trp/-Leu/-His/-AlaA1, A6, A7, A10, A11, A14, A15, A17 
A1, A6, A7, A10, A14, A37
A20, A24, A25, A26, A29, A32, A33
A36, A37
β-galactosidase activityNDA1, A6, A7, A10, A14, A37
α-galactosidase activityNDA1, A6, A7, A10, A14, A37

AH109 cells only grow under stringent conditions when they contain both Rp and A1 expressed as GAL4DBD and GAL4AD fusion proteins, respectively (Fig. 1A–C; pKGIRp, pGAD10 A1), showing that these proteins interact. To see if A1 and Rp interact irrespective of their fusion partners, GAL4AD or GAL4DBD, a domain swap was undertaken. A1 and Rp were inserted into pGBT9 (pGBT9 A1) and pGADT7 (pGADT7 Rp), respectively, and shown to still interact in AH109 cells (Fig. 1A–C; pGBT9 A1, pGADT7 Rp). The yeast two-hybrid assay also revealed that A1 can homodimerize in AH109 cells (Fig. 1D,E; pGBT9 A1, pGAD10 A1). Furthermore, the interaction of A1 with Rp is specific because A1 does not interact with laminin, p53 (Fig. 1A–C; pGBKT7Lam, pGAD10 A1 and pGBKT753, pGAD10 A1) or mCtBP2 (Fig. 1D,E; pGBT9 A1/pGAD10mCtBP2).

A1 is identical to a novel gene called UXT

The sequence of A1 is shown in Fig. 2A. It consists of a 546-nucleotide cDNA, which includes a partial poly(A) tail. There is an unbroken reading frame of 162 amino acids that is continuous with the vector GAL4DBD and terminates with TGA (Fig. 2A) at nucleotide 487. Inspection of the sequence shows that the first ATG (Fig. 2A) fits the Kozak consensus for translation initiation [19], suggesting that the gene normally encodes a putative protein of 157 amino acids with a predicted molecular mass of 18.2 kDa. Interrogation of the NCBI nucleotide database revealed the identity of A1 with a novel gene designated UXT (AF092737) that encodes a 157-amino-acid protein (Fig. 2A). A1 is subsequently referred to as UXT.

Figure 2.

  (A) Sequence of clone A1. The nucleotide sequence of clone A1 is shown. Below is the sequence of UXT (lower case) and regions of identity are indicated by -. Below the nucleotide sequence is shown the primary amino acid sequence using the single letter code. The bold nucleotide and amino acid sequence is partial pGBT9 GAL4 DBD vector nucleotide and amino acid sequence. Bold boxed nucleotide sequences show the predicted translation initiation site for UXT/clone A1 and the translation termination site. The boxed amino acid sequence represents translation of predicted 5′UXT noncoding leader sequence that maintains the reading frame of GAL4 AD and UXT. (B) Northern blot analysis of UXT expression. Human MTN™ blots (Human, Human II and Human III; Clontech) containing 2 µg per lane of the indicated poly(A+) RNAs were hybridized to a 32P-labelled UXT (A1) probe. Filters were washed stringently, 0.1× NaCl/Cit, 0.1% SDS, 65 °C and bands were visualized by autoradiography. (C) RT-PCR analysis of total cellular RNA derived from the indicated cells using human/mouse-specific Evi1 (HME1/HME2), Uxt (HMUXT5/HMUXT3) and Gapdh (GAPDH5′/GAPDH3′) primers. The expected size fragments for all three genes: Evi1 (467 bp); Uxt (278 bp) and Gapdh (451 bp) are indicated by arrows. M indicates the 1kb hyperladder (Bioline).

The tissue distribution of UXT was examined using northern blot analysis. This shows that UXT produces an abundant transcript of ∼ 750 bp in all tissues examined, with the highest expression levels in heart, skeletal muscle, pancreas, peripheral blood leukocytes, thyroid and lymph node (Fig. 2B). The transcript size is consistent with A1 being almost full length, allowing for an additional 39 5′ nucleotides described for UXT and a 200 nucleotide poly(A) tail.

The tissue distribution of UXT shows that it is expressed in the same tissues as Evi1, including lung, kidney, ovary and heart. RT-PCR was performed to confirm that both genes are expressed simultaneously in the same cells. Evi1 is abundantly expressed in the leukaemia cell line DA-3 [20] and primary mouse embryo fibroblasts (MEFS), but not in lymphoma-derived monocytic U937 cells (Fig. 2C). UXT is expressed in all cell types examined (Fig. 2C), confirming that transcripts for both genes coexist in cells in which Evi1 is expressed, including leukaemia cells where Evi1 has been activated and in cells where Evi1 is normally expressed such as fibroblasts.

UXT binds full-length Evi1

UXT/Evi1 binding was confirmed using a glutathione S-transferase (GST)-pull down assay. GST–Rp and GST–UXT fusion proteins were expressed and purified from Escherichia coli strain pLysS cells using bacterial expression vectors (see Experimental procedures). 35S-Methionine-labelled in vitro-translated UXT and Evi1 proteins were produced using the expression vectors pCDNA3–UXT (Experimental procedures) and pRC/CMV FL [6], respectively (Fig. 3A). GST pull-down assays were performed with combinations of bacterially expressed and 35S-methionine-labelled proteins and the results are shown in Fig. 3. These confirm that UXT binds Rp (Fig. 3B, lane 5) and furthermore also interacts with full-length Evi1 (Fig. 3B, lane 7). Formation of the complexes are specific as they do not occur with GST alone (Fig. 5B, lanes 2 and 3).

Figure 3.

  (A) SDS–PAGE of in vitro translated products showing 35S-labelled UXT (lane 1) and Evi1 (lane 2). (B) Analysis of GST pull-down assays by SDS–PAGE. White box indicates bacterially derived GST protein, stippled box represents in vitro translated UXT and bacterially derived GST–UXT fusion protein. Grey box represents bacterially derived GST–Rp fusion protein. Evi1 zinc-finger domains and repressor domains are indicated by black and grey boxes, respectively. Hatched box shows acidic domain. (C) Interaction of Evi1 and UXT (A1) in mammalian cells. Various combinations of pcDNA3Evi1myc (4 µg) and pcDNA33HAUXT (1 µg), indicated by +, were transiently transfected into BOSC-23 cells as described previously [2]. One-tenth of whole-cell extracts were utilized directly for western blot analysis and the remainder was immunoprecipitated with α-myc (Santa Cruz Biotechnology, IP α-9E10) prior to western blot analysis. Whole-cell extracts and immunoprecipitation α-9E10 extracts were resolved by SDS–PAGE (10%) and sequentially probed with α-haemagglutinin (Boehringer Mannheim, Mannheim, Germany; WB α-12CA5) and α-myc (WB α-9E10) mAb. Proteins were visualized by ECL™ (Amersham Pharmacia Biotech). Protein size was estimated by comparison with Full Range Rainbow™ molecular mass markers (Amersham Pharmacia Biotech, not shown). Evi1myc and HAUXT fusion proteins are indicated by arrows.

The UXT/Evi1 interaction was also confirmed in mammalian cells using coimmunoprecipitation. UXT was inserted inframe with three copies of the HA epitope tag into the expression vector pcDNA3 to create pcDNA33HAUXT (Experimental procedures). Cells from the human embryonic kidney cell line BOSC-23 were transiently transfected with various combinations of the C-myc epitope-tagged Evi1-encoding vector pcDNA3EVI1myc (A. Coyle & C. Bartholomew, unpublished) and pcDNA33HAUXT, and portions of the cell extracts were examined either directly by western blot analysis or immunoprecipitated prior to western blotting. Western blot analysis of whole-cell extracts with either HA-specific α-12CA5 or C-myc-specific α-9E10 shows the expected size epitope-tagged 21 kDa UXT (HAUXT) and 145 kDa Evi1 (EVI1myc) proteins, respectively, in cells transfected with the corresponding expression vectors (Fig. 3C). Immunoprecipitation of cell extracts and western blot analysis with α-myc reveals EVI1myc in cells transfected with pcDNA3EVI1myc, as expected (Fig. 3; IP α-9E10). In addition, western blot analysis of α-myc-immunoprecipitated cell extracts with α-HA shows HAUXT only in those extracts that also contain EVI1myc, confirming that these two proteins form a complex. Examination of the UXT–Evi1 interaction at the endogenous level using the same method must await new reagents to be developed.

UXT suppresses Evi1-mediated transformation

Next we investigated whether UXT has an effect on Evi1 biological activity by examining the impact of UXT expression on Rat1 and RatFL (Evi1-transformed cells) cell transformation. The retroviral expression vector p50MHAUXTzeo, containing UXT fused inframe to a HA tag, was created. Recombinant retrovirus produced in BOSC-23 cells was used to infect Rat1 and RatFL cells and cell populations expressing UXT were selected. Neither UXT expression in Rat1 cells nor empty vector controls were transforming (Fig. 4A, lanes 1,3,4), whereas Evi1 was (Fig. 4A, lane 2). However, enforced expression of UXT reduced the number of transformed colonies produced by RatFL cells by 50% (Fig. 4A, lane 5). The empty vector control had no effect in RatFL cells (Fig. 4A, lane 5). Cell extracts prepared from the various cell populations confirmed that they each produce the expected UXT HA-tagged fusion protein (Fig. 4B). The effect of UXT on Evi1 transcriptional repressor activity was also examined but no significant changes were observed (data not shown).

Figure 4.

  (A) Colony formation of Rat1 and RatFL cell populations infected with the indicated retroviral vectors. Numbers were determined for colonies > 0.1 mm derived from plating 103 cells. Error bars indicate the average number of colonies observed from three independent assays. Schematic representation of Evi1 and UXT proteins are as described in the legend to Fig. 3. The HA epitope tag is shown as a striped box. (B) Western blot analysis with α-haemagglutinin (as described in Fig. 3) of whole-cell extracts derived from Rat1 cells (1), p50MHAUXTzeo infected BOSC 23 (2), Rat1 (3) and RatFL (4) cells. The HAUXT fusion proteins are indicated by an arrow.

An Evi1 Rp domain mutant lacking UXT-binding activity enhances Rat1 transformation activity

The UXT-binding region of Rp was determined using a series of deletion mutants (Experimental procedures) using the yeast two-hybrid assay. Binding was retained when the Rp fragment (514–724) was deleted from the N-terminus to amino acid 634 (fragment 634–724), but lost upon C-terminal deletions between 715 and 695 (data not shown). A series of refined deletion mutants were created from the C-terminus of the 634–715 fragment to determine the minimum deletion required to lose UXT-binding activity. Yeast two-hybrid assays revealed that Rp UXT binding is lost upon deletion of amino acids 706–707 (Fig. 5A–C, quadrant 7–9). Western blot analysis with α-GAL4DBD confirmed expression of equally abundant correct size proteins (fragments were subcloned from pGBT9 to pGBKT7 for this purpose; data not shown).

Figure 5.

 Interaction of UXT deletion mutants with Rp and mCtBP2 in yeast cells. (A,D) Yeast AH109 cells were transformed with the combination of plasmids shown. The growth of single yeast colonies containing these plasmids are shown in (B) and (E) on SD plus l-histidine and adenine and (C) and (F) on SD lacking l-histidine and adenine.

The ability of an Rp-deletion mutant lacking only amino acids 706 and 707 (Rp Δ706–707) to bind UXT and mCtBP2 was examined using the yeast two-hybrid assay. Results confirm that this mutant is unable to bind UXT (Fig. 5D–F, quadrant 2) but retains the ability to bind mCtBP2 (Fig. 5D–F, quadrant 5).

The transforming activity of Evi1 containing the UXT-binding mutant Rp domain was investigated in Rat1 cells. The Rp domain of the previously described vector, p50MRpWTneo, which has identical transforming activity to p50M4.6neo [14], was substituted for RpΔ706–707 to create p50MRpΔ706–707neo and populations of Rat1 cells expressing this gene were selected and tested for transformation using a soft agar colony assay. The results show that Evi1Δ706–707 generates the same number of transformed Rat1 cell colonies as wild-type Evi1 (data not shown). However, the mutant protein produces a higher proportion of larger colonies (Fig. 6A). Figure 6B shows the percentage of total colonies that are > 0.3 mm in diameter, generated from populations of Rat1 cells expressing either WT or mutant forms of Evi1. Approximately 14% of colonies generated by Evi1Δ706–707 are > 0.3 mm in size, whereas only 3–4% of Evi1 transformed colonies achieve these dimensions.

Figure 6.

  (A) Photograph of colonies formed by Rat1 cell populations expressing the indicated Evi1 proteins using a Leica GZ6 microscope. (B) Histogram showing the percentage of soft agar colonies > 0.3 mm generated by populations of Rat1 cells expressing either wild-type (Evi1) or mutant (Evi1Δ706–707) Evi1. The percentage was determined by counting total number of colonies generated and the number of colonies > 0.3 mm. Colony numbers were determined from two independent experiments (1 & 2). Control 1 and 2 are parental Rat1 cells. Error bars indicate variation between three independent assays for each experiment. (C) Histogram showing total number of soft agar colonies > 0.3 mm produced per 1000 Rat1 cells expressing the indicated proteins. Control is empty p50MXZEO retroviral vector. Error bars indicate the average number of colonies observed from three independent assays. No colonies are observed in Rat1 cells in the absence of Evi1 (not shown).

To see if enforced expression of UXT inhibits the transforming activity of Evi1Δ706–707, the colony-forming activity of Rat1 cells expressing both exogenous proteins was examined. Populations of Rat1 cells infected with both p50MRpΔ706–707neo and p50MHAUXTzeo were selected and the production of macroscopic colonies in soft agar assessed. The results show that transformation is suppressed by ∼ 50% in cells containing Evi1Δ706–707 and UXT (Fig. 6C). The empty vector control, p50MXZEO, has no effect on Evi1Δ706–707 transforming activity.

Discussion

We have shown that Evi1 can interact with at least two proteins present in human adult kidney cells via the Rp domain. One interaction is with the human homologue of mCtBP2, HuCtBP2, and its significance has already been documented [14]. Surprisingly, our library screens did not isolate HuCtBP1, which recognizes and binds the same conserved PLDLS motif as HuCtBP2, despite PCR analysis of the human kidney library demonstrating that HuCtBP1 was present (data not shown). Evi1 has previously been shown to associate with both HuCtBP1 [21] and mCtBP1 (E. Ritchie and C. Bartholomew, unpublished). The most likely explanation is that no appropriate HuCtBP1 clones in the library are expressed inframe with GAL4AD.

Evi1's interactions with UXT, an 18.2 kDa protein, have not been described previously. Enforced UXT expression in RatFL cells moderately suppresses cell transformation. The Evi1 UXT-binding mutant (Evi1Δ706–707) retains the ability to bind mCtBP2 and Rat1 cell-transformation activity is enhanced. This shows that UXT binding: (a) is not required for Evi1 interaction with mCtBP2; (b) is not required for Evi1-mediated cell transformation; and (c) has an inhibitory effect on Evi1 biological activity. Interestingly, enforced expression of UXT with Evi1Δ706–707 in Rat1 cells still moderately suppresses cell transformation. The data suggest that at least part of the suppressor activity is mediated by its interaction with UXT but that enforced UXT expression has a general growth-suppressive activity that is independent of Evi1. In this regard, it would be interesting to see if UXT expression suppresses cell transformation by other oncogenes too.

Recent studies suggest that Evi1 is a survival factor. It is able to protect cells against chemically induced apoptosis [22] and promote survival of haemopoietic stem cells [23]. Therefore, negative regulation of Evi1 biological activity might compromise its survival function, reducing cell transformation and/or proliferation by decreasing the ability of cells to proliferate optimally in new environments, for example, the anchorage-independent growth of Rat1 fibroblasts in soft agar displayed in the transformation assay.

UXT is located on human chromosome Xp11 and was originally identified when searching for the X-linked genes responsible for Renpenning syndrome, Prieto syndrome and Sutherland–Haan syndrome that map to this region, although it does not appear to be involved in their development [24]. As shown here, UXT is abundantly and ubiquitously expressed in all tissues examined. Based on an EST database search it has been suggested that UXT is overexpressed in tumour cells [24] and might have a role in tumorigenesis. In this regard, it is interesting to note that acute basophilic leukaemia is associated with translocation t(X,6)(p11,q23) [25] and therefore UXT itself is a candidate for the disease-associated Xp11 gene or alternatively its ubiquitously active promoter may deregulate expression of a novel leukaemia gene located on 6q23.

The function of UXT is not known, but it has previously been shown to associate with several other proteins, each of which has a role in the regulation of gene transcription. These include the leucine-rich pentatricopeptide repeat-motif-containing (LRPPRC) protein [26] and the transcriptional coactivator CITED2 [27]. UXT is also known as ART-27, a transcriptional coactivator that interacts with the N-terminus of the androgen receptor [28]. Interestingly, enforced expression of UXT/ART-27 in LNCaP prostate cancer cells inhibits proliferation and furthermore its expression is downregulated in human prostate cancer [29]. This proliferation-suppressive activity is consistent with the inhibition of Evi1-mediated Rat1 cell transformation mediated by UXT in our studies. Both these experimental observations are in direct contradiction to the earlier interpretation derived from interrogation of the EST database [24]. Furthermore, UXT expression has also been shown to be elevated in bladder, breast, ovary and thyroid tumours suggesting it may be a tumour marker [30]. This apparent discrepancy may be reconciled if UXT effects on cell transformation are tissue specific or that either up- or downregulation contributes to oncogenesis. Alternatively, there are at least two naturally occurring UXT splice variants, transcript variant 1 encoding a 157-amino acid protein (NM153477) and transcript variant 2 encoding an N-terminal truncated protein of 145 amino acids (NM004182), which might have opposing effects on cell transformation. In this regard, it will be interesting to investigate both the form of UXT and the structural integrity of the gene coding sequences in tumour cells where its expression is elevated.

Several lines of evidence indicate that UXT regulates cell proliferation. UXT is a target gene for the E2F family of transcription factors that regulate transition through the G1/S phase boundary of the cell cycle [31]. Both E2F1 and E2F6 inhibit UXT gene expression [31,32]. Furthermore, E2F6 corepresses UXT and other genes with common functions in tumour suppression, suggesting that it might have a similar activity [32], consistent with its ability to inhibit cell transformation.

UXT has also been isolated as STAP1 and classified as a member of the α class prefoldin family [33]. UXT (STAP1) is a component of a large protein complex that can regulate transcription in HeLa cells, consistent with its interaction with the Evi1 transcription factor observed here. UXT has also been shown to be located in centrosomes and its overexpression disrupts centrosome structure in human U2OS cells [30]. It has been suggested that UXT abnormality may cause dysfunction of the centrosome contributing to malignant transformation [30]. Centrosomes play a key role in cell proliferation, serving both to nucleate polarized microtubule arrays for mitotic spindle organization and cytokinesis and providing a multiplatform scaffold with protein-docking sites for integrating cellular regulatory events [34]. This raises the possibility that Evi1 might contribute to the regulation of cell proliferation by interacting with a component of centrosomes.

Our results show that UXT inhibits Evi1-mediated cell transformation in addition to the previously described inhibition of cell-proliferation activity and therefore may be a negative regulator of cell growth. UXT interacts with Evi1 and may be a direct negative regulator of its biological activity. The precise molecular mechanism by which UXT reduces Evi1 activity is unknown. UXT might mediate its negative control of cell proliferation and transformation by directly interacting with and regulating the activity of factors that stimulate this biological activity such as Evi1. In addition, UXT mediates its negative activity on cell proliferation and transformation indirectly, by an independent mechanism.

Experimental procedures

Construction of plasmids

pKGIRp was created by inserting a EcoRI/BamHI Rp domain fragment from pGBT9Rp [14] into pKG1 [18]. pGBT9A1 was created by excising A1 as a BglII fragment from pGAD10 A1and inserting into a BamHI site of pGBT9. pGADT7Rp was created by excising a EcoRI/BamHI Rp fragment from pKG1Rp and inserting into the corresponding site of pGADT7 (Clontech, Mountain View, CA). pGEXUXT was created by PCR amplification with 5′- and 3′-oligonucleotides CGCTGGATCCCGGGAGGAGCCCATCATG and GGAAGAATTCTCAAATTCCAGGAAAAAACCA, respectively, and insertion of UXT fragments into BamHI/EcoRI of pGEX2 (Promega, Madison, WI). pGEXRp was created by PCR amplification with 5′- and 3′-oligonucleotides AAGCGGATCCCGCATTCTCTCAATCAATG and AAGCTGAATTCGTAGCGCTCTTTCCCCT and insertion of Rp fragments into BamHI/EcoRI of pGEX1 (Promega). pcDNA3UXT was created by inserting a HindIII/BamHI UXT PCR fragment amplified from pGAD10 A1 with oligonucleotides AATTCAAGCTTGCGCAATGAAGGTGAAGG and AATTCGGATCCTCAATGGTGAGGCTTCTC. pcDNA33HAUXT was created by simultaneous ligation of a NcoI/NotI-digested UXT PCR fragment, amplified from pGAD10 A1 with 5′- and 3′-oligonucleotides GAATCCATGGCGACGACGCCCCCTAAGCG and GAATTGCGGCCGCCTCAATGTGAGGCTTC, respectively, with a NcoI/EcoRI fragment containing three copies of the HA epitope from S3H-ERK2 (gift from W. Kolch, Beatson Institute, Glasgow, UK) and EcoRI/NotI-digested pcDNA3 (Invitrogen, Carlsbad, CA). p50MHAUXTzeo was created by PCR amplification of pGAD10 A1 with 5′- and 3′-oligonucleotides AGCTTGCGGCCGCATCATGTACCCATACGATGTTCCAGATTACGCTGCGACCCCCCTAAGCG and GCTGAATTCTCAATGGTGAGGCTTC which was digested with NotI/EcoRI and inserted into the corresponding site of p50Mxzeo [14]. Rp domain deletion mutants were created by inserting EcoRI/NotI-digested PCR fragments generated using the following 5′-oligonucleotide (E634) AGCTGAAATTCCCCTTCTTCATGGACCCCATT and 3′-oligonucleotides (N724) AATTGCGGCCGCTCAGTAGCGCTCTTTCCCCTT (N715) AATTGCGGCCGCTCAGTTCTCTGGCAGGGTGTT or EcoRI/BamHI-digested PCR fragments generated with 3′-oligonucleotides (B713) AGCTTGGATCCTATGGCAGGGTGTTGGGAGGAGC, (B711) AGCTTGGATCCTAGGTGTTGGGAGGAGCTCGGAA (B709) AGCTTGGATCCTAGGGAGGAGCTCGGAAGCTGAA (B707) AGCTTGGATCCTAAGCTCGGAAGCTGAACATGGA (B705) AGCTTGGATCCTAGAAGCTGAACATGGAGGGCAC, into EcoRI/NotI-digested pGBT9N [14] or EcoRI/BamHI-digested pGBT9. pGBT9RpΔ706–707 was created by site-directed mutagenesis (QuickChange XL system, Stratagene, La Jolla, CA) of pGBT9Rp [14] with oligonucleotides 5′-CCCTCCATGTTCAGCTTCCCTCCCAACACCCTGCC and 3′-GGCAGGGTGTTGGGAGGGAAGCTGAACATGGAGGG. The same primers were used for site directed mutagenesis of p50MRpWTneo [14] to create p50MRpWTΔ706–707neo.

Cell culture, transfections, CAT and β-galactosidase assays

RatFL cells have been described previously [11]. Rat1, RatFL, Bosc-23, HEK293 and primary mouse embryo fibroblasts were all maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, sodium pyruvate and glutamine. U937 cells were maintained in RPMI-1640 supplemented with 10% heat-inactivated fetal bovine serum, sodium pyruvate and glutamine. DA-3 cells were similarly maintained in the presence of 10% WEHI-3-conditioned medium. Procedures for transfections, production of helper-free recombinant retrovirus, retroviral infections and growth in soft agar, CAT and β-galactosidase assays have all been described previously [6]. Cells infected with zeocin containing retroviral vectors were selected and maintained in 1 mg·mL−1 zeocin™ (Invitrogen).

Yeast two-hybrid assay

For the primary screen, AH109 competent cells were prepared and transformed as described by the suppliers (Clontech) with 1 mg pKGIRp and 0.8 mg human kidney cDNA plasmid library (Clontech). Total numbers of transformed colonies were estimated by growing an aliquot of transformed cells for 3 days, at 30 °C on synthetic dropout medium (SD), 0.67% yeast nitrogen base without amino acids (Difco Laboratories, Sparks, MD), 0.06% CSM-HIS-LEU-TRP (Bio101, Inc., Irvine, CA), 2% glucose, pH 5.8, 1.5% select agar (Life Technologies, Grand Island, NY) and 20 μg·mL−1l-histidine HCl (Sigma, St. Louis, MO; H-9511). The remaining cells were grown for 14 days on SD without l-histidine. Growth of any colonies was subsequently examined under stringent conditions on SD lacking both l-histidine and adenine by substituting CSM for DO supplement (-ADE,-HIS,-LEU,-TRP; Clontech).

Plasmid DNA was isolated from yeast colonies by scraping into 1 mL TE buffer, pelleting cells and resuspending them in 0.5 mL 10 mm K2HPO4 pH 7.2, 10 mm EDTA, 50 mm 2-mercaptoethanol, 0.25 mg·mL−1 zymolase, at 37 °C for 30 min, mixing with 0.1 mL 25 mm Tris/HCl pH 7.5, 25 mm EDTA, 2.5% SDS, at 65 °C for 30 min followed by 10 min at 0 °C with 166 µL 3 m KOAc. E. coli strain DH5α electroporation-competent cells were prepared according to Sambrook et al. [35], electroporated as described by Clontech and transformed colonies selected on Luria–Bertani plates containing 50 µg·mL−1 ampicillin. Plasmid DNAs were prepared using a NucleoSpin® plus miniprep plasmid extraction kit (Clontech). α- and β-galactosidase assays, respectively, were performed as described by Clontech.

GST pull-down assay

Bacterial cultures containing pGEX expression vectors were induced with 1 mm isopropylthio-β-d-galactoside for 3 h and cells were resuspended and sonicated in NETN buffer (20 mm Tris pH 8.0, 100 mm NaCl, 1 mm EDTA, 0.5% NP40). Extracts were cleared in a microfuge at 4 °C and GST-fusion proteins bound to glutathione Sepharose by mixing with glutathione Sepharose slurry equilibrated with NETN at 20 °C for 30 min followed by centrifugation and washing three times in NETN. 35S-Labelled in vitro translated proteins were produced using TNT-coupled reticulocyte lysates (Promega). GST pull-down assays were performed by incubating GST-fusion protein/glutathione Sepharose conjugate (5 mg), 100 mg E. coli cell extract and in vitro translated protein in NETN, 4 °C, overnight. Extracts were washed three times in excess NETN, 1× in excess MTPBS (150 mm NaCl, 16 mm Na2HPO4, 4 mm NaH2PO4, pH 7.3) and bound protein eluted from complex in 50 mm Tris/5 mm reduced glutathione.

Immunoprecipitation

Cells were scraped into 0.25 mL of immunoprecipitation buffer [36], rapidly frozen, thawed at 0 °C for 1 h, then microfuged at 11 000 g for 10 min at 4 °C. Supernatant was removed, and 25 µL was aliquoted as whole-cell extract for western blotting and the remainder incubated o/n with anti-(c-myc α-9E10) serum (Santa Cruz Biotechnology, Santa Cruz, CA) at 4 °C and subsequently incubated with 50 µL of 50% slurry of rabbit anti-(mouse IgG)-coated protein A Sepharose beads for 2 h at 4 °C. Beads were washed three times in immunoprecipitation buffer and prepared for western blot analysis.

RT-PCR

Total RNA (0.2 µg, prepared using the RNazol™B method) was amplified using the Calypso™ RT-PCR system (BioGene Ltd., Kimbolton, UK) according to the manufacturers instructions. The coupled reaction was performed in a MJ Scientific thermal cycler at 50 °C for 30 min, followed by amplification by 30 cycles of 30 s at 94 °C, 30 s at 55 °C, 1 min at 72 °C, and a final 10 min at 72 °C extension time using the following human/mouse-specific primers: HME1 CCAGATGTCACATGACAGTGGAAAGCACTA; HME2 CCGGGTTGGCATGACTCATATTAACCATGG; UXT 5′-GACAAGGTATATGAGCAGCTG; UXT 3′-TTGATATTCATGGAGTCCTTG; Gapdh5 ACCACAGTCCATGCCATCAC; Gapdh3 TCCACCACCCTGTTGCTGTA. PCR products were resolved by agarose gel electrophoresis (NuSieve® GTG® agarose, FMC).

Sequencing

A Licor automated sequencer was used for sequence determ/ination using SequiTherm EXCELII (Cambio, Cambridge, UK) and appropriate IRD-800 labelled primers.

Site-directed mutagenesis

The QuickChange XL system (Stratagene) was used according to the manufacturer's instructions.

Western blot analysis

Whole-cell extracts were prepared as described previously [14]. Proteins were examined by SDS/PAGE, transferred to Hybond™-ECL nitrocellulose, incubated with appropriate antibodies and visualized with an ECL western blotting detection system (Amersham Pharmacia Biotech, Piscataway, NJ). Protein sizes were estimated by comparison with Full Range Rainbow™ molecular mass markers (Amersham Pharmacia Biotech).

Acknowledgements

This study was funded by Glasgow Caledonian University PhD studentship (RM) and The Leukaemia Research Fund (98/10).

Ancillary