Transcriptional regulation of the human ALDH1A1 promoter by the oncogenic homeoprotein TLX1/HOX11

The homeoprotein TLX1, which is essential to spleen organogenesis and oncogenic when aberrantly expressed in immature T cells, functions as a bifunctional transcriptional regulator, being capable of activation or repression depending on cell type and/or promoter context. However, the detailed mechanisms by which it regulates the transcription of target genes such as ALDH1A1 remains to be elucidated. We therefore functionally assessed the ability of TLX1 to regulate ALDH1A1 expression in two hematopoietic cell lines, PER-117 T-leukemic cells and human erythroleukemic (HEL) cells, by use of luciferase reporter and mobility shift assays. We showed that TLX1 physically interacts with the general transcription factor TFIIB via its homeodomain, and identified two activities in respect to TLX1-mediated regulation of the CCAAT box-containing ALDH1A1 promoter. The first involved CCAAT-dependent transcriptional repression via perturbation of GATA factor-containing protein complexes assembled at a non-canonical TATA (GATA) box. A structurally intact homeodomain was essential for repression by TLX1 although direct DNA binding was not required. The second activity, which involved CCAAT-independent transcriptional activation did not require an intact homeodomain, indicating that the activation and repression functions of TLX1 are distinct. These findings confirm ALDH1A1 gene regulation by TLX1 and support an indirect model for TLX1 function, in which protein-protein interactions, rather than DNA binding at specific sites, are crucial for its transcriptional activity.

Confirmation that the gene product of TLX1 is an oncoprotein has come from mouse models, which have shown that enforced expression of TLX1 impairs cell differentiation and leads to malignancy. [11][12][13][14][15] Current models for the mechanism by which TLX1 promotes leukemia are based on its ability to act indirectly, either by enhancing chromosome instability 16,17 or by regulating gene expression through specific protein-protein interactions with key cellular regulatory molecules such as the protein serine/threonine phosphatases PP1 and PP2A, and the transcriptional coactivator/acetyltransferase, CREB-binding protein (CBP). [18][19][20] Thus, TLX1 may mediate its transforming function by simultaneously inhibiting the phosphatase activity of PP1/PP2A to promote cell cycle progression via upregulation of pathways such as those downstream of E2F and MYC, 19 and sequestering CBP at heterochromatin to accomplish a differentiation block. 20 TLX1 has also long been suspected to act as a sequence-specific transcription factor 21,22 that preferentially binds to the core sequence TAA T /GTG in vitro. 22,23 However, with the exception of its associations with heterochromatic satellite 2 DNA 24 and its own promoter, 25 no direct target genes for TLX1 have been convincingly identified. Individual genes suspected to be regulated by TLX1 have been described in various settings, including spleen development (Aldh1a1, Wt1), 26,27 erythroid differentiation (ALDH1A1, c-Kit, Vegfc), 15,28 and T-cell leukemia (ALDH1A1, FHL1, NR4A3). 26,29,30 Nevertheless, the regulatory role that TLX1 plays in such cases still remains to be determined.
The best characterized TLX1 target gene, ALDH1A1 (aldehyde dehydrogenase 1A1) belongs to a subfamily (class 1A) of ALDH genes whose main biological role is the con-version of the aldehyde form of vitamin A (retinal) to its biologically active form, retinoic acid. 31 ALDH1A1 apparently has normal roles in embryonic development, 32 and in the renewal/differentiation of hematopoietic stem cells (HSCs), 33 where it is known to be highly expressed. 34 ALDH1A1 is further implicated in regulating the polarity of HSC differentiation by favoring the development of a myeloid rather than a lymphoid cell fate. 30,35 In agreement with this, ALDH1A1 expression can discriminate between acute myeloid (AML) and acute lymphoid leukemia, 36 and while we have demonstrated aberrant ALDH1A1 expression in T-ALL, 30 ALDH1A1 is reportedly down-regulated in AML. 37 Thus, ALDH1A1, which is regulated by TLX1 in its normal chromosomal context 26,30 is of interest due to its associations with both normal development and leukemogenesis. Here, we explored the molecular mechanism(s) by which TLX1 regulates the ALDH1A1 gene and find that it occurs in a non-DNA binding fashion through protein-protein interactions. We further show that TLX1 interacts directly with the general factor TFIIB via its homeodomain, indicating a role for TLX1 in gene regulation via the basal transcriptional machinery.

Cell culture and expression plasmids
The PER-117 and ALL-SIL T-cell lines, and erythroleukemic cell line HEL, were cultured as previously described. 30 The coding regions of human TLX1 and TFIIB were amplified by RT-PCR from ALL-SIL cDNA generated by Thermoscript RT (Invitrogen, Carlsbad, CA, USA) using PfuTurbo DNA Polymerase (Stratagene, La Jolla, CA, USA) and primers containing an Nhe I restriction site. The resulting products were cloned into the Nhe I site of the pCINeo mammalian expression vector (Promega, Madison, WI, USA) and sequence verified.
For measuring the effect of TLX1 on ALDH1A1 promoter activity, transfections were similarly performed with the additional inclusion of 15 µg of expression plasmid, either pEF-BOS/TLX1, pEF-BOS/TLX1ΔH3 or pEF-BOS as control. 40 In this case, transcriptional activity was defined as the log (base 2) transformation of the ratio of luciferase activity (in relative light units; RLU) from pGL3-Basic derivatives relative to β-galactosidase activity from pSV-β-Gal. Statistical analysis was performed in SPLUS 2000 using a mixed effects model with day of experiment as a random effect and luciferase reporter plasmid and expression plasmid as fixed effects. Interactions between contrasts comparing pEF-BOS/TLX1 with pEF-BOS, and contrasts comparing ALDH1A1 constructs with pGL3-Basic, were examined. This revealed the extent to which the effect of adding TLX1 was different for the various promoter constructs compared to pGL3-Basic.
NP-40 was added to a final concentration of 0.5% and the cells vortexed for 10 s. The nuclei were pelleted by centrifugation at 6500 g for 1 min and resuspended in 100 µL Buffer C (20 mM HEPES, pH 7.9, 420 mM NaCl, 1 mM EDTA, 1 mM EGTA, 20% glycerol) with protease inhibitors. The nuclear suspension was stirred vigorously on ice for 30 min. The sample was centrifuged at 13 000 g for 10 min, and aliquots of the nuclear extract were frozen immediately in liquid nitrogen and stored at -80°C until required. The protein concentration of nuclear extracts was determined by the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA).

Western blotting and immunoprecipitation of TLX1 complexes
Western blotting was performed with 50 ug of nuclear extract electrophoresed through a 12% SDS-PAGE gel and transferred to a Hybond ECL membrane (AP Biotech, Little Chalfont, UK). The membrane was blocked overnight in TBS-Tween buffer (1 mM Tris, 7.5 mM NaCl, 0.05% v/v Tween-20) containing 3.5% (w/v) gelatin and incubated with 1:1200 diluted rabbit anti-TLX1 antiserum (sc-880, Santa Cruz Biotechnology) (1 h, RT). The membrane was incubated with HRP-conjugated secondary antibody (1 h, RT) prior to visualization of bands by enhanced chemiluminescence (AP Biotech). For immunoprecipitation, nuclear extract from ALL-SIL cells (5×10 6 ) was precleared with the addition of protein A/G agarose (Santa Cruz Biotechnology) for 15 min at 4°C. Pre-cleared extracts were incubat- ed with 5 µg of affinity-purified rabbit anti-TLX1 polyclonal antibody (sc-880; Santa Cruz Biotechnology), or no antibody (as control), for 4 h at 4°C with constant gentle rocking. Immune complexes were bound to protein A/G agarose beads, centrifuged and washed with 1.2 mL of cold IP wash buffer (100 mM Tris-HCl pH 7.4, 1% NP40 and 1% deoxycholic acid) containing 500 mM LiCl (once) and 1.2 mL of IP wash buffer (four times).

MALDI-TOF mass spectrometric analyses
Immunoprecipitated proteins were resolved by SDS-PAGE and visualized by silver staining. 41 Bands were excised directly from gels into 96-well microtiter plates (Titertek, Huntsville, AL, USA), destained, 42 and in-gel trypsin digestions performed according to Shevchenko et al. 41 Peptides were extracted with 10 mg/mL α-cyano-4-hydroxycinnamic acid in 50% acetonitrile containing 0.1% trifluoroacetic acid, and aliquots of 0.5 uL applied directly onto a target plate and allowed to air dry. Tryptic peptide masses were then obtained using a Voyager-DE STR MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA, USA). Known trypsin autocleavage peptide masses (842.51, 1045.56 and 2211.1 Da) were used for internal calibration of each spectrum. The peptide masses were used to search the Swiss-Prot and NCBI nr protein databases using the MS-FIT database tool. A protein was considered to be identified if a minimum of five measured peptide masses matched calculated tryptic peptide masses and if the peptides identified by these matches provided at least 20% sequence coverage of the identified protein.
GST-pulldown assay TLX1 and TFIIB cDNAs in pCIneo were transcribed in vitro with T7 RNA polymerase. The products were labeled with [ 35 S]-methionine (Amersham Biosciences) using the TNT coupled transcription-translation system (Promega). GST-TLX1 fusion proteins were expressed from pGEX-6P-1 as described previously. 43 The GST-pulldown assay was performed by incubating 15 µg of GST, GST-TLX1 or GST-TLX1ΔH3 immobilized on glutathione sepharose beads (20 µL) with 5 µl of proteins translated in vitro and labeled with [35S]-methionine in 500 µL of binding buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 0.5% NP-40, 1 mM DTT, 0.5 mM PMSF, 0.05% BSA) for 18-20 h at 4°C with continuous rotation. Bound proteins were washed three times with 500 µL of cold binding buffer and eluted in SDS sample buffer. The eluted proteins were resolved on 12% SDS-PAGE gels and visualized by autoradiography.

Functional analysis of the human ALDH1A1 promoter
Ectopic expression of TLX1 was previously shown to modulate endogenous ALDH1A1 expression. 26,30,38 To pursue the molecular mechanism(s) underlying this regulatory effect, we cloned 1020 bp of the human ALDH1A1 5'-flanking region into the pGL3-Basic luciferase reporter vector. This sequence comprised nucleotides -978 to +42 relative to the transcriptional start site (Figure 1). Among the conserved promoter elements identified were a TATA-like sequence (GATA box) at -33 and a single CCAAT box at -74, which has previously been shown to be functionally important for ALDH1A1 expression. 44 Transient transfection of a series of deletion constructs (-978/+42, -303/+42, -201/+42, -146/+42, -91/ +42, -50/+42 and +1/+42; Figure 1) into PER-117 and HEL cells (which both lack TLX1 expression) showed that the ALDH1A1 promoter was functional in both cell lines and that the sequences between -50 and -91 and between -50 and -146 were required for maximal promoter activity in PER-117 and HEL cells, respectively ( Figure 2A). Consistent with previous results in Hep3B liver cells, 44 deletion of the CCAAT box at -74 caused a dramatic reduction in promoter activity, suggesting that it is a key element for ALDH1A1 expression in multiple cell types.

Regulation of the ALDH1A1 promoter by TLX1
To determine the effect of TLX1 on the transcriptional activity of the ALDH1A1 promoter, transactivation assays were carried out in PER-117 and HEL cells co-transfected with an expression plasmid containing TLX1 or empty vector (pEF-BOS) as a negative control. The results were expressed as a ratio of the normalized transcriptional activity of each of the two promoter constructs with and without TLX1 ( Figure 2B). Western blot analysis confirmed that TLX1 protein was expressed in the transfected cells ( Figure 2C). Consistent with our previous findings, 15 on ALDH1A1 was reversed in the transcriptional assay as compared to regulation of the endogenous gene. In PER-117 cells, TLX1 repressed ALDH1A1 promoter activity ( Figure  2B), and this required the sequence between -50 and -91, which contains the CCAAT box. Deletion of the CCAAT box (-91ΔCAT) completely abolished TLX1-mediated repression ( Figure 2B), suggesting that downregulation may possibly occur via this site. Interestingly, the shortest constructs (-50/+42 and +1/+42) were mildly stimulated by TLX1 in both PER-117 and HEL cells, (Figure 2B) indicating the presence of a separate, positive-acting TLX1response sequence localized to the region between +1 and +42. No effect by TLX1 was observed on the region between -201 and -303, which contains a conserved TLX1 in vitro binding site at -257 ( Figure 1), suggesting that TLX1 binds to a distinct recognition sequence in the ALDH1A1 promoter in vivo, or alternatively, that it acts via a non-DNA binding mechanism. In either case, these data indicate that TLX1 operates via at least two mechanisms in respect to the ALDH1A1 promoter; a general transactivating activity via an element located between +1 and +42 and a strong, cell linespecific repressive activity via an element located between -91 and -50.

The homeodomain is required for TLX1-mediated repression
To assess whether the homeodomain of TLX1 is required for the transcriptional activities of TLX1, a mutant TLX1 expression vector (TLX1ΔH3) was employed, which lacks the DNA recognition helix (helix 3) of the homeodomain. Whereas, TLX1 repressed the activity of the basal ALDH1A1 promoter (-91/+42) 4fold in PER-117 cells, TLX1ΔH3 not only lacked the ability to negatively regulate ALDH1A1, but switched to become an activator, stimulating transcription by approximately 5-fold ( Figure  3). A similar result was obtained when the promoter of another target gene, FHL1, 38 was used ( Figure 3). By contrast, the positive regulation of the -91/+42 construct observed in HEL cells was comparable when either TLX1 or TLX1ΔH3 was used (5.5-and 8-fold induction, respectively). Thus, the homeodomain is crucial for TLX1-mediated repression but is not required for TLX1-mediated activation, indicating that these two activities are distinct.

TLX1 does not affect formation of transcriptional complexes at the -74 CCAAT site
We opted to focus on the mechanism by which TLX1 mediates CCAAT-dependent repression. Mobility shift assays were therefore performed to identify whether TLX1 directly binds to the CCAAT box or interferes with the binding of other factors at this site. Nuclear extracts derived from PER-117 or HEL cells, either expressing empty pEF-BOS (as control) or pEF-BOS/TLX1, were incubated with a radiolabeled probe (ALDH CAT) spanning the CCAAT motif (-85 to -60; Figure 1). As shown in Figure 4A, incubation of nuclear extract from PER-117 or HEL cells with ALDH CAT resulted in the appearance of at least two specific DNA-protein complexes, together with a third non-specific complex (C2). The strongest, complex C1, appeared to be common to the two cell types, while additional complexes P and H were unique to PER-117 and HEL cells, respectively. These complexes were inhibited by the addition of a 35-fold molar excess of unlabeled self-competitor (ALDH CAT) but not by a 35-fold molar excess of CCAAT mutant competitor (ALDH CATMut; CCAAT to GACTG). Contrary to expectation, in both PER-117 and HEL cells the mobility shift pattern was identical regardless of TLX1   TLX1 alters DNA-protein complex formation at the -33 GATA box TLX1 may directly or indirectly inhibit CCAAT-dependent transcription via the basal apparatus. The ALDH1A1 promoter lacks a canonical TATA box but does possess a related GATA box (GATAAA) utilized by a number of genes whose transcriptional initiation involves interplay between TFIID and GATA factors. [45][46][47][48] EMSAs were thus performed to identify whether TLX1 directly binds to the GATA box or interferes with the binding of other factors at this site. Nuclear extracts derived from PER-117 or HEL cells, either expressing empty pEF-BOS (as control) or pEF-BOS/TLX1, were incubated with a radiolabeled probe (ALDH GATAB) spanning the GATA box (-42 to -19; Figure 1). As shown in Figure 4B, incubation of nuclear extract from PER-117 cells lacking TLX1 expression with ALDH GATAB resulted in the appearance of four specific DNA-protein complexes, C1, C2, P1 and P2. HEL nuclear extracts only produced three specific complexes (C1, C2 and H), the strongest of which (C2) appeared to migrate similarly in both cell types. These complexes were inhibited by the addition of a 35-fold molar excess of unlabeled self-competitor (ALDH GATAB) but not by a 35fold molar excess of GATA box mutant competitor (ALDH GATABMut; GATAAA to CTCAGC). Strikingly, in PER-117 but not HEL cells, expression of TLX1 resulted in a significant alteration in complex formation. Of the four specific PER-117 complexes, the formation of two (C1 and P1) was completely inhibited by TLX1, while the intensity of the remaining two (C2 and P2) was enhanced. Thus, TLX1-mediated regulation of ALDH1A1 in T cells, but not erythroid cells, is associated with an alteration of transcription factor binding at the GATA box.
Complex formation at the GATA box in both PER-117 and HEL cells was also strongly inhibited when a 35-fold molar excess of an unla-beled consensus GATA oligonucleotide (GATA; containing WGATAR with different flanking sequences to ALDH GATAB) was used ( Figure  5A). This indicated that GATA factors are present in all complexes formed in this assay. However, TLX1 does not appear to bind DNA at this site, since a 35-fold molar excess of an unlabeled consensus satellite 2 oligonucleotide (TLX1), capable of being bound by TLX1, 24 could not compete for complex formation ( Figure 5A). In agreement with this conclusion, no supershift or inhibition of complex formation was observed following addition of TLX1 antibody ( Figure 5B). Notably, however, PER-117 complex C1, which is abolished by TLX1, re-formed in the presence of TLX1 antibody, indicating that TLX1 directly contributes to the disruption of this low mobility complex.

TLX1 interacts with TFIIB
To help understand the transcriptional regulatory function of TLX1, we searched for binding partners using an immunoprecipitation strategy in leukemic T cells (ALL-SIL) that  aberrantly express TLX1 as a consequence of a 10q24 chromosomal translocation. A TLX1 antibody was employed to isolate naturally occurring nuclear protein complexes, which were separated by SDS-PAGE and silver stained ( Figure 6A). Excised gel slices representing discrete molecular mass intervals were digested with trypsin and analyzed by MALDI-TOF mass spectrometry in order to determine the identity of the protein bands. Table 1 summarizes the eight proteins detected. Among these were the centromeric proteins CENP-E and CENP-F, an intriguing finding given that TLX1 has previously been localized to centromeric regions. 24 Of particular interest was the detection of the general transcription factor TFIIB, which was identified by the presence of 15 peptides with sequence coverage of 49% (Table 1). To confirm that TLX1 could physically interact with TFIIB in a specific manner, we performed a GST pulldown assay ( Figure 6B). Glutathione-Sepharose beads containing GST-TLX1, GST-TLX1ΔH3, or GST alone were incubated with in vitro translated 35 S-labeled TFIIB or TLX1 protein. The latter was included as a positive control since TLX1 has previously been shown to homodimerize. 24 Following extensive washing, retention of TFIIB was observed with GST-TLX1 but not with control GST beads ( Figure 6B), demonstrating that TLX1 has the capacity to interact with TFIIB. Interestingly, retention of TFIIB, but not TLX1 control, was greatly diminished when GST-TLX1ΔH3 beads were used, indicating that helix 3 of the TLX1 homeodomain contributes strongly to the TFIIB interaction.

Discussion
TLX1 has previously been characterized as a DNA-binding factor that preferentially associates with the core sequence TAA T /GTG in vitro 22,23 and the similar sequence T A /GATTC present in satellite 2 DNA. 24 In addition, TLX1 can switch, in a cell type-and promoter contextdependent manner, between roles as activator and repressor, 26,30,38 however the mechanism(s) responsible for these divergent roles is poorly understood. In this study, we further investigated the function of TLX1 by analyzing its ability to transcriptionally regulate ALDH1A1, a gene previously identified as being TLX1dependent in developing mouse spleen as well as in several cell lines (PER-117, HEL, NIH-3T3), 26,30,49 and which contains a predicted TLX1 binding site, conserved between human and mouse, at -257 upstream of the transcriptional start site. In the first instance, our data confirmed ALDH1A1 as a regulatory target of TLX1, with the polarity of effect observed in terms of activation/repression being heavily dependent upon cell type. In PER-117 T cells, transient luciferase reporter assays with nested deletions of the ALDH1A1 promoter revealed that TLX1-mediated repression occurred in a CCAAT box-dependent manner involving an element located between -91 and -50. By contrast, TLX1 stimulated transcription in a CCAAT-independent manner from a proximal location (-91/+42) in human erythroleukemic  (HEL) cells. Transactivation, which was also observed in the T-cell background when a minimal promoter sequence was used (-50/+42), was abolished in HEL cells with the addition of extra DNA sequence (-146/-92) suggesting that a specific factor bound to this region is capable of abrogating the stimulatory potential of TLX1. Thus, TLX1 possessed two activities in respect to the ALDH1A1 promoter, namely CCAAT-dependent repression, which was cell type-specific and cryptic CCAAT¬independent activation, which was unmasked by deleting upstream regulatory sequences. Neither of these activities involved the TLX1 recognition site at -257, which initially suggested that TLX1 either bound a distinct sequence in vivo, as is the case with the regulation of its own promoter, 25 or that it acted via a non-DNA binding mechanism.

Article
Remarkably, the effect of TLX1 on ALDH1A1 promoter activity in the transient reporter assays was inversely related to its previously observed effect on endogenous ALDH1A1 levels in PER-117 and HEL cells. 30 This phenomenon has been observed previously in respect to other putative gene targets of TLX1, namely NR4A3, KIT and FHL1. 15,29,38 Although puzzling, the finding that a homeoprotein can act as a repressor or activator of transcription depending on promoter context is a common one. In many cases the activity that predominates has been found to be highly dependent on the nature of the cis-regulatory DNA sequence since this can, in turn, affect the interaction of homeoproteins with co-regulatory molecules such as TALE (three amino acid loop extension) homeoproteins, CBP (CREB-binding protein) or Groucho. [50][51][52][53][54] CCAAT-dependent repression by TLX1 was found to require the DNA recognition helix (helix 3) of the homeodomain. Intriguingly, the TLX1ΔH3 mutant lacking this helix was not only incapable of repressing ALDH1A1 transcription in PER-117 cells but switched to become an activator of transcription. This indicated that the repression/activation activities of TLX1 are separable with a structurally intact homeodomain being absolutely required for transcriptional repression, but not activation. Given these findings, a reasonable assumption was that TLX1 would repress ALDH1A1 transcription by directly binding promoter DNA at or near the CCAAT box. This crucial element is likely generally required for ALDH1A1 expression, since it was also identified as the major positive element in the Hep3B cell line and in Hepa-1 mouse hepatoma cells where it was bound by NF-Y and CCAAT/enhancer binding protein β (C/EBPβ), respectively. 44,55 Moreover, the CCAAT box is capable of being bound by the CCAAT binding transcription factor (CTF1/NFIC) with which TLX1 is known to interact in a functional manner. 56 However, EMSA assays using PER-117 (or HEL) nuclear extracts revealed that TLX1 did not directly bind the CCAAT box, nor did it affect protein   TLX1  TFIIB  GST  GST-TLX1  GST-TLX1ΔH3  GST-TLX1ΔH3   GST  GST-TLX1   Input (50%)  TLX1  TFIIB complex formation at this site. We therefore hypothesized that TLX1 may abrogate CCAATdependent transcription of ALDH1A1 by acting through the basal transcriptional apparatus. Supporting this notion, a previous study demonstrated that TLX1 is capable of repressing transcription via the RNA polymerase II holoenzyme in a manner that is DNA-independent, yet requires helix 3 of the homeodomain. 49 Indeed, we found that TLX1 altered the formation of GATA-containing protein complexes at the non-canonical TATA (GATA) box located at -33 on the ALDH1A1 promoter and that this occurred specifically in PER-117 cells where repression by TLX1 was observed. The GATA box is a dual regulatory site capable of binding both GATA factors and members of the basal transcriptional machinery, which is overrepresented among erythroid-specific gene promoters. Studies to date have indicated that the GATA box is a specialized/weakened TATA box 57 whose activity depends on interplay between binding of GATA factors and general transcription factors such as TFIID. [45][46][47][48] Based on our data, we therefore postulate that TLX1 can operate as a transcriptional repressor by altering the balance between specific factors binding at the GATA box without actually binding DNA itself. This, at least in part, may involve direct protein contacts by TLX1 since inclusion of anti-TLX1 antibody into EMSA binding reactions reduced TLX1-associated inhibition of one of the GATA box complexes identified. Intriguingly, the related HEX protein has been reported to interact with GATA2 to inhibit its binding to the flk-1/KDR gene, 58 suggesting that direct antagonism of GATA factors may be a feature of NKL homeoprotein function more generally. We further showed that TLX1 specifically interacts with TFIIB, a member of the general transcriptional machinery. TFIIB, which can bind promoter DNA in a sequence-specific manner at a TFIIB recognition element (BRE), plays a central role in pre-initiation complex (PIC) assembly, providing a bridge between promoter-bound TFIID and RNA polymerase II. This points to a mechanism to account for how TLX1 can inhibit basal transcription in a non-DNA binding manner, both in this study and that of Owens et al. 49 Specifically, TLX1 may directly affect repression of the ALDH1A1 promoter, which contains a BRE adjacent to the GATA box, by inhibiting the rate/extent of PIC formation via contacts with TFIIB. Notably, the TLX1-TFIIB interaction was reduced in the absence of homeodomain helix 3. Thus our results provide one potential mechanism to explain the switch in activity from repression to activation observed with the TLX1ΔH3 mutant.

Article
Significantly, it has been recently reported that TLX1, like many other homeoproteins, 52 can form a mutually inhibitory complex with the co-activator/histone acetyltransferase molecule CBP. 20 This interaction leads to sequestration of CBP at repressive chromatin domains, which is consistent with our previous finding that TLX1 can localize to heterochromatin via binding at satellite 2 DNA. 24 The TLX1-CBP interaction was also shown to require helix 3 of the TLX1 homeodomain, thereby suggesting an alternative, although not mutually exclusive, explanation for TLX1mediated repression and its reversal in the TLX1ΔH3 mutant. Riz et al. 20 further showed a correlation between the presence of TLX1 and a lack of CBP-associated acetylation of GATA1. As is the case for GATA2 and GATA3, 59,60 the DNA binding and transcriptional activity of GATA1 is heavily dependent on its acetylation status. 61 Taken together, we suggest a model in which interactions between TLX1 and both TFIIB and CBP lead to the transcriptional repression observed in our transient luciferase assays. Whereas contacts with TFIIB may inhibit the formation of a functional PIC, binding to CBP may prevent it functioning as a coactivator and acting to acetylate GATA factors. This would explain the substantially altered formation of GATA-containing protein complexes observed at the ALDH1A1 promoter GATA box, despite the lack of DNA binding by TLX1. It is conceivable that a general ability of TLX1 to indirectly regulate target genes by altering GATA factor activity, whether via CBP or an alternative mechanism, may be consequential for its role as an oncoprotein. This is particularly in view of evidence linking both a blockage in T-cell development and leukemogenesis to incapacitated GATA function. 62,63 Interestingly, a non-DNA binding mode of action for TLX1 that involves activity regulation of other transcription factors is reminiscent of another important T-ALL oncoprotein, SCL/TAL1, which induces leukemia by recruiting the co-repressor/histone deacetylase mSin3A to inhibit the transcriptional activity of E47/HEB. 64 There is little evidence to date showing that TLX1 can bind directly to natural DNA sequences to regulate target gene expression. Instead, TLX1 has been shown to operate indirectly by interacting with other proteins, most notably the phosphatases PP1 and PP2A to regulate gene cascades in various pathways such as RB/E2F and p107/MYC. 18,19 Consistent with this paradigm, our data confirm that TLX1 is capable of regulating ALDH1A1 expression in a non-DNA binding manner by affecting transcriptional complexes at the proximal ALDH1A1 promoter, although clearly, additional cis-regulatory elements (that may influence TLX1 protein interactions) are required to recapitulate the effect of TLX1 on endogenous ALDH1A1 gene expression. We showed that TLX1, like other homeodomain transcription factors including tinman/Nkx2-5, abdominal-A, Nkx6.1 and Vnd/Nkx2-2 51,53,54,65 is capable of acting as a bi-functional transcriptional regulator, whose activation and repression activities operate in a cell-type specific manner and via two distinct mechanisms. The first involves the ability of TLX1 to repress transcription, possibly by perturbing interactions between CCAAT box-binding transcriptional activators and proteins (GATA factors/basal transcriptional machinery) assembled at a non-canonical TATA (GATA) box. This activity does not appear to involve direct DNA binding, although a structurally intact homeodomain is required, presumably in order for TLX1 to interact with TFIIB and/or CBP. Chromatin immunoprecipitation assays to identify the specific transcription factor(s) bound at this composite CCAATbox/TLX1 responsive element and at the GATA box before and after TLX1 expression are required to substantiate this hypothesis. The second activity involves the ability of TLX1 to stimulate transcription through as yet unidentified regulatory elements in the proximal ALDH1A1 promoter and does not require an intact homeodomain. Our understanding of the role of TLX1 in normal development and in T-cell leukemogenesis is crucially dependent on deciphering the mechanisms by which TLX1 is capable of regulating gene expression. Future work to characterize TLX1 target genes and to fully define the protein participants involved in TLX1-mediated gene regulation will represent important steps towards this goal.