Human Ikaros Function in Activated T Cells Is Regulated by Coordinated Expression of Its Largest Isoforms*

The Ikaros gene is alternately spliced to generate multiple zinc finger proteins involved in gene regulation and chromatin remodeling. Whereas murine studies have provided important information regarding the role of Ikaros in the mouse, little is known of Ikaros function in human. We report functional analyses of the two largest human Ikaros (hIK) isoforms, hIK-VI and hIK-H, in T cells. Abundant expression of hIK-H, the largest described isoform, is restricted to human hematopoietic cells. We find that the DNA binding affinity of hIK-H differs from that of hIK-VI. Co-expression of hIk-H with hIk-VI alters the ability of Ikaros complexes to bind DNA motifs found in pericentromeric heterochromatin (PC-HC). In the nucleus, hIK-VI is localized solely in PC-HC, whereas the hIK-H protein exhibits dual centromeric and non-centromeric localization. Mutational analysis defined the amino acids responsible for the distinct DNA binding ability of hIK-H, as well as the sequence required for the specific subcellular localization of this isoform. In proliferating cells, the binding of hIK-H to the upstream regulatory region of known Ikaros target genes correlates with their positive regulation by Ikaros. Results suggest that expression of hIK-H protein restricts affinity of Ikaros protein complexes toward specific PC-HC repeats. We propose a model, whereby the binding of hIK-H-deficient Ikaros complexes to the regulatory sequence of target genes would recruit these genes to the restrictive pericentromeric compartment, resulting in their repression. The presence of hIK-H in the Ikaros complex would alter its affinity for PC-HC, leading to chromatin remodeling and activation of target genes.

The Ikaros gene is alternatively spliced to generate multiple isoforms. In murine lymphoid cells the most commonly expressed isoforms are IK-VI, which was initially thought to encode the full-length protein, and IK-V, which lacks the first N-terminal zinc finger (14,15); designated Ik-1 and Ik-2, respectively, in the nomenclature of Georgopoulous et al. (15). Ikaros is hypothesized to bind to DNA control elements of target genes and to aid in their recruitment to centromeric foci. This results in activation or repression of the target genes (16,17). Larger Ikaros isoforms like IK-V and IK-VI are thought to act synergistically and to be responsible for Ikaros function in chromatin remodeling and regulation of gene expression. Smaller Ikaros isoforms lacking N-terminal zinc fingers do not bind DNA. Heterodimers between large and small isoforms bind DNA poorly, suggesting that small isoforms can act as dominant negative inhibitors of Ikaros (18,19). Malignant transformation is hypothesized to be a direct consequence of altered (or diminished) function of the larger Ikaros isoforms because of overexpression of the smaller, dominant negative ones (18,19).
Previous studies of Ikaros function have been performed using murine T cell systems. Human studies identified a number of additional splice forms (6,18,20,21). Among these was IK-H, a splice variant of IK-VI that includes an additional 60 bases between exons 2 and 3 (Fig. 1A), thus making it the longest identified Ikaros isoform. (IK-H is designated Ik1ϩ in the nomenclature of Payne et al. (21,22)). Whereas IK-H protein is abundantly expressed in primary human B, NK, myeloid, and erythroid cells, it is barely detectable in primary murine hematopoietic cells (21,22).
Here we report the first functional studies of human Ikaros (hIK) 2 proteins. For our analyses we used T cells, thus allowing for more valid comparisons between our human data and that obtained in murine studies. We report that the largest human Ikaros isoforms, hIK-VI and hIK-H, exhibit distinct DNA binding abilities and subcellular localization patterns. Increased expression of the largest Ikaros isoform (hIK-H) during T cell activation determines the DNA binding specificity of Ikaros complexes toward repetitive sequences located within PC-HC. We propose a model whereby coordinated expression of the largest Ikaros isoforms regulates Ikaros function in chromatin remodeling. Our results expose unique properties of human Ikaros proteins and suggest that Ikaros function in human hematopoietic cells may occur through more complicated mechanisms than in the mouse.

EXPERIMENTAL PROCEDURES
Cells-Human CCRF-CEM (CEM) and MOLT-4 human leukemia T-cell lines, the Ramos B cell line, the human 293T endothelial kidney cell line, and the murine the NIH/3T3 (3T3) fibroblast cell line were obtained from American Type Culture Collection (ATCC). The murine 7OZ3 pre-B cell line, BAL17 B-cell line, and VL3-3M2 thymocyte cell line are a generous gift from the laboratory of Stephen Smale, HHMI, UCLA, Los Angeles, CA. Peripheral primary T cells were isolated from human donor peripheral blood (American Red Cross), or spleen of C57BL/6 mice using the nylon wool column method (23). T-cell activation with PMA and anti-CD3 has been described previously (23).
Antibodies-Specific antibody against hIK-H protein was generated by immunizing rabbits with KLH-conjugated peptide TYGADDFRDFHAIIIPKSF. The resulting rabbit antiserum was used at 1:1000 dilution. Antibodies used to detect the C terminus (IK-CTS) and the N terminus (IK-NTS) of murine and human Ikaros have been described previously (24). Anti-Ikaros antibodies were visualized using FITC-goat antirabbit IgG antibodies (Jackson ImmunoResearch, West Grove, PA). The HA-specific mouse monoclonal antibody HA.11 (Covance Research Products, Harrisburg, PA), visualized with Texas Red goat anti-mouse IgG (Jackson Immuno Research), was used to detect HA-tagged Ikaros isoforms. Anti-Helios antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA).
Plasmids, Transfection, and Retroviral Transduction-cDNAs for hIK-VI and hIK-H, were amplified by RT-PCR from human peripheral T-cell total RNA and cloned into the mammalian expression vector pcDNA3 (Invitrogen). Alanine substitution mutants for the hIK-H region N were generated using the QuikChange method (Stratagene, La Jolla, CA). For retrovirus generation, hIK-H or HA-tagged hIK-VI or hIK-H (HA tag in N terminus) were amplified by PCR and cloned between BglII and EcoRI sites of the MSCV IRES GFP (MIG) vector (a generous gift from the laboratory of David Baltimore, California Institute of Technology, Pasadena, CA). 293T cells were transfected via the calcium phosphate method. CEM, VL3-3M2 or 3T3 cells were infected with amphotropic retrovirus.
Confocal Microscopy-CEM, VL3-3M2, or 3T3 cells were infected with amphotropic retrovirus and analyzed by confocal microscopy as described previously (25). Images were acquired at room temperature by a Leica TCS-SP MP Confocal and Multiphoton Microscope with a Leica DM-LFS body (upright fixedstage microscope) using a ϫ100 Leica HX PLAPO (Planapo) oil immersion lens with numerical aperture of 1.4 (Heidelberg, Germany).
PCR conditions were 95°C for 5 min, followed by 30 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 45 s. Final PCR products were analyzed on agarose gels (2%) with ethidium bromide staining. All samples within a particular set (anti-Ikaros antibodies, IgG, and total chromatin control) were analyzed at the same time and electrophoresed on the same agarose gel. Results are representative of three independent experiments.

IK-H Is Differentially Expressed in Mouse and Human
Lymphoid Cells-Ikaros isoform H (IK-H) contains an additional 20 amino acids (region N) between exons 2 and 3 ( Fig.  1A) that is not present in IK-VI (21,32). Because abundant expression of IK-H has been detected in human, but not murine hematopoietic cells (22), it was important to study human Ikaros function in a cell type that expressed this isoform. We examined expression of Ikaros isoforms in human and murine lymphoid cell lines using an Ikaros antibody (IK-CTS) that recognizes the C terminus of both mouse and human Ikaros. IK-H was expressed in human T and B cell lines at a level comparable to IK-VI (Fig. 1B, lanes 1-3, whereas its expression was 10-fold lower than expression of IK-VI and IK-V in murine cell lines (Fig. 1B, lanes 4 -6). Identical results were obtained using another antibody (IK-NTS) that recognizes the N terminus of mouse and human Ikaros (data not shown).
To determine if the differing patterns of Ikaros isoform expression that we observed in murine and human cell lines were characteristic of normal T cells, Ikaros expression was examined in both resting and activated mouse and human primary T cells. In resting, human T cells we found expression of hIK-VI and hIk-V, but not hIK-H; however, following activation, all three isoforms were expressed at comparable levels ( Fig. 1C, lanes 1-4). Thus, the expression of Ikaros proteins in activated, primary, peripheral human T cells (Fig. 1C, lanes 3 and  4) is similar to that in human T-cell lines (Fig. 1B, lanes 1-3). Consistent with previous reports, in primary mouse T cells the expression of IK-H was 10-fold lower than expression of mIK-VI or mIK-V (Fig. 1C, lanes 5 and 6), in both resting and activated cells (22,33). Thus, human, but not murine, activated T cells express abundant IK-H.
To verify that the top two bands seen in human T cells were hIK-H and hIK-VI, each isoform was expressed in 293T cells (which do not express endogenous Ikaros), and used as size markers (Fig. 1D). Overexpressed isoforms co-migrated with endogenous proteins in the MOLT-4 human T-cell line, confirming that the two top bands were hIK-H and hIK-VI.
A polyclonal antibody against IK-H was prepared using an N region peptide as the epitope. Using MOLT-4 nuclear extract, we show that the IK-H antibody detects only the hIK-H isoform, whereas the IK-CTS antibody detected all of the three large human Ikaros isoforms (Fig. 1E). In studies reported here we will use the IK-H antibody to detect expression and localization of the hIK-H isoform.
Specific Amino Acids within Region N Determine the Reduced DNA Binding Ability of IK-H-The potential binding sites for Ikaros within human pericentromeric DNA have not been identified. In mice, the DNA sequence of PC-HC is uniform for all chromosomes. In humans, PC-HC contains different sequences, often unique for each chromosome (34). We designed DNA probes derived from published DNA sequences of human PC-HC and used them to test the DNA binding ability of individual Ikaros isoforms in an electrophoretic mobility shift assay (EMSA) (34 -40). Human hematopoietic cells express multiple Ikaros isoforms. Thus, to study DNA binding ability of individual isoforms, we used nuclear extracts of 293T cells transduced to express hIK-VI or hIK-H. The 293T cell line does not express Ikaros and it has been shown that transduced Ikaros has identical phosphorylation status and DNA binding abilities in 293T cells as in hematopoietic cells (41,42). Thus, 293T cells are an established model for studying DNA binding (25), protein-protein interactions (43), or phosphomimetic mutants of individual Ikaros isoforms (41,42). Both the hIK-VI and the hIK-H isoforms bound the IKBS4 probe that contains two optimal Ikaros binding sites ( Fig. 2A, lanes 1-3). Both isoforms also exhibited similar DNA binding affinities on probes derived from the ␥ satellite region of chromosome 8 (␥ sat 8) and the beta D repeat ( Fig. 2A, lanes 4 -9), both of which contained two Ikaros binding sites. However, the DNA binding ability of hIK-VI was stronger than that of hIK-H toward the probes derived from pericentromeric (PC-HC) that contains a single Ikaros consensus binding site: the ␥ satellite 8-2 region of human chromosome 8 (␥ sat 8-2): 8-fold stronger; the consensus ␣ satellite DNA sequence (␣ Sat): 10-fold stronger; and the CENP-B box: 6-fold stronger (Fig. 2B, lanes 1-9). A similar difference in DNA binding ability was observed when a probe derived from pJ alpha box was used (data not shown), while no binding was observed to the probe derived from ␥ satellite chromatin of human chromosome X (data not shown). These results reveal DNA binding specificities of human Ikaros within PC-HC and identified potential motifs that target Ikaros to the pericentromeric region of particular chromosomes.
To establish whether actual binding of Ikaros isoforms to pericentromeric heterochromatin has occurred in vivo in 293T cells, we performed ChIP assays. Results showed that hIK-VI, expressed individually in 293T cells, binds in vivo to the pericentromeric repeats that contain two Ikaros binding sites (␥ sat 8), as well as to the repeats that contain a single Ikaros binding site (␥ sat 8-2, ␣ sat and CENP-B) (Fig. 2C). ChIP assays show that the hIK-H isoform, expressed individually, binds in vivo a pericentromeric region that contains two Ikaros binding sites (␥ sat 8), but not regions with a single Ikaros binding site (Fig.  2C). These data confirm that results observed by EMSA correlate to in vivo binding of Ikaros isoforms in 293T cells.
We compared the DNA binding ability of murine and human Ikaros isoforms. The largest difference in binding ability was observed using a probe containing two Ikaros bindings motifs (from murine ␥ satellite A PC-HC) separated by 40 base pairs ( Fig. 2D, probe IKBS40; lanes [1][2][3][4]. Human IK-VI showed a DNA binding affinity over 10-fold that of murine IK-VI and 8 times that of hIK-H. Possible explanations for the observed data include an increased ability of hIK-VI to form multimers, which would enable it to bind to the probe containing two distant consensus sites, or increased affinity of hIK-VI toward a single Ikaros consensus site. To distinguish between these possibilities we tested the ability of Ikaros isoforms to bind a probe containing a single Ikaros binding site (supplemental Fig. S1A, lanes 1-4). The DNA binding ability of hIK-VI was 10-fold that of hIK-H, whereas mIK-VI was unable to bind the single site probe. These data suggest that the increased DNA binding ability of hIK-VI compared with mIK-VI and hIK-H is at least partly caused by increased affinity for the DNA contained in a single consensus Ikaros binding site.
Human IK-H has identical amino acid sequence to hIK-VI except for the presence of region N, yet hIK-H has decreased DNA binding ability compared with hIK-VI. If the presence of specific amino acids within region N interferes with DNA binding, mutation of these should give hIK-H the ability to bind DNA with affinity similar to hIK-VI. Alanine-scanning mutations of region N were performed, and mutants were tested for DNA binding ability. Initially, two consecutive amino acids within region N were mutated to alanine. Results (Fig. 2D, lanes 5-14 and supplemental Fig. S1A, lanes 5-14) identified two pairs of amino acids whose mutation to alanine gave hIK-H the ability to bind DNA. Point mutations of glycine, phenylalanine, and to a lesser extent, histidine, were able to restore the DNA binding ability of mutated hIK-H when either IKBS40 (Fig. 2E) or the single site probe (supplemental Fig. S1B) was used in EMSA. Mutation of tyrosine did not seem to affect DNA binding ability. Thus, three specific amino acids within region N (Fig. 2F) are responsible for decreased DNA binding of hIK-H compared with hIK-VI.
Human IK-H Exhibits Dual Centromeric and Noncentromeric Localization-In murine lymphocytes, Ikaros localizes to centromeric heterochromatin where it is hypothesized to recruit genes destined for inactivation (16). We used confocal microscopy to examine subcellular localization of the largest human Ikaros isoforms (Fig. 3). To distinguish the localization pattern of hIK-VI from hIK-H, the CEM leukemia T-cell line was transduced to express HA-tagged hIK-VI. Subcellular localization of hIK-VI was detected using antibodies against the HA tag (red), whereas localization of endogenous hIK-H was detected with isoform-specific IK-H antibodies (green).
In CEM T cells, the hIK-VI isoform displayed the punctate distribution pattern (Fig. 3A) observed for pericentromeric localization of Ikaros in murine and human lymphoid cells (16,18,20,25,41). Surprisingly, hIK-H showed both centromeric and non-centromeric localization (Fig. 3A). The co-stained image revealed patches of hIK-H co-localizing with hIK-VI (yellow in Fig. 3A overlay), but the majority of hIK-H was distributed outside of centromeric heterochromatin (green in Fig.  3A overlay). It appears that the hIK-H located outside of centromeric heterochromatin is localized to specific nuclear structures, because it does not exhibit a diffuse nuclear staining pattern typical for other transcription factors.
The unexpected localization pattern observed for hIK-H could be caused by the unique properties of this isoform, or to the unique characteristics of human heterochromatin. To distinguish between these two possibilities, we studied localization of human Ikaros isoforms in the murine VL3-3M2 cells, a T-cell line that has been used to document centromeric localization of murine Ikaros (41,44). The localization patterns of both hIK-VI and hIK-H isoforms in VL3-3M2 cells (Fig. 3B) closely resembles the localization patterns observed for these isoforms in human cells (Fig. 3A). These data suggest that the localization pattern observed for hIK-H is caused by unique properties of the hIK-H protein rather than species-specific differences in heterochromatin.
Centromeric localization of Ikaros has been reported to correlate with its ability to bind DNA; a 3-4-fold decrease in DNA binding affinity results in the loss of centromeric heterochromatin localization (25,41). The decreased DNA binding ability of hIK-H, along with its subcellular distribution, raised the question of whether hIK-H is able to localize to the centromeric region in the absence of the hIK-VI isoform. To address this question, we used NIH 3T3 fibroblasts, which do not express Ikaros. These are an established in vivo system used to study introduced Ikaros isoforms or phosphomimetic mutants (25,28,41,42). In 3T3 cells, hIK-VI retained its centromeric localization, surprisingly without non-centromeric distribution. This was documented with both anti-IK-H and anti-HA tag antibodies (Fig. 3C). These data suggest that: 1) hIK-H is able to localize to centromeric regions of heterochromatin without forming complexes with hIK-VI and 2) non-centromeric localization of hIK-H observed in human lymphoid cells is tissue specific (observed in T cells, but not fibroblasts), but not species-specific (observed in both mouse and human cells).
Alanine mutations of specific amino acids within the N region were able to restore the DNA binding ability of hIK-H to a level comparable to hIK-VI (Fig. 2). We tested whether restored DNA binding ability alters subcellular localization of hIK-H. HA-tagged hIK-H mutants were expressed in the CEM T-cell line and their localization was compared with endogenous hIK-H. Both alanine mutants retained non-centromeric distribution (Fig. 3, D and E, left panels), although this distribution appeared to be slightly reduced as compared with endogenous wild type hIK-H (Fig. 3, D and E, middle panels). In contrast, deletion of either the ten N-terminal amino acids (DEL1) or the ten C-terminal amino acids (DEL2) of region N abolishes non-centromeric localization of hIK-H (Fig. 3, F-G). DEL1 and DEL2 mutants were able to bind DNA comparably to hIK-VI (data not shown). These data suggest that DNA binding and subcellular localization of hIK-H might be determined by different domains within Region N.
Human IK-H Regulates the DNA Binding Ability of the hIK-VI Isoform-Our experiments showed that hIK-H isoform has a strong DNA binding affinity toward DNA containing two consensus Ikaros binding sites, but a weak affinity toward probes containing a single Ikaros binding site (Fig. 2). This raised the question of whether the presence of hIK-H can interfere with the ability of hIK-VI to bind DNA. To answer this question, we tested the ability of recombinant hIK-VI to bind probes containing one (␣ sat) or two consensus binding sites (␥ sat 8) in the presence or absence of the hIK-H isoform. An increasing amount of recombinant hIK-H was added to a DNA binding reaction that contained a constant amount of hIK-VI and DNA probe, and the effect of hIk-H on Ikaros DNA binding was measured by gel shift assay. Results showed that hIK-H protein acts synergistically with hIK-VI to bind ␥ sat 8 probe (Fig. 4A). In contrast, the presence of a small amount of recombinant hIK-H (compared with hIK-VI) abolished completely the ability of hIK-VI to bind the ␣ sat probe (Fig. 4B). These data show that, depending on the DNA target sequence, hIK-H can serve to potentiate or to inhibit the DNA binding function of the other large Ikaros isoform. This suggests that in vitro, hIK-H does not function as a typical dominant-negative isoform (similar to the small Ikaros isoforms that lack DNA binding zinc fingers), but rather as an additional control mechanism for determining the specificity of Ikaros DNA binding.
Human IK-H Binds DNA as a Multimeric Complex-Previous experiments were performed with recombinant Ikaros proteins. We wanted to study the DNA binding activity of Ikaros in the more complex environment of human lymphoid cells where multiple Ikaros isoforms are abundantly expressed. First, we tested whether hIK-H associates with other Ikaros isoforms in lymphoid cells. Immunoprecipitation reactions were performed with nuclear extract from the MOLT-4 T-cell line and using increasing amounts of anti-IK-H antibodies (Fig. 5A). Amounts of the various Ikaros isoforms within immunoprecipitation pellets and supernatants were determined by Western blot. IK-H antibody was able to immunoprecipitate hIK-VI and hIK-V (Fig. 5B, lanes 1-4) from MOLT-4 T cells, thus confirming a stable interaction between hIK-H and each of these proteins. IK-H antibodies depleted all of the hIK-V isoform from MOLT-4 T cell extract, but substantial amounts of hIK-VI remain unassociated with hIK-H, as evidenced by their presence in immunoprecipitation supernatant (Fig. 5B, lanes  5-8). Similar results were obtained with nuclear extracts from the CEM T-cell line (data not shown).
The next step was to test whether human Ikaros isoforms bind DNA as homodimers or in complexes with each other. We performed gel shift assays using nuclear extract of activated peripheral human T cells. Gel shift assays performed with the ␥ satellite 8 probe (Fig. 5B) revealed the presence of two DNA binding complexes that contain Ikaros (Fig. 5B, lane 2, arrows a  and b). Polyclonal antibodies against the common C-terminal domain or the common N-terminal domain of Ikaros com-pletely altered migration of the more abundant and rapidly migrating complex (Fig. 5B, lanes 3 and 4). The presence of antibodies against hIK-H produced a supershift, suggesting that hIK-H also binds DNA (Fig. 5A). However, in the presence of antibodies against hIK-H, the rapidly migrating complex was still abundant suggesting that other large Ikaros isoforms (e.g. hIK-VI and hIK-V) were the proteins mostly responsible for DNA binding to this particular probe. Similar results were obtained when nuclear extracts from the human T cell leukemia cell lines MOLT-4 and CEM were used (data not shown).
The above results suggest that hIk-H binds DNA in activated T cells primarily as a part of a multimeric complex with other Ikaros isoforms. We wanted to know whether the presence of hIK-H in multimeric Ikaros complexes affects binding specificity of these complexes toward specific DNA sequences present in PC-HC. To address this question we determined if expression of hIK-H interferes with the ability of Ikaros complexes to bind PC-HC probes to which hIK-H is unable to bind. Activated T cells endogenously express similar levels of hIK-VI and FIGURE 4. Interaction between hIK-VI and hIK-H regulates DNA binding affinity of Ikaros complexes. Gel shift assays were performed using (A) ␥ sat 8, a probe with two consensus Ikaros binding sites or (B) ␣ sat, a probe with a single Ikaros binding site. GST-hIK-VI (3 g) was mixed for 30 min at 4°C with increasing amounts of GST-hIK-H (0.3, 0.6, 1.2, 2.5 g, lanes 6 -9), and DNA binding was assayed by gel shift. As a control (lanes 2-5), GST-hIK-VI was also mixed with the same amount of GST-hIK-VI.
hIK-H (Fig. 1). So we compared the DNA binding ability of Ikaros complexes in nuclear extract from activated T cells to that of hIK-VI and hIK-H expressed separately in 293T cells.
Results show that the ability of Ikaros complexes containing hIK-H (activated T cells, Fig. 5C, lanes 3, 6, and 9) to bind PC-HC probes was reduced 5-fold (␥ sat 8-2) or abolished (␣ sat and CENP-B probes) as compared with that of complexes without hIK-H (Fig. 5C, lanes 1, 4, and 7). When both hIK-VI and hIK-H were co-expressed in 293T cells (data not shown) results were similar to those observed with nuclear extracts of activated T cells. This provides evidence that the presence of hIK-H, rather than other factors unique to activated T cells, modifies the ability of Ikaros complexes to bind pericentromeric DNA.
To confirm these data, in vivo DNA binding of Ikaros isoforms in activated T cells was compared with that in 293T cells that express only hIK-VI, or both hIK-VI and hIK-H by ChIP assay (Fig. 5D). Results were similar to the data obtained with EMSA. Co-expression of hIK-VI and hIK-H in both activated T cells and 293T cells abolished DNA binding ability of Ikaros complexes to PC-HC regions that contain a single Ikaros binding site (␥ sat 8-2, ␣ sat, and CENP-B), while expression of hIK-VI alone preserved DNA binding ability of Ikaros to these regions. The in vivo DNA binding ability of co-expressed hIK-VI and hIK-H isoforms in 293T cells was similar to that observed for physiologically expressed Ikaros isoforms in activated T cells. These data provide further validation of 293T cells as a model for the study of Ikaros function. These data, along with results presented in Fig. 4, suggest that hIK-H can function in determining the binding specificity of Ikaros complexes for PC-HC.
Preferential Binding of hIK-H to Regulatory Elements of Ikaros Target Genes-Next, we wanted to determine whether Ikaros isoforms exhibit unique DNA binding specificities with respect to regulatory sequences of genes that are known targets of Ikaros. Probes derived from upstream regulatory elements of five genes whose expression has been shown to be regulated by Ikaros were used in DNA binding experiments with nuclear extracts from resting and activated T cells. The presence of various Ikaros isoforms and/or hIK-H was documented by supershift with anti-IK-CTS and anti-IK-H antibodies, respectively. DNA binding reactions using upstream regulatory elements for IkCa1, Granzyme B, VPAC-1 receptor, FAAH, and STAT4 genes showed the presence of protein complexes containing Ikaros isoforms (evidenced by supershift with anti-IK-CTS antibodies) bound to regulatory elements in activated, but not in resting T cells (Fig. 6 and data not shown). In DNA binding reactions with probes for IKCal and Granzyme B, antibodies against hIK-H produced a complete supershift, similar to that produced by anti IK-CTS ( Fig. 6A and lanes 3 and 4). Similar results were seen for the FAAH and STAT4 probes (data not shown). This suggests that hIK-H binds strongly to the upstream elements of these genes and that most complexes contain hIK-H. However, when the probe derived from the upstream region of the VPAC-1 receptor was used, the Ikaros protein complex was supershifted with anti-IK-CTS antibodies (Fig. 6C, lane 3), but not with anti-IK-H antibodies (Fig. 6C,  lane 4). This suggests that the hIK-H isoform binds poorly to the upstream region of VPAC-1 receptor gene and that most of the proteins bound to this region are hIK-VI or hIK-V isoforms.
ChIP assay was used to study in vivo binding of hIK-H to the regulatory elements of Ikaros target genes in activated T cells (Fig. 6D). Results showed that Ikaros proteins bind to the It has been demonstrated that Ikaros positively regulates expression of IkCa1 (27), Granzyme B (26,30), and STAT4 (29) genes, but negatively regulates expression of the VPAC-1 receptor gene (28). The above data demonstrate that Ikaros isoforms have differing abilities to bind upstream elements of Ikaros target genes. These results also suggest that the hIK-H isoform has the potential to affect expression of the target gene in protein complexes bound to the upstream regulatory element.

DISCUSSION
In hematopoietic cells, Ikaros binds upstream regulatory regions of target genes and aids in their recruitment to PC-HC.
This process leads to either activation or repression of transcription of these target genes. Thus, DNA binding ability and centromeric localization are the essence of Ikaros activity. In mouse, ␥ satellite repeats are abundant at the centromeres of all murine chromosomes, and Ikaros has been shown to bind motifs derived from their sequence (25). In contrast, human PC-HC contains a variety of DNA motifs. These DNA sequences exhibit much higher diversity than their murine counterparts. Most motifs in human PC-HC are present only at specific loci of individual chromosomes. The presence of diverse motifs in human PC-HC poses a challenge for appropriate Ikaros function. In mouse, differential expression of isoforms controls Ikaros DNA binding ability. Here, we report that expression of hIK-H, the largest Ikaros isoform, provides an additional mechanism for regulating the DNA binding activity that allows Ikaros to function in the more complex heterochromatin environment present in human T cells.
hIK-H Modulates the DNA Binding Affinity of Ikaros Complexes-The 20-amino acid N region that distinguishes hIK-H from hIK-VI decreases the ability of hIK-H to bind Ikaros binding sites within PC-HC (Figs. 2 and 3). Whereas both hIK-VI and hIK-H have similar affinity for DNA sequences containing two Ikaros consensus sequences (TGG-GAA/T), hIK-H exhibits lower affinity toward DNA containing a single recognition site (Fig. 2). This difference becomes more pronounced when probes contain weaker "core" recognition sites (e.g. GGGA or GGAA). A possible reason for the decreased DNA binding ability of hIK-H could be that the N region creates conformational changes in the Ikaros protein leading to decreased DNA binding affinity. Alternatively, amino acids within region N could interact with other protein(s) whose presence interferes with the ability of hIK-H to bind DNA. Our data do not distinguish between these two possibilities, but we have identified specific amino acids responsible for altered DNA binding affinity (Figs. 2 and 3). Using probes derived from various repetitive sequences from human satellite DNA we have also determined potential binding targets for Ikaros within PC-HC (Figs. 2 and 5).
Co-expression of hIK-H and hIK-VI enhanced the ability of Ikaros proteins to bind DNA containing strong recognition sites (Fig. 4A), but abolished binding to weak sites (Fig. 4B). The in vitro experiments with purified GST-hIK-H shown in Fig. 4 demonstrate that the ability of hIK-H to influence DNA binding by hIK-VI does not require the presence of additional proteins.
hIK-H Exhibits a Unique Subcellular Localization Pattern-Amino acids within region N also determine the unique subcellular localization of hIK-H, which includes both centromeric and non-centromeric locations (Fig. 3). Surprisingly, double alanine mutations in region N restore to hIK-H the DNA binding characteristics, but not the subcellular localization pattern, of hIK-VI (Figs. 2 and 3). In contrast, deletion of 10 amino acids from either the N or C terminus of region N abolishes noncentromeric localization of hIK-H (Fig. 3). These data suggest that DNA binding and subcellular localization of hIk-H are determined by different domains within region N. The dual, centromeric and non-centromeric localization of hIK-H appears to be tissue-specific (Fig. 3). The simplest explanation for the observed results is that hIK-H interacts with an unknown protein via a domain located within region N and that this interaction determines its non-centromeric localization. We hypothesize that some fraction of the hIK-H within the cell associates with other Ikaros family members, binds DNA, and localizes in PC-HC. The remaining hIK-H interacts with an unknown protein(s) in a tissue-specific manner and localizes in non-centromeric regions of the nucleus where it might bind DNA and/or have other functions, unique to this isoform. It is possible that deletion of ten amino acids within region N significantly alters the conformation of hIK-H and/or abolishes a protein-protein interaction site, thereby restoring to mutant hIK-H the centromeric-only localization observed for hIK-VI.
hIK-H as a Regulator of Ikaros Function-All of the murine chromosomes contain abundant ␥ satellite repeats at pericentromeric heterochromatin. These repeats contain multiple consensus Ikaros binding sites (25). Human centromeres consist of heterogeneous repetitive DNAs. Most of these repeats are located at specific loci of different chromosomes (45). Thus, human chromatin presents a unique challenge for Ikaros function during cellular proliferation and as a regulator of gene expression. We propose that hIK-H provides a mechanism whereby Ikaros function is maintained in the more complex human chromatin environment.
First, Ikaros complexes must maintain accuracy and fidelity in binding to heterogeneous human PC-HC during cellular proliferation. While some human repeats contain motifs with multiple Ikaros consensus binding sites (e.g. ␥ satellite at chromosome 8), many contain a single Ikaros binding site and/or lower affinity binding site (e.g. CENP-B, satellite-3, etc.). We show that hIK-VI binds single-site probes with higher affinity than its murine (Fig. 2) counterpart. This ability allows Ikaros to bind to PC-HC of different chromosomes as a hIK-VI homodimer. We also show that expression of hIK-H can regulate DNA binding affinity of Ikaros complexes (Figs. 5 and 6).
During T-cell activation, expression of hIK-H is up-regulated. The presence of hIK-H increases the ability of Ikaros complexes to bind certain repetitive motifs in PC-HC, while abolishing the ability to bind others. We suggest that these changes in Ikaros activity contribute to chromatin remodeling leading to the "active state" that is characteristic of cycling cells. Support for this model comes from our observation that hIK-H and hIK-VI are similarly expressed in activated T cells (Fig. 1C), in the Ramos human B-lymphoma cell line, and the CEM early T-cell line (Fig. 1B), but not in quiescent T cells (Fig. 1C) or B cells (data not shown). Furthermore, we have also found that induction of cell cycle arrest in human B cell lymphoma is accompanied by loss of expression of hIK-H, while expression of hIK-VI remained unchanged. 3 Second, hIK-H functions to directly regulate expression of target genes by regulating the affinity of Ikaros complexes for PC-HC. The mechanism by which human Ikaros activates or represses target genes is still unknown. For the murine system, convincing models have been proposed to explain how Ikaros can function as an activator (potentiator) or repressor of gene expression (46). Post-translational modifications of Ikaros have an important role in determining Ikaros function (42,47). Here we present data showing that hIK-H forms part of Ikaros complexes that bind to the upstream regulatory sequence of the four genes whose expression is positively regulated by Ikaros (Fig. 6, A and B and data not shown). In contrast, hIK-H is absent from Ikaros complexes that causes transcriptional repression of the VPAC-1 receptor (Fig. 6, C and D). Based on data presented here, we propose a model for the mechanism by which Ikaros regulates expression of target genes in human cells. We propose that the presence of hIK-H in the Ikaros complex is a critical factor in the regulation of target genes (Fig. 7). In the absence of hIK-H, hIK-VI recruits the target gene into PC-HC, while binding the PC-HC tightly at multiple sites, thus creating an environment non-permissible to transcription (Fig. 7A). Increasing levels of hIK-H in the Ikaros complex results in the recruitment of the target gene into less restricted forms of heterochromatin. Thus allowing activating proteins to bind additional upstream 3 Z. Gurel and S. Dovat, unpublished data. sites and resulting in transcriptional activation of target genes (Fig. 7, B and C). This model postulates a critical role for hIK-H expression, as well as post-translational modification of Ikaros isoforms (e.g. phosphorylation, sumoylation), in determining Ikaros activity. A limitation of the proposed model is that it is based on data obtained from the only five genes whose expression has been shown to be regulated by Ikaros in activated human T cells. As more Ikaros target genes are identified, the validity of this model can be tested further.
The unique properties of hIK-H that we describe are because of the 20 amino acids of region N. This raises the possibility that other Ikaros isoforms containing region N might have distinct biological properties. Ikaros isoform X (IK-X), an Ikaros isoform that lacks exon 5, is selectively expressed during myeloid differentiation and in myeloid lineage cells. A form of IK-X that includes region N (IK-Xϩ) is abundant in both murine and human myeloid lineage cells and has been postulated to play a role in myeloid differentiation (21,22).
The differing features of hIK-VI and hIK-H, coupled with the disparity between IK-H expression in human and mouse tissues, suggest that the mechanisms by which Ikaros regulates hematopoiesis in humans are distinct and more complex than in the mouse. Determining the function of Ikaros proteins will aid in understanding the role of Ikaros as a tumor suppressor and shed light on the mechanisms of malignant transformation and immune response.