Phosphorylation-Dependent Interaction of SATB1 and PIAS1 Directs SUMO-Regulated Caspase Cleavage of SATB1

Abstract

Special AT-rich sequence-binding protein 1 (SATB1) is a tissue-restricted genome organizer that provides a key link between DNA loop organization, chromatin modification/remodeling, and transcription factor association at matrix attachment regions (MARs). The SUMO E3 ligase PIAS1 enhances SUMO conjugation to SATB1 lysine-744, and this modification regulates caspase-6 mediated cleavage of SATB1 at promyelocytic leukemia nuclear bodies (PML NBs). Since this regulated caspase cleavage occurs on only a subset of SATB1, and the products are relatively stable, proteolysis likely mediates cellular processes other than programmed cell death. However, the mechanism for the spatial and temporal regulation of SATB1 sumoylation and caspase cleavage is not known. Here we report that these processes are controlled by SATB1 phosphorylation; specifically, PIAS1 interaction with SATB1 is inhibited by phosphorylation. Mutagenesis studies identified interaction of the PIAS SAP (scaffold attachment factor-A/B/acinus/PIAS) motif with SATB1 N-terminal sequences. Notably, phosphorylation of SATB1 at threonine-188 regulates its interaction with PIAS1. Sequences near this phosphorylation site, LXXLL (residues 193 to 197), appear to be conserved among a subset of SUMO substrate proteins. Thus, this motif may be commonly involved in interaction with the PIAS SAP, and phosphorylation may similarly inhibit some of these substrates by preventing their interaction with the ligase.

Posttranslational protein modification is a critical means to regulate cellular processes. The conjugation of SUMO to specific substrates mediates numerous cellular processes, including intracellular transport, protein stabilization, regulation of transcription, and apoptosis (18). Conjugation of SUMO to target molecules requires a single E1 to activate SUMO and a unique E2 conjugating enzyme (Ubc9) that, in combination with one of the few known E3 ligases, directs conjugation and ensures target specificity (36, 45, 51, 57). Known E3 ligases for SUMO conjugation include the protein inhibitor of activated STAT (PIAS) family of proteins (PIAS1, PIAS2 [PIASx, α and β forms], PIAS3, and PIAS4 [PIASy]), Ran binding protein 2 (RanBP2), the polycomb group protein (Pc2), and topoisomerase I- and p53-binding protein (TOPORS) (31). Additional SUMO E3 ligases likely will be identified in the near future. PIAS family members influence the function of many transcription factors and of proteins involved in signal transduction, through their action as SUMO E3 ligases, by recruiting transcriptional corepressors or coactivators, and/or by blocking the DNA-binding activity of transcription factors (54, 57, 59). Numerous domains on PIAS recognize distinct sequences or conformations on target proteins, unique DNA structures, or specific “bridging” molecules to mediate these various functions (see Fig. 2D). The PIAS SIM (SUMO interaction motif) recognizes SUMO moieties of modified substrates and alters subnuclear targeting and/or assembly of transcription complexes (15, 23, 35). The PIAS SAP motif interacts with AT-rich DNA, including MAR elements, but also has a role in substrate recognition (2, 25, 45, 67), as well as in cellular localization of partner proteins, regardless of SUMO status (28, 57, 67). The PINIT domain contributes to substrate selectivity (45, 59). The SP-RING motif (Siz/PIAS-RING) interacts with Ubc9 and has a role in substrate recognition (6, 13, 23, 25, 52). The function of the poorly conserved serine/threonine-rich C terminus is unknown.

FIG. 2.

The N-terminal SAP domain of PIAS1 interacts directly with SATB1. (A) Jurkat WCEs (1,000 μg) were immunoprecipitated with preimmune (pre-I) serum, anti-PIAS1 (rabbit) antibody, or anti-SATB1 (12-9) antibody, and the precipitates were examined by SDS-PAGE and Western blotting, using a peptide anti-SATB1 (16-9) or anti-PIAS1 (goat) antibody. (B) Jurkat cells were treated for 2 h with anti-Fas; then WCEs were prepared and analyzed as described above by coimmunoprecipitation and immunoblotting to the antibodies indicated. To standardize results in lanes 4 and 5, equivalent amounts of SATB1(FL) were precipitated from cells with SATB1 or PIAS1 antibodies (lanes 6 and 7). (C) pcDNA3.1/His-SATB1(FL) and EGFP-PIAS1(FL) were coexpressed 24 h in Nalm-6 cells. WCEs (1,000 μg) were immunoprecipitated with anti-EGFP antibody, anti-SATB1 (12-9) antibody, anti-PIAS1 antibody, or pre-I serum. Precipitates were fractionated by SDS-PAGE, immobilized on transfer membranes, and detected with anti-PIAS1 or anti-SATB1 (12-9) antibody. The middle panel is a longer and expanded exposure of the top panel. (D) Schematic representation of PIAS1 functional domains. PIAS family members isolated in the y2h screen employing SATB1 as bait (and the amino acids encoded by the clones) are depicted. (E) [³⁵S]Met-SATB1(FL) was incubated with equivalent amounts (0.2 μg) of the GST fusion proteins shown or with GST alone (control). The [³⁵S]Met-labeled pulled-down proteins were fractionated by SDS-PAGE adjacent to an aliquot of the radioactive test protein (input, 10% of experimental amount). Gels were dried and exposed to Kodak Biomax MR single-emulsion film for 2 days. (F) [³⁵S]Met-SATB1(FL) and [³⁵S]Met-luciferase were incubated with equal amounts of the GST-PIAS1 fusion proteins shown, or with GST alone, and analyzed as described above for panel E.

The scarcity of SUMO conjugating and ligating enzymes suggests that the substrate confers additional levels of regulation. Reversible posttranslational modification (phosphorylation, acetylation) of specific substrate residues could affect sumoylation by recruiting isopeptidases, blocking/recruiting E3 ligase(s), sequestering substrate to a site that is conducive or not to its sumoylation, or by preventing SUMO modification in some unknown manner. Treatment of cells with drugs that enhance phosphorylation of the tumor protein p53 at ser-20 impairs both the association of Ubc9 with p53 and subsequent SUMO conjugation (29, 39). Similarly, inhibition of phosphorylation at ser-63 and ser-73 of c-Jun prevents its modification by SUMO (39). Phosphorylation of Elk-1 (ETS [E twenty-six]-like kinase 1) and PML regulates SUMO conjugation by unknown mechanisms (1). More recently, the SUMO consensus modification site (ψKXE) itself has been shown to be more complex than originally thought. Residues downstream of the ψKXE motif provide additional specificity determinants. NDSMs [negatively charged amino acid-dependent SUMOylation motif; ψKXEXX(D/E)₄] provide a negative charge to two or more of the four residues immediately following the ψKXEXX, and this promotes SUMO conjugation to substrates (71). The NDSM appears to be a variation of the PDSM (phosphorylation-dependent SUMOylation motifs; ψKXEXXSP) (1, 71). PDSMs encode a positively acting proline-directed phosphorylation site (SP) that provides a means to control the SUMOylation status of numerous substrates, particularly those with a role in transcriptional regulation (1, 19).

SATB1 interacts at MAR elements and is a key regulator involved in T-cell development. SATB1 is expressed at very high levels in thymocytes, but also in aggressive breast cancer cells, brain, and myeloid progenitors where it regulates transcription of genes that directly impact cellular proliferation and activation (7, 17, 42, 61). An important means of regulating immune cell function is via caspase cleavage, which usually leads to inflammation or immune-cell death following disruption of essential cellular functions; however, it has recently become apparent that caspases also contribute to specific aspects of immune-cell development, activation, and differentiation by nonapoptotic means (27). Upwards of 400 proteins are known to undergo cleavage by caspases (33). SATB1 is cleaved by caspase-6 early following induction of apoptosis and also under nonapoptotic conditions that permit stimulated B lymphocytes to enter the cell cycle (11, 14, 44). Caspase cleavage is regulated by covalent conjugation of SUMO to SATB1, a process which is augmented by the E3 ligase PIAS1 and enhanced in the presence of intact PML NBs (64).

It is unknown how the MAR-associating protein SATB1 detects cellular signals that lead to its nuclear relocation, modification by SUMO, and subsequent caspase cleavage, an event that regulates additional cellular processes (44). Here we show that the SUMO E3 ligase PIAS1 functionally interacts with N-terminal sequences of SATB1, specifically an LXXLL motif, to mediate subnuclear relocalization of this global gene regulator into PML NBs and to enhance nonapoptotic caspase cleavage. Phosphorylation regulates this interaction and therefore subsequent downstream events. Additionally, the interaction domains of many PIAS partner proteins encode LXXLL motifs that have the potential to be regulated by phosphorylation.

MATERIALS AND METHODS

Plasmid construction.

SATB1 cDNA was inserted into the pLexA vector (Clontech Laboratories, Inc.) as described previously (64). Deletion constructs and specific site mutations of SATB1, as well as insertion of SUMO-1 or -3 in frame with the C terminus of SATB1, were done using standard molecular biology techniques (4). SATB1 cDNA and deletions/mutations were subsequently inserted into pBluescript II (K/S) (Stratagene), pEGFP-C1, or pcDNA3.1/His for additional studies.

Ubc9, SUMO-1, and SUMO-3 were amplified by PCR from the Jurkat cDNA library using the primers as described previously (6, 65). The N- and the C-terminal halves of PIAS1 [PIAS(1-307), PIAS1(316-651)], as well as the SAP and RING motifs, were excised from the full-length protein. Complementary DNAs encoding truncated PIAS3, PIASy, and EDD were excised from pB42AD with NotI and XhoI, the ends were filled in with Klenow large fragment, and the clones were inserted in frame with glutathione S-transferase (GST) in pGEX-2T or pGEX-4T (Amersham Pharmacia Biotech) [= pGEX-PIAS3(185-619) and pGEX-PIASy(370-510)] or with enhanced green fluorescent protein (EGFP) in the mammalian expression vector pEGFP-C1 (Clontech). All clones were analyzed for correct sequences. The pGEX and pET-28 constructs were transfected into the bacterial strain BL21(DE3) for fusion protein preparation (see below).

Yeast two-hybrid library screen and β-galactosidase protein interaction assays.

Yeast two-hybrid (y2h) library screening was done as described previously (64). To assess interaction of select proteins, a nutritional plate assay was used in which one partner was cloned as a fusion with the GAL4 DNA binding domain (DBD) and the other was cloned as a fusion with the activation domain (AD). The two constructs were expressed in Saccharomyces cerevisiae of opposite mating types; the yeasts were then mated and plated on media to select for cells expressing both interaction partners (SD-His/Trp/Ura, where D is dextrose). Subsequently, the yeast was grown on inducing media (SGR-his/Trp/Ura/Leu, where G is galactose and R is raffinose) to establish whether the protein partners interacted (YPH; Clontech). In another assay, the relative strengths of interaction between SATB1(58-763) and the positive y2h clones were measured using the quantitative liquid β-galactosidase assay (YPH; Clontech). Independent colonies from representative y2h-positive clones were grown overnight in selective media employing dextrose:raffinose (0.5%:2%; raffinose is a noninducing sugar) as the carbon source. The cells were washed and then induced 6 h in galactose:raffinose (2%:0.5%). Each independent colony was assayed in duplicate, using o-nitrophenyl-β-d-galactopyraniside (ONPG) (Sigma) as the substrate. The assay was repeated at least three times for each y2h clone. Positive controls were pLexA-Pos (a single plasmid control) and pLexA-53 plus pB42AD-T (tests for interaction of p53 with the large T antigen). Negative controls were pLexA plus pB42AD, pLexA-53 plus pB42AD, and pLexA plus pB42AD-T, none of which produce fusion proteins that can activate the β-galactosidase promoter.

Protein pulldown assays.

Glutathione S-transferase fusion proteins were prepared from the pGEX-cDNA or pET-28-cDNA construct and purified on glutathione Sepharose 4B beads (Amersham Pharmacia Biotech, Inc.) using standard protocols. GST-PIAS fusion proteins were quantified by comparing dilutions of each GST fusion to known amounts of bovine serum albumin (BSA) on SDS-PAGE gels that were subsequently stained with Coomassie blue. Subsequently, 0.2 μg of the GST fusion was used for each pull-down assay. SATB1 and SATB1 deletions were labeled with [³⁵S]methionine ([³⁵S]Met) in the T_NT-coupled transcription/translation system (Promega). Protein pulldown assays were done as described previously (36, 37). Precipitated products were suspended in SDS sample buffer (20 μl), mixed on a vortexer, boiled 5 min, and mixed again. Samples were collected by brief centrifugation and fractionated on denaturing SDS polyacrylamide gels. The gels were dried and exposed to Kodak Biomax MR single-emulsion film.

In vitro SUMO-1 conjugation assay.

SATB1 proteins with mutations and deletions were labeled with [³⁵S]Met in the T_NT-coupled transcription/translation system (Promega) and used directly in the in vitro conjugation assay as described previously (9). Recombinant AOS1/UBA2 was used as the source of E1 in these studies. Recombinant GST-Ubc9, His-tagged-PIAS1, GST-SUMO-1Δ4C, and His-tagged-SUMO-3 were prepared by standard protocols. Labeled proteins were incubated with recombinant E1 (120 ng), Ubc9 (15 ng), GST-SUMO-1 (1 μg), PIAS1 (60 ng), and an ATP source at 37°C for 2 h. Reactions were analyzed by denaturing SDS-PAGE. After fixing and staining, gels were dried and exposed to KODAK BioMax MR single-emulsion X-ray film.

Antibodies.

SATB1 antibodies were prepared by injecting duplicate pairs of rabbits with recombinant SATB1(58-763) (12-9) or with N-terminal peptides of SATB1 (13-9, EATQGKEHSEMSN, residues 34 to 48, or 16-9, KTATIATERNGKPEN, residues 475 to 489) using standard protocols. Antibodies to PIAS1 were purchased from Santa Cruz Biotechnology, Inc., or AbCam, Inc. Horseradish peroxidase-conjugated anti-rabbit immunoglobulin (IgG) and anti-mouse IgG were from Zymed. Goat anti-mouse IgG-TRITC (1:400) was from Southern Biotechnology Associates. Anti-EGFP (monoclonal or polyclonal) was purchased from Clontech. Anti-FLAG was from Sigma Chemical Company.

Cell culture and immunoprecipitations.

Jurkat cells were obtained from the American Type Culture Collection (ATCC) and maintained as directed in RPMI-1640 medium plus 10% fetal bovine serum (FBS) (HyClone) and penicillin-streptomycin (Invitrogen). Cells were collected in phosphate-buffered saline (PBS), and whole-cell extracts (WCE) were prepared in hypotonic gentle lysis buffer (GLB = 10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 10 mM EDTA, 0.5% Triton X-100, 1× Halt protease inhibitor cocktail, EDTA-free [Pierce]) or in CellLyticM cell lysis reagent (Sigma Chemical Co.). Immunoprecipitations were performed in a minimal volume with 500 to 1,000 μg WCE by standard methods. Bound proteins were released in SDS sample buffer, fractionated by SDS-PAGE, and transferred to ImmobilonP transfer membrane (Millipore). After immunoblotting, specific proteins were detected using the enhanced chemiluminescent detection reagents (ECL; Amersham Pharmacia Biotech, Inc.) according to the manufacturers' instructions. Membranes were exposed to Hyperfilm (Amersham Pharmacia Biotech). For some experiments, cells were treated with calyculin A (100 nM, 45 min, 37°C; Sigma Chemical Co.) or alternatively with calphostin C (5 μM, 30 min, room temperature, in light; Calbiochem), or the apoptosis inducer etoposide (100 μM) for the times indicated in the figures.

³²P_i labeling.

Jurkat or transiently transfected Nalm-6 cells were grown to 1 × 10⁶ cells/ml, harvested, and washed once with phosphate-free medium (ICN Biochemicals, Irvine, CA) followed by growth in phosphate-free medium supplemented with 200 μCi of ³²P_i (Perkin-Elmer) and labeled for 4 h. Cells were then treated with calyculin A or calphostin C as described above, and cell lysates were prepared and immunoprecipitated with antibodies as described in the figure legends. Precipitates were examined by SDS-PAGE, and gels were then fixed, stained, destained, dried, and exposed to single-emulsion film for 1 to 4 days using standard protocols.

Microscopy.

For confocal microscopic studies, cells were prepared as previously described (64) but with the following modification: fixed cells were treated for 30 min with 0.5 μg/ml DAPI in PBS, washed three times in PBS, and then permeabilized as previously described. DAPI imaging was done at 790 nm from a Coherent Inc. Chameleon two-photon laser.

RESULTS

Interaction of distinct domains of the SUMO-1 E3 ligase PIAS1 with SUMO-modified or unmodified SATB1.

In a yeast two-hybrid (y2h) screen, SATB1(58-763) interacted with SUMO-1, Ubc9, and E3 ligase enzymes, including PIAS family members, as well as with numerous additional partners (Table 1) (L. K. Durrin, unpublished data) (64). PIAS family members identified by the y2h assay shared in common only their SUMO interaction motifs (SIMs) (Fig. 1D), and it was through this motif that they interacted with SUMO-conjugated SATB1 as verified by yeast liquid β-galactosidase, and plate nutritional, assays (37, 60) (Fig. 1A to C).

TABLE 1.

SATB1 interaction partners

Protein	Interaction strengtha	Minimal interaction domainb	GenBank accession no.
SUMO/Ub conjugation
SUMO-1	627	6-101	U83117
Ubc9	814	3-158	U66818
EDD	330	2,481-2,799	AF015051
TOPORS	108	288-628	AF098300
Polycomb (Pc2)	1,636	290-558	AF013956
PIAS family members
PIAS1	2,359	9-651	AF167160
PIAS3	1,449	185-619	AB021868
PIASy	1,186	370-510	AF077952

^aExpressed as units of β-galactosidase activity; strongly interacting positive control (pLexA-p53/pB42 AD-T) = 1,000; negative control (pLexA-laminC, pLexA) = 15; at least three replicates each.

^bAll y2h proteins interact with SATB1 at sites including their C termini, except for TOPORS (1,045 residues total), which interacts through a 340-residue internal site.

FIG. 1.

SATB1 modified by SUMO-1 interacts with PIAS family members through their C-terminally localized SIMs. (A) Schematic diagram of SATB1 functional domains. NLS, nuclear localization sequence (amino acids 20 to 40) (41); NMTS, nuclear matrix targeting sequence (residues 224 to 278) (56); MAR-BD, matrix attachment region binding domain (64). SATB1(58-763), SATB1(58-222), and SATB1(223-763) were each coexpressed with SUMO-1, Ubc9, PIAS family members, or ubiquitin Hect domain protein EDD(2,481-2,799) (control), and lysates were used in the y2h liquid β-galactosidase assay to measure interaction strength relative to controls described in Table 1, footnote a. (B) A modified y2h screen was done using SATB1(58-763) constructs (WT, K744R [mutated in SUMO conjugation site], K744R plus SUMO-1 or -3 [SUMO expressed in-frame at the C terminus]) and PIAS(1-651) (wild-type) or PIAS(1-651)-I459R (mutated in SIM) interaction partners. Negative interaction controls were pLexA vector only or pLexA-laminC. (C) SIMs of proteins that interacted with SATB1 in the y2h screen. Asterisks indicate protein motifs not within SATB1 interaction sites.

In mammalian cells, SUMO conjugation to SATB1 occurs at a very low, nearly undetectable level (64); notwithstanding, to discover a functional significance for the in vivo interaction of endogenous SATB1 and PIAS1, whether mediated through the SUMO moiety or via other motifs, coimmunoprecipitation assays were preformed with WCEs prepared from Jurkat cells (T cells; prototypical high-SATB1-expressing cells), treated or not with anti-Fas to induce apoptotic cleavage of SATB1 (11, 14, 64). Anti-PIAS1 exclusively precipitated endogenous, non-SUMO-conjugated SATB1 (103 kDa) (Fig. 2A to C). However, anti-PIAS1 did not precipitate the C-terminal region of SATB1 (i.e., the apoptotic cleavage product of SATB1 [peptides 254 to 763]), although this domain was precipitated with anti-SATB1(16-9) that recognized a C-terminal epitope of SATB1 (Fig. 2B, lanes 1 to 5). Conversely, SATB1 precipitation of endogenous PIAS1 was not detected, possibly due to low levels of PIAS1 expression in Jurkat cells (Fig. 2A). Next, after overexpression of SATB1 and EGFP-PIAS1 in the pre-B-cell line Nalm-6 (which expresses nearly undetectable levels of endogenous SATB1), the interaction of PIAS1 and non-SUMO-conjugated SATB1 was verified, as was the interaction of SATB1 with PIAS1 (Fig. 2C). Therefore, PIAS1 recognized SATB1, and vice versa, through domains in addition to the PIAS1 SIM and the SUMO moiety of conjugated SATB1.

To identify the PIAS1 domain that functionally recognized SATB1, GST pull-down assays were used. Although the SP-RING domain of PIAS binds the E2 conjugating enzyme Ubc9 and plays a role in substrate recognition and catalytic addition of SUMO moieties (6, 13, 23, 25, 52), substrate specificity is likely encoded by another PIAS functional domain(s). PIAS serves as an E3 ligase for many substrates, and these partner proteins associate with diverse motifs of the PIAS protein (57, 59). Therefore, full-length (FL) [³⁵S]methionine-SATB1 was incubated with equivalent amounts (0.2 μg) of near-full-length GST-PIAS1 (residues 9 to 651), the N- and C-terminal peptides of PIAS1, the SAP motif (1 to 75), and the SP-RING domain (309 to 436), as well as with PIAS3(185-619), with PIASy(370-510), or with GST alone, bound to microtiter plates. The PIAS1 (SAP) motif was the minimal motif that effectively recognized SATB1 (Fig. 2E, lane 5), whereas the C-terminal fragments and GST interacted weakly or not at all (Fig. 2E, lanes 6 to 8). The PIAS family members [PIAS3(185-619) and PIASy(370-510)], lacking their N-terminal SAP motifs, associated poorly with SATB1 [less than 5% that of PIAS1(FL); L. K. Durrin, unpublished data]. Thus, the PIAS1 SAP motif was a direct recognition site for SATB1. Specificity of PIAS interaction with SATB1 was shown by use of the control, [³⁵S]Met-luciferase, which did not interact with either the N- or C-terminal fragments of GST-PIAS1 (Fig. 2F, compare lanes 2 and 3 with lanes 6 and 7) or with GST alone (Fig. 2F, lane 8).

Functional PIAS1 association at the N terminus of SATB1.

Since PIAS1 did not recognize the C terminus of endogenous SATB1 (residues 254 to 763) (Fig. 2B, lanes 5 and 7), GST pull-down assays were employed, using SATB1 deleted of C-terminal sequences, to identify the N-terminal 222 amino acids of SATB1 as a motif that was recognized by full-length PIAS1 and by the SAP domain (Fig. 3A, lanes 5 and 6). To further narrow the interaction site, SATB1 deletions from the N terminus exhibited progressively weaker association with PIAS1 until deletion of amino acids 1 to 222 from SATB1 entirely abrogated interaction (Fig. 3B, lanes 11 and 12). Thus, the SAP motif of PIAS1 recognized non-SUMO-conjugated SATB1 exclusively within residues 149 to 222.

FIG. 3.

PIAS1 associates functionally at the N terminus of SATB1. (A) GST-PIAS1(FL) or the GST-PIAS1(SAP) (0.2 μg) was incubated with [³⁵S]Met-SATB1-(1-454) or -(1-222), respectively. Pulled-down proteins were fractionated by SDS-PAGE adjacent to an aliquot of the [³⁵S]Met test protein (as in Fig. 2E). Gels were examined as described in Fig. 2E. (B) GST-PIAS1(FL) and the GST-PIAS1(SAP) (0.2 μg) were incubated with [³⁵S]Met-labeled SATB1 deletions (Δ1-57, Δ1-97, Δ1-148, Δ1-222) and analyzed as described above. (C) SATB1(FL), SATB1 deletions (Δ1-57, Δ1-97, Δ1-148, Δ1-222), SATB1-K744R, and p53(FL) (control) were labeled with [³⁵S]Met and used in an in vitro SUMO conjugation assay. Products were fractionated by SDS-PAGE; gels were fixed, dried, and exposed to Kodak Biomax MR single-emulsion film for 3 days. (D) SATB1 functional domains are described in Fig. 1A. Data from Fig. 3A to C are summarized, showing constructs that interacted with PIAS1 (PIAS1 int.) and/or that were conjugated to SUMO (SUMO conj.) in vitro. The bottom construct depicts SATB1 mutated at K744 to prevent SUMO modification. (E) SUMO conjugation assays using recombinant PIAS1(9-651) or PIAS1(1-651) as E3 ligases to compare enhancement of SUMO conjugation in an in vitro assay with the substrates [³⁵S]Met-SATB1(FL) or [³⁵S]Met-p53(FL). Products were examined as described for Fig. 3C.

To address the functional significance of the PIAS1:SATB1 interaction, a series of N-terminal SATB1 deletions were employed in an in vitro SUMO conjugation assay (64). PIAS1 enhanced SUMO conjugation to K744 of SATB1 constructs deleted of amino acids 1 to 57, 1 to 97, or 1 to 148, but further deletion of residues 1 to 222 prevented this posttranslational modification (Fig. 3C, top panel). As controls, SUMO conjugation to p53 is known to be enhanced by PIAS family members (54) whereas mutation of the SUMO consensus site of SATB1, K744R, precluded SUMO modification (Fig. 3C, bottom panels) (64). Thus, the same region of SATB1 (residues 149 to 222) that interacted with the SAP motif of PIAS1 was functionally required for in vitro SUMO modification at SATB1-K744 (summarized in Fig. 3D).

The PIAS1 clone isolated in the y2h screen, with SATB1 as bait, was deleted of the eight N-terminal residues. Deleted constructs [PIAS1(9-651) and PIAS1(9-315)] both interacted strongly with SATB1 in GST pull-down assays (Fig. 2E, lane 3) (L. K. Durrin, unpublished data). However, PIAS1(9-651) failed to significantly stimulate SUMO conjugation to SATB1, to p53, or to other targets examined when added to an in vitro SUMO conjugation assay (Fig. 3E, compare lanes 2 and 3 to lanes 5 and 6) (L. K. Durrin, unpublished data).

Inhibition of in vivo PIAS1:SATB1 interaction by phosphorylation.

SATB1 is a phosphoprotein that has been shown to be modified by protein kinase C (PKC) but not by casein kinase II (CKII) (11, 46). To investigate whether SATB1 phosphorylation controls its interaction with PIAS1, Jurkat cells were treated with the protein phosphatase 2A (PP2A) and PP-1 inhibitor calyculin A or alternatively with calphostin C, a specific inhibitor of protein kinase C. SATB1 in lysates prepared from calyculin A-treated cells showed an upward shift in mobility on SDS-PAGE, but PIAS1 did not (slower-migrating, higher-molecular-weight forms of proteins are phosphorylated [3, 12, 24, 32, 38, 68]). SATB1 and PIAS1 from untreated cells, or those hypophosphorylated following exposure to calphostin C, fractionated at approximately 103 kDa (SATB1) or 75 kDa (PIAS1) (Fig. 4A, lanes 1 to 3). In coimmunoprecipitation assays using WCEs prepared from control or treated Jurkat cells, PIAS1 interacted with unphosphorylated SATB1 (plus calphostin C) to a greater extent than it did with the SATB1 control or SATB1 that was significantly modified (plus calyculin A) (Fig. 4A, lanes 5 to 7). Thus, phosphorylation appears to regulate interaction of PIAS1 with SATB1.

FIG. 4.

Phosphorylation of SATB1 inhibits PIAS1 interaction, in vivo, and SATB1 is phosphorylated at multiple residues. (A) WCEs (40 μg) from control Jurkat cells or those treated with calyculin A (100 nM, 45 min, 37°C), or with calphostin C (500 nM, 30 min, 24°C, in light) (lanes 1 to 3). Jurkat WCEs (1,000 μg) were immunoprecipitated with anti-PIAS1 or preimmune serum. Cell lysates and immunoprecipitates were fractionated by SDS-PAGE, transferred to Immobilon P transfer membranes, and immunoblotted with anti-SATB1 (12-9) or with anti-PIAS1. (B) Jurkat WCEs (1,000 μg) prepared from control, calyculin A-treated, or calphostin C-treated cells were immunoprecipitated with antiphosphothreonine, followed by analysis of precipitates by immunoblotting to anti-SATB1. (C) Jurkat cells (top panel) or alternatively Nalm-6 cells (bottom panels) transiently expressing SATB1(1-202), SATB1(1-185), or SATB1(1-202)-S185A were labeled with ³²P_i in phosphate-free media for 4 h, 37°C, and then treated with calyculin A or calphostin C. WCEs were immunoprecipitated with anti-SATB1 (13-9) or anti-EGFP. Precipitates were examined by SDS-PAGE; gels were fixed, stained, destained, dried, and exposed to single-emulsion X-ray film for 3 days at −80°C. (D) Nalm-6 cells expressing C-terminal deletion constructs of SATB1 were treated with calyculin A; lysates were analyzed by gel electrophoresis and immunoblotted to anti-SATB1 (13-9). The line drawing illustrates the deletions and site of threonine phosphorylation. (E) Sequence of the SATB1 tryptic peptide previously identified to harbor a site(s) of phosphorylation (top) (46). All serines and threonines in this tryptic peptide, either individually, in pairs, or in multiples, were mutagenized to alanines in the context of SATB1(1-202) (middle panel) or SATB1(148-202) (bottom panel) and used in the calyculin A assay described in the legend for panel D.

Phosphorylation of SATB1 at multiple residues in vivo.

Endogenous SATB1 from Jurkat cell extracts is phosphorylated at serine and threonine, but not tyrosine, residues (46). The specific protein kinase C (PKC) inhibitor, calphostin C, completely inhibits basal levels of SATB1 phosphorylation whereas the MAPK inhibitor PD98059 does not (46). Mass spectroscopy of in vitro phosphorylated SATB1 identified S185 as one site of PKC phosphorylation (11, 46). Since PIAS1 recognized SATB1 within residues 149 to 222, regulators of this protein-protein interaction likely also reside within this region; however, S185 phosphorylation alone did not control either PIAS1 interaction with SATB1 or downstream events (see below). Thus, to investigate the phosphorylation status of additional residues, cell extracts were prepared from Jurkat cells (control or treated with calyculin A or calphostin C) and immunoprecipitated with antiphosphothreonine. Results confirmed that endogenous SATB1 is a phosphoprotein that undergoes PKC-mediated phosphorylation on a threonine residue (Fig. 4B). These results were confirmed in a separate investigation in which Jurkat cells were labeled with ³²P_i and then treated with calyculin A or calphostin C. Cell extracts were immunoprecipitated with anti-SATB1(13-9) (Fig. 4C, top panel).

Next, to identify an additional, functionally relevant, SATB1 N-terminal phosphorylation site(s), C-terminal deletions of SATB1 (residues 1 to 222, 1 to 202, 1 to 185, 1 to 150, 148 to 202), prepared as fusions with EGFP, were transiently expressed in Nalm-6 cells and then treated with calyculin A. Immunoblot analysis of cell lysates revealed an upward shift in mobility of calyculin A-treated SATB1(1-222) and SATB1(1-202) but not of shorter constructs SATB1(1-185) or SATB1(1-150) (Fig. 4D). The results did not preclude phosphorylation of S185, since adjacent amino acids may contribute to the protein kinase recognition site. The region from 185 to 202 partially overlapped with the tryptic fragment (residues 176 to 189) (Fig. 4E, top) previously demonstrated by mass spectroscopy to contain a phosphorylation site(s) (46). Mutation of any site alone (S185, T187, or T188) in the EGFP-SATB1(1-202) or EGFP-SATB1(148-202) constructs did not abolish phosphorylation, nor did mutation of two or three of these sites within the 1-202 construct (Fig. 4E, middle panel). However, when both S185 and T188 were mutated in the 148-202 construct, phosphorylation did not occur, nor was it detected by ³²P_i incorporation (Fig. 4E, bottom panel). Thus, both S185 and T188 of SATB1 were phosphorylated in vivo. Additionally, since the triple mutation carried by SATB1(1-202)-S185A/T187A/T188A was phosphorylated (Fig. 4C, bottom right panel, and E, middle panel) another site (not identified) within amino acids 1 to 184 of SATB1 must also be modified, dependent upon phosphorylation of S185 and/or T188.

Phosphorylation regulates localization of SATB1 to PML NBs.

Endogenous SATB1 and EGFP-SATB1 localize diffusely within the nucleus, as detected by fluorescent and confocal microscopy (14, 56, 64). However, in Jurkat cells stably expressing low levels of exogenous SUMO-1, SATB1 partially colocalized into PML NBs (see Fig. 5B, top panels) (64). Association of target proteins at NBs is frequently influenced by noncovalent interactions between their SUMO moieties and the SIMs of PML NB components, or vice versa (58). SATB1 lacks an identified SIM (During, unpublished data); hence, it is unknown if SATB1 sumoylation directs it to PML NBs or whether some other protein targets it to these structures, where it is modified and then cleaved by caspase-6. PIAS1, in its role as an E3 ligase, is a likely candidate to effect the latter scenario. Thus, to establish the contribution of PIAS1 for subnuclear targeting of SATB1, and to establish the role of phosphorylation in this process, EGFP-SATB1(1-202) was transiently expressed in MCF-7 cells, treated with calyculin A or calphostin C, and examined by confocal microscopy. MCF-7 cells do not express detectable levels of endogenous SATB1, and they are large, attached cells that provide clear distinction between protein localization in cellular compartments. SATB1 remained diffusely localized in the nucleus when highly phosphorylated (plus calyculin A), whereas unphosphorylated SATB1 (plus calphostin C) colocalized with nuclear PML NBs in greater than 50% of the transfected MCF-7 cells (Fig. 5A). The failure of SATB1 to target to PML NBs in all transfected cells likely reflects the normal cell-cycle-associated dispersal of PML NBs (10).

FIG. 5.

Phosphorylation regulates localization of SATB1 to PML NBs. (A) MCF-7 cells expressing exogenous EGFP-SATB1(1-202) were treated with calyculin A or calphostin C and then fixed and stained with anti-PML. The fluorescent signals from 100 cells were analyzed (repeated in three separate studies) with an upright confocal microscope to identify cells with SATB1 localized to PML NBs. Representative paired images of PML (red) and SATB1 (green) and the merged images are shown. Control was untreated cells. Bar, 10 μm. (B) Jurkat cells expressing low levels of exogenous SUMO-1 (top panel) were examined by confocal microscopy, as described previously (64). MCF-7 cells expressing EGFP-SATB1(1-202) with specific residues mutated were fixed and stained with anti-PML and examined by confocal microscopy. (C) MCF-7 cells transiently expressing EGFP-SATB1(1-202), wild-type or mutant form, were examined by confocal microscopy. (D) One hundred cells (repeated in three separate experiments) were examined for localization of EGFP-SATB1(1-202) into nuclear dots. The percentage of cells displaying the dotted pattern was graphed for each mutant. The Roman numerals correspond to sections of panel C.

Treatment of cells with drugs can have numerous effects in addition to the specific effects being investigated (10). Thus, to distinguish specific effects of phosphorylation on SATB1 localization, a series of EGFP-SATB1 N-terminal phosphorylation site mutation constructs were transiently expressed in MCF-7 cells and examined by fluorescent microscopy. EGFP-SATB1(1-202)-T188A localized into PML NBs in approximately 30% of cells expressing the fluorescent construct, whereas T188E (mutated to mimic phosphorylated SATB1) remained diffusely localized in greater than 95% of transfected MCF-7 cells (Fig. 5B). Mutations carried by adjacent serine or threonine residues, S185A or T187A, or even more distant serines, S117A/S124A or S172A, did not by themselves direct SATB1 redistribution into PML NBs (Fig. 5D) (J. T. Tan and L. K. Durrin, unpublished data). However, T188A in conjunction with site mutagenesis at S185, or at upstream serines, significantly increased the number of cells with SATB1 localized into PML NBs (51% or greater) (Fig. 5C and D). These results suggested that T188 was a primary phosphorylation site that was essential to regulate targeting of SATB1 into PML NBs. To delineate the sequences around T188 that contributed to subnuclear targeting of SATB1, the deletion construct, EGFP-SATB1(1-193)-S185A/T188A that removed an LXXLL motif (L is leucine and X is any amino acid; residues 193 to 197) from SATB1 remained diffusely arranged throughout the MCF-7 nucleus, in contrast to SATB1(1-202)-S185A/T188A, which localized predominantly into PML NBs (Fig. 5C and D). In addition, EGFP-SATB1(1-202)-LXXAA (last two leucines mutated to alanines) (Fig. 6B, panel i) likewise exhibited fewer than 2% of cells with SATB1 in the PML NBs (Fig. 4C, panel vi, and D). This suggested a role for the SATB1-LXXLL motif in sequestration of a factor(s) with intranuclear targeting capability, and phosphorylation controls this process.

FIG. 6.

Identification of a PIAS1 recognition motif on SATB1 and additional substrates. (A) WCEs (1,000 μg) prepared from Nalm-6 cells transiently expressing pcDNA3.1/HIS-SATB1 (WT or mutant) were harvested and coimmunoprecipitated with anti-PIAS1. Precipitates were fractionated by SDS-PAGE, electrotransferred to Immobilon-P membranes, and probed with anti-SATB1 (top panel). IgG serves as a loading control. The blot was stripped and reprobed with anti-PIAS1 (middle panel). WCEs (20 μg) were immunoblotted to anti-SATB1 (bottom panel). (B) (panel i) Shown is the region of SATB1 recognized in vivo by PIAS1. Serines, threonines, and leucines that were mutated for specific studies are shown, as is the SATB1(1-193) deletion of the LXXLL motif. (panel ii, top) Alignment of SATB1, SATB2, and CHD3-B. Symbols depict residues that are identical (asterisks), strongly conserved (diamonds), or weakly conserved (filled circles). (panel ii, bottom) Substrate motifs recognized by PIAS-SAP are aligned, and specific interaction sites are indicated at the right of the alignment; ND, not determined. The region around the LXXLL motif is predicted to be helical. (panel iii) Alignment of LXXLL motifs from protein-protein interactors (47).

PIAS1 interacts with SATB1, in vivo, in a phosphorylation-dependent manner.

To identify whether the N-terminal SATB1 LXXLL motif was recognized by PIAS1 in vivo, and to establish whether phosphorylation at SATB1-T188 regulated this interaction, Nalm-6 cells transiently expressing full-length SATB1, either wild type (WT) or carrying mutations to S185A/T188A, T188E, or LXXAA, were harvested and used in coimmunoprecipitation assays with anti-PIAS1. PIAS1 reproducibly interacted with wild-type SATB1 and with SATB1 that could not be phosphorylated (S185A/T188A), but it did not associate with SATB1 mutated to mimic the phosphorylated state (T188E) (Fig. 6A, compare lanes 2 to 4). Nor did PIAS1 recognize SATB1 when the LXXLL motif was mutated (Fig. 6A, lane 5). These results suggested that PIAS1 normally recognized and bound to the region of SATB1 including the LXXLL motif, and this association was regulated by phosphorylation at SATB1 T188.

Identification of a general PIAS1 interaction site.

PIAS family members interact directly with at least 70 target proteins, and they enhance conjugation of SUMO to many of these substrates (45). In addition to SATB1, we identified CHD3-B, a specific variant of the chromatin modification/remodeling factor chromodomain, helicase/ATPase, and DNA binding 3 (CHD3/Mi2α), as a PIAS1 interaction partner that undergoes PIAS-enhanced SUMO conjugation at lysine-1971 (L. K. Durrin, unpublished data). CHD3/Mi-2α is a component of the chromatin-remodeling NuRD complex that associates with SATB1 at MAR elements (73). The E3 ligase PIAS1-SAP domain recognized CHD3-B through a C-terminal region (residues ∼1866 to 2000) (Durrin, unpublished data) that encoded an LXXLL motif (residues 1869 to 1873) (Fig. 6B). Finally, the SATB family member SATB2 undergoes PIAS1-augmented SUMO conjugation at K-233 and K-350 (8). Examination of restricted regions of these three proteins identified a common motif that centered on the LXXLL pentamer (Fig. 6B, panel ii), and these sequences were evolutionarily conserved in the SATB and CHD families (66). Next, to establish the prevalence of the LXXLL motif for substrate recognition by PIAS, restricted recognition sites on additional PIAS partner proteins were examined. Usually, SUMO conjugation to the substrates examined was enhanced by PIAS (Oct4 is the lone exception). In some proteins examined, a functional unpredictability exists between whether PIAS enhances SUMO conjugation to the substrate and whether there is a direct involvement of PIAS in SUMO-independent cellular relocalization and/or activation/repression of the target. Often these two consequences of PIAS action on substrates are interdependent (45, 51, 57). The substrates examined, and their restricted sites of interaction with PIAS, are shown in Fig. 6B (panel ii). Descriptions of the proteins examined can be found in references 45, 47, and 51. In general, an LXXLL pentamer was identified at the core of known PIAS recognition sites; at least one amino acid at position +2 or +3 of this motif was polar (the first L of LXXLL is +1); +7 was typically acidic (D or E) although in a few cases the residue was polar. In general, the sequences around the LXXLL core were enriched in polar residues, especially at the N terminus. These proximal residues differed from those observed in coactivators of nuclear receptors and in nonnuclear-receptor-based complexes (Fig. 6B, panel iii) (47, 53).

Phosphorylation regulates caspase cleavage of SATB1 in vivo.

To examine the role of phosphorylation in the regulation of SUMO-initiated caspase cleavage of endogenous SATB1, phosphorylation was inhibited with calphostin C in Jurkat cells, followed by induction of caspase cleavage with anti-Fas. Hypo-phosphorylated, endogenous full-length SATB1 was completely digested by about 6 to 8 h treatment with anti-Fas. SATB1 not treated with calphostin C was digested about 50% after 8 h (Fig. 7A). Thus, phosphorylation, or the potential to be phosphorylated, greatly reduced the catalytic association of PIAS1 at the N terminus of SATB1.

FIG. 7.

Phosphorylation regulates caspase cleavage of SATB1 in vivo. (A) Jurkat cells treated with calphostin C (5 μM, 30 min, 24°C, in light) and then with anti-Fas for the times indicated were collected and immediately lysed with SDS-PAGE loading buffer and then fractionated by SDS-PAGE and immunoblotted to anti-SATB1. Lane −, untreated control cells. (B) Nalm-6 cells transiently overexpressing EGFP-SATB1(FL) constructs (wild-type [wt], T188A, T188E/LXXAA, or sumoylation site mutant [K744R]) were treated with etoposide for the indicated times, harvested as described above, and examined by immunoblotting to anti-SATB1 or anti-PARP (as shown) antibody. Only the 89-kDa cleavage product of PARP (116 kDA) was detected by anti-PARP antibody.

Next, to examine effects of specific phosphorylation on caspase cleavage of SATB1, Nalm-6 cells expressing full-length SATB1, either wild type (WT) or carrying a mutation at T188A, were treated with the caspase inducer etoposide for up to 8 h. Caspase cleavage of SATB1-T188A occurred very efficiently (proteolysis was detected even at the zero time point), compared to that of the wild-type control (Fig. 7B). Cleavage of the wild-type SATB1 and the apoptosis indicator, endogenous PARP, occurred only in response to etoposide induction at 2 h or later (Fig. 7B). In contrast, SATB1 mutated both at the PIAS1 interaction site and at T188 to mimic phosphorylated protein (SATB1-T188E/LXXAA) was cleaved very poorly. The residual proteolysis observed suggested that not all caspase-6 cleavage was regulated. SATB1 mutated at the SUMO conjugation site (K744R) or at the caspase cleavage site (251-VEMD-254; SATB1-[Cs-mut]) did not undergo proteolysis, as has been shown previously (Fig. 7B) (64).

DISCUSSION

To understand the mechanism(s) that controls SUMO conjugation to the tissue-restricted MAR-binding protein, SATB1, we identified specific in vitro and in vivo interactions between the PIAS1 SAP domain and SATB1 sequences that included an LXXLL motif. Protein interaction at this domain was further relevant for subnuclear localization of SATB1 to PML NBs (Fig. 5), where the E3 ligase PIAS1 enhances SUMO conjugation to SATB1, an event that culminates in SATB1 cleavage by caspase-6 (64). Additional LXXLL motifs were encoded by SATB1 (i.e., three were within and around the CUT repeats) (Fig. 3D) and a reverse-oriented motif was immediately upstream of the site of SUMO conjugation (residues 724 to 728); the function of these latter LXXLL motifs, if any, is unknown. An LXXLL motif (amino acids 19 to 23) is conserved within PIAS family member SAP domains (within helix α2; see below) but generally not within additional proteins with SAP motifs (2), suggesting the LXXLL sequences are not essential for SAP function. Indeed, the LXXLL motif within the PIAS SAP has not been associated with specific protein associations, although it contributes to transcriptional repressive activity of the androgen receptor (15, 16, 30). The LXXLL motif was originally recognized to mediate nuclear receptor (NR)-coactivator or coactivator-coactivator interactions (47). However, many examples of protein-protein recognition that involve the LXXLL motif have been reported (for a review, see reference 47). One limited means of LXXLL protein association is mediated via homo- or heterodimerization; mutation of the latter two leucines of this motif to alanines has been reported to prevent homo- or heterodimerization of SHP or DAX1-DAX1A, respectively (21).

In general, the integrity of the LXXLL is required for binding to substrates; nonetheless, both the properties of the immediately adjacent amino acids and the number of LXXLLs within a coactivator influence target selection and binding affinity (5, 47, 50, 74). Properties of the binding partner also influence protein associations. The consensus motif identified in SATB1 and other PIAS-SAP target proteins differed significantly in immediately adjacent residues from those reported for NR-coactivator and protein-protein LXXLLs (Fig. 6B, compare panels ii and iii) (47, 53). In our system, the LXXLL-proximal amino acids, as well as their spacing, could influence specificity and affinity of substrates for interaction with and usage of distinct PIAS family members. Indeed, the E3 ligase function of PIAS proteins may have developed as an addendum to its function as an activator/repressor of transcription and now provides substrate specificity for SUMO conjugation.

The predicted conformation of the LXXLL motif in all proteins examined in this study is alpha-helical (Fig. 6B), which is consistent with the reported structural arrangement of LXXLL motifs (47). The PIAS SAP motif is an amphipathic, four-helix bundle that binds A/T-rich DNA and also has a role in protein recognition (43, 62). These conformations are consistent with known protein-protein interaction interfaces (34). Additionally, the first helix of PIAS1 (encompassing residues 5 to 11 [α1]) (43) appeared to have a role in augmenting SUMO conjugation, since deletion of the eight N-terminal residues of PIAS1 [PIAS1(9-651)] greatly repressed its E3 ligase activity toward all substrates examined, including SATB1 and p53 (Fig. 3E); however, this deletion did not dramatically alter substrate recognition (Fig. 2E and 2F). This result is unexpected, since it is generally acknowledged that, in many cases, association of the PIAS SAP motif with transcription factors regulates downstream events independently of SUMO E3 ligase activity toward the substrate (45, 51). However, in analogy to the ubiquitination system, it is believed that the ubiquitin E3 ligase MDM2 suppresses p53 transcription by direct binding to the N-terminal transactivation domain of p53 and/or by promoting degradation of p53 through its ubiquitin E3 ligase activity (20). Using a mouse model system that eliminated the E3 ligase function without affecting p53 binding, it has been established that both physical interaction of Mdm2-p53 and Mdm2-mediated ubiquitination are required to repress p53 and to mediate early events in mouse development (20). It is still unknown whether binding and E3 ligase activities of PIAS proteins are separable events or, as with Mdm2 and p53, correlated.

Our observations point to a new role for the evolutionarily conserved N terminus of SATB1: it associates with the multidomain PIAS that in turn affects SATB1 localization and posttranslational sumoylation. Previously it was suggested that the N terminus of SATB1 facilitates SATB1 homodimerization and interaction with sumoylated PML and PML NBs, in addition to association with a number of transcription factors (26, 46, 48, 49). Homodimerization of SATB1 is supposedly critical for SATB1 to localize to MAR elements. Disruption of the dimer by apoptotic caspase cleavage leads to SATB1 dissociation from chromatin and concomitant chromatin condensation and fragmentation (11). These various functions of the N terminus are allegedly mediated through a PDZ domain (11, 48). However, the Conserved Domain Database search used to identify this as a PDZ domain employed an E value that is of statistically low significance (i.e., E = 1.8, which is similar to that of random sequences) (11). We further employed Position-Specific Iterated (PSI) BLAST to confirm that this N terminus is not homologous to the PDZ motif. In addition, many studies have failed to confirm homodimerization in vivo (40, 48, 69, 70). Using in vivo coimmunoprecipitation of FLAG-SATB1(FL) with EGFP-SATB1(FL), and vice versa, or plate y2h assays, we failed to detect in vivo SATB1 homodimerization (Durrin, unpublished data). We suggest instead that many of the functions attributed to the SATB1 N terminus might be mediated through alternative N-terminal sequences (i.e., LXXLL) or through interaction of SATB1 with PIAS or another protein with multifunctional domains. For example, the four-helix bundle conformation of the PIAS1 SAP motif is capable of binding to A/T-rich DNA fragments (i.e., MAR elements) and also conveys substrate specificity (43). The SP-RING domain binds Ubc9 and is involved in substrate recognition but, along with the SAP motif, also targets PIAS to PML NBs and the nuclear periphery (22, 45, 51, 52, 54, 55, 63). The SIM motif provides a mechanism to assemble large multicomponent protein complexes.

PIAS recognition of additional substrates may be regulated by phosphorylation, since at least p53 and c-Jun harbor serines that are major sites of phosphorylation immediately upstream of their LXXLL motifs (Fig. 6B, panel ii; gray boxes denote residues that undergo phosphorylation). SUMO conjugation of p53, c-Jun, and Elk-1 is inhibited by phosphorylation (at Ser-20 of p53, Ser-63 and Ser-73 of c-Jun and Ser-383 of Elk-1) (29, 39, 72). Figure 8 depicts a model for phosphorylation control of PIAS1 interaction with SATB1 that results in nuclear relocalization, SUMO conjugation to SATB1-K744, and caspase cleavage of the substrate.

FIG. 8.

Schematic model of phosphorylation regulation of PIAS1 interaction with SATB1. PIAS1 recognized the N terminus of SATB1 near the LXXLL motif, leading to relocalization of SATB1 to PML NBs and SUMO conjugation (64). The SUMO-modified SATB1 is cleaved by caspase-6. However, SATB1 is PKC phosphorylated at T-188 after cellular signaling and this impedes PIAS interaction and subsequent posttranslational SUMO conjugation and nuclear redistribution of SATB1.

An important goal of studies of transcriptional regulators of cell-specific gene expression, such as SATB1, is an understanding of the mechanisms involved in initiating a program of development. Insight into SATB1 function has already been reported in many papers which examine the processes (i.e., chromatin remodeling, gene expression, etc.) that are controlled by SATB1. A less investigated area of interest is to examine how SATB1 itself is regulated during hematopoietic development. Included in the latter category is investigation of how posttranslational modifications, including acetylation, phosphorylation, sumoylation, etc., affect the function of SATB1. Since only a small subset of SATB1 molecules undergoes SUMO-regulated caspase cleavage, it is likely that the SATB1 proteolytic products generated are relevant for a specific function(s) in a restricted population of cells or at a specific stage of hematopoietic development. Therefore, this paper provides novel insight into a potentially important biological process and provides the foundation for future studies. Also, this work is the first to demonstrate a consensus site for interaction of PIAS1 with substrates (Fig. 6), and this will lead to investigation of the mechanism of the multiple functions of PIAS.