Function of the C. elegans T-box factor TBX-2 depends on SUMOylation

T-box transcription factors are critical developmental regulators in all multi-cellular organisms, and altered T-box factor activity is associated with a variety of human congenital diseases and cancers. Despite the biological significance of T-box factors, their mechanism of action is not well understood. Here we examine whether SUMOylation affects the function of the C. elegans Tbx2 sub-family T-box factor TBX-2. We have previously shown that TBX-2 interacts with the E2 SUMO-conjugating enzyme UBC-9, and that loss of TBX-2 or UBC-9 produces identical defects in ABa-derived pharyngeal muscle development. We now show that TBX-2 is SUMOylated in mammalian cell assays, and that both UBC-9 interaction and SUMOylation depends on two SUMO consensus sites located in the T-box DNA binding domain and near the TBX-2 C-terminus, respectively. In co-transfection assays, a TBX-2:GAL4 fusion protein represses expression of a 5xGal4:tk:luciferase construct. However, this activity does not require SUMOylation, indicating SUMO is not generally required for TBX-2 repressor activity. In C. elegans, reducing SUMOylation enhances the phenotype of a temperature-sensitive tbx-2 mutant and results in ectopic expression of a gene normally repressed by TBX-2, demonstrating that SUMOylation is important for TBX-2 function in vivo. Finally, we show mammalian orthologs of TBX-2, Tbx2, and Tbx3, can also be SUMOylated, suggesting SUMOylation may be a conserved mechanism controlling T-box factor activity. Electronic supplementary material The online version of this article (doi:10.1007/s00018-013-1336-y) contains supplementary material, which is available to authorized users.


Introduction
t-box proteins are a family of transcription factors found in all multicellular animals where they play important roles in the development of a variety of tissues [1,2]. the defining feature of this family is the conserved t-box DNabinding domain, and t-box factors are grouped into distinct sub-families based on sequence conservation within this domain. In many cases, the level of t-box factor activity is crucial to normal function. For example, reduced expression of the human tbx2 sub-family genes TBX3, TBX4, and TBX5 resulting from loss of one functional allele results in ulnar-mammary syndrome, small patella syndrome, and holt-Oram syndrome, respectively [3][4][5][6][7]. In contrast, over-expression of the tbx2-subfamily genes TBX2 and TBX3 is found in a number of human cancers [8]. Despite their developmental and clinical importance, relatively little is known about the mechanism by which t-box factors function.
Abstract t-box transcription factors are critical developmental regulators in all multi-cellular organisms, and altered t-box factor activity is associated with a variety of human congenital diseases and cancers. Despite the biological significance of t-box factors, their mechanism of action is not well understood. here we examine whether SUMOylation affects the function of the C. elegans tbx2 sub-family t-box factor tBX-2. We have previously shown that tBX-2 interacts with the e2 SUMO-conjugating enzyme UBC-9, and that loss of tBX-2 or UBC-9 produces identical defects in aBa-derived pharyngeal muscle development. We now show that tBX-2 is SUMOylated in mammalian cell assays, and that both UBC-9 interaction and SUMOylation depends on two SUMO consensus sites located in the t-box DNa binding domain and near the tBX-2 C-terminus, respectively. In co-transfection assays, a tBX-2:GaL4 fusion protein represses expression We are interested in the role that SUMOylation plays in t-box factor activity. SUMOylation is the covalent and reversible post-translational attachment of the small ubiquitin-like modifier peptide (SUMO) to specific lysine residues in target proteins [9,10], and it has been implicated in diverse processes, including modifying function, nuclear localization, and sub-nuclear localization of transcriptional regulators [11]. SUMOylation of transcription factors is typically associated with repression [12], but it has also been implicated in transcriptional activation by some factors [13,14]. the SUMO conjugation pathway is analogous to the ubiquitination pathway and involves an e1-activating enzyme (aos1/Uba2) and an e2 conjugating enzyme (Ubc9) sufficient for specific SUMO attachment in vitro [15,16]. In addition, a variety of e3 ligases have been identified that promote SUMO transfer from e2 to specific substrates in vivo. Ubc9 recognizes the ΨKX(D/e) SUMO consensus site (where Ψ is a large hydrophobic amino acid and K is the residue attached to SUMO) [17,18], and many SUMOylation substrates have been identified by their interaction with Ubc9 in yeast two-hybrid screens [19]. SUMOylation also occurs at non-consensus sites, and non-covalent SUMO/substrate or e3 ligase/substrate interactions are involved in directing SUMOylation at these sites [9].
We hypothesize that function of the C. elegans t-box factor tBX-2 depends on SUMOylation [20]. tBX-2 is the sole C. elegans member of the tbx2 subfamily and is necessary for formation of anterior pharyngeal muscles. In yeast two-hybrid assays, tBX-2 interacts with the e2 SUMO conjugating enzyme UBC-9, and loss of UBC-9 produces pharyngeal phenotypes identical to those resulting from tbx-2 loss-of-function. In addition, sub-nuclear localization of a tBX-2::GFP fusion protein is altered when SUMOylation is reduced.
here, we ask if tBX-2 is SUMOylated and whether SUMOylation affects tBX-2 activity in vivo. We first used the two-hybrid assay to map interaction sites between tBX-2 and UBC-9 and found two SUMO consensus sites in tBX-2 that mediate interaction with UBC-9. One of these sites is located near the tBX-2 C-terminus, while the other is located in a highly conserved region of the t-box DNa binding domain. We next showed that tBX-2 is SUMOylated in mammalian cell assays, and that tBX-2 SUMOylation depends on both of these UBC-9 interaction sites. We then examined tBX-2 transcriptional activity and found that in mammalian cells a tBX-2-GaL4 DNabinding domain (GaL4-DBD) fusion protein represses expression of a GaL4-responsive reporter, but surprisingly this repression did not require SUMOylation. to determine whether SUMOylation is important for tBX-2 activity in vivo, we asked if tbx-2 and ubc-9 interact genetically. We found that reduction of SUMOylation enhances the effect of a tbx-2 hypomorphic mutant on embryonic viability and pharyngeal muscle development, and that repression of a downstream target of tBX-2 depends on SUMOylation. Finally, we examined SUMOylation of two mammalian orthologs of tBX-2 and found that human tBX2 and mouse tbx3 can also be SUMOylated. We suggest SUMOylation is a common mechanism regulating activity of t-box transcription factors.

RNai analyses
Feeding RNai was performed as previously described [26] using plasmids obtained from Geneservice containing genomic fragments of ubc-9 or smo-1 cloned into L4440 [27]. to assess enhancement of the tbx-2(bx59) mutant phenotype, N2 or OK0660 [tbx-2(bx59)] L4 hermaphrodites raised at 16 °C were transferred to plates seeded with RNai feeding E. coli or OP50 and incubated at 25 °C for 24 h. these animals were transferred to fresh feeding plates at 25 °C and allowed to lay eggs for 4 h. Progeny embryos were transferred to fresh feeding plates and counted. Larvae and terminally arrested embryos were counted 24 h later to assay embryonic lethality, or examined by DIC microscopy after hatching to assess the pharyngeal phenotype.
to examine D2096.6::gfp expression, OK0666 [cuEx553] L4 hermaphrodites were transferred to plates seeded with ubc-9 RNai feeding E. coli or OP50 and grown 20 h at 20 °C. these animals were transferred to fresh feeding plates, and GFP expression was examined in progeny embryos and larvae.

Microarray and data analysis
Mixed stage populations of N2 and OK0660 [tbx-2(bx59)] animals grown at 25 °C were treated with bleach/sodium hypochlorite to isolate embryos [21]. aliquots of embryos were examined to verify comparable age distributions, and RNa was isolated using tRIzol (Invitrogen) and further purified using RNeasy Kit (Qiagen) following manufacturers' protocols.
total RNa from two independent populations of N2 embryos and three independent tbx-2(bx59) embryos were labeled and hybridized to affymetrix C. elegans Genome GeneChips by the UIC Core Genomic Facility (CGF). the microarray data was analyzed using the R statistical programming language, using the Bioconductor suite of tools [29], and the affy package. Normalization to correct for chip-to-chip variation was done using the Robust Multiarray averaging (RMa) method of microarray normalization [30]. Microarray results were pre-filtered using the genefilter function (25 % of the probes have a measured intensity of at least 100 on the original scale and the coefficient of variation is between 0.7 and 10 on the original scale) [31]. the limma package [32] was used to calculate differential expression using the limma linear model fit, eBayes smoothing of standard errors, and Benjamini-hochberg (Bh) multiple test correction with a false discovery rate of 5 % [33]. Probes were matched to genes using the affymetrix-to-WormBase ID table for WS210 (http//:www.wo rmbase.org). Probes mapping to more than one gene were discarded. When one or more probes mapping to a gene were differentially expressed, that gene was considered to be differentially expressed. One GeneChip hybridized with tbx-2(bx59) RNa exhibited high variation in the spiked in control probes (tBXa) compared to the other samples, and data from this chip was not included in our analysis.

Microscopy
Worms were visualized using a Zeiss axioskop microscope equipped for DIC and fluorescence microscopy, and images were captured using an axiocam MRm camera and axio-Vision software.

tBX-2 interacts with UBC-9 via two SUMO consensus sites
We previously showed using yeast two-hybrid assays that tBX-2 specifically interacts with the e2 SUMO-conjugating enzyme UBC-9 [20], and we used this assay to identify sites in tBX-2 mediating this interaction. tBX-2 contains several sites matching the SUMO consensus site ΨKX(D/e) (Supplementary table 1) [17,18]. the two highest-scoring matches to this consensus are an LK 231 Ie sequence located near the C-terminus of the t-box DNa binding domain and a VK 400 Ke sequence located near the tBX-2 C-terminus (Fig. 1a). LKIe is located in a conserved region of the t-box, and a SUMO consensus site is found at this position in many t-box factors, including all members of the tbx2 sub-family [34]. VKKe is located in a region that is not highly conserved among t-box factors, although high scoring SUMO consensus sites are found near the C-terminus of tBX-2 proteins from C. elegans, C. briggsae, and C. remanei, suggesting this site may be functionally conserved (Fig. 1b). We mutated each of these two sites in C. elegans tBX-2 to all alanines either in single mutants (LKIe→aaaa or VKKe→aaaa) or in a double mutant (LKIe/VKKe→aaaa) and tested whether these mutants affected the ability of a tBX-2 bait to interact with UBC-9 prey. Interactions were scored in plate assays for histidine prototrophy and ß-galactosidase (ß-gal) expression, and the level of interaction was quantified by measuring ß-gal activity.
We found that UBC-9 interaction with tBX-2 was affected by mutations affecting both the LKIe and VKKe sites. UBC-9 interacted with both the tBX-2 LKIe→aaaa and tBX-2 VKKe→aaaa single mutants in plate assays, but this interaction was reduced to 63 and 22 % of the levels observed for wild-type tBX-2, respectively (Fig. 1b).  [62]. high-scoring SUMO consensus sites are indicated in grey, and identical residues are marked with asterisks. c Quantification of ß-galactosidase activity in yeast expressing the indicated tBX-2 protein or the empty pLexa as bait and UBC-9 prey in replicate samples from three independent experiments (n = 7). Differences between mutants and wild-type tBX-2 or different mutants (bracket) are statistically significant at *p < 0.05 or **p < 0.005. Error bars indicate the standard error of the mean In comparison, UBC-9 failed to interact with the tBX-2 LKIe/VKKe→aaaa double mutant in plate assays, and the ß-gal activity was close to that obtained using an empty bait plasmid (Fig. 1b, Supplementary Figure 1). as a control, we found that wild-type and all of the mutant tBX-2 proteins retained the ability to interact with an unrelated protein UNC-37 in yeast two-hybrid assays, indicating the mutant proteins were expressed (Supplementary Figure 1). thus, both the LKIe and VKKe sites can interact with UBC-9. Because mutating both of these sites reduces interaction to near background, we believe they are the primary sites in tBX-2 that mediate this interaction. We have not tested other potential tBX-2 SUMOylation sites for interaction with UBC-9 in two-hybrid assays.
tBX-2 can be SUMOylated in mammalian cell assays to determine if tBX-2 can be SUMOylated, we co-expressed full-length tBX-2 and human SUMO-1 in COS-1 cells. tBX-2 was fused to poly-histidine and pulled down using Ni 2+ -beads under denaturing conditions, while tBX-2 and SUMO-1 were tagged with V5 and ha epitope tags, respectively, for detection on Western blots. a SUMO-1ΔGG mutant lacking the C-terminal Gly-Gly motif required for conjugation to target lysine residues was used as a control to demonstrate SUMO conjugation.
When co-expressed with SUMO-1, wild-type tBX-2 formed several more slowly migrating bands detectable with both anti-ha and anti-V5 (Fig. 2a). these bands likely represent mono-and multi-SUMOylated tBX-2, although proteins SUMOylated at different sites can also migrate at different positions due to the branched nature of the SUMOylated protein [35]. In comparison, no SUMOylated tBX-2 was detected when co-expressed with SUMO-1ΔGG. as for many SUMOylated proteins, we found only a fraction of tBX-2 (~10 %) is SUMOylated in these assays. Similar results were obtained when tBX-2 was co-expressed with human SUMO-2 or SUMO-3 (Supplementary Figure 2). We next asked how mutations in the UBC-9 interaction sites affected tBX-2 SUMOylation. SUMOylation of tBX-2 LKIe→aaaa and tBX-2 VKKe→aaaa were reduced to approximately 30 and 70 % of the level of wild-type tBX-2, respectively, while SUMOylation of the tBX-2 LKIe/VKKe→aaaa double mutant was further reduced to a level comparable to background (Fig. 2a, lanes 2 and 9). these results indicate that tBX-2 can be SUMOylated, and that the LKIe and VKKe sites for UBC-9 interaction are required for tBX-2 SUMOylation.
We next mutated the SUMO-conjugated lysine residues in the LKIe and VKKe sites to arginine, which is a conservative substitution that cannot be conjugated to SUMO, and we examined SUMOylation in COS-1 cells (Fig. 2b).
the tBX-2 K231R mutant affecting LKIe exhibited reduction of the more slowly migrating SUMOylated products, but these products were not eliminated, while the fastest migrating band appeared unaffected. In comparison, overall SUMOylation of the tBX-2 K400R mutant affecting VKKe was strongly reduced, and the fast migrating form of SUMOylated tBX-2 was nearly completely eliminated. SUMOylation of the tBX-2 2KR double mutant containing K231R and K400R was similar to that of the tBX-2 K400R single mutant, but this mutant was still SUMOylated above background levels (compare Fig. 2b, lanes 2 and 9). Because both of these conservative mutations affect the pattern of tBX-2 SUMOylation, we believe lysine residues in both the LKIe and VKKe SUMO consensus sites are SUMOylated.

Fig. 2 SUMOylation of tBX-2 is mediated via two SUMO consensus sites. Western blots of Ni-Nta pulled-down wild-type and mutant tBX-2/V5/hIS probed to detect tBX-2 (bottom) or SUMO-1 (top). Combinations of proteins (grey boxes) were expressed in COS-1 cells.
a Wild-type tBX-2 and mutants with SUMOylation sites converted to all alanines. b Wild-type tBX-2 and mutants with SUMO acceptor lysines converted to arginines. the position of the fastest migrating SUMOylated form of tBX-2 is indicated (bar and circle) and the position of molecular weight markers are indicated in kDa (bars). Signal in the lower panel in a was detected using a StORM Molecular Imager and clearly demonstrates more slowly migrating tBX-2 bands when co-transfected with SUMO-1. tBX-2/V5/hIS is ~52 kDa, and ha-SUMO-1 is ~13 kDa. the fastest migrating SUMOylated form migrates somewhat slower than predicted by its molecular weight, which is a common feature of SUMOylated proteins tBX-2 is a transcriptional repressor in mammalian cells C. elegans tBX-2 is most closely related to the mammalian t-box repressors tbx2 and tbx3 [20]. We wanted to ask if tBX-2 functions similarly to repress transcription and, if so, whether this activity depends on SUMOylation. Because mutations affecting the LKIe SUMOylation would likely affect DNa binding, we asked if tBX-2 fused to the heterologous GaL4 DNa binding domain (tBX-2:GaL4) could repress expression of the 5xGaL4:tk:luc reporter. this reporter contains five copies of the GaL4 binding site upstream of thymidine kinase promoter:luciferase reporter, and tBX-2:GaL4 repressed expression of this reporter up to fivefold (Fig. 3). SUMOylation is most often associated with transcriptional repression, and we expected that mutation of the LKIe and VKKe SUMO sites would reduce this repressor activity. however, we found that the tBX-2 LKIe/VKKe→aaaa double mutant repressed 5xGaL4:tk:luc similarly to wild-type tBX-2. Co-expressing SUMO-1 did not affect repression of 5xGaL4:tk:luc with either wildtype or mutant tBX-2 (Supplementary Figure 3). thus, SUMOylation is not required for tBX-2:GaL4 repressor activity in COS-1 cells.
tBX-2 function is SUMO-dependent in C. elegans mab-22(bx59) is a temperature-sensitive mutant that exhibits defects in male tail ray formation and partially penetrant larval lethality. bx59 has recently been identified as a missense mutation in tbx-2 (King Chow, pers. comm.), and we subsequently refer to this mutation as tbx-2(bx59). We examined the viability and pharyngeal morphology of tbx-2(bx59) mutants produced from hermaphrodites shifted to the non-permissive temperature (25 °C) at the L4 stage. Forty-four % of these animals arrested at the L1 stage (n = 64) with variable pharyngeal abnormalities (Fig. 4a-c). these phenotypes are similar to those observed in tbx-2(RNAi) animals and are not as severe as those observed in tbx-2 null mutants [20,36], and we conclude that tbx-2(bx59) is a hypomorphic allele.
We hypothesize that tBX-2 function depends on SUMOylation. to test this hypothesis, we asked if inhibiting SUMOylation by reducing UBC-9 or the SUMO Fig. 3 Dose-dependent transcriptional repression by tBX-2:GaL4. Relative luciferase activity in experiments co-transfecting increasing amounts of wild-type and mutant tBX-2:GaL4 with the 5xGaL4:tk:luc reporter. Data shown is the average of three assays and is representative of multiple independent experiments. Error bars indicate standard deviation. Statistically significant differences from control transfections are marked (*p < 0.05) Fig. 4 Pharyngeal defects in tbx-2(bx59) mutant are enhanced by reduced SUMOylation DIC micrographs of the pharynx of L1 larvae of the indicated genotypes raised at the non-permissive temperature (25 °C). a Wild-type N2. b tbx-2(bx59) exhibiting a mild pharyngeal defect. c tbx-2(bx59) exhibiting a more severe pharyngeal defect. tbx-2(bx59); ubc-9(RNAi) (d) and tbx-2(bx59); smo-1(RNAi) (e) L1 larvae exhibiting very severe defects resembling those of tbx-2 null mutants. Arrowheads mark the extent of pharyngeal tissue. anterior is left. the frequency of these phenotypes is indicated in table 1 protein SMO-1 by RNai could enhance the phenotype of tbx-2(bx59) mutants. Using the RNai-feeding method, we found that ubc-9(RNAi) produced a relatively low frequency of embryonic arrest in a wild-type background (table 1) [26]. In comparison, tbx-2(bx59); ubc-9(RNAi) double mutants exhibited a synergistic increase in the frequency of arrested embryos (table 1). tbx-2(bx59); smo-1(RNAi) double mutants also exhibited an increased frequency of embryonic lethality compared to each single mutant; however, the smo-1(RNAi) lethality alone was higher making it difficult to determine if the double mutant lethality was more than additive (table 1).
tBX-2 and SUMOylation are required for repression of D2096.6 gene expression to identify genes downstream of tBX-2, we used microarrays to compare mRNa levels in populations of wild-type and tbx-2(bx59) embryos grown at 25 °C. We found 1,276 protein coding genes that are differentially expressed in tbx-2(bx59) (Bh corrected p ≤ 0.05) (Supplementary table 2). a total of 1,030 of these genes (80.7 %) are upregulated in tbxc-2(bx59), consistent with our hypothesis that tBX-2 functions as a transcriptional repressor.
We focused on the gene D2096.6, which had previously been shown to be specifically expressed in the pharyngeal muscles, marginal cells, and epithelial cells under control of the Foxa-family transcription factor Pha-4 [37,38]. We observed an approximately 1.8-fold increase in D2096.6 expression in tbx-2(bx59) mutants in our microarray (Bh corrected p = 0.03). While several candidate t-box binding sites are located upstream of D2096.6, our preliminary characterization of this promoter suggests it is indirectly regulated by tBX-2.
to determine how tBX-2 regulates D2096.6 expression, we compared expression of a D2096.6::gfp reporter in wild-type and tbx-2 mutants. Consistent with previous studies [37], we observed that a D2096.6::gfp reporter was expressed in wild-type embryos specifically in the pharynx at beginning approximately at the bean stage when the pharyngeal primordium forms. expression was typically observed in one to two cells in the pharynx in one and onehalf fold embryos (Fig. 5), and no expression was observed outside the pharynx. the number of GFP-expressing cells increased and animals hatched as L1s with GFP expression in pharyngeal muscles, marginal cells, and epithelial cells (Fig. 5d). In comparison, in tbx-2(bx59) and tbx-2(ok529) embryos D2096.6::gfp was expressed in more cells in the pharynx, and expression was observed in many cells outside the pharynx, including body wall muscles and hypodermal cells (Fig. 5b, c; table 2). ectopic D2096.6::gfp expression continued into the L1 larval stage where it was observed in bodywall muscle, hypodermal, and gut cells (Fig. 5e, f). these results indicate tBX-2 is an upstream regulator that represses D2096.6 expression both temporally and spatially.
to ask if SUMOylation is necessary for tBX-2 function, we examined D2096.6 expression in animals where activity of UBC-9 was reduced using feeding-RNai. the most severely affected ubc-9(RNAi) animals have a highly disorganized morphology that makes it difficult to identify specific tissues [20]. therefore we characterized D2096.6::gfp expression in older embryos that had undergone morphogenesis and the surviving L1 larvae. ubc-9(RNAi) resulted in D2096.6::gfp expression in posterior body wall muscles in embryos in a pattern similar to that which we have observed in tbx-2(bx59) and tbx-2(ok529) embryos (Fig. 6a, b). In larvae, we observed expression in body wall muscles and hypodermal cells in the posterior of the worm similar to the expression pattern we see in tbx-2 mutants (Fig. 6 c, d). thus, SUMO-dependent mechanisms repress D2096.6::gfp expression and the similarities in the pattern of ectopic expression in ubc-9(RNAi) and tbx-2 mutants strongly suggests tBX-2 function depends on SUMOylation.
Mammalian tbx2 subfamily members can be SUMOylated to determine if SUMOylation is a conserved mechanism regulating t-box factor activity, we asked if mammalian orthologs of tBX-2 could be SUMOylated. human tBX2 or mouse tbx3 tagged with poly-histidine and V5 were co-expressed in COS-1 cells with or without ha:SUMO-1  tbx-2(ok529); cuEx553 b 20 (35) ubc-9(RNAi); cuEx553 75 (48) and pulled down using Ni 2+ -beads under denaturing conditions similarly to C. elegans tBX-2. When co-expressed with SUMO-1, both tBX2 and tbx3 formed two more slowly migrating SUMOylated bands (Fig. 7). this data indicates that other tbx2-subfamily members can be SUMOylated, and we hypothesize that SUMOylation may be a common regulatory method of t-box factor activity.

Discussion
here we show that C. elegans tBX-2 and its mammalian orthologs human tBX2 and mouse tbx3 can be SUMOylated, and that tBX-2 SUMOylation depends on two SUMO consensus sites that mediate interaction with the e2 SUMO conjugating enzyme UBC-9. We further demonstrate that C. elegans tBX-2 can function as a transcriptional repressor when fused to a heterologous DNa-binding domain; however, mutations that eliminate SUMOylation do not affect this repressor activity. Finally, we provide genetic evidence that SUMOylation is required for tBX-2 function in vivo by showing that reduction of SUMOylation enhances the phenotype of a hypomorphic tbx-2 mutant and phenocopies the loss of tbx-2 on expression of one gene that is downstream of tBX-2.
two tBX-2 SUMO consensus sites interact with UBC-9 and mediate SUMOylation tBX-2 contains two predicted high-scoring SUMO consensus sites. VKKe located near the C-terminus and LKIe located within the t-box DNa-binding domain. each of theses sequences interacts with the e2 SUMO-conjugating enzyme UBC-9 in yeast two-hybrid assays, and they are the only sites that can mediate this interaction. Our data strongly suggests that both of these sites are SUMOylated. We observed multiple SUMOylated forms of tBX-2 in COS-1 cells, and mutations affecting either SUMO site reduce the amount of SUMOylated tBX-2. however, there are differences in how specific mutations in these sites affect tBX-2 SUMOylation. For VKKe, mutation of the acceptor lysine (K400R) results in a more severe reduction in SUMOylation than mutating this site completely to alanines. In comparison, mutating the LKIe site to alanines results in a large decrease in SUMOylation whereas mutation of the acceptor lysine (K231R) has a moderate affect and preferentially affects the more slowly migrating SUMOylated forms of tBX-2. Mutations affecting only the acceptor lysine likely retain interaction with UBC-9 (Sampson et al. 2001), whereas mutations converting the SUMO site to alanines eliminate UBC-9 binding. this difference likely underlies the different effects we observed on tBX-2 SUMOylation when these sites are mutated.
t-boxes have a highly conserved structure when bound to DNa [39][40][41][42], and tBX-2 LKIe is located within the α3 helix that spans the DNa backbone. While SUMOylation can occur in α helices [43,44], this is an unusual 2˚ structure for UBC-9 interaction and SUMOylation, as UBC-9 has been shown to bind SUMO consensus sites in extended loops [18,45]. however, recent evidence indicates that some t-boxes have significant structural flexibility that might allow SUMOylation at LKIe. the tbx20 t-box exists as a molten globule with an unstable tertiary structure allowing flexibility between 2° structural domains [46]. Likewise in tBX5, the 3 10 -helix located just C terminal to the α3 helix is unstructured in the absence of DNa [42]. While the α3 helix remains structured when tBX5 is not bound to DNa, the LKIe site would be more accessible to bind UBC-9.
LKIe was also investigated as a potential SUMOylation site in the human t-box factor tBX22 [47]. Interestingly, mutation of this site eliminated tBX22 SUMOylation, however, this is believed to result indirectly from possible effects on DNa binding as was observed with several DNa-binding defective mutants. Biochemical analyses of SUMOylated t-box factors is necessary to explicitly determine if this site is SUMOylated in different proteins.
tbx-2 genetically interacts with ubc-9 and smo-1 tbx-2(bx59); ubc-9(RNAi) and tbx-2(bx59); smo-1(RNAi) animals exhibit penetrant embryonic arrest and enhanced pharyngeal defects, indicating that reduced SUMOylation affects tBX-2 activity in vivo. In particular, the Fig. 7 SUMOylation of mammalian tbx2 subfamily members Western blots of Ni-Nta pulled-down human tBX2/V5/hIS and mouse tbx3/V5/hIS probed to detect tBX2 and tbx3 (bottom) or SUMO-1 (top). Combinations of proteins expressed in COS-1 cells are indicated (grey boxes). SUMOylated forms of tBX2 and tbx3 are marked with arrowheads, and MW weight markers are indicated (kDa). a cross-reacting background band was detected in all lanes using anti-ha enhanced pharyngeal phenotype of these animals resembles that of tbx-2 null mutants, strongly suggesting that tbx-2 and SUMOylation function in the same pathway to specify pharyngeal muscle fate [48]. these observations are consistent with the hypothesis that tBX-2 function is SUMO-dependent.
In comparison, we do not know why enhanced embryonic lethality was observed in tbx-2(bx59); ubc-9(RNAi) and tbx-2(bx59); smo-1(RNAi) animals. Neither tbx-2 null mutants nor tbx-2(RNAi) animals exhibit embryonic lethality [20], indicating lethality does not result from loss of zygotic tbx-2(bx59) activity. One possibility is that tbx-2(bx59) may have a partial gain-of-function character, and reducing SUMOylation deregulates this activity. tbx-2(bx59) mutates a conserved histidine residue within the dimerization domain of the t-box to a tyrosine (h145Y; accession CCD69847), and mutations affecting this domain in human tBX1 result in gain-of-function associated with some cases of DiGeorge and velocardiofacial syndromes [49]. alternatively, decreased SUMOylation of a parentally provided, RNai-resistant protein or mRNa tbx-2 gene product might be required for viability. Finally, decreased SUMOylation may affect SUMO-dependent activity of both tBX-2 and another factor with a partially redundant activity required for embryogenesis.
SUMOylation of transcription factors is usually associated with transcriptional repression, and it can promote recruitment of chromatin remodeling and histone modifying complexes [12,50]. Indeed, SUMOylation of the C. elegans ets-domain factor LIN-1 leads to interaction with MeP-1 and the NuRD chromatin repressor complex [51]. however, our results argue that SUMOylation affects tBX-2 function by a different mechanism. Unlike LIN-1, repressor activity of tBX-2:GaL4 fusion protein in mammalian cells is independent of SUMOylation, and in extensive yeast twohybrid screens tBX-2 has not been observed to interact with MeP-1 [20].
SUMOylation at the LKIe site in the t-box would likely affect the ability of tBX-2 to bind sites in the genome and to interact with other factors binding tBX-2 regulated promoters. If this is the case, our assays for tBX-2:GaL4 repressor activity would be insensitive to SUMOylation, since they depend on a heterologous DNa-binding domain targeting a synthetic promoter. Indeed, SUMOylation has been shown to affect DNa-binding activity of specific transcription factors [52][53][54].
It is difficult to predict how SUMOylation at the VKKe site would affect tBX-2 activity. t-box proteins are poorly conserved outside of the DNa-binding domain, and the function of the tBX-2 C-terminus is unknown.
Interestingly, our preliminary results indicate VKKe is located near an interaction site for the Groucho-family corepressor UNC-37 (t. Crum and P. Okkema, unpublished). SUMOylation can regulate interaction with Groucho-family proteins [55,56], and we hypothesize SUMOylation at VKKe has a similar function. Groucho-interaction motifs are enriched in t-box factors from C. elegans, Drosophila, and humans [57], and several t-box factors have been shown to interact with Groucho-family proteins [58][59][60][61]. SUMOylation may be a common mechanism for regulating t-box factor interaction with Groucho.