Saos2 human osteosarcoma cells expressing the tetracycline inducible transactivator rTA (Saos2-TetOFF, Clontech) were grown in DMEM (10% fetal bovine serum, 1% penicillin/streptomycin, and 1 ug/ml G418). For transfections, the cells were grown to at least 70% confluence. 2 × 10⁶ cells were used per transfection with each construct (Amaxa Nucleofactor).

48 hours after transfection with DNA, Saos2-TetOFF cells were lysed in 1ml Buffer A (100mM NaH₂PO₄, 10mM Tris-Cl, 6M GuHCl, pH=8.0) supplemented with 10mM N-ethyl maleimide and 7% β-mercaptoethanol. The lysates were then sonicated, spun down for 15 min., and incubated with 40μl of Ni-NTA agarose beads (Qiagen) for 1 hour. The beads were then washed with Buffer A once, Buffer B (100mM NaH₂PO₄, 10mM Tris-Cl, 8M urea, pH=8.0) twice, and Buffer C (100mM NaH₂PO₄, 10mM Tris-Cl, 8M urea, pH=6.3) twice. 50mM TRIS was used to wash beads, which were then resuspended in 2X NuPAGE LDS-sample buffer and 250mM imidazole.

48 hours after transfection, GFP was visualized and the cultured cells were rinsed with phosphate-buffered saline, scraped with lysis buffer (50mM Tris-HCl pH 7.5, 1mM EDTA, 200mM NaCl, 0.1% Nonidet P-40,) supplemented with a protease inhibitor cocktail (Complete Mini, EDTA-free, Roche), 15mM N-ethyl maleimide, 20mM iodoacetamide, and sonicated. 4X NuPAGE® LDS Sample Buffer (Invitrogen) was added to these lysates after normalization of protein concentrations. These protein extracts were run on 4-12% Bis-Tris gels and MOPS SDS running buffer. Proteins were transferred onto PVDF membranes and blocked in 5% nonfat powdered milk in TBS/Tween at room temperature for 1 hour. The blots were probed with primary antibody diluted in 5% nonfat powdered milk in TBS/Tween and then with secondary antibody diluted in TBS/Tween, both at room temperature for one hour. Primary antibodies used were anti-V5 (1:2000, Serotec), anti-cMyc (1:1000, Santa Cruz). Secondary antibody used were goat anti-mouse horseradish peroxidase (1:3000, Bio-Rad). After application of each antibody, the blots were rinsed three times with TBS/Tween under similar conditions. The blots were developed by ECL (Pierce).

Transfected cells were seeded on glass coverslips coated with 0.1% gelatin. 48 hours after transfection, cells were washed three times in PBS and fixed with 4% paraformaldehyde in PBS for 15 minutes at room temperature. A 5 minute incubation in 0.2% Triton in PBS was used to permeabilize the cells before blocking unspecific sites using 3% BSA in PBS for 1 hour. Primary antibodies diluted in blocking solution were applied for 1 hour. Primary antibodies used: anti-V5 (1:100, Serotec), anti-cMyc (1:100, Santa Cruz). Secondary antibodies diluted in blocking solution were applied for 30 minutes. Secondary antibodies used: Alexa Fluor® 543 goat anti-mouse IgG and Alexa Fluor® 488-conjugated anti-rabbit IgG (1:400, Invitrogen). Nuclei were stained using DAPI and coverslips were mounted on glass slides using Mowiol (Calbiochem). Images were acquired using a Zeiss 510 META confocal microscope.

Nuclear extracts were prepared using a Nuclear Extract Kit (Active Motif). Recombinant SUMO-DLX3 fusion protein was produced using the SUMOpro kit (LifeSensors). A probe containing the DLX3 consensus binding site (GGGGGATAATTGCTGG), was radiolabeled using the High Prime DNA Labeling Kit (Roche) and [γ-³²P]dCTP. Nuclear extracts or recombinant proteins were pre-incubated in 1X gel shift binding buffer (Promega) for 15 minutes at 4°C, with an excess of unlabeled probe for competition assays or with appropriate antibody (anti-DLX3 or anti-cMyc) for supershift assays. After this pre-incubation, each sample was supplemented with 5×10⁴ DPM of radiolabeled probe and incubated for 30 minutes at 4°C. The binding reactions were resolved on 6% DNA retardation gels (Invitrogen). The gels were dried and DNA-protein complexes were visualized by autoradiography.

To determine the functionality of the two potential SUMOylation sites in DLX3, we co-expressed V5DLX3 (DLX3 tagged with a V5 epitope) with wildtype SUMO1 in Saos2-TetOFF cells. As a control, we also co-expressed V5DLX3 with a mutated form of SUMO1 lacking the C-terminal double glycine which forms an isopeptide bond with the target protein (SUMO1-ΔGG). SUMO1 and SUMO1-ΔGG are tagged with both a His tag and a c-Myc tag. Western blot was performed to analyze the size pattern of c-Myc tagged and V5-tagged proteins in these cells (Fig. 2A). Results monitored with anti-cMyc showed a low molecular weight band, around 15 kDa, corresponding to the monomer of SUMO1 or SUMO1-ΔGG, as well as a smear corresponding to all endogenous proteins SUMOylated by SUMO1, but not SUMO1-ΔGG (Fig. 2A, left panel). The anti-V5 blot revealed two major bands for V5DLX3 when co-expressed with SUMO1: the lower band corresponding to the normal size of V5DLX3, and the upper corresponding to a V5DLX3 derivative with additional 15-20 kDa (Fig. 2A, right panel). This upper band was not detected with SUMO1-ΔGG. These observations suggest that V5DLX3 is SUMOylated by SUMO1. To confirm that the upper complex detected is indeed SUMOylated V5DLX3, we performed the same western blot analysis after purifying the protein extracts with a Ni-column (Ni-NTA, binding His-tagged SUMO1) to pull-down all SUMOylated proteins. Using this approach, we detected SUMOylated V5DLX3 among the proteins purified on the Ni-column (Fig. 2B). These results demonstrate that V5DLX3 is SUMOylated by SUMO1. Moreover, the fact that we detected only one band for SUMOylated V5DLX3 suggests that only one lysine is involved in DLX3 SUMOylation.

To test which of the two potential SUMOylation sites in DLX3 is actually involved in SUMOylation, we mutated the lysine residues at positions 83 and 112 into arginines (Fig. 3A). Thus, we generated two single mutants (DLX3^K83R and DLX3^K112R) as well as a double mutant in which the two lysine residues were mutated (DLX3^2K). In Saos2-TetOff cells, we co-expressed SUMO1 or SUMO1-ΔGG with DLX3^WT, DLX3^K83R, DLX3^K112R or DLX3^2K, respectively. Western blot analysis was performed using anti-cMyc and anti-V5 antibodies, both on whole extracts and on Ni-column purified protein fraction (Fig. 3B). This assay revealed that mutation in position 83 does not preclude SUMOylation of V5DLX3, since a band corresponding to SUMOylated DLX3 appears for both DLX3^WT and DLX3^K83R. This band, however, was absent for DLX3^K112R. As predicted, the double mutant DLX3^2K is not SUMOylated. These observations demonstrate that lysine K112 is the only SUMOylation site in DLX3.

In order to assess the effect of SUMOylation on DLX3 subcellular localization, we performed immunohistochemical analysis on Saos2-TetOFF cells co-transfected with V5DLX3 and SUMO1 or SUMO1-ΔGG (Fig. 4A). Immunostaining using anti-cMyc antibody showed that SUMO1 was present in both the cytoplasm and the nucleus: the nuclear expression was rather diffuse, however, it tended to be stronger at the nuclear periphery, while in the cytoplasm it formed dense aggregates (Fig 4A, c). SUMO1-ΔGG exhibited a completely different distribution pattern: it was distributed throughout the cytoplasm in a filament-like pattern and was also present in the nucleus (Fig 4A, g). The distribution pattern of V5DLX3 in the nucleus appeared the same in the presence of wild-type and mutant SUMO1 (Fig. 4A, b and f). This observation suggests that SUMO1 does not dramatically affect DLX3 subcellular localization. However, we observed a partial co-localization between DLX3 and SUMO1 at the periphery of the nucleus (Fig. 4A, d). We used cell fractionation and western blot analysis to corroborate our immunohistochemical observations (Fig 4B). The anti-cMyc blot revealed that the cytoplasm contained both monomeric SUMO1 and SUMO1 conjugated with target proteins, while in the nucleus SUMO1 was essentially present in its conjugated form. SUMO1-ΔGG, that can only be monomeric, exhibited a much stronger expression in the cytoplasm than in the nucleus. The anti-V5 blot showed that both V5DLX3 and SUMOylated V5DLX3 were accumulated almost exclusively in the nuclear fraction. These data confirm that SUMO1 does not dramatically affect DLX3 subcellular localization. We then investigated whether DLX3 subcellular localization was affected by its inability to be SUMOylated. To address this question, immunohistochemical analysis was carried out with Saos2-TetOff cells transfected with V5DLX3^WT or V5DLX3^K112R. As shown in Figure 4C, both wild type and K112R mutant were exclusively located in the nucleus. Thus, preventing DLX3 from being SUMOylated does not affect its subcellular localization. Taken together, these data suggest that SUMOylation does not play a significant role in determining the distribution pattern of DLX3 in cells.

To determine the role of SUMOylation on DLX3 function, we next explored the effect of SUMOylation on DLX3 binding activity. We first asked if SUMOylated DLX3 was able to bind DNA. To address this question, we first generated a recombinant SUMO-DLX3 fusion protein and tested its ability to bind to the DLX3 consensus binding site in an Electrophoresis Mobility Shift Assay (EMSA), using a probe containing a DLX3 binding site. A protein-DNA complex was formed, that could be competed using an excess of non-radioactive probe (self competitor), but not using an excess of a mutated non-radioactive probe (mutant competitor) (Fig. 5A, lanes 2-4). To confirm that this protein-DNA complex contained SUMO-DLX3, we performed a supershift assay using anti-DLX3 antibody and were able to detect a protein-DNA-antibody complex (Fig. 5A, compare lanes 5 and 6). Even though the SUMO-DLX3 fusion protein does not perfectly mimic the tertiary structure of SUMOylated DLX3, this strategy is commonly used in the field [Ouyang et al., 2009] and gives a preliminary suggestion that having SUMO bound to DLX3 does not seem to affect its ability to bind DNA. To test this hypothesis in a more physiologically meaningful context, we performed EMSA using nuclear extracts from cells expressing V5DLX3 with SUMO1 or SUMO1-ΔGG (Fig. 5B, inset). In both cases (wild-type and mutant SUMO1), we could detect a protein-DNA complex formed between DLX3 and the probe (Fig. 5B, arrow). In order to detect if SUMOylated DLX3 is involved in a complex with the probe, we performed a supershift assays using anti-cMyc antibody (recognizing SUMO1 and SUMO1-ΔGG). As shown in Figure 5B, a partial supershift could be detected in the presence of anti-cMyc when DLX3 was expressed with SUMO1, but not with SUMO1-ΔGG (Fig. 5B, arrowhead, compare lanes 3 and 5). We performed a similar experiment in which we prepared nuclear extracts from cells expressing SUMO1 with DLX3^WT or DLX3^K112R (Fig. 5C, inset), and performed an EMSA as described above. As expected, using anti-cMyc antibody, we could detect a partial supershift for DLX3^WT but not for DLX3^K112R (Fig. 5C, compare lanes 3 and 5). These observations demonstrate that when SUMO1 is bound to DLX3, DLX3 is still able to bind to its consensus binding site. We also showed that DLX3^K112R is able to bind DNA (Fig 5C, arrowhead, lanes 4 and 5), demonstrating that preventing DLX3 from being SUMOylated does not affect its ability to bind DNA.