A Novel SUMO1-specific Interacting Motif in Dipeptidyl Peptidase 9 (DPP9) That Is Important for Enzymatic Regulation^*

Antibodies

Goat anti-Uba2 were previously described (36). Rabbit anti-PIAS2 antibodies and mouse GST antibodies were purchased from Sigma and Santa Cruz Biotechnology, respectively. Mouse anti-GMP1 antibodies were obtained from the Developmental Studies Hybridoma Bank. Rabbit anti-DPP9 antibodies were purchased from Abcam. For ELISA we produced goat anti-DPP9 specific antibodies by injecting a goat with 500 μg of full-length DPP9 on days 0, 40, and 70. Goat anti-DPP9 antibodies were affinity purified from the serum.

Plasmids

SUMO1 and SUMO2 were cloned into pET11 (37, 38). DPP8 and DPP9 were subcloned into a pFASTBacHT plasmid (Invitrogen), baculoviruses were generated according to the instructions of the Bac to Bac Baculovirus expression system, using SF9 cells. DPP9 was cloned into pcDNA3.1 vector for expression of N-terminal HA- or FLAG-tagged DPP9 (using the BamHI and NotI restriction sites). Single point mutations in SUMO1, SUMO2, or DPP9 were generated using primers for site-directed mutagenesis.

Peptides

Peptide sequences are included in supplemental “Materials and Methods”. All peptides (>80% purity) were purchased from Genescript. Protein purification methods are described in supplemental “Materials and Methods”.

SUMO Pull-down Screen

SUMO proteins were coupled to CNBr-Sepharose (Sigma) at a concentration of 1 mg of protein/ml of beads. To identify SUMO-interacting proteins, 25 g of frozen HeLa cell pellets were resuspended in 50 ml of TB (20 mm Hepes/KOH, pH 7.3, 110 mm potassium acetate, 2 mm magnesium acetate, 0.5 mm EGTA), supplemented with protease inhibitors (1 μg/ml of leupeptin, pepstatin, and aprotinin, and 2 mm DTT) and homogenized by douncing. Cell lysates were centrifuged at 10,000 × g for 15 min, and again at 100,000 × g for 1 h. 10 ml of supernatant was incubated at 4 °C with 200 μl of SUMO1/SUMO2 or control beads that were inactivated with ethanolamine. After 2 h of incubation, beads were washed and eluted in TB supplemented with protease inhibitors and increasing NaCl concentrations (0.25, 0.5, or 1 m). Eluted proteins were precipitated with methanol/chloroform, resuspended in LDS sample buffer, separated on a 4–12% NuPAGE gradient gel (Invitrogen), and stained with colloidal Coomassie Brilliant Blue. In-gel tryptic digestion, desalting, and LC-MS/MS analysis were performed as reported previously (39). All pkl files corresponding to one sample were merged into a single mascot generic data file and searched against the human IPI protein database (version 3.26) using Mascot with Mascot Server (version 2.2). The following settings were used: digestion with trypsin allowing one miss cleavage, carbamylation of cysteine as fixed modification, oxidation of methionine as variable, 150 ppm MS mass accuracy, and 0.3 Da for fragmentation masses.

SUMO Pull Downs with Recombinant Proteins

500 ng of recombinant proteins were incubated for 1 h at 4 °C with bead-immobilized SUMO1, SUMO2, or ovalbumin in TB supplemented with 0.5% Tween 20 and 0.2 mg/ml of ovalbumin. After 1 h, beads were washed in TB containing 0.5% Tween 20 and protease inhibitors, and eluted with sample buffer. For peptide competition assays: recombinant DPP9 (500 ng) was incubated with peptides from the SUMO1 peptide library (0.1 mg/ml) for 1 h prior to incubation with the SUMO-immobilized beads.

ELISA

A 96-well plate (Immuno 96 MicroWell^TM Solid Plates, Nunc) was coated with 600 ng of purified recombinant proteins per well, overnight at 4 °C. Next, wells were blocked with ELISA buffer (TB supplemented with 3% BSA) for 1 h at room temperature. After blocking, recombinant interacting proteins were added to the wells. Following 2 h of incubation at 25 °C, wells were washed four times. Interacting proteins were detected using goat anti-DPP9 or mouse anti-GMP (SUMO1) antibodies. After incubation with secondary antibodies, wells were washed with ELISA buffer and ddH₂O. Reactions were developed with 3,3′,5,5′-tetramethylbenzidine-substrate reagent set from BD Biosciences (OptElA substrate). Absorbance was measured at 420 nm on an Appliskan microplate fluorimeter (Thermo Scientific) and Skanit software. Experiments were performed at least three times, in triplicates. For competition assays: SUMO1 was incubated with the DPP9-arm or a control peptide for 30 min prior to incubation with immobilized DPP9 or immobilized ovalbumin.

Immunoprecipitations

HEK293T cells were transfected with pcDNA3 vectors using the calcium-phosphate precipitation method. 48 h later, cells were harvested and dounced in ice-cold TB supplemented with 0.2% Tween 20. The homogenate was centrifuged at 100,000 × g at 4 °C for 20 min. Supernatants were pre-cleared for 30 min with protein G-coupled beads, followed by immunoprecipitation for 2 h at 4 °C with mouse anti-FLAG beads (SIGMA). After extensive washing with TB containing 0.2% Tween 20, bound proteins were eluted with 0.5 mg/ml of FLAG peptide in the same buffer.

Size Exclusion Chromatography

200 nm purified DPP9 wild-type or V285A mutant in TB buffer were loaded on an analytical Superdex 200 column (General electric) for size exclusion chromatography.

Kinetics Assays

Unless otherwise stated: purified recombinant DPP9 (25 nm) was incubated with varying concentrations of GP-AMC, in TB supplemented with 0.2% Tween 20, 0.2 mg/ml of BSA, 1 mm dithiothreitol. For activation assays: DPP9 was incubated with 2.5 μm SUMO proteins, RanGAP or RanGAP modified with SUMO1 for 2 h at 4 °C prior to the assay. Fluorescence was measured using the Appliskan microplate fluorimeter (Thermo Scientific) with 380-nm (excitation) and 480-nm (emission) filters and Skanit software, and analyzed using Prism software. Each experiment was performed at least three times, in replicates of four.

DPP9 Interacts with a Novel Interaction Surface on SUMO1: E67-interacting Loop (EIL)

To study the preferential association of DPP8 and DPP9 with SUMO1 but not SUMO2, we aimed to identify the surfaces on SUMO1 involved in these interactions. First we analyzed the interaction of DPP9 with SUMO1 mutants in residues that were previously shown via NMR and crystal structures to interact with SIM containing proteins (14, 15, 17, 20). We constructed the following mutants: SUMO1K37AV38A and SUMO1V38AK39A and analyzed their capacity to pull-down DPP8 and DPP9 from cell lysates. For control, we analyzed the binding of known SIM-containing proteins, Uba2 and PIAS2. Neither of these interacted with SUMO1K37AV38A and SUMO1V38AK39A. In striking contrast, both endogenous and recombinant DPP8 and DPP9 still interacted with these SUMO1-SIG mutants (Fig. 2, A and B). Furthermore, in pull down experiments with immobilized SUMO1, we showed that recombinant Uba2 and DPP9 can simultanously interact with SUMO1 (supplementary Fig. 1). These results show that the interaction of both DPP8 and DPP9 with SUMO1 involves a surface different from the conventional SIG of SUMO1.

Open in a separate window

FIGURE 2.

DPP9 interacts with the EIL on SUMO1 independently of the SIG. A, SUMO1 and SUMO1-SIG mutant proteins (SUMO1K37AV38A and SUMO1V38AK39A) were immobilized on Sepharose beads and incubated with HeLa cell extracts. Ovalbumin (Ov)-coupled beads were used for control. Bound proteins were analyzed by immunoblotting using specific antibodies. For detection of DPP9, 0.1% input was loaded. For DPP8, Uba2, and PIAS, 0.2% input were loaded per lane (B). Shown is a pull-down assay as in A with recombinant DPP9, DPP8, Uba2, or GST-PIAS3. Input lane contains 50 ng of protein. C, recombinant DPP9 was incubated with 15-mer peptides of a peptide library covering he SUMO1 amino acid sequence. The DPP9-peptide solutions were then incubated with SUMO1-immobilized beads. Following extensive washing steps, bound DPP9 was eluted with sample buffer and detected DPP9-specific antibodies. Input lane contains 50 ng of protein D. Recombinant DPP8 and DPP9 were incubated with SUMO1-immobilized beads. To elute bound proteins, beads were incubated with either SUMO1-peptide 5 (overlapping SIG) or SUMO1-peptide 9. E, SUMO1 was mutated in amino acids covering the sequence of SUMO1 peptide 9. Recombinant SUMO proteins were immobilized on ELISA slips and incubated with recombinant DPP9. Interacting DPP9 was detected using a DPP9 specific antibody. Experiments were repeated three times in triplicates, shown is a representative, including error bars. F, SUMO pull-down assay as described in B, using recombinant DPP9, Uba2, and GST-PIAS3 and bead-immobilized SUMO1 or SUMO1 mutants. G, shown is a three-dimensional structure of SUMO1 (blue) bound to a SIM peptide (green) (Protein Data Bank accession code 2ASQ). Highlighted in red is the E67-interacting loop (EIL), which is located on the opposite side (Phe⁶⁶-Gly⁶⁸) of the SIG.

To identify the surface of SUMO1 that is important for its interactions with DPP9, we analyzed a library of short 15-mer peptides that covered the complete SUMO1 sequence. In total 12 peptides were tested, each peptide contained a 7-amino acid overlap with the carboxyl terminus of the previous peptide. The peptides were mixed with DPP9 prior to incubation with immobilized SUMO1 beads. If a peptide covers the surface of SUMO1 that associates with DPP9, it may bind to DPP9 and consequently block the SUMO1-DPP9 interaction. Most peptides did not efficiently compete with the DPP9-SUMO1 interaction (Fig. 2C). Also a peptide that includes Val³⁸ and Lys³⁹, corresponding to the β-sheet of the SIG (peptide 5, labeled SIG), did not affect binding of DPP9 to SUMO1. In contrast, preincubation of DPP9 with a peptide covering amino acids 61–75 of SUMO1 (SLRFLFEGQRIADNH, peptide 9) strongly reduced the binding of DPP9 to the SUMO1 beads (Fig. 2C). Next we tested whether SUMO1-peptide 9 can displace SUMO1 from the respective DPP8-SUMO1 or DPP9-SUMO1 complexes. For this, recombinant DPP8 or DPP9 were first incubated with bead immobilized SUMO1, to allow complex formation. SUMO1-peptides 5 or 9 were then added and the release of bound DPP8 or DPP9 from the SUMO beads was analyzed by SDS-PAGE (Fig. 2D). Elution was possible with SUMO1-peptide 9 but not with the control peptide that corresponds to the SIG (peptide 5).

Next, single amino acids in SUMO1 between phenylalanine 64 and histidine 75 (corresponding to SUMO1 peptide 9) were mutated to alanines. SUMO1 mutants were immobilized on wells of 96-well plates and their interaction with DPP9 was analyzed in ELISA, and compared with their interaction with wild-type SUMO1 (Fig. 2E). In this assay, SUMO1 mutations covering the sequence between phenylalanine 66 to glutamine 69 as well as a triple mutation SUMO1R63A,R70A,H75A resulted in reduced binding of DPP9. Replacement of glutamic acid 67 of SUMO1 with alanine (S1E67A) led to the most drastic decrease in the SUMO1-DPP9 interaction (Fig. 2E). Finally, we immobilized SUMO1, SUMO1E67A, and SUMO1K37AV38A on beads and tested for interaction with DPP9 and the SIM-containing proteins Uba2 and GST-PIAS3. As shown in Fig. 2F, Uba2 and PIAS3 interacted with SUMO1E67A but not with SUMO1K37AV38A. On the other hand, DPP9 interacted with the SUMO1 SIG mutant but not with SUMO1E67A. These results verify that the sequence covering Phe⁶⁶-His⁷⁵, and specifically glutamic acid in position 67 of SUMO1 is important for the DPP9-SUMO1 interaction, but not the SIG (Fig. 2F). We termed the sequence covering Phe⁶⁶-His⁷⁵ as the E67-interacting loop (EIL) to differentiate this sequence from the well characterized SIG that mediates binding to the SIM. The EIL in SUMO1 is located to the loop that connects the third and fourth β-sheets of SUMO1. The EIL is spatially separated and distinct from the SIG motif, on the “opposite” side of the SIG (Fig. 2G).

Gain of Binding by Point Mutations in SUMO2

To get more insight into the preferential interaction of DPP9 with SUMO1, we analyzed the homology between the SUMO paralogs in the newly identified EIL, and flanking regions. Sequence alignment shows that glutamic acid 67 in SUMO1 is replaced by aspartic acid in SUMO2; because both amino acids are negatively charged, we assumed that this could not explain the strong preference in binding of DPP9 to SUMO1 (Fig. 3A and supplemental Fig. 2). The alignment highlighted four amino acids that are not conserved in SUMO1 and SUMO2. We mutated single amino acids in SUMO2 to the corresponding ones in SUMO1 and tested their interaction with recombinant DPP9.

Open in a separate window

FIGURE 3.

A single point mutation in SUMO2 leads to gain of interaction with DPP9. A, alignment of SUMO1, SUMO2, and SUMO3 in the amino acid sequence corresponding to the EIL in all paralogs. B, recombinant DPP9 or Uba2 (for control) were incubated with bead-immobilized SUMO1, SUMO2, or SUMO2 mutants (S2). Bead-bound proteins were eluted and detected using specific antibodies. Input lane contains 50 ng of protein, eluted fractions correspond to 10 (DPP9) and 50% (Uba2). C, SUMO proteins were immobilized on ELISA slips and incubated with recombinant DPP9. Interacting DPP9 was detected using a DPP9-specific antibody. Experiments were repeated three times in triplicates, shown is a representative, including error bars.

Strikingly, mutation of a single amino acid in SUMO2, aspartic acid 71 to the corresponding histidine residue in SUMO1, resulted in a full gain of binding to DPP9 (Fig. 3, B and C). In contrast to DPP9, Uba2 interacted with all the SUMO2 variants to the same extent. Taken together, these results show that the negative charge in SUMO2 by aspartic acid 71 prevents interaction with DPP9.

SUMO1 Binds to an Arm-like Structure in DPP9

In parallel, we aimed to identify the surface in DPP9 that interacts with SUMO1. For this we constructed several DPP9 mutants, concentrating first on amino acids in the propeller domain of DPP9, assuming that it is important for protein-protein interactions. Mutants were expressed and purified from HEK293T cells and tested for binding to immobilized SUMO1 beads. Using this approach, we identified a cluster of hydrophobic amino acids in DPP9 that are essential for binding to SUMO1. Replacement of valine 285, isoleucine 288, or valine 290 by alanines resulted in a strong loss of interaction with SUMO1 (Fig. 4A). Strikingly, the single point mutation DPP9-V285A completely abolished the DPP9-SUMO1 association. Mutation in neighboring residues, such as in valine 287, did not affect the binding. To better understand where the SUMO1-binding surface is localized, we turned to a published homology model of DPP9, based on solved structures of other members of the DPPIV family: DPPIV, DPPX, and fibroblast activation protein α (40). A similar model was published by (41). Homology models were successfully used for analyzing and designing inhibitors with increased specificity for DPP8 and DPP9 over DPPIV (42). Crystal structures of members of the DPPIV family show that they all form homodimers, where each monomer is build of two domains, a barrel-like α/β hydrolase and an eight-blade β propeller. An extended arm-like structure projects from blade 4 of the propeller, and is located next to a side opening in DPPIV, which is formed between the hydrolase and the propeller (43–49).

Open in a separate window

FIGURE 4.

SUMO1 interacts with an ”arm“ motif in the propeller of DPP9. A, HA-tagged DPP9 mutants were expressed in HEK293T cells and purified using anti-HA beads with agarose. Purified DPP9 proteins (input) were then incubated with SUMO1-immobilized beads. Bound DPP9 was eluted and analyzed by Western blotting using anti-HA antibodies. B, alignment of the amino acid sequences corresponding to the arms of DPP9, DPP8, and DPPIV. C, a homology model of DPP9 (published by Ref. 40), shown from the ”back“ side and the ”front“ (180° rotation) side, which also reveal the so-called side opening. Color code: the hydrolase domain is dark green, the eight-bladed propeller is light green, and the extended SUMO-binding arm is highlighted in red. D, results of an ELISA. Prior to the assay SUMO1 was incubated with varying concentrations of a peptide corresponding to the arm of DPP9 (or a control peptide). The SUMO1/peptide mixture was then added to immobilized DPP9. Interacting SUMO1 was quantified using SUMO1-specific antibodies. Control wells were coated with ovalbumin instead of DPP9, remaining binding reflects background binding.

Both published DPP9 homology models predict that valine 285, isoleucine 288, and valine 290 locate to the extended arm of DPP9 (40, 41) (Fig. 4, B and C). We therefore asked whether a peptide corresponding to the extended arm of DPP9 (²⁸⁵VEVIHVPSPALEERKTDSYR³⁰⁴) would inhibit the SUMO1-DPP9 interaction. We tested this hypothesis by incubating SUMO1 with the DPP9-arm peptide prior to incubation with immobilized DPP9. As shown in Fig. 4D, the DPP9-arm peptide competed with DPP9 for binding to SUMO1 in a concentration-dependent manner. Binding of SUMO1 to DPP9 was unaffected in the presence of the same concentrations of a control peptide, or a shorter peptide corresponding to only part of the arm (Fig. 4D and data not shown). Taken together, we conclude that the arm of DPP9 mediates its interaction with SUMO1, and define this surface as the SUMO-binding arm (SUBA) of DPP9.