Highly homologous proteins exert opposite biological activities by using different interaction interfaces

We present a possible molecular basis for the opposite activity of two homologues proteins that bind similar ligands and show that this is achieved by fine-tuning of the interaction interface. The highly homologous ASPP proteins have opposite roles in regulating apoptosis: ASPP2 induces apoptosis while iASPP inhibits it. The ASPP proteins are regulated by an autoinhibitory interaction between their Ank-SH3 and Pro domains. We performed a detailed biophysical and molecular study of the Pro – Ank-SH3 interaction in iASPP and compared it to the interaction in ASPP2. We found that iASPP Pro is disordered and that the interaction sites are entirely different: iASPP Ank-SH3 binds iASPP Pro via its fourth Ank repeat and RT loop while ASPP2 Ank-SH3 binds ASPP2 Pro via its first Ank repeat and the n-src loop. It is possible that by using different moieties in the same interface, the proteins can have distinct and specific interactions resulting in differential regulation and ultimately different biological activities.

The ASPP proteins interact with different apoptosis-related proteins such as the p53 protein family, Bcl2 and NFκ B 2,7,13,[20][21][22][23][24] . iASPP Pro interacts with iASPP Ank-SH3 in cells and phosphorylation of iASPP by B1/CDK1 on S84 and S113 inhibits this interaction 25 . Phosphorylation of iASPP results in its re-localization to the nucleus and inhibition of p53 activity. Inhibition of iASPP phosphorylation in melanoma cells restored p53 function and suppressed the melanoma growth 25 . ASPP2 is regulated by an interaction between its Ank-SH3 domains and Pro domain, which regulates the intermolecular interactions of ASPP2 with its different protein partners by an autoinhibitory mechanism 18,26 . The binding sites of p53, Bcl2 and NFκ B to ASPP2 Ank-SH3 are different, while the binding sites of ASPP2 Pro to ASPP2 Ank-SH3 overlaps the binding sites to all three proteins 27 . Competition experiments showed that ASPP2 Pro competes with peptides from p53, Bcl2 and NFκ B for binding ASPP2 Ank-SH3 18,26 . In Helicobacter pylori -infected cells, H. pylori protein CagA binds ASPP2, which results in ASPP2 binding to p53 28,29 . In these cells the interaction of ASPP2 with p53 is inhibited in the presence of ASPP2 726-782, which is derived from ASPP2 Pro, possibly because of this regulatory mechanism 30 .
Intrinsically disordered proteins (IDPs) or regions (IDRs) are highly flexible and lack a defined 3D structure 31,32 . IDPs and IDRs play crucial roles in many cellular processes such as transcriptional regulation, translation, recognition and signal transduction 31,32 . IDPs and IDRs can bind their partners with high specificity but low affinity 32 and their binding to their partners is often regulated by post translational modifications such as acetylation, phosphorylation and methylation 31 . In many proteins that are regulated by auto-inhibition, the inhibitory region is highly disordered and has many phosphorylation sites 33 .
Despite the sequential and structural homology between their Ank-SH3 domains, the ASPP proteins have opposite activities in regulating apoptosis. The N-terminal domain that is unique to ASPP2 is not responsible for this difference 34,35 . To gain insight into the molecular mechanism behind this difference, we performed a detailed biophysical and molecular study of the Pro -Ank-SH3 interaction in iASPP and compared it to the interaction in ASPP2. We developed new protocols for expressing and purifying iASPP Pro. Using biophysical and computational methods we show that iASPP Pro is disordered, like ASPP2 Pro, and that the purified iASPP Pro and iASPP Ank-SH3 interact with each other in vitro. Peptide array screening revealed the exact binding sites between the iASPP domains. Our results show that the Pro-binding regions in iASPP Ank-SH3 are different than the Pro-binding regions in Figure 1. The ASPP protein family. (a) All the ASPP protein family members contain a Proline rich (Pro) domain, four ankyrin repeats (Ank) and an SH3 domain. ASPP2 and ASPP1 also contain a putative α -helical domain at their N-termini. The N-terminal part of ASPP2 has the structure of a β -Grasp ubiquitin-like fold (UBL) 19 . The iASPP fragments used in this study are iASPP Pro and iASPP Ank-SH3; (b) Backbone alignment of the crystal structures of ASPP2 Ank-SH3 (920-1121), PDB:4A63 56 (orange) and iASPP Ank-SH3 (607-828), PDB: 2VGE 7 (red). The alignment shows the structure similarity between ASPP2 Ank-SH3 and iASPP Ank-SH3. ASPP2 Ank-SH3 18 , revealing selectivity and specificity between the ASPP proteins. This sheds light on the molecular basis for the difference in activity between the ASPP proteins.

Development of a new protocol protocol for the expression and purification of iASPP
Pro. HLT-iASPP Pro was expressed in E.coli Rosetta2 (Novagen) as described in materials and methods. HLT-iASPP Pro initially showed a high tendency to aggregate. Following the screening of conditions as described in materials and methods 36 , aggregation was minimized in 50 mM phosphate buffer pH = 7, 300 mM NaCl, 10% glycerol and 0.001% Tween 20. HLT-iASPP Pro was purified in two steps, including affinity chromatography using a Nickel Sepharose column (Fig. 2a) followed by size exclusion chromatography (Fig. 2b). An imidazole gradient was used for eluting HLT-iASPP Pro from the Nickel column. HLT-iASPP Pro eluted in 100% elution buffer containing 300 mM Imidazole. Unspecific bound contaminations eluted in 10%-20% elution buffer. The 602 residues iASPP-Pro was expressed with it truncated forms. These impurities, which eluted with the full protein from the Nickel column, were separated from the full length protein by using size exclusion chromatography. The full protein eluted first from the size exclusion column and was successfully separated from its truncated forms. We did not cleave the HLT tag to avoid the aggregation of the protein. The final concentration of HLT-iASPP Pro was 15 μ M.
iASPP Pro is intrinsically disordered. To characterize the structural properties of iASPP Pro, we used a combination of computational and experimental tools. Several disorder prediction servers predicted iASPP Pro (iASPP 1-602) to be mostly disordered (Fig. 3a). The CD spectra of purified HLT-iASPP Pro (Fig. 3b) did not show any characteristic secondary structure, indicating intrinsic disorder. HLT-iASPP Pro (MW = 75.9 kDa) eluted in size exclusion chromatography experiments as a wide peak with an elution volume corresponding to a 250 kDa globular protein (Fig. 3c). This further indicates that iASPP Pro is disordered since disordered proteins elute earlier than globular proteins of the same MW, due to their extended unfolded nature. Our results imply that iASPP Pro is intrinsically disordered, like ASPP2 Pro.
iASPP Ank-SH3 binds iASPP Pro in vitro. After purifying the recombinant HLT-iASPP Pro and iASPP Ank-SH3 we tested whether the recombinant proteins interact using Nickel affinity pulldown assay. Nickel-NTA beads were incubated for one hour with HLT-iASPP Pro or with buffer or HLT alone. Then the beads were incubated with iASPP Ank-SH3 for two hours. After three washes the proteins were eluted from the Nickel-NTA beads. iASPP Ank-SH3 was retrieved by Nickel-NTA beads that were incubated with HLT-iASPP Pro but not by Nickel-NTA beads that were incubated with buffer or HLT (Fig. 4) indicating that both iASPP domains interact with each other.

Discussion
Despite the sequential and structural homology between their Ank-SH3 and Pro domains, the ASPP proteins have opposite activities in regulating apoptosis. To gain insight into the molecular mechanism behind this difference, we performed a detailed biophysical and molecular study of the Pro -Ank-SH3 interaction in iASPP and compared it to the interaction in ASPP2. In order to enable these quantitative biophysical studies, we developed for the first time a protocol for expressing and purifying a stable recombinant full-length iASPP Pro at concentrations sufficient for biophysical studies. We also made some modifications to the known protocol for producing recombinant iASPP Ank-SH3, mainly using the HLT tag 7 . Having the two recombinant proteins in hand at high purity and concentrations enabled us to show that iASPP Pro is intrinsically disordered like ASPP2 Pro 18 . Unlike other IDPs, iASPP Pro does not mediate interactions with other partner proteins 32 .

The binding interface between the iASPP domains.
Our results show that iASPP Ank-SH3 and the full iASPP Pro 1-602 interact in vitro. It was shown before that the interaction of iASPP Pro with iASPP Ank-SH3 in cells is inhibited by phosphorylation of iASPP Pro on S84 and S113 by B1/CDK1 25 . However it is not clear from the cellular studies or from our in vitro studies if the interaction between the iASPP domains is intramolecular or intermolecular 18,25 . The fact that iASPP Pro is intrinsically disordered might support the intramolecular possibility because many proteins that are regulated by autoinhibitory mechanism have intrinsically disordered regulatory domains. Many of these regulatory domains undergo alternative splicing, such as in the ASPP proteins, and have many phosphorylation sites 33,38 . In any case, the final regulatory outcome is the same regardless of whether the domain-domain interaction is intramolecular or involves dimerization. iASPP 1-478 and specifically iASPP 1-240 but not iASPP 249-482 were previously shown to bind iASPP Ank-SH3 in a pull down assay 25 . We further mapped this interaction using peptide arrays and identified the exact regions in iASPP Pro that bind iASPP Ank-SH3. The two tightest binding peptides were iASPP 60-74, which is derived from the binding region iASPP 1-240, with K d of 35 ± 2 μ M, and iASPP 540-562, which represents a previously unknown binding site for iASPP Ank-SH3 in iASPP Pro, with K d of 34 ± 2 μ M. iASPP 60-74 is located close to the iASPP phosphorylation sites, indicating possible regulation by phosphorylation. Other iASPP 1-240 derived peptides that bound iASPP Ank-SH3 in the peptide array are not derived from iASPP phosphorylation sites or the sequence between these two sites (Table 1). Peptides derived from iASPP 249-482, and specifically iASPP 308-410, also bound iASPP Ank-SH3 in the peptide array experiment (  32,38 . The affinity of SH3 domains to their proline rich ligands is also known to be weak 39,40 . Mutations in iASPP Ank-SH3 showed that iASPP Ank-SH3 N813 and Y814 and to a lesser extent T722 and L724 are important for binding iASPP 1-240 25 . Here we found that the peptide iASPP 800-814 that includes two of these residues bound iASPP Pro in the peptide array. iASPP 739-753, which is derived from the fourth ankyrin repeat, and iASPP 764-778, which is derived from the RT loop in the SH3 domain, also bound iASPP Pro.   The binding interfaces in iASPP vs. ASPP2. Comparing the Pro -Ank-SH3 interaction interfaces in iASPP and ASPP2 18 revealed distinct and specific binding sites, which are totally different from each other. iASPP Pro binds the fourth Ank repeat (iASPP 739-753), RT loop (iASPP 764-778) and C-terminal residues (iASPP 800-814) of iASPP Ank-SH3. ASPP2 Pro binds the first Ank-repeat (ASPP2 931-961) and n-src loop (ASPP2 1083-1096) of ASPP2 Ank-SH3 18 (Fig. 7). The interface in iASPP Pro that mediates the interaction with its Ank-SH3 domain is much larger than the corresponding interface in ASPP2 Pro, probably because iASPP Pro is more than twice longer than ASPP2 Pro. Our results can explain previous data reported in the literature regarding the interactions of ASPP2 and iASPP with p53. Energy assessment for p53 Core domain (p53CD) complexes with the Ank-SH3 domains of ASPP2 and iASPP 27 showed the complex of p53CD with iASPP Ank-SH3 had a higher interaction energy than the complex of p53CD with ASPP2 Ank-SH3 17 . This indicates that the binding mode is indeed different. Previous NMR experiments showed that the SH3 domains of iASPP and ASPP2 have a very similar binding interface with the Proline rich and core domains of p53 (p53 Pro + CD), which includes both the RT and n-src loops, while ASPP2 interacts with p53 Pro + CD also via the  loops between Ank repeats 2-3 and 3-4 41 . The crystal structure of ASPP2 Ank-SH3 complex with p53CD shows that the fourth Ank repeat and the RT and n-src loops of ASPP2 mediate this interaction 17 . However, the ASPP2 derived n-src loop peptide bound p53CD much tighter than peptides derived from the fourth Ank repeat and the RT loop of ASPP2 42 . Docking studies performed for the iASPP Ank-SH3 -p53CD complex showed that the interaction between them is mediated mostly by the RT loop of iASPP. Mutational studies based on these results showed that only mutations in the iASPP RT loop, but not mutations in the n-Src loop, abolished the inhibitory effect of iASPP Ank-SH3 on p53-mediated expression of apoptosis-related genes 43 . These results are in line with our previous results, showing that the n-Src loop but not the RT loop is the binding site of ASPP2 Ank-SH3 to its Pro domain, while it is the opposite for the iASPP Ank-SH3-Pro interaction. This may indicate that the n-Src loop is a general binding interface of ASPP2, while the RT loop is a general binding interface of iASPP.
The possible molecular basis for the opposite activity of iASPP and ASPP2. Previous studies performed in our lab suggested that an important parameter in the molecular basis for the different activity of the ASPP proteins is the different electrostatic potential of the surfaces of their Ank-SH3 domains. The binding interface of ASPP2 has a higher negative charge compared to its homologous surface in iASPP 27 . Examining the sequences of the ASPP proteins revealed that indeed the first Ank repeat and n-src loop of ASPP2 are more negatively charged than these areas in iASPP, while the fourth Ank repeat of iASPP is more negatively charged than this area in ASPP2. The RT loops of both ASPP proteins have the same charge, but there is a difference in the location of their negatively charged residues.
Other examples of highly homologous proteins that bind differentially to their ligands were reported. Sequence variability in the RT and n-src loops in SH3 domains can lead to binding specificity, for example the HIV-1 Nef protein interacts only with the SH3 domain of the Hck protein but not with similar src family members due to one different residue in its RT loop 44 . There are other examples of structurally homologous proteins that bind similar ligands in different ways, but the activity of the proteins is not opposite. For example the retinoid transport proteins bind and transport retinoids in different sub-cellular areas and tissues. Although they have very similar architectures they bind the retinoids in different ways 45 .
Our results suggest that fine-tuning of the interaction interface can lead to different and even opposite activities of highly homologous proteins. By using different moieties in the same interface, the proteins can have distinct and specific interactions resulting in differential regulation and ultimately different biological activities. Our results also shed light on the possible molecular basis for the difference in activity between the ASPP proteins, where iASPP is anti-apoptotic while ASPP2 is pro-apoptotic. The different binding interfaces to the regulatory Pro domains in these highly similar structures may be part of the reasons for the different biological activities. As misregulation of the ASPP proteins is responsible for many cancerous transformations, understanding the molecular basis for the opposite activity of the ASPP proteins will provide the basis for developing future anti-cancer lead compounds that inhibit iASPP or mimic the apoptotic activity of ASPP2.

Methods
Plasmids preparation. The plasmid containing iASPP 1-828 was a kind gift of Prof. Xin Lu, Ludwig Institute for Cancer Research, Oxford, UK. DNAs encoding fragments of iASPP were amplified by PCR using primers introducing an upstream EcoRI restriction site and a downstream NotIrestriction site: (1) iASPP Pro 1-602, 5´-GTGTACCGGAATTCAAGACAGCGAGGCATTCCAGAGC-3´, 5´-GTA AGAATGCGGCCGCCTACGGCTGCTCTGTGGGC-3´; (2) iASPP Ank-SH3 603-828, 5´-GTG TAC CGG AAT TCA ACA GAG CAT GGA GAT GCG CTC TG-3´, 5´-GTA AGA ATG CGG CCG CCT AGA CTT TAC TCC TTT GAG GCT TCA CCC TG-3´. PCR products were separated on 1% agarose gel and purified using GFXTm PCR DNA and gel band purification kit (Amersham). The PCR products were heated to 97 °C, cooled down to 37 °C and then cleaved by EcoRI and NotI (New England Biolabs) and purified again. The insert fragments were ligated (ligation master mix TAKARA) into the ampicillin resistant pET-Based vector pHis parallel 2HLT (HLT tag-The lipoyl domain fusion tag containing the N-terminus His tag and an optimized Tobacco Etch Virus (TEV) protease cleavage site 36 ). The ligation reaction product was transformed into competent Escherichia coli Stable3 bacteria (Invitrogen) and bacterial colonies were screened for the presence of the gene using PCR. Positive colonies were further verified by sequencing. The analyzed sequence of the vector containing domains of the iASPP gene was compared and found identical to the iASPP sequence from Swiss-Prot entry Q8WUF5 (IASPP_HUMAN).
Peptide array screening. The CelluSpots TM peptide micro-arrays were synthesized by INTAVIS Bioanalytical Instruments AG, Köln, Germany. The 15 residue peptides were acetylated at their N-termini and attached to a cellulose membrane via their C termini through an amide bond. For screening the binding of the array to HLT-iASPP Ank-SH3 the array was first washed for 4 h at room temperature with 50 mM Tris-HCl pH = 8 at an ionic strength (IS) of 150 mM adjusted by NaCl, 0.05% Tween20 and 2.5% (W/V) skimmed milk (buffer D) for blocking unspecific binding. HLT-iASPP Ank-SH3 was dissolved in 20 mM Phosphate buffer pH = 7.0 IS = 150 mM, 5 mM β ME, 10% glycerol and 0.02% NaN 3 and 2.5% skimmed milk. 5 ml of 5 μ M of the protein were incubated with the arrays at 4 °C with shaking overnight. After three washes with TBST, the array was incubated with anti His HRP conjugated antibody at room temperature for 1 hour. The antibody was dissolved in buffer D. Then the array was washed again three times with TBST. Immunodetection was performed using chemiluminiscence (ECL reagents). Peptide array screening for binding HLT-iASPP Pro was also preformed as above with change of the proteins buffer. HLT-iASPP Pro was dissolved in 50 mM Phosphate buffer pH = 7.0, 10% glycerol, 0.02% NaN3, 0.001% Tween 20, IS = 150 mM and 2.5% skimmed milk.
Peptide synthesis, labeling and purification. Peptides were synthesized on a Liberty Microwave Assisted Peptide Synthesizer (CEM) using standard Fmoc chemistry and HOBT/HBTU as coupling reagents. Trp was added at the N-termini of the peptides, when required, for measuring the peptide Scientific RepoRts | 5:11629 | DOi: 10.1038/srep11629 concentration using UV spectroscopy. The peptides were labeled with 5(6)-carboxyfluorescein at their N termini as described 54 and cleaved from the resin as described 55 . The peptides were purified on a MERCK-Hitachi HPLC using a reverse-phase C8 preparative column with a gradient of ACN/TDW. MALDI TOF mass spectrometry and analytical HPLC were used to verify the identity and purity of the peptides.
Fluorescence anisotropy-binding studies. Binding of the iASPP Pro-derived peptides to iASPP Ank-SH3 was measured by fluorescence anisotropy using a PerkinElmer Life Sciences LS-50b spectrofluorimeter equipped with a Hamilton microlab M dispenser. Titration of iASPP Ank-SH3 into iASPP Pro derived fluorescein-labeled (FL) peptides was performed at 10 °C in 20 mM Hepes buffer, pH 7.3, 43 mM NaCl (total ionic strength = 50 mM), 10% glycerol and 5 mM β -mercaptoethanol. The excitation wavelength was set at 480 nm and the emission measured at 530 nm. The labeled peptide, dissolved in 1 ml buffer to a final concentration of 100 nM, was placed in the cuvette. 250 μ M iASPP Ank-SH3 was placed in the dispenser and aliquots of 10-20 μ l were added at 1.5 min intervals. The solution was then stirred for 30 sec, and the fluorescence and anisotropy were measured. Data were analyzed using the program Origin8 (OriginLab) and were fit to 1:1 binding model: