NCSR Demokritos

Research Staff:

Metaxia Vlassi, PhD, Research Director (Researcher A’)

Nastazia Lesgidou, MSc, PhD-student

RESEARCH INTERESTS

• Protein structure
• Sequence-structure relationships
• Dynamics of protein structure
• Molecular dynamics simulations
• Conformational preferences of amino-acid repeats
• Protein-protein interactions
• Kinases
• Intrinsically disordered proteins

 

RESEARCH ACTIVITIES:

Our research activities include in silico structural studies of proteins aiming at elucidating sequence-structure-function relationships under physiological and/or pathological conditions. Emphasis is on amino-acid repeat containing proteins and their role in protein-protein interactions and on proteins mainly related to diseases towards an in-depth understanding of the atomic details of their function. The approach currently used includes techniques, such as homology modelling and molecular dynamics (MD) simulations, including specialized analyses of protein MD trajectories. 

RECENT RESEARCH HIGHLIGHTS:

Background & Summary

It is now well established that proteins are not static but rather dynamic molecules and that structural flexibility is essential to protein function. However, studying the structure and dynamics of proteins experimentally, is time consuming and in some cases even prohibitive. In silico techniques, such as homology modeling and especially MD simulations, offer a very powerful alternative tool. In the lab we have used such techniques to address various issues related to protein structure-function relationships.  For example, we have used in silico techniques to elucidate the conformation of highly disordered amino-acid repeats serving as general protein-protein interaction modules (see below), as well as to study the structure of proteins linked to diseases (e.g. kinases) aiming at understanding the atomic details of their function and/or to predict the consequences of gene alterations, including gene mutations linked to diseases:

 

 

1. A. Proteins linked to diseases:

• Tyk2 kinase:

The non-receptor tyrosine kinase 2 (Tyk2), encoded by the TYK2 gene, is involved in signal transduction in response to various cytokines as part of the STAT signaling pathway and is associated with the pathogenesis of several diseases, including several autoimmune and inflammatory diseases. A single nucleotide polymorphism of the TYK2 gene (rs34536443), resulting in a Pro to Ala substitution at amino-acid position 1104 of the resulting protein (P1104A), has been found to confer protection against autoimmune diseases. However, the atomic details of the protective mechanism remain unclear. To elucidate the structural and functional consequences of this amino-acid alteration, we employed multiple, independent all-atom solvated MD simulations (3 x 100ns each) and explored the structure and dynamics of the catalytic domains of both the wild type Tyk2 and P1104A variant, in their apo-forms. Comparative analyses of the MD trajectories including, monitoring of geometric parameters, hallmarks of the kinase activation state, as well as principal component and cross correlation analyses (Figure 1), revealed that this amino-acid change has long-range effects that restrict the dynamics of the resulting protein in favor of inactive conformations. This work is in line with the notion that reduced Tyk2 activity confers protection against autoimmune diseases and added to knowledge in support of the idea of targeting Tyk2 and its pathways as a therapeutic approach against these diseases (More details in  Lesgidou et al, Bioinformatics 2018).

Figure 1. Cross-correlation analyses of the concatenated MD trajectories of Tyk2 kinase and of its P1104A variant suggest highly correlated atomic motions and communication within the catalytic domain of wtTyk2, which are however, largely disrupted in the case of the variant. This in turn, is expected to negatively affect the enzymatic activity of this variant, in line with the protective role of the related TYK2 gene polymorphism against autoimmune diseases (details in Lesgidou et al, Bioinfomatics  2018).

• SRPK1 kinase:

Serine/arginine-rich protein-specific (SRPK1) kinase specifically phosphorylates its substrates (SR proteins) at serine residues located in regions rich in arginine-serine repeats (RS-repeats, see also below) and emerges as a potential therapeutic target in cancer.  Using homology modeling combined with all-atom solvated MD simulations, we studied SRPK1 in complex with a short RS peptide from one of its substrates (LBR; see also RS-repeats below) that complies with a sequence motif identified as docking motif for SRPK1. Our in silico results combined with literature and biochemical data (contributed by our collaborators in AUTh), showed for the first time, that even short RS domains may be constituents of docking motifs for SRPK1 and that SRPK1 uses the same, distal to the active site, acidic groove to recognize its RS substrates irrespective of their length (Figure 2); however, the substrate feeding mechanism must be length dependent (For details see Sellis et al., 2012).  This work shed light on then atomic details of substrate recognition by SRPK1 and contributed to a better understanding of its function.

 

Figure 2. MD model of SRPK1 kinase in complex with a RS-peptide from one of its substrates (human LBR, in stick-model), identified as docking motif. Important SRPK1 residues predicted to be involved in molecular recognition are labeled. The model, in combination with literature and biochemical data showed for the first time, that SRPK1 uses the same, distal to the active site, acidic docking groove (in red in left panel) to recognize its RS-substrates irrespective of the repeat length, indicating that even short RS domains may serve as docking motifs for this kinase. However, the  substrate feeding mechanism must be substrate specific (details in Sellis et al., 2012).

In a subsequent work, we investigated in silico the role of SRPK1 cysteine residues, based on the observation that SRPK1 function is regulated through the formation of several disulfide bonds.  Our analysis predicted that the experimentally observed disulfide bonds most probably involve  cysteine residues located outside the catalytic domain of the kinase (see Koutroumani et al 2017). More recently, we performed an in silico analysis of the amino-acid sequence of a less abundant SRPK1 isoform in humans, SRPK1a (identified by our collaborators in AUTh).  Our analysis supports the conservation of SRPK1a in mammals and provides information on the evolutionary history of the SRPK1 gene, thus paving the way for future research probing into the contribution of SRPK1a to SRPK1 functions (details in Vlassi et al, 2019).

• CHK2 kinase:

Checkpoint kinase 2 (CHK2) is encoded by the tumor suppressor CHEK2 gene, mutations of which are associated with increased cancer risk. Of these, large genomic rearrangements (LGRs) have been rarely reported. In a recent work, we predicted the structural consequences of two rare, novel CHEK2 LGRs (p.Glu107-Lys197del and p.Asp265-His282del) identified among Greek breast cancer patients (by our collaborators in INRaSTES, NCSR “D”). Our predictions shed light  on the structural/functional consequences of these mutations at the protein level and contributed to better understanding their role in carcinogenesis (Figure 3) (Apostolou et al,  2018).

Figure 3. Prediction of the structural/functional consequences of two rare, novel deletion mutants of the CHEK2 gene, identified among Greek breast cancer patients: The CHECK2 p.Glu107_Lys197del mutation (Top) deletes a  part (in red) of the  FHA domain of the resulting protein that normally mediates CHK2 dimerization (in green & cyan). This deletion, although it does not remove residues of the dimer interface per se, it is predicted to prevent dimerization by impairing the structural integrity of this domain (encircled). The p.Asp265_His282del mutation (Bottom) results in the removal of the regulatory helix (αC), which is vital for kinase activity. Both mutations are therefore predicted to result in loss of function, in line with increased cancer risk associated with these CHECK2 mutations (Apostolou et al,  2018)

• Akt kinase:

Akt kinases are associated with cancer and therefore understanding the details of their action is central. Activated Akt2 has been previously implicated in acting on SR proteins. However, it has been questioned whether this action is direct or it is mediated by co-existing SRPK1 activity. To address this issue we followed the same approach as for SRPK1 kinase and used 3D-modeling combined with molecular dynamics (MD) simulations to study Akt2 in complex with short, overlapping LBR RS-containing peptides complying with the minimum Akt recognition consensus sequence. Our in silico results (Figure 4) combined with biochemical data (contributed by our collaborators in AUTh) provided evidence that Akt kinases directly phosphorylate RS domains, although differentially compared to SRPK1 (details in Voukalis et al 2016). We propose that the balance and the concerted action of SRPK1 and Akt kinases constitute a fine-tuning mechanism that modulates the activity and function of SR proteins.

Figure 4. In silico modeling of the interaction of a 10-mer, RS peptide from the SRPK1 substrate, LBR (in stick model), in complex with Akt2 kinase (in ribbon model). (Right panel): Details of the interaction in the final model of the binary Akt/LBR-peptide complex, obtained by averaging six 3D-models derived from equivalent independent MD simulations. This work combined with biochemical data, provided evidence that Akt kinases also phosphorylate RS domains directly and suggests that the balance and the concerted action of SRPK1 and Akt kinases constitute a fine-tuning mechanism that modulates the activity and function of SR proteins (details in Voukalis et al 2016).

• TSC1/TSC2 complex in tuberous sclerosis:

Mutations of the TSC1 and TSC2 genes, encoding for the TSC1/TSC2 protein complex, have been associated with tuberous sclerosis. In order to predict the structural/functional consequences of two such mutations (TSC1 c.737G>A/p.Arg246Lys and TSC2 c.4942A>T/p.Ile1648Phe) identified in tuberous sclerosis patients from Greece (by our collaborator: Voutsinas’ group, IB-E, NCSR “D”), we carried out comparative in silico structural studies of their corresponding proteins, respectively. Our MD results predicted that the R246K amino-acid change is unlikely to affect the structure or interactions of the resulting TSC1 protein (Figure 5), supporting the idea in the literature, that the corresponding nucleotide substitution (TSC1 c.737G>A) may affect splicing, instead. On the contrary, based on a 3D-model of the catalytic domain of TSC2 that we constructed, we predicted that the I1648F amino-acid change most probably affects the structural integrity of the catalytic domain of TSC2 and therefore it is expected to result in loss of function of the resulting protein  (details in Avgeris et al., 2017).

 

Figure 5. Prediction of the structural consequences of the hTSC1 c.737G>A (p.Arg246Lys)  mutation identified in tuberous sclerosis patients from Greece, at the protein level.  The produced 3D-models (energy minimized average models from 3 MD simulations for each protein) of the core domain of wt hTSC1 and R246K proteins are very similar, indicating that this amino-acid change is unlikely to have an impact on the structure or interactions and therefore the function of the corresponding TSC1 protein. This observation adds to the idea in the literature, that this TSC1 nucleotide change may affect splicing, instead. (Avgeris et al., 2017),  

1. B. Conformational Preferences of amino-acid Repeats, Role in Protein Interactions 

Background & summary

Sequence repetition is very common in proteins involved in multiple protein interactions, with a wide variety of functions. Despite their abundance and importance however, questions about the conformational preferences of many classes of repeats and the structural determinants of repeat-mediated interactions, remain open. In particular, tandem repeats of polar amino-acids (such as RS-repeats or polyglutamine stretches) are predicted to be highly disordered and their conformations are difficult to determine experimentally. In addition, even hydrophobic amino-acid repeats, such as TPRs, exhibit conformational flexibility. We have employed mainly MD simulations to elucidate the conformational preferences of such repeats:

• PolyQ tracts

Tandem repeats of glutamine residues (polyQ) have been found in many proteins and have both physiological and pathological functions depending on their length.  Although it is well established that extended polyQ stretches are connected to diseases, the non-pathological role of wild-type polyQs is unclear. It has been noted however, that polyQ-containing proteins are biased toward functions related to transcriptional regulation. In the lab, we studied the role of a polyQ domain located at the N-terminal tail of the TPR-repeat containing protein Ssn6 (see also TPR-repeats below), which associates with the Tup1 protein to form a global transcriptional co-repressor complex that is conserved across species. More specifically, using first, a combination of size exclusion chromatography and circular dichroism, we found that, truncation/deletion of the polyQ domain results in non-physiological TPR-mediated self-associations of Ssn6 that may prevent its interaction with Tup1. Yeast-two hybrid experiments (contributed by our collaborator: D. Tzamarias, ITE-Crete) confirmed this prediction. Subsequently, using in silico techniques, such as ab-initio folding by replica exchange MD (REMD) simulations, we showed that this polyQ domain is structurally flexible and capable to adopt at least two alternative conformations relative to the TPR super-helix of Ssn6 (fig6_animated gif & Figure 6).

Figure 6. 3D-Modeling of the polyQ (Qx16) containing N-terminal tail of the TPR protein Ssn6. In silico ab-initio folding via REMD simulations revealed at least two alternative conformations of this polyQ tail relative to the TPR super-helix (in green ribbons), suggesting a crucial role of this domain in the formation of the Ssn6-Tup1 general transcriptional co-repressor and therefore its function. Regulation of gene expression through alternative conformations of structurally flexible polyQ tracts may not be restricted to the Ssn6-Tup1 complex but may also apply to other polyQ protein interactions involved in transcription (see details in Tartas et al, 2017).

Based on our models, we propose the following role of the polyQ tail in this particular protein interaction:  in the absence of Tup1, through transient interactions with its TPR super-helix, the N-terminal polyQ tail of Ssn6 protects its Tup1-binding site from non-physiological TPR-mediated self-associations, and/or stabilizes the Ssn6 TPR structure, which is essential for Tup1 interaction and function.  We propose that such modulation mechanism(s) of protein interactions may not be restricted to the Ssn6-Tup1 complex but may also apply to other polyQ protein interactions involved in transcription. In total, this work adds to the idea that the wild-type function of polyQ tracts is to modulate protein interactions (details in Tartas et al, 2017).

• Short tandem repeats in the inhibitory domain of the mineralocorticoid receptor: prediction of a beta-solenoid structure

The human mineralocorticoid receptor (MR) is one of the main components of the renin-angiotensin-aldosterone system (RAAS) that regulates the body exchange of water and sodium. The MR has two trans-activating ligand-independent domains and one inhibitory domain (ID), which modulates the activity of the former. Although it is known that prior folding of the MR-ID is required for binding of several transcriptional co-regulatory proteins, its detailed 3D-structure was unknown. Using in silico approaches and mainly replica exchange MD simulations we modelled the 3D-structure of five consecutive repeats identified in the human MR-ID. Based on our model (Figure 7), we propose that the repeat region of the MR-ID is compatible with a β-helical fold (β-solenoid) that serves as scaffold for multiple intra- and inter-molecular interactions (including dimerization) of the MR receptor. These interactions are most probably modulated by phosphorylation-dependent conformational changes regulated by specific kinases, thus playing an important role in the coordination and sequential interactions of various MR partners and therefore in the specificity and the (patho)physiological function of this receptor  (For details see: Vlassi et al, 2013 ).

 

Figure 7. Structure prediction of tandem short 10-aminocid repeats identified in the inhibitory domain of the mineralocorticoid receptor. Our 3D-model suggests that the repeat ensemble of the MR-ID is compatible with a β-solenoid fold serving as scaffold for multiple intra- and inter-molecular interactions of the MR receptor that are modulated by phosphorylation-dependent conformational changes regulated by specific kinases, thus playing an important role in the specificity and the (patho)physiological function of this receptor (Details in Vlassi et al, 2013).

 

• Arginine-Serine (RS) Repeats and phosphorylation:

Arginine-serine (RS) repeats are found in several proteins in metazoans with a wide variety of functions. RS-repeats mediate protein-protein interactions, many of which are regulated by SRPK1-mediated phosphorylation (see also above). Using ab initio folding through MD simulations we studied the conformation of the short RS-domain of a SRPK1 substrate (lamin-B receptor, LBR). The MD results showed that, unphosphorylated RS repeats adopt short, transient helical conformations, whereas Ser-phosphorylation induces more compact (Arg-claw-like) structures (Figure 8), irrespective of the repeat length, probably serving in recognition of basic partners of LBR (e.g. histone H3).

Figure 8: 3D-model of four tandem fully RS-repeats in their phosphorylated form as obtained by an abi-initio folding through an 200 ns MD simulation. The interaction between the guanidinium groups of almost all arginines and one phoshoserine results in the formation of an Arg-claw-like structure, exposing the remaining phosphoserines to the solvent probably facilitating recognition of positively charged protein partners (details in Sellis et al. 2012).

In addition, in the same work, as already mentioned, we showed, for the first time, that even short RS-domains may be constituents of recognition platforms for SRPK1 (see also above and Figure 2). In total, our results shed light on the conformational preferences of an important class of sequence repeats before and after their phosphorylation, as well as on aspects of their recognition by SRPK1 and support the idea that the RS repeats share a common recognition mechanism by SRPK1, irrespective of their length (Sellis et al. 2012).   In addition, in a subsequent work, as already mentioned, we found that RS-domains, at least of the LBR protein, are also directly phosphorylated by Akt kinases (see above, Figure 4 and Voukalis et al 2016). We propose that, modulation of the extent of phosphorylation of RS-domains by the concerted action of SRPK1 and Akt kinases may represent a fine-tuning mechanism for regulating gene expression.

·         Tetratricopeptide Repeats (TPR):

TPRs have been found in a wide variety of proteins in all organisms and are known to mediate multiple protein interactions. As already mentioned, an example of a TPR-mediated interaction is that of the TPR-containing protein, Ssn6 with the transcriptional repressor, Tup1 and this interaction is conserved across species. In the past, using the first crystallographic structure of a TPR protein as template, we produced a 3D-model of the TPR portion of the Tup1-binding domain of Ssn6, which combined with mutagenesis data (contributed by our  collaborators: T. Tzamarias’ group, IMBB, Crete) revealed that, in its bound form the Tup1-binding domain of Ssn6 follows the typical TPR coiled-coil configuration (Gounalaki et al 2000). Our subsequent research efforts, including a combination of circular dichroism (CD) spectroscopy, limited proteolysis mapping and in silico techniques, revealed that in the absence of Tup1, the Tup1-binding TPR-moiety of Ssn6 is partially unfolded, whereas a conformational change involving a disorder-to-helix transition and stabilization of the typical TPR coiled-coil configuration occurs upon Tup1 binding (details in Palaiomyltiou et al 2008). In total these two studies, reinforce the notion of an induced-fit mechanism for this particular protein interaction and suggest that folding-coupled-to-binding may be more common in TPR-mediated interactions than previously believed. In a more recent work, as already mentioned above, we found that the non-TPR polyQ rich N-terminal tail of Ssn6 has an important, regulatory role in this particular TPR-mediated protein interaction (Figure 6) ( details in Tartas et al 2017).

SOFTWARE  

The GROMITA-GUI   (http://bio.demokritos.gr/gromita)

In the framework of an internal program of NCSR “D” (Demoerevna), we have developed the GROMITA software (Sellis. et al 2009), which is a graphical user interface (GUI) to GROMACS-4: a suite of programs for performing and analyzing molecular simulations. The original GROMITA version, v.1, is compatible with GROMACS 4.0.x versions. An updated version, (v.3.01), of the program, compatible with GROMACS-4 versions 4.5.x., has been subsequently developed. The GROMITA versions are available upon request (license agreement required).

 

LIST OF SELECTED RELATED PUBLICATIONS 

Vlassi, M., Kyritsis, K. A., Vizirianakis, I. S., Giannakouros, T., Aivaliotis, M., Nikolakaki, E. (2019). Data on the expression of SRPK1a in mammals. Data in Brief, 25 (2019) 104210: 1-6. doi:10.1016/j.dib.2019.104210

Lesgidou,N., Eliopoulos,E., Goulielmos, G.N., Vlassi, M* (2018) Insights on the alteration of functionality of a tyrosine kinase 2 variant: a molecular dynamics study. Bioinformatics, 34 (17), pp i781–i786 doi:10.1093/bioinformatics/bty556

Apostolou, P, Fostira, F, Mollaki, V, Delimitsou, A, Vlassi, M, Pentheroudakis, G, Faliakou, E, Kollia, P, Fountzilas, G, Yannoukakos, D, Konstantopoulou, I (2018) Characterization and prevalence of two novel CHEK2 large deletions in Greek breast cancer patients. Journal of Human Genetics 63(8), 877-886 doi:10.1038/s10038-018-0466-3

Koutroumani M, Papadopoulos GE, Vlassi M, Nikolakaki E, Giannakouros T (2017) Evidence for disulfide bonds in SR Protein Kinase 1 (SRPK1) that are required for activity and nuclear localization. PLoS ONE 12(2): e0171328.  doi:10.1371/journal.pone.0171328

Tartas, A, Zarkadas, C, Palaiomylitou, M, Gounalaki, N, Tzamarias, D., Vlassi, M* (2017). Ssn6-Tup1 global transcriptional co-repressor: Role of the N-terminal glutamine-rich region of Ssn6. PLoS ONE 12(10): e0186363 doi:10.1371/journal.pone.0186363 (*Corresponding author)

Avgeris, S, Fostira, F, Vagena, A, Ninios, Y, Delimitsou, A, Vodicka, R, Vrtel, R, Youroukos, S, Stravopodis, DJ, Vlassi, M, Astrinidis, A, Yannoukakos, D, Voutsinas, GE (2017). Mutational analysis of TSC1 and TSC2 genes in tuberous sclerosis complex patients from Greece. Scientific Reports |7: 16697 doi:10.1038/s41598-017-16988-w

Voukkalis, N, Koutroumani, M, Zarkadas, C, Nikolakaki, E, Vlassi, M*, Giannakouros, T (2016). “SRPK1 and Akt protein kinases phosphorylate the RS domain of Lamin B receptor with distinct specificity: A combined biochemical and in silico approach” PLoS ONE 11(4): e0154198. doi:10.1371/journal.pone.0154198

Vlassi, M*, Brauns, K, Andrade-Navarro MA (2013) Short tandem repeats in the inhibitory domain of the mineralocoricoid receptor: prediction of a beta-solenoid structure. BMC Struct Biol 13:17  doi:10.1186/1472-6807-13-17

Sellis, D., Drosou, V., Vlachakis, D., Voukkalis, N., Giannakouros, T., Vlassi, M* (2012) Phosphorylation of the arginine/serine repeats of lamin B receptor by SRPK1-Insights from molecular dynamics simulations.  Biochim Biophys Acta -General Subjects 1820(1):44-55. doi:10.1016/j.bbagen.2011.10.010

Sellis, D., Vlachakis, D., Vlassi, M* (2009) GROMITA: A fully integrated graphical user interface to Gromacs 4.  Bioinformatics & Biology Insights 2009:3, 99-102. doi:10.4137/bbi.s3207. PMCID:PMC2808185

Palaiomylitou, M., Tartas, A, Vlachakis, D., Tzamarias, D., Vlassi, M* (2008) Investigating the structural stability of the Tup1 interaction domain of Ssn6: Evidence for a conformational change in the complex.  Proteins 1;70 (1):72-82.  doi:10.1002/prot.21489

Gounalaki, N., Tzamarias, D., Vlassi M* (2000). Identification of residues in the TPR domain of Ssn6 responsible for interaction with the Tup1 protein. FEBS Lett 473(1):37-41. doi:10.1016/S0014-5793(00)01480-0 (Note: This w