|Home | About | Journals | Submit | Contact Us | Français|
Intrinsically disordered proteins (IDPs) carry out important biological functions and offer an instructive model system for folding and binding studies. However, their structural characterization in the absence of interactors is hindered by their highly dynamic conformation. The cyclin-dependent-kinase inhibitor (Cki) Sic1 from Saccharomyces cerevisiae is a key regulator of the yeast cell cycle, which controls entrance into S phase and coordination between cell growth and proliferation. Its last 70 out of 284 residues display functional and structural homology to the inhibitory domain of mammalian p21 and p27. Sic1 has escaped systematic structural characterization until now. Here, complementary biophysical methods are applied to the study of conformational properties of pure Sic1 in solution. Based on sequence analysis, gel filtration, circular dichroism (CD), electrospray-ionization mass spectrometry (ESI-MS), and limited proteolysis, it can be concluded that the whole molecule exists in a highly disordered state and can, therefore, be classified as an IDP. However, the results of these experiments indicate, at the same time, that the protein displays some content in secondary and tertiary structure, having properties similar to those of molten globules or pre-molten globules. Proteolysis-hypersensitive sites cluster at the N-terminus and in the middle of the molecule, while the most structured region resides at the C-terminus, including part of the inhibitory domain and the casein-kinase-2 (CK2) phosphorylation target S201. The mutations S201A and S201E, which are known to affect Sic1 function, do not have significant effects on the conformational properties of the pure protein.
Many naturally occurring proteins have been shown to lack rigid three-dimensional structure in vitro, existing instead as dynamic ensembles of inter-converting conformations. Many of these proteins acquire an ordered structure only upon binding to specific intracellular partners or by functional regulation via posttranslational modifications.1-9 In isolation, these proteins exhibit a highly dynamic structure that is resembling more the denatured rather than native state of “normal” globular proteins. Intrinsically disordered proteins (IDPs) have attracted a great deal of interest since it became clear that their lack of structural specificity is of physiological importance and does not simply result from improper handling in vitro.10-12 Indeed, structural disorder characterizes a broad class of regulatory proteins, which all share the feature to bind multiple interactors.13-19 Intrinsic disorder is a common property of many protein types especially among the proteins involved in signal transduction and transcription regulation. Interest towards this feature is increasing also because structural disorder contributes to the gap between functions and structures in protein databases.
The dynamic properties of IDPs are considered to be instrumental to adaptation to different interaction surfaces and to favor rapid formation and dissociation of the complexes, as required for efficient intracellular regulatory networks.13 Thus, it seems that the flat energy landscape that governs conformational transitions of these proteins in solution is a selected feature for which their amino acidic sequences have been optimized, in order to keep the polypeptide chain in a sort of totipotent state that can then differentiate into distinct non-covalent complexes depending on the environmental conditions.20,21 Computational methods predict that IDPs are much more prevalent in domains from eukaryotic or viral proteins than in bacterial or archea proteins.22
The yeast protein Sic1 is a central regulatory protein of the yeast cell cycle, acting as inhibitor of the cyclin-dependent protein kinases by forming ternary complexes with kinases and their cognate cyclins.23 A well established role of Sic1 is to control the timing of entrance in S phase during the yeast cell cycle by inhibiting the complex Cdk1-Clb5/6, whose activity is required for the G1-to-S transition.24 The inhibition is released upon ubiquitin-dependent Sic1 degradation, which is, in turn, triggered by Sic1 phosphorylation at multiple sites in its N-terminal region.25,26 However, Sic1 is involved in several other intracellular processes such as exit from mitosis,27,28 coupling cell growth to cell cycle,27-30 and signaling within the Tor31 and Hog32 stress-response pathways. As for other regulatory IDPs, the pleiotropic phenotype associated to Sic1 could be related to highly promiscuous interactions enabled by an intrinsically disordered structure.
Sic1 is a 284-amino acid protein with theoretical molecular mass 32,223 Da and isoelectric point 7.89. Its minimal fragment required for in-vivo inhibition of the Cdk1-Clb5/6 activity has been mapped to the last 70 amino acids.33 This C-terminal domain has been shown to be structurally and functionally homologous to the inhibitory domains of p21 and p27,34 mammalian tumor-suppressor proteins and well characterized members of the IDP class.1,23,35,36 A crystallographic structure of the inhibitory domain of human p27 bound to the Cdk2-cyclin A complex has been reported.37 Although very poor sequence similarity links Sic1 to p21 and p27, molecular modeling of the Sic1 inhibitory domain could be performed based on the alignment of Sic1 predicted and p27 experimental secondary structure.34 The model docked onto the Cdk2-cyclin A complex displays a well formed interaction surface with proper contacts that should allow formation of a stable ternary complex. Such an interaction has also been shown experimentally,38 indicating evolutionary conservation of the key structural features for intermolecular recognition. On the basis of this evidence, Sic1 can be considered as a member of the IDP protein class, at least considering its C-terminal domain. However, in spite of intensive studies on Sic1 function in vivo, no structural characterization or computational models on the full-length protein have been reported.
Another factor that could play a role modulating Sic1 conformational transitions, besides interactions with other proteins, is its posttranslational modification via phosphorylation. In particular, the recently detected phosphorylation on S201 by CK2, different from the phosphorylation events in the N-terminal region, does not seem to control Sic1 intracellular levels but, rather, to affect Sic1 affinity for intracellular partners.27,28,38,39 Indeed it has been shown that a Sic1 peptide encompassing S201 binds more strongly to the mammalian Cdk2-cyclin A complex in its phosphorylated than in its non phosphorylated form.38 Mutations of the residue S201 that either impair (S201A) or mimic (S201E) phosphorylation by CK2 alter the coordination between cell growth and cell-cycle progression in yeast cultures in exponential phase.27 However, no data are available yet on the effects that this modification might have on the conformational properties of the protein.
Structural characterization of proteins in disordered conformation is technically difficult, but it is important to better understand folding transitions to ordered states. Increasing evidence indicates that residual structure characterizes globular proteins under various denaturing conditions,40 as well as various IDPs in the absence of binding partners.15 A promising approach to this technically challenging problem is offered by a multiparametric analysis, which employs a set of complementary biophysical methods sensitive to distinct structural features.41 In this study, such an approach is applied to the investigation of Sic1 structural properties and conformational transitions in vitro by the use of bioinformatics, gel filtration, CD, ESI-MS, and limited proteolysis. The results show that Sic1 is a highly disordered protein, nevertheless endowed of a partially collapsed structure. The most structured region maps at the C-terminus, partially overlapping to the inhibitory domain.
Disorder predictions for the Cki Sic1 were performed by PONDR®VLXT,42 PONDR®VSL2,43 PONDR®VL3,42,44 IUPred45 and RONN.46 PONDR® (Predictors of Natural Disordered Regions) is a neural network based on local amino acid composition, flexibility and hydropathy. It is the merger of three predictors, one trained on Variously characterized Long disordered regions and two trained on X-ray characterized Terminal disordered regions.7,44,47
PONDR®VL3 predictor is a feed forward neural network that was trained on 152 Long disordered regions characterized by Various methods.43, 48 This is one of the most accurate predictors of long disordered regions. The most recent advance in the PONDR® collection are the VSL predictors (trained on Variously-characterized, Short and Long disordered regions), based on compositional differences of short and long disordered regions. PONDR®VSL2, which uses support vector machines (SVMs), has achieved high accuracy and improved performance on short disordered regions, while maintaining high performance on long disordered regions.49
IUPred is an algorithm that evaluates intrinsic disorder based on the energy resulting from inter-residue interactions.21 This predictor is based on the observation that, while the structure of ordered proteins is stabilized by a large number of inter-residue interactions, intrinsically disordered proteins do not have sufficient inter-residue interactions.45,50 Finally, RONN was designed to predict disordered structures based on functional alignments.46
The output of these predictors is a position-dependent score for disorder propensity, which varies between 0 and 1. A threshold of 0.5 for prediction of structural disorder is generally applied.51 Cumulative distribution functions (CDF) were generated using PONDR®VLXT output scores as described by Oldfield and coworkers.3 Charge-hydropathy (CH) plots were generated as described by Uversky and coworkers.3,51 Ordered and disordered proteins plotted in the CH-space were shown to be separated by a linear boundary, with disordered proteins above the boundary and ordered proteins below.3 This boundary is described by the following relationship,7
where <H> and <R> are the mean hydropathy and the mean net charge of the given protein, respectively, whereas <H>b is the “boundary” mean hydropathy value, below which a polypeptide chain with a given <R> will be most probably unfolded. The mean hydropathy, <H>, is defined as the sum of the normalized hydrophobicities of all residues divided by the number of residues in the polypeptide. The mean net charge, <R>, is defined as the net charge at pH 7.0, divided by the total number of residues. CH-plot and CDF analyses were performed using software developed at Molecular Kinetics Inc. and accessible at http://www.PONDR.com.
The predictor for α-helix-forming Molecular Recognition Feature (α-MoRF) is based on the observation that predictions of order in otherwise highly disordered proteins correspond to protein regions that mediate interaction with other proteins or DNA. This predictor focuses on short (~20 residues) binding sites within long regions of disorder that are likely to form helical structure upon binding.6,52 It follows a stacked architecture, with a first level where PONDR®VLXT is used to identify sequences with helical propensity and then a second level to evaluate the likelihood for such sequences to constitute binding sites based on attributes of the sequence itself and the surroundings.
The amino acid composition of Sic1 and that of a set of disordered proteins was compared with a set of ordered proteins.53 The fractional difference in composition between Sic1 (or a set of disordered proteins) and a set of ordered proteins was calculated for each amino acid as (CX-Corder)/Corder, where CX is the content in a given amino acid of Sic1 (or a set of intrinsically disordered proteins for the DisProt database54) and Corder is the corresponding value in a set of ordered proteins.
Acetonitrile G CHROMASOLV for HPLC (super gradient grade), ammonium hydrogen carbonate, ammonium acetate (7.5 M solution), and 2,2,2-trifluoro ethanol (TFE) were of highest purity and were purchased from Sigma (Sigma Aldrich, St. Louis, MO, USA).
The Escherichia coli DH5α strain has been used for cloning and BL21 Rosetta (Novagen, EMD Chemicals Inc., Darmstadt, D) for protein expression. Standard recombinant DNA manipulations were performed according to Sambrook.54 The entire open reading frame of the SIC1 gene has been cloned by excision from the plasmid pIVEX2.4a-SIC155 and subsequently modified at its 5' and 3' terminus by back-to-back PCR55 to adapt its sequence to functional expression in pET21a vector and to add a C-terminal His6 tag. Thus the 855 bp BamHI-BamHI fragment from pIVEX2.4a-SIC1 was cloned into the BamHI restriction site of the pET21a vector and the DNA was inserted into E. coli DH5α cells, resulting in clone pET21[T7tag SIC1]. Site-directed mutagenesis was performed employing the TripleMaster® PCR System (Eppendorf) on the entire pET21[T7tag SIC1]. The DNA BamHI-BamHI insert was first modified to fuse in frame the C-terminus of Sic1 and the esahistidine tag. The oligonucleotides were: 5'-ACG TTG GCC ATG CTC TTG ATC CCT AGA T-3', 5'-ACG TTG GCC ACC ACC ACC ACC ACC ACC-3'. These primers result in the deletion of the downstream BamHI site and part of the pET21 multiple cloning site, while introducing a recognition site for MscI (shaded) The resulting clone is pET21[T7tag SIC1HIS6], which was further modified to delete the T7 tag sequence (coming from the pIVEX vector) fusing in frame the SIC1 sequence and the promoter. The oligonucleotides were: 5'-ACT CCG TCG ACC CCA CCA AGG TCC-3', 5'-ACG TGT CGA CGG AGT CAT ATG TAT ATC TCC TTC-3'. These primers result in the deletion of the upstream BamHI site and introduce a recognition site for SalI (shaded). The resulting clone is pET21[SIC1HIS6].
Sic1 site-directed mutagenesis on Ser201 to alanine (S201A) or glutamic acid (S201E) was performed by back-to-back PCR on the entire pET21[SIC1HIS6]. The oligonucleotides used for mutagenesis were: 5'-CCA GAT CTC AAG AAG AGG AAG ACG AGG AAG-3', 5'-GTC TTC CGC TTC TTG AGA TCT GGC GTC ATT TTT CG -3' for S201E; and 5'-CCA GAT CTC AAG CAG AGG AAG ACG AGG AAG-3', 5'-GTC TTC CTC TTC TTG AGA TCT GGC GTC ATT TTT CG -3' for S201A. The primers provide a single base mismatch (bold) to mutate the amino acid 201 and a triplet mismatch (underlined) to generate, without altering the translation code, a BglII restriction site (shaded) not present in the wild type sequence and used to generate sticky-end PCR products to be self-ligated for circularization by the T4 DNA ligase.56 The correct identity of the resulting recombinant plasmids pET21[SIC1HIS6S201A] and pET21[SIC1HIS6S201E] was checked by DNA sequencing on both strands.
Transformed E. coli cells BL21-Rosetta were cultured in 1L low-salt Luria-Bertani broth containing 100 mg/L ampicillin and 34 mg/L chloramphenicol at 37°C until OD600 ~0.5, and induced for 2 h, unless otherwise stated, by 200 μM isopropyl thio-β-D-galactoside (IPTG) at 30 °C. E. coli cells expressing recombinant Sic1-His6 were harvested by centrifugation and resuspended in 1/200 volume of lysis buffer, 50 mM Na2HPO4, pH 8.0, 10 mM imidazole, 300 mM NaCl, containing protease inhibitors cocktail for use in purification of histidine-tagged proteins (Sigma Aldrich, St. Louis, MO, USA). Cells were then either directly extracted or stored at -20 °C. Protein extraction and heat treatment were carried out at once by incubating cells suspension at 99 °C for 10 min. The crude extracts were then cooled 10 min on ice and centrifuged 10 min at 10000 rpm to separate cellular debris and insoluble proteins. Recombinant Sic1, recovered as soluble protein, was further purified from the supernatant by immobilized-metal affinity chromatography (IMAC) on Ni2+/NTA beads as described (Qia expressionist Handbook, Qiagen) and eluted in 50 mM Na2HPO4, pH, 8.0, 250 mM imidazole, 300 mM NaCl.
In order to evaluate the effects of heat treatment on the protein conformation, an alternative protocol has been developed. Cells were lysated by sonication by eight 15-s pulses on a Fisher sonicator model 300, relative output 50%, on ice. Sic1 was purified from the soluble fraction by sequential metal-affinity, desalting (PD-10, General Electric-Amersham Biosciences, Uppsala, Sweden), and cation-exchange (CM-Sephadex General Electric-Amersham Biosciences, Uppsala, Sweden) chromatography. Sic1 is eluted from the CM column by 50 mM Na2HPO4, pH 6.2, 600 mM NaCl. Such preparations were used to test the effect of heat treatment by CD and ESI-MS. The purified protein from either protocol was finally obtained in the desired buffer by either gel filtration on PD-10 columns or several cycles of dilution and ultrafiltration on Vivaspin 500 centrifuge units (cutoff 5,000 Da, Sartorius Biolab, D).
The final Sic1 preparations were of at least 95% purity, as judged by desitometric analysis of SDS-PAGE gels. Protein concentration was determined by absorbance at 280 nm on a Nanodrop spectrophotometer (Nanodrop Technologies, Inc.,Wilmington, DE, USA) using the calculated molar extinction coefficient of 8,480 M-1 cm-1. SDS-PAGE were carried out by the Laemmli protocol57 on 12% (for intact protein analysis) or 16% (for proteolysis experiments) acrylamide gels, subsequently stained by GelCode Blue (Pierce, Illinois, IL, USA). Western blots onto nitrocellulose filters (Schleicher and Schuell BioScience, Inc., Keene, NH, USA) were decorated by purified anti-Sic1 polyclonal antibodies28 or anti-His6 polyclonal antibodies (Santa Cruz Biotechnology, Inc.) at 1:1,000 dilution. An enhanced chemiluminescence system (ECL Western Blotting Substrates, Pierce, IL, USA) was used for antibodies detection.
Gel filtration was performed on an ÄKTApurifier liquid-chromatography system (General Electric-Amersham Biosciences, Uppsala, Sweden), using a prepacked, 30×1 cm SuperdexTM 75 HR column (General Electric Life Science, GE Co., USA). Chromatography was carried out at room temperature in 50 mM Na2HPO4, pH 8.0, 200 mM NaCl at a flow rate of 0.5 ml/min and monitored by absorbance at 280 nm. Calibration was performed using the following standards (2 mg/ml): horse cytochrome c (12,400 Da), horse myoglobin (17,600 Da), chicken ovalbumin (45,000 Da), and bovine serum albumin (66,400 Da) (Sigma Aldrich, St. Louis, MO, USA).
Trypsin (Promega Corporation, Madison, WI, USA) and chymotrypsin (Sigma Aldrich, St. Louis, MO, USA) stock solutions were prepared by dissolving the commercial powders in 1 mM HCl at a final concentration of 1 mg/ml and stored at -80 °C. Reactions were carried out at room temperature in 50 mM ammonium acetate, pH 6.5, 100 mM NaCl. Aliquots were removed over a time course and the reaction was stopped by the addition of SDS-PAGE loading buffer and immediate boiling for 3 min.
Bands of interest were excised from the polyacrylamide gels, cut into small pieces and destained by repeated washing cycles alternating 50 mM ammonium hydrogen carbonate and pure acetonitrile. After complete destaining, gel particles were dehydrated by acetonitrile, covered with a trypsin solution (12.5 ng/ml in 50 mM ammonium hydrogen carbonate, pH 8.0) and incubated 1 h on ice. The excess of liquid was removed, gel pieces were covered with a solution of 50 mM ammonium hydrogen carbonate, pH 8.0 and incubated overnight at 37 °C. Tryptic peptides were extracted from gel particles by repeated cycles of alternating incubations in pure acetonitrile and 1% formic acid. Samples were lyophilised, resuspended in 1% formic acid, and desalted by ZipTip (Millipore, Carrigtwohill, Ireland) before ESI-MS analysis.
CD spectra were recorded on a spectropolarimeter J-815 (JASCO corporation, Japan), equipped with a Peltier thermoregulation system, in a 1-mm pathlength cuvette. Reported spectra are acquired with data pitch 0.5 nm, averaged over three acquisitions, and smoothed using the Savitzky-Golay algorithm.58 Deconvolution of CD spectra were performed by the CDPro software package58 using the reference protein set SDP448, containing both native and denatured proteins. The reported percent values of secondary-structure content are averaged over the results of three different algorithms: SELCON3, CDSSTR and CONTIN.59 Spectra at variable temperature were acquired with the same parameters as above, and temperature steps of 20 °C and scanning velocity 20 nm/min. Measurements at fixed wavelength were performed with data pitch 0.2 °C and temperature slope 1 °C/min. The buffer solutions without the protein were used as blank.
ESI-MS experiments were performed on a hybrid Quadrupole-Time-of-Flight (q-TOF) mass spectrometer (QSTAR ELITE, Applied Biosystems, Foster City, CA, USA) equipped with a nano-electrospray ionisation sample source. Metal-coated borosilicate capillaries (Proxeon, Denmark) with medium-length emitter tip of 1-μm internal diameter. The instrument was calibrated with a standard solution containing cesium iodide (Cs+ 132.9 Da) and the synthetic peptide ALILTLVS (MH+ 829.5 Da, Applied Biosystems, Foster City, CA, USA). Peptide identification was performed using the MASCOT software with following parameters: 2 missed cleavages, peptide tolerance 0.6 Da, MS/MS tolerance 0.4 Da, peptide charges 2+ and 3+. Only monoisotopic masses were considered as precursor ions.
Spectra of intact Sic1 were acquired in the 500-2,000 m/z range, with accumulation time 1 sec, ion-spray voltage 1,200-1,400 V, declustering potential 60-80 V, and instrument interface at room temperature. Spectra were averaged over a time period of 2 min. Spectra of Sic1 tryptic peptides were acquired in the 400-1,500 m/z range, with accumulation time 1.5 s, ion-spray voltage 1,200 V, declustering potential 80 V, with active ,,information dependent acquisition“ (IDA), using rolling collision energy to fragment peptides for MS/MS analysis. Liquid chromatography coupled to MS analysis was performed on a nano-HPLC instrument (Tempo, Applied Biosystems, Foster City, CA, USA) using a VYDAC MS C18 column, 0.075×150mm, 5 μM particle size, 300 Å pore size (Alltech, Springfield, KY, USA) at a flow rate of 250 nL/min, equilibrated in 0.1% formic acid, and developed by a linear gradient 0-60% acetonitrile in 1% formic acid.
Figures 1 and and22 summarize Sic1 structural features predicted by sequence analysis. Figure 1 reports predicted conformational properties of this protein. The consensus secondary-structure prediction derived from several different algorithms hints to the existence of large portions in random-coil conformation, particularly within the N-terminal half of the molecule (Jnet line). The prediction also includes extended segments with a predominance of α-helices and with a higher content in regular secondary structure within the C-terminal moiety. Prediction of solvent accessibility60 indicates that very few positions have strong burial propensity, and also these are clustered in the C-terminal half of the molecule (SUBacc line). The GLOBE algorithm42 predicts a non-globular structure for the entire protein (data not shown), based on a large difference between the number of predicted exposed residues (226) and the number of expected exposed residues (120).
Disorder prediction by several different algorithms indicates a high degree of intrinsic disorder in Sic1 (Figure 2A). The disorder profiles for all the predictors employed, as well as the curve corresponding to the consensus score distribution, are mostly located above the generally applied threshold of 0.5,51 clearly showing that Sic1 is a highly disordered protein, with particularly high disorder propensity in the N-terminal region. The output of the PONDR®VLXT predictor has been used to generate a CDF curve (Figure 2B). CDF analysis summarizes the per-residue disorder predictions by plotting PONDR scores against their cumulative frequency within the protein sequence. The ordinate of each point of the graph gives the proportion of residues with a PONDR score less or equal to the abscissa. This analysis is based on a binary disorder classifier which predicts whole proteins as either mostly disordered or mostly ordered, based on whether a CDF curve is above or below a reference boundary curve.3 Figure 2B shows that the Sic1 curve lies below the boundary, confirming that Sic1 is a mostly disordered protein.
This conclusion is consistent with the results of the Sic analysis by another binary disorder classifier, the CH signature. Simultaneous observation of low mean hydropathy (<H>) and high mean net charge (<R>) is typical of IDPs.3 Analysis of the Sic1 amino acid sequence shows that this protein is characterized by <H> = 0.3834, <R> = 0.0035, and <H>b = 0.4145, thus fulfilling the requirement for the natively unfolded proteins, namely that <H> < <H>b (0.3834 < 0.4145).51 Thus, evaluation of conformational properties by disorder predictors, CDF analysis and CH-plots suggests a mostly disordered structure for the whole Sic1 molecule. This is generally taken as evidence that the protein belongs to the class of native coils or native pre-molten globules.7 However, the spot corresponding to Sic1 lies in the close proximity of the boundary in the CH-plot (not shown), indicating that this protein might have partially collapsed structure.
Figure 2C compares the results of the Sic1 compositional profiling 52 with that of the intrinsically disordered proteins from the DisProt database.53 Residues that are less abundant in Sic1 (or in a set of disorder proteins) than in typical ordered proteins are shown by negative bars (CX<Corder) and those that are enriched are shown by positive bars (CX>Corder). Compositional profiling of Sic1 reveals that in comparison with typical ordered proteins, Sic1 is noticeably depleted in order-promoting residues, including W, Y, C, I, L, V, M, and H, and is highly enriched in disorderpromoting residues, such as T, R, Q, S, N, P, E, and K (Figure 2C).
The Sic1 sequence has also been analyzed by α-MoRF, a predictive tool specifically developed to reveal short interaction-prone segments in disordered proteins that could undergo a coil-to-helix transition induced by intermolecular contacts.34 This algorithm identifies potential binding sites in three putative α-helical intermolecular recognition features: residues 75-93, 142-160, and 240-258 (Figure 1). Altogether, these results suggest that the whole Sic1 protein is mainly disordered, with an intrinsic propensity for ordered structure (Figure 1).41,61
The recombinant protein Sic1-His6 was expressed in BL21-Rosetta E. coli cells. Maximal accumulation of the intact product is seen after 2-h induction by IPTG at 30 °C. The protein is purified essentially in two steps involving heat treatment and IMAC (Figure 3). It was reasoned that an intrinsically disordered protein might be enriched in the soluble fraction of heat-treated crude extracts since no global unfolding and, therefore, no big changes in aggregation propensity are expected to take place in response to the treatment. Indeed, preparation of crude extracts by incubation at 99 °C for 10 min, followed by removal of insoluble material, results in substantial purification of Sic1, as shown by the gel in Figure 3. The effects of heat on the protein conformation are fully reversible, as indicated by CD and ESI-MS analysis on Sic1 purified by an alternative protocol that does not involve heat treatment (see Materials and methods). The spectra of such protein preparations before and after heat treatment are identical (data not shown).
As typical of IDPs, Sic1 turns out to be prone to proteolytic attack (Figure 3). Its partial degradation is not completely prevented even by the presence of protease inhibitors throughout the purification procedure. Degradation is enhanced by increasing duration or temperature of IPTG induction, as well as by sample aging or scaling up of the purification procedure. The protein also tends to aggregate in pure water (data not shown). Another feature that Sic1 shares with other IDPs is an anomalous migration on SDS-PAGE. Indeed, the apparent molecular weight derived from SDS-PAGE is ~37 kDa, higher than the isotopically averaged value calculated for Sic1-His6 based on the amino acid sequence (33,046 Da) and measured by mass spectrometry (33,050 +/- 2 Da). Such a behavior could result from poor interaction with SDS due to the anomalous amino acid composition of IDPs,3,44 in particular a lower content in hydrophobic residues and a higher content of charged residues than average.62-64
In order to get a first experimental indication about Sic1 conformational properties, the hydrodynamic behavior of the purified protein has been investigated by gel-filtration chromatography in 50 mM sodium phosphate, pH 8.0, 200 mM NaCl. Protein standards with known globular structure were used for calibration in the molecular-weight range 12.4-66.4 kDa and, thus, bracketing Sic1 molecular weight. As can be seen in Figure 4, the expected logarithmic relationship between mass and elution time is observed for the standards. Under these conditions, Sic1 elutes as a single symmetric peak between ovalbumin and bovin serum albumin. Its apparent molecular weight is ~50 kDa, ~50% higher than the real one. Thus, Sic1 in solution has a much larger hydrodynamic radius than expected for a protein of its size in a compact, globular conformation, indicating that the protein structure is, at least partially, disordered. In particular, these results indicate an increase of ~1.5-fold in the hydrodynamic volume of the protein, relative to its hypothetical compact, globular structure. Such a degree of compactness has been ascribed to proteins in their molten-globule state.65
Sic1 conformational properties have been further investigated by far-UV CD and ESI-MS analysis, which offer complementary information on protein secondary and tertiary structure, respectively. These techniques have different buffer requirements, since volatility is needed for ESI-MS and high transparency for CD measurements down to 190 nm. Thus same ionic strength and pH have been maintained for both kinds of experiment by using 50 mM ammonium acetate, pH 6.5 as MS-ESI buffer and 50 mM sodium phosphate, pH 6.5 as CD buffer.
The far-UV CD spectrum of Sic1 (Figure 5) is characterized by a shoulder at 222 nm and an intense negative minimum at ~202 nm. These features indicate that Sic1 in solution is mostly disordered but contains some helical structure (~9%), in agreement with the predictions reported in Figure 1. No significant spectral changes are observed varying NaCl concentration between 0 and 300 mM (Figure 5, upper inset). Also mutations in position 201 do not have significant effects on the secondary structure content, as indicated by the CD spectra of the mutants S201A and S201E compared to the wild-type protein (Figure 5, lower inset). Thus, the effects of these mutations that have been described in vivo and in vitro can not be simply interpreted in terms of different conformational properties of the protein itself. CD measurements down to 200 nm in 50 mM ammonium acetate, pH 6.5 (ESI buffer) gave very similar results to those reported in Figure 5 (data not shown).
Nano-electrospray is a mild ionization technique, which is known to preserve non-covalent interactions responsible for protein conformation and supramolecular complexes.66-70 In addition, charge-state distributions (CSDs) obtained by ESI-MS have been shown to be sensitive to the compactness of protein tertiary structure71 while quite insensitive to secondary-structure content.72,73 Thus, ESI-MS offers a global probe for protein conformational analysis which is nicely complementary to CD data, and has also been already applied to conformational analysis of IDPs.74 The ESI-MS spectrum of intact Sic1 in 50 mM ammonium acetate, pH 6.5, shows a single, broad CSD centered on the 28+ ion (Figure 6A). The CSD of the protein under denaturing conditions such as 1% formic acid (Figure 6B) is bimodal (maxima at 26+ and 39+), indicating coexistence of two distinct conformational states. The further addition of 50% acetonitrile reduces the component centered on the 26+ ion and results in a minor shift of the highly charged component to 43+ (Figure 6C). We assign the latter to the fully unfolded protein and the component centered on the 26+/28+ ion to a partially folded form of the protein. Indeed, a completely folded globular structure for a protein of this size is predicted to result in an average charge state of 14+.75 Thus, Sic1 can be brought to a more extended conformation by the addition of denaturing agents, indicating that the protein in ammonium acetate contains some degree of tertiary structure. Together with the CD data reported above, these results suggest that pure Sic1 in solution populates a molten-globule-like state.
Disordered regions of IDPs are characterized by an intrinsic propensity to form ordered secondary structure, which can be enhanced by the addition of TFE, a compound known to stabilize protein secondary structure by promoting the burial of polypeptide backbone groups.76 Conformational transitions induced by TFE in other IDPs take place at relatively low concentrations of the co-solvent (10-30%).4,65 Figure 7 illustrates the effect of increasing TFE concentrations on the far-UV CD spectrum of Sic1 wild-type and S201A, S201E mutants. The results reveal a pre-transition baseline in the TFE effect up to a concentration of ~20% and a highly cooperative coil-to-helix transition above 20% TFE, characterized by an isodichroic point around 200 nm. The transition slowly tends to a plateau above 50% TFE. The spectra at 50% TFE and above are typical of a highly helical structure, with well formed negative minima at 222 and 208 nm and a positive maximum at 190 nm. Deconvolution results give ~40% helical structure for the protein at 65% TFE. The protein tends to precipitate at TFE concentrations above 65%. As shown in the figure inset, the TFE response of the S201A and S201E mutants is very similar to that of the wild-type protein.
Mesophilic IDPs also have a peculiar response to heat, frequently showing folding induced by heating above room temperature.77,78 Figure 8 shows the heat dependence of Sic1 far-UV CD, indicating a progressive increase in helix content going from 20 to 80 °C. The spectra display an isodichroic point indicative of a two-state transition. The effect is modest and the spectrum reverts to the initial features after re-cooling at room temperature (inset Figure 8).
The action of proteases is affected by the accessibility and flexibility of the substrate polypeptide chain, making disordered regions an easy target for proteolytic attack.79-83 Indeed, it has been established that sites most prone to proteolysis in folded proteins are typically found at flexible and exposed loops83,84, and are typically absent in ordered regions rich in regular secondary structure.41,85-88 That is why, at low enough enzyme-to-substrate ratios, it is possible to discriminate among putative cleavage sites based on their susceptibility to proteolysis, gaining insights on the structure and dynamics of distinct regions of the protein structure. Limited proteolysis has been used for identification and characterization of IDPs.78
This technique has been employed here to map hypersensitive sites and resistant fragments of Sic1, which would correspond, respectively, to the most disordered and the most structured regions of the molecule in solution. It is generally advisable to compare results obtained by proteases of different specificity, in order to discern structure effects from sequence effects.65 Thus, even the spontaneous degradation taking place during the purification procedure can add useful information to deliberate proteolysis experiments carried out by the addition of specific proteases. Relevant to this regard is that spontaneous degradation gives rise to a very reproducible pattern of bands.
Five major bands (a-e) tend to accumulate below the band of intact Sic1, as detectable on Coomassie-stained gels (Figure 9). These fragments were first characterized by western blot. All of them are recognized by anti-Sic1 polyclonal antibodies, supporting their identification as Sic1 degradation fragments (data not shown). The three larger ones are also recognized by anti-His6 antibodies (Figure 9) and, thus, contain an intact C-terminus. This result indicates that bands a-c are produced by progressive chopping from the N-terminus providing a first indication of particular instability of the N-terminal region of this protein. Bands d and e were in-gel digested by trypsin, and the resulting peptides were analysed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). The Sic1 peptides that could be identified in each band are shown in Figure 9. Band d contains fragments from both the N- and C-terminal moieties and, therefore, seems to derive from cleavage events approximately in the middle of the molecule combined to N-and C-terminal trimming. Band e contains fragments from the C-terminal region 178-260 (178-185, 194-233 and 236-260). Since the Sic1 sequence following residue 260 is extremely rich in basic residues, it is expectable that tryptic digestion products from this region are lost. Therefore, it is likely that band e extends beyond fragment 178-260 on the C-terminal side. On the other hand, the Sic1 sequence preceding residue 178 (PGTPS) corresponds to a peak in the predicted disorder profile (Figure 2A) and it is expected to be a good proteolysis target. Thus, the smallest fragment out of the five major degradation products seems to correspond approximately to the C-terminal half of the molecule, trimmed at both ends.
Figure 10 shows the time course of Sic1 degradation under conditions optimized to monitor proteolysis by trypsin (Panel A) and chymotrypsin (Panel B). Consistent with a mostly disordered protein structure, it was necessary to use quite low enzyme-to-substrate ratios (1:2,000-1:5,000), in order to detect intermediates of the degradation process. The starting samples contain already some degradation products, reflecting some unavoidable variability among protein preparations and difficult control of spontaneous degradation through scaled-up protein purification. The degradation patterns obtained with the two proteases are similar to each other and to those produced by spontaneous degradation in crude extracts. It is possible to detect accumulation of early fragments around 30 kDa, intermediate fragments around 16 kDa, and late fragments around 6 kDa. The latter persist after prolonged incubation times (> 20 min), when all the fragments at higher molecular weight have already disappeared. These results indicate that there are major differences in the susceptibility of cleavage sites within the Sic1 molecule and that some regions of the protein are more protected than others. The similar patterns observed with proteases of different specificity suggest that protection is caused by local protein conformation and not directly by sequence features.
The late fragments (bands f from trypsin and band g from chymotrypsin digestion) have been processed by tryptic in-gel digestion and the resulting peptides were analyzed by MS/MS. The results are reported in Figure 10. All the peptides identified from these bands derive from the C-terminal region of Sic1. These peptides span the region 178-233 in the case of trypsin and 198-233 in the case of chymotrypsin, and still contain numerous potential target sites for both proteases. Altogether the results from spontaneous and deliberate digestion indicate that proteolysishypersensitive sites cluster at the N-terminus and in the middle of the molecule, consistent with the disorder prediction reported above. The C-terminal region, at least from position 198 to 233, seems to be the most structured part of the Sic1 molecule in the absence of interactors.
This work reports the first structural characterization of full-length Sic1, a central regulator of the yeast cell cycle. The first conclusion of this work is that Sic1 is an intrinsically disordered protein. Such a conclusion emerges from a consensus of bioinformatic predictions and experimental data, supporting its identification as a new member of the IDP class. Our results are in agreement with a recent NMR report on the isolated N-terminal 90 residues of the protein, which was published during revision of the present paper.89 In that work, chemical shift dispersion and nuclear Overhauser effect indicate a highly disordered structure of the Sic1 N-terminal fragment.
The second important conclusion of this work is that the disordered state of Sic1 in the absence of interactors is characterized by the presence of some intrinsic tertiary and secondary structure that makes it clearly different from a completely unfolded protein. The protein displays global compactness intermediate between that of a fully unfolded and a folded globular protein and can be ”denatured” by acids and organic solvents to a fully unfolded state. Thus, the properties of Sic1 in solution are reminiscent of those of molten globules or pre-molten globules.76,90 The isolated N-terminal fragment, too, displays some structural compactness and transient secondary and tertiary structure.89
As observed for many other IDPs, Sic1 undergoes coil-to-helix transitions induced by TFE and heat. The pronounced response of IDPs to TFE addition seems to reflect their potential to form ordered secondary structure,1 while the response to heat could be due to predominance of electrostatic interactions in the folding transitions of these proteins with peculiar charge-hydrophobicity features.91 More structural investigation will be needed to relate these transitions to the physiologically relevant conformational changes of Sic1.
This work focuses further on the localization of intrinsically structured regions within the Sic1 molecule. The the most structured moiety of the Sic1 molecule is located near the C-terminus, while the N-terminus turns out to be the most disordered region. Structured regions or subdomains have been characterized in other IDPs, such as stathmin/OP18,92 GCN4,23,36,93 and the Sic1-related p21 and p27.1,94 The intrinsic structure of IDPs could play a role counteracting protein degradation in vivo and offering locally structured regions that may serve as nuclei for the formation of interaction surfaces. This has been shown to be the case in the sequential mechanism of cyclin/kinase complex recognition by the inhibitory domain of p27.36,95 Consistently, the Sic1 proteolysis-resistant fragment identified in this work (178-233) overlaps with its inhibitory domain (215-284).
Thus, although Sic1, p21 and p27 are related disordered proteins, the domain that they have in common is the least disordered part of the protein. Also in p21 and p27, the inhibitory domain is the most structured region.96 However, no homology outside such domain links these proteins to Sic1. This fact could reflect relatively constant kinds of interactions involving the inhibitory domain, which might be “specialized” for binding at kinase active sites and to cognate cyclins,97 rather than adaptable to a variety of different interactors. Furthermore, the target kinases and cyclins, too, are evolutionary conserved, thus imposing similar selective pressure on the inhibitory domains.
The proteolysis-resistant C-terminal fragment of Sic1 also includes residue S201, the CK2 phosphorylation target. Sic1 phosphorylation at this site has been shown to affect the G1-to-S transition and the coordination between cell size and cell cycle in vivo, as well as to increase Sic1 affinity for the cyclin-kinase complex in vitro. More detailed structural information is needed in order to interpret the effect of phosphorylation at this site on the interaction with the cyclin-kinase complex. However, the results on the mutant proteins S201E and S201A reported here do not reveal appreciable effects of these substitutions on Sic1 conformation in the unbound state, as observed instead in other instances, like p2798 and a calmodulin-derived peptide.1,99
The remarkable disorder of the Sic1 N-terminal region is likely related to the high density of phosphorylation sites, as suggested by the correlation observed for several other regulatory proteins.100 Multiple Sic1 phosphorylation within this intrinsically disordered region mediates dynamic complex formation with the ubiquitin ligase Cdc4.89,101 The highly disordered regions of Sic1 could also mediate promiscuous interactions. Continuously growing evidence of multiple intracellular functions of Sic1 actually hints to the possibility that its molecular partners might be a large number, as also indicated by the recent in-vivo mapping of the yeast interactome.102
It should also be underscored that some degree of disorder can persist in IDPs even within supramolecular complexes.102 Such “fuzziness” in protein-protein interactions could be beneficial, conferring adaptability, versatility and reversibility to the binding reaction.51 Thus, it is also possible that Sic1 regions with highest disorder propensity remain locally unstructured in the bound state, thus increasing the entropy of the bound state relative to a more typical ordered complex and contributing favorably to binding.89,103 This work offers a contribution to the characterization of Sic1 in its unbound state, which is needed to interpret transitions in the presence of interactors.
By describing the structural properties and the conformational transitions of Sic1 in the absence of interactors, this work sets the bases for the interpretation of molecular-recognition properties of this IDP. Our findings highlight structural features shared by mammalian and yeast Ckis. A conservation of order/disorder features emerges, although with inverted module sequences, beyond the relationship inferable by sequence similarity. This structural conservation likely reflects conserved interaction networks, which might be revealed by future work on Ckis complexes with intracellular partners.
We thank Paola Coccetti for providing the pIVEX[sic1] plasmid, and Vera Codazzi for technical assistance. We thank Luigi Casella for helpful discussions and critical comments on the manuscript. This work was supported by grants MIUR-Italbionet to L.A., Miur Cofin 2006 to MV, 12-1-181 and 12-1-209 from the University Milano-Bicocca (“Fondo Ateneo per la Ricerca”) to R.G., R01 LM007688-01A1 to V.N.U. and GM071714-01A2 from the National Institutes of Health and by the Program of the Russian Academy of Sciences for the "Molecular and cellular biology" to V.N.U. We gratefully acknowledge the support of the IUPUI Signature Centers Initiative. M.Š. is recipient of a postdoctoral fellowship “Assegno di Ricerca” of the University Milano-Bicocca.