|Home | About | Journals | Submit | Contact Us | Français|
Here we discuss studies of the structure, folding, oligomerization and amyloid fibril formation of several proline mutants of human stefin B, which is a protein inhibitor of lysosomal cysteine cathepsins and a member of the cystatin family. The structurally important prolines in stefin B are responsible for the slow folding phases and facilitate domain swapping (Pro 74) and loop swapping (Pro 79). Moreover, our findings are compared to β2-microglobulin, a protein involved in dialysis-related amyloidosis. The assessment of the contribution of proline residues to the process of amyloid fibril formation may shed new light on the critical molecular events involved in conformational disorders.
Proline residues play a prominent role in protein folding [1,2], protein mis-folding, and aggregation . They are key to attaining the functional state of proteins . Prolines also play a role in domain-swapping [5,6] and in protein aggregation to amyloid fibrils [7,8,9]. Peptidyl-prolyl cis/trans isomerases are enzymes that catalyze the cis/trans isomerization of peptide bonds preceding prolines (Figure 1). The cis/trans isomerization of the peptide bond acts as a molecular switch controlling several physiologically important processes, such as opening of the pore of a neurotransmitter-gated ion channel  or the formation of α-synuclein inclusions  in Parkinson’s disease.
Amyloid fibril formation is a generic property among most proteins [12,13]. The cystatins, which are protein inhibitors of lysosomal cysteine cathepsins, are a model for studies of amyloid fibril formation. The cystatin family consists of three types of inhibitory proteins, namely, stefins (type-1), cystatins (type-2) and kininogens (type-3). Stefins are intracellular proteins present in the cytosol [14,15], including stefins A and B in humans  and stefins, A, B and C in bovidae [17,18]. Human stefin B [19,20,21,22,23], chimeric stefins  and cystatin C [25,26] have been used as suitable model proteins to study protein folding and amyloid fibril formation. Human stefin B is a small globular protein consisting of 98 amino acids with no disulfide bonds; its native sequence possesses a free Cys residue at position 3. To avoid intermolecular disulfide bridge formation, this Cys is changed into Ser for all in vitro studies (hereafter referred to wild-type protein, wt). This cytoplasmic protein is supposed to act primarily as a cysteine protease inhibitor , scavenging and inhibiting accidentally released cathepsins from the lysosome. In addition, stefin B also resides in the nucleus  where a number of alternative functions have been proposed. Stefin B (also termed cystatin B) gene mutations, either dodecamer repeats resulting in lower protein production or missense mutations leading to misfolding, cause a progressive myoclonus epilepsy of type 1 (EPM1) with slow signs of neurodegeneration [28,29]. Similarly to cystatin C, stefin B protects neurons from excessive oxidative stress [30,31] and protein misfolding . Alternative functions, such as amateur chaperone function, have also been suggested from both experimental data and bioinformatic analysis . A breakthrough in the understanding of the structure of cystatins and their mechanism of interaction with papain-like cysteine proteases, including lysosomal cathepsins, was provided by the three-dimensional (3D) structures of chicken cystatin monomer [34,35] and human stefin B-papain complex .
Our in vitro studies of stefin B folding revealed several slow phases [37,38], which were accompanied by dimerization of the protein. We were able to determine the crystal structure of a stefin B tetramer, which is composed of two domain-swapped dimers . Of note, in stefin B, the proline residue at position 74 in the tetramer is in a cis conformation . These structures were crucial for further development in the study of proteolysis and its inhibition, and represent the basis for understanding the mechanism of amyloid-fibril formation through 3D-domain swapping.
The cis-to-trans proline isomerization  is a slow process, dependent on pH. Stefin B has in total five proline residues at positions 6, 11, 36, 74 and 79. We have examined in more detail prolines at positions 74 and 79, and both have proved to be structurally relevant. When Pro 79 was mutated into a Ser in a stefin B-Y31 variant (with Y at site 31), the protein oligomerized predominantly as a tetramer which could be crystallized . If Pro 74 was mutated into a Ser in the same variant, it underwent a transition to an oligomeric molten globule state [19,39]. We also studied the stefin B-Y31 P36G mutant, which rendered the protein less stable . Historically, while the stefin B-Y31 variant  was observed and characterized first, the E31 variant is now referred to as wild-type since it is the most abundant.
To put our work in a wider context, we describe another protein where prolines dictate folding and amyloid fibril formation, β2-microglobulin (β2m). Interestingly, β2m shares with stefin B the same number of prolines, at positions 5, 14, 32, 72 and 90, and we compare the two systems in our Discussion and Conclusions sections.
In the early folding studies using a stefin B-Y31 variant, we observed that the protein, in contrast to stefin A, undergoes slow folding phases that are a repetition of the fast folding phases . The amplitude of the slow phases is about 25%–30%. This can be explained by the existence of a population of molecules in the denatured state with either one or two non-native (cis) proline isomers that undergo similar, albeit slow, folding as the fast folding molecules with native (trans) proline configuration in the denatured state. However, the final oligomeric state of the slow phase proves to be dimeric, thus the dimer can stabilize a structurally important proline in a trans conformation. If a proline were cis, one would expect a higher amplitude of the slow phases, amounting to 70%, not only 30%. When we used size exclusion chromatography (SEC) to study the Y31 variant and its P36G mutant, we showed that 70% and 75% of molecules were monomers, respectively and the rest were dimers, whereas for the P79S mutant of the same variant, 100% were dimeric. This points to Pro 79 trans to cis isomerization as the very likely cause for the slow phase of folding towards a dimer. Nevertheless, in the tetrameric structure of the stefin B-Y31 P79S mutant, two domain-swapped dimers form the tetramer in which the Pro 74 was found in cis conformation .
As we observed that the wt stefin B (as defined in the Introduction) is more stable and less prone to forming a molten globule, we studied the role of all five single-point proline mutants of stefin B. Using multiple sequence alignment of several stefins, we identified common amino acid substitution of the prolines in human stefin B (Figure 2). The five prolines at positions 6, 11, 36, 74 and 79 were mutated to leucine, serine, aspartic acid, serine and serine, respectively. All mutant proteins were produced in an Escherichia coli (E. coli) expression system (Supplementary Figure S1) and were shown to retain their inhibitory activity (Supplemenatry Figure S2).
The stability and exposure of hydrophobic patches were confirmed by measuring ANS (1-anilinonaphthalene-8-sulfonic acid) fluorescence spectra (Figure 3A). Together with far UV-CD (ultraviolet circular dichroism) spectra (Figure 3B) we observed that P6L, P11S, P36D and P79S have hydrophobic exposure and secondary structure similar to the wild-type (wt) protein. The highest ANS binding is observed for the P74S mutant indicating a molten globule-like state. However, this was not observed consistently for the P74S mutant of the wt  by measuring CD spectra. It may well be that the molten globule intermediate forms only under destabilizing conditions or upon freeze-thawing cycles.
SEC data on the wt and its proline mutants show (Supplementary Figure S3) that dimers are the main oligomeric form of the wt stefin B and P36D mutant—when frozen and unfrozen once. The amount of the higher oligomers increases upon freeze-thawing. Then come the tetramers and higher oligomers. An estimate from the surface area of the peaks indicates 65% dimers, 20% tetramers, 5% monomers and 10% of higher oligomers. A similar distribution of the oligomers was obtained for P6L and P11S of the wt stefin B. In the case of P79S, the tetramer peak amounted to a higher percentage of around 30% tetramers, 55% dimers and 15% of higher molecular weight species. The tendency to form oligomers is high for the P74S mutant of the wt (>50%), in accordance with its tendency to transform into an oligomeric molten globule as observed previously for the Y31 variant . The P79S mutant of the the Y31 variant was predominantly in the form of tetramers . Taken together, stabilization of the dimer of stefin B is sensitive to Pro 36, whereas stabilization of the tetramer is sensitive to Pro 79. When cis to trans transition is facilitated by the Pro mutation at the two sites, respectively, dimers and tetramers are populated to a higher amount.
The prediction of human stefin B stability upon single-point mutations of proline residues in the protein sequences (UniProt ID: P04080 and its mutant C3S)  as well as the monomeric (1STF:I  and 4N6V:chain0  and tetrameric 2OCT:chainA ) protein structures, was performed at pH 7.0 and 25 °C using a support vector machine (SVM)-based tool, I-Mutant2.0 . Of note, the protein stability increased only for the P6L mutant, whereas all remaining mutations, namely P36D and three Pro to Ser mutations at positions 11, 74 and 79, decreased protein stability (Table 1).
Substituting proline at position 74 with a serine in the sequence of the wt stefin B did not affect the protein structure and stability to any significant extent, as shown by urea and thermal denaturation . In fact, the mutant was slightly more stable, which is in contrast to the prediction in Table 1 (one, however, has to bear in mind that the changes in stability in both cases: prediction and experiment, are rather small and within the standard error of 1.4 ± 0.1 kcal/mol of the I-Mutant2.0 program). The exchange of a proline would be expected to lead to a more stable protein, due to higher flexibility—i.e., entropic contribution to stability, however, enthalpic contribution and hydration effects increase or decrease the stability.
When the fibrillation rate of the P74S mutant was compared to the fibrillation rate of the wt-like protein, however, it was shown that P74 is essential not only for stefin B tetramer formation but also for amyloid fibrillation. Indeed, when Pro74 was replaced with Ser, the lag phase was extended up to 10 times with a smaller final yield (Figure 4A,B). CD spectra show that this mutant adopts a folded structure, thus these differences are not the result of a change in the overall fold of the mutant. Transmission electron microscopy (TEM) results (Figure 4C,D) reveals that P74S remains in the form of granular aggregates (Figure 4D), whereas the wt protein formed amyloid fibrils after 7 days of incubation (Figure 4C). Moreover, when the effects of peptidyl-prolyl isomerase cyclophilin A (CypA) were examined, it was shown that CypA prolongs the lag phase and increases the final yield and length of the fibrils. On the other hand, the inactive cyclophilin A R55A caused a prolonged lag phase, but did not lead to an increase in the final fibril yield . Although the fibrils formed in the presence or absence of CypA had the same shape and morphology, the presence of CypA provides a higher yield of stefin B fibrils .
The crystal structure of the monomer of stefin B (Figure 5A,B) determined in a complex with papain was one of the first structures of cystatins and as such represented a cornerstone in our ability to understand the mechanism of its inhibitory action on proteases of the papain family . The monomer is a typical α/β protein, with a well-formed β-sheet of 5 β-strands and an α-helix (residues 12 to 37). As an interesting point, a monomer of stefin B crystallized at pH 10 , showed a 4-dimensional arrangement in the crystal cage, resembling a channel.
The monomeric structure of stefin B  was also crucial for further understanding the mechanism of amyloid-fibril formation through 3D-domain swapping. One mechanism for oligomerization is a 3D domain-swapping mechanism [26,46,47] where an intramolecular interface from one monomer becomes an intermolecular interface between subunits in the oligomers . First, the crystal structure of cystatin C domain-swapped dimer was determined , closely followed by the NMR-derived structure of stefin A . In the domain-swapped dimer of stefins, each stefin fold is made of strand 1, the α-helix and strand 2 from one monomer and strands 3–5 from the other monomer. In addition, the stefin B tetramer has been shown to have two domain-swapped dimeric units which interact through loop-swapping, also termed “hand-shaking” (Figure 5C,D).
In the case of stefin B it has been demonstrated that prolines play an important role in domain swapping, as they control the rigidity of loops between secondary structure elements. The trans conformer of Pro 74 is found widely conserved among stefins and cystatins [26,49], whereas the cis conformer is reported in the structure of the stefin B tetramer only . This isomer is particularly important as it brings the Ser 72-Leu 80 loop in the close vicinity of the N-terminal trunk. In the loop swap of two domain-swapped dimer units, the loop position from residues Ser 72 to Leu 80 is provided by Pro 74 and Pro 79. Pro 79 contributes to the rigidity of the loop through its trans conformation.
For comparison with our model protein stefin B data, we reviewed the literature on the role of proline isomerization in the structure, folding and amyloid fibril formation of β2-microglobulin (β2m). Furthermore, we predicted how chosen proline mutations may influence the stability of this protein.
β2m is a 99 amino acid long protein containing the light-chain of the major histocompatibility complex I (MHC I) . β2m is present at the surface of almost all cells. Upon dissociation from MHC I it is catabolized in the kidneys. Therefore, in patients who suffer from chronic kidney insufficiency and undergo dialysis treatment, the concentration of β2m increases up to 60-fold causing dialysis-related amyloidosis (DRA), i.e., insoluble β2m amyloids form and accumulate in the joints and in connective tissues [51,52]. Intriguingly, high concentrations of β2m cannot completely clarify the onset of amyloid precipitation, as in vitro studies have shown that this protein stays soluble and monomeric at neutral pH even when concentrations are more than 100 times higher than in patients exposed to dialysis [52,53].
Chiti et al.  have shown that β2m folds via two structurally different intermediates on its way to the globular native state. One of these, termed I1, is populated within 5 ms and contains a disorganized hydrophobic core with several hydrophobic residues exposed to solvent. The other one, termed I2, forms within ms from the I1 species and shows a native-like secondary structure with side chains packed in the hydrophobic core.  This species further folds to a globular native state within an interval ranging from seconds to minutes at 30 °C. Further studies demonstrated that the slow folding of I2 which precedes the native state is rate limited by trans to cis isomerization of the His 31-Pro 32 peptide bond .
β2m fibril formation starts rapidly at low pH with lag-dependent kinetics where dimers, trimers and tetramers are formed [56,57]. Studies of the kinetics of fibril formation have shown that monomers form a nucleus consisting of six β2m polypeptide chains, whereas fibrils are formed in the elongation phase by adding monomers . Even though several β2m oligomeric species have been characterized  the linkage between oligomers and fibrils remains unknown.
Oligomerization is considered as a crucial step towards self-association of proteins into amyloid fibrils. Moreover, oligomers are believed to be toxic in several types of amyloid-related neurodegenerative diseases [59,60,61,62]. Exploring the molecular mechanisms leading to the formation of oligomers is a great challenge, as it would help in developing strategies to suppress amyloid-related diseases. Toxicity is not restricted to pathological proteins alone, it is instead related to a common structural/conformational property of the prefibrillar oligomers [61,62]. The mechanism through which β2m causes DRA remains poorly understood. It has been reported that β2m forms nonselective, long-lived and voltage-independent ion channels in phospholipid bilayers and that their appearance is tightly correlated with DRA . These channels can bind Congo red and zinc, hence it was suggested by the authors that their structure includes β-sheets . On a separate note, it is also not clear whether it takes the full-length protein to develop the pathophysiology or whether fragments can cause it. In order to clarify this issue, Mustata and coworkers designed K3, which is a digestion fragment of the full length β2m (Ser 20-Lys 41) . It is known that this peptide forms amyloid fibrils under a wide range of conditions . Combining solid state NMR, atomic force microscopy and X-ray diffraction, the characteristic amyloid conformation was elucidated; thus showing that K3 has adopted a U-shaped β-strand-turn-β-strand motif . Interestingly, this motif had already been reported as a universal amyloid feature and hence it was speculated that it might play a role in toxicity [65,66,67]. Moreover, the same authors have proven by channel modelling that this K3 oligomer can constitute the structure of the channel. These results, together with fluorescence measurements in kidney cells which have shown channel-mediated calcium uptake, indicate that the β2m related DRA can be mediated by ion channels formed by the K3 fragment . These data add weight to the so called “channel hypothesis”; these channels lead to Ca2+ influx which can cause apoptosis and alter signaling, hence changing the plasma membrane and electrical properties of the neuron.
β2m has been widely used as a powerful model for exploration of the structural molecular mechanisms of fibril formation from a full-length protein in vitro. Natively, this protein folds into a β-sandwich fold consisting of 2 β-sheets, one containing 4 strands and the other 3, which are covalently linked by a disulphide bond between 2 cysteines (residues 26 and 81) . It contains five peptidyl-prolyl bonds and one of them (His 31-Pro 32) exhibits a thermodynamically unfavorable cis-isomer conformation in the native state .
A huge body of evidence has shown that ~60% of the sequence of β2m is highly amyloidogenic [69,70,71]. Nevertheless, the natively folded protein is not prone to aggregation [72,73] which implies that the folded structure strongly affects its amyloidogenic potential. The partial unfolding in vivo therefore appears to be a mandatory step leading to aggregation as it provides the exposure of aggregation-prone regions of the sequence. β2m spontaneously forms fibrils in vitro at pH < 3.0 with low ionic strength (<50 mM NaCl) when stimulated by agitation . In addition to setting amyloid fibrillation at low pH conditions, in order to cause partial unfolding and drive amyloid fibrillation of β2m at neutral pH, a plethora of conditions has been suggested, such as adding glycosaminoglycans, detergents, denaturants or by using ultrasonication and elevated temperature [53,74,75,76,77,78]. These intrinsic and extrinsic factors increase the concentration of a partially unfolded intermediate in which the natively cis-configured proline 32 in the polypeptide chain is isomerized to a trans isomer [53,79]. In addition, solid-state NMR studies have shown that amyloid fibrils which form from acid-denaturated β2m at pH < 3 contain a trans Pro 32 as well [46,80]. It should be noted that this is not the only structural change reported to be associated with amyloid formation as β-sheets in the protein turn from antiparallel in native β2m to parallel in the amyloid . However, cis-trans isomerization of Pro 32 is considered as a crucial trigger for the transition of soluble monomeric β2m to its misfolded amyloidogenic species [7,8]. This hypothesis is supported by the observation that in the ΔN7 variant of β2m, where the first 7 N-terminal residues are truncated, the cis-Pro 32 conformer is destabilized in such a manner that only the trans-Pro 32 exists at neutral pH .
So far, it has been proven that a single region, approximately 10 residues long (60–70), is crucial for elongation of the full-length protein under certain conditions . Aromatic residues are widely present in this region, which most probably contributes to the propensity of β2m to aggregate . Studies of the full-length protein sequence at low pH have shown that shifting certain residues, especially Leu 23, His 51 and Val 82 with Pro which acts as β-sheet breaker, causes a lowering in fibril elongation kinetics. Moreover, when comparing intact protein at low pH and peptide studies in the context of the effects of sequence alteration on the fibril growth kinetics, results are surprising. Namely, isolated fragments including residues 20–40, 60–70 and ~80–99 all form amyloid fibrils [71,82], whereas in the full-length protein chain mutation of residues, only the ~60–70 region has altered fibril formation kinetics [81,83]. Results of NMR studies explain this observation; the acid-unfolded non-native structure of β2m is stabilized by the disulphide bond and includes gathering of hydrophobic residues in two regions (29–51 and 58–78) , meaning that a single strain of 10 residues might have a strong impact on the aggregation potential of the entire protein. It is speculated that this might be a result of an evolutive twist . Namely, this sequence includes aromatic residues such as Phe 55, Trp 60, Phe 62, Tyr 63 and Leu 65, which are important for interaction with the MHC I heavy chain  and hence for regulation of immune system.
As mentioned above, the cis Pro 32 conformer is proposed as an essential residue for β2m nucleation at neutral pH and P32G and P32V mutants have been used to show this. Namely, both mutants adopt trans Gly or Val 32, respectively, but cannot form amyloid-like fibrils spontaneously, even though P32G can elongate preformed seeds more efficiently than wt β2m [8,55]. These acyclic amino acids favor the trans conformation at the peptide bond, but it is obvious that they cannot completely imitate the unique conformation of Pro 32 . Moreover, variants such as P5G and ΔN7 also affect isomerization of the Pro 32 peptide bond, facilitating fibril nucleation at pH 7.0 [53,84]. On the other hand, β2m can form oligomers and fibrils at neutral pH by addition of Cu2+ and 1 M urea . Namely, peptide bond isomerization at Pro 32 can be initiated by the coordination of a metal ion causing the rapid formation of oligomers [7,53]. Therefore, the isomerization of Pro 32 has been constantly shown as a key initial step in β2m amyloid fibrillation .
In summary, aggregation of β2m into amyloid structures may be achieved via a numerous routes as β2m forms amyloid fibrils at both pH 2.5 and 7.0 . There are many avenues that might finally lead to a better understanding of the assembly pathways in different conditions. In both cases, interactions between specific hydrophobic and aromatic residues may lead to fibrillation. However, fibrils formed at neutral pH contain a highly-charged surface , which could be neutralized at low pH. This might explain the fact that fibrils form much more rapidly under acidic conditions and provides support for a convergent mechanism of assembly at acidic and neutral pH. Another important hallmark of amyloid fibrillation which is considered as a key to amyloid formation is the destabilization of the N-terminal region; a double variant P32G/17A which combines a trans peptide bond at Pro 32 with the destabilization of the N-terminal region forming fibrils spontaneously at pH 7 . It remains to be elucidated whether the assembly pathways are similar and how they converge in the form of a common fibrillar structure.
β2m stability was assessed using I-Mutant2.0  upon Pro to Ser mutations in positions 5, 14, 32, 72 and 90 (numbering according the processed form of the protein—UniProt ID: P61769) and mutations P32G  and P32L on the primary and tertiary structures [87,88], respectively. Of note, all five Pro to Ser mutations destabilize the protein (−2.24 to −0.99 Kcal/mol) as well as the β2m-P32G mutation (−2.62 to −1.74 Kcal/mol) (Table 2). On the other hand, the β2m P32L mutant exhibits a destabilizing effect on its primary structure (−2.05 Kcal/mol) and a stabilizing effect on the tertiary structure of its monomeric  (1LDS:A, 0.79 Kcal/mol) and dimeric (3LOW:A , 1.05 Kcal/mol) forms, respectively (Table 2).
3D structures of the β2m monomer, dimer and tetramer are also known (Figure 6A–D). Some studies suggest that different reagents can trigger different oligomerization pathways . Namely, the crystallographic structure of the reductant-triggered β2m dimer  was different from the dimer and hexamer triggered by copper [7,89] suggesting that different conditions alter the protein structure in different ways, leading to different results. Liu et al.  suggested that dimerization of β2m may occur via a relatively uncommon run-away domain swap with a covalent linkage where β strands are exchanged between two subunits, creating two interfaces. One is called the closed interface, and the other the open interface due to a new β-sheet that contributes to the stability of dimer. Another hallmark of this phenomenon is the rearrangement of the disulfide bonds as they serve as an intermolecular bond to stabilize the dimer. Moreover, in cases where natively-folded proteins form amyloids, a newly formed cross-β spine is required for the fibril ensemble and in the cases of domain-swapping, studies have shown that the hinge loop is essential for forming the cross-β spine. The LSFSKD structure (residues 53–58 of human β2m) acts as a typical steric zipper structure. Upon reduction of the intramolecular disulfide bond, the β2m monomer can assemble as “closed-ended” oligomers or “open-ended” runaway domain-swapped oligomers. The formation of intermolecular disulfide bonds stabilizes the domain-swapped oligomers. The self-association of hinge loops into a zipper spine allows the transformation from oligomers into fibrils; as the oligomer grows, the loop regions between swapped domains can slide slightly to fit into a particular frame. Based on these findings they postulated the so-called “domain-swapped zipper-spine model” of a β2m fibril .
In brief: the recombinant wild-type like, C3S E31-stefin B and its corresponding proline mutants were expressed in E. coli and purified by carboxymethylated (CM) papain Sepharose affinity chromatography. The unbound material was eluted with 0.01 M Tris-HCl containing 0.5 M NaCl at pH 8.0. Stefin B protein was eluted with 0.02 M triethanolamine (TEA) buffer at pH 10.5 and was fast refolded into a stronger buffer of a neutral pH. Furthermore, in the cold room, gel-filtration on Sephacryl S-200 was performed using phosphate 0.01 M buffer pH 7.5., 0.12 M NaCl. For analytical purposes, size-exclusion chromatography was used. Using a Superdex 75 column (Pharmacia, Uppsala, Sweden), stefin B eluted as a set of oligomers: monomers, dimers, tetramers and higher oligomers.
All other methods: expression, isolation and purification of stefin B wt and mutants as well as the conditions to follow fibril fluorescence by ThT fluorescence, were the same as previously described [20,90].
Fluorescence was measured using a model 1.2× fluorimeter from PTI-Photon Technology International (Birmingham, NJ, USA) with a thermo unit for temperature control. 1-anilinonaphthalene-8-sulfonic acid (ANS) fluorescence was measured using an excitation wavelength of 370 nm and spectra were recorded from 400 to 600 nm. The entrance and exit slits for the excitation light-beam were 3 nm, 2 nm and 2 nm, respectively. Measurements were made in a 10-mm micro-cuvette at 25 °C. Thioflavin T (ThT) fluorescence was measured using an excitation wavelength of 440 nm and emission wavelength of 482 nm.
CD spectra were measured using an Aviv model 62A DS CD spectropolarimeter (AVIV, Lakewood, NJ, USA). Far-UV CD spectra were recorded in a 0.1 cm cell. Protein concentration was 34 μM or lower for the far-UV CD. For the far UV CD the bandwidth was 1 nm, and the step of measurement was 1 nm, with data integration time 4 s. Measurements were performed at 25 °C.
The oligomeric state and purity of the protein samples was determined by size-exclusion chromatography (SEC) using a Superdex 75 FPLC column (Pharmacia, Uppsala, Sweden). The flow rate was 0.5 mL/min and typically a 100 μL of 50 μM sample of the protein was applied. Buffer was 10 mM potassium phosphate, pH 7.0, with 0.15 M NaCl added—if not otherwise specified.
Studies on both stefin B and β2m indicate that there is a link between oligomerization and cis to trans isomerization of certain Pro residues. For β2m, in the monomer, Pro 32 is found in a cis conformation. In this case, cis to trans isomerism leads directly to fibril formation, whereas in stefin B the trans to cis isomerization leads to the off-pathway tetramer  so that yet another transition from cis to trans is needed for fibril elongation . In conclusion, cis to trans isomerization of a critical proline may act as a switch towards amyloid fibrils, starting with domain-swapping. Neighboring residues of the proline undergoing cis/trans isomerism are also important for the regulatory switch, such as lysine or serine/threonine residues, in prion and phosphorylated Tau, respectively [92,93].
The authors acknowledge the financial support from the Slovenian Research Agency (research core funding No. P1-0048 to Dušan Turk, and P1-0140 to Boris Turk and the project P1-0048 to Eva Žerovnik). Samra Hasanbašić was given a Fellowship via CMEPIUS (for student exchange at the Jožef Stefan International Postgraduate School).
Supplementary materials can be found at www.mdpi.com/1422-0067/18/3/549/s1.
Eva Žerovnik and Samra Hasanbašić reviewed literature data and wrote the paper, Ajda Taler-Verčič, Veronika Stoka and Samra Hasanbašić performed experiments on stefin B proline mutants and wrote their parts; Dušan Turk, Ajda Taler-Verčič, Veronika Stoka, Selma Berbić and Eva Žerovnik contributed to the Discussion.
The authors declare no conflict of interest.