Stepwise disassembly and apparent nonstepwise reassembly for the oligomeric RbsD protein

Abstract

Many cellular proteins exist as homo-oligomers. The mechanism of the assembly process of such proteins is still poorly understood. We have previously observed that Hsp16.3, a protein exhibiting chaperone-like activity, undergoes stepwise disassembly and nonstepwise reassembly. Here, the disassembly and reassembly of a nonchaperone protein RbsD, from Escherichia coli, was studied in vitro. The protein was found to mainly exist as decamers with a small portion of apparently larger oligomeric forms, both of which are able to refold/reassemble effectively in a spontaneous way after being completely unfolded. Disassembly RbsD intermediates including pentamers, tetramers, trimers, dimers, and monomers were detected by using urea-containing pore gradient polyacrylamide gel electrophoresis, while only pentamers were detected for its reassembly. The observation of stepwise disassembly and apparent nonstepwise reassembly for both a chaperone protein (Hsp16.3) and a nonchaperone protein (RbsD) strongly suggests that such a feature is most likely general for homo-oligomeric proteins.

Many cellular proteins exist as homo-oligomers. Outstanding questions regarding such oligomerization phenomena include why they form such oligomers, how the subunits are assembled, and what is the relationship between subunit folding and assembly. Recent researches, including those from our own lab, have suggested that many homo-oligomeric proteins are able to modulate their biological activities via adjusting their oligomeric states, realized by disassembly/reassembly (Fu and Chang 2004; Liu and Chang 2004; Fu et al. 2005; Hong et al. 2005; Jiao et al. 2005; Zhang et al. 2005). It follows that understanding the molecular mechanism of their disassembly and reassembly process is of great importance. Although it is conventionally believed that subunit folding must precede the subunit assembly to provide the surface required for the specific intersubunit recognition, the molecular mechanism for the open polypeptides to fold and assemble into oligomers is far from clear (Jaenicke 1991; Jaenicke and Lilie 2000). To understand such mechanism, the refolding and reassembly of oligomeric proteins are more often investigated. In this aspect, it is also important to study the disassembly and unfolding processes for such oligomeric proteins (Tanford 1968; Sánchez del Pino et al. 2002). In such studies, to detect and characterize the intermediates of disassembly and reassembly are essential but often difficult due to their inherent instability, as well as the lack of effective techniques to separate and identify them individually.

In our studies on the disassembly/reassembly mechanisms of the homo-nonameric Hsp16.3, a protein from Mycobacterium tuberculosis exhibiting chaperone-like activity, it was discovered, remarkably, that the protein undergoes stepwise disassembly and nonstepwise reassembly (Abulimiti et al. 2003).

This report represents our continued efforts in understanding the disassembly/reassembly mechanism of oligomeric proteins. This time a nonchaperone oligomeric protein, the Escherichia coli RbsD, was chosen as a model protein. The determination of the crystal structure of the homologous RbsD protein from Bacillus subtilis revealed that the protein exists as a decamer consisting of two pentameric rings (Kim et al. 2003). Additional studies demonstrate that the RbsD protein is an enzyme that catalyzes the alteration of ribose configuration, converting the pyran to the furan form (Ryu et al. 2004).

In this report, the urea-induced disassembly and reassembly processes of the RbsD protein, particularly with the intermediates involved, were effectively examined using a specially designed method based on urea-containing pore gradient polyacrylamide gel electrophoresis (PAGE) (Goldenberg and Creighton 1984; Creighton 1986). Our data demonstrate that the RbsD oligomers are able to refold/reassemble spontaneously after being completely unfolded by urea. More interestingly, the disassembly/reassembly pattern of RbsD resembles that of the molecular chaperone Hsp16.3, being stepwise for the disassembly and apparent nonstepwise for the reassembly.

Results

The RbsD protein, existing mainly as decamers, is able to refold and reassemble spontaneously

To investigate the refolding and reassembly process of RbsD protein, its oligomeric size in solution was first determined by using nondenaturing pore gradient PAGE, a powerful method commonly used in our laboratory (for example, see Gu et al. 2002; Abulimiti et al. 2003). Results presented in Figure 1B, lane 2, demonstrate that the protein exists in two oligomeric forms. The dominant form has a native molecular mass around 150 kDa, correlating to the size of decamers, which is consistent with what was revealed by crystal structure determination (Kim et al. 2003). The minor form (as indicated by the bold arrow in Fig. 1B) mobilizes much slower than the dominant 150-kDa form, whose nature needs to be further studied.

Figure 1.

Oligomeric state analysis of the purified RbsD protein. (A) SDS-PAGE results of the purified RbsD protein. Lane 1, the molecular mass marker; lane 2, the RbsD protein. (B) Examination of the oligomeric size of RbsD via nondenaturing pore gradient (4%–30%) PAGE (150 V for 24 h). Lane 2 is the native RbsD protein at the concentration of 0.5 mg mL⁻¹; lane 3 is the RbsD protein (0.5 mg mL⁻¹) after 18 h of refolding/reassembly (4°C) from the completely unfolded/disassembled states (treated with 8 M urea for 72 h at 4°C). The refolding/reassembly was obtained by adding 50 mM phosphate buffer (pH 7.5) into the urea-treated protein solution (4 mg mL⁻¹) to a final urea concentration of 1 M. Lane 4 is the completely unfolded/disassembled protein (4 mg mL⁻¹) by 8 M urea for 72 h at 4°C. An equal amount of proteins (10 μg) was loaded in lanes 2–4. Lane 1 is the molecular mass marker of BSA (monomer, 67.5 kDa; dimer, 135 kDa; trimer, 202 kDa; and tetramer, 270 kDa).

The capacity of the RbsD protein to recover its oligomeric structure after being completely denatured in the presence of 8 M urea for 72 h was examined by nondenaturing pore gradient PAGE and size-exclusion chromatography. The completeness of unfolding/disassembly is monitored by circular dichroism (CD) spectroscopy (Fig. 2B, curve 3) and nondenaturing pore gradient PAGE (Fig. 1B, lane 4).

Figure 2.

Examination of the refolding/reassembly capacity for the fully unfolded RbsD polypeptide chain. (A) The superimposed elution curves of analytical size-exclusion chromatography for the native (solid line) or 8 M urea-treated for 72 h at 4°C (dotted line) RbsD protein (both at 4 mg mL⁻¹), with the latter sample being undiluted and loaded onto the column directly. (B) Far-UV CD spectra for the native (curve 1) or 8 M urea-treated for 72 h at 4°C (curve 3), and 8 M urea-treated, then 1:8-diluted (curve 2) RbsD protein. All the samples were prepared in 50 mM sodium phosphate (pH 7.5) with the final protein concentration of 0.5 mg mL⁻¹.

Data presented in Figures 1B and 2 demonstrate that RbsD is able to refold and reassemble effectively in a spontaneous way after the urea was diluted. First, the oligomeric size of the protein appears to be fully recovered to its native form as shown by both nondenaturing pore gradient PAGE analysis (Fig. 1B, lane 3) and size-exclusion chromatography (Fig. 2A). The high recovery (∼85%) of the secondary structure of RbsD was clearly indicated by the far-UV CD spectroscopy analysis (Fig. 2B), where the spectra of the native (curve 1) and the refolded (curve 2) RbsD differ little from each other.

Disassembly intermediates of RbsD can be effectively detected by urea-containing pore gradient PAGE

The unfolding/disassembly process of the RbsD oligomers was examined by using both urea-containing pore gradient PAGE (the urea was added to prevent the extremely effective refolding/reassembly of the protein intermediates) and far-UV CD spectroscopy. To capture the unfolding/disassembly intermediates, the protein sample here was treated with 8 M urea for only 18 h, instead of 72 h (where the protein was apparently converted to monomers; Fig. 1B, lane 4).

The pore gradient PAGE data presented in Figure 3A demonstrate the existence of five dominant forms for RbsD when treated with urea of 4 M or higher concentrations (lanes 4–8). Upon further electrophoresis for another 2.5 (Fig. 3B) and 4.5 h (Fig. 3C), the five dominant bands (lanes 4–8) migrated further to positions below the nondisassembled RbsD decamers (lanes 0–3), which have moved to a position in the gel where the pore size is supposedly close to the diameter of the decamers, thus blocking them from migrating further during the extended 2.5 or 4.5 h. In other words, the five dominant bands represent forms of RbsD that have radii smaller than the decamers, implicating that they are disassembly intermediates, whose slower rates of migration than the nondenatured proteins (Fig. 3A, cf. lanes 4–8 and lanes 0–3) on the gel were probably due to their stronger interaction with the polyacrylamide. This claim is supported by the fact that such intermediates were found to be always retarded relative to the native protein in a straight percentage gel, as shown by data presented in Figure 3D. As the structure of RbsD was revealed to be a decamer consisting of two stacked pentameric rings (Kim et al. 2003), the five disassembly intermediates, repeatedly observed in this study (also see results presented in Fig. 4A), most likely represent the pentameric, tetrameric, trimeric, dimeric, and monomeric forms of RbsD. Such oligomeric intermediates are unlikely to be small aggregates formed from the fully unfolded monomers of RbsD, in view of the fact that no oligomers of any size were detected when the fully unfolded monomers of RbsD (formed as a result of full denaturization in the presence of 8 M urea for 72 h) were examined in a straight concentrated polyacrylamide gel (Fig. 5A,B, lane 2 in both) or in a native pore gradient gel (Fig. 1B, lane 4).

Figure 3.

The RbsD protein produces disassembly intermediates when treated with urea. (A) The urea-treated RbsD protein separated by urea-containing (6 M) pore gradient (4%–15%) PAGE (12 mA for 3 h) and visualized by silver staining. Lanes 0–8 are the protein samples (0.5 mg mL⁻¹) treated at urea concentrations of 0, 1, 2, 3, 3.5, 4, 5, 6, 7, and 8 M for 18 h at 4°C, respectively, with an equal amount of loaded protein (10 μg). (B,C) Identical urea-containing pore gradient PAGE as that in A, except that the electrophoresis was continued for another 2.5 (B) or 4.5 (C) h. (D) The protein samples were identical as that in A–C but were separated in a straight percentage (6%) gel PAGE system containing 6 M urea. The electrophoresis was also performed at 12 mA for 3 h. (E) Far-UV CD spectra of RbsD for the identical protein samples as those analyzed in A. Ellipticities at shorter wavelengths were disturbed because of the increased noise by the high concentrations of urea in solution.

Figure 4.

The urea-induced disassembly of RbsD protein occurs in a stepwise manner. (A) Time course study of the disassembly of RbsD protein (0.5 mg mL⁻¹) when treated by 4 M urea at 4°C by urea-containing (4 M) PAGE (20 mA for 2 h). (Lanes 2–10) The protein samples treated at various lengths of time (1, 3, 6, 9, 15, 24, 36, 48, and 72 h, respectively); (Lane 1) the native RbsD protein as a control. An equal amount of protein (10 μg) was loaded in each of the lanes. (B) Far-UV CD spectral examination of samples similarly treated as in A. Ellipticities at shorter wavelengths were disturbed because of the increased noise by the high urea concentrations in solution.

Figure 5.

The reassembly of the RbsD protein is a fast and apparently nonstepwise process. Urea-containing (1 M in A, 2 M in B) PAGE (6% gel) analysis results for the refolding/reassembly of RbsD proteins, which were prepared by diluting the completely unfolded/disassembled proteins (8 M urea for 72 h at 4°C, corresponding to lane 2 in A or B, respectively) into final urea concentrations of 1 M or 2 M, and then maintaining at 4°C for various time (5 sec to 18 h, corresponding to lanes 3 and 4 in A or B, respectively). The final concentrations of diluted proteins were identical (0.5 mg mL⁻¹). The loading buffer also contained 1 M or 2 M urea, respectively, and electrophoresis was performed at 20 mA for 2 h. Lane 1 in both panels is the native RbsD protein as a control. An equal amount of protein (10 μg) was loaded in each lane. (C) Far-UV CD spectral examination of the secondary structural recovery of the RbsD protein samples when urea was diluted. The protein samples were completely unfolded/disassembled in 8 M urea for 72 h at 4°C, and then were diluted to various final urea concentrations (1–8 M) and maintained for 18 h at 4°C, with their final protein concentrations being identical (0.5 mg mL⁻¹). Ellipticities at shorter wavelengths were disturbed because of the increased noise by the high urea concentrations in solution.

The RbsD protein is apparently able to maintain its native oligomeric sizes (both decamers and the larger forms) despite the presence of high concentrations of urea (up to 6 M) in the gel, as judged from the migration patterns of the untreated protein (Fig. 3A–C, lane 0), and those treated with low concentrations of urea (see Fig. 3A–C, lanes 1–3), on the gel. This phenomenon was also observed in results presented below (see Figs. 4A, 5A,B). This appears to result from a weakened effect of urea on such gel system.

The disassembly intermediates lost almost all their secondary structures as revealed by far-UV CD spectroscopy

The level of secondary structure remaining in the disassembly intermediates (as detected in results presented in Fig. 3A–C) was then examined using Far-UV CD spectroscopy. Results presented in Figure 3E clearly demonstrate an almost complete loss of the protein secondary structures in those disassembly intermediates. Similar results were also obtained when the disassembly intermediates of the time course study were subject to far-UV CD spectroscopic examination (see Fig. 4).

One may consider that the oligomers were significantly populated only in the gel system (as a result of the tendency for a gel to effectively concentrate a protein) while remaining largely monomeric in the solution conditions used in the CD experiment. This possibility can be apparently ruled out by the fact that the fully unfolded monomers (formed as a result of full denaturization in the presence of 8 M urea for 72 h) were resolved only as a single band, with no visible oligomers of any size, in both the straight concentrated gel (Fig. 5A,B, lane 2 in both) and the native pore-gradient gel (Fig. 1B, lane 4).

Urea-induced disassembly of RbsD occurs in a stepwise manner

The results presented in Figure 3 suggest that the presence of 4 M urea will enable the RbsD protein to unfold in solution. To “snapshot” the unfolding/disassembly process of RbsD, this relatively low concentration of urea was applied for the denaturization treatment. The far-UV CD spectroscopy analysis and urea-containing PAGE results showed that the complete unfolding of RbsD (in 4 M urea solution) took ∼15 h (Fig. 4A[lane 6],B), with the t_1/2 being ∼5.5 h (Fig. 4B), and the complete disassembly took ∼72 h (Fig. 4A, lane 10). Results of the time course study for the disassembly of RbsD, presented in Figure 4A, elucidate a time-dependent sequential conversion of the RbsD intermediates from pentamers to monomers. For the disassembly, the RbsD decamers are apparently first disassembled into pentamers, whose concentration initially increases with time (from lane 3 to lane 5) and decreases afterward (from lane 5 to lane 10), eventually all the oligomeric intermediates are converted into monomers (see lane 10). In this electrophoresis, the disassembly intermediates (pentamers, tetramers, trimers, and dimers) migrated at a slower rate than that of the native decamers, as similarly observed in the results presented in Figure 3. Moreover, the results also clearly demonstrate that the larger forms of RbsD were first disassembled into decamers before the latter to be further diassembled (Fig. 4A, lanes 1,2).

Reassembly of RbsD is a fast and apparently nonstepwise process

Data presented above demonstrate that RbsD does not seem to disassemble in the presence of up to 3 M urea (Fig. 3) and the protein was able to reassemble effectively upon removal of the denaturant (Figs. 1 [lane 3], 2). Detection of reassembly intermediates of RbsD was then attempted by diluting the urea from 8 M to 3 M, 2 M, or 1 M and applying the diluted samples to electrophoresis analysis in gels containing the corresponding level of urea (i.e., 3 M, 2 M, or 1 M). It turned out that the protein was indeed able to reassemble quickly (within 5 sec of dilution) and fully into the dominant decamers and the minor larger forms when the urea concentration was diluted to 1 M (Fig. 5A), but only partially to 2 M even after 18 h of dilution (Fig. 5B), and not at all to 3 M (data not shown). Here the monomer formed from the denaturization (Fig. 5A,B, lanes 2) migrated only slightly faster than the native decamer (Fig. 5A,B, lanes 1), not being as fast as expected for a well-folded monomer, also suggesting that such monomers somehow interacted with the polyacrylamide gel, as described above.

In light of these results, reassembly intermediates, if they exist, should have been detected in Figure 5B, when the sample was diluted to a final urea concentration of 2 M, where the reassembly was incomplete even up to 18 h after dilution. The lack of tetramer, trimer, and dimer intermediates but the presence of pentamers in Figure 5B strongly suggests that such reassembly of RbsD is a nonstepwise process, where the monomers apparently assemble directly into pentamers. The reassembly process did not seem to occur in the gel judging from the following observations: Only monomers were detected when the completely unfolded RbsD sample (with 8 M urea) was directly applied to electrophoresis analysis in two different types of gels: nonpore gradient (Fig. 5A,B, lanes 2) and pore gradient (Fig. 1B, lane 4).

The protein secondary structure, as detected by far-UV CD spectroscopy (Fig. 5C), was found to be effectively recovered when the urea was diluted to 1 M, but only partially to 2 M and not at all to 3 M or above. This is remarkably consistent with what was observed by gel electrophoresis (Fig. 5A,B).

Discussion

We have previously observed that Hsp16.3, a homo-nonameric small heat-shock protein exhibiting chaperone-like activity, apparently undergoes stepwise disassembly but nonstepwise reassembly (Abulimiti et al. 2003). This report represents our continuous study in understanding such disassembly/reassembly processes for a homo-oligomeric protein that does not exihibit chaperone-like activity. For this purpose, the RbsD protein was chosen as a model system and the urea-induced disassembly and the subsequent reassembly processes were examined. Our major discoveries include the following: (1) The RbsD protein mainly was found to exist as decamers with a small portion of apparently larger oligomeric forms, both of which were able to be reassembled very effectively in a spontaneous way after being completely denatured; (2) pentamers, tetramers, trimers, and dimers were detected as the intermediates for the disassembly of the RbsD decamers; (3) pentamer was apparently found to be the only intermediate for the reassembly of RbsD.

Although it is generally observed that small, single-domain proteins are usually able to refold spontaneously (Dobson and Karplus 1999), few reports have addressed the issue of reassembly of oligomeric proteins made of such small, single-domain subunits. In this report, we provided evidence to show that the RbsD decamers, made of such small, single-domain subunits, are indeed able to undergo spontaneous refolding effectively. In addition to this, the reassembly of RbsD was also found to occur effectively. Interestingly, similar efficient reassembly was also observed in other homo-oligomers made of small, single-domain proteins, such as the GroES, being homo-heptamers composed of 10-kDa subunits (Seale et al. 1996), and Hsp16.3, being homo-nonamers composed of 16.3-kDa subunits (Abulimiti et al. 2003).

In contrast, larger proteins composed of multiple domains often refold inefficiently. Even for the most widely studied E. coli chaperonin GroEL, being a large oligomeric protein composed of 14 identical three-domain 57-kDa subunits, attempted refolding of the urea-unfolded protein via a simple dilution was unsuccessful, only leading to the formation of protein aggregates (Price et al. 1993).

In light of the available data on the reassembly of various homo-oligomeric proteins, it appears that it is the size of each subunit, not the final size of the complete oligomers that determines the efficiency of such reassembly processes. This conclusion is supported by the effective recovery of RbsD decamers and its larger oligomeric forms (as reported here), GroES heptamers (Seale et al. 1996) and Hsp16.3 nonamers (Abulimiti et al. 2003). In contrast, proteins made of larger subunit but having a lower total molecular size seem to reassemble less effectively; such proteins include the liver-specific homo-dimeric methionine adenosyltransferase isoform III (MAT III, with the subunit size of 43.7 kDa) (Sánchez del Pino et al. 2002) and the creatine kinase homodimer (subunit being 40.0 kDa) (Grossman et al. 1986).

The phenomenon that the disassembly intermediates are able to maintain their oligomeric states despite an almost complete loss of secondary structures (Figs. 3, 4) cannot be explained by the conventional thinking of protein oligomerization, where the folded structures are believed to be required in providing the surfaces for specific intersubunit recognition (Jaenicke and Rudolph 1986). Our observation indicates that the interaction existing between the unfolded subunits in the disassembly intermediates are highly resistant to the presence of high concentrations of urea. One possible explanation for this may come from the information provided by the determined crystal structure of the decameric RbsD from B. subtilis, in which a kind of strong subunit interaction was formed involving two Lys residues and a chloride ion (Kim et al. 2003).

The only similar case that we can find in the literature is that of the complex formed between trypsin and pancreatic trypsin inhibitor (also a protein). This protein assembly was found not to be dissociated into its constituent peptide chains even by treatment with denaturing agents such as 8 M urea or 6 M guanidine hydrochloride (Berg et al. 2002). This phenomenon was originally explained by assuming the formation of covalent bonds between the two subunits (Travis and Salvesen 1983). Later determination of crystal structures of the protein complex revealed only noncovalent bonds at the interface of the subunits (Perona et al. 1993; Helland et al. 1998; Berg et al. 2002).

The most outstanding observation of this study is that RbsD, a nonmolecular chaperone, apparently undergoes stepwise disassembly (Figs. 3, 4) and nonstepwise reassembly (Fig. 5). These features are illustrated by the diagram presented in Figure 6. For the disassembly process, the decameric protein is first dissociated into a pair of pentamers, with the pentamers being further dissociated into lower order oligomers and monomers in a stepwise manner (Fig. 6A). For the reassembly, the unfolded monomers apparently first reassociate into pentamers, with the pentamers further associating into decamers in a nonstepwise manner. It would be difficult to speculate the simultaneous bringing of five monomers together in the right orientation to form the pentamers. A more likely scenario is such that the apparently nonstepwise reassembly is actually “stepwise,” with the step leading to the formation of the dimers from monomers being kinetically rate-limiting, and the steps leading to the formation of the trimers, tetramers, and pentamers being all fast (as shown in Fig. 6B). In such a pathway, the concentrations of dimers, trimers, and tetramers would be kinetically insignificant and thus escaped from detection in the gel electrophoresis. If this was indeed true, the reassembly process can be roughly considered as the reversal of the disassembly pathway.

Figure 6.

The stepwise disassembly and apparent nonstepwise reassembly model for RbsD. U, unfolded; N, native.

In view that the homo-nonameric Hsp16.3 also displays a similar disassembly/reassembly pattern (Abulimiti et al. 2003), such stepwise disassembly and apparent nonstepwise reassembly might be expected to be a general feature for homo-oligomeric proteins. The biological implications of such disassembly and reassembly mechanisms, e.g., in processes of oligomeric assembly and transformation of oligomeric states, await further investigation.

Materials and methods

Gene clone, overexpression, and protein purification

The rbsD gene was amplified by PCR using genomic DNA from the E. coli K12 strain as templates, with upstream primer (5′-GCGCATATGAAAAAAGGCACCGTTCTTAAT-3′) and downstream primer (5′-AATGGATCCCCTCAGAACGTCACGCCAGCAC-3′) that contain NdeI and BamHI sites, respectively. The gene was cloned into the pET11a vector (Novagen) for protein overexpression in E. coli BL21(DE3) cells that were cultured in Luria-Bertani medium containing 0.1 mg mL⁻¹ ampicillin (Sigma) and induced by 1 mM isopropyl-D-thiogalactopyranoside (IPTG) (Sigma) at an OD of 0.3 at 37°C for 10 h. Cells were lysed by sonication in buffer A (50 mM Tris-HCl at pH 7.3, 0.1 M NaCl) containing 1 mM PMSF (Boehringer Mannheim), and the lysates were centrifuged at 20,000g (Biofuge Stratos, Heraeus) (4°C, 45 min). The obtained supernatant was immediately precipitated by ammonium sulfate (30% fractionation). After the precipitation was resolved and dialyzed in buffer A, the protein was loaded onto a DEAE-Sepharose FF 16/25 column (Amersham Pharmacia Biotech). The pooled unbound fractions were dialyzed against buffer B (50 mM Tris–HCl at pH 8.6) and applied to Source Q15 10/40 column (Amersham Pharmacia Biotech) strong ion-exchange resin pre-equilibrated with buffer B, and then eluted by using a linear gradient ranging from 0.05 to 0.3 M NaCl. The RbsD-containing fractions were pooled and dialyzed against deionized water. The purity of RbsD was >95% as estimated by densitometry analysis of Coomassie brilliant blue–stained SDS-polyacrylamide gels (Fig. 1A). Protein concentration was determined by using the Bradford reagent (Bio-Rad) with bovine serum albumin (BSA) (Sigma) as a standard.

Nondenaturing pore gradient PAGE

The gels, having a vertical linear gradient (e.g., of 4%–15%, 4%–30%), were prepared in a 125 mm × 100 mm × 1 mm mold as described (Yamaoka et al. 1993). All the electrophoresis was performed at 20°C and visualized either by Coomassie brilliant blue staining or by silver staining. Coomassie staining is acquiescent in the text; otherwise, silver staining will be stated.

For urea-containing pore gradient PAGE, the same concentration of urea (e.g., 6 M for the results presented in Fig. 3A) was added into both the staking and the separating gels.

Far-UV circular dichroism

Circular dichroism (CD) spectra were recorded on a J-715-150L spectropolarimeter (JASCO) equipped with a water bath. All the experiments were performed at 20°C and in 50 mM sodium phosphate buffer (pH 7.5), using a quartz cuvette of 0.1-cm path length. Spectra were the average of five scans from 200 to 250 nm, and the buffer and urea base lines were subtracted. Shorter wavelengths were not analyzed because of increased noise.

Size-exclusion chromatography

Analytical size-exclusion chromatography was performed on a ÄCTA FPLC system using a prepacked Superdex 200 10/30 column (all from Amersham Pharmacia Biotech). For each analysis, a 100-μL protein sample was loaded and eluted at a flow rate of 0.5 mL min⁻¹ with 50 mM phosphate sodium buffer (pH 7.5).