|Home | About | Journals | Submit | Contact Us | Français|
Elongated trimeric adhesins are a distinct class of proteins employed by phages and viruses to recognize and bind to their host cells, and by bacteria to bind to their target cells and tissues. The tailspikes of E. coli phage K1F and Bacillus phage Ø29 exhibit auto-chaperone activity in their trimeric C-terminal domains. The P22 tailspike is structurally homologous to those adhesins. Though there are no disulfide bonds or reactive cysteines in the native P22 tailspikes, a set of C-terminal cysteines are very reactive in partially folded intermediates, implying an unusual local conformation in the domain. This is likely to be involved in the auto-chaperone function. We examined the unusual reactivity of C-terminal tailspike cysteines during folding and assembly as a potential reporter of auto-chaperone function. Reaction with IAA blocked productive refolding in vitro, but not off-pathway aggregation. Two-dimensional PAGE revealed that the predominant intermediate exhibiting reactive cysteine side chains was a partially folded monomer. Treatment with reducing reagent promoted native trimer formation from these species, consistent with transient disulfide bonds in the auto-chaperone domain. Limited enzymatic digestion and mass spectrometry of folding and assembly intermediates indicated that the C-terminal domain was compact in the protrimer species. These results indicate that the C-terminal domain of the P22 tailspike folds itself and associates prior to formation of the protrimer intermediate, and not after, as previously proposed. The C-terminal cysteines and triple β-helix domains apparently provide the staging for the correct auto-chaperone domain formation, needed for alignment of P22 tailspike native trimer.
The elongated adhesin proteins that govern the attachment of many viruses and phages to cells use their extended lateral surfaces to recognize cell surface polysaccharides and lipopolysaccharides. Examples include the Adenovirus penton fiber, T4 short tail fiber, Bordetella pertussis toxin, the bacteriophage associated Hyaluronate lyase (Hylp2), and a variety of phage tailspikes.1-8 The majority of these proteins are multimeric, folding and assembling into very stable structures that can survive the diverse environments they are exposed to.9 The structures of three bacteriophage tailspikes have been determined to high resolution (Fig. 1): Salmonella P22 tailspike, Bacillus Ø29 appendage (tailspike), and E. coli K1F tailspike.10-13 These three proteins have major β-helix domains, with triple β-helix regions, and complex trimeric C-terminal domains. Among these three proteins, the folding, assembly and off-pathway aggregation pathways of the Salmonella P22 tailspike have been characterized both in vivo and in vitro. They include monomeric, dimeric and protrimer partially folded intermediates.14-18
Many phage structural proteins require chaperones to assist in folding and assembly. The first identified chaperone, GroEL/ES was found associated with phage λ capsid morphogenesis.19 Other early reports of chaperone that aid folding and assembly are associated with phage T4 head assembly,20 T5 tail assembly,21 and P22 coat protein folding.22-24
The initial efforts to identify a chaperone for P22 tailspike folding and assembly were unsuccessful,25 and the ability of the purified fully denatured chains to refold and assemble the native biologically active state in vitro argued against such a function.16-18,26
Recent reports reveal that K1F and Ø29 tailspikes utilize an intra-molecular chaperone domain (IMC). An emerging class of chaperones are the IMC/auto-chaperone segments of proteins that are necessary for the efficient folding and assembly of their own chains. They were initially identified as N-terminal pro-peptides or pro-sequences needed for maturation of exported proteases such as Subtilisin.27 Chen and Inouye refer to this class as type I.28 Another class, type II, is represented by the C- and N-terminal pro-peptides of collagen in which the pro-peptides align and tether the strands before the collagen triple helix is formed. These registration sequences are subsequently cleaved off.29 Recently, non-cleaved type II IMC-like domains have been reported in β-helical trimeric autotransporter proteins. The autotransporter Pertactin protein of Bordetella, folds the C-terminus domain in the outer membrane, creating a channel for the rest of the chains to move through, and serving as the template for the folding of the N-terminal “passenger” domain.30,31
Recent studies of the assembly of tailspikes of the Bacillus subtilis phage Ø29 and E. coli phage K1F have shown the processes to be governed by their C-terminal domains.13,32 The crystal structures of these adhesins are close analogs to the tailspike of the Salmonella bacteriophage P22 (Fig. 1).
The phage P22 tailspike is predominantly β-sheet (Fig. 2).10,11 The N-terminal domain (residues 1–124) is comprised of anti-parallel β-sheets forming a trimeric mushroom or dome-like structure that is required for binding to the phage capsid. The major structural domain (residues 143–539) is comprised of 13 rungs of parallel β-sheets in a β-helical conformation (Fig. 2A and C). The length of the domain is presumably for the binding and cleavage of the lipopolysaccharide on the host cell surface.11,33 The three β-helical domains are bound through predominantly hydrophilic interfaces.10
After the 13th rung the parallel β-helix motif terminates and the three chains twist around each other and intertwine to form an interdigitated triple β-helix domain (residues 540–546) (Fig. 2C). The chains (residues 547–612) then separate to form 3 sides of a triangular β-prism, essentially an oligomeric left handed β-helix. The subunit interfaces constitute a single buried hydrophobic core (Fig. 2C). The chains then twist by a loop-short α-helix-loop that put three pairs of cysteines—C613 and C635—in a distinctive even plane, forming a ring-like conformation, termed the cysteine annulus (Fig. 2B and C). The tailspike ends in a triple bladed motif termed the caudal fin.
The structures of the tailspikes of Bacillus phage Ø29, and E. coli phage K1F share similarities with the P22 tailspike (Fig. 1). The large structural domain region contains the elongated binding and catalytic domain, β-helical for P22 and Ø29 (Fig. 1A and B), and for K1F, three β-propellers and a β-barrel (Fig. 1C). Further along is the shaft region with different β-sheet structures: a triple β-helix in the P22 tailspike (Fig. 1A); a triple β-prism for Ø29 (Fig. 1B), and have similar elements to the a triangular β-prism/triple β-helix of K1F (Fig. 1C).12,13
These three tailspikes have a trimeric bladed configuration in their C-terminal domains (Fig. 1, bottom panels). The tailspikes of Ø29 and K1F have “tentacles” that extended up to the triple β-helix shaft. P22 tailspike lacks these tentacles.
When the Ø29 appendage and K1F tailspike are expressed without the C-termini, the resulting tailspike chains aggregated.13,34 Ø29 and K1F mutants that retain the C-termini domain trimerize successfully. These results show that the C-termini domains are important for folding and assembly.
The C-termini of the Ø29 and K1F chains acting as IMC/autochaperones raise the question of the role of the C-terminal trimeric domain in the folding and assembly of the P22 tailspike. Unlike Ø29 and K1F, the P22 C-terminal domain is not cleaved after native trimer folding. However, the isolated C-terminal domain (residues 537–666) of P22 tailspike can function as an independent oligomerization domain.35 When three maltose binding protein were tethered to a trimeric P22 tailspike C-terminal domain, the chimera chains assembled into a P22 tailspike-like trimer. This suggests a type II IMC function is exhibited by the C-terminal domain of P22 tailspike.35
In earlier searches for cellular chaperones for P22 tailspike folding and assembly, infected cells were treated with iodoacetamide (IAM) with the intent to block ATP synthesis, in the hope of trapping chaperone complexes.25 IAM did indeed result in blocking intracellular tailspike folding and assembly. However, it turned out to be due to direct reaction of the IAM with reactive C-terminal cysteines in tailspike folding intermediates.25 These cysteine thiols are not reactive in the native state, and in fact are strongly hydrogen bonded.36
The in vivo and in vitro refolding pathway of the P22 tailspike proceeds through a number of folding intermediates (Fig. 2D). The polypeptide chains emerge from the ribosome or out of the chaotropic agent and proceed to form a single chain partially folded intermediate, in which the β-helical domain is somewhat structured.17 The intermediate further folds to form a species competent for chain/chain association yielding the metastable protrimer. The protrimer then proceeds to the mature thermostable native tailspike trimer.16-18,26,37-40
The partially folded monomeric intermediates are very sensitive to temperature, and shift from productive folding to aggregation a few degrees above physiological temperature.15,41,42 As a result of this thermolability of the early folding intermediate, the chain is the locus of many temperature sensitive folding (tsf) mutants.43,44
The metastable protrimer is trapped in the cold, and can be distinguished from the native state by its retarded migration in native-PAGE. The protrimer has not acquired the detergent, thermal or protease resistance, found in the native state.37,38 The protrimer can be converted into the native state in the absence of exogenous proteins or factors.
Robinson and King reported that the protrimer intermediate formed in vitro contains transient inter-chain disulfide bonds, which must be reduced during maturation to the native state.45 The addition of reducing reagents to P22 tailspike folding intermediates retarded mobility of monomeric intermediate through native-PAGE, also suggesting the presence of intra-chain disulfide bonds, presumably formed from the reactive cysteines.46
The unusual reactivity of the C-terminal cysteine thiols may be due to interactions with side chains or backbone atoms within some distinctive local conformation, as seen in the reactive site of cysteine proteases. The thiol group of cysteine is deprotonated by nearby basic residues such as histidine before starting cleavage.47,48 Similarly the reduction of disulfides, formed during folding or assembly, may be performed by local side chain, or backbone interaction within folding intermediates. Disulfide exchange between intra- and inter-disulfide bonds among P22 tailspike folding intermediates support this hypothesis.49
The presence of disulfide bonds in intermediates, yet missing from the native state has been carefully studied for bovine pancreatic trypsin inhibitor.50-54 Similarly, the registration peptide of collagen utilizes transient intra-disulfide bonds for nucleation of C-terminal domain to stabilize the chain association stage.55-57 The collagen zipper-like trimer folding proceeds toward the N-terminal domain with using inter-disulfide bonds formation. At the end of trimer folding, both N-terminal and C-terminal domains are processed by procollagen peptidase.
The recent identification of the auto-chaperone function in phage Ø29 and K1F adhesins suggests that the P22 tailspike cysteine annulus may be acting to organize and get in register the tailspike chains for protrimer assembly. In that case one might expect mutants of these cysteine residues to retard folding, without absolutely blocking the overall reaction. Haase-Pettingell et al. showed that the cysteine to serine mutants in vivo retarded the rate of formation of native tailspike.58 The single serine mutations for each C-terminal cysteine decreased folding yields and kinetics, but did not completely inhibit trimer folding. However, double mutations of both cysteines for serine residues C613 and C635 sharply lowered the yield of native trimer.46,58
The homologous tailspike from Det7 of Salmonella enterica lacks cysteines in this region, Seckler and coworkers have suggested that the transient disulfide bonds are not physiological but an artifact of the relatively slow in vitro refolding process.59 If they do not form disulfide bonds, the in vivo reactivity to IAA (iodoacetic acid) nonetheless indicates an unusual state of cysteine residues during folding and subunit assembly. This conformation dependent reactivity is likely to be a reporter or surrogate for the conformation of the C-terminal portion of the chain needed to carry out its auto-chaperone function.
As noted above, the unusual reactivity of at least two of the eight tailspike cysteines during folding, presumably represents activation by local interactions in some of the partially folded species. If the cysteine annulus motif represents an auto-chaperone function of this domain, it should form early in folding rather than later, as we previously proposed.38,39 The reactivity of those cysteines is likely to serve as a reporter for folding of this region of the chain. Here we examine more carefully the reactivity of the cysteine residues during refolding in vitro, following conformational intermediates separated by native-PAGE, and using mass spectrometry to identify folding and assembly intermediates with reactive cysteines thiols. The results suggest that the conformation maintaining the cysteine thiols in a reactive state forms very early in folding, prior to chain association, and that our earlier model needs to be revised.
Previous work has shown that in vitro refolding of P22 tailspike proceeds through a series of folding and aggregation intermediates. Spectroscopic techniques have not resolved the monomeric and oligomeric productive intermediates or the multimeric off-pathway species. Though native-PAGE is not a traditional method for separating conformational intermediates, the 666 residues of the tailspike chain, and the complexity of the folding and aggregation pathway have made it preferable to other biophysical techniques (Fig. 3).60-63 Utilizing native-PAGE, Betts and King resolved intermediates species on the productive pathway—two partially folded monomeric species MSLOW and MFAST, partially folded dimeric intermediates, protrimer intermediate, and of course the native trimer (Fig. 3A).62 At higher temperatures, the refolding chains partition to multimeric aggregation intermediates, which have slightly different electrophoretic mobility from the related productive intermediates (Fig. 3B). At intermediate temperatures both productive and off-pathway intermediates are populated (Fig. 3C). At 0–4°C (Fig. 3A), two partially folded monomeric species MSLOW and MFAST are precursors to both productive pathway intermediates and non-productive off-pathway aggregation intermediates.63 The complexity of the gel patterns is not an artifact of the method, but captures the population of multiple intermediates required for the chains to reach both the native trimer state, or aggregated inclusion body state.44,62,64
All eight cysteines are reduced and completely unreactive in the native tailspike.36,45 The inhibition of tailspike folding by the in vivo labeling of C-terminal cysteines with radioactive iodoacetamide (IAM), as well as the characterization of intra-disulfide bond in vitro refolding intermediates, did not unambiguously identify which intermediates exhibited redox-active cysteine thiols.25,45 To identify the partially folded in vitro intermediates with the reactive cysteines thiols, we utilized alkylation with IAA. P22 tailspike at 2.0 mg/mL was unfolded in acid urea pH 3, and refolding was initiated by dilution to 0.1 mg/mL into ice cold refolding buffer, 100 mM Tris, 2 mM EDTA, pH 7.6 for 15 min (residual concentration of urea 0.8 M).63 Then the temperature was shifted to 24°C for further refolding and samples taken at various times. For the initial experiments, tailspike was refolded with or without IAA.
As noted above, the complexity of the band pattern represents the presence of both productive and off-pathway intermediates within the samples (Fig. 4A). The kinetic species resolved in this gel were monomeric intermediates, dimeric intermediates, native trimer, protrimer intermediate, as well as trimeric, and multimeric aggregates. In the left lanes at early times, the partially folded monomeric and dimeric intermediates predominate. Note that due to the time taken to load the native gels, and initiate electrophoresis, the “0” time sample has incubated for additional minutes, accounting for the presence of multiple intermediate species. After 60 min of refolding at 24°C (right-hand lane), significant amounts of native trimer has accumulated. In native gels the productive protrimer forms the trailing edge of the off pathway trimeric aggregation intermediate band and is not fully resolved as seen in Figure 3C.
In Figure 4B, IAA has been added to the folding reaction at various times of incubation—0, 5, 30 and 60 min—corresponding to the samples in Figure 4A. These samples were further incubated with IAA as indicated in the top of Figure 4B [Refolding time (min) + IAA labeling time (min) to equal a total of 60 min]. Reaction of IAA with refolding tailspike and further incubation at 24°C sharply reduced the formation of native tailspike (Fig. 4B, 0 time). Though failing to yield native tailspikes, the samples exposed to IAA at early times accumulated higher molecular weight oligomeric aggregation species, visible at the top of the gel. In addition, the band corresponding to partially folded monomeric intermediate was not depleted upon incubation and appeared to accumulate. The simplest interpretation is (1) alkylated chains are unable to proceed through the productive pathway to the native state as found in vivo;25 (2) polymerization to high molecular weight off-pathway aggregates is not inhibited; and may be promoted; in the presence of IAA.
P22 tailspike folding intermediates which have reacted with IAA should have an additional net charge due to the negatively charged carboxymethylated group. These modified chains can be expected to be fractionated from the unreacted chains. Two-dimensional (2-D) native-PAGE was employed in this fractionation.
During 4°C incubation, partially folded monomers, dimers, and protrimer intermediates accumulated due to suppression of aggregation at low temperature.63 To selectively observe IAA labeled intermediates, labeling and subsequently non-reducing 2-D native-PAGE were performed under the 4°C condition (Fig. 5). The gel slice on the top of the gel is the lane of the first-dimensional (1-D) native-PAGE separation placed horizontally to show the species present in the refolding reaction after 3 h at 24°C in the absence of IAA. For the samples of Figure 4B, 30 mM IAA was added after 2 h refolding, and incubated for an additional 1 h at 4°C. Comparison of the 1-D pattern reveals that the formation of the native trimer was clearly inhibited, and monomeric species have accumulated (Fig. 5B, top gels). In Figure 5 after separation through in the 1-D native-PAGE, the gel lanes were excised and put on a non-reducing 2-D native gel (in this case without further treatment), and electrophoresed in a second dimension producing the observed diagonal pattern.
In Figure 5A, the most intense spot in the middle of the diagonal line corresponds to native trimer (NT). The species with retarded migration in the upper left of the gel represent trimeric, and higher order aggregates. On the lower right on the diagonal line are two monomeric intermediates, slower migrating monomeric MSLOW species and faster migrating monomeric MFAST species.
In Figure 5B, with the applied sample lane on top, refolding samples treated with IAA were electrophoresed in the first dimension. The inhibition of native tailspike formation and accumulation of slower migrating monomeric MSLOW intermediate was clear, in comparison with untreated control (Fig. 5A, top). The fractionated bands were then electrophoresed through the 2-D native-PAGE. While most bands were on the diagonal, additional spots migrated below the diagonal line indicating increased charge, or compactness (Fig. 5B). One of the intense spot migrated with the same mobility as faster migrating monomeric MFAST species in 1-D native-PAGE. The faster migrating species has the negatively charged carboxymethylated group introduced by the IAA, which may account for the change of migration, or the reaction with the thiol may have altered the conformation and thus the migration.
Regardless of the precise origin of the altered mobility, these results confirmed that cysteine side chains with unusual chemical reactivities occur in a monomeric folding intermediate in the pathway. These results are similar to previously reported by Danek and Robinson for tailspike monomeric complex treated with oxidized glutathione.65 Given that the cysteine reactivity must depend on some distinct chain conformation, this sensitive precursor may represent species with the C-terminal chaperone domain folded into a competent conformation.
Systematic study of temperature sensitive folding (tsf) mutants of the P22 tailspike established that the folding defect destabilizes a thermolabile partially folded intermediate leading to off-pathway aggregation.42,43,66 Since chaperones typically function to suppress off-pathway aggregation, we wanted to examine the relationship of the two classes of partially folded monomeric intermediates to the productive and aggregation pathways.
In order to investigate productive and non-productive monomeric intermediates, non-reducing 1-D native-PAGE and non-reducing 2-D native-PAGE were employed. P22 tailspike was denatured, refolded at lower temperature without additional IAA to observe free cysteine side chain interactions, and applied to non-reducing 1-D native-PAGE. To follow the effects that temperature had on the monomeric intermediates, the 1-D native gel slice was incubated in vitro at 24°C restrictive temperature for 0 h, 1 h and 8 h. All gels were stored on wet ice to halt folding until applied to the non-reducing 2-D native-PAGE (Fig. 6). For clarity a graphic depiction of the 1-D gel is positioned on the 2-D gel.
Both slower migrating monomeric MSLOW and faster migrating monomeric MFAST intermediates were observed on a diagonal in the non-reducing 2-D native gel (Fig. 6A: “8 hours on ice”). The arrow points to an additional faster migrating monomeric MFAST intermediate below the MSLOW (Fig. 6A, arrow). This shifted monomeric species below the diagonal line may be produced by intra-disulfide bond formation, or folding compaction of slower migrating monomeric MSLOW intermediates. After 1 h incubation at 24°C sample, the slower migrating monomeric MSLOW band decreased while the intensity of faster migrating monomeric MFAST increased (Fig. 6B, “1 hour at 24°C,” arrow). After 8 h incubation at 24°C, no MSLOW or MFAST monomeric intermediates were visible (Fig. 6C, “8 hours at 24°C”). We concluded that all monomeric species entered the aggregation pathway and were unable to enter the 2-D native gel. It appears that the slower migrating MSLOW intermediate is either intra-disulfide bonded, or collapsed to form faster migrating MFAST species, then proceeded down the aggregation pathway. There was little evidence for thiol reactivity among the higher molecular weight multimeric intermediates, confirming previous observations.45,60
The prior experiments indicated that all partially folded monomeric intermediates upon extended incubation eventually shifted onto the aggregation pathway. However, it is evident that under the refolding conditions one of the monomeric species is the kinetic productive intermediate and the other is the off-pathway aggregation precursor. In the auto-chaperone model, the chain that had reached the required C-terminal conformation would be productive. Because the state of the reactive cysteines can be a report on this conformation, we would expect different reactivities between the productive and off-pathway intermediates. To this end, we employed non-reducing and reducing 2-D native-PAGE to investigate the state of the reactive thiol groups exposed in folding intermediates.
P22 tailspike was denatured, refolded, then applied to non-reducing 1-D native-PAGE. The excised lanes were incubated with or without 0.1 M DTT for 1 h at 24°C before loading onto 2-D native-PAGE (Fig. 7). As seen in Figure 7A, during the 2-D native-PAGE in the absence of DTT, most of the intermediates remained on the diagonal line with the small amount of smeared MSLOW unlike the MSLOW shift in Figure 6B. This difference may be caused by the 7 h incubation of 1-D native gel on wet ice before 2-D native electrophoresis (Fig. 6B). Even at very low temperature in the native gel, monomeric intermediates may alter their conformations.
In the right Figure 7B, in which the second dimension is run on native-PAGE after DTT treatment, a number of new off-diagonal spots were observed (Fig. 7B). One spot appeared at the native trimer position (Fig. 7B, large open circle) migrating above MSLOW species. This confirmed that the MSLOW species was productive and proceeds to the native state in the presence of DTT. It also should be noted that the formation of a small amount of native trimer is seen below the protrimer position, as previously reported (Fig. 7B, small open circle).45 Further analysis of the samples by 2-D SDS-PAGE showed that native like bands were not dissociated into monomers in the presence of SDS, confirming the existence of native tailspike trimer (data not shown). This supported the interpretation that the MSLOW intermediate represents the chain conformation needed for productive folding and assembly. The conversion of MSLOW to native trimer was limited, with significant levels of MSLOW species still remaining after 1 h DTT treatment at 24°C (Fig. 7B).
The presence of DTT sharply reduced the level of the MFAST species but it was not converted to native trimer. Instead, aggregated multimeric species appear above the MFAST species (Fig. 7B, triangles on side of gel), which suggests that the MFAST is a disulfide bonds containing species kinetically trapped in a non-productive conformation. Upon reduction it proceeds down the aggregation pathway.
In the position below the dimeric intermediates, MSLOW monomeric species appeared, presumably due to dissociation of some dimeric intermediates rather than the productive protrimer. Since the productive and off-pathway dimeric intermediates were not resolved in the first dimension, it was difficult to interpret in the second dimension.
From these results, we concluded that tailspike folding under reducing condition is due to the presence of a partially folded monomeric intermediate with critical thiols either correctly disulfide bonded or in some critical environment in the productive conformation. This thiol environment is presumably not maintained in the species that aggregates.
In the native state the cysteine annulus and triple β-helix are intimately packed and symmetrically aligned (Fig. 2). It seemed unlikely that the early partially folded precursors would have this packing prior to their association and assembly. If the C-terminal domain is acting as an auto-chaperone, one may expect the protrimer be arranged in a conformation closer to the native tailspike. To pursue this question we compared the resistance of the partially folded tailspike monomeric species and protrimer to trypsin digestion, using mass spectroscopy to analyze the proteolytic fragments.
Burying of cysteine residues in the folding of the C-terminal domain would be coupled with the formation of tertiary structure. To compare cysteine reactivity in monomeric and protrimer folding intermediates, a two stage thiol labeling was employed, with initial exposure to IAA followed by reduction and subsequent labeling with IAM. Free thiol groups before in-gel trypsin digestion would be carboxymethylated with initial IAA treatment. Thiol groups involved in disulfide bonds and/or are buried in tertiary structure would not react with IAA. The species would be partially denatured by the organic solvent, such as acetonitrile for dehydration of gel, these disulfides would be reduced with DTT, and the resulting thiols and buried thiol would react with IAM during a part of in-gel digestion.
The mass spectrometry of such labeled tryptic fragments identifies reactive thiols due to difference of molecular mass between carbamidomethylated and carboxymethylated cysteines in the same peptide. We will first describe the protection of the tailspike chain from tryptic digestion during folding and assembly, and then the thiol labeling results.
Trypsin fragments of the treated samples were generated and identified by mass spectrometry. The native tailspike trimer is stable to enzymatic digestion, while its folding and assembly intermediates are at least partially susceptible. The summary of the recovery of the peptides is depicted (Fig. 8).
As shown in Figure 4 partially folded monomeric tailspike wild type (WT) chains accumulated after reaction with IAA. For the tryptic analysis, IAA was added at 30 min of refolding tailspike and incubated for 2 h at 4°C. After non-reducing 1-D native-PAGE, the accumulated predominant monomeric band was excised and used for in-gel trypsin digestion. 80% of total peptide of WT monomeric intermediate was recovered.
To compare with protrimer intermediate, the band corresponding to WT protrimer from the refolding reaction was excised from non-reducing 1-D native-PAGE. This gel band was incubated with IAA for 1 h to label reactive thiols in the protrimer intermediate. The gel band was applied for in-gel trypsin digestion in order to generate peptides from the protrimer intermediate. In this case only 48% of total peptide of WT tailspike protrimer intermediate was recovered.
The difference in peptide recovery between monomeric and WT protrimer intermediates is shown using the native trimer structure to depict the recovered peptides (Fig. 8A, “P22 tailspike monomeric intermediate vs native protrimer”). Yellow peptides were identified in both monomeric and protrimer intermediate. Red peptides were recovered only from the monomeric intermediate, and thus considered protected in the protrimer. White peptides were not identified in either species. In comparing peptide recovery from protrimer with partially folded monomeric species, the C-terminal domain of protrimer intermediate appears to be protected from trypsin digestion (Fig. 8A, red color region).
The protrimer in the in vitro refolding of wild type tailspike is a transient species and does not accumulate. To increase the concentration of this species, we used a mutant in the triple β-helix domain 540–548, which accumulated protrimer intermediate.39 In this truncated tailspike variant, ΔN N547Y, the N-terminal 108 residues are deleted. The intense protrimer band of ΔN N547Y was excised after non-reducing 1-D native-PAGE, then analyzed similarly to WT protrimer.
Forty four percent of total peptide of ΔN N547Y protrimer intermediate was recovered (Fig. 8B, “P22 tailspike monomeric intermediate vs ΔN N547Y Protrimer”). There were some differences in peptide recovery between WT protrimer intermediate and ΔN N547Y protrimer. The N-terminal β-helix near the head binding domain was protected in ΔN N547Y protrimer (Fig. 8B, red color region). The protection of the C-terminus begins at the cysteine annulus in the WT protrimer, and in the ΔN N547Y extends into the triangular β-prism region.
The WT protrimer intermediate and protrimer mutants may have some conformational differences, but both protrimers have stable C-terminal regions. Our interpretation is that the C-terminal domains of partially folded monomeric intermediates are relatively free and flexible and thus sensitive to the protease. Then this region begins to assemble, and collapse to a protease insensitive compact structure in the protrimer and native trimer.
Kim and Robinson proposed that the disulfide exchange of transient tailspike disulfide bonds was a key step in the formation of the protrimer.49 In the auto-chaperone model these steps would be associated with that function, similar to the disulfide exchange in procollagen folding and assembly. The earlier studies identifying transient disulfide bonds in the C-terminal cysteines were unable to resolve whether they were inter-subunit or intra-polypeptide.46
To identify the cysteine residues participating in putative transient disulfides, double alkylation labeled peptides from monomeric species and ΔN N547Y protrimer in prior experiment were compared (Table 1). From mass spectrometry of the IAA treated monomeric intermediate, three carboxymethylated cysteines C496, C613 and C635 were identified, these were in peptides in 484–497, 571–615 and 632–648 (columns 1 and 4 of Table 1). We used negatively charged modification to label exposed cysteines before the tryptic digestion. The negatively charged (IAA labeled) cysteine side chains may partially block trypsin access to the positively charged cleavage sites (K607 and K611). As a result, IAA labeled cysteine containing peptide were probably not completely digested into small peptides, and we could identify as a long peptide. We could not find (K607)VNHK611DCR peptide containing one uncleaved site. Thus, IAA labeled cysteine inducing local structural alteration would involve both K607 and K611. The two peptides containing cysteine 613 and cysteine 635 are located in the C-terminal cysteine annulus, and peptide 484–497 containing cysteine 496 is at the end at β-helix (column 2 of Table 1). Sather and King confirmed reactivity of monomeric intermediate cysteines 496, 613 and 635 to IAM in vivo.25
Two IAM labeled cysteines C169 in 164–170 and C267 in 266–274 at β-helical domain were recovered (column 1 and 4 of Table 1). C169 and C267 are among the stacked side chains that appear to fold early and rapidly.17,40 The labeling of these two cysteines by IAM indicates they were disulfide bonded in the monomeric intermediate, or were already buried and unreactive in early folding intermediates.
Two predicted peptides, 286–291 from β-helix domain containing two cysteines C287 and C290, and 447–483 from near the end of N-terminal β-helix domain containing C458 were not detected (column 4 of Table 1). Subsequently, endoproteinase Asp-N or Glu-C was used for the same experiment to generate appropriate length cysteine containing peptides. However, neither peptide was recovered (data not shown). Since the β-helix forms rapidly even in the partially folded monomeric intermediate,17 the cleavage sites flanking these peptides may be inaccessible to proteolysis.
These results indicated that three cysteines in partially folded monomeric intermediates; C496 at the end of β-helix region, C613 and C635 in the C-terminal domain are free and reactive, while other cysteines are not reactive.
We proceeded to assess the state of cysteines in the ΔN N547Y protrimer mutant. For this species only two of seven cysteine containing peptides were recovered, one containing cysteines C267 and C496, (column 5 of Table 1). The three cysteines (C169, C613 and C635) containing peptides recovered from monomeric intermediate were not recovered. This presumably reflects the protective folded and partial assembled state of the chains in the protrimer state. The peptide containing C496, resulted in a mixture with some chains labeled by IAA, and some by IAM. The C496 thiol side chains labeled by IAA were presumably exposed in initial conformation, while the C496 thiols labeled by IAM were partially buried in the protrimer intermediate. This suggests that the lower region of the β-helical domain in the protrimer was not locked in the native like conformation, but maybe more flexible. As noted above, the peptides containing cysteines C613 and C635 were not recovered, reflecting the compact conformation in both the WT and ΔN N547Y (Fig. 8, and column 5 of Table 1). This result suggested that the sequences surrounding these two cysteines were buried or otherwise protected within the protrimer intermediate.
The reactive C-terminal annulus thiols in the monomeric intermediate, C613 and C635 were labeled with IAA, but were protected in the protrimer. These results indicated the C-terminal domain of tailspike polypeptide chain undergoes significant conformational change prior to the assembly into the protrimer, consistent with this region having an auto-chaperone function as established for Ø29 appendage and K1F tailspike.
The C-terminal region of tailspike of P22 is a very complex topological fold, with its interdigitated regions, triple β-helix, cysteine annulus, and triangular bladed motif (Figs. 1 and 2).2). The P22 tailspike in vivo folding and assembly pathway has a t1/2 of 5 min at 30°C, indicating complex kinetic processes.37 Inspection of the C-terminal triple β-helix, triangular β-prism and cysteine annulus regions suggests that formation of these intimately intertwined domains may pose kinetic and/or thermodynamic barriers. Isolated β-helical (residue 109–544) domains do not form stable trimers but are in reversible trimer-monomer equilibrium.18 It seems likely that this wrapped feature of the C-terminal of the tailspike is responsible for the more than 40°C increase in thermostability in the transition from the protrimer to the native trimer.14,38
The discovery that related adhesins have auto-chaperone domains (Fig. 1) provides a unifying model; these regions fold early and chaperone the chain assembly process to produce the intertwined domain with correct chain register. The presence of unusually reactive cysteine residues and perhaps transient disulfide bonds in the C-terminus presumably reflects some distinct local conformation that reflects the auto-chaperone function required for its in-register assembly and chain interdigitation.
The native tailspike trimer is completely resistant to tyrpsin digestion. However, the C-terminal region of the partially folded monomeric intermediate was susceptible to trypsin digestion. In contrast the same regions were protected in the two protrimer intermediates tested. This suggests that the protrimer intermediates have trimer-like folded C-terminal domains, while the N-terminal domains are not fully assembled. This is consistent with an auto-chaperone function within the C-terminal sequences directing the chains into the protrimer configuration.
In this auto-chaperone model the C-terminal interdigitated region forms early in protrimer formation. Our previous description of this region serving as a molecular clamp, implied C-terminal association as a late step in protrimer formation.39 From results reported here we conclude trimer assembly is initiated from the C-terminal domains, directing trimer assembly similar to the role to the C-terminal registration peptides in pro-collagen triple helix precursor folding (Fig. 9).56,57
The tri-bladed motif at C-terminal domain of P22 tailspike is similar in conformation to the other tailspike superfamily C-terminal auto-chaperone domains, though without sequence homology.13,32 The P22 auto-chaperone domain remains part of mature protein after trimer folding, while those of Ø29 and K1F are not. Why these tailspike are different in terms of proteolysis is unclear.
The IMC/auto-chaperone functions have been reported in both C-terminal and N-terminal domains of many proteins, and for bacterial monomeric, or trimeric autotransporter proteins.67,68 As noted earlier, for Bordetella pertussis monomeric autotransporter, the type I IMC/auto-chaperone domain at the end of N-terminal β-helix helps N-terminal “passenger domain” folding upon export to the cell surface.68,69 On the other hand, both Hemophilus influenzae Hia and Yersinia enterocolitia YadA adhesin trimeric auto-chaperone domains direct trimer assembly from the end of triple N-terminal β-helix into head binding domain, as type II IMCs. These sequences remain associated and are not cleaved off.70,71
Many auto-chaperones and pro-peptides utilize transient disulfide bonds. The pro-peptide of Nerve growth factor (NGF) stabilizes folding intermediates to induce correct disulfide bond formation.72 The C-terminal pro-peptide of collagen, which initiates quaternary assembly of the collagen triple helix are stabilized by intermolecular disulfide bonds.55,57 The pro-peptide of von Willebrand factor (VWF) mediates disulfide bridge formation to complete VWF dimer folding.73 Most reported IMCs have been classified by mode of folding as type I or type II, but structural variations suggests the presence of other types of IMC classifications in protein folding intermediates.34,72-75
A variety of experiments indicate a role for transient disulfide bonds in protrimer folding and assembly.45,65 However, it has been difficult to ascertain if these are essential for efficient folding and assembly, or a byproduct of the slow folding reaction, and the proximity of the reactive 613 and 635 cysteines. The observation that double C613S/C635S are defective in folding can be interpreted as either supporting a role for transient disulfide bonds, or a role for the reactive thiol in maintaining some local critical non-disulfide bonded conformation. There are many examples of reactive cysteines in cytosols that transiently stabilize particular conformations. Li et al. has reported transient disulfide bonds-related subunits assemble in capsid protein of simian virus 40.76 Those transient disulfide bonds are well organized in their pentamer formation under the reducing environment of the cytosol. In those proteins folding intermediates, productive intra-disulfide bonds are transferred into inter-disulfide bonds. P22 tailspike folding intermediates in the cytosol exhibit a similar mechanism between non-productive intra-disulfide bond containing monomer and productive inter-disulfide bond containing protrimer.
Regardless of the exact mechanism by which the C-terminal cysteine thiols direct the chain into the productive conformation, the interpretation that these steps are part of an auto-chaperone function as demonstrated in the K1F and Ø29 tailspikes, make it much easier to understand how these chains reach their intertwined highly stable native state.
Wild type (WT), and N-terminal truncated (ΔN) N547Y of P22 tailspike were recombinantly expressed and purified from E. coli as previously described.39,58 The purity of proteins was checked by SDS-PAGE. All protein concentrations were calculated by using A280 nm = 0.98.77
P22 tailspike at 2.0 mg/mL was unfolded in acidic urea buffer (5 M urea, 50 mM sodium citrate and 2 mM EDTA, pH 3.0) for 1 h at 24°C. Unfolded tailspike was diluted to 0.1 mg/mL into refolding buffer (0.58 M urea, 100 mM Tris and 2 mM EDTA, pH 7.6) and incubated for 15 min on wet ice to accumulate productive folding intermediates.63 Refolding was performed as specified.
Addition of 0.1 M IAA stock solution to refolding P22 tailspike at each indicated time and further incubation was performed in the dark for a total of 1 h at 24°C. To prepare samples for non-reducing 2-D native-PAGE, IAA stock solution was added after 2 h refolding tailspike at 4°C, then incubated further for 1 h in dark. The final concentration of IAA was 30 mM for all experiments. At the end of labeling, all samples were mixed with ice-cold 3 × native-PAGE sample buffer (14 mM Tris-base, 109 mM glycine, 30% (v/v) glycerol, 0.1% bromophenol blue), and kept in wet ice until loaded into native-PAGE.
To obtain alkylated peptides of monomer intermediates, IAA stock solution was added to 30 min refolding tailspike reaction and incubated for additional 2 h at 4°C. The final concentration was 30 mM IAA. After non-reducing 1-D native-PAGE, coomassie blue stained monomeric band was excised, reduced and alkylated with IAM followed by in-gel digestion methods with trypsin, Asn-N and Glu-C.
For protrimer alkylation, 2 h refolding WT, and ΔN N547Y were loaded into non-reducing 1-D native-PAGE. After electrophoresis, excised protrimer bands were immediately incubated in native-PAGE running buffer with 30 mM IAA for 1 h at 4°C in the dark, then applied for in-gel digestion and mass spectrometry analysis together with monomeric bands. Tryptic peptides were identified with QSTAR Elite mass spectrometer (Applied Biosystems) installed at the Koch Institute Proteomics lab (http://web.mit.edu/ki/facilities/core.html#biopolymers).
To investigate heat stability and thiol group reactivity, refolded, and alkylated P22 tailspike was electrophoresed through non-reducing 1-D native-PAGE. Refolding and alkylation were performed as specified. After 1-D native-PAGE, the excised lane was incubated in native-PAGE running buffer in the presence, or absence of 0.1 M DTT for indicated time and temperature. After incubation, gel was loaded onto the 2-D native-gel.
All electrophoresis were performed as previously described.78 The concentrations of each acrylamide/bisacrylamide in the resolving gels were 9% for native-PAGE. The concentrations of stacking gels were 3.6% for native-PAGE. All reducing and non-reducing 2-D PAGE were performed with the Bio-Rad Criterion cassette system (Bio-Rad). Native-PAGE was performed at 150 V with a constant current of 15 mA/gel for 4 h in a 4°C cold room. All proteins separated by PAGE were visualized using Krypton™ (PIERCE), and imaged on a Typhoon™ Variable Mode Imager (GE healthcare).
The authors would like to acknowledge Dr. Ioannis Papayannopoulos and Richard Schiavoni at the MIT Biopolymers and Proteomics Core Facility for mass spectrometry, data analysis and discussions. This work was supported by the NIH R01-GM17980 and NIH R01-AI75208.
No potential conflicts of interest were disclosed.
Previously published online: www.landesbioscience.com/journals/bacteriophage/article/19775