Role for cysteine residues in the in vivo folding and assembly of the phage P22 tailspike


  • Cameron Haase-Pettingell,

    1. Department of Biology, Masschusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
    Search for more papers by this author
  • Scott Betts,

    1. Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716, USA
    2. Syngenta, Research Triangle Park, North Carolina, 27709 USA
    Search for more papers by this author
  • Stephen W. Raso,

    1. Department of Biology, Masschusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
    Search for more papers by this author
  • Lisa Stuart,

    1. Department of Biology, Masschusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
    Search for more papers by this author
  • Anne Robinson,

  • Jonathan King

    Corresponding author
    1. Department of Biology, Masschusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
    • Reprint requests to: Dr. Jonathan King, Department of Biology, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA; fax: (617) 252-1843.
    Search for more papers by this author


The predominantly β-sheet phage P22 tailspike adhesin contains eight reduced cysteines per 666 residue chain, which are buried and unreactive in the native trimer. In the pathway to the native trimer, both in vivo and in vitro transient interchain disulfide bonds are formed and reduced. This occurs in the protrimer, an intermediate in the formation of the interdigitated β-sheets of the trimeric tailspike. Each of the eight cysteines was replaced with serine by site-specific mutagenesis of the cloned P22 tailspike gene and the mutant genes expressed in Escherichia coli. Although the yields of native-like Cys>Ser proteins varied, sufficient soluble trimeric forms of each of the eight mutants accumulated to permit purification. All eight single Cys>Ser mature proteins maintained the high thermostability of the wild type, as well as the wild-type biological activity in forming infectious virions. Thus, these cysteine thiols are not required for the stability or activity of the native state. When their in vivo folding and assembly kinetics were examined, six of the mutant substitutions—C267S, C287S, C458S, C613S, and C635S—were significantly impaired at higher temperatures. Four—C290S, C496, C613S, and C635—showed significantly impaired kinetics even at lower temperatures. The in vivo folding of the C613S/C635S double mutant was severely defective independent of temperature. Since the trimeric states of the single Cys>Ser substituted chains were as stable and active as wild type, the impairment of tailspike maturation presumably reflects problems in the in vivo folding or assembly pathways. The formation or reduction of the transient interchain disulfide bonds in the protrimer may be the locus of these kinetic functions.

The ability of cysteine thiols to donate electrons and to switch between oxidized -S-S- and reduced -SH states contributes to many aspects of protein form and function, including catalysis (Lemere et al. 1995; Chapman et al. 1997; Rietsch and Beckwith 1998; Schick et al. 1998), metal binding (Christianson 1991; McCall et al. 2000), protein stability (Price-Carter et al. 1998), and protein folding.

Cysteine thiols have been particularly valuable as reporters in protein folding pathways (Creighton 1992). The presence of disulfide-bonded intermediates in the folding of the extracellular proteins bovine pancreatic RNase A and bovine pancreatic trypsin inhibitor (BPTI) provided the first means for trapping transient folding intermediates and for ordering steps in protein folding pathways. A striking finding for BPTI was the demonstration of nonnative disulfide bonds in the productive pathway (Goldenberg 1992; Darby et al. 1995). Although these intermediates are short-lived, they participate in efficient formation of the correctly folded native state. Bovine pancreatic RNase A intermediates also contain nonnative disulfide bonds. These disulfide bonds can re-arrange even in the absence of exogenous oxidants (Song and Scheraga 2000), but it is less clear if these intermediates are on the productive pathway.

Proteins that fold and assemble in the cytoplasm generally lack disulfide bonds, and it has been assumed that such proteins do not use disulfide bonds in their folding and assembly pathways. The cytoplasm is a very reducing environment governed by two systems: thioredoxin/thioredoxin reductase and glutathione/glutathione reductase; both are coupled to NADPH. In prokaryotes, disulfide bonds are found primarily in periplasmic proteins, such as alkaline phosphatase, or extracellular proteins, such as Escherichia coli enterotoxin Sta (Batisson and der Vartanian 2000; Batisson et al. 2000).

An exception to the above is the unusual role of disulfide bonds in the folding and assembly of the P22 tailspike protein, a structural protein of Salmonella phage P22. Although lacking native disulfide bonds, the tailspike uses disulfide bonds to transiently stabilize multimeric folding and assembly intermediates in the cytoplasmic formation of the mature trimer. The absence of disulfide bonds in the native state has been established both by x-ray diffraction (Steinbacher et al. 1994) and by Raman spectroscopy (Sargent et al. 1988). Although absent in the native state, the multimeric protrimer intermediate in chain folding and assembly has interchain disulfide bonds (Sather and King 1994; Robinson and King 1997). This feature may reflect the difficulty in forming the interdigitated β-sheet structures that contribute to the very high stability of the trimeric tailspike (Seckler 1998; Kreisberg et al. 2000).

Tailspike protein is the adhesin of phage P22 that recognizes the lipopolysaccharide projecting from the host Salmonella outer membrane. The major structural motif of the tailspike subunit is the β-helix or β-coil, first identified in pectate lyase (Jurnak et al. 1994). The native tailspikes are homotrimers of 666 residues per subunit. They are very stable, requiring temperatures above 80°C for thermal denaturation (Goldenberg and King 1981; Haase-Pettingell and King 1988; Sturtevant et al. 1989). The native protein is neither denatured by sodium dodecyl sulfate (SDS) nor degraded by proteases (Goldenberg and King 1981,1982). Biologically active soluble trimeric tailspikes bind to phage capsids, converting them to infectious particles (Israel et al. 1967).

The tailspike structure (Fig. 1) was determined from crystals of two fragments: the major N-terminal portion 108–666 and the C-terminal domain 1–124 (Steinbacher et al. 1994,1997a). The N-terminal domain is comprised of anti-parallel β-sheets that form a trimeric mushroom or domelike structure that is required for binding to the phage capsid. Residues 143 to 540 form the largest domain of the tailspike, an elongated parallel β-coil domain, with 13 rungs of parallel β-strands per subunit. A large loop (197 to 259) forms a “dorsal fin” between rungs B3 and C3. Three long loops between rungs 5, 7, and 8 form a smaller “ventral fin.” Between these “fins,” the elongated lateral surfaces of the tailspike β-coil motif, bind and cleave the lipopolysaccharide (LPS) of the host cell surface. The LPS is bound by W365 and W391 and cleaved by D392, D395, and E359 (Baxa et al. 1996; Steinbacher et al. 1996). In the native trimer, the three subunits are bound to each other around their threefold axis through predominantly hydrophilic interfaces (Fig. 1C) (Steinbacher et al. 1994).

After the 13th rung the parallel β-coil motif terminates and the three chains twist around each other and intertwine to form interdigitated β-sheets (Fig. 1D). This results in the formation of a triangular β-sheet prism that is essentially an oligomeric left-handed β-helix, including residues 541 to 619. In this region of the trimer, the interfaces of the three subunit chains constitute a single buried core (Seckler 1998; Kreisberg et al. 2000). The chains then separate and form terminal β-sheets (residues 620 to 666). The need for the subunits to form precisely registered interdigitated β-sheets may explain the existence of the multimeric protrimer folding intermediate in the pathway.

The native tailspike has eight non-disulfide-bonded cysteines (169, 267, 287, 290, 458, 496, 613, and 635) per subunit (Fig. 1A). These residues are buried, reduced, and unreactive in the native state. However, residues 613 and 635 of each chain are organized as a ring of six, at the C-terminal end of the interdigitated triangular β-sheet prism region of the tailspike (Fig. 1E). They are not close enough to form disulfide bonds in the native trimer (Steinbacher et al. 1994, 1997a).

The tailspike folding pathway has been well characterized both in vivo and in vitro (Fig. 2). The tailspike can be renatured in vitro from the fully denatured state in the absence of any cellular factors (Fuchs et al. 1991; Danner and Seckler 1993). The in vivo folding and in vitro refolding pathways are similar (Goldenberg and King 1982; Fuchs et al. 1991; Betts and King 1998; Betts et al. 1999). The polypeptide chains emerge from the ribosome, or out of the chaotropic agent, and proceed to form a thermolabile single chain partially folded intermediate [I]. This intermediate, in which the main β-coil motif is probably formed, folds further to form a species competent for chain/chain association yielding the protrimer. The protrimer can be trapped in the cold and can be distinguished from the native tailspike state during native gel electrophoresis, because it migrates slower than the native tailspike. The protrimer lacks the detergent, thermal, and protease resistance of the native trimer but converts to the native state in the absence of exogenous proteins or factors.

The protrimer intermediate contains interchain disulfide bonds, which must be reduced during maturation to the native state (Robinson and King 1997). The transient protrimer disulfide bonds are not just an artifact of in vitro refolding; the in vivo maturation of intracellular folding intermediates is blocked by iodoacetamide, which reacts with active cysteines in the intermediates but not with the native state (Sather and King 1994). Iodoacetamide did not label polypeptide chains truncated at residue 489, indicating that one or more of the last three cysteines (496, 613, and 635) are the critical /reactive cysteines (Sather and King 1994).

As the temperature of maturation increases, the fraction of chains that fold successfully into native tailspikes decreases both in vivo and in vitro (Goldenberg and King 1981; Haase-Pettingell and King 1988; Mitraki et al. 1993). The productive intermediate [I] is shifted to an aggregation-prone intermediate [I*]. These off-pathway [I*] species proceed through multimeric aggregation intermediates (Speed et al. 1995,1996) and accumulate in an intact but nonnative inclusion body state. These inclusion body chains, unlike the native tailspike, can be dissociated by SDS without heating (Smith and King 1981; Haase-Pettingell and King 1988). The thermolabile step is before disulfide bond formation and the off-pathway intermediates are not disulfide bonded (Speed et al. 1995).

The tailspike gene is the locus of a large set of temperature-sensitive folding (tsf) mutations (Goldenberg and King 1981; Haase-Pettingell and King 1988,1997). These mutants, which are confined to the β-coil motif, are able to fold into stable functional trimeric tailspikes at permissive temperatures. Once folded, they are as stable as the wild type, with melting temperatures (Tm's) in the region of 88°C. At the in vivo restrictive temperature of 35°C to 40°C, the chains are unable to fold into the native state but partition into inclusion bodies. Thus the tsf mutants identify residues in the tailspike, which are kinetically important for reaching the native state at the higher end of the temperature range for growth. Although six of the tailspike cysteines are in the parallel β-coil domain, tsf mutations have not been recovered at those sites.

To explore further the role of the tailspike cysteines in chain folding and assembly, we have generated site-specific substitutions and examined the properties of the resulting mutant polypeptide chains and mature trimeric proteins.


Mutagenesis and cloning

To examine the role of the cysteine side chains in the folding and stability of the tailspike, plasmids were generated in which each of the eight cysteine codons was substituted with a serine codon. The mutants were constructed using the Stratagene Quik Change Site-Directed Mutagenesis Kit as described in Material and Methods. The parent plasmid containing the wild-type tailspike sequence is denoted pET-tsp. After isolation of the desired mutant clones, the plasmids were transformed into E. coli BL21 (DE3).

The plasmid DNA was sequenced in the region of the substitution from 403 base pairs to 659, depending on the construct. All the cysteine codons were mutated to the expected serine codon. The entire C496S, C613S, and C635S plasmid genes were sequenced. All the amino acids whose codons had been changed to introduce the silent restriction site used for screening were maintained.

Purification of trimeric mutant tailspikes

The mutant polypeptide chains were all expressed at levels similar to wild-type chains and were not degraded in the cells. Although translation was essentially normal, the formation of SDS-resistant trimeric native tailspikes from the newly synthesized chains was sharply depressed at higher temperatures for some of the mutant proteins. However, all eight single Cys>Ser mutant proteins accumulated sufficient SDS-resistant native tailspikes at lower temperatures to purify. Cultures of wild type and seven of the mutants were grown at 30°C for expression and purification of native tailspikes. For C496S, expression was at 17°C. Soluble native tailspikes were purified by procedures described in Material and Methods. All eight mutant proteins behaved similarly to wild type throughout the purification steps. As shown in Table 1, recovery of soluble native tailspike ranged from 12 μg/OD600 of starting culture for wild type to 1.13 μg /OD600 of starting culture for C613S. No effort was made to recover chains from the inclusion body state.

Polyacrylamide gel fractionation of the purified proteins is shown in Figure 3. Panel A is a native gel of the tailspike proteins separated without detergent or reducing agent. The tailspikes are the predominant bands in the gel, and the mutant proteins all electrophoresed with the same mobility. Panel B shows the same proteins electrophoresed through a SDS gel. The lanes on the left side show tailspike samples mixed with the SDS and β-mercaptoethanol, and applied without heating. Because native tailspikes are not denatured by SDS in the absence of heating, they bind little SDS and the native tailspike migrates slowly, in the top third of the gel. On the right of panel B is the sample boiled in SDS and reducing agent; the tailspike chains are denatured and migrate as a 71,600 Dalton band. Both the native and SDS gels show very low levels of other protein bands.

Biological activity of the Cys>Ser proteins

The purification criteria required only that the recovered tailspikes be soluble and trimeric. We, therefore, examined their biological activities as phage adhesin proteins. A key step in the formation of infectious phage particles is the assembly of the trimeric tailspikes onto the neck of the phage capsid (Israel et al. 1967). To examine the biological activity of the purified tailspikes, we assayed their ability to bind to capsids and convert the capsids to infectious particles. This assay also tests the adhesin functions of the tailspike, because the tailspikes assembled onto capsids must function in the infection process. Thus, the overall reaction requires irreversible binding to the neck of capsid, followed by the recognition of host lipopolysaccharide (LPS) receptor and subsequent steps leading to DNA injection.

Serial dilutions of purified tailspike protein were mixed with capsids at 2 × 109 phage equivalents. The mixtures were incubated at 37°C for 1 h to allow binding of the tailspikes to the capsids. The reaction mixtures were then plated on Salmonella DB 7155 to assay for plaque-forming units (PFU), which represent infectious particles generated during the reaction. Figure 4 is the graph of tailspike binding to the capsids, plotted as PFU versus tailspike concentration. The solid triangles show the results for wild-type tailspike averaged over three experiments. At high tailspike concentration, all the capsids were converted to infectious particles, resulting in 2 × 109 PFU. As the concentration of the tailspike decreased, fewer tailspikes were available to bind to the heads, resulting in decreasing yield in PFU. The point at which the PFU line begins to break is the point at which the tailspike concentration in phage equivalents is equal to the head concentration.

As seen in Figure 4, all the mutant proteins displayed similar capsid binding and plaque-forming activity as the wild type. If the mutant tailspikes did not irreversibly bind the heads, one would expect the break in the PFU plateau to be shifted to higher tailspike concentration to ensure the capsid would receive a full complement of tailspikes (Schwarz and Berget 1989b). If the mutant tailspikes had low endorhamnosidase activity, one would expect the plateau to be significantly lower because more tailspikes would be needed to cleave the LPS (Schwarz and Berget 1989b).

Because we did not detect any significant differences between the wild type and the mutant tailspikes, we conclude that these substitutions do not greatly affect the head binding or endorhamnosidase activity of the tailspike. Apparently none of the eight tailspike cysteine residues are needed for the biological activities of the tailspike.

Thermal denaturation

The native tailspike is stable up to 88°C (Sturtevant et al. 1989) as determined by calorimetry or thermal denaturation monitored by SDS gel electrophoresis or loss of activity (Goldenberg and King 1981; Chen and King 1991b). The thermal denaturation of the tailspike proceeds through a sequential pathway so that the most sensitive measurement of thermal stability is to follow the formation of the unfolding intermediates. A trimeric unfolding intermediate (Iu) is populated when the tailspike is heated between 65–75°C in the presence of 2% SDS (Chen and King 1991a,b). This (Iu) trimeric species has its N-terminal 110 residues unfolded, leaving the β-coil region and the interdigatated C-terminal region intact (Chen and King 1991b; Danner and Seckler 1993), which includes the eight cysteines. On further heating, the trimeric unfolding intermediate (Iu) dissociates and denatures further to single chains (unfolded, U). These steps can be monitored by SDS gel electrophoresis in the cold. The N-terminal portion of the chains unfolds and becomes coated with SDS, migrating faster than the native protein in a SDS gel, due to a higher charge to mass ratio. The subsequent dissociation of the partially unfolded trimer yields the more rapidly electrophoresing SDS/polypeptide chain complexes. It was conceivable that the serine substitutions might destabilize either the β-coil region or the interdigitated C-terminus tailspike and that this could be detected during thermal denaturation.

Tailspike trimers (400 μg) were mixed with 2% SDS, 50 mM Tris pH 8 with no reducing agent and incubated at 65°C or 75°C. At various times, portions were withdrawn and mixed with cold 3× SDS sample buffer lacking mercaptoethanol. The samples were electrophoresed through a 6% SDS gel and stained. At 65°C, which is well above the restrictive temperature of growth, the wild-type native tailspike decreases and transforms to the Iu partially unfolded intermediate (Fig. 5). This unfolding intermediate was stable for the 90 to 120 min of the heating experiment. The native mutant tailspikes (open symbols) were almost indistinguishable from the native wild-type tailspike (filled symbols). Thus the in vivo folding or assembly defect detected in the 30°C to 40°C range is not due to thermolability of the native state of the mutant proteins.

By increasing the temperature to 75°C (Fig. 6), we begin to see small effects of the substitutions on unfolding. The formation and loss of the [Iu] partially unfolded trimer is shown in the left panels; the transition from [Iu] to the fully unfolded SDS-sensitive chains (U) is shown on the right. Given the elevated temperatures, these differences reflect only very small effects on stability of these species. The steep increase in the left panels depicts the Native (N) → [Iu] step; the slower decrease represents the [Iu] → U transition. The right side of Figure 6 represents the [Iu] → U unfolding step (Chen and King 1991b). None of the substituted proteins were more stable than wild type.

The wild-type transition from [Iu] → U had a first-order rate constant of 1.08 × 10−4/sec. The mutants—C613S, C635S, C169S, and C287S—did not differ significantly from wild-type 1.04 × 10−4/sec, 1.12 × 10−4/sec, 1.24 × 10−4/sec, and 1.37 × 10−4/sec, respectively. However, there were small differences in the unfolding rates for C267S, C496S, C290S, and C458. They showed [Iu] → U unfolding rate constants of 2.5 × 10−4/sec, 2.44 × 10−4 /sec, 2.57 × 10−4/sec, and 5.55 × 10−4/sec. These very small rate differences at 75°C would extrapolate to extremely small differences in stability at 40°C. They cannot account for the significant reductions in yield in the physiological temperature range. Thus, the cysteine to serine substitutions have only very minor, if any, effects on the thermostability of the Native protein.

Effect of the serine substitution on the folding or assembly of the tailspike

The initial screening for protein expression from the plasmids suggested that, at least at higher temperatures, the substitutions of serine for some of the cysteines was perturbing chain folding and assembly in vivo. Such defects had been characterized in considerable detail for the tsf mutants of the tailspike. However, none of the tsf mutants, at more than 70 sites, were substitutions of cysteines (C. Haase-Pettingell and J. King, unpubl.). To examine more carefully the effect the Cys>Ser substitutions on the intracellular folding and aggregation of the tailspike chains, we examined the production of mature tailspikes as a function of temperature.

E. coli BL21 (DE3) cells containing the tailspike pET plasmids were grown in LB supplemented with 100μg/mL ampicillin to mid-log phase at 37°C. Protein synthesis was induced with IPTG, and portions of the cultures shifted to one of four temperatures: a low temperature of 15°C or 17°C, 30°C, 37°C, and 39°C. Cultures were incubated for 2.25 h for 15/17°C and 30°C, and 1.5 h for the higher temperature (37°C and 39°C). The cells were harvested and resuspended to OD600 = 20 and lysed. The lysates were centrifuged at low speed to generate pellet and supernatant fractions. Samples were mixed with SDS sample buffer without heating and were electrophoresed through a SDS polyacrylamide gel, and the proteins stained with Coomassie blue.

An example of a gel fractionation of 30°C lysates is shown in Figure 7. Cultures expressing three Cys>Ser substitutions, wild-type, and control plasmids were fractionated into supernatant and pellet and electrophoresed through a SDS gel without heating. The central three lanes are mixtures of native trimeric and denatured monomeric tailspike chains at three different concentrations. The pET vector lacking the tailspike gene insert is in the lane to the left of the standards. This control plasmid did not express proteins corresponding to either the 71.6-kD tailspike polypeptide chains or the native tailspikes. The culture expressing the wild-type chains accumulated native trimeric tailspikes in the supernatant. One of three substitutions, C458S, also accumulated a significant level of trimeric tailspike. The C290S culture accumulated a small amount of native tailspike (enough to purify). The C496S culture failed to efficiently form native tailspikes at the higher temperature. The polypeptide chains were expressed at high levels but accumulated in the pellet fraction, presumably representing the inclusion body state within the cells. A significant fraction of the C458S polypeptide chains also accumulated in the inclusion body state.

Figure 8 summarizes the efficiency of folding as a function of temperature. The triangles represent SDS-resistant native trimeric tailspike, and the filled circles represent the off-pathway aggregation to the SDS sensitive inclusion body state, as a function of temperature for all the mutants. The wild-type chains form native tailspike with high yield up to about 35°C. In the 37°C to 40°C temperature range, the yield decreases sharply and the chains accumulate in the inclusion body state (Smith and King 1981; Haase-Pettingell and King 1997). This loss of correctly folded chains represents the thermolability of the early single-chain folding intermediate in the pathway (Goldenberg et al. 1983; Betts et al. 1999). For reference, a temperature-sensitive folding mutant tsfG244R is shown to the right of wild type. These chains partition to the inclusion body pathway at lower temperatures than wild type. The C169S chains shown below wild type display similar behavior as wild type.

Three mutants—C267S, C287S, and C458S (2nd set of panels)—exhibited a moderate defect in folding or assembly, with some native tailspike accumulated at 37°C, similar to the tsfG244R mutant. Four of the mutants exhibited more serious defects in chain folding or assembly. The two mutants C290S and C496S displayed a severe temperature-sensitive folding defect. In the case of C496S, only a small amount of the native tailspike formed even at 17°C, and all the expressed tailspike chains were aggregated at 30°C and higher temperatures. C290S chains folded to native tailspike at 17°C and 30°C but aggregated at 37°C and 39°C.

About 20% of the C613S and C635S polypeptide chains correctly folded into native tailspike at 17°C, 30°C, and 37°C. However, for the double mutant C613S/C635S, in which both cysteines were substituted with serine, no SDS-resistant tailspike was detected, and all the tailspike chains accumulated in the SDS-sensitive state (Fig. 9). This supports the view that these two cysteines may play an important role in the folding of the tailspike (Sather and King 1994; Robinson and King 1997). These in vivo results indicate that the cysteine substitutions suffer kinetic defects in chain folding or assembly.


The eight cysteine residues in the native tailspike—24/molecule—are not disulfide bonded as determined both by x-ray diffraction and Raman spectroscopy (Sargent et al. 1988; Steinbacher et al. 1994). They are not reactive with thiol reagents, do not bind metals, and are not located near the active site for LPS binding (Robinson and King 1997).

Tailspike cysteines are not needed for native stability or activity

Each of the single Cys>Ser substitutions was expressed from plasmids in E. coli. The mutant polypeptide chains accumulated in the cells, partitioned between native-like trimers and aggregated inclusion body forms. The purified trimeric mutant tailspikes maintained their ability to convert phage capsids to infectious particles, which requires all the biological functions of the tailspike. In addition, their thermal stability was similar to wild type. This provides strong evidence that these side chains make no significant contribution to the stability or function of the native tailspike.

The infectivity conferred on particles by the mutant tailspikes required their endorhamnosidase activity. The binding site of the LPS lies in the valley between the two “fins” of the tailspike. Residues Tyr365 and Tyr391 are involved in binding of the LPS and cleavage uses residue at Asp392, Asp395, and Glu359 (Steinbacher et al. 1996,1997a,b). None of the cysteines are near these residues and are probably not directly involved in the LPS-binding/cleavage site.

Cysteine-substituted tailspike chains display altered intracellular folding or assembly kinetics

Although the Cys>Ser single mutant proteins were stable and active once folded and assembled in vivo, the folding in vivo of six of the mutants was retarded with respect to wild type. Two of the mutants, C169S and C267S, showed in vivo yields very similar to wild type. The other six all exhibited depressed formation of native trimers at higher temperatures, despite relatively normal translation rates.

The four other mutants—C290S, C496S, C613S, and C635S—had sharply reduced intracellular folding efficiencies, even at lower temperatures fully permissive for wild type. These reduced yields of native trimer were not caused by defects in translation or in degradation of the chains, but rather by defects in chain folding or assembly in the higher temperature range of phage growth.

The yield of native trimer from the double mutant C613S/C635S polypeptide chains was too low to detect or isolate. This might reflect addition of the effects of each independently, or it may be that two regions of the chains or residues interact in the productive folding pathway. These two residues form a ring of six cysteines in the native trimer, as shown in Figure 1E.

Substitutions of cysteines 613, 635, and 613/635 by alanines also caused serious defects in chain folding or assembly (R. Seckler, C. Haase-Pettingell, and J. King, unpubl.). For the Cys613Ala and Cys635Ala chains expressed at 30°C, double mutant chains expressed at 20°, all the chains detected were recovered as non-native inclusion body aggregates. This is consistent with an important role for the cysteines in the folding and assembly pathway.

The C287S and C458S phenotypes were very similar to those previously described for the temperature-sensitive folding mutants of the tailspike. Both classes of mutants do not destabilize the native trimeric state or reduce its activities but do display in vivo folding and assembly defects at higher in vivo temperatures (Smith and King 1981; Goldenberg et al. 1983; Haase-Pettingell and King 1988,1997). Nonetheless, at least two aspects of the cysteine mutants suggest that they represent a class of defects different from the tsf mutants: (1) Although tsf mutants have been isolated at 70 different sites throughout the β-coil region, none of these substitutions are of cysteine residues (Haase-Pettingell and King 1997; C. Haase-Pettingell and J. King, unpubl.) and (2) The great majority of the tsf sites are exposed surface residues, with buried residues occurring very rarely. The cysteine residues in contrast are almost all buried.

The function of transient disulfide bonds in tailspike maturation

The altered kinetics of in vivo folding or assembly supports the model that the cysteine thiols are necessary for the efficient folding and assembly of the tailspike chains. Tailspike refolding is not sensitive to added metal ions or to metal chelators such as EDTA and EGTA, arguing against metal ligand binding as the role for the cysteines. The exception is cadmium, which poisons refolding (Robinson and King 1997). Cadmium is a very strong ligand for the thiol group, supporting a specific role for cysteines in the folding and assembly pathway.

Sather and King (1994) showed that iodoacetamide reacts within cells with newly synthesized folding intermediates to block intracellular tailspike folding and assembly. However, once the chains have reached the native trimeric conformation, they were no longer reactive with iodoacetamide. Examination of the in vitro pathway revealed that the protrimer intermediate has interchain disulfide bonds, which are reduced in the transition to the native trimer (Robinson and King 1997).

The region of the chain from residues 540 to 620 forms a triangular β-sheet prism in which the intersubunit contacts are primarily hydrophobic residues interacting in the buried core of the prism tailspike (Seckler 1998; Kreisberg et al. 2000). This is very different from the subunit/subunit contacts that occur in the 120 to 540 region of the chain and different from the general subunit interfaces found in multi-subunit proteins (Fig. 1). It is difficult to imagine that these hydrophobic residues, in the folding intermediate that proceeds the protrimer, could be fully solvent exposed. However, they cannot be in the same buried conformation as they will be after interdigitation. Thus, we suspect that the conformation of at least the 540 to 640 region of the polypeptide chain, and perhaps a larger region, undergoes significant rearrangement.

The protrimer is the most likely candidate for the species in which the interdigitated β-strands actually wrap around each other. It is not surprising that this species might require additional stabilization or registration in preparation for forming the complex interstrand triangular β-prism in this region of the tailspike. One model for this would be the interchain disulfide bonds formed through the terminal registration peptides of procollagen and used in setting the correct registration of the three collagen chains. Once assembled, these are cleaved off and are absent from the mature collagen molecule. Following the procollagen model, we propose that the transient disulfide bonds in the protrimer serve to ensure the correct register for the three regions of tailspike chain that are going to form the interdigitated β-sheet prism in the native trimer.

The simplest interpretation of the Cys>Ser substitution results is that the absence of these cysteines interferes with the formation of the transient disulfide bonds that stabilize the protrimer. It is unclear which cysteines participate in these transient disulfide bonds. As shown in Figure 1E, C613 and C636 form a ring of six thiols at the C-terminal end of the triangular β-prism domain of the trimer. These seem to be natural candidates supported by the observation that the double mutant is seriously defective in chain folding or assembly.

However, the individual substitution of C290 and C496 also caused kinetic defects. Sather and King found that chain fragments lacking the C-terminal cysteines C496, C613 and C635 were not radiolabeled by iodoacetamide in vivo. In addition, in vitro labeling with iodoacetic acid followed by proteolysis indicated one or more of the three are the reactive thiols (Sather and King 1994). Determining which of these residues are critical to the folding will require further investigation.

Mechanism of transient cysteine oxidation and reduction

Within cells, cysteine oxidation and reduction reactions may be catalyzed by cellular components such as thioredoxin or glutathione or other potential sources of reducing or oxidizing equivalents. However, tailspikes can be refolded in vitro in the absence of exogenous factors (Seckler et al. 1989; Fuchs et al. 1991), and yields are sensitive to redox state (Robinson and King 1997). Although it is likely that dissolved oxygen is the oxidant, the source of reducing power during the course of the in vitro refolding reaction has not been identified.

The yield of refolded native tailspikes under optimal conditions are usually in the range of 50% to 60%. As a result, the refolding solutions always contain a significant concentration of misfolded tailspike polypeptide chains whose cysteine residues are available as potential reducing agents for productively folding chains. The very dilute radioactive chains reported by Robinson and King (1997) were inhibited under reducing conditions. However, at higher chain concentrations and other buffer conditions, the yield was most inhibited under oxidizing conditions (S. Betts and J. King, unpubl.; F. Fürst and R. Seckler, pers. comm.). Because both oxidation and reduction are required for maturation in the pathway in Figure 2, it is difficult to predict the dependence of yield on redox conditions, in absence of knowledge of the bonds formed or of the donors/acceptors.

Because there is no net change in oxidation state from the beginning to the end of the reaction, there is no formal requirement for exogenous electron donors/acceptors. Thus, it is possible that the reaction is cyclic and internally catalyzed by the folding intermediates themselves. Inteins (Evans and Xu 1999) and green fluorescent protein (Reid and Flynn 1997) are recent examples of polypeptide chain covalent chemistry that are internally catalyzed by regions of the polypeptide chain. It is possible that the protrimer disulfide oxidation and reduction is also a property of the chains themselves. This will require more careful analysis of the biochemical steps involved during both in vivo and in vitro folding of the tailspike chains.

Materials and methods


Ampicillin at 100 μg/mL was used. The 1% LB agar and LB recipes previously reported by Fane and King (1991). Super broth 32 gm Tryptone, 20 gm Yeast extract, 5 gm NaCl and, 7 mL of 1N NaOH in 1 L. These recipes can be found on the King home page,

Molecular biology and cloning

Design of mutagenesis primer was performed as instructed by the Stratagene QuikChange Site-Directed Mutagenesis Kit instructions. All primers were desalted and purified by the manufacturer (Gene Link, Elmsford, NY) and used directly in the mutagenesis reactions. The plasmid containing the C613S/ C635S double mutant of the tailspike was constructed by A. Robinson, J. Batista, and J. King (unpubl.).

The single Cys>Ser mutation was generated by the Stratagene QuikChange Site-Directed Mutagenesis Kit as directed. The primer was annealed to the pET–11a plasmid (Novagen, Madison, WI) with the tailspike gene inserted, called pET-tsp (A. Robinson and J. King, unpubl.). Once the restriction digestion was confirmed the plasmids were transformed into Stratagene Epicurian Coli BL21-Gold (DE3). Tailspike production was examined at 30°C and 39°C.

The criteria for a candidate plasmid to be sequencing was that the mutant plasmid generated the expected restriction pattern and produced tailspike protein. The tailspike plasmid DNA was sequenced by the Massachusetts Institute of Technology Biopolymer laboratory using a Perkin Elmer Applied Divisions model 377 DNA sequencer or by the Joslin Diabetes Center Core using a ABI 373 Fluorescent DNA Sequencer.

Gel electrophoresis and laser densitometry

SDS gel electrophoresis was performed according to King and Laemmli (1971). Detailed instruction can be accessed from The Coomassie-stained gel was analyzed using the Molecular Dynamics Personal Densitometer and ImageQuant software.

Tailspike expression and purification

Recombinant tailspike was expressed in E. coli BL21-gold (DE3) containing the mutant or wild-type pET-tsp. Four liters of LB broth with 100μg/mL of Ampicillin were inoculated with a 1/50 overnight culture, initially grown at 37°C to an OD600 of between 0.6 and 1.3. The tailspike production was induced by the addition of IPTG to a final concentration of 0.5 mM. The cultures were shifted to 30°C and grown for 3 h. (C496S was shifted to 17°C and grown for 4 to 5 h). The cells were harvested by centrifugation at 4K rpm (4700 × g) for 20 min in a Sovall RC3 + with H6000A rotor. The pellets were resuspended in lysis buffer (50 mM Tris pH 8, 25 mM NaCl, 2mM EDTA, and 40 mM octylglucopyranoside (Sigma, St. Louis, MO) and frozen at −20°C.


Fresh lysozyme (Sigma, St. Louis, MO) was added to the thawing samples to a final concentration of 0.1 mg/mL from a 10 mg/mL stock in ddH20. In addition, PMSF (Sigma, St. Louis, MO) was added to a final concentration of 1 mM from a 100 mM stock in ethanol; samples were incubated in a 37°C shaking water bath for 30 min. The lysate was refrozen in liquid N2 and thawed at 37°C for 30 min. DNase 1 (Sigma, St. Louis, MO) was added to 100μg/mL with the addition of 100mM MgS04 for 20 min at 37°C to decrease viscosity due to the released DNA. Insoluble material was removed by two centrifugation steps: first in a Sovall GSA rotor at 9K rpm for 30 min (13K × g). Then the supernatant was further cleared by a centrifugation of 60 min at 35K rpm (96K × g) in the Beckman Ti45 rotor.


The supernatant was brought to 35% w/v ammonium sulfate saturation. The resulting precipitate was collected and dialyzed against buffer B (50 mM Tris pH 7.6, 25 mM NaCl, and 2 mM EDTA). The tailspike was further purified on Pharmacia FPLC using 7 to 8 mL DEAE matrix. A 25 mM to 80-mM NaCl gradient was passed over the column. Tailspike was eluted at 80 mM NaCl. The column was washed with 1 M NaCl to remove residual protein. The fraction with the tailspike, as monitored by OD280 and by PAG electrophoresis, was pooled. The pooled fractions were brought to 35% saturation with ammonium sulfate. The precipitate was collected by centrifugation and dialysis against 50 mM Tris (pH 7.6) 2 mM EDTA with no NaCl. The sample was applied to a hydroxylapapatite column (Biorad, Hercules, CA). Many proteins bind to this matrix while native tailspike chains pass through. Tailspike was eluted with washes of Tris EDTA buffer. A final wash with 40 mM NaPO4 (pH 7.6) eluted off all protein. The fractions containing the largest peak of tailspike were pooled and the purified tailspike was concentrated by bringing to 40% saturation with ammonium sulfate protein concentration was determined 1OD278 = 0.983mg (Sauer et al. 1982).

Head binding assay

Head purification.

To decrease the background from input phage, a lysogen was used to make phage heads lacking tailspikes. Salmonella DB7136 (su) containing the lysogen with the phage genotype of 9amN110/13amH101/c2ts30 (an amber in the tailspike gene 9 to eliminate the tailspikes and an amber in the lysis gene 13, which delays lysis, resulting in the cells filling with heads, and with a temperature-sensitive gene in the repressor). One liter of super broth was inoculated with a DB7136 containing the 9amN110/13am H101/c2ts30 and grown at 30°C until the cell density was 4 × 108 cells/mL. To release the repressor, the culture was shifted to 42°C. To ensure a heat shock response, an equal volume of 57°C broth was added, and the culture was shaken for about 10 min at 42°C. The culture was shifted to 39°C and allowed to grow for 3.5 h. The cells were collected by centrifugation in a Sorvall RC 3B+ 4K rpm for 20 min (4700 × g). The pellet was resuspended in Tris Mg buffer (50 mM Tris pH 8 100mM MgCl2) and 40 mM octylglucopyranoside. The culture was lysed by the addition of CHCl3. The debris was pelleted by centrifugation for 30 min at 9K rpm in a Sovall GSA (13K × g). Supernatants were collected, and the resulting pellet was treated with CHCl3 and frozen/thawed at 30°C and the debris pelleted. The supernatant was pooled and recentrifuged at 60 min at 35K in the Ti45 rotor (96K × g). The heads were recovered in the pellet and resuspended in Tris Mg buffer to stabilize the condensed DNA in the head. The suspended pellet was applied to a CsCl step gradient and centrifuged at 24K rpm (79K × g) for 4 h in a Beckman sw27.1 rotor. A milky white band between density 1.4 and 1.55 g/mL was harvested from the gradient and dialyzed against Tris Mg buffer. The resulting head concentration was 1 × 1013 heads /mL (12 mL total volume).

Head binding assay.

P22 phage heads at 4 × 109 phage equivalence/mL were mixed with purified tailspikes. The starting concentrations were adjusted to 25 μg/mL and then serially diluted. Samples were incubated at 37°C for 1 h and then diluted and plated for PFU at 30°C. LB agar petri plates and Salmonella DB7155 plating bacteria were prepared fresh for the experiments. See more details on the King laboratory web site,

Thermal unfolding

Tailspike protein was diluted to 400μg/mL into 50mM Tris pH8, 2% SDS lacking reducing agent. Samples were incubated at 65°C and 75°C and aliquots were taken at 0, 3, 6.5, 10, 15, 20, 30, 45, 60, 75, 90, and 120 min and mixed with iced SDS 3× SDS sample buffer with no reducing agent. These samples were electrophoresed through a 6% SDS gel in the cold on the same day. The gel was stained with Coomassie blue and quantified using a Molecular Dynamics Densitometer.

Temperature of partitioning

To examine the effect the Cys>Ser substitution had on the folding and aggregation of the tailspike, the tailspike maturation was examined at 4 temperatures of growth. For each mutant, 125 mL of LB Amp (100μg/mL) were inoculated with an overnight culture and grown to mid-log at 37°C. The cultures where removed and tailspike protein production was induced by the addition of IPTG to 0.5 mM. Four aliquots of 25 mL each were transferred to 125-mL baffle flasks and shaken at 17°C (15°C in one case), 30°C, 37°C, and 39°C. The higher temperatures (37°C and 39°C) were grown for 1.5 h and the lower temperature for 2.25 h. At the end, the cultures were iced and the OD600 was determined from a 1/10 dilution. The cells were collected by (12K × g) 10K rpm spin for 10 min in a Sorvall ss34. The cells were resuspended to an OD600 of 20 in lysis buffer (50mM Tris pH 8, 25 mM NaCl, 2 mM EDTA, and 40 mM octylglucopyranoside) and 5mM (reduced glutathione) GSH. Samples of 0.5 mL were frozen at −20°C. The samples were thawed at room temperature on a rocking platform. PMSF at 10mM and fresh lysozyme to 100μg/mL were added and incubated for 20 min at room temperature. DNase was added to decrease the viscosity at 1μg/mL and MgSO4 at 100 mM incubated at room temperature, incubating for 20 min. The partitioning of the tailspike was accomplished at 4°C by a 3-min centrifugation in the Eppendorf mini-centrifuge. The pellet was resuspended in the lysis buffer and electrophoresed through on a 71/2 % SDS polyacrylamide gel. The gels were stained with Coomassie blue and quantified as above.

Table Table 1.. Yield of tailspike from the purification
MutantCell density (total OD600)Tailspike (total mg)Tailspike/cell density (μg/OD600)
Figure Fig. 1..

Structure of the tailspike adhesin. The chain on the left (panel a) represents a single tailspike subunit separated from its partners. The space-filling atoms are the cysteine residues. The native trimeric structure is to its right (panel b). The structure was solved as two separated domains: residues 110–666 (Steinbacher et al. 1994) and 1–108 (Steinbacher et al. 1997a). To the upper right (panel c) is a cross section showing the three equivalent β-coils organized around a threefold axis. Panel d is a cross section through the interdigitated region showing the chains forming a single triangular β-prism, with the walls made of sheets formed from three different chains. Panel e is a cross section through the end of this region, showing cysteines C613 and C635 organized as a ring of six thiols.

Figure Fig. 2..

The folding and aggregation pathway of the P22 tailspike. Newly synthesized chains released from ribosomes form a single-chain partially folded intermediate [I]. These fold further to a species that can recognize itself to form the protrimer intermediate, in which the chains are associated but not fully folded (Goldenberg and King 1982; Fuchs et al. 1991; Betts et al. 1999). The protrimer, a disulfide-bonded species, then transforms to the native tailspike. It seems likely that the protrimer intermediate is required for the formation of the interdigitated region of the molecule (Kreisberg et al. 2000). Interchain disulfide bonds stabilize the protrimer but are reduced in the transition to the native trimer (Robinson and King 1997). The early single-chain species is thermolabile, and as temperature increases forms a species [I*] that associates into multimers that polymerize into the kinetically trapped inclusion body state. These species are not disulfide bonded (Speed et al. 1995,, 1996).

Figure Fig. 3..

Native and sodium dodecyl sulfate (SDS) polyacrylamide gels of purified Cys>Ser tailspike proteins. (A) 6.6 μg/mL purified wt and mutant tailspikes were loaded on to a native gel. (B) Tailspike samples were applied to an SDS polyacrylamide gel without heating (left of gel B) and after heating at 100°C for 3 min (right side of gel B).

Figure Fig. 4..

The capsid binding and cell recognition of the mutant tailspike shows wild-type behavior. Serial dilutions of mutant tailspike chains were mixed with heads at 2 × 109 phage equivalent/mL, incubated at 37°C for 1 h, and plated on Salmonella typhimurium (DB7155) lawns for determining infectious particles or plaque-forming units (PFU). Panel A: (filled triangles) wild type, (shaded circles) C169S, (shaded squares) C635S, and (shaded diamonds) C267S. Panel B: (filled triangles) wild type, (shaded circles) C290S, (shaded squares) C287S, and (shaded diamonds) C458S. Panel C: (filled triangles) wild type, (shaded diamonds) C496S, and (shaded circles) C613S. The lines were drawn by smooth curve fits to clarify the figure.

Figure Fig. 5..

Thermal unfolding of mutant tailspike proteins at 65°C. Purified mutant tailspikes were mixed with 2% sodium dodecyl sulfate (SDS) at pH 8 without reducing agent and heated at 65°C. Samples were withdrawn and added into iced SDS sample buffer without reducing agent and electrophoresed through a 6% SDS polyacrylamide gel. In both panels filled triangles represent wild-type native tailspike (N). The filled circles are the wild-type unfolding intermediate [Iu]. The symbols that are shaded with solid lines are the unfolding of the mutant Native trimers to the unfolding intermediates [Iu], and the open symbols with dashed lines is the appearance of the mutant unfolding intermediates [Iu]. Panel A: (diamonds) C169S, (squares) C267S, (circles) C287S, (triangle) C290S, and (X) C458S. Panel B: (diamonds) C496S, (circle) C635S, and (square) C613S. Lines are to guide the eye.

Figure Fig. 6..

Thermal unfolding of the mutant tailspike proteins at 75°C. Purified mutant tailspike proteins were mixed with 2% sodium dodecyl sulfate (SDS), 50 mM Tris pH 8 without reducing agent and heated at 75°C. Samples were transferred into iced SDS sample buffer without reducing agent at various times. The resulting samples were electrophoresed through a SDS polyacrylamide gel (6%). Panels A1 and B1 show the very rapid conversion of the trimer to the partially unfolded intermediate [Iu], and the slower dissociation to the fully denatured state [U]. The initial sharp increase in the first 10 min, is the melting of the N terminus of the native tailspike to the stable [Iu] species. The decrease represents the melting of the relatively stable [Iu] (as in seen Fig. 5) to the unfolded tailspike (U). Panels A2 and B2 show the appearance of the fully denatured tailspike chains [U] from the partially unfolded [Iu] species. In all panels, the wild type is represented by the filled triangles. Panels A1 and A2: (circles) C168S, (diamonds) C267S, (shaded triangles) C287S, and (squares) C458S. Panels B1 and B2: (circles) C290S, (diamonds) C496S, (shaded triangles) C613S, and (squares) C635S. The panel 1 lines are to guide the eye. In Panel 2, the lines are the best fit to a single exponential equation, yielding the first order rate constants reported in the text.

Figure Fig. 7..

Partitioning of newly synthesized tailspike chains between native and inclusion body states. Lysates containing proteins expressed from plasmid vectors in E. coli were electrophoresed through a sodium dodecyl sulfate (SDS) polyacrylamide gel. Native tailspike trimers are present in the supernatants in the left lanes; the pellets in the right lanes show the aggregated nonnative forms of the tailspike chains. The centrifugation of the lysates was for 3 min in a microfuge as described in Materials and Methods.

Figure Fig. 8..

Folding and assembly of mutant tailspike chains in vivo as a function of temperature. Escherichia coli cultures expressing Cys>Ser mutant tailspike chains were grown at 17°C, 30°C, 37°C, and 39°C. Samples were harvested and a pellet/supernatant separation was performed. “% total tailspike chains” were quantified from the Coomassie-stained sodium dodecyl sulfate (SDS) gels as seen in Figure 7. The formation of native tailspike trimers (triangles) decrease as a function of temperature with the resulting increase in aggregated chains (circles). The top panels show the yields for wild type and for a well-characterized temperature-sensitive folding mutant tsf G244R. (Haase-Pettingell and King 1988)

Figure Fig. 9..

Tailspike chains containing the C613S/C635S double mutant fail to assemble into soluble sodium dodecyl sulfate (SDS)–resistant trimers. BL21 (DE3) cells expressing the C613S/C635S double mutant and wild-type tailspike were grown and protein expression was induced and samples shifted at 17°C, 30°C, 37°C, and 39°C. Samples were harvested and a pellet/supernatant separation was performed. The yield of trimeric native tailspike (triangles) decreased as a function of temperature with concomitant increase in SDS-sensitive tailspike chains (filled circles). The C613S/C635S failed to assemble into SDS-resistant trimers at all temperatures.


We thank Mike Terry for his technical help in designing the C496S primers. The work has been funded by National Institutes of Health grant (GM17980 to J.K.) and National Science Foundation Engineering Research Center Initiative (8803014).

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC section 1734 solely to indicate this fact.