J. Mol. Biol. (1991) 221. 1007-1014
Mapping Transition States of Protein Unfolding bY Protein Engineering of Ligand-binding Sites Javier Sancho,
Cambridge
Elizabeth
M. Meiering
and Alan R. Fersht
M.R.C. Unit for Protein Function and Design IRC for Protein Engineering, Cambridge University Chemical Lensfield Rd, Cambridge CBZ IEW, U.K. (Received
18 March
1991; accepted 29 May
Laboratories
1991)
We describe a method for probing the integrity and relative orientation of structural elements that are indirectly linked by ligands in protein complexes during protein folding. The effect of 3’-GMP on the rate constants of unfolding of wild-type barnase and several mutants has been studied. By comparing the rates of unfolding of wild-type and mutant proteins, we show that the interaction between His102 and 3’-GMP is fully retained in the transition state compared with the folded state, while the interaction between Glu60 and the ligand is partly retained and that of Lys27 is broken. Our data suggest that the transition state has a partly formed ligand binding site in which the guanine binding loop containing Glu60 and the loop containing His102 are formed at the sides of the p-sheet but the docking of the N terminus of the second a-helix containing Lys27 on the /?-sheet is disrupted. The active site of barnase in complexes is thus partly retained in the transition state of unfolding. Although the ligand could in principle perturb the unfolding pathway, there is independent evidence that indicates that similar structural changes occur upon unfolding of unligated barnase. Keywords: protein; engineering; stability; pathway; barnase
1. Introduction The folding of a protein is a transition from a disordered state into a highly ordered and more compact one, the difference in energy between these two states being usually small (Pace, 1975). Theoretical studies on protein folding have shown that there must be, for each protein, a defined pathway of folding (Levinthal, 1968). The folding pathways of a few proteins are currently being studied, e.g. P-subunit of Trp synthase (BlondElguindi & Goldberg, 1990), bovine pancreatic trypsin inhibitor (Creighton, 1988), T4 lysozyme (Segawa & Sugihara, 1984a,b). The folding pathway is frequently multiphasic, containing several intermediates and transition states. The highest energy transition state for folding is, in examples studied so far, that of the final major step (Creighton, 1988; Matouschek, et al., 1989). This is sometimes termed the transition state for the folding pathway. This is also the transition state of highest energy in the unfolding reaction, and is conveniently studied by unfolding kinetics (Matouschek et al., 1989). Our laboratory has undertaken the study of the structure of the transition state of unfolding of a model
protein as a step towards understanding the general principles of protein folding. Our model protein is barnase, a small extracellular ribonuclease from Bacillus .amyloliquefaciens. The virtues of this protein for the study of non-covalent interactions in the pathway of folding have been described (Matouschek et al., 1989). Barnase belongs to a family of extracellular ribonucleases of both prokaryotic and eukaryotic organisms with which it has sequence and structural homology (Hill et al., 1983; Hartley, 1980). These ribonucleases share a common folding motif which consists of an N-terminal a-helix packed against a C-terminal antiparallel b-sheet. The enzymes are guanylspecific or guanyl-preferential endoribonucleases. Comparison of the crystal structures of the ribonucleases and their complexes with inhibitors has revealed that there is a highly conserved guanine binding motif (Sevcik et al., 1990), which is formed by a loop between two strands of the b-sheet. To date, no structures of barnase complexed with inhibitors have been published. However, the sequence of barnase is approximately 80% identical to that of binase, a ribonuclease from Bacillus intermedius, and the X-ray structure of this enzyme 1007
002%2836/91/191007-OS
$03.00/O
J. Sancho et al.
1008
complexed with 3’-GMP has been solved (Pavlovsky et al., 1988). Owing to the high sequence and structural homology of barnase and binase, it is likely that the two enzymes bind the inhibitor in very similar ways. In previous work, a number of structural characteristics of the transition state of folding of barnase have been reported (Matouschek et al., 1989). The integrity of different parts of the protein was probed by making non-disruptive deletions to remove a defined interaction that) stabilizes native barnase. Kinetic and equilibrium folding studies of wild-type barnase and mutants were then used to follow the formation of the interaction during folding. The folding of various parts of the protein has been characterized using this approach. The folding of the active site, however. has not been well characterized because residues in this region are exposed to solvent and make few simple defined intramolecular which interactions. Here, we use an approach probes the integrity of the active site during the unfolding reaction by the ability of the transition state of the protein to make intermolecular interactions with a ligand that binds to the active site. Thus, the ligand is used as an external probe of the structure of the active site. The energy of interaction between a residue within the protein and a ligand that binds to the protein can be quantified in both the folded state and transition state using kinetic data of unfolding of wild-type and mutant) barnase in the presence of ligands. We present data that suggest that the active site of barnase is partly retained in the t’ransition state of unfolding.
2. Materials
and Methods
(a) Chemicals 2-(N-morpholino)ethanesulfonic acid (Mes) and 3’.GMP disodium salt were from Sigma. SP-Trisacryl was from IBF and dialysis tubing of M, cutoff 3500 was from Spectrum Medical Industries. Inc. Ampholine PAGplate gels were from Pharmacia. Aristar grade urea was from BDH. All other reagents were analytical grade and purchased from BDH or Fisons. Stock buffer solutions and urea solutions were prepared as in Kellis et al. (1989) and used throughout these studies.
(c) 1 .rea denaturation A rapid mixing device (Matouschek rt ~1.. 1989) was used to mix urea solutions containing 3’ GMP with barnase solutions containing the same conc*rntration of ligand. Both the urea and the enzyme solutions were made in 50 mM-HMes/NaMes (pH 6.3). Tn eaelr caxperiment 10 ~018 of urea solution wpre mixed with I vol. of enzyme solution (90 to 420 C(M) and the change in fiuowscencr (A,, = 290 nm. iern = 315 nm) rrcordetl. Thr urea dependence of the rate const’ant of unfolding of tht barnase-3’-GMP complex was determined in the prrsrnc~r of 15.6 mM-3’-GMP. (d) Renaturation One
vol.
of
a solution containing 7 n-urea. and 3’.GMP (0. 1.5. 4 or I3 rnM) in 50 mM-Mes (pH 63) was mixed wit,h 10 ~01s of l)ufl%r solution containing the same concentration of ligand but no urea. The fluorescence changes were recorded as bclforr.
480 p&r-barnase
(e) Data fitting All data were fitted to t,he corresponding theorrt,ical Enz$ttrr (Elsevicr- Riosoft. equations with the program Cambridge).
3. Results (a) Injuence
of ligands on the rate wn.strrn.f qf unfolding
The rate constant of a chemical procaess is predicted by transition state theory (Eyring, 1935) as being directly proportional to the ~~quilibrium constant between the ground state and the transition state. In the context of protein folding, fi)r a two-state model, the ground state is thr> folded protein for the unfolding reaction, and the unfolded protein for a refolding reaction. Transition state theory is best applied to the quantification of differences in energy. The kinetics of barnase unfolding can be analysed in terms of transition state theory (Eyring, 1935; Matouschek et al.. 1989. 1990). When the unfolding reaction occurs in the presrncbe of a ligand the following scheme applies:
#
K? F+L
(b) Recombinant
r.zpprrimmts
-t
Mapping Transition States of Protein Unfolding bY Protein Engineering of Ligand-binding Sites Javier Sancho,
Cambridge
Elizabeth
M. Meiering
and Alan R. Fersht
M.R.C. Unit for Protein Function and Design IRC for Protein Engineering, Cambridge University Chemical Lensfield Rd, Cambridge CBZ IEW, U.K. (Received
18 March
1991; accepted 29 May
Laboratories
1991)
We describe a method for probing the integrity and relative orientation of structural elements that are indirectly linked by ligands in protein complexes during protein folding. The effect of 3’-GMP on the rate constants of unfolding of wild-type barnase and several mutants has been studied. By comparing the rates of unfolding of wild-type and mutant proteins, we show that the interaction between His102 and 3’-GMP is fully retained in the transition state compared with the folded state, while the interaction between Glu60 and the ligand is partly retained and that of Lys27 is broken. Our data suggest that the transition state has a partly formed ligand binding site in which the guanine binding loop containing Glu60 and the loop containing His102 are formed at the sides of the p-sheet but the docking of the N terminus of the second a-helix containing Lys27 on the /?-sheet is disrupted. The active site of barnase in complexes is thus partly retained in the transition state of unfolding. Although the ligand could in principle perturb the unfolding pathway, there is independent evidence that indicates that similar structural changes occur upon unfolding of unligated barnase. Keywords: protein; engineering; stability; pathway; barnase
1. Introduction The folding of a protein is a transition from a disordered state into a highly ordered and more compact one, the difference in energy between these two states being usually small (Pace, 1975). Theoretical studies on protein folding have shown that there must be, for each protein, a defined pathway of folding (Levinthal, 1968). The folding pathways of a few proteins are currently being studied, e.g. P-subunit of Trp synthase (BlondElguindi & Goldberg, 1990), bovine pancreatic trypsin inhibitor (Creighton, 1988), T4 lysozyme (Segawa & Sugihara, 1984a,b). The folding pathway is frequently multiphasic, containing several intermediates and transition states. The highest energy transition state for folding is, in examples studied so far, that of the final major step (Creighton, 1988; Matouschek, et al., 1989). This is sometimes termed the transition state for the folding pathway. This is also the transition state of highest energy in the unfolding reaction, and is conveniently studied by unfolding kinetics (Matouschek et al., 1989). Our laboratory has undertaken the study of the structure of the transition state of unfolding of a model
protein as a step towards understanding the general principles of protein folding. Our model protein is barnase, a small extracellular ribonuclease from Bacillus .amyloliquefaciens. The virtues of this protein for the study of non-covalent interactions in the pathway of folding have been described (Matouschek et al., 1989). Barnase belongs to a family of extracellular ribonucleases of both prokaryotic and eukaryotic organisms with which it has sequence and structural homology (Hill et al., 1983; Hartley, 1980). These ribonucleases share a common folding motif which consists of an N-terminal a-helix packed against a C-terminal antiparallel b-sheet. The enzymes are guanylspecific or guanyl-preferential endoribonucleases. Comparison of the crystal structures of the ribonucleases and their complexes with inhibitors has revealed that there is a highly conserved guanine binding motif (Sevcik et al., 1990), which is formed by a loop between two strands of the b-sheet. To date, no structures of barnase complexed with inhibitors have been published. However, the sequence of barnase is approximately 80% identical to that of binase, a ribonuclease from Bacillus intermedius, and the X-ray structure of this enzyme 1007
002%2836/91/191007-OS
$03.00/O
J. Sancho et al.
1008
complexed with 3’-GMP has been solved (Pavlovsky et al., 1988). Owing to the high sequence and structural homology of barnase and binase, it is likely that the two enzymes bind the inhibitor in very similar ways. In previous work, a number of structural characteristics of the transition state of folding of barnase have been reported (Matouschek et al., 1989). The integrity of different parts of the protein was probed by making non-disruptive deletions to remove a defined interaction that) stabilizes native barnase. Kinetic and equilibrium folding studies of wild-type barnase and mutants were then used to follow the formation of the interaction during folding. The folding of various parts of the protein has been characterized using this approach. The folding of the active site, however. has not been well characterized because residues in this region are exposed to solvent and make few simple defined intramolecular which interactions. Here, we use an approach probes the integrity of the active site during the unfolding reaction by the ability of the transition state of the protein to make intermolecular interactions with a ligand that binds to the active site. Thus, the ligand is used as an external probe of the structure of the active site. The energy of interaction between a residue within the protein and a ligand that binds to the protein can be quantified in both the folded state and transition state using kinetic data of unfolding of wild-type and mutant) barnase in the presence of ligands. We present data that suggest that the active site of barnase is partly retained in the t’ransition state of unfolding.
2. Materials
and Methods
(a) Chemicals 2-(N-morpholino)ethanesulfonic acid (Mes) and 3’.GMP disodium salt were from Sigma. SP-Trisacryl was from IBF and dialysis tubing of M, cutoff 3500 was from Spectrum Medical Industries. Inc. Ampholine PAGplate gels were from Pharmacia. Aristar grade urea was from BDH. All other reagents were analytical grade and purchased from BDH or Fisons. Stock buffer solutions and urea solutions were prepared as in Kellis et al. (1989) and used throughout these studies.
(c) 1 .rea denaturation A rapid mixing device (Matouschek rt ~1.. 1989) was used to mix urea solutions containing 3’ GMP with barnase solutions containing the same conc*rntration of ligand. Both the urea and the enzyme solutions were made in 50 mM-HMes/NaMes (pH 6.3). Tn eaelr caxperiment 10 ~018 of urea solution wpre mixed with I vol. of enzyme solution (90 to 420 C(M) and the change in fiuowscencr (A,, = 290 nm. iern = 315 nm) rrcordetl. Thr urea dependence of the rate const’ant of unfolding of tht barnase-3’-GMP complex was determined in the prrsrnc~r of 15.6 mM-3’-GMP. (d) Renaturation One
vol.
of
a solution containing 7 n-urea. and 3’.GMP (0. 1.5. 4 or I3 rnM) in 50 mM-Mes (pH 63) was mixed wit,h 10 ~01s of l)ufl%r solution containing the same concentration of ligand but no urea. The fluorescence changes were recorded as bclforr.
480 p&r-barnase
(e) Data fitting All data were fitted to t,he corresponding theorrt,ical Enz$ttrr (Elsevicr- Riosoft. equations with the program Cambridge).
3. Results (a) Injuence
of ligands on the rate wn.strrn.f qf unfolding
The rate constant of a chemical procaess is predicted by transition state theory (Eyring, 1935) as being directly proportional to the ~~quilibrium constant between the ground state and the transition state. In the context of protein folding, fi)r a two-state model, the ground state is thr> folded protein for the unfolding reaction, and the unfolded protein for a refolding reaction. Transition state theory is best applied to the quantification of differences in energy. The kinetics of barnase unfolding can be analysed in terms of transition state theory (Eyring, 1935; Matouschek et al.. 1989. 1990). When the unfolding reaction occurs in the presrncbe of a ligand the following scheme applies:
#
K? F+L
(b) Recombinant
r.zpprrimmts
-t
