ARTICLES
`
`
`28, Takahashi, £. 4 geophys. Res. 91, 9367-9382 (1986).
`29.
`Ito, E. & Takahashi, E. Nature 328, 514-517 (1987).
`30, Herzberg, C. 4 Geophys. Res. 95, 15 779-15 803 (1990).
`31. Herzberg, C. T. Earth planet. Sci. Lett. 67, 249-260 (1983).
`32. White, R. S.
`in Oceanic and Continental Lithosphere, 4 Petrol. spec. Lithosphere issue (ed.
`Menzies, M. A. & Cox, K, G.) 1-10 (1988).
`33.
`Jordan, 8. H.
`in Oceanic and Continental Lithosphere, 4 Petrol spec. Lithosphere issue 11-38
`(1988).
`34. Galer, S. J. G. Earth planet, Sci, Lett. 105, 214-228 (1991).
`35,
`Ito, £., Harris, D. M. & Anderson, A. T. Geochim. cosmochim. Acta 47, 1613-1624 (1983).
`36, Tatsumi, Y. 4 geophys. Res. 94, 4697 4707 (1989),
`37. Davies, J. H. & Stevenson, D. H. 4 geaphys. Res. 97, 2037-2070 (1992).
`38. Peacock, S. M. Phil, Trans, R. Soc, A335, 341-353 (1991).
`39. Green, D. H. Earth planet. Sci, Lett, 19, 37-53 (1973)
`40. Wyllie, P. J. in Magmatic Processes: Physiochemical Principles, Geochem. Soc. spec. publ. 1 (ed.
`Mysen, B. 0.) 107-119 (1987).
`41. Liu, L. Phys, Earth planet Inter. 49, 142-167 (1987).
`42. Kanzaki, M. Phys. Earth planet. inter. 66, 307-319 (1991).
`43.
`Ito, E. & Weidner, D. Geophys. Res, Lett. 13, 464-466 (1986).
`44. Eggler,D.H Carnegie inst Washington Yo. 72, 464-467 (1973).
`45. Hodges, F. N. Carnegie inst. Washington Yb. 72, 495-497 (1973).
`46. Kato, T. & Kumazawa, M. / geophys. Res. 91, 9351-9355 (1986).
`47. Stolper, E.. Walker. D., Hager, B. H. & Hays, J. F. 1 geophys, Res. 86, 6261-6271 (1981).
`48. Yamamoto, K. & Akimoto, S. Am. J Sci. 277, 288-312 (1977).
`49. Liu, L. G. Tectonophys. 164, 41-48 (1989).
`50. Tera, F. et al, Geochim. cosmachim, Acia 50, 1535 1550 (1986).
`51. Kesson, S. £. & Ringwood, A. E. Chem. Geal 78, 83-96 (1989).
`52. Akimoto, S. & Akaogi, M. Phys. Earth planet, Int. 23, 268-275 (1980).
`53. Ringwood, A. E. & Major, A. Earth planet. Sci. Lett, 2, 130-133 (1967).
`54. Ahrens, T. J. Nature 342, 122-123 (1989).
`55. Liu, L. Phys. Earth planet.
`Int. 42, 255-262 (19864).
`56. Luth, R. W. EOS 72, 199 (1991).
`57. Gasparik, T. 4 geophys. Res. 95, 15 751-15 769 (1990).
`58. Meade, C. & Jeanloz, R. Nature 251, 68-72 (1991).
`
`|. Earth planet, Sci Lett. 3, 197-203 (1967).
`59. Kushiro,|, Syona, Y. & Akimoto, S.
`60. Sudo, A. & Tatsumi, Y. Geophys. Res. Lett. 17, 29-32 (1990).
`61. Heubner, J. S. & Papike, J. J. Am. Miner. 55, 1973-1992 (1970)
`62. Trgnnes, R. G, Takahashi, E. & Scarfe, C. M. EOS 69, 1510 (1988).
`63. Trgnnes, R. G. EOS 71, 1587 (1990).
`64. Modreski, P. T. & Boettcher, A. L. Am. J Sei 273, 385-414 (1973).
`65. Foley, S. Geochim. cosmochim. Acta 56, 2689.-2694 (1991).
`66. Hart. R.. Dymond, J. & Hogan, L. Nature 278, 156-159 (1979)
`67. O'Nions, R. K. & Oxburgh, FE. R. Nature 306, 429-431 (1983),
`68. Jochum, K. P., Hofmann,A. W., Ito, E., Seufert, HM. & White, W.M. Nature 306, 431-436 (1983).
`69. Navon,©., Hutcheon, |. D., Rossman, G. R. & Wasserburg, G.
`|. Nature 335, 784-789 (1988).
`70. Menzies, M. A. Nature 335, 769-770 (1988).
`71. Bailey, D. K. Nature 296, 525-530 (1982).
`72, Harlow, G. E, & Veblen, D. R. Science 251, 652-655 (1991).
`73. Bell, D. R. & Rossman, G, R. Science 255, 1391-1396 (1992)
`74, Smyth. J. R.. Bell. D. R. & Rossman, G. R. Nature 351, 732.735 (1991).
`75. Young, T. E.. Green, H. W., Hofmeister, A. M. & Walker, D. EOS 72, 144 (1991).
`76, Clemens,J. D. & Vielzeuf, D. Earth planet. Sci, Lett. 86, 287-306 (1987).
`77. Ryabchikov, |. D. & Boettcher, A. L. Am. Miner. 65, 915-919 (1980i.,
`78. Schneider. M. £. & Eggler, D. H. Geochim. cosmochim. Acta 50, 711-724 (1986)
`79. Massune, H. J. & Schreyer, W. Contr. Miner. Petrol. 96, 212-224 (1987).
`80. Ringwood, A. E. Chem. Geol. 82, 187-207 (1990).
`81. Finger, L. W. & Prewitt, C. T. Geophys. Res. Lett. 16, 1395-1397 (1989).
`82. Finger, L. W., Hazen, R. M. & Prewitt. C. [. Am. Miner, 76, 1-7 (1991).
`83. Kudoh,Y., Finger, L. W., Hazen, R. M,, Prewitt, C. T. & Kanzaki, M. Carnegie Inst. Washington Yb.
`89, 89-91 (1989)
`84. Robie, R. A., Hemingway, B. & Fisher, |. R. US. Geol. Surv. Bull, 1452, (1978).
`85. Helgeson, H. C., Delaney, J. M., Nesbitt, N. W. & Bird, D, K. Am. J Sci. 278-A (1978).
`86. Anderson, D. L. Theory of the Earth (Blackwell, Bostor, 1989).
`
`ACKNOWLEDGEMENTS.| thank M. Kanzaki, C. Herzberg and 7. Gasparik for communications about
`their work, Q. Williams, K. Collerson, G. Ranalli and P. Ulmer, for discussions,J, Holloway, S. Peacock
`and R. Sweeney for comments, B. Buehlmann for final drafting, U. Stidwill for typing and the
`Schweizerische Nationalfonds for financial support.
`
`ARTICLES
`
`The folding of hen lysozymeinvolvespartially
`structured intermediates and multiple
`pathways
`
`Sheena E. Radford’, Christopher M. Dobson’ & Philip A. Evans‘
`* Oxford Centre for Molecular Sciences and Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QR, UK
`+ Cambridge Centre for Molecular Recognition and Department of Biochemistry, University of Cambridge, Tennis Court Road, Cambridge CB? 1QW, UK
`
`
`Analysis of the folding of hen lysozyme showsthat the protein does not becomeorganized in a single
`cooperative event but that different parts of the structure become stabilized with very different
`kinetics. In particular, in most molecules the a-helical domain folds faster than the B-sheet domain.
`Furthermore,different populations of molecules fold by kinetically distinct pathways. Thus, folding
`is not a simple sequential assembly process but involves parallel alternative pathways, some of
`which may involve substantial reorganization steps.
`
`
`UNDERSTANDING how a globular protein folds requires
`detailed characterization of partially organized intermediates
`formed during the folding process. Information about such
`species is becoming available through studies of disulphide
`formation in oxidative refolding'’, protection from hydrogen
`exchange’, the effects of amino-acid substitutions’”’, stable
`partially folded states'°-'? and peptide analogues of folding
`intermediates'*’*. Nonetheless, no clear unifying view of the
`nature of folding mechanismshas yet emerged and uncertainties
`remain over several fundamental issues.
`Here we describe a detailed study of the refolding of hen
`lysozyme. The thermodynamics and kinetics of folding of this
`protein under equilibrium conditions accord well with a cooper-
`ative two-state model'®!’. Under conditions far from equili-
`brium, however, refolding occurs in at least two stages, acquisi-
`tion of a native-like far ultraviolet (UV) circular dichroism (CD)
`302
`
`spectrum preceding formation of the tertiary structure, as
`monitored by near-UV CD'*'’, In addition, a very fast phase
`has been detected by tryptophan absorption measurements””.
`These results suggest transient population of partially folded
`intermediates but
`their detailed interpretation requires com-
`plementary information about
`the behaviour of individual
`residues, such as can be obtained from hydrogen-exchange
`labelling techniques**?'. Using a variantof these experiments”*,
`based on competition between the exchange and folding pro-
`cesses, we have recently confirmed the existence of folding
`intermediates for hen lysozyme and demonstrated that the two
`structural lobes that characterize the native structure are also
`distinct folding domains”. Here we describe pulsed hydrogen-
`exchange labelling experiments, using nearly half of the 126
`amide hydrogens in the molecule as probes of refolding,
`and stopped-flow CD measurements in both far- and near-UV
`MATURE APHBRERIE049
`© 1992 Nature Publishing Group
`  
`


` 
`Apotex Inc. et al. v. Amgen Inc. et al., IPR2016-01542
`Pace 1
`
`Amgen Exhibit 2049
`Apotex Inc. et al. v. Amgen Inc. et al., IPR2016-01542
`Page 1
`
`

`

`ARTICLES
`
`
`spectral regions. The combination of these data has allowed us
`to construct a detailed analysis of events during refolding.
`
`Evidence for partially folded intermediates
`Figure 1 shows the results of pulsed labelling experiments
`obtained for individual amides, grouped according to their
`distribution in the native state secondary structure. If folding
`occurred by a simple two-state mechanism all the curves would
`be coincident. The fact that this is not the case demonstrates
`unequivocally that partially structured intermediates are popu-
`lated during folding.
`The different labelling curves are not only distinct for the
`different sets of amides, they are also not monophasic. Each
`curve was modelled well by a sum of two exponentials (Fig. 1
`and Table 1). The rates of the fast phases show no very clear
`pattern (time constant (r)=7+4ms) but those of the slower
`phases fall qualitatively into two groups, differing in their
`average time constant by a factor of about 4. The more rapidly
`protected group (7<100 ms) comprises amides in the four a-
`helical segments, the 3'° helix close to the C terminus of the
`protein, and three amides, Trp 63, Cys 64 and Ile 78, which lie
`in the loop region in the native enzyme. With the exception of
`the last group, these structural elements all occur in one of the
`two lobes of the native conformation, which weshall call the
`a-domain (Fig. 2). In contrast, amides protected more slowly
`
`(r> 100 ms) are located, with the single exception of Asn 27,
`in the other structural domain, which we designate the B-
`domain, comprising a short double-stranded and a longertriple-
`stranded @ sheet, a 3'° helix and a long loop.
`The samequalitative distinction also extends to the amplitudes
`of the kinetic phases. For amidesin the a-domain,the fast phase
`constitutes some 40+ 12%of the total whereas in the B-domain
`the corresponding amplitude is only 24+6%. This compounds
`the difference in kinetics of the second phase to ensure that
`protection of the 6-domain is substantially slower overall, thus
`accountingfor the pattern of labelling observed when the refold-
`ing of lysozyme was examined by the competition method’’.
`
`Tertiary interactions and folding domains
`The similar protection curves for nearly all amides in the a-
`domain could, in principle, mean that individual helices form
`as persistent
`structures
`independently but simultaneously.
`However, the coincidence of both kinetic phases makes this
`seem unlikely. A more plausible explanation is that the protec-
`tion from exchange monitorsstabilization of the rapidly formed
`helices by side-chain interactions, and hence that these amide
`probes actually monitor formation of the domain core, albeit
`perhaps in an embryonic form. This is consistent with the
`protection of two amides not directly invalved in secondary
`structure, Leu 17 and Tyr 23 whichlie in the loop linking helices
`
`100
`80
`60
`
`20
`
`0
`
`0
`
`Helix A
`
`40
`
`Helix B
`
`
`
`
`
`Proton(%)occupancy
`
`FIG. 1 Time courses for the protection
`of amides from exchange during the
`refolding of hen lysozyme. The curve
`drawn represents the average of a two
`exponential fit to the time course for
`individual amides in each native secon-
`dary structural element.In the case of
`the loop region,
`two averages are
`drawn, one for Trp 63, Cys 64 andlle
`78 and another for the remainder.
`METHODS. Experiments were done at
`20 °C using a Biologic QFM5 rapid mix-
`ing quench flow apparatus. Lysozyme
`(20 mg mi~*) wasinitially dissolved in
`6M guanidine deuterochloride in D0
`at pH 6.0, leading to complete denatur-
`ation and substitution of all exchange-
`able hydrogens by deuterium. Refolding
`wasinitiated by dilution of this solution
`10-fold into 20 mM sodium acetate, pH
`5.5 in HO. At the resulting pH of 5.2
`the half
`life for amide exchange is
`about 16 s (ref. 32) so that negligible
`labelling occurred during this phase.
`After variable refolding times (3.5-
`2,000 ms)
`the solution was diluted
`again with a volume 5 times that of
`the initial protein solution of 0.2M
`sodium borate, pH 10.0. This initiated
`labelling at a pH of 9.5. After 8.4ms
`the labelling pulse was terminated by
`further dilution, with a volume again 5
`times that of the initial volume of pro-
`tein solution, of 0.5M acetic acid in H,0. The final pH was about 4.0, at
`which exchange in the native protein of the 49 amides studied is very
`slow®, A sixfold lower protein concentration during refolding had no effect
`on the observed rates of protection, confirming that intermolecular associ-
`ations,
`if present, are not kinetically significant. Protein samples were
`concentrated and the buffer exchanged for 40 mM deuterated sodium ace-
`tate, pH 3.8 in D0 byultrafiltration at 4 °C. For the zero time point (100%
`labelling) a solution of lysozymein the same buffer and isotopic composition
`as in the labelling phase of the refolding experiment was heated to 80°C
`for 10 min, causing complete exchange of all amides by reversible unfolding.
`The sample was then treated identically as for the other time points. This
`made correction for the residual deuterium content of the labelling solution
`unnecessary. A phase-sensitive COSY spectrum of each sample was recor-
`NATURE - VOL 358 - 23 JULY 1992
`
`200
`
`Helix D
`
`8 sheet 120
`
`160
`
`80
`60
`40
`20
`
`D
`
`0
`
` Trp 111 indole
`
`40
`
`°
`120
`80
`Time (ms)
`
`160
`
`200
`
`~
`
`ded on a 500 MHz GE/Nicolet spectrometer at 35 °C. 256 increments over
`1,024 complex points and 32 transients were collected. Data were processed
`using the FTNMR program of Dennis Hare (Hare Research Inc.) on a SUN
`computer. Final digital resolution was 1.7 Hz per point. Acquisition and
`processing of each spectrum wasidentical. The intensities of the CaH-NH
`cross-peaks in each spectrum were taken as the sum of the absolute values
`of the four phase-sensitive components. These were normalized using the
`sum of the eight phase-sensitive peaks associated with the Tyr 23 and Tyr
`53 CdH-CeH cross-peaks. Proton occupancies at individual amide sites
`were calculated relative to those of the unfolded control sample, for which
`COSY cross-pea intensities were taken to correspond to 100% proton
`occupancy.
`
`
`
`
`annAaenS
`
`120
`
`160
`
`Helix C
`
`80
`
`129
`
`160
`
`© 1992 Nature Publishing Group
`  
`


` 
`
`303
`
`Page 2
`
`Page 2
`
`

`

`ARTICLES
`
`
`TABLE 1 Protection of amide hydrogens during the refolding of lysozyme
`
`Fast phase
`Time
`constant Amplitude
`(ms)
`(%)
`3.90
`38.3
`4.41
`50.6
`3.39
`415
`2.61
`51.5
`6.33
`44.0
`
`Slow phase
`Time
`constant
`(ms)
`69.8
`64.3
`54.1
`53.4
`79.4
`
`Amplitude
`(%)
`47.2
`39.5
`43.6
`33.9
`44.5
`
`43.6
`48.5
`66.3
`42.1
`58.3
`36.9
`40.3
`24.5
`
`19.4
`16.6
`18.2
`
`22.2
`26.5
`14.2
`24.0
`33.4
`25.6
`17.7
`178
`
`3.09
`7.76
`12.7
`483
`2.40
`4.39
`8.00
`6.68
`
`114.5
`11.2
`9.33
`
`7.14
`8.76
`6.29
`10.6
`9.29
`8.80
`15.7
`168
`
`39.7
`35.9
`22.5
`42.1
`28.1
`48.6
`45.5
`53.5
`
`56.8
`69.7
`56.1
`
`59.6
`66.0
`48.2
`40.4
`52.7
`62.4
`63.7
`67.4
`
`89.5
`83.3
`492
`59.7
`59.4
`68.7
`97.7
`53.1
`
`475
`350
`296
`
`451
`534
`151
`188
`238
`349
`375
`392
`
`Structural
`context
`Helix A
`
`Irregular
`
`Helix B
`
`irregular
`
`Small 6-sheet
`Large B-sheet
`
`Residue
`61
`63
`64
`65
`75
`76
`78
`
`82
`83
`84
`9?
`93
`94
`95
`96
`97
`99
`108
`111
`112
`115
`
`123
`124
`125
`
`Fast phase
`Time
`constant Amplitude
`(ms)
`(%)
`9.29
`58.5
`2.45
`25.4
`3.67
`37.4
`15.7
`447
`23.5
`63.8
`8.60
`54.2
`1.87
`21.0
`
`Slow phase
`Time
`constant
`(ms)
`247
`63.9
`60.4
`241
`354
`215
`48.8
`
`Amplitude
`(%)
`17.5
`60.0
`43.2
`39.7
`23.5
`25.4
`59.2
`
`Structural
`context
`Loop(irregular)
`
`26.2
`29.0
`26.1
`29.6
`34.9
`22.2
`288
`19.4
`32.7
`28.1
`34.9
`48.0
`28.2
`44,7
`
`46.7
`67.0
`43.1
`
`8.56
`9.53
`9.81
`3.54
`9.41
`4.93
`3.29
`1.43
`7,08
`2.07
`3.73
`3.89
`3.04
`8.13
`
`3.42
`2.00
`4.95
`
`65.6
`52.5
`48.3
`56.6
`35.5
`57.7
`61.0
`748
`SLT
`56.4
`47.5
`38.1
`479
`42.7
`
`35.0
`17.8
`40.4
`
`37° Helix
`
`Helix C
`
`Helix D
`
`31° Helix
`
`788
`282
`262
`528
`76.0
`64.0
`63.5
`53.5
`61.6
`40.5
`87.5
`68.4
`49.8
`51.6
`
`65.2
`33.3
`108
`
`Residue
`8
`10
`11
`12
`13
`
`17
`23
`27
`28
`29
`31
`34
`36
`
`37
`38
`39
`
`40
`42
`44
`50
`52
`53
`56
`58
`
`The time courses of change in proton occupancies were fitted to the sum of two exponentials of the form y=A e'-49 +B ee) +0 (where A and B
`are the fractional amplitudes of the two phases and K, and K, are their rate constants; C is the apparent fractional amplitude of a third phase too slow
`to be followed in our experiments). The last column gives the secondary structural context of each amide in the native state.
`
`A and B, with kinetics indistinguishable from those of amides
`in the helices (Table 1).
`It is clear that the a-domain acquires a stabilizing core when,
`in most molecules, the B-domain has not becomefixed in a
`stable conformation. To investigate further the structure of the
`a-domain at this stage we have studied the protection of tryp-
`tophan indole NHs, two of which exchange slowly enough in
`the native enzyme to be used as probes of refolding’. The
`kinetics of protection of Trp 111 resemble closely those of the
`main chain amides of the w-domain (Fig. 1). In the native
`structure this residue is in the D-helix and the side-chain NH
`is hydrogen bonded to the side-chain oxygen of Asn 27, in the
`B helix™*. Thus, its protection may monitor ‘docking’ of these
`two helices; at the very least it suggests that the tryptophan
`side-chain becomes excluded from solvent in a well developed
`hydrophobic core on this time scale. Similar behaviour was
`
`observed qualitatively for the indole NH of Trp 28 whichis also
`buried in the core of the a-domain.
`The amide NH of Asn 27, which forms a non-helical hydro-
`gen-bondto the carbonyl oxygen of Tyr 23 in the native structure,
`is the one marked exception to the general pattern of behaviour
`in the a-domain. Its protection kinetics resemble those of amides
`in the B-domain, with a slow phase time constant greater than
`400 ms. This interaction may thus become fixed only late in
`folding, when both domains are organized. Therefore, although
`in most molecules the a-domain folds independently, at least
`somedetails of its tertiary interactions form more slowly.
`Within the 6-domainthe situation is more complicated. With
`the exceptions of Trp 63, Cys 64 and Ile 78, there is a clear
`qualitative distinction betweenthe protection kinetics of amides
`in this domain and in the a-domain but the slow phase time
`constants tor individual amides vary between 150 and nearly
`
`f
`
`FIG. 2, Schematic view of the native structure of hen lysozyme. Elements
`of secondary structure in the a-domain are coloured red, as are three
`amides protected with similar kinetics, Trp 63, Cys 64 andlle 78. Elements
`of secondary structure that are in the B-domain are coloured blue. White
`regions represent mainly surface amides for which there is no information.
`The four a helices (A-D) and the two 3?° helices are labelled. The diagram
`was produced using the program MolScript*?.
`
`© 1992 Nature Publishing Group
`  
`


` 
`
`NATURE - VOL 358 - 23 JULY 1992
`
`Page 3
`
`Page 3
`
`

`

`ARTICLES
`
`
`100
`d80
`60
`40
`
`oi
`
`0
`8.5
`
`108 ms Refolding time
`« Domain
`
`|
`
`9
`
`4
`
`ti!
`
`95
`
`10
`
`105
`
`6 Domain
`
`FIG. 3 The pH dependenceof the extent of labelling during an 8.4 ms pulse
`commencing 14 ms and 108 msafter the initiation of folding. Exchange in
`proteins usually follows EX2 kinetics,
`in which the intrinsic rate of the
`chemical exchange step is slow compared with the rate of the structural
`fluctuations which expose the amide transiently to the solvent***®. In this
`situation,if the partial protection achieved in the fast phasereflects a single
`population of a marginally stable intermediate, then the extent of labelling
`would be predicted to increase substantially as the pH of the labelling pulse
`is raised?”**, If, by contrast, the mechanism of exchange is EX1, where the
`overall rate is limited by the rate of the fluctuations that expose the amide
`to solvent water, the pH dependenceof the extent of labelling, if any, would
`not be predictable in a straightforward manner. This ambiguity can be
`resolved by increasing the duration, rather than the pH, of the labelling
`pulse°®. This would result in increased labelling of any marginally stable
`intermediates regardless of the mechanism of exchange.In this figure the
`average proton occupancies for amide hydrogens in the a-domain (a, d),
`the B-domain (b, e) and the subset containing Trp 63, Cys 64 andlle 78
`(c, f) are shown as a function of the pulse pH. Vertical bars represent one
`standard deviation from the average plotted. The pH of the labelling pulse
`was varied between 9.0 and 10.4 by changing the pH of the exchange buffer.
`For these experiments the quench buffer was 1 M acetic acid in H2O solution.
`Acquisition, processing and analysis of NMR spectra were done as described
`in the legend to Fig. 1. Control experiments in which a sample of the native
`protein (90% deuterated at amide sites) was exposed to the different
`labetling pulses showed no increase in proton occupancy at the 49 sites
`considered here, confirming that these amides are stable to exchange in
`the native enzyme under these conditions.
`
`11 ms Refolding time
`o« Domain
`
`9
`
`95
`
`10
`
`10.6
`
`8.5
`
`100
`
`8 Domain
`
`100
`
`9
`
`9.5
`
`10
`
`10.5
`
`8.5
`
`9
`
`9.5
`
`10
`
`10.5
`
`Tro63, Cys64, 1le76
`
`
`
`
`Protonoccupancy(%) 85
`Trp63, Cys64, lle78
`
`pH
`
`800 ms (Table 1). This implies that the stabilizing tertiary interac-
`tions in this domain are not formed in a single cooperative step
`but that a numberofindividual! or sequential assembly processes
`are involved.
`Protection of the three amides, Trp 63, Cys 64 and He 78,
`whichlie in the long loop at the interface between the a- and
`B-domains cannot berationalized by examination of the native
`structure; Cys 64 is not hydrogen bonded and Ile 78 forms only
`a long tertiary hydrogen-bond. Thus, their protection cannot
`be associated with native secondary structure but, presumably,
`with exclusion from solvent by surrounding side-chains”*”*.
`These interactions need not be native-like. There are two disul-
`phide bridges close by (64-80 and 76-94), the latter linking the
`a- and 8-domains. Early clustering of these residues in cooper-
`ation with formation of the a-domain core could explain the
`similarity in their protection kinetics.
`
`Multiplicity of events in folding
`There are two fundamentally different possible explanations for
`biphasic kinetics in the labelling experiment:
`the fast phase
`leads either to partial protection of a particular amide in all
`molecules or to complete protection but in only a proportion
`of molecules. The first case implies formation of a marginally
`stable intermediate, followed by slower transformation into a
`stable, presumably native-like, structure. The second possibility
`involves parallel pathways such that a proportion of the protein
`folds rapidly to a structure that completely protects the amide
`concerned from exchange, whereas the remainder follows a
`route in which protective structure is acquired more slowly.
`These possibilities can be distinguished by varying the dur-
`ation and the intensity of the labelling pulse**?’. The results
`(Fig. 3) indicate that the labelling of the majority of amides in
`both domains is essentially independent of the pulse pH. In
`addition, no significant change in the extent of labelling occurs
`when the pulse length is doubled or trebled (at refolding times
`of 22 and 108 ms, respectively). These results are not compatible
`with a simple sequential folding mechanism;
`there must be
`parallel
`folding pathways, some leading to protection of
`individual amides more rapidly than others.
`The only exceptionsto the observed insensitivity to the labell-
`NATURE - VOL 358 - 23 JULY 1992
`
`ing conditions are the amides of Trp 63, Cys 64 and Ile 78 (Fig.
`3). This suggests that rapidly formed structure protects these
`amides only partially, full protection being acquired in the
`slower phase. However, the sensitivity to the pulse conditions
`is weaker than expected for a single marginally stable species.
`Instead, there must be a heterogeneous population, with varying
`levels of protection.
`
`Native and non-native structure
`The sequence of events detected in the far-UV CD (225 nm)is
`characterized by at least three kinetic phases (Fig. 4). A large
`negative ellipticity is acquired within the dead time of the
`experiment (2 ms), so that at the earliest measurable time its
`magnitude is close to that observed for the native enzyme. The
`CD at this wavelength is generally dominated by peptide groups
`in helical structure, suggesting that an average helix content
`close to that of the native enzyme is achieved in this time.
`Contributions to the CD from aromatic groups and disulphides
`cannot, however, be ruled out”.
`There follows a remarkable further developmentofthe dichro-
`ism at 225 nm, so that the average negative ellipticity becomes
`greater than that of the native state, reaching a maximum at
`around 80 ms. The kinetics of this phase do not coincide closely
`with any of the phases of exchange protection we have observed.
`It seemslikely that the excursion in the CD reflects the transient
`formation of some kind of non-native interactions although we
`cannotrule out the possibility that it arises from asynchronous
`development of effects that tend to cancel out in the native
`protein spectrum. Return of the ellipticity to its value in the
`native state then occurs in a third phase (7 ~ 300 ms) which is
`much slowerthan protection of a-helical amides but is compar-
`able to the slow phase of protection in the B-domain.
`CD experiments were also done in the near-UV region. The
`latter monitors aromatic groups in specific orientations fixed
`through tertiary interactions, so that spectra of denatured pro-
`teins, even relatively compact ‘molten globule’states, are charac-
`terized by loss of nearly all intensity at these wavelengths'””!®.
`The time developmentat 289 nm is quite different from that in
`the far-UV (Fig. 4). There is virtually no change within the dead
`time of the experimentandthe entire kinetic amplitude can be
`305
`
`© 1992 Nature Publishing Group
`  
`


` 
`
`Page 4
`
`Page 4
`
`

`

`(%)
`
`TTT TT TT 1 TTT Tt T TyTrt
`0
`500
`1,000
`1,500
`2,000
`2,500
`
`Time (ms)
`
`FIG. 4 Refolding kinetics of lysozyme as monitored by stopped flow CD at
`a 225 and b, 289 nm. The refolding conditions were as described in the
`legend to Fig. 1, except that the final folding protein concentrations were
`0.2 and 1.0mg mi’, respectively. The experiments were done using a
`Jobin-Yvon CD6circular dichrograph equipped with a Biologic SFM3 stopped-
`flow module. The dead time was about 2 ms. In eact case the data have
`beenfitted to a sum of two exponentials. The insets shaw an expansion of
`the plots during the first 400 msof folding.
`
`fitted to a sum of two exponentials, the amplitudes and time
`constants being 32+2%,
`r=10+2ms,
`and 68+2%,
`r=
`28548 ms. These parameters resemble those observed for pro-
`tection of amides in the B-domain (Table 1).
`The dichroism at 289 nm is associated principally with the
`six tryptophan residues of lysozyme”. Of these, residues 62 and
`63 are in an irregularly structured part of the B-domain, close
`to the domain interface, and the others are all in the a-domain.
`This is interesting in view of the apparent synchrony of their
`forming fixed tertiary contacts and the protection of amides in
`the B-, rather than the a-domain. Thus, although the a-domain
`can fold more rapidly than, and therefore independently of, the
`B-domain, the results suggest thatits tertiary structure does not
`become fully organized until cooperative interactions spanning
`the two domains are established.
`
`Folding pathways
`Theearliest event observed in refolding is the acquisition within
`2 msof a far-UV CD of similar intensity to that of the native
`protein. This almost certainly reflects formation of a large
`amount of a@-helical structure, at least some of it presumably
`native-like, yet no protection from hydrogen exchange is
`apparentat this stage (Fig. 5). This suggests that although the
`306
`
`0
`
`11
`
`108
`Time (ms)
`
`500
`
`1,000
`
`FiG. 5 Summary of events during the refolding of lysozyme. Formation of
`the native enzymeis virtually complete in 2 s (ignoring the very slow phase
`that was not recorded in our experiments) and each parameter is therefore
`assigned a value of 100% at this point. The curves shownare thefits to
`the near (289 nm;solid line) and the far (225 nm; dashed line) UV CD data
`and protection of the @- (dotted line) and the 8-domains (dashed and dotted
`line), respectively. The latter two curves represent the averages of two
`exponential fits to the time courses for individual amides in each folding
`domain. For various different folding times the magnitude of each parameter
`as a percentage of that observedin the native protein is also shown: CDa,6,
`Ellipticity at 225 nm; 3°, ellipticity at 289 nm; a, protection of amides in the
`a-domain; 8, protection of amides in the B-domain.
`
`average distribution of backbone conformations has become far
`from random, specific structural elements remain extremely
`labile. Similar behaviour has been observed in otherearly folding
`intermediates*® and denatured states****". This result is con-
`sistent with the near-UV CD which detects no specific interac-
`tions involving aromatic chromophoresat this time. A change
`in the environment of these groups on this time scale has,
`however, been observed by stopped-flow absorption studies'*”,
`suggesting that these rapid folding events include somekind of
`hydrophobic collapse as well as helix formation.
`In the next phase there is rapid protection of a-domain amides
`in about 50%, and of B-domain amides in about 30%, of
`molecules. The protection data cannot reveal directly whether
`8-domain assembly occurs independently or in conjunction with
`that of the a-domain in individual molecules in this phase.
`NATURE - VOL 358 - 23 JULY 1992
`
`© 1992 Nature Publishing Group
`  
`


` 
`
`Page 5
`
`ARTICLES
`
`
`a
`
`0
`
`
`=
`
`-2
`
`el
`3
`mH
`oD
`_
`go
`Q
`
`-3
`
`-
`es
`FS
`{
`2S
`7 aoe
`= 0
`100
`200
`300
`400
`“a
`Time (rns)
`
`
`
`
` Time (ms}
`
`
`
`Nativestructure
`
`— -4
`oO
`
`1-5
`
`= &
`
`b
`
`ul
`&Oo
`
`1= a
`
`0
`
`500
`
`1,000
`
`1,500
`
`2,000
`
`2,500
`
`6 7
`
`
`
`i
`‘
`ie
`
`Page 5
`
`

`

`ARTICLES
`
`
`extent independently, in most of the population the conforma-
`However, the presence of a phase with a similar time constant
`tion of the remainder of the molecule, though not necessarily
`in the near-UV CD which, as discussed above, may reflect the
`wholly disordered, is far from native-like until it becomesfixed
`formation of cooperative structure spanning both domains, sug-
`in the cooperative tertiary structure.
`gests that roughly synchronous structure formation in both
`A major unresolved issue is why different molecules fold by
`domains occurs in at least a substantial proportion of these
`molecules.
`kinetically distinct pathways. This could reflect heterogeneity
`of the unfolded state?°**?*; for example native-like X-Pro pep-
`The remainder of the population folds more slowly and the
`tide bond isomers may be required for folding??”***”. It is
`greater amplitudeof the fast phase and higherrate of the slower
`possible that this could indeed accountfor the lack of protection
`phase of protection for a-domain amides mean that in most
`in around 17% of molecules even after 2s of refolding, but
`molecules this domain becomes substantially folded before for-
`neither the rates nor amplitudes of the distinct phases resolved
`mation of a stable B-domain. In the resulting intermediates the
`in our experiments are consistent with proline isomerization.
`pH dependenceof labelling shows that for most amides in the
`The multiple pathways could reflect other heterogeneity in the
`a-domain the protection factors are in excess of 500 whereas
`denatured state or they mayarise from aninitial rapid collapse
`those in the B-domain are less than 20. However, the slower
`of the protein, leading to various intermediates, some amenable
`development of near-UV CD intensity and of protection ofat
`to further folding whereas others needfirst to reorganize, poten-
`least one amide, Asn 27, in the a-domain suggest that it is not
`tially a relatively slow process in a partially condensed state. A
`fully native-like but has some characteristics of a molten globule
`similar observation of parallel refolding pathways has been
`at this intermediate stage. The pattern of protection has some
`made for cytochromec (ref. 40).
`resemblance to that observed in the stable molten globule state
`The four disulphide bridges persist in the denatured state and
`of the homologous protein guinea-pig a-lactalbumin. In that
`might havea significant influence on the refolding mechanism,
`case, fewer than 20 amides have protection factors greater than
`either through their isomerism in the denatured state or in
`10; again these occurin the a-, rather than the 6-domain (C.-L.
`intermediates, or through constraints they place on conforma-
`Chyan, C. Wormald, C.M.D., P.A.E. and J. Baum, manuscript
`tional freedom during refolding. In @-lactalbumin persistence
`in preparation). Similar, marginal protection is also characteris-
`of substantial residual structure in its molten globule state is
`tic of other stable partially folded proteins''’**, suggesting that
`consistent with a number of different disulphide pairings“,
`these are, in general, more labile structures than the kinetic
`suggesting that at least some intermediates can form indepen-
`in