http://www.diva-portal.org
Postprint
This is the accepted version of a paper published in IUBMB Life - A Journal of the International Union of Biochemistry and Molecular Biology. This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Citation for the original published paper (version of record):
Gianni, S., Jemth, P. (2014)
Conserved nucleation sites reinforce the significance of phi analysis in proteinfolding studies.
IUBMB Life - A Journal of the International Union of Biochemistry and Molecular Biology, 66(7):
449-452
http://dx.doi.org/10.1002/iub.1287
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Permanent link to this version:
http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-228447
Conserved nucleation sites reinforce the significance of phi analysis in protein folding studies
Stefano Gianni1,2 and Per Jemth3
1Istituto Pasteur - Fondazione Cenci Bolognetti, Dipartimento di Scienze Biochimiche
“A. Rossi Fanelli” and Istituto di Biologia e Patologia Molecolari del CNR, Università di Roma “La Sapienza”, P.le A. Moro 5, 00185, Rome, Italy.
2Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
3Department of Medical Biochemistry and Microbiology, Uppsala University, BMC Box 582, SE-75123 Uppsala, Sweden.
Running title: Assessing the robustness of Φ value analysis
Correspondence: Stefano.Gianni@uniroma1.it; Per.Jemth@imbim.uu.se
Summary
The only experimental strategy to address the structure of folding transition states, the so-called Φ value analysis, relies on the synergy between site directed mutagenesis and the measurement of reaction kinetics. Despite its importance, the Φ value analysis has been often criticized and its power to pinpoint structural information has been questioned. In this Hypothesis we demonstrate that comparing the Φ values between proteins not only allows highlighting the robustness of folding pathways, but also provides per se a strong validation of the method.
Keywords: Protein Folding, kinetics, mutagenesis, homologous proteins
Introduction
In 1975 Francis Crick stated, "it is very difficult to conceive of a scientific problem that would not be solved in the coming twenty years … except for a model of brain function and protein folding" (1). Forty years after this statement, we may conclude that Crick was very right. The major complication of both scientific problems lies in involving an extremely high number of energetically coupled interactions in a three- dimensional space. To solve the protein folding problem, for example, would imply to offer a detailed depiction of the whole pathway driving the polypeptide chain to its native conformation. Yet, experiments show that the interactions stabilizing folded states form in such a co-operative manner that, very often, only the fully native and fully denatured state may be experimentally accessible. Such a two-state mechanism represent, by definition, an intrinsic complication in studying protein folding – despite the aim is to depict a complex pathway, only the starting and ending point of the reaction can be accessed at equilibrium.
Protein folding and Φ value analysis
From the premises described in the introduction, it is easy to understand how one of the major breakthroughs in protein folding studies was obtained with the introduction in the late 80's of an experimental technique that can infer the structure of the folding transition state (2, 3). By systematically mutating side chains while probing the effect of mutation on the activation barrier and native state, it is possible to map out interaction patterns in the transition state. In fact, mutations that destabilise the transition state (slowing down the folding reaction) target contacts that are formed in
its structure. Quantitatively, the strength of the contacts is measured by the Φ value, which normalises the stability loss of the transition state to that of the native state, relative to the denatured state (4). A Φ value analysis must take into account the following points: (i) Choice of mutations. Conservative deletion mutations are associated with fewer caveats in the analysis (4, 5); (ii) Choice of experimental conditions. The stability of the protein should be adjusted by for example pH such that the observed kinetics of wild type and mutant variants are within an experimentally easily accessible window; (iii) Mutants displaying too low or too high changes in unfolding free energy on mutation should not be included in the analysis.
Once these criteria are fulfilled the quality of data in the Φ value analysis is usually good.
Ever since its introduction more than twenty years ago, the Φ value analysis has represented a key methodology in protein folding but yet, in parallel, been challenged by several authors as being an improper way of characterizing the transition state for the folding reaction. In particular, a potential problem relating to the sensitivity and accuracy of the Φ value analysis was brought up. It was argued that the most conservative mutation, deletion of a methylene group, yielded a change in free energy on mutation, ΔΔGD-N, that was too small relative to the error in the measurement to give a reliable Φ value. More drastic mutations would give significant changes in ΔΔGD-N, but the resulting Φ value would not report on a specific interaction but only return an average value close to 0.3 reflecting formation of several distinct intramolecular contacts (6). Additionally, it was argued that Φ value analysis could suffer from major effects resulting from changes in residual structure in the denatured state upon mutation, which would jeopardize the analysis (7). Another potential
problem is that mutation may change the folding pathway. To resolve this problem an alternative strategy was introduced, the so-called psi-value analysis (8). Strikingly, the conservative mutations generally employed in Φ value analysis (such as Ile to Val or Thr to Ser) were considered more drastic than replacing two adjacent side-chains with a histidine pair and conducting experiments in the presence of various concentrations of divalent cations. Most recently, Φ values were reassessed on a large database of different mutants and it was concluded to be an average scatter around a certain value (0.3) implying, similarly to the earlier criticism, that “the structural information in conventional Φ values is low” (9). Most caveats of the method were discussed in the original work (4) and later criticism were rebutted by Fersht and coworkers (5), but yet we note that some authors tend to still criticize this methodology, for example by reporting the reproducibility between different laboratories (10). Another source of debate has been represented by the so-called unusual Φ values, i.e. those positions for which the Φ value is below zero or higher than 1. Plausible models accounting for these relatively rare cases (typically below 5% of the measured values) are discussed by Weikl and Dill (11).
Φ-Φ plots demonstrate that Φ value analysis is robust
A very powerful method to compare the folding pathway of two different proteins sharing a similar structure is to make a plot of Φ versus Φ for homologous structural positions (12). Such plots can pinpoint similarities and differences of folding transition states within a protein family or other homologous proteins variants. The Φ- Φ plots demand a rigorous structural alignment between the two proteins considered.
In some cases this could be trivial and easily supported by the analysis of the primary
structures (for example in the cases of circular permutants and circularized proteins), whereas in other cases (homologous proteins with reasonable sequence identity, say
>50%) detailed analyses of aligned crystal or NMR structures must be performed.
Figure 1 shows plots of Φ versus Φ for different cases. In particular, we report a comparison between the folding pathway of proteins belonging to the same super- family but sharing a different sequence (Fig 1A) (12, 13); a protein and its circular permutant (Fig 1B) (14) or circularized variant (Fig 1C) (15); a protein characterized by different optical probes (Fig 1D) (16) or in the presence/absence of a covalently bound contiguous domain (Fig 1E) (17). Remarkably, in all these cases, it is possible to observe a clear correlation between Φ values indicating that folding pathways are robust to the chemical composition of the sequence (Fig 1A), sequence connectivity (Fig 1B and 1C), optical probes (Fig 1D) and that individual domains fold with a conserved mechanism when in isolation or in the presence of multi-domain proteins (Fig. 1E).
In this hypothesis we wish to turn the argument around to address a simple question:
if there is little structural information in Φ values, what would be the probability to obtain the correlations highlighted in Figure 1? Indeed, we note that, given the diversity of the systems considered, the plot in Figure 1 would return a linear correlation only if i) there is a conservation of the folding pathway between the different pairs considered and ii), most importantly, Φ values contain relevant structural information. The reason is that mutations represent a perturbation of the structure of the protein, by means of specific deletions of bonds. Thus, because mutations of corresponding positions in homologous proteins (protein family
members, circular permutants, circularized proteins) return a similar value of Φ, it follows that the structural perturbations have a similar effect on the folding pathway, which is reflected in its Φ value.
To address quantitatively this point we computed the p-value for each of the individual cases shown in Figure 1 and reported the values in the corresponding panel. It is evident that in all cases, it is nearly impossible for a correlation to arise by chance. Furthermore, because each of the Φ values corresponds to an independent measurement, the probability for all these events to return a linear correlation would be equal to the product of their individual probabilities. Thus, there is a probability of 10-17 that the Φ values in Fig. 1 contain low structural information and are significantly affected by any of the suggested flaws. The Φ-Φ plots thus address major issues raised against Φ value analysis, in particular the error in the measurement due to either direct experimental error (10) or a small perturbation (e.g., Ala to Gly) resulting in a low ΔΔGD-N value on mutation (6), both of which would impede a good correlation. The validity of Φ values calculated from small perturbations is important since these are much more informative (probing local structure) than a large perturbation (e.g., Phe to Ala), which is likely to probe a large region of the protein and hence report an average Φ value for several interactions.
Further detailed analyses on the errors in Φ analysis have been previously published (5, 16).
Parallel folding pathways
Another issue, which complicates Φ value analysis is the notion that fractional Φ values may arise from two different scenarios, either a homogeneous transition state
with partial bond formation (as generally assumed) or heterogeneous transition states arising from parallel pathways in which a given residue is either structured or unstructured. This difficult point was addressed experimentally by Fersht and co- workers by analyzing the so-called Brønsted or Leffler plots (ΔΔGD-‡ versusΔΔGD-N) (18). It was shown that linearity in the Brønsted plot represents a signature for partial structure formation in homogeneous transition states, with parallel pathways characterized by more complex scenarios. The hypothesis was originally tested for CI2 (18, 19) and appears to be a rather general feature in the folding of globular proteins.
While the Φ-Φ plot does not directly address these points, it clearly demonstrates conservation or divergence of folding pathways as discussed in the next section.
Φ-Φ plot as a tool to analyze folding pathways
As demonstrated in this hypothesis and before, Φ value analysis indeed provides information about the transition state of the folding reaction. We introduced the Φ-Φ plot as a tool to investigate the folding pathways in the PDZ domain family of proteins and found a clear conservation of nucleation sites for a late but not an early transition state (12) (Fig 1A). Indeed, even drastic modifications such as circular permutation and circularization show conserved nucleation sites for the late transition state in the folding reaction (Fig. 1B and C). In fact, in the PDZ domain family, whilst the native state topology defines essentially in a unique way the late stages of folding, it leaves significant freedom to the early events, a result that reflects the funnelling of the free energy landscape towards the native state (14).
Φ-Φ plots can be used whenever structurally homologous proteins are investigated (20) to give a valuable overview of the whole data set, whether or not there is a correlation. For example, three Φ value analyses of members of the small fast-folding alpha helical peripheral subunit-binding domain family have been reported (21–23).
Φ-Φ plot analysis shows that there is a correlation for two of the members (E3BD and POB) but not for the third one (BBL). Such lack of correlation suggests a diversity in folding pathways, which in the case of the peripheral subunit-binding domain family was attributed to a stronger helix propensity in helix 1 of BBL as compared to E3BD and POB (23). The extensive Φ value analyses on the homologous R15, R16 and R17 domains of α-spectrin by Clarke and co-workers also show low conservation of folding nuclei (24). They used Φ-Φ plots to compare a core-swapped mutant with the wild type domains (25) as well as a slow-folding mutant of R15 with the wild type to demonstrate that the folding nucleus is conserved (26).
Conclusions
We have used Φ-Φ plots to demonstrate that the Φ value analysis, if conducted properly (5), provides information about the protein folding reaction. A high correlation in Φ-Φ plots for a number of experimental datasets shows that the analysis is robust. Further, the Φ-Φ plot is a valuable tool to analyze Φ value data sets of homologous proteins. In the simplest case, a high correlation between different members of a protein domain family shows that folding pathways are evolutionarily conserved. Conversely, a low correlation in Φ-Φ plots within a domain family or individual members illustrates divergence and plasticity in protein folding pathways.
Acknowledgements
This work was funded by the Swedish Research Council (to P.J.) and the Italian Ministry of University and Research (PNR-CNR Aging Program 2012– 2014) (to S.G) and Sapienza University of Rome (C26A13T9NB to S.G.)
References
1. Levitt (1996) Through the breach. Curr. Opin. Struct. Biol. 6, 193–194 2. Kellis, J. T., Jr, Nyberg, K., Sali, D., and Fersht, A. R. (1988) Contribution of
hydrophobic interactions to protein stability. Nature 333, 784–786
3. Matouschek, A., Kellis, J. T., Jr, Serrano, L., and Fersht, A. R. (1989) Mapping the transition state and pathway of protein folding by protein engineering. Nature 340, 122–126
4. Fersht, A. R., Matouschek, A., and Serrano, L. (1992) The folding of an enzyme.
I. Theory of protein engineering analysis of stability and pathway of protein folding. J. Mol. Biol. 224, 771–782
5. Fersht, A. R., and Sato, S. (2004) Phi-value analysis and the nature of protein- folding transition states. Proc. Natl. Acad. Sci. U. S. A. 101, 7976–7981
6. Sánchez, I. E., and Kiefhaber, T. (2003) Origin of unusual phi-values in protein folding: evidence against specific nucleation sites. J. Mol. Biol. 334, 1077–1085 7. Cho, J.-H., and Raleigh, D. P. (2006) Denatured state effects and the origin of
nonclassical phi values in protein folding. J. Am. Chem. Soc. 128, 16492–16493 8. Krantz, B. A., and Sosnick, T. R. (2001) Engineered metal binding sites map the
heterogeneous folding landscape of a coiled coil. Nat. Struct. Biol. 8, 1042–1047 9. Naganathan, A. N., and Muñoz, V. (2010) Insights into protein folding
mechanisms from large scale analysis of mutational effects. Proc. Natl. Acad. Sci.
U. S. A. 107, 8611–8616
10. De los Rios, M. A., Muralidhara, B. K., Wildes, D., Sosnick, T. R., Marqusee, S., Wittung-Stafshede, P., Plaxco, K. W., and Ruczinski, I. (2006) On the precision of experimentally determined protein folding rates and phi-values. Protein Sci.
Publ. Protein Soc. 15, 553–563
11. Weikl, T. R., and Dill, K. A. (2007) Transition-states in protein folding kinetics:
the structural interpretation of Phi values. J. Mol. Biol. 365, 1578–1586 12. Calosci, N., Chi, C. N., Richter, B., Camilloni, C., Engström, A., Eklund, L.,
Travaglini-Allocatelli, C., Gianni, S., Vendruscolo, M., and Jemth, P. (2008) Comparison of successive transition states for folding reveals alternative early folding pathways of two homologous proteins. Proc. Natl. Acad. Sci. U. S. A. 105, 19241–19246
13. Gianni, S., Geierhaas, C. D., Calosci, N., Jemth, P., Vuister, G. W., Travaglini- Allocatelli, C., Vendruscolo, M., and Brunori, M. (2007) A PDZ domain
recapitulates a unifying mechanism for protein folding. Proc. Natl. Acad. Sci. U.
S. A. 104, 128–133
14. Ivarsson, Y., Travaglini-Allocatelli, C., Brunori, M., and Gianni, S. (2009) Engineered symmetric connectivity of secondary structure elements highlights malleability of protein folding pathways. J. Am. Chem. Soc. 131, 11727–11733
15. Nickson, A. A., Stoll, K. E., and Clarke, J. (2008) Folding of a LysM domain:
entropy-enthalpy compensation in the transition state of an ideal two-state folder.
J. Mol. Biol. 380, 557–569
16. Sato, S., Religa, T. L., and Fersht, A. R. (2006) Phi-analysis of the folding of the B domain of protein A using multiple optical probes. J. Mol. Biol. 360, 850–864 17. Batey, S., and Clarke, J. (2008) The folding pathway of a single domain in a
multidomain protein is not affected by its neighbouring domain. J. Mol. Biol. 378, 297–301
18. Fersht, A. R., Itzhaki, L. S., elMasry, N. F., Matthews, J. M., and Otzen, D. E.
(1994) Single versus parallel pathways of protein folding and fractional formation of structure in the transition state. Proc. Natl. Acad. Sci. U. S. A. 91, 10426–10429 19. Itzhaki, L. S., Otzen, D. E., and Fersht, A. R. (1995) The structure of the
transition state for folding of chymotrypsin inhibitor 2 analysed by protein engineering methods: evidence for a nucleation-condensation mechanism for protein folding. J. Mol. Biol. 254, 260–288
20. Nickson, A. A., Wensley, B. G., and Clarke, J. (2013) Take home lessons from studies of related proteins. Curr. Opin. Struct. Biol. 23, 66–74
21. Ferguson, N., Sharpe, T. D., Johnson, C. M., and Fersht, A. R. (2006) The
transition state for folding of a peripheral subunit-binding domain contains robust and ionic-strength dependent characteristics. J. Mol. Biol. 356, 1237–1247
22. Sharpe, T. D., Ferguson, N., Johnson, C. M., and Fersht, A. R. (2008) Conservation of transition state structure in fast folding peripheral subunit- binding domains. J. Mol. Biol. 383, 224–237
23. Neuweiler, H., Sharpe, T. D., Rutherford, T. J., Johnson, C. M., Allen, M. D., Ferguson, N., and Fersht, A. R. (2009) The folding mechanism of BBL: Plasticity of transition-state structure observed within an ultrafast folding protein family. J.
Mol. Biol. 390, 1060–1073
24. Wensley, B. G., Gärtner, M., Choo, W. X., Batey, S., and Clarke, J. (2009) Different members of a simple three-helix bundle protein family have very different folding rate constants and fold by different mechanisms. J. Mol. Biol.
390, 1074–1085
25. Wensley, B. G., Batey, S., Bone, F. A. C., Chan, Z. M., Tumelty, N. R., Steward, A., Kwa, L. G., Borgia, A., and Clarke, J. (2010) Experimental evidence for a frustrated energy landscape in a three-helix-bundle protein family. Nature 463, 685–688
26. Kwa, L. G., Wensley, B. G., Alexander, C. G., Browning, S. J., Lichman, B. R., and Clarke, J. (2014) The Folding of a Family of Three-Helix Bundle Proteins:
Spectrin R15 Has a Robust Folding Nucleus, Unlike Its Homologous Neighbours.
J. Mol. Biol. 426, 1600–1610 Legends To Figures.
Figure 1. Φ-Φ plots for different protein systems. Data refer to proteins belonging to the same super-family but sharing a different sequence (panel A) (12, 13); a protein and its circular permutant (panel B) (14) or circularized variant (panel C) (15); a
protein characterized by different optical probes (panel D) (16) or in the presence/absence of a covalently bound contiguous domain (panel E) (17). The lines are the best fit to a linear equation. In each plot we report the p-value, defined as the probability that the two variables are not correlated.
1.2,... ... - ... - ... ~ ...
0.8
0.6
>9 0.4
0.2
P= 0.0011 p < 0.0001 P< 0.0001 p = 0.0127 p<0.0001
