21

Volume 31, No. 4, December 2000

Reflections . . . . . . . . Syd Leach

Chemical Engineering to Protein FoldingSyd Leach reflects on his work at CSIRO then at the University of Melbourne

[. . . . continued from the previous issue of the Australian Biochemist, August 2000]

Despite much of our research fundingstemming from a substantial levy on eachbale of wool exported by Australia, ourCSIRO Biochemistry Unit was trans-formed, by 1959, into the “Division of Bio-chemistry” with broader terms of refer-ence.

In this new and stimulating environmentI was able to contribute in several ways.While others were extracting wool pro-teins using disulphide reductive methods,Lindley and I used dilute acids for extrac-tion, so as to leave the disulphide bondsintact. We noted that in common withother proteins such as ovalbumin, bovineserum albumin and haemoglobin, hydroly-sis of wool with weakly acid solutions(pH’s 1-4) led to the early release of freeaspartic acid.

To explain this curious phenomenon,we synthesised various model peptidescontaining asparagine and aspartic acidand studied the kinetics of their hydroly-sis. For example the hydrolysis of gly-L-asn and L-leu-L-asn at pH’s of 1.2 to 3.5at 70-100ºC showed that (a) the peptidebonds were more stable to hydrolysis thanthe amide side-chains of asn, (b) the ki-netics of hydrolysis were first order withsmall negative entropies of activation and(c) the mechanism of hydrolysis of thepeptide bonds adjacent to asn involvedproton transfer from the unionised sidechain -COOH group to the adjacent pep-tide bond.

In stronger acid, the normal bimolecu-lar mechanism involved the hydroxoniumion, common to acid hydrolysis of all pep-tide bonds, became more important.Hence the failure to observe preferentialaspartic acid liberation in stronger acidsolutions.

Another project, undertaken with B.Milligan, involved the mechanism of thiol/disulphide interchange during the “setting”of wool and other keratinous fibres. Thisprocess is important in commercial per-manent pleating (of woollens) and in hair-setting.

In each case “set” occurs as a result ofconformational changes in polypeptidestructure; these changes cannot occur

without disulphide bond rearrangementvia -SH/-SS interchange. Setting is en-hanced in basic solutions and with bisul-phite through the production of new tran-sient, thiol groups by hydrolytic fission ofdisulphide bonds and by ionising both thenew and original thiol groups. Thiol

anions favour -SH/-SS interchange andhence polypeptide chain rearrangementto a more stable “set” state. We notedthat the pleating or setting of yarns wasimproved by after-treatment with rea-gents such as oxidising agents (peroxide,iodate, tetrathionate) or alkylating agents(iodoacetamide) which destroy free thiolgroups and thereby inhibit disulphiderearrangements.

We followed up these practical obser-vations with tritium-hydrogen exchangerate measurements. I had previously es-tablished T/H exchange kinetics (using tri-tiated water) as a viable alternative to theD/H exchange methods established byLinderstrøm-Lang at the Carlsberg Labo-ratories in Copenhagen and widely usedelsewhere. T-H exchange has certain ad-vantages over D-H exchange, namely (i)

the high sensitivity of detection of tritiummeans that it need be used in only smallamounts (ii) there is therefore less riskof label-induced conformational change inthe protein, (iii) lower protein concentra-tions may be used (0.2 to 2.0%) and (iv)rates of exchange are slower. Exchangekinetics provide a means for detecting andquantifying H-bonded or otherwiseshielded H-atoms in folded proteins. Hatoms attached to sulphur, nitrogen oroxygen atoms and not so shielded willexchange rapidly whilst those which areshielded or involved in α-helical or β-structures will exchange more slowly.

As with most proteins we observedthat wool exhibited four classes of ex-changing hydrogen atoms. One class(Class 1) exchanged very rapidly and wasassigned to the so-called amorphous re-gion. Classes II and III H-atoms were as-signed to structures of intermediate sta-bility and these exchanged slowly with apH dependence suggestive of structuresstabilised by protonated side chains anda temperature dependence suggestive ofstructures such as α-helices of variousthermal stabilities.

Other H atoms (Class IV) exchangedonly on disruption of the keratin struc-ture by alkali or lithium bromide. We ob-served that the “setting” of wool yarns atall temperatures up to 100ºC increasedthe proportion of rapidly exchangingamorphous material. However, it is alsoknown that the highly organised β-kera-tin of silk fibroin exchanges rapidly. Nev-ertheless, prolonged setting of wool, orthe use of setting agents, was found toincrease the proportion of slowly ex-changing Class IV hydrogen atoms, pre-sumably due to conversion of the ex-tended chains to β-crystallites.

Using T-H exchange with ribonuclease,it was found that there are 50-60 H at-oms which will exchange within 24 hoursat temperatures below 42ºC. The litera-ture values previously reported were 20atoms and 60ºC.

The discrepancy may well be due todifferences in stability between a fullydeuterated protein and a “lightly” triti

Syd Leach (centre) at the 1995 Lorne Proteinconference when he chaired the opening ses-sion on Protein Evolution – with the two speak-ers Russell F. Doolittle (UCSD) (left) and H.Jörnvall (Karolinska Institute, Stokholm) (right).

22

Australian Biochemist

ated one – as well as milder dryingtreatments.This suggests important advan-tages for the tritium tracer method overthe more cumbersome D2O method. A study of the pH-dependence of T-Hexchange between pH 2.5 and 12 sug-gested more extensive H-bonding at lowpH and this was confirmed by optical ro-tatory dispersion measurements. Hereagain, the results are at odds with earlyreports in which D-H exchange was used– a discrepancy again probably attribut-able to changes in protein stability causedby extensive D-substitution.

A most exciting turning point in my re-search career came with my introductionin 1956, to Professor Harold Scheraga.Nowadays, as a close friend and colleagueof long standing, it is difficult to remem-ber the reverence with which I first ap-proached him. I spent my first sabbaticalyear with him and with Michael LaskowskiJr. in 1956, at Cornell University in Ithaca,New York.

I applied the (then) new technique ofUV difference spectroscopy to a study ofhydrogen bonding of tyrosyl groups ininsulin. Acidification or tryptic digestionproduced shifts in the tyrosyl absorptionspectra and the pH-dependence sug-gested bonding of one or more tyrosyl -OH side chains to carboxylate ion sidechains as acceptors. Models of insulin sug-gested that the B16 tyrosine was H-bonded to the B13 glutamic acid, whilethe kinetics of the trypsin-induced shiftsuggested that the B26 tyrosine was alsoH-bonded to an acceptor but one whichdid not ionise in the pH range 1.5 to 8.0.

I had later sabbatical years with HaroldScheraga as Visiting Professor (in 1964-5,1976-7 and 1981) and I was introducedto his major areas of interest: the rela-tion between intramolecular forces andprotein conformation. In this, GeorgeNémethy played an important role.

The starting point was of course thedictum that the most stable conforma-tion of a polypeptide (or indeed of anypolymer or molecule) was that of mini-mum free energy. C.B. Anfinsen1 and H.A.Scheraga had both started from this pointin the early 1960’s but Scheraga and histeam were building up a quantitative ther-modynamic approach to the ultimate ob-jective of computing 3-dimensional con-formations from a knowledge of theamino acid sequence and the location ofdisulphide links in polypeptides and pro-

teins. In 1964/5 we used hard spheremodels of the atoms in dipeptides gly-L-X where X was gly, ala, val etc., comput-ing fluctuations mainly in non-bonded vander Waal’s interactions as a function ofbond rotations around the N-Cα and Cα-C’ bonds. In this way we arrived at 2-dimensional “steric maps” which showedthose peptide conformations that weresterically permitted. The results resem-bled those of G.N. Ramachandran, show-ing for example that, of the total area ofthe steric map, gly-gly conformations werepermitted in some 52% of the area, gly-L-ala 16% and gly-L-val only 4.5%.

We then went on to construct a mo-lecular model for the preferred confor-mation of the cyclic octapeptide (residues65-72) of ribonuclease, using our crude“hard sphere” conformation maps foreach amino acid residue in the sequence.We were then emboldened to do thesame for the cyclic decapeptide gramici-din S. Surprisingly, the structure we arrivedat resembled a variant of a β-pleated sheetand gratifyingly was consistent with thex-ray diffraction data of other workers.We also predicted a low energy iso-geo-metric variant of the R.H. α-helix whichwe named the “αII-helix”– since observedin many proteins.

Nevertheless we were under no illu-sions at this stage that the precise deter-mination of preferred peptide or proteinconformations would have to involve theminimisation of their total free energy, ofwhich van der Waal’s forces were but acomponent: and even then these would

require full energy contour maps rather thanthe crude “hard sphere” maps used above.

In addition, distortions of bond-angles,bond-lengths, amide group planarity, H-bond energies, dipole, hydrophobic andsolvent interactions were later to takeninto account. In one investigation, we stud-ied the influence of amino acid side chainson the free energy of helix-coil transitions,taking into account the entropies of un-folding per residue, the entropies of in-ternal rotation of each type of side chainand including hydrogen bonding require-ments. For example, the lower stability ofthe L.H. α-helix relative to the R.H. α-helix was seen to be clearly due to inter-actions of the Cβ atoms of the side chainswith atoms in the adjacent peptide back-bone.

These first faltering steps were soonfollowed by more sophisticated treat-ments along the road towards true freeenergy minimisation including all of thesefactors. Every two years there are CASP(Critical Assessment of Structure Predic-tion) trials in which a three-dimensionalprotein structure which has been near-ing completion by x-ray and/or NMRmeasurements (but which has been de-liberately withheld from publication), ishanded over to contending researchgroups which challenges the validity of theirvarious theoretical predictive methods.

In this competition, I understand thatin 1999 Harold Scheraga’s group demon-strated signal success with their mostrecent ab initio predictive calculations. Theeventual prediction of biopolymer con-formation, deduced from sequence, willbe I believe, the most daunting, demand-ing but ultimately the most intellectuallytriumphant achievement in physical bio-chemistry.

1 Anfinsen was awarded a Nobel prize in 1972 for theseinsights and their impact on protein science. Doubt-less with many others, I recognised his influence innominatinghim for this distinction and Dr Anfinsen latertold me that the Nobel Committee had chosen mycitation in the award ceremony!

Erskine House and Beach at Lorne – the venue for many Lorne Protein Conferences and, later,Genome Conferences. This site is to be redeveloped in the next year or so.

Reflections . . . . . Syd Leach (contin.)

23

Volume 31, No. 4, December 2000

proud to have lectured to, and hopefullyinfluenced, E. Blackburn, M.J. Gething andother notable biochemists.

In 1972-73, I spent a sabbatical year withC.B Anfinsen at the N.I.H. in Bethesda andthis gave me new insights and enthusiasmin the field of structural immunology. Wewere soon synthesising peptide antigensin my Melbourne laboratory, designed todelineate sequential antigenic determi-nants on the surface of proteins of knownconformation such as myoglobins (human,sperm whale and bovine).

We showed that true antigenicity israrely to be found in small peptide frag-ments and that determinants in proteinswere invariably “topographic”. Our stud-ies with monoclonal antibodies raised tothese proteins bore out this contention.Popular views hitherto had been that thefull antigenicity of any given protein is at-tributable only to certain special “patches”on the surface of the protein. Ourpeptide / antibody studies showed con-clusively that the entire surface of a pro-tein can elicit antibody production; theextent to which peptides correspondingto the various regions on the protein sur-face can simulate the intact protein inreacting with specific monoclonal antibod-ies or T-lymphocytes depends upon theirsize and conformation. The latter can bestabilised by artificial cross-linking thusproducing more rigid structures whichrecognised T-cells more readily. Such find-ings provide guidelines in the design ofpeptide vaccines. Workers in other im-munology laboratories (notably Drs. J.Berzofsky and S. Smith-Gill at N.I.H.) hadcome to similar conclusions and we col-laborated in a joint review for AnnualReviews of Immunology. Our continuousstudies in the field of structural immu-nology – stemming from an original train-ing in “pure” physical chemistry, thus inmany ways became complementary to thestudies in cellular immunology across theroad at the Walter and Eliza Hall Insti-tute, initiated by Sir MacFarlane Burnettand continued so ably by Sir GustavNossal and his fellow scientists.

In 1976, with Tony Burgess and TheoDopheide, I founded the first Lorne Con-ference on Protein Structure and Func-tion. Melbourne had long been a centrefor medical, biochemical and especiallyprotein science with several notable in-stitutes with research interests in com-

mon. Our aim was to bring as many ofthese powerful research groups togetherat informal Protein Conferences.

At our first meeting in 1976 we hadabout 36 participants and one overseasinvited speaker. When I retired as Chair-man of the Organising Committee in 1985there were about 200 participants, includ-ing a substantial number of overseas dele-gates.

Subsequently, Dr Robin Anders took upthe reins and, with a Committee repre-senting most of the major research insti-tutes in Melbourne, carried the meetingforward to the point where, in February2000, at the 25th meeting, there were 450participants. Of these, 83 were from coun-tries other than Australia. Dr Anders andthe Committee saw fit, after 1985 to ini-tiate a series of “Leach Lectures”.

These were presented, year by year, byR.L. Baldwin, P.M. Colman, N.A. Nicola, J.M.Guss, A.W. Burgess, R. Huber, A.A. Clarke,E.H. Fischer, J.N. Varghese, D. Metcalf, L.Hood, N. Hoogenraad, G.W. Tregear, R.Y.Tsien and, in the year 2000, by R.F.Doolittle.

In recent years I have been working,with my colleague of many years Eliza-beth Minasian, and in collaboration withcolleagues at the Ludwig Institute forCancer Research on the conformationsof peptides of biological importance. Thishas culminated in the recent publicationof a paper on new techniques for the ef-ficient analysis and visualisation of pep-tide and protein conformational relation-ships.2

2 M. F. O’Donohue, E. Minasian, S. J. Leach, A.W. Bur-gess, H. R. Treutlein. PEPCAT- A New Tool for theConformational Analysis of Peptides: J Comput Chem21: 446-461 (2000).

SBMB

“In 1968, I left CSIRO to take up theChair of Biochemistry at the Univer-sity of Melbourne, vacated by theretirement of Professor V.M. Trikojus. Iwelcomed the opportunity to teach aswell as introduce young researchworkers to new areas of science. I wasfortunate to attract some fine youngminds to my laboratory, among themA.W. Burgess, N.A. Nicola, Y. Pattersonand many others. I am also

Reflections . . . . . Syd Leach (contin.)