I N M E M O R I A M
Chandrasekharan Ramakrishnan (1939–2019): The student behind the Ramachandran map
Professor Chandrasekharan Ramakrishnan (Figure 1), who generated the first version of the iconic Ramachandran map as a PhD student, passed away in Bangalore on March 13, 2019. He was on the faculty of the University of Madras first (1964–1972) and subsequently at the Molecular Bio- physics Unit, Indian Institute of Science (1972–2001), an association that continued almost until his demise. Apart from his contributions in the Ramachandran map project as a PhD student, he made several important contributions to peptide and protein conformational analysis. To mention a few, his studies of hydrogen bond geometry,1cyclic peptide conformations2 and protein structural data analysis3,4 were influential in furthering our understanding of peptide and protein structures.
The Ramachandran map was generated in Madras (now Chennai) in the early 1960s,5when the first atomic-level struc- tures of globular proteins, myoglobin and hemoglobin, were being worked out in Cambridge using X-ray crystallography.
Although no protein structures were used by Ramachandran and co-workers in their analysis of stereochemically allowed polypeptide chain conformations, the Ramachandran map and the powerful principles embedded in it have been routinely used to assess the stereochemical quality of protein structures, over the past four decades. George Rose and Janet Thornton
and co-workers point out that this is one of the very few instances, in the biological sciences, where a discrepancy between theory and experiment is used as a quality check on the experimental result.6,7The Ramachandran map also rev- ealed that only about 30% of polypeptide backbone conforma- tional (ϕ-ψ) space is stereochemically accessible, defining useful limits to take into account in protein folding studies.
C. Ramakrishnan (fondly called CR by all who knew him) was a kind, modest and engagingly honest man. I learnt the principles behind the Ramachandran map and heard the story of the work at Madras during the years that I spent in his group as a graduate student. The circumstances under which the Ramachandran Map (originally named in the clas- sic book “The Structure and Action of Proteins” by Dick- erson and Geis) emerged constitutes a story that bears retelling. In the mid 1950s, G.N.Ramachandran and Gopinath Kartha published the first of their classic papers on the triple helical structure of collagen, derived from fiber diffraction data.8,9Shortly thereafter, Francis Crick and Alexander Rich criticized the structure arguing that the nonbonded atomic distances in the Ramachandran–Kartha model were too short to be correct.10 In Rich's words: “Around this time (24 September, to be exact) a paper came out by Ramachandran and Kartha which proposed a coiled coil structure for colla- gen: but it was not possible to build it satisfactorily because of bad van der Waals contacts—the structure was wrong”.11 The Cambridge critique spurred Ramachandran to re-exa- mine the observed nonbonded contact distances in organic crystal structures, setting in motion the analysis of the stereo- chemically allowed conformations of polypeptide chains, based on hard sphere contact criteria. V. Sasisekharan, then a young faculty member at Madras and C. Ramakrishnan, then a graduate student, began addressing the problem of estimat- ing allowed limits for inter-atomic contacts from available crystal structures Interestingly, the shortest distances for vari- ous pairs of nonbonded atom types, as noted in small mole- cule crystal structures, are considerably shorter than the sum of their van der Waals radii.12For example, the Bondi radii for C and N are 1.7 and 1.55 Å, respectively. While the sum of their van der Waals radii is 3.25 Å, the shortest approach F I G U R E 1 Professor Chandrasekharan Ramakrishnan
Received: 30 July 2019 Revised: 4 September 2019 Accepted: 4 September 2019 DOI: 10.1002/pro.3723
Protein Science.2019;1–3. wileyonlinelibrary.com/journal/pro © 2019 The Protein Society 1
distance between C and N as derived from small molecule structures is usually 2.9 Å, with rare instances of even 2.8 Å which is considerably shorter than what was assumed ini- tially. This shorter nonbonded distance was evidently not expected by Rich and Crick,10 who seem to have assumed the shortest possible distance between two Cα atoms to be 3.6–4 Å and between C and O to be 3.2–3.4 Å. Interestingly, these assumptions will“disallow”even the Paulingα-helix.12 These considerations and other features like inter-chain hydrogen bonds validated the Ramachandran-Kartha model of collagen.
The Madras group then decided to work out “allowed and disallowed”combinations ofϕandψ, which are torsion angles about N-Cα and Cα-C single bonds, respectively.
They considered a model system of two-linked peptide units both in the trans conformation, with the peptide plane dimensions taken from the work of Pauling and Corey. They decided to work out two steric maps—one by considering Alanine in the L-configuration at the central Cαatom5and the other with Glycine.13 In CR's telling, they initially assumed that the shortest approach between nonbonded atoms cannot be shorter than sum of the van der Waals (Bondi) radii of the atoms involved. The steric map gener- ated with this assumption did not have even a single ϕ-ψ combination allowed! Clearly, the assumption was wrong.
The nonbonded contact distances derived from experimental
crystal structures of small molecules were undoubtedly more appropriate.
It is important to recollect the circumstances under which CR conducted his seminal work. He had to generate atomic coordinates of two-linked peptide units, using standard bond lengths and angles proposed by Pauling and Corey, for the full range ofϕandψvalues at 1 intervals, leading to 360 * 360 conformations. Generation of these conformations and checking for short-contacts between all possible combina- tions of nonbonded atoms for each of theseϕ-ψ combina- tions would take insignificant time in present day computers. But, in the early 1960s, CR had no computers available. All he had was a Marchant calculator, capable of performing only four basic operations, addition, subtraction, multiplication and division with no memory storage! More- over, generation of conformations by varying ϕ and ψ required extensive use of trigonometric functions which were not available on the Marchant calculator. CR used Clark's Tables to obtain values of trigonometric functions, which he then typed-in painstakingly into the Marchant cal- culator, innumerable times, to perform all the calculations.
With no memory available in the calculator, one can imagine him writing down all the numbers occurring in the steps of calculations in his lab notebook and re-entering the numbers into the calculator for days on end. CR was fond of rec- ounting that it took him one full year of dedicated effort
F I G U R E 2 Early Ramachandran maps for Ala (left) and Gly (right). Taken from the PhD thesis of Ramakrishnan.14The torsion angle conventions were very different from the present day convention. Therefore, the Ramachandran maps from CR's thesis look quite different from present day Ramachandran maps
2 SRINIVASAN
every single day, from morning to evening, to complete the calculations.
The very first Ramachandran maps for Ala and Gly taken from the PhD thesis14of CR are shown in Figure 2.
In the 1950s and 1960s, Ramachandran's group worked in an environment far removed from that prevailing at Cam- bridge or Caltech, the centers from which structural biology emerged. The contributions of the Madras school, of which Ramakrishnan was a key member, to structural biology are therefore a great inspiration to researchers working in India.
Determination, dedication and concentration, qualities which Ramakrishnan had in abundant measure, undoubtedly con- tributed to the development of the Ramachandran Map, which we all use today. CR was a teacher par excellence.
His course on Biomolecular Conformations trained genera- tions of students in the Molecular Biophysics Unit at the Indian Institute of Science. His lectures were characterized by the remarkable clarity and simplicity of his expression.
CR as a researcher worked with personal passion, unflinching enthusiasm and for the sheer joy of doing research. He seemed unconcerned about fame, recognition and grants. His child-like display of excitement when his FORTRAN programs worked kept him going. He was a rare Professor who worked at the computing work bench, just like a PhD student all his life. He worked from home even during the last years, with failing health. A short while before his death, he was very keen to generate Ram- achandran maps for bond parameters observed in small pep- tides. What he did once in early 1960s, he did more than 2000 times using his FORTRAN program.15
With CR's passing, we lost a passionate and a pure scien- tist and an exceptionally kind and gracious human being.
ACKNOWLEDGEMENTS
I thank Professors George Rose and Brian Matthews for an opportunity to highlight C. Ramakrishnan's work in the form of this article forProtein Science. I thank Prof. P. Balaram for his help in writing this tribute, Prof. R. Sowdhamini for her comments and Mr Ravi Kumar Boyapati for his help with reproducing figures. I thank Ms. Ashraya Ravikumar for the proof-reading.
CONFLICT OF IN TERE ST
None declared.
Narayanaswamy Srinivasan Molecular Biophysics Unit, Indian Institute of Science,
Bangalore
Correspondence Narayanaswamy Srinivasan, Molecular Biophysics Unit, Indian Institute of Science, Bangalore 560012, India.
Email: [email protected]
OR CI D
Narayanaswamy Srinivasan https://orcid.org/0000-0002- 6566-6239
REF ER ENC ES
1. Mitra J, Ramakrishnan C. Studies on hydrogen bonds. Part IV. Proposed working criteria for assessing qualitative strength of hydrogen bonds. Int J Pept Protein Res. 1981;17:401–411.
2. Manjula G, Ramakrishnan C. Stereochemical studies on cyclic peptides. Part X. Conformational analysis of hydrogen bonded cyclic pentapeptides. Int J Pept Protein Res. 1979;13:353–362.
3. Ramakrishnan C, Soman KV. Identification of secondary struc- tures in globular proteins: A new algorithm. Int J Pept Protein Res.
1982;20:218–237.
4. Soman KV, Ramakrishnan C. Occurrence of a single helix of col- lagen type in globular proteins. J Mol Biol. 1983;170:1045–1048.
5. Ramachandran GN, Ramakrishnan C, Sasisekharan V. Stereochem- istry of polypeptide chain configurations. J Mol Biol. 1963;7:95–99.
6. Rose GD. Ramachandran maps for sidechains in globular proteins.
Proteins. 2019;87:357–384.
7. Laskowski RA, MacArthur MW, Moss DS, Thornton JM.
PROCHECK: A program to check stereochemical quality of pro- tein structutes. J Appl Crystallogr. 1993;26:283–291.
8. Ramachandran GN, Kartha G. Structure of collagen. Nature. 1954;
174:269–270.
9. Ramachandran GN, Kartha G. Structure of collagen. Nature. 1955;
176:593–595.
10. Rich A, Crick FHC. The structure of collagen. Nature. 1955;176:
915–916.
11. Riddihough G. Structure of collagen. Nat Struct Mol Biol. 1998;5:
858–859.
12. Ramachandran GN, Sasisekharan V. Structure of collagen. Nature.
1961;190:1004–1005.
13. Ramakrishnan C, Ramachandran GN. Stereochemical criteria for polypeptide and protein chain conformations II. Allowed confor- mations for a pair of peptide units. Biophys J. 1965;5:909–933.
14. Ramakrishnan C. X-ray studies on biological fibres [PhD thesis].
India: University of Madras; 1965.
15. Ravikumar A, Ramakrishnan C, Srinivasan N. Stereochemical assess- ment of (ϕ,ψ) outliers in protein structures using bond geometry- specific Ramachandran steric-maps. Structure. 2019; (in press).
How to cite this article: Srinivasan N.
Chandrasekharan Ramakrishnan (1939–2019): The student behind the Ramachandran map.Protein Science. 2019;1–3.https://doi.org/10.1002/pro.3723
SRINIVASAN 3
