G. N. Ramachandran of Madras Proposed the Collagen Triple Helix in 1954 and Co-Authored the 1963 Plot That Crystallographers Still Use to Check Protein Structures

On 7 August 1954 the journal Nature printed a two-page letter from the physics department of the University of Madras. G. N. Ramachandran and Gopinath Kartha proposed that collagen, which makes up about a quarter of all the protein in most animals, is built from three helical chains packed side by side. Nine years later, in July 1963, Ramachandran, C. Ramakrishnan and V. Sasisekharan published a chart of the backbone angles a protein chain can take without its atoms colliding. It is now called the Ramachandran plot, and crystallographers still use it as a check on the protein structures they solve.

Both pieces of work were done in Madras (now Chennai). The plot was calculated in 1960 to 1962 on an electric desk calculator, because the laboratory had no computer.

A physicist from Ernakulam with a Cambridge doctorate

Gopalasamudram Narayana Ramachandran was born in Ernakulam on 8 October 1922. After an honours degree in physics he joined the Indian Institute of Science in Bangalore to work with C. V. Raman, according to the obituary by N. Yathindra in Acta Crystallographica. His master’s degree by research was on the optics of mixed materials, and his DSc came from Madras University. Yathindra also credits him as the first to record X-ray topographs, a way of studying surface features of crystals.

From 1947 to 1949 he was at the Cavendish Laboratory in Cambridge, then headed by Lawrence Bragg, on an 1851 Exhibition scholarship. His doctorate there, under W. A. Wooster, was on how X-rays scatter diffusely from crystals and what that says about their elastic constants. In 1952, after a short spell as assistant professor in Bangalore, he became professor and head of physics at Madras. Yathindra writes that this is where he turned his attention to using X-ray diffraction on biological molecules.

Collagen from a shark ray, a rat and a kangaroo

Manju Bansal, his last graduate student, writes in Resonance that Ramachandran’s group began recording X-ray pictures of collagen fibres in 1952, at the suggestion of the crystallographer J. D. Bernal during a visit to Madras. The 1954 letter names the sources: a shark ray, rat tail tendon and kangaroo tail tendon. M. Vijayan, writing for the International Union of Crystallography, adds that the original collagen sample came from the Central Leather Research Institute, which stood next to the laboratory.

Earlier models by William Astbury and by Linus Pauling and Robert Corey had not matched the X-ray pattern, Bansal notes. Collagen was also awkward chemically. Roughly a third of its amino acids are glycine, the smallest one, and it carries a lot of proline and hydroxyproline, rings that restrict how a chain can bend.

Three chains, and glycine at every third place

Bansal quotes the 1954 letter’s own description: the model “consists of nine amino-acid residues per unit cell”, arranged as three helical chains, each with a pitch of 9.5 angstroms and three residues per turn. The chains are held together by hydrogen bonds. Every peptide bond is in the “trans” form, and of the three carbon atoms in each turn, one has no room for a side group at all, so it can only belong to glycine.

That last point linked structure to chemistry. Bansal calls Ramachandran the first to see that collagen’s unusually high share of glycine and of proline and hydroxyproline must have a structural reason. By 2001, Bansal notes, more than 400 mutations in different collagens had been linked to human disease, and the overwhelming majority replace a glycine, which is what a model with glycine at the core of the triple chain predicts would disturb it.

In 1955 the group revised the model. Stretched fibres gave a pattern pointing to ten residues in three turns, so Ramachandran and Kartha, in Nature on 24 September 1955, wound the three chains around a common axis in a rope-like coiled coil. Bansal writes that Ramachandran told a lecture audience the idea came from the Moon, which turns on its axis as it circles the Earth and so always shows the same face.

The argument with Rich and Crick

On 12 November 1955 Alexander Rich and Francis Crick published their own paper on collagen in Nature. They objected to a feature of the Madras model: it had two hydrogen bonds between chains for every three residues, and the atoms came closer together than Rich and Crick thought possible. Srinivasan and colleagues, in the Protein Science obituary of Ramakrishnan, quote Rich’s own recollection that, because of bad contacts, “the structure was wrong”. Rich and Crick, and separately Cowan, McGavin and North, proposed a version with one hydrogen bond per three residues.

Ramachandran accepted the point about the second bond. He later suggested, with R. Chandrasekharan, that extra bonds could link the chains through water molecules, and Barbara Brodsky wrote in 1999 that high-resolution crystal structures of collagen-like peptides have since shown such water-mediated bonds. Shoulders and Raines, in the Annual Review of Biochemistry, say that in 1955 the structure was refined by Rich and Crick into the form accepted today, with a single hydrogen bond between chains per triplet and a repeat of ten residues in three turns. Bansal’s 2006 review in the National Medical Journal of India takes a different emphasis, saying recent work has confirmed the Ramachandran and Kartha triple helix as essentially correct apart from details of the hydrogen bonds. Brodsky adds that Rich and Crick, and a separate group at King’s College London, proposed related triple-helical structures independently in 1955. The scientific community in the West called the structure the “Madras triple helix”, Bansal writes.

How a dispute over close contacts produced the plot

The criticism had a second result. Srinivasan and colleagues, drawing on Ramakrishnan’s own account, say the Cambridge objection sent Ramachandran back to the distances at which atoms actually sit in crystals. V. Sasisekharan, then a young faculty member, and Ramakrishnan, then a graduate student, took it up. Their first attempt used the textbook size of each atom as the closest allowed approach, and the resulting map had no permitted combination of angles at all. Small-molecule crystal structures showed that real atoms get closer than that: for carbon and nitrogen, the textbook radii add up to 3.25 angstroms, yet the closest approach seen in small-molecule crystals is usually 2.9. The Protein Science obituary adds that the distances Rich and Crick assumed would have ruled out even Pauling’s alpha helix.

The calculation was set up for two peptide units, the flat links of a protein backbone, joined at one carbon atom. Two angles, called phi and psi, describe the twist at each joint. Ramakrishnan’s Resonance article says every pair of values from minus 180 to 180 degrees, at fixed intervals, was tested and labelled fully allowed, partially allowed or disallowed, in 1960 to 1962. Srinivasan and colleagues add that Ramakrishnan had only a Marchant calculator, which does the four basic operations, plus printed trigonometric tables, and that by his own telling the work took a full year of daily effort.

The paper, “Stereochemistry of polypeptide chain configurations”, appeared in the Journal of Molecular Biology, volume 7, pages 95 to 99, in July 1963, with Ramachandran, Ramakrishnan and Sasisekharan as authors. Srinivasan and colleagues write that the chart showed only about 30 per cent of the phi and psi combinations to be stereochemically accessible. The map was first given that name, they add, in the textbook The Structure and Action of Proteins by Dickerson and Geis.

A prediction that protein structures later confirmed

When the plot was published, Ramakrishnan writes in Resonance, not one protein structure had been solved, though the angles of the alpha helix and beta sheet, proposed for fibrous proteins, already fell in the allowed regions. Vijayan’s IUCr report says the structure of only one globular protein was known. Either way, the plot waited for the structures to arrive. By 2001, Ramakrishnan notes, more than 15,000 were known and the plot served as a checkpoint during and after each solution. That count includes structures solved by X-ray crystallography, the method behind Venkatraman Ramakrishnan’s ribosome structures.

Srinivasan and colleagues quote George Rose, Janet Thornton and co-workers on why this is unusual: it is one of very few cases in biology where a mismatch between theory and experiment is used as a quality check on the experimental result. The same angles are used to read the shape of a protein, since helices and strands each cluster in their own part of the chart. Ramachandran’s group also applied the idea to nucleic acids and polysaccharides, as Yathindra’s obituary and Resonance both record.

Recognition

Ramachandran was elected a Fellow of the Royal Society in 1977. In 1999 the International Union of Crystallography gave him its fifth Ewald Prize, for work on anomalous scattering, on fibre structures such as collagen and, in its words, “foremost” for the macromolecular conformation work and the plot, “which even today remains the most useful validation tool”. M. Vijayan accepted the medal at the Glasgow congress on his behalf. Ramachandran died in Chennai on 7 April 2001, aged 78.

On 7 August 2004, 50 years after the collagen letter, a symposium in New Delhi marked the anniversary, supported by the Council of Scientific and Industrial Research and the Department of Science and Technology, with the President of India, Abdul Kalam, taking part. Vijayan’s report on it calls Ramachandran “arguably the most outstanding scientist to have worked in independent India”.