Mutate one residue and see what breaks: p53 R273H

In Lesson 1 you saw MDM2 holding p53 quiet. That is one way a tumour switches p53 off. The more common way is simpler and more brutal: change one letter of p53 itself. TP53 (the gene that encodes p53) is mutated in about half of all human cancers, and most of those mutations sit in one region, the DNA-binding core domain (a domain is a separately folded region of a protein), the part p53 uses to grip the genes it switches on. A handful of positions are hit so often they are called hotspots. Two of them, Arg248 and Arg273 (the arginines at positions 248 and 273 of the protein chain), touch the DNA directly.

In this lesson you will do what a structural biologist does with a patient's mutation: open the wild-type (normal) structure, find exactly what the residue (amino acid) touches, make the mutation in silico (on the computer), and check what is lost. You will mutate Arg273 to histidine (R273H), one of the most frequent p53 mutations in cancer, in Edit with AI (Structure Studio, opened from the 3D structure card), find the interactions again, and relax the mutant with an energy minimisation (a calculation that lets the atoms settle into a low-strain arrangement). In the name R273H, R and H are the one-letter codes for arginine and histidine, and 273 is the position.

We will use PDB 1TSR (entry 1TSR in the Protein Data Bank, the public archive of solved 3D structures): the human p53 core domain bound to DNA, solved by X-ray diffraction at 2.2 Å by Cho, Gorina, Jeffrey and Pavletich in 1994. It is the structure that first explained why the hotspot mutations cause cancer. X-ray diffraction works out atom positions from how a crystal scatters X-rays; 2.2 Å is the resolution, the finest detail the data can show (smaller is sharper; one angstrom, Å, is 0.1 nanometres, about the width of an atom).

What you will make

The p53 core domain (chain B, one of three protein copies in the crystal, light blue) on its DNA site (light orange). The two teal side chains (amino-acid R groups) reaching into the DNA are the hotspot arginines: Arg248, which lies in the minor groove (the narrower of the two grooves that spiral along the double helix), and Arg273, which reaches the sugar-phosphate backbone. The grey sphere is the structural zinc ion, which helps hold the fold together. Chains A and C of the crystal are hidden for clarity. PDB 1TSR.

Wild type. Arg273 (teal) reaches the DNA backbone and its NH2 nitrogen (one of the nitrogens at the tip of arginine) hydrogen-bonds to a phosphate oxygen of DT12 (D for DNA, T for thymine, nucleotide 12) on DNA strand E (dashed line, 2.53 Å in our run). A hydrogen bond is a weak attraction between a hydrogen on one N or O atom and another N or O nearby. Grey sticks (bonds drawn as rods): nucleotides DT11 to DG13, phosphorus in orange.

R273H, modelled in Edit with AI. Same view, same DNA. His273 (magenta) is shorter and stops short of the backbone: no hydrogen bond, and its closest nitrogen is 4.30 Å from that phosphate oxygen in our run. One atom-level contact gone.

1. Open the structure Main chat

Start a new chat in RBAi:

Open PDB 1TSR as an interactive 3D structure. Tell me the protein and DNA chains, which protein chain makes the sequence-specific contacts with the DNA (Arg248 in the minor groove and Arg273 to a backbone phosphate), and where the zinc atoms are. Do not change the structure.

What you should see: three p53 core-domain chains (A, B, C), two DNA strands (E, F), one zinc ion per protein chain, X-ray diffraction at 2.2 Å. In our run RBAi identified chain B as the copy making the sequence-specific DNA contacts (the contacts that recognise the particular DNA sequence p53 binds). Use chain B for everything below. (The other two copies in the crystal are not bound to the DNA site the same way. Asking which chain to use is a real research step, not a formality.)

2. Make the overview figure Main chat

You do not write any PyMOL (the molecular-graphics program RBAi runs for you) for this. You describe the figure and RBAi builds the scene. In the prompt, cartoon draws each chain as a ribbon, sticks draw every bond as a rod and a sphere draws an atom as a ball. In the main chat:

Make a publication-quality overview figure of PDB 1TSR, the p53 core domain bound to DNA. Show only chain B of the protein as a light-blue cartoon and the two DNA strands (chains E and F) as a light-orange cartoon. Show the side chains of the two hotspot arginines, Arg248 and Arg273 of chain B, as teal sticks, and the chain B zinc ion as a grey sphere. Hide the other two protein copies. White background, orient on chain B and the DNA. Use the PyMOL figure renderer.

Optional — reproduce it exactly. The prompt above is all you need. If you would rather copy our figure pixel-for-pixel than let RBAi frame its own, you can also paste the exact PyMOL scene below into the main chat and ask RBAi to render it with the PyMOL figure renderer (Edit with AI can run it too), because both run PyMOL for you. The core of the scene we used:

cmd.show("cartoon", "chain B and polymer.protein")   # p53 core, light blue
cmd.show("cartoon", "chain E+F")                     # DNA, light orange
cmd.show("sticks", "chain B and resi 248+273 and not name N+C+O")  # teal
cmd.show("spheres", "chain B and resn ZN")           # zinc, grey
cmd.orient("chain B and polymer.protein or chain E+F")

(The full scene also sets the colours, a white background and ambient occlusion, soft shading that darkens tucked-in crevices to give depth.) That is the first figure above; the plain prompt gives you an equivalent figure, and the scene just reproduces ours exactly. Rotate the 3D card yourself and find where each arginine meets the DNA.

3. Find what the hotspot arginines touch Edit with AI

On the 3D structure card, click Edit with AI and ask:

Find the interactions between Arg273 and Arg248 of chain B and the DNA (chains E and F): hydrogen bonds and salt bridges. List each with the residue, atom names, the DNA nucleotide and atom, and the distance.

What you should see: dashed lines on the structure. In our run RBAi found two hydrogen bonds, both from Arg273: its NH2 to the phosphate oxygen OP2 of DT12 on strand E (2.53 Å) and to the O5′ (the oxygen joining that phosphate to its sugar) of the same nucleotide (3.20 Å). It also noted that PyMOL's salt-bridge rule (a salt bridge is an attraction between a positive and a negative charge) only pairs protein side chains, so it checked arginine-to-phosphate contacts directly: Arg273 qualifies; Arg248 makes no direct hydrogen bond (its nearest phosphate oxygen is 4.01 Å away).

4. Look more closely at Arg248 Edit with AI

A missing hydrogen bond is not the same as no contact. Ask (version 0 is the original structure; Edit with AI saves each change as a new numbered version):

In the original structure (version 0), which DNA atoms and waters are within 4 A of the guanidinium group of Arg248 in chain B? Is it in the minor groove?

What you should see: in our run, Arg248's guanidinium (the nitrogen-rich, positively charged group at the tip of arginine) sits 3.7 to 3.9 Å from sugar atoms of both strands (DT11 and DT12 on strand E, DT14 on strand F), a crystallographic water (a water molecule seen in the crystal, HOH 57 in the file) sits 2.27 Å from its NH1, and RBAi confirmed it lies on the minor-groove face (the minor-groove atoms are about 4.5 Å away, the nearest major-groove atom 8.0 Å). Arg248 is packed into the groove rather than bonded to one atom. That is why R248 mutations also break DNA binding, even though this rule-based count lists no hydrogen bond for it.

5. Make the mutation Edit with AI

Mutate Arg273 of chain B to histidine (R273H), then find the hydrogen bonds and salt bridges between His273 and the DNA again, and tell me the closest distance from His273 to the DT12 phosphate of chain E.

What you should see: a new version of the structure (the original stays one undo away). In our run RBAi built His273 using the rotamer with the lowest steric strain (5 of 6), then found no hydrogen bonds and no salt bridges between His273 and the DNA. A rotamer is one of the few standard shapes a side chain can twist into, and steric strain is how much it bumps into neighbouring atoms. His273's closest approach to the DT12 phosphate is 4.30 Å (His NE2, a nitrogen in histidine's ring, to OP2), too far for a hydrogen bond. The 2.53 Å contact is gone.

The two close-ups above were made in the main chat with the same camera and the same DNA. For the mutant panel the scene script places His273 at the exact coordinates (atom positions) from the Edit with AI version (we downloaded that version's structure file), so the picture is the model you analysed, not a new guess.

6. Relax the mutant Edit with AI

A rotamer dropped into a crystal structure can sit slightly wrong. An energy minimisation lets the atoms settle into the nearest low-strain arrangement. Ask:

Energy-minimise the R273H mutant and report the energy before and after and the RMSD.

It runs in the background (OpenMM, a molecular-simulation program, with the Amber14 force field and implicit solvent, heavy atoms restrained). A force field is a set of equations and numbers describing how atoms push and pull on each other; implicit solvent treats the surrounding water as a smooth medium instead of separate molecules; and heavy atoms restrained means every atom except hydrogen is held near where it started. Our first run on the whole crystal stopped: the force field had no parameters for one residue of another protein chain (SER 390 in the file). That is a normal thing to meet with real files, so we fixed the input rather than the result:

The minimisation stopped on SER 390. Starting from the R273H version, remove protein chains A and C (keep chain B, its zinc, and DNA chains E and F), then energy-minimise that complex and report the energy before and after and the RMSD.

What you should see: in our run the minimisation finished in 99 s. The energy went from 40,045 to −71,690 kJ/mol and the heavy atoms moved by an RMSD of 0.20 Å (root-mean-square deviation: a kind of average of how far the atoms moved). RBAi also added 1,959 hydrogens and 2 missing heavy atoms, removed 384 waters and kept the zinc fixed. A large drop is normal when freshly added hydrogens and small clashes relax, so the absolute numbers mean little on their own. The tiny RMSD is the useful part: the structure barely moved.

Then check the result, which is the point of the exercise:

In the minimised version, find the hydrogen bonds between His273 and the DNA again, and give the closest distance from His273 to the DT12 phosphate of chain E.

In our run there were still no hydrogen bonds, with the closest His273 atom (NE2) 4.06 Å from the phosphate oxygen. Relaxing the model did not bring the contact back.

The ideas behind the picture

• Contact versus structural mutations. Arg248 and Arg273 are contact hotspots: the protein folds, but loses a hand on the DNA. Other hotspots, such as R175, are structural: they destabilise the fold itself. Knowing which kind a mutation is changes how you might try to rescue it.
• Why arginine. Arginine's long side chain ends in a positively charged guanidinium with five N–H groups. It can reach a negatively charged phosphate and hold it with a hydrogen bond plus charge attraction. Histidine is shorter, mostly uncharged at pH 7, and in our model it cannot reach.
• Backbone versus base contacts. Arg273 grips the sugar-phosphate backbone, which is the same in every DNA sequence. It does not read the sequence; it anchors the protein so other residues (like Arg248 in the minor groove and residues in the major groove) can.
• Reading a mutation like a researcher. Measure what the wild-type residue touches, model the change, measure again, and only then reason about biology.

What the picture shows, and what it doesn't

• What you can point to: in the crystal, Arg273 hydrogen-bonds to a DNA phosphate; in our model of R273H, the histidine does not reach it.
• What is a model, not an experiment: His273 is a modelled rotamer placed into the wild-type structure. It is not an experimental mutant structure, and other rotamers or small backbone shifts are possible. The minimisation is a computational relaxation of that model with a force field. Its energies are not a measured stability or binding affinity.
• What it cannot tell you: how strongly R273H p53 still binds DNA, which genes it fails to switch on, or the extra (gain-of-function) behaviours mutant p53 shows in cells. Those come from biochemistry and cell biology. R273H's effect in cells involves much more than one lost contact.

Your turn

You removed one contact. Now think like the person designing a rescue drug or a follow-up experiment.

1. In Edit with AI, make R273C instead (arginine 273 to cysteine, another common cancer mutation at the same position) and measure its closest distance to the DT12 phosphate. Is it closer or further than His273?
2. Try R248Q (arginine 248 to glutamine) and ask what Gln248 still touches in the minor groove.
3. Write two sentences: which of the two positions do you think is more sensitive to mutation, and what experiment (not a model) would you need to prove it?

Then connect it back to Lesson 1: MDM2 silences healthy p53 from outside; a contact mutation like R273H silences it from inside. Two different structural reasons for the same broken guardian.

Try it

Open RBAi, paste the first prompt, click Edit with AI on the 3D structure, and make the mutation yourself. Every change is a version you can undo, so try a few residues and compare what each one loses.