Heme

Heme is an iron-porphyrin cofactor essential for the catalytic activity of cytochrome P450 (CYP450) enzymes, including drug-metabolizing enzymes like CYP2C9, CYP2C19, and CYP3A4.

Structure

A heme group consists of:

• An iron (Fe²⁺/Fe³⁺) atom at the center
• A porphyrin ring — a planar, aromatic scaffold with four nitrogen atoms coordinating the iron
• The protein provides a fifth coordination bond via a conserved cysteine residue (the “proximal” ligand)
• The sixth coordination site faces the substrate binding pocket (the “distal” site)

The Heme-Thiolate Bond

In CYP450 enzymes, the conserved cysteine (Cys435 in CYP2C9) forms a thiolate (–S⁻) bond with the heme iron. This heme-thiolate arrangement is the defining feature of the CYP450 superfamily — present in bacterial CYPs from 3 billion years ago and in human enzymes today.

Why It Matters for ESM-2 Predictions

The heme-thiolate chemistry is under extreme evolutionary constraint: mutations that disrupt the heme coordination site are catastrophically damaging across all CYP450s, in all organisms. ESM-2, trained on 250 million sequences including thousands of CYP450s, learned this universal conservation signal.

This explains the anomalous result in our CYP2C9 benchmark: the heme-binding domain (positions 428-437) achieves Spearman rho = 0.811 — far above the typical active-site prediction performance — because evolution has preserved this coordination chemistry for ~2 billion years.

The SRS5 Contrast

Substrate Recognition Site 5 (SRS5) is a different type of active-site residue: it determines which substrates CYP2C9 accepts. These residues have evolved to accommodate different substrates across the CYP superfamily. ESM-2 sees this evolutionary variability as tolerance, achieving only rho = 0.422 at SRS5. This contrast — heme (rho = 0.811) vs. SRS5 (rho = 0.422) — illustrates the fundamental distinction between conserved catalytic machinery and evolvable substrate specificity.