Biosynthetic steroids

The steroid nucleus — cyclopentanoperhydrophenanthrene — is four fused rings made almost entirely of methylene and methine groups that are chemically indistinguishable from one another. Synthetic chemistry works where there is a functional group to grip; here there is none. That is why industrial production of corticosteroids and sex hormones still rests on two microbial steps that seventy years of process chemistry have not displaced.

Step one: cut the side chain, keep the nucleus

The feedstock is phytosterols — sitosterol, campesterol, stigmasterol — recovered from soybean oil processing and from tall oil in pulp mills. What separates them from a cholesterol-like skeleton is the alkyl side chain at C17, which has to go.

Actinobacteria of the genus Mycobacterium do this routinely: to them a sterol is a carbon source, and they strip the side chain through successive rounds of beta-oxidation. The difficulty is that a wild strain does not stop there — it goes on to open the ring system. An industrial strain is built by deliberately knocking out the ring-degrading enzymes: inactivating 3-ketosteroid Δ1-dehydrogenase (KstD) fixes the product as androstenedione (AD), inactivating 3-ketosteroid 9α-hydroxylase (KshA) gives androstadienedione (ADD), and retaining one activity yields 9-OH-AD. Choosing the product is choosing which gene is off.

Step two: put oxygen where nothing else can

Glucocorticoid anti-inflammatory activity requires an oxygen at C11. That is an unactivated tertiary C–H buried in the middle of the molecule, and there is no selective chemical way to oxidise it. Sarett’s classical cortisone synthesis needed more than thirty steps.

The answer came at Upjohn in 1952: a culture of Rhizopus arrhizus hydroxylates progesterone at the 11α position in a single fermentation. The catalyst is a cytochrome P450 — a heme-thiolate monooxygenase that activates molecular oxygen, cleaves the O–O bond and inserts one oxygen atom into a C–H bond while the binding pocket holds the substrate in a fixed orientation. Regioselectivity here is geometric rather than electronic: what decides the site is not which bond is most reactive but which bond is pointed at the activated oxygen. For the 11β configuration that hydrocortisone requires, Curvularia lunata is used instead. The same principle covers the 1,2-dehydrogenation that converts hydrocortisone to prednisolone.

What sets the ceiling

The limit is phase behaviour rather than biology. Sterols are effectively insoluble in water — aqueous concentrations sit in the micromolar range, while the process needs to run at tens of grams per litre. The reaction therefore happens at a phase boundary and its rate is governed by mass transfer, not by the enzyme’s kcat. Hence the whole toolkit of cyclodextrin complexation, two-phase systems with an immiscible solvent, emulsions and fine substrate dispersion — and hence industrial fermentations measured in days.

The second limit is P450 coupling. The enzyme needs reducing equivalents from NAD(P)H delivered by redox partners, and a fraction of turnovers is wasted: instead of hydroxylation the cycle releases superoxide or hydrogen peroxide. Uncoupled turnover burns cellular energy and damages the producer through oxidative stress. On top of that, steroid intermediates and products are themselves toxic to the organism, which caps achievable titre independently of how good the catalysis is.