Carboxylic Acid vs Tetrazole: Reading a Classic Bioisostere

A bioisostere swap is the medicinal chemist's polite con: replace one functional group with another that keeps the property that matters — the binding interaction, the acidity — while quietly fixing something that doesn't, like membrane permeability or metabolic liability. The textbook example is swapping a carboxylic acid (−COOH) for a 1H-tetrazole. This post shows why that swap works, read straight off the electron density by Hilbeon.

The premise is that the two groups are nearly interchangeable where it counts. Benzoic acid has an aqueous pK_a of about 4.2; 5-phenyltetrazole comes in around 4.5. At physiological pH both are essentially fully ionized, so the anion that docks into a target's positively charged pocket is preserved — same charge, same salt bridge, same recognition. Yet the tetrazole version is frequently more permeable and more metabolically stable. That is not folklore: the blockbuster “sartan” antihypertensives — losartan, valsartan, candesartan, irbesartan — all carry a tetrazole exactly where a carboxylic acid would otherwise sit. The question is the interesting one: if the acidity is matched, where does the permeability advantage come from?

The answer lives in the anion

Whatever the two neutral acids do, the species that has to cross a membrane — and pay the price for being charged — is the deprotonated anion. To cross a lipid bilayer, an ion must shed part of its hydration shell. The energetic cost of that stripping, the desolvation penalty, scales with how tightly the molecule grips its water, which in turn scales with local charge density: a −1 charge crammed onto a couple of atoms holds water far more fiercely than the same −1 smeared thinly over many atoms.

So the right comparison is not the two acids but the two conjugate bases — benzoate and 5-phenyltetrazolate — and the right question is purely electronic: where does the −1 charge actually go? That is a question about the wavefunction, and it is exactly what Hilbeon was built to answer. Anions, though, are notoriously hard: their loosely bound electron cloud puffs out far from the nuclei, and a standard basis set simply has no functions out there to describe it. You need diffuse-augmented basis sets — the aug-cc-pVDZ family and friends — or the anion's energy and charge distribution are both wrong. Hilbeon now ships those bases, so we can put the two anions on an honest footing.

The reveal: two oxygens vs four nitrogens

Here is the whole story in one sentence. The carboxylate has nowhere to hide its charge: the −1 piles onto just two oxygens, each carrying an ESP charge of roughly −0.81. The tetrazolate, by contrast, is a five-membered ring with four nitrogens, and the −1 spreads over all four — the two most negative N reach only −0.60, and the remaining two sit near −0.24 and −0.25. Same total charge, very different geometry of charge.

That single difference — peak local charge of 0.81 on two atoms versus 0.60 spread over four — is the bioisostere's secret. Lower local charge density means a weaker grip on the surrounding water, which means a smaller desolvation penalty to slip into a membrane, which means better passive permeability — and all of it at matched aqueous acidity, so the binding interaction the chemist worked so hard for is untouched. The tetrazole isn't a better acid; it's a more delocalised one, and delocalisation is what crosses membranes.

The receipts

Three numbers carry the argument: the acidities are matched in water, the gas-phase trend exposes the delocalisation, and the charge maps show where it comes from.

Observable	Carboxylic acid / benzoate	Tetrazole / tetrazolate
Aqueous pKa (experiment)	4.20	~4.5
Hilbeon gas-phase deprotonation energy (kcal/mol)	354.7	342.7
Charge delocalisation: max |q| (atoms sharing −1)	0.81 (2 O)	0.60 (4 N)

Read the table top to bottom. In water the two pK_a values are essentially equal — that is the bioisostere premise, and it is why the swap preserves binding. In the gas phase, where there is no solvent to even the score, the tetrazole is the intrinsically stronger acid by 12.0 kcal/mol, precisely because its anion is stabilised by spreading the charge over four nitrogens. The two facts are the same fact seen twice: the very delocalisation that makes tetrazolate easier to form in the gas phase is the delocalisation that makes it easier to desolvate in a membrane. The charge-delocalisation row is the mechanism made visible.

How to do it yourself

Every number above came from talking to the Hilbeon assistant in plain language — no input decks, no basis-set syntax to memorize:

build benzoic acid from SMILES OC(=O)c1ccccc1, then deprotonate it
run RHF/aug-cc-pVDZ on the benzoate anion
report the ESP atomic charges and render the charge map
repeat for 5-phenyltetrazole -> 5-phenyltetrazolate
compare the gas-phase deprotonation energies

Prefer scripts? Every one of those steps is also a call in Hilbeon's MCP / HTTP API, so screening a whole set of acid bioisosteres is a loop in a notebook or a CI job. Hilbeon runs on its own GPU-accelerated integral engine (McMurchie–Davidson + Rys — no third-party integral libraries), numerically validated to tight tolerances of roughly 1e−9 Ha and bit-identical across machines, even for the diffuse-heavy anions that break lesser setups, so the charges you are reading are the real ones.

The takeaway

A bioisostere is a trade, and the classic carboxylic-acid–to–tetrazole swap is a good one because it pays in permeability without spending acidity. Hilbeon shows you the ledger: matched aqueous pK_a, a gas-phase trend that exposes the tetrazolate's extra stability, and a charge map that explains both at once — −0.81 on two oxygens versus a gentle −0.60 across four nitrogens. Lower local charge density, smaller desolvation penalty, easier passage through a membrane, same binding charge at the target. That is the sartans' design choice, and it is visible in nothing more than the electron density. Point the same engine at your own acidic warts and see which bioisostere actually moves the charge before you make a single analogue.

References

• Bioisosterism, a review — N. A. Meanwell, J. Med. Chem. 2011, 54, 2529 (and the 2018 update, J. Med. Chem. 2018, 61, 5822)
• Tetrazole as a carboxylic-acid bioisostere — C. G. Wermuth review of acid surrogates; tetrazole (acidity & bioisosterism)
• The sartans — losartan, valsartan, candesartan, irbesartan (all carry a tetrazole)
• Desolvation and passive permeability — the lipophilicity / charge-density rationale for ion transport across membranes
• The acid side of the same coin — What ionization state is your molecule at pH 7.4?