Physical Chemistry · Anno MMXXVI

Zeno

How much of an enzyme reaction runs through the quantum tunnel, and how much of its isotope effect does not
Sabrina Hackfort · Independent Researcher · ORCID 0009-0007-7458-2487
❦ ❦ ❦
Built on the Hackfort Equation

ZENO computes the quantum fraction of the Hackfort Equation, the part of a reaction rate that exists only because of tunneling:

η = 1 − 1/κ,   κ = kwith tunneling / kwithout

When hydrogen is replaced by deuterium, an enzyme slows down. This kinetic isotope effect is the classic fingerprint of quantum behaviour, and it is often read as a direct measure of tunneling. It is not. Most of it comes from zero-point energy: the C–H bond vibrates higher than the C–D bond and loses that vibration on the way over the barrier. Only a part comes from the tunnel itself. ZENO keeps the two apart:

KIE = KIEZPE · κH / κD

zero-point part · tunneling part (symmetric Eckart barrier, effective mass exponent s)

The Old Assumption and the Real Enzymes

A textbook shortcut says that deuterium lowers the tunneling frequency by √2, as if only the transferred atom moved. In the language of ZENO this is a mass exponent s = 1. Published simulations of four enzymes tell a different story: at the transition state, much of the donor–acceptor frame moves with the hydrogen, and s lies between 0.07 and 0.31. The plate shows what this means for the tunneling part of the isotope effect.

mass exponent s →
barrier frequency ν̃ →
old assumption s = 1
Heavy isotope
Height: the tunneling factor κH/κD (or κH/κT) at 300 K for a barrier of 14.5 kcal/mol, on a logarithmic scale, over the mass exponent s and the barrier frequency ν̃. The back edge (s = 1) is the old assumption. Red: enzymes from published QM/MM simulations, placed at the s their own data require. They all sit near the front, where the tunneling factor is small. Drag to turn the plate.

What the Simulations Say

Four enzymes, each with a published ensemble of QM/MM simulations including multidimensional tunneling. The quantum fraction η comes straight from their tunneling coefficient κ. The split of the isotope effect comes from the same papers, which report it with and without tunneling.

EnzymeTκHηtunnel part of KIE
Formate dehydrogenase278–318 K6,3–4,30,84–0,77×1,09–1,16 (H/T)
Dihydrofolate reductase, E. coli278–318 K3,77–2,840,73–0,65×1,07–1,14
Dihydrofolate reductase, thermophilic278–338 K5,25–2,000,81–0,50×1,02–1,08
Thymidylate synthase278–313 K9,7–16,30,90–0,94×2,1–2,7 (H/T)

Sources: Roca et al., PCCP 2018 · Pu, Ma, Gao, Truhlar, J. Phys. Chem. B 2005 · Pang et al., JACS 2006 (via Masgrau & Truhlar, Acc. Chem. Res. 2015) · Kanaan et al., JACS 2011. For a proton transfer with deep tunneling, aromatic amine dehydrogenase, the tunneling part rises to ×5 (Masgrau et al., J. Phys. Chem. B 2007).

The reading is clear. In these enzymes between half and almost all of the rate goes through the tunnel, yet the tunnel contributes only a small factor to the isotope effect. Isotope effects and tunneling are different quantities. Treating one as the other gives wrong numbers, and the first version of this work made exactly that mistake; it is withdrawn and documented.

From Physics to Medicine

Deliberate deuteration can make a drug last longer. Its real benefit is smaller than the isotope effect suggests, because the body switches to other metabolic routes. ZENO-DX models this switching. It was tested on the approved drug deutetrabenazine with independently measured inputs only: the isotope effect in human liver microsomes and the interaction study with the CYP2D6 blocker paroxetine. For the active metabolite β-HTBZ it predicts the gain in exposure correctly (×2,2 predicted, ×2,36 measured). For α-HTBZ it does not (×1,2 predicted, ×2,32 measured); the body's isotope effect there is larger than the one measured in vitro. An earlier, more favourable version of this test used an assumed isotope effect and counted a calibration value as independent. It is withdrawn.

What It Does Not Claim

ZENO uses a one-dimensional Eckart barrier. Its barrier frequency is an effective fitted parameter, not a computed one. With the published data, its prediction of how κ changes with temperature cannot yet be told apart from a constant κ: the simulations scatter too much. The mass exponent s and the zero-point share f depend on the enzyme, so without system-specific quantum chemistry the total isotope effect is only a range. The biodiversity and cryptography readings of the Hackfort Equation are not part of ZENO and remain untested.

Code

Rust core and Python reference give identical numbers. Every literature value is stored with a verbatim quote from its source. The repository and paper are being prepared for publication.