From Latte to Lab: A Toy Model of Enzyme Inhibitor Screening
Building enzyme screening workflows with everyday chemistry
Lately, we’ve been focused on building AI tooling to enable game-changing workflows in the pharmaceutical industry, which has pushed our hardware team headfirst into the deep end of wet-lab research and drug development.
Now, I may be lactose intolerant, but my brain conveniently forgets this when ice cream is involved. Thankfully we stock Lactaid at the office, right next to the coffee, tea, and energy drinks. Queue the pizza parties. When we were discussing the possibility of a pharma-centric lab demonstration, and the low hanging fruit of an AI-guided enzyme inhibitor screening demo (a crucial step in drug development workflows), my go-to supplements came to mind first. A trip to Whole Foods and a week of testing later, we’re now able to speak firsthand about enzyme inhibitor testing.
Our toy model of enzyme inhibitor testing mirrors common procedures used in early drug discovery campaigns, but leverages safe, commercially available chemicals to develop relevant testing and analysis frameworks within the scope of our automated laboratory.
The test
Caffeine is known to interfere with some enzymes, but is it impacting the efficacy of my Lactaid? A previous study (Horne, ACS Omega, 2020) showed that caffeine does inhibit E. coli beta-galactosidase, a lab-grade bacterial enzyme in the lactase family. However, commercial dietary supplements contain lactase derived from fungal sources, not bacteria. So, the question remains as to whether this inhibition translates to commercial products like Lactaid.
To find out, we used:
lactase (from Lactaid) as the enzyme,
ONPG (o-Nitrophenyl β-D-galactopyranoside) as the substrate, and
caffeine as the drug candidate for inhibition.
Lactase hydrolyzes ONPG to produce galactose and o-nitrophenol, the latter of which is optically yellow, enabling color change to serve as a proxy for enzyme activity. We monitored the reaction optically in a standard 96-well plate with varying concentrations of lactase and caffeine. Instead of using a plate reader, we leveraged our existing lab infrastructure to capture images every few seconds over a roughly 100-minute experiment.
The results
It turns out that our autonomous imaging platform is surprisingly powerful for capturing enzyme kinetics. Custom python code processed each image, plotted reaction curves, and analyzed the resulting dataset with minimal user intervention. When the testing was done, a few clear patterns emerged: wells with more lactase changed color faster, wells without lactase stayed relatively steady, and wells with more caffeine appeared to yellow slower than their caffeine-free counterparts.
Overall, we observed:
0 - 40 mM caffeine: essentially no effect
~ 60 mM caffeine: slight inhibition
~ 80 mM caffeine: ~ 20 % inhibition
Fitting a dose–response curve suggests that achieving 50% inhibition would require caffeine concentrations far higher than anything tested here (nearly 500 mM). As a result, we can only conclude that the apparent inhibition constant (Ki) necessary for a thorough kinetics analysis is much greater than 82.4 mM. Drawing more precise kinetic conclusions would require substantial expansion of the methodology.
That brings us to some important caveats. This was a small, exploratory experiment, not a regulatory-grade enzyme study. Our chemicals came from pills, not purified sources, we ran a small number of variations with no technical replicates, and we need to extrapolate for our estimated inhibition constant. Caffeine vs. lactase is an interesting trend to consider, but drawing scientific conclusions would require bolstering our dataset significantly.
If we were to approach this in a scientifically rigorous way, a full kinetic study is a must. For that we’d need to:
Test multiple substrate concentrations
Add technical replicates and randomize well positions
Obtain high fidelity measurements of early-time linear rates
Increase caffeine concentration high enough to approach 50% inhibition
Use purified lactase and caffeine of known concentrations, rather than impure supplements
The big takeaway
In this experiment, it took extremely high levels of caffeine, dramatically beyond the concentration found in your morning latte (~ 50x), to reduce lactase activity by less than 20%. So, if we’re drawing conclusions from a single exploratory study (please don’t), it appears we have nothing to worry about simultaneously chugging coffee and devouring Caffé Panna here at Mirror Physics HQ.