Behind the papers

AfG3PS

Given that this was the longest and largest project I've led from the desk and the bench, I could share tons of stories and anecdotes. To keep this post to a reasonable length, let's talk about luck and serendipity. Key moments in this project came down to luck and sharp eyes. For example, I stumbled across the fact that AfG3PS makes (Z)-configured products by accident while doing an experiment that did not even make it into the final paper. Towards the end of the project, I happened to have a preparation of ca. 1:1 geranyl/neryl pyrophosphate from a quasi-unsuccessful synthesis on hand, which I had no use for. So I figured that this sample would make a great control experiment to show that AfG3PS prefers (E)-substrates over (Z)-substrates, leaving the latter untouched (at least that was my working hypothesis at the time). Imagine my surprise and shock when I saw the crude NMR of the reaction product which very clearly had (Z)-product in it! At that point, I immediately recognized the (Z)-product for what it was because I had happened to have made phytol glycerol recently, which gives an (E)/(Z)-mixture and presented ideal reference NMR data. So my control experiment gave the opposite of the expected outcome. Fortunately for me, my very early decision to make all synthetic substrate analogues as (E)/(Z)-mixtures paid off massively. With those substrates in hand and knowning what to look for and where, interrogating AfG3PS's alkene selectivity across the substrate scope was quick and straightforward. Truth be told, without that "failed" control experiment, there's a chance I might have missed this key observation and chalked the minor amounts of (Z)-isomer in most crude products up to impurities in the baseline. But as it happened, I had done the right experiment at the right time. Obviously, one only realizes these things in hindsight. There is no way to plan out unforeseen events and breakthroughs. By definition, exploratory science happens off the beaten path, on deer trails you did not know existed but happened to spot while walking past.

N7X

N7-Xanthosine

Another serendipitous finding! Back in 2021, when we were trying to develop new strategies to shift glycosylation equilibria, we discovered that something weird was happening in some of our reactions. The UV spectra made no sense, mass balances didn't add up and overall the whole system did not behave according to the thermodynamics it should. It took a while (and a 3-day 13C NMR) for us to figure out that our enzymes had significant promiscuous activity and made the "non-native" ribosylation isomer of xanthine, N7-xanthosine. This nucleoside had a few more surprises for us in store and with each new result came new questions: Why does that compound have such unusual absorption properties? Why do some nucleoside phosphorylases make the compound while others (mostly) don't? Is that reactivity limited to xanthine or could other, similar nucleobases yield non-native regiomeric nucleoside? The list goes on... For that reason, we decided to publish an "unfinished" story with lots of potential for future studies.

Biased borate eserification

Biased borate esterification

This project was a ton of fun. And a ton of frustration. We (as in Lea and I) initially ran across biased borate esterificiation by complete accident in the summer of 2021. We were screening buffer systems for the PUB module and thought we could just give borate (pKa = 9.2) a shot as that seemed a sensible choice of buffer at pH 9. Borate being a relatively temperature-insensitive and non-complexing buffer would have been added boni. In borate buffer, PUB behaved really weirdly. At 200 mM borate, almost no reaction happened. Just to make sure we hadn't messed up that experiment, we did a follow-up experiment where we titrated in borate with external glycine buffering. The result was quite straightforward: Borate slowed down the reaction rate massively and led to lower apparent equilibrium states. We also did a quick experiment to show that preincubation of the phosphorylase in borate does not denature the enzyme (as it retained activity). At that point we were convinced that we had observed a highly unusual and pretty cool effect, but ultimately decided to let it rest and focus on the task at hand (Lea also needed to finish her master thesis and diving down this rabbit hole could/would have been a major distraction). Fast forward about three months to the fall of 2021: Lea had submitted her thesis, I was mostly finished with the paper about PUB and could not help myself from wanting to figure out what was happening with our phosphorolysis system in the presence of borate. After all, this effect seemed to be completely unprecedented, while similar observations in other biocatalytic systems in the literature fell way short of a satisfying description of what was actually happening. I was hooked. That process led me on a four-month detour from my main project, which ended up being an immensely rewarding process. Figuring out roughly what was happening (bromouridine forms borate esters) was relatively straight forward. Figuring out what exactly was happening (they're 2',3'-borate esters which act as pseudo-non-competitive enzyme inhibitors and drain mass balance from the phosphorolysis equilibrium system) took me eliminating several working hypotheses. Figuring why it was happening (entropically favored formation of nucleoside borate esters vs. borate esters of the sugar phosphate) was only possible through fruitful collaborations. Felix (Brandt) nicely backed up a lot of experimental observations with DFT data (although these highly charged systems were quite challenging to describe); and his MD simulations helped us rationalize the mechanism for enzyme inhibition. Sebastian really helped our understanding of the energetic driving forces with key NMR experiments. Sarah and I independently (and unsuccessfully) gave transglycosylations a shot and she (successfully) carried conditions from analytical-scale experiments over to preparative reactions. We learned so much during that project and I firmly believe it led to everyone involved becoming a better scientist. Also, we discovered and elucidated a pretty weird mechanism and found a way to shift glycosylation equilibria simply through the addition of borate. That is, arguably, quite neat. One more anecdote: Coming up with the (supplementary) equations to describe the thermodynamic behavior of these systems was not trivial. It may seem that way in hindsight but in the moment it was certainly not. Describing the kinetics through thermodynamically controlled formation of an enzyme inhibitor (equation (S35)) was relatively straightforward, but the coupled equilibrium systems (equation (S18)) were not. I spent an entire day on my institute desk in fall trying to work out the solution from the set of fundamental equations (mass balances, equilibrium constants and simplifications; equations (S1) to (S7)) but simply could not. I eventually yielded and thought "ah well, I'll just take this problem home and put it on my whiteboard. Perhaps I'll solve it through mere exposure." So those equations went on my whiteboard at home, where I walked past them multiple times a day. They stayed there for a month, which turned into two months, which turned into more than three months. I thought about erasing them a few times but for some reason they stayed up there. I'm glad they did. One Saturday afternoon in early spring, I had a real Eureka-moment. Everything suddenly fell into place, I solved, checked and re-checked the thing in less than an hour and had fitted all relevant experimental data before dinner time. Sometimes, patience pays off?

Isometic poins

Isometric points

How can I provide value to the community? What's an interesting topic that most people may not have heard or thought about? What would be useful to write about? - Those were questions I was asking myself when I first considered topics for an article for ChemBioChem's ChemBioTalent section. The invitation to the 2022 ChemBioTalent cohort came in January of 2022 and I was immensely honored. All of the people from 2018 ChemBioTalents that I knew were now at least Assistant professors - being award the same honor was a great confidence-booster. So I knew I was not going to turn this invitation down. However, I also knew that we did not have a research paper in the pipeline that we could submit to ChemBioChem this year. In addition, I was scheduled to become to a dad in July so the extent of my productivity for the second half of the year was a bit in question. At the same time, I did not want to write a review ... i) because there are already way to many reviews out there and most of them are not good and ii) because writing a review I would have been happy with simply would have been too great of a time investment. So, how can I provide value to the community, concisely? I let the question rest. For a while. It took until fall of that year for me to realize that there is a central and very (!) overlooked topic that I have something valuable to say about that hasn't been said before. Meet isometric points. Turns out, our previous work on isosbestic points carries over really well to all sorts of spectroscopic techniques. Doesn't matter what the technique is or what the data describe, as long as there's a recurring intercept and mass balances apply, the same simple math can be used to describe these systems. In fact, since writing the article, I have come across a few even weirder examples of isometric points in X-ray absorption plots (10.1021/jacs.2c09477) and circular dichromism diagrams (anie.201508555). I really hope that the vocabulary I laid out in my short introductory review will be helpful to the community to talk about these phenomena and that the simplified math will make highly convoluted spectra more tractable.

HHDHs

G-type halohydrin dehalogenases

This one was a while in the making and my favorite part of this project was also the most challenging one. Jenny and I had originally worked on that project when I was still a student in Braunschweig but we only managed to finish it up by the time I was back as a postdoc (admittedly, that sounds like a much larger gap than it actually was). What I enjoyed most about this project – and what kept me heads-down and zoned-out busy for an entire Saturday afternoon – was the analysis of the product mixture of 5d (which included four different isomers). Without reference data or clean separation on chiral GC (we tried. hard.), this one came down to assigning signals of four different compounds from a mixture of said for compounds in very similar quantities. An NMR riddle! If you like the IOCB Prague NMR challenge (http://nmr-challenge.uochb.cas.cz/), you would have loved this problem as I did!

PUB: Alt. reagents

PUB: Alternative assay reagents

We took a slight detour with this project and it did not go as intended. This really short paper exclusively reports on negative data. Still, we felt like it was important to report these data as others might have the same idea and us reporting these results could easily save them a lot of frustration. The main weakness of the PUB module (if we can even call it that) is the rather modest difference of the extinction coefficients between bromouridine and bromouracil (5.4 /mM cm at 304 nm and pH 9). We figured we could probably improve on that by using other uridine analogues, such as iodouridine and ethynyluridine. After all, our reasons not to use those in the first place were i) their slightly higher pKa values and ii) limited commercial availability. We were open to accept trade-offs on either front if it meant significantly higher extinction coefficients. So we bought some iodouridine and Charity made some ethynyluridine from it. We characterized those compounds and found that our initial hypothesis held up: They showed greater extinction coefficient differences during their phosphorolysis. However (and that was rather unexpected), both compounds were completely unsuited to replace bromouridine in the PUB module. Not only was their conversion by the key phosphorylase much (!) slower, but neither compound showed an isosbestic point of phosphorolysis under moderately alkaline conditions. The reason for that was one we could have never have foreseen: They possess too great of a gap between the pKa of the nucleoside and the nucleobase. See, for a transformation to exhibit an isosbestic point, it generally needs to feature two (although cases with more than two are possible, if highly unlikely) UV-active components that have intercepting absorption spectra. Bromouridine and bromouracil (both have a pKa right around 8) are present to ca. 90% as their anions at pH 9. The isosbestic point we used in the PUB module is the one generated by the conversion of 90% bromouridine anion to 90% bromouracil anion. The iodouridine/iodouracil and ethynyluridine/ethynyluracil pair each have a considerable (0.18 and 0.16) gap between their pKa values. Combined with the generally higher pKa values, we effectively observed spectra resulting from a non-constant degree of deprotonation, resulting in non-intercepting spectra and highly non-linear absorption responses. That ended our hopes of improve assay reagents right there, but we learned something in the process and hope that others may profit from these lessons.

PUB

This project was born solely out of necessity but turned into one of our most productive and impactful endeavors. When I came off my PhD and took on a new project, I also took on new chemistry, new challenges and new methods. The key enzymes of my new project(s) are ones that, as a byproduct of their main reaction, release inorganic pyrophosphate. Since reaction monitoring based on the more interesting substrates and products of that transformation was not really feasible, we figured we could probably just detect the released pyrophosphate (a relatively common biomolecule, after all) and use that for kinetic experiments. However, a long hard look in the literature and a few trial runs in the lab quickly told us that all available methods to detect pyrophosphate in biochemical assays either did not work particularly well or were complete unsuited to the types of questions and throughput that we wanted to work with. Unfortunately, those problems carried over to the detection of orthophosphate (which is easily available from pyrophosphate through quantitative enzymatic hydrolysis). So, although we did not necessarily want to at the time, Lea and I had to back to the drawing board and invent a new method. Fortunately for us, we did not have to look very far, because my first idea worked out right away: "How about we just use a nucleoside phosphorolysis to detect orthophosphate? If we use a pyrimidine* with a sufficiently low pKa, we should be able to run that assay continuously." Turns out, 5-bromouridine works fantastically as an assay reagent and after brief optimization we had a running system in our hands. Since we knew we were probably going to use that method a lot going forward, we spent a good amount of time on additional characterization, benchmarking and trial runs. We tested realistic use cases, different modes of application, different scales and really just played around with conditions to identify potential weak spots. Truth be told, we did not really find any. The method, which we eventually called PUB (for simplicity's sake; its long name is more than unwieldy), is incredibly robust and versatile. To this day, I am glad we spent that much time on optimization and bench tests as I still use PUB today and massively profit from knowing its limits and capabilities. I cannot recommend this method enough for all scenarios involving phosphate detection and I'm more than happy to help out anyone interested in giving the method a shot for their applications. *Why a pyrimidine, you ask? Because pyrimidines are much more soluble than purines and display a far greater redshift upon deprotonation than purines do. In addition, they have much higher equilibrium constants of phosphorolysis (giving a much greater and more linear dose-response relationship) than purines. I will admit, though, that our initial choice of 5-bromouridine (while it was driven by very rational considerations) did involve a fair bit of luck.

Colos

Coloring Chemistry

Have you ever printed out an interesting article in greyscale and taken it with you on the train, only to discover that you stand no chance of deciphering Figures 3 and 4 because it's all a homogenous grey blur? - I have. While that's annoying, I have the luxury of going back online and looking at the original colored version of that figure on the journal website. Granted, I am unlikely to do that, but I could if it was important enough. However, some people can't. Color vision deficiencies are much more common in society that one would think. Around 4% of the human population have some sort of color vision deficiency. Although true color blindness is rare, protanomalies (reduced ability to see/discern red) and deuteranomalies (reduced ability to see/discern green) are quite common. Both of those conditions result in people being unable to discern between shades of red and green - which are really (!) common in the natural world around us and occur very often as part of design elements in scientific figures. By mindlessly using color in scientific illustrations, we - the scientific community - actively excludes people with color vision deficiencies from access the literature. Using colors (including reds and greens) is not problematic per se, but using color as an information-coding element is. But I digress. All of that is in the full text. You came here for untold stories behind the paper. Angewandte has, in recent history, developed a somewhat unusual stance on color use. As one of the last outlets to distribute printed copies, color at Angewandte still costs money. Therefore, a good portion of Angewandte's content over the past decade does not feature color. However, as the proportion of colored illustrations increased in recent years, so did the proportion of articles featuring illustrations inaccessible to people with color vision deficiences. In 2022, around 70% of all articles published across the chemical sciences featured such problematic illustrations. The rate at Angewandte was even higher. That, coupled with the opinion-type formats run by the journal, made Angewandte the perfect address for Fabio and me to publish an article to sensitize the chemical community to the issue of mindful color usage. However, publishing that article did not go particularly smoothly. In fact, publishing that article involved (among other things) a desk-rejection as well as us, ironically, having to publish the article behind a paywall because the terms of the DEAL agreement did not foresee a case like this. However, we got there in the end and the article is now (and hopefully indefinitely) available free to read.

Nucleside diversification

Nucleoside diversification

I am convinced that the best projects start with an element of serendipity. This project was no different. It started when I (completely by chance) saw that Daniel's group had recently published on cyclopropylated nucleosides. This happened at a time (late 2019) when we had just come off our projects looking into the thermodynamics of nucleoside phosphorolysis and transglycosylation processes. I was therefore immediately curious if we could get our hands on some of Daniel's material to see if our enzymes converted those types of compounds. So I contacted Daniel and found out that, unfortunately, they didn't have any of the particular compound of interest to us left. However, they had 3 mg of a degradation product, which was an interesting compound in its own right. Instead of a cyclopropyl group, that nucleoside had a methyl group at its 4' position. To me, that felt like the perfect opportunity: Testing previously unknown substrates for nucleoside phosphorylases, a chance to test our understanding of transglycosylation thermodynamics on an uncharted system, and the opportunity to address a synthetic challenge in diversifying late-stage nucleoside synthons. I was hooked. So I travelled to Braunschweig, got the compound, and got right to work on it. All my initial experiments on that compound worked! I screened a small selection of phosphorylases that we had already sitting around and immediate got a hit. I could show that the phosphorolysis of this nucleoside analogue is thermodynamically controlled, with enthalpy and entropy values almost identical to the ones of natural uridine. I could also show that transglycosylations (both to pyrimidines and purines!) worked and got preliminary HRMS data. The project almost felt too easy. I reached out to Daniel again, who brought in Anna to make more of that compound. Then Covid hit. Everything shut down and the project was put on hold. We thought we would have to delay indefinitely, but when work slowly resumed, Anna got right back to it. She finished the synthesis (hats off to smoothly pulling off 10 steps) that summer and I got to work on more kinetic data and the full range of thermodynamic data was accessible to us. We brought in Sebastian for some really insightful NMR experiments and Rita to explain why seemingly identical active site architectures discrimited so dramatically for the presence or absence of the 4'-methyl group. The whole project was done around February of 2021 (I got slightly delayed by writing my thesis as well as a grant proposal which ended up securing my first independent funding later that year) and the preprint appeared on the day of my PhD defense. Looking back, this project was the first true interdisciplinary effort I coordinated and all of us learned a ton in the process. And it all started the way all cool projects do: With an element of serendipity.