Around 2012 my group got first involved with photoenzymes – that year, I coincidentally came across a Highlight Commentary by Wolfgang Gärtner, one of my former PhD supervisors (1). The commentary was about a paper published in Nature in 2008 by the groups of Rienk van Grondelle and Marie Louise Groot. They analyzed the conformational changes of a light-dependent protochlorophyllide oxidoreductase (LPOR)(2), an enzyme involved in plant greening. To cut a long story short – Marco Kaschner (Fig. 1a), a PhD student in my lab at that time, was looking for a side project that dealt with enzymes and light. Hence, we thought we will give LPORs a try. Back then, it was generally assumed that LPORs are only found in oxygenic phototrophs such as plants and cyanobacteria, while phototrophic microbes such as anoxygenic phototrophs lacked those enzymes. More importantly, no crystal structure of an LPOR had yet been solved, which is still true as of 2019, and I thought having new enzymes from alternate sources on hand might help to remedy that situation. With that in mind, I searched genome databases for an LPOR outside of oxygenic phototrophs – and, voila, readily found one candidate gene in the anoxygenic phototrophic proteobacterium Dinoroseobacter shibae. Marco, together with a talented undergraduate student, Judith Schneidewind (back then Judith Krause), subsequently cloned the D. shibae gene, expressed and purified the protein. After having solved the problem of obtaining the commercially not available protochlorophyllide (Pchlide) substrate (Fig. 1b), we set out to test the function of the recombinant protein and were immediately able to show light-dependent activity. At that time, I really thought we were onto something and contacted the group of Arndt von Haeseler, evolutionary biologists from Vienna, to look at the phylogeny and evolution of our newly found proteobacterial LPOR. After that, it still took Marco and Judith (who was back for her MSc work) about 2 years before we had a paper ready for publication that verified that functional LPORs are found in anoxygenic phototrophic microbes (3). Having done all that preliminary work, I submitted a grant proposal to the Deutsche Forschungsgemeinschaft (DFG) to work on the structure, function and phylogeny of the LPOR enzyme system, i.e. using a hitherto untapped source of LPORs for crystallization trials. Luckily, our work was funded and I could hire Judith to work on the project for another three years, this time pursuing her PhD.
When she started, we thought that apart from novel proteobacterial LPORs, we should also include some reference enzymes from oxygenic phototrophs in our study – and this is how we ended up working with the Thermosynechoccus elongatus LPOR (TeLPOR) that is the subject of our present paper. Since everything had worked out so well for her thus far, it was finally time for a bit of bad luck. Judith, with the help of the group of Renu Batra-Safferling, a crystallography working group from Forschungszentrum Jülich, tried thousands of different crystallization conditions, but got basically nothing useable. At that time, Judith became quite frustrated. Together with Andreas Stadler (Forschungszentrum Jülich), a SAXS and Neutron scattering (4) expert, we frequently went to the Synchrotron in Grenoble to collect small-angle X-ray scattering (SAXS) data. Since we had an LPOR sample ready, incidentally TeLPOR, we took it with us. At the end of a 24-hour shift, we put our TeLPOR apoprotein sample into the beam and, surprisingly, the data looked very good – no detectable radiation damage and most importantly no visible aggregation. Having been awake for more than 20 hours, I just thought “OK, nice, let’s see how far we get”. We took the data home and analyzed it. Interestingly, unlike other short-chain dehydrogenase enzymes, which are often oligomeric, TeLPOR seemed to be a monomer. It got more and more interesting when we obtained the first ab initio models that showed a bowling-pin like shape for the molecule. Back then, I thought that we might still get the structural insight, even though at low resolution, that we set out to acquire initially. Things got trickier when we checked the homology models that we had available back against our SAXS data. They simply did not fit the data well. We soon realized that the main problem was the missing C-terminus, which was not present in the published models, but was suggested to be α-helical in all of our predictions. Luckily, friends of mine, Marco Bocola and Mehdi D. Davari, computational biologists from the RWTH Aachen, had just started to cooperate with Marco Garavelli from Bologna on a computational spectroscopy/photochemistry project dealing with LPORs (5). Therefore, I contacted them and asked them, if they could help us with modelling full-length TeLPOR. When Marco Bocola started to add a C-terminal helix to our models, the fit against the experimental SAXS data immediately got better. We were all very happy with that and thought, well, we might get a physically feasible model of the apoprotein soon enough, but what about the substrate/cofactor/enzyme holoprotein complex? Hence, Judith optimized conditions to assemble it in vitro as we had SAXS beamtime coming up soon. Andreas took a sample along and measured it (Fig. 1c). He passed the data on to us, and I was quite surprised that it looked completely different from our apoprotein data, either the holoprotein monomer was much larger than the apoprotein, or the protein oligomerized! To verify one or the other we needed independent data. At that time, Frank Krause from the company Nanolytics in Potsdam called me. We had worked together in the past. He told me that they had recently developed a novel technique called multi-wavelength absorbance analytical ultracentrifugation (MWA-AUC) (6) that allowed them to extract complete UV/Vis spectra from the sedimenting material during the AUC run – and they needed a real-life system to test it on. Well, I thought, this would be perfect to study the oligomerization state of our TeLPOR holoprotein. So, we sent them samples! After Frank had analyzed the runs, we had our proof – TeLPOR dimerized upon holoprotein formation. It subsequently still took about a year until we had a manuscript ready to submit, in which we checked, modelled and simulated all alternative possibilities that we thought likely (some were even suggested during the review process, and we would like to thank the anonymous reviewer for making us aware of them).
In the end, our story shows how science can develop from the interplay of luck, coincidences and most importantly the interactions of dedicated people – and that is what makes research fun. We think that the consensus model for the cyanobacterial TeLPOR enzyme represents an important initial step towards understanding the structure and function of this so far structurally elusive enzyme family.
Figure 1: Working in the dark with photoenzymes is fun. a, Marco in the dark room purifying protein. b, Judith extracting the Pchlide substrate from de-etiolated wheat seedling (we later on had to switch to a different Pchlide source, because of mouldy problems. c, Andreas recording TeLPOR SAXS data at BM29 in Grenoble, holding the first 3-dimensional model that he “generated” in his hand.
(1) Gärtner W. Enzyme catalysis "reilluminated". Angew. Chem. Int. Ed. Engl. 2009. 48: 4484-4485.
(2) Sytina OA, Heyes DJ, Hunter CN, Alexandre MT, van Stokkum IH, van Grondelle R, Groot ML. Conformational changes in an ultrafast light-driven enzyme determine catalytic activity. Nature. 2008. 456: 1001-1004.
(3) Kaschner M, Loeschcke A, Krause J, Minh BQ, Heck A, Endres S, Svensson V, Wirtz A, von Haeseler A, Jaeger KE, Drepper T, Krauss U. Discovery of the first light-dependent protochlorophyllide oxidoreductase in anoxygenic phototrophic bacteria. Mol. Microbiol. 2014. 93:1066-1078.
(4) Stadler AM, Schneidewind J, Zamponi M, Knieps-Grünhagen E, Gholami S, Schwaneberg U, Rivalta I, Garavelli M, Davari MD, Jaeger KE, Krauss U. Ternary complex formation and photoactivation of a photoenzyme results in altered protein dynamics. J. Phys. Chem. B. 2019. 123: 7372-7384.
(5) Gholami S, Nenov A, Rivalta I, Bocola M, Bordbar AK, Schwaneberg U, Davari MD, Garavelli M. Theoretical model of the protochlorophyllide oxidoreductase from a hierarchy of protocols. J. Phys. Chem. B. 2018. 122: 7668-7681
(6) Pearson JZ, Krause F, Haffke D, Demeler B, Schilling K, Cölfen H. Next-generation AUC adds a spectral dimension: development of multiwavelength detectors for the analytical ultracentrifuge. Methods Enzymol. 2015. 562: 1-26.