The Selmer lab uses integrated structural biology in two different areas, protein synthesis coupled to antibiotic resistance and evolution of protein structure and function. Below you find examples of our past and present research.
Ribosome biogenesis & antibiotic resistance
Translation is a fundamental process in all living organisms, and increased understanding of this process can be used for development of new or modified antibiotics. We ask detailed questions regarding mechanisms of biogenesis, function and inhibition of the ribosome and study molecular mechanisms of antibiotic resistance. We use X-ray crystallography and cryo-EM combined with other structural methods, biochemistry and biophysics to answer these questions.
Ribosomal RNA modification
Posttranscriptional modifications of rRNA play are important in ribosome biogenesis as well as in antibiotic resistance. We are interested in the structure and function of these enzymes, how they recognize their transiently available targets during bacterial ribosome assembly and in the function of the resulting modifications.
On-going work includes cryo-EM studies of ribosomal assembly intermediates and modification-deficient ribosomes.
Many RNA-methylating enzymes consist of a Rossman-fold methyltransferase domain in combination with a substrate-recognition domain. We have solved structures of 23S RNA modification enzymes RlmM (Punekar et al., NAR 2012) and RlmJ (Punekar et al., NAR 2013) and together with Anthony Forster, Uppsala University, clarified their substrate requirements. RlmJ in vitro can methylate a small RNA substrate mimicking H72 or 23S RNA, while RlmM requires a larger substrate and most probably recognizing RNA tertiary structure.
We have also solved the structure of erythromycin resistance enzyme ErmE (Stsiapanava et al., Sci Rep 2019), which N6A methylates 23S RNA nucleotide 2058 in the macrolide binding site, preventing drug binding. Despite showing relatively low level of sequence identity, we conclude that several different enzymes methylating the same position are showing highly similar structures.
To start to get insights into structures of early assembly intermediates that could act as MTase substrates during biogenesis of the 50S ribosomal subunit, we used cryo-EM to determine structures of LiCl core particles (Larsson et al., Biomolecules 2022). Salt-induced removal of ribosomal proteins from ribosomal subunits was early shown to produce particles that were assembly competent. We could now show that salt-washed particles are highly similar to in vivo assembly intermediates and visualize dependencies between complex tertiary RNA structures and RNA-protein interactions.
Elongation factor G and fusidic acid
Fusidic acid (FA) is an antibiotic that locks EF-G to the ribosome in translocation and ribosome recycling. The crystal structure of EF-G with FA on the 70S ribosome (Gao et al. Science 2009) shows how the drug binds in a pocket between domains I, II and III.
Staphylococcus aureus is one of the main targets of FA treatment. In collaboration with Suparna Sanyal, Uppsala University, we have worked out how EF-G from this organism evolves resistance. The crystal structure of EF-G from S. aureus (Chen et al., FEBS J 2010) allowed us to map all known resistance mutations on the structure. Only few mutations are close enough to the FA binding site to directly have an impact on drug interactions. Many of the mutations instead can be classified as affecting EF‐G–ribosome interactions, EF‐G conformational dynamics or EF‐G stability, in agreement with that FA only binds with high affinity to EF-G in a specific state on the ribosome. PDB entry: 2XEX
Biochemical studies and structures of S. aureus EF-G (Kiran Koripella et al., JBC 2012) allowed us to clarify the reasons for fitness loss of the F88L mutant and compensation in the double mutant F88L/M16I. PDB entries: 3ZZ0, 3ZZT, 3ZZU.
The clinically most prevalent type of FA resistance involves an EF-G binding resistance protein FusB (or its homologues FusB/C/D). We solved the crystal structure of FusB and mapped its binding site to domain IV of EF-G (Guo et al., Open Biol 2012). PDB entries: 4ADN, 4ADO. FusB-mediated rescue from FA inhibition during both translocation and recycling was confirmed in vitro in collaboration with the Sanyal group.
An on-going PhD project funded by the Uppsala Antibiotic Center aims to clarify the molecular details of this intriguing resistance mechanism.
Aminoglycoside modification enzymes
A common mechanism of aminoglycoside resistance is the covalent modification of drug molecules by resistance enzymes. We were intrigued by the bisubstrate specificity of ANT3"(9) aminoglycoside adenylyl transferase AadA and decided to elucidate how it achieves specificity for the two dramatically different substrates streptomycin and spectinomycin.
Using a combination of X-ray crystallography, bioinformatics and binding studies, we could show that the different active-site residues are essential for binding and modification of each substrate, while the catalytic reaction occurs at the same site. Crystal structures of AadA with ATP, magnesium and streptomycin explains why ANT3"(9) enzymes have characteristic sequence differences compared to ANT(9) enzymes that perform the same reaction on only spectinomycin (Chen et al., Acta Cryst D 2015; Stern et al., JBC 2018). A crystal structure of ANT(9) from Enterococcus faecium allowed visualization of the spectinomycin binding site, clarifying that spectinomycin and streptomycin extend in different direction from the catalytic site (Kanchugal P & Selmer, AAC 2020).
Structural understanding of protein evolution
We use structural biology in combination with other techniques to understand how evolution acts on protein structure to generate new functions or adaptation to new conditions. In a project funded by the Knut and Alice Wallenberg foundation, we study "novel" functional proteins, enzymes evolving additional or alternative functions and proteins involved in adaptation to new conditions, such as adaptation of herring to low salt conditions in the Baltic. These studies involves proteins from species ranging from bacteria to herring.
Bacteriophage-encoded SAM lyases
A bacteriophage-encoded SAM-degrading enzyme (SAMase) was first discovered in phage T3 in the 1960s and annotated as a SAM hydrolase. In a study led by Dan Andersson, several new phage-encoded SAMases with extremely low sequence similarity to the T3 SAMase were discovered based on their ability to rescue an ilvA knockout strain (Jerlström-Hultqvist et. al., Nature Ecology and Evolution 2018). Degradation of SAM was causing up-regulation of the methionine biosynthetic pathway, where the promiscuous activity of MetB provided rescue of isoleucine biosynthesis.
We have characterized the structure and mechanism of one of these SAMases, Svi3-3 (Guo, Söderholm, Kanchugal P, Isaksen et al., eLife 2021), originating from phage DNA isolated in Svandammen, Uppsala. In collaboration with Johan Åqvist, Dan Andersson, Adolf Gogoll and co-workers, we could show that Svi3-3 and T3 SAMase are SAM lyases and not hydrolases as stated since the 1960s. DFT calculations predicted that the enzyme provides a water-free active site that promotes an intramolecular lyase reaction, and this could be experimentally proven using TLC and NMR. PDB entries: 6ZM9, 6ZMG, 6ZNB
We now aim to further characterize this unexplored family of enzymes with extramely low degree of sequence conservation and their role in the arms race between bacteriophage and their host bacteria.
"Giant" surface proteins of Lactobacillus kunkeei
Lactobacillus kunkeei is the dominating bacterium in the honey gut of bees. In collaboration with the group of Siv Andersson, Uppsala University, we elucidate the structure and function of "giant" extracellular proteins of 3000-8000 amino acids to which L. kunkeei devotes a significant part of their genome.
Phosphoglucomutase 5 from Clupea harengus
In collaboration with the groups of Leif Andersson and Per Jemth, we have investigated the structure and function of PGM5 from Baltic and Atlantic herring (Gustafsson et al., Biomolecules 2020).
We have solved the first structure of PGM5, an intriguing protein that is homologous to the enzymatically active PGM1. In human, PGM5 has been shown to have a structural role, while PGM1 is a well characterised phosphoglucomutase.
The active site of PGM5 is highly similar to PGM1 and binds substrate in a close to identical manner. Yet, we could not detect any phosphoglucomutase activity suggesting that PGM5 also in herring is a structural protein and not an active enzyme.
PGM5 is one of the proteins where a single amino acid substitution has been linked to salinity and/or spawning time of herring. The surfaced-exposed location of this conservative mutation Ala330Val suggests that it may affect a protein-protein interaction rather than the intrinsic properties of PGM5.
Evolution of bifunctionality and specialization in HisA
In a fruitful collaboration with Wayne Patrick, now at Victoria University of Wellington, and Dan Andersson, Uppsala University, we have worked out the structural mechanism of how the amino acid biosynthetic enzyme HisA from Salmonella enterica can evolve bifunctionality to also take the role of TrpF and further specialize on either of the two functions. The initial structural characterization of HisA showed a two-step binding mechanism of substrate (Söderholm et al., JBC 2015). PDB entries: 5AHE, 5AHF, 5A5W.
Structural and biochemical characterization of a large set of real-time evolved HisA mutants (Newton et al., PNAS 2017) showed that bifunctionality is based on two distinct conformations of the active-site loops. Protein dynamics using NMR (Patrik Lundström, Linköping University) allowed us to prove the critical role of a single point mutation in providing the flexibility essential for bifunctionality. Interestingly, we observed two different mechanisms of how specialisation on the TrpF reaction evolved: stabilisation of the required active-site loop conformation and closing of the active site to prevent binding of the larger HisA substrate. Biochemistry and in vivo experiments showed that beyond ae activity threshold, improved enzymatic activity did not lead to increased growth rate. PDB entries: 5G1T, 5AC7, 5AC8, 5L9F, 5AB3, 5G1Y, 5L6U, 5G4E, 5G4W, 5G2I, 5G2W, 5G2H, 5G5I
Based on a serendipous finding in the previous study, we also solved the crystal structure of IGPS (or TrpC) from Pseudomonas aeruginosa, the enzyme catalysing the step after TrpF in tryptophan biosynthesis, and characterized its surprising substrate promiscuity (Söderholm et al., JBC 2020). Even though IGPS is classified as a decarboxylase, decarboxylation not completely essential. We showed that the enzyme in addition to its well characterized formation of indole-3-glycerol phosphate (IGP) from the natural substrate also can catalyse formation of the same product from decarboxylated substrate. PDB entry: 6Y88