Adaptation to genome decay in the structure of the smallest eukaryotic ribosome – Nature.com

Isolation of E. cuniculi ribosomes

To improve our understanding of the evolution of proteins and nucleic acids in intracellular organisms, we set out to isolate E. cuniculi spores from infected mammalian cell cultures to purify their ribosomes and determine the structure of these ribosomes. Large quantities of microsporidian parasites are challenging to produce because microsporidians cannot be cultured in a growth medium. Instead, they grow and reproduce only inside their host cells. Therefore, to produce E. cuniculi biomass for the ribosome purification, we infected mammalian kidney cell line RK13 with E. cuniculi spores and cultivated these infected cells for several weeks to allow for E. cuniculi to grow and reproduce. Using approximately half a square meter of the infected cells monolayer, we could purify about 300mg of microsporidian spores and use them for ribosome isolation. We then broke the purified spores with glass beads and isolated crude ribosomes using a stepwise fractionation of lysates with polyethylene glycol. This allowed us to obtain approximately 300g of crude E. cuniculi ribosomes for structural analyses.

We then used the obtained ribosome sample to collect cryo-EM images and process these images with masks corresponding to the large ribosomal subunit, the head of the small subunit, and the body of the small subunit. In doing so, we collected snapshots of ~108,000 ribosomal particles and calculated cryo-EM maps at 2.7 resolution (Supplementary Figs.13). We then used the cryo-EM maps to build the model of rRNA, ribosomal proteins and the hibernation factor Mdf1 bound to E. cuniculi ribosomes (Fig.1a, b).

a The structure of E. cuniculi ribosomes in complex with the hibernation factor Mdf1 (pdb id 7QEP). b The map of the hibernation factor Mdf1 bound to E. cuniculi ribosomes. c Secondary structure diagrams compare rRNA reduction in microsporidian species with known ribosome structures. The panels indicate the location of rRNA expansion segments (ES) and ribosomal active centers, including the decoding site (DC), the sarcin-ricin loop (SRL), and the peptidyl-transferase center (PTC). d The electron density corresponding to the peptidyl-transferase center of E. cuniculi ribosomes shows that this catalytic site has the same structure in the parasite E. cuniculi and its hosts, including H. sapiens. e, f The electron density corresponding to the decoding center (e) and schematic structures of the decoding center (f) illustrate that E. cuniculi have U1491 residue instead of A1491 (E. coli numbering) in many other eukaryotes. This variation suggests that E. cuniculi may have sensitivity to the antibiotics targeting this active site.

Compared to the previously determined structures of V. necatrix and P. locustae ribosomes (both structures represent the same family of Nosematidae microsporidians and are very similar to each other)31,32, E. cuniculi ribosomes underwent further degeneration of numerous rRNA and proteins segments (Supplementary Figs.46). In rRNA, the most prominent changes include the complete loss of the 25S rRNA expansion segment ES12L and partial degeneration of helices h39, h41, and H18 (Fig.1c, Supplementary Fig.4). In ribosomal proteins, the most prominent changes include the complete loss of protein eS30 and truncations in proteins eL8, eL13, eL18, eL22, eL29, eL40, uS3, uS9, uS14, uS17, and eS7 (Supplementary Figs.4, 5).

Thus, the extreme genome reduction in Encephalotozoon/Ordospora species is reflected in the structure of their ribosomes: E. cuniculi ribosomes have experienced the most drastic loss of protein content among eukaryotic cytoplasmic ribosomes to be structurally characterized, and they are devoid of even those rRNA and protein segments that are widely conserved not only in eukaryotes but across the three domains of life. The structure of E. cuniculi ribosomes provided the first molecular model of these changes and revealed evolutionary events that were overlooked by both comparative genomics and structural studies of molecules from intracellular organisms (Supplementary Fig.7). Below, we describe each of these events, along with their possible evolutionary origin and their potential impact on ribosome function.

We next observed that, aside from large rRNA truncations, E. cuniculi ribosomes possess rRNA variations in one of their active sites. While the peptidyl-transferase center of E. cuniculi ribosomes has the same structure as in other eukaryotic ribosomes (Fig.1d), the decoding center differs due to the sequence variation in the nucleotide 1491 (E. coli numbering, Fig.1e, f). This observation is important because the decoding site of eukaryotic ribosomes typically contains residues G1408 and A1491, compared to bacteria-type residues A1408 and G1491. And this variation underlies different sensitivity of bacterial and eukaryotic ribosomes to the aminoglycoside family of ribosome-targeting antibiotics and other small molecules targeting the decoding site33,34,35. In the decoding site of E. cuniculi ribosomes, the A1491 residue is replaced with U1491, potentially creating a unique binding interface for small molecules targeting this active center. The same A14901 variation is present in other microsporidians, such as P. locustae and V. necatrix, suggesting its wide occurrence in microsporidian species (Fig.1f).

Because our samples of E. cuniculi ribosomes were isolated from metabolically inactive spores, we tested the cryo-EM maps of E. cuniculi for the presence of previously described hibernation factors that bind ribosomes under stress or starvation conditions31,32,36,37,38. We docked previously determined structures of hibernating ribosomes in the cryo-EM maps of E. cuniculi ribosomes. For this docking, we used Saccharomyces cerevisiae ribosomes in complex with the hibernation factor Stm138, P. locustae ribosomes in complex with the factor Lso232, and V. necatrix ribosomes in complex with factors Mdf1 and Mdf231. In so doing, we found the cryo-EM density corresponding to the hibernation factor Mdf1. Similar to Mdf1 binding to V. necatrix ribosomes, Mdf1 also binds E. cuniculi ribosomes, where it blocks the ribosomal E site, possibly helping inactivate ribosomes when parasites sporulate and become metabolically inactive (Fig.2).

Mdf1 blocks the ribosomal E site, which appears to help inactivate ribosomes when parasites sporulate and become metabolically inactive. In the structure of E. cuniculi ribosomes, we found that Mdf1 forms a previously unknown contact with the ribosomal L1-stalk (the part of the ribosome that helps release deacylated tRNAs from the ribosome during protein synthesis). These contacts suggest that Mdf1 dissociates from the ribosome using the same mechanism as deacetylated tRNAs, providing a possible explanation of how ribosomes can remove Mdf1 to reactivate protein synthesis.

Our structure, however, revealed a previously unknown contact between Mdf1 and the ribosomal L1-stalk (the part of the ribosome that helps release deacylated tRNAs from the ribosome during protein synthesis). Specifically, Mdf1 exploits the same contacts as the elbow-segment of deacylated tRNA molecules (Fig.2). This previously unknown molecular mimicry suggests that Mdf1 dissociates from the ribosome using the same mechanism as deacetylated tRNAs, explaining how ribosomes can remove this hibernation factor to reactivate protein synthesis.

While building the rRNA model, we found that E. cuniculi ribosomes possess anomalously folded rRNA segments, which we termed molten rRNA (Fig.3). In ribosomes across the three domains of life, rRNA folds into structures in which most rRNA bases are either base-paired and stacked with each other or interact with ribosomal proteins38,39,40. However, in E. cuniculi ribosomes, the rRNA appears to defy this folding principle by transforming some of their helices into unfolded rRNA stretches.

Structure of the helix H18 of the 25S rRNA in S. cerevisiae, V. necatrix, and E. cuniculi. Typically, in ribosomes across the three domains of life, this linker is folded into an RNA helix, which comprises between 24 and 34 residues. By contrast, in microsporidia this rRNA linker is being progressively reduced to two, single-stranded uridine-rich linkers that comprise just 12 residues. Most of these residues are exposed to the solvent. This figure illustrates that microsporidian parasites appear to defy a common principle of rRNA folding, in which rRNA bases are typically paired with other bases or involved in rRNAprotein interactions. In microsporidia, some rRNA segments adopt unfavorable folding, in which former rRNA helices are turned into single-stranded segments that are stretched out almost into a straight line. Having these unusual stretches allows microsporidian rRNA to connect distant segments of rRNA using the minimal number of RNA bases.

The most striking example of this evolutionary transformation can be observed in the helix H18 of the 25S rRNA (Fig.3). In species ranging from E. coli to humans, the base of this rRNA helix contains 24-32 nucleotides that form a slightly irregular helical structure. In the previously determined structures of ribosomes from V. necatrix and P. locustae31,32 the base of helix H18 is partially unwound yet the base-pairing of nucleotides is preserved. In E. cuniculi, however, this rRNA segment is turned into the minimal-length linkers 228UUUGU232 and 301UUUUUUU307. Unlike typical rRNA segments, these uridine-rich linkers are neither folded into a helix nor they are involved in extensive contacts with ribosomal proteins. Instead, they adopt a solvent-exposed and fully unfolded structure in which rRNA strands are stretched into an almost straight line. This stretched conformation explains how E. cuniculi can use just 12 RNA bases to fill the 33 -long gap between rRNA helices H16 and H18while other species require at least twice as many rRNA bases to fill this gap.

Thus, we could show that at the expense of energetically unfavorable folding, microsporidian parasites have invented a strategy to reduce even those rRNA segments that remain widely conserved across species from the three domains of life. Apparently, by accumulating mutations that transform rRNA helices into short poly-U linkers, E. cuniculi could evolve unusual rRNA segments that comprise the minimum possible number of nucleotides that is required to connect distant segments of rRNA. This helps explain how microsporidia have accomplished the phenomenal reduction of their essential molecular structure without losing its structural and functional integrity.

Another anomalous feature of E. cuniculi rRNA is the emerging of bulgeless rRNA (Fig.4). Bulges are non-base-paired nucleotides that flip out from RNA helices rather than being buried inside a helix41. Most rRNA bulges serve as a molecular glue by helping to bind adjacent ribosomal proteins or other rRNA segments. Some bulges serve as a hinge that allows rRNA helices to bend and adopt an optimal folding for productive protein synthesis41.

a rRNA bulges (S. cerevisiae numbering) that are missing in the structure of E. cuniculi ribosomes but present in most other eukaryotes b Comparison of the ribosome interior of E. coli, S. cerevisiae, H. sapiens and E. cuniculi illustrates that microsporidian parasites lack many ancient, highly conserved rRNA bulges. These bulges stabilize ribosome structure, therefore their absence in microsporidia suggests decreased stability of rRNA folding in microsporidian parasites. c Comparison with the P-stalk (L7/L12-stalk in bacteria) illustrates that the loss of rRNA bulges can occasionally co-occur with the emergence of new bulges in the vicinity of the lost ones. The helix H42 in 23S/28S rRNA possesses an ancient bulge (U1206 in S. cerevisiae), which is estimated to be at least 3.5 billion years old due to its conservation across the three domains of life. In microsporidia, this bulge has been eliminated; however, a new bulge (A1306 in E. cuniculi) has evolved in the vicinity of the lost one.

Strikingly, we observed that E. cuniculi ribosomes lack most of the rRNA bulges found in other species, including more than 30 bulges that are conserved in other eukaryotes (Fig.4a). This loss eliminates many contacts between ribosomal subunits and adjacent rRNA helices, occasionally creating large hollow voids within the ribosome interior, making E. cuniculi ribosomes more porous compared to the more conventional ribosomes (Fig.4b). Notably, we found that most of these bulges are also lost in the previously determined structures of V. necatrix and P. locustae ribosomes, which was overlooked by previous structural analyses31,32.

Occasionally, the loss of rRNA bulges is accompanied by the evolution of new bulges near the lost ones. For example, the ribosomal P-stalk contains a bulge U1208 (in S. cerevisiae) that is conserved from E. coli to humans and is therefore estimated to be 3.5 billion years old. During protein synthesis, this bulge helps the P-stalk to move between open and closed conformations so that the ribosome can recruit translation factors and deliver them to the active site42. In E. cuniculi ribosomes this bulge is missing; however, a new bulge (G883) is located just three base pairs away, possibly helping restore the optimal flexibility of the P-stalk (Fig.4c).

Our finding of bulgeless rRNA shows that rRNA minimization is not limited to the loss of rRNA elements on the surface of the ribosome but may affect the very core of the ribosome, creating a parasite-specific molecular defect that has not been observed in free-living species.

Having modelled canonical ribosomal proteins and rRNA, we found three segments of the cryo-EM map not accounted for by the conventional ribosome components. Two of these segments had a size of small molecules (Fig.5, Supplementary Fig.8). The first segment was sandwiched between ribosomal proteins uL15 and eL18 at a location normally occupied by the eL18 C-terminal truncated in E. cuniculi. Although we could not determine the identity of this molecule, the size and shape of this density island would be explained well by the presence of a spermidine molecule. Its binding to the ribosome is stabilized by microsporidia-specific mutations in protein uL15 (Asp51 and Arg56), which appear to increase the ribosome affinity to this small molecule as they allow uL15 to wrap around this small molecule in the ribosome structure (Supplementary Fig.8, Supplementary Data1, 2).

a The cryo-EM map indicates the presence of the extra-ribosomal nucleotide bound to the E. cuniculi ribosome. In the E. cuniculi ribosome this nucleotide occupies the same space as the 25S rRNA nucleotide A3186 (S. cerevisiae numbering) in most other eukaryotic ribosomes. b In the E. cuniculi ribosome structure, this nucleotide is being sandwiched between ribosomal proteins uL9 and eL20, stabilizing contacts between these two proteins. cd Analyses of eL20 sequence conservation in microsporidian species. A phylogenetic tree of microsporidian species (c) and a multiple sequence alignment of protein eL20 (d) illustrate that the nucleotide-binding residues F170 and K172 are conserved in most canonical microsporidia (aside from S. lophii), except for the early-branched microsporidia, in which the rRNA expansion ES39L is preserved. e The plot shows that the nucleotide-binding residues F170 and K172 are only found in eL20 from microsporidian parasites with highly reduced genomes and not in other eukaryotes. Overall, these data indicate that microsporidian ribosomes have evolved a nucleotide-binding site that appears to bind AMP molecules and use them to stabilize proteinprotein interactions in the ribosome structure. The high degree of conservation of this binding site among microsporidia and its absence in other eukaryotes indicates that this site may provide a selective advantage for microsporidia survival. Therefore, the nucleotide-binding pocket in microsporidian ribosomes appears not to be a vestigial feature or the ultimate form of rRNA degeneration, as previously suggested32, but a useful evolutionary innovation that allows microsporidian ribosomes to directly bind small molecules, utilizing them as molecular building blocks for ribosome assembly. This finding makes microsporidian ribosomes the only known ribosomes that use single nucleotides as a structural building block. f A hypothetic evolutionary path of the nucleotide-binding acquisition.

The second small molecule density was located at the interface of ribosomal proteins uL9 and eL30 (Fig.5a). This interface was previously described in the structure of S. cerevisiae ribosomes as a binding site of the 25S rRNA nucleotide A3186 (part of the rRNA expansion segment ES39L)38. In P. locustae ribosomes, where ES39L is degenerated, this interface was shown to bind an unidentified single nucleotide31, and it was hypothesized that this nucleotide represents the ultimate form of rRNA reduction in which the ~130-230 base-long rRNA expansion ES39L was reduced to a single nucleotide32,43. Our cryo-EM maps confirmed the idea that the density can be accounted for by a nucleotide. However, the higher resolution of our structure revealed that this nucleotide is an extra-ribosomal molecule, likely AMP (Fig.5a, b).

We next asked whether the nucleotide-binding site has been evolved or preexisted in E. cuniculi ribosomes. Because the nucleotide-binding is primarily mediated by residues Phe170 and Lys172 in the ribosomal protein eL30, we assessed the conservation of these residues in 4,396 representative eukaryotes. Similar to the aforementioned case of uL15, we found that Phe170 and Lys172 residues are highly conserved only in canonical microsporidia, but are absent in other eukaryotes, including non-canonical microsporidia Mitosporidium and Amphiamblys in which the rRNA ES39L segment is not reduced44,45,46 (Fig.5ce).

Collectively, these data supported the idea that E. cuniculi, and possibly other canonical microsporidians, have evolved the ability to effectively trap abundant small metabolites in their ribosome structures in order to compensate for the rRNA and protein reduction. In doing so, they have evolved the unique ability to bind extra-ribosomal nucleotides, illustrating a previously unknown and ingenious ability of parasitic molecular structures to compensate their degeneration by trapping small abundant metabolites and using them as structural mimics of degenerated RNA and protein segments.

The third unmodeled segment of the cryo-EM map we found within the large ribosomal subunit. The relatively high resolution of our maps (2.6) revealed that this density belongs to a protein with a unique combination of bulky side chains residues, which allowed us to identify this density as a previously unknown ribosomal protein, which we termed msL2 (microsporidia-specific protein L2) (Methods, Fig.6). Our homology search revealed that msL2 is conserved in the microsporidian branch of Encephalitozoon and Ordospora species but is absent in other species, including other microsporidians. In the ribosome structure, msL2 occupies a void formed by the loss of the rRNA expansion ES31L. In this void, msL2 helps stabilize rRNA folding and likely compensates for the ES31L loss (Fig.6).

a Electron density and model of the microsporidia-specific ribosomal protein msL2 found in E. cuniculi ribosomes. b Most eukaryotic ribosomes, including 80S ribosomes of S. cerevisiae, possess the rRNA expansion ES19L, which has been lost in most microsporidian species. The previously determined structure of microsporidian ribosomes from V. necatrix showed that the loss of ES19L in these parasites was compensated by the evolution of a new ribosomal protein, msL1. In this study, we discovered that E. cuniculi ribosomes also evolved an additional RNA-mimicking ribosomal protein as an apparent compensation for the loss of ES19L. However, msL2 (currently annotated as hypothetical protein ECU06_1135) and msL1 have different structure and evolutionary origin. c This finding of the birth of evolutionary unrelated ribosomal proteins msL1 and msL2 illustrates that ribosomes can achieve an unprecedented level of compositional diversity, even within a small group of closely related species, if they accumulate a deleterious mutation in their rRNA. This finding may help shed light on the origin and evolution of mitochondrial ribosomes, which are known for their severely reduced rRNA and exceptional variability of protein composition among species.

We next compared msL2 protein with the previously described protein msL1the only known microsporidia-specific ribosomal protein that was found in V. necatrix ribosomes31. We wanted to test whether msL1 and msL2 are evolutionary related to each other. Our analysis showed that msL1 and msL2 occupy the same cavity in the ribosome structure, but have distinct primary and tertiary structure, suggesting their independent evolutionary origin (Fig.6). Thus, our finding of msL2 provided evidence that close groups of eukaryotic species can independently evolve structurally distinct ribosomal proteins to compensate for the loss of rRNA segments. This finding is remarkable because most cytoplasmic eukaryotic ribosomes have invariant protein content, comprising the same set of 81 families of ribosomal proteins47. The birth of msL1 and msL2 in distinct microsporidian branches in response to the loss of the rRNA expansion segments suggests that degeneration of parasitic molecular structures forces parasites to seek compensatory mutations that may eventually lead to gain of compositional diversity of these structures in distinct groups of parasites.

Finally, when our model building was complete, we compared the composition of E. cuniculi ribosomes with the composition that was predicted based on genome sequence27. Previously, the E. cuniculi genome was predicted to lack several ribosomal proteins, including eL14, eL38, eL41, and eS30 due to the apparent absence of their homologs in the E. cuniculi genome27,48. The loss of multiple ribosomal proteins was also predicted in most other intracellular parasites and endosymbionts with highly reduced genomes49. For example, while most free-living bacteria contain the same set of 54 families of ribosomal proteins, only 11 of these protein families have detectable homologs in each of the analyzed genomes of host-restricted bacteria49. Supporting this idea, the loss of ribosomal proteins was observed experimentally in microsporidians V. necatrix and P. locustae, which both lack proteins eL38 and eL4131,32.

Our structure revealed, however, that only eL38, eL41, and eS30 are in fact lost in E. cuniculi ribosomes. The protein eL14 was retained, and our structure revealed why this protein could not be detected through homology search (Fig.7). In E. cuniculi ribosomes, most of the eL14-binding site is lost due to degeneration of the rRNA expansion ES39L. In the absence of ES39L, eL14 loses most of its secondary structure, and only 18% of the eL14 sequence is identical between E. cuniculi and S. cerevisiae. This poor sequence conservation is remarkable because even S. cerevisiae and H. sapiensorganisms that are separated by 1.5 billion years of evolutionpossess more than 51% identical residues in eL14. This extraordinary loss of conservation explains why E. cuniculi eL14 is currently annotated as hypothetical protein M970_061160 rather than ribosomal protein eL1427.

a Microsporidian ribosomes have lost the rRNA expansion ES39L, which partially eliminated the binding site for ribosomal protein eL14. In the absence of ES39L, microsporidian protein eL14 underwent a loss of secondary structure, in which former rRNA-binding -helices degenerated to minimal-length loops. b Multiple sequence alignment shows that protein eL14 is highly conserved among eukaryotic species (where it shares 57% sequence identity between yeast and human homologs), but poorly conserved and divergent among microsporidia (where no more than 24% of residues are identical to eL14 homologs from S. cerevisiae or H. sapiens). This poor sequence conservation, along with changes in the secondary structure, explains why homologs of eL14 have never been found in E. cuniculi and why this protein was thought to have been lost in E. cuniculi. Instead, E. cuniculi eL14 was previously annotated as the hypothetical protein M970_061160. This observation reveals that the diversity of microsporidian genomes is currently overestimated: some genes that are currently thought to have been lost in microsporidia are actually retained, although in a highly diverged form; conversely, some genes that are thought to encode microsporidia-specific proteins (e.g., hypothetical protein M970_061160) do in fact encode highly divergent proteins that can be found in other eukaryotes.

This finding illustrates that rRNA degeneration can lead to drastic loss of sequence conservation in adjacent ribosomal proteins, rendering these proteins undetectable for homology search. Hence, we may overestimate the actual extent of molecular degeneration in organisms with small genomes because some proteins that are viewed as lost are in fact preserved, though in a highly altered form.

More:
Adaptation to genome decay in the structure of the smallest eukaryotic ribosome - Nature.com