Keywords

1.1 Causes and Effects of Reductive Evolution

There is a common bias to describe evolution as a march toward decreased entropy and increased complexity. After all, the regular ordering of atoms to create larger and more complex systems is an intrinsic feature of life on earth. While this suggests that a gradual increase in organism complexity is inevitable, complexity is balanced in many environmental niches by selective pressures that favor rapid and efficient reproduction, leading to the elimination of superfluous traits. This process of ablation is known as “reductive evolution” and can result in simpler organisms deriving from more complex ancestors.

Reductive evolution typically results from efficient nutrient usage. Macroscopic examples of this include the loss or atrophy of eyes in a wide variety of cave fish or the somewhat ironic loss of a digestive tract in tapeworms (Castro 1996; Morris et al. 2012). At the microscopic level, reductive evolution usually involves eliminating extraneous metabolic processes. Many freshwater chrysophytes, for example, have switched from autotrophy/mixotrophy to heterotrophy in order to combat limited carbon availability and have lost photosynthetic pathways and large swaths of their genomes along the way (Olefeld et al. 2018; Majda et al. 2021).

Gene loss is the most common example of reductive evolution and is often made feasible by co-occurring organisms. One interesting case of this is the loss of the catalase-peroxidase protein, KatG, in the marine cyanobacteria Prochlorococcus spp. (Scanlan et al. 2009). KatG serves an important role in protecting many cyanobacteria from hydrogen peroxide (Perelman et al. 2003), which builds up in oceans due to photooxidation of dissolved organic carbon (Cooper et al. 1988). In a few hours of direct sunlight, enough hydrogen peroxide can be produced to kill off Prochlorococcus cultures (Morris et al. 2011), indicating that they have a very high sensitivity for the molecule. It is surprising then that Prochlorococcus spp. would have lost katG. Instead, co-occurring cyanobacteria have retained katG and scavenge surrounding hydrogen peroxide from marine environments (Petasne and Zika 1997). The loss of katG is therefore only possible because organisms in the natural community provide a protective function.

Interactions between Prochlorococcus spp. and other marine cyanobacteria inspired the “Black Queen Hypothesis,” which posits that natural selection for genomic streamlining breeds dependencies on co-occurring organisms (Morris et al. 2012). This contrasts with the “Red Queen Hypothesis,” inspired by Lewis Carroll’s Through the Looking-Glass, which postulates that competition breeds coevolution (Van Valen 1973). As a corollary to the Black Queen Hypothesis, the more dependent an organism is on other organisms, the more thoroughly a genome will be streamlined. We might therefore hypothesize that genome reduction scales with metabolic dependence on other organisms, i.e., the average genome size of evolutionarily related phototrophs > heterotrophs, and facultative parasites > obligate parasites, which seems to be the case (de Castro et al. 2009; Merhej et al. 2009; Clark et al. 2010; Majda et al. 2021) (Fig. 1.1).

Fig. 1.1
figure 1

Microsporidia have the smallest known eukaryotic genomes. Logarithmic plot of the number of annotated protein coding genes as a function of the respective organism’s genome size. All entries present in NCBI (https://www.ncbi.nlm.nih.gov/) were included, but the data were broadly filtered to remove untenable outliers, partial sequences, and nucleomorphs. Because of the broad filtering, some partially sequenced or annotated entries are still present. The plot was generated using source code from https://github.com/smsaladi/genome_size_vs_protein_count. Eukaryotes are colored in different shades of red, with microsporidia in black. Prokaryotes and viruses are represented in shades of green

Full size image

Parasites are some of the greatest beneficiaries of reductive evolution, and nowhere is this more conspicuous than in microsporidia. As obligate intracellular parasites, microsporidia have dramatically reduced many elements of their genomes. In the sections below, we describe the various factors facilitating genome reduction and outline elements of the genome that are absent in microsporidia. We then delve more deeply into the effects of genome ablation at the protein and RNA level by comparing aspects of ribosome structure, function, and maturation in microsporidia to other eukaryotes.

1.2 The Price of a Large Genome

The cost of genome replication is threefold and requires payment in time, nutrients, and space. All three costs increase with genome size, although there is some variation between prokaryotes and eukaryotes. In this section, we discuss the impact of each of these factors on genome replication and describe how they contribute to reduction in microsporidian genome sizes.

1.2.1 Time

Time is required both to collect materials for DNA synthesis and to physically duplicate the genome. The amount of time required for genome replication depends largely on the catalytic rate of the DNA polymerase. In E. coli, DNA polymerase III copies around 1000 nucleotides/second (Kelman and O’Donnell 1995; Naufer et al. 2017). On the other hand, the equivalent yeast polymerase, Pol ϵ, has a maximal catalytic rate of only 350 nt/s (Ganai et al. 2015). Yeast replication is further decreased to 50 nt/s by proofreading, lagging strand synthesis, etc. Fortunately, eukaryotes are able to offset slower catalytic rates and considerably larger genomes by segregating genetic material into different chromosomes and amplifying from multiple origins of replication. Consequently, yeast and E. coli grown in ideal conditions have replication times commensurate to their genome sizes: 90–120 min for 12 Mbp in yeast (Salari and Salari 2017), versus 40 min for 4.6 Mbp in E. coli (Fossum et al. 2007). Interestingly, replication rates for many cancerous human cells are on the order of only 20 h (Pereira et al. 2017), despite having genomes 250 times larger than yeast. This shows that various factors contribute to dramatically decrease the necessary time for eukaryotic replication, but that total replication time typically increases with increasing genome sizes. Thus, it is beneficial for intracellular parasites like microsporidia to reduce their genome size in order to decrease doubling times. Unfortunately, very little is currently known about microsporidian polymerases, or even whether their chromosomes harbor multiple origins of replication. One study on the microsporidia Nematocida parisii determined that their population doubles in around 140 min (Balla et al. 2016). Although there are a variety of confounding factors, such as growth occurring in infected nematodes rather than in an optimized broth, the replication rate of the 4.3 Mbp N. parisii genome is considerably slower than in yeast (about 1/4 the rate). This suggests that the catalytic rate of the polymerase is slower and/or that N. parisii has fewer chromosomes and origins of replication per Mbp than yeast.

1.2.2 Nutrients

Nitrogen and phosphorous are key elements in DNA and are considered the limiting nutrients for growth in most ecosystems (Ågren et al. 2012; Elser 2012). The biosynthesis of DNA is thus an extremely resource-intensive investment. In fact, comprehensive estimates for the ATP requirements for DNA replication suggest that it costs as much as 500 high-energy bonds/bp in diploid eukaryotes (Lynch and Marinov 2015). While this estimate includes indirect costs such as the production of nucleosomes to stabilize the DNA, most expenses scale linearly with genome size. The larger the genome, the more NTPs are required, and the less high-energy bonds are available for alternative functions like protein production or cell defense. Many organisms therefore pass through a cell-cycle checkpoint, called START, which acts as a nutrient-sensing step to assess available resources prior to replication (Foster et al. 2010). Cells lacking requisite nutrients enter a quiescent state until conditions are more favorable for DNA biosynthesis.

Nutrient limitations are even more restrictive for obligate intracellular parasites. Indeed, microsporidia are almost completely reliant on their hosts and are metabolically inactive in nutrient-poor, extracellular environments (Weiss and Becnel 2014). The hijacking of host systems allows them to bypass much of the innate cost of DNA replication, and simply importing nucleotides instead of synthesizing their own reduces the ATP requirements per base pair by nearly 50% (Lynch and Marinov 2015). Intriguingly, microsporidia have opted to eliminate the majority of enzymes required for nucleotide biosynthesis (Dean et al. 2016) and have instead expanded families dedicated to nucleotide import (Cuomo et al. 2012). This indicates that microsporidia have increased import proteins but greatly decreased biosynthetic pathways, facilitating a net decrease in overall genome size (Dean et al. 2016). Similar trends are identifiable in microsporidia for many other central eukaryotic pathways, such as glycolysis or fatty acid metabolism (Wiredu Boakye et al. 2017).

1.2.3 Space

Although space is perhaps the least conspicuous cost for DNA, many studies have noted and discussed the intricate relationship between genome size and cell size in eukaryotes (Gregory 2001; Cavalier-Smith 2005). The crux of this argument lies in the relatively invariant karyoplasmic ratio, i.e., the ratio of the nuclear volume to cytoplasm is important for cell function, and is generally conserved (Huxley 1925; Trombetta 1942; Cavalier-Smith 2005). The nuclear size is in turn proportional to the total volume of the chromatin (Cavalier-Smith 2005). Although the underlying causes of this effect are still being determined (Cantwell and Nurse 2019; Blommaert 2020), a decrease in genome size will generally lead to a decreased nuclear size, catalyzing a decrease in cell size. The reverse also holds true, where a decrease in cell size will herald a decrease in genome size. This relationship was cleanly demonstrated in a eukaryotic phytoplankton by Malerba et al. (2020). In this study, 72 different Dunaliella tertiolecta lineages with cell volumes spanning two orders of magnitude were placed under selective pressures favoring smaller cells. After 100 generations, lineages that were initially much larger displayed an up to 11% decrease in genome size, while smaller lineages were unaffected. This suggests that (1) selective pressures favoring smaller cells indirectly select for smaller genomes and (2) lineages with larger genomes contain a set of superfluous genes that can be lost, while smaller lineages are already operating at closer to the minimal genome (Malerba et al. 2020).

For intracellular parasites like microsporidia, the space available within their hosts directly restricts the number of spores produced. Cells infected with microsporidia are often saturated with spores (Weiss and Becnel 2014; Grigsby et al. 2020), suggesting the host cell walls limit the number of spores created per infection. In fact, the spatial costs of DNA are twofold, as not only does DNA indirectly determine the size of the spores or meronts, but it also takes up valuable real estate within the cell. It is therefore extremely beneficial for microsporidia to minimize genome size, and it is unsurprising that they are some of the physically smallest eukaryotes. As a consequence of cell-wall limitations to genome size, microsporidian species that exit via exocytosis may have less stringent spatial costs than lytic species. Mature spores are constantly being shed in exocytosed species, increasing the effective available space compared to lytic species. Currently, only one pair of species can be used as an example: Nematocida displodere is primarily released via cell lysis, while N. parisii can be exocytosed in vesicles (Luallen et al. 2016). Although the genome of N. displodere is, in fact, smaller than the genome of N. parisii (Luallen et al. 2016), more data are required to determine whether the “spore release method” contributes to genome size variation between related microsporidians.

1.3 Paths of Reductive Evolution in Microsporidia

Microsporidia are characterized by many unique and interesting features, such as a lack of innate mobility (Weiss and Becnel 2014) and a fishing-line like infection apparatus (Han et al. 2017). Despite their innovations, microsporidia are perhaps most frequently referenced for their exquisitely small genomes (Keeling and Slamovits 2004; Corradi et al. 2010; Corradi and Slamovits 2011) and minimized macromolecular complexes (Melnikov et al. 2018a; Barandun et al. 2019; Ehrenbolger et al. 2020). Microsporidian genomes are indeed very small and have the honors of claiming both the smallest known eukaryotic genome (Corradi et al. 2010) and one of the highest known eukaryotic gene densities (Fig. 1.2) (Keeling and Slamovits 2004; Keeling 2007). The genome of Encephalitozoon intestinalis, for example, is only 2.3 Mbp (Corradi et al. 2010). That is only half the size of the E. coli genome (4.6 Mbp) and 1/65,000 the size of Paris japonica (150 Gbp), a flowering perennial with the largest confirmed eukaryotic genome (Pellicer et al. 2010).

Fig. 1.2
figure 2

Microsporidia have one of the most gene-dense eukaryotic genomes. Gene density across different kingdoms was calculated by dividing the number of annotated protein coding genes by the genome size of the respective organism in kilobase pairs. All entries present in NCBI (https://www.ncbi.nlm.nih.gov/) were included, but the data were broadly filtered to remove untenable outliers, partial sequences, and nucleomorphs. Because of the broad filtering, some partially sequenced or annotated entries are still present. Eukaryotes are colored in different shades of red, with microsporidia in black. Prokaryotes and viruses are represented in shades of green

Full size image

Early studies on microsporidia noted the absence or modification of several cellular structures characteristic of eukaryotes. For example, microsporidia lack peroxisomes, have unstacked Golgi bodies, and have highly reduced mitochondria called mitosomes (Corradi and Keeling 2009; Vávra and Ronny Larsson 2014). These observations led to speculation that microsporidia represent an ancient and unsophisticated eukaryotic lineage. They were therefore classified as Archezoa, with the prevailing hypothesis stating that they diverged prior to endosymbiosis of the mitochondrial ancestor (Cavalier-Smith 1983). This theory was disproven when further genetic analyses demonstrated that a subset of genes found in eukaryotic mitochondria have been transferred to microsporidian chromosomes (Germot et al. 1996; Katinka et al. 2001), indicating that microsporidia diverged after endosymbiosis and are therefore simplified organisms derived from more complex ancestors. Likewise, the small genomes of microsporidia are not a representation of a primitive ancestral state but are instead the result of minimization of multifarious genomic features. In this section, we describe several features affecting genome size, such as gene loss, intron minimization/removal, reductions in gene length, deletions of redundant genes, and the shortening of intergenic regions (IGRs) (Fig. 1.3a).

Fig. 1.3
figure 3

Mechanisms of genome compaction in microsporidia. (a) Schematic representation of a relatively expanded (top) and compacted genome (bottom). Different genomic elements are colored, and the processes leading to their compaction (lower panel) are labeled on top. (b) The size of intergenic sequence (IGS) regions correlates with the directionality of adjacent genes, likely due to the presence of transcriptional control elements upstream of the transcriptional start site (e.g., enhancers, promotors). Gene directionality is indicated with arrows and 5′ or 3′ labels. Transcriptional control elements and their binding partners (e.g., transcription factors, RNA polymerases etc.) are shown as symbolic cartoons and colored in shades of green

Full size image

1.3.1 Non-coding Regions

Microsporidian genomes are similar to those of other eukaryotes in structure and organization. Multiple linear chromosomes can be segregated into telomeres, subtelomeres containing ribosomal DNA (rDNA) and repetitive elements, and gene-rich cores (Dia et al. 2016). Variation is more localized to individual elements of the genome, like coding sequences and intergenic regions. The regions between genes are essential for efficient transcription and contain binding sites for various promoters and enhancers, which are often thousands of nucleotides away from the gene they enhance. It is intriguing then that many microsporidia have tiny IGRs, with E. intestinalis averaging only 115 bp between genes (Corradi et al. 2010). The genes themselves are an average of 1.04 kbp (Corradi et al. 2010). By taking into account the gene density (1.16 kbp/gene) (Corradi et al. 2010), we can determine that coding regions account for as much as 90% of the E. intestinalis genome. To put this in perspective, around 70% of the yeast genome codes for proteins (Dujon 1996) and only 2% of the human genome is protein coding (Piovesan et al. 2019). The low ratio of non-coding to coding sequences suggests that microsporidia have extremely streamlined IGRs. In fact, contrary to most eukaryotes, non-coding regions in microsporidia have higher sequence conservation than coding regions (Corradi et al. 2010; Corradi and Slamovits 2011; Whelan et al. 2019), indicating that the remaining bases form important molecular recognition motifs.

Most regulatory elements are found upstream of the 5′ end of a gene. Tellingly, the length of microsporidian IGRs appears to correlate with the directionality of adjacent genes (Fig. 1.3b) (Keeling and Slamovits 2004). For Encephalitozoon cuniculi, regions wedged between the termini of two genes (the 3′ ends) are about 20% shorter than regions between parallel genes (one 3′, and one 5′ end), while regions abutting divergent 5′ ends are a further 20% longer on average. This pattern is indicative of severe reductive selection operating on IGRs (Keeling and Slamovits 2004), as zero, one, or two sets of upstream transcription factors need to bind between convergent, parallel, and divergent genes, respectively.

Several other factors suggest that Encephalitozoon spp. are operating at the limit of IGR reduction. Firstly, the length of IGRs sometimes dips into negative values, i.e., genes overlap one another (Katinka et al. 2001; Akiyoshi et al. 2009; Corradi et al. 2010). Secondly, multiple studies have noted that transcripts initiate in upstream genes and read through into downstream genes (Williams et al. 2005; Corradi et al. 2008; Gill et al. 2010), suggesting that transcriptional start sites and termination sequences are often located within adjacent genes. Finally, microsporidia produce many multigene transcripts, which surprisingly encode both sense and antisense genes (Peyretaillade et al. 2009; Corradi and Slamovits 2011; Watson et al. 2015). These transcripts, known as “noncontiguous operons,” are thought to regulate protein expression levels and result from evolutionary pressure to minimize genome size (Sáenz-Lahoya et al. 2019). These three examples provide evidence that microsporidia trim and eliminate IGRs wherever possible and have adapted more spatially efficient mechanisms to regulate protein expression levels.

Microsporidian parsimony is not only directed toward IGRs but also impacts other non-coding regions like introns. Splicing machinery and introns appear to have been convergently eliminated in at least three microsporidian genera: Edhazardia, Nematocida, and Enterocytozoon (Keeling et al. 2010; Desjardins et al. 2015). Even when introns are retained, they are reduced in both number and length (Lee et al. 2010; Campbell et al. 2013). In E. cuniculi, for example, a total of 36 introns have been identified, ranging from only 23 to 76 bases in length (Lee et al. 2010). The splicing efficiency for many of these introns is very low, often around 10–25%, and many putative introns display no active splicing (Grisdale et al. 2013; Campbell et al. 2013; Desjardins et al. 2015). For comparison, the yeast genome contains at least 300 introns ranging from around 100 to 1000 bases (S**ola et al. 1999; ** the “eukaryotic” or “e” proteins of the ribosome. It is somewhat surprising then that the drastic microsporidian reduction in ESs is not accompanied by a concomitant loss in the number of ribosomal proteins (Fig. 1.5).

To better understand the proteinaceous changes to microsporidian ribosomes, we have collected sequences from genomes available on MicrosporidiaDB (Aurrecoechea et al. 2011) and compared their conservation in Fig. 1.8. Caution is advised while drawing conclusion from these data, as many microsporidian genomes are derived from incomplete assemblies and microsporidian proteins are rapidly evolving. It is therefore very likely that some of the proteins marked as absent were simply not identified via our methods. That said, microsporidia have retained most of the ribosomal proteins found in yeast, and only a few of the 80 yeast proteins are potentially absent in many microsporidian species. Remaining proteins have a 38% average sequence identity to yeast homologs and are often considerably shorter (Fig. 1.8a and b). Some proteins have lost loops or linkers, while others have been truncated at the N- or C-terminus. Additionally, low levels of sequence identity can be used to demarcate proteins that have structurally diverged from yeast (Fig. 1.8c).

Fig. 1.8
figure 8

Microsporidian ribosomal protein phylogeny, identity, and structure relative to their yeast homologs. (a) Ribosomal protein phylogeny generated by using protein sequences conserved in all listed microsporidian species. Connected to the microsporidian phylogeny (black) is a simplified tree for other non-microsporidian species, based on (James et al. 2006, 2013; Haag et al. 2014). For the microsporidian phylogenetic tree, the protein sequences were obtained by performing translated nucleotide blast (tblastn) searches with an E-value cutoff of 0.05, using the S. cerevisiae sequences or verified microsporidian hits as query and MicrosporidiaDB (Aurrecoechea et al. 2011) as database. For P. locustae and V. necatrix, protein sequences were obtained from (Barandun et al. 2019; Ehrenbolger et al. 2020) or local genome databases. For the non-microsporidian species, sequences were obtained from https://www.ncbi.nlm.nih.gov/. Proteins were aligned using MUSCLE 3.8.31 (Edgar 2004) and trimmed using trimAl (Capella-Gutiérrez et al. 2009) with the –gappyout option. The trimmed alignments were then concatenated using FASconcat 1.11 (Kück and Meusemann 2010). The phylogenetic tree was constructed with RAxML 8.2.12 using the model PROTGAMMAILGF, determined with ProtTest 3.4.2, and 1000 bootstrap replicates. The sequence identity heatmap was constructed using MUSCLE 3.8.31 (Edgar 2004) and Clustal-Omega (Sievers et al. 2011). The S. cerevisiae sequences were set as reference, except for eL28 and msL1, where H. sapiens and V. necatrix were used. The different shades of blue describe the percentage identity of the protein sequence compared to the reference. The row for S. cerevisiae contains viability data, color coded for lethal (dark yellow), slow-growing (yellow), and normal-growing (cream) ribosomal gene knockouts (Giaever et al. 2002; Gao et al. 2015). A black dot is used to mark genes that are duplicated in the yeast genome. Only single gene knockouts were performed in the referenced study. The sequence of eS31* was modified by removing the ubiquitin moiety to create the mature protein. (b) Difference in length between the V. necatrix or P. locustae and S. cerevisiae ribosomal proteins. (c) Comparison of the region around ES4 between the S. cerevisiae (left, PDB 4 V88), V. necatrix (middle, PDB 6RM3), and P. locusate ribosome (PDB 6ZU5). Selected ribosomal proteins are colored and labeled with name and N- and C-termini in shades of red. The lost eL38 and the gained msL1 are shown in shades of green. (d) The same view is shown as in (c) with selected proteins colored solid and the ribosome structure transparent

Full size image

Genome-wide knockout screens have been performed in yeast, which allows us to identify essential ribosomal proteins (Giaever et al. 2002; Gao et al. 2015) (Fig. 1.8a). These studies further noted knockouts that led to slow-growth defects. It is important to mention, however, that yeast have duplicated the majority of ribosomal genes. Some deleterious effects may have therefore been ameliorated by the presence of paralogs during single-gene deletion studies. Nevertheless, comparisons between gene conservation and essentiality reveal several interesting results. Firstly, as might be expected, many of the essential genes in yeast were not duplicated. Secondly, essential genes are still extant in almost all microsporidia. Instances of their loss, such as uL16 in Enterocytozoon hepatopenaei, are more likely a result of incomplete genome assemblies or low sequence conservation. This is evinced by the isolation of purported losses. Only in the case of uL23 are essential genes unidentifiable in a related cluster of microsporidia (Trachipleistophora hominis and Pseudoloma neurophilia). Numerous studies have demonstrated the essentiality of uL23 for the formation of the polypeptide exit tunnel (Kaur and Stuart 2011; Polymenis 2020). We therefore find it more probable that its absence is a matter of incomplete genome assemblies; however, a genuine absence would undoubtedly provide useful insights into evolutionary strategies developed by microsporidia to minimize the ribosome exit tunnel. Thirdly, all of the yeast proteins unidentifiable in most microsporidia are nonessential (see eL28, eL38, eL41, P1, and P2), as are some of the frequently missing proteins (eS12, eS25, and eL29). The nonessential eL38 is present in all earlier branching eukaryotes and is absent in all but two microsporidian species (M. daphinae, Amphiamblys sp.), suggesting a relatively recent loss of this ribosomal protein (Barandun et al. 2019). These findings demonstrate that microsporidia have typically retained essential proteins and eliminated nonessential ones.

The nonessential protein eL41 is the only yeast subunit absent in all sequenced microsporidia (Fig. 1.8a) (Barandun et al. 2019; Ehrenbolger et al. 2020). It is remarkably short in other eukaryotes, only ~25 amino acids, and forms a small bridge between the LSU and the SSU (Tamm et al. 2019). Deletions of eL41 are easily tolerated, with knockout yeast strains displaying growth rates similar to wild-type strains (Giaever et al. 2002). More in-depth analyses have revealed that eL41 plays a role in translational efficiency (Dresios et al. 2003; Meskauskas et al. 2003). Ribosomes lacking eL41 had both lower translational fidelity and slower rates of peptidyltransferase activity. This suggests that the removal of eL41 in microsporidia may be another factor contributing to their markedly high rate of missense mutations (Melnikov et al. 2018b). The deletion of eL41 may also result in a slower translation rate, although no information is currently available on the kinetics of microsporidian ribosomes.

The ribosomal stalk proteins, which also have a purported role in translational efficiency (Wawiórka et al. 2017), are reduced in most microsporidia. A typical eukaryotic ribosomal stalk is composed of uL10, two subunits of P1, and two subunits of P2. All five protomers contain a highly conserved, C-terminal SDDDMGFGLFD motif, preceded by a long and flexible linker (Choi et al. 2015). This organization and motif is found in organisms as diverged as humans and the archaeon Pyrococcus horikoshii (Ito et al. 2014). During active translation, the C-termini of the pentamer bind to and recruit the essential elongation factor EF1α, which delivers charged aminoacyl-tRNA to the ribosome. It is proposed that the five redundant motifs aid in the rapid and efficient recruitment of the correct aminoacyl-tRNA, by greatly increasing the local concentrations of EF1α (Wawiórka et al. 2017). Additionally, this kinetic model of decoding suggests that ribosomal pausing leads to the acceptance of near-cognate anticodons, resulting in missense mutations. It is therefore interesting that the majority of microsporidia do not to encode P1, and some may have lost P2 (Fig. 1.8), implying a single EF1α-binding motif is present. Previous work has demonstrated that P1 and P2 are nonessential in eukaryotes only because uL10 retains an EF1α binding domain (Santos and Ballesta 1995; Remacha et al. 1995). On the other hand, the prokaryotic equivalents to P1/P2 are required for translation (Huang et al. 2010), as prokaryotic L10 lacks the binding motif. Remarkably, the uL10 homologs for microsporidian clades have lost the linker and the SDDDMGFGLFD motif (data not shown). Some microsporidia therefore have no identified proteins that can recruit EF1α to ribosomes. This finding may indicate that the translation rate and fidelity are much lower in microsporidia. Alternatively, microsporidia might have developed novel proteins or binding motifs to recruit EF1α. This possibility is of particular interest, as the C-terminal motif utilized by eukaryotes and archaea is a common target for potent toxins like ricin (Choi et al. 2015; Fan et al. 2016). A unique motif would represent an attractive target for therapeutics or pesticides.

1.5.3 Retained and Gained Ribosomal Proteins

Most microsporidian proteins have relatively low sequence identity to yeast proteins (Fig. 1.8). This is not entirely unexpected, as even proteins from two closely related Nematocida species share only ~70% of their amino acid sequence (Balla and Troemel 2013). A noticeable outlier in this divergence is eS31, an essential protein located in the beak of the SSU. Interestingly, eS31 is always produced as a fusion with a ubiquitin moiety. The ubiquitin acts as a chaperone protein to assist in the production and folding of eS31 and is cleaved off before eS31 is incorporated into ribosomes (Martín-Villanueva et al. 2019). The high sequence identity for eS31 derives from this ubiquitin moiety, as a realignment without ubiquitin results in much lower values (see eS31 vs eS31* in Fig. 1.8). Another highly conserved protein is eL15, which is present in all sequenced microsporidia. Little is known about eL15’s function other than that it is essential; however, it is a structural protein that is mostly buried and is therefore likely to have many conserved intermolecular interactions. Additionally, eL15 seems to mediate concentrations of other core ribosomal proteins, and its dysregulation leads to various cancers and diseases (Wlodarski et al. 2018; Ebright et al. 2020). Despite the lack of focused studies, the high conservation of eL15 in microsporidia evinces a high level of functional significance, which is not amenable to mutations in sequence or structure.

In addition to retaining most ribosomal proteins, microsporidia have also gained at least one novel subunit. The microsporidia-specific ribosomal protein (msL1) binds to V. necatrix ribosomes in a gap left by the loss of four ESs (Fig. 1.5) (Barandun et al. 2019). Although the specific role of this protein is unknown, it may be required to stabilize the ribosome in the absence of ESs. Genomic erosion in organelles, such as mitochondria, has resulted in a similarly minimized rRNA. In response, many mitochondria have acquired unique proteins used to patch unstable ribosomes (Petrov et al. 2019). It is likely that msL1 serves a similar patching function in microsporidia where rRNA reduction led to structural instability.

1.5.4 Conserving Energy by Utilizing Ribosome Hibernation Factors

Translational costs are high, and an estimated 30 ATPs are required for the biosynthesis and attachment of each amino acid (Wagner 2007). Such costs are unsustainable in nutrient-poor conditions. Organisms therefore express proteins known as hibernation factors, which bind to and inhibit ribosomes when nutrients are scarce (Prossliner et al. 2018). These factors allow cells to sequester intact ribosomes instead of degrading them (Brown et al. 2018; Trösch and Willmund 2019). The ability to inactivate ribosomes and recover them post-quiescence is of vital importance to microsporidia, as they spend a significant portion of their lifecycle as metabolically inactive spores (Weiss and Becnel 2014).

Microsporidia encode multiple hibernation factors, including the late-annotated short open reading frame 2 (Lso2), and microsporidian dormancy factors (MDF) 1 and 2 (Barandun et al. 2019; Ehrenbolger et al. 2020). All three of these proteins block active sites of the ribosome (Fig. 1.6) and are incompatible with active translation. In yeast, Lso2 is important for recovery of ribosomes post-starvation (Wang et al. 2018), and roughly 10% of ribosomes isolated from starved yeast are bound by Lso2 (Wells et al. 2020). Microsporidian ribosomes isolated from spores, on the other hand, displayed an approximately 92% occupancy rate, indicating that the vast majority of ribosomes in spores are in an inactivated state (Ehrenbolger et al. 2020). MDF1 and MDF2 have not been biochemically characterized; however, their high occupancy in spores and mechanisms of binding indicate that they are likely hibernation factors (Barandun et al. 2019). While MDF1 is broadly conserved in eukaryotes, MDF2 may be species-specific. Orthologs have thus far only been identified in V. necatrix, Nosema ceranae, and Nosema apis. The high occupancy of these factors bound to spore-stage ribosomes, and the fact that microsporidia have potentially evolved species-specific hibernation factors, demonstrates that sequestration of ribosomes during the spore stage is crucial. Although hibernation factors are not specifically associated with reductive evolution, they provide an additional example of the mechanisms by which microsporidia conserve energy.

1.6 Microsporidian Ribosome Assembly

In eukaryotes, ribosome biogenesis is a multidimensional process requiring the action of all three RNA polymerases (Pol) and a complex repertoire of over 300 assembly factors and snoRNAs (Woolford and Baserga 2013; Ebersberger et al. 2014; Klinge and Woolford 2019). The pathway starts in the nucleolus, a subcompartment of the nucleus, where the transcription of a precursor ribosomal RNA (pre-rRNA) initiates a co-transcriptional maturation pathway. In yeast, the precursor contains the rRNAs of both the small subunit (18S) and the large subunit (5.8S, 25S). These rRNAs are flanked by four transcribed spacer regions, two external and two internal (ETS, ITS; Fig. 1.9a). The third rRNA of the large subunit (5S) is transcribed from a different locus and is not part of this long precursor RNA. Assembly factors associate in a co-transcriptional manner with the rRNA precursor, including the transcribed spacers, to assist in the folding and enzymatic processing of the pre-rRNA and to incorporate ribosomal proteins. Several co-transcriptional endonucleolytic cleavage events are required to process the spacers and release the partially matured pre-ribosomal particles. Maturation then continues in the nucleus, where the pre-mature rRNA ends (e.g., 5′ ETS or ITS2) are further processed and degraded. After a controlled export through the nuclear pore complex, the last ribosomal maturation steps and quality control events occur in the cytoplasm.

Fig. 1.9
figure 9

Compaction of the microsporidian rDNA locus to a prokaryotic-like organization. Schematic representation of a single rDNA locus, below a diagram indicating the genomic distribution of all rDNA loci, from (a) S. cerevisiae, (b) microsporidia (nucleotide sizes from V. necatrix), and (c) E. coli. The genes and known spacer sizes are indicated and drawn to scale for comparative purposes

Full size image

The transcribed spacers are not present in the mature ribosome, but are essential elements required to recruit ribosome assembly factors. The level of spacer processing is also used to demarcate the maturation stage of this complex particle (Klinge and Woolford 2019). In addition, eukaryotic ribosomal expansion segments, which are part of the mature ribosome, are also involved in recruiting specific assembly factors. Genome compaction in microsporidia has not only removed rRNA elements, such as eukaryotic ESs, but also drastically affected the transcribed spacers of the ribosomal precursor (e.g., removal of ITS2; Fig. 1.9b). While the pre-ribosomal and ribosomal RNA have been minimized, the number of ribosomal proteins associated with the mature microsporidian ribosome has been less affected (see Sect. 1.5.2) (Barandun et al. 2019; Ehrenbolger et al. 2020). This raises the question of whether ribosome assembly factors and the maturation pathway have been similarly reduced overall, or if specific assembly factor categories have been more impacted by genome reduction than others. Have microsporidia lost ribosome assembly factors with a role in maturing eukaryotic-specific RNA or protein elements? The following section discusses the impact of reductive evolution on the organization of rDNA loci and the maturation of pre-rRNA in microsporidia.

1.6.1 Impact of Genome Compaction on Number and Localization of the rDNA Loci

In most organisms, the ribosomal RNAs are transcribed from one or more polycistronic ribosomal DNA loci. The number of rDNA loci increases considerably from prokaryotes to eukaryotes: from a single rDNA locus in slow-growing bacteria (e.g., Mycobacterium tuberculosis) to ~150–200 copies in yeast (Petes 1979) to more than 10,000 in some plants (Kobayashi 2014). In S. cerevisiae, the primary model organism to study eukaryotic ribosome assembly, all rDNA loci are clustered head to tail on a single chromosome (Petes 1979) (Fig. 1.9a). Within eukaryotes, the size of one pre-rRNA coding locus varies substantially. These size variations are mostly due to differences in the lengths of external and internal spacer elements or eukaryotic-specific ribosomal expansion segments. Eukaryotic rDNA sizes range from the minimal microsporidian version with approximately 4.5 kbp (calculated from the V. necatrix sequences), which has lost many regulatory spacers and ESs, to ~9.1 kbp in yeast or ~ 43 kbp in humans, which contain long ETSs and extensive intergenic spacer regions.

In microsporidia, the rDNA organization and localization within the genome differ between species. While other eukaryotes contain large numbers of clustered rDNA repeats, microsporidia are left with fewer and often not clustered rDNA genes. Twenty-two rDNA copies have been reported for E. cuniculi, located on both telomeric ends of its 11 chromosomes (Brugère et al. 2000; Katinka et al. 2001; Dia et al. 2016). Forty-six partial and polymorph rDNA loci have been found in N. ceraneae (Cornman et al. 2009), and similar to the rDNA loci in N. bombycis, they appear to be distributed over all chromosomes (Liu et al. 2008). While the individual loci are scattered throughout different chromosomes in many microsporidian species, in N. apis, the rDNA genes cluster as repeats head to tail (Gatehouse and Malone 1998), which is more similar to the classical arrangement observed in other eukaryotic organisms.

In most eukaryotes, the 5S encoding gene is dispersed throughout the genome and is not adjacent to the other three rRNAs. One exception to this observation is S. cerevisiae, where the 5S rRNA gene clusters in the intergenic spaces between rDNA repeats (Fig. 1.9a). Both arrangements have been observed in microsporidia. Similar to yeast, in N. bombycis, the 5S gene is located next to the rDNA locus (Huang et al. 2004). Other species, such as E. cuniculi and E. intestinalis, have dispersed the 5S throughout the genome. In these two microsporidia, three copies for the 5S have been detected (Katinka et al. 2001; Corradi et al. 2010), in contrast to the 22 rDNA loci. While the rDNA locus is transcribed by RNA Pol I, the 5S rRNA is transcribed by RNA Pol III (Ciganda and Williams 2011). The microsporidian transcription machinery includes elements for RNA pol I, II, and III (Katinka et al. 2001), indicating that the use of separate polymerases for 5S and rDNA transcription may be retained in microsporidia.

The comparatively small number of rDNA repeats in microsporidia may be a result of their diminutive cell size, simple genomes, and low proteomic complexity. Fewer and shorter genes might require a reduced number of ribosomes, which in turn can be synthesized from fewer rDNA repeats. Indeed, a strong positive correlation between genome size and the number of rDNA repeats in eukaryotes has been noted (Prokopowich et al. 2003). Although this correlation exists, in general, only a fraction of all rDNA repeats are transcriptionally active. The actual rRNA synthesis rate is more so determined by the rate of RNA polymerase recruitment. A yeast strain with only 42 rDNA repeats, compared to the original 142 repeats, grows as well as wild type because two times more RNA polymerases are recruited to the rDNA locus (French et al. 2003). In addition to a potentially reduced need for ribosomes and increased RNA polymerase recruitment to a single locus, the simplified microsporidian rDNA gene organization might allow for a more streamlined ribosome maturation. Fewer pre-rRNA processing steps might be required than in other eukaryotes, due to missing pre-rRNA elements such as internal transcribed spacer 2 (ITS2).

1.6.2 Loss and Minimization of Transcribed Spacers

In many microsporidian species, genome compaction and gene fusion led to a reduction in the total number of ribosomal RNAs from four to three, which represents a reversal of the evolutionary trend seen in eukaryotic ribosomes. The eukaryotic 5.8S rRNA sequence and the 5′ end of the prokaryotic large subunit gene are homologous (Jacq 1981). In typical eukaryotes, ITS2 separates the 5.8S from the remainder of the large subunit rRNA gene (Fig. 1.9b). Early branching microsporidia like M. daphnia and Chytridiopsis typographi still contain highly reduced versions of ITS2 and thereby preserve the traditional eukaryote-specific separation of the 5.8S from the LSU gene (Corsaro et al. 2019). In all later-branching microsporidia, ITS2 has been removed (Vossbrinck and Woese 1986). The reductive evolution in these organisms led to a complete loss of ITS2 and fusion of the 5.8S rRNA with the LSU rRNA (23S), which has created a unique eukaryotic rDNA locus (Fig. 1.9b) with prokaryotic features (Fig. 1.9c). The remaining ITS has been reduced to a surprisingly short sequence in some microsporidia. While N. bombycis (Huang et al. 2004) contains an ITS of ~179 nt, other microsporidians, such as V. necatrix or N. apis (Gatehouse and Malone 1998), compacted this element to only ~33/34 nt. The intergenic spacer regions are important signal sequences for co-transcriptional endonucleolytic processing of the pre-rRNA fragment. Together with an apparent reduction of the 5′ and 3′ ETS regions and the removal of ITS2, the shortening of the ITS has significant implications for the ribosome maturation process, which is tightly controlled by ribosome assembly factors binding to these regions.

1.6.3 Impact of rDNA Compaction on Ribosome Biogenesis Factors

In 2014, Ebersberger et al. performed an evolutionary analysis of 255 yeast protein factors involved in ribosome biogenesis and included four microsporidian species in their analyses (Ebersberger et al. 2014). From these initial factors, 244 were proposed to be present in the last common ancestor shared with the microsporidia. Remarkably, only about half of them could be identified in microsporidia, which was highlighted as “the most remarkable gene loss” observed among the eukaryotic supertaxa (Ebersberger et al. 2014). Although extensive lists of factors involved in yeast ribosome biogenesis existed at the time, the precise functions or binding sites of most of these factors were unknown due to a lack of structural and biochemical data. During the decade since, our knowledge of fungal ribosome biogenesis has advanced to a detailed structural and functional description of the individual factors. This is mainly due to the technical progress made in cryo-EM, which provided high-resolution information and enabled the study of previously inaccessible pre-ribosomal particles from the fungi S. cerevisiae or Chaetomium thermophilum. These structures now provide an updated and comprehensive picture of fungal ribosome maturation and depict the intricate interaction network of assembly factors and ribosomal proteins bound to pre-ribosomal rRNA elements (Barandun et al. 2018; Klinge and Woolford 2019). They show how ribosome maturation proceeds in a hierarchical manner through several different conformational states to produce the final mature eukaryotic ribosome (Klinge and Woolford 2019). The emerging structural data on the fungal biogenesis process, together with recent studies on the microsporidian ribosomes (Barandun et al. 2019; Ehrenbolger et al. 2020), allows us to give a few selected examples of why expansion segment and transcribed spacer removal or shortening might have enabled assembly factor loss (or vice versa).

In yeast, the 5′ ETS is 700 nt long and is involved in the co-transcriptional recruitment of up to 27 ribosome biogenesis factors and the formation of an assembly platform for the SSU. In microsporidia, the exact size and structure of the 5′ ETS pre-rRNA fragment are not known. However, several factors that typically bind to this region have not been identified in microsporidia (Fig. 1.10). One of the first and largest multi-subunit complexes bound to the newly synthesized 5′ ETS is UtpA (Fig. 1.10b) (Hunziker et al. 2016). UtpA is a 7-subunit complex in yeast but appears to be absent or drastically reduced in microsporidia. The UtpA binding site on the 5′ ETS is shared with Utp18, a subunit of another early binding biogenesis complex, UtpB. The potential absence of Utp18 and the entire UtpA complex (Fig. 1.10a) suggests microsporidia may contain a shorter 5′ ETS sequence, which recruits a minimal small subunit assembly platform. Alternatively, assembly factors may be too divergent to be identified.

Fig. 1.10
figure 10

A reduced set of ribosome biogenesis factors and selected examples of assembly factor and expansion segment loss in microsporidia. (a) Presence and conservation of ribosome assembly factors in selected eukaryotes and microsporidia. The protein sequences were obtained by performing translated nucleotide blast (tblastn) or protein blast (blastp) searches with an E-value cutoff of 0.05, using the S. cerevisiae sequences and MicrosporidiaDB (Aurrecoechea et al. 2011) as database. For P. locustae and V. necatrix, protein sequences were obtained from local genome databases. For the non-microsporidian species, sequences were obtained from https://www.ncbi.nlm.nih.gov/. For phylogenetic tree calculation, see legend of Fig. 1.8. Many biogenesis factors display significant sequence similarity (e.g., WD40 domain proteins). It was therefore common for the same open reading frame to be identified as homologous to multiple different biogenesis factors. In such cases, we selected hits with the lowest E-value. It should be noted that the figure thus serves as only a guide to general trends for absent and present proteins, since annotations may be inaccurate. Proteins that were not identifiable (NI) are shown in cream. Biogenesis factors are clustered based on known or predicted binding regions within the 5’ ETS, SSU, ITS2, or LSU of the pre-rRNA. Conservation correlates between members of the same complex. Example complexes are labelled below (a) in shades of gray. (b-d) Structures of S. cerevisiae pre-ribosomal particles denoting selected maturation factors that are often absent in microsporidia, as highlighted in (a). Pre-SSU structures from PDB 5WLC (Barandun et al. 2017) (b), PDB 7AJU (Lau et al. 2021) (c), and a pre-LSU-structure from PDB 6C0F (Sanghai et al. 2018) (d) are displayed with expansion segments missing in V. necatrix, colored in shades of orange and yellow (SSU) or shades of blue and green (LSU), and selected biogenesis factors colored as in (a). These examples demonstrate a correlation between ES reduction and the loss of biogenesis factors that typically bind to those ESs

Full size image

During pre-rRNA maturation, several eukaryotic expansion segments of the ribosomal RNA are bound and remodeled by assembly factors. In general, the absence of an assembly factor correlates with removal of its binding site in other organisms (Fig. 1.10). One striking example includes the SSU segments es3 and es6, which are bound and stabilized by the large HEAT repeat protein Utp20 (Fig. 1.10c). Es3 and es6 are the two largest small subunit expansion segments and have been completely lost or strongly reduced in microsporidia (Barandun et al. 2019; Ehrenbolger et al. 2020). Similarly, Utp20 appears to be absent in all microsporidian species. This suggests the primary role of Utp20 in chaperoning the maturation of these two expansion segments is no longer required in microsporidia. Similarly, the loss of h41 correlates with the loss of Utp30, an assembly factor binding to this rRNA element in pre-ribosomal particles (Fig. 1.10b). In the large subunit, ES7 is bound by two assembly factors, Rrp1 and Nsa1. Again, both the ES7 and the two assembly factors seem to be eliminated from microsporidian genomes.

A key step in large subunit maturation in S. cerevisiae involves processing of the ITS2 prior to nuclear export. Absence of ITS2, the spacer separating the two LSU rRNAs, explains the absence of many ribosome assembly factors binding this RNA region, such as Cic1, Rlp7, or the Las1 complex (Woolford and Baserga 2013). ITS1 processing in yeast is catalyzed by the essential ribozyme-protein complex RNAse MRP. While microsporidia still contain a highly reduced version of RNAse MRP (Zhu et al. 2006), ITS1 has been ablated to only 33 nt. It is unclear if this short ITS region can fold into a structure recognized by the minimized RNAse MRP, or if a simpler mechanism is used.

Apart from the mature ribosome structure, genomic data, and bioinformatics, very little is known about ribosome assembly in microsporidia. By studying the process in these minimal organisms, we can learn more about the still relatively unknown role of expansion segments during the assembly process in other eukaryotic organisms. The compaction of rRNAs together with the removal of transcribed spacer regions appears to have significantly affected the assembly process in microsporidia. A more thorough analysis of how expansion segment removal correlates with assembly factor loss will be required to understand the process in microsporidia and relate loss and compaction to a potential functional role in other eukaryotes.

1.7 Conclusion and Future Perspectives

Genome reduction and size appear to correlate with the degree of metabolic dependence on other organisms. Consequently, an obligate intracellular lifestyle provides a plausible explanation for the loss of redundant metabolic pathways and the invention of novel and more energetically efficient mechanisms of host exploitation. The drastic impact of genome compaction in microsporidia, however, has not only reduced the complexity of metabolic pathways but also affected intergenic regions, minimized gene sizes, and removed regulatory elements and features considered to be essential in eukaryotic organisms.

Genome erosion has significantly altered the microsporidian ribosomal DNA locus. By removing eukaryote-specific elements, such as ITS2 and nearly all expansion segments, the rDNA gene arrangement regressed to a prokaryote-like organization. The recent structural characterization of the microsporidian ribosome has illustrated the impact of genome reduction on the composition and assembly of this essential and ancient particle. It provided the surprising information that despite the loss of their rRNA binding site, almost all eukaryote-specific ribosomal proteins, albeit shortened, are still retained in the structure. Could limited access to primary metabolites precipitate a more compact ribosome? Are nucleotides more “rare” than amino acids, and could this be one reason why the rRNA is much more compacted than ribosomal proteins? Does the extensive rRNA loss affect the fidelity of the ribosome? Further studies are required to delineate the functional implications of ribosome compaction on protein synthesis and to reveal the suitability of ribosome-targeting antibiotics as translation inhibitors in microsporidia.

Microsporidia are of great interest in the fields of infection biology and comparative structural biology. They act as a reservoir for many unique and peculiar structures and have developed the most minimized versions of eukaryotic macromolecular complexes. Additional biochemical and structural studies in microsporidia not only will illuminate their own lifecycle but will also shed light on optional elements in many highly conserved cellular processes.