A new family of cell surface located purine transporters in Microsporidia and related fungal endoparasites

Plasma membrane-located transport proteins are key adaptations for obligate intracellular Microsporidia parasites, because they can use them to steal host metabolites the parasites need to grow and replicate. However, despite their importance, the functions and substrate specificities of most Microsporidia transporters are unknown. Here, we provide functional data for a family of transporters conserved in all microsporidian genomes and also in the genomes of related endoparasites. The universal retention among otherwise highly reduced genomes indicates an important role for these transporters for intracellular parasites. Using Trachipleistophora hominis, a Microsporidia isolated from an HIV/AIDS patient, as our experimental model, we show that the proteins are ATP and GTP transporters located on the surface of parasites during their intracellular growth and replication. Our work identifies a new route for the acquisition of essential energy and nucleotides for a major group of intracellular parasites that infect most animal species including humans.


Introduction
Microsporidia are a highly successful group of strict intracellular eukaryotic parasites that infect a broad range of animal hosts, including humans and economically important species of fish, honeybees, and silkworms (Stentiford et al., 2016;Vávra and Lukesˇ, 2013). Microsporidia can only complete their life cycle inside infected eukaryotic cells. In the external environment they exist only as resistant thick-walled spores. New infections are then initiated by spore germination followed by transit of the parasite through a unique polar tube infection apparatus into a eukaryotic host cell (Vávra and Lukesˇ, 2013). As a result of this lifestyle, Microsporidia have undergone dramatic genomic and cellular streamlining including the complete loss of biosynthetic pathways for de novo synthesis of purine and pyrimidine nucleotides (Dean et al., 2016;Dean et al., 2018;Heinz et al., 2014;Heinz et al., 2012). At the organelle level, reduction is dramatically illustrated by their minimal mitochondria (called mitosomes) which have lost the organelle genome and the capacity to generate ATP (Freibert et al., 2017;Goldberg et al., 2008;Williams et al., 2002). While glycolysis is conserved in some, but not all (Wiredu Boakye et al., 2017) Microsporidia, it appears to be mainly active in spores and is not used during intracellular growth and replication (Dolgikh et al., 2011;Heinz et al., 2012;Williams et al., 2014). The loss of indigenous pathways for making ATP and nucleotides means that Microsporidia are now entirely dependent upon the host cells they infect for the energy, cofactors and nucleic acid building blocks that they need to complete their life cycle (Dean et al., 2016;Nakjang et al., 2013). Hence parasite transport proteins must play critical roles in servicing the energetic and metabolic demands imposed by parasite growth and replication, but there is currently little functional data for Microsporidia transporters (Dean et al., 2016;Dean et al., 2014).
Microsporidia have far fewer (~2000-3500) protein-encoding genes than free-living eukaryotes, providing strong evidence that gene loss, particularly for metabolism, is a pervasive feature of their evolution (Cuomo et al., 2012;Nakjang et al., 2013). Genes for candidate surface-located transporters that have been retained against this background of gene loss, are likely to be important for supporting parasite growth and replication (Heinz et al., 2012;Nakjang et al., 2013). We previously identified 10 gene families encoding candidate surface transport proteins that are conserved in all published Microsporidia genomes (Heinz et al., 2012;Nakjang et al., 2013). The conserved families include the experimentally characterised nucleotide transporters (NTT) that are expressed on the cell surface of intracellular Microsporidia where they can import host purine nucleotides, including ATP, GTP and NAD + (Dean et al., 2016;Dean et al., 2014;Dean et al., 2018;Tsaousis et al., 2008). Intriguingly, the NTT transporters did not transport pyrimidine nucleotides in these experiments (Dean et al., 2018;Heinz et al., 2014), even though genome analyses suggest that Microsporidia can no longer make these substrates for themselves (Dean et al., 2016;Dean et al., 2018;Heinz et al., 2014;Tsaousis et al., 2008). These data suggest that additional transporters must exist to supply Microsporidia with the pyrimidines that they need to grow and replicate.
In the present study, we have characterised a family of Microsporidia Major Facilitator Superfamily (MFS) transport proteins which were discovered in Nematocida spp. (Cuomo et al., 2012) and subsequently shown to be present in all Microsporidia for which genomes are available (Cuomo et al., 2012;Dean et al., 2018;Heinz et al., 2014;Heinz et al., 2012;Nakjang et al., 2013;Watson et al., 2015). Some of the Nematocida proteins share a Pfam domain (PF03825) with the NupG transporter of Escherichia coli (Xie et al., 2004), suggesting (Cuomo et al., 2012) that, like NupG, they might be purine and pyrimidine nucleoside transporters, potentially solving the parasite eLife digest Microsporidia are a group of microscopic parasites that spend part of their lives inside the cells of a broad range of animal hosts, including humans. These parasites are considered to be related to fungi, some of which also live within the cells of other species and are known as fungal endoparasites. One of the shared characteristics of these parasites is that they cannot make nucleotides, molecules that are both the main source of energy of the cell and also the building blocks of DNA. Instead, they take nucleotides, or the materials needed to make nucleotides, from their host cells. Once Microsporidia have depleted a host cell, they turn into spores that can survive outside the host until they invade a new cell, starting the cycle anew.
Microsporidia have proteins on their surface, including nucleotide transporter family proteins (NTT), that enable them to import nucleotides from their host into themselves. Although most fungal endoparasites are also thought to steal nucleotides from their hosts, many do not have NTT proteins, raising the question of how they import the nucleotides. A group of proteins called the Major Facilitator Superfamily (MFS) consists of proteins that were thought to transport the materials cells need to make nucleotides (which are also called nucleotide precursors). Members of this family are found throughout Microsporidia and related fungal endoparasites.
These proteins could explain how fungal endoparasites take nucleotides from their hosts. To test this hypothesis, Major et al. infected mammalian cells with Microsporidia and then checked where two MFS proteins were located during infection. This showed that the proteins were on the surface of the endoparasites, implying that they could be nucleotide precursor transporters. Next, Major et al. genetically modified Escherichia coli bacteria so they would produce MFS proteins, and showed that the proteins could transport two types of nucleotides. Together these results show that MFS proteins could be responsible for nucleotide transport in fungal endoparasites.
In addition to humans, Microsporidia and related fungal endoparasites infect a wide range of animals, including pollinating insects, which have ecological and economic importance. Given that Microsporidia can only survive if they take nucleotides from their hosts, knowing more about the proteins that import the nucleotides could lead to new cures for Microsporidia infections. pyrimidine deficit discussed above. To test this hypothesis and to investigate the evolution and role (s) of these MFS transporters in Microsporidia, we characterised the expression, cellular location and functional characteristics of the homologous proteins from Trachipleistophora hominis (Heinz et al., 2012;Nakjang et al., 2013;Watson et al., 2015), a model species that can be maintained in cell culture and was originally isolated from an HIV/AIDS patient . Our data provide no evidence that the T. hominis proteins can transport the pyrimidine nucleoside uridine, a known substrate for NupG (Xie et al., 2004), or the pyrimidine nucleotides CTP or UTP, but demonstrate that they do transport the purine nucleotides ATP and GTP. These data reveal that T. hominis, and potentially other Microsporidia, have at least two distinct transport systems for importing the ATP and GTP from infected host cells, that they need to complete their intracellular lifecycles.

Results and discussion
Microsporidia encode multiple paralogues of MFS transporters that are distantly related to E. coli NupG We used BlastP to search for sequences related to the Microsporidia conserved protein family # c_456 (Nakjang et al., 2013) which contains the Nematocida spp. putative NupG-like nucleoside/ H+ symporter (Cuomo et al., 2012), in the genomes of Microsporidia and their endoparasitic relatives among the Rozellomycota (Corsaro et al., 2016), including species of Rozella (James et al., 2013), Mitosporidium (Haag et al., 2014) and Amphiamblys, (Mikhailov et al., 2017). Recent phylogenetic analyses suggest that Rozellomycota is a paraphyletic group containing a 'core Microsporidia' clade (Corsaro et al., 2016), although others have argued for a different taxonomy with an expanded definition for the Microsporidia (Bass et al., 2018). The core Microsporidia (the term we use here) contains all the species traditionally classified as Microsporidia and which share diagnostic features of the group, including a coiled polar tube, a polaroplast, and the absence of a mitochondrial genome (Corsaro et al., 2016;Mikhailov et al., 2017). A total of 63 members of the protein family c_456 were identified among the Rozellomycota and core Microsporidia (Figure 1-source data 1). Phylogenetic analysis including homologues from prokaryotes and eukaryotes recovered the Microsporidia transporters as a separate clade with homologues from Mitosporidum, Amphiamblys and Rozella (Figure 1, Figure 1-figure supplements 1 and 2); providing evidence that the genes were present in the common ancestor of these species. Homologues of the Microsporidia proteins were also detected in other eukaryotes including free-living fungi, Oomycetes, Metazoa, Euglenozoa, Alveolata and Stramenopiles. The broad distribution of the genes across major lineages of eukaryote including free-living species suggests that this family of transport proteins may be ancestral among eukaryotes. In particular, the tree topology provides no compelling support for an origin of the Microsporidia genes through LGT from prokaryotes as previously suggested (Cuomo et al., 2012), and there is no evidence from the tree for a specific relationship to the E. coli NupG and XapB (Nakjang et al., 2013;Xie et al., 2004) transporters (Figure 1-figure supplements 1 and 2).
Detailed phylogenetic analysis identified two clades (A and B) of Microsporidia proteins (Nakjang et al., 2013) originating from a gene duplication in the common ancestor of the core Microsporidia and some Rozellomycota (Figure 1, Figure 1-figure supplement 1). Clade A contains at least one copy from every Microsporidia genome sampled. Clade B contains sequences from most Microsporidia apart from the Encephalitozoon/Nosema/Ordospora and Enterocytozoon/Vittaforma lineages, which appear to have lost the genes for these transporters (Figure 1). The majority of the Microsporidia sequences from clade A contain an indel (insertion or deletion) between the 7th and 8th transmembrane domains (TMD) compared to other proteins (Nakjang et al., 2013)

Antonosp s ora r r lo l l cus cu cu tae t t (ALOC_0053)
Anncaliia alg l l e g g rae r r (H312_00601) Anncaliia alg l l e g g rae r r (H312_01875) 66

Sp S S ra r r g a uea loph o ii (SLOPH_467)
Sp S S ra r r g a uea loph o ii (SLOPH_466) 100 100

Tr T T ac r r hip i i le l l is i i to t t p o hora r r homin mi mi is i i (THOM_3170)
Va V V vra r r ia culicis i i (VCUG_01849) 100 100  (Nakjang et al., 2013) providing the raw material (new genes) for functional divergence among paralogues (Dean et al., 2018;Heinz et al., 2014).

Tr T T ac r r hip i i le l l is i i to t t p o hora r r homin mi mi is i i (THOM_1681)
We used the four T. hominis sequences and related proteins from Microsporidia and Rozellomycota as search queries in HMMer searches against the Pfam domain database (Figure 1-source data 1). The Microsporidia members of clade A, one sequence from Mitosporidium daphniae and four from Rozella allomycis matched the MFS profile PF03825 found in nucleoside/H + symporters including E. coli NupG. This profile did not match any of the other transporter homologues, including the proteins from Amphiamblys sp. (Figure 1-source data 1). Microsporidia sequences from clade B were characterised by matches to the more general MFS transporter profiles MFS_1 (PF07690) and MFS_2 (PF13347) (Figure 1-source data 1). The alignment for the indel in sequences from clade A did not generate a significant hit (all e-values ! 13) in a HHPred profile-profile search of gene families in the protein domain databases (data not shown). The Microsporidia MFS transporters are predicted to have 12 alpha helical TMD, with both termini facing the cytoplasmic side ( Figure 1-figure supplement 4). This is a typical conformation (Reddy et al., 2012) as previously reported for the E. coli NupG and XapB nucleoside/H+ symporters (Nørholm and Dandanell, 2001;Vaziri et al., 2013;Xie et al., 2004) and is consistent with the Microsporidia proteins being functional MFS transporters.
In summary, all the Microsporidia proteins are members of the major facilitator family (MFS) of transport proteins, but there is no compelling evidence for a particularly close relationship to NupG, or strong indication from trees or bioinformatics concerning their substrates or transport mechanisms. In the rest of the manuscript we refer to the four T. hominis transporters as T. hominis ThMFS1-4 with the following locus tags and uniprot accessions (Heinz et al., 2012): in clade A, ThMFS1: THOM_0963 -L7J Â 55; in clade B, ThMFS2: THOM_1192 -L7J Â 19; ThMFS3: THOM_1681 -L7JVD0; ThMFS4: THOM_3170 -L7JT12.

ThMFS transporters are expressed during infection and localize to the cell surface of intracellular parasites
We used RNA-Seq and measured transcript abundance (expressed as transcripts per million, TPM) at six time points representing different stages in the parasite lifecycle, during a synchronised intracellular infection of rabbit kidney (RK13) cells by T. hominis (Dean et al., 2018). ThMFS1 mRNA was characterised by a relatively high abundance at the first time point (3 hr) followed by a gradual decrease at 14, 22 and 40 hr ( Figure 2A) and then a dramatic increase after 40 hr during spore     antisera from one of the ThMFS (red) and rat antisera to T. hominis mitochondrial Hsp70 (ThmitHsp70) (green), with the latter labelling parasite mitosomes and acting as a control for the fixation and permeabilization protocol (Freibert et al., 2017;Goldberg et al., 2008). DAPI (4',6-diamidino-2-phenylindole) (blue) was also added to stain parasite and host nuclear DNA ( Figure 2C). During intracellular infection  the surface plasma membrane of T. hominis is exposed to the host cell cytosol or to the lumen of an intracellular compartment called a sporophorous vesicle (SPOV) (Cali and Tokvarian, 2014). In the infective spore stage, the plasma membrane is protected from the external environment by a thick spore coat (Vávra and Lukesˇ, 2013). The antisera to ThMFS1 and ThMFS3 gave clear and specific labelling of the surface of parasites consistent with a plasma membrane location (Heinz et al., 2014) in samples taken during the time course of infection ( Figure 2C). By contrast, no parasite specific IFA signals were detected for the antisera to ThMFS2 and ThMFS4 at any stage of the infection cycle, despite observing consistent labelling of the mitosomes and nuclei of parasites (Figure 2-figure supplement 4). The reasons for the lack of ThMFS2 and ThMFS4 signals are not known but the expression levels of these two genes were much lower than for ThMFS1 and ThMFS3 (Figure 2A), so it is possible that the antibodies we used were insufficiently sensitive to detect expression in our model system or the proteins were not expressed under the experimental conditions used. Consistent with these possibilities, the antisera for ThMFS2 and ThMFS4 gave no parasite-specific signal in western blot analyses on total protein extracts from T. hominis infected RK13 cells and T. hominis spores, in contrast to the strong signals for antisera to ThMFS1 and ThMFS3 (Figure 2-figure supplement 5). The antisera for ThMFS2 and ThMFS4 also generated relatively weaker signals against the peptides used as antigens in dot blots suggesting the antibodies may be ineffective (Figure 2-figure supplement 6).
The antisera for ThMFS1 and ThMFS3 showed different patterns of parasite labelling during the infection ( Figure 2C), suggesting that they have evolved functional differences following gene duplication, as previously shown for the paralogous NTT transporters of T. hominis (Dean et al., 2018). We detected no labelling of the parasite surface by the antisera to ThMFS1 in replicate slides at 3 hr post-infection, despite transcript levels for the gene being relatively high at this point ( Figure 2A) and clear labelling of mitosomes and nuclei on the same slides ( Figure 2C). The ThMFS1 antisera gave strong labelling of the surface of parasites at 14 hr, 22 hr and 40 hr post-infection ( Figure 2C). At 70 hr and 96 hr post-infection, strong signal was located in the SPOV membrane surrounding groups of T. hominis ( Figure 2C) sporonts and spores (Hollister et al., 1996;Watson et al., 2015). This suggests that ThMFS1 has a role in supporting parasite development within the SPOV. The spores inside the SPOV were not labelled with either of the antibodies or by DAPI, suggesting that the developing spore coat may exclude access of these reagents. A lack of signal from spores in these types of experiments has previously been observed using antibodies to the ThNTT transporters and to several mitosomal proteins (Dean et al., 2018;Freibert et al., 2017;Goldberg et al., 2008). Proteomics data for purified T. hominis spores (Heinz et al., 2012) has shown, however, that ThMFS1 is present in spores. We detected strong parasite surface labelling by antisera to ThMFS3 from 3 hr up to 40 hr post-infection, but no labelling by antisera to ThMFS3 of either parasites or SPOV membrane in slides for 70 hr and 96 hr ( Figure 2C). The absence of signal for ThMFS3 on the surface of mature vegetative cells at 70 hr appears to be stage specific because fields of highly infected RK13 cells that contain a mix of mature and early stage parasites (through reinfection), show strongly labelled smaller cells (Figure 2-figure supplement 7). Labelling of small cells characteristic of earlier time points, with antisera for ThMFS1 and ThMFS3 at 96 hr post-infection, are also consistent with the initiation of new infections from germinating newly differentiated spores (Figure 2-figure supplement 8). Consistent with the absence in IFA of a detectable signal for ThMFS3 in the later stages of parasite development, ThMFS3 was not detected in proteomics data for purified T. hominis spores (Heinz et al., 2012).

Trachipleistophora hominis MFS proteins transport purine nucleotides but not uridine
Upon their discovery, it was suggested (Cuomo et al., 2012) that the Nematocida homologues of ThMFS transporters might be purine and pyrimidine nucleoside transporters like the E. coli NupG transporter. NupG has a broad specificity for nucleosides (Nørholm and Dandanell, 2001;Patching et al., 2005;Vaziri et al., 2013;Xie et al., 2004) including the pyrimidine nucleoside uridine (Nørholm and Dandanell, 2001;Xie et al., 2004). We used heterologous expression in E. coli strains (Dean et al., 2018;Heinz et al., 2014;Tsaousis et al., 2008) to test if the ThMFS1-4 proteins could transport radiolabelled uridine, using the E. coli NupG transporter as a positive control. Because expression of eukaryotic proteins in bacteria can be improved by codon optimisation in the bacterial host (Gustafsson et al., 2012) we tested expression of native and E. coli codon-optimised ThMFS in different E. coli strains. The E. coli strain GD1333, which lacks the two endogenous nucleoside transporters NupG and NupC (Nørholm and Dandanell, 2001), was used as the expression host in these assays ( Figure 3A). In contrast to the positive control expressing recombinant NupG, none of the ThMFS proteins transported [ 14 C]-uridine above background levels for the empty vector control (ptrc99a) ( Figure 3A). These data suggest that uridine is not a substrate for transport by ThMFS1-4.
We recently characterised the substrate specificities and evolution of a family of Microsporidia nucleotide transport proteins (NTT, Pfam profile PF03219) that import ATP and GTP for parasite growth and replication inside infected host cells (Dean et al., 2018;Heinz et al., 2014). These transporters (family # c_336 in Nakjang et al., 2013) are also members of the MFS protein superfamily (clan CL0015; Dean et al., 2018). To investigate if the structural similarity between the ThMFS and NTT transporters is reflected in their transport properties, we tested if the ThMFS proteins could transport radiolabelled purine and pyrimidine nucleotides when expressed in E. coli ( Figure 3B and C). All four ThMFS transported ATP above background, whereas E. coli expressing NupG did not transport ATP ( Figure 3B). Further experiments demonstrated that all four ThMFS proteins can also transport radiolabelled-GTP, but not radiolabelled-CTP or UTP ( Figure 3C). Consistent with the specific uptake of [a 32 P]-ATP and [a 32 P]-GTP, uptake of both substrates by E. coli cells expressing ThMFS transporters was time-dependent ( Figure 4). Our data demonstrate that ThMFS1-4 can transport purine and thus their substrate specificity overlaps with that of the previously characterised Microsporidia NTT nucleotide transporters (Dean et al., 2018;Heinz et al., 2014). The lack of a positive control in these assays makes it difficult to interpret the negative results for transport of the two tested pyrimidines.
The E. coli NupG transporter is a H + symporter that uses a proton gradient to drive the transport of uridine and other nucleosides (Xie et al., 2004). We have previously shown that some Microsporidia NTTs have evolved into symporters (Dean et al., 2018) capable of mediating the net import of nucleotides. To test if nucleotide uptake by ThMFS proteins is also proton dependent, we investigated [a 32 P]-ATP and [a 32 P]-GTP uptake by ThMFS1 and ThMFS3 in the presence of the protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP). ThMFS1 and ThMFS3 are the most abundant transcripts in RNA-Seq ( Figure 2A) and both are detected on the cell surface of actively growing parasites ( Figure 2C). The addition of CCCP had no effect on nucleotide uptake by E. coli expressing ThMFS1 or ThMFS3 ( Figure 5), suggesting that they are not H + -symporters. By contrast, CCCP did inhibit import of nucleotides by Protochlamydia amoebophila PamNTT5, a known H + -symporter of GTP (Haferkamp et al., 2006) (Figure 5). The nucleotide transport activity of some Microsporidia NTT (e.g. ThNTT4) were previously shown to be sensitive to CCCP treatment in the same expression system (Dean et al., 2018).

Conclusions
Previous studies have shown that Microsporidia lack the genes needed to make primary metabolites including nucleotides, and that they have a limited capacity to make their own energy (Dean et al., 2016;Williams et al., 2014). This raises the question of how these intracellular parasites obtain the enormous amounts of ATP and other nucleotides that they need to support their rapid growth and replication. For example, it is suggested that it takes at least 10 9 ATPs (Phillips and Milo, 2009) to make a single E. coli, and the ATP requirement to make the larger cells of Microsporidia is likely to be considerably higher. Previous work suggests that Microsporidia can use surface-located NTT nucleotide transporters to import purine nucleotides including ATP and GTP during their intracellular growth (Dean et al., 2016;Dean et al., 2018;Tsaousis et al., 2008). In the present study, we show that Microsporidia have a second set of MFS transporters that they can potentially use to supplement their energy and nucleotide budget. Thus, all four T. hominis ThMFS transporters were able to transport ATP and GTP when expressed in E. coli ( Figure 3C Figure 2C). The functional relevance of the transporters ThMFS2 and ThMFS4 is unclear as we were unable to detect any evidence of their protein expression in our model system. It was originally suggested that the Nematocida homologues of the T. hominis ThMFS transporters might have similar transport properties to the E. coli MFS NupG (Cuomo et al., 2012). NupG is an MFS H + symporter that uses the proton gradient to drive transport of purine and pyrimidine nucleosides (Xie et al., 2004). However, as we show in the present study, the levels of shared sequence similarity between NupG and Microsporidia sequences are generally low. Moreover, characterised members of the MFS superfamily that includes the Microsporidia sequences and NupG, transport a broad spectrum of ions and solutes using a variety of different mechanisms (Yan, 2015). This diversity makes it difficult to make reliable functional inferences based upon sequence similarity (Finn et al., 2016;Saier et al., 2016;Yan, 2015). This is exemplified by recently published data (Dean et al., 2018) for Microsporidia NTT transporters where substrate and mechanism vary among closely related paralogues (Dean et al., 2018). In this study, we show that ThMFS1-4 do not transport uridine, a known pyrimidine nucleoside substrate of NupG (Xie et al., 2004), and that ThMFS1 and ThMFS3, are not inhibited by CCCP, a classic inhibitor of H + symporters like NupG (Xie et al., 2004). Moreover, although T. hominis has a putative uridine kinase (Heinz et al., 2012), most Microsporidia genomes appear to lack the kinases needed to utilise uridine and other nucleosides in their metabolism (Dean et al., 2016), even if they could import them.
In recent years, a number of endoparasitic microbial eukaryotes related to Microsporidia have been identified and their genomes sequenced (Galindo et al., 2018;Haag et al., 2014;James et al., 2013;Mikhailov et al., 2017;Quandt et al., 2017). Phylogenomic analyses suggest that these taxa form a monophyletic sister group to the true fungi, for which the name Rozellomycota has been suggested (Corsaro et al., 2016), and that core Microsporidia form a distinct clade within Rozellomycota (Galindo et al., 2018;Mikhailov et al., 2017;Quandt et al., 2017). Like the Microsporidia, many of the Rozellomycota lack genes for making nucleotides de novo and some have a limited capacity for making their own ATP (Dean et al., 2016;Dean et al., 2018;Galindo et al., 2018;James et al., 2013;Mikhailov et al., 2017;Quandt et al., 2017). This suggests that they all depend to some degree upon host energy and nucleotides for their own intracellular growth and replication. Recent work has shown that Rozella and Microsporidia possess horizontally acquired nucleotide (NTT) transport proteins that they can use to import host ATP, and in the case of Microsporidia other purine nucleotides and NAD + (Dean et al., 2018). The topology of the NTT tree (Dean et al., 2018) suggests that NTT transporters were acquired by lateral gene transfer into the common ancestor of Rozella and Microsporidia and subsequently vertically inherited (Dean et al., 2018;Major et al., 2017). Analysis of the draft genomes of Amphiamblys sp. (Mikhailov et al., 2017), Metchnikovella incurvata (Galindo et al., 2018), Mitosporidium daphnia (Haag et al., 2014) and Paramicrosporidium saccamoebae (Quandt et al., 2017), suggest that they lack NTT genes. Based upon the topology of published trees (Dean et al., 2018;Galindo et al., 2018;James et al., 2013;Mikhailov et al., 2017;Quandt et al., 2017) which place these species as internal branches of the Rozellomycota, it appears that they have lost NTT genes during their independent evolution. By contrast, all these species contain multiple homologues of the ThMFS transporters ( Figure 1, Figure 1-source data 2) suggesting that these transporters provide an alternative and hitherto unrecognised way for endoparasites to exploit the host nucleotide pool. (C) Uptake assays for the four radiolabelled nucleotides (a 32 P-) ATP, GTP, CTP and UTP using the same expression system as in A and B. The same control (empty ptrc99a plasmid) was used for each tested substrate. For the Rozella NTT (RozNTT) cloned in pET16b, the control was the empty plasmid pET16b (Dean et al., 2018). N = 3 for each condition and the error bars represent standard deviations. Significant differences at p<0.05 (oneway ANOVA) between controls (empty plasmids, ptrc99a or pET-16b) and individual transporters are shown with * (ns: non-significant). DOI: https://doi.org/10.7554/eLife.47037.023 The following source data is available for figure 3: Source data 1. Raw data and their processing to calculate the transport of the tested nucleoside or nucleotides by ThMFS1-4 and control transporters expressed in E. coli (worksheet 1: panels A and B; worksheet 2: panel C  Figure 4. Time course of ATP and GTP uptake by E. coli cells expressing recombinant ThMFS1-4 proteins. Each ThMFS transporter was assayed using both E. coli-expression vector systems (pET16b or ptrc99a) with the results shown being taken from the experiment with the highest transport activity. In each experiment, the corresponding empty plasmid was used as control for background transport. The indicated substrates were all used at 0.5 mM.
(A) Uptake assay for the ThMFS1 (THOM_0963) protein expressed using pET16b in E. coli Rosetta2(DE3)pLysS and the native (not codon optimised) ORF. (B) Uptake assay for the ThMFS2 (THOM_1192) protein expressed with the ptrc99a plasmid system in E. coli GD1333 and the E. coli codon optimised synthetic ORF. (C) Uptake assay for the ThMFS3 (THOM_1681) protein expressed with pET16b system in E. coli Rosetta2(DE3)pLysS and the native ORF. (D) Uptake assay for the ThMFS4 (THOM_3170) protein expressed with the ptrc99a plasmid system in E. coli GD1333 and the E. coli codon optimised synthetic ORF. N = 3 for each condition with the error bars representing standard deviations. All 8 min time points for specified transporters and nucleotides were significantly different at p<0.05 (one-way ANOVA) from controls (empty plasmids, ptrc99a or pET-16b). DOI: https://doi.org/10.7554/eLife.47037.025 The following source data is available for figure 4: Source data 1. Raw data and their processing to calculate the transport of the tested nucleotides by ThMFS1-4 expressed in E. coli (worksheet 1: panels A and C; worksheet 2: panels B and D Continued on next page

RNA-Seq data generation and analyses
To initiate the synchronised infection by T. hominis, non-infected RK13 cells grown on 150 cm 2 round tissue culture dishes were incubated with freshly purified spores as previously described (Watson et al., 2015) before processing for RNA-Seq analysis as described (Dean et al., 2018;Watson et al., 2015). Briefly, 2 hr after the addition of T. hominis spores to the RK13 cells monolayer, the host cell monolayer was extensively washed with PBS in order to remove spores, followed by the addition of fresh culture medium. Two tissue culture dishes were processed independently for RNA purification, sequencing and mRNA quantification (Watson et al., 2015) at six different time points post-infection: 3 hr, 14 hr, 22 hr, 40 hr, 70 hr and 96 hr (Dean et al., 2018). Previous studies (in triplicate) indicated that the variation between independent replicates for this experimental design was low (Watson et al., 2015). Library preparation and sequencing were carried out using the Illumina stranded preparation kit and run on two lanes of an Illumina HiSeq 2500, which [a 32 P]-nucleotide import by ThMFS1, ThMFS3, and by the Protochlamydia amoebophila symporter NTT5 (PamNTT5) (Haferkamp et al., 2006) was compared in the absence (control, set to 100%) and presence of the protonophore CCCP. The GTP and ATP H + -symporter PamNTT5 was used as a positive control for CCCP inhibition (Haferkamp et al., 2006). N = 3 for each condition with the error bars representing standard deviations. The significant reduction of transport between control (no CCCP) and CCCP treatments at p<0.05 (one-way ANOVA) is indicated with *. DOI: https://doi.org/10.7554/eLife.47037.027 The following source data and figure supplement are available for figure 5: Source data 1. Raw transport data and their processing to investigate the impact of the protonophore CCCP on nucleotide import in E. coli expressing ThMFS1, ThMFS3 or the control PamNTT5 ( produced 27.5 M paired-end reads from both the RK13 host cells and T. hominis transcripts. CutAdapt (Martin, 2011) was used to remove adapter sequences and low quality 3' sequence regions (set at the q À20 threshold). Two approaches were used to quantify transcripts in the resulting dataset. Trimmed reads were mapped to the T. hominis genome using TopHat2 (Kim et al., 2013) and transcripts assembled and quantified using Cufflinks (Trapnell et al., 2012) as previously described (Dean et al., 2018;Watson et al., 2015). Transcript abundances were presented in FPKM (Fragments Per Kilobase per Million mapped reads) (Figure 2-source data 2) (Dean et al., 2018;Watson et al., 2015). In a second approach (presented in Figure 2A, Figure 2-source data 2), predicted T. hominis transcripts were quantified by pseudoalignment of reads using kallisto (Bray et al., 2016). The results were then analysed using sleuth (Pimentel et al., 2017) with transcript abundances presented as Transcripts Per Kilobase per million mapped reads (TPM) (Wagner et al., 2012). The RNA-Seq data were submitted to GenBank and listed at the NCBI under the existing BioProject PRJNA278775 with the BioSample accession numbers SAMN11265032-SAMN11265043 (one accession number for each of the two samples per time point post infection).

Antisera generation
Pairs of peptides were selected from hydrophilic segments of the ThMFS1-4 amino acid sequences and used to generate custom-made antisera targeting each of the four ThMFS1-4 candidate transporters ( Figure 2-figure supplement 6). Anti-peptides antisera were produced in rabbits by Bio-Genes GmbH (Germany) and commercially affinity purified against the peptide antigens. Two rabbits for each pair of peptides were processed in parallel. To test the antisera specificity for their respective peptides, we used western blots with 300 ng of peptide directly spotted onto nitrocellulose membranes, affinity purified antibodies were diluted 1:1000 and secondary goat anti-rabbit antisera conjugated to HRP were diluted 1:10000 ( Figure 2-figure supplement 6). Western blot detection was processed with a chemiluminescent substrate (Thermo Scientific -Pierce ECL) following the manufacturer's instructions. Image development was processed with a Biorad gel imager ChemiDoc XRS+ with images taken every 10 s.

Immunofluorescent microscopy
The same T. hominis infected RK13 cells used for the synchronised infection experiments (see RNA-Seq section) were also grown on 13 mm glass coverslips and then fixed in methanol/acetone (50:50 at À20˚C for at least 10 min) (Dean et al., 2018;Goldberg et al., 2008;Heinz et al., 2014;Tsaousis et al., 2008) at the different time points post-infection. Rat antisera against ThmtHSP70 (locus tag THOM_3057) were used to label T. hominis mitosomes and both host and parasite nuclei were stained with DAPI, as previously described (Dean et al., 2018;Freibert et al., 2017;Goldberg et al., 2008;Heinz et al., 2014;Tsaousis et al., 2008). Dilutions of antisera were as follows: anti-ThmtHSP70 1:200; anti-ThMFS1-4 1:50. For anti-ThMFS2 and anti-ThMFS4 1:10 and 1:2 dilutions were also tested but no IFA signal was observed in any of the tested conditions. For ThMFS1 and ThMFS3 both rabbit antisera gave parasite-specific IFA signals. Blocking and antibody incubations were performed in 5% milk prepared from skimmed milk powder in PBS, with PBS used for washing steps. Cells were stained with DAPI (Molecular Probes) and mounted in Vectashield Hard set (VectorLabs). Peptide competition assays were performed with the addition of solubilized pairs of peptides in a 200-fold molar excess compared to the affinity-purified antibody. The IgG molar mass was considered as 150,000 Daltons, and the molar masses of the blocking peptides were taken into consideration for the calculations. For ThMFS1, the final antibody concentration was 5.2 mg/ml and the concentrations of peptide 1 and peptide 2 were 13.6 mg/ml and 11.2 mg/ml, respectively. For ThMFS3, the final antibody concentration was 3.3 mg/ml and the concentrations of peptide 1 and peptide 2 were 8.3 mg/ml and 8.5 mg/ml, respectively. As expected, the control mitosomal ThmitHsp70 signal was not affected by any of these conditions (Figure 2-figure supplement  9). Images were taken with a Zeiss Axioimager II epifluorescence microscope using a 63x objective lens and images processed with ImageJ (Schneider et al., 2012). Quantification of the ThMFS antibody fluorescence signal was determined for fields including intracellular stages of the parasite. Axiovision software and Image J[Fiji] with the Bioimport plugin was used to determine relative fluorescence for each specific antibody including the mitosome-specific antibody to mitHsp70. Background signal was determined in a similar manner and subtracted from the specific signal.

Western blot analyses on total protein extracts from cell cultures and purified spores
For total protein extracts, confluent monolayers of either T. hominis-infected or non-infected RK13 cells were washed three times with PBS, and lysed with ice cold 2% SDS-PBS lysis buffer containing protease inhibitor cocktail (Sigma P8340), PMSF (Sigma), MgCl2 (0.6 mM final concentration), and Benzonase (25U, Novagen). T. hominis spores were purified from the culture medium from T. hominis infected RK13 cultures using Percoll (Sigma) density gradient centrifugation. Purified spores were lysed by boiling for 10 min in 2% SDS-PBS lysis buffer with protease inhibitors. The protein concentration of the samples for Western blotting was determined with a BCA assay (Pierce BCA Protein Assay Kit). The samples were boiled for 5 min in Laemmli loading buffer and the indicated amount of total protein extracts were loaded per lane for SDS-PAGE. To estimate the size of the detected proteins and to monitor protein migration and transfer we used the PageRuler prestained protein ladder, 10 to 180 kDa (ThermoFisher Scientific). Western blot analysis used the indicated affinity purified rabbit antibodies (diluted 1:1000) in combination with HRP-conjugated secondary goat antirabbit antisera (diluted 1:10000) (Sigma). Image development was processed using a Biorad gel imager ChemiDoc XRS+.

Heterologous expression and transport assays in E. coli cells
The four native ThMFS genes were PCR amplified from purified genomic DNA and cloned into the pET-16b expression vector. Sequences of cloned genes were verified and submitted to GenBank when distinct from the previously deposited genome sequence-derived ORF (Heinz et al., 2012). These have the following accession numbers: ThMFS2_native: MH824667, ThMFS3_native: MH824668. Codon-optimized genes for expression in E. coli that contained a C-terminal 2x HA-tag were produced synthetically (GeneArt, Thermo Fisher Scientific) and then cloned into the ptrc99a plasmid. These sequences have the following GenBank accession numbers: ThMFS1_synthetic: MH824663; ThMFS2_synthetic: MH824664; ThMFS3_synthetic: MH824665; ThMFS4_synthetic: MH824666. Proteins were expressed as described previously for Microsporidian NTTs (Dean et al., 2018;Heinz et al., 2014;Tsaousis et al., 2008). T7-based constructs were used for expression in E. coli Rosetta2(DE3)pLysS and trc-based constructs were expressed in E. coli GD1333 (kindly provided by Prof. Gert Dandanell) (Nørholm and Dandanell, 2001). Briefly, a starting culture from a single colony was grown in LB at 37˚C overnight and used to inoculate a 50 ml culture grown at 37˚C until an OD 600 of 0.4-0.6. Protein expression was induced by addition of 1 mM IPTG and the culture was incubated for and additional 16 hr at 18˚C. Cells were harvested at 5000 x g for 10 min and were washed twice with PBS to remove residual medium. The cells were re-suspended in PBS with the OD 600 adjusted to 5/ml. Cultures were kept in PBS at room temperature until used for uptake assays. Transport of nucleosides and nucleotides was assayed using 0.5 mM or 2 mM of radiolabelled substrate, as described in the main text/Figures. a 32 P-radioisotopes were present at 1-2 mCi/ml and used as described (Audia and Winkler, 2006), and [U-14 C] isotopes were used at 2.5-5 mCi/mmol as described (Nørholm and Dandanell, 2001). To test if a proton gradient affected transport by E. coli cell expressing the ThMFS1 and ThMFS3 transporters, we added 250 mM of CCCP to dissipate the H + gradient across the bacterial membranes prior to transport assays (Dean et al., 2018;Haferkamp et al., 2006).