Retortamonads from vertebrate hosts share features of anaerobic metabolism and pre-adaptations to parasitism with diplomonads.

Although the mitochondria of extant eukaryotes share a single origin, functionally these organelles diversified to a great extent, reflecting lifestyles of the organisms that host them. In anaerobic protists of the group Metamonada, mitochondria are present in reduced forms (also termed hydrogenosomes or mitosomes) and a complete loss of mitochondrion in Monocercomonoides exilis (Metamonada:Preaxostyla) has also been reported. Within metamonads, retortamonads from the gastrointestinal tract of vertebrates form a sister group to parasitic diplomonads (e.g. Giardia and Spironucleus) and have also been hypothesized to completely lack mitochondria. We obtained transcriptomic data from Retortamonas dobelli and R. caviae and searched for enzymes of the core metabolism as well as mitochondrion- and parasitism-related proteins. Our results indicate that retortamonads have a streamlined metabolism lacking pathways for metabolites they are probably capable of obtaining from prey bacteria or their environment, reminiscent of the biochemical arrangement in other metamonads. Retortamonads were surprisingly found do encode homologs of components of Giardia's remarkable ventral disk, as well as homologs of regulatory NEK kinases and secreted lytic enzymes known for involvement in host colonization by Giardia. These can be considered pre-adaptations of these intestinal microorganisms to parasitism. Furthermore, we found traces of the mitochondrial metabolism represented by iron‑sulfur cluster assembly subunits, subunits of mitochondrial translocation and chaperone machinery and, importantly, [FeFe]‑hydrogenases and hydrogenase maturases (HydE, HydF and HydG). Altogether, our results strongly suggest that a remnant mitochondrion is still present.

Introduction

Retortamonadida (Metamonada:Fornicata), represented by genera Retortamonas and Chilomastix, are bi- and quadriflagellates with a conspicuous cytostome employed to ingest food constituted by bacteria and organic particles. The recurrent flagellum is decorated by lateral vanes, passes through the cytostomal pouch and propels food particles into the cytopharynx, whereas the smooth anterior flagellum (or flagella) serves cell locomotion. Under adverse conditions or to disperse, Retortamonadida retract their flagella, develop a sturdy cell wall composed of filamentous material, and encyst [1]. Despite having ultrastructural similarities, Retortamonadida are polyphyletic, with Chilomastix and insect Retortamonas species being sister lineages and branching among Carpediemonas-like organisms, whereas Retortamonas species from vertebrate hosts are in fact sister to Diplomonadida (Fig. 1) [2,3].

Fig. 1

Retortamonads fall into two separate groups among Fornicata, with vertebrate commensals R. caviae and R. dobelli branching sister to Diplomonada. A, schematic tree based on SSU rDNA phylogeny from [1], also corroborated by multi-gene analyses [2]. B, phase-contrast light micrographs of retortamonads from this work; scale bar = 2 μm.

Most often, retortamonads are found as endocommensals in the digestive tracts of vertebrates and invertebrates, and a single known free-living species, Chilomastix cuspidata, inhabits hypoxic water sediments. Only two species are considered potentially pathogenic, C. mesnili and C. gallinarum [1]. The subjects of this work are known as vertebrate commensals. R. caviae is found in mammals (guinea-pigs) and R. dobelli in amphibians and reptiles [4].

There are electron-microscopical observations of hydrogenosome-like organelles in Chilomastix spp. but no evidence of such organelle has been presented for any Retortamonas species, raising hypotheses about their amitochondriate status [1,5]. Indeed, the ancestor of Fornicata had been adapted to anoxic conditions, but apparently retained a rather complex mitochondrial metabolism including anaerobic ATP production. Modern fornicates have evolved into a wide range of parasitic and free-living lineages that possess a spectrum of reduced mitochondrion-related organelles (MROs) [[6], [7], [8]]. Functions that reside in the MROs of Fornicata include succinyl-CoA synthase or acetyl-CoA synthase acting in ATP production via substrate-level phosphorylation; [FeFe]‑hydrogenase (HYD) producing molecular hydrogen as the ultimate electron sink; the glycine cleavage system (GCS) and serine hydroxymethyltransferase (SHMT) that partake in one‑carbon pool metabolism; iron‑sulfur cluster assembly complex ISC; and the mitochondrial translocation machinery TOM/TIM. Some of these functions were gradually lost or gained in some fornicate lineages alongside a switch to endobiotic and parasitic lifestyle.

To inspect the presence and metabolic capabilities of mitochondria in vertebrate retortamonads, we obtained transcriptomic data from xenic monoeukaryotic cultures of two Retortamonas species, R. dobelli and R. caviae. By means of similarity searches and phylogenetic inference we construct a model of their reduced metabolism and find sequence evidence for MROs in both species. We also describe pre-adaptations for parasitism further developed by the infamous diplomonad pathogens Giardia and Spironucleus.

Results

Retortamonads employ a streamlined economical core metabolism

After removal of prokaryotic and eukaryotic contaminants (see Material and Methods), the transcriptomic data of R. caviae and R. dobelli consisted of 45,847 and 28,032 contigs, respectively (Trinity assembly; Table 1). These respectively translated to 8380 and 6222 non-redundant predicted proteins longer than 100 amino acids. The transcriptomes were also assembled using rnaSPAdes, but the resulting assemblies were much more fragmented, and we did not analyze them further, besides searching them for longer contigs. In the Trinity-assembled transcriptomes, only about 30% of the ODB_v10 BUSCO conserved orthologs were present, but these figures are comparable to that of the Giardia intestinalis Assemblage A protein set that is based on genomic data (Supplementary Fig. S1). To address whether our data seem incomplete because of a lack of conservancy of BUSCO orthologs in Metamonada rather than a lack of sequence coverage, we searched for 72 proteins annotated as ribosome-associated in G. intestinalis. We were able to recover both retortamonad orthologs of all but 6 ribosomal proteins and only in one case (RPL35a) we could not find an ortholog in either R. caviae or R. dobelli (Supplementary Table S1). These results suggest that at least highly expressed proteins have a good representation in our transcriptomic data.

Table 1

General characteristics of the presented transcriptomes.

	R. caviae BELLA		R. dobelli CERAT-3R
	Trinity	SPAdesa	Trinity	SPAdesa
# contigs	45,847	221,479	28,032	116,528
N50	1,791	1,599	1,701	2,403
total length	34,978,710	119,389,613	20,712,416	70,747,992
mean GC%	31.75	30.09	35.85	34.83
predicted proteins	8,380	24,164	6,222	13,818

SPAdes assembly statistics for contigs >200 bp.

We investigated which biological pathways are represented in retortamonad transcriptomes that could illuminate the core metabolism of these bacterivorous flagellates (Fig. 2A, Supplementary Table S1). To present a rigorous picture, we considered phylogenetic relationship to diplomonads and Dysnectes brevis in order to identify true retortamonad sequences, and homolog annotations and InterPro domain structure to assign their functions. We found most enzymes of glycolysis, pyruvate hub and storage carbohydrate metabolism quintessential for substrate-level ATP production and conservation. Notably, two glycolytic enzymes, phosphofructokinase (PFK) and pyruvate, phosphate dikinase (PPDK), use pyrophosphate (PPi) as substrate during phosphorylation (the former phosphorylates d-fructose 6-phosphate, the latter AMP to ATP). PPi is a by-product of some hydrolytic reactions (e.g. nucleic acid and aminoacyl-tRNA synthesis) and specificity for this lower-energy phosphate molecule in core metabolism suggests this resource is economically used [9]. Furthermore, enzymes specific for PPi perform physiologically reversible phosphorylation depending on PPi availability. In Metamonada, the presence of PPi-dependent PFK and PPDK is not unusual [10], whereas most eukaryotes employ ATP-dependent enzymes for the respective glycolytic and gluconeogenetic reaction [11]. Functionally, PPi-PFK seems to replace ATP-dependent fructose bisphosphatase, whereas PPi-PPDK and ATP-pyruvate kinase apparently have non-overlapping metabolic functions and both are retained in retortamonads.

Fig. 2

Overview of the pathways found in vertebrate retortamonads. A, core metabolic pathways reconstructed from transcriptomic data showing enzymatic steps where substrate-level phosphorylation yields chemical energy in the form of ATP. Abbreviations: 2OG – 2-oxoglutarate, FA – fatty acid, MVA – mevalonate, OAA – oxaloacetate, orn – ornithine, pyr – pyruvate, R5P – ribose 5-phosphate, (m)THF – (5,10-methylene)-tetrahydrofolate. Numbers denote enzymes as follows. Arg dihydrolase pathway: 1 – Arg deiminase, 2 – orn carbamoyltransferase, 3 – carbamate kinase, 4 – orn decarboxylase. Storage sugars: 1 – phosphoglucomutase, 2 – glucose-phosphate uridylyltransferase, 3 – glycogen synthase, 4 – glycogen branching enzyme and glucanotransferase, 5 – glycogen debranching enzyme, 6 – glycogen phosphorylase, α- and β-amylase, 6B – ADP-sugar diphosphatase. Glycolysis: 1 – glucokinase, 2 – glucose-phosphate isomerase, 3 – PPi-phosphofructokinase, 4 – aldolase, 5 – triose-phosphate isomerase, 6 – glyceraldehyde-phosphate dehydrogenase, 7 – phosphoglycerate kinase, 8 – phosphoglycerate mutase, 9 – enolase. Pentose-phosphate cycle: 1 – transketolase, 2 – ribose-phosphate epimerase, 3 – ribose-phosphate isomerase. Pyruvate hub: 1 – pyruvate kinase, 2 – pyruvate, phosphate dikinase, 3 – pyruvate:ferredoxin oxidoreductase, 4 – acetyl-CoA synthase, 5 – alcohol dehydrogenase, 6 – aldehyde dehydrogenase, 7 – Ser ammonia lyase, 8 – malic enzyme, 9 – malate dehydrogenase, 10 – PEP carboxykinase, 11 – methylmalonyl-CoA:pyruvate transcarboxylase. Transamination: 1 – Asp transaminase, 2 – Glu dehydrogenase, 3 – Ala transaminase, 4 – asparaginase, 5 – Asp-ammonia ligase. Cys synthesis: 1 – Ser O-acetyltransferase, 2 – Cys synthase. Fatty acid elongation: 1 – acyl-CoA synthase, 2 – ketoacyl-CoA synthase, 3 – ketoacyl-CoA reductase, 4 – hydroxyacyl-CoA dehydratase, 5 – trans-enoyl-CoA reductase, 6 – acyl-CoA thioesterase. Terpenoids: 1 – acetyl-CoA acetyltransferase, 2 – HMGC synthase, 3 – HMGC reductase, 4 – mevalonate kinase, 5 – phosphomevalonate kinase, 6 – diphosphomevalonate decarboxylase, 7 – IPP isomerase, 8 – IPP synthase, 9 – undecaprenyl diphosphate synthase, 10 – farnesyltransferase. B, summary of mitochondrial translocon subunits found in retortamonads. Color code same as in A. C, summary of components of the Fe-S cluster assembly systems ISC and CIA found in retortamonads. For simplicity, only one subunit is drawn per complex, although most subunits are typically multimeric. Color code same as in A, arrows designate cluster assembly steps and transfer among the subunits and to the client proteins. How exactly clusters are assembled with IscA is a matter of debate.

Another energy resource that appears to be utilized by retortamonads is arginine, based on the presence of the arginine dihydrolase pathway (Fig. 2A). By decomposition of arginine to CO2, ammonia and putrescine, the cells are probably capable of generating ATP while maintaining a source of polyamines. This is one of the main sources of ATP also in Giardia and anaerobically grown Hexamita [12,13], and to certain extent also Trichomonas vaginalis, and Spironucleus salmonicida [14,15].

Among anabolic pathways, we also found enzymes of the pentose phosphate cycle, terpenoid synthesis, fatty acid elongation and lipid synthesis. The pentose phosphate cycle is a set of interconversion reactions that yield the sugar components of nucleotides, ribose and deoxyribose. In retortamonads as well as diplomonads, the cycle functions without the transaldolase step, analogously to some bacteria (Fig. 2A) [16]. Notably, PPi-phosphofructokinase partakes in these reactions, with a gain of one PPi molecule per “cycle” (Fig. 2A). Terpenoids are synthesized de novo via the mevalonate pathway with the precursor (acetoacetyl-CoA) derived from the pyruvate pool. Terpenoid derivatives are apparently required for protein tethering in the cellular membranes of retortamonads and possibly other lipidic compounds typical of eukaryotic cells (see farnesyltransferases and prenyl-protein modifying enzymes in Supplementary Table S1). Retortamonads seem to have lost the capacity for de novo synthesis of fatty acids (C14-C18), so it might be that these are taken from the bacterial prey. Instead, retortamonads possess a suite of enzymes for fatty acid elongation and phospholipid synthesis (Fig. 2A, Supplementary Table S1), similar to Giardia [17]. No fatty acid desaturases were found. The intriguing (methyl)malonyl-CoA:pyruvate transcarboxylase fusion enzyme from Giardia [18] has orthologues in both retortamonads and might be involved in fatty acid elongation by providing malonyl-CoA.

Amino acid synthesis pathways are almost absent from our data, except for Cys synthesis (from Ser via serine O-acetyltransferase and cysteine synthase) and Met salvage (via S-adenosylmethionine synthase and S-adenosylhomocysteine nucleosidase). The ability to synthesize Cys is in line with the central role of this amino acid in oxidative stress protection in Giardia [19] and modulation of their cell surface [20], possibly conserved in retortamonads as well. Using a series of transamination reactions linked to the pyruvate hub, several other amino acids can be catabolized or converted, including Asp, Glu, and Ala, partly contributing to the core energy metabolism (Fig. 2A). Most amino acids are thus provided by the prey bacteria or obtained from the environment. This limited amino acid synthesis capacity is shared with Giardia and other diplomonads (Supplementary Table S1) [21,22].

Among metabolite transporters, we could only find sequence evidence for ABC and major facilitator superfamily (MFS) proteins. The ABC transporters from retortamonads belong to the A, C, D, and G subfamilies with limited information on their substrate specificity, given their weak phylogenetic affiliation with reference sequences. ABCA homologs might act as cholesterol/phospholipid flippases or transporters of other lipidic compounds; ABCC homologs are known as drug conjugate transporters and no specificity could be assigned; ABCD homologs belong to a broad fatty acyl-CoA transporter family; and ABCG homologs have a broad spectrum of specificity ranging from pigments, lipids, waxes, volatiles and xenobiotic compounds (Supplementary Fig. S2). MFS transporters include transporters of sugars and sugar phosphates (clades II, V and VII; Supplementary Fig. S2, Supplementary Table S1), nitrogen compounds and perhaps amino acids (clades III, IV and VI), riboflavin and unknown solutes (clade I). As with ABC transporters, specificities are also difficult to assess for MFS transporters, since functional data are lacking, and phylogenetic signals and reference annotations are mostly weak. We did not detect any mitochondrial carrier proteins even using HMM profiles, which is consistent with their apparent lack in diplomonads.

There is a mitochondrion in retortamonads involved in hydrogen and cofactor production

We searched the retortamonad transcriptomes for hallmark MRO proteins to determine if this organelle is present. We found most of the mitochondrial translocon subunits known from Giardia (namely subunits of complexes TOM and TIM23), mitochondrial chaperones, and subunits of the processing peptidase (Fig. 2B) [[23], [24], [25]], indicative that proteins are translocated into this compartment. Furthermore, we found homologs of Giardia spp., Spironucleus salmonicida and Dysnectes brevis mitosomal/MRO proteins described in previous works (Table 2) [26,27]. This is in contrast with available ultrastructural reports and also our later attempts that failed to identify mitochondria in retortamonads [28]. As pointed out before [8], few mitochondrial proteins of fornicates possess a clear mitochondrial targeting presequence albeit homolog sequence alignments support their N-terminal completeness. This is also true for retortamonads (Supplementary Table S1).

Table 2

MRO-localized proteins in retortamonads, Dysnectes brevis and diplomonad representatives Giardia intestinalis Assemblage A (WB), Giardia muris Roberts-Thomson, and Spironucleus salmonicida ATCC50377.

Pathway	Abbr.	Function	R. caviae BELLA	R. dobelli CERAT3RET	Dysnectes brevis	Giardia intestinalis WB	Giardia muris	Spironucleus salmonicida	REF
translocation	Tom40	protein translocon	TRINITY_DN116128	NODE_2681	Dysnectes_Gene.9220	GL50803_17161	GMRT_12633	SS50377_17947	[26]
translocation	MOmp35	protein translocon	—	—	—	GL50803_14939	GMRT_11472	—	[26]
translocation	Tim44	protein translocon	—	—	Dysnectes_Gene.6028	GL50803_14845	GMRT_10149	—	[26]
translocation	Tim17/23	protein translocon	—	—	—	GL50803_10452	GMRT_10619	—	[25]
translocation	tim14/pam18	protein translocon	TRINITY_DN155484	TRINITY_DN96875	—	GL50803_300001	GMRT_10830	SS50377_16563	[26]
translocation	pam16	protein translocon	TRINITY_DN116568	TRINITY_DN96540	Dysnectes_Gene.10499	GL50803_19230	GMRT_13588	—	[26]
translocation	Hsp70	chaperone	TRINITY_DN115501	TRINITY_DN79477	Dysnectes_Gene.1294	GL50803_14581	GMRT_12714	SS50377_17850	[26]
translocation	Mge1_GrpE	exchange factor	TRINITY_DN117516	TRINITY_DN47121	Dysnectes_Gene.1980	GL50803_1376	GMRT_14282	SS50377_11178	[26]
translocation	dnaJ-typeIII	chaperone	TRINITY_DN116144	TRINITY_DN79881	Dysnectes_Gene.9975	GL50803_9751	GMRT_14124	SS50377_17851	[26]
translocation	DnaJ-Jac1	chaperone	TRINITY_DN99286	TRINITY_DN73315	Dysnectes_Gene.7890	GL50803_17030	GMRT_11701	—	[26]
translocation	Hsp60/Cpn60	chaperone	TRINITY_DN116520	TRINITY_DN5016	Dysnectes_Gene.1999?	GL50803_103891	GMRT_14232	SS50377_14834	[26]
translocation	Cpn10	co-chaperone	—	—	—	GL50803_29500	GMRT_10354	—	[26]
translocation	MPPb	mitochondrial processing peptidase	NODE_16301	TRINITY_DN79029	Dysnectes_Gene.12240	GL50803_9478	GMRT_10827	—	[26]
translocation	VAP	vesicle-associated protein	—	—	?	GL50803_15985	GMRT_10767	SS50377_10497	[26]
redox		NADH oxidase	TRINITY_DN115043	TRINITY_DN79263	Dysnectes_Gene.8823	GL50803_9719	GMRT_12238, GMRT_16222	SS50377_12178	[26]
redox		NADH oxidase	TRINITY_DN117413	TRINITY_DN79173	Dysnectes_Gene.5055, Dysnectes_Gene.6431	—	—	SS50377_16942
redox/ISC		NADPH-P450 reductase GiOR-1	—	—	?	GL50803_91252	GMRT_14862	—	[26]
redox/ISC		NADPH-P450 reductase GiOR-2	—	—	?	GL50803_15897	—	—	[26]
redox	PRX	peroxiredoxin	TRINITY_DN38938	TRINITY_DN80435	Dysnectes_Gene.1251	GL50803_16076, GL50803_15383a, GL50803_14521a	GMRT_20713, GMRT_23775, GMRT_23890	SS50377_12593, SS50377_13435, SS50377_15339	[26]
redox	TXR	thioredoxin reductase	TRINITY_DN113934, TRINITY_DN102420	TRINITY_DN80271, TRINITY_DN57461	Dysnectes_Gene.2142	GL50803_9827	GMRT_15171	SS50377_14835	[26]
ISC	Fd	Ferredoxin_Yah1	TRINITY_DN113917	TRINITY_DN99387	—	GL50803_27266	GMRT_fx013	SS50377_11494	[26]
ISC	Fra	frataxin	TRINITY_DN114123	TRINITY_DN78607	—	—	—	SS50377_10981
ISC		IscU_N/NifU-like	TRINITY_DN115747	TRINITY_DN106642	Dysnectes_Gene.1864	GL50803_15196	GMRT_12363	SS50377_11862	[26]
ISC		NifU	TRINITY_DN96174	TRINITY_DN47383	Dysnectes_Gene.3874	GL50803_32838	GMRT_12117	SS50377_19167	[26]
ISC		IscS	TRINITY_DN141819	TRINITY_DN81761	Dysnectes_Gene.654	GL50803_14519	GMRT_12737	SS50377_15482	[26]
ISC		IscS	TRINITY_DN115446	TRINITY_DN80249	Dysnectes_Gene.4329	—	—	SS50377_16780
ISC	Grx5	glutaredoxin 5	TRINITY_DN114492	TRINITY_DN57215	Dysnectes_Gene.1665	GL50803_2013	GMRT_10532	—	[26]
ISC		rhodanese repair	—	—	—	GL50803_27910	—	—	[27]
ISC		cdc25-rhodanase domain	TRINITY_DN116264	TRINITY_DN72193, TRINITY_DN67314	Dysnectes_Gene.1366	GL50803_4369a	GMRT_13177	SS50377_11382
ISC		iscA/hesB	TRINITY_DN114783	TRINITY_DN75416	Dysnectes_Gene.2212	GL50803_14821	GMRT_14040	—	[26]
unknown		hypothetical, armadillo repeat	—	—	—	GL50803_16386	GMRT_10254	—	[27]
unknown		hypothetical, sorting nexin-22/24	—	—	—	GL50803_16596	GMRT_12734	—	[26]
unknown		hypothetical, DUF1624 + TMD	—	—	Dysnectes_Gene.5026	GL50803_17236	GMRT_12999	—	[26]
unknown		hypothetical, G.intestinalis-specific	—	—	—	GL50803_8148, GL50803_8358, GL50803_12229, GL50803_12999, GL50803_17276	—	—	[26,27]
unknown		hypothetical	—	—	—	GL50803_3491	GMRT_13073	SS50377_15743	[26,27]
unknown		hypothetical	—	—	Dysnectes_Gene.12635	GL50803_4768	GMRT_12055	SS50377_15024	[26]
unknown		hypothetical	—	—	—	GL50803_4852	GMRT_12235	—	[27]
unknown		hypothetical	—	—	—	GL50803_7035	GMRT_11511	—	[27]
unknown		hypothetical	—	—	—	GL50803_9296	GMRT_15786	—	[26]
unknown		hypothetical	—	—	—	GL50803_10971	GMRT_14213	—	[27]
unknown		hypothetical	—	—	—	GL50803_22587	GMRT_10293	—	[27]
unknown		MLF1IP	TRINITY_DN116329	TRINITY_DN80152	—	GL50803_16424	GMRT_11758	SS50377_12584	[26,27]
unknown		guanylate kinase	TRINITY_DN12251	TRINITY_DN62004	Dysnectes_Gene.1097	GL50803_7203	GMRT_15800	SS50377_16947, SS50377_16948	[26]
transporter	ABCA	ABC transporter	TRINITY_DN14848, TRINITY_DN116477	TRINITY_DN80473_g1, TRINITY_DN80473_g3	Dysnectes_Gene.256	GL50803_3470, GL50803_16575a	GMRT_12500	SS50377_14356	[26]
transporter	ABCA	ABC transporter	TRINITY_DN113921, TRINITY_DN114109, TRINITY_DN114710	TRINITY_DN79535, TRINITY_DN57692	Dysnectes_Gene.10941, Dysnectes_Gene.18923, Dysnectes_Gene.13067, Dysnectes_Gene.15987, Dysnectes_Gene.22435	GL50803_21411, GL50803_113876a, GL50803_94478a	GMRT_16378, GMRT_15176	SS50377_17655, SS50377_14280	[26]
transporter	ABC	ABC transporter	—	—	—	GL50803_17165	GMRT_12185	SS50377_12476	[26]
transporter	ABC	ABC transporter	—	—	—	GL50803_87446	GMRT_12182	SS50377_11649	[26]
transporter	MFS	putative transporter	—	—	—	GL50803_17296, GL50803_6132a	GMRT_15249, GMRT_11675	SS50377_19082, SS50377_jh029, SS50377_14281	[26]
transporter	MFS	putative transporter	TRINITY_DN115802, TRINITY_DN116697_g3, TRINITY_DN116697_g4, TRINITY_DN91660	TRINITY_DN80264_g1, TRINITY_DN80264_g4, TRINITY_DN80595_g2, TRINITY_DN80595_g3, TRINITY_DN39094	Dysnectes_Gene.1944	GL50803_17342, GL50803_112063a, GL50803_114777a	GMRT_15581, GMRT_15602	SS50377_19153, SS50377_17880, SS50377_17525, SS50377_16939, SS50377_16984	[26]
hydrogenase	HYDA	2Fe-2S hydrogenase	TRINITY_DN117283	TRINITY_DN119378	Dysnectes_Gene.5365	GL50803_6304	GMRT_12611	SS50377_12678 cluster	[26]
hydrogenase	HYDA	2Fe-2S hydrogenase	—	—	Dysnectes_Gene.5440	—	—	SS50377_16154
hydrogenase	HYDA	2Fe-2S hydrogenase	TRINITY_DN116842	TRINITY_DN80127	—	—	—	SS50377_14286
hydrogenase	HYDE	hydrogenase maturase subunit	TRINITY_DN117330	TRINITY_DN79253	Dysnectes_Gene.508	—	—	SS50377_14972
hydrogenase	HYDF	hydrogenase maturase subunit	TRINITY_DN115593	TRINITY_DN77151	Dysnectes_Gene.931	—	—	SS50377_16918
hydrogenase	HYDG	hydrogenase maturase subunit	TRINITY_DN113532	TRINITY_DN80475	Dysnectes_Gene.1629	—	—	SS50377_17876
one-carbon pool	GCSH	glycine cleavage system	—	—	Dysnectes_Gene.912	—	—	—
one-carbon pool	GCSL/DLD	glycine cleavage system	—	—	Dysnectes_Gene.6035	—	—	—
one-carbon pool	GCSP1	glycine cleavage system	—	—	Dysnectes_Gene.516	—	—	—
one-carbon pool	GCSP2	glycine cleavage system	—	—	Dysnectes_Gene.483	—	—	—
one-carbon pool	GCST	glycine cleavage system	—	—	Dysnectes_Gene.2916	—	—	—
one-carbon pool	SHMT	serine hydroxymethyltransferase	TRINITY_DN115899	TRINITY_DN117879	Dysnectes_Gene.14360	—	—	SS50377_17865
one-carbon pool	FolC	folate-polyglutamate synthase	TRINITY_DN115540	TRINITY_DN80349	Dysnectes_Gene.3784	—	—	SS50377_13420
pyruvate hub	ACS	acetyl-CoA synthase	—	—	—	—	—	SS50377_10513
terpenoids	ACAT	acetyl-CoA acetyltransferase	TRINITY_DN117046	TRINITY_DN81271	Dysnectes_Gene.3562+ Gene.18206	GL50803_3287	GMRT_15802	—	[26]
phospholipids	pgsA	cardiolipin synthase	TRINITY_DN114306	TRINITY_DN121187	Dysnectes_Gene.4476	GL50803_7259	GMRT_14375	SS50377_15571	[26]
phospholipids	pssA	CDP-DAG:serine O-phosphatidyl-transferase	—	—	Dysnectes_Gene.7470	GL50803_17427	GMRT_12707	SS50377_11355	[26]
lipid metabolism	peroxi/mito LCCL	long-chain CoA ligase	TRINITY_DN111751	TRINITY_DN79286	Dysnectes_Gene.2121	GL50803_9062, GL50803_30476a	GMRT_11146	SS50377_12250, SS50377_17594	[26]
lipid metabolism	mito LCCL	long-chain CoA ligase	TRINITY_DN114893, TRINITY_DN113703, TRINITY_DN114303, TRINITY_DN5467	TRINITY_DN60450, TRINITY_DN46300, TRINITY_DN79110	Dysnectes_Gene.4299, Dysnectes_Gene.4566, Dysnectes_Gene.9801, Dysnectes_Gene.13650	GL50803_21118, GL50803_113892	GMRT_15304, GMRT_15553	SS50377_14794, SS50377_16242, SS50377_17330, SS50377_17341	[26]

Not reported as mitosomal, but phylogenetically related to a mitoprotein.

Retortamonad MROs could be responsible for hydrogenase activity, iron‑sulfur (Fe-S) cluster assembly and one‑carbon pool recycling. One or both HYDA homologs we found in each retortamonad could be MRO-localized, based on our recovery of hydrogenase maturases E, F and G (Fig. 3, Table 2, Supplementary Table S1). These proteins have N-terminal extensions comparable to those of their MRO-localized homologs from Spironucleus salmonicida. Consistently, the expression of the ISC type Fe-S assembly complex suggests that the mitochondrion of retortamonads has the capacity to form and deliver Fe-S clusters, with hydrogenase being one of the recipients (Fig. 2C). Though we could not find the mitochondrial ATM1 Fe-S cluster export protein, Giardia and all other metamonads also lack this protein and must exchange Fe-S clusters otherwise [24].

Fig. 3

Phylogeny of retortamonad hydrogenase subunit A, giardins and serine hydroxymethyltransferase. Sequences from R. dobelli and R. caviae are highlighted in white font. Sequences with mitochondrial localization predicted by at least three algorithms (see Material and Methods) are in blue. Support values of the ultra-fast bootstrap are shown where >85. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

One‑carbon pool uses 5,10-methylenetetrahydrofolate (mTHF) as a cofactor to transfer methyl groups. To recharge mTHF with a methylene group, eukaryotes metabolize Ser by SHMT (yielding Gly) and subsequently Gly by the glycine cleavage system (GCS). The accumulation and use of THF in retortamonads is corroborated by the presence of not only SHMT (Figs. 2A and 3) but also folate-polyglutamate synthase folC that is involved in the formation and cellular retention of THF [29]. However, none of the other conventional enzymes that utilize mTHF (namely GCS, thymidylate synthase, and methionine synthase metH) have been found in the transcriptomes of R. dobelli and R. caviae (Supplementary Table S1), obscuring the biological function of this cofactor. Folate-independent activity has also been reported for SHMT with Ser and Thr [30], and while the absence of threonine aldolase in diplomonads could rationalize the presence of SHMT when GCS is lacking, folC would be functionally redundant under this scenario.

Retortamonads encode some of the diplomonad virulence factors

Vertebrate retortamonads represent the closest known relatives to diplomonads, a taxon that includes infamous human and animal parasites of the genera Giardia and Spironucleus. To find possible pathogenic adaptations in R. caviae and R. dobelli, we searched for Giardia virulence factors triggered in the parasites by host recognition and attachment [31,32]. Some of these factors have a general function in protein translation and homeostasis (Supplementary Table S1), or in oxidative stress response – such as rubredoxin‑oxygen oxidoreductase. All these factors are expressed by retortamonads as well, including secretory cysteine proteases related to those from Giardia that partake host tissue disruption and immune response modulation (Supplementary Fig. S3) [32]. Many of these virulence factors could be considered pre-adaptations for parasitism, as they are widely present in free-living fornicates as well (Fig. 4). As another example, we found evidence for bactericidal/permeability-increasing proteins encoded by retortamonads that could be involved in coping with gram-negative bacteria or effecting mutualistic relationships (Supplementary Fig. S3) [33,34]. We also found evidence for NEK kinases that, although not as diversified as in diplomonads, probably underlie central biological processes not only in retortamonads but also Carpediemonas-like organisms (CLOs) in general (Fig. 4, Supplementary Fig. S3) [35]. In contrast, we failed to find any retortamonad homologs of Giardia variant-specific surface proteins (VSPs) and growth factor receptors, suggesting these might have evolved specifically in diplomonads as adaptation to parasitism. Alternatively, retortamonad VSPs could be expressed specifically inside a host and not in culture.

Fig. 4

Presence and absence of marker proteins from amino sugar pathways (encystation), Golgi apparatus and pathogenesis-related processes in Giardia. For amino sugar metabolism, the listed markers include enzymes of amino sugar synthesis (GNPAT – glucosamine-phosphate acetyltransferase; PAGM – phosphoacetylglucosamine mutase; UAP – UDP-N-acetylglucosamine pyrophosphorylase) and catabolism (HEX – hexosaminidase; NAGK – N-acetylglucosamine kinase). For Golgi, the markers represent beta-coatomer adaptin B-COP and two vacuolar protein sorting-associated proteins Vps26 and Vps35. For pathogenesis-related virulence factors, giardins, bactericidal/permeability-increasing proteins (BPI), NEK kinases, secretory cysteine proteases (SCP), variant-specific surface proteins (VSP) and lysozymes are listed. * – among many homologs, note that the presence of only one NEK clade is shown, branching close to human Nek2 involved in cell division. “micro” – microscopic evidence. Cysts were reported for several Spironucleus species, but not specifically for Spironucleus salmonicida (“sp.”), for which genomic sequence data are available [15].

In support of a predatory nature of vertebrate retortamonads, we identified an expansion in a lysozyme family specific to R. caviae and R. dobelli. This family is most closely related to glycosyl hydrolase family 25 and counting factors of Dictyostelium discoideum (Supplementary Fig. S3). Lysozymes are ubiquitous enzymes utilized to target bacterial cell wall, and interestingly, while widespread among free-living Fornicata, we found no evidence for lysozymes in diplomonads (Fig. 4). Unlike their ancestors, therefore, diplomonads apparently gave up on their peptidoglycan lytic activity during their transformation to parasitism.

Also worth noting are enzymes of the amino sugar metabolism, which is activated in Giardia during encystation [36]. These enzymes (i.e., glucosamine-phosphate N-acetyltransferase, N-acetylglucosamine kinase, phosphoacetylglucosamine mutase, UDP-N-acetylglucosamine diphosphorylase, hexosaminidase, and glucosamine-6-phosphate deaminase; Supplementary Table S1) were also present in retortamonad transcriptomes, consistent with the appearance of encysted cells in the culture. This probably means that the cyst wall is built by similar mechanisms in both retortamonads and Giardia (and possibly other fornicates; Fig. 4) and thus represents a conserved biological process to survive adverse conditions and spread among hosts. Surprisingly, cysts are infrequently observed among CLOs, so far documented in Hicanonectes telescopos and Iotanema spirale [37,38]. Another example of a biological process that is not directly obvious from ultrastructural observations is the presence of a Golgi apparatus. This has a form of dispersed vesicles in Giardia trophozoites and only develops into conventional complexes during encystation [39]. Based on the presence of Golgi-specific markers (adaptin B-COP and vacuolar sorting-associated proteins Vps26 and Vps35) [40] in our transcriptomic data, we hypothesize that there is a functionally transient Golgi apparatus also in retortamonads (Fig. 4, Supplementary Table S1), probably linked to encystation. This feature is potentially conserved in all CLOs, despite only having been observed in Carpediemonas membranifera [41]. Lastly, we were able to identify homologs of beta- and delta-giardins, which are integral components of Giardia's remarkable ventral disk [39,42]. These structural proteins contain striated fibre assemblin domains. They have only been known from diplomonads, but our phylogenetic analysis strongly supports a broader taxonomic distribution among Carpediemonas-like organisms (Fig. 3), including retortamonads. This suggests an alternative ancestral role for these proteins, possibly in building comb-like structures seen in insect retortamonads (see e.g. [1]). Our data thus highlight several biological processes and structures that have not been detailed in retortamonads, some of which might be considered pre-adaptations for parasitism in diplomonads.

Discussion

The putative presence of [FeFe]‑hydrogenase with maturases in the mitochondrion of retortamonads would be consistent with their classification as “hydrogenosomes”, only without ATP production in the organelle. This metabolic set-up was also hypothesized for Dysnectes brevis and the ancestor of diplomonads by Leger et al. [8], and hence retortamonads typify a missing step in the reductive evolution of MROs in Fornicata. With this set-up, retortamonads likely use hydrogenases to sink superfluous electrons from yet unclear metabolic reactions, which can be problematic in anoxic environments [43,44]. One question that remains to be answered are the roles of SHMT, folC and one‑carbon pool in organisms that probably lack GCS, metH and thymidylate synthase, i.e. enzymes known to utilize mTHF. This seems to be the case also for the fish parasite Spironucleus salmonicida, that too has a putatively MRO-localized SHMT (SS50377_17865) [8,15], and potentially other anaerobes [7]. We could be missing mTHF-utilizing enzymes due to their low expression or sequence divergence, and we cannot exclude that there might be alternative enzymatic reactions that require this cofactor. Another viable explanation is that mTHF is in fact not utilized, and SHMT and folC have a THF-independent function, as illustrated by the threonine aldolase activity of SHMT [30].

Here, we also investigated how retortamonads gain and distribute energy and nutrients. In this respect, they are reminiscent of their parasitic diplomonad relatives – retortamonads apparently rely on substrate-level phosphorylation via both glycolysis and arginine catabolism to generate ATP, while recycling and incorporating other amino acids, fatty acids, and nucleotides into their primary metabolism. To our knowledge, this is among the first general accounts on core metabolism in members of Fornicata outside Diplomonada, although similar reconstructions have been published for other anaerobic protists [6,[45], [46], [47], [48], [49], [50]].

The striking similarity of the core metabolism in retortamonads with that of diplomonads led us to inspect whether there are parasite-like adaptations in R. caviae and R. dobelli. Indeed, the pathogenicity of retortamonads for humans is marginal, even doubtful (reviewed in [1]), consistent with their lack of VSPs that confer antigenic variation in Giardia to avoid host immune reaction [51]. Considering the differences in the microbial communities of the natural environment of R. caviae and R. dobelli (intestine of a herbivorous rodent versus a carnivorous frog), it might seem surprising that their metabolisms have not diverged substantially. However, metamonads have been found already quite similar in terms of orthologous protein group content [50]. It is clear that inhabiting a nutrient- and prey-rich milieu could have enabled many metamonad anaerobes to survive with an efficient, minimal metabolism and that retortamonads as vertebrate commensals possess few specific adaptations towards their host. This ancestral efficiency might have been the reason why some diplomonads (such as Trepomonas and Hexamita spp.) reverted to free-living lifestyle by regaining crucial genes through lateral gene transfer [22]. It remains to be determined if retortamonads form a mutualistic, metabolic interaction with any of the bacteria from their environment. Exchange of nutrients or essential cofactors emerges as an important phenomenon in various environmental settings [52] and strongly influences species composition [53]. For retortamonads, it could be beneficial to trade some of their end metabolites (i.e. acetate or putrescine), although we cannot exclude that they are strictly predatory, given their strong complement of lysozymes.

Conclusions

Retortamonads Retortamonas dobelli and R. caviae currently represent the closest sister lineage to Diplomonada with in-depth transcriptomic data available, making them important to better understand the evolutionary trajectory of this clade. Retortamonads exhibit a streamlined metabolic architecture much alike that of diplomonads and apparently lack virulence factors to cope with vertebrate immune systems, corroborating their commensalism. Due to their lesser divergence, retortamonads might be instrumental in linking molecular functions in diplomonads to homologous processes in other eukaryotes.

Material and methods

Cultivation

Retortamonads studied here, R. dobelli strain CERAT3RET isolated from the frog Ceratophrys ornata and R. caviae strain BELLA isolated from guinea pig Cavia porcellus, were obtained from our culture collection and maintained in Dobell-Laidlaw diphasic serum-Ringer-egg media [54] at 27 °C, being subcultured weekly. We did not see R. caviae grow significantly better at 37 °C.

Nucleic acid extraction, library preparation and sequencing

Total RNA was isolated from 200 μL culture suspension using TRI Reagent (Sigma-Aldrich) according to the manufacturer's protocol, followed by Dynabeads Oligo(dT) (Thermo Fisher Scientific, Waltham, MA, USA) mRNA enrichment. Sequencing libraries were prepared by BIOCEV's Core Sequencing Facility in Vestec, Czech Republic, using NEBNext Ultra II Directional RNA Library Prep (New England Biolabs, Ipswich, MA, USA), which included another step of mRNA selection. Final library quality control and 2 × 100 bp sequencing were performed by Macrogen Europe (Amsterdam, the Netherlands) on an Illumina HiSeq4000 sequencer at 5 Gbp per-sample depth.

Transcriptome assembly and decontamination

The obtained reads were adapter- and quality-trimmed using Trimmomatic v0.36 [55] and bbduk (part of the BBTools suite; jgi.doe.gov/data-and-tools/bbtools/). Cleaned reads were assembled using Trinity v2.6.5 [56] with default parameters and rnaSPAdes v3.13.0 [57] with k-mer lengths up to 81 (-k 21,33,55,77,81). From these assemblies we removed obvious bacterial and eukaryotic contaminants, i.e. contigs with more than 80% identity and at least 50% query coverage in nucleotide space to any NCBI nt sequence with bacterial, plant, fungal, or animal provenance were discarded along with reads mapping to these query contigs. Next, protein-coding sequences were predicted with TransDecoder v5.5.0 (github.com/TransDecoder), and another decontamination took place in protein space, again at 80% identity and 50% coverage level, with minor manual curation during the downstream analyses. Lateral gene transfer was not considered unless retortamonads constituted a single clade with no close bacterial hit.

The difference in transcriptome sizes was partly caused by one-sided sample cross-contamination (R. dobelli reads in R. caviae data), the consequence of which was largely mitigated by removal of sequences from the R. caviae dataset that showed >99% identity to ones from the R. dobelli dataset, and manual inspection of alignments.

Phylogenetic analyses and inference of localization

We performed a phylogeny-informed bioinformatic analysis of predicted proteomes of both retortamonad species to retrieve homologs of selected proteins previously identified in diplomonads and other fornicates. NCBI nr/nt databases and EukProt [58], as well as an in-house database compiled from transcriptomic and genomic data of Metamonada representatives [8,21] were used to build single-gene datasets of these proteins, and their homologs from our retortamonad data were identified by BLAST and, where increased sensitivity was necessary, HMMER. ABC and MFS transporter datasets were enriched by sequences from curated databases UniProt and TCDB [59]. Single-gene datasets were aligned by MAFFT v7.450 [60] using the automatic refinement and trimmed to remove sites with >50% gaps. Maximum likelihood (ML) tree search was performed in PhyML v2.2.4 [61] under the LG model with no bootstrapping. Besides, functional domains were identified using the InterProScan v1.1.4 online search of Geneious v2020.1.2, last accessed in May 2020 [62,63] in both the query sequences and the retortamonad hits and compared.

For more comprehensive phylogenies, datasets were aligned by MAFFT using the L-INS-i refinement and a maximum of 1000 iterations, followed by trimAl v1.4 [64] trimming at sites with >70% gaps (−gt 0.3). ML trees were inferred by IQ-TREE v 1.6.12 [65] using the LG+F+G model and the ultrafast bootstrapping strategy (1000 replicates).

Mitochondrial targeting was predicted using algorithms that evaluate sequence N-termini (MitoFates, TargetP, MultiLoc2, NOMMPred, PProwler and Cello; [61,[66], [67], [68], [69], [70]]).

Funding

This project has received funding from the (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. 771592) and the Centre for research of pathogenicity and virulence of parasites (registration no. CZ.02.1.01/0.0/0.0/16_019/0000759). Computational resources were supplied by the project “e-Infrastruktura CZ” (e-INFRA LM2018140) provided within the program Projects of Large Research, Development and Innovations Infrastructures. TP was supported by the (UNCE 204069) and IČ was supported by the (18-18699S).

Data availability

Reads and transcriptomic assemblies were deposited in GenBank under BioProject PRJNA659466. Phylogenies of putative retortamonad mitochondrial proteins are available in the Supplementary Table S2.

Authors’ contributions

Conceptualization VH; Data curation ZF, SCT; Funding acquisition VH, PD, IČ, TP; Investigation ZF, MV, PS, TP; Methodology ZF, SCT, MV; Project administration VH; Supervision VH, ZF, IČ, PD; Validation ZF, MV; Visualization ZF; Writing – original draft ZF; Writing – review & editing ZF, TP, SCT, VH, IČ, PD. All authors read and approved the manuscript.

Declaration of Competing Interest

None.