An Evolutionary-Focused Review of the Holosporales (Alphaproteobacteria): Diversity, Host Interactions, and Taxonomic Re-ranking as Holosporineae Subord. Nov

Abstract

The order Holosporales is a broad and ancient lineage of bacteria obligatorily associated with eukaryotic hosts, mostly protists. Significantly, this is similar to other evolutionary distinct bacterial lineages (e.g. Rickettsiales and Chlamydiae). Here, we provide a detailed and comprehensive account on the current knowledge on the Holosporales. First, acknowledging the up-to-date phylogenetic reconstructions and recent nomenclatural proposals, we reevaluate their taxonomy, thus re-ranking them as a suborder, i.e. Holosporineae, within the order Rhodospirillales. Then, we examine the phylogenetic diversity of the Holosporineae, presenting the 20 described genera and many yet undescribed sub-lineages, as well as the variety of the respective environments of provenance and hosts, which belong to several different eukaryotic supergroups. Noteworthy representatives of the Holosporineae are the infectious intranuclear Holospora, the host manipulator ‘Caedimonas’, and the farmed shrimp pathogen ‘Candidatus Hepatobacter’. Next, we put these bacteria in the broad context of the whole Holosporineae, by comparing with the available data on the least studied representatives, including genome sequences. Accordingly, we reason on the most probable evolutionary trajectories for host interactions, host specificity, and emergence of potential pathogens in aquaculture and possibly humans, as well as on future research directions to investigate those many open points on the Holosporineae.

Overview and Purposes

The Holosporales are an alphaproteobacterial order that was taxonomically described in the last decade [1], but research on their members has a long-lasting tradition in microbiology, in particular on two noteworthy representatives, namely the eponym Holospora and ‘Caedimonas’, both living intracellularly in ciliate protists. Holospora spp. are highly infectious bacteria inhabiting the nucleus of Paramecium hosts [2, 3], and their description dates back to the seminal work by Mardukhey Wolf-Vladimir Hafkin in the late nineteenth century [4]. ‘Caedimonas varicaedens’ confers its Paramecium hosts the so-called ‘killed trait’, similarly to the gammaproteobacterium Caedibacter taeniospiralis, and was indeed ascribed to the Caedibacter genus before molecular phylogenies showed their unrelatedness [5–7]. Long before understanding that the causative agents were actually bacteria, the killer trait was identified as early as 1938 in the pioneering investigations by Tracy Sonneborn [8].

Over the last three decades, extensive molecular surveys of host-associated bacteria showed that Holospora and ‘Caedimonas’ are phylogenetically close and form a conspicuous lineage together with several other bacteria [5, 9–12]. Currently, this lineage is formally described at the order rank (Holosporales) (https://lpsn.dsmz.de/order/holosporales) following Szokoli and co-authors [1]. As motivated in details below, here we will propose to rank it as a suborder (i.e. Holosporineae). Until the dedicated section, here, we will use taxonomic terms according to Szokoli and co-authors unless otherwise specified.

The Holosporales are likely very ancient (over 1 billion years) [13], and their extant members characterised so far are solely host-associated, with the vast majority of such hosts being unicellular eukaryotes, i.e. protists [14]. The potential host association status is unknown for additional uncharacterised members of the Holosporales, known only by their DNA sequences obtained in environmental screening and/or metagenomic studies (e.g. [15–27]), in several cases with more or less strong indications of association to eukaryotic hosts, e.g. [28–37]. Accordingly, it seems probable that the association with protists (and other eukaryotes) dates back to the last common ancestor of the Holosporales.

Their evolutionary antiquity, diversity breadth, and ancient host-association make the Holosporales a noteworthy subject of study and research, bringing them together with other well-known bacterial lineages with ancient host-association, such as Rickettsiales, Chlamydiae, and Legionellales [38–40]. The common features of those lineages (including the Holosporales) led some authors to define them as ‘professional symbionts’, underlining their ability to adapt to varied hosts and in some way presumably take control of the interactions [41, 42]. Therefore, the study of Holosporales is particularly relevant for comparisons with other professional symbionts, in order to highlight evolutionary convergent or lineage-specific traits. For example, the evolutionary origin and possible conservation of functional features of the killer trait of ‘Caedimonas’ has a broad significance, as it may be seen as a host addictive manipulation phenomenon [43]. Last but not least, differently from other professional symbiont lineages [44–46], the Holosporales do not encompass renowned human pathogens, but they do include the members of the genus ‘Candidatus Hepatobacter’ [10, 47, 48], which affect farmed shrimps and other crustaceans, causing necrotising hepatopancreatitis (NHP), a disease recognised by the World Organization for Animal Health [49].

The research interest on the Holosporales is shown by numerous research papers produced in recent years (e.g. [14, 34, 50–58]). However, to our best knowledge, there is not yet a dedicated review piece focused on the Holosporales as a whole and on their evolutionary and functional features, possibly due to their relatively recent recognition as a high-rank and independent lineage [1, 59]. Here, we aim to fill this gap, dealing with the following subjects:

• Taxonomical overview and proposed revisions

• Diversity of representatives and the respective hosts

• Evolution of the lineage with a focus on the interactions with the hosts, informed by genomics

The Taxonomy of the Holosporales: Historical Overview and Proposed Revisions

The phylogenetic reconstructions of the Holosporales and the consequent taxonomic views have changed multiple times across recent years. Although not all taxonomic proposals were formally validated, this inevitably led to some incongruence among different studies published within short timings from one to another, each referring to a different taxonomical version, with a substantial risk to mislead non-specialists. Here, we will review in chronological order the main changes in the taxonomy of Holosporales, showing in particular the current ambiguities and the reasons that led us to propose here a novel revision that, in our view, should settle the issue for long.

The taxonomic affiliation of Holospora was initially uncertain [60]. Early molecular phylogenies based on 16S rRNA gene sequences indicated its relatedness to the order Rickettsiales [61], forming a distinct early-diverging lineage together with other bacteria associated with ciliates and other protists (including ‘Caedimonas’) [5, 9, 11, 62, 63], as well as with metazoans [10]. Accordingly, Görtz and Schmidt created a novel family for these bacteria within the Rickettsiales, namely the Holosporaceae [64], including also few other symbionts of ciliates that were not yet characterised molecularly. Curiously, molecular phylogenies later disproved the affiliation to the Holosporaceae of some of the latter, namely Lyticum and Pseudolyticum, both actually belonging to the ‘Candidatus Midichloriaceae’ (Rickettsiales) [65, 66], while molecular data are still lacking for others (Pseudocaedibacter and Tectibacter) [67, 68], making their actual affiliation uncertain to date.

Later on, based on the strong synapomorphies in the morphology and life cycle among Holospora and Holospora-like bacteria (HLB) [3, 4], Boscaro and colleagues informally proposed to restrict the Holosporaceae only to those bacteria, while treating all other Holosporaceae sensu Görtz and Schmidt as incertae sedis within the Rickettsiales [69].

In the same years, the phylogeny and taxonomy of these bacteria were influenced by multiple studies. On one hand, multiple other members of the Holosporaceae sensu Görtz and Schmidt were identified, e.g. [29, 70–75], with Hess and colleagues formally proposing to elevate at the family rank a sublineage including multiple symbionts of amoebas, namely the ‘Candidatus Paracaedibacteraceae’ [75]. Moreover, seminal phylogenetic studies with extended molecular markers [76–78] cast significant doubt on the actual affiliation of the Holosporaceae sensu Görtz and Schmidt to the Rickettsiales, which had relied on the 16S rRNA gene. Specifically, Ferla and colleagues made a first informal proposal to elevate them at the order rank [77].

Accounting for the above findings, Szokoli and colleagues formally re-organised the taxonomy and systematics of these bacteria [1], re-describing the Holosporales as an order, with four families, which is the taxonomy formally accepted to date (https://lpsn.dsmz.de/order/holosporales). These families are the ‘Candidatus Paracaedibacteraceae’ sensu Hess and co-authors, the Caedibacter/Nucleicultrix clade (soon after described as ‘Caedimonadaceae’ by Schrallhammer and colleagues, jointly with the description of ‘Caedimonas’ as a distinct genus from Caedibacter [7]), the Holosporaceae (revised in order to include HLB and their close relatives, thus narrower than sensu Görtz and Schmidt, but broader than the proposal by Boscaro and colleagues), and the newly described ‘Candidatus Hepatincolaceae’. The latter encompass mostly arthropod-associated bacteria [1, 79, 80] but have been recently shown to be phylogenetically unrelated to the Holosporales [81]. Below, we will propose a systematic revision for them as well.

Soon afterwards, the phylogenetic placement of the Holosporales was reassessed by Muñoz-Gómez and co-authors, who provided substantial evidence that the placement within Rickettsiales was an artefact due to compositional biases and that the Holosporales are actually related to, and possibly nested within, another broad and ancient alphaproteobacterial order, namely the Rhodospirillales [59]. Accordingly, they proposed to down-rank the Holosporales at the family level (to be named once again Holosporaceae) within the Rhodospirillales and to down-rank their families recognised at the time as subfamilies (i.e. Holosporodeae, ‘Candidatus Paracaedibacteriodeae’, and ‘Candidatus Hepatincolodeae’).

While the taxonomic revision proposed by Muñoz-Gómez and co-authors was not formally validated, it was adopted by a number of successive studies, e.g. [54, 82], in parallel to the version by Szokoli and co-authors in others, e.g. [55–57]. The presence of these two alternative versions is quite unfortunate, especially when considering that both use the term Holosporaceae with different meanings.

This taxonomic scenario is further complicated by the revisions recently proposed by Chuvochina and co-authors based on the Genome Taxonomy Database (GTDB) [83]. These consist in subdividing the members of the Holosporales into three orders roughly corresponding to the three families accounted by Szokoli and co-authors, namely Holosporales, ‘Candidatus Paracaedibacterales’, and ‘Caedimonadales’, and in proposing two additional families, namely the ‘Candidatus Hepatobacteraceae’ (Holosporales) and ‘Candidatus Nucleicultricaceae’ (‘Caedimonadales’). Those novel proposals are currently treated as non-standing heterotypic synonyms (https://lpsn.dsmz.de/). The discrepancies with respect to other classifications are partly due to the fact that the inference of the underlying phylogenetic backbone did not counteract the known artefacts evidenced by Muñoz-Gómez and co-authors, thus resulting in incorrect placing and splitting of the Holosporales (sensu Szokoli et al.). However, they are also due to different and more stringent thresholds used to delineate taxonomic ranks. It seems worth to consider that, given such thresholds, Chuvochina and co-authors also proposed to elevate at the order rank several lineages that are affiliated to the Rhodospirillales according to the currently validated taxonomy [84], namely Acetobacterales, Azospirillales, Geminicoccales, Oceanibaculales, Reyranellales, Thalassobaculales, and Tistrellales [83].

We believe that the current taxonomic ambiguities on the Holosporales should be settled by an evolutionarily and biologically meaningful proposal that meets the following criteria: (i) consistency with most up-to-date phylogenetic reconstructions, (ii) consistency and compatibility with most up-to-date (and ideally also foreseen) taxonomic classification schemes of Alphaproteobacteria, (iii) compliance to the formal standards for validation. The most credible phylogenetic backbone for the Holosporales is the one obtained by Muñoz-Gómez and co-authors, namely phylogenetically nested within the Rhodospirillales sensu Hördt et al., as confirmed by successive studies [40, 85]. While the taxonomic proposal by Muñoz-Gómez and co-authors accounts for this phylogeny, it also implies the down-ranking of the families of Holosporales as subfamilies, which does not account for the great phylogenetic diversity of those, the same that led Chuvochina and co-authors to rank them as separate orders. We believe that the most suitable solution is intermediate between those extremes. Accordingly, we formally propose to down-rank the Holosporales sensu Szokoli et al. (corresponding to the Holosporaceae sensu Gortz and Schmidt and sensu Muñoz-Gómez et al.) as a suborder within the Rhodospirillales sensu Hördt et al., namely Holosporineae. Accordingly, the rank and composition of the included families should be kept unchanged as per Szokoli et al. This novel taxonomic proposal has the additional advantage that it would suit well to potential future revisions of high-order taxonomy of the Alphaproteobacteria, including subdividing the Rhodospirillales into multiple orders, should a ‘splitter’ view similar to the one by Chuvochina and co-authors eventually prevail. Indeed, in such a case, the Holosporineae could be meaningfully re-elevated at the order rank, but, differently from other previous proposals, without the need of further changes in their inner subdivisions, thus simplifying revisions and favouring nomenclatural consistency over time. From now on, here, we will refer to this lineage as Holosporineae. Moreover, accounting for the most recent phylogenetic evidence of an independent branching of the ‘Candidatus Hepatincolaceae’ and Holosporineae within the Rhodospirillales [81], here, we formally propose to move this family out of the Holosporineae. Those taxonomic revisions are presented at the end of the text.

Diversity of the Holosporineae and Their Hosts

All the characterised Holosporineae (i.e. Holosporales sensu Szokoli et al.) were consistently retrieved as intracellular bacteria in multiple eukaryotic hosts, predominantly protists [1, 3, 7, 14, 41, 54, 57, 59, 75]. Therefore, here, we will assume that all the Holosporineae are host-associated and will treat those representatives sequenced in environmental screening studies as putatively associated with unknown hosts. Accordingly, their environmental provenance would actually reflect the one of the respective hosts, notwithstanding the possibility of detecting temporarily free transmission forms of the bacteria, such as the infectious forms of Holospora [2, 3]. Among the Holosporineae putatively associated with Metazoa, only the members of the ‘Candidatus Hepatobacter’ genus were investigated in detail and confirmed as intracellular in the host cells [10, 47, 48]. On the other hand, all the others were sequenced from samples coming from the host gut [28, 32, 36, 86–90] or skin [91–94]. In those cases, a hypothetical intracellular association with the target host is still to be verified, while an alternative association with microbial eukaryotes, possibly ingested by the animal or part of its stable gut community, seems also plausible [95].

In this section, we will review the phylogenetic diversity of the Holosporineae, with a focus on experimentally characterised and taxonomically described species over specimens known only from metagenomic screenings. The Holosporineae encompass three families, namely Holosporaceae, ‘Caedimonadaceae’, and ‘Candidatus Paracaedibacteraceae’ (Fig. 1). Most phylogenetic inference studies, both on the 16S rRNA gene or by phylogenomics, indicate that Holosporaceae and ‘Caedimonadaceae’ are sister groups [1, 14, 50, 54, 58, 69, 74, 75, 78, 95].

Fig. 1.

Cladogram showing the phylogenetic relationships among the Holosporineae, based on a combination of published reference studies, with inconsistent/unresolved relationships represented as polytomies [1, 14, 54, 57, 58, 75]. For presentation purposes, the branches leading to the ‘fast-evolving Holosporaceae’ are shown longer than the others. For each organism/clade, the most typical hosts are shown on the right-hand side, with name labels on the first occurrence of each host (see main text for further details). Branches without text labels indicate unnamed bacteria found through environmental screening studies in association with the shown putative hosts. Dotted branches indicate the grouping, for space constraints, of organisms that descend from the previous node in the tree, but with unconfirmed reciprocal phylogenetic relationships. The three families of the Holosporineae are shown by coloured backgrounds, and two subgroups of the Holosporaceae are encircled by dashed rectangular shapes. “Ca.” is an abbreviation for ‘Candidatus’, while ‘HLB’ for ‘Holospora-like bacteria’

The Holosporaceae are mostly associated with ciliates and other protists [96] but also with arthropods, and encompass 13 described genera (Table 1). Among them, the monophyletic lineage made up of Holospora and the HLB (three other described genera, namely ‘Candidatus Goertzia’, ‘Preeria’, and ‘Candidatus Hafkinia’) is prominent (highlighted in Fig. 1) [3, 69, 97, 98]. These bacteria are associated with ciliates, prevalently of the genus Paramecium, and are characterised by distinctive apomorphies, including morpho-functional differentiations linked with an infectious lifecycle that involves invasion of the host nuclei [2]. Such peculiarities allowed Hafkin to notice these bacteria for the first time over 100 years ago [4] and consented a reliable assignment of new specimens to this lineage based on microscopy observations [60, 68]. Eventually, molecular data revealed the phylogenetic breadth of this genus, with five species accordingly recognised to date, namely H. obtusa, H. undulata, H. curviuscula, H. acuminata, and ‘Candidatus Holospora parva’, hosted by Paramecium caudatum, P. bursaria, P. aurelia, and P. chlorelligerum (Table 1) [61, 69, 99–101]. Molecular phylogenies are not always fully consistent with morphology, leading to a recent revision of H. undulata to encompass also the former species H. elegans and H. recta [102] and to the re-classification of the former H. caryophila into a novel genus as ‘Preeria caryophila’, being more distant from Holospora than the other HLB genera (Fig. 1) [97]. On the other hand, ‘Candidatus Hafkinia simulans’ is the closest relative of Holospora spp. but is hosted by the brackish water peniculid ciliate Frontonia salmastra rather than by its close relative Paramecium [98]. ‘Candidatus Goertzia’ includes three species, namely ‘Candidatus Goertzia infectiva’, ‘Candidatus Goertzia shahrazadae’, and ‘Candidatus Goertzia yakutica’, respectively hosted by P. jenningsi, P. multimicronucleatum, and P. putrinum (Table 1) [69, 103, 104]. Further, Holospora/HLB are possibly awaiting to be fully characterised in association with other ciliates (e.g. Trithigmastoma cucullulus and Prorodon teres) [4]. Most of Holospora and HLB species were found infecting a single host species [4], but this species-specificity is not always so sharp (e.g. ‘Preeria’ found in species of P. aurelia complex as well as in P. caudatum) (Table 1) [97], consistent with the incongruence between bacterial and host phylogenies [96, 98]. Uncharacterised Holospora-related bacteria were also detected in the gut of the freshwater fish Panaque nigrolineatus [105] and in biofilm [106] while ‘Candidatus Goertzia’-like ones in the microbiome of cladocerans [35].

Table 1.

List of described and molecularly validated Holosporineae species

Species	Family	Host(s)	Location(s)	Genome available	Reference
Holospora undulata	Holosporaceae	Paramecium caudatum	Intranuclear and infectious	Yes	[60]
Holospora obtusa	Holosporaceae	Paramecium caudatum	Intranuclear and infectious	Yes	[60]
‘Holospora curviuscula’	Holosporaceae	Paramecium bursaria	Intranuclear and infectious	Yes	[307]
‘Holospora acuminata’	Holosporaceae	Paramecium bursaria	Intranuclear and infectious	No	[308]
‘Candidatus Holospora parva’	Holosporaceae	Paramecium chlorelligerum	Intranuclear and infectious	No	[100]
‘Candidatus Goertzia infectiva’	Holosporaceae	Paramecium jenningsi	Intranuclear and infectious	No	[69]
‘Candidatus Goertzia shahrazadae’	Holosporaceae	Paramecium multimicronucleatum	Intranuclear and infectious, occasionally cytoplasmic	No	[103]
‘Candidatus Goertzia yakutica’	Holosporaceae	Paramecium putrinum	Intranuclear and infectious	No	[104]
‘Candidatus Hafkinia simulans’	Holosporaceae	Frontonia salmastra	Intranuclear and infectious	No	[98]
‘Preeria caryophila’	Holosporaceae	Paramecium biaurelia, Paramecium octaurelia, Paramecium novaurelia, Paramecium caudatum, Paramecium sp.	Intranuclear and infectious	No	[97]
‘Candidatus Fujishimia apicalis’	Holosporaceae	Euplotes octocarinatus	Cytoplasm, mostly apical	No	[52]
‘Candidatus Hydrogenosomobacter endosymbioticus’	Holosporaceae	Cyclidium-like scuticociliate	Cytoplasm, close to hydrogenomes	Yes	[53]
‘Candidatus Paraholospora nucleivisitans’	Holosporaceae	Paramecium sexaurelia	Cytoplasm and nucleus	No	[71]
‘Candidatus Mystax nordicus’	Holosporaceae	Paramecium nephridiatum	Cytoplasm, sometimes aggregating with mitochondria	No	[56]
‘Candidatus Gromoviella agglomerans’	Holosporaceae	Paramecium polycaryum	Cytoplasm, sometimes forming aggregates	Yes	[58]
‘Candidatus Cytomitobacter primus’	Holosporaceae	Diplonema japonicum	Cytoplasm, occasionally possibly inside mitochondria	Yes	[50]
‘Candidatus Cytomitobacter indipagum’	Holosporaceae	Diplonema aggregans	Cytoplasm, occasionally possibly inside mitochondria	Yes	[50]
‘Candidatus Cytomitobacter rhynchopi’	Holosporaceae	Rhynchopus asiaticus	Cytoplasm	No	[108]
‘Candidatus Ignotibacter abundans’	Holosporaceae	Diplonema aggregans	Cytoplasm	Yes	[54]
‘Candidatus Hepatobacter penaei’	Holosporaceae	Litopenaeus vannamei and other crustaceans	Cytoplasm	Yes	[47]
‘Candidatus Hepatobacter paralithodis’	Holosporaceae	Paralithodes platypus	Cytoplasm	No	[48]
‘Candidatus Bealeia paramacronuclearis’	Holosporaceae	Paramecium biaurelia	Cytoplasm, in proximity of the macronucleus	Yes	[1]
‘Caedimonas varicaedens’	‘Caedimonadaceae’	Paramecium biaurelia, Paramecium novaurelia, Paramecium caudatum, Paramecium duboscqui, Spirostomum ambiguum, Euplotes sp., Peridinium cinctum	Cytoplasm or nucleus, depending on the host species	Yes	[7]
‘Candidatus Paracaedimonas acanthamoebae’	‘Caedimonadaceae’	Acanthamoeba sp.	Cytoplasm	Yes	[7]
‘Candidatus Nucleicultrix amoebiphila’	‘Caedimonadaceae’	Hartmanella sp.	Intranuclear and infectious	Yes	[74]
‘Candidatus Paracaedibacter symbiosus’	‘Candidatus Paracaedibacteraceae’	Acanthamoeba sp.	Cytoplasm	Yes	[11]
‘Candidatus Odyssella thessalonicensis’	‘Candidatus Paracaedibacteraceae’	Acanthamoeba sp.	Cytoplasm	Yes	[62]
‘Candidatus Odyssella acanthamoebae’	‘Candidatus Paracaedibacteraceae’	Acanthamoeba sp.	Cytoplasm	Yes	[11]
‘Candidatus Finniella lucida’	‘Candidatus Paracaedibacteraceae’	Orciraptor agilis	Cytoplasm	No	[75]
‘Candidatus Finniella inopinata’	‘Candidatus Paracaedibacteraceae’	Viridiraptor invadens	Cytoplasm	Yes	[75]
‘Candidatus Finniella dimorpha’	‘Candidatus Paracaedibacteraceae’	Euplotes daidaleos, Euplotes eurystomus, Euplotes octocarinatus	Cytoplasm	No	[52]
‘Candidatus Intestinibacterium nucleariae’	‘Candidatus Paracaedibacteraceae’	Nuclearia delicatula	Cytoplasm	No	[148]
‘Candidatus Intestinibacterium parameciiphilum’	‘Candidatus Paracaedibacteraceae’	Paramecium biaurelia	Cytoplasm	No	[95]
‘Candidatus Captivus acidiprotistae’	‘Candidatus Paracaedibacteraceae’	Unnamed protists from acidic mine drainage	Cytoplasm	No	[63]
‘Candidatus Parafinniella ignota’	‘Candidatus Paracaedibacteraceae’	Euplotes sp.	Cytoplasm	No	[52]
‘Candidatus Bodonicaedibacter vickermanii’	‘Candidatus Paracaedibacteraceae’	Bodo saltans	Cytoplasm, in proximity of the nucleus	Yes	[57]

Holospora and HLB display long branches in molecular phylogenies, indicative of fast sequence evolution, and their close relatives display quite long branches as well. Accordingly, the members of whole clade, including Holospora and HLB, have been defined as ‘fast-evolving’ Holosporaceae (highlighted in Fig. 1) [58], which include the majority of the characterised Holosporaceae. Besides holosporas, several fast-evolving Holosporaceae live in association with ciliates as well (Table 1) [96, 107], namely ‘Candidatus Fujishimia apicalis’, symbiont of Euplotes octocarinatus [52], ‘Candidatus Hydrogenosomobacter endosymbioticus’, symbiont of an anaerobic Cyclidium-like scuticociliate [53], ‘Candidatus Paraholospora nucleivisitans’, symbiont of Paramecium sexaurelia [71], ‘Candidatus Mystax nordicus’, symbiont of Paramecium nephridiatum [56], and ‘Candidatus Gromoviella agglomerans’, symbiont of Paramecium polycaryum [58]. Other members of this clade were characterised in association with marine diplonemids (Table 1), namely ‘Candidatus Cytomitobacter primus’, symbiont of Diplonema japonicum [50], ‘Candidatus Cytomitobacter rhynchopi’, symbiont of Rhynchopus asiaticus [108], as well as ‘Candidatus Cytomitobacter indipagum’ and ‘Candidatus Ignotibacter abundans’ (formerly ‘Candidatus Nesciobacter abundans’), symbionts of the same strain of Diplonema aggregans [54]. Uncharacterised fast-evolving Holosporaceae were found in association with the prairie dog flea Oropsylla hirsuta [29], in the gut of the hemipteran Pyrrhocoris apterus [89], and in the microbiome of the marine bryozoon Bugula neritina [33]. Other members of this clade were detected in multiple environments and sources, such as freshwater lakes [17, 22, 24, 109], hypersaline microbial mat [18], activated sludge [110], acidic mine drainage [25], ocean depths [111], soil [112], and hospital dental units [113].

The genus ‘Candidatus Hepatobacter’ is phylogenetically proximate to the fast-evolving Holosporaceae (Fig. 1). ‘Candidatus Hepatobacter penaei’ thrives inside the epithelial cells of the hepatopancreas of the marine shrimp Litopenaeus vannamei (Table 1) [10, 47, 49] and possibly other akin crustaceans, such as L. setiferus, L. stylirostris, Farfantepenaeus aztecus, and F. californiensis [114]. On the other hand, ‘Candidatus Hepatobacter paralithodis’ was found in the hepatopancreatic epithelium of the crab Paralithodes platypus [48]. Phylogenetically close bacteria were found in possible association with amphipods (Melita plumosa) [115], sponges (Xestospongia muta) [31], and oribatid mites (Achipteria coleoptrata) [34]. Thus, notwithstanding the uncertainties on the actual hosts for the latter bacteria, it is possible to wonder whether the members of this subclade could present a marked preference towards arthropod and other metazoan hosts rather than protists (Fig. 1), being an exception among the whole Holosporineae.

Early diverging Holosporaceae are characterised by overall shorter branches in molecular phylogenies. The only described species is ‘Candidatus Bealeia paramacronuclearis’, found as symbiont of two P. biaurelia strains (Fig. 1; Table 1), in one case in coexistence with the Rickettsiales bacterium ‘Candidatus Fokinia cryptica’ [1]. ‘Candidatus Bealeia’-allied bacteria were found in freshwater particles associated with blooms of dinoflagellate Alexandrium monilatum [116]. Other early-diverging Holosporaceae were detected from various origins, including the stomach of the catfish Pelteobagrus fulvidraco [90], the intestine of zebrafish [88], aquatic moss [117], soil [118–122], mine drainage and water [16, 123], freshwater and marine sediments [124–126], and volcanic cinder deposit [127].

Further sequences assigned to Holosporaceae-related bacteria were found in the microbiota of Daphnia cf. pulex [37] and in olive oil production pomace [128].

The ‘Caedimonadaceae’ are hosted by diverse protists, such as ciliates, dinoflagellates, and amoebas [7, 11, 74, 129]. This is the least rich family in terms of described genera and species, including only three monotypic genera (Fig. 1; Table 1). ‘Caedimonas varicaedens’ is the most-studied member. This bacterium was re-described recently [7], but it was known for decades, being previously placed within the gammaproteobacterial genus Caedibacter, due to shared traits such as R-bodies and killer trait [7]. Former species now part of ‘Caedimonas varicaedens’ were Caedibacter varicaedens, Caedibacter caryophilus, and ‘Caedibacter macronucleorum’ [67, 130, 131]. ‘Caedimonas’ was found as symbiont of multiple Paramecium species, in particular, P. caudatum, members of the P. aurelia complex such as P. biaurelia and P. novaurelia, and P. duboscqui, being the causative agent of the killer effect [6, 67, 131–133]. On the other hand, it was also detected as symbiont of other ciliates (Table 1), such as Spirostomum ambiguum [134] and Euplotes sp. [52], as well as in the dinoflagellate Peridinium cinctum [6, 129], though without evidence of a killer effect. Further relatives of ‘Caedimonas’ were retrieved from various origins, such as urban aerosol [135] and sludge [136]. ‘Candidatus Paracaedimonas acanthamobae’ (formerly ‘Candidatus Caedibacter acanthamoebae’ [11]) is phylogenetically close to ‘Caedimonas’ and was originally described as symbiont of Acanthamoeba [11]. ‘Candidatus Paracaedimonas’ bacteria were repeatedly detected in acanthamoebas worldwide [137, 138] and were also found in soil [139] and bioreactors [140]. Moreover, sequences of bacteria more distantly allied to ‘Caedimonas’ and ‘Candidatus Paracaedimonas’ were retrieved from drinking water [15].

The other and more diverging described representative of the ‘Caedimonadaceae’ is ‘Candidatus Nucleicultrix amoebiphila’ (Fig. 1; Table 1), symbiont of the amoebozoan Hartmanella sp. [74]. This bacterium displays an infectious cycle, as was able to invade Acanthamoeba castellanii in laboratory experiments. Sequences of bacteria that are phylogenetically allied to ‘Candidatus Nucleicultrix’ were found in soil [141, 142] and in a lake [143].

Further members of the ‘Caedimonadaceae’ were detected in the gut of the beetle Harpalus pensylvanicus [86], as well as in lakes [26], seawater [144], marine sediments [145, 146], and drinking water [147].

The family ‘Candidatus Paracaedibacteraceae’ displays a high phylogenetic diversity, with seven genera and 11 species described (Fig. 1; Table 1). The breadth of their collective host range is comparable to the Holosporaceae and possibly even wider in terms of relative frequency of each main host lineage. Ascertained hosts are protists belonging to various lineages, including amoebas, cercozoans, ciliates, nucleariids, and euglenozoans [11, 57, 62, 69, 75, 95, 148]. The most long-term known representatives of the family belong to the genus ‘Candidatus Paracaedibacter’ and are found intracellularly in Acanthamoeba, including clinical isolates [11, 149–152], which led some authors wondering about their role in pathogenicity for the eye of the amoebas. ‘Candidatus Paracaedibacter’ is paraphyletic with respect to ‘Candidatus Odyssella thessalonicensis’, which is a symbiont of Acanthamoeba as well [62] and is more closely related to ‘Candidatus Paracaedibacter acanthamoebae’ than the two of them to ‘Candidatus Paracaedibacter symbiosus’ (Fig. 1) [1, 75]. Specifically, ‘Candidatus Paracaedibacter acanthamoebae’ and ‘Candidatus Odyssella thessalonicensis’ share a 16S rRNA gene identity of 97.8%, while, respectively, having 92.5% and 91.8% identity with ‘Candidatus Paracaedibacter symbiosus’ [1]. Considering the commonly accepted genus threshold for the 16S rRNA gene (94.5%) [153], and that the original description did not designate any type species for the genus ‘Candidatus Paracaedibacter’ [11], we propose to elect ‘Candidatus Paracaedibacter symbiosus’ as type and to move ‘Candidatus Paracaedibacter acanthamoebae’ into the genus ‘Candidatus Odyssella’ as ‘Candidatus Odyssella acanthamoebae’ comb. nov. (Table 1) and we will be referr to this bacterium as such from now on (see taxonomic revision at the end). Relatives of ‘Candidatus Paracaedibacter’/ ‘Candidatus Odyssella’ were detected in many environments and sources, namely lakes [22, 24], soil [154, 155], groundwater [156, 157], drinking water [23, 158], sludge [159, 160], wastewater [161], acidic mine drainage [25], textiles [162], human faeces [163], hospital dental units [113], and hydrocarbon [164].

The ‘Candidatus Paracaedibacteraceae’ includes several other clades with respective phylogenetic relationships not yet fully and consistently resolved (Fig. 1). ‘Candidatus Finniella’ bacteria belong to one of those clades and are hosted by cercozoans, namely ‘Candidatus Finniella lucida’ symbiont of Orciraptor agilis and ‘Candidatus Finniella inopinata’ of Viridiraptor invadens [75], and by ciliates, namely ‘Candidatus Finniella dimorpha’ symbiont of multiple Euplotes spp. (Table 1) [52]. Relatives of ‘Candidatus Finniella’ were retrieved in a possible association with Hydra vulgaris [70] and from multiple additional sources, such as lakes/rivers [21, 24, 27, 165], drinking water [23, 166], acidic mine drainage [16, 25], and electronic waste aerosol [167].

Another clade of the ‘Candidatus Paracaedibacteraceae’ is the one encompassing the genus ‘Candidatus Intestinibacterium’ (formerly ‘Candidatus Intestinusbacter’) (Fig. 1; Table 1), with the described species ‘Candidatus Intestinibacterium nucleariae’ symbiont of the opisthokont Nuclearia delicatula [148] and ‘Candidatus Intestinibacterium parameciiphilum’ symbiont of P. biaurelia [95]. Members of the same clade were retrieved from several, prevalently aquatic, sources, such as lakes [17, 19–21, 24, 26, 27, 168, 169], a river [170], seawater [143], a sulphidic spring [171], subsurface water [172], drinking water [173], wastewater [161], biofilm [157, 174], microbial mat [175], and even human skin [91, 94]. Moreover, bacteria related to both ‘Candidatus Intestinibacterium’ and ‘Candidatus Finniella’ were retrieved in an aquaculture after applying silver nanoparticles [176].

‘Candidatus Captivus acidiprotistae’ is another representative of the family (Table 1), found as endosymbiont in unnamed protists from an acidic mine drainage [63]. A close relative is an unnamed symbiont of the euglenozoan Petalomonas sphagnophila (Fig. 1) [72]. Further relatives were found in soil [177–179], sediment [180], acidic mine drainage [181], and an acidic pit lake [169]. The other two described members of the ‘Candidatus Paracaedibacteraceae’ are ‘Candidatus Parafinniella ignota’, symbiont of Euplotes ciliates [52] in coexistence with other intracellular bacteria [182], and ‘Candidatus Bodonicaedibacter vickermanii’ (formerly ‘Candidatus Bodocaedibacter vickermanii’), symbiont of the free-living kinetoplastid Bodo saltans (Fig. 1; Table 1) [57].

Another quite conspicuous clade of ‘Candidatus Paracaedibacteraceae’ includes only uncharacterised bacteria, frequently derived from the gut of various animals (Fig. 1), in particular termites (e.g. Reticulitermes speratus, R. santonensis, and Coptotermers curvignatus) [28, 32, 183–188], as well as the sea cucumber Apostichopus japonicus [36], the fish Ctenochaetus striatus [189], and springbok antelopes [87].

Further ‘Candidatus Paracaedibacteraceae’ bacteria were retrieved from human skin [93], the blowhole of the bottlenose dolphin Tursiops truncatus [92], the epithelium of Hydra magnipapillata [30], lakes [21, 24, 190], a river [191], biofilm [192, 193], drinking water [147, 194], groundwater [195], and wastewater [196, 197].

Host Interactions, Genomics, and Evolution

The interactions between the Holosporineae and their hosts, both in terms of mechanisms and effects, are overall still poorly understood, also due to the inherent experimental limitations in handling host-associated bacteria. In the last 15 years, a growing number of Holosporineae genomes have been sequenced, belonging to each of the three families, namely Holosporaceae, three ‘Caedimonadaceae’, and five ‘Candidatus Paracaedibacteraceae’ (Table 2). While many genus-level sublineages are not yet sequenced, available genomes have represented a major advance, allowing a deeper understanding of the functional features of the Holosporineae and of their evolution. Leveraging on the available observational, experimental, and genomic data, below, we will present a comprehensive account of the current knowledge on those subjects, focusing in particular on the most-studied cases, namely Holospora, ‘Caedimonas’, and the NHP determinant ‘Candidatus Hepatobacter penaei’.

Table 2.

List of Holosporineae with assembled genomes, and respective assembly statistics

Species	Host	Size	GC%	Number of contigs/scaffolds	N50	Plasmid number	Accession	Reference
Family Holosporaceae
Holospora undulata subsp. undulata	Paramecium caudatum	1.4 Mbp	36	203	10.9 kbp	Not determined	GCA_000388175.3	[226, 309]
Holospora undulata subsp. elegans	Paramecium caudatum	1.3 Mbp	36	152	13 kbp	Not determined	GCA_000648275.1	[226]
Holospora obtusa	Paramecium caudatum	1.3 Mbp	35	91	24.4 kbp	Not determined	GCA_000469665.2	[226]
‘Holospora curviuscula’	Paramecium bursaria	1.7 Mbp	37.5	152	40.5 kbp	Not determined	GCA_002930195.1	[101]
‘Candidatus Hydrogenosomobacter endosymbioticus’	Cyclidium-like scuticociliate	826.7 kbp	41.5	1	826.7 kbp	Absent	GCA_021654655.1	[310]
‘Candidatus Cytomitobacter primus’	Diplonema japonicum	622.4 kbp	30	1	622.4 kbp	Absent	GCA_008189405.1	[54]
‘Candidatus Cytomitobacter indipagum’	Diplonema aggregans	628 kbp	29.5	1	628 kbp	Absent	GCA_008189285.1	[54]
‘Candidatus Ignotibacter abundans’	Diplonema aggregans	616.1 kbp	30	1	616.1 kbp	Absent	GCA_008189525.1	[54]
‘Candidatus Gromoviella agglomerans’	Paramecium polycaryum	590 kbp	32	1	590 kbp	Absent	GCA_021065005.1	[58]
‘Candidatus Hepatobacter penaei’	Litopenaeus vannamei	1.1 Mbp	50	5	334.2 kbp	Not determined	GCA_000742475.1	[78]
‘Candidatus Bealeia paramacronuclearis’	Paramecium biaurelia	1.9 Mbp	43	1	1.9 Mbp	6	GCA_036670005.1	[14]
Family ‘Caedimonadaceae’
‘Caedimonas varicaedens’	Paramecium biaurelia	1.7 Mbp	42	142	20.2 kbp	Not determined	GCA_001192655.1	[311]
‘Candidatus Paracaedimonas acanthamoebae’	Acanthamoeba sp.	1.7 Mbp	38	1	1.7 Mbp	5	GCA_000743035.1	[78]
‘Candidatus Nucleicultrix amoebiphila’	Hartmanella sp.	1.8 Mbp	39.5	1	1.8 Mbp	Absent	GCA_002117145.1	Schulz et al., unpublished
Family ‘Candidatus Paracaedibacteraceae’
‘Candidatus Paracaedibacter symbiosus’	Acanthamoeba sp.	2.7 Mbp	41	50	1.7 Mbp	2	GCA_000757605.1	[78]
‘Candidatus Odyssella thessalonicensis’	Acanthamoeba sp.	2.8 Mbp	42	20	388.1 kbp	Not determined	GCA_000190415.2	[76]
‘Candidatus Odyssella acanthamoebae’	Acanthamoeba sp.	2.5 Mbp	41	1	2.5 Mbp	1	GCA_000742835.1	[78]
‘Candidatus Finniella inopinata’	Viridiraptor invadens	1.8 Mbp	44	28	174.7 kbp	Not determined	GCA_004210305.1	[59]
‘Candidatus Bodonicaedibacter vickermanii’	Bodo saltans	1.4 Mbp	40.5	1	1.4 Mbp	Absent	GCA_014896945.1	[57]

Holospora and HLB are probably the most deeply investigated among the Holosporineae and have been subject of dedicated extensive reviews over the years [2–4]. Indeed, these bacteria present a distinct set of apomorphic features, which allow them to engage in a dimorphic infectious life cycle interacting with their ciliate hosts, which in most of the characterised cases, though not all [98], belong to genus Paramecium [4, 69, 97, 100, 104]. The reproductive form of the typical holosporas is shaped as a short Gram-negative rod [69, 198, 199] and actively multiplies within the host nucleus, which may be the micronucleus or, more frequently, macronucleus of the ciliate, depending on the bacterial and host species (Fig. 2) [2, 4]. Under starvation or other conditions, which may be linked to the arrest of host protein synthesis [2, 200], the reproductive form differentiates into an elongated infectious form, which presents an extended electron-dense periplasm with a translucent tip (Fig. 2) [4, 69, 198, 200, 201]. The infectious forms do not divide and are typically released into the external medium through exocytic vesicles [202–206], formed by different mechanisms according to the species [207]. In extreme cases, infectious forms may cause the host cell lysis, possibly as a consequence of the deliverance of lipopolysaccharide, thus leading to their own release [2]. The infectious form is able to survive apart from the host for several days [208]. If ingested by a novel host cell, it gets activated by the acidification of the host digestive vacuole and is able to escape from the vacuole with the periplasmic tip ahead and then interacting with the host membrane trafficking systems and cytoskeleton, in order to reach its target nucleus [209–214]. It can enter the nucleus without causing its disruption and therein will produce back novel reproductive forms, thus fuelling the Holospora life cycle [2, 3, 215, 216].

Fig. 2.

Subcellular location and ultrastructure of ‘Candidatus Holospora parva’, symbiont of the ciliate P. chlorelligerum, modified from [100]. A Densely packed bacteria inside the macronucleus of the host, stained with a red fluorescent probe specific for ‘Candidatus Holospora parva’. B, C Ultrastructure of the bacteria. In B, both reproductive forms (RF) and infectious forms (IF) are shown, with the white arrows indicating the enlarged periplasm of the latter form. In C, a magnification of the details of an infection form is presented, namely the cytoplasm (c), the periplasm (asterisks), and its apical tip (t). The black arrow indicates fine fibrous material that may be present on the surface of some infectious forms of this bacterium. Scale bars: A 20 μm; B 2 μm; C 1.5 μm

The molecular determinants of the infection are only partly understood. Some studies, mainly on H. obtusa, were aimed at the characterisation of stage-specific proteins and their possible involvement in different infection phases, such as binding to actin or to the macronucleus [212, 217–224]. The identified proteins have little homology with those of other organisms and were labelled by their molecular weight (e.g. 89 kDa, 63 kDa, or 5.4 kDa proteins), thus preventing more extensive comparative studies to date.

As described above, Holospora and HLB present the typical traits of specialised parasites, namely an infectious life cycle with the possibility to harm their hosts, including hampering sexual processes [206, 225]. An ongoing parasitic interaction is also supported by genomic evidences of quite pronounced scavenging of metabolites by the bacteria from their hosts [101, 226], as well as by the presence of host resistance mechanisms of the ciliates against the infection [227–231]. As such, Holospora/HLB and Paramecium have been repeatedly used as experimental models for the evolution of host-parasite interactions, including plasticity and trade-offs between transmission modes, infectivity, and virulence in the parasite [232–235], as well as between resistance and fitness in the host [236]. Further studies investigated local adaptation [237–239], effect of environmental variations [240–243], impact of host growth and lifespan on the growth, infectivity and virulence of the parasite [244, 245], reciprocal effects of parasite traits and host dispersal [246–249], and competition among parasites [250]. At the same time, the effect of Holospora and the interplay with its host are more complex than a purely parasitic interaction. Indeed, in particular as a reproductive form, it was shown to induce positive effects on the host, namely protection against environmental stress, such as temperature, salinity, and osmotic variations [2, 251–253]. This has been tentatively linked to an enhanced heat-shock protein expression both of the host and the bacterium, which could make the host more reactive to stressors [254–256]. These findings could explain the observation of a higher frequency of bacteria in hosts sampled from brackish with respect to freshwater environments [134].

In any case, the remarkable traits of the interaction between Holospora/HLB and Paramecium/other ciliates are suggestive of a significant (co)evolutionary specialisation, which however can be difficult to trace by comparative analyses, considering the sharp differences with respect to the other closely-related Holosporineae. Nevertheless, it seems interesting to notice two different cases, which show somehow intermediate traits and could thus be seen as potentially reminiscent of some ancestral steps in the evolution of the peculiar infectious nuclear tropism. The first case is the HLB ‘Candidatus Goertzia sharazharadae’, which, besides being resident within the host macronucleus as typical for this bacterial lineage, was recurringly found free (i.e. without being enclosed in host-derived membranes) in the host cytoplasm [103]. The other instance is the one of ‘Candidatus Paraholospora nucleivisitans’, a fast-evolving Holosporaceae bacterium that is closely related to Holospora and HLB clade, although not its direct sister lineage. ‘Candidatus Paraholospora’ was observed in the host cytoplasm or, alternatively, in the nucleus, otherwise more rarely in both locations in the same host cell [71].

It is worth considering that many other Holosporaceae and Holosporineae in general display some relation with host organelles including the nucleus, up to being intranuclear as well. The early-diverging Holosporaceae bacterium ‘Candidatus Bealeia paramacronuclearis’ is loosely co-localised with the host macronucleus from its outside [1], similarly to the position of the ‘Candidatus Paracaedibacteraceae’ bacterium ‘Candidatus Bodonicaedibacter vickermanii’ with respect to its host nucleus [57]. On the other hand, the fast-evolving Holosporaceae bacteria ‘Candidatus Cytomitobacter primus’ and ‘Candidatus Mystax nordicus’ were found in proximity or even aggregation with host mitochondria [50, 56], with the former bacterium possibly able to enter within these organelles. Moreover, both ‘Candidatus Mystax’ and ‘Candidatus Gromoviella agglomerans’ (also a member of fast-evolving Holosporaceae) can form bacterial aggregates within the respective host cells [56, 58], the latter also with potential lethal division effects on the host. Finally, ‘Candidatus Hydrogenosomobacter endosymbioticus’ was found in association with host hydrogenosomes [53].

The endonuclear localisation is not exclusive of the Holosporaceae among the Holosporineae, as the same condition is typical also of some members of the ‘Caedimonaceae’. For instance, the intracellular localisation of ‘Caedimonas’ appears to be correlated with the host species, being intramacronuclear in Paramecium caudatum and Paramecium duboscqui [130, 131], while cytoplasmic in the species of the P. aurelia complex [7, 132]. The case of ‘Candidatus Nucleicultrix amoebiphila’ is even more distinctive, since this bacterium presents a complex infectious life cycle that is highly reminiscent of the one of Holospora and HLB [74]. Similarly to the latter, the effect of ‘Candidatus Nucleicultrix’ on the host is variable and was shown to be negligible for its natural host, the amoeba Hartmanella, but lethal for experimentally infected Acanthamoeba castellanii. Overall, it seems legitimate to speculate that the ability for interacting with host nuclei could have been ancestral in the Holosporaceae + ‘Caedimonadaceae’ lineage, being successively lost (or remaining unnoticed to date) in some of the descendants. Alternatively, multiple parallel evolutionary events of this trait are also possible, considering its independent evolution in other phylogenetically unrelated bacteria [257]. On the other hand, and remarkably, it seems more parsimonious to infer that the highly specialised infectious life cycles of Holospora + HLB and of ‘Candidatus Nucleicultrix’ have most likely arisen independently from such a hypothetical ‘permissive’ ancestral ability to colonise host nuclei.

‘Caedimonas varicaedens’, the other deeply investigated member of the Holosporineae, has been studied particularly for the distinctive killer trait that it confers to its Paramecium hosts, similarly to the gammaproteobacterium Caedibacter taeniospiralis [5–7]. A decades-long history of investigations was focused on the killer trait and its determinants, also before the discovery that ‘Caedimonas’ and Caedibacter were phylogenetically apart. As such, these two bacteria were treated jointly in many studies and in reviews on the subject [6, 258, 259], which will be summarised below. Part of the intracellular bacterial population ceases to divide and produce proteinaceous coiled ribbons (Fig. 3) that are light-refractile and thus called R-bodies [260, 261]. The bacteria bearing R-bodies can be discharged extracellularly through their host cytoproct [262] and, when ingested by a novel host, are lysed in the digestive vacuoles leading to the release of the R-bodies, which under those acidic conditions unroll and break the vacuolar membrane (Fig. 3) [261, 262]. The unrolling of R-bodies ultimately leads to host death, with different lethal symptoms according to the bacterial genotype, including reversion of the normal rotation direction of the cell, formation of large vacuoles, or paralysis [258]. Although required for the killer trait [263], the R-bodies are not directly toxic, rather their role is in the disruption of the host vacuolar membrane, allowing the delivery of the actual toxin produced by the bacterium [264–266]. This toxin is probably a protein but was yet not conclusively identified [6, 259]. The potential toxic activity of ‘Caedimonas’ towards eukaryotes other than Paramecium has been only seldom explored until now, with still inconclusive evidence [267]. On the other hand, paramecia hosting ‘Caedimonas’ are resistant, probably thanks to an antidote produced by the bacterium. Interestingly, in both ‘Caedimonas’ and Caedibacter, the determinants of R-bodies (as well as probably toxin and antidote) are encoded on plasmids and/or linked to phage particles and prophage induction [260, 268–272]. As such, the multipartite interactions involving ciliates, bacteria, and plasmids/phages were meaningfully defined as involving ‘extrachromosomal elements of extrachromosomal elements of Paramecium’ [273]. Being encoded on mobile elements, R-bodies and other determinants of the killer trait are likely horizontally transmissible, which would explain their presence in unrelated bacteria, such as Caedibacter and ‘Caedimonas’, as well as others (see below).

Fig. 3.

Ultrastructure of R-bodies, modified from [306]. A Coiled R-bodies (black arrows) within the bacterial symbiont cells in the Paramecium cytoplasm. B Isolated R-body in the process of unrolling in a telescopic fashion. Scale bars: A 100 nm; B 1 μm

Thanks to the killer trait, the paramecia bearing ‘Caedimonas’ display competitive advantages [133, 274], but at the same time, the bacteria can be parasitic for taking ATP for energy and other metabolites from their hosts [130, 274, 275], somehow comparably to Holospora. Besides, the peculiar effect of the killer trait was tentatively envisioned as a phenomenon of addictive manipulation, namely a way by which the bacterium indirectly prevents the host from getting rid of it, by ‘punishing’ it thanks to the action of bacteria released by still-infected neighbouring host cells [43]. This is posited to be analogous to the reproductive manipulation of arthropods exerted by Wolbachia and other bacteria, in particular cytoplasmic incompatibility, by which infected males sterilise the crosses with uninfected females, thus favouring the reproduction of infected females, the only ones that transmit the bacteria to the progeny [276]. Following these lines of thought, it is worth to consider that, although R-bodies and their genetic determinants are best studied in Caedibacter and ‘Caedimonas’, they are found in a wide range of phylogenetically unrelated bacteria [258, 277], which can employ their R-bodies in interaction with eukaryotes [278–281]. Interestingly, these other bacteria equipped with R-body genes include two other Holosporineae, affiliated to different families, namely ‘Candidatus Bealeia paramacronuclearis’ (Holosporaceae) and ‘Candidatus Finniella inopinata’ (‘Candidatus Paracaedibacteraceae’) [14]. R-bodies were not observed in either of these bacteria [1, 75], suggesting that their expression is conditional, and leaving still open which is their role (if any) in the interaction with the respective hosts, in particular, a possible addictive manipulation alike to the one exerted by ‘Caedimonas’ through the killer trait. Herein, it seems worthwhile to consider the potential implications in the evolution of Holosporineae as a whole, since phylogenetic reconstructions of the R-body genes are compatible both with a vertical inheritance from the ancestor of this lineage (followed by losses in other representatives) or with a recent exchange between the equipped members [14].

The other more investigated case study among the Holosporineae is ‘Candidatus Hepatobacter penaei’ (Holosporaceae), known for being the causative agent of NHP [49]. NHP is mostly known in the Pacific white shrimp L. vannamei, which is the main farmed shrimp worldwide [282], but the disease (or its causative agent) was detected also in other shrimp species, such as L. setiferus, L. stylirostris, F. aztecus, F. californiensis, F. duorarum, Penaeus monodon, Fenneropenaeus merguensis, and Melicertus marginatus, as well as the American lobster Homarus americanus [49, 114]. ‘Candidatus Hepatobacter penaei’ resides and multiplies exclusively inside the tubular epithelial cells of the host hepatopancreas [283], and its involvement in NHP was experimentally demonstrated [284]. This bacterium is pleomorphic, being observed as a coccoid/short rod-shaped form, and as a long helical rod with eight long periplasmic flagella [285, 286], which might be involved in bacterial motility, adherence to host cells, and virulence [287]. The bacterial infection heavily damages the hepatopancreas, causing detachment of tubular cells, melanisation, and necrosis of the tubules, strong intracellular haemocytosis, and oedema [49]. The disease ultimately affects multiple organs and functions, as the observed signs include, among many others, anorexia, lethargy, abdominal muscular atrophy, soft exoskeleton, decreased growth rate, empty intestines, erosion of appendages, darkening, and lesions in the cuticle [288, 289]. NHP chronically causes mortalities of up to 50 − 95% in affected postlarval stages and well as in juveniles and broodstock [290], with significant economical impacts in terms of production losses and management costs [291]. At the appropriate progression stages, it can be effectively counteracted by antibiotic treatments [292]. The transmission of the NHP is not entirely clarified but can occur rapidly in densely populated farms [49], and a possible role of microalgae, other crustaceans (Artemia), or zooplankton as vectors was hypothesised [293, 294]. The disease is common especially in the south of the USA, as well as in Central and South America, and was shown to occur typically after persistently high water temperature (29–35 °C) and salinity (30–40‰) during summer, with the bacterium reaching over 15% prevalence in farms [295], but less than 1% in the wild [282]. Available molecular diagnostic tools include a multilocus sequence analysis [114], as well as a qPCR on flgE flagellar gene [290].

It is noteworthy that the recently described relative ‘Candidatus Hepatobacter paralithodis’ also affects a crustacean, namely the blue king crab P. platypus, although with very low prevalence in the wild [48]. Several traits are in common with NHP, as this bacterium as well is localised only within the epithelial cells of the host hepatopancreas, mostly free from host vacuoles. Two main morphotypes are present, namely rounded forms, often found in chains, and elongated rods devoid of flagella. The hepatopancreas structure and function are impaired by the bacteria, with hypertrophy and desquamation of infected cells, softening of tubules, granuloma, and necrosis. However, very little external signs of the disease were observed, besides lethargy. The geographical and climatic pattern is different from NHP, as ‘Candidatus Hepatobacter paralithodis’ was found in a much colder area of the Northern Hemisphere (Sea of Okhotsk), but seasonal patterns in the disease emergence were suspected as well [48].

Taking into account the more deeply investigated cases presented above and the available data on other representatives, it is thus possible to summarise the available knowledge into an evolutionary scenario for the interactions between the Holosporineae and their hosts. In quite ancient times, the free-living ancestors of this bacterial lineage acquired the ability to interact with eukaryotic hosts, which were most likely unicellular aquatic ones, and, based on relative environmental frequency of the current available representatives and its phylogenetic patterns, may have been more specifically freshwater [95]. It seems likely that, analogous to other professional symbionts [296–299], secretion systems and effectors have likely played a pivotal role in the establishment and successive development of these interactions. In the case of the Holosporineae, the most credible candidate is the type VI secretion system, conserved in almost all the genomes sequenced, including the smallest ones, with the exception of Holospora spp. Despite the lack of experimental data on its functioning and on possible secreted molecules, this high conservation is highly indicative of the key role of this apparatus in the lifestyle of Holosporineae. The type VI secretion system of Holosporineae is probably a non-standard one, considering the apparent lack of genes for some main components (outer membrane complex TssD and inner tube TssJ) [54, 58]. The case of Holospora could be explained by the markedly specialised infectious life cycle of these bacteria [2, 3], which likely involves equally specialised effector molecules [212, 220, 222, 224], possibly making the ‘canonical’ ones among the Holosporineae superfluous.

As a matter of fact, while only few members of the Holosporineae were experimentally shown to invade novel hosts [3, 74, 284], indirect evidence from compared host and symbiont phylogenies clearly indicates the recurrent ability of these bacteria of host transfer and host species shift along their evolutionary history, e.g. [1, 14, 52, 54, 56, 75]. Flagella likely play an important role in transmission and invasion of novel hosts, as hypothesised for other professional symbionts [40, 300–302], potentially working also as an additional secretion system [303]. As described above, a role of flagella in host invasion could be the case for ‘Candidatus Hepatobacter penaei’, which is the only Holosporineae bacterium observed bearing flagella but likely applies also for several other representatives equipped with flagellar genes, which could express them conditionally, similarly to what many Rickettsiales are thought to do [40, 300]. From an evolutionary perspective, a still relevant open point is the transition from putative ancestral protist hosts to Metazoans. This is a common trait observed among professional symbionts, having significant medical or veterinary impacts, as it can pave the way for the emergence of dangerous pathogens [44–46]. In the case of the Holosporineae, this has occurred relatively rarely, with the only ascertained documented case being the ‘Candidatus Hepatobacter’ lineage [47, 48]. It is yet to be determined whether this is only somehow accidental, or whether the molecular genetic set of the Holosporineae is for any reason less permissive than other professional symbiont lineages for such transition.

Comparative genomics quite sharply indicates metabolic dependence of the Holosporineae on their hosts, particularly in terms of obtaining metabolic precursors, such as amino acids or nucleotides, and cofactors [14]. With respect to their free-living ancestors, all the Holosporineae appear to have experienced variable degrees of genome reduction, from relatively large sizes (2.5–3 Mb) in certain ‘Candidatus Paracaedibacteraceae’ [76, 78], down to less than 600 kbp in some fast-evolving Holosporaceae [54, 58]. This suggests that the reduction was probably rather gradual and/or recent, with possible lineage-specific patterns, with the larger gene repertoires suggestive of more complex yet uncharacterised regulatory mechanisms and interaction mechanisms with the hosts. Reductive trends are particularly marked among the Holosporaceae, likely with concurrent specialisation towards certain life cycles and/or hosts, in particular in the extremely reduced fast-evolving members. Indications of possible multiple losses of a biosynthetic ability (i.e. for biotin) as a consequence of the independent acquisition of the respective transporters in different sub-lineages of Holosporineae were obtained [14], reminiscently of the Rickettsiales [40]. Nevertheless, the data available so far suggest that the major evolutionary steps which resulted in host-dependence occurred only once in the common ancestral evolution of all Holosporineae, rather than independently in the different sub-lineages.

Regarding interaction mechanisms and effects on the host, the presence in other Holosporineae of the genes for the R-bodies, involved in the killer trait conferred by ‘Caedimonas’ to its hosts, raises the question of whether similar mechanisms might be more pervasive among the Holosporineae and whether they might also have some implications in earlier evolutionary steps of this lineage [14]. This is even more so if we consider such phenomena as instances of addictive manipulation exerted by the bacteria on their hosts [43] and if we also take into account other cases among Holosporineae which could indicate addiction. These pertain ‘Candidatus Cytomitobacter spp.’ (Holosporaceae) and ‘Candidatus Bodonicaedibacter vickermanii’ (‘Candidatus Paracaedibacteraceae’), which, while devoid of R-bodies, could not be eliminated by antibiotics or cause the death of the host as well if removed, respectively [50, 57].

Finally, considering that multiple Holosporineae belonging to different families are pleomorphic, including the abovementioned Holospora/HLB, ‘Caedimonas’, and ‘Candidatus Hepatobacter’, as well as ‘Candidatus Finniella’ species [52, 75], it is interesting to wonder whether these are purely secondary lineage-specific adaptations or rely also on an ancestral morpho-functional ‘flexibility’ in the lifestyle of the ancestral Holosporineae.

Final Remarks and Perspectives

Here, we provide an in-depth review of the literature on a broad, diverse, and ancient alphaproteobacterial lineage living in obligate association with eukaryotic hosts, mostly protists, aiming for a comprehensive dedicated resource for interested researchers. With the purpose of offering a common and stable nomenclature ground, we also propose a taxonomic revision based on phylogenetic and taxonomic considerations. Specifically, while recently these bacteria were commonly referred to either as an order (Holosporales) or a family (Holosporaceae) [1, 59], we move them as a suborder within the Rhodospirillales, namely the Holosporineae, keeping the internal taxonomic substructure of the Holosporales sensu Szokoli et al. [1].

We put forward that the knowledge on the diversity, host range, environmental, and geographical spread of the Holosporineae, though growing steadily through the years with continuous novel reports, is probably still an underestimate to date. Indeed, their most frequent hosts, protists, are still neglected in studies on associations with bacteria as compared to multicellular eukaryotes [41] and, despite being ecologically and geographically widespread [304], have little individual biomass, which may frequently hinder the detection of their associated bacteria, Holosporineae included, in environmental metagenomic screening studies.

Even in the most studied Holosporineae Holospora and ‘Caedimonas’, the interactions with the hosts, including mechanisms and reciprocal effects, are still insufficiently understood and almost entirely unknown in most of the other members of this lineage. This is quite unfortunate, not only for the sake of scientific and evolutionary curiosity on this neglected lineage, but also for several other relevant reasons. One of those is the still underappreciated, but well-plausible, impact that Holosporineae may have on aquatic ecosystems, based on the marked effects they have on certain widespread hosts, such as the competitive advantages conferred to the ‘killer’ paramecia by ‘Caedimonas’ [6]. Moreover, we underline the still not well-exploited convenience of using Holosporineae in comparative studies with other more renowned bacterial lineages sharing lifestyle features, such as the other professional symbionts [41]. This applies in particular to the Rickettsiales [40, 299], as these two lineages are independent instances of evolutionarily long-lasting associations with eukaryotes among the Alphaproteobacteria, thus offering a way to discern traits that are common, and thus more likely fundamental, from lineage-specific ones.

It is also worthwhile to consider the direct economical impact of those Holosporineae that affect farmed crustaceans [49, 291]. Additionally, we should not overlook that some Holosporineae are frequently associated with human pathogenic amoebas [11, 62, 137, 138, 149, 150, 152] and that DNA of others was found associated to humans [91, 93, 94, 163]. The latter findings may indicate a yet-to-be-confirmed direct association with human cells or the presence of undetected skin protists/fungi intracellularly hosting the bacteria. All these studies suggest a hypothetical role of the bacteria in the diseases [11], with potential analogies with the Wolbachia symbionts of filarial nematodes [305], and thus deserving further targeted investigations.

To sum up, accounting for evolutionary, ecological, economical, and possibly sanitary reasons, we highlight the need for future investigations to reveal the diversity of the Holosporineae and elucidate their functional interactions with eukaryote, which may be hopefully fostered and sustained by the recent and possible future increased availability of genomic sequences of these bacteria.

Taxonomic Proposals

Description of the Species ‘Candidatus Odyssella acanthamoebae’ comb. nov.

‘Candidatus Odyssella acanthamoebae’ (O.dys.sel’la. a.can.tha.moe’bae, N.L. fem. pl. n.). This corresponds to the description of ‘Candidatus Paracaedibacter acanthamoebae’ [11], with the following modifications. Phylogenetic position, family ‘Candidatus Paracaedibacteraceae’; intracellular symbiont of Acanthamoeba sp. UWC9 and other Acanthamoeba strains [11, 149–151].

Emended Description of the Genus ‘Candidatus Paracaedibacter’ Horn et al. 1999

‘Candidatus Paracaedibacter’ (Pa.ra.cae.di.bac’ter, N.L. masc. s. n.). The genus contains only one described species, namely ‘Candidatus Paracaedibacter symbiosus’ [11]. Is the type genus of the family ‘Candidatus Paracaedibacteraceae’ [75].

Emended Description of the Family ‘Candidatus Hepatincolaceae’ Szokoli et al. 2016

‘Candidatus Hepatincolaceae’ (He.pat.in.co.la’ce.ae, N.L. fem. pl. n.). The description of ‘Candidatus Hepatincolaceae’ [1] is emended as follows. The family currently contains three genera, ‘Candidatus Hepatincola’ [12], ‘Candidatus Tenuibacter’ [79], and ‘Candidatus Tardigradibacter’ [81], and is affiliated to the order Rhodospirillales. The type genus is ‘Candidatus Hepatincola’.

Description of the Suborder Holosporineae subord. nov.

Holosporineae (Ho.lo.spo.ri’ne.ae, N. L. fem. n., Holospora type genus of the suborder; suff. -ineae ending to denote a suborder; N.L. fem. pl. n. Holosporineae, the suborder of the genus Holospora). The description is the same as that given previously for Holosporales [1], with some modifications. Defined by phylogenetic analyses based on SSU rRNA gene sequences and on concatenated conserved protein-coding ortholog genes. The suborder contains three families (Holosporaceae, ‘Caedimonadaceae’, and ‘Candidatus Paracaedibacteraceae’). The suborder Holosporineae is a member of the order Rhodospirillales.

Author Contribution

M.C. and G.P. planned and organised the work. M.C. performed the literature search, wrote the main manuscript text and prepared the figures. All authors reviewed the manuscript.

Funding

Open access funding provided by Università degli Studi di Pavia within the CRUI-CARE Agreement.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing Interests

The authors declare no competing interests.