Introduction: what is the PVC superphylum?

The application of 16S rRNA-based phylogenetics and molecular ecology methods revealing wholly new groups within the domain bacteria encouraged the definition of new phylum-level divisions within the bacteria. Such divisions or phyla, detected by their forming distinct clades in phylogenetic trees, were quantitatively defined by shared 16S rRNA sequence similarity of less than 85 % when their members were compared (Hugenholtz et al. 1998a, b). These phyla even included ‘candidate’ divisions represented only by uncultured environmental community members (Hugenholtz et al. 1998a, b; Dunfield et al. 2012). However, some separate phyla were soon recognized as related, forming looser higher level clades in phylogenetic trees (Wolf et al. 2001). Such superphylum-level groups included one defined as the PVC superphylum (Wagner and Horn 2006), named for its member phyla Planctomycetes, Verrucomicrobia and Chlamydiae, but now known to include the uncultured phylum OP3 and the phylum Lentisphaerae. The planctomycetes are widely distributed environmental bacteria (Fuerst and Sagulenko 2011) occurring in soil and in marine and freshwater, including wastewater, and can be aerobic or anaerobic including the obligately anaerobic ammonium-oxidizing anammox planctomycetes (Fuerst and Sagulenko 2011). Members of phylum Verrucomicrobia (Hedlund et al. 1997) occur widely in soil, in marine habitats and even in the human and animal model intestine. The anaerobic Akkermansia species may play an important role in mucin degradation and in maintaining a healthy intestinal lining (Belzer and de Vos 2012; Everard et al. 2013). An extremophile within verrucomicrobia is the thermoacidophilic methane oxidizer Methylacidiphilum (Dunfield et al. 2007; Pol et al. 2007). OP3 organisms, recently named phylum Omnitrophica (Rinke et al. 2013) were originally detected in Obsidian Pool of the Yellowstone National Park hot spring system (Hugenholtz et al. 1998a, b) but are now known to include organisms from diverse habitats including anoxic sediments such as flooded paddy soils and marine sediments(Glockner et al. 2010). The lentisphaerae contain the cultured marine species Lentisphaera araneosa and L. marina (Cho et al. 2004, 2013) and the anaerobic inhabitant of the human gastrointestinal tract Victivallis vadensis of WWE2 subgroup II (Zoetendal et al. 2003) as well as a cultured anaerobic sludge organism Oligosphaera ethanolica (Qiu et al. 2013) and uncultivated organisms from landfill leachate in WWE2 subgroup 1 (Limam et al. 2010) and non-human primate faeces (Yildirim et al. 2010). The phylum Poribacteria of obligate marine sponge symbionts (Fieseler et al. 2004) was at one stage thought to be a member of the superphylum but this view appears to have to be revised in light of recent indel sequence analysis of superphylum members (Gupta et al. 2012).

What are the supporting data for a PVC superphylum?

The PVC superphylum was first proposed on the basis of 16S rRNA analysis which indicated a clustering of the phyla Planctomycetes, Verrucomicrobia, Chlamydiae, Lentisphaerae, OP3, and Poribacteria (Wagner and Horn 2006). There was also support from common ancestry of nucleotide transporter genes for at least the planctomycete Rhodopirellula baltica and chlamydiae. Much earlier, a relationship just of planctomycetes and chlamydiae had been proposed on the basis of 16S rRNA sequence comparison (Weisburg et al. 1986), and concatenated ribosomal protein analysis with R. baltica as representative planctomycetes confirmed this link (Teeling et al. 2004). Concatenated protein phylogenetics and indel analysis in turn linked verrucomicrobia to chlamydiae (Griffiths and Gupta 2007). Genome-derived phylogenetics using concatenated ribosomal proteins including anammox planctomycetes confirmed a relationship between planctomycetes and chlamydiae, and also found some orthologous group proteins such as a 60 kD cysteine-rich outer membrane protein and a DNA-binding protein to be shared (Strous et al. 2006). In a commentary review on the first description of the methane-oxidizing verrucomicrobia member Methylacidiphilum infernorum (Hou et al. 2008), Radhey Gupta announced the bioinformatic discovery of a signature protein [homologous to the hypothetical protein CT421.2 (NP_219933) in Chlamydia trachomatis] uniquely present in all available sequences from the Planctomycetes, Verrucomicrobia-Lentisphaerae and Chlamydiae phyla of the superphylum (Gupta 2008). 23SrRNA phylogenetics has supported a relationship between the phyla Planctomycetes, Chlamydiae, Verrucomicrobia and Lentisphaerae (Pilhofer et al. 2008) and confirmed membership of the OP3 phylum of this PVC superphylum as well as the four ‘core’ phyla with cultured members (Glockner et al. 2010). Phylogenomic analysis using concatenated proteins and indel analysis confirmed the superphylum but excludes phylum Poribacteria from the group (Fig. 1) (Gupta et al. 2012). Most recently a phylogenomic analysis using 38 concatenated genes has confirmed the PVC clade as including phyla Planctomycetes, Verrucomicrobia, Chlamydiae, Lentisphaerae and OP3, with a naming of the OP3 phylum as Omnitrophica (Rinke et al. 2013).

Fig. 1
figure 1

a Neighbor–joining distance tree for the sequenced cultured species belonging to the PVC group of bacteria based upon concatenated sequences for 16 conserved proteins. Members of the so far uncultured PVC phylum OP3 are not included. The numbers on the node indicate % statistical support for different nodes in the ML and NJ analyses, respectively. The scores that were less than 50 % are not shown and represented by (–). The letters in the circles mark separate clades for the phylum Planctomycetes phylum (P), phylum Chlamydiae (C), phylum Lentisphaerae (L), phylum Verrucomicrobia (V), and within this phylum class Opitutae (O1), and family Opitutaceae (O2). Reproduced from Fig. 1 in (Gupta et al. 2012)

Full size image

The case for the PVC superphylum as exceptions to the bacterial definition

There are substantial arguments supporting a perspective that members of the PVC superphylum form important exceptions to our concept and definition of the bacteria. This is so since many of their members display considerable differences from most other bacteria in a number of features, going far beyond the 16S rRNA and other gene sequence-based differences defining them as a clade. Concerning structural features, most dramatically in phyla Planctomycetes, Verrucomicrobia and Lentisphaerae is the occurrence of compartmentalization of the cell by endomembranes separating cell cytoplasm into at least two major regions, one of which contains the nucleoid. This appears to be a shared cell plan where an intracytoplasmic membrane (ICM) divides the cell cytoplasm into at least two major compartments (Fig. 2) (Lindsay et al. 2001; Fuerst 2005; Lee et al. 2009). These two compartments comprise an outer ribosome-free paryphoplasm (which can also extend into other more central parts of the cell via invagination of the intracytoplasmic membrane) and an inner pirellulosome or riboplasm containing ribosomes and DNA-containing nucleoid.

Fig. 2
figure 2

The unity and variety of cell plans in phylum Planctomycetes, illustrated for genera Pirellula, Gemmata and Kuenenia. The simplest plan occurs in Pirellula and related members of the Pirellula group such as Blastopirellula, as well as in Planctomyces and Isosphaera. It comprises two major compartments, the ribosome-free paryphoplasm (P, in yellow) and the pirellulosome (PS, in blue) containing ribosomes (R) and the DNA of the nucleoid (in red). The cytoplasmic membrane (dark blue in inset box) is closely apposed to the cell wall (light blue in inset box), and an intracytoplasmic membrane (ICM, in green) showing no continuity with the cytoplasmic membrane separates the paryphoplasm from the pirellulosome compartment. In organisms within the Gemmata clade such as G. obscuriglobus, and in anaerobic authotrophic anammox planctomycetes such as Kuenenia suttgartiensis, there is a third compartment inside the pirellulosome. In Gemmata, this is a region enclosing the nucleoid and some ribosomes and is surrounded by a nuclear envelope consisting largely of two membranes, except where connection is made between the outer membrane of the nuclear envelope and the ICM. The inner membrane of the nuclear envelope is indicated in purple wile the outer membrane is indicated in green consistent with its continuity with the ICM. In anammox planctomycetes such as Kuenenia, a ribosome-free anammoxosome organelle (brown) surrounded by a single membrane lies within the pirellulosome—this organelle contains enzymes operative in anaerobic ammonium oxidation and its envelope contains F-ATPase suggesting that a membrane potential may be generated across the anammoxosome membrane. Boxes show expanded regions of membranes to indicate the relationships of different membranes shown as bilayers without any potentially inserted membrane proteins—cytoplasmic membrane in dark blue, intracytoplasmic membrane in green, inner membrane of nuclear envelope in purple, and anammoxosome membrane in brown. Modified from Fig. 1 in Fuerst and Sagulenko (2013)

Full size image

Complex compartmentalization in planctomycetes can involve three cell compartments and complexity resembling that of the basic eukaryote cell plan. Thus, some planctomycetes have further compartments.

Gemmata obscuriglobus and related members of the Gemmata clade have a double-membrane-bounded nuclear compartment within the pirellulosome which is of course bounded by its intracytoplasmic membrane separating it from the paryphoplasm (Fuerst and Webb 1991; Lindsay et al. 2001; Wang et al. 2002). The nuclear compartment contains a characteristically condensed nucleoid as well as ribosomes, so both cytoplasm inside the nuclear compartment and cytoplasm outside it but still within the pirellulosome contain ribosomes. And the anaerobic ammonium-oxidizing anammox planctomycetes contain an additional ribosome-free anammoxosome organelle within the pirellulosome, bounded by a single membrane (Lindsay et al. 2001; van Niftrik et al. 2004; van Niftrik and Jetten 2012).

Thus in both the case of Gemmata clade strains and anammox planctomycete species, there is a third compartment in addition to paryphoplasm and pirellulosome, lying within the pirellulosome. In such cases transport mechanisms for passage of molecules including RNA and proteins across several internal membranes must exist, making these bacteria quite distinct from other bacteria without closed internal membrane-bounded regions. The case for existence of compartments in PVC organisms such as the planctomycetes will be outlined below.

Importantly concerning the exceptional nature of PVC organisms is the absence of peptidoglycan from phylum Planctomycetes, some members of phylum Verrucomicrobia and species of phylum Chlamydiae so far as has been determined, though some genes for peptidoglycan synthesis can be present. Chlamydiae possess all the genes needed for peptidoglycan synthesis and they can even stimulate a host response normally elicited by peptidoglycan suggesting some may be synthesized during infection (Welter-Stahl et al. 2006). Phylum Planctomycetes members possess cell walls composed mostly of protein which can be resistant to boiling in 10 % SDS (König et al. 1984; Liesack et al. 1986; Stackebrandt et al. 1986). Interestingly, chlamydial outer membrane proteins can be cysteine-rich and associated disulfide-crosslinking might substitute for structural functions of peptidoglycan (Yen et al. 2005). Some planctomycete cell wall protein is even cystine-rich by chemical analysis consistent with disulfide bridging, and it has been suggested disulfide-crosslinked proteins may stabilize planctomycete walls (Liesack et al. 1986). Neither planctomycetes nor chlamydiae possess the cell division protein FtsZ and in chlamydiae at least a MreB protein homolog might substitute for FtsZ in some way (Ouellette et al. 2012). Concerning other exceptional molecular features in the PVC group, there is a true eukaryotic tubulin in one genus—Prosthecobacter—of phylum Verrucomicrobia (Jenkins et al. 2002a, b), and this has recently been shown to form microtubules with five protofilaments rather than the 13 characteristic of eukaryotic microtubules (Pilhofer et al. 2011). Phylogenetics and structural biology suggests that if lateral gene transfer was responsible for the eukaryotic homology (Schlieper et al. 2005) it was probably a very ancient event and the Prosthecobacter tubulin may be close to an ancestral form of this cytoskeletal protein (Martin-Galiano et al. 2011; Pilhofer et al. 2011). Since the tubulin family protein FtsZ is also found in verrucomicrobia (Yee et al. 2007) and may even be present in the same strain of Prosthecobacter as contains the BTubA/BtubB tubulin homologs (Pilhofer et al. 2007), it is conceivable that these are bacteria in which the early gene duplication giving rise to both tubulins and FtsZ arose. Two other unusual features found in PVC members are the unique ladderane lipids of anammox planctomycetes found nowhere else in nature (Sinninghe Damste et al. 2005), and the sterols found in the ‘nucleated’ G. obscuriglobus planctomycete, synthesized by what appears to be the simplest pathway known (Pearson et al. 2003), though some other unrelated bacterial species can also produce sterol (Patt and Hanson 1978; Schouten et al. 2000; Bode et al. 2003; Volkman 2003). PVC members also display exceptional properties relative to other bacteria concerning function. The anammox planctomycetes not only have unique lipids, they possess a unique ability to oxidize ammonium anaerobically using distinctive biochemistry which appears to depend on the ribosome-less anammoxosome compartment and potentially on a membrane potential set up across the anammoxosome membrane. Such a mitochondrion-like internal organelle in a chemoautotroph is unique among the bacteria. In planctomycetes C1 transfer enzymes with homology to those of some other rare bacterial groups and archaeal methanogens have been found. There is some evidence that they are distinct phylogenetically from those of both other Bacteria and Archaea, though they may also be the result of a presumably ancient horizontal gene transfer (Chistoserdova et al. 2004). These C1 transfer enzymes do not appear to function in either methanotrophy or methane formation in planctomycetes, but in phylum Verrucomicrobia, Methylacidiphilum can perform methane oxidation at pH1 and at high temperature (Hou et al. 2008).

Finally, an endocytosis-like process of protein uptake resembling that of eukaryote endocytosis has been found in the planctomycete G. obscuriglobus (Lonhienne et al. 2010). This process is correlated with compartmentalization in cells of this species since protein accumulates specifically in the ribosome-less paryphoplasm compartment, and like the eukaryote process it involves receptor mediation (evidenced both by saturation and competition between distinct proteins for uptake), and protein degradation in the paryphoplasm. Even more relevant evolutionarily, the process may occur via a mechanism of vesicle formation from cytoplasmic membrane and is correlated with a membrane-coating (MC) protein homologous to those in eukaryotes such as clathrin (Lonhienne et al. 2010; Santarella-Mellwig et al. 2010). No comparable process is known to occur in any other bacteria. It is however possible that other planctomycete species will be found to have such an ability and that macromolecules other than proteins will also be found to be incorporated from the external milieu by a similar mechanism. The process in G. obscuriglobus is in any case of great significance for understanding possible modes of evolution of minimal endomembrane systems in early analogs or homologs of the first eukaryote cells.

The case against the PVC superphylum as exceptions to the bacterial definition

Despite possessing some dramatically unusual features for members of domain bacteria, PVC superphylum members have many characters in common with classical bacteria. So, although peptidoglycan is absent from the phylum Planctomycetes and so far there is no evidence for its presence as a cell wall component in phylum Chlamydiae, nevertheless there are at least some of the genes needed for peptidoglycan synthesis in planctomycetes (to a greater or lesser extent) (Jogler et al. 2012) and most of the genes for such synthesis in phylum Chlamydiae members. And in cells of phylum Verrucomicrobia we find evidence for peptidoglycan itself from meso-diaminopimelate analysis (Albrecht et al. 1987; Schlesner 1987). Interestingly from an evolutionary perspective, Verrucomicrobium spinosum has recently been found to possess the l, l-diaminopimelate aminotransferase pathway for synthesis of this peptidoglycan component, an unusual pathway used also by members of phylum Chlamydiae and plants (McCoy et al. 2006). Regarding genes for synthesis of typical components of a characteristically Gram-negative cell wall, there is a wide distribution in planctomycetes of genes for components of the LPS insertion complex, lipid A synthesis, outer membrane protein synthesis and insertion, OMPs and OMP assembly and evidence for expression of OMPs (Sutcliffe 2010; Speth et al. 2012). Finally, even in the most structurally divergent phylum Planctomycetes, homologs to domain bacteria members appear to dominate assignable ORFs deduced from the genome sequence, e.g. in R. baltica (Glockner et al. 2003), though hits of ORFs to Eucarya and Archaea genes can be as high as a dramatic 8 and 9 % relative to only 0.4 % for E. coli (Glockner et al. 2003). In several verrucomicrobia, genes for nitrogen fixation, a characteristically bacterial ability, are found (Khadem et al. 2010; Wertz et al. 2012).

The Shine-Dalgarno sequence in the 5′ untranslated region of mRNA is used for translation initiation in prokaryotes, but as found for several other diverse bacteria and Archaea, the percentage of genes using the Shine-Dalgarno sequence can be quite low in planctomycetes and even amount to less than 20 % of all genes (e.g. 17 % of all genes in R. baltica) (Palleja et al. 2009). This may reflect use of alternative translation initiation mechanisms. In Blastopirellula marina and for rRNA genes at least, there are similarities (e.g. in the Pribnow box) but also perhaps some differences (e.g. absence of −35 upstream sequence) from other bacteria in promoters (Liesack and Stackebrandt 1989). Interestingly in R. baltica the genes for purine, biotin and amino acid biosynthesis are not organized in operons, contrasting with the organization in other bacteria (Glockner et al. 2003).

The homology versus analogy controversy concerning PVC exceptional features

The similarities of PVC organisms to other bacteria may support one side in a controversy about the relevance of this group to the evolution of eukaryotes and the potential relationship of their eukaryote-like features to those of eukaryotes (McInerney et al. 2011; Devos 2012; Fuerst and Sagulenko 2012). PVC organisms may well be ‘curious’ exceptions to bacterial definitions but may not be relevant to eukaryote evolution if such exceptions are analogous rather than homologous, the result of convergent evolution to adaptations performing similar functions as those performed by similar eukaryote features (McInerney et al. 2011). Analogy would imply that no evolutionary model needs to be found for relations with eukaryotes. In that case however, exactly what molecular mechanisms underlie the analogous eukaryote features becomes in itself a very interesting evolutionary problem to solve. It could however be argued (see below) that assuming homology for the purposes of experimental design initially is a more efficient generator of hypotheses to be tested concerning the origins of these eukaryote-like features, and this has already born fruit in the case of the tubulins of Prosthecobacter and testing ability to form eukaryote-like microtubules (Pilhofer et al. 2011) and the successful test of the ability of planctomycete G. obscuriglobus to perform an endocytosis-like uptake of proteins (Lonhienne et al. 2010), correlated with presence of homologs of clathrin-like MC proteins associated with endocytic vesicles (Lonhienne et al. 2010; Santarella-Mellwig et al. 2010). Assumption of analogy on the other hand has yet to yield any testable experimental predictions or new test results.

Arguments supporting PVC members as exceptions to classical bacteria

In order to approach understanding just how and why PVC members may be considered exceptional relative to other bacteria, it is illuminating to outline some arguments which may be proposed supporting such a perspective, or which are necessary to understand to grant PVC members an exceptional status. This can also be used as a way of exploring some sources of controversy surrounding appreciation of the complex internal structure possible in many species of the group, including the apparent enclosed cell compartments of important model organisms such as G. obscuriglobus, Candidatus Kuenenia stuttgartiensis and Planctomyces limnophilus.

Argument 1—the planctomycete intracytoplasmic membrane is not a cytoplasmic membrane, and the paryphoplasm is not a periplasm (and thus comprises a genuine intracellular compartment unique among bacteria)

In some recent perspectives, some researchers have placed planctomycetes and PVC organisms in a much more bacterial context concerning cell organization than is warranted by the existing ultrastructural data. It should be more widely understood that the planctomycete intracytoplasmic membrane is not a cytoplasmic membrane understood as the major membrane associated with the cell wall and enclosing all cytoplasm, and the paryphoplasm is not a periplasm such as occurs in Gram-negative bacteria. Therefore, planctomycete cells are indeed compartmentalized in a way unique for bacteria. The arguments for this interpretation are as follows. Firstly, in planctomycetes (at least Pirellula staleyi, B. marina, G. obscuriglobus, and P. limnophilus) an outermost membrane layer can be seen just under the cell wall via TEM of sectioned cryosubstituted cells or freeze-fracture replicas (Lindsay et al. 1997, 2001), exactly as is expected for the position of a classical cytoplasmic membrane of either a Gram-negative or a Gram-positive bacterium in relation to the wall. This membrane is thus not absent from these organisms as is implied by the view which some have proposed assigning the status of outermost cytoplasmic membrane of the cell to an internal membrane, here referred to as the ICM. Though it is the outermost membrane, it is not the outermost layer of the cell boundary—that layer is the proteinaceous wall itself, isolatable after boiling in detergent as a rigid but non-peptidoglycan sacculus. If the cytoplasmic membrane were equivalent to outer membrane, it would be the part of the cell wall, indeed the outermost membranous layer of that wall—this is not the case, as it clearly lies interior to the wall in both sections and freeze-fracture replicas. The planctomycete internal ICM appears completely separate from the outermost cytoplasmic membrane which is closely apposed to cell wall. Because the cytoplasmic membrane is closely apposed to the wall, it is difficult to visualize in sectioned cells. Secondly, RNA, though apparently not ribosomal, appears to be present in the paryphoplasm of organisms such as Pirellula, Blastopirellula, Gemmata and Isosphaera consistent with its cytoplasmic rather than periplasmic nature (Lindsay et al. 1997; Lindsay et al. 2001). Thirdly, when cells of the planctomycete G. obscuriglobus are fixed in glutaraldehyde in moderately high osmolarity buffer, cells shrink like Gram-positive cells under hyperosmolar conditions and plasmolytic retraction of the cytoplasmic membrane characteristic of behaviour of Gram-negative bacteria under such conditions does not appear, consistent with close bonding of cytoplasmic membrane to wall external to it facilitated by glutaraldehyde cross-linking (Lindsay et al. 1995). Fourth, during the endocytosis-like uptake of protein by G. obscuriglobus, outermost membrane associated with the wall forms vesicles entering the paryphoplasm and containing the incorporated protein, consistent with this membrane comprising the cytoplasmic membrane and analogous with the role of that membrane in eukaryotes. The ICM is not involved with uptake.

Fifth, in the anammox planctomycete Candidatus K. stuttgartiensis, F-ATPase is found via immunogold localization mainly in the outermost membrane of the Kuenenia cell and the membrane of the anammoxosome organelle within the pirellulosome, as well as to a lesser extent in the intracytoplasmic membrane around the pirellulosome (van Niftrik et al. 2010) and see also the article by van Niftrik et al. in this issue). The status of the outermost cell membrane as cytoplasmic membrane was considered confirmed on the basis of this data.

If we accept the separation of cytoplasmic membrane from the internal ICM membrane, then it is clear that the internal membranes of PVC superphylum members do not arise from the cytoplasmic membrane. They are thus not analogous to internal membranes of other bacteria such as photosynthetically active membranes of bacteria such as Rhodobacter or the cyanobacteria, which arise from infolding of the cytoplasmic membrane in a way visible via ultrastructural membrane continuity. Planctomycete internal membranes are not mere invaginations of cytoplasmic membrane, though they might conceivably be in a form of communication with cytoplasmic membrane via vesicle formation as occurs in eukaryote cells (see below). Finally, the paryphoplasm compartment is not a periplasm, which in planctomycetes must lie between the cytoplasmic membrane and wall as in other bacteria, and not between the cytoplasmic membrane and the intracytoplasmic membrane enclosing the riboplasm and nucleoid region.

Concerning the case in other PVC phyla, as indicated from both cryosubstitution and freeze-fracture replica TEM, V. spinosum in phylum Verrucomicrobia displays a clear cytoplasmic membrane close to the wall and separate from the intracytoplasmic membrane as in planctomycetes (Lee et al. 2009). This is consistent with a shared cell plan with planctomycetes.

Argument 2—PVC members have a more complex cell plan than other bacteria including features resembling a eukaryote plan, and such cell compartmentalization is widely distributed among PVC superphylum species consistent with an ancestor already exceptional to a classical bacterial cell

Consistent with their exceptional status, PVC phyla members can possess a more complex cell plan than bacteria (plotting to the simplest eukaryote plan). In members of at least 3 PVC phyla, cells are compartmentalized by at least one major internal membrane. These form two major compartments, the paryphoplasm and the pirellulosome (Fig. 2). In contrast to the naked and in cryofixed cells uncondensed nucleoids of classical bacteria, there are single-membrane-bounded condensed nucleoids in the compartmentalized cells of species in phyla Planctomycetes, Verrucomicrobia and Lentisphaerae. In the planctomycete Gemmata clade only so far, a complex structure with several even more dramatic eukaryote-like features is observed. In G. obscuriglobus, these include a double-membrane-bounded nuclear compartment and ribosomes bound to internal membranes (Fig. 3), as well as the condensed nucleoids and single membrane enclosure of pirellulosome shared with other planctomycetes (Fuerst and Webb 1991; Lindsay et al. 2001). The condensed nucleoids in G. obscuriglobus share a liquid crystalline structure with the chromosomes of eukaryote dinoflagellates (Yee et al. 2012), and such structure may conceivably be correlated with special DNA-binding proteins. The homologs of the DNA-binding HU proteins of typical bacteria have unusual N or C-terminal extensions in Gemmata (Yee et al. 2011). Anammox planctomycetes such as Candidatus K. stuttgartiensis also possess a third compartment within the shared planctomycete and PVC cell plan, but the ribosome-less physiologically active anammoxosome is bounded only by a single membrane [(van Teeseling et al. 2013) see van Niftrik et al. in this Special Issue].

Fig. 3
figure 3

Transmission electron micrographs of sectioned cryosubstituted cell of Gemmata obscuriglobus showing a nuclear body (NB) surrounded by a double membrane nuclear envelope (NE) surrounding a condensed nucleoid (N). The boxed area on the cell is expanded on the right to show more clearly the two membranes comprising the nuclear envelope and the ribosomes (arrowheads) bound to each of the these membranes. Bar on the left is 1 μm and in the expanded box on the right 50 nm. Modified from Fig. 2 in Fuerst and Sagulenko (2013)

Full size image

The evidence for the reality of such cell compartments in PVC members has been derived from transmission electron microscopy of sectioned cryosubstituted cells in the genera Gemmata (Fig. 3), Pirellula, Blastopirellula, Zavarzinella, Planctomyces, Isosphaera, Kuenenia, Brocadia, and Scalindua in phylum Planctomycetes (Fuerst and Webb 1991; Lindsay et al. 2001; Kulichevskaya et al. 2009; Jogler et al. 2011), and from at least four of the five subdivisions recognized in phylum Verrucomicrobia (Hugenholtz et al. 1998a, b) including the genera Chthoniobacter, Verrucomicrobium and Prosthecobacter in phylum Verrucomicrobium, as well as from one genus Lentisphaera of the phylum Lentisphaerae (Lee et al. 2009). TEM examination of freeze-fracture replicas of cells in planctomycetes such as G. obscuriglobus, B. marina and P. staleyi and use of fluorescent dyes to stain internal membranes (in the case of G. obscuriglobus) supports such compartmentalization (Fuerst and Webb 1991; Lindsay et al. 2001). In G. obscuriglobus and B. marina, 3D reconstruction from serially sectioned cryosubstituted cells is consistent with closed compartments especially with respect to the nuclear compartment of G. obscuriglobus (Lindsay et al. 2001). Electron tomography of the anammox planctomycetes Candidatus K. stuttgartiensis, Candidatus Brocadia fulgida, Candidatus Anammoxoglobus propionicus, and Candidatus Scalindua spp. has confirmed the enclosed nature of the anammoxosome and pirellulosome compartments in these anammox planctomycetes (van Niftrik et al. 2008). The restricted distribution of protein taken up by an endocytosis-like process in G. obscuriglobus (see below) is also consistent with separation of paryphoplasm from a distinct pirellulosome compartment. It should be noted however that there are published interpretations of planctomycete structure which reject existence of any compartmentalization, at least in G. obscuriglobus (Santarella-Mellwig et al. 2013). The validity of such a view inconsistent with previous published studies needs to be rigorously tested by other approaches.

The phylogenetic distribution of compartmentalization among at least three of the five confirmed PVC superphylum phyla strongly suggests that the common ancestor of the PVC superphylum was also compartmentalized in a way exceptional among bacteria. Recent genomic analysis revealing correlation of both some proteins apparently wholly restricted to the superphylum (Gupta 2008; Santarella-Mellwig et al. 2010) or unusual protein domains conserved in the phyla Planctomycetes, Verrucomicrobia and Lentisphaerae known from ultrastructural studies to contain members with compartmentalized cells, is consistent with this at a molecular level (Kamneva et al. 2012). However in other PVC members such as the compartmentalized anammox planctomycete Kuenenia, one of the most distinct of such domains, DUF1501, is missing, suggesting loss in some lineages and that other domains can also function in compartment formation. Horizontal gene transfer appears to have been important among superphylum members (Kamneva et al. 2012). Development of cell compartments may nevertheless have been an ancient event within the superphylum and one correlated with evolution of specific molecular mechanisms, rather than an invention separately evolving in each phylum. The phylogenetic significance and conservation of compartment organization type is also suggested by the occurrence of double-membrane bounded nuclear compartments only in strains clustering with the Gemmata clade (Wang et al. 2002), suggesting that an ancestor of the Gemmata clade may have also had such compartments. Of the other PVC phyla, cells of chlamydiae and ‘chlamydia-like’ species within the phylum Chlamydiae, and of the uncultured OP3 species, have not yet been examined via cryosubstitution TEM needed to reveal compartments in verrucomicrobia, so we cannot yet exclude occurrence of internal compartments in those groups. Similarly, the newly described anaerobic phylum Lentisphaerae species O. ethanolica shows distinct indications of potential compartments in chemically fixed cells and should be reexamined using cryotechniques (Qiu et al. 2013). Interestingly, Rhabdochlamydia crassificans infecting terrestrial arthropods such as isopods have electron-translucent oblong structures in their cytoplasm and a ‘five-layered cell wall’ which could also be interpreted as including layers and membranes additional to wall and cytoplasmic membrane (Kostanjsek et al. 2004). There have been other hints of potential complex internal structure in the ‘crescent bodies’ observed for Parachlamydia acanthamoeba (Greub and Raoult 2002) that need reexamination for test of complex cell structure with EM cryotechniques rather than classical chemical fixation methods common in chlamydiology. There are even chlamydia-like organisms with developmental elementary body stages in Estrella lausannensis (Lienard et al. 2011) and Criblamydia sequanensis (Thomas et al. 2006) resembling in morphology the prosthecate extension-bearing cells found in phylum Verrucomicrobia such as V. spinosum. The basis for the ‘star’ arms of these stages, whether via exo- or endo-skeletal structures, should be investigated, since it may reveal homologies between chlamydia and verrucomicrobia not yet discovered.

Argument 3—in the planctomycete Gemmata structural complexity is correlated with a functional ability exceptional for bacteria—protein uptake via endocytosis

Cells of G. obscuriglobus among the planctomycetes exhibit the remarkable property unique among the bacteria of ability to incorporate proteins from the external milieu (Lonhienne et al. 2010). This has been demonstrated clearly by use of GFP in the external medium as a marker protein for uptake (Fig. 4). Highlighting this property as an exceptional one for bacteria, and linking it to the compartmentalization of these organisms, is that the mechanism of this uptake involves incorporation only into the paryphoplasm compartment of the cell (Fig. 5). Like the analogous eukaryote process, it is receptor-mediated and energy-dependent, associated with vesicle formation from the cytoplasmic membrane, and correlated with a degradation of incorporated protein. Such degradation is conceivably of nutritional and adaptive value in native microbial communities. At a molecular level, the endocytosis-like protein uptake appears correlated with the presence of homologs of eukaryote MC proteins such as clathrin mentioned above, which can be found via antibodies to one of these proteins to be associated with membrane vesicles from cell fractions positive for incorporated protein (Lonhienne et al. 2010) and also with internal vesicles of intact cells (Santarella-Mellwig et al. 2010). The mechanism involves endosome-like vesicle formation from infolding of the cytoplasmic membrane, a mechanism consistent with existence of outermost membrane in contact with external medium and thus with the concept of a true cytoplasmic membrane distinct from the wholly internal intracytoplasmic membrane in planctomycetes.

Fig. 4
figure 4

Fluorescent micrograph showing uptake of green fluorescent protein (green) by cells of Gemmata obscuriglobus and localization of fluorescent protein in outer regions of the cell consistent with paryphoplasm. Bar is 2 μm. Micrograph courtesy of Dr. Evgeny Sagulenko

Full size image
Fig. 5
figure 5

Transmission electron micrograph of sectioned cell of Gemmata obscuriglobus showing GFP protein taken up after incubation of cells with purified GFP protein, localized via anti-GFP antibody immunogold labeling as indicated by colloidal gold particles (arrowheads), which are located only in the paryphoplasm compartment of the cell. Modified from Fig. 1a in Fuerst and Sagulenko (2010)

Full size image

It is conceivable that such a process is the key to understanding how complex compartmentalization and endomembranes evolved, in a scenario similar to that proposed by De Duve for endomembrane system evolution in early eukaryotes (de Duve 2007). This applies whether the process is analogous or homologous to the eukaryote one. In any case, structural compartmentalization in at least this model planctomycete species is correlated with a cell biology function involving processes restricted to one cell compartment and perhaps dependent on molecular mechanisms restricted to that compartment.

Endocytosis in a planctomycete suggests a perspective on the first type of nutrition which may have led to evolution of eukaryote life-styles and endomembranes. A planctomycete may have evolved a new way to utilize proteins—via endocytosis rather than secreted enzymes and involving development of an endomembrane system.

There is thus suggested a conceivable selective pressure leading to evolution of endomembrane systems characteristic of the eukaryote lineage, whether via the PVC lineage or its ancestor or not. Such a selective adaptation might also link the origin of compartmentalization and the origin of the first eukaryote endomembrane system.

Of course, anammox planctomycetes also possess functional compartmentalization, with the anammoxosome containing not only enzymes important for specific anammox ammonium oxidation but also a membrane-bound ATP synthase consistent with generation of a membrane potential (van Niftrik et al. 2010).

Argument 4—PVC superphylum members can possess true eukaryote homologs, including eukaryote-specific proteins important in cell biology

One dramatic example of a true eukaryote homolog in PVC members is that of what appears to be the closest known homolog of the eukaryote cytoskeletal protein tubulin occurring in members of genus Prosthecobacter in the phylum Verrucomicrobia (Jenkins et al. 2002a, b). These BtubA and BtubB tubulin homologs resemble tubulins of eukaryotes much more closely than the other bacterial tubulin homolog FtsZ, and are capable of forming simple microtubules of five protofilaments in vitro (Pilhofer et al. 2011). Although some analyses have suggested the verrucomicrobial tubulins originated via horizontal gene transfer from eukaryotes, phylogenetic comparison with eukaryote tubulin classes suggests that such a transfer if it occurred must have been a very ancient one, since they branch quite deeply within the eukaryote tubulin clade (Martin-Galiano et al. 2011; Pilhofer et al. 2011). Such trees leave open the alternative possibility of tubulin invention in verrucomicrobia or its ancestors and gene transfer in the bacteria-to-Eukarya direction as a contribution to the ‘tools of eukaryality’.

Another important example of eukaryote homologs in PVC superphylum members is the occurrence of the homologs of eukaryote MC proteins in PVC members but not in most if not all other bacteria (Santarella-Mellwig et al. 2010). Such proteins in eukaryotes play key roles in formation of vesicles in eukaryote endomembrane trafficking e.g. clathrin forming clathrin cages during endocytosis, and the COPI and COPII proteins needed for vesicles budding from ER or Golgi, and may account also for the stabilization of nuclear envelope membrane curvature associated with nuclear pore formation. The homology of such proteins with vesicle complex-forming MC proteins has suggested a mechanism for the ancient origin of MC proteins via common origin with ancient membrane-curving proteins, in the ‘protocoatomer’ hypothesis (Devos et al. 2004; DeGrasse et al. 2009). Additional published examples of potential homologs with eukaryote proteins in PVC members (Staley et al. 2005) remain to be investigated, including homologs of eukaryotic integrins (Jenkins et al. 2002a, b) and remarkably in the case of the ‘Hell’s Gate’ globin in the verrucomicrobial methanotroph Methylacidiphilum, a mammalian type of neuroglobin (Teh et al. 2011). With approaches based on secondary structure similar to those used to reveal the MC protein homologs, many more may be discovered in the ‘hypothetical protein space’.

Argument 5—dominance of hypothetical proteins in planctomycete proteomes

We cannot assume absence of homology with eukaryote-specific proteins in PVC members, especially in planctomycetes where eukaryote-like structure is most pronounced. This is so because hypothetical proteins, those with no known homologs to be found in the genome database, can comprise a substantial proportion of the proteome and may hide structural homologs not clear at primary sequence level. The proportion of ORFs in genomes annotated as hypothetical proteins is substantially higher in planctomycete members than in other bacteria of other major phyla. Genes with function prediction are only 54–60 % of total genes in planctomycete species with complete genomes such as P. limnophilus, P. staleyi and Isosphaera pallida while in a range of non-PVC bacteria annotated via the same protocol such as Rhizobium leguminosarum in phylum Proteobacteria, Pedobacter heparinus in phylum Bacteroidetes and Thermomicromonospora curvata in phylum Actinobacteria between 65 and 72 % of total identified genes have function predictions (Fuerst and Sagulenko 2012). Interestingly, this phenomenon may also apply to some other members of the PVC superphylum, since in L. araneosa in phylum Lentisphaerae, only 49 % of genes have been reported to have predicted function (Thrash et al. 2010). There may be cryptic eukaryote homologs to be discovered in this unexplored ‘hypothetical protein’ space and we cannot exclude the potential existence of such homologs which may only be detectable through analysis at the secondary or tertiary structure level reflecting deep relationships rather than primary sequence matches via local alignment search approaches (Santarella-Mellwig et al. 2010).

Future research programs should explore PVC members from a perspective of considering them exceptions to the bacteria

There is a vast field of study for our emerging PVC superphylum research community regarding PVC species and their cell biology, and one of the most fruitful approaches will be to further investigate the exceptions they display to other bacteria. In ecology we can now explore new modes of nutrition not known in other bacteria e.g. direct protein uptake. This might even point eventually to the selective origin of the exceptional status of these organisms, since these novel nutritional abilities correlated with compartmentalization may well have provided a competitive advantage to the species in which they first evolved.

In cell biology, we need to test how far eukaryote homologies apply, and to explore these organisms experimentally using eukaryote cell biology as a model for what processes, molecules and macromolecular complexes we might predict to be present on this basis. For example, in Gemmata and other planctomycetes we know little of the endomembrane system and the underlying mechanisms of its organization, of ribosome binding to internal membranes and its basis, internal co-translational protein secretion, the function of tubulin-like cytoskeletal proteins in verrucomicrobia, and the functions of the diversity of MC proteins found widely in PVC members beyond Gemmata where they may function in endocytosis. Features of molecular cell biology such as the location of and relation between transcription and translation and the possibility of mRNA processing should be investigated, along with potentially new mechanisms for secretion and transport across internal membranes (Fuerst and Sagulenko 2013). Interestingly, compatible with potential unusual features of its molecular cell biology and translation systems, G. obscuriglobus has a striking specific insertion in its L17 ribosomal protein not found elsewhere in bacteria (Kamneva et al. 2010). We also need ultrastructural and experimental tests of the hypothesis (Fuerst and Webb 1991) of nucleocytoplasmic transport via pores in the nuclear envelope in G. obscuriglobus.

In genomics, we need to explore sequence and structural space of the ‘hypothetical protein universe’ in PVC proteomes—which may be hiding deep homologies with Eukaryotes. Much more comparative genomics needs to be performed with existing and emerging genome sequences, with completion of those remaining only in draft form so far.

Concerning the cell wall and its evolution in these organisms, we need to know much more about the proteinaceous wall of planctomycetes, both its components and how it functions, and what role might be played in planctomycetes by genes homologous with those in other bacteria concerned with synthesis of peptidoglycan and Gram-negative outer membrane components. The marine verrucomicrobia related to the Opitutaceae such as species Cerasicoccus (Yoon et al. 2007a, b, c) Pelagicoccus, (Yoon et al. 2007a, b, c) Puniceicoccus (Choo et al. 2007) and Coraliomargarita (Yoon et al. 2007a, b, c) are apparently unable to synthesize peptidoglycan (as supported by absence of muramic acid and resistance to beta-lactam antibiotics). Via analysis of their cell wall composition and associated synthetic genes, these might form one portal to our understanding of cell wall evolution in PVC organisms and perhaps bacteria in general.

For planning a future research program on the PVC superphylum, I would argue that assuming a working hypothesis of evolutionary homology of eukaryote-like features found in PVC superphylum species with those of eukaryotes is likely to form a more productive basis for discovery than assumption of analogy and convergence without any evolutionary link. This working hypothesis can provide a heuristic logic of discovery for researchers on fundamental cell biology and evolutionary biology of PVC member species. Such a logic involves rigorous experimental tests of the implications of the hypothesis, which is of course capable of being refuted by such experimental data. The alternative however of assuming analogy does not immediately imply a simple road for experimental investigation, since the molecular mechanisms will be likely to be completely distinct from eukaryote pathways and more difficult to unravel. From a strategic point of view in terms of resources available to a so far not adequately funded PVC community, it makes sense to follow an experimental program which promises more fruitful productivity. Already we have seen in the case of the endocytosis protein uptake in G. obscuriglobus the potential results of an experimental approach assuming there may well be cell biology correlates of a eukaryote-like compartmentalized cell structure. One can predict that this will not be the last such result if other cell biology implications of PVC cell structure are seriously tested experimentally rather than merely by computational and genomic analysis. If relationships between PVC members and eukaryotes are ancient, deep or in any case distant, assuming such relationships are indeed genuine, many computational approaches may be difficult and bumpy roads, requiring sophisticated secondary and tertiary structure analysis at the least to reveal relationships. The experimental road poses other problems but is likely to be the one leading to more intelligent searches for genomic and proteomic signals of the PVC superphylum’s evolutionary relevance and even potential homology with eukaryotes.