Various systems of gymnamoebae have been based mostly on the morphological features (Schaeffer 1926; Lepssi 1960), sometimes combined with the nuclear division pattern (Singh 1952; Chatton 1953; Page 1976) and the modes of locomotion (Jahn and Bovee 1965; Jahn, Bovee, and Griffith 1974). All these systems could hardly pretend to reflect phylogenetic relationships among the higher taxa. For convenience, all amoeboid protists with lobose pseudpodia were included in the class Lobosea (Levine et al. 1980; Bovee 1985). Based on the ultrastructure and peculiarities of the life cycle, Page and Blanton (1985) suggested a dichotomy of the classes Lobosea and Heterolobosea. Within the class Lobosea, Page (1987, 1991) recognized four orders of naked amoebae (Euamoebida, Leptomyxida, Acanthopodida, and Loboreticulatida) and groued them in the subclass Gymnamoebia. At least two groups (the class Heterolobosea and the order Euamoebida) were viewed as more or less natural (Page 1987). The phylogenetic relationships within and between these groups, however, remained largely hypothetical becasue of the low resolution of the morphological taxonomy.
One of the most enigmatic points concerns the phylogenetic origin of amoebae. The first molecular data confirmed the polyphyly of amoebae (Clark and Cross 1988), in agreement with the morphology-based distiction between the classes Lobosea and Heterolobosea (Page and Blanton 1985). Comparison of small subunit (SSU) rRNA gene sequences of Acanthamoeba castellanii (Lobosea) and Naegleria gruberi (Heterolobosea) shows that both species branch separately, the first one emerging within a radiation of eukaryotes (Gunderson and Sogin 1986), the second one branching in the middle part of the SSU tree (Baverstock et al. 1989). The independent origin of Naegleria was later confirmed by molecular studies on other Vahlkampfiidae (Hinkle and Sogin 1993). The analysis of SSU rRNA also suggested a lack of any specific relationship between A. castellanii and Entamoeba histolytica (Sogin 1989), leading to the exclusion of Entamoeba from the Lobosea and to the creation of a new parvkingdom of Entamoebia (Cavalier-Smith 1993). Among all examined Gymnamoebia, only Hartmannella vermiformis shows a relationship to A. castellanii (Gunderson, Goss, and Sogin 1994; Weekers et al. 1994). The position of both species in the upper part of the SSU tree was considered as representative for gymnamoebae. However, the SSU rRNA of Vannella anglica, another Gymnamoebia, did not associate with those to the two species (Sims, Rogerson, and Aitken 1999).
The SSU rRNA sequence was also used to examine the phylogeny of other amoeboid protists. In particular, these data showed that the euglyphid filose amoebae (Bhattacharya, Helmchen, and Melkonian 1995) and the anaerobic pelobiontid amoeboflagellate Phreatamoeba balamuthi (Hinkle et al. 1994) have an independent origin. According to Simpson and others (Simpson et al. 1997; Walker et al. 2001), P. balamuthi should be considered to belong to the genus Mastigamoeba; nevertheless, despite their morphological similarity, the SSU sequences of P. balamuthi and Mastigamoeba invertens do not group together. Later, this discrepancy was interpreted as an aritifact because of the heterogenous evolutionary rates (Stiller and Hall 1999). On the other hand, M. invertens was shown to branch among the earliest eukaryotes in the tree based on the gene encoding the largest subunit of RNA polymerase II (RPB1) (Stiller, Duffield, and Hall 1998).
In view of these data, based pricipally on ribosomal DNA sequnces, the polyphyly of amoeboid protists seems to be well established. However, the results of some recent reanalyses of SSU rRNA sequences question the solidity of this hypothesis. Maximum likelihood reanalysis of SSU rRNA sequences (Cavalier-Smith and Chao 1996) showed that E. histolytica, P. balamuthi, A. castellanii, and H. vermiformis branch together, in opposition to the previous results of the same data set. Moreover, in some maximum likelihood (ML) and maximum parsimony (MP) trees, Lobosea appears as a sister group to Archamoebae and Mycetozoa, supporting the inclusion of all the three groups in the phylum Amoebozoa (Cavalier-Smith 1998). The close relationships between gymnamoebae were later confirmed by the study of E. histolytica and Endolimax nana SSU rDNA sequences (Silberman et al. 1996). The lobose amoebae grouped together also in the ML analysis of the SSU rRNA of leptomyxid amoebae (Amaral Zettler et al. 2000), although the limited number of nonamoeboid species used in this study does not allow any general conculsion.
A common origin for Amoebozoa was also proposed based on the analysis of combined protein data (Baldauf et al. 2000). The grouping of Acanthamoeba with Mycetozoa (Dictyostelium and Physarum) is supported by protein sequences, like actin (Bhattacharya and Weber 1997; Philippe and Adoutte 1998) and action-related proteins ARP2 and ARP3 (Kelleher, Atkinson, and Pollard 1995; Schafer and Schroer 1999), as well as by similarities in genome organization between Acanthamoeba and Dictyostelium (Iwamoto et al. 1998). However, the position of gymnamoebae in other proteins trees is extremely variable. For example, E. histolytica branches at the base of the EF-1alpha tree (Baldauf and Palmer 1993) and as a sister group to Euglenozoa in the EF-2 tree (Moreira, Le Guyader, and Philippe 2000), whereas A. castellanii branches with plants in the RPB1 tree (Stiller and Hall 1997). Given that in most cases, the sequences of both amoebae are not available for the same protein, their mutual relationships cannot be inferred.
The System of Gymnamoebia sensu stricto
Traditionally, all naked amoebae possessing lobose pseudopodia are included in the subclass Gymnamoebia that comprises four orders and three incertae sedis families (Page 1987). In our analysis, six independent lineages of gymnoamoebae appear. The majority of the species (11) branch in a clade grouping mostly species from the orders Euamoebida and Leptomyxida. As these species represent the most typical gymnamoebae, we propose considering this clade as a representative for the subclass Gymnamoebia. Among other amoeba lineages, two are composed of more than one species (Acanthamoeba castellanii + Balamuthia mandrillaris and Entamoeba histolytica + Endolimax nana + Phreatamoeba balamuthi), whereas three others are single species lineages (Gephyramoeba sp., Filamoeba nolandi, and Vannella anglica).
The clade Aanthamoeba + Balamuthia corresponds most probably to the order Acanthopodida, which traditionally includes a single family Acanthamoebidae (Sawyer and Griffin 1975; Page 1987). The close relationship between both genera has been demonstrated using rRNA data (Stothard et al. 1998; Amaral Zettler et al. 2000).
From a morphological point of view their grouping is quite unexpected. Balamuthia mandrillaris was initially described as a leptomyxid (Visvesvara et al. 1990). However, more detailed study of B. mandrillaris showed that the species differs fundamentally from other leptomyxids (Gephyramoeba and Leptomyxa) in the morphological, physiological, and antigenic characteristics (Visvesvara, Schuster, and Martinez 1993). At the same time, it has been observed that B. mandrillaris possesses a MTOC similar to those seen in A. castellanii (Visvesvara, Schuster, and Martinez 1993). The presence of this basic cellular feature in both species reconciles somehow the molecular and morphological data. It will be interesting to compare our data with the sequence of Stereomyxa, which also possesses similar MTOC's (Von Benwitz and Grell 1971).
Another well-supported clade of amoeboid protists is composed of E. histolytica + E. nana + P. balamuthi. This clade was shown previously in rRNA trees (Cavalier-Smith and Chao 1977; Cavalier-Smith 1998, 2000; Silberman et al. 1999). The clade exists only if the sequence of E. nana is included. If E. nana is omitted, E. histolytica and P. balamuthi branch separately. This explains why the relationship between P. balamuthi and Entamoebidae has not been suggested by Hinkle et al. (1994). In fact, P. balamuthi is a pelobiont, i.e., a free-living amoeboflagellate that lacks mitochondria and Golgi bodies. It shares with Entamoebidae general features, such as a anaerobic life mode and a remarkably simplified intracellular organization (Silberman et al. 1999), yet no shared derived morphological features have been identifed so far.
Among the other three lineages composed of single species (Gephyramoeba sp., Filamoeba nolandi, and Vannella anglica), the first two are quite unusual gymnamoebae. Based on the morphological characteristics, they are classified, respectively, in the suborder Leptoramosina (order Leptomyxida) and incertae sedis family (Page 1987). In our analyses, both species tend to group together, but their grouping is weakly supported. The case of V. anglica is much more puzzling. Similar to a previous study (Sims, Rogerson, and Aitken 1999), our analysis of the molecular data shows this species as an independent lineage branching in the middle part of the rRNA tree. Yet, the vannellids are quite typical gymnamoebae (Page 1987) and no obvious morphological or ultrastructural feature could distinguish Vannella from the other Gymnamoebia. As the unexpected position of Vannella in the trees cannot be attributed to a significantly fast rate of evolution, other explanation will have to be found.