Acanthamoeba spp. are the most ubiquitous protozoa among environments such as soil, air, freshwater and ocean sediment (De Jonckheere, 1991). Sometimes the amoebae can be carriers or vectors in the dispersion and dissemination of pathogenic microbes such as Legionella spp., Listeria spp., Mycobacterium spp., and Vibrio spp. (Jadin, 1973; Ly and Muller, 1990; Field, 1991; Thom et al., 1992; Kong and Chung, 1996; Chung et al, 1997). Furthermore, several members of the genus Acanthamoeba play direct roles as human pathogens causing life-threatening granulomatous amebic encephalitis and vision-threatening keratitis (Sisson et al., 1995).

Recent molecular systematic studies provided a new classification of the genus Acathamoeba. Stothard et al. (1998) classified 53 strains of Acanthamoeba spp. into 12 Rns sequence types based on sequence analysis of 18S rDNA. The results obtained were inconsistent with some previous species designation and indicated that the taxonomy of Acanthamoeba should be revised. Although 18S rDNA sequence data are useful for identification and differentiation of Acanthamoeba isolates, generation of sequence data is too labor-intensive and expensive for routine identification and classification of Acanthamoeba. Riboprinting, PCR-RFLP analysis of 18S rDNA, is a simpler and less expensive substitute for 18S rDNA sequence analysis (Clark and Diamond, 1997). Chung et al., (1998) applied riboprinting to subgenus classification of Acantamoeba. The results were congruent with those of sequence analysis. However, 18S rDNA of some Acanth-amoeba strains contains an intron, which therefore can generate misleading results (Chung et al., 1998).

Mitochondrial DNA (mtDNA) is a molecular clone within mitochondriate eukaryotic cells like Acanthamoeba. Although variation of the mitochondrial genome size limited the availability of the RFLP analysis of whole mitochondrial genome as a phylogenetic tool, usefulness of mitochondrial small subunit (SSU) rDNA remains unknown. In this study, we evaluated PCR-RFLP of Acanthamoeba mitochondrial SSU rDNA as a phylogenetic tool and compared the results with those from nuclear riboprinting analysis.

Only a single PCR product was amplified from each Acanthamoeba strain (Fig 1). The PCR products of 25 Acanthamoeba strains were approximately 1,550 nucleotides in length as predicted.

Mitochondrial riboprint patterns of 25 Acanthamoeba strains using eight restriction endonucleases are shown schematically in Fig 2. The patterns of A. divionensis AA2 and A. paradivionensis AA1 were identical with all enzymes used. The Chang strain which was previously assigned to A. castellanii had identical riboprints with BH-2, the type strain of A. hatchetti. Vil3, the type strain of A. quina had identical mitochondrial riboprints with the Ma strain of A. castellanii.

Proportions of co-migrating fragments and estimated genetic distance between amoebae strains are shown in Table 2, and the dendrogram constructed based on the estimated genetic distance is presented in Fig 3. Acanthamoeba astronyxis Ray & Hayes and A. tubiashi OC-15C of morphological group 1 (Pussard and Pons, 1977) were the earliest branching strains. Among the remaining strains, the most divergent were A. culbertsoni A-l and A. healyi OC-3A, which belong to morphological group 3 and are known to be highly virulent. Acanthamoeba griffini, with an intron in its nuclear 18S rDNA (Gast et al., 1996; Ledee et al., 1996) branched between virulent strains A-1 and OC-3A and avirulent strains GE-3a, Reich Oak Ridge and PD2S of morphological group 3. Strain P23 which is the neotype strain of A. polyphaga of morphological group 2 formed a cluster with the Oak Ridge strain (the type strain of A. royreba; morphological group 3). RB-F-1, the type strain of A. stevensoni was the earliest branching strain among the remaining 15 strains of morphological group 2. The Castellani, Ma, Vil3, and Jones strains formed a cluster separated from the remaining ten strains. The Nagington, 1652, L3a, Singh, AA2 and AA1 strains formed another cluster.

Analysis of mtDNA, using either sequence or RFLP data, may permit the identification of problematic specimens (Wells and Sperling, 1999). Among mitochondrial genes, the SSU rRNA gene has been widely and successfully applied to investigation of relationships at subgeneric levels as well as among very distantly related groups (Flook and Rowell, 1997). Although RFLP analyses of the mitochondrial genome have been used in phylogenetic and taxonomic studies of Acanthamoeba spp. (Byers et al., 1983; Yagita and Endo, 1990), the results were somewhat ambiguous partly because of the quite difference of genome size among strains. Recently, riboprinting of nuclear 18S rDNA has been widely applied to classification of morphologically indistinguishable protozoa including Acanthamoeba and recognized as an inexpensive and rapid tool for taxonomic and phylogenetic studies (Clark et al., 1995; Clark and Diamond, 1997; Chung et al., 1998). However, some strains of Acanthamoeba have introns in their nuclear 18S rDNA (Stothard et al., 1998) and identification of these strains can be problematic with nuclear riboprinting. PCR-RFLP analysis of mt SSU rDNA (mt riboprinting) has not been used previously for phylogenetic and taxonomic studies of Acanthamoeba.

In this study, we obtained similar but not identical results from mt riboprinting as we obtained previously with nuclear 18S riboprinting. A dendrogram constructed in this study (Fig 3) coincided generally with the grouping of Pussard and Pons (1977) based on the morphological features of cysts. Some of the results strongly support the observations of Chung et al. (1998) based on nuclear 18S riboprinting. For example, AA2 and AA1, the type strains of A. divionensis and A. paradivionensis respectively, showed the identical mt riboprints with all restriction enzymes used. Both strains were isolated simultaneously from soil of France by the same group of investigators (Pussard and Pons, 1977), and differed slightly in mor-phology. Furthermore, they showed the same mt genome RFLP, alloenzyme (unpublished data) and nuclear riboprint patterns (Chung et al., 1998). As Chung et al. (1998) proposed, both strains should be a species and closely related with the Singh strain, the type strain of A. rhysodes. Also, the Chang strain, previously assigned to A. castellanii revealed the identical mt riboprints with BH-2, the type strain of A. hatchetti. Both strains were pathogenic for mice (unpublished data) and were found to be closely related based on nuclear riboprinting. In addition, these strains had lots of comigrating DNA fragments on mtDNA RFLP analyses and revealed very similar alloenzyme patterns (unpublished data).

In contrast to our previous studies, L3a, the type strains of A. lugdunensis, which clustered with the Castellani, Vil3, Jones, and Ma strains in a dendrogram reconstructed by 18S riboprinting, formed a cluster with the Nagington, 1652, Singh, AA2, and AA1 strains here. The Neff, SH621, and P23 strains also branched at different positions in this dendrogram. Estimated divergence between A. astronyxis and A. tubiashi was lower in this study than in 18S riboprinting analysis. These differences may result from a difference in evolutionary rate between the mitochondrial and nuclear genes.

In spite of the likely difference in evolutionary rate between the mitochondrial and nuclear genomes, similar results with mt and nuclear riboprint patterns were observed with several strains assigned to different species. These results suggest that the taxonomic status of some species of the genus Acanthamoeba should be revised. This study has shown that mt riboprinting is another tool for the phylogenetic study of Acanthamoba. In addition, the mt riboprinting confirms the basic features of the evolutionary history of the genus as revealed in studies of the 18S rDNA. One important advantage is that it avoids problems caused by isolates with introns within nuclear 18S rDNA.