Molecular Studies of Parasitic Plants Using Ribosomal RNA
Daniel L. Nickrent and R. Joel Duff. Department of Plant Biology, Southern Illinois University, Carbondale, Illinois, USA 62901-6509
Nickrent, D. L. and R. J. Duff. 1996. Molecular studies of parasitic plants using ribosomal RNA. Pp. 28-52. In: M. T. Moreno, J. I. Cubero, D. Berner, D. Joel, L. J. Musselman, C. Parker (eds.), Advances in Parasitic Plant Research, Junta de Andalucia, Dirección General de Investigación Agraria, Cordoba, Spain.
Molecular evolutionary and phylogenetic studies of parasitic flowering plants have been advanced through the use of nuclear and plastid-encoded ribosomal RNA genes. Phylogenetic analysis of all families of Santalales using nuclear 18S rDNA sequences supports the basal position of Olacaceae, the monophyly of Opiliaceae, the distinctiveness of Loranthaceae and Viscaceae, and the sister relationship between Santalaceae (including Eremolepidaceae) and Viscaceae. These data do not support a close relationship between Santalales and holoparasites in Balanophoraceae, Hydnoraceae, and Rafflesiaceae. High nucleotide substitution rates in these plants hinder their placement in the global angiosperm phylogeny. Some support is obtained for the placement of Hydnoraceae with the paleoherbs (Magnoliidae). Analysis of 18S rDNA sequences for ten genera of Scrophulariaceae and Orobanchaceae indicate the latter family is paraphyletic. The holoparasites Orobanche, Harveya, Hyobanche and Lathraea appear on a clade with substitution rates higher than related hemiparasites. Plastid-encoded 16S rDNA was PCR amplified and sequenced from representatives of Balanophoraceae, Hydnoraceae, and Rafflesiaceae, thereby providing preliminary evidence for the existence of a plastid genome. The 16S rRNA sequence of Cytinus ruber (Rafflesiaceae or Cytinaceae) has mutations at 8.6% of the 1497 sites, yet structural integrity (and likely functionality) is maintained. Formal relative rate tests document that Cytinus is more divergent than any of the previously sequenced plant 16S rRNAs. These holoparasitic plants should be viewed as valuable model organisms that can increase our understanding of the mode and tempo of evolutionary change at the molecular level.
Additional key words: Nuclear rDNA, Phylogeny, Evolution
Within angiosperms, haustorial parasitism has evolved independently at least seven times and more likely at least nine times. Parasitic families are restricted to three of the six dicot subclasses of Cronquist (1988): Magnoliidae, Rosidae and Asteridae (Fig. 1). Strangely, true parasitism has apparently never evolved in monocots (Kuijt, 1969). All parasitic angiosperms develop haustorial attachments to their hosts, yet they vary widely in their nutritional dependence. Chlorophyllous hemiparasites can be found in Laurales (Cassytha), Polygalales (Krameria), and all families of Santalales. The evolution and biochemistry of two orders, Solanales (Cuscuta) and Scrophulariales, are of interest because some species are photosynthetic hemiparasites whereas other species have lost photosynthesis and become holoparasitic. Holoparasites represent the most extreme manifestation of the parasitic mode since they lack photosynthesis and must rely upon the host for water, and inorganic and organic nutrients. Four families (sensu Cronquist, 1988) are represented entirely by holoparasites: Lennoaceae, Balanophoraceae, Hydnoraceae, and Rafflesiaceae.
This paper will focus on the use of ribosomal RNA genes in studies of parasitic flowering plants, specifically molecular phylogenetic and molecular evolutionary analyses of several of the groups shown in Fig. 1. The three distinct genomes of plants (nuclear, plastid and mitochondrial) each contain a complement of the ribosomal RNA genes (rDNA). This paper will deal specifically with projects involving plastid 16S and nuclear 18S rDNA. The goal is not only to demonstrate the utility of these molecular markers in documenting phylogenetic history, but also to show how parasitic plants represent unique models that can be used to study molecular evolutionary and genetic processes.
Nuclear rRNA in Parasitic Plants
In eukaryotes, nuclear rDNA loci occur at nucleolar organizing regions (NORs) on one or several chromosomes. The characteristics and evolution of nuclear rDNA has been extensively discussed (Appels and Honeycutt, 1986; Jorgensen and Cluster, 1988; Hamby and Zimmer, 1992) and only the following brief review will be given. Ribosomal cistrons are repeated hundreds to thousands of times in tandem arrays within the genome. The genes that code for 18S, 5.8S and 26S rRNA occur within a cistron, each separated by spacers (Fig. 2A). The intergenic spacer (IGS) occurs between repeat types of each cistron and is composed of a nontranscribed spacer (NTS) bounded by external transcribed spacers (ETS) on each end. The internal transcribed spacer (ITS) exists in two pieces, ITS-1 between the 18S and 5.8S rDNA and ITS-2 between the 5.8S and 26S rDNA. After transcription and processing, the 18S rRNA resides in the small ribosomal subunit and the 5.8S and 26S rRNAs reside in the large ribosomal subunit of the mature ribosome. These rRNA molecules fold into higher-order structures that are directly involved in the translation process.
Sequence similarity between the individual cistrons within a single organism is generally very high, possibly due to unequal crossing over during meiosis, gene conversion, slippage, transposition, and RNA mediated changes (Arnheim, 1983; Dover, 1982; 1987). This homogenizing process that occurs in multigene families has been called concerted evolution (Brown et al., 1972; Zimmer et al., 1980; Sanderson and Doyle, 1993) and ribosomal loci represent an extreme case. This characteristic has also made these genes useful for examining deep phylogenetic divergences (Cedergren et al., 1988; Sogin et al., 1986). The ribosomal cistron is not invariant over its entire length, but is a mosaic of slowly and rapidly evolving regions (Fig. 2A). Variable and conserved domains exist on both 18S and 26S rDNA. The actual utility of 18S rDNA in molecular phylogenetic studies of angiosperms has only recently been realized (Nickrent and Soltis, 1995; Soltis et al., submitted). Spacer regions, such as ITS, have recently attracted much interest since their level of variability is high enough to allow phylogenetic analysis within and among genera (Baldwin, 1995; Nickrent et al., 1994).
Molecular Phylogenetic Studies of Santalales
One of the earliest studies to use 18S rRNA sequences in a phylogenetic analysis of angiosperms examined parasitic Santalales (Nickrent and Franchina, 1990). Using direct rRNA sequencing with reverse transcriptase, sequences from representatives of 10 families were analyzed. Although only three parasitic families were examined, the analyses supported the monophyly of Santalales, the sister group relationship between Viscaceae and Santalaceae and the basal position of Olacaceae within the order. This study also showed that 18S rRNA contained sufficient variation to conduct phylogenetic analyses in angiosperms. Since this initial work, 18S sequences of parasites in Santalales have increased rapidly, mainly owing to the advent of the polymerase chain reaction (PCR). Small-subunit rDNA is readily amplified from genomic DNA samples and can then be purified and directly sequenced. Sampling within Santalales has also steadily improved such that currently genomic DNA exists for representatives of all families of the order and, in some cases, all genera within a family.
The following section will discuss the results of preliminary phylogenetic analyses of Santalales using complete 18S rDNA sequences. The order is here defined as the following families: Olacaceae, Loranthaceae, Misodendraceae, Opiliaceae, Santalaceae and Viscaceae. The holoparasite families Balanophoraceae, Rafflesiaceae and Hydnoraceae are often classified in or near Santalales, however, since these relationships are not clear, these families will be treated separately (see below). The methods used to extract DNA, PCR amplify and sequence 18S rDNA, conduct multiple sequence alignments, and generate minimum-length Fitch parsimony trees (via PAUP - Swofford, 1993) are discussed in Nickrent and Soltis (1995). Figure 3 shows the results of parsimony analysis of 18S rDNA sequences from 62 members of Santalales and several outgroup taxa. Specific results are discussed below.
Olacaceae. The family has traditionally been considered the most primitive of the order based upon the presence of two ovular integuments in some taxa and the presence of both autotrophic and hemiparasitic members. Kuijt (1968, 1969) considered the Olacaceae the "plexus" from which all other families in the order were derived. The combination of primitive and specialized features and the very high number of monotypic genera prompted Sleumer (1984) to suggest the family differentiated early during the Cretaceous, prior to the separation of the continents. The Olacaceae is certainly the most problematic one in the order, for indeed extreme variability can be seen in morphological features such as habit (nonparasitic and parasitic), flower sexual condition (unisexual, perfect), petal fusion (distinct, connate), ovary position (hypogynous, epigynous, perigynous), ovular integuments (0, 1, 2), embryo sac type (monosporic, bisporic), and even cotyledon number (2, 3, 4 and 8). Such variation has prompted the erection (and subsequent subsumption) of numerous splinter families such as Aptandraceae, Cathedraceae, Erythropalaceae, Heisteriaceae, Octoknemaceae, Schoepfiaceae, Scorodocarpaceae, Strombosiaceae, and Tetrastylidiaceae. Sleumer (1935 and amended in 1984) divided the family into three subfamilies: Anacolosoideae (formerly Dysolacoideae), Olacoideae, and Schoepfioideae.
At present only seven of the ca. 28 genera in the family are represented in Fig. 3. All genera except Schoepfia are basal in the order, which is concordant with previous taxonomic systems. The genera do not form a monophyletic clade, hence the family is paraphyletic. Schoepfia screberi Gmelin forms a clade with Misodendron brachystachyum DC and this clade appears basal to the Old World Loranthaceae. This same relationship is supported by analyses of the chloroplast gene rbcL (Nickrent, 1996) that included five genera of Olacaceae. Schoepfia is distinct from other olacaceous genera in possessing aliform-confluent parenchyma (Sleumer, 1984) and by its ratio of tracheid and vessel features (Reed, 1955). Reed (1955) also noted similarities in the pollen of Schoepfia with Santalaceae (a more advanced family of the order). Taken as a whole, it is worth considering possible phylogenetic affinities between Schoepfia and the root parasitic Loranthaceae such as Atkinsonia and Nuytsia. Additional sampling within Olacaceae is required to further address the possible polyphyly of the family. Assistance by colleagues is requested to acquire genera from tropical South America, Africa, and Indomalaya.
Misodendraceae. Misodendron consists of ca. 10 mistletoe species parasitic on Nothofagus of southern South America. The genus is unusual in having wind-dispersed fruits (via enlarged, plumed staminodes). Relatively little published information on phylogenetic affinities of this monogeneric family exists, although Kuijt (1968, 1969) derives the family from Olacaceae. Results of analyses of 18S rDNA (Fig. 3) and rbcL sequences (Nickrent and Soltis, 1995; Nickrent, 1996) show that Misodendron clusters with Schoepfia (Olacaceae), although connected to it by a long- branch. Given this molecular and biogeographic information, Misodendron may represent a relictual taxon that diverged early from loranthaceous or olacaceous stock present on the Gondwanan landmass.
Loranthaceae. Complete 18S rDNA sequences currently exist for 23 of the ca. 75 genera in this family. Results of phylogenetic analyses (Fig. 3) show that the family (with the inclusion of Schoepfia and Misodendron) is monophyletic An apparent feature of the tree is that significant genetic differentiation has occurred between clades composed of New World mistletoes (e.g. Gaiadendron, Ligaria, Psittacanthus, etc.) and Old World mistletoes (Loranthus, Tapinanthus, Amyema, etc.). The two mistletoes endemic to New Zealand (Tupeia and Alepis) show affinity with Lysiana, an Australian genus. Gaiadendron is classified with Atkinsonia in tribe Elytrantheae (Danser, 1933) and shares with it (and Nuytsia) N=12, the base chromosome number for the family. These three genera are considered the most primitive in the family (Barlow, 1983). The 18S rDNA data do not strongly support a basal position in the family for Gaiadendron (see, however, rbcL analyses in Nickrent, 1996). Although we have DNA for the root parasite Nuytsia, we have yet to generate any sequences. Tissue for another root-parasitic genus, Atkinsonia, has yet to be obtained.
By examining branch lengths, it is apparent that fewer mutations exist among loranthaceous than viscaceous genera. The molecular data strongly support the concept of Barlow (1983) that these two mistletoe families are distinct. Furthermore, more mutations occur among Old World than New World genera of Loranthaceae. A number of relationships support current concepts, such as associations between Amyema and Diplatia, Oryctanthus and Dendropemon, etc. Although 18S rDNA does provide some indications of intergeneric affinities, continued sequencing using a faster rate molecule is required to fully resolve relationships in this family. Sequences for rbcL for three genera (Gaiadendron, Moquinella and Tupeia) indicate that this molecule will also provide insufficient phylogenetic signal to address relationships among all genera. Our lab is currently exploring the utility of two chloroplast genes (ndhF and matK) as well as 26S rDNA which contains more rapidly evolving domains (expansion segments).
Opiliaceae. Four of the possible nine genera of this family have been sequenced for 18S rDNA: Agonandra, Cansjera, Champereia, and Opilia. Results of this analysis (Fig. 3) represents the family as a monophyletic clade between the Santalaceae and Loranthaceae. Opiliaceae is one of the most coherent families within the order as reflected by the overall similarity in its wood structure (Reed, 1955). Similarities between Opiliaceae and Santalaceae can be seen in floral morphology (Fagerlind, 1948), embryology (Johri and Bhatnagar, 1960), and haustorial anatomy (Kubat, 1987). Although Hiepko (1979, 1982) published a taxonomic revision of the Old World members of the family, no intergeneric phylogenetic inferences were made, therefore the classification of Sleumer (1935) must be used. Therein, Agonandra (with Gjellerupia) was placed in its own tribe, Agondandreae. The remaining seven genera were classified within tribe Opilieae. rDNA analyses place Opilia, not Agonandra, at a basal position in the family, however, this is supported by only a few steps. In contrast, rbcL analyses (Nickrent, 1996) place Agonandra at the base of the clade in support of the Sleumer (1935) classification.
Santalaceae and Eremolepidaceae. Pilger (1935) divided the family into three tribes, the Anthoboleae (2 genera - Anthobolus and Exocarpos), Osyrideae (= Santaleae - 22 genera), and Thesieae (6 genera). Little subsequent taxonomic work has been conducted on higher-level relationships in the family. This analysis utilized sequences from 11 of the 30 genera of Santalaceae. The family does not form a monophyletic group but a grade that eventually culminates in Viscaceae (Fig. 3). Sequences for only two genera (Osyris and Santalum) were used in the rbcL analysis by Nickrent (1996), however, they formed a monophyletic group (with Eremolepidaceae - see below) sister to Viscaceae. Several relationships are worth noting. Buckleya and Pyrularia, two relictual genera that have representatives in eastern North America and eastern China, form a clade at the base of the family. The morphologically similar north temperate genera Geocaulon and Comandra, form a clade with Thesium. An unresolved polytomy that includes Osyris, Nestronia and Santalum in one clade and Dufrenoya and Dendrotrophe in another is also present.
Three santalalean genera, Antidaphne, Eubrachion and Lepidoceras, were placed in the family Eremolepidaceae by Kuijt (1988) and allied with primitive Loranthaceae. The results of this 18S rDNA sequence analysis places two representatives (Antidaphne and Eubrachion - DNA has yet to be obtained from Lepidoceras) within the Santalaceae (Fig. 3). Eubrachion forms a clade with Exocarpos, the one representative of tribe Anthoboleae, and Antidaphne occupies a position intermediate between Santalaceae and Viscaceae. Results of analyses of rbcL and 18S rDNA sequences (Nickrent and Soltis, 1995; Nickrent, 1996) place both Antidaphne and Eubrachion on a clade with Santalaceae. These results support the statements made by Wiens and Barlow (1971) that Eremolepidaceae is not closely related to Viscaceae and that the three genera are not related to each other based upon karyological and morphological evidence. Embryological features for these two families, summarized by Bhandari and Vohra (1983), support an association between Eremolepidaceae and Santalaceae. Taken together, these results are intriguing since, prior to the present molecular study, the only aerially parasitic Santalaceae were Old World genera such as Phacellaria, Dendromyza, and Cladomyza. The status of Eremolepidaceae ultimately depends upon the placement of the third genus, Lepidoceras, which has resided with viscaceous mistletoes, the Loranthaceae (Kuijt 1968) and later Eremolepidaceae (Kuijt 1988).
Viscaceae. Analyses of nuclear 18S rDNA sequences of the Viscaceae confirms its monophyly, derivation from Santalaceae, and advanced position in the order (Fig. 3). Contrary to Bandahri and Vohra (1983), the family is distinct from Loranthaceae and, relative to it, exhibits increased evolutionary rates. This can be seen by examining branch lengths on Fig. 3, especially for advanced hemiparasites such as Arceuthobium. Sequences from both 18S rDNA and rbcL have been obtained from several representatives from each of the seven genera in the family. The three species of Arceuthobium (Old and New World species) form a clade that is sister to two Australian species of Notothixos. Ginalloa arnottiana Korth. (from Borneo) clusters with two species of Korthalsella (from Hawaii and New Zealand). The next clade is composed of the New World genera Dendrophthora and Phoradendron. The last clade is composed of two species of Viscum. The order of branching for the above major clades is not resolved following bootstrap analysis, hence only the relationships between Phoradendron and Dendrophthora and Korthalsella and Ginalloa are strongly supported.
These results differ in several ways from the relationships proposed by Wiens and Barlow (1971) who used mainly chromosomal and biogeographical information. They proposed two major lines, one composed of Viscum, Notothixos and Ginalloa, the second composed of Phoradendron, Dendrophthora, Korthalsella and Arceuthobium. Given their similar distributions and shared karyotype of 12 or 13 large chromosomes, it is logical to derive a relationship between Ginalloa and Notothixos, as was done by Wiens and Barlow (1971). This relationship is not supported by either rDNA (Fig. 3) or rbcL (Nickrent, 1995; Nickrent and Soltis, 1995; Nickrent, 1996) sequence analysis which both link Korthalsella with Ginalloa. The array of chromosome numbers seen in Viscum (N=10, 11, 12, and 13) suggests that extensive genetic variance exists for this feature. The basal position of this genus following bootstrap analysis (results not shown), although not strongly supported, suggests radiation of the major viscaceous lines from a Viscum-like ancestor. Further studies of the molecular evolution in this family using 26S rDNA or other molecules is required to elucidate relationships in this family.
Molecular Phylogenetic Studies of Rafflesiales and Balanophoraceae
Over the past century, traditional means of classifying Balanophoraceae, Hydnoraceae, and Rafflesiaceae have met with difficulty owing to the extreme reduction and/or modification of morphological structures that have accompanied the evolution of these lineages. Moreover, the features that remain are often enigmatic, i.e. they are so unusual (derived) as to confound assessment of homology via character state transformation series. The disagreement regarding classification of these families can be seen by comparing three current systems for angiosperms. Cronquist (1981, 1988) placed Balanophoraceae in Santalales and classified Hydnoraceae and Rafflesiaceae together in the sister order Rafflesiales. The system of Takhtajan (1980, 1987) was similar in that Hydnoraceae and Rafflesiaceae were placed in Rafflesiales, however, this order was derived from within subclass Magnoliidae, not Rosidae. Takhtajan (1980) stated that Balanophoraceae (in Balanophorales) were "probably near to and derived from Santalales, but the affinity is not fully clear." The system of Thorne (1992) differed little from the above by classifying Balanophorales near Santalales and Rafflesiales in superorder Rafflesianae.
Resolution of phylogenetic relationships in angiosperms has become increasingly refined using molecular markers such as rbcL and 18S rDNA (Chase et al., 1993; Nickrent and Soltis, 1995). For this reason, the analysis of appropriately selected molecular markers holds great promise in placing the holoparasites within the global phylogeny of all angiosperms. Unfortunately, the chloroplast genome of these plants is likely highly modified, hence genes such as rbcL are not available for analysis (see below). As already demonstrated for Santalales, 18S rDNA sequences are appropriate molecular markers for addressing phylogenetic relationships. Complete 18S rDNA sequences have been obtained for representatives of Balanophoraceae, Hydnoraceae, and Rafflesiaceae. Analyses of sequences from these three holoparasite families showed increased nucleotide substitution rates as was first observed in 18S rDNA of Arceuthobium. Using relative rate tests, it was found that the holoparasites were, on average, 3.5 times faster than nonparasitic and hemiparasitic plants (Nickrent and Starr, 1994). Elevated substitution rates in genes normally considered extremely conservative, such as nuclear ribosomal genes, complicates phylogenetic analysis. For example, when these divergent sequences are included in an analysis with nonparasites, the resulting topologies frequently show aberrations such as "long-branch attractions" (Felsenstein, 1978) that artifactually links clearly unrelated taxa with faster rates. When fewer than 20 nonparasitic "outgroup" taxa are used, the parasites with long-branches migrate to the base of the angiosperm clade, intermediate between the angiosperms and the outgroup Gnetales. This same result is seen using parsimony, neighbor-joining and maximum likelihood methods. Further discussion of rates tests will be given in the section "Molecular Evolutionary Studies - Relative Rate Tests."