J Mol Evol (2005) 60:434–446 DOI: 10.1007/s00239-004-0152-2

Divergent Histories of rDNA Group I Introns in the Family

Dawn Simon,1 Jessica Moline,1 Gert Helms,2 Thomas Friedl,2 Debashish Bhattacharya1

1 Department of Biological Sciences and the Roy J. Carver Center for Comparative Genomics, University of Iowa, 312 Biology Building, Iowa City, IA 52242-1324, USA 2 Albrecht-von-Haller-Institut fu¨ r Pflanzenwissenschaften, Abteilung Experimentelle Phykologie und Sammlung von Algenkulturen, Universita¨ tGo¨ ttingen, 37073 Go¨ ttingen, Germany

Received: 14 May 2004 / Accepted: 7 November 2004 [Reviewing Editor: Dr. W. Ford Doolittle]

Abstract. The wide but sporadic distribution of group Introduction I introns in protists, plants, and fungi, as well as in eubacteria, likely resulted from extensive lateral Group I introns are autocatalytic RNAs that encode transfer followed by differential loss. The extent of conserved primary and secondary structures essential horizontal transfer of group I introns can potentially for self-splicing (Kruger et al. 1982; Cech 1985). The be determined by examining closely related species or conserved RNA elements are readily aligned and can genera. We used a phylogenetic approach with a large be used for phylogenetic reconstruction to uncover data set (including 62 novel large subunit [LSU] the evolutionary history of these sequences (e.g., rRNA group I introns) to study intron movement Bhattacharya et al. 1994; Hibbett et al. 1996; Haugen within the monophyletic lichen family Physciaceae. and Bhattacharya 2004; Haugen et al. 2004a, b, Our results show five cases of horizontal transfer into 2005). Group I introns are distributed in organellar homologous sites between species but do not support and ribosomal (r)DNA in protists, plants, and fungi, transposition into ectopic sites. This is in contrast to as well as in eubacteria (Turmel et al. 1993; Bhat- previous work with Physciaceae small subunit (SSU) tacharya et al. 1994; Gargas et al. 1995; Bhattacharya rDNA group I introns where strong support was 1998; Nishida et al. 1998; Schroeder-Diedrich et al. found for multiple ectopic transpositions. This dif- 1998; Perotto et al. 2000). Their scattered distribution ference in the apparent number of ectopic intron in some lineages has been interpreted as evidence for movements between SSU and LSU rDNA genes may rampant lateral transfer of group I introns (DePriest in part be explained by a larger number of positions and Been 1992; Hibbett 1996; Cho and Palmer 1999). in the SSU rRNA, which can support the insertion Remarkable examples of horizontal movement be- and/or retention of group I introns. In contrast, we tween distantly related taxa (e.g., fungi to plants, suggest that the LSU rRNA may have fewer accept- [Vaughn et al. 1995], red algae to brown algae able positions and therefore intron spread is limited [Bhattacharya et al. 2001], algae to amoebae [Tunnel in this gene. et al. 1995]) have been described, but the scarcity of these events suggests that they may be relatively Key words: rRNA group I introns — — infrequent or poorly sampled until now (for excep- Physciaceae — Large subunit rRNA — Small subunit tion, see Cho et al. 1998). In other cases, detailed rRNA phylogenetic analyses have provided strong evidence for intron vertical inheritance over millions of years with few losses (e.g., Bhattacharya et al. 1994; Be- sendahl et al. 2000; Simon et al. 2003; Haugen et al. Correspondence to: Debashish Bhattacharya; email: dbhattac@ 2004a). It therefore seems clear that group I introns blue.weeg.uiowa.edu experience a variety of evolutionary fates, which are 435 likely dependent upon a number of factors (e.g., et al. 2003). This suggests that the distribution in the autocatalytic ability [see Haugen et al. 2004a], selec- rRNAs of the highly structured and relatively large tive pressure [see Goddard and Burt 1999], ecological (e.g., 180- to 400-nt) group I introns may reflect factors [see Friedl et al. 2000; Haugen et al. 2005]). divergent probabilities of insertion, efficient splicing, The relative roles of vertical and horizontal trans- and loss. Here we use phylogenetic methods and a mission in the evolution of group I introns remain an large data set containing 62 novel LSU rDNA group important but poorly understood aspect of the biol- I introns from 23species to study their horizontal ogy of mobile elements. transfer within the monophyletic Physciaceae with The rapidly accumulating data suggest that in or- the goal of contrasting these data for SSU and LSU der to gain a more realistic picture of the significance rDNA introns. of horizontal transfer in group I intron evolution, it is important to focus on lower taxonomic levels, such as species and/or genera, and ultimately within and be- Materials and Methods tween populations. By narrowing the taxonomic scope, we can focus on a small window of evolu- DNA Extraction, PCR, and Sequencing tionary time and sample densely within the target Total genomic DNA was extracted from lichen in the ascomycete group. There have, however, been relatively few family Physciaceae as described previously (Helms et al. 2001). The examples in which group I intron horizontal transfer PCR conditions were as follows: 95C for 4 min, followed by 35 has been studied in detail among closely related spe- cycles of 95C for 30 s, 50C for 30 s, and 72C for 2 min. cies (exceptions include Holst-Jensen et al. 1999; Amplification concluded with a 10-min extension period at 72C. Friedl et al. 2000; Nikoh and Fukatsu 2001). Several sets of primers were used to screen the LSU rDNA for group I introns. The primer sequences are as follows (the numbers Lichens are a symbiosis of photosynthetic cells refer to the corresponding position in the E. coli LSU rRNA); 5¢-40 (photobiont) and fungi (mycobiont). Whereas both (5¢CTCGCATCGATGAAGAACGCA3¢), 3¢-148 (5¢AATGACR lichen bionts often have rDNA group I introns, there CTCAAACAGGCATG3¢), 5¢-148 (5¢CATGCCTGTTTGAGYG is no evidence until now to suggest a close evolu- TCATT3¢), 3¢-562 (5¢TTGGTCCGTGTTTCAAGAC3¢), 5¢-562 tionary relationship between these sequences. This (5¢GTCTTGAAACACGGACCAA3¢), 3¢-1245 (5¢ACCACCAAG ATCTGCACTAGAG3¢), 5¢-1252 (5¢GATCTTGGTGGWAGT suggests that horizontal transfer may not occur be- AGC3¢), 3¢-1825 (5¢GAGCACTGGGCAGAAATCAC3¢), 5¢-1825 tween the mycobiont and the photobiont (Friedl et al. (5¢GTGATTTCTGCCCAGTGCTC3¢), 3¢-2252 (5¢TTTAACAGA 2000; Bhattacharya et al. 2002). The fungal compo- TGTGCCGCC3¢), 5¢-2252(5¢GGCGGCACATCTGTTAAA3¢), nent of the Physiaceae family of lichens comprises 3¢-2746 (5¢GATTCTGRCTTAGAGGCGTTC3¢), and 3¢-2904 one of the most group I intron-rich lineages known (5¢ACAAAGGCTTAATCTCAG3¢). The primers were designed to ensure coverage of the entire LSU rDNA. The following primer (Bhattacharya et al. 2000, 2002), making them a good pairs were used: 5¢-40/3¢-148, 5¢-148/3¢-562, 5¢-562/3¢-1245, 5¢-1252/ model to address many questions regarding group I 3¢-1825, 5¢-1825/3¢-2252, and 5¢-2252/3¢-2746 (or 3¢-2904). If the intron evolution. Previous work in our lab has fo- PCR products were larger than the expected size (based on the cused on the small subunit (SSU) rDNA gene in the corresponding Saccharomyces cerevisiae coding region), they were Physciaceae mycobiont and has identified multiple purified and directly sequenced. Alternatively, the PCR products were cloned into the pGEM-T vector (Promega) prior to cases of intron lateral transfer between neighboring sequencing. Both strands were sequenced using an ABI 3100 and distant ectopic rDNA sites (Bhattacharya et al. automated sequencer (Applied Biosystems) at the Center for 2002). In this paper, we expand our analyses to the Comparative Genomics at the University of Iowa. large subunit (LSU) rDNA in this taxonomic group to address one critical question: Are there significant differences in the phylogeny of group I introns in the Reconstruction of Group I Intron Trees SSU versus the LSU rDNA gene? Specifically, we are interested in understanding whether the observed A total of 82 group I intron sequences were included in the phy- logenetic analyses. These sequences were both novel (62) and ex- differences in the folding property and structure of tracted from GenBank (20). Intron secondary structures were the rRNAs correlate with different intron evolution- constructed manually using previously published foldings as a ary histories. It is now well known that the SSU guide (Michel and Westhof 1990; Bhattacharya et al. 1994). The ribosome is composed of discrete structural domains less well-conserved peripheral regions of the structure were esti- (Ramakrishnan and Moore 2001). In contrast, the mated using m-fold V3.1 (Zuker 2003). The secondary structure shown in Fig. 1A illustrates the results of this process. Juxtaposi- LSU ribosome is more monolithic with a significant tion of conserved secondary structure elements offers the only fraction of buried RNA (Ban et al. 2000). Perhaps viable approach to aligning highly divergent group I intron se- reflecting this difference, the distributions of group I quences (Haugen et al. 2005). Whereas many introns at the same introns in the primary, secondary, and tertiary site can be machine-aligned, this process is generally not useful structure of the SSU and LSU rRNAs differ, with the when introns from multiple different insertions sites are compared (Haugen et al. 2004b). In our alignment, BioEdit v5.0.9 (Hall 1999) introns in the SSU rRNA having a more uniform was used to manually align and delimit conserved secondary distribution and those in the LSU rRNA being more structure elements (e.g., P3, P7; see Fig. 1). The final data set highly clustered (Jackson et al. 2002; Bhattacharya contained 122 aligned characters; 118 of these sites were parsimony 436

Fig. 1. Intron secondary structure models. A Secondary structure each of the genic positions analyzed in this study. These structures of intron at position L779 in Physcia aipolia, drawn according to were used to guide the intron alignments. The regions denoted with conventions of Golden et al. (1998). The exon sequences are shown dashed lines indicate structural variability between introns at this in lowercase, whereas nucleotides encoding the catalytic intron are position. For example, at L779 some introns have an extension at in capitals. The conserved paired elements P1–P9 are also labeled. P2 (P2.1), whereas others do not. Note how the 3regions of intron B Schematic figures summarizing the intron secondary structure at variability are limited to the peripheral loops.

informative (PI). This alignment is available at http://www.biol- sites, and transition/transversion ratio = 2) to calculate bootstrap ogy.uiowa.edu/debweb/. support values (2000 replications). By using these two approaches, Hierarchical likelihood ratio tests were done to identify the we had bootstrap estimates from a parameter-rich and a parame- best-fit model for subsequent analyses (MODELTEST v3.06; Po- ter-poor model. This is important for highly divergent data sets, sada and Crandall 1998). This approach identified the general time because in these cases more complex models (i.e., often identified reversible model (Rodriguez et al. 1990) with estimations of by likelihood ratio tests) can retrieve the ‘‘wrong answer’’ (Posada nucleotide frequencies (A = 0.1855, C = 0.2896, G = 0.3172, and Crandall 2001b). In addition, the ability of model selection T = 0.2077) and the shape parameter of the gamma distribution methods to choose the correct model is compromised when, as is (G = 0.8797) to accommodate rate variation across sites as the the case here, the number of phylogenetic characters is small (Po- best-fit model. Bayesian analysis (MrBayes V3.0b4; Huelsenbeck sada and Crandall 200la). A third set of bootstrap values (500 and Ronquist 2001) using the GTR + G model was performed with replications) was obtained using unweighted maximum parsimony. the intron data set. Metropolis-coupled Markov chain Monte Ten heuristic searches with random-addition-sequence starting Carlo (MCMCMC) from a random starting tree was initiated in trees (10 rounds) and tree bisection–reconnection (TBR) branch the Bayesian inference and run for 2 million generations. Four rearrangements were done to find the optimal parsimony tree. Best- chains (one heated, three cold) were run simultaneously and sam- scoring trees were held at each step. In all cases, gaps were treated pled every 100 generations. After discarding the first 50,000 trees as missing data. (‘‘burn-in’’), a consensus phylogeny was made with the remaining A selection of group I introns from commonly occupied posi- trees. Posterior probabilities for each node were inferred from this tions in the SSU rDNA, as well as from all LSU rDNA positions consensus tree. Two-thousand neighbor-joining (NJ) bootstrap described in this paper, were used to infer a second phylogeny. An replicates (Felsenstein 1985) were analyzed using the GTR + G alignment of 122 characters was submitted to a pairwise distance model. In addition, we used the single parameter Jukes–Cantor analysis using the Jukes–Cantor model. Missing data and gaps model (equal nucleotide frequencies, equal divergence rates across were excluded from each pairwise comparison. This distance matrix 437 was then used as input for the NJ tree building method. Bootstrap The single exception was the 779 intron data set for which a non- analysis (2000 replications) was done using both the Jukes–Cantor Physciaceae representative is not known. In this case, the root was model and the GTR + G model (nucleotide frequencies: placed on the longest branch. A = 0.1859, C = 0.3043, G = 0.3038, T = 0.1855; G =11.0255). To test the topological congruence of the intron and host trees, In addition, a third set of bootstrap values was obtained using alternative intron tree topologies were created with MacClade unweighted maximum parsimony (2000 replications). Finally, V4.05 (Maddison and Maddison 2002) either by generating all Bayesian analysis was done as described above. All parsimony and possible trees (i.e., for introns at sites L1090 and L2563, we studied distance phylogenetic analyses were implemented with PAUP* 15 topologies) or by positioning the group of interest on every V4.0b10 (Swofford 2003). The evolutionarily distantly related IE other branch in the tree. The site-by-site likelihoods for these pools subgroup introns (Suh et al. 1999; Bhattacharya et al. 2001) were of trees were calculated using PAML V3.13 (Yang 1997) and the used to outgroup-root the phylogenies. GTR + G model with the default settings. Thereafter, the approximately unbiased (AU)) test was implemented using CONSEL V0.1f (Shimodaira and Hasegawa 2001). Comparison of Intron and Host Trees

To more closely examine the evolution of the group I introns in this Ribosome TertiaryStructures study, subsets of the intron data set were assembled for separate analyses. Each rDNA position with at least four Physciaceae in- The structures were obtained from the Protein Data Bank and trons at that site was subjected to further study. Because introns at manipulated using DS Viewer Pro 5.0 (Accelyrs). All known Phy- each position are more closely related to each other than to those sciaceae intron positions were mapped onto these structures. For from other positions, it was possible to refine the individual group I the SSU ribosome, the Thermus thermophilis 30S (1J5E [Protein intron alignments and increase the number of characters for the Data Bank: http://www.rcsb.org/pdb/]) was used, whereas the phylogenetic analyses. The final LSU rDNA (L) intron data sets Haloarcula marismortui 50S (1JJ2) was utilized for the LSU ribo- included the following number of characters: L779 (144 total, 56 some (Wimberly et al. 2000; Klein et al. 2001). PI), L798 (166 total, 102 PI), L1025 (126 total, 67 PI), L1090 (130 total, 27 PI), L1094 (175 total, 86 PI), L1921 (148 total, 76 PI), L2449 (135 total, 54 PI), and L2563 (136 total, 31 PI). As described Results and Discussion above, MODELTEST was used to determine the best-fit model for each data set. They were as follows: L779 (TVM + G), L798 (K80 + G), L1025 (TrN + G), L1090 (K81 + G), L 1094 (GTR + G), Diversityof LSU rDNA Group I Introns in L1921 (TrN + G), L2449 (K81uf + G), and L2563(HKY + G). Physciaceae Fungi Maximum likelihood (ML) analyses were done for each data set using PAUP* and the appropriate model. Starting trees were built In order to study the evolution of group I introns stepwise (10 random additions) and optimized by TBR. Minimum within the Physciaceae, the LSU rDNA gene from 53 evolution (ME) and NJ bootstrap analyses (2000 replications) were also done using the specified evolutionary models and heuristic species was screened using PCR for the presence of searches with 10 random additions for the ME approach. Bayesian introns. At least one intron was found in 23taxa. A phylogenetic inference was also done as described above. total of 62 introns were identified at 11 distinct rDNA To construct the host trees corresponding to the intron data positions. Of these, there was one novel insertion site sets, LSU rDNA and ITS sequences from each species were con- (L779). All other introns have previously been catenated. In each case, the entire ITS plus 5.8S rRNA was used; however, the chosen region of the LSU rRNA genes was variable reported in other species (Cannone et al. 2002). based on the availability of sequence data. For example, for the 798 The results of the intron screen are summarized in host tree, the GenBank accessions for the outgroup taxa (Grove- Table 1. sinia pyramidalis and Botrytis tulipae) were lacking about 500 nt in Regarding our PCR approach for identifying in- comparison to the Physciaceae species. In addition, because LSU tron-containing lichen fungi, it should be noted that rDNA sequencing was focused on intron discovery, the same re- gion of this gene was not available for all Physciaceae species. The rDNA occurs in tandem repeats. Reported gene copy region of the LSU rDNA used in the construction of the host numbers in fungi range from 60 in the basidomycete phytogeny generally corresponds to the PCR primers used to am- Coprinus (Cassidy et al. 1984) to 220 in the ascomy- plify the intron (i.e., for introns 779, 798, 1025, 1090, and 1094, the cete Neurospora (Russell et al. 1984). It has previ- region spanning 5¢-562/3¢-1245). The total number of characters in ously been suggested that rDNA sequences in these data sets was as follows: L779 (1025 total, 121 PI), L798 (657 total, 122 PI), L1025 (995 total, 97 PI), L1090 (817 total, 38 PI), mushroom-forming fungi (homobasdiomycetes) are L1094 (1061 total, 94 PI), L1921 (657 total, 102 PI), L2449 (658 heterogeneous (Hibbett 1996; Lickey et al. 2003). total, 37 PI), and L2563 (662 total, 19 PI). MODELTEST resulted Because PCR preferentially amplifies small frag- in the identification of the following best-fit evolutionary models: ments, the presence of introns may go undetected if L779 (SYM +I + G), L798 (SYM + G), L1025 (SYM + G), the rDNA copies containing these sequences in taxa L1090 (SYM + G), L1094 (SYM +I + G), L1921 (TrNef +I + G), L2449 (SYM + I+ G), and L2563(SYM + G). Again, ML occur in low numbers. In this case, our screening trees were made with each data set as described above and we also procedure would identify a dominant rDNA frag- implemented ME and NJ bootstrap analyses (2000 replications ment that lacks introns but rare intron-containing each). The Bayesian inference was done as described above except alleles could possibly go undetected using an ethidi- that the data were divided into two partitions: the first corre- um bromide-stained agarose gel. For this reason, we sponded to ITS sequences and the second to the LSU rDNA. Site- specific rates were calculated for each partition and incorporated chose not to unambiguously identify taxa as being into the analysis. Outgroup rooting was done in each case using a intron-less or to reach conclusions about absolute non-Physciaceae species containing an intron at the same position. rates of intron loss and gain in the studied species. 438

Table 1. Taxa used in this study: GenBank accession numbers are given for all sequences included in the phylogenetic analyses; novel intron sequences determined in this study are in boldface

Taxon Family Introns rRNA accession No. ITS accession No.

Chaetothyriales Exophiala calicioides Herpotrichiellaceae S1506 AB007686 — Eurotiales Paecilomyces tenuipes Trichocomaceae L2066 AB044642 — Metarhizium anisopliae Trichocomaceae L2449 AF197124 — jenensis L1025 AF279391 — Helotiales Grovesinia pyramidalis Sclerotiniaceae L798 AJ226081 Z81433 Botrytis tulipae Sclerotiniaceae L798 AJ226078 Z99668 Hypocreales Cordyceps prolifica Clavicipitaceae L1921 AB044640 AB027370 Cordyceps bassiana Clavicipitaceae L1921 AB044638 — Cordyceps kanzashiana Clavicipitaceae L1921 AB044639 AB027371 Cordyceps sp. Clavicipitaceae L2563AB044641 — Metarhizium anisopliae Clavicipitaceae S943AF487276 — Tolypocladium inflatum Clavicipitaceae S1199 AB044634 — Fusarium solan f. xanthoxyli Nectriaceae S1199 AF150492 — Lecanorales Acarospora complanata Acarosporaceae L800, L1094 AF356654 — Lecanora dispersa Lecanoraceae S516, S788, S1516 L37734 — Nephroma arcticum Nephromataceae L800 AF286828 — Neofuscelia pulla L1090 AJ421433 AY037005 Melanelia stygia Parmeliaceae L1094 AJ421434 AF115763 Massalongia carnosa Peltigeraceae L800 AF286827 — Amandina punctata Physciaceae L1094 AY773918 AF540492 L2449 AY773934 runcinata Physciaceae L1094 AY773919 AJ421249 Buellia aethhalea Physciaceae L1921 AY773927 AF540496 Buellia alboatra Physciaceae L1090 AY773915 AF408677 L1921, L2066 AY773928 L2449 AY773933 Buellia geophila Physciaceae L779, L1025, L1090 AY773917 AF540499 L2563 AY787754 Buellia georgei Physciaceae L779, L798, L1025 AY773910 AJ421416 Buellia griseovirens Physciaceae L2449 AY773935 AF540500 Buellia muriformis Physciaceae L1025 AY773925 AF540501 Buellia penichra Physciaceae L2449 AY773936 AF540503 Dimelaena oreina Physciaceae L2563 AY773940 AJ421417 Diplotomma epipolium Physciaceae L779, L1090 AY773916 AF540509 Diplotomma epipolium Physciaceae S1210, S1516 AJ506969 — Phaeophyscia ciliata Physciaceae S1516 AF224457 — Phaeophyscia orbicularis Physciaceae L1090 AY773926 AF540528 Physcia aipolia Physciaceae L779, L798 AY773914 AY787753 L1921, L2066 AY773929 Physcia dimidiata Physciaceae S1210 AJ507615 — Physcia stellaris Physciaceae L779, L798, L1094 AY773912 AJ421421 Physcia stellaris Physciaceae S516 AJ421421 — Physcia tenella Physciaceae L779, L798, L1094 AY773908 AF540538 Physconia distorta Physciaceae L798, L1025, L1094 AY773913 AF540522 L1921 AY773930 Physconia enteroxantha Physciaceae L779, L798, L1025 AY773909 AF540523 L1921 AY773931 L2449 AY773937 Physconia grisea Physciaceae L779, L1025, L1094 AY773921 AF540524 Physconia perisidiosa Physciaceae L798, L1094 AY773911 AF540525 Physconia perisidiosa Physciaceae S788, S1210 AJ421689 — Rinodina atrocinerea Physciaceae L779, L1094 AY773923 AF540544 L1921, L2069 AY773932 Rinodina cacuminum Physciaceae S516, S788 AJ421690 — Rinodina capensis Physciaceae L779, L1025 AY773924 — L2449, L2563 AY773938 Rinodina milvina Physciaceae L779, L1025, L1094 AY773922 AF540546

(Continued) 439

Table 1. Continued.

Taxon Family Introns rRNA accession No. ITS accession No.

Rinodina tunicata Physciaceae L800, L1025, L1094 AY773920 AF540551 L2449, L2563 AY773939 Orbiliales Arthrobotrys superba Orbiliaceae S1506 U51949 — Phyllachorales Verticillium dahliae Mitosporic Phyllachorales S943AY056821 — Protomycetales Protomyces pachydermus Protomycetaceae S1506 D85142 — Saccharomycetales Arxula adeninivorans Dipodascaceae L2449 Z50840 — Trichotheliales guentheri L1090 AF279405 — Xylariales Xylaria polymorpha Xylariaceae S943AB014043— incertae sedis Coccotrema pocillarium Coccotremataceae L1025 AF274093AF329167 Basidiomycota Bensingtonia thailandica Agaricostilbomycetidae S1199 AB040116 —

Characterization of LSU rDNA Group I Introns in the Bayesian support. Group I introns at the remaining Physciaceae three positions (L800, L1921, L2066) do not form monophyletic lineages and there is poor support for The Physciaceae introns in this study ranged in size internal splits in these subtrees. This phylogeny also from 177 to 431 bp, with a mean size of 233 bp. None shows good support for the monophyly of the IE of the introns contain homing endonuclease (or other subgroup introns (L2066, L2069, L2249, L2563). significant) open reading frames (see Haugen et al. Their monophyly indicates that IE introns represent a 2004b). Group I introns have traditionally been distinct evolutionary lineage, a result that is consis- classified into subgroups based on the presence or tent with numerous other studies (Bhattacharya et al. absence of particular paired RNA elements (Michel 2001; Bhattacharya et al. 2002; Haugen et al. 2004). and Westoff 1990). Most of the introns we have The introns at position L1921 in the Physciaceae and found could be classified as members of either the IC Cordyceps spp. are paraphyletic, possibly indicating or the IE subgroups, however, those at position independent insertions of evolutionarily divergent L1921 shared characteristics of both subgroups. introns at this rDNA site. This paraphyly is, however, Figure 1A shows the secondary structure of a typical not statistically supported here. Nonetheless, there is intron identified in this study. This particular ribo- support for independent origins based on secondary zyme can be thought of as a ‘‘minimal intron,’’ con- structure differences. The Cordyceps spp. introns can taining all of the essential structural elements (P1–P9) clearly be classified as belonging to secondary struc- but lacking extensions in the terminal loops. How- ture subtype IC, whereas the Physciaceae introns ever, many of the introns in this study contain have characteristics of IC and IE introns (see above extensions of one or more elements. This is variable and Nikoh and Fukatsu 2001). The finding that in- both within and between introns at different positions trons at each position are more related to each other (see Fig. 1B). than to those at other sites remains true even when a broad sample of SSU rDNA introns is included in the Phylogeny of Physciaceae LSU rDNA Group I Introns analysis (see Fig. 3). The lack of support for many nodes on these trees is a consequence of a small The Bayesian 50% majority rule consensus tree pre- number of characters, large number of taxa, and the sented in Fig. 2 generally supports the monophyly of high rate of divergence in group I introns. Lack of introns at each rDNA insertion site. This result has bootstrap support, particularly at the deep splits, is a been found in a number of previous studies (e.g., typical result for group I intron phylogenies (Bhat- DePriest and Been 1992; Bhattacharya 1998; Nikoh tacharya et al. 1994, 2002; Nikoh and Fukatsu 2001; and Fukatsu 2001, Bhattacharya et al. 2002; Haugen Lundbolm et al. 2004). et al. 2005). There is strong support for the mono- phyly of the L779, L1090, and L1094 clades, whereas Comparison of Host and Intron Trees there exists moderate support for the introns at sites L1025, L2249, and L2563. Introns at position L798 To better understand the dynamics of group I intron are monophyletic but have neither bootstrap nor inheritance, we examined the evolution of introns at 440

Fig. 2. Majority-rule consensus tree inferred from a Bayesian Cantor models) bootstrap analysis (2000 replications each), analysis (GTR + G model) of LSU rDNA group I introns from the whereas those below the branches result from an unweighted Physciaceae and other fungi. The branch lengths are proportional maximum parsimony bootstrap analysis (500 replications). Only to the number of substitutions per site (see scale). This tree was bootstrap support values ‡60% are shown. The thick branches rooted on the branch leading to the IE subclass introns. The values indicate clades that have ‡95% posterior probability in the above the branches show the results of NJ (GTR + G/Jukes– Bayesian inference. each site. Hence, direct comparisons were made of significance of topological differences between the the host and intron ML trees for each LSU rDNA intron and the host trees was tested using the AU test intron position. For some intron positions, only two (Shimodaira and Hasegawa 2001; significance at p < or fewer Physiaceae introns are known, rendering 0.05). The results of this analysis are shown in Fig. 4 unnecessary these comparisons. In total, 8 of the 11 (left side). For four of the intron data sets (L798, intron positions were examined in this manner. The L1090, L1921, and L2563), we found no significant 441

Fig. 3. A phylogeny of LSU and SSU rDNA group I introns inferred with the NJ method and the Jukes– Cantor model. The tree is rooted on the branch leading to the IE introns. The branch lengths are proportional to the number of substitutions per site (see scale). The numbers above the branches are NJ (Jukes–Cantor/ GTR + G) bootstrap values (2000 replications), whereas those below the branches result from an unweighted maximum parsimony bootstrap analysis (2000 replications). Only bootstrap values ‡60% are indicated. The thick branches indicate clades that have ‡95% posterior probability in the Bayesian inference. differences between the host and the intron trees at L779 have not yet been found in taxa outside of the (Fig. 4). However, in some cases this may reflect poor Physciaceae (Cannone et al. 2002). It is therefore taxon sampling, leading to the absence of robust reasonable to assume that introns have been inserted phylogenetic support for particular nodes in the in- at this position after the origin of the Physciaceae. tron tree, rather than true vertical inheritance (e.g., Because the L1025 introns form a moderately well- for L2563, 13 of 15 possible intron trees were not supported monophyletic group with the L779 introns significantly different from each other). (see Fig. 2), a potential source for the initial insertion For the remaining comparisons (right side of may have been an existing L1025 intron. The sub- Fig. 4), significant differences were found between the sequent evolution within the Physciaceae likely in- host and intron trees, supporting intron horizontal volved a combination of vertical transmission, transfers between species. More specifically, we sug- horizontal transfer, and loss. gest, in the case of the L1025 intron, a horizontal transfer from the Physconia clade to Rinodina capensis and, for LI094, a transfer from Physcia Evolution of the Physciaceae Group I Introns stellaris/tenella (or another closely related species) to Physconia distorta. The implied transfer within the There are thought to be two general mechanisms for L2449 group occurs from the Rinodina clade to horizontal transfer of group I introns. The first is Physconia enteroxantha. Finally, there have been two DNA-based and relies on an intron-encoded endo- putative horizontal transfers within the L779 group. nuclease that recognizes a specific (14- to 40-nt) se- This group is particularly interesting because introns quence (Lambowitz and Belfort 1993). A double- 442 443

Fig. 5. Primary and tertiary structures of SSU (A) and LSU (B) rRNA. On the tertiary structures, white ladders denote RNA, proteins are dark gray ribbons, and the black spheres indicate sites of intron insertion in the Physciaceae. The intron positions in the primary structures show the nucleotide prior to the insertion site based on the homologous position in the E. coli gene. The 30S ribosomal subunit is of Thermus thermophilis (1J5E; Wimberly et al. 2000), whereas the 50S ribosomal subunit is of Haloarcula marismortui (1JJ2; Klein et al. 2001).

b strand break in the target sequence is followed by a Fig. 4. Congruence of LSU rDNA intron and host (LSU conversion event resulting in the insertion of the in- rDNA + ITS) phylogenies. The trees were constructed using the ML method with the best-fit model determined using MODEL- tron and flanking exon sequences (Belfort and Perl- TEST. In each case, the Physciaceae introns were rooted with at man 1995). This mechanism is termed ‘‘homing.’’ least one non-Physciaceae fungal intron. The bootstrap values Alternatively, reverse splicing occurs at the RNA le- (2000 replicates) are shown (when ‡60%) on the branches and were obtained using distance methods (ME/NJ) with ML-derived vel and results in the integration of the free intron parameters and the best-fit model. The thick branches indicate into a heterologous or homologous site in the tran- clades that have ‡95% posterior probability in a Bayesian script (Woodson and Cech 1989). Subsequent reverse inference. The diagonal lines indicate a significant difference between the intron and the host trees. Significance was deter- transcription and recombination could result in the mined using the AU test, with dashed lines indicating p < 0.05 intronÕs incorporation into genomic DNA. Efficient and solid lines p < 0.01. reverse splicing requires binding of the intronic 444 internal guide sequence (IGS) with 4–6 bp of the 5¢ interface and tRNA binding sites; Wilson and Noller exon (Woodson and Cech 1989; Thompson and 1998). Presumably, the rRNA must unfold to allow Herrin 1994; Roman and Woodson 1995, 1998). The the intronÕs splice site to be recognized, therefore absence of homing endonucleases in the Physciaceae making flexibility a necessity (Kjems and Garret introns is consistent with reverse splicing being the 1991; Woodson and Cech 1991; Jackson et al. 2002). likely mechanism of intron spread in this group. This explanation is consistent with our data. Com- Our detailed analysis of the Physciaceae suggests parison of the rRNA subunits shows that the SSU that horizontal transfer to the homologous site is a rRNA in general both is more flexible and has intron- prominent force in group I intron evolution in the containing sites dispersed throughout the molecule, LSU rDNA. Although strong evidence for recent whereas the LSU is more structurally rigid and has transpositions into ectopic rDNA sites was not de- introns that are essentially confined to the interface tected (except possibly for the L779 and L1025 in- region (see Fig. 5) (Ramarkrishnan and Moore 2001; trons), interspecific horizontal transfer is implicated Jackson et al. 2002). Examination of Fig. 5 shows in at least four of the eight studied intron positions. that introns are clearly more clustered on the LSU This general result has also been reported for rDNA than the SSU rDNA. There are a greater number of introns in a group of parasitic fungi (Cordyceps spp.; intron-containing positions in the SSU rDNA, which Nikoh and Fukatsu 2001). These results are particu- could theoretically cause this discrepancy. However, larly interesting in light of previous work with the in this case in which the same DNAs were screened Physciaceae SSU rDNA introns (Bhattacharya et al. for both SSU and LSU rDNA group I introns, we are 2002). This group of introns, much like those found confident that our result does not reflect oversam- in the LSU rDNA, form monophyletic lineages based pling of the SSU rDNA gene (see Bhattacharya et al. on insertion site. However, unlike the LSU introns, 2002). Rather our results reflect real differences in the there is strong phylogenetic support for a close evo- frequency of distinct intron sites and their distribu- lutionary relationship between introns at ectopic sites tion in these coding regions. This suggests that a such as between the S114 and S303 introns and the smaller number of acceptable sites in the LSU rRNA S1046 and S1052 introns. In addition, there is mod- may result in intron transpositions being more erate support for the monophyletic origin of the infrequent in this gene than in the SSU rRNA. S287–S1199 and the S1516–S1506 group I introns. Lichen thalli, rather than simply being composed We speculate that inherent differences between the of single strains of fungi and algae (or cyanobacteria), ribosomal subunits may explain this difference in the are actually diverse assemblages of organisms. This in number of apparent ectopic transpositions (i.e., hor- itself may increase the probability of lateral transfer izontal transfer to a heterologous site). between organisms. It has been known for some time The observation that introns tend to be physically that multiple strains of algae can inhabit the same clustered on the rDNA suggests that there are pref- lichen (Friedl 1987), but it has also become apparent erential sites for intron insertion and/or retention (see that more than one mycobiont genotype can exist in Fig. 5) (Jackson et al. 2002; Bhattacharya et al. 2003). the same thallus (Murtagh et al. 2000; Dyer et al. Furthermore, the observation that the extent of 2001). For example, using RAPD markers it was re- clustering in the LSU rDNA is statistically more cently shown that fungal spores from the same thallus significant (p < 0.001) than in the SSU rDNA (p < in the lichen Ochrolechia parella were polymorphic 0.01) implies that there may be fewer acceptable sites (Murtagh et al. 2000). For horizontal transfer of ge- for intron insertion in the LSU rRNA (Bhattacharya netic material to be detected, both physical transfer et al. 2003). This may be the key to understanding and stable incorporation into the genomic DNA must differences in intron evolutionary patterns between occur. The close proximity of multiple strains of fungi the SSU and the LSU rDNA. inhabiting a single thallus may therefore substantially The particular factors controlling intron location, increase the likelihood of physical transfer. It has however, are not obvious. It was previously hypoth- previously been hypothesized that viruses may act as esized that introns would be found at rRNA sites on the external vector in such cases (Dujon 1989; Bhat- accessible regions of the ribosome (Turmel et al. tacharya et al. 1996; Friedl et al. 2000). Viral infec- 1993). The reason for this preference was thought to tion specificity may explain why between-mycobiont be twofold: prevention of intron interference during and between-photobiont group I intron transfer is ribosomal assembly and increased access for reverse often implicated, but there is no evidence yet for in- splicing (Woodson and Cech 1989; Turmel et al tron movement between fungi and algae (Friedl et al. 1993). However, this has turned out not to be the 2000; Bhattacharya et al. 2002). case, because sites containing introns are no more Horizontal transfer clearly plays an important role solvent accessible than on average (Jackson et al. in the evolution of group I introns and, indeed, in 2002). An alternative explanation may be that introns evolutionary biology as a whole. Our work has led to are retained in structurally flexible sites (such as at the two important conclusions regarding this phenome- 445 non. First, we have found that within the Physciacae Cassidy JR, Moore D, Lu BC, Pukkila PJ (1984) Unusual orga- LSU rDNA, group I intron horizontal transfers do nization and lack of recombination in the ribosomal-RNA not occur to ectopic sites (with the possible exception genes of Coprinus cinereus. Curr Genet 8:607–613 Cech TR (1985) Self-splicing RNA: implications for evolution. Int of the L779 and L1025 introns). In contrast, transfer Rev Cytol 93:3–22 into homologous sites appears to be quite common. Cho Y, Palmer JD (1999) Multiple acquisitions via horizontal Second, these results contrast with data from the SSU transfer of a group I intron in the mitochondrial cox1 gene rDNA of a common set of species (Bhattacharya during evolution of the Araceae family. Mol Biol Evol 16:1155– et al. 2002). We believe that this is indicative of dif- 1165 Cho Y, Qiu Y-L, Kuhlman P, Palmer JD (1998) Explosive invasion ferent evolutionary forces acting on group I introns of plant mitochondria by group I intron. Proc Natl Acad Sci in these two genes. Specifically, the tertiary structure USA 95:14244–14249 of the two ribosomal subunits may dictate the dis- DePriest PT, Been MD (1992) Numerous group I introns with tribution and number of acceptable intron sites, with variable distributions in the ribosomal DNA of a lichen . the LSU rDNA having fewer available sites and, J Mol Biol 228:315–321 Dujon B (1989) Group I introns as mobile genetic elements: facts therefore, a lower probability of ectopic transfer than and mechanistic speculations—A review. Gene 82:91–114 is found in the SSU rDNA. Dyer PS, Murtagh GJ, Crittenden PD (2001) Use of RAPD-PCR DNA fingerprinting and vegetative incompatibility tests to investigate genetic variation within lichen-forming fungi. Sym- Acknowledgments. This work was supported by a grant awarded biosis 31:213–229 to D.B. from the National Science Foundation (MCB 01-10252) Felsenstein J (1985) Confidence intervals on phytogenies: an ap- and a grant from the Deutsche Forschungsgemeinschaft to T.F. (Fr proach using the bootstrap. Evolution 39:783–791 905/7-1,2). D.S. was partially supported by a Stanley Fellowship Friedl T (1987) Thallus development and phycobionts of the and an Avis E. Cone research fellowship from the University of parasitic lichen Diploschistes muscorum. Lichenologist 19:183– Iowa and J.M. was supported by a NSF-REU grant (MCB-03- 191 26864) awarded to D.B. We also thank Peik Haugen (Iowa) for Friedl T, Besendahl A, Pfeiffer P, Bhattacharya D (2000) The helpful discussions. distribution of group I introns in lichen algae suggests that lichenization facilitates intron lateral transfer. Mol Phylogenet Evol 14:342–352 References Gargas A, DePriest PT, Taylor JW (1995) Positions of multiple insertions in SSU rDNA of lichen-forming fungi. Mol Biol Evol Ban N, Nissen P, Hansen J, Moore PB, Steitz TA (2000) The 12:208–218 complete atomic structure of the large ribosomal subunit at Goddard MR, Burt A (1999) Recurrent invasion and extinction of 2.4 A˚ resolution. Science 289:905–919 a selfish gene. Proc Natl Acad Sci USA 96:13880–13885 Besendahl A, Qiu Y-L, Lee J, Palmer JD, Bhattacharya D (2000) Golden BL, Gooding AR, Podell ER, Cech TR (1998) A preor- The cyanobacterial origin and vertical transmission of the ganized active site in the crystal structure of the Tetrahymena plastid tRNALeu group-I intron. Curr Genet 37:12–23 ribozyme. Science 282:259–264 Belfort M, Perlman PS (1995) Mechanisms of intron mobility. J Hall TA (1999) BioEdit: a user-friendly biological sequence align- Biol Chem 270:30237–30240 ment editor and analysis program for Windows 95/98/NT. Bhattacharya D (1998) The origin and evolution of protist group I Nucleic Acids Symp Ser 41:95–98 introns. Protist 149:113–122 Haugen P, Bhattacharya D (2004) The spread of LAGLIDADG Bhattacharya D, Surek B, Rusing M, Damberger S, Melkonian M homing endonuclease genes in rDNA. Nucleic Acids Res (1994) Group I introns are inherited through common ancestry 32:2049–2057 in the nuclear-encoded rRNA of Zygnematales (Charophyceae). Haugen P, Runge HJ, Bhattacharya D (2004a) Long-term evolu- Proc Natl Acad Sci USA 91:9916–9920 tion of the S788 fungal nuclear small subunit rRNA group I Bhattacharya D, Friedl T, Damberger S (1996) Nuclear-encoded introns. RNA 10:1084–1096 rDNA group I introns: origin and phylogenetic relationships of Haugen P, Reeb V, Lutzoni F, Bhattacharya D (2004b) The evo- insertion site lineages in the green algae. Mol Biol Evol 13:978– lution of homing endonuclease genes and group I introns in 989 nuclear rDNA. Mol Biol Evol 21:129–140 Bhattacharya D, Lutzoni F, Reeb V, Simon D, Fernandez F (2000) Haugen P, Simon D, Bhattacharya D (2005) The natural history of Widespread occurrence of spliceosomal introns in the rDNA group I introns. Trends Genet 21:111–119 genes of ascomycetes. Mol Biol Evol 17:1971–1984 Helms G, Friedl T, Rambold G, Mayhofer H (2001) Identifi- Bhattacharya D, Cannone JJ, Gutell RR (2001) Group I intron cation of photobionts from the lichen family Physciaceae lateral transfer between red and brown algal ribosomal RNA. using algal-specific ITS rDNA sequencing. Lichenologist 33: Curr Genet 40:82–90 73–86 Bhattacharya D, Friedl T, Helms G (2002) Vertical evolution and Hibbett DS (1996) Phylogenetic evidence for horizontal transmis- intragenic spread of lichen-fungal group I introns. J Mol Evol sion of group I introns in the nuclear ribosomal DNA of 55:74–84 mushroom-forming fungi. Mol Biol Evol 13:903–917 Bhattacharya D, Simon D, Huang J, Cannone JJ, Gutell RR (2003) Hoist-Jensen A, Vaage M, Schumacher T, Johansen S (1999) The exon context and distribution of Euascomycetes rRNA Structural characteristics and possible horizontal transfer of spliceosomal introns. BMC Evol Biol 3:7 group I introns between closely related pathogenic fungi. Mol Cannone JJ, Subramanian S, Schnare MN, Collett JR, DÕSouza Biol Evol 16:114–126 LM, Du Y, Feng B, Lin N, Madabusi LV, Muller KM, Pande Huelsenbeck JP, Ronquist F (2001) MrBayes: Bayesian inference N, Shang Z, Yu N, Gutell RR (2002) The Comparative RNA of phylogeny. Bioinformatics 17:754–755 Web (CRW) site: an online database of comparative sequence Jackson S, Cannone J, Lee J, Gutell R, Woodson S (2002) Dis- and structure information for ribosomal, intron, and other tribution of rRNA introns in the three-dimensional structure of RNAs. BMC Bioinformatics 3:15 the ribosome. J Mol Biol 323:35–52 446

Kjems J, Garrett RA (1991) Ribosomal RNA introns in archaea the ribosomal ribonucleic acid genes in various wild-type strains and evidence for RNA conformational changes associated with and wild-collected strains of Neurospora. Mol Gen Genet splicing. Proc Natl Acad Sci USA 88:439–443 196:275–282 Klein DJ, Schmeing TM, Moore PB, Steitz TA (2001) The kink- Schroeder-Diedrich JM, Fuerst PA, Byers TJ (1998) Group-I in- turn: a new RNA secondary structure motif. EMBO J 20:4214– trons with unusual sequences occur at three sites in nuclear 18S 4221 rRNA genes of Acanthamoeba lenticulata. Curr Genet 34:71– Kruger K, Grabowski PJ, Zaug AJ, Sands J, Gottschling DE, Cech 87 TR (1982) Self-splicing RNA: Autoexcision and autocyclization Shimodaira H, Hasegawa M (2001) CONSEL: for assessing the of the ribosomal RNA intervening sequence of Tetrahymena. confidence of phylogenetic tree selection. Bioinformatics Cell 31:147–157 17:1246–1247 Lambowitz AM, Belfort M (1993) Introns as mobile genetic ele- Simon D, Fewer D, Friedl T, Bhattacharya D (2003) Phylogeny ments. Annu Rev Biochem 62:587–622 and self-splicing ability of the plastid tRNA-Leu group I intron. Lickey EB, Hughes KW, Peterson RH (2003) Variablility and J Mol Evol 57:710–720 phylogenetic incongruence of an SSU nrDNA group I intron in Suh SO, Jones KG, Blackwell M (1999) A group I intron in the Artomyces, Auriscalpium, and Lentinellus (Auriscalpiaceae: nuclear small subunit rRNA gene of Cryptendoxyla hypophloia, Homobasdiomycetes). Mol Biol Evol 20:1909–1916 an ascomycetous fungus: evidence for a new major class of Lundblad EW, Einvik C, Rønning S, Haugli K, Johansen S (2004) group I introns. J Mol Evol 1999 48:493–500 Twelve group I introns in the same pre-rRNA transcript of the Swofford DL (2003) PAUP*: Phylogenetic analysis using parsi- myxomycete Fuligo septica: RNA processing and evolution. mony (*and other methods) 4.0b8. Sinauer Associates, Sun- Mol Biol Evol 21:1283–1293 derland, MA Maddison DR, Maddison WP (2002) MacClade 4.05. Sinauer Thompson AJ, Herrin DL (1994) A chloroplast group I intron Associates, Sunderland, MA undergoes the first step of reverse splicing into host cytoplasmic Michel F, Westoff E (1990) Modelling of the three-dimensional 5.8S rRNA: implications for intron-mediated RNA recombi- architecture of group I catalytic introns based on comparative nation, intron transposition and 5.8S rRNA structure. J Mol sequence analysis. J Mol Biol 216:585–610 Biol 236:455–468 Murtagh GJ, Dyer PS, Crittenden PD (2000) Sex and the single Turmel M, Gutell RR, Mercier JP, Otis C, Lemieux C (1993) lichen. Nature 404:564 Analysis of the chloroplast large subunit ribosomal RNA gene Nikoh N, Fukatsu T (2001) Evolutionary dynamics of multiple from 17 Chlamydomonas taxa. Three internal transcribed group I introns in nuclear ribosomal RNA genes of endopara- spacers and 12 group I intron insertion sites. J Mol Biol sitic fungi of the Cordyceps. Mol Biol Evol 18:1631–1642 232:446–467 Nishida K, Suzuki S, Kimura Y, Nomura N, Fujie M, Yamada T Turmel M, Cote V, Otis C, Mercier JP, Gray MW, Lonergan KM, (1998) Group I introns found in Chlorella viruses: biological Lemieux C (1995) Evolutionary transfer of ORF-containing implications. Virology 242:319–326 group I introns between different subcellular compart- Perotto S, Nepote-Fus P, Saletta L, Bandi C, Young JP (2000) A ments (chloroplast and mitochondrion). Mol Biol Evol 12:533– diverse population of introns in the nuclear ribosomal genes of 545 ericoid mycorrhizal fungi includes elements with sequence sim- Vaughn JC, Mason MT, Sper-Whitis GL, Kuhlman P, Palmer JD ilarity to endonuclease-coding genes. Mol Biol Evol 17:44–59 (1995) Fungal origin by horizontal transfer of a plant mito- Posada D, Crandall KA (1998) MODELTEST: testing the model chondrial group I intron in the chimeric CoxI gene of of DNA substitution. Bioinformatics 14:817–818 Pepperomia. J Mol Evol 41:563–572 Posada D, Crandall KA (200la) Selecting the best-fit model of Wilson KS, Noller HF (1998) Molecular movement inside the nucleotide substitution. Syst Biol 50:580–601 translational engine. Cell 92:337–349 Posada D, Crandall KA (2001b) Simple (wrong) models for com- Wimberly BT, Brodersen DE, demons WM, Morgan-Warren RJ, plex trees: a case from Retroviridae. Mol Biol Evol 18:271–275 Carter AP, Vonrhein C, Hartsch T, Ramakrishnan V (2000) Ramakrishnan V, Moore PB (2001) Atomic structures at last: the Structure of the 30S ribosomal subunit. Nature 407:327– ribosome in 2000. Curr Opin Struct Biol 11:144–154 339 Rodriguez F, Oliver JL, Marin A, Medina JR (1990) The general Woodson SA, Cech TR (1989) Reverse self-splicing of the stochastic model of nucleotide substitution. J Theor Biol Tetrahymena group I intron: implication for the directionality 142:485–501 of splicing and for intron transposition. Cell 57:335–345 Roman J, Woodson SA (1995) Reverse splicing of the Tetrahymena Woodson SA, Cech TR (1991) Alternative secondary structures in IVS: evidence for multiple reaction sites in the 23S rRNA. RNA the 5¢ exon affect both forward and reverse self-splicing of the 1:478–490 Tetrahymena intervening sequence RNA. Biochemistry Roman J, Woodson SA (1998) Integration of the Tetrahymena 30:2042–2050 group I intron into bacterial rRNA by reverse splicing in vivo. Yang Z (1997) PAML: A program package for phylogenetic Proc Natl Acad Sci USA 95:2134–2139 analysis by maximum likelihood. CABIOS 13:555–556 Russell PJ, Wagner S, Rodland KD, Feinbaum RL, Russell JP, Zuker M (2003) Mfold web server for nucleic acid folding and Bretharte MS, Free SJ, Metzenberg RL (1984) Organization of hybridization prediction. Nucleic Acids Res 31:3406–3415