Mycologia, 92(4), 2000, pp, 674-684.@ 2000 by The Mycological Society of America, Lawrence, KS 66044-8897

Northern Illinois

Department of Nematology, University of California,Davis, California 95616

A COX2 molecular phylogeny of the Peronosporomycetes

Deborah S. S. Hudspeth

Department of Biological Sciences,University, DeKalb, Illinois 60115

Steven A. Nadler

Michael E. S. Hudspeth!Department of Biological Sciences, Northern IllinoisUniversity, DeKalb, Illinois 60115

Abstract: The evolutionary history of the mitochon-drial COX2 locus has been used to infer the phylo-genetic relationships among 15 peronosporomyceteand a hyphochytriomycete species. This molecularphylogenetic analysis at both the ordinal and genericlevels provides strong evidence for the recognition ofthe Saprolegniomycetidae and the Peronosporomy-cetidae as natural groups, and for the monophyly ofthe Saprolegniales, Leptomitales and Pythiales. Athree amino acid insertion/deletion event (indel)has been identified as a putative synapomorphy forthe Saprolegniales. Parsimony mapping of 12 mor-phological and biochemical characters on the COX2molecular phylogeny yields an hypothesis for peron-osporomycete ancestral states and shared-derived fea-tures.

Key Words: Achlya, APhanomyces, Apodachlya, cy-tochrome c oxidase, Dictyuchus, Hyphochytrium, La-genidium, Leptolegnia, Leptomitus, Oomycetes, Peron-ophythora, Peronosporomycetidae, PhytoPhthora, Plec-tosPira, Pythiopsis, Pythium, Saprolegnia, Saprolegni-omycetidae, Thraustotheca

INTRODUCTION

The Peronosporomycetes, traditionally referred to asoomycetes, are a class of fungus-like heterotrophsplaced in the Kingdom Straminipila (for initial as-semblage see Patterson 1989). The diversity of thisKingdom, whose sole synapomorphy is the appear-ance of tripartite tubular hairs (stramenopiles) atsome point during the life cycle of its members, is

Accepted for publication january 13, 2000.1 Email: mykes@niu.edu

...

reflected by the inclusion of the autotrophic heter-okont chromophytic algae (chlorophylls a and c); theheterotrophic unicellular slopalinids (Proteromonas;Leipe et al 1996), bicosoecids (Cafeteria; Leipe et al1994) and Developayella (Leipe et al 1996); and ad-ditional heterotrophic fungus-like heterokont groupssuch as the thraustochytrids, labyrinthulids and hy-phochytriomycetes. Despite the apparent extreme di-versity of its members the monophyly of the strami-nipiles has been strongly supported by several molec-ular systematic studies using data derived from smallsubunit r-RNA (SSURNA) and/or large subunit r-RNA (e.g., Forster et al 1990, Leipe et al 1994, Vande Peer et al 1996, Van der Auwera et al 1995) andactin (Bhattacharya et al 1991, Bhattacharya andEhlting 1995) gene sequences.

The Peronosporomycetes are a ubiquitous groupof over 65 recognized genera (500-800 describedspecies) with representatives found in most moisthabitats (Dick 1990, 1995). The primary economicimpact of the Peronosporomycetes results from thephytopathogenic genera, which include as hosts awide variety of leguminous and cereal crops, and for-est and rosaceous fruit trees. As the causative agentsof downy mildews, white rusts, and a variety of rootrots and late blights they are a significant detrimentto agriculture. Similarly, infection of roe, fingerlings,or freshwater and marine crustaceans and molluscansis of increasing concern to the rapidly expandingfield of aquaculture. Yet despite their economic im-pact, no formal cladistic analysis involving more thana few peronosporomycetes is available, and the evo-lutionary relationships among the diverse generawithin the class remain unresolved.

Over the last 25 yr the major taxa of the Peronos-poromycetes have undergone significant revision.The most traditional organization placed the taxa infour major orders: the Saprolegniales (Dick 1973a),the Leptomitales (Dick 1973b), the Lagenidiales(Sparrow 1973), and the Peronosporales (Water-house 1973). Several major reassessments of peron-osporomycete systematics followed (e.g., Cavalier-Smith 1981, 1986, 1987, Beakes 1987) with the mostdetailed reorganizations proposed by Dick et al(1984) and Dick (1990, 1995). In generating themost recent of these taxonomic revisions Dick (1995)has provided a testable set of relationships. Based pri-~

674

marily on overall similarities of morphological char-acters, three subclasses (Peronosporomycetidae, Sap-rolegniomycetidae, and Rhipidiomycetidae) wereerected with requisite subdivision and redistributionof existing orders.

Given the paucity of morphological and biochem-ical data for inferring peronosporomycete relation-ships, molecular-based approaches are very promis-ing for understanding peronosporomycete phyloge-ny. Unfortunately, published molecular phylogeneticstudies have typically included only one to five gen-era, and were primarily intended to assess the rela-tionships of peronosporomycetes to other major eu-karyotic groups. One of those studies (Forster et al1990) used three SSURNA sequences to representthe Peronosporomycetidae and Saprolegniomyceti-dae; these authors noted that the three sequenceswere very similar to each other and suggested thatless conserved DNA sequences may be needed. Sub-sequently, Lee and Taylor (1992) inferred relation-ships among several Phytophthora species using morevariable rDNA internal transcribed spacer sequences.Neither of these studies, however, was intended toinfer a peronosporomycete phylogeny per se, nor hasa formal cladistic analysis of this group been pub-lished to date. Based on these reports and other stud-ies where SSURNA data were sometimes unable toresolve closely related taxa (e.g., Nadler and Hud-speth 1998), we examined alternative loci prior toinitiating our molecular phylogenetic analysis of the

Peronosporomycetes.Several factors suggested that the use of mtbNA

would be helpful in establishing a peronosporomy-cete phylogeny. Foremost among these were: the ex-tensive physical (Hudspeth et al 1983, Klimczak andPrell 1984, McNabb et a11987, McNabb and Klassen1988, Shumard-Hudspeth and Hudspeth 1990) andgenic (Shumard et al 1986, Shumard-Hudspeth andHudspeth 1990, Hudspeth 1992; Hudspeth and Hud-speth 1996) characterizations of a wide variety of per-onosporomycete genomes; the ability to highly purifyperonosporomycete mtDNA to obtain primary se-quence data (Hudspeth et a11980, 1983); the encod-ing of a variety of respiratory complex subunits rang-ing from very highly conserved to poorly conservedpolypeptides; and, most recently, the availability fromthe Fungal Mitochondrial GeI'fbme Project of a vari-ety of complete gene maps and sequences (Paquin etal 1997) including the complete mtDNA gene mapof the peronosporomycete Phytophthora infestans. Al-though widely used for other organisms prior use ofmtDNA for phylogenetic analyses of the Peronospo-romycetes had been limited. Sachay et al (1993) usedCOX2 and COX] deduced amino acid data from Phy-toPhthora to infer a closer relationship with plant

HUDSPETH ET AL: PERONOSPOROMYCETE PHYLOGENY 675

rather than eufungal mitochondrial polypeptides.More recently Paquin et al (1995) and Chesnick etal (1996) used P. infestans amino acid data (NDH5and NDH4L) to infer a noneufungal straminipilousrelationship for the Peronosporomycetes.

In this study we have constructed an hypothesis forthe evolutionary history of the peronosporomycetemitochondrial locus ( COX2) encoding subunit 2(COlI) of cytochrome c oxidase. By extension, DNAand deduced amino acid sequence data from 15 per-onosporomycete taxa and a hyphochytriomycetehave been used to infer the phylogenetic relation-

ships among peronosporomycete species. Specifically,we ask if there is molecular support for the recogni-tion of the Saprolegniomycetidae and the Peronos-poromycetidae as natural groups; and, whether thereis support for the placement of the Leptomitaleswithin the Saprolegniomycetidae as recently pro-posed (Dick 1995). The analyses presented here rep-resent the first rigorous cladistic analysis of this eco-nomically important, but greatly understudied classof the Straminipila.

MATERIALS AND METHODS

Strains and media.-Straminipile taxa used in this studywere obtained from the following sources (ATCC cataloguenumbers in parentheses): Achlya ambisexualis E87 (11400),A. Barksdale, New York Botanical Garden, New York, NewYork; APhanomyces euteiches A466 S.E. Holub, University ofWisconsin, Madison, Wisconsin; Leptomitus ladeus (38076),J. Aronson, Arizona State University, Tempe, Arizona; Pyth-ium ultimum 67-1 (32939), j.G. Hancock, University of Cal-ifornia, Berkeley, California; Saprolegnia ferax (36051), LB.Heath, York University, Toronto, Ontario, Canada. Dictyu-chus sterilis (44891), Hyphochytrium catenoides (18719), La-genidium f:)ganteum (36492), Leptolegnia caudata (48818),PeronoPhythora litchii (28739), PlectosPira myriandra(64139), Pythiopsis cymosa (26880), Thrau~totheca clavata(34112) were all purchased directly from ATCC, Manassas,Virginia. Apodachlya pyrifera and Sapromyces elongatus DNAswere provided by G. Klassen, University of Manitoba, Man-itoba, Canada (McNabb and Klassen 1988).

Cultures were routinely maintained on potato dextroseagar, cornmeal agar, or Emerson's YpSs agar (Difco Labo-ratories, Troy, Michigan). Mycelia for DNA isolation weretypically propagated in 8-L or 16-L aerated carboys of PYGbroth (Griffin et al 1974) at ambient temperatures. Hy-phochytrium catenoideswas propagated in stationary 2 L PYGfor DNA preparation.

Escherichia coli (]M83; F-, ara, /1(lac-proAB), rpsL,F80dlacZDMI5) containing plasmid clones was cultured inLB broth (Maniatis et al 1982) supplemented with 50 fLg/mL ampicillin. Stocks and clones were cryopreserved in25% glycerol at -70 C.

DNA techniques.-Total DNA was prepared from late log-phase mycelia and the mtDNA purified according to the

676

protocol of Shumard et al (1986). Purified mitochondrialDNAs were obtained (Hudspeth et al 1980) as the upperband in a CsCl gradient (p = 1.65g/mL), containing 100

j.lg/mL bis-benzimide (Hoechst 33258; Calbiochem, LaJol-la, California). Plasmid DNAs were isolated by the alkalinelysis method of Birnboim and Doly (1979).

Cloning and PCR COIl-encoding EcoRI fragments ofLeptomitus lacteus, APhanomyces euteiches, and Saprolegniaferax, were cloned into pUC18 and transformed into JM83following established procedures (Maniatis et al1982). Theperonosporomycete forward (GGCAAATGGGTTTTCAA-GATCC) and reverse (CCATGATTAATACCACAA-ATTTCACTAC) primers for COX2 amplification were de-signed from aligned complete COX2 sequences of Phyta-phthora megasperma (Sachay et al1993), A. euteiches, L. lac-teus, and S. ferax. PCR amplifications (25 j.lL) wereoptimized empirically in reactions containing 300 ngmtDNA, 3 mm MgC12, 12.5 nmol each primer, and 0.25units Taq polymerase (Promega, Madison, Wisconsin) inthe buffer supplied by the manufacturer. Reactions wereperformed using the following cycling conditions: 96 C, 4min; 96 C, 30 s; 50 C, 30 s; 72 C, 1 min; repeated 24 timesfrom step 2; 72 C, 4 min. The fragment size and specificityof each reaction was assessed by agarose gel electrophoresis.

PCR products were ligated into the pGEM-T vector (Pro-mega, Madison, Wisconsin) according to the manufactur-er's suggested guidelines and transformed into E. coliJM83(Maniatis et aI1982). White colonies were screened for ap-propriately sized inserts from a small scraping of agar-growncells following boiling and removal of cellular debris by cen-trifugation. A 7.5-j.lL aliquot of the supernatant was ampli-fied by PCR using the original reverse primer and a forwarddirected internal primer (GGTAGTCAATGGTATTGG) de-duced from the original four aligned peronosporomyceteCOX2 sequences. The sizes of predicted PCR products wereconfirmed by agarose gel electrophoresis.

Sequencing and molecular techniques. DNA sequences ofpeR-derived clones were confirmed by the use of differentclones for each strand. DNA sequences were determinedusing the .:l Taq Cycle Sequencing Kit with 7-deaza-dGTP ter-mination mixes and [a-32p] dATP (ICN, Costa Mesa, Cali-fornia) according to the manufacturer's suggested guide-lines (Amersham Corp., Arlington Heights, Illinois). Ad-justm~nts were made in the annealing temperature of thelabeling reactions based on the theoretical T m of each prim-er used, and again for the termination reaction based onthe predicted T m of the extended primer. For example, withan initial primer Tm of 60 C the following sets of parameterswere used: labeling at 95 C, 20 s; 55 C, 20 s; repeated 39times, and termination at 96 C, 45 s; 64 C, 20 s; 72 C, 1min; repeated 29 times. Ends of inserts were determined byuse of pGEM-T specific primers of our own design(GGCCAGTGAATTGTAATACGACTCandGACACTAT~GAATGCTCAAGCTATGC). DNA transfer hybridizationsand radiolabeled gene probes were prepared as previouslydescribed (Shumard-Hudspeth and Hudspeth 1990).

Sequence data management and Phylogenetic analysis. Se-quencing data files were organized and maintained usingthe PCGENE program group (Intelligenetics Inc., Moun-tain View, California). Clustal V, based on the alignment

MYCOLOGIA

algorithm developed by Higgins and Sharp (1989), wasused for preliminary multiple alignments of both nucleo-tide and amino acid sequences (Higgins et al 1992). Finalalignments were adjusted manually to ensure that codonalignments were maintained. For phylogenetic analysis eachhomologous sequence position was treated as a discretecharacter with four possible unordered states (G, A, T, orC). Gaps were treated as-"tnissing data. PAUP* (test version4.0.0d63) was used to infer maximum parsimony trees fromthese character-state data. Gaps (one nine-nucleotide indel)were treated as missing data in initial analyses; however, theparsimony analyses were repeated with this indel recoded(indel present/absent as the alternative states) as a singlecharacter (Swofford 1993, Crandall and Fitzpatrick 1996).Heuristic parsimony searches (with tree-bisection-reconnec-tion branch swapping) were performed using 1000 repli-cates of random stepwise addition of taxa. Reported valuesfor consistency indices exclude uniformative characters.Bootstrap resampling (2000 replicates) was used to estimatethe relative reliability of inferred monophyletic groups.PHYLIP (version 3.3, Felsenstein 1993) was used for themaximum-likelihood analyses. Program options that wereinvoked are noted in the text where appropriate.

The nucleotide sequence data presented in this workhave been deposited in GenBank as AF086687-AF086701 asa phylogenetic data set. Accession numbers for publishedsequences used to determine outgroups are as follows:Cyanidium Z48930, Prototheca U02970, Chondrus Z47547,Marchantia M68929, Allomyces U41288, SchizosaccharomycesX54421, Neurospora KO0825, and Aspergillus X15441. Align-ments and trees have been deposited in TreeBASE as S446and M657.

RESULTS

The peronosporomycete COX2 locus.-In order to de-velop a molecular phylogenetic hypothesis for thePeronosporomycetes, we characterized the COX2 lo-cus of representative genera. Purified mtDNAs fromA. euteiches, L. lacteus, and S. ferax were restricted byEcoRI, and COX2-containing restriction fragmentswere identified by DNA transfer hybridization usinga radiolabeled COX2 clone from P. megasperma (Sa-chay et al 1993). The appropriate COX2 fragmentswere then electroeluted, ligated into pUC19, andtransformed into JM83. Recombinant plasmids wereisolated and the sequence of the COX2 r~gion deter-mined. These sequences were then aligned with thecomparable region from P. megasperma 695T (Sachayet al1993). This alignment, representing three majororders of peronosporomycetes, provided an initial es-timate of peronosporomycete COX2 nucleotide se-quence diversity, which enabled confident design ofPCR primers for the efficient amplification of addi-tional taxa. Our forward primer was positioned on ahighly conserved N-terminal loop sequence regionwhereas the reverse primer exploited the presence of

highly conserved or invariant residues in the copperbinding region (CUA) of the C-terminal globular do-main (Iwata et al 1995). Use of these primers gen-erated COX2 PCR products from an additional 11peronosporomycete genera as well as the hyphochy-triomycete representative H. catenoides. Of the 15peronosporomycete taxa analyzed herein eight are

type species.Readily identifiable within the 15 deduced peron-

osporomycete and H. catenoides corr residues arethose implicated in the CUA binding region of corr(Iwata et al 1995, Tsukihara et al 1995). These resi-dues as found in the Para coccus denitrificans maturepolypeptide (Iwata et al 1995) are Asp178, His181,Cys216, GIU218, Cys220, His224, and Met227. The Asp183and His186 of the processed peronosporomycete sub-unit (Sachay et a11993) correspond to the Asp178 andHis181 of P. denitrificans, and are conserved in all per-onosporomycete and hyphochytriomycete sequencesdetermined. All but one of the remaining conservedresidues were included in the reverse PCR primingsequence and all are found in the four COX2 se-quences for which the entire nucleotide sequence isavailable.

Alignment of the deduced corr amino acid se-quences of the peronosporomycetes, hyphochytrio-mycete, and two algal outgroups identified a three-codon indel (insertion/ deletion event). This de-duced tripeptide sequence, which is predicted to liewithin the mitochondrial intermembrane space(Holm et al 1987), is notably absent in all membersof the Saprolegniales. It is, however, present as a po-tentially useful diagnostic feature within the Leptom-itales (Saprolegniomycetidae) as -Tyr-Thr-Asp-, andwithin both the Peronosporales and the Rhipidiales(Peronosporomycetidae) as -Leu-Glu-Phe/Tyr-. TheH. catenoides sequence is uniquely -Gln-Thr-Lys-.

Codon usage for the 15 examined peronosporo-mycete COX2 loci (TABLE I) is governed by the ex-pected A/T third position bias previously reportedfor the PhytoPhthora mitochondrial genome (Sachayet al 1993, Paquin et al 1997) as well as for a varietyof other peronosporomycete genera (Hudspeth andHudspeth 1996). In two of the three six-fold degen-erate codon families (Leu and Arg) a first and thirdposition A or T is preferred with neither the LeuCTCand LeuCTG nor the ArgCGC codons used. Similarly,neither the SerTCC nor the CY~GC codons are em-ployed in this region of COX2. Within the Thr codonfamily ThrAcc is used twice as frequently by the Sap-rolegniales s. s. (Achlya, Thraustotheca, Saprolegnia,Dictyuchus, and Pythiopsis) than by any other group.This same group also shows a stronger preference forArgCGT versus other groups. In the Gly family GlYGGAis used three times more frequently within the Sap-

HUDSPETH ET AL: PERONOSPOROMYCETE PHYLOGENY 677

rolegniales s. 1. (Aphanomyces, Leptolegnia, and Plec-tosPira) than by any other group.

Outgroup selection.-In the absence of any priorCOX2 phylogenetic analyses that included stramino-pile genera we considered a variety of potential out-groups for our study. It was determined from boththe morphological literature (Dick 1990) and SSUR-NA analyses (Cavalier-Smith 1993, Van der Auwera etal 1995) that the hyphochytriomycetes are basal tothe Peronosporomycetes, and that the rhodophyte(Cyanidium) and chlorophyte (Prototheca) represen-tatives would be more distantly related. PreviousSSURNA analyses using chrysophytes (Gunderson etal 1987), diatoms (Forster et al 1990), fucophytes(Bhattacharya and Stickel 1994), and hyphochytrio-mycetes (Van der Auwera et a11995) had consistentlyplaced these taxa in a position basal to the Peronos-poromycetes. Similarly, the analysis of 150 eukaryoticSSURNA sequences (Cavalier-Smith 1993) includedthe Peronosporomycetes as part of the Heterokonta,a clade also including chrysophytes, diatoms, and fu-cophytes. To insure an appropriate outgroup choice,taxa previously included in peronosporomycete stud-ies [e.g., plants and the more distantly related eufun-gi (Sachay et aI1993)] were included in a maximumparsimony (MP) analysis of COIl sequences. De-duced COIl sequences from several of these candi-dates were aligned with the deduced amino acid se-quences from the 15 peronosporomycetes and Hy-phochytrium catenoides (this study). Amino acid se-quences from four ascomycetes, one chytridiomycete,two rhodophytes, one chlorophyte, and one bryo-phyte were subsequently aligned and the entire dataset of 24 taxa analyzed by maximum parsimony. Thestrict consensus of the 21 most parsimonious treesfound in a heuristic parsimony search (FIC. 1) clearlysupported that the hyphochytriomycetes, rhodophy-tes, and chlorophytes were valid outgroup choices asinferred from COIl sequences. Thus, Hyphochytriumcatenoides, Cyanidium caldarium, and Prototheca wick-erhamii were included as outgroups in all our analy-ses. The rhodophyte Chondrus crispus was excludedbecause it uses a nonstandard genetic code, whichcomplicates amino acid based parsimony analysis.

Amino acid anaZysis.-Deduced amino acid sequenc-es of 14 peronosporomycete genera and H. catenoides(this study) were aligned with data from P. megasper-ma (Sachay et al 1993), C. caldarium (GenBankZ48930), and P. wickerhamii (Wolff et al 1994) foramino acid analysis. One hundred fourteen of the193 sites in the final alignment varied and, of these,81 were phylogenetically informative in MP analysis.MP analysis of the amino acid data set (gaps treatedas missing data) using a heuristic search strategy

UUU F 177 90 92uue F 19 10 8UUA L 251 91 93UUG L 6 2 -

DUU FUUC FUUA LDDG L

CUD LCDC..LCUA .LCDG L

AUD IADC IAUA IAUG M

GUD VGUC VGUA VGUG V

DAD YDAC YDAA -DAG -

CAD HCAC HCAA QCAG Q

AAU NAAC NAAA KAAG K

GAD DGAC DGAA EGAG E

93912

6 6LL'LL

15030

1 1-

732

25100

724

24100

2438

8457

IIIM

vvvv

361452

14611

481

483

-586

946

91. 6

00

yy

973

-919

982

707

463

928

8020

yielded six most-parsimonious trees, each 304 stepswith a consistency index (CI) of 0.718. The minimumbranch lengths supporting internal nodes of all sixtrees ranged from 1 to 22 synapomorphies. The onlydifferences among the six trees involved the place-ment of Pythiopsis and Saprolegnia within the Sapro-legniomycetidae clade. Analysis of the amino aciddata with the indel recoded as one additional char-acter yielded the same six trees by parsimony. Boot-

=--..

955

914

79 9121 9 -

34

67

33

82

18

57

43

100

100

27

33

83

75

482

2291 203

919

714

25100

49

492

100

321

"'"

eeueeeeCAeeG

413

542

4 675

412

584

362

75pppp

582121

8

1 205

2

762

22100

74 62 778

15

ACUACCACAACG

712

252

679

231

1181541

1

577

3626100

541

405

100

38100

5356 677

224

100

AAAA

GCUGCCGCAGCG

563

365

603

344

100

56 854

485

44431

47

8812

1002900

75

cc

UGUUGCUGUUGG

1008020-

w

-

~

strap resampling was performed to assess relative lev-els of support for monophyletic groups (FIG. 2). Mostnodes are reliably supported (recovered in 70% ormore of the replicates). Polytomies and nodes notsupported. by bootstrap are in some instances re-solved by the nucleotide analysis. It is notable thatthe Leptomitales, although not resolved as a mono-phyletic group in the amino acid analysis, is includedwith other taxa of the subclass Saprolegniomycetidae

FIG. 1.

680

~

FIG. 2. MP fifty-percent majority-rule bootstrap consen-sus tree obtained from analysis of amino acid data set. Num-bers represent bootstrap percentages of clades.

biased towards third position A's or T's (Sachay et al1993). Thus, A--7G and T--7C transitions would bestrongly selected against at third, and to a lesser ex-tent first, codon positions.

Prompted by recent suggestions that maximumlikelihood (ML) may frequently be a more accurateand robust method for inferring phylogeny (Huel-senbeck and Rannala 1997) ML analysis was per-formed using global rearrangement and jumble op-tions of DNAml in the PHYLIP program package(Felsenstein 1993) which recovered the single treeshown in FIG. 4 (In likelihood = -5127.96820).

DISCUSSION

A molecular phylogenetic hypothesis.- The consistenttree topologies generated by MP and ML combinedwith bootstrap values for nodes in both the aminoacid and nucleotide MP trees lends support to thismolecular hypothesis for the Peronosporomycetes. Inessence, all tree topologies support recognizing thePeronosporomycetidae and the Saprolegniomyceti-dae as natural groups. In addition, both the Leptom-itales and the Saprolegniales are included within theSaprolegniomycetidae as suggested by Dick (1995).The Saprolegniales in turn form two clades-the Sap-rolegniales s. s. consisting of Achlya, Thraustotheca,Dictyuchus, Pythiopsis, and Saprolegnia, and a sisterclade containing APhanomyces, Leptolegnia, and Plec-tosPira. When combined these two clades constitutethe more traditional Saprolegniales s. 1. (Dick 1973a).

MYCOLOGIA

FIG. 3. MP fifty-percent majority-rule bootstrap consen-sus tree obtained from analysis of nucleotide data set. Num-bers represent bootstrap percentages of clades.

Within the Saprolegniales s. s. the topology for Ach-lya, Thraustotheca, Dictyuchus, and Saprolegnia isidentical to that inferred from ITS sequence datapreviously used in examining the evolution of zoos-porangial emptying mechanisms for these genera(Daugherty et al 1998). In the Peronosporomyceti-dae all studied members of the Pythiales (Dick 1990)form a clade.

Two other molecular features discovered in thisstudy also support these interpretations. Absence ofthe indel appears to be a diagnostic character for theSaprolegniales, whereas codon usage analysis of theCOX2 locus for ArgCCT> GlYccA' and Thr ACC also sup-ports the separation of the Saprolegniales s. s.

MaPPing of nonmolecular characters.-In an attemptto compare the COX2 molecular phylogeny with thenonphylogenetic classical interpretation of the mor-phological and biological characters of the Peronos-poromycetes, 11 taxonomic characters and the aminoacid gap as a character (see list in FIG. 4, only un-ambiguous character mappings are noted on thetree) were mapped using parsirriony on the MLCOX2 tree. Since no data for these characters existoutside of the Peronosporomycetes outgroup taxacould not be used to assist in the polarity determi-nation. However, when considering the most parsi-monious distribution of these character states on theML tree (FIG. 4), ancestral (plesiomorphic) statesand putative shared-derived features could be in-ferred in certain cases. These unambiguous map-

Protothaca

Cyanidium

I 0.1 I

FIG. 4. ML COX2 tree with taxonomic characters listed. Numbers on the branches refer to characters unambiguouslymapped by parsimony. Characters and character states are as follows; states in bold are apomorphic (derived) as inferred bymapping. 1. Lipid globules: numerous and centric only = I; centric, eccentric, subcentric or subeccentric, numerous orsingle = 2 (Dick 1995); 2. Oosporogenesis: centripetal = 1; centrifugal = 2 (Dick 1995); 3.. Ooplast: solid and hyaline = I;fluid and granular = 2 (Dick 1995); 4. Use ofSO4-2: unable to use = 1; can use = 2 (Reischer 1951a); 5. Absolute requirementfor organic nitrogen source: needs organic nitrogen = 2; uses inorganic nitrogen = 1 (Reischer 1951 b); 6. Strict saprophytes= 2; not strict saprophytes = 1 (Dick 1990); 7. Oospore: elaborate exospore membrane = 1; rarely ornamented intra-episporemembrane = 2; gametic membrane = 3 (Dick 1995); 8. Amino acid gap: 3 aa missing = 1; 3 aa present = 2, (this work);9. Cellulin granules: present = 2; none present = 1 (Dick 1990); 10. Strongly oxidative = 2; not notably oxidative = 1(Natvig 1982); 11. Fermentative = 2; nonfermentative = 1 (Dick 1990); 12. K-bodies: morphotypes as labelled in Powell andBlackwell (1995).

pings provide hypotheses for the evolutionary inter-pretation of certain classical peronosporomycetecharacteristics. For taxa wher~ molecular data arelikely to remain difficult to obtain, assessments ofwhich traditional character states represent putativesynapomorphies (versus shared ancestral states) areimportant for developing testable phylogenetic hy-potheses.

As interpreted from the COX2 ML tree, a solid andhyaline ooplast and the ability to use 504-2 as a solesource of sulfur appear to be ancestral (plesiomorph-

681HUDSPETH ET AL: PERONOSPORO PHYLOGENYMYCETE

Genus Order Subclass

Outgroup

Rhipidiales

Outgroup..J

ic) states retained by the Leptomitales, Pythiales, andRhipidiales; by the tenets of cladistics, shared ances-tral states are not informative about phylogenetic re-lationships. Other ancestral states inferred in thisanalysis include rarely ornamented intra-episporemembrane, absence of cellulin granules, and non-markedly oxidative metabolism. In contrast, a fluidand granular ooplast, and the loss of the ability touse 804-2 as a sole source of sulfur ~re inferred tobe shared-d.erived characters for the Saprolegniales(FIc. 4). Parsimony mapping suggests that three

,.

682

states are putative synapomorphies for the Leptomi-tales: a gametic membrane in the oospore, presenceof cellulin granules, and strongly oxidative metabo-lism (FIG. 4). Fermentative metabolism and an elab-orate exospore membrane appears to be derivedwithin the Rhipidiales, however, additional taxa areneeded to assess if members of this taxon are mono-phyletic. Character-state distributions for lipid glob-ules and oosporogenesis are consistent with cladesrepresenting Peronosporomycetidae and Saproleg-niontycetidae. However, ancestral states are equivocalfor both characters, thus it is uncertain whichclade(s) is supported by a shared-derived state ineach case. Mapping of K-body morp11Otypes (Powelland Blackwell 1995) on the COX2 tree does not yieldany unambiguous hypothesis regarding which statesmay be shared-derived features.

Considering only the presence or absence of thethree amino acid indel, loss of these residues is aputative synapomorphy for the Saprolegniales (FIG.4). With consideration of the presence and type ofamino acids found in Hyphochytrium (the sister-taxonto the Peronosporomycetes in the molecular trees),amino acid states in Leptomitales (YTD), and Pythi-ales plus Rhipidiales (LEF /Y) are also shared-derivedstates. The distribution of the character mappings forstrict saprophytic nutrition and the requirement fororganic nitrogen are identical, which is consistentwith the idea that these features may be correlatedas a consequence of organismal ecology. However, forthese latter two characters parsimony mapping yieldsequivocal reconstructions for ancestral lineages deepin the tree such that interpretation of the derivedversus ancestral condition is not possible by this ap-proach.

An heuristic search of the 12 taxonomic charactersfor the 15 oomycete taxa yields 5250 trees (maxtreesset to 10000) each with a length of 18 and a CI of0.857 (excluding uninformative characters). Both thestrict consensus and the 50% majority-rule consensusof these trees yield identical topologies (trees notshown). Characters in these analyses were treated asunordered states and the tree was midpoint rootedbecause these data were unavailable for the outgrouptaxa. Since the indel character is derived from themolecular data set a separate heuristic search exclud-ing it was also performed. The exclusion of the indelmade no difference in either tree topology or thenumber of trees, but treelength was reduced to 16and CI to 0.833. Although the consensus trees fromthese analyses were poorly resolved, it is notable thatthe Saprolegniales and Leptomitales were recoveredas monophyletic groups.

This work has provided the first molecular phylo-genetic analysis, at both the ordinal and generic

'-,'

MVCOLOGIA

level, of this economically important group of stra-minopiles. Monophyletic groups have been hypoth-esized at the ordinal level for the Saprolegniales, Lep-tomitales, and the Pythiales. However, additional rep-resentatives of the Rhipidiales will be required to as-sess monophyly for this order. Additional studies havealready been initiat~d that involve the addition ofmore diverse taxa, and the inclusion of other geneticloci. It is anticipated that a more comprehensive mo-lecular phylogeny for this group can contribute to animproved understanding of the Peronosporomycetesby providing an expanded evolutionary frameworkfor interpreting their biological characteristics.

ACKNOWLEDGMENTS

This work was supported in part by Public Health Servicegrant GM51110 from the National Institutes of Health,grant DEB-9807937 from the National Science Foundation,and by the Plant Molecular Biology Center and the De-partment of Biological Sciences at Northern Illinois Uni-

versity.

LITERATURE CITED

Beakes CW. 1987. Oomycete phylogeny: Ultrastructural per-spectives. In: Rayner ADM, Brasier CM, Moore D, eds.Evolutionary biology of the fungi. Cambridge: Cam-bridge University Press. p 405-421.

Bhattacharya D, EhltingJ. 1995. Actin coding regions: genefamily evolution and use as a phylogenetic marker.Arch Protistenkd 145:155-164.

-, Stickel SK. 1994. Sequence analysis of duplicatedactin genes in Lagenidium giganteum and Pythium ir-regulare (Oomycota). J Mol Evol 39:56-61.

-, -, Sogin ML. 1991. Molecular phylogeneticanalysis of actin genic regions from Achlya bisexualis(Oomycota) and Costaria costata (Chromophyta). ]Mol Evol 33:525-536.

Birnboim HC, Doly]. 1979. A rapid alkaline extraction pro-cedure for screening recombinant plasmid DNA. NuclAcids Res 7:1513-1523.

Cavalier-Smith T. 1981. Eukaryote kingdoms: seven or nine?BioSystems 14:461-481.

-. 1986. The kingdom Chromista: origin and system-atics. In: Round FE, Chapman DJ, eds. Progress in phy-cological research. Vol. 4. Bristol, UK: Biopress. p 309-347.

-. 1987. The origin offungi and pseudofungi. In: Ray-ner ADM, Brasier CM, Moore D, eds. Evolutionary bi-ology of the fungi. Cambridge: Cambridge UniversityPress. p 405-421.

-. 1993. Kingdom Protozoa and its 18 phyla. MicrobiolRev 57:953-994.

Chesnick JM, Tuxbury K, Coleman A, Burger G, Lang BF.1996. Utility of the mitochondrial nad4L gene for algaland protistan phylogenetic analysis. ] Phycol 32:452-456.

Crandall KA, FitzpatrickJF Jr. 1996. Crayfish molecular sys-tematics: using a combination of procedures to esti-mate phylogeny. Syst BioI 45:1-26.

Daugherty ], Evans TM, Skillom T, Watson LE,Money NP.1998. Evolution of spore release mechanisms in theSaprolegniaceae (Oomycetes): evidence from a phylo-genetic analysis of internal transcribed spacer sequenc-es. Fungal Genet BioI 24:354-363.

Dick MW. 1973a. Saprolegniales. In: Ainsworth GD, Spar-row FK, Sussman AS, eds. The fungi. An advanced trea-tise. Vol. !VB. New York: Academic Press. p 113-144.

-. 1973b. Leptomitales. In: Ainsworth GD, SparrowFK, Sussman AS, eds. The fungi. An advanced treatise.Vol. !VB. New York: Academic Press. p 145-158.

-.1990. Phylum Oomycota. In: Margulis L, Corliss]O,Melkonian M, Chapman D], eds. Handbook of Protoc-tista. Boston: Jones and Bartlett Publishers. p 661-685.

-.1995. Sexual reproduction in the Peronosparomycetes(chromistan fungi). Can] Bot 73(Suppll):S712-S724.

-, Wong PTW, Clark G. 1984. The identity of the oom-ycete causing 'Kikuyu Yellows', with a reclassification ofthe downy mildews. Bot] Linn Soc 89:171-197.

FelsensteinJ. 1993. PHYLIP (Phylogeny Inference Package)version 3.5c. Distributed by the Author. Seattle: Uni-versity of Washington, Department of Genetics.

Forster H, Coffey MD, Elwood H, Sogin ML. 1990. Se-quence analysis of the small subunit ribosomal RNAsof three zoosporic fungi and implications for fungalevolution. Mycologia 82:306-312.

Griffin DH, Timberlake WE, Cheney ]C. 1974. Regulationof macromolecular synthesis, colony development, andspecific growth rate of Achlya bisexualis during bal-anced growth. J Cen Microbiol 80:381-388.

Gunderson JH, Elwood H, Ingold A, Kindle K, Sogin ML.1987. Phylogenetic relationships between chlorophytes,chrysophytes, and oomycetes. Proc Natl Acad Sci USA84:5823-5827.

Higgins DG, Bleasby AJ, Fuchs R. 1992. CLUSTAL V: Im-proved software for multiple sequence alignment. Ca-bios 8:189-191.

-, Sharp PM. 1989. Fast and sensitive multiple se-quence alignments on a microcomputer. Cabios 5:151-153.

Holm L, Saraste M, Wikstrom M. 1987. Structural modelsof the redox centres in cytochrome oxidase. EMBO] 6:2819-2823.

Hudspeth MES. 1992. The fungal mitochondrial genome-a broader perspective. In: Arora DK, Mukerji KG, Elan-der RP, eds. Handbook of applied mycology. Vol. 4. Bio-technology. New York: P. Marcel Dekker. p 213-241.

-, Hudspeth DSS. 1996. Mi,lPchondrial genomes ofthe zoosporic fungi. In: Dayal R, ed. Advances in zoo-sporic fungi. New Delhi: MD Publications. p 173-199.

-, Shumard DS, Bradford CJR, Grossman LI. 1983.Organization of AchlyamtDNA: a population with twoorientations and a large inverted repeat containing therRNA genes. Proc Natl Acad Sci USA 80:142-146.

-, -, Tatti KM, Grossman LI. 1980. Rapid puri-fication of yeast mitochondrial DNA in high yield.Biochim Biophys Acta 610:221-228.

HUDSPETH ET AL: PERONOSPOROMYCETE PHYLOGENY 683

Huelsenbeck JP, Rannala B. 1997. Phylogenetic methodscome of age: testing hypotheses in an evolutionary con-text. Science 276:227-231.

Iwata S, Ostermeier C, Ludwig B, Michel H. 1995. Structureat 2.8 A resolution of cytochrome c oxidase from Par-acoccus denitrijicans. Nature 376:660-669.

Klimczak LJ, Prell HH. 1984. Isolation and characterizationof mitochondrial DNA of the oomycetous fungus Phy-tophthora infestans. Curr Genet 8:323-326.

Lee SB, Taylor JW. 1992. Phylogeny of five fungus-like pro-toctistan Phytophthora species, inferred from the inter-nal transcribed spacers of ribosomal RNA. Mol BioIEvol 9:636-653.

Leipe DD, Tong SM, Goggin CL, Slemenda SB, PienjazekNJ, Sogin ML. 1996. 16S-like rDNA sequences from De-velopayella elegans, Lallyrinthuloides haliotidis, and Pro-teromonas lacertae confirm that the stramenopiles are aprimarily heterotrophic group. Eur J Protistol 32:449-458.

-, Wainwright PO, GundersonJH, Porter D, PattersonDJ, Valois F, Himmerich S, Sogin ML. 1994. The stra-menopiles from a molecular perspective: 16S-like rRNAfrom Labyrinthuloides minuta and Cafeteria roenbergen-sis. Phycologia 33:369-377.

Maniatis T, Fritsch EF, SambrookJ. 1982. Molecular clon-ing: a laboratory manual. Cold Spring Harbor, NewYork: Cold Spring Harbor Laboratory.

McNabb SA, Boyd DA, Belkhiri A, Dick MW, Klassen GR.1987. An inverted repeat comprises more than three-quarters of the mitochondrial genome in two speciesof Pythium. Curr Genet 12:205-208.

-, Klassen GR. 1988. Uniformity of mitochondrialcomplexity in Oomycetes and the evolution of the in-verted repeat. Exp Mycol 12:233-242.

Nadler SA, Hudspeth DSS. 1998. Ribosomal DNA phyloge-ny of the Ascaridoidea (Nemata: Secernentea): Impli-cations for morphological evolution and classification.Mol Phylogenet Evoll0:221-236.

Natvig DO. 1982. Comparative biochemistry of oxygen tox-icity in lactic acid-forming aquatic fungi. Arch Micro-bioI 132:107-114.

Paquin B, Laforest MJ, Forget L, Roewer I, Wang Z, Long-core J, Lang BF. 1997. The fungal mitochondrial ge-nome project: evolution of fungal mitochondrial ge-nomes and their gene expression. Curr Genet 31:380-395.

-, Roewer I, Wang Z, Lang BF. 1995. A robust phylog-eny using the mitochondrially encoded NAD5 proteinsequence. CanJ Bot 73(Suppll):SI89-S195.

Patterson DJ. 1989. Stramenopiles: chromophytes from aprotistan perspective. In: Green JP, Leadbeater BSC,Diver WC, eds. The chromophyte algae: problems andperspectives. Oxford, UK: Clarendon Press. p 357-379.

Powell MJ, Blackwell WH. 1995. Searching for homologousultrastructural characters in zoosporic fungi. Can] Bot

73(Suppll):S693-S700.Reischer HS. 1951a. Growth of Saprolegniaceae in synthetic

media. I. Inorganic nutrition. Mycologia 43:142-155.~. 1951b. Growth of Saprolegniaceae in synthetic me-

684 ~YCOLOG~

dia. II. Nitrogen requirements and the role of Krebs- - - -- -- - ---

cyCle acIds. MyCOIOgIa 'iz):z)1~5Z1S.Sachay DJ, Hudspeth DSS, Nadler SN, Hudspeth MES.

1993. Oomycete mtDNA: PhytoPhthora genes for cyto-chrome c oxidase u,se an unmodified genetic code andencode proteins most similar to plants. Exp Mycol17:7-23.

Shumard DS, Grossman U, Hudspeth MES. 1986. Achlyamitochondrial DNA: gene localization and analysis ofinverted repeats. Mol Gen Genet 202:16-23.

Shumard-Hudspeth DS, Hudspeth MES. 1990. Genic rear-rangements in Phytophthora mitochondrial DNA. CurrGenet 17:413-415.

Sparrow FK. 1973. Lagenidiales. In: Ainsworth GD, SparrowFK, Sussman AS, eds. The fungi. An advanced treatise.Vol. NB. New York: Academic Press. p 159-163.

Swofford DL. 1993. PAUP: Phylogenetic analysis using par-simony, version 3.1.1. Champaign, Illinois: The IllinoisNatural History Survey.

Tsukihara T, Aoyama H, Yamashita E, Tomizaki T, Yama-

Note added in proof: While this manuscript was in review both SSU-rRNA (Dick et aI, 1999. Mycol Res 103:1119-1125)and LSU-rRNA (Riethmuller et aI, 1999. Can J Bot-Rev Can Bot 77:1790-1800) phylogenies appeared.

guchi H, Shinzawa-Itoh K, Nakashima R, Yaono R,Yoshikawa S. 1995. Structures of metal sites of oxidizedbovine heart cytochrome c oxidase at 2.8 A Science269:1069-1074. , .

Van de Peer Y, Van der Auwera G, De Wachter R. 1996. Theevolution of stramenopiles and alveolates as derived by"substitution rate calibration" of small ribosomal sub-unit RNA. J Mol Kyol 42:201-210.

Van der Auwera G, De Baere R, Van de Peer Y, De Rijk P,Van den Broeck I, De Wachter R. 1995. The phylogenyof the Hyphochytriomycota as deduced from ribosomalRNA sequences of Hyphochytium catenoides. Mol BioIF.vol 12'n71--h'1R

Wolff G, Plante I, Lang BF, Kuck U, Burger G. 1994. Com-plete sequence of the mitochondrial DNA of the chlo-rophyte alga Prototheca wickerhamii. Gene content andgenome organization. J Mol BioI 237:75-86.

Waterhouse GM. 1973. Peronosporales. In: Ainsworth GD,Sparrow FK, Sussman AS, eds. The fungi. An advancedtreatise. Vol. IYB. New York: Academic Press. p 165-183.