1959

Mol. Biol. Evol. 18(10):1959–1969. 2001q 2001 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038

Evolutionary History of the impala Transposon in Fusarium oxysporum

Aurelie Hua-Van, Thierry Langin,1 and Marie-Josee DaboussiInstitut de Genetique et Microbiologie, Universite Paris-Sud, Orsay, France

Impala is an active DNA transposon family that was first identified in a strain of Fusarium oxysporum pathogenicto melon. The 10 copies present in this strain define three subfamilies that differ by about 20% at the nucleotidelevel. This high level of polymorphism suggests the existence of an ancestral polymorphism associated with verticaltransmission and/or the introduction of some subfamilies by horizontal transfer from another species. To gain insightsinto the molecular evolution of this family, impala distribution was investigated in strains with various host spec-ificities by Southern blot, PCR, and sequencing. Detection of impala elements in most of the F. oxysporum strainstested indicates that impala is an ancient component of the F. oxysporum genome. Subfamily-specific amplificationsand sequence and phylogenetic analyses revealed five subfamilies, several of which can be found within the samegenome. This supports the hypothesis of an ancestral polymorphism followed by vertical transmission and inde-pendent evolution in the host-specific forms. Highly similar elements showing unique features (internal deletions,high rates of CG-to-TA transitions) or being present at the same genomic location were identified in several strainswith different host specificities, raising questions about the phylogenetic relationships of these strains. A phyloge-netic analysis performed by sequencing a portion of the EF1a gene showed in most cases a correlation betweenthe presence of a particular element and a close genetic relationship. All of these data provide important informationon the evolutionary origin of this element and reveal its potential as a valuable tool for tracing populations.

Introduction

Fusarium oxysporum is a complex of soil-bornefungi responsible for diseases in more than 120 plantspecies. This asexual species complex is composed ofnumerous strains classed into several formae specialesbased on pathogenic criteria (Armstrong and Armstrong1981). Each forma specialis groups strains that are path-ogenic to one particular species or group of plants. For-mae speciales are further subdivided into physiologicalraces according to cultivar specificity. Phylogenetic re-lationships in F. oxysporum have been the subject ofmany studies undertaken to investigate evolutionarytrends among species and to determine the dynamics ofthe different formae speciales and physiological races(Gordon and Martyn 1997). Genetic distances amongstrains have been evaluated through analyses of patho-genicity, vegetative compatibility group (VCG), chro-mosomal features, rDNA restriction fragment lengthpolymorphism, mtDNA, and other molecular markers(Jacobson and Gordon 1990; Appel and Gordon 1995,1996; O’Donnell et al. 1998; Alves-Santos et al. 1999).Most studies reported to date have focused on an indi-vidual forma specialis, seeking to characterize the di-versity therein, especially as it relates to physiologicalraces. In many cases, it was found that formae specialesare genetically heterogeneous, and can sometimes havea polyphyletic origin (O’Donnell et al. 1998).

More recent studies have used new tools such asfingerprinting with repeated sequences (usually trans-posable elements), proven to be valuable for several fun-

1 Present address: Institut de Biotechnologie des Plantes, Univ-ersite Paris-Sud, Orsay, France.

Key words: transposable element, filamentous fungi, Fusariumoxysporum, Tc1-mariner superfamily, evolution, repeat-induced pointmutation.

Address for correspondence and reprints: Marie-Josee Daboussi,Institut de Genetique et Microbiologie, Universite Paris-Sud, 91405Orsay cedex, France. E-mail: daboussi@igmors.u-psud.fr.

gal species, including Magnaporthe grisea (Hamer et al.1989; Dobinson, Harris, and Hamer 1993; Farman, Tau-ra, and Leong 1996), Mycosphaerella graminicola(McDonald and Martinez 1990), Erisyphe graminis(O’Dell et al. 1989), Cryphonectria parasitica (Milg-room, Lipari, and Powell 1992), and F. oxysporum. Inthis species, fingerprinting allowed researchers to distin-guish formae speciales (Namiki et al. 1994), to track theorigin of new infestation (Mouyna, Renard, and Brygoo1996), to detect a given forma specialis (Fernandez etal. 1998), and to identify races within a forma specialis(Chiocchetti et al. 1999). Moreover, PCR assays basedon several transposable element insertion sites provideda useful diagnostic tool for quickly identifying formaespeciales and races.

Transposable elements appear especially abundantwithin F. oxysporum that exhibit a high degree of ge-netic variability. This is illustrated by numerous trans-poson families (about 17) characterized thus far, repre-senting the major classes of retroelements and DNAtransposons (Julien, Poirier-Hamon, and Brygoo 1992;Daboussi and Langin 1994; Mouyna, Renard, and Bry-goo 1996; Okuda et al. 1998; Gomez-Gomez et al. 1999;Hua-Van et al. 2000; Mes, Haring, and Cornelissen2000). One of these active DNA transposon families,named impala, is composed of few (8–10) elements andis typically 1,280 bp long with 37-bp inverted terminalrepeats (ITRs). Impala contains a single open readingframe encoding a transposase of 340 amino acids (aa).This transposase is related to those found in elementsbelonging to the widespread Tc1-mariner superfamily(Langin, Capy, and Daboussi 1995; Hua-Van et al.1998). In strain FOM24 (herein called M24), in whichimpala was first identified, about 10 different impalacopies have been characterized. Three subfamilies,named E, D, and F, have been detected (Hua-Van et al.1998). The E and D subfamilies are represented by sev-eral copies, which are autonomous, inactive, or truncat-ed (Hua-Van et al. 1998; Hua-Van et al. 2001). Withineach subfamily, nucleotide divergence between the full-

1960 Hua-Van et al.

length elements is relatively low (around 1%), while thetruncated elements are more polymorphic. The F sub-family contains only one full-length-but-inactivated el-ement. These three subfamilies differ by as much as20% at the nucleotide level (Hua-Van et al. 1998). Thisresult is intriguing when compared with the 0.3%–5%polymorphism observed within the F. oxysporum com-plex for the nia and EF1a genes and internal trans-scribed sequence or intergenic sequence of ribosomalDNA (Avelange 1994; Appel and Gordon 1996;O’Donnell et al. 1998). Two non-mutually-exclusive hy-potheses may be proposed to explain the evolutionaryorigin of impala subfamilies. First, the presence of dif-ferent subfamilies in a genome might be the result ofancestral polymorphism. The subfamilies present in thecommon ancestor would then be expected to be presentin genetically diverse strains associated with diversifi-cation of these pathogens. Another possibility is the oc-currence of one or more horizontal transfers. Suchevents have been described for other transposable ele-ments (Kidwell 1992; Robertson and Lampe 1995a), no-tably for mariner elements, for which horizontal transferappears to have played a major role in evolution (Gar-cia-Fernandez et al. 1995; Lohe et al. 1995; Robertsonand Lampe 1995b). In this case, the foreign impala el-ement(s) might be expected to be present in a smallnumber of related strains, all derived from the same an-cestor in which the transfer occurred.

In order to understand the dynamics of impalawithin the complex, a collection of F. oxysporum strainswith different host specificities were analyzed by South-ern blot, PCR, and sequencing. This analysis showedthat the impala family is widespread within F. oxyspo-rum strains. A total of five subfamilies were identified,and most of these were present in several strains withdifferent host specificities. On several occasions, similarinactive copies with unique sequence characteristicswere detected in strains with various host specificities.Relationships of these strains were inferred through phy-logenetic analysis of translation elongation factor(EF1a) gene sequences. This revealed that strains con-taining very similar impala copies are usually closelyrelated. The implications of these results for the evolu-tionary origin of strains and the use of impala as a toolare discussed.

Materials and MethodsFungal Strains

Fungal strains used in this study are listed in table1. Strains were obtained from the following sources: (1)C. Alabouvette, Institut National de Recherche Agron-omique, Dijon, France; (2) T. R. Gordon, Department ofPlant Pathology, University of California, Davis, Calif.;(3) D. Fernandez, Institut de Recherche pour le Devel-oppement, Montpellier, France; (4) J. Guadet, Institut deGenetique et Microbiologie, Universite Paris-Sud, Or-say, France; and (5) R. C. Ploetz, Tropical Research andEducation Center, University of Florida, Homestead,Fla.

DNA Extraction and Southern Blot Analysis

DNA extractions were conducted as previously de-scribed in Langin et al. (1990). Ten micrograms of ge-nomic DNA was digested with EcoRI, separated by0.8% agarose gel electrophoresis, transferred on nylonmembrane (Amersham) using a vacuum blotter, and hy-bridized using probes labeled with the Pharmacia T7Quick Prime Kit according to standard procedures (Sam-brook, Fritsch, and Maniatis 1989).

Polymerase Chain Reaction and Primer Sequences

Three pairs of primers were used to specificallyamplify elements of the E subfamily (SpE5 [AGAA-CACAACCCTGCCACGG] and SpE3 [TCCGGGCCG-TATGCACAGAG]) or the D subfamily (SpD5 [AAG-GGCTTACGCACTCACAG] and SpD3 [CACGGGCA-GTGTGAACAGAC]) or any impala elements (NS5[CGATGCCTCGAGGCAAGGAA] and NS3 [TCT-CTGCCTTCATCAACGCCC]). Elements at particularlocations were amplified using primers nested in theflanking regions, about 100 bp from either side of theelements: SFA5 (TTGCCCCCACTTTGTGTCTG) andSFA3 (TATGCGACTTAGAAGTCGGC) for impA,SFC5 (ACCACTTGTTAGATGCTCGG) and SFC3(ATCTAAACAAGGGGTGCGGC) for impC, and SFD5(ATCGTGTGTATTCCTGATCC) and SFD3 (ACAGC-CCGTTTCCCTCACGC) for impD. For impB and impGamplifications, a primer in the flanking region and aninternal primer were used: SFB5 (CAATTCATTGCA-GCCCACTG) and SpE3 for impB, or SFG3 (TCTGAG-GAAGAAACTGATC) and SpG5 (TCCTGTTGA-ATGTTGGAGGG) for impG. The latter primer primeson the solo-LTR Han, inserted in impG (Hua-Van et al.2000). Amplification of the EF1a fragment was con-ducted using primers EF-1 and EF-2 described inO’Donnell et al. (1998). Amplifications were done usingstandard procedures in a PTC100 thermocycler (MJ Re-search, Waldham, Mass.) using the following program:30 cycles of 1 min at 948C, 30 s at 608C, and 1 min 30s at 728C, followed by a final extension of 10 min at728C.

Inverse PCR

To determine if deleted copies or ripped copies orig-inated from the same event, the insertion points of thesecopies were cloned by Inverse PCR in strain MK, fol-lowing a procedure adapted from Gloor et al. (1993). Onehundred nanograms of genomic DNA was digested withrestriction enzyme for 30 min at room temperature in atotal volume of 20 ml. After 10 min inactivation at 658C,3 ml was used in a ligation experiment (10 ml final vol-ume with Promega ligase and buffer; 30 min at roomtemperature). The ligation product inactivated for 10 minat 658C was used in a PCR experiment using the follow-ing PCR program: 2 cycles of 958C for 1 min, 588C for15 s, and 728C for 5 min; 35 cycles of 958C for 15 s,588C for 15 s, and 728C for 3 min; and 728C for 10 min.

For deleted copies, we used the restriction enzymeXbaI and primers DDV5 (AAAACGCGTGCCTAATC-

Evolution of impala Transposable Elements 1961

Table 1Strains Used in this Study and impala Distribution

STRAIN ABB.a F. SP. HOST ORIGINb SB

PCRb

NS SpE SpD

FOM24 . . . . . . . .FOM150. . . . . . .FOM7 . . . . . . . . .FOMK419 . . . . .FOMP2. . . . . . . .FOM1127. . . . . .

M24M150M7MKMPM11

melonismelonismelonismelonismelonismelonis

MelonMelonMelonMelonMelonMelon

FranceFranceFranceMexicoCaliforniaNew York

(1)(1)(1)(2)(2)(2)

111111

1111*11

111221

1111*22

FO5. . . . . . . . . . .FO47. . . . . . . . . .FOLn3 . . . . . . . .FOLn86248 . . . .FOLn88321 . . . .FOL15 . . . . . . . .

S5S47Ln3Ln86Ln88L15

NPNPlinilinilinilycopersici

SoilSoilFlaxFlaxFlaxTomato

FranceFranceFranceFranceFranceTunisia

(1)(1)(1)(1)(1)(1)

211111

211111*

211112

221**211*

FORL28 . . . . . . .FORL4 . . . . . . . .FOA1 . . . . . . . . .FOA4 . . . . . . . . .FOA5 . . . . . . . . .

RL28RL4A1A4A5

rad-lyco.rad.-lyco.albedinisalbedinisalbedinis

TomatoTomatoDate palmDate palmDate palm

FranceCanadaMoroccoMoroccoMorocco

(1)(1)(3)(3)(3)

11111

1*11

NDND

111

NDND

1*21

NDND

FOA438 . . . . . . .FOB . . . . . . . . . .FOCa cal . . . . . .FOCa O1 . . . . . .FOCa lago . . . . .

A438BCacCaOCa1

albedinisbulbigenumcanariensiscanariensiscanariensis

Date palmVanillaOrn. palmOrn. palmOrn. palm

Morocco

CaliforniaFranceCanary

(3)(4)(1)(1)(1)

11211

ND1

ND2

ND

ND1

ND2

ND

ND2

ND2

NDFOCi. . . . . . . . . .FOCu2 . . . . . . . .FOD11 . . . . . . . .FOG . . . . . . . . . .FOP108 . . . . . . .

CiCuD11GP108

cicericubensedianthigladioliphaseoli

ChickpeaBanana treeCarnationGladiolusCom. bean

Costa RicaItaly

Brazil

(1)(5)(1)

11111

111

ND11

111

ND11

1**22

ND1

FOP204 . . . . . . .FO var read . . . .FOR4 . . . . . . . . .FOR5 . . . . . . . . .FOVr1. . . . . . . . .

P204

R4R5V

phaseoliredolensraphaniraphanivasinfectum

Com. bean

RadishRadishCotton

Columbia

FranceGermanyUSA

(4)(1)(1)(3)

12111

ND21

ND1

ND21

ND1

ND22

ND2

NOTE.—f. sp. 5 formae speciales; SB 5 Southern blot; NP 5 nonpathogenic; ND 5 not determined; orn. 5 ornamental; com. 5 common; rad.-lyco. 5 radicislycopersici; 1 5 detection of hybridization signal or PCR amplification; 2 5 absence of signal or amplification; * 5 fragment shorter than expected; ** 5 fragmentlarger than expected.

a The abbreviated names of the strains (used throughout the text).b Strains were provided by different laboratories (numbers in parentheses). (1) C. Alabouvette, Institut National de Recherche Agronomique, Dijon, France; (2)

T. R. Gordon, Department of Plant Pathology, University of California, Davis, Calif.; (3) D. Fernandez, Institut de Recherche pour le Developpement, Montpellier,France; (4) J. Guadet, Institut de Genetique et Microbiologie, Universite Paris-Sud, Orsay, France; (5) R. C. Ploetz, Tropical Research and Education Center,University of Florida, Homestead, Fla.

c Primer pairs used in PCR experiments correspond to ‘‘nonspecific’’ NS3 and NS5 primers (NS), allowing amplification of any impala element, or primersSpE3 and SpE5, specific for subfamily E (SpE), and primers SpD3 and SpD5, specific for subfamily D (SpD). These primers are described in Materials andMethods.

d redolens strains are considered as a species distinct from Fusarium oxysporum (see O’Donnell and Cigelnik 1997).

GGG) and DDV3 (CTCGGGCTAGACCGATCAGTC).For ripped copies, enzymes PstI and HindIII and primersRIPK5OUT (ATAACCAGCACGCCATAATTCG) andRIPK5IN (TTAAACACGAAGAAAAACGCGG) al-lowed amplification of 59 flanking sequences. We de-signed primers SF3-Del (CAAACAGAGCTTAACCT-CAACCGC) and SF5-RIP (TCACTTGCTGTTC-TCCCCAG), located in these flanking sequences, andused them along with a primer specific for impala (SpD5or RIP5OUT, respectively) in a direct PCR experiment.An amplification product of the expected size was ob-tained in all strains containing these deleted or rippedcopies (MK, L15, RL28 and MK, L15 and Ci).

Cloning and Sequencing

Impala PCR products were cloned using the Pro-mega pGEMT-easy kit. For each strain, several individ-

ual clones were sequenced using the ABI PRISM Dye-Terminator Cycle Sequencing Ready Reaction Kit (Ap-plied Biosystems) on an ABI373 Automated DNA se-quencer (Applied Biosystems). Sequencing primerscorresponded to M13 forward and M13 reverse primers.EF1a sequencing was performed either after cloning asdescribed above or directly on PCR products usingprimers EF-1 and EF-2. The nucleotide sequences re-ported in this paper are deposited in GenBank underaccession numbers AF363407–AF363439 (impala) andAF363394–AF363406 (EF1a).

Phylogenetic Analysis

Multiple DNA alignments were obtained with theClustal V program (Higgins and Sharp 1988). Maxi-mum-parsimony analyses were performed using theheuristic algorithm and default options of PAUP 4.0

1962 Hua-Van et al.

FIG. 1.—Southern blot analysis of strains within the Fusarium oxysporum complex. The DNA was digested with EcoRI, which does notcut within impala. A, Hybridization with the impala-160 probe. B, Strains hybridized with the impEDF probe, consisting of an internal part ofeach of the three impala subfamilies (NS PCR products; see Materials and Methods and fig. 3).

(Swofford 2000). We carried out 100 or 1,000 bootstrapreplications of maximum parsimony with a full heuristicsearch and default options.

ResultsCopy Number and Genomic Distribution of impala

Southern blot experiments using a copy of imp160or a mix of PCR products (impEDF; see below) asprobes were performed on 32 strains belonging to 14formae speciales (table 1 and fig. 1). These experimentsrevealed that all but three strains contained impala ele-ments. The number of hybridization signals (#10) in-dicated that impala was always present in low copynumbers. The variation in intensity of the hybridizationsignals (fig. 1A) probably reflects the presence of severaldivergent subfamilies in the same genome, as demon-strated in strain M24 (Hua-Van et al. 1998). Similar var-iations (fig. 1B) were obtained with the use of the mixedprobe impEDF, composed of 840-bp impala internalPCR fragments obtained from copies of impE, impD,and impF, which are representative of the three knownsubfamilies (see below), indicating that other subfami-lies may exist.

PCR experiments were carried out on a set of 23strains, representing 13 formae speciales, using a set ofspecific primers for the major impala subfamilies. Non-specific (NS) primers able to amplify all types of impalawere used as a positive controls (fig. 2A). As shown infigure 2B and table 1, all strains showing impala se-quences by Southern blot gave PCR products using NSprimers. Using the specific primers (SpE and SpD pairs),we were able to detect elements belonging to E and toD subfamilies in 17 and 10 strains, respectively. Eightstrains contained both subfamilies, and one strain con-tained none of these subfamilies, although impala ele-ments were detected with NS primers. Several strainsgave additional PCR products of unexpected sizes.Shorter fragments were obtained in formae speciales

melonis (MK), lycopersici (L15), and radicis-lycopersici(RL28) using the NS primers and the SpD primerswhich are specific for the D subfamily, strongly sug-gesting the presence of internally deleted elements ofthe D subfamily in these strains. In formae speciales lini(Ln3) and ciceri (Ci), a longer fragment was obtainedwith the SpD primers but not with the NS primers. Thiscould reflect the existence of more complex rearrange-ments within the impala elements.

When the PCR products were hybridized with theprobe impEDF (see above and fig. 2A and B), stronghybridization signals were observed in some strains (i.e.,M24 and MK), while other strains (i.e., MP and RL4)contained elements that gave very weak hybridizationsignals. These elements may represent new subfamiliesor highly polymorphic members of a known subfamily.

Sequence Variation and Phylogenetic Analysis

The different PCR products obtained with the NSprimers were cloned, and some of these clones werecompletely sequenced. A phylogenetic tree deducedfrom these sequences by maximum parsimony is shownin figure 3. The clades corresponding to the known sub-families E, D, and F were strongly supported by boot-strapping (Hua-Van et al. 1998). The E subfamily (100%bootstrap) is composed of several groups of highly sim-ilar elements (around 4% nucleotide divergence; diver-gence can reach 12% for more divergent members), ex-cept for elements RL4-110, M11-6, and impB, which aresupported by long branches. The D subfamily (100%bootstrap) is also very homogeneous in sequence (2.3%sequence divergence) but is characterized by the pres-ence of several internally deleted elements. The F sub-family (98% bootstrap), for which only one elementfrom strain M24 had been described, was also found intwo other strains and appeared less homogeneous (14%nucleotide divergence).

Evolution of impala Transposable Elements 1963

FIG. 2.—PCR amplification of different types of impala elements. A, Locations of primers and sizes of the expected PCR products. B, Gelanalysis and hybridization results for the three primer pairs (NS, SpE, and SpD). DNA was transferred to nylon membrane and hybridized withthe impEDF probe. EtB 5 ethidium-bromide-stained gels.

The phylogenetic analysis resulted in the identifi-cation of two additional subfamilies: one of them, calledthe K subfamily (63% bootstrap), is composed of fivesequences obtained from four different formae speciales.The two subclades within the K subfamily received100% bootstrap support, and one of these, composed ofthree sequences, was supported by the longest nodewithin the tree. These three elements presented a partic-ular mutation pattern (see below) that can explain thishigh nucleotide divergence (an average of 14% betweenthe two subclades). Nevertheless, these copies appearedcloser to the two other sequences within the K subfamilyand have been considered to belong to this subfamily.

The fifth subfamily, P, contained only one member iden-tified in strain MP. The subfamilies differ from each oth-er by 20%–30% nucleotide divergence, with the P sub-family being the most divergent (26%–30%).

Active Versus Inactive Elements

Potentially active elements (i.e., containing no stopcodons or frameshift mutations over the 800 bp se-quenced) were found within the E and D subfamilies(fig. 3), similar to those described previously from strainM24 (Hua-Van et al. 1998). The present study also re-vealed the existence of such elements in the F subfamilybut not in the K and P subfamilies.

1964 Hua-Van et al.

FIG. 3.—Phylogenetic tree based on impala sequences using par-simony. This tree is one of the five most-parsimonious trees foundusing the heuristic search algorithm of PAUP* 4.0. All trees wereidentical in topology except for minor differences within clade E. MP2-26 was used as an outgroup. Parsimony bootstrap values based on 100replicates are indicated on the nodes. Only bootstrap values .50% areshown. Subfamilies discussed in the text are indicated by capital letterson the right side of the figure. 1 5 copies with an uninterrupted openreading frame that could be active; * 5 copies with high rates of CG-to-TA mutations; D 5 internally deleted copies; impE, impD, impB,and impF are M24 elements.

FIG. 4.—Structure of the internally deleted elements. The top se-quence corresponds to the full-length element (impD), and the bottomsequence corresponds to the deleted elements. The 7-bp repeats foundflanking the deletion in the full-length element are framed. Numbersrefer to the nucleotide positions of the repeats relative to the full-lengthelement.

Most of the sequences represented inactive copies,as deduced from the presence of stop codons and/orframeshifts. They were found in every subfamily.Among these, we identified two particular types of in-active elements. The first type was characterized by in-ternal deletions. Three sequences with deletions,RL28D22, L15D5, and MKD208 (belonging to the Dsubfamily), were identified in three different formae spe-ciales (fig. 3). Surprisingly, each of these possessed thesame 327-bp deletion, and these truncated elementswere highly similar at the nucleotide level. The deletionoccurred within a 7-bp motif present at each deletionpoint in full-length elements (fig. 4). Translations re-vealed that the reading frame was maintained in two ofthese copies. Using inverse PCR, we determined that allthree copies were inserted at the same genomic position.Thus, these copies result from vertical transmission ofan ancestral deleted copy.

Inactive copies of the second type were identifiedin subfamilies E and K (RL4-110, M11-6, Ci-16, MK-28, and L15-16). These elements appeared to have beenextensively mutated relative to other copies of the samesubfamily. This high level of polymorphism, reflectedby the long branches of the tree, appeared to be pri-marily due to CG-to-TA transitions (table 2), reminis-

cent of the repeat induced point mutation (RIP) processfirst discovered in Neurospora crassa (Selker 1987). Ananalysis of the 39 and 59 sequence context of C-T mu-tations (including G-A mutations, counted as C-T mu-tations on the complementary strand) revealed that mu-tations occurred primarily into (A/T/C)pCp(A/G), withA or T preferred in the 59 flank and A preferred over Gin the 39 flank (table 2). We used a ‘‘de-RIP’’ approach(Cambareri et al. 1998) consisting of removal of all ob-served CG-to-TA transitions to investigate whether sim-ilarities of these copies to members of their familiescould be improved. In phylogenetic trees including these‘‘de-RIPped’’ sequences, RL4-110(dRIP) grouped withthe small clade containing impE, Ln88-23, S47-35, R-8, and Ln3-1 (99.7% identity with impE). M11-6(dRIP)and Ci-16(dRIP) remained more related to their‘‘ripped’’ relatives (data not shown). Although we pre-viously reported the existence of a ‘‘ripped’’ copy(impB) in strain M24 (Hua-Van et al. 1998), the patternof transition was quite different between impB, RL4-110, and M11-6, all belonging to the E subfamily. Incontrast, the patterns of mutation in the three ‘‘ripped’’elements belonging to the K subfamily—Ci-16, MK-28,and L15-16—were very similar (no more than 2% nu-cleotide divergence among these elements). We deter-mined that these three copies were inserted at the samegenomic position and thus derived from one ancestralcopy in which RIP occurred.

Genomic Positions of impala Copies

Using PCR, we investigated the presence of impalaelements at the same genomic position in the strainspreviously used for PCR-detection of impala. Primersannealing to flanking regions were used for the detectionof impA, impC, impD, and impE. For the 39 deleted copyof impB, a primer hybridizing in the 59 flanking regionand an impala-specific internal primer (SpE3) wereused. For the detection of the impG element, we chosea 39 impala internal primer and a 59 primer nested insolo-LTR Han that is inserted in the 59 end of impG(Hua-Van et al. 1998, 2001). Impala elements were de-tected in eight of the 20 strains analyzed, and thesestrains have various host specificities (see table 3). ImpAwas never detected via PCR; a short fragment corre-sponding to the empty site was always amplified. Ab-sence of impE was also associated with the amplificationof an empty site. In contrast, empty sites were never

Evolution of impala Transposable Elements 1965

Tab

le2

Per

cent

ages

ofC

G-t

o-T

AT

rans

itio

nsan

dN

ucle

otid

eC

onte

xtin

Deg

ener

ate

Cop

ies

CO

PIE

S

%T

RA

NS

ITIO

Na

CG

-TO

-TA

%a

No.

39F

LA

NK

OF

MU

TA

TE

DC

b

AC

GT

59F

LA

NK

OF

MU

TA

TE

DC

b

AC

GT

impB

c ...

....

.R

L4-

110c

....

M11

-6c

....

..C

i-16

d..

....

.

98 100 97 80

74 99 82 64

85 158

108 82

15 30 17 27

11 13 9 13

3 1 7 5

14 44 28 25

28 45 28 30

0 0 2 3

12 23 10 28

2 2 5 6

aP

erce

ntag

esar

eba

sed

onth

eto

tal

num

ber

ofnu

cleo

tide

diff

eren

ces.

bD

ata

incl

ude

only

chan

ges

that

are

not

adja

cent

toon

ean

othe

r.c

Rel

ativ

eto

impE

.d

Rel

ativ

eto

Ci-

36.

Cop

ies

L15

-16

and

MK

-28

are

not

repo

rted

here

beca

use

ofth

eir

high

sim

ilar

ity

toC

i-16

.

Table 3Presence of impala Copies at the Same Position as in M24in Other Strains

impASFA5SFA3

impBSFB5NS3

impCSFC5SFC3

impDSFD5SFD3

impESFE5SFE3

impGSFG5SFG3

M7. . . . .K . . . . . .B . . . . . .D11 . . . .Ln3 . . . .Ln88 . . .R . . . . . .V . . . . . .

22222222

10110001

1?????1?

10000000

22211122

11000000

NOTE.—1 indicates presence of the copy (1.5-kb PCR product); 2 indicatesabsence of the copy (PCR product of 200 bp); 0 indicates absence of PCRproduct amplification; ? indicates that results are unclear due to aspecific am-plifications.

amplified with the impD primers, and the impD elementwas detected only in strain M7. Complete absence ofamplicons may be interpreted as the absence of one orboth sequences complementary to the primers or an or-ganization of the region different from that present inM24, or it could be due to the absence of the entireregion. ImpB was amplified in four strains. The 39 endof this copy was truncated in M24. In order to determineif impB was also truncated in these strains, we per-formed a PCR using a primer lying in the deleted partof impB. No amplification was obtained, suggesting thatimpB-homologous elements also possess 39 deletions(data not shown). Moreover, the sequencing of the dif-ferent amplified fragments of impB revealed the samemutational pattern observed in M24 impB (these se-quences, B-impB, V-impB, and D-impB, were includedin the tree shown in fig. 3).

Phylogenetic Relationships Among Strains

During the course of this study, we found severalsituations in which strains with different host specifici-ties contained elements with similar mutations or ele-ments that were present at the same genomic position.The presence of such elements might reflect a close ge-netic relationship between the strains. These resultsraised questions about the phylogenetic relationshipsamong strains of F. oxysporum responsible for diseasesin different hosts. To better understand these relation-ships, a phylogenetic analysis based on DNA sequencesof a portion of the EF1a gene was conducted, as inO’Donnell et al. (1998). The aligned sequences wereanalyzed by maximum parsimony and bootstrapping(fig. 5). The analysis showed the existence of three mainclades. We found that except for MK, all strains con-taining one or several copies of impala present at thesame position as in strain M24 were grouped in oneclade (73% bootstrap). MK and L15, both containingsimilar deleted and highly mutated copies, appearedvery close to each other (97% bootstrap). The two otherstrains containing either a mutated copy (Ci) or a deletedcopy (RL28) arose in the same clade (86%) (fig. 5).

1966 Hua-Van et al.

FIG. 5.—One of the five most-parsimonious trees illustrating phy-logenetic relationships among Fusarium oxysporum strains, based onthe EF1a gene. The topology shown is obtained with midpoint rooting.Analysis was based on 609 nt from position 24 to position 633 relativeto published sequences (O’Donnell et al. 1998) and was performedexcluding all uninformative characters. Numbers represent bootstrapvalues from 1,000 replicates (below nodes, in bold) and branch lengths(above nodes). Asterisks indicate sequences used in O’Donnell et al.(1998); or obtained from the GenBank database. The presence of dif-ferent types of elements is indicated by the different geometric shapeson the right.

DiscussionImpala Is an Ancient Component of the F. oxysporumComplex

Distribution of the impala family within the F. ox-ysporum complex as investigated by Southern blot andPCR revealed that impala elements were present in mostof the strains we tested. Impala thus appears to be anancient component of the F. oxysporum genome. Am-plification using specific primers and DNA sequencingrevealed that subfamilies that had diverged by 20%–30% were frequently present within the same genome.This supports the hypothesis of an ancient presence ofthe different subfamilies in the F. oxysporum genome.The coexistence of these subfamilies in several strainsis consistent with the vertical transmission of these cop-ies present in a common ancestor. These subfamilies arealways represented by a small number of copies, sug-gesting that impala is not a very active family or thatselection acts to eliminate additional copies. However,occasional activity is evidenced by the presence of sev-eral copies of the same subfamily in some strains, char-acterized by strain-specific polymorphism, as observed

for copies Cu-12 and Cu-15 (forma specialis cubense)and Ci-32 and Ci-172 (forma specialis ciceri).

Diversity Among Subfamilies

Potentially active elements (with no interruption inthe sequenced coding frame) have been identified insubfamilies E, D, and F. Inactive elements were foundin all subfamilies and represent the majority of the se-quences. Inactivity mainly results from stop codons andframeshift mutations. However, we detected other alter-ations resulting from large deletions or a high rate oftransitions. All of these data suggest that the impalafamily is genetically diverse.

Three elements, containing the same internal de-letion, were identified in three strains with different hostspecificities. These elements were located at the samechromosomal position and clearly arose by verticaltransmission from a common ancestor that containedthis deleted copy. Location of the deletion between tworepeats of 7 bp suggests mechanisms similar to thoseput forward for several other deleted transposable ele-ments (O’Hare and Rubin 1983; Streck, MacGaffey, andBeckendorf 1986; MacRae and Clegg 1992). One is ho-mologous recombination between short repeats, fromone element or from copies on sister chromatids (Streck,MacGaffey, and Beckendorf 1986). Another is interrup-tion of the gap repair mechanism that takes place afterexcision of the element (Kurkulos et al. 1994; Hsia andSchnable 1996; Rubin and Levy 1997).

Diversity among subfamilies is also illustrated bythe identification of several highly mutated elements inthe E and K subfamilies. Alterations (CG-to-TA transi-tions) resemble those associated with RIP, a mutationalprocess described in N. crassa (Selker et al. 1987). Thisprocess acts on duplicated sequences in the same nucle-us during the sexual phase. RIP-like processes have alsobeen identified in other ascomycetous fungi, Aspergillusfumigatus (Neuveglise et al. 1996), M. grisea (Nakay-ashiki et al. 1999), and Podospora anserina (Hamann,Feller, and Osiewacz 2000). Aspergillus fumigatus, likeF. oxysporum, is an asexual fungus. Although the typeof alteration is the same, the RIP processes in all ofthese species differ in target preference. In N. crassa,C-to-T transitions occur with decreasing frequencies inCpA . CpT . CpG . CpC (Cambareri et al. 1989).C-to-T transitions are found mainly in (A/T)pCp(A/T)in M. grisea. In F. oxysporum mutations into CpA andCpG were preferred, and we also observed in the 59flanks of mutated C’s a preference for A, T, and, to alesser extent, C. The de-RIPped sequences confirmedthat RL4-110 and M11-6 belong to the E subfamily(99.8% and 98% identity with impE, respectively). Thede-RIPped copy Ci-16(dRIP) still showed 7% nucleotidedivergence from Ci-36, suggesting that Ci-16 is not aripped version of Ci-36 and comes from another copy,not identified during this work. This putative elementcould be a divergent member of the K subfamily or amember of another new subfamily. Taking into accountthat subfamilies usually diverge from each other by20%–30%, the three ripped elements can be considered

Evolution of impala Transposable Elements 1967

to belong to the K subfamily. Whereas the distributionsof RIP in copies impB, RL4-110, and M11-6 are differ-ent, impB and its homologs, as well as the three copiesin the K subfamilies, share nearly identical mutationalpatterns. In these two cases, the mutational process ap-pears to have occurred in a common ancestor of thestrains, since all copies homologous to impB are locatedat the same genomic position, as are the three rippedcopies of subfamily K. Ripped copies appear to be theonly remnant of impala in strains M11 and RL4 (seehybridization results in fig. 2B). Strains M24 and Cicontain ripped and nonripped elements from the samesubfamily. At least in these strains, the RIP-like processmay have acted transiently, with a possible reinvasionof impala.

Impala Evolution and Phylogenetic RelationshipWithin the F. oxysporum Complex

The presence of highly similar impala elements instrains belonging to different formae speciales raised thequestion of the evolutionary origin of these strains. Weconstructed a phylogeny with the nuclear gene EF1a,chosen because of its demonstrated utility for phylogenyreconstruction within this complex (O’Donnell et al.1998). We showed that except for strain MK, discussedbelow, strains containing impala elements at the samegenomic position are genetically related on the basis ofthe EF1a gene tree. This result provides evidence thatstrains may be more closely related than assumed bytheir host specificities. For instance, the group of strainsrelated to M24 exhibits pathogenic specificities to awide range of plants. This suggests that pathogenicityshifts are relatively frequent and raises questions con-cerning the genetic bases of such a versatile pathogenicbehavior. The polyphyletic origins of some formae spe-ciales, notably of melonis (O’Donnell et al. 1998), werealso confirmed in this study. Strain MK was previouslyfound to be different from other melonis strains basedon VCG and mtDNA haplotype analysis (Jacobson andGordon 1990). We showed here that this strain, collectedin Mexico, is actually closely related to strain L15, apathogen of tomato collected in North Africa.

The phylogenetic data help clarify the evolutionaryhistory of the impala family. The presence of copies atthe same genomic location in closely related strainsstrongly suggests that these copies were inserted in acommon ancestor. Absence of these copies in other re-lated strains might then indicate that they were lost, ei-ther by excision (for impA and impE) or by deletion orrearrangement of the region (for impD). Stochastic lossis also suggested by the fact that some strains lack somesubfamilies and by the fact that some rare strains appearto be devoid of impala. However, the existence of im-pala elements belonging to more widely divergent sub-families is not completely excluded. Too-divergent cop-ies would be undetectable in standard Southern experi-ments or by PCR amplification.

Another mechanism that may play a role in trans-posable element evolution is horizontal transfer. This issuggested for several transposable elements, with the

most convincing examples being illustrated by the P el-ement from Drosophila (Daniels et al. 1990; Clark andKidwell 1997; Silva and Kidwell 2000) and by the wide-ly spread mariner element (Lohe et al. 1995; Robertsonand Lampe 1995b). Members of this family have beenfound to be more similar among distantly related speciesthan within a species. In the absence of strong phylo-genetic support, horizontal transfer is not easily dem-onstrated. The widespread distribution of the differentsubfamilies is likely the result of an ancestral polymor-phism. The horizontal transfer of one or more subfam-ilies is not excluded, but investigations on closely relat-ed species will be required for assessment of this ques-tion. Within the F. oxysporum complex, the distributionof different elements (ripped, deleted, or at a given po-sition) is also suggestive of a vertical transmission, ex-cept in the case of impG in strain MK. This strain doesnot appear to be closely related to the other melonisstrains on the basis of the EF1a analysis. Nevertheless,it contains the copy impG, detected only in a smallgroup of melonis strains. Moreover, this strain containsa molecular marker (the sfo gene, homologous to theyeast SUR1 gene; Hua-Van et al. 2000), which has thusfar been detected only in melonis strains (unpublisheddata). This gene is absent in all non-melonis strains,even those closely related to M24, and it is located 10kb upstream of impG in M24. The presence of this geneand impG in MK suggests that the whole region couldbe conserved between MK and M24. However, this re-gion, which is extremely rich in repeated sequences, ischaracterized by several apparent recombination eventsand is apparently subject to rapid reorganization (Hua-Van et al. 2000). Therefore, the possibility that the entireregion surrounding impG in M24 could be present inMK seems rather surprising and raises the possibility ofthe genetic transfer of a large piece of DNA from theM24 group to MK. However, further analyses are re-quired to test this hypothesis.

Impala as a Marker of Genetic Relationships

Identification of closely related elements (by theirsequences or by their genomic positions) in differentstrains raises the question of whether impala could beused as a marker for inferring genetic relationships. Thestudy of specific fixed impala elements appears to be agood strategy to detect close relationships. However, theexample of strain MK indicates that some exceptionscan occur. Therefore, the use of impala as a marker forgenetic relationships can be very helpful in detectingunsuspected relationships, but it has to be used alongwith, or confirmed by, other markers, such as single-copy nuclear genes.

Acknowledgments

We are grateful to K. O’Donnell for helpful com-ments on the manuscript. We thank J. M. Daviere forhis help in sequencing, and C. Gerlinger for her tech-nical assistance.

1968 Hua-Van et al.

LITERATURE CITED

ALVES-SANTOS, F. M., E. P. BENITO, A. P. ESLAVA, and J. M.DIAZ-MINGUEZ. 1999. Genetic diversity of Fusarium oxys-porum strains from common bean fields in Spain. Appl.Environ. Microbiol. 65:3335–3340.

APPEL, D. J., and T. R. GORDON. 1995. Intraspecific variationwithin populations of Fusarium oxysporum based on RFLPanalysis of the intergenic spacer region of the rDNA. Exp.Mycol. 19:120–128.

———. 1996. Relationships among pathogenic and nonpatho-genic isolates of Fusarium oxysporum based on the partialsequence of the intergenic spacer region of the ribosomalDNA. Mol. Plant Microbe Interact. 9:125–138.

ARMSTRONG, G. M., and J. K. ARMSTRONG. 1981. Formae spe-ciales and races in Fusarium oxysporum causing wilt dis-eases. Pp. 391–399 in P. NELSON, T. A. TOUSSON, and R. J.COOK, eds. Fusarium: diseases, biology, and taxonomy.Pennsylvania State University Park, University Park, Pa.

AVELANGE, I. 1994. Recherche d’un outil de caracterisationvisant a construire une classification phylogenetique englo-bant l’ensemble des souches de l’espece Fusarium oxys-porum. Ph.D. dissertation, Universite Paris-Sud, Orsay,France.

CAMBARERI, E. B., R. AISNER, and J. CARBON. 1998. Structureof the chromosome VII centromere region in Neurosporacrassa: degenerate transposons and simple repeats. Mol.Cell. Biol. 18:5465–5477.

CAMBARERI, E. B., B. C. JENSEN, E. SCHABTACH, and E. U.SELKER. 1989. Repeat-induced G-C to A-T mutations inNeurospora. Science 244:1571–1575.

CHIOCCHETTI, A., I. BERNARDO, M. J. DABOUSSI, A. GARIBAL-DI, M. LODOVICA GULLINO, T. LANGIN, and Q. MIGHELI.1999. Detection of Fusarium oxysporum f. sp. dianthi incarnation tissue by PCR amplification of transposon inser-tions. Phytopathology 89:1169–1175.

CLARK, J. B., and M. G. KIDWELL. 1997. A phylogenetic per-spective on P transposable element evolution in Drosophila.Proc. Natl. Acad. Sci. USA 94:11428–11433.

DABOUSSI, M. J., and T. LANGIN. 1994. Transposable elementsin the fungal plant pathogen Fusarium oxysporum. Genetica93:49–59.

DANIELS, S. B., K. R. PETERSON, L. D. STRAUSBAUGH, M. G.KIDWELL, and A. CHOVNICK. 1990. Evidence for horizontaltransmission of the P transposable element between Dro-sophila species. Genetics 124:339–355.

DOBINSON, K. F., R. E. HARRIS, and J. E. HAMER. 1993. Grass-hopper, a long terminal repeat (LTR) retroelement in thephytopathogenic fungus Magnaporthe grisea. Mol. PlantMicrobe Interact. 6:114–126.

FARMAN, M. L., S. TAURA, and S. A. LEONG. 1996. The Mag-naporthe grisea DNA fingerprinting probe MGR586 con-tains the 39 end of an inverted repeat transposon. Mol. Gen.Genet. 251:675–681.

FERNANDEZ, D., M. OUINTEN, A. TANTAOUI, J. P. GEIGER, M.J. DABOUSSI, and T. LANGIN. 1998. Fot1 insertions in theFusarium oxysporum f. sp. albedinis genome provide di-agnostic PCR targets for detection of the date palm patho-gen. Appl. Environ. Microbiol. 64:633–636.

GARCIA-FERNANDEZ, J., R. J. BAYASCAS, G. MARFANY, M. A.MUNOZ, A. CASALI, J. BAGUNA, and E. SALO. 1995. Highcopy number of highly similar mariner-like transposons inplanarian (Platyhelminthe): evidence for a trans-phyla hor-izontal transfer. Mol. Biol. Evol. 12:421–431.

GLOOR, G. B., C. R. PRESTON, D. M. JOHNSON-SCHLITZ, N. A.NASSIF, R. W. PHILLIS, W. K. BENZ, H. M. ROBERTSON, and

W. R. ENGELS. 1993. Type I repressors of P element mo-bility. Genetics 135:81–95.

GOMEZ-GOMEZ, E., N. ANAYA, M. I. RONCERO, and C. HERA.1999. Folyt1, a new member of the hAT family, is activein the genome of the plant pathogen Fusarium oxysporum.Fungal Genet. Biol. 27:67–76.

GORDON, T. R., and R. D. MARTYN. 1997. The evolutionarybiology of Fusarium oxysporum. Annu. Rev. Phytopathol.35:111–128.

HAMANN, A., F. FELLER, and H. D. OSIEWACZ. 2000. The de-generate DNA transposon Pat and repeat-induced point mu-tation (RIP) in Podospora anserina. Mol. Gen. Genet. 263:1061–1069.

HAMER, J. E., L. FARRALL, M. J. ORBACH, B. VALENT, and F.G. CHUMLEY. 1989. Host species-specific conservation of afamily of repeated DNA sequences in the genome of a fun-gal plant pathogen. Proc. Natl. Acad. Sci. USA 86:9981–9985.

HIGGINS, D. G., and P. M. SHARP. 1988. CLUSTAL: a packagefor performing multiple sequence alignment on a micro-computer. Gene 73:237–244.

HSIA, A. P., and P. S. SCHNABLE. 1996. DNA sequence analysessupport the role of interrupted gap repair in the origin ofinternal deletions of the maize transposon, MuDR. Genetics142:603–618.

HUA-VAN, A., J. M. DAVIERE, F. KAPER, T. LANGIN, and M. J.DABOUSSI. 2000. Genome organization in Fusarium oxys-porum: clusters of class II transposons. Curr. Genet. 37:339–347.

HUA-VAN, A., F. HERICOURT, P. CAPY, M. J. DABOUSSI, and T.LANGIN. 1998. Three highly divergent subfamilies of theimpala transposable element coexist in the genome of thefungus Fusarium oxysporum. Mol. Gen. Genet. 259:354–362.

HUA-VAN, A., J. PAMPHILE, T. LANGIN, and M. J. DABOUSSI.2001. Transposition of autonomous and engineered impalatransposons in Fusarium oxysporum and a related species.Mol. Gen. Genet. 264:724–731.

JACOBSON, D. J., and T. R. GORDON. 1990. Variability of mi-tochondrial DNA as an indicator of relationships betweenpopulations of Fusarium oxysporum f.sp. melonis. Mycol.Res. 94:734–744.

JULIEN, J., S. POIRIER-HAMON, and Y. BRYGOO. 1992. Foret1,a reverse transcriptase-like sequence in the filamentous fun-gus Fusarium oxysporum. Nucleic Acids Res. 20:3933–3937.

KIDWELL, M. G. 1992. Horizontal transfer. Curr. Opin. Genet.Dev. 2:868–873.

KURKULOS, M., J. M. WEINBERG, D. ROY, and S. M. MOUNT.1994. P element-mediated in vivo deletion analysis ofwhite-apricot: deletions between direct repeats are stronglyfavored. Genetics 136:1001–1011.

LANGIN, T., P. CAPY, and M. J. DABOUSSI. 1995. The trans-posable element impala, a fungal member of the Tc1-mar-iner superfamily. Mol. Gen. Genet. 246:19–28.

LANGIN, T., M. J. DABOUSSI, C. GERLINGER, and Y. BRYGOO.1990. Influence of biological parameters and gene transfertechnique on transformation of Fusarium oxysporum. Curr.Genet. 17:313–319.

LOHE, A. R., E. N. MORIYAMA, D. A. LIDHOLM, and D. L.HARTL. 1995. Horizontal transmission, vertical inactivation,and stochastic loss of mariner-like transposable elements.Mol. Biol. Evol. 12:62–72.

MCDONALD, B. A., and P. J. MARTINEZ. 1990. DNA restrictionfragment polymorphisms among Mycosphaerella gramini-cola (anamorph Septoria tritici) isolates collected from asingle wheat field. Phytopathology 80:1368–1373.

Evolution of impala Transposable Elements 1969

MACRAE, A. F., and M. T. CLEGG. 1992. Evolution of Ac andDsl elements in select grasses (Poaceae). Genetica 86:55–66.

MES, J. J., M. A. HARING, and B. J. CORNELISSEN. 2000. Foxy:an active family of short interspersed nuclear elements fromFusarium oxysporum. Mol. Gen. Genet. 263:271–280.

MILGROOM, M. G., S. E. LIPARI, and W. A. POWELL. 1992.DNA fingerprinting and analysis of population structure inthe chestnut blight fungus, Cryphonectria parasitica. Ge-netics 131:297–306.

MOUYNA, I., J. L. RENARD, and Y. BRYGOO. 1996. DNA poly-morphism among Fusarium oxysporum f. sp. elaeidis pop-ulations from oil palm, using a repeated and dispersed se-quence ‘‘Palm.’’ Curr. Genet. 30:174–180.

NAKAYASHIKI, H., N. NISHIMOTO, K. IKEDA, Y. TOSA, and S.MAYAMA. 1999. Degenerate MAGGY elements in a sub-group of Pyricularia grisea: a possible example of suc-cessful capture of a genetic invader by a fungal genome.Mol. Gen. Genet 261:958–966.

NAMIKI, F., T. SHIOMI, T. KAYAMURA, and T. TSUGE. 1994.Characterization of the formae speciales of Fusarium ox-ysporum causing wilts of cucurbits by DNA fingerprintingwith nuclear repetitive DNA sequences. Appl. Environ. Mi-crobiol. 60:2684–2691.

NEUVEGLISE, C., J. SARFATI, J. P. LATGE, and S. PARIS. 1996.Afut1, a retrotransposon-like element from Aspergillus fu-migatus. Nucleic Acids Res. 24:1428–1434.

O’DELL, M., M. S. WOLFE, R. B. FLAVELL, C. G. SIMPSON,and R. W. SUMMERS. 1989. Molecular variation in popula-tions of Erysiphe graminis on barley, oats and rye. PlantPathol. 38:340–351.

O’DONNELL, K., and E. CIGELNIK. 1997. Two divergent intra-genomic rDNA ITS2 types within a monophyletic lineageof the fungus Fusarium are nonorthologous. Mol. Phylo-genet. Evol. 7:103–116.

O’DONNELL, K., H. C. KISTLER, E. CIGELNIK, and R. C.PLOETZ. 1998. Multiple evolutionary origins of the fungus

causing Panama disease of banana: concordant evidencefrom nuclear and mitochondrial gene genealogies. Proc.Natl. Acad. Sci. USA 95:2044–2049.

O’HARE, K., and G. M. RUBIN. 1983. Structures of P trans-posable elements and their sites of insertion and excision inthe Drosophila melanogaster genome. Cell 34:25–35.

OKUDA, M., K. IKEDA, F. NAMIKI, K. NISHI, and T. TSUGE.1998. Tfo1: an Ac-like transposon from the plant pathogenicfungus Fusarium oxysporum. Mol. Gen. Genet. 258:599–607.

ROBERTSON, H. M., and D. J. LAMPE. 1995a. Distribution oftransposable elements in arthropods. Annu. Rev. Entomol.40:333–357.

———. 1995b. Recent horizontal transfer of a mariner trans-posable element among and between Diptera and Neurop-tera. Mol. Biol. Evol. 12:850–862.

RUBIN, E., and A. A. LEVY. 1997. Abortive gap repair: under-lying mechanism for Ds element formation. Mol. Cell. Biol.17:6294–6302.

SAMBROOK, J., E. F. FRITSCH, and T. MANIATIS. 1989. Molec-ular cloning: a laboratory manual. 2nd edition. Cold SpringHarbor Laboratory Press, New York.

SELKER, E. U., E. B. CAMBARERI, B. C. JENSEN, and K. R.HAACK. 1987. Rearrangement of duplicated DNA in spe-cialized cells of Neurospora. Cell 51:741–752.

SILVA, J. C., and M. G. KIDWELL. 2000. Horizontal transferand selection in the evolution of P elements. Mol. Biol.Evol. 17:1542–1557.

STRECK, R. D., J. E. MACGAFFEY, and S. K. BECKENDORF.1986. The structure of hobo transposable elements and theirinsertion sites. EMBO J. 5:3615–3623.

SWOFFORD, D. L. 2000. PAUP*. Phylogenetic analysis usingparsimony (*and other methods). Version 4.0b4a. Sinauer,Sunderland, Mass.

THOMAS EICKBUSH, reviewing editor

Accepted June 20, 2001