A complete list of the strains and their accession numbers is shown in the Table. In brief, we studied the following isolates: L. giganteum (ATCC 36492 = ATCC 52675 used as biological control and ATCC 48336 isolated from mosquito larvae); L. humanum (ATCC 76726 isolated from the skin of dead humans); L. giganteum pathogenic to mammals (culture collection at the Biomedical Laboratory Diagnostics, Michigan State University (MTLA): MTLA 01 = ATCCMYA-4933 type strain, MTLA-03, MTLA-04 = ATCCMYA-4934 (from a US cat), MTLA-05 = ATCCMYA-4935, MTLA-10, MTLA-12, MTLA-14, MTLA-15, MTLA-16, MTLA-17, and MTLA-18 from dogs); L. ajelloi (MTLA-06 = ATCCMYA-4936 type strain, MTLA-07 = ATCCMYA-4937, MTLA-19, MTLA-20, MTLA-21, MTLA-22, MTLA-23, isolated from dogs); L. vilelae (MTLA-24 = DNA extracted from fixed feline tissue and MTLA-25 cat strain); and L. albertoi (MTLA13 = ATCCMYA-4932 type strain).

The isolates were grown on brain-heart infusion (BHI) (DIFCO, Detroit, MI, USA), corn meal agar (CMA) (BBL, Sparks, MD, USA), 2% Sabouraud dextrose agar (SDA), 2% Sabouraud dextrose broth, in triplicate. Cardinal temperatures of growth (25°C, 30°C, and 37°C measured during 3 days’ incubation) and their relation to the production of sexual and asexual structures were evaluated on the above media. Development of the different stages of vesicle and zoospore formation, cleavage, and release were assessed on colonized grass leaves in water cultures containing Ca⁺⁺ and Mg⁺⁺. Briefly, the evaluated strains (Table) were subcultured on SDA plates at 37°C for 24 h. After incubation, 5 × 5–mm diameter blocks were cut from the advancing edges of the culture and placed on top of a 2% water agar plates. Sterile 4 × 10–mm grass blades were laid on top of each block and incubated at 37°C for 24 h (L. giganteum = MTLA-01 from mammals). The grass blades were then collected and placed in a beaker that contained 50 mL of sporulation solution (made of 2 solutions: mix no. 1 contained NH₄₂HPO₄ [66.04 g], KH₂PO₄ [68.05 g], and K₂HPO₄ [87.09 g] in 500 mL H₂O; and mix no. 2 contained CaCl₂.2H₂O [18.38 g] and MgCl₂.6H₂O [25.42 g] in 250 mL H₂O). The sporulation solution was obtained by mixing 0.5 mL of solution no. 1 plus 0.1 mL of solution no. 2 in 1.0 L of distilled water. The beaker with the 50-mL of sporulation solution plus the parasitized grass blades was incubated at 37°C, and the development of sporangia and zoospores microscopically was evaluated every 30 min for the following 6 h (or longer when needed).

We evaluated the morphologic features of L. giganteum strains after subculture on BHI, CMA, and SDA at different intervals during 5 days’ incubation at 37°C or at room temperature (25°C). Briefly, 4 × 4–mm agar blocks were cut from the above media and mixed with 1 drop of lactophenol cotton blue (phenol 20 mL, lactic acid 20 mL, glycerol 40 mL, and distilled water 20 mL). The morphologic features of their hyphal and other structures developed in these media, including oogonia, were microscopically investigated. We evaluated the development of sporangia and zoospores in sporulation medium (see above). After incubation at the induction temperatures, the beaker containing 50 mL of sporulation medium plus the parasitized grass blades was inspected on an inverted microscope. If vesicles and zoospores developed, we removedthe grass blade with >10 vesicles, using tweezers, and placed it on a glass slide. Immediately, 5 μL of Merthiolate (0.02%) was added to the grass blade to stop the movement of the zoospores and facilitate measurements. Alternatively, grass blades containing vesicles with zoospores were also transferred to plates holding Culex pipiens larvae.

The strains of L. giganteum from mammals (MTLA01, MTLA03, MTLA04, MTLA05, and MTLA10) (Table) were used to evaluate their capability to infect mosquito larvae of C. pipiens. Instar 3 C. pipiens larvae were collected and placed in 6-well plates (1 plate per experiment; Corning Inc., Corning, NY, USA) containing 5 mL of water and fed every other day with fish food flakes (TetraMin, Lancaster, PA, USA) until the end of the experiment. We transferred fresh grass blades, with >10 vesicles per blade containing numerous zoospores, to the plates holding 5 mosquito larvae per well (total of 30 larvae per plate). The plates were incubated at 25°C and observed daily by using an inverted microscope for the next 3 weeks. The criteria to select putative infected mosquito larvae were based on their swimming capabilities. We closely inspected slower swimmers for filamentous structures outside and between the larval segments and inside their bodies. Every trial included its positive (L. giganteum biological control ATCC 36492 = ATCC 52675) and negative controls (larvae without exposure to L. giganteum species), and the experiment was repeated twice (total 60 C. pipens larvae).

The 20 strains investigated (Table) were inoculated into 250-mL flasks containing 100 mL of Sabouraud dextrose broth and incubated for 72 h at 37°C on a shaker rotating at 150 rpm. In addition, 1 formalin-fixed tissue sample from a cat was also used to extract total genomic DNA following the company protocols (QIAGEN, Valencia, CA, USA). After incubation, the cultures were killed with Merthiolate (0.02%, wt/vol) and then filtrated to separate the hyphal cell mass. The hyphal cell mass was then transferred to a mortar and ground in the presence of liquid nitrogen. DNA from the disrupted hyphae was treated with sodium dodecyl sulfate and proteinase K and then incubated at 60°C for 1 h, and genomic DNA of the hydrae was extracted with phenol, chloroform, isoamyl alcohol (Sigma, St. Louis, MO. USA). PCR was conducted by hot start amplification using the following primers: the universal primers for internal transcribed spacers (ITS): ITS1 = TCCGTAGGTGAACCTGCGG and ITS4 = TCCTCCGCTTATTGATATGC; cytochrome oxidase II (COXII): COX.LagF = 5′-CCACAAATTTCACTACATTGA-3′ and COX.LagR = 5′-TAGGATTTCAAGATCCTGC-3′; heat shock protein 90 (HSP90): HSP90.LagF = 5′-CAACCTBGGHACSATYGCCAAG-3′ and HSP90.LagR = 5′-ACRAAMGACARGTAYTCVGGCA-3′; cell division cycle 42 (CDC42): CDC42.LagF = 5′-GTSCCVACYGTVTTYGANAAYTA-3′ and CDC42.LagR = 5′-GCWSWGCAYTCVASRTAYTT-3′; Tubulin: TUB.LagF = 5′-GGTGGTGGTACCGGTTC-3′ and TUB.LagR = 5′-GACACACGCTTGAACATC-3′. In addition, the following primers were constructed to PCR amplify some of the L. ajelloi and L. vilelae strains: CDC42.Lag2F = 5′-GTGCCGACYGTGTTYGABAAC-3′ and CDC42.Lag2R = 5′-CTSTTCKGTTGTRATBGG-3′; and HSP90.Lag2F = 5′-GAGGCCTTCGTGGAAGCG-3′ and HSP90.Lag2R = 5′-GTTGTTCATTTTCTTGCGGG-3′. The PCR temperature-cycling parameters were as follows: 10 min at 95°C and 1 min for subsequent cycles, annealing for 1 min at 60°C, and elongation at 72°C for 2 min. The parameter was repeated for 40 cycles, followed by a final elongation of 2 min at 72°C for 7 min. The amplicons were ligated into pCR 2.1-TOPO vector (Invitrogen, Carlsbad, CA. USA), purified, and then sequenced by using BigDye Terminator chemistry in an ABI Prim 310 genetic analyzer (Perkin-Elmer, Foster City, CA. USA).

Genomic DNA sequences of the studied strains’ complete ITS and the partial gene sequences of the selected 4 exons (CDC42, COXII, HSP90, and TUB) were aligned with other Lagenidium and Pythium DNA sequences available at the National Center for Biotechnology Information by using ClustalW version 1.81 (http://www.clustal.org)with default settings and their alignments visually inspected. Pythium spp. groups a–c, e–g, and j (15) (Table) were used as outgroups. Phylogenetic and molecular evolutionary analyses were conducted by using MEGA6 (http://www.megasoftware.net). The aligned sequences were exported for parsimony analysis by using a heuristic search with tree bisection reconnection branch swapping (MEGA6) and distant analysis by neighbor-joining (MEGA6). We coded large insertions as 1 event by excluding all but 1 nt per insertion. The generated gaps were treated as missing data. Neighbor-joining analysis used either uncorrected distances or maximum-likelihood estimates of distance with a time-reversible model (6ST), and empirical base frequencies with no rate variation among sites, or a shape parameter of 0.5 γ-distribution with 4 rate categories. Branch support was estimated as percentage of parsimony tree (1,000 resampling, heuristic, branch swapping) or neighbor-joining trees (1,000 resampling, maximum-likelihood distances) within each branch. Concatenated DNA sequences of the 5 loci investigated were used in Bayesian analysis. We conducted the test in MrBayes version 3.2.1 ×64 (http://mrbayes.sourceforge.net) using the general time reversible + I + γ model, with 2 chains (1 heated), 2 runs, sampling every 100th generation for 1 × 10⁶ generations, and exclusion of the first 2.5 × 105 samples (the burn-in) before analysis. Support for branches was estimated as the percentage of parsimony tree (1,000 resampling, heuristic, nni branch swapping) or neighbor-joining tree (1,000 resampling, maximum-likelihood distances) containing the branch, as well as by determining the Bayesian probability estimated as the percentage of Bayesian trees possessing the branch after discarding the burn-in sample.

Using Pythium spp. groups a–c, e–g, and j (Table) as outgroups, we sorted the concatenated DNA sequences of 4 protein-coding genes and 1 ribosomal region in Bayesian and maximum-likelihood phylogenetic analyses 21 mammalian pathogenic Legenidium strains into 4 strongly supported clades in the combined analysis (Figure 1). In the largest of these clades, 11 of the 21 strains (Table) recovered from infected mammals form a monophyletic clade that includes other L. giganteum strains previously used to develop biological control agents of mosquitoes (Figure 1). Also included in this clade is an L. giganteum strain recovered from nematodes in Taiwan (Lsp1). The remaining 10 mammal-infecting Lagenidium strains were resolved with strong support into 3 independent clades that are basal to L. giganteum and identified in this study as L. ajelloi, L. albertoi, and L. vilelae (Figures 1, 2). These new species will be fully described elsewhere. Other clades of Lagenidium species, as yet unnamed, are also apparent in the phylogeny (Lspa to Ld; Figures 1, 2).

The 11 L. giganteum strains from mammalian hosts grew faster at 37°C and 30°C than at 25°C, whereas L. giganteum biological control strains (ATCC 36492 = ATCC 52675 and ATCC 48336) grew well at 25°C and poorly at 30°C and almost failed to develop at 37°C in growth experiments conducted in CMA and SDA media (Figure 3). The biological control strains barely developed at 37°C but survived, as shown by their subsequent growth, after being transferred to 25°C.

We tested the ability of the mammalian L. giganteum zoospores to infect mosquito larvae in 5 L. giganteum mammal strains (MTLA01, 03, 04, 05, and 10; Table) and 2 L. giganteum biocontrol strains (ATCC 48336 = Lgm2, ATCC 36492 = Lgm3; Table). The emerging mammalian pathogenic strains exhibited mosquito-infection capabilities similar to those of the strains approved by the US Environmental Protection Agency for mosquito control (Figure 4). At least 1 larvae per well was found infected. Infected larvae exhibited abnormal swimming behavior and died within 2 days (Figure 4). Of the 60 C. pipiens larvae (2 trials, 30 larvae per experiment), 12 were found infected at the end of both experiments (instar 3 or instar 4). Uninfected larvae continued their normal life cycle. Postmortem examination of infected larvae showed extensive infections with numerous hyphae spreading internally and emerging between segments (Figure 4, panels A–E). Under our experimental conditions, L. giganteum ATCC 36492 infected 18 of the 60 C. pipens larvae. In unexposed controls all larvae survived, larvae swam normally, and we found no evidence of infection. The identity of mammalian L. giganteum strains experimentally infecting mosquito larvae was confirmed by culture, DNA extraction, sequencing, and phylogenetic analysis at the conclusion of each experiment.

Submerged colonies of both L. giganteum types (biological control and mammalian strains) are colorless or yellow with few aerial mycelia that present irregular radiating and undulating patterns on any of 3 media (CMA, BHI, or 2% SDA) (Figure 5, panels A, D). Typically, the elongated mycelium comprises ovoides or spherical 30- to 40-μm diameter segments forming chains and connected by broad to slender isthmuses; hyphae also form long irregular segments of >400 μm (Figure 5, panels B, C, E, F). We did not find oogonia in the evaluated strains. In liquid sporulation medium, hyphal segments develop 1–2 discharge tubes that produce terminal vesicles containing reniform zoospores 9–10 μm wide × 10–12 μm long (Figure 5, panel G). Zoospores were released in both mammal and mosquito L. giganteum strains after the discharge vesicle was disrupted (Figure 5, panel H), which then swam for 10–20 minutes before encysting. Germ tubes developed from encysted zoospores within 10–20 minutes after encystment.