Genomic signatures of heterokaryosis in the oomycete pathogen Bremia lactucae

Abstract

Lettuce downy mildew caused by Bremia lactucae is the most important disease of lettuce globally. This oomycete is highly variable and rapidly overcomes resistance genes and fungicides. The use of multiple read types results in a high-quality, near-chromosome-scale, consensus assembly. Flow cytometry plus resequencing of 30 field isolates, 37 sexual offspring, and 19 asexual derivatives from single multinucleate sporangia demonstrates a high incidence of heterokaryosis in B. lactucae. Heterokaryosis has phenotypic consequences on fitness that may include an increased sporulation rate and qualitative differences in virulence. Therefore, selection should be considered as acting on a population of nuclei within coenocytic mycelia. This provides evolutionary flexibility to the pathogen enabling rapid adaptation to different repertoires of host resistance genes and other challenges. The advantages of asexual persistence of heterokaryons may have been one of the drivers of selection that resulted in the loss of uninucleate zoospores in multiple downy mildews.

Fig. 1

Genome and assembly features of B. lactucae. a Estimation of genome size of heterokaryotic isolate C82P24 by flow cytometry. The nuclei of B. lactucae have two peaks calibrated relative to the reference nuclei of Oryza sativa (Os, 2 C = 867 Mb). Nuclei of isolate C82P24 (P24) were estimated to be 305 Mb (P24 1, 2 C) and 599 Mb (P24 2, 4 C). Another 38 isolates all have similar sizes (Supplementary Table 1). b Extensive collinearity between B. lactucae and P. sojae displayed as a SyMap plot. c Comparison of heterozygosity in 54 isolates of 22 oomycete species (Supplementary Table 9). Dots, representative of a single isolate, are joined by bars to aid interpretations of each species. d High quality of B. lactucae assembly demonstrated by inclusion of k-mers from paired-end reads in the assembly. Colors indicate presence of k-mers in the assembly, relative to reads. Black: the distribution of k-mers present in the read set but absent in the assembly. Red: K-mers present in the read set and once in the assembly. Purple: K-mers present in the read set and twice in the assembly. The first peak depicts heterozygous k-mers and the second peak depicts homozygous k-mers. A high-quality consensus assembly will contain half the k-mers in the first peak, the other half of which should be black due to heterozygosity, and all the k-mers in the second peak should be present only once, which therefore should be red. Very few duplicated k-mers were detected in the SF5 assembly. K-mers derived from repeat sequences have higher multiplicity and are not plotted. Source data for panel c is provided in the Source Data file

Fig. 2

Comparative long terminal repeat retrotransposon (LTR-RT) analysis. a Comparison of percent divergence of LTR elements in 15 oomycete assemblies. Distribution of percent divergence of LTR elements is shown for 12 downy mildew (B. lactucae, H. arabidopsidis, P. effusa, P. tabacina, P. viticola, and S. graminicola) and three Phytophthora (P. infestans, P. ramorum, and P. sojae) assemblies. Statistics of these assemblies are included in Supplementary Table 2. LTR elements of B. lactucae are less diverged than elements in other downy mildew assemblies. b Counts of unique LTR-RTs harvested and annotated from each assembly surveyed. Larger assemblies (panel c) are observed as having higher counts of LTR-RTs. Bars are ordered by the percent of the assembly masked displayed in panel c. Only partial elements could be found for P. halstedii. c Scatterplot demonstrating the percentage of the assembly sequence that is masked by annotated LTR-RTs and partial elements. Colors are retained from panel b. The percentage of the assembly masked increases with assembly size. B. lactucae is an outlier as it has a medium assembly size, but the highest masked percentage. Source data for all panels are provided in the Source Data file

Fig. 3

Polyphyly of downy mildews and paraphyly of Phytophthora spp. Phylogenetic maximum likelihood tree based on the analysis of 18 BUSCO protein sequences across 29 Peronosporaceae species rooted with Pythium ultimum as the outgroup. The 20 Phytophthora species were selected to represent assemblies from the nine published Phytophthora clades indicated by the number in brackets. No assembly from Phytophthora clade 9 was available. Downy mildew clades 1 and 2 are shown in red and blue, respectively. Support for nodes is shown as percent bootstrap values from 1000 iterations of the nucleotide/protein alignments. The output nucleotide and protein trees only disagreed in the order of Phytophthora clade 1 species and are indicated by -- nucleotide support. Scale is the mean number of amino acid substitutions per site. Nucleotide and protein decatenated alignments and output trees are provided as Source Data

Fig. 4

Heterokaryosis in B. lactucae. Example alternative allele frequency plots of SNPs detected in four field isolates of B. lactucae. a A unimodal distribution with a 1:1 ratio of reads supporting alternative and reference alleles seen in the homokaryotic SF5 isolate. b A trimodal distribution with peaks at 1:1, 1:3, and 3:1 ratios of reads supporting alternate alleles in the heterokaryotic C82P24 isolate, consistent with two nuclei being present in approximately equal proportions. c A bimodal distribution with two peaks at 1:2 and 2:1 ratios of reads supporting alternative alleles observed in the heterokaryotic isolate C04O1017, consistent with three nuclei being present in equal proportions. d The complex distribution observed in isolate C15C1689, consistent with an uneven mixture of multiple nuclei in a heterokaryotic isolate. Allele distributions of 31 isolates are shown in Supplementary Fig. 4. Sequencing statistics for these isolates are provided in Supplementary Table 8

Fig. 5

Sexual progeny from a cross between homokaryotic and heterokaryotic isolates form two half-sib groups. Kinship analysis based on SNPs segregating in sexual progeny generated by crossing SF5 (homokaryotic) with C82P24 (heterokaryotic). The first cluster delineates the majority of the offspring as one group of siblings derived from the same two parental nuclei (Subpopulation 1, SP 1). The second cluster delineates the remaining offspring as a second group of siblings derived from a different nucleus in C82P24 (Subpopulation 2, SP2). Relatedness of these two groups is consistent with having one parental nucleus in common derived from SF5. Relatedness of single-spore asexual derivatives of both isolates is also shown. Single-spore derivatives of C82P24 had a high relatedness to all other C82P24 derivatives and the original isolate. These derivatives and C82P24 were equidistant to all offspring, indicating the heterokaryotic C82P24 isolate had not been separated into homokaryotic components by generating single-spore derivatives. Source data is provided in the Source Data file

Fig. 6

Genomic and phenotypic instability of the heterokaryotic isolate C98O622b. a Relatedness analysis of ten asexual single-spore derivatives of C98O622b placed them into three genomic groups. One group of derivatives, A to F, were heterokaryotic and highly similar to C98O622b. The other two groups, derivatives G to I and derivative J, were each homokaryotic, less similar to C98O622b than the heterokaryotic group was, and even less similar to each other. Combining reads in silico of isolates G to I did not change their relatedness to other isolates; combining reads of any of G to I with J scored similarly high in relatedness to C98O622b as derivatives A to F. b Phenotypic differences between heterokaryotic and homokaryotic derivatives of C98O622b compared with the original isolate. Derivatives A to F were virulent on both Dm4 and Dm15; however, derivatives G to I were avirulent on Dm4 and virulent on Dm15, while derivative J showed the reverse virulence phenotype. c Alternative allele frequency plots of four C98O622b derivatives showing that derivatives A to F are heterokaryotic and G to J are homokaryotic. Alternative allele frequency plots of the derivatives nine derivatives are shown in Supplementary Fig. 7. Derivative G is not presented as inadequate coverage was obtained (Supplementary Table 8). d Alternative allele frequency plots of heterokaryotic derivatives based only on SNPs unique to each homokaryotic derivative. In a balanced heterokaryon such as derivative B, SNPs unique to each homokaryon are observed at frequencies of 0.25 and 0.75, consistent with the presence of each nucleus in a 1:1 ratio. In an unbalanced heterokaryon, such as derivative C, SNPs unique to homokaryotic derivatives G, H, and I are present at frequencies of approximately 0.17 and 0.83, while SNPs unique to derivative J are present at frequencies of 0.33 and 0.66; this is consistent with twice as many nuclei of J as those of G, H, and I. Similar distributions are observed for derivatives A, D, and E, indicating that they are unbalanced heterokaryons (Supplementary Figs. 8, 9). Source data for panel a is provided in the Source Data file

Fig. 7

Differences in fitness between heterokaryotic and homokaryotic derivatives of C98O622b. a Growth of four single-spore derivatives on the universally susceptible lettuce cv. Green Towers (GT). Heterokaryons exhibit higher growth mass per lettuce seedling and DNA quantity collected per ml of sporangia suspension. Area under the curve (AUC) measurements demonstrate significantly faster sporulation of heterokaryon derivative B compared with homokaryon derivative I (p < 0.01, n = 4). Error bars depict standard error over four replicates. b Growth curves of six heterokaryotic isolates (black lines) versus four homokaryotic isolates (red lines) on differential lettuce lines NunDM15 (Dm15) and R4T57D (Dm4) demonstrating that viable homokaryons sporulate faster on selective hosts than heterokaryons. Measurements were taken 7, 11, 15, and 21 days post inoculation (dpi) and sporulation was measure on 20 cotyledons per observation. Source data for all panels are provided in the Source Data file

Fig. 8

The multinucleate architecture of B. lactucae. Lettuce cotyledons infected with B. lactucae stained with 4’,6-diamidino-2-phenylindole (DAPI) to render nuclear DNA fluorescent. a Densely multinucleate coenocytic mycelium growing between spongy mesophyll cells of a non-transgenic lettuce cotyledon five days post infection (dpi), prior to sporulation. Two of several multinucleate haustoria that have invaginated the host plasmalemma are indicated (h). The larger plant nuclei fluoresce purple. Auto-fluorescent chloroplasts are visualized as green. b Infected lettuce cotyledon stably expressing DsRED stained seven dpi at the onset of sporulation. The multinucleate stem of a sporangiophore is visible exiting a stoma. Two multinucleate spores are visible on the cotyledon surface (arrowed). Small DAPI-stained bacterial cells are also visible. Scale bars in each represent 15 μm