Multiomic approaches reveal novel lineage-specific effectors in the potato and tomato early blight pathogen Alternaria solani

The effectome of the necrotrophic fungal pathogen, Alternaria solani, was determined using multiomics. In total, 238 effector candidates were predicted from the A. solani genome, and apoplastic effectors constitute most of the total candidate effector proteins (AsCEPs). Comparative genomics revealed two main groups of AsCEPs: lineage-specific and conserved effectors. RNA-Seq analysis revealed that the most highly expressed genes encoding AsCEPs were enriched with lineage-specific forms. Two lineage-specific effector genes, AsCEP19 and AsCEP20, were found to form a ‘head-to-head’ gene pair located near an AT-rich region on the chromosome. To date, AsCEP19 and AsCEP20 have been found only in a few fungal species. Phylogenetic inference revealed that AsCEP19 and AsCEP20 were likely acquired by the common ancestor of A. solani and A. tomatophila via horizontal gene transfer, probably mediated by long terminal repeat retrotransposon. RT-qPCR analysis showed that AsCEP19 and AsCEP20 are tightly coexpressed in a host-specific manner and that they are upregulated at advanced stages of A. solani infection only in solanaceous hosts. Transient expression of AsCEP19 and AsCEP20 in Nicotiana benthamiana plants showed that these effectors could promote Phytophthora infestans infection. AsCEP19 and AsCEP20 were required for the full virulence of A. solani on host potato, because deletion of this gene pair significantly reduced the size of necrotic lesions on potato leaves. Transient expression of AsCEP20 could elicit plant cell death depending on the presence of its signal peptide, indicating that AsCEP20 is a necrosis-inducing apoplastic effector with the mature form localized specifically in chloroplasts. Our work provides a better understanding of the function and evolution of necrotrophic fungal effectors, and helps explain the high aggressiveness of A. solani against solanaceous crops.


Background
Alternaria solani and A. tomatophila (formerly named A. linariae) are sister species in phylogeny; both are necrotrophic pathogens that cause early blight in potato (Solanum tuberosum) and tomato (S. lycopersicum) plants (Woudenberg et al. 2014;Adhikari et al. 2017). In general, species within section Porri of the genus Alternaria have developed a certain level of host specificity, and are

Open Access
Phytopathology Research † Jinhui Wang and Siyu Xiao contributed equally to this work *Correspondence: zhujiehua@hebau.edu.cn; bdyzh@hebau.edu.cn destructive to their primary host plants (Ozkilinc et al. 2017). For example, A. solani and A. tomatophila are well adapted to solanaceous crops, such as potato and tomato. A. solani varies slightly from A. tomatophila in virulence in that A. solani is equally aggressive to both potato and tomato plants, but A. tomatophila is more aggressive to the latter (Gannibal et al. 2014). Knowledge of the genetic and molecular basis for differences in pathogenicity between Alternaria lineages, especially for the high aggressiveness of A. solani and A. tomatophila on solanaceous crops, is still limited. Previous research on pathogenicity of Alternaria species has mainly focused on phytotoxins, plant cell wall-degrading enzymes and melanin (Tsuge et al. 2013;Meena et al. 2017). However, it is still unclear what type of effector proteins are secreted by early blight fungus and how these effectors participate in pathogenicity. Until the present study, no effector has been reported from A. solani or A. tomatophila.
Pathogenic variation among lineages could be partially explained by the presence-and-absence variation (PAV) of genes. Lineage-specific genes usually play roles in determining virulence, dictating the host range, shaping host specificity, and enabling host jumping (Borah et al. 2018;Fouché et al. 2018;Sánchez-Vallet et al. 2018;de Vries et al. 2020). Fusarium oxysporum has diverged into two lineages that infect humans or plants by the acquisition of small chromosomes, which are rich in lineage-specific genes encoding effectors or other virulence factors (Ma et al. 2010;Zhang et al. 2020). Acquisition of lineage-specific genes through horizontal gene transfer (HGT) is an important driving force in fungal evolution. It allows fungal pathogens to acquire new virulence factor genes from other species. The most well-known HGT event between phytopathogenic fungi is related to ToxA, which encodes a hostspecific toxin (HST) that can interact with the product of the Tsn1 gene in wheat (Adhikari et al. 2009). Acquisition of ToxA by Pyrenophora tritici-repentis led to increased virulence and severe tan spot epidemics (Friesen et al. 2006). It has been confirmed that ToxA has been subjected to HGT between three different wheat pathogenic fungi, Parastagonospora nodorum, P. tritici-repentis, and Bipolaris sorokiniana (McDonald et al. 2019). HGT is particularly important for those pathogenic fungi that are considered to be strictly asexual, such as Verticillium dahliae, F. oxysporum, and A. alternata (Reinhardt et al. 2021). It has been reported that a lineage-specific fragment encoding virulence factors was horizontally transferred from F. oxysporum f. sp. vasinfectum, the pathogen causing Fusarium wilt of cotton, to V. dahliae strain Vd991, and this HGT event makes strain Vd991 a hypervirulent race on cotton (Chen et al. 2018a, b). HGT of conditionally dispensable chromosomes has been revealed in Alternaria species, and therefore different pathogenic lineages can carry extra chromosomes which harbor gene clusters for the biosynthesis of HSTs (Tsuge et al. 2013;Rajarammohan et al. 2019;Wang et al. 2019).
Transposable elements (TEs) are often involved in HGT events, and TE insertions activate repeat-induced point (RIP) mutations in fungal genome to silence the inserted TEs, thereby accumulating G/C to A/T mutations and gradually forming AT-rich regions (Frantzeskakis et al. 2020). The pre-existing AT-rich regions could serve as preferred integration sites for other TEs, followed by various types of mutations, including point mutations, insertions, duplications, recombination, and deletions, while subjected to host or environmental selection pressure, and eventually become lineage-specific regions as shelters for lineage-specific genes (Torres et al. 2020). Typical genomic features of such lineage-specific regions have been reported, such as frequent association with AT-rich repetitive sequences, presence of TE remnants, low gene density, and faster evolution, which are useful characteristics for identifying lineage-specific genes in fungal genomes .
Three genome assemblies of A. solani (strains BMP0185, HWC168, and NL03003) and one genome assembly of A. tomatophila (strain BMP2032) have been published so far (Dang et al. 2015;Wolters et al. 2018;Zhang et al. 2018). The de novo sequence assemblies of A. solani BMP0185 and A. tomatophila BMP2032 are still highly fragmented (Dang et al. 2015), because the repetitive genomic regions are difficult to assemble using only short reads. The completeness of the genome sequence of A. solani HWC168 is better, because it was assembled using Illumina paired-end and mate-pair reads (Zhang et al. 2018). The genome sequence of A. solani NL03003 was assembled using long reads generated from PacBio SMRT sequencing, and NL03003 is the only completely assembled A. solani genome reported to date (Wolters et al. 2018). A contiguous genome assembly is essential, when the aim is to identify novel effectors or other virulence factors, which are often associated with lineagespecific genomic regions. Nevertheless, analysis of these complete or draft genomes provide insights into the biology of early blight pathogens, and these data are valuable genomic resources for comparative genomic studies.
Here, effector candidates were predicted from the A. solani genome, and genome comparisons were preformed to assess the evolutionary conservation of the effector genes in Alternaria species. RNA-Seq data were used to identify the effector genes that were highly expressed in planta. A series of assays, including phylogenetic inferences, gene expression analysis, Agrobacterium-mediated transient expression, and gene knockout were performed to reveal the roles of lineage-specific effectors in the virulence of this early blight pathogen.

Effector candidates in A. solani
The genome of A. solani HWC168 was predicted to possess 11,951 protein-coding genes (Additional file 1: Table S1), of which 238 (1.99%) were considered to be candidate effector proteins (CEPs) (Additional file 1: Table S2). Of the A. solani CEPs (AsCEPs), 157 (65.97%) and 28 (11.76%) were classified as apoplastic and cytoplasmic effectors, respectively, and 53 (22.27%) could be dual localized effectors ( Fig. 1a and Additional file 1: Table S2). Obviously the apoplastic effectors constitute the major group of the total effector (p-value 8.38e −7 , χ 2 test). Based on the RNA-Seq data of A. solani (Additional file 1: Table S3), 209/238 (87.8%) AsCEP genes were defined as effector genes expressed in planta but with different expression levels (Additional file 1: Table S2): 10 AsCEP genes were expressed at high levels in potato leaves and ranked within the top 200 among all proteincoding genes, while 111 and 88 AsCEP genes had moderate and low expression levels, respectively (Fig. 1b). Conserved long range synteny was observed between the genomes of strains HWC168 and NL03003 (Additional file 2: Figure S1); furthermore, all AsCEP genes predicted in strain HWC168 have homologs in strain NL03003 and exhibit high collinearity (Fig. 1c), indicating a high level of conservation in gene content and genome structure.

Conservation of AsCEP genes within the genus Alternaria
Among the 29 Alternaria species whose genomes have been sequenced (Additional file 1: Table S4), our PAV analysis of the genes encoding AsCEPs displayed 'twopeaks' (Fig. 2a). The first peak represents a cluster of 30 (12.6%) AsCEP genes that were only identified in A. solani or were also found in a sister species, suggesting that their functions might be associated with pathogenic traits within a very narrow lineage. The second peak consists of 101 (42.4%) genes encoding AsCEPs that were commonly found in at least 27 (90%) Alternaria species and represent a cluster of conserved effectors in the genus. Interestingly, the most highly expressed AsCEP genes were enriched with lineagespecific effectors (p-value 3.68e −3 , Fisher's exact test), suggesting that A. solani has evolved a unique effector arsenal that appears to rely heavily on lineage-specific effectors ( Fig. 2a and Additional file 1: Table S2). Of the Histogram showing the expression levels of AsCEPs genes. AsCEPs genes were divided into four subsets according to the Log 2 (TPM + 1) values. c Collinearity of AsCEP genes between genomes of strains NL03003 and HWC168. Lines connect AsCEP gene homologs lineage-specific AsCEP genes that are highly expressed, AsCEP17 encodes a hypothetical protein that did not return any hit in UniProtKB. AsCEP1 and AsCEP13, AsCEP19 and AsCEP20 encode hypothetical proteins which have homologs in Cochliobolus heterostrophus and Corynespora cassiicola, respectively. Of the conserved and highly expressed AsCEP genes, AsCEP138 encodes a virulence factor, hydrophobin, while other genes encode hypothetical proteins with unknown function. A. alternata, A. arborescens, A. gaisen, A. tomatophila, and A. tenuissima share at least 190 (79.8%) effector genes with A. solani, while A. gansuensis has the fewest effector genes (118) in common with A. solani (Fig. 2b).

Novel lineage-specific AsCEP genes
Of the lineage-specific AsCEP genes showing high levels of expression, AsCEP19 and AsCEP20 lie adjacent to each other, constituting a bidirectional (head-tohead, H2H) gene pair which is located within a small GC-equilibrated region (15 kb) on chromosome 3 (CP022026.1.3) of strain NL03003 (Fig. 3). The small GC-equilibrated region is flanked by two AT-rich regions, which are located between positions 199,848 and 242,843 (43 kb) and positions 257,850 and 284,655 (26.8 kb) on this chromosome, respectively (Fig. 3). The distance between the 3' end of AsCEP19 and nearby AT-rich region is only 1,012 bp (Additional file 1: Table S5). Two similar (94% identity) gypsy-family long terminal repeat (LTR) elements were found in the Histogram showing the counts of AsCEPs genes that were found in one to twenty-nine (X-axis) Alternaria species, and heatmap showing the conservation of the most highly expressed AsCEPs genes. Yellow columns represent effector genes found in more than 26 or less than three Alternaria species. The number indicated on each column represents the count of AsCEP genes. Colors in heatmap represent the Log 2 (TPM + 1) values, and the numbers in grey boxes represent the number of Alternaria species in which each gene was found. b The number of AsCEPs genes that have homologs in other Alternaria species flanking AT-rich regions, which are located from base pairs 217,219 to 223,363 and from 272,436 to 279,499, respectively. Within the Alternaria genus, AsCEP19 and AsCEP20 only present in A. solani and sister species A. tomatophila, which also cause early blight of potato and tomato.

Homologs of AsCEP19 and AsCEP20
Both AsCEP19 and AsCEP20 are small secreted proteins that are rich in cysteine (9/99 and 8/102, respectively) and are most likely to be homologs of the putative small secreted proteins, which have been reported in the C. cassiicola strain Philippines (CCP), a fungal pathogen causing leaf fall disease on rubber trees (Lopez et al. 2018). The CCP genome contains two genes (BS50DRAFT_638061 and BS50DRAFT_64582) homologous to AsCEP19 and one gene (BS50DRAFT_627105) homologous to AsCEP20 (Fig. 4a). In contrast to AsCEP19 and AsCEP20 that form an H2H gene pair, the CCP homologs are located on different scaffolds. Genes homologous to AsCEP19 have only been found in CCP so far, but genes homologous to AsCEP20 have also been discovered in some Colletotrichum species (Fig. 4a). AsCEP19 has no known domain, but AsCEP20 contains a fungal calcium-binding domain PF12192 (HMMER search e-value 1.5e −9 ), suggesting that AsCEP20 is a distant homolog of CBP1, a well-known virulence factor identified in the human fungal pathogen Histoplasma capsulatum (Sebghati et al. 2000). AsCEP20 also contains the six conserved cysteine residues, a characteristic of CBP1 (Batanghari et al. 1998). In addition, AsCEP20 and its homologs found in other phytopathogens contain two more cysteine residues (Additional file 2: Figure S2).

Putative origins of AsCEP19 and AsCEP20
The gene trees (Fig. 4a) and species tree ( Fig. 4b) indicates that AsCEP19 and AsCEP20 do not have a vertical descent origin. Within the Alternaria genus, AsCEP19 and AsCEP20 are only present in A. solani and sister species A. tomatophilia, and their homologs have been found in very few fungal species outside this genus (Fig. 4a). It is still unclear whether AsCEP19 and AsCEP20 arose through HGT or have de novo origins as 'orphan' genes. Although the CCP genome harbors homologs of AsCEP19 and AsCEP20, these homologs were weakly expressed at non-infection (spore suspension) or infection (in planta) stages. The total RNA-Seq read counts of the AsCEP19 homologs, BS50DRAFT_638061 and BS50DRAFT_64582, were all less than 30 in all the RNA-Seq samples, and no significant upregulation was detected in planta ( Fig. 4a). Only six read pairs were mapped to the AsCEP20 homolog BS50DRAFT_627105 in all samples, suggesting its insignificant role in CCP. Low expressions of the CCP homologs were similar, as reported (Lopez et al. 2018). However, as a remote homolog of AsCEP20, the CBP1 gene is actively expressed (Log2 fold change > 9, adjusted p-value < 0.001) in the yeast phase (parasitic form) of H. capsulatum; this result is consistent with previous work (Gilmore et al. 2015).

Expression profiles of AsCEP19 and AsCEP20 in solanaceous and non-solanaceous hosts
Expression levels of AsCEP19 and AsCEP20 were significantly upregulated in the leaves of host plants upon A. solani infection (Fig. 5) when compared with those in A. solani grown on potato dextrose agar (PDA) plates. 3) were shown in colors. Protein-coding genes and long terminal repeat (LTR) elements were shown as arrows and rectangles in colors, respectively. The GC content was shown along the selected region. In the 5' to 3' orientation, AsCEP19 is located between positions 244,287 and 243,852, and AsCEP20 is located between 246,214 and 246,637 Interestingly, this H2H gene pair showed positive correlation in expression under all conditions, with the 95% confidence interval of Pearson's correlation coefficient of 0.82-0.98 (p-value 1.06e −6 ). AsCEP19 and AsCEP20 were significantly upregulated at 24, 48, and 72 h postinoculation (hpi) in potato leaves (Fig. 5a), furthermore, high expression levels were also observed in the other two solanaceous hosts, tomato (Solanum lycopersicum) (Fig. 5b) and chili pepper (Capsicum annuum) (Fig. 5c), where AsCEP19 and AsCEP20 were consistently expressed at 48, 72, and 96 hpi, indicating that both of these effector genes play important roles in the interactions between A. solani and solanaceous hosts. In contrast, AsCEP19 and AsCEP20 showed low expression levels in the non-solanaceous host, Arabidopsis thaliana (Fig. 5d), where the average Ct values of the two genes were greater than 34.59 and 35.34, respectively. The different expression patterns of AsCEP19 and AsCEP20 in solanaceous and non-solanaceous hosts suggest that the expression of this gene pair is controlled strictly in a host-specific manner.

Transient expression of AsCEP19 and AsCEP20 in N. benthamiana
To better understand the roles of AsCEP19 and AsCEP20 in pathogenicity, the two effector genes were transiently expressed in N. benthamiana leaves via agroinfiltration. Only the full length (FL) form of AsCEP20 was able to trigger plant cell death in N. benthamiana, but not the signal peptide-deleted (ΔSP) form, indicating that the SP is essential for AsCEP20 to induce plant cell death (Fig. 6a). Plant cell death induced by AsCEP20 FL was also observed in tomato leaves (Fig. 6b). Neither the FL nor the ΔSP forms of these two effectors could suppress Fig. 4 Gene trees of AsCEP19 and AsCEP20 and species tree of the genus Alternaria. The best amino acid substitute models were determined using ProtTest. Maximum-likelihood trees were constructed using RAxML. a Phylogenetic trees constructed according to amino acid sequences of AsCEP19, AsCEP20 and their homologous proteins. The numbers shown next to the branches indicate the percentage of bootstrap support values (1000 replicates). The branch lengths indicate the evolutionary distance as the number of base substitutions per site. The rectangular boxes represent the changes in the gene expression during fungal infection. NA, data not available; b Phylogenetic tree constructed according to the alignments of 1,216 single-cpoy protein-coding genes from 37 fungal genome datasets. The 1,216 orthogroups were aligned using MAFFT, and these multiple alignments were trimmed and concatenated into a super-matrix INF1-triggered plant cell death in N. benthamiana (Additional file 2: Figure S3). Transient expression of AsCEP19 and AsCEP20 in N. benthamiana could promote P. infestans infection. No necrotic lesions were observed on the zone agroinfiltrated with the effector genes (AsCEP19 or AsCEP20) for 48 h. After inoculation with P. infestans, necrosis was observed at the inoculation site, and the necrotic zone enlarged gradually but not exceeded the agroinfiltration zone. The size of necrotic lesions on leaf tissue infiltrated with the AsCEP19 FL or AsCEP20 FL construct was significantly larger than that inoculated with the GFP control (Fig. 6c). Although AsCEP20 FL induces plant cell death, which suggests that it can act as a necrotrophic effector in apoplastic space, we cannot exclude the possibility that AsCEP20 might have other functions in the cytoplasm. To investigate whether AsCEP19 and AsCEP20 have a potential targeting site within plant cells, AsCEP19-YFP and AsCEP20-YFP fusion proteins were transiently expressed in N. benthamiana leaves via agroinfiltration. AsCEP19-YFP was observed in multiple subcellular compartments including the plasma membrane, nucleus, and cytoplasm, which resembled the free GFP control (Fig. 7). Interestingly, AsCEP20-YFP has a specific subcellular localization that was only found in the chloroplasts, suggesting that these organelles might be potential targets.

AsCEP19 and AsCEP20 contribute to the full virulence of A. solani
Gel electrophoresis of the PCR amplicons (Additional file 2: Figure S4) and their sequencing results (Additional file 1: Table S6) confirmed that the entire region spanning the H2H gene pair (AsCEP19 and AsCEP20) in A. solani HWC168 had been replaced by the hygromycin phosphotransferase gene (HPH). The deletion of AsCEP19 and AsCEP20 (ΔAsCEP19 + AsCEP20) from A. solani HWC168 led to reduced virulence on potato leaves (Fig. 8). The necrotic lesion (Fig. 8a) caused by the gene deletion mutant ΔAsCEP19 + AsCEP20 (1.81 ± 1.16 mm) was much smaller than that of the The expression levels of AsCEP19 and AsCEP20 in planta at different time points after inoculation were compared with those during mycelial growth of A. solani on PDA. Actin gene (ACTB) of A. solani was used as reference gene. Error bars represents means ± SD from five biological replicates. Two-tailed Student's t tests, *P < 0.05, **P < 0.01, and ***P < 0.001, ns means not significant Fig. 6 Transient expression of AsCEP19 and AsCEP20 in N. benthamiana and tomato leaves. a The full-length (FL) AsCEP20 (AsCEP20 FL ) triggered obvious necrotic lesion at seven days post-infiltration. The GFP and INF1 plasmid constructs were infiltrated as negative and necrosis control, respectively. b AsCEP20 FL elicited plant cell death in tomato leaves. c AsCEP19 FL and AsCEP20 FL promoted P. infestans infection in N. benthamiana. AsCEP19 FL , AsCEP20 FL , and GFP (empty vector) were individually agroinfiltrated into leaves for 48 h, and then the infiltrated leaves were detached from plants and inoculated with 10 μL zoospore suspension (5 × 10 4 zoospores/mL) of P. infestans. At 5 days post-inoculation (dpi), lesion size was measured, and the leaves were stained with DAB (Two-tailed Student's t test, * P < 0.05, ± SD, n = 6) wild-type (7.57 ± 1.74 mm), the p-value of a paired t-test (t = 16.16) was 2.5e −11 (Fig. 8b). Thus, this H2H gene pair contributes significantly to the full virulence of A. solani on host potato plants.

Discussion
In this study, effector candidates were identified in A. solani using a multiomics approach. Most of the AsCEPs were predicted to be apoplastic effectors (Fig. 1a), consistent with the lifestyle of A. solani in that the necrotrophic fungal genome is enriched with apoplastic effectors (Lo Presti et al. 2015;Sperschneider and Dodds 2022). On the whole, most of the AsCEP genes appear to be conserved and their homologs were also found in many other closely related species (Fig. 2a, b), however, based on the actual expression levels of AsCEP genes in planta, it was evident that A. solani largely relies on few lineage-specific effectors (Figs. 1b and 2a), but so far, their functions remain unclear. Of the lineage-specific AsCEP genes showing high expression levels, AsCEP19 and AsCEP20 appear to have been acquired by the common ancestor of A. solani and A. tomatophila via HGT based  Variation in virulence between the gene pair (AsCEP19 and AsCEP20)-knockout mutant and wild-type of A. solani on potato leaves. a Detached potato leaves inoculated with conidial suspensions of the wild-type strain (left) and the ΔAsCEP19 + AsCEP20 mutant (right). The sizes of necrotic lesions were measured at 5 dpi. b The average size of necrotic lesions caused by the ΔAsCEP19 + AsCEP20 mutant and the wild-type strain. Error bars indicate the standard deviation of the mean, and asterisks indicate statistical significance determined by paired t-test (t = 16.16, p-value = 2.5e −11 , df = 16, n = 17), with three (***) P < 0.001 on phylogenetic analysis and characteristics of gene loci, however, given the limited presence of AsCEP19 and AsCEP20 in the fungal kingdom, other possibilities, such as de novo gene birth, cannot be excluded (Figs. 3 and 4).
Both AsCEP19 and AsCEP20 have a high content of cysteine residues that might be involved in disulfide bond formation to enhance stability in a protease-rich apoplast environment . Furthermore, it is known that lineage-specific effector genes preferentially associate with AT-rich and gene-poor chromosomal regions (Testa et al. 2016), and AsCEP19 and AsCEP20 are no exceptions (Fig. 3). The presence of LTRs in nearby AT-rich regions suggests that this putative HGT event was very likely mediated by an LTR retrotransposon. However, we still lack conclusive evidence of the donor organism of AsCEP19 and AsCEP20. At present, both homologs of AsCEP19 and AsCEP20 are only known from the necrotrophic fungus C. cassiicola strain Philippines (CCP), a highly virulent isolate from the rubber tree (Lopez et al. 2018). In contrast to AsCEP19 and AsCEP20 organized in an H2H arrangement near AT-rich regions, the CCP homologs are located in conserved core genomic regions but on different scaffolds. None of the CCP homologs were truly expressed in host plants (the rubber tree), so it is possible that these two effectors were previously required for pathogenicity but no longer needed for infection of rubber tree and have become silenced. Avirulence (Avr) gene silencing has been shown to be an efficient mechanism for Phytophthora pathogens to evade effector-triggered immunity (Dong and Ma 2021). AsCEP20 contains a fungal calcium-binding domain, PF12192, which was originally identified in CBP1, a critical virulence factor for H. capsulatum known to cause histoplasmosis in human (Sebghati et al. 2000). AsCEP20 shares conserved cysteine residues involved in disulfide bond formation in CBP1 (Beck et al. 2008). Although the sequence similarity suggests that AsCEP20 is a distant homolog of CBP1, more studies need to be conducted to determine whether it still retains the function of CBP1: calcium binding and uptake.
Genes in H2H pairs are often coexpressed and functionally related (Chen et al. 2010). Overall, the H2H gene pair we identified here (AsCEP19 and AsCEP20) was highly expressed in planta but not in vitro. A previous study showed that A. solani conidia germination and germ tube elongation were observed at 12 hpi; penetration was first observed at 24 hpi and increased at 36 hpi (Dita et al. 2007). The expression level of the H2H gene pair was only slightly upregulated at 12 hpi but this was not significant (Fig. 5a). Thus, AsCEP19 and AsCEP20 do not belong to the first wave of secreted effectors, therefore their functions are probably insignificant in the initial colonization period when attached conidia germinate and grow on the potato leaf surface (0-12 hpi). However, the expression level of the H2H pair was upregulated markedly at 12-24 hpi, and remained high (Log2 fold change > 7) at 48 and 72 hpi. This result indicates that these two effectors play important roles at advanced stages, including penetration and hyphae spread subepidermally in plant leaves. Furthermore, the high expressions of AsCEP19 and AsCEP20 in other solanaceous hosts (Fig. 5b, c), chili pepper and tomato, were also confirmed, but not in the non-solanaceous host A. thaliana (Fig. 5d). Taken together, this coexpressed H2H gene pair is strictly regulated and involved in the interaction of A. solani and solanaceous hosts, and probably plays key roles to facilitate A. solani infection at post-penetration stages.
Transient expression in N. benthamiana showed that both AsCEP19 and AsCEP20 can promote the infection of P. infestans, indicating their roles in virulence (Fig. 6c). Their function in fungal virulence was further determined by comparing the pathogenicity between the gene deletion mutant and wild-type strain (Fig. 8). Our results confirmed that the H2H gene pair was required for the full virulence of A. solani towards host potato plants. Based on the phenotypes incurred by transient expression of AsCEP20 in N. benthamiana and tomato, AsCEP20 is probably a necrotrophic effector that requires its SP to induce plant cell death (Fig. 6a, b). Similar findings have also been reported in Rhizoctonia solani (Wei et al. 2020) and F. graminearum (Yang et al. 2021). AsCEP20 was predicted to be an apoplastic effector, but surprisingly, AsCEP20 ΔSP was specifically localized in chloroplasts (Fig. 7). The chloroplast is prime target of effector proteins since it is the major source of reactive oxygen species generation and a key component of early immune responses (de Torres Zabala et al. 2015). We cannot rule out the possibility that AsCEP20 might have a dual functionality. It is likely that AsCEP20 is secreted into the apoplastic space and subsequently re-enters the host cell cytoplasm, as similar traits have been reported in the Zymoseptoria tritici effector Zt6 (Kettles et al. 2018). Further studies are needed to uncover how AsCEP20 localizes to chloroplasts.
Effector genes acquired via HGT commonly play important roles in the evolution of fungal pathogens, by providing novel genetic materials involved in pathogenicity or associated with lineage specific niche adaptation Frantzeskakis et al. 2020). In contrast to core effector genes that are relatively conserved across fungal species, the effector genes acquired via HGT by A. solani might favor its host's adaptation to solanaceous hosts. HGT is a significant source of genetic variability, and is particularly important for those fungal species that lack sexual reproduction (Wang et al. 2019;Reinhardt et al. 2021), such as A. solani.

Conclusions
In this study, 238 effector genes were predicted from the A. solani genome. Comparative genomics and transcriptomics analysis revealed that A. solani has developed a sophisticated effector arsenal and relies on certain lineage-specific effectors to interact with solanaceous hosts. A pair of lineage-specific effector genes, AsCEP19 and AsCEP20, was identified and they seem to have arisen by HGT. Upregulation and coexpression of AsCEP19 and AsCEP20 during A. solani infection suggests their important roles in promoting the infection of solanaceous hosts and they might be functionally related. Deletion of this gene pair confirmed that AsCEP19 and AsCEP20 were required for the full virulence of A. solani towards host potato plants. Transient expression of AsCEP19 and AsCEP20 can facilitate the infection of P. infestans in N. benthamiana. AsCEP20 FL elicits plant cell death indicating that it is a necrotrophic effector, and AsCEP20 ΔSP can specifically localize to chloroplasts.

Computational prediction of effector genes in A. solani
First, gene prediction was performed on the genome of A. solani HWC168 (GCA_002837235.1) using BRAKER2 v2.1.6 (Hoff et al. 2015). The proteome of closely related species A. alternata SRC1lrK2f (GCA_001642055.1) was used as a reference genome. In addition, RNA-Seq data (PRJNA574559) from our previous study were also used to improve gene prediction (Additional file 1: Table S3). The RNA-Seq datasets were generated from potato leaves that were sprayed with spores of A. solani HWC168 and subsequently detached at 48 hpi. The transcriptome of A. solani HWC168 was assembled de novo from RNA-Seq data using Trinity v2.12.0 (Grabherr et al. 2011); then alternative splicing variations were determined using PASA v2.4.1 (Haas et al. 2003). Finally, a consensus gene structure of A. solani HWC168 was generated by EVi-denceModeler v1.1.1 (Haas et al. 2008) using evidence from BRAKER2 gene prediction and de novo transcripts.
Genes encoding CEPs were identified from the protein-coding genes of A. solani HWC168. A bioinformatics pipeline was established, which was modified from the general pipeline for prediction of candidate secreted effector proteins described in a previous review (Dalio et al. 2018). The pipeline consists of four steps and is executed in the following order: i) prediction of putative secreted proteins using SignalP v6.0 (Teufel et al. 2022) and WoLF PSORT (Horton et al. 2007); ii) removal of membrane proteins with TMHMM v2.0 (Krogh et al. 2001); iii) removal of glycosylphosphatidylinositol (GPI) anchors using NetGPI (Gíslason et al. 2021); and iv) identification of CEPs by EffectorP v3.0 (Sperschneider and Dodds 2022). These AsCEPs were functionally annotated by BLASTp against the UniProtKB database, and protein domains were identified by SMART searching (e-values < 1e −5 ).

Synteny conservation between A. solani strains
Intraspecific variation in A. solani is still unknown, A. solani strains from different origins might possess dramatic structural variations in local genome architecture, especially in the regions where the effector genes are located. Therefore, long-range synteny between the genome assemblies of A. solani strains was analyzed using MCScanX (Wang et al. 2012), and local gene collinearity between two strains were further examined. Although three A. solani strains are available in NCBI and JGI MycoCosm, unfortunately the genome assembly of strain BMP0185 (JGI) consists of more than 2,000 contigs which was too fragmented to be included in the analysis. Therefore, synteny conservation between the genomes of strains NL03003 and HWC168 was evaluated.

Identification of PAV of AsCEP genes in Alternaria species
Genome sequences of Alternaria species available from GenBank and JGI MycoCosm (as of March 2022) were used to determine gene orthologs using OrthoFinder v2.3.8 (Emms and Kelly 2019). To prioritize the selection of genome assemblies, the completeness of genomes was estimated using BUSCO (Manni et al. 2021), and complete or near complete genome assemblies were chosen for ortholog clustering. Of the Alternaria genome assemblies lacking gene annotations in GenBank, these assemblies were annotated using BRAKER2 v2.1.6 (Hoff et al. 2015). The ortholog data matrix was converted into a binary matrix, then the presence of AsCEP genes in other Alternaria species was calculated. To determine whether lineage-specific effector genes sit within or close to ATrich RIP mutation hotspots, AT-rich and GC-equilibrated regions were identified using OcculterCut v1.1 (Testa et al. 2016). Any TEs within the AT-rich regions were detected using RepeatModeler v2.0.1 (Flynn et al. 2020).

Presence of AsCEP19 and AsCEP20 outside of Alternaria
To test whether AsCEP19 and AsCEP20 originated via HGT, the gene trees of AsCEP19 and AsCEP20 were compared with the species tree of Alternaria. Potential homologs of AsCEP19 and AsCEP20 were determined by BLASTp against UniProtKB and NCBI nr, and any BLASTp-hit with identity greater than 40% and coverage more than 70% over the query was retained.
The sequences of AsCEP19 and AsCEP20 and their homologs were aligned using MAFFT v7.453 (Standley 2013). To construct a species tree, Alternaria species were compared with each other and 12 fungal species from the order Pleosporales were used as the outgroup. Protein-coding genes were clustered into orthogroups using OrthoFinder v2.3.8 (Emms and Kelly 2019). In total, 1,216 single-copy orthogroups from all the fungal genomes were aligned using MAFFT, and then the multiple alignments were trimmed and concatenated into a super-matrix. The best amino acid substitute models for the alignments of AsCEP19, AsCEP20, and the concatenated super-matrix were determined using ProtTest v3.4.1 (Darriba et al. 2011). Maximum-likelihood trees were constructed using RAxML v8.2.12 (Stamatakis et al. 2005).

RNA-Seq data analysis
To determine whether homologs of AsCEP19 and AsCEP20 might play potential roles in fungal virulence, the RNA-Seq datasets were retrieved from NCBI SRA (Additional file 1: Table S3), and the expression levels of homologs were determined. Adaptor and quality trimming of raw RNA-seq datasets were performed using fastp v0.20.1 (Chen et al. 2018a, b). The trimmed reads of each sample were mapped to the corresponding fungal genomes (Additional file 1: Table S3) using HISAT2 v2.1.0 (Kim et al. 2015). Mapping results were converted to BAM format and then sorted using Samtools v1.12 (Li et al. 2009). Reads counts for each gene were calculated using the htseq-count tool in HTSeq v0.11.2 (Anders et al. 2015). Differential expression of AsCEP19 and AsCEP20 homologs between samples at infectious and noninfectious phases was determined using the R package DESeq2 v1.28.1 (Love et al. 2014).

Fungal isolates, plants, and culture conditions
A. solani strain HWC168 was grown on PDA plates at 25 °C in the dark, and mycelia were harvested after 8 days. To induce sporulation, strain HWC168 was grown on tomato juice agar plates in the dark at 25 °C for 8 days, and aerial hyphae were scraped off with a scalpel. The plates were subsequently exposed to UV light for 10 min and then kept in the dark at 25 °C/20 °C (12 h/12 h) for 3 days. For obtaining conidial suspensions, conidia were harvested with sterile double distilled (dd)H 2 O and centrifuged at 1,970 g for 10 min, then conidia were diluted to 10 5 conidia/mL. P. infestans was grown on rye agar in the dark for 10 days, and mature sporangia were harvested with 2 mL of sterile ddH 2 O. Sporangial suspensions were kept in the dark at 4 °C for 2 h, and then at 18 °C in the dark for 1 day to release zoospores. Potato plants (cv Favorita) were grown in a greenhouse at 24 °C with a 16/8 h light/dark cycle for 8 weeks. N. benthamiana plants were grown in a growth chamber at 25 °C and 50% relative humidity with 12/12 h light/dark cycles for 5 weeks. A. thaliana plants were grown on MS culture medium at 4 °C for 3 days, and then placed in a growth chamber at 22 °C with 16/8 h light-dark cycles. Tomato (cv Monkeymaker) and chili pepper (cv Chaotianjiao) plants were grown in a greenhouse at 24 °C with 16/8 h light-dark cycles.

Gene expression analyses using reverse transcription-quantitative polymerase chain reaction (RT-qPCR)
Expression profiles of AsCEP19 and AsCEP20 in strain HWC168 were determined during different stages of infection. Leaves of potato, A. thaliana, tomato, and chili pepper were surface-sterilized with 70% ethanol for 30 s, rinsed three times with sterile ddH 2 O, dried on filter paper, and transferred to wet filter papers placed on 1% water agar in Petri dishes. Leaves were inoculated with 20 μL of A. solani conidial suspension or sterile ddH 2 O (mock inoculation). RNA was extracted from mycelia grown in PDA and also from detached potato leaves inoculated with conidia at 12, 24, 48, and 72 hpi using EasyPure Plant RNA Kit (TransGen Beijing, China), and from detached Arabidopsis thaliana, tomato, and pepper leaves inoculated with conidia at 48, 72, and 96 hpi, and then treated with DNase I (TransGen). First-strand cDNA was synthesized from mRNA using TransScript First-Strand cDNA Synthesis SuperMix (TransGen) according to the manufacturer's suggestions. The gene encoding β-actin (ACTB) was used as an internal control. qPCR was performed on a C1000 thermal cycler equipped with a CFX96 real-time PCR detection system (Bio-Rad, CA, USA). PCR was performed with Mag-icSYBR Mix (CoWin BioSciences, MA, USA) and specific primers (Additional file 1: Table S7). Relative gene expression in the samples was calculated by the ddCt method (Livak and Schmittgen 2001). Two-tailed Student's t tests were used for comparisons between means.

Agrobacterium-mediated transient expression
To determine the phenotypic alterations induced by AsCEP19 and AsCEP20, two forms of protein, FL and ΔSP, were transiently expressed in N. benthamiana plants by agroinfiltration. The SP cleavage sites in AsCEP19 and AsCEP20 were predicted using SignalP. The INF1 elicitin of P. infestans (Kamoun et al. 1998) was used to induce plant cell death. Nucleotide sequences of AsCEP19 FL , AsCEP19 Δsp , AsCEP20 FL , AsCEP20 Δsp , and INF1 were obtained by gene synthesis (Stargene, Wuhan, China), and all sequences were tagged with GFP at their C terminus. Then, nucleotide sequences fused with the GFP coding sequence were cloned into the plant expression vector pCAMBIA1301. The recombinant vectors were sequenced to ensure correct insertion. A. tumefaciens strain EHA105 transformed with a recombinant vector was grown overnight at 28 °C in LB medium containing appropriate antibiotics. A. tumefaciens cells were pelleted, washed, and resuspended in infiltration buffer (10 mM MgCl 2 , 10 mM MES, 150 μM acetosyringone). Five-week-old N. benthamiana leaves were infiltrated with Agrobacterium using a syringe without a needle. To determine whether AsCEP19 and AsCEP20 might promote the infection of pathogens, AsCEP19 or AsCEP20 were expressed in N. benthamiana leaves via agroinfiltration prior to pathogen inoculation. Agrobacterium carrying GFP and effector gene (AsCEP19 or AsCEP20) plasmid constructs were injected to the left and right side of the leaf, respectively. Leaves were detached at 48 h after agroinfiltration, and subsequently inoculated with 10 μL zoospore suspension (5 × 10 4 zoospores/mL) of P. infestans and kept in Petri dishes for 5 days. The detached leaves of N. benthamiana were decolorized using 95% ethanol in a water bath at 100℃ for 15 min, then stained with DAB. The sizes of necrosis lesions on the leaves of N. benthamiana were measured and compared using paired Student's t test (Fig. 6c).

Subcellular localizations of AsCEP19 and AsCEP20 in N. benthamiana leaf cells
Determining the subcellular localization of fungal effectors in host plant cells can reveal clues about the mechanism of virulence. The AsCEP19 and AsCEP20 gene sequences were tagged at the C terminus with the yellow fluorescent protein-coding gene (YFP), and were further cloned into pCAMBIA1301 for expression in N. benthamiana. The recombinant vectors were sequenced to ensure correct insertion and then were transformed into A. tumefaciens strain GV3101. A. tumefaciens cells were resuspended in infiltration buffer. In addition, A. tumefaciens harboring a pCAMBIA1301-YFP plasmid was infiltrated into N. benthamiana leaves as control. The infiltrated leaves were visualized with a LEICA FCS SP8 fluorescence microscope (Leica, Germany) at 3 days post-infiltration.

Gene deletion by homologous recombination
A gene deletion mutant of A. solani HWC168 was generated for the contiguous region spanning the gene loci AsCEP19, AsCEP20, and the region between them. To replace the target region with HPH gene, a 1009 bp downstream flanking sequence of the AsCEP19 gene (AsCEP19RR) and a 914 bp downstream flanking sequence of the AsCEP20 gene (AsCEP20RR) were amplified with primer pairs 1920-RR-F/R and AsCEP20-RR-F1/ R1, respectively (Additional file 1: Table S6). HPH gene fragment was amplified from the vector pEASY-HPH with primers HPH-F3/R1 (Additional file 1: Table S6). All PCR reactions were performed using Super Pfx DNA polymerase (CWBIO, China). Three PCR amplicons were fused in the order 'AsCEP19RR-HPH-AsCEP20RR' , and then cloned into a pUC19 vector using pEASY-Uni Seamless Cloning and Assembly Kit (TransGen, China). The fused fragment was subsequently amplified from the recombinant plasmid by PCR for transformation. Protoplasts of A. solani were prepared and transformed with the PCR product of the fused fragment using the polyethylene glycol (PEG4000)-mediated method (Goswami 2012). The transformants were cultured on hygromycin-resistant PDA for three generations, and the gene knockout mutants were identified by PCR using primer pairs 1920-LRHPH-F/R and 1920-RR-F-HPH/1920-RR-HPH-R (Additional file 1: Table S6), which targets the regions of AsCEP19RR fused with HPH and AsCEP20RR fused with HPH, respectively. To compare the virulence between the gene knockout mutant and wild-type strain in host potato, detached potato leaves (n = 17) were inoculated with conidial suspensions (10 5 conidia/mL) of the tested strains. The necrotic lesions were measured at 5 dpi, and paired t-test was performed in R v4.1.3 to determine whether there was any significant difference in virulence between the mutant and wild-type strain. Similar results were obtained from two independent experiments.