Stacking Resistance Genes in Multiparental Interspeciﬁc Potato Hybrids to Anticipate Late Blight Outbreaks

Resistance Genes in Multiparental Interspeciﬁc Potato Hybrids to Anticipate Late Outbreaks. Abstract: Stacking (pyramiding) several resistance genes of diverse race speciﬁcity in one and the same plant by hybridization provides for high and durable resistance to major diseases, such as potato late blight (LB), especially when breeders combine highly efﬁcient genes for broad-spectrum resistance that are novel to the intruding pathogens. Our collection of potato hybrids manifesting long-lasting LB resistance comprises, as a whole, the germplasm of 26 or 22 Solanum species (as treated by Bukasov and Hawkes, respectively), with up to 8–9 species listed in the pedigree of an individual hybrid. This collection was screened with the markers of ten genes for race-speciﬁc resistance to Phytophthora infestans ( Rpi genes) initially identiﬁed in S. demissum ( R1, R2, R3a, R3b, and R8 ), S. bulbocastanum/S. stoloniferum ( Rpi-blb1/ Rpi-sto1, Rpi-blb2, Rpi-blb3 ) and S. venturii ( Rpi-vnt1 ). The hybrids comprised the markers for up to four-six Rpi genes per plant, and the number of markers was signiﬁcantly related to LB resistance. Nevertheless, a considerable portion of resistance apparently depended on presently insufﬁciently characterized resistance genes. Bred from these multiparental hybrids, the advanced lines with the stacks of broad-speciﬁcity Rpi genes will help anticipate LB outbreaks caused by rapid pathogen evolution and the arrival of new pathogen strains.


Introduction
Persistent and unrelenting, late blight (LB) of potato (Solanum tuberosum L.) caused by the oomycete Phytophthora infestans (Mont.) de Bary levies a permanent tax on potato growers: up to $10 billion is lost annually as direct crop losses and costs of chemical protection; the losses rise dramatically in the years of epidemic disease development [1][2][3]. The most economical and environment-friendly way to effectively contest and contain LB is to breed new cultivars with durable resistance. Durable resistance is empirically defined as resistance efficient over long periods of widespread crop cultivation under conditions favorable to disease, a compromise between plant defense capacity and the evolutionary potential of the pathogen [4]. Such resistance is reached by transferring the genes for resistance to P. infestans (Rpi genes) into cultivated potato [5]. Wild potatoes readily supply the necessary germplasm, and multiple Rpi genes have already been introgressed into marketable cultivars by the marker-assisted sexual and somatic hybridization or with the technologies of genetic engineering [3,[5][6][7][8][9][10][11]. This resistance is gained slowly and with hard labor-and can be disappointedly lost, sometimes within few years, due to the rapid evolution of P. infestans genome and the arrival of new pathogen strains with novel repertoire of (a)virulence genes (Avr genes) [1][2][3]6,10,11].
An efficient strategy to overcome or at least alleviate this problem, when aiming at long-lasting and durable LB resistance, is to combine in one plant the Rpi genes that recognize several Avr genes. This strategy is called gene stacking, or pyramiding. Such gene pyramids will remain sustainable and effective as long as at least one Rpi component of the pyramid can recognize the corresponding Avr gene of the pathogen and trigger the defense response. Theoretically, a pyramid of four resistance genes would withstand pathogen invasion-on condition that both the resistance gene pyramids and the colonizing pathogen population(s) would concurrently fulfill several criteria. First, the stacked resistance genes should be highly effective and not leaky. Second, the best resistance genes and their combinations are those truly novel to the infecting pathogen population. Third, the pathogen genome should only rarely recombine, a criterion easily met only in a primarily asexual population. Fourth, the resistance will stay durable at a low level of gene flow due to pathogen migration [1,6,[12][13][14][15][16].
In the case of potato, the most evident way to achieve long-lasting resistance against P. infestans is to recruit new Rpi genes into breeding schemes and to stack as many Rpi genes as possible into a single cultivar. The genetic diversity of cultivated potatoes that may provide such resources has been substantially pauperized in the process of conventional breeding [6,[9][10][11]. Therefore within the last two decades, combining multiple resistance genes into a single plant genotype has heavily relied on the identification and cloning of Rpi genes of interest from the vast resource offered by wild Solanum species. Particularly inviting sources of germplasm enhancement are insufficiently explored South American wild potatoes, which have not been conspicuously involved in practical breeding, and the species that were never before reported to resist LB [7,[9][10][11][17][18][19][20].
In the past centuries preceding the informed breeding, many cultivated genotypes in Mexico and South America had already harbored a significant contribution of wild germplasm [9]. Current germplasm enrichment by identifying and introgressing new Rpi genes and new alleles of already known Rpi genes must also include careful study of the gene pools presently used by breeders. Among other things, such exploration would lower the chance to undermine the efforts of breeders if they deploy Rpi genes that have already been broken by local pathogen strains [3,6,9,10,[12][13][14][15][16].
The search for new Rpi genes and new alleles of previously characterized Rpi genes (allele mining) brings us to the mission of a wider span: identification of the full complement of Solanum genes contributing to the resistance to P. infestans [7,10,17,[19][20][21]. For more than three decades, this field was successfully searched using various DNA markers [5,7,9,[21][22][23]. Later, over 20 Rpi genes were identified and cloned from wild Solanum species. Recent breakthrough technologies of resistance gene enrichment sequencing (RenSeq) and the diagnostic version of this technology (dRenSeq) have opened new vistas to comprehensive exploration of Solanum Rpi genes and their introduction to advanced breeding schemes [24,25]; in addition, these technologies facilitate the wide-ranging characterization of allelic diversity enabling the evolutionary analysis of Rpi genes and prediction of new sources of these genes in genetic collections.
The multiparental potato hybrids described in this paper were obtained by remote crosses and combine genetic material from 20 wild and two cultivated Solanum species as treated by Hawkes [26], with up to 8-9 species reported per single hybrid pedigree. For over a decade, many of these hybrids and derived advanced lines have manifested high LB resistance. They are prospective donors containing pyramids of broad specificity genes that nowadays are readily involved in breeding, such as Rpi-blb1 = Rpi-sto1, Rpi-blb2, Rpi-vnt1, R2 = Rpi-blb3, etc. An important advantage of these breeding donors is that the introgressed Rpi genes maintain the genetic environment inherited from parental forms, including race-nonspecific resistance genes [5,18]. Rather than single genes, the remote crosses would transfer whole clusters of genes combining the Rpi genes of diverse race specificity and even the genes for resistance concurrently to several pests. These hybrid characteristics would ensure the stability of future cultivars and slow down the onset of more adapted pathogen forms in potato stands [5,17,18]. Here we present the evidence obtained with the markers of ten Rpi genes characterized in more detail. Some data presented below have been reported earlier [27,28] at the Euroblight workshops (https://agro.au.dk/forskning/internationale-platforme/euroblight/).

Plant Material
The plant material explored in this study is predominantly represented by multiparental interspecific hybrids. The pedigrees of these hybrids combine from two to nine species of Solanum L., section Petota Dumort. (Table 1). The sample under study includes ten hybrids with high field resistance to LB bred by I.M. Yashina at the Russian Potato Research Center (Korenevo, Moscow region), hereinafter Yashina's hybrids [29], by crossing demissoid potato varieties and/or breeding lines, which were backcrosses comprising the genetic material of S. andigenum Juz. & Buk. (=S. tuberosum ssp. andigena Hawkes), S. chacoense Bitt. and S. chilotanum (Buk. & Lechn.) Hawkes (=S. tuberosum ssp. tuberosum L.). Hereinafter, the names of Solanum species in the pedigrees of hybrids (Table 1) are listed according to Hawkes [26] and those of cultivars follow the information provided by breeders. Ten hybrids originally obtained by V.A. Kolobaev at the Institute of Plant Protection (Pushkin, St. Petersburg), hereinafter Kolobaev's hybrids [30], were bred using the accessions of wild Solanum species from the VIR collection, which were previously recognized as the sources of high LB resistance: S. berthaultii Hawkes, S. pinnatisectum Dunal., S. polytrichon Rydb., S. simplicifolium Bitt., and S. verrucosum Schlechtd. Thirty seven hybrids produced by E.V. Rogozina at VIR, hereinafter Rogozina's hybrids [30,31], represent two-species hybrids and backcrosses with the participation of South American species S. alandiae Cárd. and S. okadae Hawkes & Hjert., which have not been previously involved in potato breeding. They also include the hybrids obtained by crossing potato cultivars and breeding lines comprising the genetic material of cultivated and wild potato species It is significant to emphasize that the development of these hybrids involved many South American species rarely used by the Russian breeders. Many of these hybrids bred over several decades are particularly important as they possibly preserved the Rpi alleles that could have been lost in the world collections of wild Solanum species due to genetic drift and loss of individual accessions.  [26,32,33]. In addition to all these hybrids, our study also included several registered varieties, some of which come from complex interspecific hybrids: Alouette (https://varieties.ahdb. org.uk/varieties/view/Alouette; [25]), Sarpo Mira and Sarpo Axona (http://sarpo.co. uk/portfolio/; https://pomidom.ru/sarpo-mira-potatoes/), Mastenbroeck-Black potato differential set plants R5, R8 and R9 [34] and others. These varieties were used to verify SCAR markers of the Rpi genes; besides, they also served as positive and negative controls for PCR screening. Other cultivars, some of which are also interspecific hybrids, were employed as controls when assessing plant LB resistance ( Table 2).

Series in the Section
As a whole, this collection (Tables 1 and 2) contains potato hybrids with their pedigrees representing nine series of Solanum L. section Petota Dumort. and listing 22 wild and four cultivated Solanum species as treated by Bukasov [32], which correspond to 20 wild and two cultivated Solanum species as treated by Hawkes [26] and 15 wild and one cultivated species as treated by Spooner [33]. The pedigrees of some individuals include as many as nine Solanum species. To verify SCAR markers of Rpi genes we also employed the accessions of wild Solanum species in the VIR collection.

Resistance to Pathogens
Late blight resistance of leaves was evaluated in long-term field trials under conditions of natural infestation in two European regions of the Russian Federation, i.e., the Northwest ( In the Northwest region, the growing seasons during the period of field trials were different: in 2016 and 2017, abundant precipitation and cool temperatures were favorable for the early manifestation and development of LB; in 2014, 2019 and 2020, moderate rainfall and unstable temperatures delayed the appearance of disease. In the Central region, dry weather early in the 2014 growing season delayed the LB progress; however, heavy rainfall and a drop in temperature early in August provided extremely favorable conditions for the LB development on potato haulms and later, damage to tubers. Through the following six years (2015-2020), the weather conditions were favorable for a fairly early (the middle of June) LB development, and later LB epiphytoty. Within this period, the air temperatures in June and the first half of July were below the long-term values. In addition, significant precipitation was recorded annually.
Pathogen population at two sites was represented by numerous diverse and highly aggressive complex races of P. infestans comprising seven to eleven virulence genes [35].
The field assessment of the partial LB resistance of potato plants was carried out every 10-12 days, and these data were used to calculate the area under the disease progress curve (AUDPC) in the course of the growth season and the corresponding yield losses caused by the early destruction of leaves (%). To evaluate the LB resistance level, the calculated yield losses were converted into 1-9-point scores, where 9 points correspond to the highest resistance level [35].
Resistance to LB in the laboratory tests was evaluated with detached leaves according to the Eucablight protocol (www.euroblight.net/). Detached leaves of plants grown in a greenhouse were infected with a highly virulent and aggressive isolate of P. infestans N161 (race 1.2.3.4.5.6.7.8.9.10.11, mating type A1) isolated in the Moscow region (the collection of the Institute of Phytopathology), using cv. Santé as a reference [35]. The aggressiveness of N161 in the Lapwood test [36] with Santé tubers exceeded the indices registered with all isolates collected in the potato stands under study. The experimental data for LB resistance were transformed to 1-9-point scores.

SCAR Markers for Resistance Genes
All SCAR markers (Table 3, Figure 1) were derived from the sequences of already well-characterized Rpi prototype genes deposited in the NCBI Genbank (https://www. ncbi.nlm.nih. gov/nucleotide/). Most markers were already reported elsewhere, and some were designed or modified following multiple alignment of the prototype gene sequences, their structural homologues and anonymous genome fragments lifted from the NCBI Genbank using BLAST and Vector NTI Suite 8 package (Invitrogen, Carlsbad, CA, USA). In the case of R2/Rpi-blb3 and Rpi-blb1/Rpi-sto1, more than one marker was used to recognize the particular gene. Wherever possible, marker specificity was verified against wild species that were the initial sources of the prototype genes in the NCBI Genbank, including amplification, cloning and sequencing the marker amplicons and phylogenetic analysis of the marker sequences. To this end, multiple alignments of nucleotide sequences assembled using a combination of the Martinez and Needleman-Wunsch algorithms were performed with SeqMan, Lasergene 7.0 Sequences. The phylogenetic analysis was performed with MEGA6 (https://www.megasoftware.net/). Table 3. SCAR markers of Solanum Rpi genes (see also Figure 1).

LB Resistance of the Multiparental Potato Hybrids
In the field experiments, 50 hybrids and cultivars were assessed for their LB resistance in the span of seven years (2014-2020). Ten Yashina's hybrids, ten Kolobaev's hybrids and 23 Rogozina's hybrids were evaluated together with seven standard cultivars. For the sake of comparison, another seven cultivars (Alouette, Atzimba, Elizaveta, Nayada, Priekul'skij rannij, Svitanok kievskij and Zagadka Pitera) were tested in field trials for four years within the 2014-2020 period. The cvs. Alouette (8-9), Sarpo Mira (8) Table 2). The data from field trials and laboratory assessments run in parallel for many years are in good agreement (Spearman's correlation coefficient r = 0.75 at p < 0.05).
Based on the evidence from long-term field trials and laboratory assessment, 55 hybrids and cultivars were grouped in the following way (Figures 2 and 3). Full coupling grouping using Euclidean distance and k-means clustering gave similar results. By the hierarchical classification, the sample of 55 hybrids and cultivars is separated into three groups with a similarity level >0.4. The k-means method also formed three disjoint subsets: each cluster consists of similar objects, and objects from different clusters differ significantly from each other. Cluster 1 comprises potato genotypes, which are moderately resistant to LB in the field trials and moderately susceptible in laboratory tests: cvs Nayada and Zagadka Pitera
To recognize the Rpi-R2/Rpi-blb3 genes, we used three SCAR markers corresponding to different regions of this gene (Figure 1, Table 3). Marker Rpi-R2-686 covers about half of the Rpi-R2-1137 sequence, and the evidence for these two markers matches in most cases ( Table 4). The third marker Rpi-blb3-305 usually follows Rpi-R2-1137. The Rpi-R2/Rpi-blb3 family of genes in the cluster on chromosome 4 has been reported in many Mexican species [20,40,47,48], and to distinguish the input of particular germplasms in the interspecific hybrids should become the goal of future studies. It is difficult to explain the presence of Rpi-R2 markers in S. chacoense × S. okadae hybrid (135-3-2005)-especially when other segregants of this combination are free of these markers. Table 4. Markers of Rpi genes in multiparental interspecific hybrids and reference potato cultivars (1/0-presence/absence of markers).               Rpi-sto1-890
The marker Rpi-R3a was previously reported in several Demissa and Longipedicellata species and also in S. microdontum [21]. The functional Rpi-R3a analogues were found in several series of Petota species with the effectoromics technology [47], whereas the complete Rpi-R3a cdc was cloned from S. stoloniferum (Genbank accession HQ731037). Two genes, Rpi-R3a and Rpi-R3b, are located in one cluster on chromosome 11, and their markers go together in most though not all hybrids (Table 4).
Using the effectoromics technology to mine cv. Sarpo Mira, Rietman et al. [40] reported five Rpi genes: Rpi-3a, Rpi-3b, Rpi-R4, Rpi-Smira1 (Rpi-R9) and Rpi-Smira2 (Rpi-R8). Our marker analysis of this cultivar confirmed the presence of genes Rpi-3a, Rpi-3b, and Rpi-R8. All these genes were most probably transferred from S. demissum and S. stoloniferum (https://pomidom.ru/sarpo-mira-potatoes/). Now, let us turn to one more gene assayed with several SCAR markers: Rpi-blb1/Rpi-sto1. Two markers, Rpi-blb1-821 иRpi-sto1-890, which cover different regions of the gene sequence (Figure 1), perfectly concurred in a range of Bulbocastana and Longipedicellata accessions [49] and now in most hybrids containing the genetic material of these species (Table 4). In addition to the predictable presence of the markers Rpi-blb1-821 and Rpi-sto1-890 in such hybrids, these markers were unexpectedly found in the Atzimba × S. alandiae hybrid 39-1-2005. Only single marker Rpi-blb1-821 was found in cvs Priekulskiy rannij and Svitanok kievskij. Previously this marker was also reported in a highly resistant accession VIR5399 of S. microdontum [49].The short marker Rpi-blb1-226 usually accompanied two longer markers of the gene; however, Rpi-blb1-226 alone was found in four genotypes that contained Longipedicellata genetic material (113 (50/1  Our collection lacks the hybrids with the genetic material of S. venturii. However, the Rpi-vnt1 analogues and pseudogenes are widely distributed in South American Tuberosa species, including S. microdontum and S. okadae [45]. Indeed, we registered one allele of this gene, Rpi-vnt1-3, in two thirds of hybrids containing the germplasm of S. alandiae and S. microdontum: the comparison of this allele sequence to that of the prototype Rpi-vnt1 gene indicated 92-98% identity [50]. In addition, the S. alandiae genome comprised the structural homologues of R2/Rpi-blb3, R8, R9a, Rpi-vnt1 and Rpi-blb2; respective homologues were 94-99, 94-99, 86-89, and 91% identical with the prototype genes [50]. It is also relevant to mention that the complete Rpi-vnt1-like sequence was cloned from S. microdontum ssp. gigantophyllum (Genbank accession GU338312). We failed to find the marker Rpi-vnt1.3-612 in all hybrids comprising S. okadae genetic material (Table. 4), whereas this marker was found in the S. okadae accession k-25397-1 different from the accession к-20921 used as the male parent of the hybrids [50].

LB Resistance is Enhanced by Pyramiding Rpi Genes
The numbers of Rpi genes combined in particular potato hybrids are clearly in line with plant LB resistance in the field experiments. We compared LB resistance in field trials in cultivars and hybrids in two contrasting subsets of potato genotypes: those containing only one Rpi gene and those with five genes. The former subset of nine genotypes comprises six cultivars (Desiree, Bintje, Alpha, Negr, Eersteling, and Robijn) and three hybrids (134-3-2006, 2585-80, and 97.1.17), wherein only one Rpi gene, either Rpi-R2/Rpi-blb3 or Rpi-R8, was found (Table 4). In the latter subset of 18 genotypes five-six genes were recognized (Tables 4 and 5). Two subsets significantly differ in their LB resistance in field trials by the Mann-Whitney criterion: U observed = 33 < U critical = 42 at p < 0.05. The Spearman' correlation coefficient (R observed = 0.514 > R critical = 0.382 at p < 0.05) is another proof of statistically significant relationship between the number of Rpi genes and LB resistance in these subsets of potato cultivars and hybrids.
Mundt [14] demonstrated that under optimal conditions, a stack of four efficient resistance genes would provide a durable protection against the pathogen. We therefore focused on the genotypes that comprised four and more Rpi genes per plant (Table 5). Over 80% of these hybrids, together with the cultivars derived from multiparental hybrids, manifest significant and long-lasting field resistance to LB (6 points and higher). The predominant resistance genes of these genotypes are demissoid Rpi-R3b (with the frequency of 0.79), Rpi-R2/Rpi-blb3 (0.74), Rpi-R8 (0.66), and Rpi-R3a (0.59); the frequencies of other genes are 0.41-0.44 (Table 5).

Conclusions
High and long-lasting LB resistance is a major prerequisite for sustainable potato production. In this project, a considerable collection of potato interspecific hybrids and standard cultivars was assayed with SCAR markers for ten Rpi genes, and plant LB resistance was evaluated in the field trials and laboratory tests with detached leaves. These hybrids combine several Rpi genes that are currently in high demand with potato breeders, such as Rpi-R2/Rpi-blb3, Rpi-blb1/Rpi-sto1, Rpi-blb2, and Rpi-vnt1. The level of LB resistance manifested by these hybrids is significantly related to the number of Rpi genes stacked in a single hybrid. This evidence seems to support the concept of pyramiding Rpi genes for durable LB resistance. However, when the patterns of gene stacking are examined with SCAR markers, it seems proper to focus on several caveats.
First, a considerable portion of resistance manifested by the investigated hybrids was not associated with the markers used in this study, and we believe that such resistance depended on some new or insufficiently characterized Rpi genes, which are not recognized by the markers employed to screen the hybrids. To exemplify such possibility, S. chacoense germplasm is found in many hybrids examined in the present study (Tables 2 and 4), and some of their LB resistance could be related to the Rpi-chc1 gene [7]. Indeed, screening such hybrids with the marker for this gene developed in our laboratory produced the positive signal in hybrids 2372-60, 2522-173 and 2584-7 but not in 2359-13. Among five S. okadae k-20921 × S. chacoense k-19759 hybrids, only 135-3-2005 was positive, other four segregants of this hybrid and the accession S. chacoense k-19759 itself responded negatively (M. Beketova, personal communication). Another possibility would link such resistance to other defense pathways, including non-specific tolerance.
Second, in such a complex assortment of genetic material, the gene stacks may comprise several alleles of one and the same gene introgressed from different Solanum species, e.g., S. chacoense, S. demissum, S. pinnatisectum, S. phureja, S. stoloniferum, etc. [7,20,47,51,52]. It is not always possible to distinguish such alleles. At least, in this study, by using the markers that reliably discriminate between demissum and stoloniferum alleles of Rpi-R1 [53], we demonstrated that nine hybrids combining demissum and stoloniferum germplasms comprised only the former allele of Rpi-R1 and were devoid of the latter.
Third, the SCAR markers employed in this study do not stretch over the full-size sequences of candidate genes, especially in the case of short markers Rpi-R3b-378 and Rpi-blb3-305. The changes in the candidate gene under study beyond the region covered by the particular marker would render this gene inactive. Perhaps, the presence of pseudogenes would explain the occurence of markers of Rpi genes in the standard cultivars believed to be devoid of such genes: Rpi-R1 in cv. Magellanes, Rpi-R2 in cv. Robijn, Rpi-R8 in cvs Alpha, Desiree, and Eersteling, Rpi-blb2 in cvs Magellanes and Early Rose, and Rpi-vnt1 in cvs Bintje and Early Rose (Table 4). Similarly, when the presence of markers in the hybrids is not supported by their pedigrees, such discrepancy can be explained by the presence of inactive homologues. In support of these suggestions, the BLAST search recognized the homologues of all these genes except Rpi-vnt1 in a true S. tuberosum cv. Solyntus [54] (the corresponding Genbank accessions CP055238, CP055237, CP055242, CP055241, and CP055239).
Fourth, even when the complete sequences of candidate genes are assessed (e.g., with the dRenSeq technology [25]), the proof for their functionality must be obtained by independent methods, such as effectoromics [40,47,55].
There are two ways to combine a sufficient number of Rpi genes of broad specificity towards diverse pathogen races and in this way to develop the basis of long-lasting and durable LB resistance: to stack several efficient genes in a single potato genotype or to produce a mosaic of Rpi genes in a potato stand combining several cultivars. When bred from the multiparental hybrids, the advanced lines with the stacks of broad-specificity Rpi genes will become prospective breeding donors immediately at hand when new pathogen strains arrive with Avr genes virulent to existing potato cultivars [1,13,14]. These breeding strategies usually aim at supporting and expanding the genetic diversity in potato stands.
Developing such sources of resistance to combat future pathotypes is called pre-emptive, or anticipatory breeding [56,57]. In the case of P. infestans, with its extremely plastic genome [58] and rapid changes in the repertoire of Avr genes [1,59], the advanced lines bred from multiparental hybrids would help withstand LB outbreaks caused by rapid pathogen evolution and invasion of new pathotypes.
By their productivity (0.89-1.25 kg of tubers per plant), most tested hybrids were comparable to cv. Sarpo Mira, the international standard of LB resistance, and considerably overtook the susceptible standard cv. Bintje. However, within the selection of highly resistant genotypes with 4+ markers of Rpi genes per plant (Table 5), it is difficult to relate tuber yield immediately to plant resistance and the number of resistance genes.
In many aspects, the success of pyramiding Rpi genes depends on the breeder's appraisal of the agricultural ecosystem as a whole [60] and the knowledge of potato Rpi genes and Avr genes of P. infestans in the particular potato stands. In the latter case, rapid and efficient assessment of Rpi and Avr gene profiles with dRenSeq and PenSeq technologies [25,59] seems most hopeful as regards the prediction of crop losses and evaluation of breeders' efforts.