Genetic mapping identifies loci that influence tomato resistance against Colorado potato beetles

The Colorado potato beetle (CPB; Leptinotarsa decemlineata Say), the most economically important insect pest on potato (Solanum tuberosum L.), also feeds on other Solanaceae, including cultivated tomato (Solanum lycopersicum L.). We used tomato genetic mapping populations to investigate natural variation in CPB resistance. CPB bioassays with 74 tomato lines carrying introgressions of Solanum pennellii in S. lycopersicum cv. M82 identified introgressions from S. pennellii on chromosomes 1 and 6 conferring CPB susceptibility, whereas introgressions on chromosomes 1, 8 and 10 conferred higher resistance. Mapping of CPB resistance using 113 recombinant inbred lines derived from a cross between S. lycopersicum cv UC-204B and Solanum galapagense identified significant quantitative trait loci on chromosomes 6 and 8. In each case, the S. galapagense alleles were associated with lower leaf damage and reduced larval growth. Results of both genetic mapping approaches converged on the same region of chromosome 6, which may have important functions in tomato defense against CPB herbivory. Although genetic mapping identified quantitative trait loci encompassing known genes for tomato acyl sugar and glycoalkaloid biosynthesis, experiments with acyl sugar near-isogenic lines and transgenic GAME9 glycoalkaloid-deficient and overproducing lines showed no significant effect of these otherwise insect-defensive metabolites on CPB performance.

Tomato recombinant inbred lines (RILs), which are generated by repeated selfing of F2 progeny derived from a cross between two tomato species also have been used to identify QTLs. RILs derived from a cross of S. lycopersicum with Solanum galapagense S.C. Darwin & Peralta were used to identify QTLs related to fruit quality (fruit weight and soluble sugars) and morphological traits 22,23 . RILs originating from a cross of S. lycopersicum × Solanum cheesmaniae (L.Riley) Fosberg, which is very close genetically to S. galapagense 24 , identified QTLs related to salt stress tolerance 25 and rootstock effects that help to elucidate the physiological mechanisms behind salt tolerance 26,27 . Experiments with RILs derived from S. lycopersicum × Solanum pimpinellifolium L. identified acylsugar content and trichomes as resistance traits against two-spotted spider mites (Tetranychus urticae C.L. Koch) 28 .
The ability of CPB to grow on tomato plants can be used to study natural variation in resistance against this pest, with the ultimate goal of increasing pest resistance in the field. In particular, genetic mapping populations facilitate the identification of tomato loci that provide enhanced resistance to CPB feeding. In this study, we analyzed tomato resistance variation against CPB in S. lycopersicum cv. M82 × S. pennellii LA0716 introgression lines, as well as in RILs derived from S. lycopersicum cultivar UC-204B × S. galapagense LA0483 22,29 , with the objective being the genetic mapping of QTLs that provide CPB resistance.

Results
Colorado potato beetle performance on S. pennellii and S. lycopersicum parental lines. CPB larval mass was lower on S. pennellii plants than on S. lycopersicum M82 plants at both 5 and 10 days after the start of infestation (Fig. 1a). Differences in survivorship were found only after ten days of feeding, when CPB survival was lower on S. pennellii (Fig. 1b). Among the surviving insects, adult mass (p = 0.08, t-test) and adult emergence rate (p = 0.06, t-test) also tended to be lower in S. pennellii plants, but these effects were not significant at the p < 0.05 level.
Colorado potato beetle feeding on S. pennellii introgression lines. As differences in CPB performance on S. pennellii introgression lines were greater after 10 days of plant growth than on younger seedlings, this growth stage was used to investigate the genetic basis of tomato defense against CPB. After 10 days of feeding, larval mass ranged from 5 mg to 142 mg (Fig. 2a), whereas survival ranged from 8% to 68% in the different S. pennellii introgression lines (Fig. 2b). A positive correlation between larval mass and survival (Kendall's correlation rank, τ = 0.3, p < 0.01) was found. Lines where CPB had significantly lower larval mass or survival compared to M82 (Fig. 2, red bars), as well as lines where larval mass or survival was greater than on M82 in lines (Fig. 2, green bars) were identified. Introgressions causing both significantly lower larval mass and survival were found Colorado potato beetle feeding on S. lycopersicum × S. galapagense recombinant inbred lines. After 5 days of feeding on S. lycopersicum × S. galapagense RIL leaflets, CPB larval mass ranged from 0.4 to 3.7 mg (Fig. 3a), whereas survival ranged from 0 to 100% (Fig. 3b). There was a positive correlation between larval weight and survival (Kendall´s correlation rank, τ = 0.3, p < 0.001). Damage in the leaflets varied as show in Fig. 3c and was positively correlated with larval weight (τ = 0.6, p < 0.001) and survival (τ = 0.3, p < 0.001). There is evidence for transgressive segregation in the population, as CPB larvae performed significantly better on the most sensitive RILs than on either of the parents.
Single nucleotide polymorphism (SNP) marker data for the S. lycopersicum × S. galapagense RIL population were obtained from RNAseq data (Supplementary Table 1, positions refer to tomato reference genome build 2.50). Composite interval mapping showed two QTLs with non-additive effects for leaf damage level (Fig. 4a, red line) on chromosomes 6 and 8 that explain 21% and 11% of the total variance for this trait, respectively. The chromosome 6 QTL region overlaps with S. pennellii IL 6-2 ( Fig. 4b), which also affects CPB resistance. RILs with at least one S. galapagense (S.g.) allele had approximately 0.5 units less damage than RILs with both S. lycopersicum (S.l.) alleles (Fig. 4c). A QTL explaining 13% of the variance in larval mass (Fig. 4a, black line) was also found  (Fig. 3). These results are also consistent with the predicted effects of the RIL genotypes for the QTL on chromosomes 6 and 8.
The leaf damage QTL on chromosome 6 is in a 1.73 Mb region with 237 annotated genes (Supplementary Table 2). On chromosome 8, the leaf damage QTL is located in a 640 kb region with 76 annotated genes (Supplementary Table 3) This location on chromosome 8 is distinct from the resistance locus identified with S. pennellii introgression lines (IL 8-1-3; Fig. 2c). The larval mass QTL is in a 1.12 Mb region with 150 annotated genes (Supplementary Table 4), which overlaps with S. pennellii IL 6-2 BIN 6-D. No significant QTL was found for CPB survival.
Analysis of the known genes in the three QTL mapping regions shows that there are several with potential roles in plant defense. These include MYB transcription factors, Kunitz trypsin inhibitors, and genes involved in ethylene responses. Other genes in the three genetic mapping intervals were differentially expressed in prior experiments to investigate tomato defense responses [30][31][32][33][34] , including genes that had altered expression after T. urticae infestation 30 , flagellin treatment 31 , or effector triggered immunity 32 .

Effect of acylsugars on Colorado potato beetles.
Acylglucoses of S. pennellii are both deterrent and had a negative effect on survival and development of two generalist lepidopteran herbivores, corn earworm (Helicoverpa zea Boddie) and beet armyworm (Spodoptera exigua Hübner) 35 . To determine whether acylsugars also function in tomato defense against CPB, larval mass and survival, adult emergence rate, and adult mass, were assessed on five backcrossed introgression lines with variable acylsugar content 36,37 . CPB larval mass was not significantly affected by ten days of feeding in the acylsugar lines (F = 0.69 p = 0.6) (Fig. 5a). However, there was an effect on larval survival (F = 3.7, p = 0.006). Larval survival was lower in FA7/AS (27% lower), FA8/AS (24% lower) and FA2/FA7/AS (36% lower) lines compared to line FA5/AS (Fig. 5b). At the adult stage, there was no effect of acylsugars on adult mass (F = 0.69, p = 0.6). Adult emergence rate was affected by the genotype (F = 2.1, p = 0.03), but adult emergence was not significantly different on any of the acylsugar lines when compared to that of FA5/AS ( Table 1).
Effect of α-tomatine on Colorado potato beetles. Recently Cardenas et al. showed that GAME9, which encodes an AP2/ERF-type transcription factor regulating the biosynthesis of glycoalkaloids in tomato and potato, is located on chromosome 1 between TG21 and TG59 markers and forms part of a cluster of ERF-genes 20 . The location of this cluster overlaps with S. pennellii IL 1-1(accession 4028) and IL 1-2 (accession 4021) which were susceptible to CPB in our bioassays (Fig. 2).To determine whether tomatine has a detrimental effect of on CPB larvae, experiments were conducted using tomato plants (S. lycopersicum var. Micro Tom) with the GLYCOALKALOID METABOLISM 9 (GAME9) gene overexpressed or silenced. GAME9 is an APETALA/ ethylene response transcription factor located on chromosome 1 in tomato and has been shown to be directly related with the regulation of steroidal alkaloid biosynthesis. Tomato lines overexpressing this transcription factor (GAME9-OX) have 2.5-fold increase, whereas silenced lines (GAME9-RNAi) have 20-fold reduction in α-tomatine compared to wild type controls 20 . However, there was no significant difference on CPB larval mass ( Fig. 6a; F = 0.91, p = 0.43) or survival among these lines ( Fig. 6b; F = 0.48, p = 0.63).

Discussion
In this study we investigated natural variation in tomato resistance against CPB using two genetic tools developed for tomato, S. lycopersicum × S. pennellii introgression lines and S. lycopersicum × S. galapagense RILs. We identified regions in the tomato genome conferring either resistance or enhanced susceptibility to CPB. Overall, our results indicate that, although tomato acylsugar amount and/or composition may have some impact on CPB, they are almost certainly not the most important contributors to CPB resistance in these tomato populations. The tested tomato acylsugar breeding lines with high acylsugar content (FA2/FA7/AS, FA2/AS and FA7/ AS, Fig. 5) have a negative effect on larval survival, but not on larval mass or adult beetle emergence. These Some of the S. lycopersicum × S. pennellii introgression lines with significant impacts on CPB have previously-identified changes in acylsugar production. For instance, the CPB-resistant introgression line 1-4 has been shown to have an altered acylsucrose profile due to the activity of a specific acyltransferase 15,18 . However, the acylsugar levels in all of the tested introgression lines (Fig. 2) are less than 1% of those found in S. pennellii and are lower than those in the tested acylsugar breeding lines (Fig. 5), including that of the low acylsugar control FA5/AS. GAME9, a transcription factor that is essential for glycoalkaloid biosynthesis in tomato and potato 20 , was localized in the same region as S. pennellii introgression 1-1, which is more susceptible to CPB. Although GAME9 overexpressing plants have increased tomatine levels and silenced plants lack this metabolite, we found no significant effect on CPB larval mass or survival (Fig. 6). The effect of glycoalkaloids on CPB has been studied before with very variable results, ranging from antifeedant and negative effects on development 39,40 , to no effect at all 41,42 . The effect of glycoalkaloids also may be modified by other metabolites in the plant 43 . GAME9 overexpression and silencing affects other tomato metabolites, including cholesterol and β-amyrin 20 . Thus, the effect on CPB may be due to the interaction of several factors, rather than solely the high or very low levels of tomatine in the GAME9 mutant lines.
Introgression lines 6-2 and 6-2-2 were the most susceptible to CPB. Moreover, during the experiments these lines showed a weak, stressed phenotype in growth chambers, as is also described by the Tomato Genetic Resource Center (http://tgrc.ucdavis.edu/). A recent study on the expression quantitative trait loci (eQTL, defined as a chromosomal region that drives variation in gene expression patterns) showed that tomato chromosomes 4, 6 and 8 have an abundance of trans eQTLs 44 . Thus, regions on these chromosomes may control the expression of several transcripts. ILs 6-2 and 6-2-2 were identified as regions modifying the expression of genes related to defense and IL8-1-3, one of the introgression lines we found as resistant, of genes related to leaf development 44 . This altered gene expression in IL 6-2 and 6-2-2, which increases pathogen defense 45 , may also affect the defense response to insects such as CPB.
Some of the genes located within the three QTL mapping intervals that we identified using S. lycopersicum × S. galapagense recombinant inbred lines were previously reported to be involved with different types of stress responses (Supplementary Table 5). These defense-related genes can serve as candidates for further investigation of the genetic basis of natural variation in tomato resistance to CPB. Even without the identification of specific  causative loci, the chromosome 6 QTL region that influenced CPB resistance in both of our genetic mapping approaches has potential utility for breeding cultivated tomato varieties with enhanced herbivore resistance. Together, our results show that tomato resistance to CPB is a complex trait, but that there are individual QTL with strong effects. Known loci related to acylsugar production and tomatine accumulation likely do not account for the observed variation in CPB resistance. Additionally, synergy between defense compounds may be more important than any individual compound. Further research will be required to identify the actual genetic basis and molecular mechanisms that underlie the identified CPB resistance QTL.

Methods
Plant material and growth conditions. Seeds of S. lycopersicum (M82), S. pennellii LA0716, and S. pennellii introgression lines were obtained from the Tomato Genetics Resource Center (University of California, Davis). Seeds of S. lycopersicum cv. UC-204B × S. galapagense LA0483 RILs in the F7 generation were provided by Ilan Paran (The Volcani Center, Israel). The acylsugar producing tomato benchmark line CU071026 and backcrossed inbred lines with altered acylsugar chemotypes (FA2/FA7/AS, FA7/AS, FA2/AS, FA8/AS, and FA5/AS) were characterized and described previously 36,37 . Seeds of tomato lines with overexpression or silencing of the GLYCOALKALID METABOLISM 9 (GAME9) gene, as well as seeds from the wild type background (S. lycopersicum var. Micro Tom 20 ) were provided by Pablo Cardenas and Asaph Aharoni (Weizmann Institute of Science, Israel).
Tomato plants were grown in a growth chamber with a 16:8 h light:dark photoperiod, 180 mmol photons/ m 2 /s light intensity at a constant temperature of 23 °C and 60% humidity. Solanum lycopersicum × S. galapagense RILs were grown in a greenhouse with a 16:8 h light:dark photoperiod and a temperature range from 28 °C day to 20 °C night.
Colorado potato beetle growth conditions and bioassays. Ten clutches of CPB eggs were obtained from Jennifer Thaler (Cornell University), from a colony that was originally collected from potato in Tompkins County, New York and which had been maintained on potato in the laboratory for one year (about 10 generations). These eggs were used to establish a colony that was maintained on potato (Solanum tuberosum var. Russet) plants in a growth chamber with a 16:8 h light:dark photoperiod and a constant temperature of 23 °C.
Herbivory assays involving parental lines, S. lycopersicum cv. M82 and S. pennellii, were conducted using young plants with eight fully-expanded leaves. Eight CPB larvae, newly hatched and randomly selected from three egg clutches that had matured at the same time, were placed in each plant (M82 n = 7; S. pennellii n = 9) and caged in microperforated polypropylene bags (27.9 cm × 50.8 cm; PJP Marketplace). Five and ten days after infestation, larvae were counted and weighed. Adult emergence rate after pupation also was recorded. Herbivory assays involving S. pennellii introgression lines were conducted with plants at the reproductive stage (first floral bud visible). Assays were performed as described above. Ten days after infestation, surviving larvae were counted and weighed. Two or three replicate plants were used for each introgression line. For herbivory assays using S. lycopersicum × S. galapagense RILs, 119 individual RILs were sampled. A leaflet of fully expanded leaves from fruiting-stage plants were placed on petri dishes with seven newly hatched CPB larvae. After 5 days of feeding, larval survival and mass were recorded. Damage in the leaflets was also recorded on a categorical scale from 0 to 5, where 0 signified no visible damage and 5 signified at least 50% of the leaflet damaged. The experiment was repeated two additional times with four of the most sensitive RILs and two of the most resistant RILs.
Herbivory assays involving tomato lines with altered acylsugar content were performed as described above for S. pennellii introgression lines. Line CU071026 is a benchmark acylsugar breeding line, which contains five S. pennellii introgressions, and produces acylsugars at levels at approximately 15% of S. pennellii levels 46 . The other acylsugar lines were bred from CU071026. Each of those lines has all five of the S. pennellii introgressions of CU071026 and 1 or 2 additional S. pennellii introgressions that carry acylsugar QTL 36 . The resulting lines differ from CU071026 for acylsugar accumulation and/or composition 37 . Relative to CU071026, the acylsugar content of FA2/FA7/AS is 171%, FA7/AS and FA2/AS is 121%, FA8/AS is 80%, and FA5/AS is only 16%. Due to its extremely low acylsugar content, line FA5/AS was used as a low/negative acylsugar control. Ten plants were used for each line, with the exception of FA2/FA7/AS (n = 6) and FA2/AS (n = 8), for which the germination rate was lower. Ten days after infestation, surviving larvae were counted and weighed. Adult emergence rate, adult weight, and days to reach the adult stage also were recorded.
Herbivory assays with GAME9 overexpressing and RNAi-silenced plants were performed as described for S. pennellii introgression lines. Four to five plants of each line (wild type, GAME9-overexpressing and GAME9-RNAi) were used. Larval weight and survival were recorded five days after the start of infestation.
Quantitative trait locus mapping. SNP marker data for 113 S. lycopersicum × S. galapagense RILs were generated from fruit pericarp RNA sequencing data. Strand-specific RNA sequencing was performed as described previously 47 . Reads of each line were aligned to the tomato genome (v. 2.50) using STAR (v. 020201) 48 with "2-pass Mapping" method. GATK (v. 3.4-46) 49 pipeline was used to call SNPs for each sample, using "HaplotypeCaller" method with parameters: "-GQB 5 -GQB 20 -GQB 60 -GQB 99 -stand_call_conf 20.0 -stand_emit_conf 20.0". Only sites presenting homozygous differences between the two parental lines were kept for further analysis. A total of 17,567 SNPs were identified and 2,216 genomic bins were calculated and used as genetic markers for QTL analysis.
QTL analysis was performed with WinQTL Cartographer version 2.5 50 . The experimental LOD threshold was determined by permutation tests with 1000 permutations at 0.05 confidence level. Analysis was done by composite interval mapping with the following settings: map function = Kosambi, CIM program module = Model 6: standard model, walking speed, 1 CM, control marker numbers = 5, window size = 10 CM, regression method = backward regression method. Chromosomal confidence intervals to identifying candidate genes for mapped QTL were defined as two LOD unites from the peaks of the respective QTL. Statistical analysis. Statistical comparisons were conducted using R software 51 . Correlation tests were conducted using the Kendall's rank correlation tau (τ) for non-parametric data. Dunnett's test were conducted using the Multcomp package 52