Introduction

Gummy stem blight (GSB) is a severely destructive foliar disease of cucurbits caused by the ascomycete fungus Didymella bryoniae (Auersw.) Rehm and its anamorph Phoma cucurbitacearum (Fr.:Fr.) Sacc. Agriculturally significant losses in melon, the focus of this study, occur regularly in the southern United States (Sherf and MacNab 1986) and in several other regions worldwide including Asia and Europe. In melon, symptoms of this disease include spreading necrotic lesions on leaves and water-soaking of both leaves and hypocotyls resulting in the formation of stem cankers in cortical tissue that produce a characteristic brown gummy exudate. In a susceptible interaction, these lesions continue to expand, eventually girdling the stem leading to wilting and death of the plant (Sitterly and Keinath 1996). The ineffectiveness of chemical control especially during periods of high rainfall and humidity (Norton and Cosper 1989), and the widespread resistance of D. bryoniae to commonly used systemic benzimidazole fungicides (Keinath and Zitter 1998) have generated strong interest in the possibility of breeding melon varieties resistant to GSB.

Because of the agricultural importance of this disease, early searches for genetic resistance were undertaken. Although several sources of resistance to GSB have been reported in the ensuing decades (Sowell Jr. et al. 1966; Sowell Jr. 1981; McGrath et al. 1993), GSB-resistant melon varieties and breeding lines released to date, (e.g., Norton 1971, 1972; Norton et al. 1985; Norton and Cosper 1989), all derive resistance from PI 140471 (a wild melon originating in Texas) and have failed to provide adequate levels of resistance (Sowell Jr. 1981; Sitterly and Keinath 1996; Zhang et al. 1997). While resistance is clearly apparent in this accession which is a dudaim type, the level of resistance derived from this source appears to be diminished when introgressed into large-fruited commercially acceptable genetic backgrounds (Zhang et al. 1997). In an effort to find new sources of GSB resistance, 798 C. melo plant introductions were screened in greenhouse and field tests for high levels of resistance to GSB, emphasizing accessions from the humid tropics and subtropics along with accessions previously known to have resistance to other foliar diseases (Zhang et al. 1997). This study identified several accessions with resistance equal to or greater than PI 140471. From that effort, four additional wild melon accessions (PI 157082 from China, PI 511890 from Mexico, PI 482398 and PI 482399 from Zimbabwe) were selected for further characterization. GSB resistance in both PI 157082 and PI 511890 was subsequently shown to be monogenic and dominant, and furthermore, the resistance in PI 157082 was shown to segregate independently from resistance attributed to the locus Mc for GSB resistance in PI 140471 (Zuniga et al. 1999). A preliminary inheritance study of resistance derived from PI 482399 indicated that this resistance was recessive (Zuniga et al. 1997).

The present study was undertaken to provide a comprehensive understanding of the resources in C. melo available for resistance to this disease and to determine which breeding strategies will be most likely to yield success. Despite the commercial importance of the Cucurbitaceae, this family has lagged behind some other families such as the Solanaceae and Cruciferae with respect to investments in genomic resources. Basic knowledge of the genetics of important traits in this family, particularly agriculturally important traits such as GSB resistance that are found in a number of cucurbit crop species, will be essential in order to extend the impact of genomic analysis in melon to melon-breeding programs. The first objective of this study was to determine the mode of inheritance of GSB resistance found in PI 482398 and PI 482399. The second objective was to determine the genetic relationship among resistance sources PI 140471, PI 157082, PI 511890, PI 482398, and PI 482399.

Materials and methods

Germplasm and population development

All plant introduction (PI) accessions were obtained from the USDA NPGS Ames, Iowa. The GSB-susceptible genotype used as a parent in this study was ‘Cornell ZPPM 339’ (ZM) (Cornell University, Ithaca, N.Y.), a cantaloupe breeding line with resistance to zucchini yellow mosaic virus, papaya ringspot virus and powdery mildew. The GSB susceptible check for disease screens was ‘Honeydew Greenflesh’ (HD) (Asgrow Seed Co., San Juan Bautista, Calif.). Controlled pollinations were carried out in the field and greenhouse as described previously (Zuniga et al. 1999). For inheritance studies, ZM was crossed with PI 482398 and PI 482399 to generate reciprocal F1, F2 and backcross populations, as described previously (Zuniga et al. 1999). To determine whether resistance genes segregated independently, the five resistant PI accessions, PI 140471 (471), PI 157082 (082), PI 511890 (890), PI 482398 (398) and PI 482399 (399), were intercrossed to generate F1, F2 and backcross populations.

Disease screening

Resistance evaluations were made in the greenhouse according to methods described in Zhang et al. (1997) using the highly virulent D. bryoniae isolate NY1 (Keinath et al. 1995). Two susceptible control plants and two resistant control plants (471) were planted in each seedling flat to monitor inoculation efficiency and to test severity of infection. Spore suspensions were applied to stem and leaf surfaces of 3- to 4-week-old seedlings at 5×105 spores/ml using a backpack sprayer, followed by a 72-h incubation period at 25°C in a mist chamber to promote uniform development of disease symptoms. Symptom severity was measured 21 days post-inoculation using a scale from 1 to 5 based on stem damage ratings (Zuniga et al. 1999) where 1=no damage; 2=a single lesion 1–10 mm long or coalesced lesions 1–20 mm long with no girdling of the stem; 3=lesions 21–80 mm and/or girdling of the stem; 4=withered stem; 5=dead seedling. Our previous studies have indicated that classification of segregating progenies is most reliably based on stem damage ratings where resistant individuals typically score 1 or 2 and susceptible individuals usually score 4 or 5 (Zhang et al. 1997; Zuniga et al. 1999). The number of individuals falling into resistant (stem damage ratings 1–2) and susceptible (stem damage ratings 3–5) categories was determined in each segregating population, and the resulting ratios were tested for goodness-of-fit using χ2 analysis.

Results

Inheritance of resistance to GSB derived from PI 482398 (398)

Parental lines and (ZM×398) F1, F2 and backcross populations were scored for response to inoculation with D. bryoniae. Results are reported in Table 1. The susceptible parent (ZM) was uniformly susceptible with an average disease rating of 4.4. In contrast, the average disease rating for 398 was 1.19; one 398 plant out of 26 was rated a 3, possibly due to high inoculum pressure from an adjacent susceptible check plant ‘Honeydew Greenflesh’ (HD). HD typically supported severe infections, leading to death of the plant during GSB screening, and local increases in test severity were occasionally noted. Because of extreme difficulty making this cross, only a very small number of F1 seed were available, but all F1 plants were uniformly highly resistant to infection from GSB, consistent with dominant inheritance (Table 1). The F2 population showed segregation consistent with a 3 resistant:1 susceptible ratio indicating monogenic dominant inheritance. The backcross population to the resistant parent 398 was completely resistant while the backcross population to the susceptible parent ZM showed segregation consistent with a 1 GSB resistant:1 GSB susceptible ratio. Taken together, these data indicate that resistance to GSB in PI 482398 is conferred by a single dominant gene.

Table 1 Response of resistant (PI 482398) and susceptible (Cornell ZPPM 339) Cucumis melo genotypes and derived populations to inoculation with Didymella bryoniae. Symptom severity was measured 21 days post-inoculation using a 1–5 scale based on stem damage ratings where 1= no damage; 2= a single lesion 1–10 mm long or coalesced lesions 1–20 mm long with no girdling of the stem; 3= lesions 21–80 mm and/or girdling of the stem; 4= withered stem; 5= dead seedling
Full size table

Inheritance of resistance to GSB derived from PI 482399 (399)

The reactions of parental lines, (ZM×399) F1, F2 and backcross populations to GSB inoculation are indicated in Table 2. The average disease rating for 26 399 plants was 1.5. As observed with the accession 398, one 399 plant also located adjacent to a susceptible check plant was rated 3. In contrast to results obtained for 398, the F1 population derived from the cross between 399 and ZM was uniformly susceptible to infection from GSB, consistent with recessive inheritance (Table 2). Segregation observed in the F2 population was consistent with a 1 resistant:3 susceptible ratio, suggesting monogenic recessive inheritance. The backcross population to the susceptible parent ZM was uniformly susceptible, and the backcross population from the resistant parent 399 segregated in a manner consistent with a 1 resistant:1 susceptible ratio. Together, these data demonstrated that GSB resistance from PI 482399 is controlled by a single recessive gene.

Table 2 Response of resistant (PI 482399) and susceptible (Cornell ZPPM 339) C. melo genotypes and derived populations to inoculation with D. bryoniae. Symptom severity was measured 21 days post-inoculation using a 1–5 scale based on stem damage ratings where 1= no damage; 2= a single lesion 1–10 mm long or coalesced lesions 1–20 mm long with no girdling of the stem; 3= lesions 2–80 mm and/or girdling of the stem; 4= withered stem; 5= dead seedling
Full size table

Gsb-1 (formerly Mc) is distinct from other loci controlling monogenic dominant GSB resistance

The only gene designated to date for resistance to GSB in melon is Mc, a locus symbol that reflects the pathogen synonym, Mycosphaerella citrullina (Prasad and Norton 1967). Based upon results reported above and the fact that Mc is not consistent with current rules governing gene nomenclature in melon, we submitted a proposal to the Cucumis Gene Nomenclature Committee for revision of Mc derived from PI 140471 to Gsb-1. This proposal has been accepted (J. McCreight, personal communication).

A previous study established that Gsb-1 segregated independently from monogenic dominant resistance in PI 157082 (Zuniga et al. 1999). Based upon our approved proposal for nomenclature revision and this previously published result, we propose the designation Gsb-2 for the locus that controls monogenic dominant resistance to GSB derived from PI 157082.

The same study also reported monogenic dominant resistance derived from PI 511890 (890) but again, no locus designation was given. In order to determine whether this resistance was controlled by a locus distinct from Gsb-1, 471 was crossed with 890 to generate F1 and F2 populations. If Gsb-1 and the dominant GSB resistance gene in 890 occur at distinct loci, the (471×890) F1 population should be uniformly resistant and the F2 population should show segregation consistent with a 15 resistant :1 susceptible ratio. The data obtained from this experiment summarized in Table 3 confirm this hypothesis. Based upon this result, we propose the locus designation Gsb-3 for the resistance allele derived from 890.

Table 3 Gsb-1 from C. melo PI 140471 is genetically distinct from monogenic dominant resistance to gummy stem blight in C. melo PIs 157082, 511890, and 492398. Response of resistant (R) parents, (R×R) F1, and (R×R) F2 populations to inoculation with D. bryoniae. Symptom severity was measured 21 days post-inoculation using a 1–5 scale based on stem damage ratings where 1= no damage; 2= a single lesion 1–10 mm long or coalesced lesions 1–20 mm long with no girdling of the stem; 3= lesions 21–80 mm and/or girdling of the stem; 4= withered stem; 5= dead seedling
Full size table

If Gsb-1 and the newly identified monogenic dominant resistance in 398 occur at independent, unlinked loci, the same pattern of segregation should also be observed from F1 and F2 populations derived from the cross (471×398). Results shown in Table 3 are consistent with independence of these two loci. Based upon this result and the segregation data presented in Table 1, we propose the locus designation of Gsb-4 for monogenic dominant resistance to GSB derived from 398. Taken together, these data confirm that Gsb-1 is distinct from the loci we have designated Gsb-2, Gsb-3, and Gsb-4.

Gsb-2 segregated independently from Gsb-3 and Gsb-4

To confirm directly that Gsb-2 is independent of Gsb-3 and Gsb-4, 082 was crossed with 890 and 398 as above to generate F1 and F2 populations. The resulting populations were screened with GSB and results are reported in Table 4. In both crosses, susceptible segregants were recovered in the F2 populations, confirming the designation of these loci as distinct. In the case of the (082×398) F2, more susceptible individuals were recovered than expected according to a 15 resistant:1 susceptible ratio. This result still favors the hypothesis of independent loci, rather than allelism or linkage, and may be due to environmental effects that increased phenotypic screen severity. These data demonstrate that Gsb-2 is distinct from Gsb-3 and Gsb-4.

Table 4 Gsb-2 from C. melo PI 157082 is genetically distinct from monogenic dominant resistance to gummy stem blight in C. melo PI 511890 and PI 492398. Response of resistant (R) parents, (R×xR) F1 and (R×xR) F2 populations to inoculation with D. bryoniae. Symptom severity was measured 21 days post-inoculation using a 1–5 scale based on stem damage ratings where 1= no damage; 2= a single lesion 1–10 mm long or coalesced lesions 1–20 mm long with no girdling of the stem; 3= lesions 21–80 mm and/or girdling of the stem; 4= withered stem; 5= dead seedling
Full size table

Gsb-3 segregated independently from Gsb-4

To determine directly whether Gsb-3 segregated independently from Gsb-4, 890 was crossed with 398 as above to generate F1 and F2 populations and screened for response to the pathogen. Results are reported in Table 5. Again, susceptible segregants were recovered in the F2 population, confirming the designation of these loci as distinct. Taken together, this series of experiments supports the designation of four unlinked loci in melon, Gsb-1-4, each one of which has an allele which conferred monogenic dominant resistance to GSB.

Table 5 Gsb-3 from C. melo PI 511890 is genetically distinct from monogenic dominant resistance in C. melo PI 482398. Response of resistant (R) parents, (RxR) F1 and (RxR) F2 populations to inoculation with D. bryoniae
Full size table

Gsb-2 segregates independently from gsb-5

In addition to testing the allelic relationships between the monogenic dominant sources of resistance in this study, we also wanted to confirm the recessive nature of resistance from PI 482399 (399), designated gsb-5, and to evaluate its independence from dominant sources of resistance. Independent segregation of one monogenic recessive and one monogenic dominant gene in an F2 population results in an expected ratio of 13 resistant individuals:3 susceptible individuals. When 195 F2 individuals from the cross of (399×082) were screened for resistance to GSB, 160 individuals were resistant and 35 were susceptible, consistent with the expected 13:3 ratio (P=0.77) (data not shown). Similar tests are underway to confirm independence from the other dominant sources of resistance.

Discussion

This study defines four genetically distinct loci, each of which has an allele that confers monogenic dominant resistance to GSB, and a fifth locus that controls monogenic recessive resistance. These loci are designated Gsb-1 (formerly Mc) for monogenic dominant resistance from PI 140471, Gsb-2 for monogenic dominant resistance derived from PI 157082, Gsb-3 for monogenic dominant resistance derived from PI 511890, Gsb-4 for monogenic dominant resistance derived from PI 482398, and gsb-5 for monogenic recessive resistance derived from PI 482399.

The sources of GSB resistance described in our study have been determined in relationship to the highly virulent NY1 isolate of D. bryoniae, which was identified and isolated from infected melon tissue in New York (Keinath et al. 1995). While our studies were conducted only with this isolate, a range of D. bryoniae isolates collected from watermelon and melon from South Carolina to New York that included NY1 were evaluated on both watermelon and cantaloupe and showed no host-by-isolate interaction (Keinath et al. 1995). No studies to date have investigated the possibility of race-specific GSB resistance in melon, although a study using six American and two European D. bryoniae isolates found that GSB resistance in cucumber was not race-specific (St. Amand and Wehner 1995). This result may or may not be relevant to melon because the GSB resistance sources in cucumber show polygenic inheritance (St. Amand and Wehner 2001). Since our original publication, the accessions used in this study have been included in trials under heavy disease pressure around the world with no evidence of strain-specificity, suggesting that the resistance attributed to these loci may be stable across broadly different environments. Clearly, our results should be confirmed by controlled studies using well-defined fungal isolates typical of target environments before these genes are deployed in resistant varieties for use in those production environments.

The redundancy for resistance to GSB revealed by this study in melon has not been reported for other cucurbits infected by this pathogen; however, multiple dominant race-specific resistance genes exist in melon for two other agriculturally important fungal diseases, powdery mildew and Fusarium wilt. Six dominant genes (Pm-1, Pm-2, Pm-3, Pm-4, Pm-5, Pm-6) are known for race-specific resistance to powdery mildew in melon (McCreight et al. 1993), while three unlinked dominant resistance genes in melon (Fom-1, Fom-2, and Fom-3) have been reported for race-specific resistance to Fusarium wilt (Risser et al. 1976; Zink and Gubler 1985, 1986). Although not yet isolated at the molecular level, both Fom-1 and Fom-2 are tightly linked to sequences bearing NBS-LRR motifs (Brotman et al. 2002; Wang et al. 2002).

A hallmark of many dominant resistance genes, including those of the NBS-LRR type, is the hypersensitive response (HR). GSB resistance sources described in this study do not exhibit a detectable HR when challenged with D. bryoniae under the conditions employed in this study nor has this response been observed in any of the environments in which materials carrying these genes have been assessed. During the severe controlled environment screens described in this study, inoculated seedling leaves generally became infected in both susceptible and resistant individuals; however, the extent of infection and its consequences varies clearly. The resistance mechanism appears to operate mainly at the junction of leaves and stem. In susceptible plants, the infection quickly spread to the stem, causing lesions and girdling, while in resistant plants foliar lesions expanded less quickly and stem lesion formation appeared to be greatly reduced or inhibited.

The results from the study clearly indicated that in one case, resistance was recessive. Recessive fungal resistance is likely to represent a mechanism distinct from that controlled by dominant resistance genes. Recessive resistance in crop plants to fungal pathogens has been characterized at the molecular level only in the case of the mlo gene, which represents a defective negative regulator of constitutive defense (Buschges et al. 1997). Barley plants homozygous for mlo exhibit a lesion mimic phenotype, typical of a constitutive defense response. None of the parental, F2 or backcross individuals homozygous for gsb-5 exhibited a detectable lesion mimic phenotype. It is possible that gsb-5 may represent an important target for identification at the molecular level because the wild type allele at this locus may encode a gene product that is important for pathogenesis. If this is the case, then this gene may serve as a useful target for engineering of resistance via gene silencing approaches. Recessive GSB resistance controlled by the db gene (Norton 1979) has already been used to breed resistant watermelon cultivars (Norton et al. 1986, 1995) and recessive resistance in melon to powdery mildew has also been reported (McCreight 2001). Efforts are underway to set up mapping populations to locate gsb-5 and the other Gsb loci on the genetic map in melon, although a significant impediment to these studies is a relative lack of DNA polymorphism in the species.

It has been proposed that effects of genetic background and physiological stress may explain why Gsb-1 alone when used in melon varieties does not confer a high level of resistance (Zhang et al. 1997). All of the resistant PIs in this study have relatively small fruit with thin or nearly absent flesh. A major objective of our current breeding efforts is to combine these GSB resistance genes to recover higher resistance levels in horticulturally acceptable lines with large sweet fruit. There is an indication in our breeding program that resistance gene combinations give significantly higher levels of GSB resistance than single genes alone (J.D. Frantz and M.M. Jahn, unpublished results). An important difficulty that arises as we pyramid these genes is the discrimination among lines that carry several Gsb genes. Because resistance has not been identified as race-specific, progeny tests or test crosses are required to confirm segregation of multiple genes and even then it is essentially impossible to identify which of the four dominant genes are segregating. In addition to setting up the populations necessary to identify linked molecular markers, we are currently pursuing parallel introgression programs to produce inbred lines carrying one or two of the Gsb genes. The dominant genes can then be brought together in the form of an F1 hybrid, effectively pyramiding up to all four of the dominant genes. Hybrid breeding programs utilizing recessive gsb-5 will require introgression into both parents of the F1 with progeny testing along the way. Marker-assisted selection of these loci would greatly facilitate the creation of resistance gene pyramids by allowing the breeder to discriminate the homozygous allelic state from the heterozygous state and to confirm which resistance genes are present.