Self-incompatibility alleles in important genotypes for apple breeding in Brazil

The objective of this study was to identify self-incompatibility (S) alleles of advanced breeding selections of apple (Malus × domestica Borkh.). The Salleles of 42 apple genotypes were analyzed by markers using allele-specific PCR amplification and amplicons digested with restriction endonucleases. Among the screened genotypes were cultivars, advanced selections, and accessions of the Apple Germplasm Bank of Epagri (Caçador, Santa Catarina, Brazil). Two S-alleles were identified in 36 genotypes, and only one S-allele was determined in the other six genotypes. In all, eleven S-alleles were identified among all the genotypes evaluated. The S3 and S5 alleles were most frequent (30.2% and 18.6%, respectively). The identification of S-alleles using molecular markers in important apple tree genotypes is useful for determination of compatible parents for breeding programs.


INTRODUCTION
The development of new apple cultivars (Malus × domestica Borkh.) by classical breeding methods requires from 13 to 17 years of research (Sedov 2014). The process begins with the choice of parents that have traits of interest, which are then crossed to select new cultivars and their pollinizers (Denardi et al. 2019a). The possible parental combinations are restricted by the gametophytic self-incompatibility (GSI) system present in Malus (Pereira-Lorenzo et al. 2018). The GSI of a fertile plant is its inability to produce zygotes after self-pollination or pollination among individuals that have S-alleles in common (Muñoz-Sanz et al. 2020). The S-locus is responsible for determining self-incompatibility and is positioned on chromosome 17 of the apple genome (Maliepaard et al. 1998). Crosses between genetically compatible plants (even between species) are required to generate as many plants as possible with the greatest genetic variability (De Franceschi et al. 2016). The S-alleles of several apple cultivars and genotypes have not been genotyped, making it difficult to choose compatible parents for planned crosses.
Traditionally, the presence of S-alleles was determined indirectly by pollination and pollen tube growth tests (Bošković and Tobutt 1999), but this methodology is strongly influenced by the environment and requires replications in different growing seasons to ensure the reliability of this identification (Breen et al. 2016). The use of genetic markers to identify S-alleles, such as allele-specific TL Brancher et al. primers, provides information on their distribution among apple genotypes (Long et al. 2010, Akbari et al. 2016, Larsen et al. 2016, Kasajima et al. 2017 and allows breeders to plan crosses between compatible genotypes (Morita et al. 2009, Breen et al. 2016).
In the GSI mechanism, if the pollen has the same S-allele as the pistil, the developed pollen-tubes are recognized and rejected by a pistil-specific ribonuclease (S-RNase) encoded by the S-locus. The S-RNase is always expressed in the pistil, but when the S-allele of the pollen is not the same as either of the two S-alleles expressed in the pistil, the S-RNases are inactivated by at least two genes specifically expressed in pollen -S-locus F-box Brother genes (Sassa et al. 2007). To date, 57 S-alleles of the Malus S-locus encoding a different S-RNase have been identified (Kim et al. 2016).
The only active public apple breeding program in Brazil is at the Agricultural Research and Rural Extension Company of Santa Catarina (Epagri), located at the Experimental Stations of Caçador and São Joaquim, SC. For breeding crosses, the Epagri Apple Breeding Program uses selections and cultivars resulting from their own crosses and/or those developed in other countries as parents. It is crucial for this breeding program to choose fully compatible parents for the planned crosses. The objective of the study was to use genetic markers to identify the S-alleles of 42 apple genotypes/cultivars used as parents in the Epagri Apple Breeding Program.

MATERIAL AND METHODS
Among the apple genotypes used in the Epagri Apple Breeding Program, a total of 42 were tested (Table 1). Of these, six are cultivars developed by Epagri, 20 are advanced selections developed by the Epagri Apple Breeding Program, and 16 are accessions from the Epagri Apple Germplasm Bank. These genotypes were grown in experimental orchards and on the premises of the Apple Germplasm Bank at the Epagri Experimental Station in the municipality of Caçador in the Midwestern region of the state of Santa Catarina (lat 26° 49' 5'' S, long 50° 59' 12'' W, alt 940 m asl).
Young and healthy leaves were collected from the 42 apple genotypes and deep frozen at -20 ºC in plastic bags until DNA extraction, which was performed according to the protocol proposed by Lefort and Douglas (1999) with modifications (Revers et al. 2005), using 0.1 g of ground plant tissue. Each polymerase chain reaction (PCR) contained 1 U of Taq DNA polymerase, 1x enzyme buffer, 2.00 mM MgCl 2 , 0.2 mM dNTPs, 1 μM of each primer (forward and reverse), and 50 ng of genomic DNA, with a final volume of 15 μL. Primers for the identification of 16 S-alleles of apple trees were used: S 1 , S 2 , S 3 , S 4 , S 5 , S 6 , S 7 , S 9 , S 10 , S 16 , S 19 , S 20 , S 22 , S 23 , S 24 , and S 26 ( Table 2).
The PCRs were performed in a T100™ thermocycler (BioRad ® , California, USA) programmed for 3 min at 94 °C, followed by 30 cycles of 94 °C for 1 min, annealing at 54-62 °C (depending on the primer characteristics; see Table 2) for 1 min, and 72 °C for 1 min, followed by a final extension step (72 °C for 7 min).

TL Brancher et al.
For discrimination of the S 4 , S 16 , and S 22 alleles, part of the PCR product (10 μL) was digested by the restriction enzyme TaqI (for 1 h in a 65 °C water bath). Likewise, for identification of the S 20 allele, 10 μL of the PCR product was digested by the restriction enzyme NarI (for 4 h in a 37 °C water bath). For the S 10 modified and S 3 /S 5 /S 10 primers, PCR programming and restriction enzyme digestion are described in Table 2.
As a positive control for the presence of each S-allele, cultivars previously characterized for the respective S-allele were used (Table 3). The only exception was the S 16 allele since no genotype is maintained by Epagri with this pre-identified allele. In addition, the same cultivars were used for primer optimization.
After PCR and respective digestions with restriction enzymes, if necessary, the amplification products were analyzed by 3% agarose gel electrophoresis using the 50 bp DNA marker to help identify the size of the PCR product. The gels were stained with GelRed ® (Biotium, California, USA) and then observed and photographed with Kodak Gel Logic 212 Pro (Carestream, New York, USA), for registration and interpretation. The samples with bands that coincided with the size of the respective S-allele amplifications ( Table 2) were considered to be present.

RESULTS AND DISCUSSION
At least one S-allele was identified in each genotype characterized (Table 4). The genotypes 'Castel Gala' (Epagri Apple Breeding cultivar) and 'Galaxy' were identified as S 2 S 5 . They are sport mutations of 'Gala' for early budding and skin color, respectively (Hawerroth et al. 2018, Denardi et al. 2019a, and as originally expected, had the same genotype as the original cultivar (Matsumoto et al. 1999).
Among the genotypes tested, the genotypes 21-300-21, 21-361-75, Co-op 24, M-11/92, 'MacFree', and 'SCS416 Kinkas' manifested only one of the S-alleles identified with the primer set used in this study: S 9 S ? , S 9 S ? , S 2 S ? , S 2 S ? , S 20 S ? , and S 9 S ? , respectively. New markers are required for identification of the second S-allele of these genotypes, for example, through use of the markers developed by Larsen et al. (2016), which identify S-alleles found at a lower frequency among apple cultivars. It is noteworthy that these low-frequency S-alleles were not evaluated initially in the present study because they are not commonly found in genotypes developed in Brazil (Albuquerque Junior et al. 2011). The cultivar 'SCS416 Kinkas' (S 9 S ? ) is the result of the cross between 'Fuji' (S 1 S 9 ) and PWR37T133 (S-alleles unknown), while M-11/92 (S 2 S ? ) is a descendant of the cross between M-41 [Anna ♀ (S 3 S 29 ) × NJ-56 ♂ (S-alleles unknown)] and Gala (S 2 S 5 ). Both genotypes have unknown S-alleles in their genealogy and were able to exhibit S-alleles other than those we attempted to genotype in this study.
For the selection M-13/91, a S 3 S 5 genotype was detected, which is different from the S 5 S 10 previously reported by Albuquerque Junior et al. (2011). This confirms the pedigree of 'M-13/91' ['Mollie's Delicious' (S 3 S 7 -♀) × 'Princesa' (S 3 S 5 -♂)], whose parents do not carry the S 10 allele. Using the FTC12 and FTC228 primers, an allele size corresponding to the S 10 allele (209 bp) occurred in the selection M-13/91. However, when using the 'S 10 modified' marker recommended by Kitahara and Matsumoto (2002), the amplification of the expected region in 'M-13/91' was not confirmed. In contrast, for 'SCS417 Monalisa' (S 2 S 10 ), the amplification product generated by the 'S 10 modified' marker treated with restriction enzyme NarI generated specific fragments (185 and 97 bp), indicating the presence of the S 10 allele. Likewise, when using the 'S 3 /S 5 /S 10 ' marker for genotyping of the selection M-13/91, fragments characteristic of S 3 (423 bp and 264 bp) and S 5 (399 bp and 273 bp) alleles were amplified, but not of S 10 (382 bp). In addition, the final amplified product was 382 bp for the cultivar 'SCS417 Monalisa' when using the 'S 3 /S 5 /S 10 ' marker, a size characteristic of the S 10 allele (Larsen et al. 2016). Based on the sequences of the available S-alleles (Benson et al. 2013), alleles S 3 (GenBank code: U12200.1) and S 10 (GenBank code: AB052683.1) have a percentage of identity of 96%. The S 3 and S 5 alleles (GenBank code: U19791.1) have a sequence identity of 77%, and the S 5 and S 10 alleles, 90%. Therefore, the data suggest that the primers FTC12 and FTC228 match homologous sequences at the three alleles (S 3 , S 5, and S 10 ), impairing the use of these primers for identification of the S 10 allele, thus explaining the discrepancy in genotyping compared to results of Albuquerque Junior et al. (2011).
According to the official pedigree, the cultivar 'SCS425 Luiza' is a descendant of 'Imperatriz' (S 3 S 5 -♀) and 'Cripps Pink' (S 2 S 23 -♂) (Denardi et al. 2019b), and is a sibling of 'SCS427 Elenise' and 'M-10/09' (Table 1). The S-allele genotypes were S 3 S 23 and S 5 S 23 for the cultivar 'SCS427 Elenise' and selection M-10/09, respectively. However, 'SCS425 Luiza' exhibited the genotype S 5 S 9 , which was not expected, based on its genealogy. So, the presence of the S 9 allele in 'SCS425 Luiza' indicates that there may have been cross contamination during the development of the cultivar (pollen contamination, seed exchange between crosses at sowing, or hybrid exchange at planting) or that this cultivar is not the result of the cross 'Imperatriz' × 'Cripps Pink'. Consequently, the true pedigree of 'SCS425 Luiza' must be determined. Similar results have been reported in the literature. Sakurai et al. (2000) admitted the possibility of cross contamination during the development of the cultivar 'Kent' (S 3 S 9 ). In this case, this cultivar was theoretically a descendant from 'Cox's Orange Pippin' (S 5 S 9 -♀) × 'Jonathan' (S 7 S 9 -♂). For that reason, these authors suggested that 'Jonathan' is not the true pollen donor of the cultivar 'Kent'. In this sense, the genotyping of the S-alleles in the present study generated information that show possible errors in the genealogy previously registered by the Epagri Apple Breeding Program. Thus, S-allele Genotypes in the groups are incompatible -they have the same S-alleles. * At least semi-compatible genotypes. ** Using triploid genotypes as female parent.
genotyping by markers can be used as an auxiliary tool in the characterization of the descendants of apple crosses, taking the expected segregation of S-alleles into account.
There are 11 cultivars from the Epagri Apple Breeding Program that had already been genotyped ( The compatibility levels between the genotypes based on identification of the S-alleles are presented in Table 5. Crosses using the genotypes within groups (Table 5) are impossible because they are incompatible. In the crosses between groups, there is at least semi-compatibility between the groups and the other genotypes, except for triploid genotypes. Pollen from triploid plants is sterile because the chromosomes are unequally divided during meiosis (Sedov et al. 2017). For that reason, triploid plants can only be used as a female parent.
Eleven different S-alleles were identified in the 42 genotypes evaluated (Figure 1). The S 3 and S 5 alleles were most frequently identified (30.2% and 18.6%, respectively). One of the reasons for the higher frequency of these alleles is that 26 of the 42 genotypes tested were direct or indirect descendants from the cultivars 'Imperatriz' (S 3 S 5 ) (Albuquerque TL Brancher et al. Junior et al. 2011), 'Golden Delicious' (S 2 S 3 ) (Matsumoto et al. 1999), and/or 'Gala' (S 2 S 5 ) (Matsumoto et al. 1999). For a long time, these three genotypes served as the basis for the crosses of the Epagri Apple Breeding Program. A consequence of selection through breeding is that the bottleneck in genetic variability is indirectly reflected in the higher frequency of a few S-alleles, such as S 3 and S 5 in this situation. Larsen et al. (2016) showed a higher frequency of S 3 alleles (28%) among 432 genotypes of the genus Malus. In European apple cultivars, Dreesen et al. (2010) identified S 2 , S 3 , S 5 , and S 9 as the most common S-alleles. Meanwhile, Hegedűs (2006) reported that the S 2 , S 3 , S 5 , S 7 , S 9 , and S 10 alleles were the most frequent among the commercial apple cultivars, due to the extensive use of the genotypes 'Golden Delicious', 'Delicious', 'Jonathan', 'McIntosh', and 'Cox's Orange Pippin' in apple breeding programs around the world.
According to Halász et al. (2011) andDe Franceschi et al. (2016), the presence of the S 2 , S 3 , and S 5 alleles is associated with resistance to apple scab (Venturia inaequalis). Apparently, none of these S-alleles are linked to the gene of vertical resistance against apple scab (Rvi6), but somehow they are linked to different resistance levels, close to genes of minor effect on horizontal resistance (Halász et al. 2011). Indirect selection for these alleles can be performed by breeders, and most of the parents used for generating scab-resistant plants have at least one of these alleles (S 2 , S 3 , and S 5 ), explaining their higher frequency in elite genotypes of the Epagri Breeding Program.
The identification of the S-alleles in the genotypes evaluated allows breeders to plan crosses. Furthermore, it provides important information for other breeding programs, which can use the genotypes evaluated in this study as a genetic source in their research.