Validation of K46 , a Pup1 -linked marker, using a selection of Sri Lankan rice ( Oryza sativa L.) germplasm for marker assisted selection towards phosphorous deficiency tolerance

Phosphorous deficiency (PD) tolerance is a pivotal trait that is advantageous if present in modern day high performing rice varieties. However most of the frequently grown mega rice varieties lack this trait leading to expensive application of artificial phosphate fertilizer results secondary consequences such as environmental pollution and higher cost of production. Marker assisted breeding (MAB) for PD tolerance in rice is often hailed as the pragmatic solution to tackle this problem. The genetic basis of PD tolerance in rice has been dissected using a wide cross made between PD tolerant rice landrace Kasalath and a sensitive landrace Nipponbare . A major QTL known as Pup1 has been identified, molecularly characterized and underlying polymorphisms were detected. A major INDEL region within Pup1 QTL within Kasalath background is conferred to control the PD tolerance which is genetically ‘null’ in Nipponbare background. The DNA marker K46 which is present within Pup1 has been developed to identify this INDEL region in Kasalath like genotypes. The PD tolerance in Sri Lankan rice germplasm has been recently studied using a core panel of rice genotypes that are important in rice breeding programs. It is important to figure out the genomic landscape of the INDEL region of Pup1 QTL in these Sri Lankan rice genotypes. Therefore the present study was conducted to characterize the K46 DNA marker locus of the previously characterized for PD tolerance rice cultivars in Sri Lanka. The K46 specific primer-pair was assayed across 30 selected Sri Lankan cultivars and the PCR products were sequenced. The resulted DNA sequences were aligned with the Kasalath reference sequence for K46 locus. Further cluster analysis of the identified SNPs resulted four distinct haplotypes in which nine cultivars were grouped with Kasalath like haplotype, two unique haplotypes and the null haplotype. However, there is no strong association between the haploype class and the PD tolerance score of the cultivars implying the potentially novel PD tolerance mechanisms that require further studies.


INTRODUCTION
Since green revolution, profitable rice farming is believed to be entirely dependent on the application of artificial fertilizer. The higher and expedited responding of rice cultivars to fertilizer application was considered as a key trait (MacDonald et al., 2011). However, currently it is believed that the application of fertilizer itself cannot increase the rice production due to many reasons such as fixing nutrients by the soil systems and higher cost associated with the fertilizer inputs (Arai and Sparks, 2007). Moreover prolonged and handful applications of fertilizer could dramatically pollute the environment (Cordell et al., 2009) and also contribute to the health hazards (Chandrajith et al., 2010). Phosphorous (P) is the most problematic nutrient in rice growing soils. Phosphorous gets fixed by Fe 3+ and Al 3+ ions present in the soils and thereby becoming limitedly available to the plants (Shen et al., 2011;Holford and Mattingly, 1976;Bieleski, 1973). In fact most of the rice growing soils lack the optimum levels of P for growth and development of rice (Fairhurst et al., 1999). Because of these reasons farmers tend to apply more P fertilizer, but consequently they pollute the environment and increase the cost of production. Governments of the developing countries spend lot of money on fertilizer imports and in Sri Lanka it approximately accounts for 0.3 billion US dollars annually (Central Bank, Sri Lanka, 2014). The situation is further aggravated when government funded fertilizer subsidies are placed due to numerous socio-economic and political reasons (Cordell et al., 2009). Organic rice growers also face difficulties as the popular organic manure such 46 Aluwihare et al. as crop stubbles and straw lack sufficient levels of P (Sirisena and Wanninayake, 2014;Kumaragamage and Indraratne, 2011). Therefore, an alternative strategy to overcome the P fertilizer led crisis is required.
To solve this problem, the plant research community considers breeding of P deficiency (PD) tolerant rice varieties as the most promising solution (Cordell et al., 2009;Wissuwa and Ae, 2001a;Rose et al., 2011). The PD tolerance is found to be quantitatively inherited with a significantly higher level of heterosis ranging from -60.6 to 148.6 (Majumder et al., 1989). It was an encouraging sign for the rice geneticists and 17 years ago detailed quantitative genetic studies were begun to dissect the underlying genetics of PD tolerance in rice (Chin et al., 2011;Wissuwa et al., 1998;Doerge and Churchill, 1996). A major quantitative trait locus (QTL) controlling PD tolerance was identified on the rice chromosome 12 using recombinant inbred lines (RILs) generated from a cross between a PD tolerant landrace Kasalath and sensitive landrace Nipponbare (Wissuwa and Ae, 2001b;Wissuwa et al., 2002). Later this QTL was fine mapped and labeled as Pup1 (Wissuwa et al., 1998;Heuer et al., 2009;Chin et al., 2010;Chin et al., 2011). The Pup1 QTL in Kasalath which is tolerant to PD, is 278 kb in length and Nipponbare which is sensitive to PD is only 145 kb in length (Gamuyao et al., 2012;Heuer et al., 2009) due to an INDEL. Thus, this major INDEL in the Pup1 locus is a key determinant in conferring PD tolerance in Kasalath-like rice germplasm (Chin et al., 2010). The molecular markers have been developed to introgress the Pup1 alleles conferring PD tolerance into PD sensitive rice genotypes (Chin et al., 2010;Heuer et al., 2009;Chin et al., 2011). The DNA marker K46, present inthe major INDEL within Pup1 QTL and produces a 523 bp allele in Kasalath genetic background and a null allele (i.e. no band) in Nipponbare background (Chin et al., 2010).
However, the presence of K46 locus is not validated across a diverse rice germplasm collection to confirm it as a diagnostic marker for PD tolerance. In Sri Lanka, although MAB is still in its infancy, PD tolerance in rice and the importance of using genomic information in breeding is highlighted (Kottearachchi and Wijesekara, 2013). An important set of landraces and improved varieties (here in after collectively referred to as a panel of cultivars) were characterized for PD tolerance based on a three-tier indexing system (Aluwihare et al., 2015). However, no attempts are reported in characterizing the Pup1 locus to lay a foundation towards the MAB for PD tolerance. Therefore, the present study was conducted to characterize a set of rice cultivars previously evaluated for PD tolerance in Aluwihare et al., (2015) for the DNA sequence polymorphism in K46 marker locus to lay a foundation for MAB to produce better performing rice varieties under low P conditions.

Materials and methods
A set of 30 rice cultivars were obtained from Rice Research and Development Institute (RRDI), Bathalagoda (Bg), Sri Lanka (Appendix 1). Seeds were germinated, immature leaf samples were collected from young seedlings and DNA was extracted using the Dneasy Plant Minikit® (Qiagen, Solna, Sweden). Extracted DNA was confirmed for the successful PCR amplification using matK, a universal plant DNA barcoding primer pair (Hollingsworth et al., 2011). Then PCR amplification was carried out using K46 specificprimer pair (PF: 5'TGAGATAGCCGTCAAGATG CT and PR: 5'AAGGACCACCATTCCATAGC; Chin et al., 2010). The PCR conditions were as follows, initial denaturation at 94 °C for min, 35 cycles of 94 °C for 0.5 min, 57 °C for 1.5 min, 72 °C for 2 min and final extension at 72 °C for 10 min. The PCR products were resolved in 1.5 % agarose gel electrophoresis and purified using a PCR clean-up system (Promega Corporation, Madison, WI, USA). The purified products were subjected to DNA sequencing using ABI 3500 series Genetic Analyzer (Applied Biosystems®, USA)

Data Analysis
The K46 locus sequence of Kasalath (GenBank accession no: AB458444.1) was retrieved from GenBank (www.ncbi.nlm.nih.gov/genbank/) and was used as the reference sequence. The amplified K46 DNA sequences of rice cultivars were aligned with the reference sequence using ClustalW software (www.ebi.ac.lk). The DNA polymorphism of K46 locus among the selected rice cultivars were subjected to cluster analysis by using pair-wise Simple Matching Coefficients (Webb and Copsey, 2011) calculated based on the shared single nucleotide polymorphisms (SNPs). A dendrogram was constructed by unweighted pairwise group method using average (UPGMA) based hierarchical clustering (Nei and Kumar, 2000). The haplotypes and PD tolerance scores of the rice cultivars (Aluwihare et al., 2015) were subjected to association analysis using the Statistical Package SPSS 16 (SPSS Inc., 2007;Landau and Everitt, 2004).

RESULTS
PCR amplification of K46 locus produced a distinct band having an approximate size of 500 bp for 15 out of 30 rice cultivars ( Figure 1). When these PCR products were subjected to DNA sequencing, the exact length was revealed as 523 bp which is compatible with the K46 locus in Kasalath (Chin et al., 2010). A total of six SNPs were detected at the 47 th , 225 th , 226 th , 247 th , 339 th and 420 th positions along the 523 bp region, indicating the overall degree of polymorphism was 1.15 %. Out of the six SNPs, four transition and two transversion mutations were detected. All six SNPs were biallelic and were distributed along the locus without any apparent pattern ( Figure 2). The DNA sequence polymorphism in the K46 locus of 16 cultivars including Kasalath revealed three clusters (i.e. three distinct haplotypes) at 84.33 % of sequence similarity. Cluster one (C1) contained nine rice cultivars including Kasalath and therefore, was referred to as the Kasalath-like cluster, cluster two (C2) included two cultivars Pokkali and At 354 and cluster three (C3) contained five rice cultivars. The SMC based UPGMA clustering revealed that the Kasalath-like cluster (C1) was 84.33 % similar to C2 and C3 was only 58.33 % similar to the other two clusters ( Figure 3). The set of 15 cultivars in which the priming site of K46 locus is absent was considered as the C4. When the overall score of PD tolerance of the cultivars was compared with the K46 haplotype, it was found that, PD tolerance score of the cultivars were not significantly associated with the K46 haplotypes (Table 1; p < 0.05).

DISCUSSION
Molecular breeding (i.e. MAS or MAB) is a theoretically sound and logical timely paradigm shift from the conventional breeding (Collard and Mackill, 2008;Xu and Crouch, 2008). However, complex polygenic traits provide a great deal of skepticism in molecular breeding when it is put into the practice as they are also controlled by the environment (Xu and Crouch, 2008). Although PD tolerance is an important trait in rice breeding, it is a complex trait controlled by genetic and environmental factors. The characteristic feature of plants under P deprived conditions is the enhancement of the expression of specific genes to uptake more Pi from roots in P deficient soil conditions. Pup1 has been identified as the major QTL for P deficiency tolerance in rice causes enhanced Pi uptake under P deprived conditions (Wissuwa et al., 2002). The DNA marker K46 was developed along with another set of markers to specifically target the INDEL region of the Pup1 in Kasalath-like rice genomic background and to introgress the Pup1 QTL effectively into the missing backgrounds (Chin et al., 2010;Chin et al., 2011, Pariasca-Tanaka et al., 2014. However, the present study has clearly indicated that the employment of K46 in molecular breeding for improved PD tolerance is quite inefficient (Table  1). Maker validations for country specific programs is a prerequisite before launching MAB into routine practice given the nature of higher diversity of rice germplasm (Ni et al., 2002;Gao, 2003) and utilization of a narrow set of genotypes from the global rice germplasm for initial discovery of markers.

Suwandel
The absence of K46 locus in the cultivars strongly imply the fact that they also lack the INDEL region of Pup1 as in Nipponbare yet possess unrevealed molecular mechanisms to confer the tolerance for PD tolerance. However the possibility of the presence of paralogous K46 like loci in elsewhere of the genome of these cultivars cannot be simply ruled out (Thiel et al., 2009). Therefore further studies are required to locate the K46 locus of the genomes of these cultivars using techniques such as fluorescent in-situ hybridization (FISH) and linkage mapping. The utilization of FISH and linkage mapping to genomically locate the target loci is common in genetics (Rastogi et al., 2013;Wissuwa et al., 1998, Wissuwa, 2003. Interestingly there are two PD sensitive rice varieties Bg 357 and Bg 300 which do not have the K46 locus and show the sensitivity to PD (Aluwihare et al., 2015). The PD sensitive rice cultivar At 306 contains the Kasalath-like haplotype for K46 locus indicating that the mere presence of Kasalath-like haplotype is not guaranteeing the PD tolerance by default. The presence of two unique K46 haplotypes in the studied set of rice germplasm reveals the molecular diversity within K46 locus may be related to the overall degree of PD tolerance. The association analysis between the class of haplotype and the categories of the traits is a routine diagnostic procedure in genomic studies to find the marker trait associations (Bush and Moore, 2012). The chi square value, the Cramer's V coefficient and the associated probabilities indicated the strength and the significance of the associations (Sokal and Rohlf, 1987). In the present study it is clearly shown that K46 haplotypes are not significantly associated with the score of PD tolerance (p > 0.05) ( Table 1). Therefore it is safe to conclude that neither the sheer presence of K46 locus in a rice cultivar nor the underlying haplotypic variation provide the necessary basis for marker-assisted selection. However chi-square association analysis of this nature often suffers the inherent weakness of the smaller sample size and statistical demand to have more stringent chi square thresholds (McDonald, 2014). However, in practical genetics and breeding research designs such kind of bigger sample sizes cannot be achieved as certain haplotype vs. trait category combinations are naturally rare and not available in abundance for studying. Although these complexities are present in many polygenic eukaryotic traits such as PD tolerance in rice, the value of having DNA markers as selections tools in breeding cannot be undermined (Xu and Crouch, 2008). Therefore marker validation is very much required and Xu and Crouch, (2008) clearly explained what steps are required in MAB to go from publications to real practice. It would be compulsory to molecularly characterize the entire Pup1 region of a much larger panel of rice cultivars that are important to the country specific breeding programs. More in-depth analysis for QTLs with genome scans using novel state of the art SNP (Chen et al., 2014;Zhao et al., 2011) and genotyping by sequencing (GBS) (Poland and Rife, 2012;Arbelaez et al., 2015) platforms would be necessary to dig for any potential QTLs other than the Pup1. Although it is very expensive, such kind of and large scale research efforts are required to accurately establish the MAB platforms to produce PD tolerant rice varieties to conduct profitable rice farming with less input in the future.

CONCLUSION
The K46 marker locus within the Pup1 QTL is present only in the 15 rice Sri Lankan cultivars studied. There are four K46 haplotypes within the studied set of cultivars namely Kasalath like haplotype, two unique haplotypes and the 'null' haplotype (i.e. the INDEL region is missing). However these haplotypes are not significantly associated with the degree of PD tolerance suggesting the presence of hitherto unknown mechanisms of PD tolerance in the local germplasm.