Regression Analysis for the Identification of RAPD Markers Linked To Drought Tolerance in Sorghum

Sorghun (Sorglum bicolor) can actually withstand in dry or drought condition better than other crops, therefore it canbe grown at different agroclimatic conditions and its product can be used for different PurPoses such as food, feed and industrial raw material. However at severe condition, the productivity will also drop drastically. The aim of this research was to identify RAPD marker linked to the drought tolerance. In this research, varieties of sorghum used as research matedals were Durra, Zhengzu, the mutants of Durra and Zhengzu (from 300 Gy gamma radiation) 8-100 and Zh-30, and the F2 seeds from Zh-30 x 8-100 and B-100 x Zh-30. Drought screening was caried out using 0.3 % KI during sorghum vegetative stage. DNA extraction was done using a modified CTAB method. PCR was carried out for RAPD analysis. PCR amplification products were scored and analyzed using SAS program. The result showed that potassium iodide can be used for drought screening dudng the vegetative stage and regression analysis using the logistic method can be used to identify RAPD markers that is linked to drought tolerance in sorghum. The logistic analysis showed that band ,48-480 was linked to drought tolerance in sorghum. Ke1-words : drought, molecular marker, logistic regression, sorghum Introduction Sorghum is the fourth most important world cereal following wheat, rice and corn. It is a staple food in the drier parts of the world such as Africa, India and also China (Zidenga,2004). The grain can be used for flour to make bread, porridge and for brewing beer. The large juicy sweet stems are used for chewing and making syrup and sugar. Sorghum grain and straw, taken after harvesting can be used to feed cattle or other livestock. Other uses are for making brooms, fences, or starch production from cultivars with waxy endosperms (Purseglove, 1972). Sorghum has been planted in dry areas as *Corresponding author : Taryono, Department of Plant Breeding, Faculty of Agriculture, Universitas Gadjah Mada Jl. Flora,Bulaksumur,Yogyakarta, Indonesia, 55281 E-mai1 : taryono@faperta.ugm.ac.id it can withstand dry condition or drought better than other crops (Carualho et a1.,2000). However, in an extreme condition sorghum can also fail to reach its optimum yield because water is an essential need. Since sorghum can withstand dry condition, it is interesting to study the mechanism of its molecular activities. Subudhi efal. (2002) discovered that sorghum can grow in dry conditions due to its stay green characteristics. Observations showed that the stay green characteristics may be controlled by more than one gene. Yield is also another characteristic which is known to be controlled by more than one gene or polygenic and there is also epistasis and interaction with the environment (Yin e f al., 2003).If more than one gene is involved then identifying or locating them becomes difficult, due to the inheritance of the characteristics cannot be easily identified from the phenotype. The quantitative trait


Introduction
Sorghum is the fourth most important world cereal following wheat, rice and corn.
It is a staple food in the drier parts of the world such as Africa, India and also China (Zidenga,2004).The grain can be used for flour to make bread, porridge and for brewing beer.The large juicy sweet stems are used for chewing and making syrup and sugar.Sorghum grain and straw, taken after harvesting can be used to feed cattle or other livestock.Other uses are for making brooms, fences, or starch production from cultivars with waxy endosperms (Purseglove, 1972).Sorghum has been planted in dry areas as can also fail to reach its optimum yield because water is an essential need.
Since sorghum can withstand dry condition, it is interesting to study the mechanism of its molecular activities.Subudhi efal. (2002) discovered that sorghum can grow in dry conditions due to its stay green characteristics.Observations showed that the stay green characteristics may be controlled by more than one gene.Yield is also another characteristic which is known to be controlled by more than one gene or polygenic and there is also epistasis and interaction with the environment (Yin e f   al., 2003).If more than one gene is involved then identifying or locating them becomes difficult, due to the inheritance of the characteristics cannot be easily identified from the phenotype.The quantitative trait is controlled by allelic differences but also affected by the environment.From the possibility of many genes involved, Kearsey and Pooney (1996) coined the term polygenes and more recently it is known as QTL (Qu an tit atiae Tr ait Lo ci).
Quantitative hait loci actually describes an area on the chromosome which is explained by linkage with one or two markers close to the QTL of a specific trait.The markers used are usually DNA markers (Gupt4 2002), because DNA marker is not affected by the envfuorunent.According to Mulengwa et al. (2005), molecular markers can provide a powerful tool for crop improvement.To detect molecular markers, several techniques such as hybridization, polymerase chain reaction (PCR) or based on DNA sequence can be explored (Collard et al., 2005).Random amplified polymorphic DNA (RAPD) is PCR based techniques which produce a dominant marker and can be used for many organisms without prior knowledge of its nucleotide sequence (Zidenga,2004;KumN, et a1.,2006).
From statistical perspective, methods for QTL mapping are generally based on least square, maximum likelihood and Bayesian approaches (Wang ef a1.,2006).Gupta (2002) mentioned three methods to detect QTL l.e single marker analysis, simple interval mapping (SIM) and composite interval mapping (CIM).The single marker analysis is the earliest and simplest methods to detect QTL.Single marker analysis can be approached using the regression method (Haley and Knott,1992).Kalbehdari and Robinson (2007) reported that linear regression has been able to be used with single and multiple markers.to stimulate the effect of drought stress by inhibiting current assimilation.Using small p)ots of wheat, a solution of magnesium or sodium chlorate with 4% active ingredient was sprayed on the plants surface.The chemical which was a desiccant killed all photosynthetic tissue including leaves, leaf sheath, glumes and awns.When plants were devoid of chlorophyll then it could proceed grain filling only with plant reserves.Potassium iodide could also be used as it has been done in Australia (Farag,2004).
The objective of this research was to identify RAPD markers that are linked to drought tolerance in two sorghum varieties (from ICRISAT and China), their mutant and non-mutant populations.

Materials and Methods
Materials used to identifl RAPD markers linked to drought tolerance x'ere sorghum varieties Durra which '"vas introduced from ICRISAT and Zhengzu from China, their mutants 8-100 and ZH-30 and also the F, generation from Zh-30 x 8-100 and B-100 x Zh-30 (Figure 1) to produce 50 plants that were needed f or the DNA analvsis.The mutants used are from the \110 generation.
The mutant 8-100 was used as the male parent and is more tolerant to drought, higher in plant height (85-165 cm) and produced yellowish grain.The mutant Zh-30 rlas used as the female parent and n'as Iess tolerant to drought, shorter (80-130 cm) and produced white grains better for human consumption.The sorghum varieties have been used by PATIR-BATAN (The Center for application of Isotope and Radiation-National Nuclear Energy Centre) for plant breeding program using mutation.Mutation was induced using Gamma ray from Cobolt-60.
The F2 population that were used for this reaearch corsisted of 12 population (E-P).
Those 12 population were from crosses of Zh-30 x 8-100 and 8-100 x Zh-30.Population with codes E -M were from crosses of Zh-30 x 8-100 and N -P were from crosses of B-L00 xZh-30.

Ez.t aluation of drought tolerance
The plant materials used in this experiment were non-mutant Durra (A), non-mutant Zhengzu, mutant Durra (C) and mutant Zhengzu (D).The seeds were planted in plots and ar, anged in completely randomized design.Treatment with potassium iodide (KI) at 0.3% w / v) was carried out on the sorghum leaves at the vegetative stage or 5 weeks old plant by spraying a KI solufion and irrigation was kept normal.The selection criteria for tolerant and non-tolerant plants was observed foom the change in the colour of leaves.If the leaves turned brown then the plant is noted as non-tolerant.If the leaves turned vellow then the plant is tolerant.A minimum of 50 plants were observed and used for the DNA analysis.

DNA analysis
DNA was isolated from 3-4 week old plants using a modified CTAB (Cetyl trimethyl Ammonium Bromide) method.
The leaves were cut for samples (85 mg) and only one replication was sampled for DNA extraction.The DNA was purified using Ceneclean@ III kit (MP Biomedicals) and it was quantified using GeneQuant to measure the amount of extracted DNA.Dilution was done to obtain a final DNA concenration of 2.5 ng/pl for PCR analysis.
The PCR reaction was done with a total volume of 10 pl for each sample using the PCR kit AmpliTaq@DNA Polymerase Stoffel LJ.Biotech.
Fragment (Applied Biosystem).The six RAPD primers consisted of Proligo A8, A13, A20, D3,87, and T14 were used to amplify DNA template.Amplification was done using a PCR machine Perkin Elmer 9600.Initial heating was done at 94.C for 3 min, incubation at 95"C for 1 min, then followed by 45 cycles with each cycle at 94'C for 30 sec, annealing at 37.C for 30 sec and elongation at 72"C for 1 min 30 sec followed by final elongation at 72"C for 7 min.
To analyze the PCR product, DNA cocktails were electrophored on the agarose gel.
The gel was made with 76 nrl Psd HrO, "1.4 g agarose (1.75 % ) and 4 mI 20 x TBE buJfer which was a rnix of 0.45M tris-HCl pH8,0.45Mboric acid, and 20mM EDTA.The mixture was then diluted properly through warm boiling and when the solutionbecame warm, itwas added with 5 pl ethidium bromide.The electrophoresis buffer was made from 5 I of H,O, 250 ml of 20 X TBE and 250 pl of ethidium bromide.Electrophoresis was done after adding 2 pl Gl3 in each well containing amplified DNA from the PCR.As a marker, 10 gl of 100 bp DNA ladder was used.Electrophoresis was carried out for 2.5 h at 120 V.
Visualization of electrophoresis product was done using theFotodyne Image Analyzer with a UV light.Scoring of visualized bands was done by scoring 1 if the band was presence and 0 for no observable band.
Scoring was done for each individual plant.

Data analysis
The scoring data was used to estimate the linkage between the RAPD markers and drought tolerance in sorghum using regression analysis of SAS program and logistic procedures (Anonym, 2002).

Results and Discussion
Drought experiment Observation of plants that were tolerant or not tolerant to Potassium Iodida (KI) was done three days after spraying because the effect of KI did not appear immediately.With this method drought tolerant sorghum could be differentiated from the sensitive plant due to its stay green characteristics.With such technique, to screen drought tolerant cereals, it was not neccesary to be carried out in dry condition and saves cost needed from large area of land (Tdnstra et al.,'1996).
The use of KI caused a gradual loss of chlorophyll such as plants in a drought stress situation (Farag, 2004).Potassium iodide was sprayed at the vegetative stage because at this stage sorghum is tolerant to drought stress and the stress would not have an effect on yield.This was important even though yield was not one of the characteristics measured but the grains would be used again in further experiments.The use of KI simulated drought and has been used by several researchers such as Royo et al. (2008) inTriticale trujilo.Potassium iodide was given 10 days after anthesis and the reduced yield due to KI was almost the same as planted in dry condition.
According to Collard et al. ( 2005), the ratio expected for a dominant marker in an F2 population is 3:1 (B-:bb).The use of a dominant marker, however, could not differentiate a homozygote and heterozySote genotypes.The segregation ratio of the marker could be seen with a Punnet square.
In an F3 population, the ratio must be 3:2:3 (Zubay, 1987).By using a dominant marker the ratio became 5:3.If the gene controlling drought tolerance was controlled by a single dominant gene, then the ratio of the phenotype would be a ratio of 5 tolerant individuals : 3 non tolerant individual.
In this experiment, besides the mutant and non mutant parents, the F3 populations chosen were population L and M which showed segregation in drought tolerance (Table 1) and enough number of plants needed to carry DNA analysis.A minimum of 50 plants were needed to carry out QTL analysis.A segregation in the drought tolerant characteristics or a ratio that was not according to Mendelian ratio might be due to interaction of genes such as epistasis, linkage, crossing, selfing or the effect of the environment (Van Oosterom et a1.,1996).

Identification of RAPD market linked to drought tolemnce ifl sotghum
The simplest type of population used for QTL analysis was an F2 population which was easy to construct and required a short time to produce and had all the combination of alleles from the parent.In this experiment, the F3 populations were used because the number of seeds in the F2 populations were not enough for a QTL analysis which must be more 50 individual per population (Collard  et a1.,2005).However, F3 population have been known to be used as evaluation for F2 plants.Gupta (1996) mentioned progeny test of F2 individuals (F3 population) could still be used in making maps when the phenotype did not reflect genotype such as disease resistance or other characteristicss which were polygenically controlled.PCR analysis using primer A8 showed a specific band in the durra population.Mutant Durra can be identified using primer A13 with the presence of bands at 480 bp, 450 bp and 420 bp while mutant Zhengzu had a specific band at 380 bp (Figure 2).Using primer A20, a band at 650 bp showed for the mutant Durra population while primer D3 showed a specific band at 520 bp for mutant Zhengzu and can be seen in its F3 population.Primer 87 showed a specific band at 650 bp for mutant Zhengzu and 420 bp for mutant Durra which could be used to differentiate individuals in the F3 population.Primer T14 showed a band at420 bp which occured only in Zhengzu population. of marker types, therefore linear regression approach seems very similar to analysis of variance (Kearsey, 1998).Such methods did not locate the QTL but simply confirm that the "eyeballed" location indicated a real effeci.Logically, i{ the marker and the eTL were on different chromosomeg there will be no regression, but if they were on the same chromosome, the regression will be significant and show that the marker was close to the QTL.The easiest way to evaluate the fittest of the regression would be from the coefficient of determination ([{) .With a regression approach of RFLP markers ,Masood et al. ( 2004) showed several markers linked to some agronomic characteristics in rice.with 87 with R'? = 0.2555.According to the R'z value, primer A8 could be more likely used as a marker for drought tolerance in sorghum (Table 2).The correlation between markers and quantitative characteristics could also be observed from its regression coe{ficient (b) for each marker.Regression analysis based on the model : Y = a + b1X1 + b2X2 +...bjxj...bnxn + d + e, where b were partial regression coefficients which show a relationship between variable Y and X,, d was the residual effect and e wass error from Y due to variation in the environment (Yirk et a1.,2009).
According to Xu ef aI.
(1,998), rnany characteristicss in plants and animals which were economically important were expressed in binary forms.Binary characteristics which cannot be explained by Mendelian inheritance were called complex binary traits and could be explained using the probit model.In a probit model, the residual error was assumed to be in normal distribution while logistic regression used a maximum likelihood estimation after transforming the dependent variable into a logit variable.In this experiment both the regression method and the maximum likelihood method were used to find out the relationship between markers and drought tolerance in sorghum.
Regression analysis for QTL mapping have been used by many researchers such as Haley and Knott (1992) using F2 ppopulatiorL Rebai (1997) and Xu ef al.( 2005) using binary data.The results all showed that simple regression method was similar to using a maximum likelihood or probit model.From the six primers used (A8, AL3, A20, D3, 87, T1.4) which produced 96 polymorphic bands or markers then could be used to search RAPD markers linked to drought tolerance in sorghum using the stepwise program.Using the prcc reg step in SAS as it has been done by Haley and Knott (1992), marker 480 bp from primer A8 showed the most significant chance of being a marker I.J.Biotech.
for drought tolerance in sorghum.A similar result showed using the proc logistic step in SAS version 9 (Table 3).
The estimates (b) or the regression coeffiecients which were significant (from all 96 bands produced) in two markers or bands from primer A8 at 700 bp and 480 bp.The estimate or the value of b was larger in 480 bp which showed that this band could explain better or could detect more accurately the drought tolerance characteristics (Y) in sorghum.The result from the logistic regression analysis using stepwise program showed that the linear model for the relationship between drought tolerance characteristics (Y variable) and the bands from each primer (X variable) for Durra and Zhengzu sorghums was: y = -2.3409+ 1.5034 (A.8-700 bp) + 2.3959 (4.8_480 bp).
It showed that the use of .48-400bp band would provide more chance to screen drought tolerant lines than A8-700b, although both alleles (marked by the bands at .48-480 and A8-700) could be situated at the same chromosome for drought tolerance.48-480 bp band was probably located more close to the QTL due to a higher coefficient of 2.3959 show that it can be used as a stronger marker for the tolerant variety.This result is the same as in Figure 2 where a specific band occur at 480 for tolerant variety 8-100 (specific band showed by arrow).
Although this finding has not been applied to assist selectiory regession approach from practicle point of view would be the approach of choice because it offered greater speed and flexibility (Knott, 2005).It was relatively fast to be implemented, easy to be generalized to experimental population and easy to include cofactor.Regression approach showed good robustness properties against non-normality and the estimation procedue was also distribution free (Rubai, 1997).
Regression based QTL will continue to play a role, given its status as a simple framework in which to fit complicated model (Knott, 2005).
The use of molecular marker such as RAPDwould be more benefit to assist selection because molecular markers in general were scattered through the genome and their associations with various agronomic traits could be useful to help crop improvement program (Muthusamy et a1.,2008).Tanksley (1993) mentioned five inherent properties of molecular markers that was distinguished from morphological markers.The phenotype of most morphological markers could only be determined at the whole plant level, whereas molecular loci could be assayed at the whole plant, tisuue, and cellular level.Allele frequency tended to be much higher at the molecular loci compared with morphological markers.Alleles at morphological loci interacted in dorninanttecessive manner that limit the identification of heterozygotes genotypes, while some rnolecular loci exhibited a codominant made of inheritance.Only fewer espistatic and pleitropic effects are observed with molecular markers and large number of polymorphic markers could be generated and monitored in a single cross only.It was believed that the use of molecular marker can speed up new cultivar development.
From this experiment, it could be concluded that screening for drought tolerance using 0.3% KI on sorghum was efficient because screening could be done three days after treatment and not dependent on the season or weather, and RAPD could be used to obserwe variation in sorghum and (3) QTL analysis using regression can be used to find specific marker for drought tolerance in sorghum.
Indonesian Jounal of Biotectmology is a biannual international joumal devoted to tlle to tlle publication of articles on all aspects of biotechnology including agricultural biotechnology, health and medicine, social welfare, and environmental biotechnology and bioproccess technology.The scope of Indonesian Journal of Biotechnology is wide and multidisciplinary.It provides a forum in-tended to help realixe innovatiorL set challenges and provide a proper Blobal perspective on all matte$ conceming biotechnology.

SUBSCRIPTION
The annual subscdption p ce is IDR 100.000,00 not including delivery cost-Subsc ption is available ftom Research Center for Biotechnology, Universitas Gadjah Mad4 Jl.Teknika Utar4 Barek, Yogyakada 55281, Indonesia.All payments should be made in IDR and send to Dr. Widodo, M.Sc at BaIk Muamalat UGM Yogyakarta, account number : 0183921842 or Dra.Eruni Triastuti at Baik Negara Indonesia (BM) UGM Yogyakarta, account number | 0038633702

INSTRUCTIONS TO THE AUTHORS Submission
All manuscripts should describe original research which has not been reported or submitted elsewhere.Please send an oddnal and soft copies of the manuscript to the editorial olf ce.Manuscripts should be type double -spaced on one side of quarto or A4 papet and prepared with left and right margins of at least 2.5 cm.
The manuscdpt must be w tten in clear and concise English and sould comprise : The following inforrnation should be provided : (a) a concise and informative title, (b) names of author(s), one given name may be in full an aste sk should indicate the author responsible for correspondence, @ full nailint address, telephone, fax and e-mail address of the corresponding author.

Abstract and kErnods
The abstract should be written in one paragraph, comprehensible without reference to the text and not exceed 200 words.The rnain f ndings and new a]ld important aspects of the study should be emphasized.Up to f ve ke,'words or pfuases may may submitted Ior indexing purposes.

Text
Trne ifittoductionshotld state the purpose of the paper.lt should not review the subject exteDsiv ely.Moterials and Methods *oltldbe described in detail suJf cient to enable the work to be repeated, and suppliers oI materials and equipment givery iJ necessary.R€s lfs should be presented in logical sequence, with tables and illustation as approp ate.Disc ssiofl should emphasize the implication of the f ndings, tlrcir limitations, and rclevance to other studies.It should not be a recapitulation of the /esrlfs.

Number
Use Arabic numerals, not Roman or ordinal nurnbeN, throughouL including tables and f gures (e. t. croup 3, not the third goup or goup I ; table 1, not table I).Avoid bednning a sentence with a number or symbol-

References
The Harvard System is used.References in the tax should be quoted in the following for single,two or more than two authors respectively: Gula (1945), Perment and Breem (971), Celaly et al. (1990).Multiple papers refeEed to together in the text should be given in chronological order.
The references should be listed alphabetically at the end of the paper.The journal titles should be abbreviated to conform to the style oI Index Medicus.The layout should be as follows.

TabIes
TyPe each table on a separate sheet, number tlrcm consecutively and provide them with a brief title.When results are expressed as percentate, state the absolute value(s) that conespond to 100%.

Illustratiofi
Photographs should be gloss p nts with good conhast.Photographs must have intemal scale markers ald symbols, and arrows or letters should contrast with the back$ound.Where photo$aphs of gel, autoradiotrams etc, have been processed to enhance their quality, this should be stated.Costs of colour printing will be charged to the author.

Statistical methods
Tests of statistical signiJ cance should be identif ed, and references should be cited.Statements about the statistical signil cance of the results should be accompanied by indications the level of sitnif cance.
Blum ef nl.(1983) developed a technique Figure 1.A diagram of plant materials used in the tesearch

Table 3 .
Result from Maximum Likelihood Estimation using proc log step of SAS paramerer or tuit"* stlfn"d e, ' cr'isq