DNA fingerprinting with homologous multilocus probes and search for DNA markers associated with yield attributes in silkworm , Bombyx mori

RFLP clones harbouring multi-copy DNA sequences were isolated from the Pst I sub-genomic library of the indigenous silkworm race, Nistari, and were used for DNA fingerprinting studies in 13 stocks of silkworm, Bombyx mori L. Six multilocus probes produced 180 RFLP markers that showed a high level (98%) of polymorphism and are highly useful in molecular mapping, genotype characterization and marker assisted selection (MAS). The dendrogram derived from UPGMA analysis clearly divides the 13 silkworm stocks into two major clusters: highand low-yield stocks. Furthermore, adopting multiple regression analyses, the RFLP marker(s) associated with characters of economic importance were identified, a first of its kind for any species of insect of commercial importance. The results obtained create an opportunity of using germplasm stocks directly for isolating specific RFLP band(s) and use it for MAS in breeding programs.


INTRODUCTION
The Restriction Fragment Length Polymorphism (RFLP) profiling (Botstein et al., 1980;Tanksley et al., 1989) proved to be a very important technique for identi fying genotypic differences.It is applied to both plant and animal systems (Jeffreys et al., 1985(Jeffreys et al., , 1991;;Miller & Tanksley, 1990;Song et al., 1990;Jung et al., 1993), especially because of their reproducibility (Paull et al., 1998).The relationship between different species as well as other taxonomic issues has been examined with improved methodologies for generating RFLP profiles (Virk et al., 1996;Salimath et al., 1995;Lu et al., 1996).The polymorphic DNA markers also helped in the con struction of a linkage map for various plant (Tanksley et al., 1992;Paterson et al., 1995) and insect species (O'Brien, 1993) including the silkworm, Bombyx mori (Shi et al., 1995).The banded krait minisatellite (Bkm-2 (8)) has also been used successfully as a probe in gener ating highly polymorphic RFLP profiles for several stocks of B. mori (Nagaraju et al., 1995).Probes were also developed from homologous genomes by the diges tion of genomic DNA with specific restriction enzymes for identification of clones (Jung et al., 1993).However, such a system has not been utilised for assessing genetic variability among the different stocks and breeds of B. mori.The present report is based on the result of such an attempt.As indicated elsewhere, the system is based mainly on the generation of multi-copy DNA sequence probes (Sulaiman et al., 1995).
B. mori, commercially, contributes more than 90% to total silk production (Currie, 1996) and is represented by a large number of stocks having a wide range of yield potential (Chatterjee & Datta, 1992;Chatterjee, 1993).Conventional studies were made to understand the genetics of yield components, but until now, no attempt has been made to identify specific DNA marker(s) con tributing to the wide variability in yield potential.Further more, the low yielding silkworm stocks are less susceptible to diseases and abiotic stresses, while the high yielding breeds and hybrids need strict disease manage ment and control of abiotic factors (Goldsmith, 1991;Chatterjee et al., 1993) for fuller realization of yield potential.Indian silk-cocoon producers include a large section of small and marginal farmers, for whom strict disease management and high input of fertiliser and disin fectants are a well-defined economic constraint.Thus, molecular analysis of these yield components and identi fication of DNA markers associated with high germ-load tolerance and high yield potential are of great significance for establishing work on directional molecular breeding of the silkworm.The relevance of DNA markers for molecular breeding and conventional breeding has been extensively highlighted elsewhere (Mitra et al., 1999), and is of great relevance for India and other countries of the third world practicing sericulture.
The identification of specific DNA markers associated with such attributes of economic importance, including that of disease resistance, is in itself an important issue.One of the approaches that has been adopted by Michael more et al. (1991) was the generation of bulk-segregants based on specific yield component or disease resistance/susceptibility among the F2 population.The F2 population is raised from the cross between two stocks, divergent for the specific character under study.Another approach is to establish a similar association on the basis of correlation between DNA profiles and specific pheno typic characters (Virk et al., 1996;Yonash et al., 2000).The present report is based on the result of analyses fol- lowing the second approach and is first of its kind for any species of insect having economic importance.

Silkworm stocks
The present study was based on thirteen races and breeds of Indian, Chinese and Japanese origin (Table 1), and these stocks are maintained in the lab for more than 10 generations.The mean estimates of ten yield components and the status of voltinism for the thirteen stocks utilized, given in Table 2, are taken from published manual of Central Silkworm Germplasm Resource Centre (Thangavelu et al., 1997).

DNA extraction
The genomic DNA was isolated from posterior silk glands of a minimum of ten four days old, fifth instar larvae for each of the thirteen stocks mentioned earlier.To study inheritance, the genomic DNA was extracted from individual moths of BMI-0017 (low yielding Indian race) and BBE-JAP (high yielding Japanese breed) as well as their F1 hybrid progeny.The silkglands or the moths were crushed in liquid nitrogen, subjected to lysis in the presence of proteinase-K at 37°C and followed by the normal method of phenol-chloroform extraction with ethanol precipitation.The DNA initially obtained was re-extracted in the same way after RNase A treatment and finally dissolved in 10mM TE buffer [Tris-Na2EDTA (ethylene-diamine-tetra-acetic acid, di-sodium salt)].Table 2. Average estimates (Thangavelu et al., 1997) 2) indicate Indian race and Indian breed, respectively.EGGNO = mean number of eggs laid by one mother moth; TLD = total larval duration; VLD = duration of fifth instar; LWT, PWT, CWT, and SHWT denote weight of single matured larva, pupa, cocoon, and cocoon shell, respectively; SR% = ratio of weight of shell to the weight of cocoon; YLDNO and YLDWT = number of cocoons realized /10,000 larvae reared and their weight.VOLT = voltinism; ND and D denote non-diapause and diapause, respectively.

DNA probes
A size selected (0.5-2 kb) Pst I silkworm genomic library was prepared in the pUC18 using genomic DNA from BME0017 (Nistari), one of the oldest Indian races of silkworm.Multiple copy recombinant clones were selected by colony hybridization using 32P-labelled total silkworm genomic DNA as a probe (Sulaiman et al., 1995).For Southern hybridization, probes with insert length varying between 0.8 to 1.8 kb, were labeled with a 32P CTP using a random primer labeling kit (Amersham Int., Buckinghamshire, UK) (Sambrook et al., 1989).

Restriction digestion, electrophoresis and Southern hybridi zation
10-12 pg of DNA per sample was digested to completion with different restriction enzymes (New England Biolabs, Hert fordshire, England).Four restriction endonucleases such as Bam HI, Eco RI, Eco RV and Hind III were used for the DNA finger printing studies.Digested DNA was separated by electropho resis on 0.8% agarose gel in 1 X TAE (40mM Tris, 20 mM acetate and 1mM EDTA) buffer.The gel fractionated DNA samples were vacuum blotted to a Hybond-N Nylon membrane (Amersham Int., Buckinghamshire, UK).Subsequent to baking at 80°C for 2 h, the membranes were hybridized (Sambrook et al., 1989) at 65°C for 16 h, and washed several times at 65°C in descending concentration of 2.0 X to 0.1 X SSC (sodium chloride/sodium citrate buffer) containing 0.1% SDS (sodium dodecyl sulphate).Subsequently, the blots were exposed with Kodak X-ray film with intensifying screens for 1-3 days at -70°C.

Data analysis
Binary scoring (0 and 1) was done from RFLP profiles observed for 13 different silkworm stocks using six different probes in autoradiograms.For all DNA profiles, only distin guishable fragments greater than 2.0 kb were scored.Pairwise comparison of DNA fingerprint lanes was done only within a gel.
Estimation of similarity index.For clustering, the similarity index (S) was calculated after Nei and Li (1979), as the fraction of shared fragments between pairs of silkworm stocks.For silk worm stocks, x and y, S = 2Nxy/Nx+Ny where Nxy was the number of common fragments in the two silkworm stocks, while Nx and Ny are the number of fragments scored in stocks x and y, respectively.For the thirteen different genotypes, S was con verted to a genetic distance as D = 1-S.The dendrogram was constructed using the "UPGMA" (Unweighted pair-group method using arithmetic average) method in the PHYLIP pro gramme (Felsenstein, 1989).
Correlation and multiple regression analysis.All DNA bands scored were considered as independent variables while yield components of silkworm stocks were taken as dependent variables.First, Pearson correlation coefficients were deter mined for all DNA markers.DNA markers showing positive correlation with yield components were identified and analyzed using multiple regression analyses.The provision of stepwise multiple regression analysis available with SPSS 10.0 program was used with > 0.45 and < 0.099 as F-value to enter and remove, respectively.6 for their significant association with LWT and SR%, respectively; b -the arrow-heads point out the polymorphic bands between the two parents and illustrate the detection of homo-or heterozygous status from the segregation of the marker(s) in Fi individuals.The duplex-band of ~ 5.0 kb in lane 2 appears very prominent in comparison to the faint band in lane-2 ofF inFig.1b.

DNA fingerprinting analysis
A total of 1056 recombinant plasmids were obtained from the partial genomic library developed from Pst I digest of silkworm genomic DNA.On dot blot screening, 245 recombinant plasmids showing signal of moderate intensity were selected as putative multiple copy clones.Of these, twenty-five clones were randomly selected and screened for their polymorphism on two selected diver gent silkworm stocks, BMI-0017 (low yielding) and BBE-JAP (high yielding).Four restriction enzymes, Bam HI, Eco RI, Eco RV and Hind III were used for initial screening (Fig. 1a, b).Among the above four restriction enzymes, Eco RI and Eco RV produced maximum poly morphism (85%) when compared to Bam HI or Hind III (60%).Southern hybridization showed that the cloned fragments of all 25 recombinant plasmids were distrib uted in the size range of 2 to 23 Kb (Fig. 1a,b).However, six of these pBmN clones (138(A), 319(B), 445(C), 803(D), 948(E), and 966(F)), with an insert size varying from 0.8 to 1.8kb and a high level of polymorphic pattern between these BMI-0017 and BBE-JAP stocks, were sub sequently selected for fingerprint analysis in the thirteen silkworm stocks (Fig. 3a,b).
The six multilocus probes produced 180 RFLP markers across the 13 silkworm stocks and of these, 38 fragments were not present in more than one stock.The number of unique fragments was found to be highest for BME-0005, while, no unique fragment could be detected for BBI0081 and BBI0082.Inheritance pattern for the few selected multilocus probes such as pBmN 138, 319, 803 and 966 were studied in the F1 progeny of BMI-0017 X BBE-JAP cross (Fig. 2a,b).The pattern presented in Fig. 2a,b, clearly substantiates the inheritance of the RFLP markers from parents to offspring.From the profiles of F1 indi viduals and their parents, presented in Fig 2a,b, it is also evident that before planning to utilize specific markers for MAS program, it is possible to determine whether a marker in a parent is homo-or heterozygous.For example, the presence of the marker A-8 in all F1 indi viduals suggests that it is homozygous in the parent-1.On the other hand, the absence of the band A-3 in three of the five F1 individuals tested indicated it to be in heterozy gous condition in parent-2.Likewise, the first two markers at the top of Fig 2b (indicated with arrowhead), could be detected in all the offspring, showing its homo zygous status in parent 2 because this is the only parent possessing these markers.The distribution of the last two marked fragments exhibits their heterozygous status in  5 and 6).parent 2.However, mention should be made of the very prominent duplex-band of ~ 5.0kb in lane 2 of Fig. 2b.The band in the same position in Fig. 1b (lane 2 of F) appears very faint, possibly due to a bleaching effect in the autoradiogram as evident from black smear in the left corner.However, it may be also noted that DNA template used for work presented in Fig. 1 is from a minimum of 10 individuals, isolated en-masse, while that in Fig. 2 is from a single parent of the particular cross.

Genetic variation
Each hybridising fragment from the DNA fingerprint for 13 silkworm stocks (Fig. 3a,b) was scored as a piece of independent data and a genetic similarity index (S) for each pair of silkworm stocks was calculated (Table 3), which revealed the similarity index ranging from 0.23 to 0.75 with the average of 0.40.It is also of interest to note that the highest similarity was found between the Chinese stock, BME-0052, and the Indian stock, BMI-0002.

Genetic relationship among silkworm stocks
The dendrogram derived from UPGMA analysis clearly depicts the presence of two major clusters (Fig. 4).The Cluster-A includes all the five Indian stocks of low yield potentials (Table 2) and two Chinese stocks, almost of similar yield status.The Cluster-B includes all improved Indian breeds and one Chinese stock, BBE-CHI, of medium yield of 8.03 kg cocoons/100 broods.The den drogram further reveals that BME-0005 enters the cluster combination at the eleventh (last) step of agglomeration while in the preceding step, cluster-pair of BBI-0095 and BBE-CHI, joined the small sub-cluster of BBI-0044: BBI-0081: BBI-0082: BBE-JAP.
Results of stepwise regression analyses with the selected 70 fragments and eleven dependent variables (ten yield components and voltinism) are presented in Table 5.It is evident that within the limit fixed for F-to enter and F-to remove, the analysis selected only three fragments for EGGNO, CWT and SHWT while for TLD, the number of alleles contributing positive effect is 9, which is the maximum.Some fragments have significant con tributory effect in the formation of regression equation for more than one character.For example, D5 was selected for EGGNO, TLD and VLD while F24 appeared signifi cant for TLD and LWT.Further, the relative importance of some bands differs for different yield components, e.g., D5 was selected at step-2 for EGGNO and TLD, while it was selected at the step-1 for VLD.Summarizing the data in Table 5, it may be said that 27 fragments are being selected out of 70 fragments analyzed for multiple regres sion, three of which are marked in Fig. 3a,b.
To test the efficacy of the selection of fragments through MRA, the status of marker(s) selected at the first step of stepwise regression in the thirteen silkworm stocks was examined by t-test (Table 6).Result for all the yield components, except EGGNO, show that the group mean for those stocks where the band is absent signifi cantly differs from the group mean for those stocks where the band is present.Shi et al. (1995) showed that the silkworm genome with comparatively higher GC content can produce high poly morphism when restriction enzymes with six base pair recognition sequences are used.The presented results cor roborate the above finding.Similar level of polymor phism was also observed in plants (Neuhusen, 1992;Bark & Havey, 1995) and in insects (Musca domestica; Blanchetot, 1991).

Level of polymorphism
The level of polymorphism (98%) presently realized is higher than that realised with various PCR based markers such as RAPD (94%; Nagaraja & Nagaraju, 1995), ISSR (86%; Reddy et al., 1999b) and SSR (77%; Reddy et al., 1999a) or that realised with the use of Bkm-2(8) probe (Nagaraju et al., 1995).Landry et al. (1987) comparing various sources of probes for their efficiency in detecting RFLP in lettuce concluded that an increased detection of polymorphism in multiple copy sequences could be the result of independent mutation in the different copies of the gene in the presence of a relevant selection pressure.Noli et al. (1997) and Hahn et al. (1995) also observed a higher level of polymorphism in barley and maize, with homologous probes against that realised with RAPD markers.

Genetic diversity
Genetic similarity estimates based on RFLPs have been shown to be good in accordance with pedigrees in several crops (Smith et al., 1990) and have been used to identify divergent parents for Fi hybrid production (Powel et al., 1996).Reliable estimates of genetic diversity within and between silkworm stocks, as revealed by the RFLP analysis (Powel et al., 1996), are important for the main tenance of germplasm stocks and selection of parents for the development of elite hybrids with higher yield poten tial.
Earlier studies showed that the number of probes required for generating sufficient number of markers for assessing genetic similarity varies from 15 (Dos Santos et al., 1994) to 100 (Messmer et al., 1993).However, in the present approach substantial numbers of bands were gen erated with only six probes.The same has been consid ered as sufficient to estimate genetic similarly as it was projected by Tivang et al. (1994) in maize, that the number of bands required for a CV of 10% was 388, 150 and 38 for closely, intermediately, and distantly related inbreeds, respectively.Pejic et al. (1998) while using the bootstrap procedure suggested that 150 bands are suffi cient for reliable estimates of genetic similarity.

Genetic relationship among silkworm stocks
As indicated in an earlier paragraph the clustering or grouping on the basis of RFLP marker clearly demarcates the yield potential of 13 silkworm stocks studied.In addi tion, the closer association between BBI-0044 and BBI-0081 is substantiated with the information on the origin of these two breeds, both of which were developed from the commercially available double hybrid of (Kokko x Seihaku) x (N124 x Cm).The greater distance of BBI-0082 from the above two breeds is supported by the fact that the latter was developed from a different Japanese hybrid namely Kinshu x Showa (Chatterjee, 1993).On the other hand, BBI-0095 was developed (Thangavelu et al., 1997) from another Japanese double hybrid [(N122 x C110) x (N124 x C124)].However, it is difficult to explain the association of BBE-CHI with BBI-0095.It is possible that sometime during the breeding process a common parent was used to infuse certain specific characters.
With regard to the relationship between the low yielding stocks, certain aspects need to be highlighted.Though BME-0005 is very close to BMI-0017 or BMI-0001 with regard to the economic character, they do not occupy close position in the dendrogram.This is not unusual if one considers the origin of these races.BMI- 0017 is a race used on the Gangetic plain for not less than 100 years (Mukherjee, 1912), whereas BMI-0001 is a race used in Southern peninsula where sericulture devel oped later.Thus, the distance reflected in the dendrogram appears justified.The position of C'Nichi is quite inter esting.Chatterjee & Datta (1992) indicated that this race is a segregant from the original hybrid Shi.Nichi.This hybrid was introduced during early part of the last cen tury (Ghosh, 1949).Other studies in this laboratory (unpublished) also show the presence of DNA markers specific to high cocoon weight and high shell weight in this race, though the expression of high cocoon weight and high shell weight could not be realised.

Association between DNA markers and yield components
Different groups of scientists have utilized different sta tistical approaches to ascertain the association(s) between molecular markers and yield attributes.For instance, Barbosa-Neto et al. (1996) and Virk et al. (1996) adopted linear regression, while Lynch (1999) and Yonash et al. (2000) utilized genetic correlation and single band/multiband analysis, respectively.
The correlation analysis enabled us to find 70 RFLP markers showing significant association with ten yield components.Pearson correlation estimates do not take into account the multiple interactions between inde pendent variables on one hand and that between inde pendent and dependent on the other.Multiple regression analysis is a relevant tool in this context, as projected by the "multiband" analysis for establishing linkage between RFLP markers and antibody response in meat-type chickens (Yonash et al., 2000).Virk et al. (1996) also adopted a similar approach in analyzing quantitative variation within rice germplasm using RAPD markers.The application of t-test (Table 6) attests significance to the difference between the two means for all components except EGGNO, thereby strengthening the adoption of MRA for establishing association between DNA markers and yield attributes.
Thus the result obtained creates an opportunity for iso lating specific bands for further characterisation of the fragment(s).Moreover, such selected markers can help in marker assisted selection (MAS) program for molecular breeding (Stomberg et al., 1994).The success of such selection programs depends exclusively on the extent of linkage between markers and the relevant loci such as QTLs (quantitative trait loci).This requires further study with the specific markers selected from such analyses.package; QTL = quantitative trait loci; RFLP = restriction frag ment length polymorphism; SDS = sodium dodecyl sulphate; SR% = ratio of weight of shell to the weight of cocoon; SSC = saline sodium citrate; TAE = Tris acetic acid; TE = Tris-EDTA; TLD = total larval duration; TRIS= hydroxymethyl-aminomethane; UPGMA = unweighted pair-group method using arith metic average; VLD = duration of fifth instar; VOLT = voltinism; YLDNO and YLDWT = number of cocoon realized /10,000 larvae reared and their weight in kg.

Fig. 2 .
Fig. 2. The hybridization pattern of Eco RV digested DNA from parents, BMI-0017 (lane-1) and BBE-JAP (lane-2) silkworm stocks and their F1 offspring (lanes 3-7) with the multilocus probe, pBmN 138. a -bands A13 and A8 marked with arrows are two of the eight bands projected in Table 6 for their significant association with LWT and SR%, respectively; b -the arrow-heads point out the polymorphic bands between the two parents and illustrate the detection of homo-or heterozygous status from the segregation of the marker(s) in Fi individuals.The duplex-band of ~ 5.0 kb in lane 2 appears very prominent in comparison to the faint band in lane-2 ofF inFig.1b.

Fig. 4 .
Fig. 4. The dendrogram based on RFLP markers illustrating the classification of silkworm stocks into two well defined sub clusters, comprising of stocks having low (sub-cluster-A) and high yield (sub-cluster-B) potential.

Table 3 .
Genetic similarity index (S) between pairs of silkworm genotypes.

Table 4 .
List ofRFLP markers showing significant positive and negative correlation with selected yield parameters.

Table 5 .
Result of multiple regression analyses (Stepwise method) for identifying significant association between yield components and specific RFLP markers.

Table 6 .
Results of t-test for the difference in the mean values realized from a particular yield component in stocks showing the presence and absence of the band reflected at the first step of MRA.