| Nucleic Acids Research | Pages |
Positional cloning without a genome map: using `Targeted RFLP Subtraction' to isolate dense markers tightly linked to the regA locus of Volvox carteri
Introduction
Materials And Methods
Strains and crosses
Volvox DNA and adaptors
Preparation of tracer and driver
RFLP Subtraction
Analyzing the TRS products
Results
The TRS strategy
Analyzing the products of TRS
Genetic analysis of the regA locus
Estimating number of clones in the TRS library that are tightly linked to the regA locus
Density of recovered markers that are tightly linked to the regA locus
Does a Volvox DNA modification system affect the nature of the TRS markers?
Discussion
Acknowledgements
References
Positional cloning without a genome map: using `Targeted RFLP Subtraction' to isolate dense markers tightly linked to the regA locus of Volvox carteri
ABSTRACT
INTRODUCTION
Our current understanding of the processes underlying basic biological phenomena such as inherited human diseases, the cell cycle and embryogenesis, owes much to the strategy of isolating genes corresponding to mutations affecting traits of interest. The power of this approach is reflected by the disproportionate amount of fundamental biological information that has come from work on the small group of experimental organisms including Saccharomyces cerevisiae, Drosophila and Caenorhabditis elegans from which it is relatively easy to isolate genes defined by mutant phenotypes. In contrast, isolating genes corresponding to important mutations in other organisms-such as those affecting organ development in vertebrates or crop disease resistance in plants-can be a formidable task. Positional cloning, a generally applicable approach for isolating genes defined by mutations, is labor and cost intensive and is usually applied to organisms with high resolution genetic maps. Such maps have recently been developed for human (1) and mouse (2), but for most organisms have yet to be constructed.
Methodology is needed to facilitate positional cloning in organisms which harbor mutations of interest, but for which neither powerful gene isolation technologies nor high resolution maps have been developed (3,4). An ideal method would directly yield region specific markers at sufficiently high density to identify overlapping genomic clones covering a gene of interest-without requiring a genetic map or chromosome walking. This, as yet unrealized, approach has been termed `chromosome landing' (5).
A genetic method, phenotypic pooling, has been used to facilitate the isolation of markers from specific loci (5-8). Phenotypic pooling, also known as bulked segregant analysis or pooled-sampling, effectively masks markers from all regions of the genome with the exception of the region immediately surrounding the mutation of interest. A composite DNA sample is constructed by mixing the DNA from strains that have the same genetic background at the locus of interest but whose chromosomes are mosaics of two different genetic backgrounds. The resulting DNA sample has useful properties resembling those of a DNA sample from an extensively backcrossed line. However, phenotypic pooling has the advantage of not requiring time consuming genetic crosses.
Phenotypic pooling has aided genome-wide marker scanning strategies, such as the Amplified Fragment Length Polymorphism method, by lowering the number of DNA samples needed for scoring thousands of markers (9-11). One recently described method, genetically directed representational difference analysis (GDRDA), uses phenotypic pooling combined with a subtractive method (12), to specifically isolate markers from a locus of interest (13). The advantage of this approach is that markers not located at the site of interest are automatically discarded. However, GDRDA generally isolates only two or fewer markers per experiment, even when the target region is large (13-16). This yield of markers is too low for efficient chromosome landing.
We recently developed a method, Restriction Fragment Length Polymorphism (RFLP) Subtraction, which isolated large numbers of randomly located RFLPs spanning the mouse genome (17). RFLP Subtraction produces many more markers than the subtractive method used in GDRDA (17). Thus, we reasoned that combining RFLP Subtraction with phenotypic pooling (for targeting a specific region) might provide enough linked markers to constitute a chromosome landing strategy.
Our interest in chromosome landing was motivated by our desire to isolate DNA markers near regA,a gene that controls cell fate in the eukaryotic green alga Volvox. Volvox provides a useful model for studying the embryonic mechanisms underlying the assignment of germ line and somatic cell fates and for understanding the evolution of cell specialization (18). Multicellular individuals develop from a single cell via embryogenesis (Fig. 1B) and are composed of [sim]2000 haploid cells of only two types: somatic ([sim]2000 cells) and germ line ([sim]16 cells) (Fig. 1A).
Figure 1. Volvox:cell types, life cycle and regA- phenotype. The regA gene determines whether a cell becomes a totipotent germ line cell or a terminally differentiated somatic cell (19,20). regA- mutations cause somatic cells to redifferentiate into germ line cells (Fig. 1C). Thus, in regA- organisms, all 2000 somatic cells, which normally die at the end of a generation, instead undergo embryogenesis. A regA- individual yields a brood of 2000 juvenile Volvox rather than the normal 16 progeny. The regA gene also exhibits an extraordinary developmental stage specific hypermutability upon UV irradiation (21). During a brief (1 h) period of embryogenesis, regA becomes [sim]105 times more mutable than its spontaneous rate of mutation. The regA product is required for suppressing the expression of germ line cell specific genes in somatic cells (22,23). We recently discovered that a major class of germ line specific genes regulated by regA is composed of genes whose homologues are coordinately regulated during the cell cycle of Chlamydomonas reinhardti, a single celled relative of Volvox (24). Based on these results we have suggested that modification of regA was required for conversion of coordinate cell stage specific gene regulation to coordinate cell type specific gene regulation during the evolution of cell specialization in the Volvocine line (24). Cloning the regA gene is necessary for understanding how the gene directs cell fate, why it becomes transiently hypermutable, and what role it played in the evolution of cell specialization. Since only a rudimentary genetic map exists for Volvox (25,26), taking a positional cloning approach would be difficult. We reasoned that TRS would provide a direct route to the regA gene and would also provide a general means to isolate phenotypically defined genes in diverse organisms. We demonstrate that TRS, without the aid of a genetic map, directly and efficiently isolated dense markers tightly linked to the regA locus.
MATERIALS AND METHODS
Strains and crosses
The wild-type Japanese Volvox strain HK10, referred to as J regA+, is a female strain of Volvox carteri forma nagariensis that was obtained from the Culture Collection of Algae at the University of Texas (Austin). The regA- Indian male strain (referred to as I regA-) is a spontaneous mutant (E. Ananiev, unpublished) of a male strain of V.carteri forma nagariensis that was originally isolated near Poona, India. The I regA+ strain was kindly provided by Richard Starr. All strains were maintained in Standard Volvox Medium (27,28).
The progeny analyzed in our experiments was obtained by mating (19) the I regA- male and J regA+ female strains. As has been previously noted (25), the viability of the zygotes of the J × I crosses was low, usually [sim]6% and part of the progeny was non-recombinant. All non-recombinants appeared genetically identical to the I regA- male parent according to analysis of chromosomal and organellar RAPD markers (E. Ananiev, H. Wu and D. Straus, unpublished). Based on these genetic data, the non-recombinant isolates are likely to be male parental contaminants. We suspect that the non-recombinants are due to contamination of the zygotes with somatic cells from the regA- parent. These cells are difficult to detect, due to their small size, but are competent to reproduce. In contrast, a previous report attributed the non-recombinants to pseudogamy or perhaps parthenogenesis (25).
Volvox DNA and adaptors
Volvox were harvested by filtering cultures through 30 µ Nitex filter (Tetko). DNA was prepared using a CTAB extraction procedure(29,30) followed by treatment with RNase A (Sigma), chloroform extraction, and precipitation with ethanol. The following oligonucleotides were purchased from the Midland Certified Reagent Corp. (Midland, TX) and purified by polyacrylamide gel electrophoresis: d(GACACTCTCGAGACATCACCGTCCA) (OL38), d(GATCTGGACGGTGATGTCTCGAGAGTG) (OL46), d(AATTCTTGCGCCTTAAACCAACA) (OL40), d(GATCTGTTGGTTTAAGGCGCAAGAA) (OL47), d(AGCACTCTCCAGCCTCTCACCGCA) (OL24), d(GATCTGCGGTGAGAGG) (OL16), d[(biotin-dX)AA(biotin-dT)TCTTGCGCCTTAAACCAAC] (OL31DB), d(GACACTCTCGAGACATCACCGTCC) (OL25) and d[(biotin-dX)GACACTCTCGAGACATCACCGTCC] (OL25B). Adaptors AD14, AD15 and AD2 are equimolar mixtures of oligonucleotides OL38 and phosphorylated OL46, OL40 and phosphorylated OL47, and OL24 and OL16, respectively. The M13 reverse and M13 -20 primers are: d(AACAGCTATGACCATG) and d(GTAAAACGACGGCCAGT).
Preparation of tracer and driver
To construct the driver, we pooled equal amounts of genomic DNA from each of 15 regA- individuals from the progeny of the cross I regA- male × J regA+ female. The pooled DNA and DNA from the J regA+ female parent were digested with Sau3A, ligated to adaptors AD15 and AD14 respectively, and a portion (60 and 75 ng respectively) of each sample was applied to a 1% agarose gel for electrophoretic sizing. The driver, derived from the size selected and pooled regA- DNA fragments (120-860 bp) and the tracer, derived from the size selected J regA+ parental DNA fragments (250-740 bp), were purified and prepared as described previously (17).
RFLP Subtraction
The sample for the first round of subtractive hybridization contained 10 µg driver, 100 ng tracer, 20 µg yeast tRNA, 2 µg OL31DB and 2 µg OL25. The hybridization was carried out in 4 µl of 20 mM EPPS/1 M NaCl/2 mM EDTA pH 8.0 at 65°C and processed as described previously (17). Samples for subsequent rounds of subtractive hybridization contained 90% of the material from the previous round of subtraction, 10 µg driver, 20 µg yeast tRNA, 2 µg OL31DB and 2 µg OL25. Subtractive hybridizations and removal of poorly hybridizing sequences were performed as previously described (17) with the following minor modifications. The hybridization times were [ge]21 h, which is [sim]5 × Cot1/2 when only the DNA of one of the pooled driver strains is considered. After four rounds of subtractive hybridization, 1 µl of the 50 µl sample was amplified using 20 cycles of PCR with the biotinylated primer OL25B. The sample used in the first step of the procedure to remove poorly hybridizing DNA was allowed to reassociate for 20 h. Restriction enzyme Sau3A was used to digest the reassociation products, which were then ligated to non-phosphorylated adaptor AD2 and subjected to affinity chromatography. The sample used for the second step of the procedure to remove poorly hybridizing DNA contained 1 µl of the sample from the first step (the total volume of which was 25 µl), 10 µg of driver, 20 µg of yeast tRNA and 2 µg of OL31DB.
Analyzing the TRS products
TRS products, digested with Sau3A, were ligated to XhoI digested pBluescript SK- after partially filling in the ends of both insert (dATP, dGTP) and vector (dCTP, TTP) using Klenow fragment of DNA polymerase (New England Biolabs). We amplified the inserts of 24 clones. The insert DNA was amplified and labeled with [[alpha]-32P]dCTP (New England Nuclear) using `M13 -20' and `M13 reverse' primers and Klenow fragment of DNA polymerase. Genomic DNA for Southern blots was isolated by CTAB extraction (30) and passed over QIAprep Plasmid Miniprep columns (Qiagen) prior to digestion with Sau3A. DNA samples (1.5 µg) were electrophoretically sized on a 1% agarose gel and then transferred and UV crosslinked to nylon membranes (Gene Screen, New England Nuclear). Filters were hybridized in phosphate-buffered 7% (w/v) SDS (31) and were washed in 0.2× SSC/0.1% SDS at 65°C. Linkage analysis was performed using MAPMAKER/EXP version 3.0 (32). Markers linked to the regA locus in a test of 16 individuals from the progeny were used to genotype [sim]50 more progeny strains.
To determine the multiplicity of each of the 24 clones in the library of TRS products, we hybridized each clone to replica filters of the TRS library. Vector DNA, flanking the amplified inserts, was removed by digestion with ApaI and AccI followed by electrophoretic separation on an agarose gel.
RESULTS
The TRS strategy
TRS relies on subtractive hybridization, which purifies DNA fragments that are present in one sample (the tracer) but absent in another sample (the driver). Purification of the target fragments that are unique to the tracer is achieved by removing from the tracer all of the sequences that are also present in the driver (Fig. 2A).
Figure 2. Subtractive hybridization and RFLP subtraction strategy. (A) Subtractive hybridization isolates target sequences that are present in one sample (the tracer) but absent in another sample (the driver). Enrichment of the target sequences is achieved by removing from the tracer fragments that are common to both the tracer and the driver. A small amount of tracer is mixed with an excess of biotinylated driver and the mixture is denatured and allowed to reanneal. Most of the tracer strands have biotinylated driver counterparts with which to anneal. Applying avidin affinity chromatography removes these strands. In contrast, the target strands have no biotinylated driver counterparts with which to anneal, and so are not removed by affinity chromatography (36,37). (B) When the tracer and driver are composed of restriction fragments in a limited size range, subtractive hybridization isolates RFLP markers that are distributed throughout an entire genome. TRS is based on RFLP Subtraction, which isolates large numbers of unique sequence DNA markers distributed randomly throughout the genome (17). In RFLP Subtraction, the tracer is a size selected fraction of restriction fragments from one strain and the driver is the corresponding fraction of restriction fragments from a related strain (Fig. 2B). Due to random DNA sequence polymorphism, some restriction sites will be present in one strain but absent in the other. As a result, some of the sequences that lie on fragments in the tracer size fraction are absent from the corresponding driver size fraction. Such RFLPs can therefore be purified by subtractive hybridization. Whereas RFLP Subtraction isolates markers covering the whole genome, TRS is designed to isolate only those RFLPs that are tightly linked to the mutation of interest. Technically, TRS is identical to RFLP subtraction except for the makeup of the driver DNA. The driver sample is constructed by phenotypic pooling so that each tracer fragment has a driver counterpart unless the tracer fragment is an RFLP at the locus of interest. To make the phenotypically pooled driver, we first crossed two polymorphic strains that have distinct genotypes at the locus of interest. We mated polymorphic strains of Volvox differing at the regA locus: a Japanese regA+ strain (J regA+) and an Indian regA- strain (I regA-) (Fig. 3). Some of the haploid progeny individuals obtained were wild-type (regA+) and some mutant (regA-). Note that we know the origin of the DNA at the regA locus in the progeny of the cross shown in Figure 3. The wild-type progeny strains inherit their regA locus from the J regA+ parent and the mutant progeny strains inherit their regA gene from the I regA- parent. Figure 3. Targeted RFLP subtraction strategy for isolating markers linked to regA. To enrich for those RFLP markers located near the regA locus, the driver is constructed from phenotypically pooled F1 progeny of the cross I regA- × J regA+. In the experiment reported here, the tracer was composed of small Sau3A fragments from the Japanese regA+ parent and the driver was made from small Sau3A fragments of 15 pooled regA- F1 progeny. Note that DNA was isolated from asexually reproducing Volvox which are haploid. (After mating, a transient diploid zygote directly undergoes meiosis yielding haploid progeny.) It is important to realize that the RFLPs obtained in this experiment result from random variation in DNA sequence, not from the lesion that caused the regA- mutation. Therefore, we do not expect to isolate an RFLP caused by a point mutation that is responsible for the regA- phenotype, but rather RFLPs that are located near the mutation. After digesting the DNA from the strains shown in Figure 3 there are three kinds of fragments. Most fragments are not polymorphic (white); some fragments are RFLPs that are not linked to regA (gray); and some fragments are RFLPs that are linked to regA (black). The object of TRS is to isolate the latter fragments. Consider the distribution of the three types of fragments in the digested DNA in Figure 3. Non-polymorphic sequences (white) reside on fragments of the same size in both of the parents and in all of the progeny. Polymorphic sequences that are unlinked to regA (gray) represent the bulk of the RFLPs. Whether a progeny strain inherits a large or small allele of an unlinked RFLP is not correlated with the strain's regA phenotype. In contrast, alleles of the RFLPs that are very tightly linked to the regA locus (black) are non-randomly distributed with respect to the regA phenotype of the progeny. Figure 3 shows, for example, a tightly linked RFLP whose small allele (black) occurs in the regA+ parent and whose large allele occurs in the regA- parent. The small allele will be inherited by the regA+ progeny and the large allele will be inherited by the regA- progeny since the RFLP is tightly linked to the regA locus. We constructed a driver sample that lacks the class of RFLPs that are represented by the small black fragment in Figure 3 by size selecting restriction fragments from pooled regA- progeny DNA. For the tracer we used the similarly sized fragments from the J regA+ parent. Except for RFLPs tightly linked to regA, the phenotypically pooled driver contains all of the sequences that occur in the tracer. Thus, applying subtractive hybridization, which removes from the tracer all of the fragments common to the driver, should purify the small alleles of the linked RFLPs (black bars in Fig. 3). In the TRS experiment described below the tracer was derived from the short Sau3A fragments (250-740 bp) of the J regA+ parent and the driver was a mixture of the short fragments from 15 of the phenotypically pooled regA- progeny of the cross I regA- × J regA+. The number of strains pooled affects both the size of the targeted region (the more strains pooled the smaller the expected size of the region) and the required reassociation time for subtractive hybridization (the more strains pooled the longer the incubation time required). Our strategy in choosing a pool size was to use the largest number of strains that would still be compatable with a practical reassociation incubation time ([sim]1 day).
Analyzing the products of TRS
To determine how efficiently TRS enriched for RFLP markers linked to the regA locus, we cloned the TRS products. We analyzed 24 randomly picked clones from a library containing [sim]3130 cloned TRS products.
We determined how many of the 24 randomly picked TRS products are RFLPs and how many are linked to regA by probing genomic Southern blots. Figure 4 shows a representative blot in which TRS clone H was labeled and hybridized to a filter containing Sau3A digested genomic DNA from progeny of the cross I regA- × J regA+. The pattern of hybridization indicates that clone H contains a fragment with the desired characteristics for a product of TRS. First, the insert sequence corresponds to an RFLP of the expected type: the sequence resides on a large Sau3A fragment (outside of the tracer range) in the I regA- parent and on a small fragment (inside the tracer range) in the J regA+ parent. Second, marker H is tightly linked to the regA locus (it cosegregates with the regA phenotype in all of the 67 progeny that we have tested).
Figure 4. A representative genomic Southern blot showing the cosegregation of regA and a TRS product. The probe, trsH, is one of the 24 randomly picked cloned TRS products that we analyzed. Each of the 24 TRS products were used to probe Sau3A digested genomic DNA from asexually grown (haploid) parents and progeny of the cross I regA- × J regA+. Probe trsH cosegregates with the regA locus in all 67 progeny analyzed. Table 1. 
No. clones in sample
Distance (d) from regA (cM)a
Estimated no. similar sequences in libraryb
3
d = 0
130 ± 70
2
0 > d < 21
90 ± 60
8
d > 21
340 ± 100
13
total RFLPs
560 ± 100
11
not RFLPs
470 ± 100
Estimated no. distinct sequences in library
1030c
where mi is the multiplicity, that is the number of clones in the library that hybridize to probe made from the ith clone in the sample (17). For example, trsH, trsJ and trsU (the 3 clones in the sample that are the closest to the regA locus) hybridized to 4, 18 and 12 clones in the library, respectively.
Table 1 summarizes the results obtained by probing Southern blots with randomly picked clones from the TRS library. Of the 24 clones tested, 13 were RFLPs in the expected size range.
Genetic analysis of the regA locus
Three of the 13 RFLP markers cosegregate with regA in all of the progeny we tested (n [le] 67). Using a multipoint mapping strategy (32) we ordered the 13 RFLP markers in the sample of 24 clones (Table 1). Figure 5 displays the map positions of the five markers in our sample that map to within 21 map units of regA. The markers in our small sample appear to be non-randomly distributed with a clustering at the regA locus. The clustering effect may reflect the size of the window determined by the recombination breakpoints lying closest to the regA locus in the pooled progeny.
Figure 5. Genetic map of the regA locus and linked TRS products. Map of the five cloned TRS products (of the 24 randomly picked clones analyzed) that map nearest to the regA locus. The clones were mapped by probing Southern blots of progeny of the cross I regA- × J regA+. The maximum likelihood map was calculated by the program Mapmaker (32) using the default values for error detection. The markers listed directly under regA cosegregated with the regA locus in all of the progeny tested. The number of progeny scored for cosegregation with the regA locus for markers H, J, U, O and W was 67, 67, 66, 59 and 59 respectively. Our Southern blot analysis revealed that one of the regA+ individuals from the progeny of the cross I regA- × J regA+ carried both alleles of markers linked to regA (labeled in Fig. 4). This biallelic strain apparently carried a stable duplication of the regA locus. Markers surrounding the regA locus are biallelic for a distance of at least 29 cM on one side of the locus and for a distance of 23 cM on the other side of the regA locus, after which only the Indian allele is observed for the two markers tested (J. Corrette-Bennett and D. Straus, unpublished results). These results are consistent with the notion that the spontaneous regA mutation in the Indian strain is recessive. To our knowledge this is the first indication of the mode of inheritance of a regA- allele. The strain, unfortunately, has been lost, although there did not appear to be a strong selection against it, as both alleles were present at comparable levels for scores of generations. 
Estimating number of clones in the TRS library that are tightly linked to the regA locus
To estimate the number of markers in the library, we probed replica filters containing [sim]3130 insert containing colonies with each of the 24 TRS products. All 24 probes showed unique non-overlapping hybridization patterns. Thus, the 24 randomly picked clones contain 24 different insert sequences. Based on the number of times each sequence is represented in the library, we estimate the total number of distinct clones in the library to be [sim]1030 (Table 1). Assuming our sample of 24 clones is representative of the whole library, we estimate that the library contains [sim]560 ± 100 distinct markers of which 130 ± 70 are expected to cosegregate with the regA locus in all 67 progeny (Table 1). We expect that 60 ± 30 (48%) of such markers lie within 1 cM of the regA locus, assuming that the distribution of completely concordant TRS product markers is the same as the predicted distribution of completely concordant markers in a series of 67 crosses. These estimates are crude due to the small number of clones that we analyzed. Nevertheless, we conclude that our library of TRS products contains numerous markers that are tightly linked to the regA gene.
Density of recovered markers that are tightly linked to the regA locus
Determining whether we have isolated markers at a physical density high enough to achieve chromosome landing will require analysis of the region covered by the markers. However, it is useful to make a rough preliminary estimate of the density of the markers using knowledge of the Volvox genome. The physical size of an average map unit in Volvox is [sim]100 kb, based on a preliminary map of RAPD markers (1300 cM; E. Ananiev and H. Wu, unpublished) and a genome size of 1.2 × 108 bp (33). Assuming that a map unit at the regA locus is of average size and that our library contains 60 ± 30 markers within 1 cM of the regA locus (covering 2 cM total), we would expect a density >15 markers per 100 kb.
However, since recombination frequency can vary greatly with position in the genome the actual number of markers recovered within a 100 kb region at the regA locus may be substantially different from the above estimate. For example, if 1 cM around the regA locus covers 1 Mb, the physical density of the recovered markers would be expected to be more than one marker per 100 kb.
Chromosome landing requires a density of recovered markers in a region that is high enough to identify overlapping genomic clones. To identify overlapping BAC or YAC clones, densities greater than one marker per 100 kb would suffice. Thus, we conclude that even if 1 cM at the regA locus covers a region of 1 Mb (10 times the size of an average cM in the Volvox genome) the density of the markers we recovered would be high enough to achieve chromosome landing. We emphasize, however, that demonstration of chromosome landing awaits physical analysis of the region surrounding the regA gene.
Figure 6. Linkage of markers to the regA locus is not an artifact of the Volvox Sau3A modification system. Many Sau3A recognition sites are not cut by Sau3A in Volvox (34). To test whether the linkage of the TRS markers (which are Sau3A fragments) to the regA locus is an artifact of DNA modification, we probed Southern blots containing genomic DNA from a subset of progeny in our mapping population that had crossovers close to the regA locus. DNA was cut with AluI, an enzyme that cuts at all restriction sites tested in Volvox (34). The probes were fragments derived from cosmids harboring the tightly linked markers trsU (shown here) and trsH. Polymorphic fragments cosegregating with the regA locus in all 17 progeny tested were detected by probes from both cosmids. Thus, the linkage of the markers to the regA locus is not a byproduct of the modification system.
Does a Volvox DNA modification system affect the nature of the TRS markers?
Since performing the initial experiments, which isolated Sau3A markers, we learned that many Sau3A sites are resistant to digestion in the Japanese strain of Volvox, perhaps due to differential DNA modification (34). How could differential modification affect the TRS subtraction experiment? Had we used an enzyme that cut at all recognition sites we might have recovered markers at an even higher density, since the number of markers is proportional to the number of fragments. However, we were concerned that differential modification might also result in artifactual linkage of the TRS markers to regA.
To determine whether linkage of our markers to the regA locus is independent of a Sau3A modification system, we mapped fragments that are physically linked to the TRS markers, but that are the products of digestion with AluI, an enzyme that appears to cut at all recognition sites in Volvox DNA(34). We used fragments of the cosmid clones containing markers trsH and trsU to hybridize to the AluI digested genomic DNA of 17 individuals from the mapping population that had crossovers close to the regA locus. Our Southern blot analysis indicates that AluI fragments from both cosmids map to the regA locus (Fig. 6). Thus, we conclude that our regA-linked Sau3A markers are not an artifact of a Volvox modification system.
DISCUSSION
Using TRS we efficiently isolated markers at a very high genetic density tightly linked to regA, a Volvox developmental gene defined solely by a mutant phenotype. Furthermore, we isolated the markers without the aid of a genetic map. Thus, the TRS method represents progress towards a generally applicable chromosome landing strategy.
We estimate that TRS isolated 60 ± 30 markers lying within 1 cM of the regA locus. Testing whether this genetic density corresponds to a physical density high enough to achieve chromosome landing will require construction and probing of large insert Volvox genomic libraries (e.g., YAC or BAC). To find the regA gene, genomic clones could be tested by complementation analysis using newly developed transformation methods for Volvox (35).
The high yield of region specific markers achieved by TRS compared with the GDRDA method (13-16) may be due to a combination of factors including the high degree of nucleotide polymorphism between our Volvox strains (estimated as >1% in ref. 16, and D. Straus, unpublished), use of fragments defined by 4 bp recognition sites (rather than 6 bp sites), sampling a large fraction of the genome, and minimization of the number of PCR and reassociation steps (during which desired products can be competitively eliminated). Other advantages of TRS include the relative simplicity of our methodology, efficient removal of repetitive sequences, and the versatility and precision of using gel electrophoretic size selection rather than PCR based size selection.
The AFLP method, combined with phenotypic pooling, also isolates very dense region specific markers (11). Whether direct isolation of linked markers using TRS or indirect screening by AFLP analysis is most useful may be reduced to a question of which method is more time and cost efficient for a laboratory. The AFLP analysis carried out by Thomas et al. required substantial preliminary genetic analysis of the region of interest, [sim]60 oligonucleotide primers, >700 analytical PCR reactions, dozens of sequencing gels, and careful analysis of [sim]40 000 sets of bands. If a small laboratory aims to analyze several genes, both labor and cost increase steeply. In contrast, the labor involved in TRS is small (the subtractions themselves require one person's labor over a period of <2 weeks) and the cost of materials for additional experiments are minimal. One significant advantage of the AFLP method over TRS is that it requires less molecular biological expertise.
Our future plans include application of TRS to organisms with complex genomes and optimization of the method to minimize the number of undesired clones in the products. The phenotypic pooling scheme we used can be simply modified for application of the method to diploid organisms. We expect TRS to work when applied to a complex genome since RFLP subtraction, the untargeted version of TRS, efficiently isolated mouse RFLPs (17). In contrast to our earlier work using RFLP subtraction, the TRS subtraction library contained, in addition to the desired products, some undesired clones containing non-polymorphic sequences and some unlinked markers. The ratio of desired to undesired products might be improved by making simple modifications such as increasing the ratio of driver to tracer (so that the DNA of each strain in the pooled driver is in large excess over the tracer) and carrying the reassociation reactions to higher Cot values.
TRS shows promise as a generally applicable method for saturating a genomic region of interest with DNA markers. By circumventing the slow and expensive process of generating high resolution genetic maps of a whole genome, our method could facilitate positional cloning in diverse organisms. TRS may prove useful for isolating genes corresponding to human diseases (using pedigrees showing linkage disequilibrium, for example) and could be used to aid genome projects by providing overlapping genomic clones from a region of interest for sequence analysis.
ACKNOWLEDGEMENTS
We thank an anonymous referee and his or her postdoctoral fellow for valuable editorial suggestions. This work was supported by grants to D.S. from the National Science Foundation.
REFERENCES
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 24 Mar 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Li, F. Wang, V. Zabarovska, C. Wahlestedt, and E. R. Zabarovsky Cloning of polymorphisms (COP): enrichment of polymorphic sequences from complex genomes Nucleic Acids Res., January 15, 2000; 28(2): e1 - e1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||