Introduction Direct analysis of the genetic information of an organism will provide an invaluable data base for exploring the biology of that organism. A part of this analysis is the establishment of a high resolution genetic map. The mouse is an attractive system since it is possible to observe many genetic markers with high cfficiency in many individuals. The genetic map of the mouse is currently approaching a resolution of an average of 3cM (approximately 6 Mbp)‘13’). An increase of resolution to 1cM would be sufficient to allow the integration of the genetic map with physical mapping and cloning techniques. Genetic mapping to this resolution directly complements restriction mapping by pulsed field gel electrophoresis (PFGE) and yeast artificial chromosome cloning. The advantages of this level of resolution are: 1) rapid mapping and cloning of genes identified by phenotype; 2) correlation of genome structure with genetic structure; 3) correlation of cloned genes with mouse phenotypes; and 4) further definition of regions of synteny between mouse and human which will allow mouse models for human diseases to be constructed. Meiotic Mapping The requirements for high resolution genetic mapping are probes for many loci, many DNA sources and a high degree of polymorphism. The latter two requirements are met by meiotic recombination analysis in the mouse, using largc backcrosses between distantly related animals. The resolution depends on the numbcr of mice used to construct the backcross and the number Interspecific panels of probes already mapped on it(233). arc the crosses of choice, rather than recombinant inbred strains, because of thc higher degree of polymorphism found between different species of mice(2,‘) . We have exploited a mapping panel constructed between two different species of mice, Mus spretus and C57BL/6. The cross used was C57BL/6xMus sprerus) FI femalex CS7BL/6 male(4). Conventionally, in this system. Mus spretus-specific restriction fragment length variants (RFLVs) are followed in backcross individuals and mapped using data accumulated on the inheritance of other loci in the panel. The panel is typed
initially using markers whose chromosomal location is accurately known: additional markers are then typed relative to these. gradually building up a detailed map of the recornbination events that have occurred in each markers are mapped, gene order r n e i o s i ~ ( ~ As . ~ )more . can be determined by comparing new segregation patterns to those already known and minimizing the numbcr of double or multiple crossovers required to explain the new probe distribution (‘Pedigree analysis’(2’)). In thesc experiments, Southern blots of reqtriction digcsts of progenitor DNAs are screened with DNA probes in order to identify RFLVs. DNA from individuals in the panel are then probed, and the pattern of inheritance of the restriction fragment variant is scored. The principles of this technique are illustrated in Fig. 1. Because repeated screening of Southern blots (which can be reused but cventually have to be replaced) is necessary, the finite supply of
Mus spretus
X
C57BL16
TI T i IT i i \1 X
F1
N2
N2
\1
C57BL/6
N2
li Pig. 1. Interspecific backcross mapping. The schematic shows a pair of chromosomes from Mus TpYetiis (a wild mouse species that has been inbred) and C57RL/6 (an estensively inbred laboratory mouse strain). Each chromosome pair from the F1 of this cross contains one chromosome of Mirs .spYetin origin and one o f CS7BL/6 origin. Recombination during meiosis can occur between these chromos o m a giving new combinations of il/lci.~ryrcfiu and CS7BL/6 alleles. Backcrossing onto CS7BL,/6puts these recornhinant chromosomes on an homogeneous background where Mzu spretus allcles can e d d y be followed in N2 individuals.
I
I
I
I
Primer A
c Primer B
Primer A
5_(( W y
L-
Strain A DNA
Strain B DNA
Primer B
PCR of genomic DNA, size separation
---+ - - Strain A
Strain B
Recombinant inbred strains
(CA'y Allele (CA)n
Fig. 2. Schematic of PCR-analysed microsatcllites. Unique sequence primers (A and R) are synthesized to allow PCK across the simple (shown as CA here) repeat (opcn boxes). Polymorphisins in the length of these repeats is detected on agarose or polydcryhmide gels (heavy bars at bottom of figure). By analysing the PCR products generated from recombinant inbred strains or interspecific backcross animals these polymorphisms can bc mapped (see text).
D N A (initially taken from organs such as spleen and thymus, although other tissues are stored and used) for each backcross individual is eventually exhausted and no further probes can be mapped. The life-span of this resource can be conserved by dividing the panel into different levels of resolution. A probe is mapped to a particular chromosome or subchromosomal region using a small number of individuals and, then, finer mapping is carried out, using only those individuals informative for that region. The polymerase chain reaction (PCR) has high sensitivity and robustness, requiring very little DNA (50 ng or less) but generating a lot of product, which can be easily detected. It also has the advantage that it can be used to generate probes directly. For these reasons it has been incorporated into a number of meiotic mapping protocols, three of which are outlined below. Love ct aZ.(') have developed a meiotic mapping method based on PCR across dinucleotide repeats (Fig. 2). This exploits the high degree of repeat length polymorphism found between strains of mice. In this system the sequence around such repeats is determined and unique PCR primers are made. Products can then be mappcd as outlined above. The main limitation of this method is the requirement for sequence information so that primers can be designed. However, the method is amenable to multiplexing, since, with some optimization, it should be possible to carry out PCR on many loci in one reaction and analyse these on one gel. A second approach called Random Amplified Polymorphic D N A (RAPD) has recently been reported In this method the 'need-to-know' by Williams et sequence problem is avoided by randomly synthesizing oligonucleotide 10mers, each of which is used to carry out PCR on genomic DNA. Each primer generates a limited number of products, which are present in some individuals but not in others, depending on the
presence or absence of the appropriate primer sequence in the genome. It has been shown that these RAPDs are inherited in a Mendelian fashion in soyabean. This method remains to be evaluated for its usefulness in mouse genome mapping but ha5 great potential, although it will require synthesizing many primers and carrying out many PCR reactions per backcross individual. As a third approach, intcrspersed repetitive element PCR (IRS-PCR)".') has the advantage that it can be used to amplify unique sequence DNA without the need to know part of that sequence. IRS-PCR is carried out using PCR primers homologous to repetitive elements; since these elements are distributed throughout the genome it is possible to generate multiple probes at random. The number of probes generated in a single PCR reaction far exceeds that obtained by the previous two methods. These probes contain flanking repetitive element D N A and varying amounts of unique sequence. The human Alu repetitive element has been used for IRS-PCR from somatic cell hybrids, since under appropriate conditions these A h primers are species specific and are very common (close to one million copies)(7). The most common mouse interspersed repetitive elements are short interspersed nuclear elements (SINES). B1 (130000-180000 copies per genome), B2 (80 000-120 000 copies) and evolutionary conserved (EC) sequences (100000 copies)- and the long interspersed nuclear element (LINE) L1 (70000-100 000 copies) (rcviewed in ref. 9). The LINE element L1 is frequently and variably truncated at its 5' end, resulting in a lower copy number for the 5' end (10-12000) compared to the 3' end (1000r)O). It is therefore preferable to use PCR primers from the 3' end and that are orientated 3' out of the repeat. We have worked mainly with B1, B2 and L1, which are all thought to be generated by a mechanism involving an RNA intermediate. In the case of B1 and B2, these are probably the 7SL RNA and tRNA, respectively. Because of the retrotransposon nature of these repeats. we reasoned that their position in the genome would vary between species of mice, reflecting their time of insertion during evolution and their evolution after insertion. Such polymorphism of Alu IRS-PCR products has been reported between human individuals("). A number of possible outcomes can be predicted when one carries out PCR on different species of mice. Considering a particular unique sequence flanked by repetitive elements, then, when a single PCR primer is used. a product will be observed. provided the elements are close enough together and are in the appropriate orientations relative to one other. If in one species a primer is missing, there will be no product: altcrnatively, when there is another priming site near, a larger product will be observed. The products that we have mapped so far fall into these categories. Additionally. when one of the primer target sequences is sufficiently mutated, binding would either be eliminated
, Sequence B
,Sequence A
c
t
, Sequence A
l6
,
Sequence B
PC R on genomic DNA using IRS primers; Resolve on agarose gels
C57BL16
Mus spretus
-
-
Backcross progeny
-
A
-
B
or occur with lower affinity. giving rise to quantitative differences in the PCR products. (See Fig. 3 for schematic illustration). We prepared primers to B1, B2 and L1 repeats and used them singly to carry out PCR on 50ng of high molecular weight genomic DNA. It was not necessary to use combinations of primers in these experiments since the products generated were sufficiently complex for our immediate purpose. Additionally, we have found that combinations of primers from different repeat families result in smaller products, as one might predict from the combined frequencies. All of these primers generate complex banding patterns when resolved on agarose gels and stained with ethidium bromide. The L1 pattern is the least complex and many differences can be seen between the products from two different species of mice (illustrated schematically in Fig. 4). The Mus ,hpvekds specific bands can bc followed in backcross animals (see above), and have allowed us to assign bands A, B, C, D and E to mouse chromosomes 3, 2, 1, 12 and 2, respectively('). We were also able to map them relative to other genes mapped in the paneli'). The B2 and B1 products were too complex to $core for variants between the two species, although differences in the patterns of ethidium bromide stained bands on agarose gels were clearly visible. Tn order to simplify the pattern we blotted gels of the products and probed them with simple 12mer oligonucleotide repeats. Since it has been found empirically that some 12mers are present in up to 10 '% of clones in cosmid libraries from mammalian genomes("), we expected that that some of these sequences would have been amplified using our primers and that the patterns revealed using these sequences as probes would be much less complex than those defined by looking at all of the bands at once. Using oligonucleotides (GATA)3, (CC'IT)3 and (GACA)3 as probes we were able to identify a number of Mus spretuJ-specific polymorphisms in B1 and B2 products and map these to subchromosomal regions('). In addition a number of common bands were found.
Fig. 3. Principle of IRS-PCR and its application to genome mapping. Two wrts of variant are illustrated. variation in product sizc (sequence A ) and presence or abscnce (sequence B) of a product due to one of the primer sequenccs being inserted after divergence of the two species. Open boxes represent repetitivc elements, primers are represented as small arrows above the boxes and the products resolved on agarosc gels as heavy bars.
In total. we have mapped 13 new loci to 9 different chromosomes('). We have demonstrated the potential of this method for generating and simultaneously mapping inter-repeat PCR products.
Prospects for Meiotic Mapping New devclopments improving the capabilities in meiotic mapping can be expected in two directions. Using the interspecific backcross described above, only Mus sprelus variants can be mapped, but mapping panels are now being constructed in which the backcross is to Mus spretus (for example [(Mus spretus x C57BL/6) F1 x Mus spretus]) which will allow the segregation of Mus dornesricus variants to be followed. These reverse backcrosses will be useful for mapping PCR products from somatic cell hybrids. which usually contain chromosomes from inbred strains rather than wild mice. The other area of improvement to be expected will involve further simplifications of the protocols of PCRbased mapping, with the goal of maximising the number of loci and the number of meiotic evcnts to be analysed. Such an increase in the power of analysis can be achieved by direct scoring of PCR products on, for instancc, one or two dimensional gel systems, by indirect scoring of PCR products which can be either spotted on filters, or again separated by one or two dimensional gels, or by 'reverse scoring', in which the PCR products are labelled and used to probe an array of oligonucleotides on a filter surface.
One-dimensional gels In order to score directly a large number of loci using one-dimensional gels wc would have to carry out PCR reactions that generate only the number of bands (for example by RAPD, see above) which can easily be resolved. Consequently we would have to run many gels. For example. to score 1000 loci wc would need SO
Segregation of Variants
Gel Analysis of PCR Products
-
-+ I-
- IE
1
3
2
A
B
C
D
E
1
S
B
B
B
B
2
B
S
B
S
S
3
B
B
S
S
B
4
S
B
B
B
S
I
12
t' i f1 B
IB
I
A
E
C
t
Analysis
PCR reactions per backcross individual. each generating 20 loci markers, resolved on 50 gels.
Fig. 4. Mapping of LINE IRS-PCR products, adapted from actual data"'. Products are rcsolved on agarose gels, stained with ethidiuin bromide and the Miis speruv (S) specific bands scored in interspecific backcross (IB) animals (1.2,3,3, etc.) The data scored as eithcr MZLY ,spwfur (S) or C57BLl6 (B) alleles are entered into the database containing data on the inheritance of other probes mappcd i n the panel. Chrornowmal and subchromosomal assignments and genetic distances are thus determined.
Two-dimensional gels The number of gel5 can be reduced if the resolution of the gel system is increased. Optimization of the PCR reactions to increase the number of products in line with the resolution of the gel system reduces the number of individual PCR reactions required. Higher resolution is offered by two-dimensional gels, in which the first dimension is a size separation and the second depends on the sequence of a marker and its behaviour in a denaturing gradient (see ref. 12). Tf. for example. one assumes that 200 sequences can be resolved on a twodimensional gel (well within the capacity of these systems('')), then. to score 1000 loci, one would have to ruii five gels.
the form of IRS-PCR products). Using two-dimensional gels, the minimum number of oligonucleotide probings is further reduced. For example, assuming 200 loci can be separated, then 5 probes, each detecting 200 loci, would be needed to score 1000 loci on one gel. Obviously in practice more probes would be required but the relative economies of the alternatives are apparent from these simple comparisons. This indirect approach could also be applied to PCR products that have been dot blotted. The minimum number of probe oligonucleotides required is a function of the number of loci present in each dot. This is not very efficient, since to reduce the number of oligonucleotide probings by decreasing the number of loci per spot and increasing the number of spots pcr individual, one has to increase the number of PCR rcactions per individual.
Indirect detection methods Further economies of scale should be possible using indirect detection methods. We have describcd above the use of one-dimensional gels coupled with indirect scoring of markers by probing Southern blots with oligonucleotides. With this level of resolution we would have to use a large number of oligonucleotide probes. For example, to score 1000 loci we would nced a minimum of SO oligonucleotide probes, each detecting 20 loci (the resolution of a gel). Fewer gcls would be rcquired than for the direct approach, since each individual represents one track on a gel (for example in
Reverse detection However, these approaches can be reversed so that the oligonucleotide probes or a library of cloned DNA is immobilised on a solid support, in a high density These are probed with the PCR products themselves. In this case, a signal in the array generated by a probe from one species but not by a probe from another can be followed and mapped in backcross individuals. A mapped PCR product can be recovered using the detected oligonucleotide as a probe. If cloned DNA was used (in the form of bacterial colonies for example), it can be simply recovered from frozen
cell stocks used to make the array. This approach is very efficient requiring one PCR reaction per individual (using IRS-PCR, more using other methods) and one hybridisation per individual, the total number of hybridisations depending on the number of backcross individuals scored. Using oligonucleotide hybridisation techniques (above and refs 13.14) we should be able to 'fingerprint' the PCR products and provide partial sequence information. It is possible that the density of information generated in this way will allow the detection of changes (mutations) in the genome, so that, for example, tumour cells could be compared with their normal counterparts. The amount of sequence information obtained depends on the number of oligonucleotide permutations used(13,14).A full discussion of sequencing b oligonucleotide h bridisation is provided by DrmanacJ') and Lehrach(14? Somatic Cell Genetics Somatic cell hybrids containing whole mouse chromosomes or irradiation fragments(") are important mapping tools, providing levels of resolution between whole chroniosomes and analysis by pulsed field gels. The use of Alu TRS-PCR to generate multiple probes from human DNA present in somatic cell hybrids has been d e ~ c r i b e d ( ~This * ~ ) . is a very powerful technique that has simplified and facilitated the characterisation and exploitation of somatic cell hybrids. It would, therefore, be useful if mouse specific IRS-PCR primers could be found. Unfortunately, many of the somatic cell hybrids currently available are on hamster backgrounds, which. because of the shortcr evolutionary distance between mouse and hamster, makes it more difficult to find species-specific primers. Added to this is the problem that these repeats are less common than their human equivalents, decreasing the density of probes that can be generated. Careful selection of primers and conditions has solved the former problem and the latter problem can be mitigated by combination of rimers from different repetitive element families(P6-18).We have identified a primer sequence from the B2 family that is divergent between mouse and h a m ~ t e d ' ~ .Using ' ~ ~ , this primer it is possible to amplify muuse but not hamster, dog or human DNA. Conditions have been optimised to generate probes from irradiation hybrids that contain relatively low amounts of mouse DNA. These experiments increase the number of products amplified by using the B2 primer in conjunction with a species-specific 3' L1 primer (ref. 17, 18 see also 16). We can now use IRS-PCR to fingerprint hybrids containing similar amounts of DNA(17),and to generate probes from a subset of the DNA present in hybrids. This allows further characterisation of the DNA content of a hybrid as well as generating new probes for mapping cxperiments. IRS-PCR can be used to couple somatic cell genetics
with physical mapping and cloning methods. Pools of PCR products from somatic cell hybrids and irradiation hybrids can be used as probes (following compctition with repetitive DNA) to screen Yeast Artificial Chromosome (YAC) libraries (pre-hybridised with genomic DNA to suppress signal due to repetitive DNA), in order to recover YACs from limited or individual chromosome regions("). In this way, contiguous YAC clones can be assemblcd for megabase regions of DNA. Cloning of large regions of mouse DNA in YAC libraries(") offers a level of resolution in the range of PFGE analysis and increases the distance which can be covered by each IRS-PCR product. This latter point is important because of the frequency of repeats and the limitation on the size of PCR products (see earlier), which might make it impossible (using this particular approach) to make contiguous clones using cosmids for example. Finer mapping at the level of cosmids is then possible. using YAC clones. IRS-PCR. using species-specific primers, can be used to generate probes from individual pulsed-field gel fragments(17).Fragments arc cut out of a pulsed field gel and DNA extracted for PCR; at this stage individual slices can be screened by dot blotting to identify fractions carrying particular fragments for which a probe already exists(17). Probes generated in this way can be used for YAC recovery and map extension. To summarise, we now have a range of PCR-based mapping tools which, coupled with the power of meiotic mapping, somatic cell genetics, and physical mapping will permit a high resolution genetic map of the mouse genome to be constructed in the shortest possible time. Application of the techniques developed for genome sequencing may allow even higher resolution mapping. In the long run, these techniques may also offer efficient ways to analyse human populations. References 1 Cox. R. D., COBEMND, N . , JEYKI'IS.N. A. .\ND LElIK4CH. H. (1991). Intcrsperaed rcpetitive clement pnlymerasc chain reaction product mapping using it mouie interspccific backcrosh. Genomrcs, in press. 2 COPELANT). N . G. A N D JENKIYS, N. A. (1991). DCVehpmcilt and applicalions of a molecular genetic linkage map of the mouse genomc. TIG, in press. 3 AVUER. P.. AXAR, L.. DANDOLO, L. AND GIXNET. J. L. (1988). Genctic analyiis of thc mouse using interspecific crosseb. TIG 1>18-23. 4 BVCIIHERG, A. M.. BFDIGIAN, H. C; .. T 4 Y l . O R . B. A., BROWNF.I.L, E., IHI E, J . NAC~ATA, S . . JEVKIXS. N. A . AND C n P E L a u , N. G. (198R). Localicaiion of -2to chromosome 11: Linkage to other proto-oncogene and growth factor loci using interspccific backcruss mice. Oncoq?nr~Keseard7 2. 149-165. 5 LOVE.J. M.. K N rCiii-. A . M.. MCALEEK.M. AND TODD, J. (1990). 'Towardconstructiun of a high resolution map of the ninuse genome using PCR-analysed microsatcllites. ,Vucl. Acids lies. 18, 4123-4130, 6 WIiLI4MS. J . G. K.. KUBLLIK.A. N., LIVAK.K. J., K.1F.4LSKI. J. A. A Y U T i m w , S. V. (19990). DNA polyrnoi-phisms amplificd by arbitrary primers are uschil as genetic markers. iVcrcZ. Acids Rrs. 18, 6511-6535. 7 "&SON, D. L. . LEDBLITER. s. A , , CORIO. L., VILIORIA. M . F.. RAMIKEZSOLIS. R., WEBBSTER. ?'. D.. LEDBFlTER, D. H. ANY CASKFY, C. T. (1989). A h polymerase chain reaction: A method lor rapid isolation of human-specific sequences from complex DNA sources. Proc. Nnrt Acnd. Sri. L:SA 86, 66864690. 8 L F D B L ~ ES.RA, , . NELSON,D. L., WARREU, S. T. A Y D L F D B E ~ ED. R , H. (1990). Rapid isolation of DNA probes within specific chromosome regions by interspersed rcpetitive sequencc polymerase chain reaction. Genornics 6 , 475-481. 9 HASTIE, N . D. (1989). Highly rcpeated DNA familics in the genome of ,Vfrrs
rnusrirli~r.In Genetic Ljariants and srrains of the i o t i ~ ~ r a ~mouse o r y (M. F. Lyon and A . G. Searlc, Ed?.). pp. 559-571. OUP, Oxford. 10 SIUUFT.D.. DEMGON. J-M.. SIMARD,L. R. AND LABUDA. D . (19YO). Alumorphs - Human DKA polymorphisms detected by polymerase chain reaction using Ah-specific primers. Genomics 7. 331-334. J. D., ZEHTKEK. G . A N D LLHR4CH, 11 CRZIG. A. G , , N l Z L I l C . D., H~HEISEI H. (ISVO). Ordering of cosmid cloncs covering the herpes simplex virus typc 1 (HSIr-I) gcnome: a teat case for fingerprinting b y hybridisation. Nitci. Acids R E L 18, 2653-2660. 12 LJTTERLINTJEN,A. (3.. SLAGBOOM, P. E.. KKOOK, D . L. AN^ VIJG..I. (198Y). Two-dimensional DNA fingerprinting of human individuals. Proc. Nat1 Acad. Sci. I:SA 86, 274-2746. 13 D~OJANAC. R., LLSNON. ti., DRVANAC, s.. L A B ~ T1.., CRKVtNJAKOV. K.AYD LFHRACH, H. (1990). Partial sequencing by oligo-hybridisation: Conccpt and application5 in genome analysis. In Elecrropkoresis, supercompiifing and the hrtmnn gr'nome, Proceedings of the First International Conference. Tallahassee, Floiidrc, 10-13 April 1990 (C. K. Cantor and H . A. Lin, Eds.). World Scientific, Singapore. in press. 14 LEHRACH, H .. DRMANAC, R.,HOHFISEL.J . . LAWIN,z., LENNOX.G,. NIZLTIC,D.. MONACO,T.. ZEHEINER.G. AND POCSTKA.A . (1SYl). Hybridiyation lingcrprinting in genome mapping and sequencing. In Genome Analysis I : Gen& end Phyricc~lMopping. pp. 39-81, Cold Spring Harbor Laboratory Pr es . i n p r e s . 15 Cox, D. R., BURMEIST~.K, M., ROYDON PKICL, E., KIM,S. A ~ MYERS, D K.M. (190). Radiation hybrid mappmg: A somatic cell genetic method for constructing high-resolution maps of mammalian chromosomes, Science 250, 245-250.
16 RASTAN, 5. (1990). Czech mouse. TIC; 6. 233-236. 17 Cox. R. D.. SIURBS.L. AND LFHRACII. H. (1991). Mouse inter-repeat polymerase chain reaction: probe generation from somatic cell hybrids. (manuscript in preparation). 18 SIMMLFR, M-C.. Cox. R. D. APID A r ~ t a .P. (1991). Adaption of the intersperscd repetitive sequence polynierabe chain rcaction to thi: isolation of mowc DNA probes from somatic cell hy-hrids on a hamster background. Gmonrics, in pres?. 19 KRAYLV, A. S . , kfAXhCSHEVA. T. V., KR4hfEROV: D . A , . RYSKOV. A . P.. SKYRARIN, K. G., B A Y E A ~ ,. A. AND G ~ O R G I EGV. ,P. (1982). Uhiquitou5 transposnn-like repeats BI and B2 of the mouse gcnome: B2 sequencing. Nucl. Acids Res. 10. 1461-1415. 20 MONACO, A. P.. LAM.V . , IXNNON.(3.. DOUGLAS. C.. Z E H E T N ~G.. K. NIZETIC. D . ~OODFEL1,IJW.P. N. AND LEHIUCH, H. (1991). Direct hybridization of Ah-PCK products from irradiation hybrids to cosmid and yeast artificial chromosome lihrarie,. (nianuacript submitted). 21 L>\RIN.2 . AND LEHRACH. H. (1990). Yeasl artificial chromosomes: an alternative approach to the molecular analysis of mciusc mutations. Genet. Rea. 56. 203-208. ~
Roger D. Cox and Hans Lehrach are at the Genome Analysis Laboratory, Imperial Cancer Research Fund. PO Box 123, Lincoln's Inn Fields, London WC1 3PX, UK.
initially using markers whose chromosomal location is accurately known: additional markers are then typed relative to these. gradually building up a detailed map of the recornbination events that have occurred in each markers are mapped, gene order r n e i o s i ~ ( ~ As . ~ )more . can be determined by comparing new segregation patterns to those already known and minimizing the numbcr of double or multiple crossovers required to explain the new probe distribution (‘Pedigree analysis’(2’)). In thesc experiments, Southern blots of reqtriction digcsts of progenitor DNAs are screened with DNA probes in order to identify RFLVs. DNA from individuals in the panel are then probed, and the pattern of inheritance of the restriction fragment variant is scored. The principles of this technique are illustrated in Fig. 1. Because repeated screening of Southern blots (which can be reused but cventually have to be replaced) is necessary, the finite supply of
Mus spretus
X
C57BL16
TI T i IT i i \1 X
F1
N2
N2
\1
C57BL/6
N2
li Pig. 1. Interspecific backcross mapping. The schematic shows a pair of chromosomes from Mus TpYetiis (a wild mouse species that has been inbred) and C57RL/6 (an estensively inbred laboratory mouse strain). Each chromosome pair from the F1 of this cross contains one chromosome of Mirs .spYetin origin and one o f CS7BL/6 origin. Recombination during meiosis can occur between these chromos o m a giving new combinations of il/lci.~ryrcfiu and CS7BL/6 alleles. Backcrossing onto CS7BL,/6puts these recornhinant chromosomes on an homogeneous background where Mzu spretus allcles can e d d y be followed in N2 individuals.
I
I
I
I
Primer A
c Primer B
Primer A
5_(( W y
L-
Strain A DNA
Strain B DNA
Primer B
PCR of genomic DNA, size separation
---+ - - Strain A
Strain B
Recombinant inbred strains
(CA'y Allele (CA)n
Fig. 2. Schematic of PCR-analysed microsatcllites. Unique sequence primers (A and R) are synthesized to allow PCK across the simple (shown as CA here) repeat (opcn boxes). Polymorphisins in the length of these repeats is detected on agarose or polydcryhmide gels (heavy bars at bottom of figure). By analysing the PCR products generated from recombinant inbred strains or interspecific backcross animals these polymorphisms can bc mapped (see text).
D N A (initially taken from organs such as spleen and thymus, although other tissues are stored and used) for each backcross individual is eventually exhausted and no further probes can be mapped. The life-span of this resource can be conserved by dividing the panel into different levels of resolution. A probe is mapped to a particular chromosome or subchromosomal region using a small number of individuals and, then, finer mapping is carried out, using only those individuals informative for that region. The polymerase chain reaction (PCR) has high sensitivity and robustness, requiring very little DNA (50 ng or less) but generating a lot of product, which can be easily detected. It also has the advantage that it can be used to generate probes directly. For these reasons it has been incorporated into a number of meiotic mapping protocols, three of which are outlined below. Love ct aZ.(') have developed a meiotic mapping method based on PCR across dinucleotide repeats (Fig. 2). This exploits the high degree of repeat length polymorphism found between strains of mice. In this system the sequence around such repeats is determined and unique PCR primers are made. Products can then be mappcd as outlined above. The main limitation of this method is the requirement for sequence information so that primers can be designed. However, the method is amenable to multiplexing, since, with some optimization, it should be possible to carry out PCR on many loci in one reaction and analyse these on one gel. A second approach called Random Amplified Polymorphic D N A (RAPD) has recently been reported In this method the 'need-to-know' by Williams et sequence problem is avoided by randomly synthesizing oligonucleotide 10mers, each of which is used to carry out PCR on genomic DNA. Each primer generates a limited number of products, which are present in some individuals but not in others, depending on the
presence or absence of the appropriate primer sequence in the genome. It has been shown that these RAPDs are inherited in a Mendelian fashion in soyabean. This method remains to be evaluated for its usefulness in mouse genome mapping but ha5 great potential, although it will require synthesizing many primers and carrying out many PCR reactions per backcross individual. As a third approach, intcrspersed repetitive element PCR (IRS-PCR)".') has the advantage that it can be used to amplify unique sequence DNA without the need to know part of that sequence. IRS-PCR is carried out using PCR primers homologous to repetitive elements; since these elements are distributed throughout the genome it is possible to generate multiple probes at random. The number of probes generated in a single PCR reaction far exceeds that obtained by the previous two methods. These probes contain flanking repetitive element D N A and varying amounts of unique sequence. The human Alu repetitive element has been used for IRS-PCR from somatic cell hybrids, since under appropriate conditions these A h primers are species specific and are very common (close to one million copies)(7). The most common mouse interspersed repetitive elements are short interspersed nuclear elements (SINES). B1 (130000-180000 copies per genome), B2 (80 000-120 000 copies) and evolutionary conserved (EC) sequences (100000 copies)- and the long interspersed nuclear element (LINE) L1 (70000-100 000 copies) (rcviewed in ref. 9). The LINE element L1 is frequently and variably truncated at its 5' end, resulting in a lower copy number for the 5' end (10-12000) compared to the 3' end (1000r)O). It is therefore preferable to use PCR primers from the 3' end and that are orientated 3' out of the repeat. We have worked mainly with B1, B2 and L1, which are all thought to be generated by a mechanism involving an RNA intermediate. In the case of B1 and B2, these are probably the 7SL RNA and tRNA, respectively. Because of the retrotransposon nature of these repeats. we reasoned that their position in the genome would vary between species of mice, reflecting their time of insertion during evolution and their evolution after insertion. Such polymorphism of Alu IRS-PCR products has been reported between human individuals("). A number of possible outcomes can be predicted when one carries out PCR on different species of mice. Considering a particular unique sequence flanked by repetitive elements, then, when a single PCR primer is used. a product will be observed. provided the elements are close enough together and are in the appropriate orientations relative to one other. If in one species a primer is missing, there will be no product: altcrnatively, when there is another priming site near, a larger product will be observed. The products that we have mapped so far fall into these categories. Additionally. when one of the primer target sequences is sufficiently mutated, binding would either be eliminated
, Sequence B
,Sequence A
c
t
, Sequence A
l6
,
Sequence B
PC R on genomic DNA using IRS primers; Resolve on agarose gels
C57BL16
Mus spretus
-
-
Backcross progeny
-
A
-
B
or occur with lower affinity. giving rise to quantitative differences in the PCR products. (See Fig. 3 for schematic illustration). We prepared primers to B1, B2 and L1 repeats and used them singly to carry out PCR on 50ng of high molecular weight genomic DNA. It was not necessary to use combinations of primers in these experiments since the products generated were sufficiently complex for our immediate purpose. Additionally, we have found that combinations of primers from different repeat families result in smaller products, as one might predict from the combined frequencies. All of these primers generate complex banding patterns when resolved on agarose gels and stained with ethidium bromide. The L1 pattern is the least complex and many differences can be seen between the products from two different species of mice (illustrated schematically in Fig. 4). The Mus ,hpvekds specific bands can bc followed in backcross animals (see above), and have allowed us to assign bands A, B, C, D and E to mouse chromosomes 3, 2, 1, 12 and 2, respectively('). We were also able to map them relative to other genes mapped in the paneli'). The B2 and B1 products were too complex to $core for variants between the two species, although differences in the patterns of ethidium bromide stained bands on agarose gels were clearly visible. Tn order to simplify the pattern we blotted gels of the products and probed them with simple 12mer oligonucleotide repeats. Since it has been found empirically that some 12mers are present in up to 10 '% of clones in cosmid libraries from mammalian genomes("), we expected that that some of these sequences would have been amplified using our primers and that the patterns revealed using these sequences as probes would be much less complex than those defined by looking at all of the bands at once. Using oligonucleotides (GATA)3, (CC'IT)3 and (GACA)3 as probes we were able to identify a number of Mus spretuJ-specific polymorphisms in B1 and B2 products and map these to subchromosomal regions('). In addition a number of common bands were found.
Fig. 3. Principle of IRS-PCR and its application to genome mapping. Two wrts of variant are illustrated. variation in product sizc (sequence A ) and presence or abscnce (sequence B) of a product due to one of the primer sequenccs being inserted after divergence of the two species. Open boxes represent repetitivc elements, primers are represented as small arrows above the boxes and the products resolved on agarosc gels as heavy bars.
In total. we have mapped 13 new loci to 9 different chromosomes('). We have demonstrated the potential of this method for generating and simultaneously mapping inter-repeat PCR products.
Prospects for Meiotic Mapping New devclopments improving the capabilities in meiotic mapping can be expected in two directions. Using the interspecific backcross described above, only Mus sprelus variants can be mapped, but mapping panels are now being constructed in which the backcross is to Mus spretus (for example [(Mus spretus x C57BL/6) F1 x Mus spretus]) which will allow the segregation of Mus dornesricus variants to be followed. These reverse backcrosses will be useful for mapping PCR products from somatic cell hybrids. which usually contain chromosomes from inbred strains rather than wild mice. The other area of improvement to be expected will involve further simplifications of the protocols of PCRbased mapping, with the goal of maximising the number of loci and the number of meiotic evcnts to be analysed. Such an increase in the power of analysis can be achieved by direct scoring of PCR products on, for instancc, one or two dimensional gel systems, by indirect scoring of PCR products which can be either spotted on filters, or again separated by one or two dimensional gels, or by 'reverse scoring', in which the PCR products are labelled and used to probe an array of oligonucleotides on a filter surface.
One-dimensional gels In order to score directly a large number of loci using one-dimensional gels wc would have to carry out PCR reactions that generate only the number of bands (for example by RAPD, see above) which can easily be resolved. Consequently we would have to run many gels. For example. to score 1000 loci wc would need SO
Segregation of Variants
Gel Analysis of PCR Products
-
-+ I-
- IE
1
3
2
A
B
C
D
E
1
S
B
B
B
B
2
B
S
B
S
S
3
B
B
S
S
B
4
S
B
B
B
S
I
12
t' i f1 B
IB
I
A
E
C
t
Analysis
PCR reactions per backcross individual. each generating 20 loci markers, resolved on 50 gels.
Fig. 4. Mapping of LINE IRS-PCR products, adapted from actual data"'. Products are rcsolved on agarose gels, stained with ethidiuin bromide and the Miis speruv (S) specific bands scored in interspecific backcross (IB) animals (1.2,3,3, etc.) The data scored as eithcr MZLY ,spwfur (S) or C57BLl6 (B) alleles are entered into the database containing data on the inheritance of other probes mappcd i n the panel. Chrornowmal and subchromosomal assignments and genetic distances are thus determined.
Two-dimensional gels The number of gel5 can be reduced if the resolution of the gel system is increased. Optimization of the PCR reactions to increase the number of products in line with the resolution of the gel system reduces the number of individual PCR reactions required. Higher resolution is offered by two-dimensional gels, in which the first dimension is a size separation and the second depends on the sequence of a marker and its behaviour in a denaturing gradient (see ref. 12). Tf. for example. one assumes that 200 sequences can be resolved on a twodimensional gel (well within the capacity of these systems('')), then. to score 1000 loci, one would have to ruii five gels.
the form of IRS-PCR products). Using two-dimensional gels, the minimum number of oligonucleotide probings is further reduced. For example, assuming 200 loci can be separated, then 5 probes, each detecting 200 loci, would be needed to score 1000 loci on one gel. Obviously in practice more probes would be required but the relative economies of the alternatives are apparent from these simple comparisons. This indirect approach could also be applied to PCR products that have been dot blotted. The minimum number of probe oligonucleotides required is a function of the number of loci present in each dot. This is not very efficient, since to reduce the number of oligonucleotide probings by decreasing the number of loci per spot and increasing the number of spots pcr individual, one has to increase the number of PCR rcactions per individual.
Indirect detection methods Further economies of scale should be possible using indirect detection methods. We have describcd above the use of one-dimensional gels coupled with indirect scoring of markers by probing Southern blots with oligonucleotides. With this level of resolution we would have to use a large number of oligonucleotide probes. For example, to score 1000 loci we would nced a minimum of SO oligonucleotide probes, each detecting 20 loci (the resolution of a gel). Fewer gcls would be rcquired than for the direct approach, since each individual represents one track on a gel (for example in
Reverse detection However, these approaches can be reversed so that the oligonucleotide probes or a library of cloned DNA is immobilised on a solid support, in a high density These are probed with the PCR products themselves. In this case, a signal in the array generated by a probe from one species but not by a probe from another can be followed and mapped in backcross individuals. A mapped PCR product can be recovered using the detected oligonucleotide as a probe. If cloned DNA was used (in the form of bacterial colonies for example), it can be simply recovered from frozen
cell stocks used to make the array. This approach is very efficient requiring one PCR reaction per individual (using IRS-PCR, more using other methods) and one hybridisation per individual, the total number of hybridisations depending on the number of backcross individuals scored. Using oligonucleotide hybridisation techniques (above and refs 13.14) we should be able to 'fingerprint' the PCR products and provide partial sequence information. It is possible that the density of information generated in this way will allow the detection of changes (mutations) in the genome, so that, for example, tumour cells could be compared with their normal counterparts. The amount of sequence information obtained depends on the number of oligonucleotide permutations used(13,14).A full discussion of sequencing b oligonucleotide h bridisation is provided by DrmanacJ') and Lehrach(14? Somatic Cell Genetics Somatic cell hybrids containing whole mouse chromosomes or irradiation fragments(") are important mapping tools, providing levels of resolution between whole chroniosomes and analysis by pulsed field gels. The use of Alu TRS-PCR to generate multiple probes from human DNA present in somatic cell hybrids has been d e ~ c r i b e d ( ~This * ~ ) . is a very powerful technique that has simplified and facilitated the characterisation and exploitation of somatic cell hybrids. It would, therefore, be useful if mouse specific IRS-PCR primers could be found. Unfortunately, many of the somatic cell hybrids currently available are on hamster backgrounds, which. because of the shortcr evolutionary distance between mouse and hamster, makes it more difficult to find species-specific primers. Added to this is the problem that these repeats are less common than their human equivalents, decreasing the density of probes that can be generated. Careful selection of primers and conditions has solved the former problem and the latter problem can be mitigated by combination of rimers from different repetitive element families(P6-18).We have identified a primer sequence from the B2 family that is divergent between mouse and h a m ~ t e d ' ~ .Using ' ~ ~ , this primer it is possible to amplify muuse but not hamster, dog or human DNA. Conditions have been optimised to generate probes from irradiation hybrids that contain relatively low amounts of mouse DNA. These experiments increase the number of products amplified by using the B2 primer in conjunction with a species-specific 3' L1 primer (ref. 17, 18 see also 16). We can now use IRS-PCR to fingerprint hybrids containing similar amounts of DNA(17),and to generate probes from a subset of the DNA present in hybrids. This allows further characterisation of the DNA content of a hybrid as well as generating new probes for mapping cxperiments. IRS-PCR can be used to couple somatic cell genetics
with physical mapping and cloning methods. Pools of PCR products from somatic cell hybrids and irradiation hybrids can be used as probes (following compctition with repetitive DNA) to screen Yeast Artificial Chromosome (YAC) libraries (pre-hybridised with genomic DNA to suppress signal due to repetitive DNA), in order to recover YACs from limited or individual chromosome regions("). In this way, contiguous YAC clones can be assemblcd for megabase regions of DNA. Cloning of large regions of mouse DNA in YAC libraries(") offers a level of resolution in the range of PFGE analysis and increases the distance which can be covered by each IRS-PCR product. This latter point is important because of the frequency of repeats and the limitation on the size of PCR products (see earlier), which might make it impossible (using this particular approach) to make contiguous clones using cosmids for example. Finer mapping at the level of cosmids is then possible. using YAC clones. IRS-PCR. using species-specific primers, can be used to generate probes from individual pulsed-field gel fragments(17).Fragments arc cut out of a pulsed field gel and DNA extracted for PCR; at this stage individual slices can be screened by dot blotting to identify fractions carrying particular fragments for which a probe already exists(17). Probes generated in this way can be used for YAC recovery and map extension. To summarise, we now have a range of PCR-based mapping tools which, coupled with the power of meiotic mapping, somatic cell genetics, and physical mapping will permit a high resolution genetic map of the mouse genome to be constructed in the shortest possible time. Application of the techniques developed for genome sequencing may allow even higher resolution mapping. In the long run, these techniques may also offer efficient ways to analyse human populations. References 1 Cox. R. D., COBEMND, N . , JEYKI'IS.N. A. .\ND LElIK4CH. H. (1991). Intcrsperaed rcpetitive clement pnlymerasc chain reaction product mapping using it mouie interspccific backcrosh. Genomrcs, in press. 2 COPELANT). N . G. A N D JENKIYS, N. A. (1991). DCVehpmcilt and applicalions of a molecular genetic linkage map of the mouse genomc. TIG, in press. 3 AVUER. P.. AXAR, L.. DANDOLO, L. AND GIXNET. J. L. (1988). Genctic analyiis of thc mouse using interspecific crosseb. TIG 1>18-23. 4 BVCIIHERG, A. M.. BFDIGIAN, H. C; .. T 4 Y l . O R . B. A., BROWNF.I.L, E., IHI E, J . NAC~ATA, S . . JEVKIXS. N. A . AND C n P E L a u , N. G. (198R). Localicaiion of -2to chromosome 11: Linkage to other proto-oncogene and growth factor loci using interspccific backcruss mice. Oncoq?nr~Keseard7 2. 149-165. 5 LOVE.J. M.. K N rCiii-. A . M.. MCALEEK.M. AND TODD, J. (1990). 'Towardconstructiun of a high resolution map of the ninuse genome using PCR-analysed microsatcllites. ,Vucl. Acids lies. 18, 4123-4130, 6 WIiLI4MS. J . G. K.. KUBLLIK.A. N., LIVAK.K. J., K.1F.4LSKI. J. A. A Y U T i m w , S. V. (19990). DNA polyrnoi-phisms amplificd by arbitrary primers are uschil as genetic markers. iVcrcZ. Acids Rrs. 18, 6511-6535. 7 "&SON, D. L. . LEDBLITER. s. A , , CORIO. L., VILIORIA. M . F.. RAMIKEZSOLIS. R., WEBBSTER. ?'. D.. LEDBFlTER, D. H. ANY CASKFY, C. T. (1989). A h polymerase chain reaction: A method lor rapid isolation of human-specific sequences from complex DNA sources. Proc. Nnrt Acnd. Sri. L:SA 86, 66864690. 8 L F D B L ~ ES.RA, , . NELSON,D. L., WARREU, S. T. A Y D L F D B E ~ ED. R , H. (1990). Rapid isolation of DNA probes within specific chromosome regions by interspersed rcpetitive sequencc polymerase chain reaction. Genornics 6 , 475-481. 9 HASTIE, N . D. (1989). Highly rcpeated DNA familics in the genome of ,Vfrrs
rnusrirli~r.In Genetic Ljariants and srrains of the i o t i ~ ~ r a ~mouse o r y (M. F. Lyon and A . G. Searlc, Ed?.). pp. 559-571. OUP, Oxford. 10 SIUUFT.D.. DEMGON. J-M.. SIMARD,L. R. AND LABUDA. D . (19YO). Alumorphs - Human DKA polymorphisms detected by polymerase chain reaction using Ah-specific primers. Genomics 7. 331-334. J. D., ZEHTKEK. G . A N D LLHR4CH, 11 CRZIG. A. G , , N l Z L I l C . D., H~HEISEI H. (ISVO). Ordering of cosmid cloncs covering the herpes simplex virus typc 1 (HSIr-I) gcnome: a teat case for fingerprinting b y hybridisation. Nitci. Acids R E L 18, 2653-2660. 12 LJTTERLINTJEN,A. (3.. SLAGBOOM, P. E.. KKOOK, D . L. AN^ VIJG..I. (198Y). Two-dimensional DNA fingerprinting of human individuals. Proc. Nat1 Acad. Sci. I:SA 86, 274-2746. 13 D~OJANAC. R., LLSNON. ti., DRVANAC, s.. L A B ~ T1.., CRKVtNJAKOV. K.AYD LFHRACH, H. (1990). Partial sequencing by oligo-hybridisation: Conccpt and application5 in genome analysis. In Elecrropkoresis, supercompiifing and the hrtmnn gr'nome, Proceedings of the First International Conference. Tallahassee, Floiidrc, 10-13 April 1990 (C. K. Cantor and H . A. Lin, Eds.). World Scientific, Singapore. in press. 14 LEHRACH, H .. DRMANAC, R.,HOHFISEL.J . . LAWIN,z., LENNOX.G,. NIZLTIC,D.. MONACO,T.. ZEHEINER.G. AND POCSTKA.A . (1SYl). Hybridiyation lingcrprinting in genome mapping and sequencing. In Genome Analysis I : Gen& end Phyricc~lMopping. pp. 39-81, Cold Spring Harbor Laboratory Pr es . i n p r e s . 15 Cox, D. R., BURMEIST~.K, M., ROYDON PKICL, E., KIM,S. A ~ MYERS, D K.M. (190). Radiation hybrid mappmg: A somatic cell genetic method for constructing high-resolution maps of mammalian chromosomes, Science 250, 245-250.
16 RASTAN, 5. (1990). Czech mouse. TIC; 6. 233-236. 17 Cox. R. D.. SIURBS.L. AND LFHRACII. H. (1991). Mouse inter-repeat polymerase chain reaction: probe generation from somatic cell hybrids. (manuscript in preparation). 18 SIMMLFR, M-C.. Cox. R. D. APID A r ~ t a .P. (1991). Adaption of the intersperscd repetitive sequence polynierabe chain rcaction to thi: isolation of mowc DNA probes from somatic cell hy-hrids on a hamster background. Gmonrics, in pres?. 19 KRAYLV, A. S . , kfAXhCSHEVA. T. V., KR4hfEROV: D . A , . RYSKOV. A . P.. SKYRARIN, K. G., B A Y E A ~ ,. A. AND G ~ O R G I EGV. ,P. (1982). Uhiquitou5 transposnn-like repeats BI and B2 of the mouse gcnome: B2 sequencing. Nucl. Acids Res. 10. 1461-1415. 20 MONACO, A. P.. LAM.V . , IXNNON.(3.. DOUGLAS. C.. Z E H E T N ~G.. K. NIZETIC. D . ~OODFEL1,IJW.P. N. AND LEHIUCH, H. (1991). Direct hybridization of Ah-PCK products from irradiation hybrids to cosmid and yeast artificial chromosome lihrarie,. (nianuacript submitted). 21 L>\RIN.2 . AND LEHRACH. H. (1990). Yeasl artificial chromosomes: an alternative approach to the molecular analysis of mciusc mutations. Genet. Rea. 56. 203-208. ~
Roger D. Cox and Hans Lehrach are at the Genome Analysis Laboratory, Imperial Cancer Research Fund. PO Box 123, Lincoln's Inn Fields, London WC1 3PX, UK.
