Proc. Nati. Acad. Sci. USA Vol. 88, pp. 9623-9627, November 1991 Genetics cDNA selection: Efficient PCR approach for the selection of cDNAs encoded in large chromosomal DNA fragments (yeast artificial chromosomes/cosmids/major histocompatibility complex) SATISH PARIMOO*t, SANKHAVARAM R. PATANJALI*, HRIDAYABHIRANJAN SHUKLA*, DAVID D. CHAPLIN*, AND SHERMAN M. WEISSMAN* *Department of Genetics, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06510; and tDepartment of Internal Medicine and Howard Hughes Medical Institute, Washington University School of Medicine, St. Louis, MO 63110 Contributed by Sherman M. Weissman, July 24, 1991

ABSTRACT Identification of coding segments in large B30H3, isolated by one of us (D.D.C.; ref. 20), encompasses fragments of genomic DNA is a recurrent problem in genome a 320-kilobase (kb) region from the human major histocom- mapping and positional cloning studies. We have developed a patibility complex (MHC) class I HLA-A locus. Yeast chro- rapid and efficient protocol to achieve this goal, based on mosomes from agarose plugs were fractionated by 1% aga- hybridization of cDNA fragments to immobilized DNA and rose contour-clamped homogeneous electric field gel elec- recovery of the selected cDNAs by the PCR. The procedure trophoresis (21) in 0.5x TBE (45 mM Tris HCl, pH 8.3/45 permits rapid cloning of cDNA fragments encoded by large mM boric acid/0.2 mM EDTA) at 14'C for 45 hr at 150 V/cm genomic DNA fragments, groups of yeast artificial chromo- with a switching interval of 30 sec in an LKB Pulsaphor somes, or cosmids and has the potential to directly enrich apparatus. A 320-kb YAC B30H3 band was excised and cDNAs encoded in chromosome segments. By this approach we digested in situ with EcoPJ, and DNA was electroeluted. have been able to identify several non-major histocompatibility Immobilization of Genomic Cosmid or YAC DNA on Nylon complex class I clones from a yeast artificial chromosome that Discs. Cosmid DNA (0.2 ng) or YAC DNA (1-2 ng) was includes the HLA-A locus. digested with EcoPJ, heat-denatured at 950C for 3 min along with 50 ng ofan Hae III digest ofbacteriophage 4X174 DNA, Large segments of DNA from complex genomes are becom- as carrier DNA, and loaded in a 0.5-,ul volume onto 6.25-mm2 ing available as yeast artificial chromosomes (YACs) (1, 2), nylon discs (Hybond, Amersham), in the presence of 1Ox cosmid or phage contigs (3, 4), fragments enriched by affinity SSC (lx SSC = 0.15 M NaCl/0.015 M sodium citrate, pH capture (5), or chromosomal segments isolated in a foreign 7.0). The DNA on discs was denatured in 0.5 M NaOH background in somatic cell hybrids (6, 7). Current estimates containing 1.5 M NaCl for 5 min. The discs were neutralized are that only a few percent of total animal cell DNA is a with 0.5 M Tris Cl (pH 7.2) containing 1.5 M NaCl for 3 min, template for mRNA and that 10-20% of all mRNAs may be either exposed to UV irradiation in a Stratalinker (Strata- expressed in any particular cell type. A large part of the gene) for 1 min (autolink mode) or baked at 80'C in vacuum information content of the DNA that is of biologic relevance for 2 hr, and stored dry at 40C until used. for the organism could be extracted from the cDNA se- Quenchers of Nonspecific Hybridization to Nylon Discs. (i) quences. Methods to identify mRNA templates include the Human chromosome 15 . DNA from a human chro- identification of sequences conserved mosome 15 (15 NSO3, ATCC 57740) was between divergent digested with EcoRI and 0.8- to 6-kb (size range) DNA was species such as man and mouse (8-11), use of these se- cloned into the pTZ18 vector (Pharmacia) at the EcoRI site. quences as probes for hybridization to cDNA libraries, direct A total of 3 x 105 recombinants was generated after trans- screening ofcDNA libraries with probes made from the entire formation and ampicillin selection. large fragment under conditions that suppress hybridization (ii) Genomic repetitive sequence library (XRLI). XRLI was due to repetitive sequences (12), and analysis of sequences prepared from the human X chromosome library around CpG-rich islands that lie at the 5' ends ofmany but not (LAOXNLO1, ATTC 57750) by digestion of various DNA all mRNA templates (13, 14). More recently PCR procedures samples with several single restriction enzymes, pooling, with consensus primers for Alu repetitive sequences in het- cloning 0.3- to 2-kb (size fraction) DNA in Charon BS vector erogeneous nuclear RNA (15, 16) and exon trapping methods (22), and probing the resulting library with 32P-labeled total that depend on in vivo splicing systems to detect DNA genomic DNA (23). Five hundred positive plaques were sequences encoding RNA splicing sites (17, 30) have been processed for isolation of phage DNA and subsequent plas- introduced. We report here a conceptually simple alternative mid preparation (22). approach, cDNA selection, that is experimentally rapid and (iii) rRNA-specific clones. clones containing hu- relatively insensitive to the size of the DNA fragments to be man rRNA gene EcoRI fragments of7.3 kb and 5.8 kb cloned screened. into pBR322 at the EcoRI site were kindly provided by D. Ward (Yale University) (24). MATERIALS AND METHODS (iv) Poly(dI)-poly(dC). Material of 724 base pairs (average length; Pharmacia) was used without any further treatment. Preparation of Random Short-Fragment cDNA Library. A (v) Yeast DNA. Yeast DNA from strain AB1380, which is human spleen cDNA library was prepared using random the parent host of YAC B30H3, was prepared (20) and hexamer primers as described (18). sonicated to an average size of 0.7-1 kb. Plasmid DNA from Preparation of Genomic Target DNA-Cosmids and YACs. the sources i-iii was sonicated, digested with EcoRI, and Cosmid DNA was prepared by standard methods (19). YAC treated with 1 unit of mung bean nuclease per pug of DNA at

The publication costs of this article were defrayed in part by page charge Abbreviations: MHC, major histocompatibility complex; PFGE, payment. This article must therefore be hereby marked "advertisement" pulsed-field gel electrophoresis; YAC, yeast artificial chromosome. in accordance with 18 U.S.C. §1734 solely to indicate this fact. TTo whom reprint requests should be addressed.

9623 Downloaded by guest on September 24, 2021 %24 Genetics: Parimoo et al. Proc. Natl. Acad. Sci. USA 88 (1991) room temperature for 30 min to generate blunt-ended frag- 300 base pairs was cloned into phosphatase-treated AgtlO at ments of DNA ranging from 0.3 to 1 kb in size. the EcoRI site. The plaques were analyzed with the probes Conditions for Quenching of Nonspecific Bindi"g of DNA to used for DNA blots. Nylon Discs and Hybridization for Selection of cDNAs. Non- specific binding to target DNA was suppressed by incubating 6.25-mm2 nylon discs with immobilized DNA at 65TC in 40 ILI RESULTS of solution containing 5x SSPE (lx SSPE = 0.15 M NaCI/ The basic protocol ofselection is outlined in Fig. 1. The nylon 0.02 M sodium phosphate/i mM EDTA, pH 7.2), 5x Den- filter discs with immobilized cosmid or YAC DNA were hardt's solution (19), 0.5% SDS, and a mixture of quenching blocked for repetitive, ribosomal, or GC-rich sequences with agents described above, after heat denaturation at 950C for 5 quenching agents and then hybridized with total random min. The final concentrations of quenching agents in the cDNA fragments prepared by PCR amplification with flank- reaction mixture were as follows: for XRLI, 0.025 ug/jl~; for ing vector primers. After an appropriate wash, the hybrid- chromosome 15 DNA fragments, 0.05 jxg/Al; for rRNA selected material was PCR-amplified, analyzed on Southern plasmid, 0.04 Ag/,ul; for poly(dI)-poly(dC), 0.02 ,ug/Al; and blots, and cloned in AgtlO. The first essential step for the for sonicated yeast DNA (for YACs only), 0.025 ,ugId. After cDNA selection was to establish methods and reagents that incubation under mineral oil for 20-24 hr, discs were briefly would reduce the coselection of repetitive and GC-rich rinsed with 5x SSPE/0.1% SDS at room temperature and sequences to a minimum. quickly transferred to a fresh tube containing 30 ul4 of Hybridization Conditions for cDNA Selection. We investi- hybridization solution of the same composition as the gated a number of conditions for hybridization and washing quenching reaction except that poly(dI)-poly(dC) was re- of nylon discs. Hybridization in SSPE, rather than in form- placed by a heat-denatured short-fragment cDNA library amide, and washing conditions as described in Materials and (PCR-amplified with primer set C, defined below) at a con- Methods proved to be optimal as far as yield and specificity centration of 10 pug/ml. After an incubation of 36-40 hr at of the selection process were concerned (data not shown). A 65°C under mineral oil, the discs were washed for two 5-min more stringent wash of 2.4 M tetraethylammonium chlo- periods with 2x SSC/0.1% SDS at room temperature, for ride/50 mM Tris HCl, pH 8/2 mM EDTA/0.1% SDS (27) at three 20-min periods with 2x SSC/0.1% SDS at 650C, for one various temperatures (50-600C) was investigated and could 20-min period with lx SSC/0.1% SDS at 65°C, for one be used at 550C where the yield is of a lesser concern-e.g., 10-min period with 0.2x SSC/0.1% SDS at 650C, and for two when using higher concentrations oftarget DNA immobilized 20-min periods with O.1x SSC/0.1% SDS at 65°C in 1.5-ml on nylon discs (data not shown). Eppendorftubes with 6001l ofwash solution. The discs were Specificity and Sensitivity of the Selection Process. To avoid transferred to fresh tubes and rinsed twice with 0.lx SSC PCR artifacts while providing the desired overall amplifica- alone at room temperature. The discs were kept wet through- tion, three sets of nested primers, with set A closest to the out the procedure and were either used immediately for the EcoRI cloning site and set C farthest from the cloning site, PCR or stored in 20 ,ul of water at 4°C overnight and then the were tested for PCR amplification. Primer set B (between sets PCR was carried out in the same tube. A and C) gave smaller-sized PCR products, compared to PCR Amplification. Two sets of primers, sets A and B, primer sets A and C (data not shown). Hence we routinely have been described earlier as primer sets 2 and 1, respec- used primer sets C and A for subsequent reamplification tively (18). Another set of primers, set C, is composed of stages (Fig. 1). The data in Fig. 2 indicate that the selection primer C1 (5'-CCACCTTTTGAGCAAGTTCAG-3') and process is quite sensitive and can detect a signal on a DNA primer C2 (5'-GAGGTGGCTTAT GAGTATTTC-3'), 18 and 20 bases away, respectively, from the EcoRI site of the AgtlO IGenornic DNA vector. on nyvncr disc< Short-fragment cDNA inserts were liberated from a AgtlO

random cDNA library by 30 cycles of PCR amplification in e -tlrrn 100 ,ul, essentially as described (18), except that 2 mM MgCl2 I~ and 0.9 ,M primer set C were used, and the reactions were carried out first by initial denaturation at 94°C for 2 min with all the components of the PCR mixture except AmpliTaq r...,ivbridjze randorr DNA polymerase (Cetus), followed by addition of the en- short fragment lo"a zyme at 80°C. PCR cycles were 94°C for 70 sec, 50°C for 90 EDNA who flanki-n * vector sequencres sec, and 72°C for 120 sec. rIIunIn& PCR amplification of selected material on nylon discs was rs-"-!, carried out by direct immersion of nylon disc in the PCR mixture and carrying out the PCR for 30 cycles as described for Wnlash cDNA library amplification except that the primer (set C) I -^on-soec~fic concentration was reduced to 0.3 ,uM. A sample (1.5-2 1.l) from eI b s~ this PCR product was taken for a second round of 30 cycles of the PCR in the prqsence of the inner primer pair (set A). OCR vatse'Do DNA Blots and Probes. Southern blot analysis of PCR ' ankina vector products was done as described (18). The probes used for or: me,!s Southern blots were HLA-A cDNA (isolated by one of us-H.S.), MHC class II DRa (3-kb EcoRI fragment) (25), y-globin 200-base-pair fragment derived by PCR amplifica- Neste- pr mc tion of a globin cosmid (26), and total genomic DNA (soni- Re-PCR w ih cated material). Other probes have been described (18). inner vector prmeo of PCR-Amplifled and -Selected Material. A of restrict, with rcO R' Cloning pool bl-eln i71A five 100-/ tubes of PCR reaction products from the second / , round of 30 cycles of PCR was concentrated on a Centricon- 1It 100 (Amicon) by following the manufacturer's instructions and digested with EcoRI, and the cDNA material greater than FIG. 1. Protocol for cDNA selection. Downloaded by guest on September 24, 2021 Genetics: Parimoo et al. Proc. Natl. Acad. Sci. USA 88 (1991) 9625

input cDNA (ng) Cosmid (pg" 1 2 3 4 5 6 7 8 9 10 1112 13 14 kb

c,? I -1.6 x n o) CD o0csa x °D Starting -1.0 co co - 0 CD N C\OCNJLO- Material A 2 §-0.5 -0.22

A B_ Ii".' 0. C B I 4911 D C

E D

FIG. 3. Selection with YAC B30H3 or cosmid mixtures. Southern blots of selected cDNAs recovered after two rounds of PCR ampli- E fication and one cycle (lanes 3-5) or two cycles (lanes 1 and 2) of selection from purified YAC B30H3 DNA immobilized on nylon discs are shown; lane 5 contains cDNAs selected from a nylon strip FIG. 2. Specificity and sensitivity of PCR selection process. from a PFGE Southern blot containing YAC B30H3. Lanes 6-12 Southern blots of PCR-amplified cDNAs (using primer set C) se- contain cDNAs recovered after two rounds ofPCR amplification and lected from an MHC class I cosmid alone (input cDNA lanes) or a one cycle of selection from a mixture of three cosmids (human MHC mixture of MHC class I and P-globin cosmids (Cosmid lanes). class I and II and f3-globin) alone (lanes 6 and 7) or in combination Starting material is also included. A constant cDNA concentration of with rodent DNA (lanes 8-10) or rodent DNA alone (lanes 11 and 12). 180 ng/30 Ml of hybridization mixture was used for various amounts Lanes 13 and 14 contain samples ofPCR products from a total cDNA of cosmids (as indicated) and a constant amount (200 pg) of each library. Concentrations of target genomic DNAs immobilized on cosmid per disc was used for various amounts of input cDNA (as nylon discs were as follows: YAC, 5 ng (lanes 1-5); cosmid mixture, indicated). All immobilized samples ofcosmid DNAs were loaded on 10 pg (lanes 6 and 8), 50 pg (lane 9), and 200 pg (lanes 7 and 10); rodent 6.25-mm2 discs and all hybridization reaction mixtures were 30 Al. DNA, 1 gg (lanes 8, 9, and 11) and 5 ,ug (lanes 10 and 12). Rodent The probes used were HLA-A (A), human p-globin (B), human DNA was immobilized on 9-mm2 discs for 1 Mug of DNA and 25-mm2 y-globin (C), total human genomic DNA (D), and MHC class II DRa discs for 5 ug of DNA. The probes for blots were as follows. (E). YAC-encoded probes: HLA-A probe (A) and clone B30.7 (B). Cosmid-encoded but not YAC-encoded probes: MHC class 11 (C) blot with class I or f3-globin probes (moderate abundant and 3-globin (D). rRNA probe that is encoded neither by cosmid nor species) with as little as 1 pg of target cosmid DNA mixture by YAC (E). The concentration of total cDNA in the hybridization (Fig. 2 A and B). Though increasing target concentration had mixture for selection was 10 Ag/ml during the first cycle of selection no or very little effect on coselection of nonspecific cDNAs and 0.25 Ag/ml during the second cycle ofselection. The lanes inA-E are identical. (Fig. 2 D and E), increasing cDNA concentration did cause coselection of a small proportion of contaminating cDNAs selected materials were probed with YAC-encoded specific (Fig. 2 B-E). However, at the standard cDNA concentration probes (Fig. 3 A and B, lanes 1-5) vs. nonencoded probes (180 ng/30 ,l), the MHC class I cosmid selected HLA-A (Fig. 3 C-E, lanes 1-5). sequences (Fig. 2A) but did not significantly coselect B3-globin To evaluate the ability of the system to select cDNAs sequences or other sequences not encoded in the cosmid corresponding to human DNA fragments contained in hybrid (Fig. 2 B-E). Also, we have routinely used lower concentra- cell lines, experiments were performed with a mixture of tions of cDNA in the second cycle of selection with a cosmids (MHC class I, MHC class II, and f3-globin) alone or considerable increase in the specificity (Fig. 3 A and B, with rodent DNA (sonicated to an average size of 3 kb). The compare lanes 1 and 2 vs. 3 and 4). For this experiment we blots of PCR products were probed with various cosmid- cloned the selected material after one cycle of selection (Fig. encoded probes (Fig. 3 A, C, and D) or nonencoded probes 3, lanes 3 and 4) in AgtlO and then used it for a second cycle (Fig. 3 B and E). The yield of specific product was reduced of selection (Fig. 3, lanes 1 and 2). The data in Fig. 3 also when large amounts of hamster DNA were included with indicate that one can use pulsed-field gel electrophoresis cosmid DNA (Fig. 3, lanes 8-10) in comparison to cosmid (PFGE)-purified YAC DNA either digested with a restriction alone (Fig. 3, lanes 6 and 7). enzyme and immobilized on nylon (lanes 1-4) or directly Clones from YAC B30H3-Selected Libraries. We chose from a thin nylon strip cut out from Southern blot of a PFGE plaques from the YAC B30H3-selected minilibrary that did gel of a YAC (lane 5). The specificity ofthe selection process not react with MHC class I, rRNA gene DNA, or repetitive is also evident with YAC B30H3 when Southern blots of DNA probes and analyzed them further on DNA blots of Downloaded by guest on September 24, 2021 %26 Genetics: Parimoo et al. Proc. Natl. Acad. Sci. USA 88 (1991)

human DNA, and two clones not hybridizing with either A E c) YAC or genomic DNAs. (kcs B 1 2 Quantitative Evaluation ofSelected Library. Libraries were prepared from once- or twice-selected cDNA samples from kb two experiments with YAC B30H3 or a cosmid mixture of 12 \ MHC class I and II and 3-globin, and the results are shown in Table 1. MHC class I (moderate abundance) cDNAs 4.0 -,- w represent an enrichment of 70-fold, as is evident from the 3.0 experiment with YAC B30H3, and a rare species sequence 2.0 - (B30.7) is enriched more than 1000-fold in one round of 1 .0 - selection and more than 7000-fold by the second round of selection. Treatment of quenching reagents to make them 0.5 - blunt-ended after EcoRI digestion reduced coselection of nonencoded cDNAs.

FIG. 4. Identification of an anonymous clone. (A) Southern blot DISCUSSION ofEcoRI-digested human genomic DNA (6 ,ug) or YAC B30H3 DNA The goal of the present experimental design is to rapidly (20 ng), probed with the clone B30.7. The extra bands in the genomic DNA lane, in comparison to the YAC B30H3 lane, could indicate that identify most of the coding sequences in large DNA frag- a YAC B30H3 does not contain the entire gene B30.7 or that the gene ments. A limitation ofthe approach is that it requires cDNA is part of a multigene family. The minor difference in mobility library representing all possible mRNA from a species to be between the common band of YAC and genomic DNA could be due able to identify all coding sequences. Although there are to the difference in the amount of DNA loaded. The presence of a serious practical obstacles to obtaining such libraries for smear in the genomic DNA probably resulted from the high GC man, the development of normalized libraries (18, 28, 29) and content ofthis clone (B30.7). (B) Southern blot of a contour-clamped the pooling and renormalization of libraries from multiple homogeneous electric field gel of YAC B30H3 (lane 1) or YAC tissues and developmental stages make it possible to ap- AlSOA10 (lane 2), probed with the clone B30.7. YAC A15OA10 does proach such a complete library in experimental animals and not encode the B30.7 sequence but encodes HSP70, factor B, to a lesser degree in man. A second limitation in principle of complement C2, etc., of chromosome 6. The arrow indicates the position of YAC B30H3. this approach is that pseudogenes that have not deviated extensively from the active gene would also select cDNA EcoRI-digested genomic DNA or YAC B30H3 DNA (Fig. clones and would have to be sorted out on some other basis. 4A). and PFGE gels oftwo YACs (Fig. 4B) to identify unique There are several advantages to the present approach over sequences encoded in YAC B30H3. We identified a clone the other approaches to the problem of identification of (B30.7) that is present in the original cDNA library at a coding sequences. The selection procedure is insensitive to of less than 1 in 1 x 105 clones. This clone is number and size of , cryptic splice sites, or signals in frequency intergenic DNA that resemble splice sites. The method would present at a frequency of 1% and 4% in the YAC B30H3- detect 5' and 3' exons and genes that lack CpG-rich islands selected library after .one and two cycles of selection, re- and introns or those genes that are not transcribed in hybrid sequence comparison spectively (Table 1). The preliminary cell lines or have highly diverged between species. The with GenBank/EMBL did not show significant homology cDNA fragments that are obtained are expected to be ran- with any known sequence (date of search, September 19, domly distributed over the mRNA and, therefore, would link 1991). Northern blot analysis with clone B30.7 indicated that exons. The approach is more robust than is screening with the clone hybridized to an RNA species of =1.0 kb in size labeled probes from large DNA fragments and is clearly less (data not shown). The RNA was induced within 4 hr of sensitive to the size of the fragment used for selection. phorbol 12-myristate 13-acetate treatment in Jurkat cells and Selection is particularly convenient for YACs where a single is present in both human spleen and thymus (data not shown). gel and blot can be used to identify cDNAs corresponding to Southern blot analyses of 10 random non-MHC class I clones a dozen or more YACs within a relatively short period of from the YAC B30H3-selected minilibrary, after two cycles time. of selection (Table 1, experiment 1), revealed six clones The results of dilution experiments and studies with re- hybridizing with both YAC B30H3 and human genomic DNA constituted mixtures ofDNAs indicate that the method could as single-copy species, one clone hybridizing with YAC be used directly to select cDNAs encoded by very complex B30H3 DNA as a single-copy species that appears to repre- mixtures of DNA, possibly even from flow-sorted chromo- sent low copy repeats in human DNA, one clone hybridizing somes or chromosome-specific libraries. The reduction in with both YAC B30H3 and yeast genomic DNA but not yield of a specific PCR product with more complex DNA Table 1. Positive plaques with various probes % positive plaques Immobilized Total cDNA material used for genomic target MHC MHC genomic Exp. screening DNA class I class II f3-Globin y-Actin rRNA DNA B30.7 I Total cDNA library 0.5 0.1 0.1 0.2 10 6.7 5.8 x 10-4 One cycle of selection* YAC B30H3 DNA 15 0.2 NT NT 14 27 1 Second cycle of selectiont YAC B30H3 DNA 30 ND ND ND 0.1 7.4 4.5 II One cycle of selectiont YAC B30H3 DNA 34 ND ND ND 2 12.7 1.5 Cosmid DNA 21 22.6 23.5 ND 2 0.5 ND Genomic target DNA was immobilized on nylon discs. Cosmid DNA was a mixture ofMHC class I and II and (3-globin DNAs. NT, not tested; ND, not detected in at least 500 plaques. *Quenching agent was sonicated plasmid DNA. tQuenching agent was sonicated plasmid DNA, digested with EcoRI and made blunt-ended with mung bean nuclease. Downloaded by guest on September 24, 2021 Genetics: Parimoo et al. Proc. Natl. Acad. Sci. USA 88 (1991) 9627

targets (Fig. 3, lanes 8-9) could be due to steric hindrance C. J., Kurnit, D. M. & Kunkel, L. M. (1986) Nature (London) because of high DNA concentration on the nylon disc or to 323, 646-650. the competition between the large number ofselected cDNAs 9. Page, D. C., Mosher, R., Simpson, E. M., Fisher, E. M. C., during the PCR. Mardon, G., Pollack, J., McGilliway, B., de la Chapelle, A. & procedure could be used to directly isolate Brown, L. G. (1987) Cell 51, 1091-1104. The selection 10. Rommens, J. M., lanuzzi, M. C., Kerem, B.-S., Drumm, polymorphic clones such as sequences containing CA repeats M. L., Melmer, G., Dean, M., Rozmahel, R., Cole, J. L., derived from a particular large fragment, to select linking or Kennedy, D., Hidaka, N., Buchwald, M., Riordan, J. R., Tsui, jumping clones corresponding to a large fragment, or to use L.-C. & Collins, F. S. (1989) Science 245, 1059-1065. linking clones to select jumping clones and vice-versa for 11. Call, K. M., Glaser, T., Ito, C. Y., Buckler, A. J., Pelletier, J., long-range physical mapping. The search for the lesions in Haber, D. A., Rose, E. A., Kral, A., Yeger, H., Lewis, W. H., genetic disorders would be simplified if one could take DNA Jones, C. & Housman, D. E. (1990) Cell 60, 509-520. corresponding to the region suspected to harbor a mutant 12. Elvin, P., Slynn, G., Black, D., Graham, A., Butler, R., Riley, gene and use this to select cDNA species that are expressed J., Anand, R. & Markham, A. F. (1990) Nucleic Acids Res. 18, in relevant tissues. If the selection approach can be applied 3913-3917. to large arrays oflysed bacterial colonies containing cosmids 13. Sargent, C. A., Dunham, I. & Campbell, R. D. (1989) EMBOJ. or yeast colonies containing YACs, it would provide a most 8, 2305-2312. convenient method to obtain expression maps of complex 14. Hanson, I. M., Poustka, A. & Trowsdale, J. (1991) Genomics 10, 417-424. genomes. 15. Corbo, L., Maley, J. A., Nelson, D. L. & Caskey, C. T. (1990) Note Added in Proof. We have confirmed that this approach can be Science 249, 652-655. used to select cDNAs directly from nylon filter carrying lysed 16. Liu, P., Legerski, R. & Siciliano, M. J. (1989) Science 246, colonies of bacteria that harbor specific cosmids. 813-815. 17. Buckler, A. J., Chang, D. D., Grau, S. L., Brook, J. D., We thank M. Weiler for technical preparation of this manuscript. Haber, D. A., Sharp, P. A. & Housman, D. E. (1991) Proc. We thank A. Bhargava for the Northern blot and H. Vasavada and Natl. Acad. Sci. USA 88, 4005-4009. S. Ganguly for MHC class I and II probes. We thank B. Brownstein 18. Patanjali, S. R., Parimoo, S. & Weissman, S. M. (1991) Proc. and P. Taillon-Miller for screening the human YAC library and S. Natl. Acad. Sci. USA 88, 1943-1947. Bronson and D. Geraghty for preliminary characterization of the 19. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1990) Molecular YAC clones. A fellowship award to S.P. from the Cancer Research Cloning:A Laboratory Manual (Cold Spring Harbor Lab., Cold Institute, New York, is gratefully acknowledged. This work was Spring Harbor, NY). supported in part by National Cancer Institute Outstanding Inves- 20. Bronson, S. K., Taillon-Miller, J. P., Chorney, M. J., Ger- tigator Grant R35 CA42556-06 (S.M.W.) and in part by National aghty, D. E. & Chaplin, D. D. (1991) Proc. Natl. Acad. Sci. Institutes of Health Grants P50 HG00201 and AI15322 (D.D.C.). USA 88, 1676-1680. 21. Vollrath, D. & Davis, R. W. (1987) Nucleic Acids Res. 15, 1. Abidi, F. E., Wada, M., Little, R. D. & Schlessinger, D. (1990) 7865-7876. Genomics 7, 363-376. 22. Swaroop, A. & Weissman, S. M. (1988) Nucleic Acids Res. 16, 2. McCormick, M. K., Shero, J. H., Cheung, M. C., Kan, Y. W., 8739. Hieter, P. A. & Antonarakis, S. E. (1989) Proc. Natl. Acad. 23. Feinberg, A. P. & Vogelstein, B. (1984) Anal. Biochem. 137, Sci. USA 86, 9991-9995. 266-267. 3. Spies, T., Blanck, G., Bresnahan, M., Sands, J. & Strominger, 24. Wilson, G. N., Hollar, B. A., Waterson, J. R. & Schmickel, R. J. L. (1989) Science 243, 214-217. (1978) Proc. Natl. Acad. Sci. USA 75, 5367-5371. 4. Spies, T., Bresnahan, M. & Strominger, J. L. (1989) Proc. 25. Das, H. K., Lawrance, S. K. & Weissman, S. M. (1983) Proc. Natl. Acad. Sci. USA 86, 8955-8958. Natl. Acad. Sci. USA 80, 3542-3547. 5. Kandpal, R., Ward, D. C. & Weissman, S. M. (1990) Nucleic 26. Collins, F. S. & Weissman, S. M. (1984) Prog. Nucleic Acids Acids Res. 18, 1789-1795. Res. Mol. Biol. 31, 315-462. 6. Nussbaum, R. L., Airhart, S. D. & Ledbetter, D. H. (1986) 27. Wood, W. I., Gitschier, J., Lasky, L. A. & Lawn, R. M. (1985) Am. J. Med. Genet. 23, 457-466. Proc. Natl. Acad. Sci. USA 82, 1585-1588. 7. Warren, S. T., Knight, S. J. L., Peters, J. F., Stayton, C. L., 28. Weissman, S. M. (1987) Mol. Biol. Med. 4, 133-143. Consalez, G. G. & Zhang, F. (1990) Proc. Natl. Acad. Sci. 29. Ko, M. S. H. (1990) Nucleic Acids Res. 18, 5705-5711. USA 87, 3856-3860. 30. Duyk, G. M., Kim, S., Myers, R. M. & Cox, D. R. (1990) Proc. 8. Monaco, A. P., Neve, R. L., Colletti-Feener, C., Bertelson, Natl. Acad. Sci. USA 87, 8995-8999. Downloaded by guest on September 24, 2021