Construction and utility of a human chromosome 22-specific Fosmid library

ELSEVIER

Genetic Analysis: Biomolecular Engineering 12 (1995) 81-84

ENETIC ALYSIS Engineering

Biomotecular

Construction and utility of a human chromosome 22-specific Fosmid library U n g - J i n K i m *a, H i r o a k i S h i z u y a a, J e s u s S a i n z b, J e f f r e y G a r n e s c, S t e f a n M . P u l s t b, P i e t e r d e J o n g c, M e l v i n I. S i m o n a aDivision of Biology, 147-75, California Institute of Technology, Pasadena, CA 91125, USA bNeurogenetic laboratory, Cedars-Sinai Medical Center and University of California at Los Angeles, Los Angeles, CA 90048, USA CHuman Genome Center, L-452, Lawrence Livermore National Laboratory, Livermore, CA 94550, USA Received 23 May 1995; revision received 25 July 1995; accepted 31 July 1995

Abstract

We have previously demonstrated the capability of the Fosmid vector based on Escherichia coil F-factor replicon to stably propagate cosmid-sized human genomic DNA fragments. Using the Fosmid vector, we have constructed and arrayed a 10 x human chromosome 22-specific library, partly by picking human positive clones from a total Fosmid library constructed using DNA from human-hamster hybrid cell line containing human chromosome 22, and partly by using flow-sorted chromosomal DNA. The clones and physical contig maps af the clones in the library will serve as a valuable resource for detailed analysis of the chromosome by providing reliable materials for high resolution mapping and sequencing. In order to efficiently build physical maps for the chromosomal regions of interest spanning several hundred kilobases to a megabase, it is necessary to rapidly identify subsets of the Fosmid clones from the library that cover such regions. In this report, we describe a method of using random amplification products derived from YAC clones to rapidly identify a subset of Fosmid clones that cover a specific genomic subregion. Keywords: Fosmid library; Human chromosome 22; Contig map; YAC clone;

1. Introduction

Fosmid and BAC (bacterial artificial chromosome) systems are reliable methods to generate large and stable genomic clones [1-2]. We have demonstrated that BAC clones can stably propagate human DNA inserts as large as 300 kb [2]. The Fosmid cloning system employs a low copy number cosnaid vector based on the Escherichia coli F-factor replicon, and provides a method of preparing mini-BACs with an average of 40 000 base pair inserts with the efficiencies comparable to other cosmid cloning systems [1]. Once cloned, the structure of Fosmids are identical to BAC clones except fo;r the smaller insert size. The Fosmid system is a method of choice to rapidly create high titer mini-BAC libraries from a small amount of genomic materials. For example, the Fosmid system is * Corresponding author. 1050-3862/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 1050-3862(95)00122-3

excellent for rapidly creating chromosome-specific miniBAC libraries from flow-sorted chromosomal DNA. The major advantage of Fosmids over other cosmid systems lies in its capability of stably propagating human DNA fragments. Highly repetitive in nature, human DNA is well known for its extreme instability in multicopy vector systems. We have found that the stability increases dramatically when the human DNA inserts are present in single copies in recombination deficient E. coli cells [1]. Therefore, Fosmids will serve as reliable substrates for large scale genomic DNA sequencing. For this reason, it is important to be able to efficiently generate Fosmid contig maps covering the regions to be sequenced. The first step in generating such contigs would be to identify subsets of Fosmid clones corresponding to the regions of interest. We report here the construction of a 10× coverge human chromosome 22-specific Fosmid library. We also present a method of using amplified products of YAC

82

U.-J. Kim et al./ Genetic Analysis: Biomolecular Engineering 12 (1995) 81-84

clones as probes to screen the library and rapidly identify a subset of the Fosmids covering a specific genomic region. The Fosmid clones in the subset were assembled into a contig using the contig9 program originally developed for contig assembly of cosmid clones based on restriction fingerprint data [3-4]. 2. Materials and methods

2.1. Construction of Fosmid library The first 3 x human chromosome 22-specific Fosmid library consisting of approximately 4500 clones was construtted using the monosomic human-hamster hybrid cell line KG-1 that carries human chromosome 22 (kindly provided by B. Emmanuel). The DNA prepared from the cell line was partially digested with Hind III and processed for packaging and transfection as described previously [1]. Colonies generated from the total cellular DNA were plated on LB-agar plate containing 25/~g/ml chloramphenicol, transferred to nylon membranes by colony lifting, and subjected to colony hybridization with p32-1abelled human placental DNA (GIBCOBRL) under non-suppressive conditions [5]. Hybridization was carried out at 58°C in 15% formamide, 0.2 M sodium phosphate (pH 7.2), lmM EDTA, 2% SDS and 1% BSA. The filters were washed in 0.1 x SSC at 42°C and the positive colonies were picked and arrayed into 96-well microtiter plates. The second part of the Fosmid library was construtted at the Lawrence Livermore National Laboratory using flow-sorted chromosome 22 DNA derived from a human-hamster cell line J640-51 [6-71. Chromosomes were isolated by the polyamine method [8], stained with Hoechst and chromomycin A3 [9] and sorted on the Livermore modular instrument for high speed chromosome sorting MoFlo [10]. The sorted chromosome 22 (2.4 million) were collected, centrifuged and used for DNA extraction. The chromosomal DNA was partially digested with MboI, treated with alkaline phosphatase and ligated to BamHI-AatlI-treated pFOS 1 vector arms [11. The ligation mixture was packaged with Gigapack II Gold packaging extracts (Stratagene). A portion of the packaging mixture was titered on host DH5-alpha MCR (GIBCO-BRL Life Technologies Inc.) by infecting the cells for 15 rain at 37°C, expressing the antibiotic resistance for 30 min at 37°C, and plating on LB-agar plates containing 25 t~g/ml chloramphenicol. The resulting library is estimated to contain 4.1 × 104 cfu's. In total, 14 500 clones were arrayed in microtiter plates. A more detailed description of the procedures for constructing chromosome-specific cosmid and Fosmid libraries will be published elsewhere (de Jong and Games, in preparation).

and A226C4 kindly provided by Marco Giovannini, and YAC bands were identified and excised from agarose gel after the separation from yeast chromosomes on a pulsed field gel. The YAC DNA was purified from the agarose gel blocks using the GeneClean kit (Biol01). The large DNA was sheared by heavy vortexing, and released from the glass beads in 40/zl TE by heating in boiling water for 10 rain. The DNA was amplified according to the SOP-PCR [11] procedure using an arbitrary oligo 5'-CGTGTGCCAACCACTGG-3' after minor modifications. A 1-/~1aliquot of the boiled YAC DNA was added to the first PCR mixture (1 /zl 10x PCR buffer (Perkin-Elmer), 1.5 #1 dNTP (2.5 mM for each nucleotide), 1.5 t~g oligo, 0.5/~1 10% Nonidet P-40, 0.5 t~l 10% Tween 40, 5 #1 24% PEG8000, 1.75/~1 H20 and 0.25/~1 TaqI polymerase). After 3 min heating at 95°C, the samples were subject to 7 thermal cycles consisting of 93°C for 3 min, 18°C for 60 min, 18°C to 50°C over 30 min, and 50°C for 30 min. The first PCR products (3/~1) were added to 10/~1 of the second PCR mix (1 #1 10x PCR buffer, 1.3/~1 dNTP, 7.45 #1 H20 and 0.25/zl TaqI polymerase), and subjected to 11 thermal cycles consisting of 93°C for 1.5 min, 45°C for 1 min and 72°C and 2 min. Finally, 50 t~l of the third PCR mix (5 /~1 10x PCR buffer, 1.5/~g oligo, 6.5 /zl dNTP, 37/zl H20 and 0.5 t~l TaqI polymerase) were added to the second PCR product, heated at 95°C for 3 rain and subjected to 30 thermal cycles consisting of 92°C for 1.5 rain, 500C for 1 min and 72°C for 2 min. YAC-AIu PCR conditions and primers have been described elsewhere

[121. 2.3. Restriction fingerprint analysis Miniprep DNAs were prepared from the Fosmid clones [1], and 1/50 aliquots of them digested with Hind III and Msp I. Using AMV reverse transcriptase (United States Biochemicals) and ot-Pa2-dATP, Hind III ends were specifically labelled, and the resulting fragments were resolved in 4.5% polyacrylamide gel containing 8 M Urea in a 40 crn height sequencing gel box (Bethesda Research Laboratories) at 65 Watts. Lambda phage DNA digested with Hinf I and end-labelled was used as a standard marker. The gels were run until the bromophenol blue dye reached the bottom. The gels were dried and exposed to phosphoscreens, which were then digitized by a Phosphoimager scanner (Molecular Dynamics). The gel images were edited to a size and format suitable for the image analysis programs (Image) ineluded in the contig9 package. The gel images were processed and band data extracted. The band data was transfered to contig9, and Fosmid clones were assembled into a contig. 3. R ~

2.2. Preparation of YAC probes and screening of the Fosmid library Yeast plugs were prepared from YAC strains 45C7

We have constructed a 10 x human chromosome 22specific Fosmid library represented by 18 500 clones. In-

U.-J. Kim et aL / Genetic Analysis: Biomolecular Engineering 12 (1995) 81-84

Fig. 1. Identification of human chromosome 22-specific Fosmid clones by colony hybridization. Total Fosmid clones derived from KG-I cell line were hybridized with l~tuman genomic DNA under a nonsuppressive condition.

itially, 4500 clones were selected from primary Fosmid clones derived from monosomic human-hamster hybrid cells carrying human clhromosome 22 by hybridizing with human genomic DNA (Fig. 1). Interspersed repetitive sequences character:istic of human DNA, especially Alu repeats, served as human-specific markers. Approximately 1% of the total transfectants from the hybrid cell line were positive to human genomic DNA (not shown). The other part of the Fosmid library was prepared from a flow-sorted chromosomal DNA. The library was ar-

O t,,.© Oc~

oO ~©t,,. t-cqO

-60 ~ ~t,,. eO

83

rayed in 96-well microtiter plates, and the first 2/3 of the library (128 microtiter plates) representing approximately 6.5 x coverage of the chromosome was gridded at high density (4 x 4 or 5 x 5 format) onto hybridization filters for efficient screening of the library by hybridization. Judging from the hybridization of the library filters to human DNA probe, at least 80% of the clones in the Fosmid library contained human DNA (not shown). Over 90% of the 100 Fosmid clones randomly chosen from the first part of the library derived from KG-1 were mapped to the subregions of chromosome 22 by fluorescent in situ hybridization, and were shown to distribute to all the chromosomal subregions on chromosome 22 (B. Birren, unpublished). We also identified 911 ribosomal clones from the first 2/3 of the library using a ribosomal probe as described previously [5] (not shown). The first 3 x library that was constructed by screening the total transfectants derived from the hybrid cell line with human genomic DNA probe was almost devoid of ribosomal clones, while at least 11% of the Fosmids derived from flow-sorted chromosomal DNA contained ribosomal repeats. This is due to the difference in the procedures of the library construction. The first 3 x library was picked by looking for Fosmid clones containing human specific Alu repeats from among the total Fosmids generated from human-hamster hybrid cell, using total human DNA as a probe, which represents a very weak ribosomal probe. The rest of the library was directly derived from flow-sorted chromosome 22, thereby without any bias against the ribosomal clones. Library screening with unique probes as well as cosmids, BACs (not shown) and YACs (see on) indicate that the first 2/3 of the library is approximately 6 x deep. The YAC DNA obtained from a pulsed field gel was amplified using the arbitrary oligonucleotide under PCR conditions that allow ubiquitous non-specific

300bp200bp,-

:iiii!ii!¸ili¸

A

"B

C

Fig. 2. (A) Pulsed field gel ele~trophoresis of yeast plugs from YAC strains A226C4 and 45C7. YAC bands were visualized by staining with EtBr. Positions of the YAC bands were inferred by previously known YAC sizes. (B) Amplification product of the YAC clones were run on 1.8% agarose gel and visualized by EtBr staining. Control lane is the amplification product of PCR without added template DNA. For unknown reason, some PCR products were always observed in the control lanes. (C) Confirmation of the amplification products by Southern hybridization with human genomic DNA.

Fig. 3. Autoradiogram of colony filter hybridization with amplified YAC DNA. Both amplification products from the two YACs were combined and used for hybridization at the same time. Only 2 filters from a total of 8 filters are shown.

84

U.-J. Kim et al. /Genetic Analysis: Biomolecular Engineering 12 (1995) 81-84 ~ALUI

CEN

022S~B

I I I

VAC A22SC4 3 0 0 ~

I I

YAC 4SC7

SIAl m

HF

OSM

I (

I

I

I ,

I r

I I

;

I

I

170kb

(eA)n

D22~30

TEL

82D3 ~

31A2 m ~4Ee

51Cll 8CA1

e4~ l 770S m

m

g7D 7 m 12G6

I

95G3

Fig. 4. Fosmid contig assembled by restriction digestion of Fosmid clones identified by amplified YAC DNA probes. Fosmids were fingerprinted as described in section 2, and assembled to initial contigs using contig9. The size of Fosmids as drawn in the figure does not necessarily reflect the actual size, but are proportionate to the number of restriction fragments. Note that there is a gap between two contigs. Localization of Fosmids with respect to the markers and YAC clones were checked by Southern hybridization of the YACs and Fosmids digested with EcoR I or Hind III using Fosmids and STS markers as specific probes. The probe l-Alul was obtained by isolating from agarose gel an AIu-PCR product band that existed in YAC A226C4 but not in YAC 45C7.

priming over the template DNA as described in section 2 (Fig. 2). Amplified DNA was labelled and used to screen Fosmid grid filters (Fig. 3). Nearly 100 clones with moderate to strong positive signals were picked, and only some of the clones with strong signals were subjected to restriction fingerprint analysis. Fig. 4 shows two Fosmid contigs covering the 300 kb region spanned by a YAC clone A226C4. The clones were preassembled into contigs via the contig9 program. The overlap relations of some of the Fosmids in the contig were separately confirmed by manually performing Southern hybridizations against each other after digesting the clones with EcoR I or Hind III (data not shown). The relative position of YAC, the clones as well as the position of the markers on the map were also determined by Southern analysis of EcoR I and Hind III fragments of the YACs and the Fosmids (not shown). From the Fosmid contig, a new microsatellite marker D22S430 was established [13]. 4. Discussion YAC clones provide efficient means to rapidly identify underlying clones from a cosmid library [14]. As global YAC contig maps over the entire genome emerge [15], relatively large YAC clones that have been mapped to specific genomic loci will provide efficient means to screen genomic libraries with more tractable and stable clones. We have demonstrated the stability of Fosmid and BAC cloning systems [1,2]. The human chromosome 22-specific Fosmid library that we have constructed will be useful for detailed characterization of human chromosome 22, especially if they have been ordered into contigs.

We demonstrated the possibility of rapidly generating a sequence ready Fosmid contig by using YAC clones as probes to identify corresponding Fosmid clones and assembled them into contigs using a restriction fingerprint analysis method. We have also been successful in screening a human genomic BAC library with YAC probes (Kim, unpublished). Since it is cumbersome to identify and cut the YAC bands from agarose gels, we have attempted to use total yeast DNA containing YAC DNA and AIu-PCR products of YAC clones with limited success. While the probes were easier to prepare, significant percentages (up to 50%) of the clones identified by these probes were false positives (unpublished). An alternative to YAC to Fosmid would be Fosmid to YAC, or screening YAC clone grids with Fosmids. In this scheme, Fosmids from the library are randomly assembled into multiple contigs using the contig9 scheme. One Fosmid from each contig is then used to probe the YAC grid filters and identify corresponding YAC clones, allowing rapid localization of a number of Fosmid contigs to specific subregions defined by the YAC contig maps. Resulting Fosmid contigs will provide materials for efficient genomic sequencing. Acknowledgments We thank Dr. Marco Giovannini for providing us with YAC clones and two cosmids containing LIF and OSM genes, and Dr. Guy Rouleau for D22S268 marker. References [1] Kim U-J, Shizuya H, de Jong P, Birren B, Simon MI. Nucleic Acids Res 1992; 20: 1083-1085. [2] Shizuya H, Birren B, Kim U-J, Mancino V, Slepak T, Tachiiri Y, Simon MI. Proc Natl Acad Sci USA 1992; 89: 8794-8797. [3] Sulston JE, Mallet F, Staden R, Durbin R, Horsnell T, Coulson A. CABIOS 1988; 4: 125-132. [4] Sulston JE, Mallet F, Durbin R, Horsnell T. CAB1OS 1988; 5:101-106. [5] Kim U-J, Shizuya H, Birren B, Slepak T, de Jong P, Simon MI. Genomics 1994; 22: 336-339. [6] Jones C, Kao F'r, Taylor RT. Cytogenet Cell Genet 1980; 28: 181-194. [7] Trask B, van den Engh G, Christensen M, Massa H, Gray JW, van Dilla MA. Somatic Cell Mol Genet 1991; 17L: 117-136. [8] van den Engh G, Trask B, Cram S, Bartholdi M. Cytometry 1984; 5: 108-117. [9] Langlois RG, Yu LC, Gray JW, Carrano AV. Proc Natl Acad Sci 1982; 79: 7876-7880. [10] Trask BJ, van den Engh G, Christensen M, Massa HF, Gray JW. Somatic Cell Mol Genet 1991; 17: 117-136. [11] Hadano S, Watanabe M, Yokoi H, Kogi M, Kondo I. Genomics 1991; i h 364-373. [12] Lengauer CH, Green ED, Cremer T. Genomics 1992; 13: 826-828. [13] Sainz J, Nechiporuk A, Kim U-J, Simon MI, Pulst SM. Hum Mol G-enet 1993; 2: 2203. [14] Baxendale S, Bates GP, MacDonald ME, Gusella JF, Lehrach H. Nucleic Acids Res 1991; 19: 1665. [15] Bellannechantelot C et al. Cell 1992; 70: 1059-1068.