A Human Genomic Library Enriched in Transcriptionally Active Sequences (aDNA Library)

Library Construction

The dinucleosomal DNA was dephosphorylated and treated with both T4 DNA polymerase and Klenow to achieve blunt ends (Satchwell and Travers 1989). Adapters (10 bp) were ligated onto the DNA, and the adapted DNA coprecipitated with λZap arms before ligation and in vitro packaging. The total number of phage was 8.3×10⁵ plaque-forming units (PFU). This number is ample for an analysis of individual clones to assess the representation of different types of DNA sequence in the aDNA library.

A random selection of 103 plaques were cored from a plate carrying ∼50,000. pBluescript phagemids were rescued by in vivo excision. Insert fragments were then released by EcoRI digestion and separated on agarose gels, allowing an estimation of insert sizes: The majority were in the range of 200–400 bp. In a separate experiment, a random selection of 77 plaques were taken from a plate of 2000 clones. Phagemid DNA was rescued as before, and insert DNA segments were sized by PCR amplification of bacterial colonies using the T7 and T3 vector primers, followed by separation on agarose gels. Combining the results of both experiments showed that a total of 170 clones (out of 180) contained insert DNA, 101 (61%) of which had an insert size between 300 bp and 500 bp, and 28 (17%) contained inserts between 200 bp and 300 bp.

The next step was to estimate the proportion of clones containing highly repetitive DNA, using hybridization to total genomic DNA. Hybridization to total genomic DNA has been estimated to detect sequences present in 50 or more copies in the genome (Shen and Maniatis 1980). Of the 103 clones tested in this manner, 24 showed significant hybridization signals and were rejected. Because 5 of these 103 clones had already been rejected as having no insert DNA, it follows that 74 (72%) of this set of clones contain low copy number sequences. In an alternative approach to defining the proportion of clones containing repeated sequences, a further set of ∼2000 plaques were transferred to a nylon membrane and hybridized with sonicated, labeled human genomic DNA. About 12% of plaques were found to show visible signals on autoradiography. This is about half the proportion found in the above Southern hybridizations, probably because the lower amount of DNA in a single plaque reduces the sensitivity of detection by hybridization. Because ∼40% of human genomic DNA is of moderate to high repetition, significant preferential rejection of such sequences has taken place in the immunoselection process.

Sequence Analysis

The 74 clones identified as low copy number were manually sequenced using T7 and T3 primers to sequence both strands. The second set of 77 randomly selected clones were sequenced by single-pass cycle sequencing using a T7 primer on an ABI 377 automated sequencer at the Human Genome Centre, Hinxton, Cambridge, UK. Five (7%) of this set were found to contain no insert following sequencing. A total of 74 + 72 = 146 sequenced clones were then analyzed further.

Identification of aDNA Clones Deriving from Transcribed Sequences

The sequences of the 146 aDNA library clones were compared with both the nucleic acid and protein databases, for the latter in all six frames. Table 1 gives the size and GenBank accession number for each sequenced clone together with the type of matching sequence found. Of the 74 clones deemed not to contain repeated sequences following hybridization to total genomic DNA, 4 were found to include Alu repeats and 5 to include L1 repeats, in each case quite short. Sixteen clones (11%, as of August 15, 1999) correspond to gene sequences defined previously, and these are listed separately in Table 2. Because the aDNA clones are comparatively short, we cannot exclude the possibility that some may derive from pseudogenes. The region of match for the large majority of these 16 clones is upstream of the coding sequences, that is, in a region expected to carry acetylated histones if the gene is active, poised, or was active previously.

Because acetylated H4 histones are expected to be concentrated at active genes, the most important question to be answered regarding the aDNA library is the proportion of clones that truly derive from genes active in K562 cells. Library inserts were therefore used to analyze the mRNA population of K562 cells. Highly expressed genes might be detected by Northern analysis, but genes expressed at low level require the sensitivity of RT–PCR amplification. aDNA clones derived from promoter or intronic sequences that carried acetylated histones would not of course be present in the mRNA population: These approaches therefore underestimate the number of clones derived from active genes.

Northern Analysis

A random selection of 34 clones from the 146 sequenced clones, none of which included highly repeated DNA, was chosen for Northern analysis. It has been estimated that mRNA accounting for ∼0.01% of the total population can be detected by Northern blots using 10 μg of total RNA per lane (Brown 1994). Only one insert, ALP7, showed a significant hybridization signal (at ∼3 kb; data not shown). The aDNA library is thus not dominated by the DNA of highly transcribed genes, as might be the case if the density of acetylated histone H4 on a gene were directly related to the frequency of transcription of the gene.

RT–PCR Amplifications

Forty clones from the aDNA library were chosen for this analysis (see Table 3), only two of which displayed sequence matches with cDNA fragments in the EST database at the time of the experiments. The remaining 38 were chosen from those showing no significant matches to sequences in any of the nucleic acid databases at that time. Computer analyses of the cloned sequences using GRAIL, GENSCAN, Fgeneh, polyAH, and CpG programs were used to direct the choice of further clones to test by RT–PCR. Eleven clones that were predicted to contain exon sequences were chosen and a further seven because they contained predicted poly(A) addition sites (possible 3′ gene fragments). Nine were chosen as potential 5′ sequences because they included predicted CpG island regions. Because there was some overlap of these three predicted features among these clones, they represent a total of 20 clones (see Table 3, column 2). The remaining 18 were randomly chosen from clones not predicted to contain any of the above three features.

Total cytoplasmic RNA from K562 cells was used as template in RT–PCR amplifications. First strand cDNA synthesis was first performed using oligo(dT) priming, but because the likely location of acetylated histones is at the 5′ end of active genes, first strand synthesis was also separately performed using random hexamer primers. Primer pairs for PCR amplification of first strand cDNA for the 40 chosen clones were designed to give products ranging from 100 bp to 230 bp and annealing temperatures were optimized. Amplification from three different templates was separately carried out for each primer pair: genomic DNA, first strand cDNA from oligo(dT) priming, and first strand cDNA from random priming. In addition, a mock RT reaction was carried out for both the oligo-d(T) and random primed reactions omitting the RT enzyme, to check for possible carryover of genomic DNA. In no case was an amplicon obtained when the RT enzyme was omitted. In some cases an amplified product was obtained from both first strand cDNAs but that from the oligo(dT) primed reactions was always more intense than from the random primed: The random primed material was therefore not used further. Figure 3 compares the amplification products from genomic DNA and oligo(dT) primed cDNA for 12 of the 40 selected aDNA clones. These 12 gave essentially single and correctly sized amplification products from both templates: All 12 aDNA clones are therefore derived from sequences transcribed in K562 cells and do not span splice junctions. Two of these, ALP27 and ALP71, represent sequences that match known cDNA fragments in the GenBank nonredundant nucleic acid and EST databases, respectively, at the time of the experiments but it was not known whether these were expressed in K562 cells. ALP80 has since been shown to match a new entry to the expressed sequence tag database (dbEST). A further 15 of the 40 clones gave correctly sized amplification products from the genomic DNA template but yielded incorrectly sized single products or multiple products from the cDNA template. Of these 15 clones, 5 display high identity to gene sequences in the nonredundant nucleic acid and EST databases (ALP31, ALP35, ALP74, ALP107, and ALP168). Failure to generate a correctly sized product from the cDNA template might be due to one (or both) primer being located in introns; so the RT–PCR experiment must be regarded as inconclusive for these 15 clones. For the remaining 13 clones, all the primer pairs but one generated the correctly sized products with genomic DNA as template but yielded no product at all with the cDNA template. Of these 13, two clones (ALP120 and ALP143) correspond to known gene sequences (Table 2). The primer pair for the final clone (ALP96) yielded no products from either genomic or cDNA templates, although this clone displays high sequence identity to a cDNA fragment in dbEST.

Figure 3.

Four percent agarose–TBE gel electrophoresis of PCR amplicons generated using oligo(dT) primed reverse transcribed total cytoplasmic RNA (RT) and genomic DNA (G) as template. Nucleic acids were extracted from K562 cells. PCR was performed using library clone specific primers. M denotes a 50n-bp DNA size marker. All clones show the presence of a single PCR amplicon of the expected size using both DNA templates. (←) The expected amplicon size.

Open in new tabDownload PowerPointLink to figure

Overall therefore, we find that 12 of 40 (30%) of this selection of aDNA clones are derived from sequences that are actively transcribed in K562 cells. This figure of 30% is much higher than would be observed if the library were a random collection of dinucleosomes from the human genome. Of these 12, 9 clones (ALP5, ALP15, ALP32, ALP60, ALP79, ALP148, ALP155, ALP157, and ALP169) are not currently present in any nucleic acid or protein database and thus represent newly defined gene sequences, transcribed in K562 cells. It is noteworthy that 7of these 9 expressed clones were among the 18 selected for testing by virtue of their lack of computer predicted gene features; thus, 7 of 18 clones (39%) having no gene features turned out to be from expressed genes. These results indicate that the aDNA library is rich in expressed sequences.

Chromosomal Mapping

To decide whether the aDNA library exhibited any strong bias toward particular chromosomes or parts thereof and to investigate the use of library clones as STSs, we attempted to map the 40 clones to individual chromosomes. The Corriel somatic cell hybrid mapping panel (no. 2) was used, which consists of 24 human–mouse monochromosomal hybrids representing human chromosomes 1–22, X, and Y in mouse cells. All primer pairs were tested for their ability to generate amplicons from human but not mouse or hamster genomic DNA before use with the hybrid panel DNA. Seven of the 40 primer pairs (ALP31, ALP32, ALP35, ALP46, ALP157, ALP160, and ALP168) were rejected at this stage because, despite varying PCR conditions, they either did not yield a clear simple amplicon from human DNA or they readily amplified mouse or hamster genomic DNA. The ALP14 primer pair was rejected following chromosomal assignment assays because it yielded an abundant product from total human genomic DNA but nothing from any of the hybrid cells. It seems that the segment of human DNA that includes clone ALP14 is not included in any of the hybrid cells. The remaining 32 primer pairs gave successful chromosomal assignments.

Table 3 (column 4) gives the chromosomal assignments: It is apparent that there is no strong preference for any individual chromosome or chromosomes among the 32 tested clones. Because K562 cells are a chronic myeloid lineage derived from leukemic cells of a female patient, they do not carry a Y chromosome. Four clones derive from each of chromosomes 1, 16, and 22: Chromosomes 1 and 22 are known to be gene rich. Chromosomes 4 and 21 have no clones assigned to them and are known to be gene poor. But with a total of only 32 assignments, no statistical significance attaches to this chromosome distribution.

More detailed chromosomal mapping was performed for these 32 clones using radiation hybrid (RH) mapping with duplicate assays against the reduced Genebridge 4 RH panel. Four of these 32 primer pairs were rejected because of ambiguous PCR amplifications with the RH panel. Chromosomal positioning data were thus obtained for 28 of the 40 clones tested for expression, and these are given in column 5 of Table 3 as map position together with chromosome length in centiRays.

Preparation of Dinucleosomes

Human K562 cells were harvested by centrifugation at 150gfor 10 min at 4°C. Cell pellets were resuspended at ∼2×10⁸ cells/ml in 250 mm sucrose, 10 mm Tris-HCl (pH 7.4), 10 mm sodium butyrate, 4 mm MgCl₂, 0.1 mm PMSF, 0.1 mmbenzamidine, and 0.1% Triton X-100 and lysed with 15 strokes of a Dounce homogenizer. Nuclei were pelleted by centrifugation, resuspended in the above buffer omitting Triton X-100 and purified by centrifugation through 6 ml of 30% (wt/vol) sucrose cushions at 2400g for 10 min at 4°C. Nuclei (∼25 mg of DNA) were resuspended in digestion buffer (10 mm NaCl, 10 mmTris-HCl at pH 7.4, 10 mm sodium butyrate, 3 mmMgCl₂, 1 mm CaCl₂, 0.1 mm PMSF, 0.1 mm benzamidine, 0.15 mm spermine, and 0.5 mm spermidine) to a final concentration of 5 mg/ml DNA. Varying amounts of MNase were added, and incubation was performed at the required temperature; reactions were terminated by addition of Na₃–EDTA to 5 mm. Nuclei were pelleted, and the supernatant (S1) retained. Pellets were resuspended in lysis buffer (10 mm Tris-HCl at pH 7.5, 10 mm sodium butyrate, 0.25 mm Na₃–EDTA, 0.1 mm PMSF, and 0.1 mm benzamidine), incubated on ice for 30 min, and recentrifuged, retaining the supernatant (S2). Nucleosome oligomers were separated by centrifugation of ∼2 mg of chromatin through 5%–30% exponential sucrose gradients containing 20 mmTris-HCl (pH 7.5), 20 mm sodium butyrate, 0.25 mmNa₃–EDTA, 0.1 mm PMSF and 0.1 mmbenzamidine at 36,000 rpm for 18 hr at 4°C in a Beckman SW40 rotor.

Immunofractionation of Dinucleosomes

Immunofractionation of purified dinucleosomes pooled from S1 and S2 was performed essentially as described by Hebbes et al. (1994). In brief, 400 μg dinucleosomes (as DNA, Input chromatin) in 10 mm Tris-HCl (pH 7.5), 50 mm NaCl, 10 mmsodium butyrate, 0.2 mm EDTA, 0.1 mm PMSF, 0.1 mm benzamidine were added to ∼100 μg of affinity purified anti-acetylated H4 antibody and incubated for 2 hr at 4°C with constant agitation. Immunocomplexes were isolated by the addition of 50 mg of prewashed protein A–Sepharose beads (Pharmacia), Unbound chromatin was removed, and following extensive washing, immunocomplexes (Bound) were released with 1.5% SDS. DNA was precipitated after phenol extraction, and proteins were recovered from the phenol phase by acetone precipitation.

aDNA Library Construction

All DNA manipulation steps below were followed by heat inactivation (10 min at 65°C unless otherwise stated), phenol–chloroform extraction, and ethanol precipitation to recover the DNA. Antibody-Bound dinucleosomal DNA fragments were first blunt-ended: Two micrograms (6.9 pmoles) of DNA was dephosphorylated in 10 mmTris-acetate, 10 mm magnesium acetate, 50 mmpotassium acetate with 1 unit of calf intestinal alkaline phosphatase (CIP; Pharmacia) per pmole of DNA, incubating for 45 min at 37°C. Further CIP was added to a final concentration of 1.5 U/pmole DNA, and incubation continued for 45 min at 55°C. CIP was heat inactivated at 85°C for 15 min. DNA pellets were resuspended in 20 μl of 50 mm Tris-HCl (pH 8.0), 10 mm MgCl₂, 50 μm each dNTP, and 10 units each of T4 DNA polymerase and Klenow (Pharmacia) and incubated for 3 hr at 20°C.EcoRI– NotI adapters were ligated to the blunt-ended DNA using a modified method from the Time Saver cDNA Synthesis Kit (Pharmacia). DNAs were then phosphorylated and end-labeled by addition of 10 μCi of [γ-³²P]dATP (4500 Ci/mmole, ICN), 5 units of T4 polynucleotide kinase and incubated for 5 min at 37°C. Phosphorylation was completed by the addition of 1.5 μl of unlabeled dATP (75 mm) and incubation for a further 25 min at 37°C. Adapted DNA was separated from excess and dimerized adaptors using a 3×270 mm Sephacryl S-300 column in 0.3m sodium acetate, 10 mm Tris-HCl (pH 8.0), and 1 mm EDTA, and fractions containing adapted DNA were pooled and precipitated. Adapted DNA fragments and λZap arms were coprecipitated with ethanol and resuspended in ligation buffer (66 mmTris-HCl at pH 8.0, 150 mm NaCl, 6.6 mmMgCl₂, 0.1 mm spermidine, 10 mm DTT). ATP was added to 0.1 mm together with 5 units of T4 DNA ligase in a final volume of 10 μl. Samples were incubated for 30 min at 16°C, a further 5 units of T4 DNA ligase were added, and incubation continued overnight at 4°C. Ligation mixtures were then separately added to Gigapack Gold III packaging extracts (Stratagene) and incubated for 2 hr at 24°C. Two hundred microliters of SM buffer [100 mm NaCl, 8.1 mm MgSO₄, 50 mm Tris-HCl (pH 7.5), 0.01% (wt/vol) gelatin] and 20 μl of chloroform were added to each, mixed, and centrifuged briefly to sediment cell debris.

In Vivo Excision of Library Phage

Single plaques were cored from plates and added to 250 μl of SM buffer with 4% chloroform and incubated for 2 hr at room temperature and then overnight at 4°C. Aliquots (200 μl) of phage samples were added to 200 μl of host XL1-Blue cells in 10 mmMgSO₄ and infected with F₁ helper phage R408 (1×10⁸ plaque-forming units). Samples were incubated for 15 min at 37°C, before the addition of 5 ml of Luria broth, and incubated for 4 hr at 37°C in a rotary shaker. Bacteria were heat-killed by incubation for 20 min at 68°C, and cell debris was pelleted by centrifugation. Supernatants containing the packaged pBluescript phagemids were stored at 4°C.

Amplification of Library Sequences for Sizing and Sequencing

Cloned dinucleosomal DNA sequences were amplified by PCR directly from bacterial colonies for use as template during automated sequencing. Single colonies of bacterial cells “picked” from agar plates were lysed by incubation at 100°C for 10 min and quenched on ice. Aliquots of these samples were subjected to hot start PCR amplification using T3 and T7 primers and 5 units of AmpliTaq DNA polymerase (Perkin-Elmer) for 25 cycles of 94°C for 30 sec, 50°C for 30 sec, and 72°C for 60 sec. Amplification products were size-analyzed by electrophoresis on 1.5% agarose gels and treated with 10 units of exonuclease I and 2 units of shrimp alkaline phosphatase by incubation at 37°C for 30 min before sequencing.

DNA Sequencing

Manual sequencing reactions were performed using the ΔTth DNA Polymerase Sequencing Kit (Cambridge Biosciences, UK) using [α-³⁵S]dATP with T7 and T3 primers. Samples were analyzed on 8% polyacrylamide sequencing gels. Automated sequencing reactions were performed at the Human Genome Mapping Project Resource Centre, Hinxton, UK, using the ABI Prism Fluorescent Dye Terminator Ready Reaction Kit with the T7 primer. Samples were analyzed on 4% polyacrylamide sequencing gels using an ABI 377 automated sequencer.

Preparation of Total RNA from K562 Cells

K562 cells (10⁷–10⁸) were pelleted by centrifugation at 4°C, washed twice in sterile PBS, and lysed by resuspension in 2 ml of 4 m guanidinium isothiocyanate, 25 mm sodium citrate (pH 7.0), 0.5% N-lauryl sarcosine, 0.1% antifoam A (Sigma), and 0.1 mmercaptoethanol. Chromosomal DNA was sheared by repeated passage of the lysate through a 23-gauge sterile needle. Lysates were layered onto an equal volume of 5.7 m caesium chloride/0.1 m EDTA cushions, RNA pelleted by centrifugation at 30,000 rpm for 18 hr at 15°C, and washed twice with the guanidinium thiocyanate solution. Contaminating genomic DNA was removed by treatment with RNase-free DNase I, and samples were stored at 80°C.

Northern Analysis

Aliquots (20 μg) of total cellular RNA extracted from K562 cells were dissolved in 10 mm phosphate buffer (pH 7.0) and 0.88 m glyoxal and electrophoresed in 1.2% agarose gels in the same buffer. Transfer to Biodyne A membranes (Pall) was by capillary action in 20× SSC, and RNA was fixed by baking at 80°C for 1 hr. A 397-bp GAPDH probe was used to confirm effective transfer by observation of the 1.4 kb GAPDH message. Following excision byEcoRI and purification, 34 aDNA library inserts were labeled with ³²P by random priming and used to probe the membranes.

RT of Total RNA

Five hundred nanograms of oligo(dT)_12–18 or 50 ng of random hexameric oligonucleotides was added to 5 μg of total cellular RNA from K562 cells, and nucleic acids were denatured by incubation for 10 min at 70°C and chilling on ice. After adding reaction mix, primer annealing for oligo(dT) reactions was at 42°C for 5 min and for random hexameric reactions was at 25°C for 5 min. Two hundred units of SuperScript II RT enzyme (GIBCO) was added and incubated for 50 min at 42°C. Mock reactions omitting the RT enzyme were also performed for both the oligo(dT) and random hexameric primed reactions. RNA was removed from the cDNA by incubation with 2 units ofEscherichia coli RNase H for 20 min at 37°C.

Amplification of First Strand cDNA

PCR amplification of the first strand cDNA used primer pairs designed from library clones: primers for the GAPDH gene were used as a positive control. Thirty cycles of 94°C for 1 min; 50°C–62°C for 1 min, and 72°C for 1 min were used with 5 units of AmpliTaq (Perkin-Elmer). Annealing temperatures were optimized for each primer set.

Amplification of Genomic DNA

DNA was prepared from K562 nuclei essentially as described byMiller et al. (1988). PCR amplification of 100-ng aliquots of genomic DNA used primer pairs designed from library clone sequences, essentially as described above.

RH Mapping and Chromosomal Assignment

PCR reactions were performed at the RH mapping laboratories of the Sanger Centre, Cambridge using Peltier Thermal Cycle (PTC) 225 DNA Engine Tetrads (MJ Research). Primer pairs were “prescreened” against 50 ng of human placental, hamster CHO cell and mouse BALB/c genomic DNA templates. Annealing temperatures and primer levels were optimized before chromosome assignment and RH mapping. Chromosome assignments were carried out using the Corriel somatic cell hybrid mapping panel no. 2 using hamster, mouse, and 0.1× TE buffer controls. Hybrid samples (30 ng) were added to the chromosome reaction mix, and PCR amplifications were performed under the conditions optimized for each primer pair during prescreening. For RH mapping, primer pairs were assayed in duplicate against the Genebridge 4 RH mapping panel. A reduced panel containing 85 hybrids was used, the controls being human and hamster genomic DNA and 0.1× TE buffer. PCR reactions were performed under the optimized conditions determined during chromosomal assignment and prescreening. PCR products were analyzed on 1.5% agarose–TBE gels and visualized over UV light. Images of each gel were digitized using a Kodak Megaview high-resolution camera and a Macintosh based image-capture system. Resulting RH retention patterns were then analyzed by the RHMAPPER program (Slonim et al. 1996) using the 1998 map framework (Deloukas et al. 1998).

Nucleic Acid Electrophoresis, Transfer, and Hybridization

Total cellular RNA was electrophoresed in 1.2% agarose gels containing 10 mm sodium phosphate (pH 7.0) with recirculation of running buffer. The RNA was pretreated with 880 mmdeionized glyoxal for 1 hr at 50°C before addition of loading buffer. Northern and Southern blot transfers used 20× SSC high salt transfer buffer (0.3 m sodium citrate, 3 m NaCl) with Biodyne B membranes (Pall) for DNA and Biodyne A membranes for RNA. Slot blotting and hybridizations were as in Hebbes et al. (1994).