Detection of small amounts of human DNA in human−rodent hybrids

When a cell with a genome fragmented by irradiation is fused with a normal unirradiated cell, the resulting hybrid undergoes extensive loss of the genetic material of the irradiated parent, and only a small chromosomal fragment may be retained: this is the basis of a method of gene mapping (Goss & Harris, 1975, 1977)-Cell hybrids made by chromosome-mediated gene transfer have a similar end result (McBride & Ozer, 1973; Miller & Ruddle, 1978). In both of these cases, the usual way of determining the total contribution of chromosomal material of each parent to the hybrid, by karyotyping, may be applicable either with difficulty or not at all.

Molecular hybridization is a method commonly used for detecting and quantitating viral DNA and other specific sequences (e.g. Y-chromosome-specific DNA; Kunkel et al. 1977) in cells. The method has been used occasionally to measure total foreign DNA (as opposed to specific sequences) in cell hybrids (Coon, Horak & Dawid, 1973; Jones, 1977), although the sensitivity was not very high, 2–5% equivalent genomes of foreign DNA. It was thought worthwhile to see how far hybridization methods could be useful as a measure of the heterologous DNA in animal cell hybrids. The results show that filter hybridization will detect the equivalent of 0·5% of a human genome, and reassociation kinetics the equivalent of 0·005–0·01%, in a rodent × human cell hybrid. The reassociation method also gives information about sequences. Application of the method to several cell hybrids containing human DNA fragments is described.

A9, an HPRT⁻ derivative of the mouse L cell, was normal laboratory stock.

Ag-Daudi clones. Various clones derived from a Sendai virus fusion of A9 with Daudi, a Burkitt lymphoma-derived B-type lymphocytic line have been described (Allderdice et al. 1973). Clones were from the laboratory stock and were karyotyped co-incidentally with these experiments (see Results).

Wg3-h is a Chinese-hamster DON line derivative lacking HPRT.

R₄93 is a hybrid cell, Wg3-h x irradiated human lymphocyte, made by the Sendai virus method. The lymphocyte had received 40 J kg^-1γ-rays before cell fusion.

R₄93(6TG) was R493, back selected in 6 μg/ml 6-thioguanine.

R₆2 is similar to R493 except that the lymphocyte had received 60 J kg⁻¹γ-rays before fusion. Chinese-hamster × lymphocyte fusions were originally made by Dr Stephen Goss.

PG19 is a HPRT” cell line derived from a spontaneous melanoma in a C57 Black mouse (Jonasson, Povey & Harris, 1977).

B₂ and Di were clones from a PG19 x MRC-5 (human fibroblast) fusion in which the MRC-5 had received 200 J kg⁻¹γ-rays before fusion. These fusions were originally made by Professor H. Harris.

Cl19S16TG is a PG19X human lymphocyte hybrid cell, described by Jonasson & Harris (J977)-

Cell lines (except Raji) were grown in monolayer culture in MEM. Hybrid cells were grown similarly but with the addition of HAT.

Raji cells were grown in suspension culture in DMEM with 20 % foetal calf serum.

Cell lines or hybrid cells were grown in roller bottles at 37 °C, trypsinized from the surface, washed twice with ice-cold PBS (20 ml per ml packed cell volume) and resuspended in the same volume of PBS. 1 vol. of sarkosyl-EDTA-Tris (3 5% sarkosyl +0·075 M Tris pH 7·5, 0·025 M EDTA) was added and the mixture stood overnight at room temperature. It was then heated at 65 °C for to min followed by incubation with 20 μg/ml heated RNase for 30 min at 37 °C, and then pronase (1–2 mg/ml, 3 h at 37 °C). CsCl was added to give a density of 1700 g cm^-3 and the solution centrifuged to equilibrium at 20 °C. DNA-containing fractions were dialysed against 1 × SSC (1 × SSC is 0·-15 M NaCl-f-0·015 M trisodium citrate pH 7·2) and stored at —70 °C. DNA for reassociation was further treated by degradation in 0·3 M NaOH at 100 °C for 15–20 min (Sharp, Pettersson & Sambrook, 1974) followed by dialysis. DNA concentrations were measured by the diphenylamine method of Burton (1956).

Sheared human placental DNA (Calbiochem) was reassociated to C₀t 1 in 0·12 M phosphate buffer at 60 °C and isolated on hydroxylapatite (HAP). C_ot 0·5–10 DNA was prepared in the same way. 5 μg reassociated DNA were transcribed with E. coll RNA polymerase (Boehringer Corp. Ltd) in 40 mM Tris-HCl, pH 7·9, 0·15 M KC1, 10 mM MgCl₂, 0·1 mM EDTA, 0·1 mM 2-mercaptoethanol and 0·15 mM GTP. 100 μCi of tritiated ATP, CTP, UTP were added, incubated at 37 °C for 3 h and the product isolated on Sephadex G50. It was freeze-dried and stored at —70 °C.

DNA (20 μg per filter) in 0·1 x SSC was denatured by adding 1 vol. 0 5 M NaOH and standing at about 30 °C for 15-20 min. It was neutralized by adding 2 vol. ice-cold neutralization mixture (12 x SSC : 1 M Tris-HCl pH 7-1 : distilled water : cone. HC1, too : 60 : 140 : 3). The neutralized DNA was applied to the filter under slight vacuum and the filter washed with 10 ml 6 x SSC: it was then removed from the filtration apparatus, air-dried for 3 h and baked at 80 °C in vacuo for 18 h. Filters were prepared in triplicate. Filters for hybridization were preincubated for 1 h at 37 °C in 50 % formamide+ 6 × SSC, the preincubation buffer removed and cRNA (90000 cpm) in the same buffer was added. Hybridization was at 50 °C for 48 h, after which the filters were washed (twice each time) with 50% formamide — 6 × SSC, 6× SSC, 2 × SSC, and then incubated with RNase (25 μg/ml final concentration) in 2 × SSC for 60 min at room temperature. Filters were washed in a large volume of 12 × SSC. Filters were then heated at 70 °C for 45 min in 1·0 M perchloric acid to hydrolyse the nucleic acid, the O.D.200 of the hydrolysate measured, and counted in a scintillation counter. Counts were corrected to a known DNA concentration from the O.D.₂₀₀.

Probes of lower repetitiveness (C_ot 1·10) were nick translated after isolation of the fraction by 2-3 repeated reassociations. Probes of higher repetitiveness (C_ot 0·1) were usually made by nick translation of total DNA and isolation of the probe by reassociation to the appropriate C₀t.

Nick translation followed the method of Rigby, Dieckmann, Rhodes & Berg (1977) or Maniatis, Jeffrey & Kleid (1975). 5 μg DNA were nick translated using DNA polymerase I (Boehringer Corp. Grade I or Miles Lab. Ltd): the DNA did not need to be nicked before the reaction. The product of the nick translation was isolated on HAP at room temperature and either used (C_at 1-10) or further processed by reassociation (C_ot 0–1). The final specific activity was > 10⁷ cpm/ μg DNA.

1 × SSC is 0·15 M NaCl + 0·015 M trisodium citrate: phosphate buffer (PB) is an equimolar mixture of Na₂HPO₄ and NaH₂PO₄, pH 6·8: 40% FSSPE buffer is 40% (v/v) formamide+ 1 M NaCl+ 1 % SDS + 0·12 M PB + 0·1 mM EDTA.

It was clear at the outset that since no single DNA sequence was known to be widely distributed in the genome, the hybridization probe would be heterogeneous and contain a spectrum of sequences. Since much evidence suggests that some repetitious sequences at least are interspersed with unique sequence DNA in a wide variety of eukaryotes, both simple (Kram, Botchan & Hearst, 1972; Davidson, Hough, Amenson & Britten, 1973; Graham, Neufeld, Davidson & Britten, 1974; Davidson, Hough, Klein & Britten, 1975; Manning, Schmid & Davidson, 1975) and mammalian (Pearson, Wu & Bonner, 1978; Saunders et al. 1972; Marx, Allen & Hearst, 1976; Deininger & Schmid, 1976), these sequences seemed most appropriate as the basis of the probe.

Therefore, the properties of a cRNA probe prepared by transcribing isolated C_ot 0–1 human placental DNA with E. coli RNA polymerase and ³H-nucleotide triphosphates (see Methods) were investigated. Human placental DNA was used as the origin of the probe rather than cell DNA to avoid any special relationship between the probe and the DNA to be analysed.

The relationship between the amount of human probe hybridized and the amount of human DNA in a cell hybrid was tested in the following way. DNA was isolated from several mouse (A9, HPRT⁻, fibroblast) × human (Daudi, Burkitt lymphoma derived B-lymphocyte) clones of known karyotype. The karyotype analysis (by banding) was kindly provided by Mr M. Burtenshaw and Dr E. P. Evans. 20 μg A₉-Daudi DNA and human DNA (separately) were denatured, bound (in triplicate) to 13-mm nitrocellulose filters and hybridized in 6 × SSC —50% formamide at 50 °C for 48 h, with cRNA to C_ot 1 human DNA (9 × 10⁴cpm). Time-course experiments (not shown) indicated that there was no additional hybridization after 48 h. The amount of human DNA in the hybrid cells was calculated from the karyotype and published measurements (Bosman, van der Ploeg, van Duijn & Schaberg, 1977) of the DNA content of each human chromosome. These values were compared with the DNA content derived from hybridization using human DNA standards. Previous experiments (unpublished) had shown that under the conditions of the hybridization there was little cross-hybridization between cell DNA and Epstein-Barr virus DNA (which the hybrid cells contain), and the hybridization was not related to the viral DNA content. The EBV DNA content of the Ag-Daudi clones did not parallel their human DNA content.

The results are shown in Table 1. About 4·2% of the added counts hybridized to total human DNA. There are 2 likely reasons for this: (1) the very stringent hybridization conditions; and (2) the heterogeneous nature of the probe, which may contain up to 50% unique and non-hybridizing sequences.

Table 1 shows that there is quite good agreement between the DNA content derived from cRNA hybridization and from karyotyping. This table also shows that at the whole chromosome level, the sequences measured are not chromosome specific. Thus CL6, whose human DNA content was derived almost entirely from human chromosomes 13,10 and 3, was as accurately estimated as P₃ whose DNA was derived from chromosomes 21, 22, 5, 3 and 1 and very small numbers of most of the others.

This suggests that satellite DNAs are not the major components being hybridized since they tend to be associated with specific chromosomes (Gosden et al. 1975). There are undoubtedly other fast-reassociating sequences with similar properties but different base sequences in the probe.

The method was extended to lower amounts of human DNA by examining the hybridization of cRNA with mouse or Chinese hamster DNA containing various percentages by weight of human DNA. 20-μg DNA mixtures were bound (in triplicate) to nitrocellulose filters and hybridized under the same conditions as before.

Fig. 1 shows the relationship between the percentage of human DNA (by weight) and hybridized counts in human-mouse or human-Chinese hamster DNA mixtures. A linear relation clearly extends down to about 99·5% mouse DNA+0·5% human DNA mixtures. Chinese-hamster-human DNA mixtures gave similar results (inset). The sensitivity of the method can be increased further by increasing the concentration of probe, although the cross-hybridization with mouse and hamster DNA increases also; 0·2% of human DNA is probably the least that this method will detect.

The filter hybridization method has been applied to several somatic cell hybrids. The cells were grown as described in Methods and DNA isolated, denatured and bound to nitrocellulose filters. Calibration mixtures of known amounts of human and Chinese-hamster or human and mouse DNA were bound to filters at the same time as the hybrid cell DNA. Filters were hybridized with 9 × 10⁴ cpm cRNA in 6 × SSC + 50% formamide at 50 °C for 48 h, and hybridized counts measured. Human DNA content was determined from the calibration. The results are given in Table 2.

These results show that whenever a large amount of human DNA was measured, e.g. CI2D with 1 human X-chromosome (2·5% of a human genome) or Di with 5–6 % human, there was reasonably good agreement between the hybridization value and cytological expected value. However, cells with a human chromosome fragment either visible (R493) or not visible (R₆2, B₂) gave hybridizable DNA values greater than expected (see Discussion).

A cRNA probe to human DNA of lower repetitiveness (C_ot 0·5–10·0) has also been tested with appropriate calibrations, and human DNA in CI2D and R_G2 investigated. The method used was identical with that used before. Values of 2·8 and 0·65 % were obtained (Table 2), in good agreement with the values obtained with the C_ot 1 probe (2 and 0·75%, respectively).

Although the human DNA sequences at the concentration used here are not saturated by the probe, the linearity of the calibration suggests that saturation is not essential provided known mixtures are hybridized at the same time as the unknowns.

Reassociation kinetics is a more sensitive and informative method of examining DNA sequences, and the potential of this method was therefore tested. The principles of reassociation kinetics are well known (Wetmur & Davidson, 1968; Britten & Kohne, 1968), although there are problems of interpretation when heterogeneous probes are used (Sharp et al. 1974), as is the case in the present experiments.

³H-human DNA probes were prepared by nick translation with DNA polymerase I in the presence of ³H-deoxynucleotide triphosphates as described in Methods.

An essential to using the probe was to establish that all sequences in the nick-translated DNA were equally labelled. This was done by comparing the nick-translated probe reassociation with that of uniformly labelled C_ot 0–1 DNA. The experiment was conducted as follows. Uniformly labelled C_ot 1 DNA was isolated from Raji cell DNA after the cells had been grown from low density in medium containing 60 μCi [³H]thymidine/ml. Human DNA was added to give an appropriate reassociation rate, in a final volume of 50 μl of 40% FSSPE buffer. Similarly 8000 cpm nick-translated C_ot 1 DNA and unlabelled human DNA were mixed in a final volume of 50 μl in 40% FSSPE buffer. Both samples were overlaid with 40 μl paraffin oil and stoppered. The DNA was denatured by heating to 105 °C for 7 min and the reassociation started by immersing the tubes in an accurately controlled waterbath at 50± 0·05 °C. Samples (12-μl) were removed at intervals into 1·0 ml 0·12 M phosphate buffer+ 0·04% SDS at 4 °C and stored at 4 °C, until all the samples had been collected. Samples were fractionated at 60 °C in a o-q-ml column of hydroxylapatite (HAP) equilibrated with CUM phosphate buffer. Single-stranded DNA was washed through the column with 8·0 ml 0·12 M PB at 60 °C and double-stranded DNA eluted with 8·0 ml of 0·4 M PB. The reassociation curve is shown in Fig. 2, from which it is clear that the line (Raji C_ot 1 DNA reassociation) is well represented by the points (nick translation C_ot 1 DNA), and therefore the nick-translated probe can be considered uniformly labelled. The reassociation became much slower after about 60% hybridization, as expected of a probe which contains about 30–40% single-stranded material (Marx et al. 1976).

The reassociation rate in 50% FSSPE buffer at 50 °C was similar to that in 0·12 M phosphate buffer at 60 °C. C₀t₀.₅ for C_Qt 1 DNA reassociation in formamide buffer was (Fig. 2) 2–3 × 10⁻² mol s l^-1, compared with the value of 2 × 10⁻² in 0·12M phosphate buffer obtained by Marx et al. (1976). The accelerating effect of high monovalent ion concentration (Britten, Graham & Neufeld, 1974), 1·18 M in this buffer, is counterbalanced by the formamide concentration and lower reassociation temperature (McConaughy, Laird & McCarthy, 1969).

Fig. 3A shows the reassociation of nick-translated, C_ot 1 DNA in the presence of various concentrations of human total DNA (0·2, 0·6, 1·0 μg/ml final cone.), and also 1500, 3000, and 4500 μg/ml PG 19 mouse DNA (see legend to figure). The reassociation with human DNA is clearly biphasic, with a rapid early phase followed by a slower phase: self-reassociation of the probe alone followed a linear path. Fig. 3A shows that the rate of reassociation of the nick-translated probe varied with the added human DNA concentration. The sequences reassociated in the fast phase (at 1 μg/ml) correspond with a C₀t of about 2 × 10⁻², in the range of ‘intermediate’ repetitive sequences: the apparently linear portion of the curve represents reassociation of less-repetitive DNAs. If the slope of the linear part of the reaction was plotted against DNA concentration then a reasonably straight line resulted. In contrast, mouse DNA, even at a 15–22000 times higher concentration did not show a progressive interaction with the probe DNA. There may be an early phase of increased reassociation, but after 3·75–5 h the rate was not significantly different from the probe alone. It was concluded therefore that up to 4500 μg mouse DNA/ml did not interact significantly with human probe DNA in the hybridization buffer.

If the faster reassociating sequences were removed from either the unlabelled human DNA sample or the probe, then the initial phase of the hybridization did not take place. This is shown in Fig. 3B where sequences reassociating at C_ot < 10⁻³ were removed from the probe, by an appropriate reassociation and HAP separation. The probe was then reassociated with 0·3 μg/ml final concentration of total human DNA in 40% FSSPE buffer at 50 °C, and the hybridization plotted as a function of time. The slope of this reassociation was in good agreement with the expected, slope derived from the linear phase of total DNA reassociation with total probe (Fig. 3 A). Probe self-reassociation followed the usual linear time course.

Experiments have also been made with a probe to C_ot 1–10 human DNA, i.e. of rather low repetitiveness, but although these probes can be used the long reassociation times to achieve a reasonable hybridization make them inconvenient to use. They are also contaminated with presumptive unique sequence DNAs in spite of multiple purification steps to remove them. Another disadvantage is that their relatively low degree of repetition in the genome could mean a loss of sensitivity in the detection of small fragments of DNA.

The reassociation method has been applied to 2 clones, R_G2, found to contain 0·6–0·7 % equivalent of a human genome by the filter method (Table 2) and CI19S16-TG, a PG19X human lymphocyte hybrid (see Discussion). Fig. 4A shows the timecourse of reassociation of 1 mg/ml R₆2 DNA with nick-translated C_ot 0–1 probe. There was a fast reassociation complete in about 1·5–2 h, followed by a rate not significantly different from the probe alone rate. It is clear by comparison with Fig. 3 A that the slower reassociating sequences are in much lower concentration than expected from the extent of the fast phase. Based on the extent of reassociation of the fast component, DNA in this clone is approximately 0·2% equivalents of a human genome, lower than the value found by filter hybridization.

In contrast, CI19S16TG does not show a fast-reassociating phase at all (Fig. 4B) and clearly this hybrid contains no very repetitive human sequences. However, the progressive reassociation rate shows that less-repetitive sequences are present, and comparison with Fig. 3 A suggests a value of 0·005–0·01% equivalents of human DNA in this hybrid.

These experiments show that molecular hybridization methods can be used under suitable conditions to measure semi-quantitatively the amount of human DNA in human somatic cell hybrids. The reassociation method is more informative about the types of human sequences in the hybrid and is more sensitive but the filter method is simpler to use. The lower limit of the method is around 0·005–0·01 % equivalents of a human genome (‘equivalents’ has the same meaning as ‘genome equivalents’ in tumour virus assays). The good agreement (Table 1) between the DNA content of A9 x Daudi hybrids determined from the karyotype and by molecular hybridization on filters confirms that the hybridization is not chromosome-specific. Although human satellite DNAs are most probably involved in the hybridization (with C_ot 0–1 probes), they are only a fraction of the sequences in the probe (Marx et al. 1976), and reassociation and filter methods can be used when these sequences are absent from the system. The reassociation method may give lower values for the DNA content than the filter method (see later), although the evidence is not yet conclusive.

A surprising feature of experiments with cell hybrids made from one irradiated human parent, is the relatively large amount of human DNA retained along with the selected gene, HPRT; e.g. in R493, R₆2, B₂ (Table 2). HPRT (a trimer of mol. wt. 75-85000 (Fenwick et al. 1977)) is probably encompassed by less than 5 × 10³ b.p., about 3 × 10⁻³ % of the X-chromosome. Hybrid cell R₆2 which had only HPRT activity (PGK⁻, αGAL⁻ or G6PD⁻ (S. Goss, personal communication)) however, contained DNA equivalent to o-6% equivalent of a human genome, about 25% of that in the X-chromosome (Table 2). The value from reassociation kinetics was 0·2%, about 8% of that in the X. This DNA reassociated in the fast-intermediate class (C_ot = 2 × 10⁻²). No slower sequences were detected (cf. Fig. 4A and Fig. 3 A, curve C). R₄93, a similar hybrid but containing a larger human fragment and 1·9% equivalents of a human genome (around 75% of an X) had HPRT and G6PD activities (S. Goss, personal communcation). These results suggest that HPRT is either normally linked to a very large piece of repetitive DNA, or that recombination can occur between a small HPRT-containing fragment and a large repetitive DNA segment. In either case, the association is close since back selection of R493 with 6-thioguanine caused loss of the HPRT and repetitive sequences together. The human repetitive sequences are clearly not spread around the Chinese-hamster genome. More needs to be known about DNA sequence distribution along the human X-chromosome for this to be resolved.

Cl19S16TG presents a very different picture. This clone, PG19 mouse melanoma x human lymphocyte (Jonasson & Harris, 1977), contained no DNA detectable by filter hybridization. It also contained no human chromosomal material identifiable by cytology, but suppression of malignancy (Harris et al. 1969; Wiener, Klein & Harris, 1974; Jonasson et al. 1977) was very strong (Jonasson & Harris, 1977), coupled with changes in cell-surface glycoproteins (Bramwell & Harris, 1978). Re-check of the malignancy suppression of this clone showed that it had not changed since Jonasson & Harris (1977). Injection of 5 x io⁵ cells into nude mice produced 0/10 tumours in 3 months.

The nature of the association between the human fragments and the mouse or Chinese-hamster parental genome in any of the clones examined is unknown (except R₄93 which had a small chromosomal fragment of possibly human origin). Among the possibilities are: (a) free existence of the fragment, synchronized to the parental cell cycle (R493 ?); (b) translocation of the fragment to a terminal position on a parental chromosome; (c) non-integrative association; and (d) integrative association.

The relation between EBV and human DNA in lymphoblastoid cell lines typifies (c) and (d) (Adams, Lindahl & Klein, 1973; Kaschka-Dierich et al. 1976; Adams & Lindahl, 1975; Lindahl et al. 1976).

Whatever the association, it is interesting that in CI19S16TG, cells with malignancy highly suppressed can contain only a small amount (around 3 × 10⁵ b.p.) human DNA.

Probes to human DNA could have other uses. One use currently under investigation is to identify human sequences in recombinant plasmids by hybridization (Grunstein & Hogness, 1975). Cell hybrids such as those used in the present experiments may provide convenient starting material for the isolation of some human DNA sequences.

I would like to thank Mr M. D. Burtenshaw and Dr E. P. Evans for karyotyping the humanmouse hybrids. I would also like to thank Prof. H. Harris for encouragement and discussion. Chinese-hamster cell lines were given by Dr S. Goss. Mr K. M. S. Townsend and Mrs F. C. Croall provided excellent technical assistance. The work was supported by the Cancer Research Campaign.