Rapid detection, cloning and molecular cytogenetic characterisation of sequences from an MRP-encoding amplicon by chromosome microdissection.

Chromosome microdissection was utilised for the analysis of cytogenetic markers of gene amplification [homogeneously staining regions (hsrs) and double minutes (dmins)] in two doxorubicin-resistant cell lines, fibrosarcoma HT1080/DR4 and small-cell lung cancer H69AR. Microdissection products from the hsr(7)(p12p15) of HT1080/DR4 were amplified and used for fluorescent in situ hybridisation (micro-FISH) analysis of drug-sensitive HT1080, resistant HT1080/DR4 and normal lymphocytes. The results demonstrated that the hsr contains a domain of DNA amplification of complex origin including sequences derived from 16p11.2-16p13.1, 2q11.2, 7q32-7q34 and 10q22. The amplification was confirmed by converting the micro-dissected probe into a microclone library for probing HT1080 and HT1080/DR4 Southerns. A micro-FISH probe from normal band region 16p11-16p13 further demonstrated amplification of 16p sequences in both HT1080/DR4 and H69AR. During the course of this analysis, Cole et al. (1992) (Science, 258, 1650-1653) published the amplification of the MRP gene in H69AR cells, which maps to chromosome 16p13.1. Our results corroborate the finding of MRP amplification in these doxorubicin-resistant cell lines, but, importantly, they provide information on the composition of the complex amplicon contributions from four different chromosomes. This study demonstrates the potential utility of chromosome microdissection for the rapid recovery of sequences from amplified regions in drug-resistant cells.

hsr(7Xpl2pl5) of HT1080 DR4 were amplified and used for fluorescent in situ hybridisation (micro-FISH) analysis of drug-sensitive HT1080. resistant HT1080 DR4 and normal lImphocytes. The results demonstrated that the hsr contains a domain of DNA amplification of complex origin including sequences dernved from 16pl 1.2-16pl3.1. 2q1 1.2. 7q32 -7q34 and 10q22. The amplification was confirmed by converting the microdissected probe into a microclone library for probing HT1080 and HT1080 DR4 Southerns. A micro-FISH probe from normal band region 16pl -16pl3 further demonstrated amplification of 16p sequences in both HT1080 DR4 and H69AR. Dunrng the course of this analysis. Cole et al. (1992) (Science. 258, 1650(Science. 258, -1653 published the amplification of the MfRP gene in H69AR cells. which maps to chromosome 16pl3.1. Our results corroborate the finding of MRP amplification in these doxorubicin-resistant cell lines. but. importantly. they provide information on the composition of the complex amplicon contributions from four different chromosomes. This study demonstrates the potential utility of chromosome microdissection for the rapid recosers of sequences from amplified regions in drug-resistant cells. Acquired resistance to chemotherapeutic agents is a frequently encountered problem in cancer chemotherapy. Treatment is often limited by the emergence of clonal tumour cell populations that display resistance not only to the drugs used in prior treatment. but to a wide range of chemotherapeutic agents (Morrow & Cowan. 1993). Model systems based on tumour cells selected in lvitro for increasing resistance to chemotherapeutic agents have been useful in determining the genetic and biochemical mechanisms of acquired drug resistance. Acquisition of the drug-resistant phenotype in tissue culture is frequently associated with amplification of specific drug resistance genes (Kellems, 1993). Drug-resistant cells which have undergone gene amplification frequently display cytogenetic alterations such as homogeneously staining regions (hsrs) or double minutes (dmins) which contain the amplified target gene. For example, acquisition of the multiple drug-resistant phenotype is frequently associated with amplification of the MDRI gene encoding the P-glycoprotein transporter (Riordan et al., 1985;Ueda et al., 1986). Interestingly. several cell lines with cross-resistance to multiple drugs and which do not exhibit MDR1 amplification or P-glycoprotein overexpression have been reported (Beck et al., 1987;McGrath & Center, 1987;Mirski et al., 1987: Slovak et al.. 1988. It is likely that amplification of genes other than MDRI may relate to drug resistance, and recently the gene MRP has been reported to be amplified in P-glycoproteinnegative cell lines (Cole et al., 1992;Slovak et al., 1993).
We have recently applied FISH analysis using probes generated by chromosome microdissection (micro-FISH) to the detection, cloning and identification of amplified sequences from human tumours (Zhang et al.. 1993). We sought to apply this technology to drug-resistant cell lines in order to detect, clone and identify amplified sequences that may be involved in the acquisition of the drug-resistant phenotype.
Two drug-resistant cell lines were used in this study. The doxorubicin-resistant fibrosarcoma cell line HT1080'DR4 displays an hsr(7)(p12pl5) which is not present in the paren-  (Slovak et al.. 1987). Similarly. the drugresistant small-cell lung carcinoma cell line H69AR (Mirski et al.. 1987) developed an hsr and an increased number of dmins relative to the parental cells. H69 (Slovak et al.. 1993).
These cell lines have been demonstrated to be negative for MDRI amplification and P-glycoprotein overexpression. but do have amplification and overexpression of the gene MRP (which maps to l6pl3.l) (Cole et al.. 1992). Recent transfection expenments support a role for MRP in confemrng the drug-resistant phenotype (personal communication from C.E. Grant, S.P.C. Cole & R.G. Deeley). In this report, we applied chromosome microdissection to the hsr of HT1080 DR4. The results corroborate the high level of amplification of 16p sequences within the hsr of HT1080 DR4. The utilisation of chromosome microdissection also allowed us to determine the complex nature of the MRP amplicon.

Materials and metbods
Cell culture HT1080 parental and HT1080DR4 cell lines were cultured as described by Slovak et al. (1987). H69 parental and H69AR cell lines were kindly provided by S.P.C. Cole (Queen's University, Kingston, Canada) and were cultured as described by Mirski et al. (1987).
Microdissection and amplification of chromosomal DNA Cell metaphases were harvested and G-banded for microdissection from tissue culture using conventional cytogenetic techniques (Trent & Thompson, 1987). Microdissection was performed with glass microneedles controlled by a micromanipulator attached to an inverted microscope as previously described by Meltzer et al. (1992). The dissected chromosome fragments were transferred to a 5 p1 collecting drop [containing 40 mM Tns -HCI. pH 7.5. 20 mM magnesium chlonrde. which the collection drop was covered with a drop of mineral oil and incubated at 96°C for 5 min. An initial eight cycles of polymerase chain reaction (PCR) (denaturation at 94°C for 1 min, annealing at 30°C for 2 min and extension at 37°C for 2 min) were conducted by adding approximately 0.3 units of T7 DNA polymerase (Sequenase version 2.0, USB) at each cycle. [Sequenase (13 units gi-') was diluted 1:8 in enzyme dilution buffer (USB) and 0.2 gl was added to 5 gil of reaction mixture.] Following this preamplification step, a conventional PCR reaction catalysed by Taq DNA polymerase was performed in the same tube. The components of the PCR reaction were added to a final volume of 50 gil [10 mM Tris-HCI, pH 8.4, 2 mM magnesium chloride, 50 mM potassium chloride, 0.1 mg ml-' gelatin, 200 giM each dNTP and 2 units of Taq DNA polymerase (Perkin-Elmer/Cetus)]. The reaction was heated to 95°C for 3 min followed by 35 cycles at 94°C for 1 min, I min at 56°C and 2 min at 72°C, with a 5 min final extension at 72°C.

Fluorescent in situ hybridisation
Amplified microdissected DNA (2 gil) was labelled with biotin-11-dUTP in a secondary PCR reaction identical to that described above except for the addition of 20 giM biotin-11-dUTP. The reaction was continued for 12 cycles of 1 min at 94°C, 1 min at 56°C and 3 min at 72°C with a 10 min final extension at 72°C. The products of this reaction were purified with a Centricon-30 filter and used for FISH. Hybridisation of the micro-FISH probes followed our procedure described previously by Meltzer et al. (1992). For each hybridisation, 100 ng of probe was used in 10 gl of hybridisation mixture containing 55% formamide, 2 x SSC and 1 gg of human Cot I DNA (BRL). The slides with metaphase spreads were denatured in 70% formamide, 2 x SSC, at 70°C for 2 min and then hybridised with probes at 37°C in a moist chamber overnight. After a series of washes and avidin/anti-avidin/ fluorescein isothiocyanate (FITC) treatments, the slide was counterstained with 0.5 mg ml-' propidium iodide (including an antifade solution) and examined with Zeiss Axiophot microscope equipped with a dual bandpass (fluorescein/ rhodamine) filter.
Microcloning A library of hsr(7)(pl2pl5) specific microclones was generated essentially as described in Guan et al. (1992). The PCR products were directly inserted into the T-tailed vector pGEM-T (Promega). For this library, 100 ng of PCR products was ligated with 400 ng of vector in a 10 il volume reaction at 12°C overnight. Ligation product (1 gil) was then used to transform Escherichia coli by electroporation. Inserts were recovered by PCR amplification of individual colonies using vector primers (T7 and pUC/M 13 reverse). Those clones which hybridised to repetitive human Cot I sequences were discarded and not used for Southern analyses.

Results
Previously, cytogenetic analysis of the doxorubicin-resistant cell line HT1O8O/DR4 demonstrated the acquisition of an hsr(7)(pl2pl 5) during drug selection (Slovak et al., 1987, Figure I Micro-FISH probe from HT1080/DR4 hsr(7)(p) hybridised to HT1080 interphase nuclei a, HT1080 metaphase chromosomes b. HTI080/DR4 interphase nuclei c and HT 1080/DR4 metaphase chromosomes d. Note the increased probe hybridisation to resistant cell chromatin and chromosomes. In d, the fluorescence signal is most intense on the HT1080/DR4 hsr(7)(p) but is also visible at several secondary sites.
.'i 1993). This marker appeared likely to carry amplified DNA and was therefore targeted for microdissection. After amplification of 20 microdissected fragments in vitro, the product was biotinylated. This micro-FISH probe was then hybridised to drug-resistant HT1080 DR4 and drug-sensitive HT1080 parental cell interphase and metaphase nuclei. As shown in Figure (7)pl2pl5). These results demonstrate that the probe recognises a chromosomal domain consistent with the hsr(7)(pl2pl5) of HT1080,DR4. and a comparison of HT1080 and HT1080 DR4 suggests that the probe hybridises to sequences which are amplified in HT1080 DR4 relative to the parental cell line. Slovak et al. (1993) used a probe for the MRP gene to document clearly the presence of MRP sequences within this hsr. However, of interest, intervening blocks of chromosomal DNA were observed which did not hybridise with either an MRP probe or a whole chromosome composite painting probe (WCP) for chromosome 16. In order to identify the chromosomal origin of sequences in the hsr(7Xpl2pl5) we examined the hybridisation pattern of the hsr (7)(pl2pl5) micro-FISH probe to previously G-banded normal lymphocyte metaphases. As shown in Figure 2, the probe hybridised to four discrete chromosome bands: 2ql l.2. 7q32-7q34. 1Oq22 and 16pl.2 -16pl13.1. The signals on chromosome 16 and 7 were consistently strongest. Observation of multiple metaphases suggested that the chromosome 16 signal was the most intense. These results indicate that the hsr(7)(pl2pl5) of HT1080 DR4 consists of amplified sequences from 16p (consistent with the results of Slovak et al.. 1993) but also contains sequences from 2q. 7q and 10q.
The micro-FISH probe from the HT1080 DR4 hsr(7) (pl2pl5) was also hybridised to metaphase chromosomes of H69AR. As shown in Figure 3. H69AR nuclei demonstrated hybridisation to multiple intrachromosomal sites, including both large marker chromosomes and smaller chromosomal regions. Numerous double minutes within the same cells also show hybridisation. The non-uniform hybridisation to the large hsr markers may indicate the presence of sequences in the H69AR amplicon not represented in the HT1080 DR4 hsr(7Xpl2pl5) probe. However, the positive hybridisation signal clearly indicates a significant extent of overlap between 7!-.. Hybridisation is apparent at 2q1 .2. 7q32-34. 10q22 and 16p 1 1.2 -13.1. Bright signals appear on 7q and 16p. and observation of multiple metaphases reveals that the 16p signal is consistently the most intense. This result suggests that the microdissected region contains sequences translocated from other sites in addition to its major contribution from 16p. sequences amplified in HT1080 DR4 and H69AR. This result corroborates the results of Slovak et al. (1993) (who used a 16 WCP) and indicates that homologous sequences have been amplified in two independently isolated doxorubicin-resistant cell lines.
Because the HT1080,DR4 hsr(7)(pI2pl5) micro-FISH probe was complex and the most intense signal localised to band 16pl 1.2-16pll3.1, we performed microdissection on this segment in normal metaphases in order to investigate the involvement of sequences from this region in the amplification events in HT1080/DR4 and H69AR. A micro-FISH probe specific for 16pll -16pl3 was hybridised to metaphase chromosomes of HT1080 DR4 and H69AR. The results shown in Figure 4 confirm that the amplicons in the H69AR hsrs and some of the H69AR dmins as well as the hsr (7) (pl2pl5) of HT1080,/DR4 contain sequences from the 16pl 1 -16pl3 region. The localisation of the signals from the l6pl l-6pI3 micro-FISH probe on HT1080DR4 and H69AR chromosomes is in agreement with that of the HT1080 'DR4 hsr(7Xpl2pl5) micro-FISH probe seen in Figure ld and Figure 3. Interestingly, the probe displayed a ladder-like' pattern of hybridisation to the hsr(7)(pl2pl5) of HT1080/DR4, similar to the observations of Slovak et al. (1993), who utilised a 16WCP. This is consistent with the presence in the HT1080'DR4 hsr(7)(pl2pl5) of segments derived from other chromosomal regions interspersed with matenal from the I6pll-6pl3 region. Based on the results in Figure 2. we conclude that these sequences are derived from 2qI1.2. 7q32-7q34 and 10q22. Fue 4 A micro-FISH probe was generated from the normal 16pl1-16p13 region and hybnrdised to HT1080 DR4 a. and H69AR b. metaphase nuclei. The results demonstrate amplification of sequences from this region in both cell lines. Note the striped appearance of the hybnrdisation pattern on the HT1080 DR4 hsr a. indicating interspersal of non-16pl 1-16pl3 sequences. Also note hybnrdisation to multiple H69AR chromosomes b. hsrs (lower two arrows) and dmins b. (upper arrow).
In order to characterise the amplification of DNA sequences in the hsr(7)(pl2pl5) of HT1080 DR4. the amplified microdissection products were converted into a microclone libary. Thirty-five independent clones were analysed. The insert size ranged from 200 to 700 bp. which is consistent with previous microclone libraries constructed with this methodology (Guan et al.. 1992). Six inserts were then eliminated which hybridised with repetitive sequence probes. The 29 remaining inserts were used as probes against Southern blots of EcoRI-digested genomic HT1080 and HT1080 DR4 DNA. Twenty-five of the 29 probes tested (86%) detected amplified restriction fragments in HT1080 DR4 relative to HT1080. Representative examples are illustrated in Figure 5. It appeared that each amplificationpositive probe detected a different restriction fragment. although some fragments were of similar size. Densitometnand DNA serial dilution expeniments revealed the level of amplification of these microclones to be in the range of 5to 10-fold, similar to that of MRP amplification in HT1080 DR4 (Slovak et al.. 1993). Six of the 25 microclones which showed amplification in HT1080 DR4 w-ere also tested on H69 and H69AR Southern blots. Three of these six detected amplified restriction fragments in H69AR relative to H69 (data not shown). These results confirm that the amplified product generated from microdissected chromosomal material from the hsr contains sequences which are amplified in HT1080 DR4 (as well as H69AR).

Discussion
Chromosome microdissection and microclone library construction provide a novel approach for the rapid detection and cloning of amplified DNA sequences from specific cytogenetically recognisable markers such as hsrs or dmins. Other approaches to the analysis of amplified DNA sequences have relied on techniques based on DNA electrophoresis such as in gel renaturation and restriction landmark genomic scanning (Roninson. 1983: Hatada et al.. 1991. These techniques have successfully identified amplified sequences. but are laborious and can be confounded by amplified sequences unrelated to the phenotype of interest. The recently reported molecular cytogenetic technique of comparative genome hvbnidisation (CGH) is able to identify directly the chromosomal origins of amplified sequences but does not directly lead to the generation of cloned probes specific for the amplicon (Kallioniemi et al.. 1992).
We sought to appil the technology of chromosome microdissection to detect and clone amplified sequences from the hsr(7)(pl2pl5) of the drug-resistant cel' line HT1080 DR4 because our previous attempts to obtain amplified sequences from HT1080 DR4 by in-gel renaturation were unsuccessful (Slovak et al.. 1991). Micro-FISH analysis utilising the probe from the HT1080 DR4 hsr confirmed the presence of amplified sequences from 16p within the hsr (Slovak et al.. 1993).
but also enabled analysis of the chromosomal origins of additional sequences within the amplicon. In addition to the major contribution from 16pl1.2-16pl3.1. the hsr also includes sequences from 2q 11.2. 7q32 -7q34 and 10q22. The contribution of 16p w-as readily confirmed by hybridisation of a 16pll-16pl3 micro-FISH probe from normal cells to HT1080 DR4 and H69AR cells. Slovak et al. (1993) utilised a chromosome 16 WCP for FISH analysis of the hsr(7) (pl2pl5) of HT1080 DR4 as well as H69AR. They reported the presence of chromosome 16 signals on multiple chromosomes and dmins in H69AR and described a striped pattern of fluorescent signal on the HT1080 DR4 hsr(7) (pl2pl5). suggesting the presence of non-chromosome 16 sequences interspersed with chromosome 16 sequences u-ithin the hsr. Our studies confirm that sequences from the specific region of 16pll-16pl3 are amplified in both cell lines and duplicate this 'ladder-like' fluorescent signal pattern on HT1080 DR4 hsr(7Xpl2pl5). Our anal sis identifies the chromosomal origins of the sequences u-hich are interspersed with the l6pll -16p13 sequences as 2q'11.2. 7q32 -7q34 and I0q22. The amplification of homologous sequences from 16p in tu-o independently isolated doxorubicin-resistant cell lineages strongly suggests that this region is inxolved in acquisition of the drug-resistant phenotype. The roles of sequences derived from other chromosomal origins in the HT1080 DR4 hsr(7Xpl2pl5) remain uncertain. These sequences may represent a record of the chromosomal events which led to the amplification of the 16p sequences. reflecting upon amplification mechanisms. It is of interest to note that 2qlI.2 (FRA2A). 7q32.3 (FRA7H). 10q22.1 (FRAIOD) and l6pl3.11 (FRA16A) are all fragile sites. which may increase the likelihood of their involvement in chromosomal rearrangements (Reeders et al.. 1991;Simpson & Cann. 1991: Spurr & White. 1991Tsui & Farrall. 1991). However. it remains possible that the chromatin interspersed between the domains of 16p sequences contain genes which contribute to the drug-resistant phenotype. Chromosome microdissection will provide a valuable technique for further investigation of the roles played by these sequences.
Of interest within the region of I6pl1 l6pl3. the gene U11RP has been mapped to 16pl3.1 (Cole et al.. 1992) and was cloned from H69AR. in which it is amplified and overexpressed. FISH analysis utilising MRP probes has demonstrated that MRP is restricted to l6p13.1 in parental H69 and HT1080 cells but localises to the hsr(7)(pl2pl5) in HT1080 DR4 and multiple hsrs and dmins in H69AR (Slovak et al.. 1993). The sequence of the MRP product shows homologY to the superfamily of transmembrane ATPdependent transport proteins. Recent transfection data Fugwe 5 Representative microclones from the HTl080/DR4 hsr(7Xp) library tested as probes on Southern blots. The majornty of the non-repetitive clones in the library recognise restnction fragments which are amplified in HT1080/DR4 relative to HT1o8o. In each blot, the left two lanes (labelled S) contain two dilutions of EcoRl-digested genomic DNA from HT1080 and the right two lanes (labelled R) contain EcoRI-digested genomic DNa from HTI080/DR4. Approximate size markers appear to the right, while a single-copy control probe hybridisation is shown below each blot. support a role for the MRP product in conferring drug resistance (personal communication from C.E. Grant, S.P.C. Cole & R.G. Deeley). Further clarification of the genetic events which have occurred in the development of the HT1080/DR4 hsr(7) (pl2pl5) will require more detailed physical mapping studies of the amplified DNA. This will facilitate identification of all of the genes encoded in the hsr so that their relationship to the drug-resistant phenotype can be systematically evaluated. In this regard, a significant advantage of chromosome microdissection-based technology is that, in addition to confirming the presence of DNA sequence amplification and identifying its chromosomal origin, it leads directly to the generation of a microclone library which is highly enriched for amplification unit probes. Eighty-six per cent of the non-repetitive microclones tested showed significant amplification in HT1080/DR4 relative to HT1080, and several showed amplification in H69AR relative to H69 as well. These microclones are valuable as entry point probes for the analysis of the amplicon structure, and can be used to define the overlap of the amplification units between independent drug-resistant cell lines. Furthermore, these microclones are of a convenient size for automated sequence analysis, which can be used to establish sequence tagged sites (STSs) useful for the isolation of large insert genomic clones such as yeast artificial chromosomes (YACs). YAC clones can then be used to establish a map of the amplicon in a manner similar to that described by Schneider et al. (1992). In contrast to the mapping of the N-myc amplicon, for which numerous probes previously existed, the physical mapping of amplicons from newly identified amplification regions will be greatly facilitated by techniques such as microdissection, which can not only confirm the presence of amplified sequences at specific chromosomal sites and identify the chromosomal origins of those sequences, but also generate a library of entry point probes for the initiation of amplicon structure analysis.
We gratefully acknowledge Dr S.P.C. Cole and Dr R.G. Deeley for their critical review of the manuscript. We also acknowledge the excelknt technical assistance in chromosome banding by Ann