67 GATA 8(2): 67-74, 1991
Clinical Applications of Fluorescence in situ Hybridization D. C. T K A C H U K , D. P I N K E L , W.-L. K U O , H.-U. W E I E R , and J. W. GRAY
We review here the application of fluorescence in situ hybridization with chromosome-specific probes to chromosome classification and to detection of changes in chromosome number or structure associated with genetic disease. Information is presented on probe types that are available for disease detection. We discuss the application of these probes to detection of numerical aberrations important for prenatal diagnosis and to detection and characterization of ~lumerical and structural aberrations in metaphase spreads and in interphase nuclei to facilitate tumor diagnosis.
Introduction Fluorescence in situ hybridization (FISH) has bec o m e an important tool for analysis of the number, size, and location of specific DNA sequences in mammalian cells [1-20]. The hybridization reaction "stains" the target sequences so that their location and size can be determined using light microscopy. D N A targets ranging from whole chromosomes down to several kilobases can be studied using current hybridization techniques. For example, whole chromosomes can be stained by FISH with probe mixtures that have homology at multiple sites along the target chromosomes [1-3]. Specific DNA sequences such as the ABL oncogene can be reliably stained using probes of only 15 kb [12]. In some circumstances, even kilobase-sized targets can be deFrom the Biomedical Sciences Division (D.T., D.P., W.L-K., U.W., J.W.G.), Lawrence Livermore National Laboratory, Livermore; and the Department of Pathology (D.T.), Stanford University Medical Center, Stanford, California, USA. Address correspondence to Dr. Joe W. Gray, Biomedical Sciences Division, Lawrence Livermore National Laboratory, PO Box 5507, Livermore, CA 94550, USA. Received 4 October 1990; revised and accepted 15 November 1990.
tected [21-24]. F I S H - b a s e d staining is sufficiently distinct that the hybridization signal can be seen both in metaphase spreads and in interphase nuclei. Chromosomes in interphase nuclei are generally organized into distinct domains so that their location, number, and integrity (in some situations) can be assessed using FISH with probes to the involved chromosomes or parts thereof (see Lichter et al. [25] for a current review). This enables detection of both numerical and structural aberrations in many situations. The ability to detect and characterize numerical and structural aberrations in metaphase spreads and in interphase nuclei has substantial clinical value. The scope of this review is to outline clinical applications of single- and multicolor F I S H to chromosome assignment in standard karyotyping, prenatal diagnosis, and tumor cytogenetics. This should enable readers to become familiar with the capabilities of this approach, to understand how to interpret results, and to know when it might be useful to implement such an assay. Protocols for target cell preparation and fixation, probe labeling, hybridization, washing procedures, and signal amplification are presented only briefly here as they are described in detail elsewhere [25, 26].
Fluorescence in situ Hybridization In F I S H , D N A from a chromosome-specific probe is chemically modified (for example, altered by nick translation so that thymine is replaced by deoxyuridine triphosphate labeled with biotin [27]). The labeled DNA is then hybridized to metaphase chromosomes or interphase nuclei for which chromosome-specific staining is desired. In situ hybridization requires denaturation of both the DNA of the target cells and the labeled probe DNA. Incubation of the probe and target DNA together at a temperature below the melting point allows the chemically modified probe DNA to bind to complimentary sequences in the target. Many potentially useful probes contain DNA seq u e n c e s that are r e p e a t e d t h r o u g h o u t the genome. In such cases, the repeated sequences must be removed [281 or unlabeled sequences, for example genomic DNA, must be added to block the binding of the repeated sequences [1-3]. The bound, chemically modified DNA is then fluorescently stained (for instance, by incubation with f l u o r e s c e i n - a v i d i n that binds specifically to biotin) to complete the staining reaction. The
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010 1050-3862/91/$03.50
68 GATA 8(2): 67-74, 1991
stained chromosomes may be counterstained with DNA-specific fluorescent dyes such as propidium iodide for further discrimination if desired. This procedure allows selected chromosomes or regions thereof to be stained with remarkable intensity and distinctness. Binding of several different probes may be detected simultaneously by properly utilizing different probe labels (for example, biotin and digoxigenin) and detection reagents (for instance, FITC-anti-digoxigenin antibody and Texas r e d avidin). Analysis proceeds by hybridizing all probes simultaneously and detecting each probe with the proper reagent. The locations of the fluorescent hybridization signals can be determined by f l u o r e s c e n c e m i c r o s c o p y using multiband filters that permit simultaneous observation of both fluorochromes [29].
Chromosome-Specific Probes The clinical utility of FISH depends on the availability of probes that bind specifically to regions of genetic or cytogenetic interest. The probes now in use fall in three general classes: (a) Probes for repeated DNA sequences found on only one chromosome type. These probes are used primarily for chromosome enumeration (reviewed in Bauman et al. [30]. (b) Whole c h r o m o s o m e probes comprised of elements with D N A sequence homology along the length of a target chromosome. These probes, typically made from chromosome-specific recombinant DNA libraries, are useful for analysis of both structural and numerical aberrations in metaphase [1-3, 17]. (c) Probes for specific loci. These are typically single c o p y p r o b e s h o m o l o g o u s to specific targets ranging in size from 15 to >500 kb, so that the signals are bright and reliable [1, 10-12, 22]. They may be used to diagnose specific genetic diseases.
Repeat-Sequence Probes All human chromosomes have been shown to carry DNA sequences that are tandemly repeated several hundred to several thousand times in the centromeric regions [31-33]. The total amount of repeated sequence typically ranges from -105 to >106 bp. On most human chromosome types, some part of the repeated sequence is sufficiently different that FISH with a probe to the variant region produces a signal that is intense and chromo-
D. C. Tkachuk et al.
some specific. Most of these sequences are in the alpha-satellite [33] or the satellite-II! families [34]. Alpha-satellite sequences are comprised of tandemly repeated -171-bp monomers, whereas satellite-III sequences are comprised of 5-bp monomers. The specificity of probes for these regions comes from the fact that there is significant variation in the repeated monomer among chromosome types. Chromosome-specific repeat sequence probes have now been isolated and cloned for human chromosomes 1, 3-12, 15-20, X, and Y [31]. However, the hybridization conditions that must be used with some of these probes must be carefully chosen and controlled to prevent hybridization to regions on other chromosome types to which they are moderately homologous (that is, to related repeats from the same family on other chromosome types). Figure la shows F I S H with an alpha-satellite probe to human chromosome i I in a metaphase spread and in an interphase nucleus.
Figure 1. Photomicrographs showing fluorescence in situ hy~ o n with chromosome-specific probes to metaphase or interphase human cells. Probe hybridization for panels a - h was detected using fluorescein so that the hybridization domains appear yellow. The cellular DNA was counterstained with propidium iodide so that the DNA not covered by the probes appears red. a. Hybridization with a probe to a tandemly repeated alpha-satellite sequence on human chromosome 11. b. Hybridization with a whole chromosome composite probe for human chromosome 4 [1]. e. Hybridization with DNA from yeast carrying an artificial chromosome with a 300-kb insert from chromosome 18. d. Hybridization with a whole chromosome probe for chromosome 21 to a metaphase spread from an individual with Down's syndrome, e. Hybridization to a metaphase spread prepared for CML cells with a whole chromosome probe for chromosome 22 [lll. The probe was produced by linker-adapter PCR amplification of DNA from chromosome 22. f. Hybridization with a probe for dihydrofolate reductase to a metaphase spread from cells in which the dihydrofolate reductase gene was heavily amplified [40]. g. Hybridization with a whole chromosome probe for chromosome 21 to interphase nuclei from an individual with Down's syndrome [1]. h. Hybridization with a probe for chromosome 9 to leukemic cells trisomic for chromosome 9. i. Hybridization with a repeat sequence probe to the Y chromosome to cells from a male recipient after bone marrow transplantation with sex-mismatched donor cells, j. Dual color hybridization to normal leukocytes with probes for the BCR and ABL genes [12]. The ABL probe was detected with fluorescein, so its hybridization domain appears green. The BCR probe was detected using Texas red, so its hybridization domains appear red. k. Dual color hybridization as described in panel i to CML nuclei [12]. Dual band-pass filters were used during photography for panels a, d, j, k, and n so that the red and green fluorescence signals were visible simultaneously. I. Hybridization with a probe for the ABL gene to nuclei from K562 cells in which the B C R - A B L fusion gene was amplified 10- to 20-fold [12].
© 1991 Elsevier Science PublishingCo., Inc., 655 Avenue of the Americas, New York, NY 10010
69 GATA 8(2): 67-74, 1991
Fluorescence in situ Hybridization
(
d
C
:~
k
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
70
GATA 8(2): 67-74, 1991
Whole Chromosome Probes F I S H with c h r o m o s o m e - s p e c i f i c c o m p o s i t e probes whose individual elements have sequence homology at many sites along one chromosome type enables an entire chromosome to be stained either in metaphase spreads or in interphase nuclei, since the various elements of the probe bind all along the target chromosome type. One useful set of whole chromosome probes was produced by subcloning the inserts from complete-digest HindlII chromosome-specific recombinant DNA Charon 21A phage libraries into B l u e s c r i b e plasmids [28, 35]. The phage libraries were produced by the National Laboratory Gene Library Project [36] from chromosomes of a single type purified by fluorescence-activated sorting. Subcloning of the phage inserts into plasmids reduced the amount of vector DNA associated with each human insert. These probes are referred to as "painting" probes, since they are used to stain (paint) whole chromosomes [1]. These probes contain families of repeat sequences (Alu and Kpn) that are on the chromosome from which the library was made and which are shared by other chromosomes. To achieve the desired staining contrast, these sequences are prevented from hybridizing by adding unlabeled blocking DNA, in many cases just human genomic DNA, to the probe mixture. Figure lb shows that FISH with these probes stains the target chromosome in metaphase spreads more or less uniformly. These probes do suffer from some deficiencies, however [35]. Some of these libraries are not complete, presumably because some chromosomal regions do not have frequently spaced HindlII sites and thus were not cloned into the Charon-21A phage. Thus, regions of some c h r o m o s o m e s are not stained by FISH with these probes. In addition, probe labeling (usually by incorporating biotinl l-dUTP or d i g o x i g e n i n - l l - d U T P during nick translation or random primer extension) can critically affect the quality of the hybridization signal. We have found it important that the probe fragments range in size from 300 to 1000 bp on a nondenaturing agarose gel. Variability in commercial nick translation kits can lead to inconsistent results. Some of the limitations of the Bluescribe composite probes can be overcome with new generation composites. One promising approach now being pursued is to produce more complete, easily labeled libraries using the polymerase chain reac-
D. C. T k a c h u k et al.
tion. In our laboratory, this is accomplished by starting with DNA from a few thousand chromosomes of a single type purified by sorting [37]. This DNA is digested to completion using one or more restriction endonucleases that recognize four bases (for instance, HpalI). An adapter double-stranded oligonucleotide is ligated onto both ends. All elements are amplified by PCR using a primer to the adapter sequence to produce a library of high complexity containing almost entirely human DNA. This can be labeled by incorporation of the modified nucleotides (biotin or digoxigenin) in the final stages of the PCR amplification. A related approach to whole c h r o m o s o m e probe generation is to amplify sequences in human-rodent hybrid cells by PCR using primers homologous to conserved parts of the Alu repeat. Several kilobase-long regions flanked by Alu sequences can be amplified using this procedure. The composite probes produced by applying this approach to hybrids carrying a single human chrom o s o m e or fragment t h e r e o f can be used as probes to selectively stain human chromosomes in human cells [38].
Locus-Specific Probes Once loci important in genetic disease have been identified, they can be studied using FISH with probes to these regions. Our studies to date suggest that probes for this purpose should target sequences ranging in size from 15 to 500 kb. Probes to targets smaller than 15 kb may not be reliably detected, especially in clinical samples. Probes to targets larger than 500 kb stain extended portions of interphase nuclei so that the hybridization signals on homologous chromosomes may overlap with each other or with hybridization domains of other probes. DNA sequences cloned into large insert phages, cosmids, or yeast artificial chromos o m e s have p r o v e d useful as specific locus probes. Figure lc, for example, shows the hybridization signals produced in interphase nuclei after hybridization with a probe for a 300-kb region of chromosome 18 cloned into a yeast artificial chromosome. Mapped individual clones and contiguous stretches of hundreds of kilobases of DNA from overlapping clones will become increasingly available over the next few years as the Human Genome Project progresses.
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
71 Fluorescence in situ Hybridization
GATA 8(2): 67-74, 1991
Cytogenetic Applications
Metaphase Analysis Chromosome analysis by classic cytogenetics relies on interpretation of banding patterns resulting from dye uptake of individual metaphase chromosomes. The whole genome can be surveyed at one time, but aberrations involving less than a few megabases of DNA, or which involve regions of the genome with indistinct banding, are difficult to detect. Frequently, identification of the components of marker chromosomes is ambiguous. FISH solves these problems because the staining is based on the DNA sequences in the target, and so can be optimized for particular applications. Chromosomes can be quickly and accurately identified after F I S H with repeat sequence or whole chromosome probes. Figure l d, for example, shows hybridization with a whole chromosome probe for chromosome 21 to a metaphase spread prepared from cells from an individual with Down's syndrome [28]. The presence of the extra copy of chromosome 21 can be determined rapidly and accurately. Even more important is the use of FISH with composite probes to detect and characterize structural chromosome aberrations such as translocations or marker chromosomes. Figure le, for example, shows FISH with a whole chromosome linker-adapter probe for chromosome 22 to cells from an individual with chronic myeiogeneous leukemia. The two derivative chromosomes resulting from the t(9;22) translocation are clearly visible [11]. We estimate that rearrangements involving only a few hundred kilobases can be detected using FISH with whole chromosome probes as long as the structural change results in the juxtaposition of the target and nontarget DNA sequences so that material stained by FISH and material not stained by FISH reside on the same chromosome. Subtle deletions and inversions, however, are impossible to detect using this approach. Structural aberration analysis can be made even more accurate by hybridizing with probes targeted to the vicinity of the locus of interest. For example, hybridization with a digoxigenin-labeled probe for a 15-kb region of the BCR gene on chromosome 22 (proximal to the CML breakpoint region) and a biotin-labeled probe for a 35-kb region of the ABL gene on chromosome 9 (distal to the CML breakpoint region) enables detection of the BCR-ABL fusion event associated with CML
[12]. The ABL probe was stained using Texas red-avidin and the BCR probe was stained using FITC-anti-digoxigenin. The juxtaposition of the red and green fluorescing hybridization domains on the short Phi chromosome signaling the B C R ABL fusion was clearly visible. Recent work on probe mapping using FISH suggests that the location of probes along chromosomes can be determined to within a few megabases [39]. This work suggests that FISH in metaphase can be used to detect inversions involving this amount of DNA. Deletions of the region targeted by the probe can be detected, of course, by the loss of hybridization domain, while amplification of the target region can be detected by an increase in the intensity or area of the hybridization domain, or by counting the number of resolvable hybridization sites. The latter method enables detection of duplications if the amplicon is large enough. Figure If, for example, shows FISH with a probe for the dihydrofolate reductase gene to cells that were grown for an extended period in methotrexate so that the dihydrofolate reductase gene was heavily amplified [40]. The large hybridization domain to the amplified (hundreds of times) gene is clearly visible.
Interphase Analysis A significant drawback to metaphase analysis is the requirement that the cells be grown in vitro. This is labor-intensive, time-consuming, and often results in the selection of cells that grow well in culture but may not be representative of the original population from which the cells were taken. The time and labor considerations are of dominant concern in prenatal diagnosis. Arguably, the most important aberrations found during prenatal screening are trisomies for chromosomes 21, 18, and 13 and aneuploides involving sex chromosomes [41]. Detection of numerical aberrations involving the autosomes is particularly important because they occur at relatively high frequency in live born infants and have serious consequences. Analysis of numerical aberrations involving these chromosomes using FISH to interphase nuclei reduces the time required for cell culture (and may eventually eliminate the need for this entirely), since only a few hundred representative cells are required [1, 3, 42]. Figure lg, for example, shows FISH with a whole chromosome probe for chromosome 21 to interphase amniocytes from a fetus
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
72 D. C. Tkachuk et al.
GATA 8(2): 67-74, 1991
with Down's syndrome. The extra copy of the number 21 chromosome is clearly visible. However, not all cells in a population trisomic for the target chromosome show the expected three hybridization domains. This is thought to be due primarily to the chance overlap of the hybridization domains. In fact, the frequencies of trisomic human amniocytes showing three domains after hybridization with whole chromosome probes for chromosomes 21, 18, and 13 range from as low as 10% to >50% [42]. The lowest frequencies were observed in samples in which the hybridization intensity was not high so that domain definition was difficult in some cells. However, the frequency of normal cells showing more than two hybridization domains is usually
Clinical Applications of Fluorescence in situ Hybridization D. C. T K A C H U K , D. P I N K E L , W.-L. K U O , H.-U. W E I E R , and J. W. GRAY
We review here the application of fluorescence in situ hybridization with chromosome-specific probes to chromosome classification and to detection of changes in chromosome number or structure associated with genetic disease. Information is presented on probe types that are available for disease detection. We discuss the application of these probes to detection of numerical aberrations important for prenatal diagnosis and to detection and characterization of ~lumerical and structural aberrations in metaphase spreads and in interphase nuclei to facilitate tumor diagnosis.
Introduction Fluorescence in situ hybridization (FISH) has bec o m e an important tool for analysis of the number, size, and location of specific DNA sequences in mammalian cells [1-20]. The hybridization reaction "stains" the target sequences so that their location and size can be determined using light microscopy. D N A targets ranging from whole chromosomes down to several kilobases can be studied using current hybridization techniques. For example, whole chromosomes can be stained by FISH with probe mixtures that have homology at multiple sites along the target chromosomes [1-3]. Specific DNA sequences such as the ABL oncogene can be reliably stained using probes of only 15 kb [12]. In some circumstances, even kilobase-sized targets can be deFrom the Biomedical Sciences Division (D.T., D.P., W.L-K., U.W., J.W.G.), Lawrence Livermore National Laboratory, Livermore; and the Department of Pathology (D.T.), Stanford University Medical Center, Stanford, California, USA. Address correspondence to Dr. Joe W. Gray, Biomedical Sciences Division, Lawrence Livermore National Laboratory, PO Box 5507, Livermore, CA 94550, USA. Received 4 October 1990; revised and accepted 15 November 1990.
tected [21-24]. F I S H - b a s e d staining is sufficiently distinct that the hybridization signal can be seen both in metaphase spreads and in interphase nuclei. Chromosomes in interphase nuclei are generally organized into distinct domains so that their location, number, and integrity (in some situations) can be assessed using FISH with probes to the involved chromosomes or parts thereof (see Lichter et al. [25] for a current review). This enables detection of both numerical and structural aberrations in many situations. The ability to detect and characterize numerical and structural aberrations in metaphase spreads and in interphase nuclei has substantial clinical value. The scope of this review is to outline clinical applications of single- and multicolor F I S H to chromosome assignment in standard karyotyping, prenatal diagnosis, and tumor cytogenetics. This should enable readers to become familiar with the capabilities of this approach, to understand how to interpret results, and to know when it might be useful to implement such an assay. Protocols for target cell preparation and fixation, probe labeling, hybridization, washing procedures, and signal amplification are presented only briefly here as they are described in detail elsewhere [25, 26].
Fluorescence in situ Hybridization In F I S H , D N A from a chromosome-specific probe is chemically modified (for example, altered by nick translation so that thymine is replaced by deoxyuridine triphosphate labeled with biotin [27]). The labeled DNA is then hybridized to metaphase chromosomes or interphase nuclei for which chromosome-specific staining is desired. In situ hybridization requires denaturation of both the DNA of the target cells and the labeled probe DNA. Incubation of the probe and target DNA together at a temperature below the melting point allows the chemically modified probe DNA to bind to complimentary sequences in the target. Many potentially useful probes contain DNA seq u e n c e s that are r e p e a t e d t h r o u g h o u t the genome. In such cases, the repeated sequences must be removed [281 or unlabeled sequences, for example genomic DNA, must be added to block the binding of the repeated sequences [1-3]. The bound, chemically modified DNA is then fluorescently stained (for instance, by incubation with f l u o r e s c e i n - a v i d i n that binds specifically to biotin) to complete the staining reaction. The
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010 1050-3862/91/$03.50
68 GATA 8(2): 67-74, 1991
stained chromosomes may be counterstained with DNA-specific fluorescent dyes such as propidium iodide for further discrimination if desired. This procedure allows selected chromosomes or regions thereof to be stained with remarkable intensity and distinctness. Binding of several different probes may be detected simultaneously by properly utilizing different probe labels (for example, biotin and digoxigenin) and detection reagents (for instance, FITC-anti-digoxigenin antibody and Texas r e d avidin). Analysis proceeds by hybridizing all probes simultaneously and detecting each probe with the proper reagent. The locations of the fluorescent hybridization signals can be determined by f l u o r e s c e n c e m i c r o s c o p y using multiband filters that permit simultaneous observation of both fluorochromes [29].
Chromosome-Specific Probes The clinical utility of FISH depends on the availability of probes that bind specifically to regions of genetic or cytogenetic interest. The probes now in use fall in three general classes: (a) Probes for repeated DNA sequences found on only one chromosome type. These probes are used primarily for chromosome enumeration (reviewed in Bauman et al. [30]. (b) Whole c h r o m o s o m e probes comprised of elements with D N A sequence homology along the length of a target chromosome. These probes, typically made from chromosome-specific recombinant DNA libraries, are useful for analysis of both structural and numerical aberrations in metaphase [1-3, 17]. (c) Probes for specific loci. These are typically single c o p y p r o b e s h o m o l o g o u s to specific targets ranging in size from 15 to >500 kb, so that the signals are bright and reliable [1, 10-12, 22]. They may be used to diagnose specific genetic diseases.
Repeat-Sequence Probes All human chromosomes have been shown to carry DNA sequences that are tandemly repeated several hundred to several thousand times in the centromeric regions [31-33]. The total amount of repeated sequence typically ranges from -105 to >106 bp. On most human chromosome types, some part of the repeated sequence is sufficiently different that FISH with a probe to the variant region produces a signal that is intense and chromo-
D. C. Tkachuk et al.
some specific. Most of these sequences are in the alpha-satellite [33] or the satellite-II! families [34]. Alpha-satellite sequences are comprised of tandemly repeated -171-bp monomers, whereas satellite-III sequences are comprised of 5-bp monomers. The specificity of probes for these regions comes from the fact that there is significant variation in the repeated monomer among chromosome types. Chromosome-specific repeat sequence probes have now been isolated and cloned for human chromosomes 1, 3-12, 15-20, X, and Y [31]. However, the hybridization conditions that must be used with some of these probes must be carefully chosen and controlled to prevent hybridization to regions on other chromosome types to which they are moderately homologous (that is, to related repeats from the same family on other chromosome types). Figure la shows F I S H with an alpha-satellite probe to human chromosome i I in a metaphase spread and in an interphase nucleus.
Figure 1. Photomicrographs showing fluorescence in situ hy~ o n with chromosome-specific probes to metaphase or interphase human cells. Probe hybridization for panels a - h was detected using fluorescein so that the hybridization domains appear yellow. The cellular DNA was counterstained with propidium iodide so that the DNA not covered by the probes appears red. a. Hybridization with a probe to a tandemly repeated alpha-satellite sequence on human chromosome 11. b. Hybridization with a whole chromosome composite probe for human chromosome 4 [1]. e. Hybridization with DNA from yeast carrying an artificial chromosome with a 300-kb insert from chromosome 18. d. Hybridization with a whole chromosome probe for chromosome 21 to a metaphase spread from an individual with Down's syndrome, e. Hybridization to a metaphase spread prepared for CML cells with a whole chromosome probe for chromosome 22 [lll. The probe was produced by linker-adapter PCR amplification of DNA from chromosome 22. f. Hybridization with a probe for dihydrofolate reductase to a metaphase spread from cells in which the dihydrofolate reductase gene was heavily amplified [40]. g. Hybridization with a whole chromosome probe for chromosome 21 to interphase nuclei from an individual with Down's syndrome [1]. h. Hybridization with a probe for chromosome 9 to leukemic cells trisomic for chromosome 9. i. Hybridization with a repeat sequence probe to the Y chromosome to cells from a male recipient after bone marrow transplantation with sex-mismatched donor cells, j. Dual color hybridization to normal leukocytes with probes for the BCR and ABL genes [12]. The ABL probe was detected with fluorescein, so its hybridization domain appears green. The BCR probe was detected using Texas red, so its hybridization domains appear red. k. Dual color hybridization as described in panel i to CML nuclei [12]. Dual band-pass filters were used during photography for panels a, d, j, k, and n so that the red and green fluorescence signals were visible simultaneously. I. Hybridization with a probe for the ABL gene to nuclei from K562 cells in which the B C R - A B L fusion gene was amplified 10- to 20-fold [12].
© 1991 Elsevier Science PublishingCo., Inc., 655 Avenue of the Americas, New York, NY 10010
69 GATA 8(2): 67-74, 1991
Fluorescence in situ Hybridization
(
d
C
:~
k
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
70
GATA 8(2): 67-74, 1991
Whole Chromosome Probes F I S H with c h r o m o s o m e - s p e c i f i c c o m p o s i t e probes whose individual elements have sequence homology at many sites along one chromosome type enables an entire chromosome to be stained either in metaphase spreads or in interphase nuclei, since the various elements of the probe bind all along the target chromosome type. One useful set of whole chromosome probes was produced by subcloning the inserts from complete-digest HindlII chromosome-specific recombinant DNA Charon 21A phage libraries into B l u e s c r i b e plasmids [28, 35]. The phage libraries were produced by the National Laboratory Gene Library Project [36] from chromosomes of a single type purified by fluorescence-activated sorting. Subcloning of the phage inserts into plasmids reduced the amount of vector DNA associated with each human insert. These probes are referred to as "painting" probes, since they are used to stain (paint) whole chromosomes [1]. These probes contain families of repeat sequences (Alu and Kpn) that are on the chromosome from which the library was made and which are shared by other chromosomes. To achieve the desired staining contrast, these sequences are prevented from hybridizing by adding unlabeled blocking DNA, in many cases just human genomic DNA, to the probe mixture. Figure lb shows that FISH with these probes stains the target chromosome in metaphase spreads more or less uniformly. These probes do suffer from some deficiencies, however [35]. Some of these libraries are not complete, presumably because some chromosomal regions do not have frequently spaced HindlII sites and thus were not cloned into the Charon-21A phage. Thus, regions of some c h r o m o s o m e s are not stained by FISH with these probes. In addition, probe labeling (usually by incorporating biotinl l-dUTP or d i g o x i g e n i n - l l - d U T P during nick translation or random primer extension) can critically affect the quality of the hybridization signal. We have found it important that the probe fragments range in size from 300 to 1000 bp on a nondenaturing agarose gel. Variability in commercial nick translation kits can lead to inconsistent results. Some of the limitations of the Bluescribe composite probes can be overcome with new generation composites. One promising approach now being pursued is to produce more complete, easily labeled libraries using the polymerase chain reac-
D. C. T k a c h u k et al.
tion. In our laboratory, this is accomplished by starting with DNA from a few thousand chromosomes of a single type purified by sorting [37]. This DNA is digested to completion using one or more restriction endonucleases that recognize four bases (for instance, HpalI). An adapter double-stranded oligonucleotide is ligated onto both ends. All elements are amplified by PCR using a primer to the adapter sequence to produce a library of high complexity containing almost entirely human DNA. This can be labeled by incorporation of the modified nucleotides (biotin or digoxigenin) in the final stages of the PCR amplification. A related approach to whole c h r o m o s o m e probe generation is to amplify sequences in human-rodent hybrid cells by PCR using primers homologous to conserved parts of the Alu repeat. Several kilobase-long regions flanked by Alu sequences can be amplified using this procedure. The composite probes produced by applying this approach to hybrids carrying a single human chrom o s o m e or fragment t h e r e o f can be used as probes to selectively stain human chromosomes in human cells [38].
Locus-Specific Probes Once loci important in genetic disease have been identified, they can be studied using FISH with probes to these regions. Our studies to date suggest that probes for this purpose should target sequences ranging in size from 15 to 500 kb. Probes to targets smaller than 15 kb may not be reliably detected, especially in clinical samples. Probes to targets larger than 500 kb stain extended portions of interphase nuclei so that the hybridization signals on homologous chromosomes may overlap with each other or with hybridization domains of other probes. DNA sequences cloned into large insert phages, cosmids, or yeast artificial chromos o m e s have p r o v e d useful as specific locus probes. Figure lc, for example, shows the hybridization signals produced in interphase nuclei after hybridization with a probe for a 300-kb region of chromosome 18 cloned into a yeast artificial chromosome. Mapped individual clones and contiguous stretches of hundreds of kilobases of DNA from overlapping clones will become increasingly available over the next few years as the Human Genome Project progresses.
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
71 Fluorescence in situ Hybridization
GATA 8(2): 67-74, 1991
Cytogenetic Applications
Metaphase Analysis Chromosome analysis by classic cytogenetics relies on interpretation of banding patterns resulting from dye uptake of individual metaphase chromosomes. The whole genome can be surveyed at one time, but aberrations involving less than a few megabases of DNA, or which involve regions of the genome with indistinct banding, are difficult to detect. Frequently, identification of the components of marker chromosomes is ambiguous. FISH solves these problems because the staining is based on the DNA sequences in the target, and so can be optimized for particular applications. Chromosomes can be quickly and accurately identified after F I S H with repeat sequence or whole chromosome probes. Figure l d, for example, shows hybridization with a whole chromosome probe for chromosome 21 to a metaphase spread prepared from cells from an individual with Down's syndrome [28]. The presence of the extra copy of chromosome 21 can be determined rapidly and accurately. Even more important is the use of FISH with composite probes to detect and characterize structural chromosome aberrations such as translocations or marker chromosomes. Figure le, for example, shows FISH with a whole chromosome linker-adapter probe for chromosome 22 to cells from an individual with chronic myeiogeneous leukemia. The two derivative chromosomes resulting from the t(9;22) translocation are clearly visible [11]. We estimate that rearrangements involving only a few hundred kilobases can be detected using FISH with whole chromosome probes as long as the structural change results in the juxtaposition of the target and nontarget DNA sequences so that material stained by FISH and material not stained by FISH reside on the same chromosome. Subtle deletions and inversions, however, are impossible to detect using this approach. Structural aberration analysis can be made even more accurate by hybridizing with probes targeted to the vicinity of the locus of interest. For example, hybridization with a digoxigenin-labeled probe for a 15-kb region of the BCR gene on chromosome 22 (proximal to the CML breakpoint region) and a biotin-labeled probe for a 35-kb region of the ABL gene on chromosome 9 (distal to the CML breakpoint region) enables detection of the BCR-ABL fusion event associated with CML
[12]. The ABL probe was stained using Texas red-avidin and the BCR probe was stained using FITC-anti-digoxigenin. The juxtaposition of the red and green fluorescing hybridization domains on the short Phi chromosome signaling the B C R ABL fusion was clearly visible. Recent work on probe mapping using FISH suggests that the location of probes along chromosomes can be determined to within a few megabases [39]. This work suggests that FISH in metaphase can be used to detect inversions involving this amount of DNA. Deletions of the region targeted by the probe can be detected, of course, by the loss of hybridization domain, while amplification of the target region can be detected by an increase in the intensity or area of the hybridization domain, or by counting the number of resolvable hybridization sites. The latter method enables detection of duplications if the amplicon is large enough. Figure If, for example, shows FISH with a probe for the dihydrofolate reductase gene to cells that were grown for an extended period in methotrexate so that the dihydrofolate reductase gene was heavily amplified [40]. The large hybridization domain to the amplified (hundreds of times) gene is clearly visible.
Interphase Analysis A significant drawback to metaphase analysis is the requirement that the cells be grown in vitro. This is labor-intensive, time-consuming, and often results in the selection of cells that grow well in culture but may not be representative of the original population from which the cells were taken. The time and labor considerations are of dominant concern in prenatal diagnosis. Arguably, the most important aberrations found during prenatal screening are trisomies for chromosomes 21, 18, and 13 and aneuploides involving sex chromosomes [41]. Detection of numerical aberrations involving the autosomes is particularly important because they occur at relatively high frequency in live born infants and have serious consequences. Analysis of numerical aberrations involving these chromosomes using FISH to interphase nuclei reduces the time required for cell culture (and may eventually eliminate the need for this entirely), since only a few hundred representative cells are required [1, 3, 42]. Figure lg, for example, shows FISH with a whole chromosome probe for chromosome 21 to interphase amniocytes from a fetus
© 1991 Elsevier Science Publishing Co., Inc., 655 Avenue of the Americas, New York, NY 10010
72 D. C. Tkachuk et al.
GATA 8(2): 67-74, 1991
with Down's syndrome. The extra copy of the number 21 chromosome is clearly visible. However, not all cells in a population trisomic for the target chromosome show the expected three hybridization domains. This is thought to be due primarily to the chance overlap of the hybridization domains. In fact, the frequencies of trisomic human amniocytes showing three domains after hybridization with whole chromosome probes for chromosomes 21, 18, and 13 range from as low as 10% to >50% [42]. The lowest frequencies were observed in samples in which the hybridization intensity was not high so that domain definition was difficult in some cells. However, the frequency of normal cells showing more than two hybridization domains is usually
