Note: Descriptions are shown in the official language in which they were submitted.
CA 0223~39~ l998-0~-l2
WO97/18326 I PCT1l~96J0l229
ULTRAHIGH RESOLUTION COMPARATIVE NUCLEIC
ACID HYBRIDIZATION TO COMBED DNA FIBERS
BACKG,ROUND OF THE I~v~NllON
s This invention relates to the detection and quantification
of the presence of a gene in a genome. In one embodiment, the
invention relates to the detection and quantification of a
human oncogene.
A recently developed method, called comparative genomi~
hybridization (CG~) (Kallioniemi et al. 1992, du Manoir et al.
1993; Joos et al. 1993) has provided a new tool to detect non-
random gains and losses of DNA sequences in genomic DNA
(obtained e.g., from tumor specimens). For CGH, genomic tumor
DNA is labeled with a hapten (e.g., Biotin) or directly with a
l~ fluorochrome (e.g., FITC). Genomic DNA prepared from normal
cells (of the patient or other persons) is dif~erently labeled
with another hapten (e.g., digoxigenin) or directly with
another ~luorochrome (e.g., TRITC or Texas red). Labeled tumor
and control DNAs are mixed in equal amounts. This mixture is
hybridized in the presence of an excess of unlabeled cotl-DNA
to normal metaphase spreads (target chromosomes) prepared from
a healthy male or female person. The excess of cotl-DNA
hybridizes to labeled interspersed and tandem repetitive
sequences present in genomic DNA as well as to such sequences
~5 present in the target chromosomes. This step is essential to
suppress the unwanted hybridization of the repetitive sequences
to the target chromosomes (Lichter et al. 1988, Pinkel et al.
1988). Following hybridization and washing steps, standard
detection procedures are applied to visualize haptenized
sequences with two different ~luorochromes (Lichter and Cremer
1992). This step is omitted in case of DNA-probes directly
labeled with fluorochromes.
Consider for example a tumor with an essentially diploid
karyotype except for a few monosomic or trisomic chromosomes or
3s chromosome segments. Labeled DNA fragments with a size of
CA 0223~39~ l998-0~-l2
WO97/18326 2 PCT~B96/01219
several hundred base pairs ~rom the tumor DNA and the normal
control DNA will hybridize with equal probability to their
respective target sequences. The labeled DNA fragments from a
chromosome or chromosome segment present in two copies in the
pseudodiploid tumor cells together with the differently labeled
fragments from genomic DNA of diploid cells yield a certain
color mixture on the respective target chromosome or chromosome
segment, e.g., yellow when the chromosome or chromosome segment
is labeled with equal numbers of green and red fluorochromes
representing the hybridized fragments from tumor DNA and normal
DNA, respectively. For a chromosome segment present in three
or higher copy numbers in the tumor, this color would become
more greenish, while the loss of the segment would result in a
more red color.
These color changes can be quantitatively recorded by
measuring fluorescence ratio profiles alon~ target chromosomes
(du Manoir 1994, Piper et al. 1994). The choice of the
appropriate equipment to measure signal intensities is
important. Detectors should allow linear intensity
~0 measurements over a wide range. CCD-camerase are particularly
useful in this respect. All data are stored digitally so that
they can be used by microprocessor for the calculation of
fluorescence ratios. In this way, a copy number karyotype can
be established.
CGH can be performed with DNA extracted from archived,
paraffine embedded tissues. Even minute amounts of genomic DNA
can be used for this purpose after amplification with
degenerate oligonucleotide primers (DOP-PCR) (Telenius et al.
1992; Speicher et al. 1993i Isola et al. 1994). It is possible
to microdissect areas containing tumor cells from a tissue
section and screen it for gains and losses of genetic materials
by CGH performed with DOP-PCR amplified DNA (Speicher et al.
1994).
The number of studies, which demonstrate the usefulness of
3~ CGH to detect DNA copy number changes in tumors, is rapidly
CA 0223~39~ 1998-0~-l2
WO97~l8326 3 PC~B96l012Jg
increasing. Studies published so far reflect already a variety
of tumors, including various forms of acute and chronic
leukemias, bladder cancer, breast cancer, colorectal cancer,
gliomablastoma, kidney cancer, neuroblastomas, prostate cancer,
small cell and non-small cell lung carcinomas, uvea melanomas
(e.g., du Manoir et al. 1993, 1994; Isola et al. 1994a,b; Joos
et al. 1993; Kallioniemi et al. 1992, 1993, 1994a,b; Muleris et
al. 1994; Ried et al. 1994; Schrock et al. 1994, Speicher et
al. 1993, 1994, 1995 and our unpublished data). Numerous
hitherto unknown regions, in particular amplification sites,
have been found in these studies and will become the focus of
efforts to clone the respective tumor relevant genes.
Different tumor entities generally show distinctly different
patterns of non-random changes and clinical follow up studies
1~ will show to which extent specific gains and losses can be
correlated with the clinical course and prognosis of a given
tumor.
The minimum size of a chromosome segment for which a
single copy number change can be detected at present by CGH is
in the order of lO Mbp (Joos et al. 1993; du Manoir et al.
1994; Piper et al. 1994). Possibly, the resolution can be
somewhat improved, when CGH is performed on prometaphase
chromosomes. For amplified DNA sequences the detection limit
of CGH is presently about 2 Mbp (number of amplification
'~ repeats times amplicon size). Still the precision with which
the borders of chromosome segments involved in gains or losses
is limited by the banding resolution of the target chromosomes.
Thus, the current status of the CGH development does not allow
to define the copy number representation of single tumor
relevant genes (e.g., oncogenes, tumor. suppressor genes). This
limits the application of CGH to the detection of gross,
unbalanced chromosomal abnormalities.
SU~RY OF THE INVENTION
This invention aids in ~ulfilling these needs in the art
3~ by providing a method of ultrahigh comparative genomic
CA 0223~39~ 1998-0~-12
W097/18326 4 PCT~B96tO1219
hybridization, which differs from state-of-the-art comparative
genomic hybridization by the following new and essential
features. The approach is performed on combed DNA fibers
instead of reference chromosomes and referred to as combed
5 fiber CGH. Comber fiber CGH allows the analysis of copy number
representation of specific sequences (represented by the combed
DNA fibers) in a genomic test DNA with an ultrahigh resolution
(in the kb-pair range instead of the Mb-pair range as
previously published methods). This improvement makes combed
fiber CGH a very useful method to study the copy number
representation of single genes or parts thereof.
Combed fiber CGH is particularly suited to eliminate
background problems in fluorescence measurements, which arise
when the fluorescence is measured from entire DNA-spots. In
the case of combed fiber CGH, the area of fluorescence
measurements is adapted to a single fiber in a way that only
hybridization dots located precisely on the DNA fiber
contribute to the measured signal derived from the test or
reference genomic DNAs. This improvement provides a very
~0 considerable advantage of combed fiber CGH as compared to a CGH
approach where fluorescence measurements are obtained from
entire DNA-spots of non-ordered target DNA se~uences attached
to a supportive matrix.
In addition to fluorescence intensity measurements, which
~5 can be carried out on individual combed target DNA fibers by
standard procedures, combed fiber CGH allows the counting of
hybridization dots located on the combed DNA fibers. By this
approach the total number of dots on a sufficiently large
series of target DNA fibers resulting from hybridized test and
a reference genomic DNA can be counted and a ratio (or a
difference) between the total dot number from the test genomic
DNA and the total dot number from the reference genomic DNA can
be calculated as a measure of the copy number representation of
the combed target DNA ~equence in the test and reference
genomic DNA. This approach has not be realized so far in CGH
CA 02235395 1998-05-12
WO g7/18326 5 PCTJIB96J01219
experiments and allows the evaluation of combed fiber CGH
experiments in situations where the resulting signals are too
weak to allow meaningful measurements of fluorescence ratios on
single DNA fibers.
Instead of 1:1 mixtures of test and reference genomic
DNAs, 1:1 mixtures of RNA preparations (or from cDNAs
synthesized ~rom RNA preparations) representing e~uivalent
numbers of test and reference cells can be hybridized under
suppression conditions (see standard CGH procedures) to combed
lo target DNA fibers e.g., cDNA fibers representing the coding
sequence of genes of interest. The evaluation of the
experiment is performed as described above. This approach
allows an estimate of the relative copy number representation
of mRNAs in test cells (e.g., tumor cells) and reference cells
Is (e.g., normal progenitor cells of the tumor cells).
Target DNA fibers, e.g., cosmids containing (part of) a
gene of interest are genetically engineered in a way that
interspersed repetitive sequences are removed to avoid problems
of insufficient suppression hybridization. (Note: only target
~0 fiber specific single copy sequences, but not interspersed
repetitive sequences contained in the hybridization mixture,
can hybridize to the combed target DNA fibers under these
precautions.)
In case the localization of individual target DNA fibers
~5 cannot be easily identified by arrays of numerous signal dots
along each fiber, several procedures can be followed for an
unequivocal target fiber identification with fluorochromes that
have an emission spectrum, which allows a clear distinction
from the emission spectrum of the fluorochromes implied in the
30 visualization of hybridized test and reference genomic DNA
fragments. a) Fibers can be stained with appropriate
fluorescent DNA stains. b) Target fiber DNA can be cloned in
the presence of fluorochrome labeled DNA nucleotides for the
dlrect visualization of the fiber or in the presence of hapten
35 modified nucleotides, e.g., BRdU, biotin, digoxigenin, for
CA 0223~39~ 1998-0~-12
WO97/18326 6 PCT~B96/01219
indirect visualization by, e g., indirect immuno~luorescence.
c) Target fiber DNA can be visualized by the addition o~ an
appropriate amount of labeled target DNA sequences to the
hybridization mixture employed for combed fiber CGH. In this
S case, the amount of labeled target sequences should not be
excessi~e in a way that suppresses the hybridization of target
specific labeled test and genomic DNA sequences. d) Target
fiber DNA containlng vector sequences can be hybridized with
labeled vector sequences added to the hybridization mixture.
If desirable, linker DNA of different length can be added to
target ~iber DNA in a way that makes it possible to identify
the course and orientation of the combed target fiber by
signals derived from the addition of labeled linker DNA
sequences to the hybridization mixture.
1~ Instead of fluorochromes with different emission spectra,
fluorochromes that differ in fluorescence lifetime can be used.
The choice o~ fluorochromes is performed with particular
reference to avoid background fluorescence from the supportive
matrix as much as possible.
~0 DETATT~n DESCRIPTION OF PR~:h~KK~ EMBODIMENT
This invention involves the use of a DNA alignment method
for the detection and quantification of multiple copies of a
gene present in a genome. By multiple copies is meant
visualization of at least about 100 copies per genome, with
very good results at about 2000 copies per genome.
Ultrahigh Resolution Comparati~e
Nucleic Acid Hybridization To Combed DNA Fibers
In the following part we describe the development of a
CGH-test providing a resolution at the single gene level. This
test can be ~ully automated and broadly used in clinical
settings.
The new test is based on the idea that, instead of enti~e
chromosomes, specific target nucleic acids (DNAs or RNAs~ are
immobillzed on a supportive matrix, such as glass or plastic
CA 0223~39~ 1998-0~-12
WO 97J18326 7 PCTIIB96J012~9
materials, in any desirable geometric format. The number of
target nucleic acid (TNA-) spots and the sequence complexity of
each TNA-spot can be chosen with regard to the specific goals
of a test (see the application examples below). The number of
TNA-spots included in a given matrix-CGH test may vary from a
few spots to hundreds or even thousands of spots (for potential
applications see below).
A typical TNA-spot may contain DNA from a single cosmid
representing a gene or part of a gene of interest or it may
10 contain a complex mixture o~ DNA representing a chromosome
segment or even an entire chromosome of interest. In the
latter case a matrix CGH test would not provide a resolution
superior to the resolution of CGH to reference metaphase
chromosomes. In the following we will construct our
15 considerations mainly to the development of a matrix CGH test
with the highest conceivable resolution, i.e., a test to detect
copy number changes in a set of selected genes.
A matrix with TNA-spots as described above can be used to
test tumor or other test DNAs for genetic imbalances down to
~0 the kbp-range. For this purpose, the hybridization probe
consisting of a 1:1 mixture of differently labeled test and
reference genomic DNAs (or RNA- or cDNA-preparations) is
hybridized under suppression conditions against the set of
immobilized TNA-spots. Measurements of the fluorescence ratio
on each individual TNA-spot should provide an estimate of the
copy number representation of the respective target sequences
in the test DNA (or test RNA) as compared to the reference DNA
(or reference RNA) (for further details of measurements see
below).
The successful development of such a test depends on three
requirements, namely the ability to firmly immobilize target
nucleic acids on a supportive matrix, e.g., glass or plastic, a
low autofluorescence of the matrix, and a sufficiently high
signal/noise ratio for se~uences specifically hybridized to a
~5 given TNA-spot. Notably, in CGH experiments with two
CA 0223~39~ 1998-0~-12
WO97/18326 8 PCT~B96/01219
differently labeled genomic DNAs the fraction of labeled DNA
fragments, which are specific for a given TNA-spot, is
generally very small. Any non-specific attachment of labeled
sequences or detection reagents to the matrix may impair or
even inhibit the measurement of meaningful fluorescence ratios.
Such adverse effects may become a limiting factor in attempts
to measure fluorescence ratios on entire TNA-spots,
particularly in cases where the specific signal is relatively
small as compared to background.
In order to minimize background problems in fluorescence
ratio measurements we make use of a procedure called "molecular
combing" (Bensimon et al. 1994), which is relied upon and
incorporated in its entirety by reference herein. By this
procedure DNA target fibers can be extended and aligned in
parallel like hairs by the use of a comb. To this end, the DNA
fibers are attached at one end to a solid surface, "combed" by
a receding air-water interface, and finally immobilized on the
drying surface. (Bensimon et al. 1994). In this way the DNA
of each TNA-spot can be represented by a series of ~'combed"
~0 target DNA fibers, all representing a specific DNA sequence of
interest, e g., a cosmid clone from a gene of interest. Using
standard fluorescence in situ suppression hybridization
techniques under appropriate stringency conditions,
complementary sequences present in the hybridization probe
~5 hybridize specifically to these target sequences.
For each TNA-spot, fluorescence is separately recorded for
both fluorochromes on a series of individual combed target DNA
fibers using an appropriate camera, such as a CCD-camera. For
evaluation, each target sequence is enclosed in a narrowly
adapted rectangular field to determine the fluorescence of the
fluorochromes applied in the labeling/visualization of the
hybridized DNA fragments. Background fluorescence is measured
in the immediate neighborhood of each measured target sequence
and subtracted. After background substraction, the
,5 fluorescence ratio is determined for each individual target
CA 0223~39~ 1998-0~-12
WO 97118326 9 PCT~B9610121g
sequence with a microprocessor. Foe each TNA-spot the
variation of fluorescence ratios obtained for a number of
combed target DNA fibers is determined. Target sequences,
which are represented in normal copy numbers in both the test
and the reference gnomic DNA, are used to obtain reliable
thresholds for ratios indicative for increased or decreased
copy number representation. For TNA-spots representing target
sequences, which are over- or underrepresented in the test-DNA,
one should then obtain correspondingly increased or decreased
fluorescence ratios, while balanced regions should yield a
ratio within the limits o~ control experiments. Since
fluorescence ratios are recorded from individual target DNA
fibers present in the TNA-spots, the new test is inert against
variations in the total amount of target-DNA in each TNA-spot.
IS In model experiments, probe mixtures consisting of
different ratios of biotin labeled and digoxigenin labeled
cosmid sequences (e.g., 1:1 (20 ng + 20 ng), 2:1 (20 ng + 10
ng), 5:1 (20 ng + 4 ng) and 10:1 (20 ng + 2 ng)) were prepared.
For each chosen ratio of sequences in the probe mixture, the
same cosmid was used as target sequence. Following comparative
hybridization some fifty target DNA fibers were evaluated as
described in the Examples. The results demonstrate highly
significant differences of mean fluorescence ratio values
obtained for the different probe mixtures.
~S The use of individual target DNA fibers provides the
possibility of another approach for the evaluation of a CGH
experiment. Instead of measuring fluorescence intensities from
entire target DNA fibers, it is possible to simply count
individual fluorescence hybridization dots. Each such dot
presumably represents the hybridization of an individual probe
DNA fragment of a few hundred bp. The coverage of the combed
target DNA sequence with differently colored dots from the 1:1
hybridization mixture of genomic DNAs is a stochastic event.
- The dot numbers for the two differently labeled genomic DNAs
counted over a series of target DNA DNA fibers therefore
CA 0223~39~ 1998-0~-12
WO97/18326 10 PCT~B96/01219
re~lect the copy number representations of the target DNA
sequence in the two genomic DNAs used for comparative
hybridization.
Notably, using such an approach it is not required to
achieve sufficient signal from the different labeled genomic
DNAs over each target sequence to allow meaningful fluorescence
measurements. Consider that the test genomic DNAs on average
yields five specific hybridization dots, while the reference
genomic DNA yields one specific dot per target DNA sequence.
Then the conclusion seems valid that the copy number
representation of the target sequence in question is five times
higher in the test genomic DNA.
The ratio of dots from test and reference genomic DNA
counted over a series of target DNA DNA fibers contained in a
1~ given TNA-spot may deviate from the actual ratio of copy number
representation of the target sequence in the test and reference
genomic DNAs for a number of reasons. The differently labeled
DNAs in the l:l hybridization mixture should be digested to the
same size distribution. The number of background dots in the
70 vicinity of the target sequence should be approximately the
same for both the test and reference genomic DNA. If
necessary, the number of dots, which are expected to result
from chance background dots on the target DNA sequence, should
be calculated from the area surrounding the target DNA sequence
and subtracted from the overall number of counted dots. Only
those background dots that are ordered e~actly in line along a
combed target DNA sequence can be confused with actual signal
dots. The proposed approach to measure dots along combed DNA
target DNA fibers thus helps strongly to minimize the number of
background dots, which could deteriorate the accuracy of
counting the number of speci~ic hybridization dots along a
series of combed target DNA fibers. This is a decisive
advantage of the proposed test as compared to a test where
fluorescence ratios are determined from an entire DNA-spot
" built up by a large number of non-combed DNA fibers.
CA 02235395 1998-05-12
WO97J18326 11 PCTJ1B96101219
Further, it should be emphasized that the proposed test
allows the comparison of fluorescence or dot number ratios over
a number of TNA~spots. TNA-spots, which contain combed Target
DNA fibers present in equal copy number in both the test and
reference genomic DNA, can serve to standardize the
fluorescence ratios and dot ratios, respectively, not only with
regard to the internal standardization in each individual TNA-
spot given by the fluorescence or dot number obtained from the
reference genomic DNA, but also with regard to the
standardization of the data between different TNA-spots
representing combed target DNA fibers present in equal and
different copy numbers in the test and the reference genomic
DNA.
In case that each target DNA DNA fiber is covered with
numerous specific dots, the target fibers can be easily
distinguished as a linear array of dots. The unequivocal
identification of the combed target DNA fibers is an absolutely
essential requirement to count small numbers of dots, where a
dot number ratio (or difference obtained by subtraction) is
~0 only meaningful when obtained from a series of target DNA
fibers. Therefore the target fibers need to be visualized by
other means when the number of dots is small, including
specific DNA fluorochromes with an emission spectrum
distinguishable from the fluorochromes used for the
~5 identification of hybridized DNA ~or RNA) fragments.
Target DNA fibers can also be visualized by the admixture
of labeled target DNA to the hybridization mixture. In this
case a third label is required in addition to the two labels
for the test and reference genomic DNA. The admixture of
labeled target sequences has to be carefully adjusted in order
to avoid too much suppression of the hybridization of the
labeled target sequences present in the test and reference
genomic DNAs. This problem can be avoided, if linkers are
adapted to the target DNA sequence, which can be visualized by
hybridization with a specific linker probe. The combed target
CA 0223~39~ 1998-0~-12
WO97/18326 12 PCT~B96/01219
DNA sequence in question would then be embraced by two
fluorescently labeled linker sequences. Alternatively, target
DNA sequences can be cloned in the presence of hapten modified
nucleotides, such as BrdU, and visualized immunocytochemica~ly.
For the choice of useful target DNA sequences one should
take into account that sequences that contain interspersed
repetitive signals, e.g., Alu elements, require suppression
hybridization, e.g., with an excess of unlabeled Cotl-DNA. It
might be preferable to use target DNA fibers, which are
entirely specific for the genomic region in question, or to
construct target DNA fibers devoid of interspersed repetitive
elements.
CGH on combed DNA fibers bears the potential for an
ultrahigh resolution CGH. ~onsider the following scenario: A
DNA fiber with known DNA sequence contains a target region of
interest comprising a few hundred base pairs. We assume that
the copy number of this target region is variable and may be
higher (or lower) in the test genomic DNA as compared to the
reference genomic DNA. We assume further that the positions of
the target and control regions along the DNA fiber are
precisely mapped and that the DNA fiber is engineered in a way
that its 5' - 3' orientation can be visualized, e.g., by probes
to linker adapters of different size. In such a scenario the
target region could be mapped by fractional length measurements
on each DNA ~iber. Accordingly, one could identify dots which
represent hybridization events to the target region and count
the number of such events on a series of DNA fibers. This
number should correlate with the representation of target
sequences in the genomic test and reference DNA. In case that
the number target sequences is increased in the test genomic
DNA as compared to the reference genomic DNA, the respective
dot ratio should be increased over this region in contrast to
other regions of the DNA fiber, for which we assume an equal
copy number representation in both the test and reference
genomic DNA.
CA 0223~39~ 1998-0~-12
WC) g7/18326 ~3 P~rn~g6)~2~g
As a model system one could prepare PCR-amplified probes
for two regions of a given DNA-fiber. DNA aliquots from each
probe could be differently labeled and various hybridization
mixtures prepared, where the differently labeled aliquots are
present in different ratios. By counting the hybridization
dots over the two target regions on a series of DNA fibers, one
could determine to which extent the dot number ratios reflect
the ratios of the dif~erently labeled probe aliquots. This
example may illustrate that an approach based on dot counting
may be feasible to determine the copy number representation of
very small target sequences where fluore5cence ratios are no
longer valid. This advantage is reemphasized by the following
scenario.
Consider single-stranded DNA sequences containing a small
1~ (300 bp) target sequence of interest. Four CGH probe mixtures
with differently labeled complementary 300 bp sequences are
applied. On each individual fiber the target sequence of
interest can only hybridized once, i.e., it will bear a dot of
one color only. If the target is represented by double-
stranded, denatured DNA, both strands serve as targets for adenatured double-stranded probe. Accordingly, the target
region can hybridize with two 300 bp ~ragments at most. The
resulting dots are either of the same color or of different
color. While the measurement of a fluorescence ratio over the
~5 target region of a single fiber is obviously meaningless, the
frequency with which the target region is covered by dots of
the same (e.g., green or red) or of different colors (e.g.,
yellow) in a series of DNA fibers should be highly informative
with regard to the frequency of differently labeled target
sequences in the hybridization mixture.
Alternatively, DNA representing a target sequence of a few
hundred base pairs could be fixed to the matrix the target spot
and a fluorescence ratio could be determined from the entire
spot. In this case, however, bac~ground could become a major
3~ problem. From these considerations we conclude that a dot
CA 0223~39~ 1998-0~-12
W097/18326 14 PCT~B96/01219
counting approach performed on a sufficient number of
individual DNA fibers may be superior or even the only feasible
way in case of ultrahigh resolution C~H.
In contrast to the largely variable size, form, and
relative position of individual target chromosomes considered
in CGH on reference metaphase spreads, the length and
orientation of a combed target DNA fibers can be strictly
controlled. These defined patterns of combed target DNA fibers
strongly facilitate their fully automated evaluation. In case
of a large number of TNA-spots, a colored print-out of
fluorescence ratio or dot ratio measurements is recommended to
facilitate the investigator's recognition of genomic regions,
which are over- or under represented. In this print-out, each
TNA-spot is represented by a colored spot. One color should
reflect TNA-spots with the range of fluorescence ratios
apparently representing sequences present in balanced copy
number in the test-DNA, a second color should reflect TNA-spots
with sequences present in increased copy number, while a third
color should reflect TNA-spots with sequences present in
~0 decreased copy number. If desirable, color intensity may
reflect the relative extent of over- or underrepresentation.
For a series of TNA-spots containing physically mapped se-
quences, the color spots could be arranged in a way that
reflects mapping positions on chromosomes. For example, a
linear array of colored spots could represent the order of
clones in a contig used for combed DNA fiber C5H. In case
that the whole chromosome c~mplement is represented by TNA-
spots, the resulting color spots could be ordered as 24 linear
arrays (representing chromosomes 1-22, X and Y). The color
spots within a given array then could represent the physical
order of clones within the respective chromosome. The
investigator then is enabled to see at one glance which
chromosomes or chromosomal subregions are present in balanced,
increased or decreased copy number.
CA 0223~39~ 1998-0~-12
WO 97/18326 ~5 PCT/IB96101219
Potential Applications Of CGH On A
Matrix with Com~ed Taraet DNA Fibers
In this section we will consider potential applications of
CGH on combed target DNA fibers. In addition to potential
S studies of tumor DNA samples, we will also include applications
in clinical genetics and cytogenetics
Tests with gene specific TNA-spots would allow the rapid
screening of test DNAs from tumor samples for copy number
changes of specific genes. Given the appropriate equipment for
lo the automated evaluation of TNA-spots, it may become feasible
to evaluate DNA spot matrices with numerous spots representing
the copy number representation of an entire set of genes of
interest at a reasonable price.
A survey of whole genomes at the highest possible level of
resolution would require such a high number of spots that such
an approach appears impractical. For practical purposes, it
may be advantageous to perform a survey of a test genomic DNA
with unknown gains and losses by a series of tests with
increasing resolution, starting by CGH on metaphase spreads and
~0 subsequently homing in on specific chromosome segments. In
addition to CGH on reference chromosomes, matrices with spots
representing composite DNA sequences of entire chromosomes,
chromosome arms or bands may be applied. CGH performed
directly on target DNA fibers provides the ultimate level of
~5 resolution for CGH and is only reasonable in cases where the
screening of a very specific subset of target DNA sequences for
copy number changes in a test genomic DNA is required. The
following examples may illustrate how such a strategy could be
applied.
CGH on reference chromosome spreads may reveal, for
example, the non-random loss of a certain chromosome segment
for a certain tumor entity. A consensus region can be defined
by the comparison of all tumors showing this deletion.
However, a considerable fraction of the tumors may not show any
detectable deletion at this level of resolution (>10 Mbp). To
CA 0223~39~ 1998-0~-12
WO 97/18326 16 PCT/IB96/01219
screen these tumors for much smaller deletions, a matrix with
TNA-spots representing the consensus region could be used. The
resolution would depend on the size and linear, genomic
distance of the target sequences represented by a given matrix.
In many cases it should be sufficient ~or screening purposes to
represent a chromosome region by a series of TNA-spots, where
each spot defines ~or example a cosmid sequence a few hundred
kb apart ~rom the target DNA sequence contained in the next
TNA-spot.
To achieve the highest, possible resolution even a whole
contig can be represented by a series of TNA-spots.
~luorescence ratio measurements performed on a high resolution
matrix representing a region of interest should help to de~ine
the cosmids, which represent the smallest deletion detectable
in this region of interest for a whole series of genic test
DNAs obtained from patients with a specific tumor entity. ~his
minimum deletion could be confirmed by FISH of the respective
cosmids to tumor nuclei.
In this way, e~forts of positional cloning o~ a suspected
tumor suppressor gene could be strongly facilitated. For such
a purpose it would not be necessary to know the precise linear
order o~ the cosmid clones representing the chromosome segment
of interest. If, say, three TNA-spots with decreased
fluorescence ratio would define the minimal detectable
deletion, one would expect that FIsH with these three clones to
extended chromatin fibers would confirm their vicinity. The
latter approach could also be used to map the linear order of
these cosmids.
Similarly, matrices with physically mapped cosmids
representing a chromosomal subregion could help to define
amplified regions. Positional cloning of the genes involved in
amplifications would be greatly ~acilitated, if the extension
o~ such amplifications could be precisely mapped. Consider
that a chromosome band has been identified as the source of the
ampli~ied sequences by CGH to reference metaphase spreads.
CA 0223~39~ l998-0~-l2
WO 97/18326 17 PCT/lB96lO1219
Applying a matrix with a series o~ physically mapped clones
representing the chromosome band in question should yield
increased fluorescence ratios for any TNA-spot representing DNA
sequences, which are in fact ampli~ied. Normal ~luorescence
ratios should be measured for all other TNA-spots. The extent
of the amplification could then be defined by the two TNA-spots
with increased fluorescence ratios, which represent the most
proximally and most distally mapping clones.
In the future, matrices containing TNA-spots with combed
lo target DNA ~ibers for the copy number representation of
oncogenes and tumor suppressor genes can be developed. The
choice of the target sequences for a given matrix will depend
on the tumor entity and the demands of the test. For example,
matrices can be specifically developed to identify gains or
losses with prognostic value (e.g., N-mye amplifications or
lp36- deletions in neuroblastomas). For some tumors, e.g.,
colorectal tumors, knowledge about the relevant genes involved
in tumor initiation and progression seems already sufficiently
advanced to consider the development of such a strategy. For
other tumors we still lack such knowledge. CGH on combed DNA
fibers may help to obtain such knowledge in the future and to
perform large scale tests performed with the aim to correlate
the patterns of relative copy number changes of genomic DNA
sequences (and changes in the number of speci~ic mRNAs) in
tumor cells with the clinical course of the disease. Ideally,
matrices should be developed, which contain TNA-spots
representing all genes that are relevant for the biological
properties of the tumor entity in question.
High resolution matrices could also open new avenues in
clinical cytogenetics. Two examples may be sufficient to
demonstrate the range of possible applications. A CGH test-
matrix could be developed to screen DNA from patients with
phenotypes suspicious for unbalanced chromosome aberrations.
Taking into account that unbalanced rearrangements often
3~ include terminal chromosome segments, a CGH-matrix containing
CA 0223~39~ 1998-0~-12
WO97/18326 18 PCT~B96/01219
TNA- spots with combed target DNA ~ibers representing cloned
sequences from each individual chromosome end may become a
great clinical value. As another example, consider the case of
a carrier-analysis for X-linked recessive diseases. A boy, who
su~fers from Duchenne muscular dystrophy, may be the first
victim of that disease in a family. In some 60~ of the cases a
deletion can be found as the cause of the mutation. The
question has then to be answered whether his disease is due to
a germ cell mutation or whether his mother is already a
carrier. The consequences for genetic counseling are totally
dif~erent and in the latter case other ~emale members, e.g.,
the sisters of the boy's mother, may also be concerned about
their carrier-status. A matrix with a series of cos~lds
spanning the entire dystrophin gene could potentially provide a
reliable and automated procedure for carrier screening in
deletion prone cases of DMD.
Comparative ~NA Hybridization On Combed
DNA Fibers: A New Method For The Assessment
0~ Expression Levels O~ Tumor Relevant Genes
While CGH on combed DNA fibers should provide in~ormation
on deleted or ampli~ied genes, it would not detect the
silencing or overexpression of genes in tumor cells as compared
to their normal counterparts. We propose an approach to study
the expression status of genes by comparative RNA hybridization
~5 on combed DNA fibers. Consider for example a scenario where an
amplification is detected in a given tumor entity. Amplicons
may be large and contain several genes. It may not be clear
which gene(s) of these genes are strongly expressed. A DNA-
spot matrix containing combed cDNA-fibers for the coding
sequences of all genes in question can be used ~or comparative
nucleic acid hybridization with di~erently labeled RNA-
preparations (or corresponding cDNA preparations) from the
tumor and a normal reference tissue. The resulting
fluorescence or dot number ratios then can provide insight in
3~ the (relative) expression status of the tested genes in the
CA 0223~39~ 1998-0~-12
Wo 97~18326 19 PCTnB96101219
tumor as compared to normal tissue. The same approach could be
used to detect point mutations which interfere with the
description of a gene.
This invention will now be described in greater detail in
the following Examples.
EXAMPh~ 1
Preparation Of Glas~3 Slides With Combed DNA-fibe:~s
Any cloned DNA, PCR-amplified DNA or other puri~ied DNA
can be used for DNA-combing depending on the purpose of the
comparative hybridization experiment. The amount of DNA
necessary per TNA-spot is small, since only a small number of
DNA-fibers is needed for evaluation is combed fiber CGH
experiments (for example, 1 ng of cosmid DNA (40kb~ contains
2.4 x 102 molecules). In the following we describe a typical
experiment with cosmid DNA.
Attachment of target DNA fibers requires glass surfaces
pretreated by salinlzation as described previously
(Bensimon et al. 1994). Prior to their attachment, cosmids can
be stained for 1 hour at room temperature with YOY~-l
~0 (Molecular probes; Cat. No. Y-3601). Staining solution was
freshly prepared as follows: l~ml DNA (lmg/ml) + 33~ml YOYO-1
diluted 1:100 in T40E~2 + 66 ~ml T40E~2; t40E~2 buffer contains 40
mM Tris-Acetate, 2 mM EDTA, pH 8.0). YoYo-1 stained cosmid DNA
was diluted in 50 mM MES-buffer, pH 5.5, and then attached to
~5 the glass surface (note that the pH is a most critical point
for successful attachment).
In our present experiments, each glass slide (22 x 22 mm)
contained only one type of combed DNA fibers. Where
appropriate, several glass slides containing DNA fibers with
different sequences were processed in parallel.
For the preparation of a single matrix with a series of
TNA spots with combed DNA fibers, salinization of the glass
surface can be restricted to the areas selected for the
positions of the TNA spots. A series of droplets, each
containing DNA fibers with the required DNA target sequence for
CA 0223~39~ 1998-0~-12
WO97/18326 20 PCT~B96/01219
a given area, can be put on these preselected areas. Target
~ibers contained in each droplet are allowed to attach to the
surface. The excess fluid with non-attached fibers is removed
taking care that the surface is kept wet.
Following DNA fiber attachment, combing was carried out as
described (Bensimon et al. 1994). Microscopic visualization of
YoYo-1 stained fibers allowed a control of the density and
direction of attached fibers. Slides with combed DNA fibers
were baked at 60~ C for at least 4 hours and treated with a
"blocking solution" consisting of 3~ BSA in 2xSSC at 37~C for
30 min. After a brief wash with 2xSSC, slides were put through
a series of 70~, 9O~, 100~ E:OH, 2 min each and air dried.
Storage o~ the dried slides is recommended in sealed boxes at
+4~C.
I~ EXAMPLE 2
Comparative Hybridization To Combed DNA-fibers
Comparative genomic hybridization (CGH) to combed DNA
fibers was essentially carried out as described elsewhere for
CGH to metaphase chromosomes (du Manoir et al. 1993, 1995).
Briefly, test and reference genomic DNAs were nicktranslated
with biotin and digoxigenin, respectively. Alternatively, the
DNAs can be labeled directly with appropriate fluorochrome
conjugated nucleotides. Combed DNA fibers were denatured for 2
min. at 72~C in 70~ FA/0.6xSSC, pH7Ø Thereafter, slides were
put through a series of ice cold ErOH (70~, 90~, 100~) and air
dried. Ten ~ml hybridization mixture (containing 500 ng each
of the test and reference genomic DNA, 50 ~mg o~ Cotl fraction
of human DNA (BRL/~ife Technologies) and S5 ~mg sonicated
salmon testes DNA (Sigma) in 50~ formamide, lxSSC and 10~
dextrane sulphate) were put on each glass slide (22 x 22 mm)
with combed DNA fibers. Another slightly smaller glass slide
(18 x 18 mm) was put on top and sealed with fixogum.
Hybridization was carried out overnight at 37~C. Washing
and detection procedures for biotin and digoxigenin labeled
3~ sequences were carried out as described (Lichter and Cremer
,
CA 0223~39~ 1998-0~-12
Wl~ 97~18326 ~ I pcTn~s6~ol2l~
1992, du Manoir et al. 1993, 1995) with minor modifications.
Slides were washed 3 x 5 min. with 50~ FA/SSC and another 3 x 5
min. with 2SSC at room temperature. Following equilibration in
~xSSC/0.1% Tween 20 at 37~C slides were incubated ~or 30 min.
r S with 3~ BSA/4SSC/0.1~ Tween 20 at 37~C ~a blocking step to
reduce background). Slides were then washed for 5 min. in
4xSSC/0.1% Tween 20 at 37~C and incubated with avidin DCS
conjugated to FITC (vector Laboratories) for 45 min. at 37~C to
visualize biotin labeled probes.
Digoxigenin-labeled probes were detected by incubation
with mouse-anti-digoxin IgG antibodies as the primary antibody
(Sigma), followed by incubation with a sheep-anti-mouse IgG
antibody conjugated with the fluorochome CY3. In between these
steps slides were washed 5 min. each in 4xSSC/0.1% Tween 20 at
l~ 37~C. If necessary, avidine-FITC signals were amplified as
described (Pinkel et al. 1986). Finally, slides were air-dried
and mounted in Vectashield (Vector Laboratories) as an
antifade.
In a series of model experiments, probe mixtures
consisting of different ratios of biotin labeled and
digoxigenin labeled cosmid sequences (e.g., 1:1 (20 ng + 20
ng), 2:1 (20 ng + 10 ng), 5:1 (20 ng + 4 ng) and 10:1 ~20 ng +
2 ng) were used for in situ hybridization to combed DNA fibers
representing the same cosmid as target sequence.
EXAMPLE 3
Evaluation Of Combed DNA Fibers
Subiected To Comparati~e Hybridization
(A) Ima~e acquisition
Grey-level images were recorded with a cooled black and
,0 white CCD camera (Photometrics) separately for each
fluorochrome. Optimal exposure times and optical settings were
established empirically and then kept constant for the entire
set of DNA fibers recorded for a given experiment. Images were
stored under FITS-format.
CA 0223~39~ 1998-0~-12
WO 97/18326 ~2 PCT/IB96/01219
(B) Imaqe processinq
(1) Measurement of fluorescence ratios
Digital images were processed by NIH image (version
1.59b9). A rectangular mask was adapted to each DNA fiber.
Within the mask two integrated fluorescence values were
obtained as the sum of the grey level values for all pixels.
Background fluorescence intensity was determined after shifting
the mask to the immediate neighborhood of a given DNA fiber.
Fluorescence intensities were corrected by background
subtraction. The fluorescence ratio for each fiber was
calculated by dividing the corrected FITC fluorescence
intensity with the corrected Cy3 fluorescence intensity. Fifty
DNA ~ibers were evaluated in a typical experiment to calculate
the mean fluorescence ratio value.
(2) Dot counting
Notably, the labeling observed along combed DNA fibers in
the experiments described above is not homogeneous. Instead
signal dots probably representing hybridized labeled DNA
fragments can be distinguished on these fibers. For dot
counting, digital images were thresholded and gravity centers
were determined.
In summary, this invention provides procedure termed
combed
DNA fiber comparative hybridization (CFCH). For this new
~5 procedure, there is provided
1) A possibility to measure the relative copy
number representation o~ DNA sequences from
labeled genomic test DNA (e.g. tumor DNA, DNA
from patients with imbalanced types of
chromosome aberrations) as compared to the copy
number representation in differently labeled
normal genomic DNA with a resolution in the
kilobase pair range, i.e. at the level of single
genes. In contrast, CGH as described before can
CA 0223~39~ 1998-0~-12
WO 97tl8326 ~3 PCTIIB96101219
detect copy number changes only in the megabase
pair range.
2) A possibility to measure copy number differences
in mRNAs transcribed from specific genes by
comparative hybridization of a mixture of mRNAs
prepared from test and reference cells.
3) The arrangement of target DNA spots with combed
DNA fibers representing genomic or cDNA test
sequences of interest on a suitable matrix for
lo the simultaneous testing of multiple sequences
for their relative copy number representation in
the hybridization mixture in a geometrical
format, which makes such an arrangement
particularly useful for automated evaluation.
The invention is particularly useful for the detection of
genetic diseases in eucaryotic cells.
CA 0223~39~ l998-0~-l2
WO97/18326 24 PCT~B96101219
REFERENCES
Bensimon, A., Simon, A., Chi~audel, A., Cro~uette, V.,
Heslot, F., Bensimon, D. Science 265:2096-2098 (1994).
du Manoir, S., Speicher, M.R., Joos, S., Schrock, E.,
s Popp, S., Dohner, H., ~ovacs, G., Robert-Nicoud, M., Lichter,
P., Cremer, T. Hum. Genet. 90:590-610, ~1993).
du Manoir, S., Schrock, E., Bentz, M., Speicher, M.R.,
Joos, S., Ried, T., Lichter, P. Cremer, T. Cytometry 19:27-41
(1995).
l0Joos, S., Scherthan, H., Speicher, M.R., Schlegel, J.,
Cremer, T., Lichter, P. Hum. Genet 90:584-589 (1993).
Isola, J., DeVries, S. Chu, L., Ghazvini, S., Waldman, F.
Am. J. Path. 145:1301-1308 (1994)
Kallioniemi, A., Kallioniemi, O.P., Sudar, D., Rutovitz,
1~D., Gray, J.W., Waldman, F., Pinkel, D. Science 258:818-821
(1992).
Kallioniemi, O.P., Kallioniemi, A., Piper, J., Isola, J.,
Waldman, F.M., Gray J.W., Pinkel, D. Genes, Chromosomes and
Cancer 10:231-243 (1994).
~0Kallioniemi, O.P., Kallioniemi, A., Sudar, D., Rutovitz,
D., Gray, J.W., Waldman, F., Pinkel, D. Sem. Cancer Biol. 4:41-
46 (1993).
Kallioniemi, A., Kallioniemi, O.P., Piper, J., Tanner, M.,
Stokke, T., Chen. L., Smith, H.S., Sudar, D., Pinkel, D., Gray,
J.W., Waldman, F. Proc. Natl. Acad. Sci. USA 91:2156-2160
(1994).
Lichter, P., Cremer, T., Borden, J., Manuelidis, L., Ward,
D.C. Hum. Genet. 80:224-234 (1988).
Lichter, P., Cremer, T.: Chromosome analysis by non-
isotopic in situ hybridization. In: Human cytogenetics: A
practical approach; eds.: Rooney, D.E., Czepulkowski, B.H.,
IRL Press, Ox~ord:157-192 (1992).
Muleris, M. Almeida, A., Dutrillaux, A.M., Pruchon, E.,
Vega, F., Delattre, J.Y., Poisson, M. Mal~oy, B., Dutrillaux,
3~B. Genes, Oncogene 9:2717-2722 (1994).
CA 02235395 1998-05-12
WO97~l8326 25 PCT~B96J01219
Pinkel, D., handegent, J.; Collins, C. Fuscoe, J.,
Segraves, R. Lucas, J., Gray, J. Proc. Natl. Acad. Sci. U.S.A.
85:9138-9142 (1988).
Piper, J., Rutovitz, D., Sudar, D., Kallioniemi, A.,
Kallioniemi, O.P., Waldman, F.M., Gray, J.W. Pinkel, D.
Cytometry 19:10-26 (1995).
Ried, T., Petersen, L., Holtgreve-Grez, H., Speicher,
M.R., Schrock, E., du Manoir, S., Cremer, T. Cancer Res.
54:1801-1806 (199~).
I0 Schrock, E., Thiel, G., Lozanova, T., du Manoir, S.,
Meffert, M.C., Jauch, A., Speicher, M.R. Nurn~erg, P. Vogel,
S., Janisch, W., Donis-Keller, H., Ried, T., Witkowski, R.,
Cremer, T. Am J. Path. 144:1203-1218 (1994).
Speicher, M.R., du Manoir, S., Schrock, E., Holtgreve-
I~ Grez, H., Schoell, B., Lengauer, C., Cremer, T., Ried, T. Hum.Mol. Genet. 2:1907-1914 (1993).
Speicher, M.R., Prescher, G., du Manoir, S., Jauch, A.,
Horsthemke, B., Bornfeld, N., Becher, R., Cremer, T. Cancer
Res. 54:3817-3823 (1994).
~0 Speicher, M.R., Jauch, A., Walt, H., du Manoir, S. Ried,
T., Jochum, W., Sulser, T. Cremer, T. Am J. Path. 146: 1332-
1340 (1995).
