Title: Detection of methylated CpG rich sequences diagnostic for malignant cells
Abstract: The present invention provides methods for determining the methylation status of CpG-containing dinucleotides on a genome-wide scale using infrequent cleaving, methylation sensitive restriction endonucleases and two-dimensional gel electrophoretic display of the resulting DNA fragments. Such methods can be used to diagnose cancer, classify tumors and provide prognoses for cancer patients. The present invention also provides isolated polynucleotides and oligonucleotides comprising CpG dinucleotides that are differentially methylated in malignant cells as compared to normal, non-malignant cells. Such polynucleotides and oligonucleotides are useful for diagnosis of cancer. The present invention also provides methods for identifying new DNA clones within a library that contain specific CpG dinucleotides that are differentially methylated in cancer cells as compared to normal cells.
Patent Number: 6,893,820 Issued on 05/17/2005 to Plass
| Inventors:
|
Plass; Christoph (Columbus, OH)
|
| Assignee:
|
The Ohio State University Research Foundation (Columbus, OH)
|
| Appl. No.:
|
775398 |
| Filed:
|
January 31, 2001 |
| Current U.S. Class: |
435/6; 435/91.2 |
| Intern'l Class: |
C12Q 001/68 |
| Field of Search: |
435/6,912
|
References Cited [Referenced By]
U.S. Patent Documents
| 6214556 | Apr., 2001 | Olek et al.
| |
| Foreign Patent Documents |
| WO 9928498 | Jun., 1999 | WO.
| |
Other References
Konishi (Journal of Oral Pathology and Medicine (1999) 28: 102-106).*
Plass et al. Oncogene. May 20, 1999. 18: 3159-3165.*
"Aberrant CpG-island methylation has non-random and tumour-type-specific patterns"
by Costello, et al., Nature Genetics, vol. 25, Feb. 2000, pp. 132-138.
"Aberrant Hypermethylation of the Major Breakpoint Cluster Region in 17p1 1.2
in Medulloblastomas but not Supratentorial PNETs" by Fruhwald, et al., Genes,
Chromosomes & Cancer, 30:38-47 (2001).
"Aberrant methylation of genes in low-grade astrocytomas" by Costello, et al.,
Brain Tumor Pathol., (2000) 17:49-56.
Abstract—"DNA Methylation in Acute Myeloid Leukemia with Evidence of Involvement
of Chromosome 11" by Rush, et al., American College of Veterinary Pathologists,
Annual Meeting, 2000.
Abstract—"Promotor Hypermethylation in Medulloblastomas: Aspects of Tumor
Biology and Potential Clinical Utility" by Fruhwald, et al., Pediatric Oncology,
San Francisco, California, Jun., 2000.
Abstract—"DNA Hypermethylation in Acute Myeloid Leukemia (AML): Nonrandom
Patterns with Preferential Involvement of Chromosome 11" by Rush, et al., The 42nd
ASH Annual Meeting, San Francisco, California, Dec. 2000.
Abstract—"Testicular Germ Cell Tumors as a Model System to Study DNA Methylation"
by Smiraglia, et al., Gordon Research Conference, Cancer Genetics and Epigenetic,
Ventura Beach, California, Feb. 2000.
Abstract—"Global DNA Methylation Changes in Primary Lung Cancer" by Dai,
et al., Gordon Research Conference, Cancer Genetics and Epigenetic, Ventura Beach,
California, Feb. 2000.
Abstract—"Identification of Hypermethylated CpG Islands in Non-Small Cell
Lung Cancer and Related Aberrant Gene Transciption" by Dai, et al., Stone Lab Meeting,
Sep. 2000.
Abstract—"The contribution of DNA methylation to oncogenesis—results
of a genome scanning approach in multiple human tumors" by Plass, Oncogenomics,
Tucson, Arizona, Jan. 2001.
"An Arrayed Human Not I-EcoRV Boundary Library as a Tool for RLGS Spot Analysis"
by Plass, et al., DNA Research, 4, 253-355 (1997).
"Methylation-specific PCR: A novel PCR assay for methylation status of CpG islands"
by Herman, et al., Proc. Natl. Acad. Sci. USA, vol. 93, Sep. 1996, pp. 9821-9826.
"A New Tool for the Rapid Cloning of Amplified and Hypermethylated Human DNA
Sequences from Restriction Landmark Genome Scanning Gels" by Smiraglia, et al.,
Genomics, 58, 254-262 (1999).
"Restriction landmark genome scanning for aberrant methylation in primary refractory
and relapsed acute mycloid leukemia; involvement of the WIT-1 gene" by Plass, et
al., Oncogene, (1999) 18, 3159-3165.
Medulloblastoma: A Developmental Abnormality of the Cerebellum. A comprehensive
analysis of genetic and epigenetic alternations. Michael C. Frühwald. The
Ph.D Thesis Ohio State University 1999 pp. i-180 Columbus Ohio USA.
Graff et al., "Mapping Patterns of CpG Island Methylation in Normal and Neoplastic
Cells Implicates Both Upstream and Downstream Regions in de Novo Methylation",
The Journal of Biological Chemistry (1997) vol. 272, No. 35, pp. 22322-22329.
|
Primary Examiner: Myers; Carla J.
Attorney, Agent or Firm: Calfee, Halter & Griswold
Goverment Interests
This invention was conducted, at least in part, with government support under
National Institutes of Health Grants No: P30 CA16058 and CA80912 awarded by the
National Cancer Institute. The U.S. government has certain rights in the invention
Claims
1. A method of identifying CpG islands which are preferentially methylated in
malignant cells contained within a tumor or neoplasm, comprising:
a) digesting genomic DNA obtained from the malignant cells with an infrequently-cutting,
methylation-sensitive, restriction enzyme to provide a set of malignant cell restriction
fragments;
b) digesting genomic DNA obtained from non-malignant, control cells with an infrequently-cutting,
methylation-sensitive, restriction enzyme to provide a set of control cell restriction
fragments;
c) attaching a detectable label to the ends of the malignant cell restriction
fragments and the control restriction fragments;
d) digesting the labeled malignant cell and control cell restriction fragments
with a second restriction enzyme;
e) separating the labeled malignant cell restriction fragments and the labeled
control cell restriction fragments, wherein the malignant cell restriction fragments
and the control cell restriction fragments are separated by electrophoresis on
two different gels;
f) digesting the restriction fragments in each of said gels with a third restriction
enzyme;
g) electrophoresing the restriction fragments in each of said gels in a direction
perpendicular to the first direction to provide a first pattern of detectable malignant
cell restriction fragments and a second pattern of detectable control cell restriction
fragments; and
h) comparing the first pattern to the second pattern to identify diagnostic control
cell restriction fragments in said second pattern which are absent or exhibit a
decreased intensity in the first pattern, wherein said diagnostic control cell
restriction fragments comprise a CpG island that is unmethylated in the DNA of
the control cells and methylated in the DNA of the malignant cells,
wherein the tumor or neoplasm is selected from the group: colon, glioma, lung,
and non-medulloblastoma primitive neuroectodermal tumors (PNET).
2. The method of claim 1 further comprising the step of determining the sequence
of at least a portion of a diagnostic control cell restriction fragment, wherein
said portion is located at or near an end of the fragment.
3. The method of claim 1 further comprising the step of obtaining a clone from
a DNA library which comprises a diagnostic control cell restriction fragment.
4. The method of claim 1 wherein the tumor or neoplasm: from colon is stage I,
II, II or IV as classified according to the American Joint Committee on Cancer
staging and from PNET is supratentorial PNET.
5. The method of claim 1 wherein the tumor or neoplasm is a primary tumor or neoplasm.
Description
BACKGROUND OF THE INVENTION
Diagnosis of cancer, classification of tumors, and cancer-patient prognosis
all depend on detection of properties inherent to cancer, or malignant cells, that
are absent in normal, nonmalignant cells. Since cancer is largely a genetic disease,
resulting from and associated with changes in the DNA of cells, one important method
of diagnosis is through detection of related changes within the DNA of cancer cells.
Such changes can be of two types. The first type of change is a genetic change
that occurs when the sequence of nucleotide bases within the DNA is changed. Base
changes, deletions and insertions in the DNA are examples of such genetic changes.
The second type of change in the DNA is an epigenetic change. Epigenetic changes
do not result in nucleotide sequence changes, but rather, result in modification
of nucleotide bases. The most common type of epigenetic change is DNA methylation.
In mammalian cells, DNA methylation consists exclusively of addition of a methyl
group to the 5-carbon position of cytosine nucleotide bases. In the process, cytosine
is changed to 5-methylcytosine. Cellular enzymes carry out the methylation events.
Only cytosines located 5′ to guanosines in CpG dinucleotides are methylated
by the enzymes in mammalian cells. Such CpG dinucleotides are not distributed randomly
throughout the genome. Instead, there are regions of mammalian genomes which contain
many CpG dinucleotides, while other areas of the genome contain few CpG dinucleotides.
Such CpG-rich areas of the genome are called "CpG islands." Most often, CpG islands
are located in the transcriptional promoter regions of genes.
Not all CpG islands are methylated However, the methylation status of CpG islands
(i.e., whether the CpG dinucleotides within a particular CpG island are methylated
or not) is relatively constant in cells. Nevertheless, the pattern of CpG island
methylation can change and, when it does, often a new, relatively stable methylation
pattern is established. Such changes in methylation of CpG islands can be either
increases or decreases in methylation.
Methylation of CpG islands in the promoter region of a few specific genes
has been observed in some types of human cancer. However, at present it is still
uncertain whether the methylation status of multiple CpG islands in the genomic
DNA of subjects suspected of having cancer can be used as a diagnostic tool for
determining whether or not tissue obtained from such subjects contain malignant cells.
SUMMARY OF THE INVENTION
The present invention relates to methods for identifying CpG islands which are
diagnostic of one or more cancers in a subject The method employs restriction landmark
genomic scanning (RLGS) techniques and comprises separately digesting genomic DNA
which has been obtained from malignant cells derived from a particular tumor tissue
and genomic DNA which has been obtained from control cells derived from healthy
tissue with an infrequently cutting restriction enzyme that is not capable of cleaving
methylated recognition sites to provide a first set of DNA restriction fragments
from the tumor tissue, referred to hereinafter as "malignant cell restriction fragments",
and a first set of DNA restriction fragments from the healthy tissue, referred
to hereinafter as "control cell restriction fragments"; attaching a detectable
label to the ends of the malignant and control cell restriction fragments; digesting
the labeled malignant and control cell restriction fragments with a second restriction
enzyme; separating each set of restriction fragments on a gel; digesting the restriction
fragments in each of the gels with a third more frequently cutting restriction
enzyme; electrophoresing each set of restriction fragments in a direction perpendicular
to the first direction to provide a first pattern of detectable malignant cell
restriction fragments and a second pattern of detectable control cell restriction
fragments; and comparing the second pattern to the first pattern to identify control
cell restriction fragments, hereinafter referred to as "diagnostic fragments",
which are absent, or exhibit an decreased intensity of label in the first pattern.
Such fragments comprise CpG islands that are methylated in the malignant cells.
Such patterns are useful for characterizing tissue which is suspected of containing
malignant cells. Preferably, each of the diagnostic fragments is then isolated
and sequenced, at least in part. In one preferred embodiment, the first restriction
enzyme is NotI. In another preferred embodiment, the first restriction enzyme is
AscI. Advantageously, the present method permits the detection of numerous methylation
sites within the entire genome. In accordance with the present method, applicants
have determined that particular CpG islands are preferentially methylated in DNA
obtained from tumor tissues of subjects diagnosed as having breast cancer, glioma,
acute myeloid leukemia, primitive neuroectodermal tumors of childhood, colon cancer,
head and neck cancer, testicular cancer, and lung cancer.
The present invention also provides isolated polynucleotides, referred to hereinafter
as "CpG diagnostic polynucleotides", and isolated oligonucleotides referred to
hereinafter as "CpG diagnostic oligonucleotides", which are useful for characterizing
tissue samples obtained from a subject suspected of having gliomas, acute mycloid
leukemia, primitive neuroectodermal tumors of childhood, or cancer of the breast,
colon, head and neck, testicle or lung. The CpG diagnostic polynucleotides and
oligonucleotides both comprise a sequence which contains CpG islands that have
been shown to be preferentially methylated in DNA that has been obtained from malignant
cells of subjects diagnosed as having breast cancer, glioma, acute mycloid leukemia,
primitive neuroectodermal tumor of childhood, colon cancer, head and neck cancer,
testicular cancer or lung cancer. The CpG diagnostic polynucleotides are from 35
to 3000, preferably, 35 to 100 nucleotides in length, and comprise from 15 to 34,
preferably 18 to 25 of the consecutive nucleotides contained with the sequences
depicted in the accompanying DNA sequence listing, or sequences which are complementary
thereto. The CpG diagnostic polynucleotides comprise two or, preferably, more CpG
dinucleotides or dinucleotides which are complementary thereto. The CpG diagnostic
oligonucleotides are from 15 to 34 nucleotides in length and comprise from 15 to
34 consecutive nucleotides contained within the sequences depicted in the sequence
listing, or sequences which are complementary thereto. The CpG oligonucleotides
comprises two or more CpG dinucleotides, or dinucleotides which are complementary thereto.
The present invention also relates to methods which employ the CpG diagnostic
polynucleotides and oligonucleotides of the present invention to characterize tissue
from patients suspected of having cancer. Such methods are based on the methylation
status of CpG islands that have been shown to be preferentially methylated in DNA
that has been obtained from tumor tissues of subjects diagnosed as having breast
cancer, glioma, acute myeloid leukemia, primitive neuroectodermal tumor of childhood,
colon cancer, head and neck cancer, testicular cancer and lung cancer. In one method,
DNA which is isolated from suspected tumor tissue from a subject is digested into
smaller fragments and reacted with a CpG diagnostic polynucleotides under stringent
hybridization conditions. The reaction products are then assayed to determine the
size or the sequence of the DNA fragment with which the CpG diagnostic polynucleotide
has hybridized. The size or the sequence of the DNA fragment to which the CpG diagnostic
polynucleotide has hybridized, hereinafter referred to as the "target DNA fragment",
indicates whether the target DNA fragment comprises methylated or nonmethylated
CpG islands. The presence of methylated CpG islands in the target DNA fragment
indicates that the DNA has been obtained from a tumor or neoplasm for which the
diagnostic CpG polynucleotide serves as a diagnostic marker.
In another method the DNA from the suspected tumor tissue is treated with a chemical
compound which converts nonmethylated cytosines to a different nucleotide base.
An example of such a compound is sodium bisulfite which converts non-methylated
cytosines to uracil. The DNA is then reacted with at CpG diagnostic oligonucleotides
under conditions which permit the CpG diagnostic oligonucleotide to hybridize with
a complementary sequence in the DNA, referred to hereinafter as the "target sequence".
The DNA is also reacted with a modified CpG diagnostic oligonucleotide. The modified
CpG diagnostic oligonucleotide comprises a sequence that is complementary to a
modified target sequence, i.e., a sequence in which the non-methylated cytosines
in the target sequence are converted to a different nucleotide base, e.g. uracil,
when treated with a chemical compound. The reaction products are then assayed to
determine whether the DNA contains sequences which have hybridized with the CpG
diagnostic oligonucleotide or with the modified CpG diagnostic oligonucleotide.
Hybridization of the sample DNA with the CpG diagnostic oligonucleotide, as opposed
to the modified CpG diagnostic oligonucleotide, indicates that the cytosines in
the target sequence are methylated and that the DNA sample has been obtained from
a tumor or neoplasm for which the CpG oligonucleotide has been shown to serve as
a diagnostic marker.
The present invention also relates to a method of identifying genes whose expression
is increased or decreased in cancer cells.
BRIEF DESCRIPTION OF THE FIGURES
FIG. 1. Methylation detection in restriction landmark genomic scanning
(RLGS) profiles. A, Diagram of the RLGS procedure showing the quantitative nature
of methylation detection on Not fragments displayed on RLGS profiles. Methylation
detection in RLGS profiles depends on the methylation sensitivity of the endonuclease
activity of NotI. Differences in digestion are assessed by radiolabelling the DNA
at cleaved NotI sites. Following further endonuclease digestion, two-dimensional
electrophoretic separation and autoradiography, the intensity of a DNA fragment
on the resultant RLGS profile quantitatively reflects the copy number and methylation
status of the Nod fragment. A priori, this allows NotI fragments containing single-copy
CpG islands to be distinguished from the abundant NotI fragments present in repeat
elements and rDNA sequences. B, A portion of an RLGS profile from normal peripheral
blood lymphocyte DNA displaying nearly 2,000 single-copy NotI fragments and 15-20
high copy-number fragments. First dimension separation of labeled NotI/EcoRV fragments
extends from right to left horizontally. Following in-gel digestion with HinfI,
the fragments are separated vertically downward into a polyacrylamide gel and autoradiographed.
To allow uniform comparisons of RLGS profiles from different samples and different
laboratories, each fragment is given a three-variable designation (Y coordinate,
X coordinate, fragment number). The central region of the RLGS profile used for
all comparisons described in this invention has 28 sections (1-5 vertically and
B-G horizontally, the 4G and 5G sections were excluded due to high density and
lower resolution of fragments). C, Enlarged view of profile section 2D, showing
the numbers assigned to each NotI fragment D, Analysis of the GC content and CpG
ratio {(number of CpGs)/(number of guanines)(number of cytosines)}(number of nucleotides
analyzed) of 210 non-redundant NotI/EcoRV clones containing the Not/HinfI fragments
seen in B and in other portions of the RLGS profile. Of 210 clones, 184 clones
were randomly chosen and 26 corresponded to fragments which were frequently lost
from tumor profiles. CpG islands have a GC content of greater than 50% and a CpG
value of 0.6 or greater, relative to bulk DNA (average CG content of 40% and CpG
ratio of 0.2). Nucleotide sequences were determined with greater than 99% accuracy
overall. An average of 377 nt/clone were analyzed (not indicative of actual CpG
island size). The average NotI/EcoRV clone size was approximately 2 kb.
FIG. 2. Fragment loss from RLGS profiles is due to methylation. Top,
portions of the RLGS profiles obtained from normal tissue and from two tumors having
NotI fragments with either decreased intensity or no change in intensity. Bottom,
Southern-blot analysis of EcoRV (NotI: -) and EvoRV/NotI (NotI: +) restriction
digested DNAs from a larger number of samples, including the samples at top. In
samples without methylation in the NotI site, the probe detects a smaller fragment
on double digestion with NotI and EcoRV. The quantitation from multiple Southern
blots using a phosphorimager allowed the determination of a lower limit of reliable
detection in RLGS profiles of 30% decreased intensity of the diploid NotI/EcoRV
fragments. Presence (+) or absence (-) of the corresponding NotI fragment is indicated.
N, normal tissue DNA; T, tumor tissue DNA A, CpG-island locus 3C1 methylation in
low-grade gliomas. B, CpG island locus 2C40 methylation in leukemias. C, CpG-island
locus 3E24 methylation in PNETs of childhood. *, EcoRV fragment of approximately
13 kb with homology to the probe. BLAST searches using the NotI-EcoRV clone sequence
identified a homologous BAC clone sequence lacking an internal NotI site, which
accounts for the 13-kb fragment on the Southern blot.
FIG. 3. Heterogeneity in CpG-island methylation across tumors. RAGS profiles
w generated from 98 primary human tumors and compared with profiles of either matched
normal DNA (58 of 98 cases) or to multiple profiles of tissue typematched normal
DNA from unrelated individuals. Loss or decreased intensity of single-copy fragments
in the tumors, relative to several neighboring unaltered NotI fragments, were detected
by visual inspection of overlaid autoradiographs and confirmed in many cases by
independent profiles of the same DNA samples. For each tumor type, the dot plots
display the total number of methylated CpG islands (of 1,184 CpG islands analyzed)
observed in each tumor. Under the assumption that the tumors are drawn from a homogeneous
distribution, with all tumors having the same frequency of methylation, the loss
distributions should be approximately Poisson. The colored curve represents the
expected distribution. BRE, breast tumors; CLN, colon tumors; GLI, gliomas; HN,
bead and neck tumors; LEU, acute mycloid leukemias; PNET, primitive neuroectodermal
tumors of childhood; TST, testicular tumors.
FIG. 4. Subsets of CpG islands are preferentially methylated. For each
tumor type, the histograms display the number of tumors in which the particular
CpG islands were methylated. Most of the 1,184 CpG islands were not methylated
in any of the tumors (histogram bar at 0 is not shown), but several CpG islands
were methylated in multiple tumors. The black line shows the expected distribution
under the null hypothesis that the CpG islands have equal frequencies of methylation.
Most of the tumor types show significant preferential methylation.
DETAILED DESCRIPTION OF THE INVENTION
In one aspect, the present invention relates to methods for identifying clones
within a DNA library that can be used for cancer diagnosis and tumor classification,
based on the methylation status of CpG dinucleotides contained within or closely
adjacent to the specific clones. Such method employs methylation-sensitive restriction
endonucleases (MSREs) and restriction landmark genomic scanning (RLGS) gels to
identify new, differentially-methylated CpG islands within malignant cells obtained
from patients diagnosed as having cancer. In accordance with the present invention,
Applicants have identified 93 clones which can be used to determine whether a tumor
biopsy from a patient contains benign or malignant cells.
To carry out such method, tissue (referred to hereinafter as "tumor tissue")
which
contains a tumor or neoplasm is obtained from a patient known to have a cancer.
In some cases, the tumor tissue is obtained from a particular type of solid tumor
which has bee surgically removed from the patient. In some cases, the tumor tissue
is obtained from the hematopoietic system, such as for example, bone marrow or
blood, of the patient The tumor tissue will have been determined to be from either
a benign or malignant tumor or neoplasm.
Separately, tissue (referred to hereinafter as "healthy tissue") which
does not contain a tumor or neoplasm is obtained from a subject. The healthy tissue,
may be obtained by surgically removing normal tissue from the patient or by surgically
removing normal tissue from a healthy control subject who does not have cancer.
The healthy tissue may also come from the hematopoietic system, such as for example,
bone marrow or blood, of a healthy control subject. The healthy tissue will have
been determined to be nor-tumorigenic or non-neoplastic.
DNA is then isolated from both the tumor tissue and healthy tissue. If the tumor
tissue is a solid tissue sample, such procedure may first comprise separating the
individual cells contained within the tissue from each other. For example, if the
tissue samples were frozen after surgical removal from a patient, cells may be
separated from one another by grinding the frozen tissue with a mortar and pestle.
DNA is then isolated from the individual cells using procedures well known to those
skilled in the art. Commonly, such DNA isolation procedures comprise lysis of the
individual cells using detergents, for example. After cell lysis, proteins are
commonly removed from the DNA using various proteases. The DNA is then commonly
extracted with phenol, precipitated in alcohol and dissolved in an aqueous solution.
In the procedures which follow, the DNA obtained from the tumor tissue is treated
separately from the DNA obtained from healthy tissue (i.e., the two DNAs are not
mixed). The DNAs are separately analyzed using a method called restriction landmark
genomic scanning (RLGS). The purpose is to analyze both DNAs separately. The two
analyses are then compared in order to identify CpG islands that distinguish cancer
cells from normal cells.
Both DNA samples are treated with restriction enzymes and the free ends that
result from the restriction enzyme cleavage are labeled. However, since the isolated
DNA is in linear pieces, there are free ends that exist before the DNA is cleaved
with the restriction enzymes. To prevent these ends from being labeled, the ends,
preferably, are blocked before restriction enzyme treatment. Such blocking can
be done by addition of dideoxynucleotides and sulfur-substituted nucleotides to
the free ends before treatment with restriction enzymes. Subsequently, when the
DNA is cleaved by restriction enzymes and labeled, only the ends resulting from
the restriction enzyme cleavage will be labeled.
After the reaction to block free ends, the DNA samples are cleaved with a first
restriction enzyme that can be characterized as an infrequently cleaving, methylation-sensitive
restriction enzyme. Examples of suitable first restriction enzymes are NotI, AscI,
BssHII and EagI. As used herein the term "infrequently cleaving" refers to a restriction
enzyme that is expected to cleave genomic DNA at intervals greater than 10 kilobases.
For example, NotI is an infrequently cleaving restriction enzyme. NotI recognizes
a nucleotide sequence of 8 base pairs (bp) in the genome (i.e., 5′GCGGCCGC3′)
and cleaves the DNA at this site. There are an estimated 4000-5000 of such NotI
recognition sequences within the human genome. It is estimated that such recognition
sequences are spaced at approximately 1 megabase (Mb) intervals within the genome.
In contrast, a frequently cleaving restriction enzyme is expected to cleave the
human genome at from 5-10 kb intervals. Such an enzyme will have approximately
100-times more cleavage sites within the human genome than infrequently-cleaving
enzymes. Such frequently cleaving enzymes usually recognize a nucleotide sequence
of less than 8 bp in the genome and cleave the DNA at that site. However, not all
restriction enzymes that have nucleotide recognition sequences of less than 8 bp
are frequently cleaving enzymes. BssHII and EagI both have 6 bp recognition sequences
but the recognition sequences for these two enzymes are spaced at intervals within
the genome that are greater than 10 kb. "Methylation sensitive" as used herein
refers to any enzyme that is unable to cleave DNA at its normal restriction site
if one or more nucleotides within the recognition sequence is methylated. For example,
the restriction enzyme NotI will cleave the 5′GCGGCCGC3′ recognition
sequence if the sequence does not contain a 5 methylcytosine. However, the NotI
enzyme will not cleave this sequence if any of the cytosines have been methylated
to become 5-methylcytosine.
Following digestion of the DNA with the first restriction enzyme, the ends
of the DNA fragments are labeled. This can be done, for example, by attachment
of nucleotides carrying a detectable label, such as a radiolabel, to the ends of
the DNA sample. Typically, attachment is accomplished by filling in the recessed
DNA ends left by cleavage with the first restriction enzyme such that the ends
become blunt (i.e., non-recessed). Such end-filling reaction may employ deoxynucleoside
triphosphates having a radiolabeled phosphate at the α phosphate position.
Such labeled phosphate is preferably
32P.
The labeled fragments from each sample are then cleaved with a second restriction
enzyme. Such second restriction enzyme preferably cleaves human DNA at average
intervals of between 5-10 kb. Such enzymes normally have a 6 bp recognition sequence.
Preferably, the second restriction enzyme is not methylation sensitive. Examples
of suitable second restriction enzymes are PstI, PvuI, EcoRV or BamHI. Cleavage
of the DNA fragments with the second restriction enzyme provides a second set of
fragments, labeled at the ends-left by cleavage with the first enzyme. Many of
such second fragments are smaller than the fragments resulting from cleavage with
the first restriction enzyme.
The DNA fragments are then separated from one another. Preferably this separation
is based on size. Preferably this separation is performed by first-dimension electrophoresis
through an agarose tube-shaped gel of approximately 60 cm in length.
After electrophoresis through the tube-shaped gel, the DNA is digested within
the gel with a third restriction enzyme. Such third restriction enzymes preferably
have recognition sequences of 4 or 6 bp. Such third restriction enzymes also have
the property of being able to cleave DNA which is embedded within agarose. One
such enzyme is HinfI.
After cleavage by the third restriction enzyme, the DNA is again separated
based on size, preferably by electrophoresis through a polyacrylamide gel. Subsequently,
the separated DNA fragments are detected based on the labeled ends of the DNA fragments.
In those cases where the fragments are radiolabeld, detection is by autoradiography
of the two-dimensional gel. Such autoradiography provides a pattern of DNA fragments
or "spots." Such pattern is called an RLGS profile.
Each fragment on the RLGS profile obtained from using the DNA from healthy tissues
is uniquely identified by its location on the autoradiograph (Y coordinate, X coordinate,
fragment number). For each fragment location on the RLGS profile obtained from
healthy tissue DNA, the identical location is observed on the RLGS profile obtained
from tumor tissue DNA.
In a fragment by fragment comparison of RLGS profiles obtained from tumor tissue
DNA with healthy tissue DNA, three different patterns are possible. First, for
a given fragment on the healthy tissue RLGS profile, there may be a corresponding
fragment at the same location, and of the same intensity, on the tumor tissue RLGS
profile. This indicates that the first restriction enzyme cleaved both DNAs at
the same sequences (FIG.
1A). This indicates that there were no differences
in methylation of the NotI nucleotide recognition sequence of that fragment between
the tumor tissue DNA and the healthy tissue DNA.
Second, for a given fragment on the healthy tissue RLGS profile, there may
be no fragment at the same location on the tumor tissue RLGS profile. Such a pattern
indicates that the first restriction enzyme did not cleave the tumor tissue DNA
at the recognition sequence required to produce that specific fragment, but did
cleave at such sequence within the healthy tissue DNA (FIG.
1A). This indicates
that there was methylation within the NotI recognition sequence in the tumor tissue
DNA but not in the healthy tissue DNA.
Third, for a given fragment on the healthy tissue RLGS profile, there may
be a corresponding fragment at the same location on the tumor tissue RLGS profile,
but the intensity of the fragment may be of decreased intensity. Such a pattern
indicates that the first restriction enzyme cleaved one of two copies (i.e., the
genome is diploid) of the tumor tissue DNA at the recognition sequence required
to produce that specific fragment (FIG.
1A). In healthy tissue DNA, the
first restriction enzyme cleaved both copies of the recognition sequence. This
indicates that there was methylation within one of two NotI recognition sequences
in the tumor tissue DNA.
Through comparisons of RLGS profiles obtained from healthy tissue DNA with
profiles obtained from a large number of different tumor tissue DNAs, loss of specific
fragments in multiple tumors can be associated with a specific type of cancer.
Loss of such fragments from RLGS profiles, therefore, can be diagnostic for cancer
in a subject. For example, loss of a specific fragment (i.e., methylation of the
first restriction enzyme site at the end of said fragment) in a high percentage
of tumor tissue DNAs from women known to have breast cancer can be diagnostic for
breast cancer in subjects suspected of having the disease. To perform such a diagnostic
analysis, DNA isolated from a patient suspected of having breast cancer would be
analyzed by RLGS, as described above, to determine whether there was loss of one
or more fragments in RLGS profiles that are known to be lost at high frequency
in women known to have breast cancer. Similarly, loss of other specific fragments
can be diagnostic for other cancers, such as for example, colon cancer, head and
neck cancer, lung cancer, testicular cancer, neuroectodermal cancer, gliomas, acute
myeloid leukemias, and others.
Loss of a specific fragment in RLGS profiles from multiple tumors can also be
diagnostic of several types of cancer, rather than a single type of cancer. For
example, loss of a specific fragment can occur in a high percentage of tumor tissue
DNAs obtained from individuals with either breast, colon or lung cancer. Loss of
such a spot from RLGS profiles using DNA obtained from a patient suspected of having
cancer would be diagnostic for either breast, colon or lung cancer in that patient.
Isolated Polynucleotides and Oligonucleotides Diagnostic for Cancer
Individual DNA clones that contain the DNA present in each spot or fragment
that makes up an RLGS profile can be obtained. This is done by constructing a DNA
library of healthy tissue DNA that has been cleaved with the same first and second
enzymes used to perform the RLGS gel analysis. Such DNA library will contain individual
clones, each clone comprising DNA that is present in a single spot of the RLGS
profile. The totality of clones within the library is representative of the combined
DNA spots in the RLGS profile.
Individual clones within the library can be identified that contain the
DNA of each spot on the RLGS profile. This can be done by taking DNA from one or
a few individual clones of the DNA library and mixing it with healthy tissue DNA,
before RLGS analysis is begun. When this mixture of DNAs is used to produce an
RLGS profile, the intensity of the spots that contain the same DNA as the individual
clones added to the mixture will be increased. By performing multiple analyses
of this type, each spot on an RLGS profile can be matched up with a DNA clone within
the library. The result of such an analysis is an ordered human genomic library
of restriction fragments containing the same subset of genomic fragments as those
displayed on RLGS profiles. In such ordered genomic libraries, an individual library
clone corresponding to any spot or fragment in an RLGS profile can be rapidly located.
To design diagnostic CpG polynucleotides and oligonucleotides, tie sequence of
the DNA within each clone (referred to hereinafter as a "diagnostic clone") that
corresponds to a spot that is absent or exhibits decreased intensity on the RLGS
profile of the DNA from malignant tumor tissue is sequenced using standard techniques.
Once sequence information is obtained, regions comprising multiple CpG dinucleotides
are located. Such regions serve as the target sequence for the CpG polynucleotides
and oligonucleotides.
The CpG polynucleotides are from 35 to 3000, preferably from 35 to 1500 nucleotides
in length and comprise two or, preferably, more CpG dinucleotides or dinucleotides
which are complementary thereto. The CpG diagnostic oligonucleotides are from 15
to 34 nucleotides, preferably from 18 to 25 nucleotides, in length and comprise
at least two CpG dinucleotides or dinucleotides which are complementary thereto.
The CpG diagnostic polynucleotides and oligonucleotides each comprise a sequence
which is substantially complementary to target sequences containing CpG islands
that are known to be preferentially methylated in the DNA from one or more types
of cancer cells. "Substantially complementary" means that there is enough complementarity
between the CpG diagnostic polynucleotides or oligonucleotides and the target sequence
so that hybridization occurs between the CpG diagnostic polynucleotides and oligonucleotides
under stringent conditions, preferably under highly stringent conditions. Such
assays include hybridization assays, such as for example Southern analysis, where
the sample DNA is reacted with the CpG diagnostic polynucleotide under stringent
hybridization conditions.
The term "stringent conditions, as used herein, is the "stringency" which occurs
within a range from about Tm-5 (5 below the melting temperature of the probe) to
about 20 C below Tm. "Highly Stringent hybridization conditions" refers to an overnight
incubation at 42 degree C. in a solution comprising 50% formamide, 5×SSC (750
mM NaCl, 75 mM sodium citrate), 50 mM sodium phosphate (pH 7.6), 5×Denhardt's
solution, 10% dextran sulfate, and 20 g/ml denatured, sheared salmon sperm DNA,
followed by washing the filters in 0.2×SSC at about 65 degree C. As recognized
in the art, stringency conditions can be attained by varying a number of factors
such as the length and nature, i.e., DNA or RNA, of the probe; the length and nature
of the target sequence, the concentration of the salts and other components, such
as formamide, dextran sulfate, and polyethylene glycol, of the hybridization solution.
All of these factors may be varied to generate conditions of stringency which are
equivalent to the conditions listed above.
Changes in the stringency of hybridization and signal detection are primarily
accomplished through the manipulation of formamide concentration (lower percentages
of formamide result in lower stringency); salt conditions, or temperature. For
example, moderately high stringency conditions include an overnight incubation
at 37 degree C. in a solution comprising 6×SSPE (20×SSPE=3M NaCl; 0.2
M NaH
2PO
4; 0.02M EDTA, pH 7.4), 0.5% SDS, 30% formamide,
100 ug/ml salmon sperm blocking DNA; followed by washes at 50 degree C. with 1×SSPE,
0.1% SDS. In addition, to achieve even lower stringency, washes performed following
stringent hybridization can be done at higher salt concentrations (e.g. 5×SSC).
Note that variations in the above conditions may be accomplished through the
inclusion and/or substitution of alternate blocking reagents used to suppress background
in hybridization experiments. Typical blocking reagents include Denhardt's reagent,
BLOTTO, heparin, denatured salmon sperm DNA, and commercially available proprietary
formulations. The inclusion of specific blocking reagents may require modification
of the hybridization conditions described above, due to problems with compatibility.
Such assays also include polymerase chain reactions (PCR) where the sample DNA
and the diagnostic CpG oligonucleotides are reacted, preferably under conditions
which result in the synthesis of a single PCR product. Computer programs, such
as for example, the "Primer3" program that can be accessed via the internet at
<URL: genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi> can be used to
determine the size and sequence of the CpG diagnostic oligonucleotides. Optimum
conditions are determined empirically.
The CpG diagnostic polynucleotides and oligonucleotides are made using standard
techniques. For example, these polynucleotides and oligonucleotides may be made
using commercially available synthesizers.
Diagnostic Methods
In another aspect, the present invention relates to methods which use the CpG
diagnostic polynucleotides and oligonucleotides to characterize tissue samples
from a subject suspected of having cancer, referred to hereinafter as test sample
DNA. To do this, DNA is isolated from the cells of the tissue sample of the patient.
Preferably, DNA that serves as a control is also obtained from healthy tissue of
the test subject or a control subject as described previously. The diagnostic methods
comprise reacting the test sample DNA with the diagnostic CpG polynucleotide or
oligonucleotide and assaying the products that are formed as the result of the
reaction. In some cases, the sample DNA is digested into smaller fragments prior
to reaction with the CpG diagnostic polynucleotides or oligonucleotides. In some
cases, a portion of the test sample DNA is first reacted with a chemical compound,
such as for example sodium bisulfite, which converts methylated cytosines to a
different nucleotide base.
Southern Blot Analysis
One such method for diagnosing cancer in a patient involves cleavage of the test
sample DNA with a methylation sensitive enzyme, then Southern blot analysis of
said cleaved DNA using a CpG diagnostic polyncleotide or oligonucleotide as a probe.
For example, the DNA from the patient and the control, healthy tissue DNA are separately
cleaved with a methylation-sensitive restriction endonuclease, such nuclease being
the same first restriction enzyme used to identify the diagnostic spot in the RLGS
profile that corresponds to the CpG diagnostic polynucleotide or oligonucleotide.
After cleavage, the test sample and control DNAs are electrophoretically separated
by size in different lanes of the same agarose gel and blotted to a membrane that
can be used in hybridization, such as for example, nitrocellulose or nylon. The
membrane is then used in a hybridization reaction with a labeled CpG diagnostic
polynucleotide or oligonucleotide. The labeled CpG diagnostic polynucleotide or
oligoneucleotide will hybridize to complementary DNA sequences on the membrane.
After hybridization, the location on the membrane where the probe hybridized to
the control and patient DNAs is visualized. Such locations will identify DNA fragments
or bands within the control and patient DNAs containing the same sequence as the
CpG diagnostic polynucleotide or oligonucleotide. Hybridization of the probe to
a fragment within the patient DNA that is of higher molecular weight than that
of the fragment within the control DNA to which the probe hybridized, indicates
that a restriction endonuclease cleavage site flanking the target sequence of the
CpG diagnostic polynucleotide or oligonucleotide was not cleaved due to methylation
Such result indicates that the tissue is from a cancer for which the CpG diagnostic
polynucleotide or oligonucleotide serves as a diagnostic tool.
A second method for diagnosing cancer in a patient involves cleavage of patient
DNA with a methylation-sensitive restriction endonuclease, such nuclease being
the same first restriction enzyme used to identify the diagnostic spot in the RLGS
profile that corresponds to the fragment Such nuclease will cleave the patient
DNA at the diagnostic recognition sequence only if the DNA is unmethylated. Using
nucleotide information derived from sequencing of the library clone corresponding
to the diagnostic spot on the RLGS gel, primers for PCR are selected that span
the diagnostic recognition sequence. Using the primers, PCR is performed on the
DNA. PCR amplification of the sequences will be successful only if the diagnostic
nucleotide sequence in the patient DNA had been methylated and was not cleaved
by the enzyme. Successful PCR amplification, therefore, is indicative of cancer
in the patient.
Methods Employing a Chemically-Modified DNA Test Sample
Another group of methods for diagnosing cancer in a patient using CpG diagnostic
polynucleotides and oligonucleotides are based on treatment of patient DNA with
sodium bisulfite which converts all cytosines, but not methylated cytosines, to
uracil. The bisulfite converted patient DNA can then be analyzed in a number of
different ways. One method of analysis is direct sequencing of the DNA to determine
whether the sequence contains cytosine or uracil. Such DNA sequencing requires
primers adjacent to the sequenced region to be made. Such primers would be based
on DNA sequence information obtained from the diagnostic RLGS spots.
Another method of analyzing bisulfite converted patient DNA is a method called
"methylation sensitive PCR" (MSR). In MSR, primers are designed to comprise a sequence
which is substantially complementary to the the CpG islands which are known to
be preferentially methylated in DNA of cells found in one or more type of tumor
tissues. Two sets of PCR primers are made to encompass this region. One set of
primers is designed to be complementary to the sequence that was changed by bisulfite
(i.e., cytosines that were originally unmethylated and changed to uracil). As discussed
above, these are the modified CpG diagnostic oligonucleotides. A second set of
primers is designed to be complementary to the same sequence that was not changed
by bisulfite (i.e., cytosines that were methylated and not changed to uracil).
As discussed above these are the unmodified CpG diagnostic oligonucleotides, i.e
the oligonucleotides which containe at least two CpG dinucleotides or dinucleotides
which are complementary thereto. Two sets of PCR reactions are then run, one reaction
with each set of primers, using DNA from the subject as the template. In the case
where cytosines within the target sequence of the subject DNA are not methylated,
the target sequence will be modified by the chemical reaction and the primers complementary
to the modified sequence, i.e., the modified CpG diagnostic oligonucleotides, will
produce a PCR reaction product while the primers complementary to the methylated
sequence, i.e., the unmodified CpG diagnostic oligonucleotides, will not produce
a PCR product. In the case where cytosines within the target sequence of the subject
DNA are methylated, the target sequence will not be altered by the reaction with
the sodium bisulfite, and the primers complementary to the unaltered sequence,
i.e., the unmodified CpG diagnostic oligonucleotides, will produce a PCR reaction
product while the modified CpG diagnostic oligonucleotides, which are complementary
to the modified target sequence (i.e., unmethylated sequence) will not produce
a PCR product.
A modification of MSR is bisulfite treatment of patient DNA and PCR amplification
of said DNA using primers designed to amplify either methylated or unmethylated
sequences. The PCR product is then digested with a restriction enzyme that will
cleave or not depending on whether said product contains uracil (rather, thymidine,
the complement of uracil; found in PCR product if original patient DNA contained
unmethylated cytosine) or cytosine (found in PCR product if original patient DNA
contained methylated cytosine).
Another technique referred to as MS-SnuPE, uses bisulfite/PCR followed by
primer extension, where incorporation of C (vs. T) denotes methylation.
Methods of Identifying Genes
In another aspect of the invention, the CpG diagnostic polynucleotides and oligonucleotides
can be used as probes to to identify genes whose expression is increased or decreased
in cancerous tissues. To do this, CpG diagnostic polynuceotides are reacted with
individual clones of the DNA library. The clones which hybridize with the CpG diagnostic
polynucleotide can then be analyzed to determine if they contain an open reading
fires that could encode proteins. To determine if the CpG diagnostic polynucleotide
hybridizes with the promoter region of a known gene, the open reading frame sequence
is analyzed by searching existing DNA databases. For example, GenBank databases
can be searched using the BLAST algorithm. If no known genes that correspond to
a library clone is found, the sequence can be searched for open reading frames
that could encode a protein. Such searching can be performed using commercially
available sequence analysis programs commonly known to those skilled in the art.
GCG is an example of one such program.
Sequences from clones of the DNA library that contain either known genes
or open reading frames can be used as probes to determine whether genes encoded
by the sequences are expressed in tumor tissues as compared to control, healthy
tissues. To do this, RNA, preferably messenger RNA (mRNA) is isolated from healthy
tissue and from tumor tissue from which it is desired to test expression. Such
RNA is examined for the presence of expressed transcripts encoded by the sequences
obtained from the library. Examination for the presence of expressed transcripts
can be performed using a number of methods. One method is Northern blotting where
the isolated RNA is separated by size using gel electrophoresis and then blotted
to a hybridization membrane. A fragment, polynucleotide or oligonucleotide from
the sequence obtained from a library clone is labeled and then used to probe the
hybridization membrane containing the size-separated RNA. Detection of hybridization
of the probe to the membrane indicates presence of a transcript encoded by the
sequence and indicates expression of the gene encoded by that sequence.
Another method to examine isolated RNA for the presence of expressed transcripts
is to use RT-PCR analysis. In such analysis, primers are designed and made that
span a region of the gene whose expression is to be tested. The isolated RNA is
reverse transcribed into DNA using reverse transcriptase. Such DNA is then amplified
with the designed primers using PCR. PCR products are visualized after electrophoresis.
The presence of PCR products on the gel indicates that the gene encompassed by
the designed primers was expressing RNA transcripts. Such analysis can identify
and determine genes whose expression is changed in cancer cells as compared to
normal, non-cancerous cells.
The following examples are for purposes of illustration only and are not intended
to limit the scope of the invention as defined in the claims which are appended hereto.
EXAMPLES
Example 1
Identification of Diagnostic Markers Using NotI and RLGS
A. Isolation and Enzymatic Processing of Genomic DNA
Tissue from solid tumors was obtained as surgical tissue samples. Where possible,
surrounding non-tumor tissue was taken and used as a control. Where it was not
possible to obtain patient-matched normal tissue, normal tissue from multiple patients
was used. Tissue samples from patients with acute myelogenous leukemis (AML) consisted
of either bone marrow aspirates or peripheral blood. Normal samples were obtained
from the same patients who were in remission after chemotherapy.
The surgically removed tissues were quickly frozen in liquid nitrogen and stored
a -80° C. prior to isolation of DNA. When DNA was ready to be isolated, 2
ml of lysis buffer (10 mM Tris, pH 8.0; 150 mM EDTA, 1% sarkosyl) was added to
100-300 mg of tissue in a 50 ml Falcon tube and frozen in liquid nitrogen. The
frozen mixture was then removed from the tube, wrapped in aluminum foil, and quickly
broken into pieces with a hammer. The broken pieces of cells were transferred to
a chilled mortar and ground to a powder with a chilled pestle. For peripheral blood
samples, cells were separated on a sterile Histopaque-1077 (SIGMA) gradient and
stored id at -80° C. before DNA isolation Cells were transferred to a 50 ml
tube and 15-25 ml of lysis buffer containing 0.1 mg proteinase K per ml of lysis
buffer was added and mixed using a glass rod. The mixture was incubated at 55°
C. for 20 min with gentle mixing every 5 min. The mixture was then placed on ice
for 10 min. Subsequently, an equal volume of PCI (phenol:chloroform:isoamylalcohol
in a ratio of 50:49:1) was added and the tubes containing the mixture were gently
rotated for 30-60 min. The tubes were then centrifuged for 30 min at 2500 rpm and
the separated, aqueous phase was transferred to a new 50 ml tube using a wide-bore
pipette. The PCI extraction was repeated one time. The collected aqueous phase
containing the DNA was transferred to dialysis tubing and dialyzed against 4 L
of 10 mM Tris, pH 8 for 2 hr. The dialysis tubing was then transferred into fresh
10 mM Tris and dialyzed overnight at room temperature. One additional dialysis
was performed in fresh 10 mM Tris for an additional 2 hr. The DNA was then transferred
from the dialysis tubing to 50 ml tubes and RNase A was added to a final concentration
of 1 μg/ml. The mixture was incubated at 37° C. for 2 hr. Subsequently,
2.5 volumes of 100% ethanol were added to the DNA and the mixture was gently rotated.
The insoluble DNA was transferred to a microfuge tube, centrifuged briefly, and
the remaining alcohol removed. The pellet was briefly dried in air. The DNA in
the pellet was resuspended to a final concentration of 1 μg/μl. Such
isolated DNA had an average size of 200-300 kb.
The isolated genomic DNA was blocked at ends where the DNA had been sheared.
Blocking was done by addition of dideoxynucleotides and sulfur-substituted nucleotides.
In a 1.5 ml tube, 7 μl of genomic DNA solution was added along with 2.5 μl
of blocking buffer (1 μl 10×buffer 1, 0.1 μl 1 M DTT, 0.4 μl
each of 10 μM dGTPαS, 10 μM ddATP, 10 μM ddTTP, and 0.2
μl 10 μM dCTPαS; buffer 1 consists of 500 mM Tris, pH 7.4, 100
mM MgCl
2, 1 M NaCl, 10 mM DTT) and 0.5 μl DNA polymerase I. The
mixture was mixed thoroughly and incubated at 37° C. for 20 min. The mixture
was then incubated at 65° C. for 30 min to inactivate the polymerase. The
reaction was then cooled on ice for 2 min. The DNA was digested with NotI by adding
to the sample, 8 μl of 2.5×buffer 2 (20×buffer 2 is 3 M NaCl, 0.2%
Triton X-100, 0.2% BSA) and 2 μl (10 U/μl) of NotI. The sample was
incubated at 37° C. for 2 hr. The DNA was then radioactively labeled. This
was done by adding to the sample 0.3 μl 1 M DTT, 1 μl [α-
32P]-GTP,
1 μl [α-
32P]-dCTP and 0.1 1 μl[α-
32P]-GTP
Sequenase ver 2.0 (13 U/μl). The mixture was incubated at 37° C. for
30 min. The DNA was then digested with EcoRV by adding to the sample 7.6 μl
second enzyme digestion buffer (1 μl 1 mM ddGTP, 1 ul 1 mM ddCTP, 4.4 μl
ddH
2O, 1.2 μl 100 mM MgCl
2) and 2 μl EcoRV (10
U/μl). The mixture was incubated at 37° C. for 1 hr. Then, 7 μl
of 6× first-dimension loading dye (0.25% Bromophenol Blue, 0.25% Xylene Cyanol,
15% Ficoll type 400) was added.
B. First Dimension Gel Set-up and Electrophoresis
To make the 60 cm long agarose tube-shaped gel, a gel holder was made. To do
this,
a sharp razor was used to cut one end of PFA-grade teflon tubing (PFA 11 thin wall,
natural; American Plastic, Columbus, Ohio) at an angle to make a bevel. The beveled
end of the tubing was fed into glass tubes (4 mm inner diameter, 5 mm outer diameter,
60 cm long). Using a hemostat, the beveled end was pulled up through the tapered
end of the glass rod until it protruded 2 to 4 cm. The tubing was cut horizontally
at the same end, leaving a 2 mm protrusion (this is the top of the gel holder).
The opposite end was cut horizontally, leaving a 5 to 6 cm protrusion from the
glass tube. The gel holder was inverted and the top protruding end was pressed
firmly against a hot metal surface (metal spatula heated by a Bunsen burner) to
fold the edges of the teflon outward onto the rim of the glass support A rubber
stopper with cored center was pulled over the top end of the gel holder until it
was just past the taper of the glass rod. A two-way stopcock was attached to a
10 ml syringe and then to the gel holder via 2 to 3 cm of flexible tubing.
