 |
Description  |
|
|
FIELD OF THE INVENTION
This invention relates to DNA sequencing using reporter-labeled DNA and,
more particularly, to a fluorescence-based system for detecting the
presence of radiant energy from different species following separation in
time and/or space, and fluorescent dyes for use therewith. The dyes are a
family of closely-related yet distinguishable fluorescent dyes. Methods
for protecting, activating, and coupling these dyes are disclosed. A set
of fluorescence-labeled DNA chain-terminators, prepared using this
methodology, are employed for the generation of fluorescence-labeled DNA
sequencing fragments. A photometric detection system capable of detecting
these fragments during electrophoretic separation and identifying the
attached fluorescent reporter is described.
BACKGROUND OF THE INVENTION
DNA sequencing is one of the cornerstone analytical techniques of modern
molecular biology. The development of reliable methods for sequencing has
led to great advances in the understanding of the organization of genetic
information and has made possible the manipulation of genetic material
(i.e. genetic engineering).
There are currently two general methods for sequencing DNA: the
Maxam-Gilbert chemical degradation method [A. M. Maxam et al., Meth. in
Enzym. 65 499-559 (1980)] and the Sanger dideoxy chain termination method
[F. Sanger. et al., Proc. Nat. Acad. Sci. USA 74 5463-5467 (1977)]. A
common feature of these two techniques is the generation of a set of DNA
fragments which are analyzed by electrophoresis. The techniques differ in
the methods used to prepare these fragments.
With the Maxam-Gilbert technique, DNA fragments are prepared through
base-specific, chemical cleavage of the piece of DNA to be sequenced. The
piece of DNA to be sequenced is first 5'-end-labeled with .sup.32 P and
then divided into four portions. Each portion is subjected to a different
set of chemical treatments designed to cleave DNA at positions adjacent to
a given base (or bases). The result is that all labeled fragments will
have the same 5'-terminus as the original piece of DNA and will have
3'-termini defined by the positions of cleavage. This treatment is done
under conditions which generate DNA fragments which are of convenient
lengths for separation by gel electrophoresis.
With Sanger's technique DNA fragments are produced through partial
enzymatic copying (i.e. synthesis) of the piece of DNA to be sequenced. In
the most common version, the piece of DNA to be sequenced is inserted,
using standard techniques, into a large, circular, single-stranded piece
of DNA such as the bacteriophage M13. This becomes the template for the
copying process. A short piece of DNA with its sequence complementary to a
region of the template just upstream from the insert is annealed to the
template to serve as a primer for the synthesis. In the presence of the
four natural deoxyribonucleoside triphosphates (dNTP's), a DNA polymerase
will extend the primer from the 3'-end to produce a complementary copy of
the template in the region of the insert. To produce a complete set of
sequencing fragments, four reactions are run in parallel, each containing
the four dNTP's along with a single dideoxyribonucleoside triphosphate
(ddNTP) terminator, one for each base (.sup.32 P-Labeled dNTP is added to
afford labeled fragments.) If a dNTP is incorporated by the polymerase,
chain extension can continue. If the corresponding ddNTP is selected, the
chain is terminated. The ratio of ddNTP to dNTP's is adjusted to generate
DNA fragments of appropriate lengths. Each of the four reaction mixtures
will, thus, contain a distribution of fragments with the same
dideoxynucleoside residue at the 3'-terminus and a primer-defined
5'-terminus.
In both methods, base sequence information which generally cannot be
directly determined by physical methods has been converted into
chain-length information which can be determined. This determination can
be accomplished through electrophoretic separation. Under denaturing
conditions (high temperature, urea present, etc.) short DNA fragments
migrate as stiff rods. If a gel matrix is employed for the
electrophoresis, the DNA fragments will be sorted by size. The single-base
resolution required for sequencing can usually be obtained for DNA
fragments containing up to several hundred bases.
To determine a full sequence, the four sets of fragments produced by either
Maxam-Gilbert or Sanger methodology are subjected to electrophoresis in
four parallel lanes. This results in the fragments being spatially
resolved along the length of the gel. The pattern of labeled fragments is
typically transferred to film by autoradiography (i.e. an exposure is
produced by sandwiching the gel and the film for a period of time). The
developed film shows a continuum of bands distributed between the four
lanes often referred to as a sequencing ladder. The ladder is read by
visually scanning the film (starting with the short, faster moving
fragments) and determining the lane in which the next band occurs for each
step on the ladder. Since each lane is associated with a given base (or
combination of bases in the Maxam-Gilbert case), the linear progression of
lane assignments translates directly into base sequence.
The Sanger and Maxam-Gilbert methods for DNA sequencing are conceptually
elegant and efficacious but they are operationally difficult and
time-consuming. Analysis of these techniques shows that many of the
problems stem from the use of a single radioisotopic reporter.
The use of short-lived radioisotopes such as .sup.32 P at high specific
activity is problematic from both a logistical and a health-and-safety
point of view. The short half-life of .sup.32 P requires that reagent
requirements must be anticipated several days in advance and that the
reagent be used promptly. Once labeled DNA sequencing fragments are
generated they are prone to self-destruction and must be immediately
subjected to electrophoretic analysis. The large electrophoresis gels
required to achieve single base separation lead to large volumes of
contaminated buffer which must be disposed of properly. The
autoradiography required to subsequently visualize the labeled DNA
fragments in the gel is a slow process (overnight exposures are common)
and adds considerable time to the overall operation. Finally, there are
the possible health risks associated with use of such potent
radioisotopes.
The use of only a single reporter to analyze the position of four bases
lends considerable operational complexity to the overall process. The
chemical/enzymatic steps must be run in separate containers and
electrophoretic analysis must be carried out in four parallel lanes.
Thermally induced distortions in mobility result in skewed images of
labeled DNA fragments (e.g. the smile effect) which in turn, lead to
difficulties in comparing the four lanes. These distortions often limit
the number of bases that can be read on a single gel.
The long times required for autradiographic imaging along with the
necessity of using four parallel lanes force one into a "snapshot" mode of
visualization. Since one needs simultaneous spatial resolution of a large
number of bands one is forced to use large gels that are typically 40 cm
or more in length. This results in additional problems: large gels are
difficult to handle and are slow to run adding more time to the overall
process.
Finally, there is a problem of manual interpretation. Conversion of a
sequencing ladder into a base sequence is a time-intensive, error prone
process requiring the full attention of a highly skilled scientist.
Numerous attempts have been made to automate the reading and some
mechanical aids do exist but the process of interpreting a sequence gel is
still painstaking and slow.
To address these problems one can consider replacing .sup.32
P/autoradiography with some alternative, non-radioisotopic
reporter/detection system. Such a detection system would have to be
exceptionally sensitive to achieve a sensitivity comparable to .sup.32 P;
each band on a sequencing gel contains on the order of 10.sup.-16 mole of
DNA. One method of detection which is capable of reaching this level of
sensitivity is fluorescence. DNA fragments could be labeled with one or
more fluorescent dyes. Excitation with an appropriate light source would
result in a characteristic emission from the dye thus identifying the
band.
The use of a fluorescent dye as opposed to a radioisotopic label would
allow one to more easily tailor the detection system for this particular
application. For example, the use of four different fluorescent dyes
distinguishable on the basis of some emission characteristic (e.g.
spectrum, life-time, polarization) would allow one to uniquely link a
given tag with the sequencing fragments associated with a given base. With
this linkage established, the fragments could be combined and resolved in
a single lane and the base assignment could be made directly on the basis
of the chosen emission characteristic.
The "real-time" nature of fluorescence detection would allow one either to
rapidly scan a gel containing spatially resolved bands (resolution in
space) or sit at a single point on the gel and detect bands as they
sequentially pass through the detection zone (resolution in time). Large
gels would not necessarily be required. Furthermore, a "real-time", single
lane detection mode would be very amenable to fully automated base
assignment and data transfer.
Several attempts to develop a fluorescence-based DNA sequencing system have
been described. One system developed by a group at the California
Institute of Technology, has been disclosed in L. M. Smith, West German
Pat. Appl. #DE 3446635 Al (1984): L. E. Hood et al., West German Pat.
Appl. #DE 3501306 Al (1985): and L. M. Smith et al., Nucleic Acids
Research, 13 2399-2412 (1985). This system conceptually addresses the
problems described in the previous section but the specifics of the
implementation appear to render this approach only partially successful.
The Cal Tech system employs four sets of DNA sequencing fragments, each
labeled with one of four fluorescent dyes. Two representative sets of
fluorescent dyes are described. Each set is comprised of dyes from at
least two different structural classes.
The emission maxima are spread over a large range (approximately 100 nm) to
facilitate discrimination between the four, but unfortunately the
absorption (excitation) maxima are also-comparably spread. This makes it
very difficult to efficiently excite all four dyes with a single
monochromatic source and adequately detect the resulting emissions.
In contrast, the use of dyes with closely spaced absorption (and
corresponding emission) peaks to enhance the excitation efficiency causes
other difficulties. A detection system for DNA sequencing must be able to
distinguish between four different dye emission spectra in order to
identify the individual labeled fragments. These emissions are typically
of relatively low intensity. Therefore, the detection system must have a
high degree of sensitivity (better than 10.sup.-16 moles DNA per band) and
selectivity, along with a means to minimize stray light and background
noise, in order to meet desired performance characteristics. The system
also must be able to frequently monitor the-detection area in order to
avoid missing any fragments that migrate through the gel past the
detection window. Such a detection system should be relatively cost
efficient to allow for multiple detection devices within a single
instrument without detrimentally affecting mill cost.
Many detection devices are known which utilize fluorescence in a detection
scheme. One such device is discussed in "Quantitative Fluorescence
Analysis of Different Conformational Forms of DNA Bound to the Dye . . .
and Separated by Electrophoresis" by Naimski et al., Anal. Biochem.,
106,471-475, 1980. In this electrophoresis/detection system a glass tube
is filled with agarose gel for separating the relatively large DNA
fragments. A scanning monochromator is then used as the detection system
for defining each of the large fragments. It is known that scanning
monochromators can accurately measure a wide range of spectral
characteristics; however, much light is lost due to the limited ability of
the monochromator and its associated optics to collect and disperse
emitted light. These detection techniques limit the fraction of light that
can be sensed and measured. Consequently, their sensitivity for low light
applications is limited. Additionally, light collected sequentially is
typically inefficient.
The detection apparatus disclosed by Smith et al. (see above) uses a series
of narrow band interference filters in order to select the wavelengths
impinging upon a single or multiple photodetectors. This type of system
has the advantage of being rather simple and inexpensive; however, it does
have substantial deficiencies. The specific system described uses a filter
photometer which can either use multiple interchangeable filters with one
photodetector or multiple stationary filters with corresponding detectors.
The first of these devices, a rotary filter with a single detector (see
FIG. 3 of Smith et al.), has the disadvantage of limiting the time period
during which each of the filtered regions can be measured. The detector
time must be shared with the different filters in order to distinguish
among different emission spectra. The Smith et al. system has additional
optical difficulties which need not be dealt with here.
More serious problems still result from using dyes which have different net
charges. The conventional sequencing gel displayed in the Smith et al.,
Nucleic Acids Research paper illustrates T-lanes produced from primers
labeled with each of four dyes. It is clear that there are significant
differential perturbations in the electrophoretic mobilities. A complete
set of sequencing fragments bearing these four dyes will, when combined,
show considerable overlap and perhaps even misordering when subjected to
electrophoresis in a single lane. This effect, combined with the
aforementioned large dynamic range in signal intensity, makes it difficult
to perform single-lane sequencing with this dye set.
Finally, the methodology used to prepare the fluorescence-labeled
sequencing fragments creates difficult sequencing conditions. For
Maxam-Gilbert sequencing, 5'-labeled oligonucleotides are enzymatically
ligated to "sticky ended", double-stranded fragments of DNA produced
through restriction cleavage. This limits one to sequencing fragments
produced in this fashion. For Sanger sequencing, 5'-labeled
oligonucleotides are used as primers. Four special primers are required.
To use a new vector system one has to go through the complex process of
synthesizing and purifying four new dye-labeled primers.
A second approach to automation of non-radiolabel DNA sequencing was
disclosed by Ansorge, W., et al., J. Biochem. Biophys. Methods, 13:315-323
(1986), in which a single fluorescent label was covalently attached to the
5' end of a 17-base oligonucleotide primer. This primer was reacted in
four vessels with the standard dideoxynucleotide sequencing chemistry
method that was modified to omit the radiolabeled nucleotide, to produce
sets of enzymatically copied DNA fragments of varying length. Each of the
four vessels contained a dideoxynucleotide chain terminator corresponding
to one of the four DNA bases which allowed terminal base assignment from
conventional electrophoretic separation in four gel lanes. Each fragment
carried a 5'-tetramethylrhodamine fluorescent label which was excited by
an argon ion laser passing through the width of the entire gel.
Fluorescent emissions of DNA bands resolved over time were collected from
the four lanes with separate, stationary means for each lane comprising
imaging optics, field apertures, light guides, filter assemblies, and
photomultipliers in series.
One advantage claimed by this approach is the need for fewer moving parts
in the apparatus due to stationary detectors which allows continuous
monitoring of the four gel lanes. This monitoring method, although more
complex than that used with one lane, reportedly offers the advantage of
determining the presence of a labeled band for base assignment relative to
the absence of bands in the remaining three lanes to improve confidence in
the assignment. In fact, the use of a single label requires the use of
four lanes for base assignment and the system as presented is incapable of
further simplification to improve the capacity or throughput of the
instrument. The system is also limited in potential accuracy by the
requirement for faithful lane-to-lane relative positioning for a single
sequence analysis. Operational complexities such as thermal gradients and
gel impurities may defeat this positional integrity to produce local gel
distortions which affect band mobilities, that may in turn, compromise
base sequence assignments.
The use of labeled primers by Ansorge et al. and Smith et al. is inferior
in other respects as well. The polymerization reactions must still be
carried out in separate vessels. All DNA fragments--be they bona fide
termination fragments or extraneous fragments--will be labeled. This is
similar to the existing system where effectively all fragments containing
incorporated adenosine nucleotides are labeled. Thus, the resulting
sequencing pattern will retain most of the artifacts (e.g. false or shadow
bands, pile-ups) encountered in the current methods.
Finally. Brumbaugh, J. A. et al. in European Patent Application 85103155.9,
published Oct. 9, 1985, disclosed a system and method for post-labeling
strands of DNA which optionally contained pre-marked nucleosides. The
pre-marking could be accomplished by covalent attachment of biotin to a
desired chain terminating nucleotide before the nucleotide was used in a
modification of the Sanger DNA chain termination method. However, the
pre-marked nucleotide was not detectable in the disclosed system. The
pre-marked strands of DNA prepared in separate vessels corresponding to
the A, T, C, and G DNA bases, were electrophoretically separated and then
exposed to a complementary binding material, typically avidin, which had a
fluorophore such as fluorescein covalently attached to it. The fluorophore
was detected and the signal presence was related to the particular vessel
or gel/lane corresponding to A, T, C, or G originally prepared. This
post-labeling method requires the preparation and subsequent
electrophoretic separation of marked DNA strands in separate vessels and
gels/lanes, respectively. There is no disclosure of any method or system
capable of labeling DNA strands differentially in the same vessel
simultaneously during the reactions of a chain termination method, or
differentiating labels during strand detection in a single gel/lane of a
suitable detection system.
SUMMARY OF THE INVENTION
This invention seeks to overcome many of the disadvantages of the prior
art. It includes a DNA sequencing system which has many of the desired
performance characteristics without many of the deficiencies previously
discussed. This invention is a system for detecting the presence of
radiant energy from different species, typically reporter-labeled DNA,
following separation in time and/or space, and identifying the species.
The system includes means responsive to the spectra of the species for
generating a first signal that varies in amplitude in a first sense as a
function of the nature of the species, means responsive to the spectra for
generating a second signal that varies in amplitude in a second sense
different than the first sense as a function of the nature of the species,
and means responsive to the first and second signals for obtaining a third
signal corresponding to the ratio of functions of the first and second
signals, the amplitude of the third signal being indicative of the
identity of the species.
The means for generating the first and second signals may include a
dichroic filter, with a transmission/reflection characteristic that varies
as a function of wavelength, means to direct the emissions to the filter,
first and second detectors positioned respectively to receive the
transmitted and reflected emissions and generate first and second signals
corresponding to the intensities of each. Preferably the dichroic filter
characteristic has a relatively sharp transition from transmission to
reflection which occurs near the center of the species emission spectra.
With the transition point located near the center, the change in amplitude
of the third signal is more evenly distributed over the range of spectra.
Typically, the species to be analyzed are DNA fragments or other molecules
covalently labeled with fluorescent materials that have closely spaced
spectra. These molecules are typically contained in an electrophoresis gel
adapted to separate them by size, charge, or other physical properties.
The system includes a laser or other radiant energy source with an output
within the excitation region of the fluorescent material.
The emitting portion of the labeled species to be detected, e.g., a DNA
fragment, can have the structure
##STR1##
wherein n=2 or 3 and R.sub.1 and R.sub.2 include H, lower alkylo halo,
lower alkoxy and cyano.
There is also described a method of detecting the presence of radiant
energy emitted from different species, following separation in time and/or
space, and determining the identity of each such species that emit,
comprising the steps of obtaining functions of the emitted energy which
vary over the wavelengths of the closely spaced spectra in different
senses, ratioing such functions, the ratio being indicative of the
identity of the species. The function of the emitted energy may be
obtained by passing the radiant energy through a dichroic filter with a
transmission/reflection characteristic which varies as a function of
wavelength.
A process is described for DNA sequence analysis according to a
modification of the Sanger chain termination method where the chain
terminator carries the reporter. Preferably, the chain terminator carries
a colored, more preferably fluorescent reporter. The chain terminator can
be one of the following structures
##STR2##
where (a) X is H, NH.sub.2, or halo, and Y is H, NH.sub.2, OH, or halo, or
(b) X.dbd.Y.dbd.OH.
or
##STR3##
where A can be a fluorescent reporter having the structure
##STR4##
where n is 2 or 3, and R.sub.1 and R.sub.2 are H, lower alkyl, halo, lower
alkoxy, and cyano, B is a heterocyclic base such as uracil, cytosine,
7-deazaadenine, 7-deazaguanine, or 7-deazahypoxanthine where the
pyrimidines are linked to the sugar part through the N.sub.1 position and
the deazapurines are linked to the sugar part through the N.sub.9 position
(purine numbering), and the dotted line is a linker and optional spacer
(group of atoms) joining the fluorescent part (A), preferably via an amide
bond, and the heterocyclic base (B) provided that if B is a pyrimidine the
linker is attached to the 5-position of that pyrimidine and if B is a
deazapurine the linker is attached to the 7-position (purine numbering) of
that deazapurine.
According to another aspect of this invention, a process of DNA sequence
analysis according to the Sanger chain termination method as modified by
this invention, is provided where each of the four chain terminators
corresponding to the four bases carries a different distinguishable
reporter. The four chain termination reactions may thus be carried out in
separate vessels and combined prior to electrophoretic analysis or carried
out in a single vessel.
The combined DNA sequencing fragments thus produced can be subjected to
simultaneous electrophoretic separation in a single lane. Excitation of
the fragments bearing their respective reporters by a single source
results in their characteristic emission thereby allowing detection and
identification.
The set of four reporters are chosen such that all four are efficiently
excited by a single source and have emission spectra that are similar but
distinguishable. The differential perturbations in electrophoretic
mobility of the attached DNA fragments are small. This requirement can
generally be satisfied if the four reporters have similar molecular
weights, shape, and charge.
These criteria can be met with reporters having fluorescent parts with the
structure
##STR5##
where n is 2 or 3 and R.sub.1 and R.sub.2 are chosen from the group H,
lower alkyl, lower alkoxy, halo, and cyano. Such reporters may be
introduced via protected, activated intermediates of the structure
##STR6##
where n is 2 or 3, R.sub.1 and R.sub.2 are H, lower alkyl, lower alkoxy,
halo, and alkoxy, R" is alkyl, R' is alkyl or aryl, and X is a good
leaving group.
The system and method of this invention have the ability to distinguish in
real time between the relatively small wavelength differences in emission
spectra, while maintaining a relatively high degree of sensitivity. The
system delivers a high portion of the usable light onto the photometric
detectors. Finally, the detection system provides continuous monitoring of
the gel containing the fluorescent species. This feature reduces the
possibility of deriving incomplete data that are typically inherent in
intermittent type detection systems. All of the above-mentioned features
are incorporated into this unique system at a relatively low mill cost.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention may be more fully understood from the following detailed
description thereof taken in connection with accompanying drawings which
form a part of this application and in which:
FIG. 1 is a partial block, partial diagrammatic layout of a system
constructed in accordance with this invention for detecting the presence
of radiant energy from different sources that each emit energy at
different but closely spaced wavelengths:
FIG. 2 is a diagrammatic representation of a single electrophoresis gel
capable of being used in the system of FIG. 1
FIG. 3 shows the relationship of the dye emission radiation and the
transmission/reflection characteristics of the dichroic filter used in the
system of FIG. 1
FIGS. 4A, 4B and 4C are flow diagrams of the data processing steps used to
evaluate the detected emission peaks
FIGS. 4 and 5 shows typical detector signals obtained as a function of time
for the labeled bases T, G, A and C
FIG. 6 is a block diagram of a fluorescence-labeled chain terminator and
FIG. 7 shows the relationship of the dye emission radiation and the
passbands of two filters used to replace the dichroic filter.
DETAILED DESCRIPTION OF THE INVENTION
The radiation from very closely spaced emission bands may be detected using
the system of this invention. These closely spaoed emissions are produced
from preselected reporter species which are irreversibly bound to the
materials that are to be analyzed in the system. Acceptable reporters are
generally one or more species chosen for their ability to emit radiation
over a narrow range of wavelengths, typically between a 50 and 100 nm
range, preferably over a 20 to 50 nm range. Preferably, the peak maxima
should be spaced no closer than 2 nm. One reporter species may be capable
of emitting energy at more than one wavelength range, depending upon the
manner of attachment to the materials of interest and the conditions of
analysis in the system. However, individual reporters with unique emission
characteristics in the system are more conventionally chosen to emit
radiation in the wavelength range to be detected. Preferred reporter
species are described below.
Although the system of this invention has broad applicability, it will be
described in a particular application of DNA sequencing.
In this preferred form of the invention, reporter-labeled DNA sequencing
fragments are produced in a single vessel. The contents of this vessel,
reporter-labeled DNA chains of varying lengths, are passed through an
electrophoresis apparatus for separation. For this purpose, as is
illustrated in FIG. 2, the electrophoresis may be carried out by a
suitable electrophoresis slab 10 arrangement having a thickness of about
0.2 mm to 0.4 mm and about 25 to 40 centimeters long. Other sizes may be
used as appropriate. This slab 10 has a suitable gel 11, typically 6% to
8% polyacrylamide sandwiched between glass or plastic supports 12. The
slabs (gels) are prepared in a conventional manner.
The slab 10 is typically placed in an upright position in a holder with the
upper end of the slab 10 extending through and into an upper container 16
holding a buffer solution 24 and downwardly into a second container 14
also holding a buffer solution 18. The buffer solution is any suitable
buffer typically a 0.1M tris-borate-EDTA, pH 8.3 may be used. In this
manner, the buffer contacts the gel at either end of the slab in order to
make electrical contact therewith. With this arrangement, a sample of the
reporter-labeled DNA fragments can be pipetted into a cavity (not shown)
that is created at the top of the gel. An electrical circuit is then
completed through the terminals 20 in reservoir containers 14 and 16. A
suitable potential is needed to obtain separations for gels of this
particular length and thickness. The positive electrode is located at the
lower end of the slab to cause the DNA fragments to migrate downwardly.
Under these conditions, as the fragments migrate through the gel they are
separated spatially into bands. The detection zone is located near the
bottom of the slab. In this zone, the fragments are irradiated by a laser
beam 31 and excitation/emission occurs as the fragments move through the
zone.
An optical arrangement for irradiating the electrophoresis slab 10 is shown
in FIG. 1. The system of FIG. 1 may be used with any fluorescent or other
type reporter system to distinguish between and measure the intensity of
closely spaced emission radiation bands. However, it will be described, as
noted above, in the preferred application of detecting the emissions from
reporter-labeled DNA fragments, where the reporter species are fluorescent
compounds. It includes the laser 30 which is selected to provide a
specific wavelength determined as a function of the excitation wavelengths
of the preferred excitable reporter species used. For example, the
specific source used for the fluorescent reporters disclosed herein is an
argon ion laser with a wavelength of 488 nm and a 0.8 mm diameter light
beam operated at about 25 to 40 mW. The laser beam passes through an
excitation filter 32 and focusing lens 33 which concentrates the beam to a
diameter of about 0.2 mm at the detection zone of the electrophoresis slab
10.
The filter 32 is selected to block out the undesired excitation wavelengths
that could otherwise interfere with the detection process. If the laser
light is very pure, this filter may be omitted. The light beam entering
the slab excites the reporter-labeled material, here fluorescently labeled
DNA fragments, as they migrate through the detection zone, causing them to
fluoresce at wavelengths shifted from the excitation wavelength. Although
the peak emission wavelengths characteristic of the particular dyes
disclosed hereinafter, when free in solution, are 505, 512, 519 and 526
nm, it is to be noted that the detection system is adaptable to
discriminate wavelengths associated with other sets of reporters with
closely spaced emission bands. Furthermore, while a laser source is
preferred since it allows a minimum of extraneous light to impinge upon
the sample, with suitable filtering and optics, other sources including a
non-coherent source such as a xenon arc lamp could be used.
The light emitted by the fluorescent species is collected by suitably
positioned collimating lens 34 which produces a collimated beam of light
for transmission to dichroic filter 38 via an emission filter 36 to
eliminate essentially all light except for the specific wavelengths
characteristic of the fluorescence. Using this filter, substantially all
of the light below 500 nm and above 560 nm is filtered out with the light
between these limits being transmitted with greater than 50% efficiency.
In accordance with this invention, the dichroic interference filter 38
enables this system to distinguish between closely spaced emission
spectra. It is oriented typically to about 45.degree. with respect to the
incident beam. Light impinging on this filter will either be reflected or
transmitted through the filter. For the emission maxima of the fluorescent
reporters, when coupled to DNA fragments, disclosed hereinafter, namely,
515, 524, 530 and 536 nm filter 38 has been chosen to have the
reflective/transmissive characteristics shown in FIG. 3. The dichroic
filter 38 is seen to have a sharp transmission/reflection transition 39
which lies approximately in the center of the fluorescence bands which are
characteristic of these four reporters. As the fluorescence spectrum
shifts from the lower to higher wavelengths, the ratio of transmitted to
reflected light decreases in a continuous manner. Although this particular
filter has been chosen to accommodate the reporters selected for this
application, a different set of reporters would require different filter
characteristics.
Light (reflected or transmitted) from the dichroic filter 38 passes through
respective focusing lenses 40 to respective detectors 42. The detectors
42, preferably are photomultiplier tubes. They are known to have a high
degree of sensitivity within the spectral bands of interest.
Alternatively, silicon photodiodes or other similar detectors may be used.
In the instance where the detectors 42 are positioned within a close,
predetermined distance from a collecting aperture, the collimating lens 34
can be omitted. When this lens is omitted, the collecting aperture may be
defined by having an opening that corresponds to the desired sensing area
of the detector 42, or an aperture can be defined by an alternate means,
such as a fiber optic face plate. In similar fashion, the focusing lenses
40 can be omitted if the detector sensing area is large enough to
sufficiently and directly collect the available light.
It should be apparent that the function of the dichroic filter 38 could
also be and preferably is served by two separate filters and appropriate
apertures placed in the light path from the emitted light source to the
detectors 42, in which each filter has in a transmission sense, two
different transmission characteristics. i.e., either the reflective or
transmissive transition characteristic of the dichroic filter. In this
manner, the two detectors 42 are still dedicated to the transmissive and
reflective characteristics provided by the dichroic filter in the previous
description. The passbands of these two filters is seen more clearly in
FIG. 7 and are labeled Filter 1 and Filter 2. The passbands of these two
filters are seen to overlap at about the center of the fluorescence bands
which are characteristic of the four reporter labeled terminators used
herein of G515, A524, C530, and T536 nm.
A system of this type using two filters is described in the copending
Robertson et al. application, the contents of which are incorporated
herein by reference. As described in Robertson et al., a pair of modules
are positioned above and below a plane in which the reporter exciting
light beam scans multiple lanes on an electrophoresis gel. Each channel
contains reporter-labeled DNA fragments. Each detection module comprises a
photomultiplier tube having a wide entrance area and a separate wavelength
selective filter positioned between its PMT and the fluorescent species in
the gel. These filters are interference filters having complementary
transmission band characteristic which simulate the dichroic filter
action. The filters permit the PMT's to generate signals that vary in
amplitude in different senses as a function of the nature of the species.
One filter largely passes the lower emission wavelengths and rejects the
high emission wavelengths while the other filter does precisely the
reverse. Transmission filters may be used with each interference filter to
reject light from off axis angles greater than a predetermined angle. The
wavelength filters have roughly complementary transmission vs. wavelength
characteristics in the emission region of the four dyes, with the
transition wavelengths occurring near the center of the species radiant
energy spectra.
The electrical signals from the detectors 42 are then passed via respective
preamplifiers 46 to analog-to-digital (A/D) converters 48 and thence to a
system controller 52. The tasks of the system controller 52 may be
performed by a small computer such as an IBM PC. A function of the system
controller which is described by the flow diagram of FIG. 4, is to compute
the ratio of the two signal functions. The dichroic filter 38 modulates
the intensity of the signals in each of the different wavelength bands
according to wavelength, i.e., for the reflected light detector, the
shorter wavelength emissions will have a lower amplitude signal value and
the longer wavelength emissions will have a higher amplitude. Thus, as a
particular reporter species, i.e., a fluorescently-labeled DNA fragment in
the preferred embodiment of the invention, passes through the detection
zone following separation in space in the gel 10, its emissions will be
varying in amplitude as a function of its wavelength and also time
(because of the movement through the gel 10). The amplitude modulated
light signals are converted to electrical signals and digitized for such
processing as described. After conversion, the digital signals are
ratioed, i.e., to obtain the quotient of the reflected to transmitted
fluorescent light. These digital signals are those representing a peak in
the light signal corresponding to a set of DNA fragments. Signals
corresponding to either the peak height or peak area are the ones ratioed.
The magnitude of the ratio signal is indicative of the identity of the
species. The function W is defined as the ratio of peak intensity in one
detector to peak intensity in the other detector, for example, the peak
intensity in the transmitted light detector divided by the peak intensity
in the reflected light detector. The magnitude of the ratio signals for
each reporter tends to fall into grouping or clusters which are uniquely
indicative of each reporter as may be seen in the illustrative Example 10
below.
This amplitude modulation and ratioing procedure, whether accomplished with
a dichroic filter or with separate filters to generate two signals that
each vary in different senses in response to the same spectra of the same
species under test, may be described mathematically. Thus, the total
electrical signal (corresponding to the detected light signal) present
during a reporter peak emission consists of components due to scattering
and stray light as well as the signal due to the fluorescence itself from
each of the different reporters. During electrophoresis, the reporter
fluorescence signal may be distinguished from other background components
because the fluorescence signals vary in time or space in a predictable
manner. This is in contrast with the background noise signal which
contributes a relatively constant signal, particularly when there is no
relative motion between the detector and gel. In a stationary gel and
detector configuration, the fluorescence signals vary in time due to the
movement of the reporter through the gel. Alternatively, the gel may
remain stationary following electrophoresis, and the detection system
moved relative to the gel or vice versa. In still another alternative, the
detection system may be moved while migration in the gel is still taking
place.
There may be seen in FIG. 5 a representation of how the two detector output
signals vary as a function of time. Each pair of peaks in the figure
corresponds to a different set of DNA fragments which correspond to the
sequence of bases occurring in the piece of DNA under test. The ratios of
each pair of peaks falls into four groupings each corresponding to a
particular DNA base. It is these ratio groupings which identify the
particular base T, C, A, or G.
In order to improve selectivity between the reporters (determined by values
of W), the change in W over all the combinations of the different
fluorescent reporter emissions must be optimized. This can be accomplished
by choosing a dichroic filter, or its equivalent as noted above, with
transmission/reflection characteristics which change substantially over
the different reporter emission spectra. However, for a closely spaced
group of reporters, it is preferable to have a relatively sharp filter
transition that occurs near the center of the reporter emissions in order
to evenly distribute the change in W for the different emission spectra
(FIG. 3).
In essence, this system provides uncommon measurement sensitivity and
selectivity performance. With this unique system, light is efficiently
directed to the two closely | | |
