3226-3232 Nucleic Acids Research, 1994, Vol. 22, No. 15
`
`Cyanine dye dUTP analogs for enzymatic labeling of DNA
`probes
`
`Hong Yu+, Jean Chao, David Patek, Ratan Mujumdar, Swati Mujumdar and Alan S.Waggoner*
`Department of Biological Sciences and Center for Light Microscope Imaging and Biotechnology,
`Carnegie Mellon University, 4400 Fifth Avenue, Pittsburgh, PA 15213, USA
`
`Received December 13, 1993; Revised April 8, 1994; Accepted April 11, 1994
`
`ABSTRACT
`Fluorescence in situ hybridization (FISH) has become
`an indispensable tool in a variety of areas of research
`and clinical diagnostics. Many applications demand an
`approach for simultaneous detection of multiple target
`sequences that is rapid and simple, yet sensitive. In this
`work, we describe the synthesis of two new cyanine
`dye-labeled dUTP analogs, Cy3-dUTP and Cy5-dUTP.
`They are efficient substrates for DNA polymerases and
`can be incorporated into DNA probes by standard nick
`translation, random priming and polymerase chain
`reactions. Optimal labeling conditions have been
`identified which yield probes with 20-40 dyes per
`kilobase. The directly labeled DNA probes obtained
`with these analogs offer a simple approach for
`multicolor multisequence analysis that requires no
`secondary detection reagents and steps.
`
`INTRODUCTION
`Since the introduction of biotinylated nucleic acid probes in early
`1980's (1), fluorescence in situ hybridization (FISH) has found
`increasing application in many areas of research and clinical
`diagnosis. Examples include molecular cytogenetics (2, 3),
`nuclear organization (4-6), gene mapping (7-12) and gene
`amplification (13), prenatal (14, 15), tumor diagnosis (16, 17)
`and pathogen detection (18-21). Fluorescence techniques have
`advantages over isotopic techniques in safety, spatial resolution,
`speed of detection and simultaneous detection of multiple
`sequences in single cells using different fluorescent labels. The
`simultaneous use of FISH probes with different fluorophores is
`especially useful for detection of chromosomal translocations,
`deletions and amplifications, for gene mapping on chromosomes
`and for chromosome enumeration (22 -25).
`It has been demonstrated that up to seven different DNA targets
`can be simultaneously visualized by using combinatorial labeling
`of DNA probes with three fluorochromes (24). This technique
`is based on labeling each individual probe with a different
`combination of up to three markers including two indirect labels
`(biotin and digoxigenin). The availability of new colors of
`
`deoxynucleoside triphosphates would extend the power of
`combinatorial labeling. Furthermore, new fluorescent dNTPs will
`be valuable when combinatorial labeling is not optimal such as
`possible mis-identification of the sequence due to fluorescence
`signals that are spatially overlapping. Instead of combinatorial
`labeling, each nucleic acid probe can be labeled with a
`fluorophore that is spectrally distinct. This approach would
`provide probes of high sensitivity with simplified labeling and
`detection of multiple sequences.
`In this study, we introduce two cyanine dye-conjugated dUTP
`analogs for direct enzymatic labeling of nucleic acid probes.
`Direct labeling significantly simplifies the hybridization procedure
`by eliminating the detection steps with affinity binding proteins.
`This time- and cost-efficiency
`is magnified in multicolor
`multisequence hybridization assays. A directly labeled probe can
`be used in combination with probes labeled with other
`fluorochromes to obtain spectrally distinct signals. As many as
`five
`target sequences plus DAPI counter-staining of the
`chromosomes can be detected with currently
`available
`fluorophores and microscope systems.
`Nick translation, random priming and polymerase chain
`reactions (PCR) are the most commonly used enzymatic DNA
`labeling methods. Biotin-dUTP, digoxigenin-dUTP, FITC-dUTP
`and other chemically modified deoxynucleoside triphosphates
`have been successfully incorporated into DNA probes by these
`enzymatic reactions
`(24, 26 - 32). We have studied the
`incorporation of the cyanine dye dUTP analogs by nick
`translation, random priming and PCR labeling reactions and have
`optimized the labeling conditions so that the maximal number
`of dyes can be incorporated into probes and the brightest
`hybridization signals can be obtained. Multicolor FISH with three
`probes directly tagged with FITC, Cy3 and Cy5 is demonstrated.
`
`MATERIALS AND METHODS
`Synthesis of cyanine dye conjugated dUTPs
`The synthesis of 5-(3-amino)allyl deoxyuridine 5' triphosphate
`(AA-dUTP) was based on a previously published procedure (1).
`The characterization of AA-dUTP was performed by ninhydrin
`
`*To whom correspondence should be addressed
`+Present address: Abbott Diagnostic Division, Dept. 9MG, Bldg. AP8, One Abbott Park Road, Abbott Park, IL 60064, USA
`
`Illumina Ex. 1052
`IPR Petition - USP 10,435,742
`
`

`

`assay (33), UV spectrum (1) and comparison to proton NMR
`spectrum provided by Dr David Ward. AA-dUTP is also
`commercially available.
`To synthesize Cy3-dUTP, AA-dUTP was dissolved in 0. 1M
`sodium borate buffer pH 8.7. An equal molar amount of
`Cy3.29OSu (a succinimidyl ester, available from Biological
`Detection Systems, 955 William Pitt Way, Pittsburgh, PA 15238)
`dissolved in small amount of neutral water was added to the AA-
`dUTP in aliquots within half an hour. Total incubation was 1
`to 2 hours at room temperature. The labeled product was purified
`by reversed-phase chromatography. The column was first washed
`with H20 and the product was eluted with increasing
`MeOH:H20 gradient. Fractions were checked on reversed-
`phase thin layer chromotography (TLC) and those containing
`cyanine-dUTP were pulled. The TLC plate was first run in
`distilled water and let dry. Then, 30:70 (v/v) MeOH:H20 was
`used to separate the dUTP analog from free dye and other
`impurities. Rf of the free dye was 0.8 and that of Cy3-dUTP
`was 0.58. In case of Cy5-dUTP, 37:63 (v/v) MeOH:H20 was
`used. Rf of the free dye was 0.70 and that of the Cy5-dUTP was
`0.51. The product was concentrated by rotary evaporation to a
`small volume and further purified by reversed-phase HPLC on
`a 250 x4.6mm C18 column (Alltech). A linear gradient of 0.05M
`phosphate buffer, pH6.0 and MeOH was used. The HPLC
`fractions were collected and rotary evaporated to remove MeOH.
`They were then desalted on disposable C18 guard columns
`(Millipore) with H20 and eluted with 50:50 (v/v) MeOH:H20.
`The final product was concentrated to a small volume and stored
`in sterile water at -20°C.
`
`Labeling probes by nick-translation
`Nick translation was performed following standard procedures
`(34). The DNase I and the E. coli DNA polymerase I reactions
`were carried out simultaneously and sometimes sequentially. In
`a simultaneous reaction, the template DNA was incubated in a
`total of 50 AI of nick translation buffer (10x from Boehringer
`Mannheim), with 20,M of each of the dNTPs, cyanine-dUTP,
`and 5 units of DNA polymerase I and the appropriate amount
`of DNase I. Several reactions with a series of dilutions of DNase
`I (Boehringer Mannheim, 10 units/,ul) were tested. The reaction
`mixtures were incubated at 16°C for 90 min and stopped by
`adding 2 Al of 0.5M EDTA and 1.25 ,u of 5% SDS. When
`sequential reactions were used, the DNA template was incubated
`with about 1/500 unit (template size- and amount-dependent) of
`DNase I (Boehringer Mannheim, 10 units/Al stock) in nick
`translation buffer at 37°C for 10 min. The fragment length was
`checked on 1 % agarose gel while the reaction was sitting on ice.
`If the fragments were not short enough, the sample tube was
`returned to the 37°C water bath and further incubated. When
`the desired fragment length was reached, the nicked DNA was
`ethanol precipitated and resuspended in sterile H20. The
`fragments were then incubated in a total of 50 yl of nick
`translation buffer with 20 AM of each of the dNTPs, cyanine-
`dUTP, 5 units of DNA polymerase I (Boehringer Mannheim)
`for a minimum of 2 hours. Typically, in order to obtain
`reasonably accurate UV/VIS spectral measurement, 2 Ag of the
`template was nick translated in each sample. Unincorporated
`Cy3-dUTP was removed by double ethanol precipitation, spin
`columns or Ultrafree filters (Millipore). The purified probes were
`in 200 1.l sterile water for spectroscopic measurements.
`
`Nucleic Acids Research, 1994, Vol. 22, No. 15 3227
`
`Labeling probes by random priming
`For random priming reactions, 2 4g of linearized plasmid probe
`was denatured at 95°C for 5 minutes and immediately cooled
`on ice. The template was incubated with random priming buffer
`(10 x from Boehringer Mannheim), hexanucleotides (Boehringer
`Mannheim, lOx hexanucleotide solution), 100 /%M of each of
`the dNTPs, cyanine-dUTP, and S units of Klenow DNA
`polymerase (Boehringer Mannheim). The reaction mixture was
`incubated at 37°C for 30 minutes to 2 hours. The labeled probes
`were purified by ethanol precipitation and brought up in 200 Al
`sterile water for spectroscopic measurement.
`
`Labeling probes by PCR
`PCR reactions were carried out in a total of 100,l mix containing
`50 mM KCl, 10 mM Tris-HCl (pH 9.0 at 25°C), 1.5mM
`MgCl2 1 % Triton X-100, 40 ng of the template DNA, 0.25,M
`of 5'-end labeled primers, 5 units of Tli or Taq DNA polymerase
`(Promega), and 60 ,uM of each of the dNTPs. In the control
`reaction, 60 AM of all the four normal dNTPs and no Cy3-dUTP
`was used. In others, 50% of dTTP was substituted with different
`amounts of Cy3-dUTP. The template used was a chromosome
`1 plasmid probe containing 900 bp insert (ATCC cat. no. 59863).
`The primer sequences
`are: GCTGCAGGTCGACTCTAG
`(65-Reverse) and GAATTCGAGCTCGCCCGG (65-Forward),
`which direct the amplification of the insert. The primers were
`labeled at 5'-end with CyS.180Su and purified to homogeneity
`by HPLC as previously described (35). PCR was performed in
`DNA Thermal Cycler (Perkin Elmer Cetus). After an initial
`denaturation at 94°C for 4 min, 25 cycles of PCR were carried
`out with denaturation at 94°C for 1 min, annealing at 52°C for
`2 min and extension at 72°C for 3 min. The labeled probes were
`purified using Prep-A-Gene (BioRad) to remove unreacted
`Cy3-dUTP and excess CyS labeled primers. The probes were
`in 150 Al sterile water for spectroscopic measurement.
`PCRs were carried out in 100 1l to provide sufficient amount
`of labeled probes for spectroscopic measurement, but it can be
`scaled down to 10 ,ul in practical labeling reactions in order to
`save material.
`Characterization of fluorescently labeled probes
`UV/VIS spectra were obtained with a HP 8452 diode array
`spectrophotometer (Hewlett-Packard) equipped with a deuterium
`lamp. Fluorescence emission and excitation spectra were taken
`on Spex Fluorolog 2 spectrofluorometer (Spex Industries, Inc.)
`equipped with a Xenon lamp. Excitation and emission slits were
`1 and 3 mm respectively, corresponding to bandpasses of 1.8
`and 5.4 nm.
`For characterizing nick translated probes, UV/VIS absorption
`spectra of the purified probes were obtained. The labeling density
`of the labeled probe measured as dye-per-kilobase ratio (d/kb)
`were calculated using Beer's law. The extinction coefficient of
`Cy3 is 150,000 L/mol.cm (36) and the average extinction
`coefficient of a base was taken as 10,000 L/mol.cm (37). For
`random priming labeled probes, the d/kb were not calculated.
`In addition to the bases in the labeled probe, the template and
`excess primers contribute significantly to 260 nm absorption.
`Although the exact d/kb could not be calculated for random
`priming labeled probes, the ratio of dye absorbance to 260 nm
`absorbance still reflected the relative amount of dye molecules
`incorporated and could be used to monitor the incorporation.
`
`

`

`3228 Nucleic Acids Research, 1994, Vol. 22, No. 15
`
`For characterizing PCR labeled probes, the probes were
`purified with Prep-A-Gene. Cy3 and Cy5 emission spectra were
`measured for each sample and Cy5 emission spectra of the control
`samples were taken. Cy3/Cy5 molar ratios were converted from
`Cy3/Cy5 fluorescence intensity ratios using calibration standards
`of known Cy3 and Cy5 molar ratio. The amplification efficiency
`is relative to the control sample.
`In situ hybridization
`The hybridization protocol for interphase nuclei and metaphase
`chromosome spreads is based on a previously published
`procedure (26). When using Cy3- and Cy5-dUTP analogs, no
`secondary reagent
`is needed for
`After post-
`detection.
`hybridization wash, the slides were stained with 25,Ag/ml Hoechst
`in Hank's Balanced Salt Solution (HBSS) for 10 minutes at room
`temperature and washed twice in HBSS for 5 minutes. The slides
`were mounted in crystal mount (Biomeda Corp.) and sealed with
`nail polish.
`Microscope imaging and spot analysis
`Images were obtained with a Zeiss fluorescence microscope
`equipped with a 1OOx oil immersion objective, a Macintosh
`Quadra 950, a mercury arc lamp and a cooled CCD camera
`(Photometrics). The imaged fields were 512x512 pixels. A
`software package based on BDS Image (Biological Detection
`Systems, Inc. Pittsburgh, PA) was used for image collection and
`processing, including registration, uniform field correction and
`signal intensity measurement. A subroutine, 'spot analysis' was
`written to quantify hybridization signals. The 'FISH signal
`intensity' is defined as the sum of all pixel intensities within the
`spot minus the overall background. It is related to the brightness
`of the probe, the efficiency of hybridization, and the efficiency
`of the detection system. The overall background intensity was
`derived by multiplying the average background pixel intensity
`by spot area. The average background pixel intensity was
`determined at the periphery of the observed spot. The peripheral
`region for each spot was determined by dilating the spot until
`its intensity began to increase or no longer decreased by a pre-
`determined threshold intensity. The program also calculated
`signal-to-background, which was defined as maximum pixel
`intensity within the spot divided by the background pixel intensity.
`For each slide sample, about 30 signal dots were analyzed and
`results averaged. The standard deviation was in the range of 10%
`to 30%.
`
`RESULTS AND DISCUSSION
`Synthesis and purification of Cy3-dUTP and Cy5-dUTP
`analogs
`Like many other succinimidyl ester dyes, the cyanine dyes react
`with primary amino groups.
`Therefore, an amino-group-
`containing linker was attached to dUTP in order to obtain a
`cyanine-dUTP analog. The common modification site on dUTP
`is the C-5 position of the pyrimidine ring, which does not
`participate
`in the base-pair hydrogen bonding. We used
`5-(3-amino)allyl-dUTP (AA-dUTP) for the synthesis of cyanine
`dye-dUTP analogs. In order to avoid cross-linking, Cy3.290Su,
`a mono-reactive form of Cy3.18 (36) was used to label AA-
`dUTP. Figure 1 illustrates the chemical structure of cyanine-
`dUTP analogs. The space between the dye moiety and the dUTP
`
`-03S
`
`soi0
`
`o-ro-r-o-0-
`o- o. a-
`
`OH
`
`m=1 for Cy3-4-dUTP
`m=2 for Cy54-dUTP
`
`Figure 1. Chemical structure of cyanine-4-dUTP analogs.
`
`A
`
`B
`
`39.50
`
`27.79
`
`Figure 2. HPLC profile of Cy3-dUTP monitored at 550 nm with visible detector.
`A: Cy3-dUTP pre-purified by TLC (Materials and Methods) B: The second peak
`in A was collected and re-injected on HPLC using the same gradient. The numbers
`shown are time (minutes) after injection.
`
`is 4 atoms and therefore the conjugates are termed Cy3-4-dUTP
`and Cy54-dUTP according to previously used nomenclature (28).
`For simplicity, Cy3-dUTP and Cy5-dUTP are abbreviations used
`in the remainder of this manuscript. It should be noted that there
`is a 6 atom space between the fluorochrome and its reactive
`group, giving a total of 10 atom space between the fluorochrome
`and the base.
`The Cy3-dUTP conjugate was purified by reversed-phase
`column as described in materials and methods. The yield of the
`coupling step was about 50%. The product was further purified
`on a C18 reversed phase HPLC column. As shown in Figure
`2A, five components were resolved which were collected
`individually and lyophilized. Each of the HPLC components of
`
`

`

`Table 1. Characterization of probes generated by nick translation
`
`15
`5
`
`[dTTP] zM
`[Cy3-4-dUTP] uM
`labeling density
`(d/kb)
`-
`FISH signal intensity NC
`signal-to-background -
`
`10
`10
`
`8
`3500
`1.87
`
`0
`20
`
`0
`30
`
`0
`40
`
`0
`50
`
`30
`29,800
`6.29
`
`19
`38,500
`5.60
`
`25
`34,000
`5.81
`
`24
`28,000
`5.41
`
`The template was nicked with DNase I to obtain fragments with average length
`of 150-250bp. In the subsequent labeling reactions, dTTP was substituted with
`increasing amounts of Cy3-dUTP. Labeling density (d/kb) of purified probes were
`obtained from UVWVIS absorption spectra. Hybridization signals were quantified
`by microscope imaging (Materials and Methods). NC=not consistent.
`
`Table 2. Characterization of probes generated by random priming
`
`[dTTP] AM
`[Cy3-4-dUTP]
`,uM
`ODd e/OD260
`FISH signal
`intensity
`signal-to-
`background
`
`50
`
`50
`1.9%
`
`1,650
`
`1.74
`
`0
`
`100
`4.1%
`
`10,400
`
`2.04
`
`0
`
`150
`4.2%
`
`15,600
`
`2.94
`
`0
`
`200
`4.2%
`
`8,400
`
`2.24
`
`Linearized template DNA was labeled as described in Materials and Method.
`dTTP was partially or completely substituted by various amounts of Cy3-dUTP.
`The order of ODdye/OD260 indicates relative incorporation since base absorption
`resulted from the template, primer and the labeled fragments is constant for all
`experiments. Hybridization signals were quantified by microscope imaging.
`
`Cy3-dUTP was tested in nick translation reaction. All the
`components incorporated
`into DNA probe equally
`well.
`Furthermore, when the probes labeled with each of the fractions
`were used in interphase nuclei hybridization, spot analysis of the
`signals showed similar intensity (data not shown). Because of
`the small quantities of these material, further analysis was not
`performed. For the consistency of the data, only the major HPLC
`component (Figure 2B) was used in all the studies described in
`this work.
`Enzymatic preparation of Cy3-labeled DNA probes by nick
`translation
`Nick translation utilizes three different enzymatic activities,
`DNase activity of DNase I, 5' to 3' exonuclease activity and 5'
`to 3' DNA polymerase activity of E.coli DNA polymerase I. The
`procedure generates a pool of partially labeled DNA fragments.
`It was in our interest not only to label DNA probes with
`cyanine-dUTP but also to provide quantitative measurement of
`the incorporation and the brightness of the labeled probes. Unlike
`non-fluorescent markers, the incorporation of the fluorescent dyes
`into DNA probes can be conveniently and quantitatively
`monitored by measuring the absorbance or the fluorescence
`emission of the labeled probes. For example, the absorption of
`the dye and bases can be obtained from UV/VIS absorption
`spectrum of a labeled probe. Labeling density measured as dye-
`per-kilobase ratios (d/kb) can then be calculated using Beer's law
`knowing the extinction coefficient of the dye and the average
`extinction coefficient of the bases. This approach was used to
`monitor the extent of cyanine-dUTP incorporation under various
`labeling conditions. While it is important to incorporate a high
`number of dyes into a DNA probe, it is also important not to
`over-label DNA because dye-dye contacts can lead
`to
`
`Nucleic Acids Research, 1994, Vol. 22, No. 15 3229
`
`fluorescence quenching (36). Overlabeling may also reduce
`hybridization efficiency of the probe. Thus, it is essential to
`quantify actual FISH signals in order to identify the optimal
`labeling density. Therefore the probes labeled at various d/kb
`were hybridized with interphase nuclei on slides and images were
`obtained with an imaging microscope. Relative signal intensities
`and the signal-to-background of the hybridization signals were
`measured using 'spot analysis' program (Materials and Methods).
`Optimization of the labeling conditions was performed using
`SstI plasmid probe, a moderately repetitive sequence of 2.5
`kilobase appearing in tandem arrays on human chromosome 19
`and less frequently on chromosome 4 (38). The pUC12 plasmid
`containing 2.5 kb SstI insert (total of 5 kb) was used in the labeling
`reactions.
`One important variable in the enzymatic labeling reactions is
`the relative substitution of dTTP by the dUTP analog. For
`instance, it has been reported that, when incorporating biotin-
`or digoxigenin-labeled nucleotides, complete substitution of dTTP
`by the modified analogs does not give the brightest signal (39,40).
`The situation is different for cyanine labeled dUTPs. Table 1
`shows labeling density (d/kb) as well as hybridization signal
`intensities of nick translated probes using different Cy3-dUTP
`concentrations. It is obvious that the probe is significantly brighter
`when all the dTTP is completely replaced by an equal amount
`of Cy3-dUTP. Intensities of the hybridization signals parallel the
`d/kb measurement, indicating that fluorescence quenching is not
`significant at the higher labeling density. The results also reveal
`that higher Cy3-dUTP concentrations do not improve labeling
`density or hybridization signal-to-background. Lower percent
`substitution results in a much lower labeling density. The sample
`with 25 % substitution did not result in measurable dye absorption
`of the labeled probe. Although the hybridization signals were
`observable, they were too weak for valid quantification.
`It is evident that 100% substitution of dTTP by Cy3-dUTP
`is necessary to obtain the brightest probes. When this condition
`is used, the labeling density is in the range of 20 to 40 d/kb in
`routine experiments. This number is somewhat lower than
`expected, since when 100% substitution is applied, approximately
`25 % of the bases (d/kb - 250) would be labeled if the complete
`template sequence is replicated. However, several studies have
`suggested that modified nucleoside triphosphate analogs are
`nucleoside
`efficiently
`incorporated
`natural
`less
`the
`than
`triphosphate (28, 39, 40). In the T7 RNA polymerase in vitro
`transcription system, truncated transcripts are generated and the
`yield of full length transcript is very low when all the ATP is
`replaced by aminohexyl-ATP, especially with AT-rich template
`(39). It is conceivable that heavy labeling may cause steric
`perturbation at the binding and/or catalytic site of the enzyme
`that would cause the polymerase to fall off the template and stop
`polymerization. For example, in nick translation with modified
`dUTP, nucleotide replacement may cease when several adenines
`are next to each other in the template. Thus, there would be
`regions in the template that are not labeled at all, while the regions
`that are labeled may have close to 25% labeling density. The
`method used to calculate the d/kb only provides an average
`labeling density throughout the complete length of the probe.
`Another important variable in nick translation reaction is DNase
`I concentration or probe length. For most in situ hybridization
`experiments, an average probe length in the range of 150-250
`was found optimal (26, 27). We found in our experiment that
`labeling density was a function of DNase I concentration. The
`
`

`

`3230 Nucleic Acids Research, 1994, Vol. 22, No. 15
`
`1 20
`
`100
`
`80 *
`
`60 X
`
`40 -
`
`20
`
`I1-
`
`AP
`
`.0
`
`0'
`@0
`
`.0
`
`0
`
`2
`
`8
`6
`4
`[Cy3-dUTPI/[TTPI
`
`10
`
`1 2
`
`function of
`yield
`Labeling density and amplification
`a
`Figure 3.
`as
`Cy3-dUTP/dTTP molar ratio in PCR. PCRs were performed and the labeling
`densities as d/kb were measured as described in Materials and Methods.
`
`labeling density increased inversely with the probe length. For
`example, 44 d/kb was obtained if the probe was nicked to an
`average length shorter than 100 bases, whereas 20 d/kb and 12
`d/kb were obtained for probes with average lengths of 150 bases
`and 500 bases respectively. According to this result, shorter
`probes have higher average number of dyes per unit probe and
`therefore might be preferred on the basis of fluorescence signal
`intensity. However, heavily labeled short probes may form
`unstable hybrids due to possible steric perturbation with hydrogen
`bonding. Therefore the hybridization and wash stringency may
`have to be adjusted to obtain the best signal-to-background.
`
`Labeling probes by random priming
`In a random primed labeling reaction, degenerate oligonucleotides
`(6 mer or longer) are used as primers to drive the synthesis of
`uniformly labeled DNA probes by DNA polymerase. For random
`primed labeling, it was also found that complete substitution of
`dTTP with Cy3-dUTP provided the brightest probe (Table 2).
`In this case, only absorption ratio of the dye and the bases instead
`of d/kb was obtained, because the unlabeled template and excess
`oligonucleotide primers were not completely removed and
`contribute significantly to the 260 nm absorption reading. Again,
`spot analysis of the hybridization signals paralleled the absorption
`ratio measurement.
`
`Labeling probes by polymerase chain reaction
`Since the introduction of polymerase chain reaction (PCR)
`technologies, numerous applications have been devised including
`fluorescence labeling of DNA probes. Labeling by PCR offers
`the advantage of simultaneous amplification and labeling of a
`specific probe or a specific region of a probe.
`We have characterized the incorporation of Cy3-dUTP into
`DNA probes via PCR. By using primers labeled at the 5' end
`with CyS, a dye that absorbs and fluoresces at wavelengths
`distinct from Cy3, we were able to accurately measure the d/kb
`of PCR amplified probes. The CyS fluorescence (680nm) or
`absorbance (650nm) signal is proportional to the number of
`amplified sequences created (amplimers) and provides an accurate
`measurement of relative amplification efficiency. At the same
`time, the internal incorporation of Cy3-dUTP was monitored at
`a separate wavelength (550 nm for absorption and 575nm for
`fluorescence). After the purification of PCR labeled probes, the
`
`Figure 4. Multicolor FISH image. Images of four fluorescence signals from a
`human male lymphocyte chromosome spreads after hybridization. The cyanine
`dye labeled probes were preparaed by nick translation. Upper-left: Hoechst
`indicating location of molecular DNA; Upper-right:
`fluorescence signal,
`hybridization signal from Cy3-labeled Chromosome 1 probe; Lower-left:
`hybridization signal from Cy5-labeled SstI probe hybridized to Chromosome 19;
`Lower-right: hybridization signal from FITC-labeled X chromosome probe
`(generous gift from Integrated Genetics).
`
`relative Cy3 to Cy5 fluorescence intensity ratios were measured
`and were converted into Cy3 to CyS molar ratios of the labeled
`probe. These ratios together with the total length of the amplified
`probe (0.9kb) were used to calculate d/kb. Although the spectral
`overlap between Cy3 emission and Cy5 absorption is relatively
`small, energy transfer from Cy3 to Cy5 could occur when Cy3
`molecules are inserted within approximately 5.5 nm of the 5'
`terminal Cy5 label (the distance for 50% energy transfer
`calculated by Dr Reddington, personal communication). A
`distance of 5.4 nm along a double helix corresponds to
`approximately 16 base pairs. Since 18 base primers are situated
`between the 5'- Cy5 energy acceptor and the closest place a Cy3
`is unlikely that energy transfer
`dUTP can be inserted, it
`appreciably causes underestimation of d/kb.
`As an amplification template, a Sp65 plasmid probe against
`a chromosome 1 repetitive sequence was used and the primers
`directed amplification of the 900 bp insert. The amplification was
`confirmed by loading a fraction of each sample on 1 % agarose
`gel. The same volume of material before amplification was loaded
`and nothing was observed on the gel (gel not shown). Because
`of the net negative charges on the dye (Figure 1), labeled probes
`had only slightly retarded mobility on the gel. However, when
`longer linkers were used between dUTP and the fluorochrome,
`the mobility shift was significant (Zhu et al., unpublished results).
`Several thermal stable DNA polymerases were investigated for
`their ability to incorporate cyanine-labeled dUTP. The experiment
`indicated that the extent of label incorporation could increase by
`a factor of two or three when using polymerases other than the
`commonly used Tag DNA polymerase (data not shown). Tli
`(Promega), Pfu (Stratagene), Deep Vent and Vent Exo- (New
`England Biolabs) were among those that incorporated most
`Cy3-dUTP per kb of synthesized DNA under conditions for each
`of the enzymes as recommended by the manufactures.
`
`

`

`The effect of relative amount of Cy3-dUTP to dTTP on the
`PCR labeled product was tested using Tli DNA polymerase. In
`this experiment, dTTP concentration was kept constant, at 30
`iM in final reaction mixture and increasing amounts of
`Cy3-dUTP, ranging from 0 and 300 ,uM final concentration were
`added. A control reaction was performed which contained 60
`liM of dTTP and no Cy3-dUTP. The relative yield of
`amplification in presence of Cy3-dUTP was compared with this
`control sample. Figure 3 shows that the labeling density (d/kb)
`increases almost linearly with the increase in Cy3-dUTP
`concentration, however, at the expense of lower amplification
`efficiency. As expected, it is the molar ratio of Cy3-dUTP to
`dTTP in the reaction mixture that governs the frequency with
`which Cy3-dUTP enters DNA. At higher dTTP concentration,
`higher Cy3-dUTP concentration is required for a specified
`labeling density. When the Cy3-dUTP to dTTP molar ratio is
`in the range of 0.5 to 10, the labeling density varies from 2 to
`20 d/kb with relative yields decreasing from 80% to 15%
`correspondingly. This labeling density with Cy3-dUTP is at the
`lower end of what can be achieved by nick translation method
`(20-40 d/kb). While 100% substitution of dTTP with Cy3-dUTP
`in nick translation provided the best labeling density, complete
`substitution of dTTP in PCR resulted
`in no observable
`amplification by both gel electrophoresis and spectroscopic
`measurements.
`When equal amounts of dTTP and Cy3-dUTP were used in
`the PCR reaction, the labeling density averaged only 3.3 d/kb.
`Assuming equivalent amount of all four bases in the sequence
`and no discrimination between modified and unmodified
`nucleoside triphosphates, the expected replacement would be 125
`d/kb. Thus it is clear that the polymerase discriminates between
`dTTP and the modified analog with a preference for the dTTP.
`is known that many DNA polymerases show little
`It
`discrimination between dUTP and dTTP in DNA synthesis under
`most circumstances (41). Therefore the discrimination is mainly
`caused by the attachment of the linker molecule and/or the
`fluorophore on the base. This is not surprising given the precise
`binding site of nucleoside triphosphates on DNA polymerase (42).
`However, the labeling density increases almost linearly with
`Cy3-dUTP concentration, demonstrating that the analog is,
`nevertheless, a good substrate for Tli DNA polymerase and other
`DNA polymerases. When the same experiment was carried out
`with Taq DNA polymerase, similar result and plot were obtained
`(data not shown).
`The fact that the amplification drops as the labeling density
`that an increasing number of labels
`increases
`suggests
`incorporated into the template inhibits DNA synthesis in
`subsequent cycles. Perhaps, heavily labeled DNA template
`strands do not provide the appropriate recognition sites required
`for the binding of DNA polymerase, especially where several
`modified nucleotides are next to each other. By contrast, nick
`translation and random priming, which translate from unmodified
`template, produce labeled probe efficiently even with 100%
`cyanine-labeled dUTP.
`Our results demonstrate that there is a trade-off between the
`labeling density and the amplification yield in PCR labeling
`reactions. For producing probes for chromosome painting or
`repetitive sequences, relatively lower labeling density may be
`sufficient. However, for single copy gene detection where
`sensitivity is extremely important, higher labeling density may
`be necessary and the yield may have to be sacrificed.
`
`Nucleic Acids Research, 1994, Vol. 22, No. 15 3231
`
`Four color FISH with three directly labeled probes
`Figure 4 demonstrates 4-color hybridization images obtained with
`human lymphocytes and probes that were directly labeled with
`Cy3-, Cy5- and FITC-dUTP.
`
`CONCLUSION
`Two cyanine-dUTP analogs, Cy3-dUTP and Cy5-dUTP have
`been synthesized and proven to be efficient substrates for DNA
`polymerases. An important parameter, cyanine-dUTP to dTTP
`molar ratio, has been optimized in standard nick translation and
`random primed labeling reactions. In both cases, complete
`substitution of dTTP with cyanine-dUTP yields the brightest
`probes and hybridization signals. In nick translation, labeling
`density of about 20-40 d/kb can be routinely obtained under
`the optimal labeling conditions. In polymerase chain reaction,
`however, complete substitution of dTTP does not result in
`detectable amount of full length DNA. The labeling density
`