IQ-D 1992 Oxford University Press
`
`Nucleic Acids Research, Vol. 20, No. 18 4831-4837
`
`Detection of single base differences using biotinylated
`nucleotides with very long linker arms
`
`Kenneth J.Livak, Frank W.Hobbs and Robert J.Zagursky
`The Du Pont Merck Pharmaceutical Company, Experimental Station, PO Box 80328, Wilmington,
`DE 19880-0328, USA
`
`Received June 8, 1992; Revised and Accepted August 12, 1992
`
`ABSTRACT
`A simple primer extension method for detecting
`nucleotide differences is based on the substitution of
`mobility-shifting analogs for natural nucleotides (1).
`This technique can detect any single-base difference
`that might occur including previously unknown
`mutations or polymorphisms. Two technical limitations
`of the original procedure have now been addressed.
`First, switching to Thermococcus litoralis DNA
`polymerase has eliminated variability believed to be
`due to the addition of an extra, non-templated base to
`the 3' end of DNA by Taq DNA polymerase. Second,
`with the analogs used in the original study, the mobility
`shift induced by a single base change can usually be
`resolved only in DNA segments 200 nt or smaller. This
`size limitation has been overcome by synthesizing
`biotinylated nucleotides with extraordinarily long linker
`arms (36 atom backbone). Using these new analogs and
`conventional sequencing gels (0.4 mm thick),
`mutations in the human ,B-hexosaminidase a and
`CYP2D6 genes have been detected in DNA segments
`up to 300 nt in length. By using very thin (0.15 mm)
`gels, single-base polymorphisms in the human APOE
`gene have been detected in 500-nt segments.
`
`INTRODUCTION
`The basis of molecular genetics is the identification and
`characterization of mutations. Thus, many techniques have been
`developed to compare homologous segments of DNA to
`determine if the segments are identical or different. We have
`described a simple primer extension assay that distinguishes
`homologous DNA segments differing by as little as a single
`nucleotide (1). DNA strands are synthesized with one of the four
`natural
`nucleotides
`replaced with an analog that
`retards
`electrophoretic
`mobility
`('mobility-shifting'
`analog). For
`example, incorporation of biotin-l l-dUTP, a commercially
`available analog of TTP, into a DNA strand causes a one
`nucleotide mobility shift when the DNA is fractionated on a
`sequencing gel. This means that, for each biotinylated residue
`incorporated, the biotin-containing DNA strand migrates at a
`position approximately one nucleotide slower than is expected
`based on the length of the DNA strand. DNAs that are the same
`length but differ in the number of analog molecules per strand
`migrate differently on a sequencing gel and thus are distinguished.
`
`The assay is sensitive-by using two analogs (in separate
`reactions) and analyzing both strands, any single nucleotide
`change can be detected. Furthermore, the assay can identify
`previously undetected mutations. Another advantage is that the
`method uses conventional molecular biology techniques. The
`primer extension reactions and gels are basically the same as those
`used in standard dideoxynucleotide sequencing.
`The ability to use mobility-shifting nucleotide analogs to detect
`single base differences was first demonstrated by examining a
`segment of the human insulin receptor gene (1). When the method
`was used to examine a number of additional DNA segments, two
`limitations of the technique became apparent. First, using the
`nucleotide analogs described in Kornher and Livak (1), the
`method best distinguishes single base differences when the DNA
`segments examined are 200 nt or less. Increasing the size limit
`would greatly increase the utility of this technique. Second, there
`was run-to-run variation in the appearance of shadow bands
`migrating just faster than the main bands. At its worst, the shadow
`bands made it difficult to distinguish homozygotes, which should
`exhibit one prominent band, from heterozygotes, for which two
`prominent bands are expected. As described in this report, these
`two problems were overcome by using a different enzyme to
`perform the primer extension reactions, by synthesizing new
`nucleotide analogs with extraordinarily long linker arms, and by
`using very thin polyacrylamide gels to reduce electrophoresis
`time.
`
`MATERIALS AND METHODS
`Synthesis of nucleotide analogs
`Biotin-36dUTP (lb). Biotin-XX-NHS ester (22.7 mg, 40 jLmol,
`Clonetech # 5002) in 200 Al of dimethylformamide was heated
`with occasional vortexing at 500C until all of the solid dissolved.
`A solution of the triethylammonium salt of dUTP-36 (la, 20
`Amol, 242 ODU at 290 nm; 2) and N-methylmorpholine (22 Al,
`200 timol) in water (133 I1) and dimethylformamide (67 jAd) was
`added. The reaction was allowed to proceed with occasional
`vortexing at 250C for 4 h. The reaction was filtered, diluted with
`5 ml of 0.1 M TEAB (aqueous triethylammonium bicarbonate),
`and loaded onto a DEAE-A25 -120 Sephadex column (2.6 x40
`cm, prewashed with 1.0 M TEAB and equilibrated with 3 column
`volumes of 0.1 M TEAB). The column was eluted with a linear
`gradient from 0.1 M (200 ml) to 1.0 M (200 ml) TEAB at a
`flow of 2 ml/min while monitoring UV absorption at 260 nm.
`
`Illumina Ex. 1053
`IPR Petition - USP 10,435,742
`
`

`

`4832 Nucleic Acids Research, Vol. 20, No. 18
`
`(CO2 was continuously bubbled into the TEAB.) The major
`peak eluting from 250 to 290 ml was collected, concentrated and
`co-evaporated twice with ethanol to afford 242 ODU (77%) of
`white foam. The material was further purified by preparative
`HPLC (1 x25 cm ODS column, 5-35% acetonitrile in 0.1 M
`TEAB (pH = 7) gradient over 30 min with a flow rate of 3
`ml/min). The major peak was collected, co-evaporated and
`lyophilized. The product was dissolved in water (1 ml) containing
`triethylamine (5 A1) and eluted through a column (1 ml) of the
`sodium form of Dowex AG 50-X8 with water. Fractions (ca.
`I ml each) containing UV absorbing material were combined and
`lyophilized to afford biotin-36-dUTP, lb, (153 ODU at 290 nm,
`12.6 pmol, 63%) as a white foam. UV (water): 212, 235 and
`290 nm. (The extinction coefficients were 16,000, 12,400 and
`12,100 based on an extinction coefficient of 12,100 at 290 nm
`previously established for this chromophore.) 31P-NMR: 6 =
`-9.2, -9.7, and -21.4. 1H-NMR: 6 = 8.15 (s, 1H, H6),
`6.27 (t, 1H, Hi'), 4.65 (m, 2H, H3' and H7"), 4.52 (s, 2H,
`CCCH20), 4.45 (dd, 1H, H8"), 4.20 (m, 3H, H4' and H5'),
`3.6-3.9 (m, 22 H, OCH2CH20), 3.42 (t, 2H, OCH2CH2N),
`3.35 (dd, 1H, H6"), 3.20 (m, 4H, CH2N), 3.00 (dd, 1H,
`H9"a), 2.77 (d, 1H, H9"b), 2.2-2.5 (m, 8H, H2' and
`CH2C=O) and 1.3-1.8 (m, 18H, CH2). (Carbons on the biotin
`subunit of the molecule were assigned double-prime numbers
`beginning with the carboxylic acid carbon.)
`
`Biotin-36-dC7P (14). The triethylammonium salt of dCTP-36 (ic,
`20 pmol, 186 ODU at 294 nm; 2) was biotinylated and purified
`as described above to afford biotin-36-dCTP, ld, (121 ODU at
`294 nm, 13.4 gmol, 65%) as a white foam. UV (water): 213,
`238, and 294 nm. (The extinction coefficients are 28,000, 15,000
`and 9,300 based on an extinction coefficient of 9,300 at 294 nm
`previously established for this chromophore.) 31P-NMR: 6 =
`-5.60 (br s, IP), -9.51 (d, 1P) and -20.63 (t, IP). 1H-NMR:
`6 = 8.22 (s, 1H, H6), 6.28 (t, 1H, HI'), 4.65 (m, 2H, H3'
`and H7"), 4.52 (s, 2H, CCCH2O), 4.45 (dd, 1H, H8"), 4.24
`(m, 3H, H4' and H5'), 3.6-3.9 (m, 22 H, OCH2CH2O), 3.42
`(t, 2H, OCH2C 2N), 3.35 (dd, 1H, H6"), 3.20 (m, 4H,
`CH2N), 3.03 (dd, 1H, H9"a), 2.81 (d, 1H, H9"b), 2.2-2.5
`(m, 8H, H2' and CH2C=O), 1.94 (s, acetate impurity), and
`1.3-1.8 (m, 18H, CH2).
`Biotin-36-dc7ATP (2b). The triethylammonium salt of
`dc7ATP-36 (2a, 10 ,mol, 127 ODU at 280 nm; 2) was
`biotinylated and purified
`as described above to
`afford
`biotin-36-dc7ATP, 2b, (25.4 ODU at 280 nm, 2.0,mol, 20%)
`as a white foam. The following changes were made: a) The
`reaction was run on one-half the above scale. b) A 1 x 19 cm
`ion exchange column was used and elution was with 150 ml of
`each buffer. c) From preparative HPLC, two equal fractions were
`collected and the faster eluting fraction was repurified by
`preparative HPLC. UV (water): 214, 239 and 280 nm. (The
`extinction coefficients were 20,300, 13,100 and 12,700 based
`on an extinction coefficient of 12,700 at 280 nm previously
`established for this chromophore.)
`PCR amplifications
`The buffers used in PCR amplifications were Taq buffer (10 mM
`Tris-HC1, pH 8.3; 50 mM KCl; 5 p4M EDTA; 0.01 % gelatin)
`or Stoffel buffer (10 mM Tris-HCl, pH 8.3; 10 mM KCl). All
`reactions contained 100 ng human genomic DNA and 200 pM
`
`each dNTP, and were performed in 50 Il reactions under mineral
`oil in a DNA Thermal Cycler (Perkin-Elmer Cetus).
`The primers used to amplify 113 bp and 498 bp segments of
`the (3-hexosaminidase ca gene were: TSl-GTGTGGCGAGAG-
`GATATTCCAGT (exon 11 primer from Myerowitz and
`Costigan
`(3)); TS6-TCCTGCTCTCAGGCCCAACCCTC
`(intron 12 primer PCA2 from Myerowitz (4)); TS1O-AAGCT-
`TCACTCTGAGCATAACAAG (intron 7 primer from Navon
`and Proia (5)); and TS14-TATCCGTGTGCTTGCAGAGUTTG
`(nucleotides 749-771 in Myerowitz et al
`(6)). Reactions
`contained Taq buffer, 2.5 mM MgCl2, 200 nM each TS primer,
`plus 1 unit AmpliTaq polymerase (Perkin-Elmer Cetus) and were
`subjected to the following regimen: 92°C (3 min); 30 cycles of
`ramp to 92°C (20 sec), 92°C (40 sec), ramp to 68°C (30 sec),
`68°C (1 min, increase by 3 sec each cycle); 72°C (5 min).
`The primers used to amplify a 297 bp segment of the CYP2D6
`gene were: CYl l-CCGCCTTCGCCAACCACT (nucleotides
`1826-1843 in Kimura et al (7)); and CY2-GAGACTCCTCG-
`GTCTCTC (primer 2 from Heim and Meyer (8)). Reactions
`contained Stoffel buffer, 5 mM MgCl2, 1 pM each CY primer,
`plus S units Stoffel fragment (Perkin-Elmer Cetus) and were
`subjected to the following regimen: 98°C (3 min); 5 cycles of
`98°C (1 min), 62°C (1 min), 72°C (1 min); 30 cycles of 95°C
`(1 min), 62°C (1 min), 72°C (1 min); 72°C (5 min).
`The primers used to amplify a 538 bp segment of the APOE
`gene were: ApoEl 1-GGCACGGCTGTCCAAGG (adapted from
`primer F6 of Emi et al (9)); and ApoE24-CACCAGGGGCT-
`CGAACC (nucleotides 4218-4202 in Paik et al (10)). Reactions
`contained Taq buffer, 1.5 mM MgCl2, 10% DMSO, 1 pM each
`ApoE primer, plus 4 units AmpliTaq polymerase and were
`subjected to the following regimen: 95°C (3 min); 35 cycles of
`95°C (1 min), 58°C (1 min), 70°C (1 min); 70°C (5 min).
`After amplification, each DNA sample was transferred to a
`fresh tube, treated with 1 unit calf intestinal alkaline phosphatase
`(Promega) at 37°C for 15 min to hydrolyze residual dNTPs, and
`precipitated by adding 50 Ml 5 M NH4-acetate and 100 ,dl
`isopropanol. After sitting at room temperature for 10 min, the
`DNA is collected by a 10 min centrifugation, rinsed with 200
`itl cold 70% ethanol, dried, dissolved in 20p1 TE-8.0 (10 mM
`Tris-HCl, pH 8.0; 1 mM EDTA), incubated at 70°C for 15 min,
`and stored at 4°C. This procedure eliminates the dNTPs and most
`of the primers from the PCR reaction.
`Primer extensions with nucleotide analogs
`For n samples, a Master Mix was prepared containing (n+ 1) x 1
`p1 H20; (n+1)x0.5 il 10xhi buffer [200 mM Tris-HCl, pH
`8.8; 100 mM (NH4)2SO4; 100 mM KCl; 20 mM MgSO4; 1%
`Triton X-100]; (n+1)x0.5 pl DMSO; (n+1)xO.1 Al
`Thermococcus litoralis (7li) DNA polymerase (Vent DNA
`polymerase, 1 U/pl; New England BioLabs); and (n+ 1) xO.9
`A labeled primer. For each primer extension reaction, 1 pul analog
`mix (1 mM biotin-36-dATP, -dCTP, or -dUTP plus 500 pM of
`each of the remaining three dNTPs) and 1 pl PCR-amplified
`template DNA were put in an Eppendorf tube. After adding 3
`pl Master Mix to each reaction, the tubes were incubated in a
`boiling water bath for 1 min, at room temperature for 1 min,
`and at 75°C for 3 min. For ApoE24 primer extensions this cycle
`was repeated four additional times. Reactions were terminated
`by adding 10lO
`formamide-dye solution (95% formamide; 12.5
`mM EDTA; 0.3% bromophenol blue; 0.3% xylene cyanol) and
`incubating in a boiling water bath for 3 min.
`
`

`

`\~~~~~~N3
`
`0 H
`
`~~~~0
`
`0
`
`X = O, R = H2
`la
`lb X = 0, R = 3.
`lc X= NH2, R= H2
`ld
`X = NH2, R = 3.
`2a X=O,R=H22.
`2b X = O, R = 3.
`
`d UTP-22.
`Biotin-36-dUTP.
`dCTP-22.
`Biotin-36-dCTP.
`dc7ATP-22.
`Biotin-36-dc7ATP.
`
`Figure 1. Biotinylated nucleotides with very long linker arms. Structures are shown
`for the intermediates and final compounds used in the mobility shift assay.
`
`Nucleic Acids Research, Vol. 20, No. 18 4833
`
`shadow bands were the result of failing to add the non-templated
`base. Variation was observed because the extent of extra
`nucleotide
`addition depends on enzyme lot, enzyme
`concentration, nucleotide concentration, exact reaction conditions,
`and possibly sequence context. This problem was overcome by
`switching to a different DNA polymerase. In order to retain the
`convenience and sensitivity of using a thermostable enzyme, the
`polymerase chosen was Tli DNA polymerase.
`Tli DNA
`polymerase possesses 3' exonuclease activity and thus does not
`leave an extra, non-templated nucleotide on DNA fragments. The
`use of Tli DNA polymerase greatly reduces problems in
`interpretation due to shadow bands.
`
`Nucleotide analogs with long linker arms
`The switch to Tli DNA polymerase led to another fortunate result.
`A series of nucleotide analogs with longer linker arms had been
`synthesized and tested for acceptance by Taq DNA polymerase.
`Figure 1 shows three analogs synthesized with a linker of 36
`atoms, the longest linker so far constructed. The analogs of
`Figure 1 were not accepted that well by Taq DNA polymerase,
`but were found to be fairly good substrates for Tli DNA
`polymerase. Using these analogs with Tli DNA polymerase
`dramatically improved the results obtained with the mobility-shift
`technique for polymorphism detection.
`
`0o'
`
`°>
`
`o"v
`
`°""
`
`0
`
`t°""NHR
`
`O N
`
`O
`
`3 HO9P30-
`HO~
`
`la-id
`
`2a-2b
`
`The labeled primers used to analyze the Tay Sachs mutations
`in the f3-hexosaminidase a gene were: TS15-TCTGGTCCCAG-
`ACATCATTC (intron 7 sequence from footnote 31 in Navon
`(5)); TS16-CCTTCCAGTCAGGGCCATA
`and Proia
`(nucleotides 1297-1279 in Myerowitz et
`(6)); and
`al.
`TS17-TTGGTGGAGAGGCITTGTATG (nucleotides 1359-1378 in
`Myerowitz et al. (6)). These TS primers were synthesized with
`a succinyl fluoroscein dye (SF-505; 11) attached to their 5' ends
`(12). The final concentrations of the primers in the extension
`reaction were 100 nM TS15, 300 nM TS16, and 100 nM TS17.
`Reactions with the fluorescent primers were terminated by the
`addition of 10 1d GENESIS loading solution (Du Pont) rather
`than the formamide-dye solution. The labeled primers used to
`analyze the CYP2D6 and APOE genes were: CY2 (PCR primer);
`ApoE24 (PCR primer); and ApoEl3-GCGCGGACATGGAG-
`GAC (nucleotides 3725 -3741 in Paik et al. (10)). These primers
`were labeled with 32P by mixing 7 yl (70 ILCi) -y-32P-ATP, 1
`t1l
`lOxkinase buffer (500 mM Tris-HCl, pH 7.5; 100 mM MgCl2;
`50 mM dithiothreitol; 1 mM spermidine), 1 j,l 10 ttM primer,
`1 4d (10 units) polynucleotide kinase (New England BioLabs)
`and incubating at 37°C for 10 min. The final concentration of
`each primer in an extension reaction was 18 nM CY2, 180 nM
`ApoE24, or 36 nM ApoEI3.
`Gel electrophoresis
`The buffer used to run the sequencing gels was a modified form
`of TBE designated MTB (IOxMTB = 163.5 g Tris base, 27.8
`g boric acid, 9.3 g Na2EDTA 2H20 per 1 liter). Except for the
`C analog gel in Figure 4, electrophoresis was performed with
`1 x MTB in both the gel and running buffer. Gels were 6% 19: 1
`acrylamide/bis-acrylamide and contained either 7 M or 8 M urea
`(7 M urea seemed to give slightly better separation). Very thin
`gels were approximately 0.15 mm thick, 32 cm long, and 20
`cm wide. They were poured from the bottom by capillary action
`using adhesive tape (Serva cat. #42927) as side spacers and 0.2
`mm combs (IBI) wedged in the top. At first, the very thin gels
`were transferred to 3MM paper (Whatman) after electrophoresis
`and dried down. At times, though, the bands were not perfectly
`straight. The waviness observed in bands can be seen in the 0.15
`mm gel of Figure 3. Thus, for the analysis of 500 nt primer
`extension products in Figure 4, the gels were covalently attached
`to the glass by treating the larger plate with -y-methacryloxy-
`propyl-trimethoxy silane (Sigma) as described by Garoff and
`Ansorge (13). After electrophoresis, the gel was covered with
`Reynolds 904 plastic film and dried directly onto the glass plate
`under vacuum.
`
`RESULTS
`DNA polymerase used in the assay
`The mobility-shift method for polymorphism detection was
`originally developed using Taq DNA polymerase to incorporate
`nucleotide analogs as part of a primer extension reaction. When
`these primer extension products were examined on a denaturing
`polyacrylamide gels, shadow bands were sometimes seen
`migrating just faster than the main bands. The appearance of these
`extra bands was finally traced to the ability of Taq DNA
`polymerase to add a single, non-templated nucleotide, usually
`A, to the 3' end of blunt-end DNA fragments (14). In fact, the
`main bands observed on gels contained this extra base; whereas,
`
`

`

`4834 Nucleic Acids Research, Vol. 20, No. 18
`
`Tay Sachs mutations
`Long linker analogs were first used in a nonradioactive assay
`for detecting mutations implicated in Tay Sachs disease.
`Figure 2A shows the segments of the /3-hexosaminidase ca gene
`analyzed and the fluorescent primers that were used to assay for
`each specific mutation. The 113-bp and 498-bp segments were
`amplified from human genomic DNA in a single reaction. After
`elimination of the dNTPs used in PCR, the amplified segments
`were used as templates in a primer extension reaction containing
`the three fluorescent primers, biotin-36-dCTP, dATP, dGTP,
`UTP, and Tli DNA polymerase. Figure 2B shows fluorescent
`primer extension products electrophoresed through a denaturing
`polyacrylamide gel and detected using the GENESIS 2000
`system. The normal individual shows a single peak for each exon
`because this person is homozygous normal at all three mutation
`sites. Heterozygotes are detected by the appearance of two peaks.
`For exon 7, the peak for the mutant allele runs faster than the
`normal peak because the mutant primer extension product has
`one less biotinylated C residue. The mutant peak for exon 12
`runs slower because this product has one additional C analog;
`and the mutant peak for exon 11 runs slower because this product
`has four additional non-analog nucleotides. Figure 2B clearly
`shows that heterozygotes can be detected for each of the three
`exons. Furthermore, this analysis demonstrates that more than
`one mutation can be assayed in a single sample by using multiplex
`PCR and multiple primers in the extension reaction. The only
`requirement is that the primers used produce extension products
`migrating at different points in the gel. The results of Figure 2B
`also show that the use of fluorescent primers provides a
`convenient method for achieving nonradioactive detection in the
`mobility shift assay.
`CYP2D6 mutation
`The use of the biotin-36 analogs also makes it possible to analyze
`primer extension products longer than 200 nt. Figure 3 presents
`an analysis using extension products of approximately 300 nt
`performed on a segment of the human CYP2D6 gene. CYP2D6
`encodes the cytochrome P450 debrisoquine 4-hydroxylase.
`Mutations in this gene cause poor metabolism of debrisoquine
`and over 25 other drugs, a phenotype inherited as an autosomal
`recessive trait by 5-10% of Caucasians. One of the most
`common mutations that inactivates the CYP2D6 gene is a G-to-
`A change that eliminates the splice acceptor site at the intron
`3/exon 4 boundary (15). As diagrammed in Figure 3, primer
`extensions across this mutation site were performed using either
`a C or a T analog. Detection of the CYP2D6 mutation is shown
`by the three individuals who exhibit two bands. These people
`are heterozygous for the mutation and thus are carriers for the
`'poor metabolizer' phenotype. The other three individuals have
`only one of the two bands, and this is the slower band with the
`C analog and the faster band with the T analog. This
`electrophoretic behavior indicates there is a C at the mutation
`site and therefore the individuals with a single band are
`homozygous for the nonnal allele. The key point is that single
`base differences were detected in primer extension products of
`297 nt. This is nearly 100 nt longer than the products that could
`be analyzed using analogs with shorter linker arms. In analyzing
`these longer extension products,
`the time required
`for
`electrophoresis becomes a significant factor. Figure 3 shows
`detection of single base differences when the products were
`analyzed on a conventional 0.4 mm sequencing gel. This required
`a 7 h electrophoresis run. By using a very thin (0.15 mm) gel,
`
`A
`
`113 bp
`
`498 bp
`
`C
`t.....
`
`80 nt_,
`
`-C-
`
`T
`
`exon 7
`
`B
`
`AK
`
`J
`
`exon 7
`80 nt
`15 or 16 C's
`
`exon 12
`111 nt
`15 or 16 C's
`
`A-TG
`127 nt _,
`
`ATAG
`
`exon 11
`
`exon 12
`
`normal
`
`compound heterozygote
`exons 7 & 11
`
`compound heterozygote
`exons 7 & 12
`
`AK
`
`exon I 1
`127 or 131 nt
`
`Figure 2. Detection of Tay Sachs mutations using fluorescence-labeled primers.
`A: Scheme for amplifying two segments (113 bp and 498 bp) of the j3-
`hexosaminidase a gene and detecting three Tay Sachs mutations (3-5) using
`the mobility shifl assay. B: As described in Materials and Methods, DNA segments
`containing the Tay Sachs mutation sites were amplified from three individuals-one
`normal and two compound heterozygotes. These PCR-amplified segments were
`used as templates for the primer extension reaction diagrammed in A. Primer
`extension products (80 nt, 127 nt, and 111 nt) incorporating biotin-36-dCTP were
`synthesized in a single reaction using the fluorescence-tagged (F) primers TS15
`(exon 7), TS16 (exon 11), and TS17 (exon 12). The extension products were
`electrophoresed on a 6% acrylamide/8 M urea gel at 24 watts in the GENESIS
`2000 DNA Analysis System (Du Pont) following the manufacturer's protocol.
`Output from the GENESIS 2000 is shown as fluorescence intensity plotted versus
`electrophoresis time. As indicated, the exon 7 and exon 11 products each contained
`15 or 16 biotinylated C residues.
`
`approximately the same separation of bands was achieved in 4
`h. Thus, the use of 0.15 mm gels enhances the convenience of
`the mobility-shift technique. One final point about the CYP2D6
`results is that the CY2 primer was used both to amplify the
`CYP2D6 segment and as the labeled primer in the extension
`reactions. This demonstrates that a nested primer is not required
`for mobility-shift analysis as long as the PCR products do not
`have excessive background.
`Apo E polymorphisms
`The APOE gene has three common alleles e2, e3, and c4, which
`were originally detected by isoelectric focusing of the apoE
`protein (reviewed in 16). Sequencing has demonstrated that the
`
`

`

`Nucleic Acids Research, Vol. 20, No. 18 4835
`
`AA
`
`Hunimn CYI12D16 Gene
`
`AG
`
`Cy,
`
`P
`
`C 97 III
`<1 6L"/C
`11
`
`milronl 3
`
`exllo z.
`
`biotin-36-dClTP
`
`hiotin-36-dUTll
`
`hiotin-36-d(TP
`
`hibiotin-36-dUT}p
`
`91
`
`.m as ur. -4
`
`- mm
`
`do. Ym
`
`-
`
`52
`SlF
`
`0.4 imim gel
`
`0.15 nim gel
`
`Figure 3. Detection of a mutation in the human CYP2D6 gene using primer extension products of about 300 nt. Primers CY 11 and CY2 were used to amplify
`a 297-nt segment of the CYP2D6 gene from the parents and 4 children of CEPH pedigree 1333. These segments were used as templates in primer extension reactions
`containing 2P-labeled CY2 and either biotin-36-dCTP or biotin-36-dUTP as the mobility-shifting analog. The left panel shows the products electrophoresed on a
`6% acrylamide/8 M urea 0.4 mmX40 cm gel run at 90 watts for 7 h. The right panel shows the products electrophoresed on a 6% acrylamide/7 M urea 0.15
`mm x 32 cm gel run at 40 watts for 4 h. After electrophoresis, the gels were transferred to 3MM paper (Whatman), dried under vacuum, and exposed to X-ray
`film. The numbers to the left show the number of modified dC (91 or 92) or dU (51 or 52) residues in the bands. All extension products are 297 nt in length,
`but the C analog products run slower because they contain approximately 40 additional biotinylated residues.
`
`E 4
`t3
`£2
`
`G
`A
`A
`< I
`codon
`112
`
`(;
`G
`A
`
`codonlo
`58
`
`_
`
`538
`
`l2p
`
`C,4
`t 3
`£ 2
`
`3'21
`
`(
`I
`Tl
`
`C
`C
`T
`
`codoll
`112
`
`co(tit1
`158
`
`494 nt
`
`cl
`
`Sc
`
`C.
`
`-t
`
`-t
`
`Cl
`
`*
`
`0
`
`t
`
`-r
`
`e 1
`
`rt,
`
`1-1
`
`r-.
`
`W
`
`Cl
`
`X
`
`w
`
`m.
`w
`
`co
`
`t
`
`55
`
`w co
`
`co w tw w
`
`54-,
`
`5 2
`
`low
`
`_M
`
`X ~
`
`Cr2
`
`Cr2
`
`.~
`
`16 ..
`1 6 2 -wlw_
`
`11
`
`Xr
`
`-t
`
`CrW
`
`biotin-36-dAITP
`
`biotin-36 dUIT
`
`biotin-36-dCTP
`
`Figure 4. Detection of polymorphisms in the human APOE gene using primer extension products of about 500 nt. Out of a screen of APOE genotypes, an individual
`was selected to represent each of the six possible genotypes. Primers ApoEl 1 and ApoE24 were used to amplify a 538-nt segment of the APOE gene from these
`six individuals. Primer extensions incorporating biotin-36-dATP, -dUTP, or -dCTP were performed across the polymorphic sites as diagrammed. The A and T analog
`products were electrophoresed on 6% acrylamide/7 M urea/! x MTB 0.15 mm x 32 cm gels using 1 x MTB as running buffer. The gels were run at 40 watts for
`6 h, then dried onto the larger glass plate and exposed to X-ray film. The C analog products were electrophoresed on a 5 % acrylamide/7 M urea/i1.2 xMTB 0.15
`mm x 32 cm gel using 0.6 x MTB as running buffer. The gel was run at 40 watts for 7 h, then dried onto the larger glass plate and exposed to a storage phosphor
`screen. The gel image was generated using the Phosphorlmager Model 400E and ImageQuant software v. 3.15 (Molecular Dynamics). The numbers to the left
`of each gel indicate the number of analog residues in each of the extension products.
`
`three alleles are due to single base differences in the codons for
`amino acids 112 and 158 (8). Figure 4 shows detection of these
`polymorphisms using primer extension products approximately
`500 nt in length. Synthesizing the bottom strand with an A analog
`or the top strand with a T analog, it is possible to distinguish
`all six genotypes for this three allele system. When the C analog
`is used in the primer extensions, however, it has not been possible
`to resolve distinct bands. Figure 4 shows the best separation we
`have been able to achieve so far with the C analog. The reason
`A and T work, but not C, is that the 538 bp amplified segment
`of APOE is 75 % GC. For the c3 allele, the A analog extension
`
`product contains 54 biotinylated A's out of 521 nucleotides
`incorporated (10.3%). For the T analog the ratio is 53/477
`(11.1 %), and for the C analog the ratio is 162/477 (34.0%). For
`the C analog products, the heterozygote bands are noticeably
`broader than for the homozygote bands, indicating that the
`different species are starting to be resolved. Thus, it seems that
`we are obtaining full length products, but have reached the limits
`of resolution for this particular gel system. The analysis of the
`C analog products does show that it is possible to reduce
`electrophoresis time even further by lowering the ionic strength
`of the running buffer.
`
`

`

`4836 Nucleic Acids Research, Vol. 20, No. 18
`
`DISCUSSION
`
`This mobility shift assay requires enzymatic preparation of a
`primer extension product which is: a) precisely full length; and
`b) fully substituted at every expected site with the mobility-shifting
`nucleotide analog. The success of the assay therefore depends
`critically on the mobility-shifting analog and on how efficiently
`it is incorporated by the DNA polymerase being used. If the
`mobility-shifting
`analog
`is
`incorporated
`slowly,
`too
`misincorporation of the other three unmodified dNTPs will
`produce a mixture of partially substituted primer extension
`products. If a site on the template strand requires adjacent
`incorporation of too many mobility-shifting
`analogs,
`the
`polymerase may not be able to extend through this site. If the
`analog
`impurities which are
`mobility-shifting
`contains
`incorporated, the primer extension product probably will not be
`electrophoretically homogeneous. An early version of this assay
`(1) used commercially available biotin-l 1-dUTP and Taq DNA
`polymerase. For unknown reasons, commercially available
`biotin-l 1-dCTP gave much poorer results than biotin- I -dUTP.
`An alkynyl analog of biotin-1 1-dCTP was prepared and found
`to give superior results. Using these biotinylated nucleotides with
`11-atom linker arms, it was difficult to resolve single base
`differences in primer extension products longer than 200 nt.
`Therefore, the use of other analogs was explored in order to
`increase the utility of the mobility shift assay.
`Nucleotide analogs which have large groups linked through
`the 5-position of pyrimidines or the 7-position of 7-deazapurines
`can be surprisingly good substrates for certain DNA polymerases
`(11; 17-19). A versatile synthetic route to a family of such
`nucleotides has been developed (2,20). These nucleotide analogs
`have a flexible hydrophilic polyethylene glycol linker attached
`to the base via an acetylenic bridge. A variety of groups have
`been attached to the terminal amino group of these nucleotides
`for testing in the mobility shift assay. In general, most of these
`nucleotide analogs were incorporated into full length primer
`extension products which appeared to be homogeneous. Although
`these properties fulfill the above criteria for a successful mobility-
`shift assay, the modified primer extension products unfortunately
`gave significantly broader bands than normal nucleic acids upon
`electrophoretic analysis. Although these broader bands could be
`due to the presence of minor impurities or other problems, we
`assume that the broader bands are an intrinsic property of these
`highly modified nucleic acids. Whatever the reason, in practice,
`nucleotides derivatized with biotin gave products with tighter
`bands than those with fluorescent dyes (21), a simple acetyl
`group, or no group on the amine at all. The relationship between
`band tightness and the structure of the mobility-shifting group
`is currently unclear, but the following structural properties might
`be expected to favor tighter bands: a) minimal hydrophobicity,
`so that multiple conformers cannot be stabilized by hydrophobic
`interactions; b) some rigidity, to minimize the number of
`conformations the mobility-shifting group can assume. (Rigidity
`too near the nucleoside, however, could prevent the mobility-
`shifting group from adopting a conformation which is acceptable
`to the polymerase.) This analysis suggests that biotin itself has
`no unique features and that superior mobility-shifting groups
`might be discovered.
`Detection of single base differences requires electrophoretic
`resolution of primer extension products containing N and N +1
`mobility-shifting nucleotide analogs. In the absence of increased
`band broadening, a larger mobility-shifting group should increase
`
`the separation of bands and permit resolution of larger primer
`extension products. In an attempt to maximize mobility shifting
`and/or define an upper size limit for acceptance of mobility-
`shifting groups by DNA polymerases, biotin with a 14-atom
`linker was attached to alkynylamino nucleotides with a 22-atom
`linker derived from hexaethylene glycol (la, lc, and 2a in
`Fig. 1). The resulting biotin-36 dNTPs (lb, Id, and 2b in Fig. 1)
`were incorporated by Taq DNA polymerase, but a significant
`fraction of the primer extension products were not full length.
`When 77i DNA polymerase was tried, however, predominantly
`full length products were observed. These biotin-36 analogs have
`the largest mobility-shifting group (MW 771) tried so far. The
`acceptance of the biotin-36 analogs by 7Ti DNA polymerase is
`especially illustrated by the following examples. In the analysis
`of the APOE segment, up to 163 biotin-36-dC residues are
`incorporated with very little premature termination observed. The
`analysis of the GC-rich CYP2D6 segment required that 5
`biotin-36-dC residues be incorporated in a row. In the analysis
`of a 288-nt AT-rich segment, 7 biotin-36-dU residues were
`incorporated in a row with little apparent stalling (data not
`shown). This ability to generate full length products is crucial
`to the use of the mobility shift technique.
`The 36-atom linker arm causes a much greater shift per residue
`making it possible to detect single base differences in products
`of 300-500 nt. However, this greater electrophoretic retardation
`also causes a problem. DNA strands containing a substantial
`number of biotin-36 residues are electrophoresing as such high
`molecular weight species that we are reaching the limits of
`resolution for a conventional sequencing gel. Thus, in the
`300-400 nt range, it should be possible to analyze most DNA
`segments. In the 400-500 nt range, however, the ability to
`resolve distinct bands on a gel will depend on nucleotide
`composition. Different electrophoresis conditions, different gel
`matrices, or different types of gels,