- Email: [email protected]
Analysis and purification of nucleic acids by ion-pair reversed-phase high-performance liquid chromatography
J. Biochem. Biophys. Methods 46 (2000) 83–93 www.elsevier.com / locate / jbbm
Analysis and purification of nucleic acids by ion-pair reversed-phase high-performance liquid chromatography Karl H. Hecker*, Stacy M. Green, Kaoru Kobayashi Transgenomic Inc., 2032 Concourse Drive, San Jose, CA 95131, USA Received 23 November 1999; received in revised form 22 August 2000; accepted 22 August 2000
Abstract Sizing of DNA fragments is a routine analysis traditionally performed on agarose or polyacrylamide gels. Electrophoretic analysis is labor-intensive with only limited potential for automation. Recovery of DNA fragments from gels is cumbersome. We present data on automated, size-based separation of DNA fragments by ion-pair reversed-phase high performance liquid chromatography (IP RP HPLC) — DNA chromatography — on the WAVE DNA Fragment Analysis System with the DNASep cartridge. This system is suitable for accurate and rapid sizing of double-stranded (ds) DNA fragments from 50 to ca. 2000 base pairs (bp). Fluorescently labeled DNA fragments are compatible with the technology. Length-dependent separation of dsDNA fragments is sequence independent and retention times are highly reproducible. The resolving capabilities of DNA chromatography are illustrated by the analysis of multiple DNA size markers. Resolved dsDNA fragments are easily collected and are suitable for downstream applications such as sequencing and cloning. DNA chromatography under denaturing conditions with fluorescently labeled DNA fragments offers a means for the separation and purification of individual strands of dsDNA. Analysis of DNA fragments on the WAVE System is highly automated and requires minimal manual intervention. DNA chromatography offers a reliable and automated alternative to gel electrophoresis for the analysis of DNA fragments. 2000 Elsevier Science B.V. All rights reserved. Keywords: Nucleic acids; Chromatography; DNA sizing; DNA purification
Abbreviations: ACN, acetonitrile; 6-FAM, 6-carboxyfluorescein; IP RP HPLC, ion-pair reversed-phase high-performance liquid chromatography; ROX, 6-carboxyrhodamine; TAMRA, N,N,N9,N9-tetramethyl-6carboxyrhodamine; TEAA, triethylammonium acetate *Corresponding author. Tel.: 11-408-514-3108; fax: 11-408-894-0405. E-mail address: [email protected] (K.H. Hecker). 0165-022X / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 00 )00133-0
84
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Analysis of nucleic acids is routinely performed by gel electrophoresis [1]. Agarose and polyacrylamide matrices are among those most commonly used for electrophoretic separation, identification, and purification. Size-based separation of linear dsDNA molecules, ranging from 100 to 60 000 bp in length, is typically performed under conditions of constant voltage on gels containing from 2.0 to 0.3% (w / v) agarose. Polyacrylamide gels are widely used for the analysis of DNA molecules because of their high resolving power; molecules differing by as little as 0.2% can be separated [1]. Polyacrylamide gels are most commonly run under denaturing conditions at concentrations typically ranging from 3.5 to 20% (w / v) with effective ranges of separation from 1000–2000 to 6–100 bp. Agarose gels are relatively easy to prepare and run. However, the technique is time consuming and considerable manual effort is required for set-up and analysis and agarose gel electrophoresis does not lend itself for automation. Polyacrylamide gels are even more of a nuisance to prepare and run than agarose gels. Data analysis for polyacrylamide gels has been automated significantly for some applications, e.g., sequencing, but preparation and clean up of gels is cumbersome and labor intensive. Furthermore, data analysis after size-based separation of nucleic acids on agarose or polyacrylamide gels involve the use of hazardous chemicals, such as ethidium bromide or radioactive isotopes, to visualize results and additional photographic steps are required for data recording. In this report we present data on ion-pair reversed-phase high-performance liquid chromatography (IP RP HPLC) — DNA chromatography — as a powerful technique providing an alternative to gel-based analysis for size-based separation of nucleic acids [2–5] and for the purification of dsDNA and single-stranded (ss) DNA. Fragments of dsDNA of up to ca. 2000 bp are resolved by IP RP HPLC and a software algorithm predicts gradient conditions. DNA fragments of interest are quantified by peak integration and easily recovered by fraction collection. IP RP HPLC analysis has the advantage over gel-based analysis of short setup time and fully automated sample analysis. Furthermore, analysis of DNA fragments may be performed under denaturing conditions leading to strand separation and strand resolution of individual strands under specific conditions. DNA chromatography was performed on the WAVE System (Transgenomic, San Jose, CA, USA). Under non-denaturing conditions dsDNA fragments, e.g., restriction fragments or PCR products, are separated in a sequence independent manner on the DNASep cartridge (Transgenomic) [3–5]. The stationary phase consists of a nonporous alkylated poly(styrene-divinylbenzene) matrix. Chromatography is performed with a two-eluent system. Eluent A is aqueous 0.1 M triethylammmonium acetate (TEAA) (Transgenomic) at pH 7.0 and eluent B is aqueous 0.1 M TEAA at pH 7.0 containing 25% by volume of the organic solvent acetonitrile (ACN). The WAVEMAKERE software program determines gradient conditions for the analysis of dsDNA. Restriction digests and PCR can be injected directly, without the need for sample preparation. The injection of up to 192 different samples can be pre-programmed. DsDNA is eluted in a sequence independent, but size-dependent manner with a gradient increasing in eluent B. Chromatograms are recorded at 260 nm (UV) and / or at the emission wavelength of the
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
85
Table 1 List of size markers Marker a
Fragment sizes (bp)
25-bp ladder (0.36 mg / ml) 50-bp ladder (0.34 mg / ml) 100 bp ladder (0.13 mg / ml) pUC18 /HaeIII digest (0.05 mg / ml) pGEM marker (1.0 mg / ml) FX174 /HinfI digest (1.0 mg / ml) FX174 /HaeIII digest (0.13 mg / ml) pBR322 /BstNI digest (1.0 mg / ml) 6-FAM-labeled DNA Ladder (2 fmol / ml / band)
12 Fragments: 25, 50, 75, 100, 125, 150, 175, 200, 225, 250, 275, 300 16 Fragments: 50, 100, 150, 200, 250, 300 350, 400, 450, 500, 550, 600, 650, 700, 750, 800 11 Fragments: 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1500 11 Fragments: 11, 18, 80, 102, 174, 257, 267, 298, 434, 458, 587 15 Fragments: 36, 51, 65, 75, 126, 179, 222, 350, 396, 460, 517, 676, 1198, 1605, 2645 20 Fragments: 24, 40, 42, 48, 66, 82, 100, 118, 140, 151, 200, 249, 311, 413, 417, 427, 500, 553, 713, 726 11 Fragments: 72, 118, 194, 234, 271, 281, 310, 603, 872, 1078, 1353 5 Fragments: 122, 383, 929, 1058, 1857 12 Fragments: 100, 200, 209, 300, 400, 500, 529, 600, 700, 800, 900, 1000
a The pBR322 /BstNI size marker was purchased from New England Biolabs (Beverly, MA). The 6-FAMlabeled DNA Ladder and the FX174 /HaeIII and pUC18 /HaeIII size markers were obtained from Transgenomic (San Jose, CA). All other size markers were purchased from Promega (Madison, WI).
fluorescent marker used for labeling of the DNA fragments. Peaks may be integrated for quantification and separated fragments collected for downstream applications. Several size markers (Table 1) and a 209-bp PCR product were analyzed to demonstrate the sensitivity and resolving power of IP RP HPLC. Gradient conditions are optimized for the separation of dsDNA fragments ranging in size from 50 to 600 bp (Fig. 1A). For comparison with the chromatographic data, 25- and 50-bp size ladders were run on a 3% agarose gel and the 25-, 50- and 100-bp ladder as well as the pUC18 /HaeIII, pGEM, and fX174 /HinfI digests were run on a 1.5% agarose gel (Fig. 1B). DNA fragments ranging in size from 48 bp in the fX174 /HinfI digest to 600 bp in the 100-bp ladder yielded uniform, well-resolved peaks when analyzed by IP RP HPLC. A plot of retention times versus fragment size demonstrates the length dependence and sequence independence of retention of DNA fragments (Fig. 1C). The median standard deviation for DNA fragments of identical lengths across markers was determined to be 60.07 min. Fragments shorter than 48 bp are not resolved under the gradient conditions used here. DNA fragments longer than ca. 1000 bp are eluted in the clean-off step of the gradient (100% eluent B). The sensitivity of online UV detection permits the identification of small amounts of DNA, down to 0.5 ng / peak. This sensitivity, coupled with the high resolving power of DNA chromatography, revealed the presence of trace amounts of unwanted DNA fragments of intermittent lengths among the expected fragments for the 25- and 50-bp
86
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 1. (A) DNA fragment sizing (50–600 bp). IP RP HPLC analysis of various double-stranded size markers (Table 1) was performed on a DNASep cartridge (4.6350 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 44), (0.5 / 49), (3.0 / 58), (5.5 / 62), (8.0 / 64), (10.5 / 66), (13.0 / 67), (13.1 / 100), (13.6 / 100), (13.7 / 44), (15.7 / 44). Flow rate, 0.9 ml / min at 508C. (B) DNA fragment sizing on agarose gels. Analysis of (A) 25- and (B) 50-bp ladder on 3% agarose gel and analysis of various double-stranded size markers on 1.5% agarose gel. (A) 25-bp ladder, (B) 50-bp ladder, (C) 100-bp ladder, (D) pUC18 /HaeIII digest, (E) pGEM Marker, (F) fX174 /HinfI digest. 13 TBE running buffer. Gel and buffer contained 0.5 mg / ml ethidium bromide. (C) Sequence independence of retention time. Plot of retention time versus dsDNA fragment size. Size markers were analyzed in triplicate and individual retention times were averaged and plotted.
ladders (Fig. 1A). Ethidium bromide staining did not provide sufficient sensitivity to detect these fragments. As expected, size-based resolution on agarose gel shows band broadening due to diffusion for smaller sized fragments. Decreased resolution of larger fragments is due to the linear relationship between the rate of migration and the logarithm of the fragment length (Fig. 1B). DNA chromatography was performed on a DNASep cartridge with markers containing dsDNA fragments of up to 2645 bp. Fig. 2 shows the results of four markers containing dsDNA fragments exceeding 600 bp in length. During the initial 23 min of each run DNA fragments ranging in size from 65 to 600 bp are resolved. At higher concentrations of ACN, fragments ranging in size from 603 bp in the FX174 /HaeIII digest to 2645 bp in the pGEM marker are eluted. Larger fragments are generally less resolved than shorter fragments, which is apparent in the broadening of peaks. As previously noted for the 25- and 50-bp ladders (Fig. 1A), spurious peaks were also
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 1. (continued)
87
88
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 2. DNA fragment sizing: larger fragments. IP RP HPLC analysis of various dsDNA size markers (Table 1), which contain fragments larger than 600 bp, was performed on a DNASep cartridge (7.8350.0 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 36), (0.5 / 44), (12.8 / 65), (25 / 69), (37.3 / 72), (49.5 / 75), (49.6 / 100), (51.6 / 100), (51.7 / 36), (53.7 / 36). Flow rate, 0.9 ml / min at 508C.
observed in the chromatogram of the FX174 /HaeIII digest (Fig. 2). Spurious peaks preceding the 194- and 234-bp fragments in the FX174 /HaeIII digest are most obvious. From our experience, these spurious products are most likely to arise from impure restriction enzyme preparations. A distinct advantage of IP RP HPLC over gel-based techniques is the ease with which resolved fragments can be quantified and recovered for downstream uses such as sequencing or cloning. Fig. 3 illustrates the separation and recovery of restriction fragments obtained by HaeIII digestion of a pUC18 plasmid. Nine fragments ranging in size from 80 to 587 bp were separated by IP RP HPLC and five of the fragments were collected and reanalyzed. DNA fragments of interest were collected in volumes ranging from 450 to 900 ml. Purity of collected fragments was assessed by injection of 50-ml aliquots of the collected fractions. Four of the collected fragments (174, 298, 434, 587) were pure, e.g., were devoid of other fragments of the digest. The 458-bp fragment, which was collected in a volume of 690 ml, contained a small amount of 434-bp fragment. The purity of the collected 458-bp fragment could be improved by either collecting the back end of the peak of interest or by collecting multiple, smaller size fractions. Recovery was determined by integration of peak areas in the original run and comparison with peak areas obtained from re-injected sample. Typically, the recovery rate of unlabeled dsDNA fragments exceeds 90% (data not shown).
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
89
Fig. 3. DNA fragment purification. IP RP HPLC analysis was performed on 20 ml pUC18 /HaeIII digest (1000 ng) on a DNASep cartridge (4.6350.0 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 30), (0.5 / 35), (3.0 / 55), (10.0 / 65), (13.5 / 65), (15.0 / 100), (16.5 / 100), (16.6 / 30), (18.6 / 30) at 508C; flow rate, 0.75 ml / min. Peaks corresponding to fragment sizes of 174, 298, 434, 458, and 587 bp were collected. Fifty-ml aliquots of the collected fractions were re-injected and analyzed using the above gradient. A 5-ml injection of pUC18 /HaeIII size marker (250 ng) is shown as reference for retention times of the individual fragments. DNA amounts per peak in the reference chromatogram of the pUC18 /HaeIII digest are as follows: 80 bp / 7.45 ng, 102 bp / 9.50 ng, 174 bp / 16.2 ng, 257 bp / 23.9 ng, 267 bp / 24.9 ng, 298 bp / 27.7 ng, 434 bp / 40.4 ng, 458 bp / 42.6 ng, and 587 bp / 54.6 ng.
DNA fragments separated by IP RP HPLC are eluted in a solution containing 0.1 M TEAA and ACN. The ACN concentration in any given fraction depends on the position in the gradient at which the fragment elutes. The presence of TEAA and ACN do not interfere with downstream applications such as enzymatic sequencing [6,7] or PCR [7]. However, DNA concentrations in collected fractions are significantly lower than in the original sample. This often necessitates concentration of samples, which is easily achieved by peak capture in polypropylene tubes, drying under vacuum, and resuspension in an appropriate buffer such as TE (10 mM Tris–HCl, 1 mM EDTA, pH 8.0). Complete drying results in complete removal of the organic solvent ACN and the ion-pairing reagent (triethylamine and acetic acid). Concentrations of purified and reconstituted DNA fragments may be accurately determined by integration of peak area [8]. Accurate quantification of purified DNA fragments, particularly PCR products, devoid of primers and dNTP is crucial for downstream applications such as sequencing and cloning.
90
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fluorescent dye molecules are frequently used for tagging of DNA [9,10]. Most commonly, a single fluorophore is covalently attached to the 59-end of one of the primers used for PCR, which leads to single-labeled PCR products. Fluorescent dye molecules are generally aromatic, hydrophobic organic molecules. The DNASep column matrix used for IP RP HPLC is also hydrophobic. Thus, the hydrophobicity of the fluorescent dye molecule generally increases retention times of labeled DNA fragments due to increased affinity of the labeled DNA fragment for the column matrix. This increase in retention time can be offset by a gradient containing a higher percentages of organic solvent, e.g., by increasing the percentage of eluent B. While retention times are affected by the presence of fluorescent dye molecules, no loss of resolution is apparent [10]. Fig. 4 shows the analysis of a 6-FAM-labeled 209-bp PCR product and a size marker with fragments ranging in size from 100 to 1000 bp. This demonstrates the compatibility of fluorescently labeled DNA fragments with analysis by IP RP HPLC. All fragments are resolved under the gradient conditions used here. Best resolution occurs for fragments ranging in size from 100 to ca. 700 bp. A 200-bp fragment is well resolved from a 209-bp fragment. DNA fragments 500, 529, and 600 bp in length give
Fig. 4. Size-based separation of 6-FAM-labeled DNA fragments. IP RP HPLC analysis of a 6-FAM-labeled 209-bp PCR product and a 6-FAM-labeled size marker (Table 1) was performed on a DNASep cartridge (4.6350.0 mm). Fifteen ml of 6-FAM-labeled DNA ladder were analyzed. Each peak corresponding to a multiple of 100 bp represents 30 fmol of labeled DNA fragment. The chromatogram was recorded at an emission wavelength of 520 nm. Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 45), (0.5 / 52), (5.3 / 59), (10.0 / 64), (19.5 / 70), (20.0 / 100), (21.5 / 100), (22.0 / 45), (24.0 / 45). Flow rate, 0.75 ml / min at 508C.
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
91
Fig. 5. Analysis and purification of single-stranded DNA. IP RP HPLC analysis of a double-labeled (ROX, FAM) 209-bp PCR product was performed under denaturing conditions at 758C. Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 35), (0.5 / 40), (10.5 / 60), (10.6 / 100), (11.1 / 100), (11.2 / 35), (12.7 / 35). Flow rate, 0.9 ml / min. (A) UV chromatogram of a 10-ml aliquot of a 50-ml PCR. The 209-bp product was generated with one FAM-labeled and one ROX-labeled primer and analyzed under denaturing conditions at 758C using the above gradient. (B) Fluorescence emission chromatogram. FAM / ROX double-labeled PCR product was analyzed as described above and chromatograms were recorded at the emission wavelengths of the FAM label (520 nm) and the ROX label (602 nm). (C) Analysis of purified FAM- and ROX-labeled ssDNA. Labeled ssDNA was collected manually, dried down, resuspended in TE buffer, and reanalyzed by IP RP HPLC under denaturing conditions.
92
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
well-resolved individual peaks. Sensitivity of detection is improved significantly over UV detection. Amounts of FAM-labeled DNA down to the low fmol range are detectable and detection is linear over several orders of magnitude [9]. The absolute increase in sensitivity relative to UV detection depends on the size of the labeled nucleic acid fragment and the specific label used. DNA fragments labeled with other commonly used fluorescent dyes such as fluorescein, ROX, and TAMRA gave resolutions comparable to those determined for 6-FAM-labeled dsDNA for fragments ranging in size from 100 to 1000 bp (data not shown). DNA chromatography with the DNASep cartridge on the WAVE System can be performed at temperatures of up to 808C. At these elevated temperatures and in the presence of ACN dsDNA denatures during analysis and the individual strands may be analyzed. A 209-bp fragment was PCR amplified using a FAM-labeled forward and a ROX-labeled reverse primer. Separation and purification of the individual strands of this PCR product is illustrated in Fig. 5. Under non-denaturing conditions, at 508C, this PCR product yields a single peak (data not shown). Under denaturing conditions, at 758C, two distinct peaks representing the individual strands are observed in the UV chromatogram (Fig. 5A). Oefner et al. reported that the retention of oligonucleotides is dependent on the hydrophobicity of the bases, introducing sequence dependence of retention, and the nature of the fluorophore [9]. Specifically, the impact on retention of a fluorophore is directly proportional to its hydrophobicity. ROX label, which is more hydrophobic than FAM label, is expected to increase retention to a greater extend than FAM label. Fig. 5B shows chromatograms recorded at the emission wavelengths specific for the ROX and FAM labels. FAM labeled ssDNA is clearly identified as the earlier eluting species. ROX labeled ssDNA has the longer retention time. The magnitude of the separation of these complementary strands is remarkable and unparalleled by any other
Fig. 5. (continued)
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
93
separation technique. In addition to the ROX- and FAM-labeled strands of the 209-bp PCR product, the respective primers are also detectable in the fluorescence chromatograms. Fig. 5C shows chromatograms of HPLC-purified ROX- and FAM-labeled ssDNA. These highly pure ssDNA molecules may be used in downstream applications such as in situ hybridization, or Northern and Southern blotting. In summary, DNA chromatography on the WAVE System using the DNASep cartridge provides an automated alternative to sizing of DNA fragments by gel electrophoresis. Chromatographic size-based separation avoids labor-intensive steps associated with gel electrophoresis such as sample preparation, gel pouring, sample loading, gel staining, photographic data recording, and gel clean-up. Analysis set-up for up to 192 samples is fast with gradient conditions being determined by the software. Sample injection and data recording is fully automated. Distinct advantages over gel-based analysis are the ease with which separated fragments can be quantified by peak integration and recovered by peak capture. Also, DNA chromatography using the DNASep cartridge can be conducted at temperatures of up to 808C, allowing the separation and recovery of fluorescently labeled individual strands of dsDNA. The degree of separation of fluorescently labeled complementary strands by DNA chromatography is unparalleled by any gel-based separation technique.
References [1] Sambrook J, Fritsch EF, Maniatis T, editors, Molecular Cloning — A Laboratory Manual, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989, pp. 6.1–6.62. [2] Huber CG, Oefner PJ, Bonn GK. Rapid and accurate sizing of DNA fragments by ion-pair chromatography on alkylated nonporous poly(styrene-divinylbenzene)particles. Anal Biochem 1995;67:578–85. [3] Munson K, Haefele R, Kuklin A, Gjerde D, Taylor P. Sizing of DNA fragments with the WAVE DNA Fragment Analysis System. Transgenomic, Inc., San Jose, CA. Application Note [ 107, 1998. http: / / www.transgenomic.com / pdf /AN107.pdf (and references cited therein). [4] Kuklin A, Davis AP, Hecker KH, Gjerde DT, Taylor PD. A novel technique for rapid automated genotyping of DNA polymorphisms in the mouse. Mol Cell Probes 1999;13:239–42. [5] Hecker KH, Turpie B, Kuklin A. Optimization of cloning efficacy by pre-cloning DNA fragment analysis. BioTechniques 1999;26:216–22. [6] Marino MM, Devaney JM, Smith JK, Girard JE. Sequencing using capillary electrophoresis of short tandem repeat alleles separated and purified by high performance liquid chromatography. Electrophoresis 1998;19:108–18. [7] Shaw-Bruha C, Kuklin A, Taylor P. PCR Amplification and Automated Sequencing of DNA Fragments Isolated with the WAVE System. Transgenomic, Inc., San Jose, CA. Application Note [ 104, 1998. http: / / www.transgenomic.com / pdf /AN104.pdf (and references cited therein). [8] Munson K, Haefele R, Taylor P. Competitive quantitative PCR on the WAVE DNA Fragment Analysis System. Transgenomic, Inc., San Jose, CA. Application Note [ 105, 1998. http: / / www.transgenomic.com / pdf /AN105.pdf (and references cited therein). [9] Oefner PJ, Huber CG, Umlauft F, Berti GN, Stimpfl E, Bonn GK. High-resolution liquid chromatography of fluorescent dye-labeled nucleic acids. Anal Biochem 1994;223:39–46. [10] Hecker KH, Taylor PD, Gjerde DT. Mutation detection by denaturing DNA chromatography using fluorescently labeled polymerase chain reaction products. Anal Biochem 1999;272:156–64.
Analysis and purification of nucleic acids by ion-pair reversed-phase high-performance liquid chromatography Karl H. Hecker*, Stacy M. Green, Kaoru Kobayashi Transgenomic Inc., 2032 Concourse Drive, San Jose, CA 95131, USA Received 23 November 1999; received in revised form 22 August 2000; accepted 22 August 2000
Abstract Sizing of DNA fragments is a routine analysis traditionally performed on agarose or polyacrylamide gels. Electrophoretic analysis is labor-intensive with only limited potential for automation. Recovery of DNA fragments from gels is cumbersome. We present data on automated, size-based separation of DNA fragments by ion-pair reversed-phase high performance liquid chromatography (IP RP HPLC) — DNA chromatography — on the WAVE DNA Fragment Analysis System with the DNASep cartridge. This system is suitable for accurate and rapid sizing of double-stranded (ds) DNA fragments from 50 to ca. 2000 base pairs (bp). Fluorescently labeled DNA fragments are compatible with the technology. Length-dependent separation of dsDNA fragments is sequence independent and retention times are highly reproducible. The resolving capabilities of DNA chromatography are illustrated by the analysis of multiple DNA size markers. Resolved dsDNA fragments are easily collected and are suitable for downstream applications such as sequencing and cloning. DNA chromatography under denaturing conditions with fluorescently labeled DNA fragments offers a means for the separation and purification of individual strands of dsDNA. Analysis of DNA fragments on the WAVE System is highly automated and requires minimal manual intervention. DNA chromatography offers a reliable and automated alternative to gel electrophoresis for the analysis of DNA fragments. 2000 Elsevier Science B.V. All rights reserved. Keywords: Nucleic acids; Chromatography; DNA sizing; DNA purification
Abbreviations: ACN, acetonitrile; 6-FAM, 6-carboxyfluorescein; IP RP HPLC, ion-pair reversed-phase high-performance liquid chromatography; ROX, 6-carboxyrhodamine; TAMRA, N,N,N9,N9-tetramethyl-6carboxyrhodamine; TEAA, triethylammonium acetate *Corresponding author. Tel.: 11-408-514-3108; fax: 11-408-894-0405. E-mail address: [email protected] (K.H. Hecker). 0165-022X / 00 / $ – see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S0165-022X( 00 )00133-0
84
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Analysis of nucleic acids is routinely performed by gel electrophoresis [1]. Agarose and polyacrylamide matrices are among those most commonly used for electrophoretic separation, identification, and purification. Size-based separation of linear dsDNA molecules, ranging from 100 to 60 000 bp in length, is typically performed under conditions of constant voltage on gels containing from 2.0 to 0.3% (w / v) agarose. Polyacrylamide gels are widely used for the analysis of DNA molecules because of their high resolving power; molecules differing by as little as 0.2% can be separated [1]. Polyacrylamide gels are most commonly run under denaturing conditions at concentrations typically ranging from 3.5 to 20% (w / v) with effective ranges of separation from 1000–2000 to 6–100 bp. Agarose gels are relatively easy to prepare and run. However, the technique is time consuming and considerable manual effort is required for set-up and analysis and agarose gel electrophoresis does not lend itself for automation. Polyacrylamide gels are even more of a nuisance to prepare and run than agarose gels. Data analysis for polyacrylamide gels has been automated significantly for some applications, e.g., sequencing, but preparation and clean up of gels is cumbersome and labor intensive. Furthermore, data analysis after size-based separation of nucleic acids on agarose or polyacrylamide gels involve the use of hazardous chemicals, such as ethidium bromide or radioactive isotopes, to visualize results and additional photographic steps are required for data recording. In this report we present data on ion-pair reversed-phase high-performance liquid chromatography (IP RP HPLC) — DNA chromatography — as a powerful technique providing an alternative to gel-based analysis for size-based separation of nucleic acids [2–5] and for the purification of dsDNA and single-stranded (ss) DNA. Fragments of dsDNA of up to ca. 2000 bp are resolved by IP RP HPLC and a software algorithm predicts gradient conditions. DNA fragments of interest are quantified by peak integration and easily recovered by fraction collection. IP RP HPLC analysis has the advantage over gel-based analysis of short setup time and fully automated sample analysis. Furthermore, analysis of DNA fragments may be performed under denaturing conditions leading to strand separation and strand resolution of individual strands under specific conditions. DNA chromatography was performed on the WAVE System (Transgenomic, San Jose, CA, USA). Under non-denaturing conditions dsDNA fragments, e.g., restriction fragments or PCR products, are separated in a sequence independent manner on the DNASep cartridge (Transgenomic) [3–5]. The stationary phase consists of a nonporous alkylated poly(styrene-divinylbenzene) matrix. Chromatography is performed with a two-eluent system. Eluent A is aqueous 0.1 M triethylammmonium acetate (TEAA) (Transgenomic) at pH 7.0 and eluent B is aqueous 0.1 M TEAA at pH 7.0 containing 25% by volume of the organic solvent acetonitrile (ACN). The WAVEMAKERE software program determines gradient conditions for the analysis of dsDNA. Restriction digests and PCR can be injected directly, without the need for sample preparation. The injection of up to 192 different samples can be pre-programmed. DsDNA is eluted in a sequence independent, but size-dependent manner with a gradient increasing in eluent B. Chromatograms are recorded at 260 nm (UV) and / or at the emission wavelength of the
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
85
Table 1 List of size markers Marker a
Fragment sizes (bp)
25-bp ladder (0.36 mg / ml) 50-bp ladder (0.34 mg / ml) 100 bp ladder (0.13 mg / ml) pUC18 /HaeIII digest (0.05 mg / ml) pGEM marker (1.0 mg / ml) FX174 /HinfI digest (1.0 mg / ml) FX174 /HaeIII digest (0.13 mg / ml) pBR322 /BstNI digest (1.0 mg / ml) 6-FAM-labeled DNA Ladder (2 fmol / ml / band)
12 Fragments: 25, 50, 75, 100, 125, 150, 175, 200, 225, 250, 275, 300 16 Fragments: 50, 100, 150, 200, 250, 300 350, 400, 450, 500, 550, 600, 650, 700, 750, 800 11 Fragments: 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1500 11 Fragments: 11, 18, 80, 102, 174, 257, 267, 298, 434, 458, 587 15 Fragments: 36, 51, 65, 75, 126, 179, 222, 350, 396, 460, 517, 676, 1198, 1605, 2645 20 Fragments: 24, 40, 42, 48, 66, 82, 100, 118, 140, 151, 200, 249, 311, 413, 417, 427, 500, 553, 713, 726 11 Fragments: 72, 118, 194, 234, 271, 281, 310, 603, 872, 1078, 1353 5 Fragments: 122, 383, 929, 1058, 1857 12 Fragments: 100, 200, 209, 300, 400, 500, 529, 600, 700, 800, 900, 1000
a The pBR322 /BstNI size marker was purchased from New England Biolabs (Beverly, MA). The 6-FAMlabeled DNA Ladder and the FX174 /HaeIII and pUC18 /HaeIII size markers were obtained from Transgenomic (San Jose, CA). All other size markers were purchased from Promega (Madison, WI).
fluorescent marker used for labeling of the DNA fragments. Peaks may be integrated for quantification and separated fragments collected for downstream applications. Several size markers (Table 1) and a 209-bp PCR product were analyzed to demonstrate the sensitivity and resolving power of IP RP HPLC. Gradient conditions are optimized for the separation of dsDNA fragments ranging in size from 50 to 600 bp (Fig. 1A). For comparison with the chromatographic data, 25- and 50-bp size ladders were run on a 3% agarose gel and the 25-, 50- and 100-bp ladder as well as the pUC18 /HaeIII, pGEM, and fX174 /HinfI digests were run on a 1.5% agarose gel (Fig. 1B). DNA fragments ranging in size from 48 bp in the fX174 /HinfI digest to 600 bp in the 100-bp ladder yielded uniform, well-resolved peaks when analyzed by IP RP HPLC. A plot of retention times versus fragment size demonstrates the length dependence and sequence independence of retention of DNA fragments (Fig. 1C). The median standard deviation for DNA fragments of identical lengths across markers was determined to be 60.07 min. Fragments shorter than 48 bp are not resolved under the gradient conditions used here. DNA fragments longer than ca. 1000 bp are eluted in the clean-off step of the gradient (100% eluent B). The sensitivity of online UV detection permits the identification of small amounts of DNA, down to 0.5 ng / peak. This sensitivity, coupled with the high resolving power of DNA chromatography, revealed the presence of trace amounts of unwanted DNA fragments of intermittent lengths among the expected fragments for the 25- and 50-bp
86
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 1. (A) DNA fragment sizing (50–600 bp). IP RP HPLC analysis of various double-stranded size markers (Table 1) was performed on a DNASep cartridge (4.6350 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 44), (0.5 / 49), (3.0 / 58), (5.5 / 62), (8.0 / 64), (10.5 / 66), (13.0 / 67), (13.1 / 100), (13.6 / 100), (13.7 / 44), (15.7 / 44). Flow rate, 0.9 ml / min at 508C. (B) DNA fragment sizing on agarose gels. Analysis of (A) 25- and (B) 50-bp ladder on 3% agarose gel and analysis of various double-stranded size markers on 1.5% agarose gel. (A) 25-bp ladder, (B) 50-bp ladder, (C) 100-bp ladder, (D) pUC18 /HaeIII digest, (E) pGEM Marker, (F) fX174 /HinfI digest. 13 TBE running buffer. Gel and buffer contained 0.5 mg / ml ethidium bromide. (C) Sequence independence of retention time. Plot of retention time versus dsDNA fragment size. Size markers were analyzed in triplicate and individual retention times were averaged and plotted.
ladders (Fig. 1A). Ethidium bromide staining did not provide sufficient sensitivity to detect these fragments. As expected, size-based resolution on agarose gel shows band broadening due to diffusion for smaller sized fragments. Decreased resolution of larger fragments is due to the linear relationship between the rate of migration and the logarithm of the fragment length (Fig. 1B). DNA chromatography was performed on a DNASep cartridge with markers containing dsDNA fragments of up to 2645 bp. Fig. 2 shows the results of four markers containing dsDNA fragments exceeding 600 bp in length. During the initial 23 min of each run DNA fragments ranging in size from 65 to 600 bp are resolved. At higher concentrations of ACN, fragments ranging in size from 603 bp in the FX174 /HaeIII digest to 2645 bp in the pGEM marker are eluted. Larger fragments are generally less resolved than shorter fragments, which is apparent in the broadening of peaks. As previously noted for the 25- and 50-bp ladders (Fig. 1A), spurious peaks were also
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 1. (continued)
87
88
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fig. 2. DNA fragment sizing: larger fragments. IP RP HPLC analysis of various dsDNA size markers (Table 1), which contain fragments larger than 600 bp, was performed on a DNASep cartridge (7.8350.0 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 36), (0.5 / 44), (12.8 / 65), (25 / 69), (37.3 / 72), (49.5 / 75), (49.6 / 100), (51.6 / 100), (51.7 / 36), (53.7 / 36). Flow rate, 0.9 ml / min at 508C.
observed in the chromatogram of the FX174 /HaeIII digest (Fig. 2). Spurious peaks preceding the 194- and 234-bp fragments in the FX174 /HaeIII digest are most obvious. From our experience, these spurious products are most likely to arise from impure restriction enzyme preparations. A distinct advantage of IP RP HPLC over gel-based techniques is the ease with which resolved fragments can be quantified and recovered for downstream uses such as sequencing or cloning. Fig. 3 illustrates the separation and recovery of restriction fragments obtained by HaeIII digestion of a pUC18 plasmid. Nine fragments ranging in size from 80 to 587 bp were separated by IP RP HPLC and five of the fragments were collected and reanalyzed. DNA fragments of interest were collected in volumes ranging from 450 to 900 ml. Purity of collected fragments was assessed by injection of 50-ml aliquots of the collected fractions. Four of the collected fragments (174, 298, 434, 587) were pure, e.g., were devoid of other fragments of the digest. The 458-bp fragment, which was collected in a volume of 690 ml, contained a small amount of 434-bp fragment. The purity of the collected 458-bp fragment could be improved by either collecting the back end of the peak of interest or by collecting multiple, smaller size fractions. Recovery was determined by integration of peak areas in the original run and comparison with peak areas obtained from re-injected sample. Typically, the recovery rate of unlabeled dsDNA fragments exceeds 90% (data not shown).
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
89
Fig. 3. DNA fragment purification. IP RP HPLC analysis was performed on 20 ml pUC18 /HaeIII digest (1000 ng) on a DNASep cartridge (4.6350.0 mm). Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 30), (0.5 / 35), (3.0 / 55), (10.0 / 65), (13.5 / 65), (15.0 / 100), (16.5 / 100), (16.6 / 30), (18.6 / 30) at 508C; flow rate, 0.75 ml / min. Peaks corresponding to fragment sizes of 174, 298, 434, 458, and 587 bp were collected. Fifty-ml aliquots of the collected fractions were re-injected and analyzed using the above gradient. A 5-ml injection of pUC18 /HaeIII size marker (250 ng) is shown as reference for retention times of the individual fragments. DNA amounts per peak in the reference chromatogram of the pUC18 /HaeIII digest are as follows: 80 bp / 7.45 ng, 102 bp / 9.50 ng, 174 bp / 16.2 ng, 257 bp / 23.9 ng, 267 bp / 24.9 ng, 298 bp / 27.7 ng, 434 bp / 40.4 ng, 458 bp / 42.6 ng, and 587 bp / 54.6 ng.
DNA fragments separated by IP RP HPLC are eluted in a solution containing 0.1 M TEAA and ACN. The ACN concentration in any given fraction depends on the position in the gradient at which the fragment elutes. The presence of TEAA and ACN do not interfere with downstream applications such as enzymatic sequencing [6,7] or PCR [7]. However, DNA concentrations in collected fractions are significantly lower than in the original sample. This often necessitates concentration of samples, which is easily achieved by peak capture in polypropylene tubes, drying under vacuum, and resuspension in an appropriate buffer such as TE (10 mM Tris–HCl, 1 mM EDTA, pH 8.0). Complete drying results in complete removal of the organic solvent ACN and the ion-pairing reagent (triethylamine and acetic acid). Concentrations of purified and reconstituted DNA fragments may be accurately determined by integration of peak area [8]. Accurate quantification of purified DNA fragments, particularly PCR products, devoid of primers and dNTP is crucial for downstream applications such as sequencing and cloning.
90
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
Fluorescent dye molecules are frequently used for tagging of DNA [9,10]. Most commonly, a single fluorophore is covalently attached to the 59-end of one of the primers used for PCR, which leads to single-labeled PCR products. Fluorescent dye molecules are generally aromatic, hydrophobic organic molecules. The DNASep column matrix used for IP RP HPLC is also hydrophobic. Thus, the hydrophobicity of the fluorescent dye molecule generally increases retention times of labeled DNA fragments due to increased affinity of the labeled DNA fragment for the column matrix. This increase in retention time can be offset by a gradient containing a higher percentages of organic solvent, e.g., by increasing the percentage of eluent B. While retention times are affected by the presence of fluorescent dye molecules, no loss of resolution is apparent [10]. Fig. 4 shows the analysis of a 6-FAM-labeled 209-bp PCR product and a size marker with fragments ranging in size from 100 to 1000 bp. This demonstrates the compatibility of fluorescently labeled DNA fragments with analysis by IP RP HPLC. All fragments are resolved under the gradient conditions used here. Best resolution occurs for fragments ranging in size from 100 to ca. 700 bp. A 200-bp fragment is well resolved from a 209-bp fragment. DNA fragments 500, 529, and 600 bp in length give
Fig. 4. Size-based separation of 6-FAM-labeled DNA fragments. IP RP HPLC analysis of a 6-FAM-labeled 209-bp PCR product and a 6-FAM-labeled size marker (Table 1) was performed on a DNASep cartridge (4.6350.0 mm). Fifteen ml of 6-FAM-labeled DNA ladder were analyzed. Each peak corresponding to a multiple of 100 bp represents 30 fmol of labeled DNA fragment. The chromatogram was recorded at an emission wavelength of 520 nm. Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 45), (0.5 / 52), (5.3 / 59), (10.0 / 64), (19.5 / 70), (20.0 / 100), (21.5 / 100), (22.0 / 45), (24.0 / 45). Flow rate, 0.75 ml / min at 508C.
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
91
Fig. 5. Analysis and purification of single-stranded DNA. IP RP HPLC analysis of a double-labeled (ROX, FAM) 209-bp PCR product was performed under denaturing conditions at 758C. Eluent A, 0.1 M TEAA, pH 7.0; eluent B, 0.1 M TEAA, 25% (v / v) acetonitrile, pH 7.0. Gradient conditions (time (min) / % eluent B): (0.0 / 35), (0.5 / 40), (10.5 / 60), (10.6 / 100), (11.1 / 100), (11.2 / 35), (12.7 / 35). Flow rate, 0.9 ml / min. (A) UV chromatogram of a 10-ml aliquot of a 50-ml PCR. The 209-bp product was generated with one FAM-labeled and one ROX-labeled primer and analyzed under denaturing conditions at 758C using the above gradient. (B) Fluorescence emission chromatogram. FAM / ROX double-labeled PCR product was analyzed as described above and chromatograms were recorded at the emission wavelengths of the FAM label (520 nm) and the ROX label (602 nm). (C) Analysis of purified FAM- and ROX-labeled ssDNA. Labeled ssDNA was collected manually, dried down, resuspended in TE buffer, and reanalyzed by IP RP HPLC under denaturing conditions.
92
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
well-resolved individual peaks. Sensitivity of detection is improved significantly over UV detection. Amounts of FAM-labeled DNA down to the low fmol range are detectable and detection is linear over several orders of magnitude [9]. The absolute increase in sensitivity relative to UV detection depends on the size of the labeled nucleic acid fragment and the specific label used. DNA fragments labeled with other commonly used fluorescent dyes such as fluorescein, ROX, and TAMRA gave resolutions comparable to those determined for 6-FAM-labeled dsDNA for fragments ranging in size from 100 to 1000 bp (data not shown). DNA chromatography with the DNASep cartridge on the WAVE System can be performed at temperatures of up to 808C. At these elevated temperatures and in the presence of ACN dsDNA denatures during analysis and the individual strands may be analyzed. A 209-bp fragment was PCR amplified using a FAM-labeled forward and a ROX-labeled reverse primer. Separation and purification of the individual strands of this PCR product is illustrated in Fig. 5. Under non-denaturing conditions, at 508C, this PCR product yields a single peak (data not shown). Under denaturing conditions, at 758C, two distinct peaks representing the individual strands are observed in the UV chromatogram (Fig. 5A). Oefner et al. reported that the retention of oligonucleotides is dependent on the hydrophobicity of the bases, introducing sequence dependence of retention, and the nature of the fluorophore [9]. Specifically, the impact on retention of a fluorophore is directly proportional to its hydrophobicity. ROX label, which is more hydrophobic than FAM label, is expected to increase retention to a greater extend than FAM label. Fig. 5B shows chromatograms recorded at the emission wavelengths specific for the ROX and FAM labels. FAM labeled ssDNA is clearly identified as the earlier eluting species. ROX labeled ssDNA has the longer retention time. The magnitude of the separation of these complementary strands is remarkable and unparalleled by any other
Fig. 5. (continued)
K.H. Hecker et al. / J. Biochem. Biophys. Methods 46 (2000) 83 – 93
93
separation technique. In addition to the ROX- and FAM-labeled strands of the 209-bp PCR product, the respective primers are also detectable in the fluorescence chromatograms. Fig. 5C shows chromatograms of HPLC-purified ROX- and FAM-labeled ssDNA. These highly pure ssDNA molecules may be used in downstream applications such as in situ hybridization, or Northern and Southern blotting. In summary, DNA chromatography on the WAVE System using the DNASep cartridge provides an automated alternative to sizing of DNA fragments by gel electrophoresis. Chromatographic size-based separation avoids labor-intensive steps associated with gel electrophoresis such as sample preparation, gel pouring, sample loading, gel staining, photographic data recording, and gel clean-up. Analysis set-up for up to 192 samples is fast with gradient conditions being determined by the software. Sample injection and data recording is fully automated. Distinct advantages over gel-based analysis are the ease with which separated fragments can be quantified by peak integration and recovered by peak capture. Also, DNA chromatography using the DNASep cartridge can be conducted at temperatures of up to 808C, allowing the separation and recovery of fluorescently labeled individual strands of dsDNA. The degree of separation of fluorescently labeled complementary strands by DNA chromatography is unparalleled by any gel-based separation technique.
References [1] Sambrook J, Fritsch EF, Maniatis T, editors, Molecular Cloning — A Laboratory Manual, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press, 1989, pp. 6.1–6.62. [2] Huber CG, Oefner PJ, Bonn GK. Rapid and accurate sizing of DNA fragments by ion-pair chromatography on alkylated nonporous poly(styrene-divinylbenzene)particles. Anal Biochem 1995;67:578–85. [3] Munson K, Haefele R, Kuklin A, Gjerde D, Taylor P. Sizing of DNA fragments with the WAVE DNA Fragment Analysis System. Transgenomic, Inc., San Jose, CA. Application Note [ 107, 1998. http: / / www.transgenomic.com / pdf /AN107.pdf (and references cited therein). [4] Kuklin A, Davis AP, Hecker KH, Gjerde DT, Taylor PD. A novel technique for rapid automated genotyping of DNA polymorphisms in the mouse. Mol Cell Probes 1999;13:239–42. [5] Hecker KH, Turpie B, Kuklin A. Optimization of cloning efficacy by pre-cloning DNA fragment analysis. BioTechniques 1999;26:216–22. [6] Marino MM, Devaney JM, Smith JK, Girard JE. Sequencing using capillary electrophoresis of short tandem repeat alleles separated and purified by high performance liquid chromatography. Electrophoresis 1998;19:108–18. [7] Shaw-Bruha C, Kuklin A, Taylor P. PCR Amplification and Automated Sequencing of DNA Fragments Isolated with the WAVE System. Transgenomic, Inc., San Jose, CA. Application Note [ 104, 1998. http: / / www.transgenomic.com / pdf /AN104.pdf (and references cited therein). [8] Munson K, Haefele R, Taylor P. Competitive quantitative PCR on the WAVE DNA Fragment Analysis System. Transgenomic, Inc., San Jose, CA. Application Note [ 105, 1998. http: / / www.transgenomic.com / pdf /AN105.pdf (and references cited therein). [9] Oefner PJ, Huber CG, Umlauft F, Berti GN, Stimpfl E, Bonn GK. High-resolution liquid chromatography of fluorescent dye-labeled nucleic acids. Anal Biochem 1994;223:39–46. [10] Hecker KH, Taylor PD, Gjerde DT. Mutation detection by denaturing DNA chromatography using fluorescently labeled polymerase chain reaction products. Anal Biochem 1999;272:156–64.