Analysis of short tandem repeat polymorphisms by electrospray ion trap mass spectrometry

Abstract

The application of electrospray ionization (ESI) ion trap mass spectrometry (MS) to the analysis of short tandem repeats (STRs or microsatellites) is described. Several equine dinucleotide STR loci were chosen as a model system to evaluate ESI ion trap as a routine instrument for rapid and reliable genoytping. With the use of specific primers STR loci were amplified from different blood samples having allele sizes between 60 and 100 bp. A new purification method based on reversible binding of PCR products to magnetic particles has proven to be directly compatible with ESI ion trap MS analysis. The sense and antisense strands of the PCR products with concentrations of ∼100 fmol/µl were measured with a mass accuracy of 0.01%. The simplicity of the purification method and the capability for automated handling together with the precise sizing of PCR products by ESI ion trap MS facilitate the large scale analysis of polymorphic STRs. Moreover, mixtures of different allele length as obtained for heterozygous samples could accurately be assigned as well as a C→G switch between the two strands of a PCR product.

INTRODUCTION

Short tandem repeats (STRs or microsatellites) consist of tandemly repeated 1–6 bp sequences with 10–60 copies (1). STR loci display several qualities that make them useful as genetic markers. They are evenly distributed in all known eukaryotic genomes, amenable to PCR and are highly polymorphic due to variations in the number of the repeat motifs present at a particular locus within a population of individuals (1–3). Numerous STRs in the human population (4,5), as well as in animals such as horse (6,7) and cattle (8,9), have been examined and validated for use in genetic mapping, linkage analysis and forensic or parentage determination (3,4,6,7,10–12). Currently, the PCR products from STR loci are analyzed on a routine basis by means of gel electrophoresis, employing either slab gels or, increasingly, gel-filled capillaries in combination with either silver staining or the introduction of fluorescent labels for detection (13,14). However, gel-based methods are still laborious, time consuming if a resolution >4 bp has to be achieved and inaccurate, since the size determination of DNA molecules relies on calculation of the electrophoretic mobility in the gel matrix relative to an internal standard (3,15).

Mass spectrometry (MS) has become an alternative method to electrophoresis in DNA analysis due to the capabilities for accurate molecular weight determination within a very short analysis time (16,17). The characterization of several polymorphic STR loci by MS has been demonstrated with matrix-assisted laser desorption ionization (MALDI) as well as with electrospray ionization (ESI) MS (18–22). So far, MALDI MS has been the preferred technique utilized in DNA analysis due to the potential for high throughput (16–26). Although MALDI MS has been improved significantly, the poor resolution achieved for large DNA molecules (e.g. PCR products >100 bp) has limited its application (27,28). In comparison to MALDI MS, molecular weight measurements with higher resolution and mass accuracy of PCR products has been achieved using ESI in conjunction with Fourier transform ion cyclotron resonance (FTICR) MS (29–34). Furthermore, new technical developments for molecular weight measurement allow the detection of DNA molecules as large as 110 MDa by ESI MS (35–37).

However, the analysis of large DNA molecules by MS remains difficult due to the increased affinity of the polyanionic DNA ribose–phosphate backbone for positively charged cations (e.g. Na⁺ and K⁺) even at very low concentration (<1 mM) (16,17). In the past, several strategies were developed to obtain sufficiently pure DNA samples. Most involved either ethanol precipitation (16,21,22,33,38–41) or centrifugation through appropriate filter systems/spin columns (30–32,34,41) followed by microdialysis (42–44) or incubation with cation exchange particles (17) for final desalting.

In this study, ESI ion trap MS as a more cost-effective and widely available ESI MS technique compared to ESI FTICR MS (16,45) has been applied to the analysis of polymorphic STR loci. The selected STRs, which consist of dinucleotide repeats, represent a highly abundant sequence motif in many eukaryotic genomes (3,46). However, the determination of such repeat motifs represents a severe challenge since resolving capabilities are required to differentiate alleles which differ by as little as 2 bp. In addition, dinucleotide repeats are prone to amplification errors resulting in truncated PCR artefacts which significantly affect the accurate and reliable allele assignment (47,48). The use of a novel solid phase approach for purification of the PCR reaction mixture which relies on reversible immobilization of PCR products on magnetic particles without special preparation of the PCR product (49) provides a simple sample preparation procedure. The rapid and efficient purification of PCR products from salts and surplus oligonucleotide primers and deoxynucleotide triphosphates (dNTPs) and the capability for this clean-up procedure to be fully automated greatly facilitates the routine application of ESI MS in DNA analysis.

MATERIAL AND METHODS

Polymerase chain reaction

PCR of the equine microsatellite loci was carried out in 50 µl volumes. Each PCR reaction contained ∼50 ng genomic DNA (extracted from whole blood samples), 5 pmol both forward and reverse primer, 1.25 U Taq polymerase (Amersham Pharmacia Biotech, Freiburg, Germany), 10× PCR buffer and 100 µM each dNTP. The conditions for thermal cycling were 95°C for 3 min and 30 cycles of 95°C for 30 s, 52–55°C for 30 s and 72°C for 30 s. The final step was a 10 min elongation at 72°C. As a control, PCR products were analyzed on 10% non-denaturing polyacrylamide gels and visualized by staining with SYBR Gold (MoBiTec, Göttingen, Germany).

Purification of PCR products

Ethanol precipitation. Prior to precipitation of the 50 µl PCR mixture, surplus primers were digested by incubation with 5 U exonuclease I (Amersham Pharmacia Biotech, Freiburg, Germany) for 1 h at 37°C. Without further purification, NH₄OAc (Merck, Germany) was added to this mixture to a final concentration of 2 M, and then 2.5 vol of 100% ethanol. After incubation overnight at –20°C, solutions were centrifuged for 30 min at 14 000 r.p.m. The precipitated pellet was washed with cold 70% ethanol and centrifuged for 15 min at 14 000 r.p.m. After lyophilization, the PCR product was resuspended in 5 µl of deionized water.

Use of Ultrafree-MC centrifugal filters. Due to the expected size of the PCR products, Ultrafree-MC filters (Millipore, Eschborn, Germany) with a nominal molecular weight limit of 10 000 were used. After incubation with exonuclease I, 50 µl of PCR mixture were applied to the filter unit together with 350 µl of deionized water. The sample was centrifuged for 10 min at 10 000 r.p.m. and washed twice with 200 µl of deionized water by centrifugation for 10 min at 8000 r.p.m. The remaining sample was lyophilized and resuspendend in 5 µl of deionized water.

Reversible immobilization of the PCR product on magnetic particles. The removal of surplus primers, dNTPs, buffer, salts and enzyme from the PCR reaction mixture was performed using the GenoPure ds kit (Bruker Daltonik GmbH, Bremen, Germany). The principle of this approach based on reversible immobilization of the PCR product on the surface of specifically designed magnetic particles in the presence of a suitable binding buffer. The amount of sample required for this procedure was 40 µl of the original PCR mixture. After removal of the supernatant and washing of the particles with two different buffers containing ethanol, the purified PCR product was eluted from the particles by incubation with 5 µl of deionized water.

Mass spectrometry

Mass spectrometric analysis was performed on an Esquire-LC quadrupole ion trap (Bruker Daltonik GmbH). The instrument has a fundamental radiofrequency (rf) of 781 250 Hz and was operated with an estimated helium pressure of 5 × 10^–3 mbar. Ions were scanned with a scan speed of 13 000 Da/s at unit resolution using resonance ejection at the hexapole resonance of one third the rf frequency. Calibration of the mass spectrometer was performed using an ES tuning mix (Hewlett Packard, Palo Alto, CA).

Piperidine (TCI Tokio Kasei, Tokyo, Japan) and imidazole (Sigma-Aldrich Chemie GmbH, Steinheim, Germany) were added to the purified sample to a final concentration of 25 mM in 50% acetonitrile. The estimated final concentration of each PCR sample was 100 fmol/µl; 25 µl aliquots were infused at a flow rate of 2 µl/min.

Using ESI in negative ion mode, mass spectra were acquired from m/z = 500 to 1500. For spectra acquisition a total of 100 scans were summarized. The multiple charged signals were deconvoluted to the neutral masses using an automated software tool.

RESULTS AND DISCUSSION

Purification of PCR products

The ability to obtain high quality mass spectra of PCR products by ESI MS is highly dependent on the efficiency of the applied purification procedure for removal of salts, as well as of surplus dNTPs and oligonucleotide primers, which are present in the reaction mixture in molar excess. So far, the most successful clean-up procedures rely on ethanol precipitation or use of spin columns followed by microdialysis as a final desalting step (16,21,30,31,34,38–44). However, these procedures are time- and labor-intensive and incur significant sample loss. Additionally, this kind of purification protocol is prolonged, since an off-line microdialysis step has proven to be more efficient in desalting than the on-line configuration (44). In order to employ ESI MS as a routine analytical technique for genotyping based on the analysis of multiple polymorphic STRs, a more rapid and simple sample clean-up procedure must be explored.

In this report, a novel solid phase method for the purification of PCR products in a single tube format has been applied prior to ESI MS. This approach makes use of magnetic particles on which the PCR products are reversibly immobilized in the presence of an appropriate binding buffer. The surplus oligonucleotide primers and dNTPs, salts and buffer components remain in solution and are easily removed following magnetic separation. After washing of the magnetic particles the captured PCR products are released from the solid phase by incubation with distilled deionized water for a few minutes.

The advantageous features of a solid phase-based purification approach involving immobilization of the analyte on magnetic particles prior to MS has been demonstrated (18,49). However, streptavidin-coated magnetic particles were utilized in these studies, which require labeling of the PCR products with biotin for immobilization. Another disadvantage of this approach is the additional denaturing step necessary to release the non-biotinylated strand of the captured PCR product prior to MS analysis.

Figure 1 illustrates a comparison of the novel magnetic bead approach for purification of PCR products prior to ESI ion trap MS with both ethanol precipitation from ammonium acetate solution and centrifugation through size exclusion filter units followed by an additional desalting step involving incubation of the purified PCR product with NH₄⁺-loaded cation exchange beads. Independent of the applied purification method, 25 mM imidazole and 25 mM piperidine in 50% acetonitrile were added to each purified sample prior to infusion into the ESI source. These additives have proven to be advantageous for ESI MS analysis of DNA samples due to the efficient suppression of residual cation adducts (32,50). Figure 1a shows the ESI ion trap mass spectrum of the 81 bp PCR product amplified from an individual homozygous at the equine STR locus HMS3 after application of the novel magnetic bead approach. The spectrum exhibited individual charge states (–22 to –35) of the two complementary strands of the PCR product allowing molecular mass determination via a charge deconvolution procedure, which is implemented in the post-processing software, with a mass accuracy of 0.02%. Detection of mass differences of approximately ±40 Da between the experimental and the predicted masses of the sense (24988.4 Da) and antisense (25560.6 Da) strands indicates a one base substitution (C→G) between the two strands (theoretical mass deviation for each strand 40.03 Da). The occurrence of a base substitution as detected here was somewhat unexpected since only those STR loci which showed a mutation rate <5 × 10^–4 were selected for genotyping (3). However, with the achieved resolution and mass accuracy such sequence variations can be readily detected and identified by ESI ion trap MS.

Figure 1.

Negative ESI ion trap MS obtained for a homozygous individual at the equine dinucleotide STR locus HMS3. Comparison of different approaches for purification of the 81 bp PCR product: (a) solid phase procedure based on reversible immobilization on magnetic particles; (b) ethanol precipitation; (c) ultrafiltration using Ultrafree-MC filters. The applied purification procedures (b) and (c) were followed by incubation with NH₄⁺-loaded cation exchange beads for removal of residual salts. A, sense (+) strand; a, antisense (–) strand; *, sense (+) strand without non-templated addition of A.

In comparison, both the resolution and signal-to-noise ratio are significantly decreased after application of either ethanol precipitation (Fig. 1b) or centrifugation through size exclusion filter units (Fig. 1c) for sample clean-up of the same PCR product. Inefficient purification of the PCR product (HMS3) from salts and surplus components (e.g. primers and dNTPs) is obvious due to the presence of a variety of additional signals and a considerable number of cation adducts detected under these conditions. As a consequence, the experimental masses calculated from the partially resolved charge states of the sense and antisense strands of this PCR product are less accurate compared to the data obtained after application of the magnetic bead procedure (Fig. 1a). Although these two methods have been described to be efficient for purification of PCR products prior to ESI MS, the present data clearly indicate that application of these procedures needs further optimization if used prior to ESI ion trap MS. Above all, sample quality and quantity suffers from a low degree of reproducibility. Thus, the advantageous features of the novel magnetic bead purification procedure are readily evident in the fact that reliable purification of PCR products was routinely achieved with a high level of purity and reproducibility within a relatively short preparation time (<20 min). Additionally, this purification approach is amenable to fully automated handling.

The ability to detect the discrete single strands of a PCR product rather than the intact dsDNA is most likely an effect of the applied purification method. As known from the literature, detection of denatured PCR products is presumably a consequence of the removal of salt below a concentration optimal for stabilization of the secondary structure of the dsDNA molecules (31,51). However, efficient removal of salts is necessary to obtain high quality ESI MS spectra. The mass accuracy obtained for the detected single-stranded species enables a more precise mass assignment, which is beneficial for characterization of mutations in DNA sequences (21,51).

In Figure 1a an additional component is present in the spectrum which could be associated with an artefact of the PCR reaction caused by the Taq polymerase. Taq has been reported to incorporate an additional dA residue at the 3′-ends of blunt-ended double-stranded PCR products, which results in a mass difference of 312.3 Da. In case of the STR locus HMS3, Taq displays a reactivity for both addition of dA on the 3′-end and correct amplification without non-specific elongation.

Analysis of STRs

In order to evaluate ESI ion trap MS as a reliable analytical method for STR-based genotyping, several equine dinucleotide STR loci were amplified from different blood samples and subjected to the above mentioned magnetic bead clean-up procedure prior to MS analysis. In general, STR loci with dinucleotide repeats have been reported to be most challenging, since amplification of these sequence motifs is prone to errors (3,48). It is assumed that the template and also the newly synthesized strand become unpaired during elongation across a dinucleotide repeat and, as a result, truncated PCR products (‘stutter bands’) are generated which affect the accurate and reliable allele sizing (47). However, the generation of such PCR artefacts is not predictable for every STR locus, since many loci were found to yield precisely defined PCR amplification products. Therefore, analytical methods are required which are capable of distinguishing PCR artefacts from the desired product. Among the selected equine dinucleotide STR loci analyzed in this work, only one marker (VHL20) was found to generate such PCR artefacts. Figure 2 illustrates the genotyping results obtained for two different individuals which were either homozygous (Fig. 2a) or heterozygous (Fig. 2b) at the marker VHL20. In Figure 2a two related series of distinct charge states were resolved corresponding to allele sizes of 79 and 81 bp (Table 1). Data obtained after electrophoretic separation of the same sample revealed that this DNA sample is homozygous at this marker with an allele length of 81 bp (data not shown). Thus, the allele with a size of 79 bp can be associated with an artificial PCR product likely to be generated during amplification of the dinucleotide repeat. It has to be noted that such an additional amplification product was also visible in the electropherogram as a light band 2 bp apart from the main band. This is consistent with the assumption that such PCR artefacts are generated to a minor degree (47). In practice, the assignment of PCR artefacts on the basis of relative signal intensities is not always an absolutely reliable method, since the mobility of DNA molecules during electrophoretic separation has been found to be highly anomalous if alteration of the molecular conformation occurs (14,46). As a consequence, PCR artefacts and true alleles were found to overlap when analyzed by gel electrophoresis (48). In contrast to this, such PCR artefacts can be clearly discriminated from the real amplified STR allele by MS. As evident from the spectra displayed in Figure 2, the molecular weight determination is not affected by the sequence of the DNA molecules. In addition, the fact that PCR artefacts are generated to a minor degree compared to the amplification of the real alleles has also been observed if STRs are analyzed by ESI ion trap MS. In Figure 2b the charge states corresponding to the PCR artefacts (* and #) and to the real alleles (A, a, B and b) of this heterozygous DNA sample could be resolved. The deconvoluted spectrum of this sample (Fig. 2b, insert) gives three pairs of molecular masses which can be assigned to the expected alleles with sizes of 71 and 73 bp as well as to a PCR artefact 2 bp shorter than the 71 bp allele. This result clearly indicates the potential of ESI ion trap MS for genotyping based on the analysis of STRs, since allele lengths which differ by as little as one dinucleotide repeat can be resolved.

Figure 2.

Negative ESI ion trap MS of the PCR products obtained from the equine dinucleotide STR locus VHL20. The PCR product in (a) is derived from a homozygous individual [(GT)₁₉] and those in (b) are derived from a heterozygous individual [(GT)₁₄/(GT)₁₅]. A and B, sense (+) strand; a and b, antisense (–) strand; * and #, PCR artefacts. The charge states are provided for A. The inserts illustrate the deconvoluted mass spectra.

Table 1. ESI ion trap mass measurement results of the amplification products from both a homozygous and a heterozygous individual at the equine dinucleotide STR loci VHL20 (Fig. 2), HTG6 and HMS3.

STR locus	Allele (repeat)	Allele size (bp)	Mass (Da)
			Strand	Predicted	Measured		Δ
Homozygous
VHL20	(GT)18a	79	+	24605.9	24607.8	#	1.9
			–	24088.8	24089.2	*	0.4
	(GT)19	81	+	25239.3	25240.6	A	1.3
			–	24691.2	24690.6	a	0.6
HTG6	(CA)14	64	+	19946.1	19945.0		1.1
			–	20099.0	20100.3		1.3
HMS3	(CA)16	81	+	25028.4	25027.3		1.1
			–	25520.6	25520.7		0.1
Heterozygous
VHL20	(GT)13a	70	+	21438.9	21439.8	#	0.9
			–	21076.9	21076.8	*	0.1
	(GT)14	71	+	22072.3	22072.3	A	0
			–	21679.3	21679.7	a	0.4
	(GT)15	73	+	22705.7	22707.1	B	1.4
			–	22281.7	22281.5	b	0.2
HTG6	(CA)14	64	+	19946.1	19943.6		2.5
			–	20099.0	20099.6		0.6
	(CA)19	74	+	22960.1	22958.0		2.1
			–	23268.4	23268.0		0.4
HMS3	(CA)20	89	+	27397.9	27397.6		0.3
			–	28094.2	28094.0		0.2
	(CA)23	85	+	29205.1	29202.4		2.7
			–	29994.5	29993.9		0.6

^aPCR artefact.

#, *; A, a, B and b correspond to the signals in Figure 2.

The analysis of unknown samples by ESI MS is, however, complicated, since PCR artefacts, if detected with high signal intensities, are likely to be interpreted as real alleles. To ensure the performance of ESI ion trap MS for STR-based genotyping, polymorphic markers have to be selected which show less tendency for generation of PCR artefacts. Table 1 lists the results obtained for analysis of the STR markers HMS3 and HTG6. In accordance with the data displayed in Figure 2, the sense and antisense strands of each PCR product were resolved and detected with a mass accuracy better than 0.01%. Since PCR artefacts are not present after amplification of these markers, an unambiguous determination of allele lengths for both the homozygous and the heterozygous DNA sample was feasible.

Individual identification

To ensure reliable assignment of different STR-based genotypes among a population by ESI ion trap MS, a correct mass measurement within a dynamic mass range is required in order to cover the complete length diversity of a polymorphic marker. DNA samples from four different individuals were selected which yielded amplification products representing five different alleles at the marker HMS3 with fragment sizes in the mass range 24–30 kDa (Table 2). It should be noted that the use of Pfu (exo^–) polymerase was advantageous for amplification of this STR locus since this polymerase appears to have no tendency for non-specific reactivity, such as the non-templated addition of dA (18,31). As a consequence, the resultant mass spectra are less complex and the interpretation is facilitated. Figure 3 shows the ESI ion trap mass spectra of the four different amplification products obtained after application of the magnetic bead purification procedure. Figure 3a shows the mass spectrum of a DNA sample homozygous at this marker, while all other DNA samples were obtained from heterozygous individuals (Fig. 3b–d) carrying alleles with differences in the number of repeat motifs (Δn) from 1 to 3 CA-repeats (Table 2). All spectra are interpretable with a resolution of the complementary strands. A minor series of additional charge states were also detected in all spectra, which correspond either to PCR artefacts, such as the above mentioned repeat slippage products (Fig. 3a), or residual impurities (Fig. 3d). However, correct interpretation of these spectra is not prevented, since the PCR products give signals with higher intensities compared to the additional detected species. As displayed in Table 2, the measured molecular weights are in good agreement with the theoretical values. With the achieved averaged mass error of 0.01% a correct determination of the number of repeats was possible for each detected allele of this marker. Again, the ability to detect the two complementary strands of the PCR product rather than the intact dsDNA molecule is advantageous for the analysis of STRs. On one hand, redundant molecular weight information is obtained due to separate detection of the sense and antisense strand of a PCR product. On the other hand, the mass accuracy and precision obtained for detection of single-stranded PCR products allow a more reliable identification of changes in the sequence. For instance, a C→G switch between the two strands of the alleles of the STR marker HMS3 was observed for nearly all tested DNA samples (Table 2).

Table 2. Allele sizes at the equine dinucleotide locus HMS3.

DNA sample	Allele (CA)n	Allele size (bp)	Mass (Da)
			Strand	Predicted	Measured		Δm
Horse 1	(CA)16	81	+	24715.1a	24714.1	A	1.0
			–	25207.4a	25206.7	a	0.7
Horse 2 (Δn = 1)	(CA)21	91	+	27727.2a	27726.9	A	0.3
			–	28374.4a	28373.9	a	0.5
	(CA)22	93	+	28329.5a	28328.2	B	1.3
			–	29007.8a	29008.2	b	0.4
Horse 3 (Δn = 2)	(CA)20	89	+	27124.8a	27124.3	A	0.5
			–	27741.0a	27740.6	a	0.4
	(CA)22	93	+	28329.5a	28328.3	B	1.2
			–	29007.8a	29008.4	b	0.6
Horse 4 (Δn = 3)	(CA)20	89	+	27084.7	27084.4	A	0.3
			–	27781.0	27780.3	a	0.7
	(CA)23	95	+	28891.9	28892.4	B	0.5
			–	29680.2	29677.7	b	2.5

ESI ion trap mass measurement results for PCR products derived from four different DNA samples (Fig. 3).

A, a, B and b correspond to the signals in Figure 3. ^aMolecular masses calculated for a one base substitution (C→G).

Figure 3.

Negative ESI ion trap MS analysis of the STR locus HMS3 from several individuals. The PCR product in (a) is derived from a homozygous individual. In (b)–(d) amplification products are derived from different heterozygous DNA samples. All samples were amplified using Pfu (exo^–) DNA polymerase and were purified by reversible immobilization on magnetic particles. A and B, sense (+) strand; a and b, antisense (–) strand; * and #, PCR artefacts.

The ESI ion trap results shown in this work clearly demonstrate the applicability of this technique to the analysis of polymorphic STRs. Encouraging genotyping results have been obtained using a novel magnetic bead purification method prior to ESI ion trap MS. Sufficient resolution and mass accuracy has been achieved to allow reliable determination of the number of repeats for dinucleotide repeat sequences as well as identification of genetic variations.

CONCLUSIONS

The analysis of polymorphic STR loci for genotyping by ESI ion trap MS has been demonstrated for the first time. A novel approach for purification of the PCR-amplified STR loci prior to mass spectrometric analysis, which relies on reversible immobilization of the PCR products to magnetic particles, proved to be very effective for separation of PCR products from salts, buffers, surplus dNTPs and oligonucleotide primers. Furthermore, since only simple pipetting and magnetic separation are involved, this clean-up procedure is amenable to fully automated handling. The two complementary strands of each PCR product were detected separately with a mass resolution and accuracy adequate for correct determination of the different allele sizes obtained for the dinucleotide repeat sequences of the selected equine STR loci. In addition, a single base substitution (C→G) between the sense and antisense strands within the sequence of STR locus HMS3 could be identified. Thus, ESI ion trap MS in combination with the novel magnetic bead purification approach is a promising alternative for STR-based genoytping compared to instruments capable of high resolution (e.g. ESI-FTICR MS) or high throughput (e.g. MALDIMS).