Anal. Chem. 1998, 70, 1203-1207
Heterogeneity in Bacillus cereus PCR Products Detected by ESI-FTICR Mass Spectrometry David S. Wunschel,† David C. Muddiman,†,§ Karen F. Fox,‡ Alvin Fox,‡ and Richard D. Smith*,†
Macromolecular Structure and Dynamics, Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99352, and Department of Microbiology and Immunology, University of South Carolina School of Medicine, Columbia, South Carolina 29209
PCR amplification of a segment of the 16/23S rDNA interspace region (ISR) from Bacillus cereus 6464 produced a mixture of products. An 89-bp product was predicted on the basis of the reported sequence. The ESI-FTICR analysis revealed three double-stranded products, differing in size by a single nucleotide corresponding to two homoduplexes of 89 and 88 base pairs and a heteroduplex of 89 and 88 nucleotide strands. These were produced from a single preparation of genomic DNA and a single primer pair. ESI-FTICR analysis of the single strands identified a deletion of a T in the coding strand and a corresponding loss of an A in the noncoding strand of this product. The ESI-FTICR analysis indicated the presence of an unreported sequence variation between rRNA operons in this organism. This report illustrates that PCR products amplified from templates differing by a single nucleotide can be resolved and identified using ESI-FTICR at the 89-bp level. Furthermore, the ESIFTICR mass measurements provided the identity of the deletion, which is indicative of interoperon variability. The polymerase chain reaction (PCR) has shown utility in many roles within the biological sciences, including microbial detection and assessment of genetic variability, by generating millions of copies from specific genetic regions. Commonly used gel electrophoretic techniques (i.e., agarose gel electrophoresis) do not provide information on slight differences in length or sequence between PCR products (e.g., single base substitutions or deletions/insertions). Therefore, the development of higher resolution techniques capable of providing precise information on genetic variability (including those contained within PCR products) without the additional steps required for full sequence analysis is an area of ongoing interest. Mass spectrometry is potentially well suited for the role of characterizing enzymatically amplified DNA. Several reports have already described the analysis of PCR products by mass spectrometry.1-9 Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry * Corresponding author. Phone: (509) 376-0723. Fax: (509) 376-2303. E-mail:
[email protected]. † PNNL. ‡ University of South Carolina. § Current address: Department of Chemistry, Virginia Commonwealth University, Richmond, VA 23284. E-mail:
[email protected]. (1) Ch’ang, L. Y.; Tang, K.; Schell, M.; Ringelberg, C.; Matteson, K. J.; Allman, S. L.; Chen, C. H. Rapid Commun. Mass Spectrom. 1995, 9, 772-774. S0003-2700(97)01156-6 CCC: $15.00 Published on Web 02/14/1998
© 1998 American Chemical Society
was predicted to have sufficient resolution for distinguishing 3-base pair (bp) deletions at the 70-mer level (m/∆m ) 15-20).6 Known deletions requiring this level of resolution, such as the detection of cystic fibrosis gene mutations of 3 bp, could be addressed. Restriction-digested PCR products containing a 3-bp deletion (∆F508) from the cystic fibrosis gene were successfully resolved and identified in human samples.1 Enzymatically digested products of 133 and 152 bp in size were detected, potentially allowing detection of a deletion in a separate region of the cystic fibrosis gene using a similar approach.10 Detection of many polymorphisms requires the detection of single base substitutions between PCR products, such as a T-to-A substitution in the β globin gene found in sickle cell anemia.11 Furthermore, identification of base substitutions demands that individual strand masses be resolved, due to the similarity in the masses of deoxyadenosine-thymidine (A-T) and deoxyguanosinedeoxycytidine (G-C) base pairs (617.39 and 618.38 Da, respectively). Not only must the individual strand masses be resolved, but also mass accuracy must be sufficient to identify the smallest possible difference in mass between bases, 9.03 Da for an A to a T. In these cases, higher resolution is clearly necessary to detect mutations producing small mass changes in PCR products. It is important to note that not all PCR amplifications can be tailored to allow analysis of individual products while still providing the necessary biological information. Multiple alleles of a given gene may result in a heterogeneous population of PCR products. (2) Tang, K.; Taranenko, N. I.; Allman, S. L.; Chen, C. H.; Chang, L. Y.; Jacobson, K. B. Rapid Commun. Mass Spectrom. 1994, 8, 673-677. (3) Bai, J.; Liu, Y. H.; Lubman, D. M.; Siemieniak, D. Rapid Commun. Mass Spectrom. 1994, 8, 687-691. (4) Naito, Y.; Ishikawa, K.; Koga, Y.; Tsuneyoshi, T.; Terunuma, H. Rapid Commun. Mass Spectrom. 1996, 9, 1484-1486. (5) Naito, Y.; Ishikawa, K.; Koga, Y.; Tsuneyoshi, T.; Terunuma, H.; Arakawa, R. J. Am. Soc. Mass Spectrom. 1997, 8, 737-742. (6) Doktycz, M. J.; Hurst, G. B.; Habibigoudarzi, S.; McLuckey, S. A.; Tang, K.; Chen, C. H.; Uziel, M.; Jacobson, K. B.; Woychik, R. P.; Buchanan, M. V. Anal. Biochem. 1995, 230, 205-214. (7) Wunschel, D. S.; Fox, K. F.; Fox, A.; Bruce, J. E.; Muddiman, D. C.; Smith, R. D. Rapid Commun. Mass Spectrom. 1996, 10, 29-35. (8) Muddiman, D. C.; Wunschel, D. S.; Liu, C. L.; Pasatolic, L.; Fox, K. F.; Fox, A.; Anderson, G. A.; Smith, R. D. Anal. Chem. 1996, 68, 3705-3712. (9) Tsuneyoshi, T.; Ishikawa, K.; Koga, Y.; Naito, Y.; Baba, S.; Terunuma, H.; Arakawa, R.; Prockop, D. J. Rapid Commun. Mass Spectrom. 1997, 11, 719722. (10) Liu, Y.-H.; Bai, J.; Zhu, Y.; Liang, X.; Siemieniak, D.; Venta, P. J.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1995, 9, 735-743. (11) Saiki, R. K.; Scharf, S.; Faloona, F.; Mullis, K. B.; Horn, G. T.; Erlich, H. A.; Arnheim, N. Science 1985, 230, 1350-1354.
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998 1203
Bacterial genomes have also been reported to carry multiple operons, of which one example is Bacillus subtilis strain 168, which possess 10 operons of ribosomal RNA genes.12,13 Sequence variation of the highly conserved 16S rRNA gene can be as much as 5.0% between different operons and would result in a mixture of PCR products for that gene.14 Therefore, detection of small length or sequence polymorphisms can be required within mixtures of products similar in size or sequence (and, therefore, similar in mass). In comparison to the MALDI-TOF studies, higher resolution measurements of PCR products have been achieved using electrospray ionization (ESI) mass spectrometry.4,5,8,9 Analyses of defined mutations in the APC gene have recently been reported for products of 57 and 44 bp, with resolution of a G-C polymorphism also being reported for the 44-bp product.4,5,9 Using Fourier transform ion cyclotron resonance (FTICR) mass spectrometry, a high-resolution measurement was achieved with mass accuracy of 0.01% for products up to 114 bp in size.8 This allowed for the putative identification of an unexpected base pair substitution, a G-C switch, in a 114-bp product. This base substitution was later confirmed by conventional sequencing.15,16 The previous report revealed an unexpected mixture of PCR products from a segment of the 16S/23S rDNA interspace region (ISR) of a B. cereus strain. This ISR had previously been used as a tool to discriminate between bacterial species on the basis of length or sequence variations.17 It was, therefore, examined as a basis for provide differentiating information on the closely related species in the B. cereus group of bacilli18 and served as a taxonomic model for the previous report.8 The amplification of the 3′ portion of the ISR in B. cereus 6464 was found to produce a mixture of three double-stranded PCR products. However, because only the double-stranded products were observed in the previous report, the sequence variation between the products could not be determined. This article describes the mass measurement of individual strands of multiple polymerase chain reaction products derived from a single region of B. cereus 6464 using ESI-FTICR MS.
dine and imidazole were added to a concentration of 25 mM. Acetonitrile was added to a final concentration of 50%. DNA Preparation and Polymerase Chain Reaction. The procedure for purification of the DNA template and conditions used for the polymerase chain reaction have been previously described.8 The primers, “forward” (5′-GTGTTCTTTGAAAACTAG-3′) and “reverse” (5′-CAAGGCATCCATCGT-3′), correspond to positions 91-108 of the spacer region between the 16S and 23S rRNA genes and 35-21 of the 23S rRNA gene of B. cereus type strain ATCC 14579.19 The predicted double-stranded masses of 89- and 88-bp templates previously sequenced in B. cereus strains are 54 856.74 and 54 239.35 Da, respectively (ATCC 14579 19 and DSM 31 20). The silica resin cleanup and microdialysis procedures were performed to eliminate residual metal cations, as described elsewhere.8,21 Successive dialysis steps were performed using a regenerated cellulose hollow fiber (200-mm-i.d. microdialysis tubing with a 13K molecular weight cutoff) and countercurrent flow of dialysis buffer (2.5 mM NH4OAc), continuously introduced at a flow rate of 0.5 mL/min. The sample was injected into the dialysis tube at 2 µL/min. After the initial desalting step, doublestranded products were evident and previously reported.8 Following storage at -20 °C for 4 weeks, an additional dialysis step was performed using the above procedure. Before analysis, piperidine and imidizole were added, to a combined final concentration of 50 mM.8 The estimated final concentration of each product during infusion was 60 fmol/µL, or 60 pM. Mass Spectrometry. The ESI Fourier transform ion cyclotron resonance mass spectrometer used for the present study has been described in detail elsewhere.22 Briefly, the PCR products were introduced into the mass spectrometer at a flow rate of 0.3 mL/ min in a pulled glass capillary with a coaxial sheath gas (SF6). A source potential of ∼2.3 kV was applied to produce a stable negative ion current. To enhance ion accumulation, broadband noise waveforms were applied from m/z 1000 to 3500 at 10 Vp-p.23-25 Alternatively, a broadband noise waveform was applied at m/z 700-1800 at 10 Vp-p.
EXPERIMENTAL SECTION Synthetic Oligonucleotides. Synthetic 53-nucleotide oligomers were purchased, differing in composition by having either an “A” or a “T”. The base composition for each was A16C18G12T7 for “A”, with a predicted molecular mass of 16 234.78 Da, and A15C18G12T8 for “T”, with a predicted molecular mass of 16 225.77 Da. Prior to analysis, equimolar amounts of oligomers were mixed together to a 20 pmol/µL final concentration, after which piperi-
RESULTS To first demonstrate the ability to resolve an A-T polymorphism in a single strand of DNA, synthetic oligonucleotides were examined. Oligonucleotides of 53 nucleotides differing in composition by a single A-T difference were examined in a single mixture. In Figure 1, the 10- charge states of each component are shown to be clearly be resolved from one another. The predicted mass for the “T”-containing strand was 16 225.77 Da, while that of the “A”-containing strand was 16 234.78 Da. The
(12) Loughney, K.; Lund, E.; Dahlberg, J. E. Nucleic Acids Res. 1982, 10, 16071624. (13) Green, C. J.; Stewart, G. C.; Hollis, M. A.; Bott, K. F. Gene 1985, 37, 261266. (14) Mylvaganam, S.; Dennis, P. P. Genetics 1992, 130, 399-410. (15) Muddiman, D. C.; Anderson, G. A.; Hofstadler, S. A.; Smith, R. D. Anal. Chem. 1997, 8, 1543-1549. (16) Krahmer, M.; Wunschel, D. S.; Fox, K. F.; Fox, A.; Nagpal, M.; Muddiman, D. C.; Smith, R. D. Proceedings of the 45th American Society for Mass Spectrometry Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, June 1-5. 1997. (17) Jensen, M. A.; Webster, J. A.; Straus, N. Appl. Environ. Microbiol. 1993, 59, 945-952. (18) Bourque, S. N.; Valero, J.; Levesque, R. Appl. Environ. Microbiol. 1995, 61, 1623-1626.
(19) Harrell, L. J.; Anderson, G.; Wilson, K. H. J. Clin. Microbiol. 1995, 33, 18471850. (20) Meile, L. EMBL GenBank DDBJ Databases, 1995, Accession No. X89896. (21) Liu, C. L.; Wu, Q. Y.; Harms, A. C.; Smith, R. D. Anal. Chem. 1996, 68, 3295-3299. (22) Winger, B. E.; Hofstadler, S. A.; Bruce, J. E.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1993, 4, 566-577. (23) Bruce, J. E.; Anderson, G. A.; Hofstadler, S. A.; Vanorden, S. L.; Sherman, M. S.; Rockwood, A. L.; Smith, R. D. Rapid Commun. Mass Spectrom. 1993, 7, 914-919. (24) Bruce, J. E.; Vanorden, S. L.; Anderson, G. A.; Hofstadler, S. A.; Sherman, M. G.; Rockwood, A. L.; Smith, R. D. J. Mass Spectrom. 1995, 30, 124133. (25) Bruce, J. E.; Anderson, G. A.; Smith, R. D. Anal. Chem. 1996, 68, 534541.
1204 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
Figure 1. ESI-FTICR mass spectra of two 53-mer oligonucleotides, showing isotopic resolution for the 10- charge state in the inset. The oligonucleotide labeled “T” has a base composition of A15C18G12T8 and a predicted molecular mass of 16 225.77 Da. The “A” 53-mer has a composition of A16C18G12T7 and a predicted molecular mass of 16 234.78 Da.
A
B
Figure 2. ESI-FTICR mass spectra of PCR products produced from B. cereus strain 6464. (A) The charge states 18- through 27- for the double-stranded species. The inset of the 23- charge state contains peaks corresponding in mass to three double-stranded products, the 89/89, 88/89, and 88/88 species. (B) The mass spectra for the single-stranded species having approximately the same mass as the doublestranded products. Four individual strands were detected, labeled 89 C, 89 N, 88 N, and 88 N, with the 89 C and 89 N corresponding in mass to the previously described 89-bp products found in B. cereus and B. anthracis (Table 1). The difference in composition between the coding (C) and noncoding (N) strand of each product is indicated for the 22- charge state (inset).
measured values of 16 225.73 and 16 234.76 Da correlated well with the predicted values and demonstrate the A-T polymorphism with the 9.03-Da mass difference. Clearly, at the 53-mer level, the resolution of FT-ICR is capable of distinguishing between the strands containing the single base polymorphism. However,
resolving two products differing by 9 mass units becomes much more challenging as mass increases. The initial mass spectral analysis of the products from B. cereus strain showed three products. The triplet appearing in each charge state, Figure 2A, provided molecular mass information Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1205
Table 1. Comparison of Double- and Single-Stranded Masses of PCR Products Generated from B. cereus 6464 double-stranded
single-stranded
product label, Figure 2A
measd mass (Da)
product label, Figure 2B
measd mass (Da)
89/89 ∆m 88/89 ∆m 88/88
54 856.6 308.4 54 548.2 308.5 54 239.7
89 C ∆m 88 C 89 N ∆m 88 N
27 603.4 304.1 27 299.3 27 253.2 312.7 26 940.5
corresponding to the intact double-stranded species. The largest and smallest products corresponded to products of 89 and 88 bp in size, respectively (Table 1). The mass of the 89-bp product fit closely with the predicted mass of an 89-bp product expected from this region.8 Interestingly, the difference in mass between the 89/89 and 88/89 as well as the 88/89 and 88/88 products corresponded to the mass of approximately a single nucleotide (Figure 2A and Table 1). This indicates that the 88/89-labeled peaks are from a double-stranded product containing one 88nucleotide and one 89-nucleotide strand. However, the mass difference did not correspond to the mass of a specific nucleotide. Instead, the observed mass difference between the 89/89 and 88/ 89 products was the average mass of an A and T or the average of a G and C, about 308 Da (Table 1). This difference of 308 mass units between the 88/89-labeled product and the other two components indicates that it consists of a mixture of more than one 88/89-nucleotide product. The ratio of the full width at halfmaximum (fwhm) for 89/89 versus 88/88 equaled 1.0, while the ratio between 88/89 and either 88/88 or 89/89 was greater than 1.2. Unfortunately, at the mass of 54 548.1 Da, resolution of a possible mixture of products present within 88/89 (Figure 2A) was not possible. Figure 2B shows the mass spectra of the single-stranded products of the reaction. The largest peak (labeled 89 C) in each charge state corresponds in mass to the 89-nucleotide “coding” strand previously observed for PCR products derived from B. anthracis and B. thuringiensis samples (Figure 2b).8 Likewise, the peak labeled 89 N corresponded in mass to the “noncoding” strand of the 89-bp products from the B. anthracis and B. thuringiensis samples. The m/z peak labeled 88 C in each charge state can be used to calculate a mass corresponding to the 89-nucleotide coding strand minus 304.1 mass units. (Figure 2 and Table 1). This difference in mass corresponds the loss of a T. Likewise, the mass calculated using 88 N from each charge state is equal to the mass of the 89-nucleotide noncoding strand minus 312.7 (Table 1). This mass difference equals the mass of an A. The individual strands of each product were observed after storage at -20 °C for 4 weeks and an additional dialysis step. The additional dialysis step may have further diluted the sample buffer and encouraged dissociation of the individual strands. However, the role of cold storage for several weeks in aiding this process is unknown. 1206 Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
DISCUSSION The amplification scheme employed was designed using conserved regions of the 16S/23S rRNA ISR and 23S rRNA.17,24 In addition, Pfu was the polymerase selected for the generation of PCR products because of the 3′-5′ exonuclease proofreading activity, which eliminates the nontemplate addition of 3′ nucleotides (adenines are preferred when Taq DNA polymerase is used for amplification). Upon analysis of the PCR products from B. cereus strain 6464, three sets of masses corresponding to double-stranded products were observed. The largest of the products corresponded in mass to an 89-bp product, while the smallest corresponded to an 88-bp product. The “middle” product appeared at a mass of one nucleotide less than the 89-bp product and appeared to be a heteroduplex of 88- and 89-nucleotide strands. Because the mass difference was an average mass of an A and T, it was ascribed to a mixture of 88/89 and 89/88 products. The fact that the peak widths for the putative “heteroduplexes” are wider than those seen for products 88/88 or 89/89 also supports this possibility. The factors that could encourage the formation of heteroduplex are not known. However, during the process of cleanup and desalting, the use of guanidinium hydrochloride or the transition to a dilute ammonium acetate buffer may have destabilized the double-stranded structure of each. Observation of double-stranded products formed with strands of slightly different length or sequence has been documented using other methods of analysis. Denaturing gradient gel electrophoresis (DGGE) and electrophoresis in gel-filled capillaries have both been used to analyze mixtures of double-stranded products where strands without exact complementarity (termed “heteroduplexes” by the authors) have been observed.26,27 Without obtaining the masses of the individual strands for each product, the determination of the base difference between products 89/89 and 88/88 was not possible. Using the mass differences, the single base pair difference between the two could clearly be assigned as a missing T in the coding strand and a missing A in the noncoding strand. The ESI-FTICR analyses demonstrated the presence of multiple products arising from a single set of primers. There are two possible explanations for this phenomenon. First, an artifactual amplification may have occurred during the PCR, resulting in a second product. Second, the products could have been generated from multiple operons, differing by a single nucleotide in that portion of the ISR. This possibility is supported by reports documenting sequence and length variability between different operons in a single organism. In the case of 16S rRNA sequences, organisms with more than one reported sequence, 48.0% were found to have variability at more than 1.0% of the positions (1 in every 100 bp). That level of variability is 10 times the estimated sequence error rate (0.1%) for Genebank as a whole.28 Therefore, even in conserved genes such as the 16S rRNA, interoperon variability is prominent. CONCLUSIONS Mass spectrometry provides the potential to examine a complex mixture of PCR products in a single analysis. The presence of multiple forms of genes (e.g., allelic forms) in a single (26) Ferris, M. J.; Ward, D. M. Appl. Environ. Microbiol. 1997, 63, 1375-1381. (27) Righetti, P. G.; Gelfi, C. Anal. Biochem. 1997, 244, 195-207. (28) Clayton, R. A.; Sutton, G.; Hinkle, P. S.; Bult, C.; Fields, C. Int. J. Syst. Bacteriol. 1995, 45, 595-599.
organism proves the necessity of obtaining information on each component within a mixture of PCR products. In the case of very slight differences between the products present, lower resolution techniques, such as commonly used gel electrophoresis, would be unable to determine the presence of multiple forms of genes. In this particular case, the identification of the single base difference could not be made on the basis of the double-stranded masses. Mass differences between the individual strands were obtained by ESI-FTICR to determine the identity of the single base difference. In order for this approach to be useful in detecting sequence variations, including those at the root of some inheritable disorders, small mass variations in each strand must be identified, a step we have now demonstrated to be practical using ESI-FTICR. Furthermore, the presence of multiple copies of genes requires that this information be obtained from within mixtures of similar products, and we anticipate extension of the present approach to such applications.
ACKNOWLEDGMENT We thank the U.S. Department of Energy, Office of Health and Environmental Research, Human Genome Program, for support of this research. Pacific Northwest National Laboratory is operated by Battelle Memorial Institute for the U.S. Department of Energy, through Contract No. DE-AC06-76RLO 1830. We also thank the Army Research Office (ARO) for support through Grants DAAl03-92-G-0255, DAAH0493-G-0506, ARO/ERDEC Grant DAAH04-95-1-0359, and a training award through the DOD AASERT program.
Received for review October 21, 1997. Accepted January 9, 1998. AC971156T
Analytical Chemistry, Vol. 70, No. 6, March 15, 1998
1207
