| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Sequenom Instruments, Mendelssohnstr. 15D, 22761 Hamburg, Germany.
2
Department of Biochemistry and Molecular Biology,
University of Hamburg, Martin-Luther-King-Platz 6, 20146 Hamburg,
Germany.
a Author for correspondence. Fax 49-40-89967610; e-mail abraun01{at}aol.com
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: cystic fibrosis • PROBE • dideoxynucleotides
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Mass spectrometry (MS) is another separation and detection scheme that has been used for decades in the identification and structural determination of small biomolecules such as metabolites. The introduction of the gentle ionization technique matrix-assisted laser desorption (MALDI) (13)(14)(15) and electrospray ionization (16) have made it possible to extend the inherent analytical advantages of MS to the analysis of far larger biomolecules. With the former, analyte molecules are cocrystallized with a matrix having a strong absorbance at the wavelength of a pulsed laser and ionized by proton transfer. With the latter, an aqueous solution containing analyte is passed through a highly charged (~2 kV) needle; the resulting highly charged droplets desolvate until multiply charged ions are emitted.
Although sizing and partial sequence analysis of recombinant and extracted proteins by MS are now in most cases routine (17), progress with DNA has been slower. Its labile phosphodiester backbone and the tendency for base cleavage from deoxyribose makes intact volatilization of larger strands challenging (18). Furthermore, the affinity of the negatively charged backbone phosphate groups for nonvolatile cation (Na+,K+) adduction spreads the molecular ion signal into several peaks, decreasing the signal-to-noise ratio and molecular mass measurement accuracy with instrumentation of insufficient resolving power to differentiate adduct from molecular ion signal. However, after careful sample preparation (19)(20) and use of low-energy-source conditions, signal from DNA up to a 500-mer has been observed with MALDI (21). Meanwhile, the potential of MS in diagnostics of clinical samples through amplified DNA has been demonstrated (19)(22)(23).
We present a new method of mutation detection combining primer oligo base extension (PROBE) with mass measurement by MALDI coupled with time-of-flight (TOF) MS. PROBE is a solid-phase method principally similar to minisequencing (10)(11)(12), and both can be performed without excessive attention to stringency. A single mutation detection primer is annealed to a target downstream of a mutation region and on addition of a polymerase, three dNTPs, and one ddNTP, undergoes an oligonucleotide extension through the mutation region. Product lengths are determined directly from their molecular mass (Mr) so that labeling the DNA is unnecessary; the high duty cycle of TOF-MS (single-shot spectrum acquisition <1 s) makes it far more amenable to high-throughput screening programs than fluorescent or radioactive methods, which require slow electrophoretic separation or hybridization procedures. The model system used here comprises three clinically relevant CFTR gene regions.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
F508 mutation, and individuals
compound-heterozygous or simple-heterozygous for other mutations
leading to CF. All wild-type and mutant alleles in the exons tested
thus far have been confirmed by standard Sanger sequencing.
pcr amplification
Oligonucleotides were synthesized in-house on a Perseptive
Expedite DNA synthesizer using standard ß-cyanoethanol phosphoamidite
chemistry (24). PCR primers, biotinylated (-bio) as
indicated, were delineated from the CFTR intron sequences
(25). Exon 9, CFEx9-F-bio: d(GAA AAT ATC TGA CAA ACT CAT
C) and CFEx9-R: d(CAT GGA CAC CAA ATT AAG TTC); exon 10, CFEx10-F-bio:
d(GCA AGT GAA TCC TGA GCG TG) and CFEx10-R: d(GTG TGA AGG GTT CAT ATG
C); exon 11, CFEx11-F: d(CAA ATT CAG ATT GAG CAT AC) and CFEx11-R-bio:
d(ACA GCA AAT GCT TGC TAG AC). Total reaction volume was 50 µL with
20 pmol of primers per reaction. Taq polymerase including
10x buffer (1x buffer: 10 mmol/L Tris-HCl, 1.5 mmol/L
MgCl2, 50 mmol/L KCl, pH 8.3) was purchased from Boehringer
Mannheim (Mannheim, Germany), and dNTPs were obtained from Pharmacia
Biotech (Uppsala, Sweden). Cycling conditions were: 5 min at 94 °C,
followed by 35 cycles of 30 s at 94 °C, 45 s at 53 °C,
and 30 s at 72 °C, with a final extension time of 2 min at
72 °C.
purification of pcr products
Amplification products were purified using Qiagen's PCR
purification kit 28106 according to the manufacturer's instructions;
purified products were eluted with 50 µL of TE buffer (10 mmol/L
Tris, 1 mmol/L EDTA, pH 7.5).
affinity capture and denaturation of double-stranded dna
Aliquots (10 µL) of the purified double-stranded PCR product
(~3 pmol) were transferred to a streptavidin-coated microtiterplate
well (~16-pmol capacity per 50-µL volume; Model 1645684, Boehringer
Mannheim), followed by 10 µL of incubation buffer (80 mmol/L sodium
phosphate, 0.4 mol/L NaCl, 4 mL/L Tween 20, pH 7.5) and 30 µL of
water. After incubation for 1 h at room temperature, the wells
were washed three times with 200 µL of washing buffer A (40 mmol/L
Tris, 1 mmol/L EDTA, 50 mmol/L NaCl, 1 mL/L Tween 20, pH 8.8) and then
incubated with 100 µL of a 50 mmol/L NaOH solution for 3 min to
denature the double-stranded DNA. Finally, the wells were washed three
times with 200 µL of 70 mmol/L ammonium citrate solution.
probe
The annealing of 100 pmol of the various detection primers [CFpT:
d(TTC CCC AAA TCC CTG); CF508: d(CTA TAT TCA TCA TAG GAA ACA CCA);
CF542: d(GAA AGA CAA TAT AGT TCT T); CF553: d(CTG AGT GGA GGT CAA)]
was performed in a hybridization oven with 50 µL of annealing buffer
(50 mmol/L ammonium phosphate buffer, pH 7.0, and 100 mmol/L ammonium
chloride) at 65 °C for 2 min, at 37 °C for 10 min, and at room
temperature for 10 min. The wells were washed three times with 200 µL
of washing buffer B (40 mmol/L Tris, 1 mmol/L EDTA, 50 mmol/L
NH4Cl, 1 mL/L Tween 20, pH 8.8) and once with 200 µL
of TE buffer. Subsequently, the extension reaction was performed using
some components of the DNA sequencing kit from US Biochemical Corp.
(Cleveland, OH; Model 70770) and dNTPs or ddNTPs from Pharmacia;
deoxypurines were used in the N-7 deaza form unless noted otherwise.
Total reaction volumes were 45 µL, consisting of 21 µL of water, 6
µL of Sequenase buffer, 3 µL of 100 mmol/L dithiothreitol solution,
4.5 µL of 0.5 mmol/L solution of three dNTPs, 4.5 µL of 2 mmol/L
solution of the ddNTP not present in the deoxy form, 5.5 µL of
glycerol enzyme dilution buffer, 0.25 µL of Sequenase 2.0, and 0.25
µL of pyrophosphatase. The reaction was pipetted on ice and incubated
for 15 min at room temperature and for 5 min at 37 °C. Finally, the
wells were washed three times with 200 µL of washing buffer B.
denaturation and precipitation of extended primer
The extended primer was denatured from the template strand after
addition of 50 µL of 50 mmol/L ammonium hydroxide in 100 mL/L
dimethyl sulfoxide solution by heating at 80 °C for 10 min in a
hybridization oven. For precipitation, 5 µL of 3 mol/L ammonium
acetate (pH 6.5), 0.5 µL of glycogen (10 g/L, cat. no. G1765; Sigma
Chemical Co., St. Louis, MO), and 110 µL of absolute ethanol were
added to the supernatant and incubated for 1 h at room
temperature. After centrifugation at 13 000g for 10 min,
the pellet was washed in ethanol:water (70:30, by vol) and resuspended
in 1 µL of 18 M
/cm H2O. Considering that both the
known amount of PCR products used and the hybridization/precipitation
efficiency were high, it was estimated that ~3 pmol of diagnostic
product was available for MS analysis.
MS
Matrix solution (0.35 µL of 0.7 mol/L 3-hydroxypicolinic acid
[26], 70 mmol/L dibasic ammonium citrate in 1:1
water:acetonitrile) and resuspended DNA/glycogen pellet (0.35 µL)
were mixed on a stainless-steel sample target and allowed to air-dry.
The target was introduced into the source region of an unmodified
Thermo Bioanalyis Vision 2000 MALDI-TOF operated in reflectron mode
with 5 and 20 kV on the target and conversion dynode, respectively.
Theoretical average molecular masses
[Mr(calc)] were calculated from atomic
compositions; experimental [Mr(exp)]
values, all determined with external calibration, are reported as
neutral deprotonated forms. Average masses of nucleotides are 313.2
(A), 304.2 (T), 289.2 (C), and 329.2 (G); those for terminal
dideoxynucleotides are 16.0 Da lower than the above respective values.
The instrument was calibrated with an external DNA standard between 5
and 18 kDa. Intense matrix signals (<<1 kDa) are not shown in the
figures because they are always present but add nothing to the
diagnostic information. No template signal that remains captured on the
microtiterplate was observed in any mass spectra.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
presents a schematic overview of the PROBE experiments for
potential sequence variations in exons 10 and 11 and intron 8 of the
CFTR gene. In exon 10, for instance, the sense strand of the
PCR product served as template; the nucleotides by which the primer was
extended are depicted in bold letters. Two specific termination
reactions (ddTTP and ddCTP, Fig. 1
, a and b) were performed to
investigate known mutations at the codons 506, 507, and 508.
|
Figure 2
, a and b, shows mass spectra of PROBE reaction products
generated from a negative control, i.e., a sample that is
homozygous-normal at each of the codons 506, 507, and 508. In the ddT
and ddC reactions, the 7288.8-Da 24-mer primer is converted entirely to
products consistent with the addition of 5 (8842.8 Da, Fig. 2a
) and 14
(11 609.7 Da, Fig. 2b
) bases, respectively; the result in Fig. 2a
is
consistent with not only wild-type, but also a genotype homo- or
heterozygous for I506S and I507. To resolve this ambiguity, these
are readily distinguished on consideration of the ddC reaction, in
which wild-type, I506S, and I507 alleles yield products of
dramatically different mass (11 609 vs 9468 and 10 653 Da,
respectively). Thus only by consideration of both the ddT and ddC
reactions can this genotype be verified as homozygous-normal.
|
The split peaks in Fig. 2c
(ddT) indicate a heterozygous genotype in
which two diagnostic products result from a single detection primer;
those primers that anneal to allele 1 (Table 1
) are again extended by 5 bases (8839.5 Da), whereas those that
anneal to allele 2 are extended by only 2 bases (7885.8 Da); the latter
can only be from F508. Again, consideration of the ddC reaction
(Fig. 2d
) for this patient is needed for an unambiguous analysis; the
11 606-Da fragment can be generated only from a wild-type, while the
10 653-Da product confirms the presence of F508. Further genotype
assignment for patients homozygous for F508 (Fig. 2
, e and f),
compound heterozygote I507/F508 (Fig. 2
, g and h), and
heterozygote for I506S (Fig. 2
, i and k) are equally definitive, even
though the two molecular species of the peak in Fig. 2h
(m = 9 Da) are not resolved.
|
PROBE was also used for simultaneous (biplex) detection of the
CFTR gene exon 11 G542X and R553X mutations in two patients.
As seen in Fig. 1c
, the ddT reaction alone is sufficient to
differentiate between mutant and wild-type states at each codon. For
the first sample, the codon 553 primer is converted only to a 9277.9-Da
product consistent with wild-type; the codon 542 primer is converted to
a pair of peaks at 9952.1 and 7371.2 Da. Thus this sample is homozygous
wild-type at codon 553 and heterozygous for the codon 542 mutation.
Similar analysis for the sample in Fig. 3
b shows it to be homozygous wild-type at codon 542 and
heterozygous for the codon 553 mutation (Table 1
, Fig. 1c
). Although
MALDI-TOF MS to date is not considered a quantitative analytical method
for DNA analysis, the relative peak heights of the reaction products
corresponding to the heterozygous position are in both of the Fig. 3
samples about one-half of those corresponding to the homozygous
position. These are the peaks with the masses 7371.2 and 9952.1 Da
(heterozygous site for codon 542, Fig. 3a
) compared with the peak with
the mass 9277.9 Da (homozygous wild-type site for codon 553, Fig. 3a
)
and the peaks with the masses 6149.4 Da and 9275.4 Da (heterozygous
site for codon 553, Fig. 3b
) compared with the peak with the mass
9940.0 Da (homozygous wild-type site for codon 542, Fig. 3b
).
|
As a third example we tested the efficiency of the method in
determining the number of thymidines in the intron 8 splice acceptor
site; the number of thymidines is relevant to the degree of exon 9
skipping during maturation of the CFTR mRNA (27). Each of
the three common alleles (T5, T7, T9) in Fig. 1d
are represented in
samples from two different patients. Mr(exp)
values from the first (Fig. 4
a) were consistent with a T5/T9 heterozygote genotype, while
those from the second (Fig. 4b
) were consistent with T7/T9. For both,
measured masses were in good agreement with
Mr(calc) values of 6890.6 (T5), 7515.0 (T7), and
8139.4 Da (T9).
|
The effect of extension by modified (28), rather than by
normal, purines is demonstrated on comparison of Fig. 4
, b and c. The
extension reaction of the template heterozygous for the T7/T9 alleles
was performed with (Fig. 4b
) and without (Fig. 4c
) N-7 deazaadenine. In
the latter, signal from fragmented (i.e., MALDI-induced depurination)
extension products is far more intense, decreasing the total
signal-to-noise ratio of molecular ions by distributing signal into
several peaks.
Calculated vs observed Mr values are listed in
Table 1
. The accuracy of Mr(exp) from all
experiments ranged from -5.6 to +10.5, in all cases within 0.1% of
Mr(calc). Thus, a definitive interpretation of
the results is allowed in each case, because the mass difference of the
diagnostic products of the PROBE method specific for wild-type or
mutation is at least one nucleotide (average mass 309 Da).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
For high-quality DNA diagnostics, sequencing is the ideal analytical procedure in terms of detection limits and specificity; however, with current methods it remains too cumbersome for routine use in high-volume clinical screening programs. The PROBE assay in principle gives the same information for known mutations as obtained by diagnostic sequencing, albeit with far less effort. A mutation detection primer is elongated with three deoxynucleotides and terminated by the dideoxynucleotide not present in the deoxy form. Thus in all tests single-sequencing ladder molecules specific to either wild-type or mutant alleles are obtained. The solid-phase minisequencing procedure (10)(11)(12), in which a diagnostic primer is extended by a single labeled base, is a related method that has been successfully utilized for detection of alterations in a variety of genes. Initially, radioactive nucleotides were used, and the incorporated radioactivity was measured (10); subsequently, multiplexed formats were achieved by using primers of different length and fluorescent dideoxynucleotides, followed by electrophoretic separation of the products (12). Minisequencing as described thus far is not applicable to loci in which several adjacent mutations are possible, such as codons 506–508 in exon 10 of the CFTR gene, or to cases where the number of identical adjacent bases is relevant, such as the CFTR intron 8 poly(T) tract.
The PROBE assay clearly differentiates known codon 506–508 mutations
and quantitates exactly the number of identical bases in a sequence of
consecutive T nucleotides, demonstrating the resolution of PROBE and MS
detection. The MS evaluation of PROBE analytes does not require labeled
DNA; the mass of the product itself provides a far more analytically
specific signal than a base-specific color found with fluorescent
detection, the product size of which can be estimated only by comparing
its relative electrophoretic mobility with that of an internal
calibrator. Using external calibration, the maximum deviation of
Mr(exp) from Mr(calc) was
0.09% (wild-type F508 ddCTP termination: expected 11 604.6 Da,
observed 11 615.1 Da). Although the resolution of the reflectron TOF
utilized in this study is too low for differentiation of detection
products with very similar masses (i.e., A vs T, m 9.0
Da, ddC termination reaction of I507 vs F508, Fig. 2h
), the
recent reintroduction of delayed ion extraction instrumentation has
enabled resolution improvements of an order of magnitude on
oligonucleotides in this mass range (30).
In conclusion, the PROBE assay with the evaluation of its products by MS is a promising replacement for conventional mutation detection systems; diagnostic product masses represent a highly definitive and unambiguous signal compared with a relative electrophoretic mobility. The MALDI-TOF technology has now been developed to a state where it can be applied to DNA diagnostics, and current efforts on improved sample preparation, automation of laser scanning and data collection, and more extensive multiplexing are directed at making this the future method of choice for high-throughput, analytically robust DNA diagnostics.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
F508 mutation of the cystic fibrosis gene by matrix-assisted laser desorption/ionization mass spectrometry. Rapid Commun Mass Spectrom 1995;9:772-774.
[ISI][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Sturm, S. Quinten, C. G. Huber, and O. Kohlbacher A statistical learning approach to the modeling of chromatographic retention of oligonucleotides incorporating sequence and secondary structure data Nucleic Acids Res., June 13, 2007; (2007) gkm338v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Rej Clinical Chemistry through Clinical Chemistry: A Journal Timeline Clin. Chem., December 1, 2004; 50(12): 2415 - 2458. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Takatsu, T. Yokomaku, S. Kurata, and T. Kanagawa A new approach to SNP genotyping with fluorescently labeled mononucleotides Nucleic Acids Res., April 15, 2004; 32(7): e60 - e60. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Meyer, A. Fredriksen, and P. M. Ueland High-Level Multiplex Genotyping of Polymorphisms Involved in Folate or Homocysteine Metabolism by Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry Clin. Chem., February 1, 2004; 50(2): 391 - 402. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Blondal, B. G. Waage, S. V. Smarason, F. Jonsson, S. B. Fjalldal, K. Stefansson, J. Gulcher, and A. V. Smith A novel MALDI-TOF based methodology for genotyping single nucleotide polymorphisms Nucleic Acids Res., December 15, 2003; 31(24): e155 - e155. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Wang, J. Milunsky, M. Yamin, T. Maher, R. Oates, and A. Milunsky Analysis by mass spectrometry of 100 cystic fibrosis gene mutations in 92 patients with congenital bilateral absence of the vas deferens Hum. Reprod., August 1, 2002; 17(8): 2066 - 2072. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Shahgholi, B. A. Garcia, N. H. L. Chiu, P. J. Heaney, and K. Tang Sugar additives for MALDI matrices improve signal allowing the smallest nucleotide change (A:T) in a DNA sequence to be resolved Nucleic Acids Res., October 1, 2001; 29(19): e91 - e91. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Amexis, P. Oeth, K. Abel, A. Ivshina, F. Pelloquin, C. R. Cantor, A. Brau, and K. Chumakov Quantitative mutant analysis of viral quasispecies by chip-based matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry PNAS, September 26, 2001; (2001) 211423298. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Abdi, E. M. Bradbury, N. Doggett, and X. Chen Rapid characterization of DNA oligomers and genotyping of single nucleotide polymorphism using nucleotide-specific mass tags Nucleic Acids Res., July 1, 2001; 29(13): e61 - e61. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Douglas, M. R. Erdos, R. M. Watanabe, A. Braun, C. L. Johnston, P. Oeth, K. L. Mohlke, T. T. Valle, C. Ehnholm, T. A. Buchanan, et al. The Peroxisome Poliferator-Activated Receptor-{gamma}2 Pro12Ala Variant: Association With Type 2 Diabetes and Trait Differences Diabetes, April 1, 2001; 50(4): 886 - 890. [Abstract] [Full Text]
|
|
|
|
|
|
K. H. Buetow, M. Edmonson, R. MacDonald, R. Clifford, P. Yip, J. Kelley, D. P. Little, R. Strausberg, H. Koester, C. R. Cantor, et al. High-throughput development and characterization of a genomewide collection of gene-based single nucleotide polymorphism markers by chip-based matrix-assisted laser desorption/ionization time-of-flight mass spectrometry PNAS, December 28, 2000; (2000) 21506298. [Abstract] [Full Text]
|
|
|
|
|
|
E. Nordhoff, C. Luebbert, G. Thiele, V. Heiser, and H. Lehrach Rapid determination of short DNA sequences by the use of MALDI-MS Nucleic Acids Res., October 15, 2000; 28(20): e86 - e86. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Sun, H. Ding, K. Hung, and B. Guo A new MALDI-TOF based mini-sequencing assay for genotyping of SNPS Nucleic Acids Res., June 15, 2000; 28(12): e68 - e68. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Muller, D. van den Boom, D. Zirkel, H. Koster, F. Berthold, M. Schwab, M. Westphal, and W. Zumkeller Retention of imprinting of the human apoptosis-related gene TSSC3 in human brain tumors Hum. Mol. Genet., March 22, 2000; 9(5): 757 - 763. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. H.L. Chiu and C. R. Cantor Mass Spectrometry of Nucleic Acids Clin. Chem., September 1, 1999; 45(9): 1578 - 1578. [Full Text] [PDF]
|
|
|
|
|
|
K. Tang, D.-J. Fu, D. Julien, A. Braun, C. R. Cantor, and H. Koster Chip-based genotyping by mass spectrometry PNAS, August 31, 1999; 96(18): 10016 - 10020. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Harksen, P. M. Ueland, H. Refsum, and K. Meyer Four Common Mutations of the Cystathionine {beta}-Synthase Gene Detected by Multiplex PCR and Matrix-assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Clin. Chem., August 1, 1999; 45(8): 1157 - 1161. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. B. Fields Mass-Spectrometric Approaches for DNA-Based Genetic Screening Clin. Chem., July 1, 1997; 43(7): 1108 - 1109. [Full Text] [PDF]
|
|
|
|
|
|
K. H. Buetow, M. Edmonson, R. MacDonald, R. Clifford, P. Yip, J. Kelley, D. P. Little, R. Strausberg, H. Koester, C. R. Cantor, et al. High-throughput development and characterization of a genomewide collection of gene-based single nucleotide polymorphism markers by chip-based matrix-assisted laser desorption/ionization time-of-flight mass spectrometry PNAS, January 16, 2001; 98(2): 581 - 584. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Amexis, P. Oeth, K. Abel, A. Ivshina, F. Pelloquin, C. R. Cantor, A. Braun, and K. Chumakov Quantitative mutant analysis of viral quasispecies by chip-based matrix-assisted laser desorption/ ionization time-of-flight mass spectrometry PNAS, October 9, 2001; 98(21): 12097 - 12102. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
