| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Nucleic Acids Research | Pages |
Dye structure affects Taq DNA polymerase terminator selectivity
Introduction
Materials And Methods
Enzymes
Nucleotide triphosphates and other materials
Synthetic oligodeoxyribonucleotides
Buffers
Product analysis
Measuring elementary rate constants
Measurement of the dissociation rate of the E·DNA complex
Selectivity assay format
Results And Discussion
Measuring dye effects on the nucleotide rate and binding constants for KlenTaq
Measuring dye effects on AmpliTaq FS selectivity
Dye effects on the pre-steady-state burst for AmpliTaq FS
Dye effects on the conformational change step
Conclusion
Acknowledgements
References
Dye structure affects Taq DNA polymerase terminator selectivity
ABSTRACT
INTRODUCTION
Fluorescent DNA sequencing can be broadly classified into either primer chemistry or terminator chemistry methods depending on whether the dye resides on the primer or on the terminator. Each method has its advantages and disadvantages. New modified forms of Taq Pol I having the F667Y substitution do not discriminate between deoxy- and dideoxynucleotides (1). Hence, primer chemistry peak height patterns show remarkable uniformity (2) which makes base calling and heterozygote detection much more reliable (3). A major disadvantage for primer chemistry methods is that four separate reactions are necessary and each reaction requires a different dye-labeled primer. Consequently, primer chemistry methods do not easily lend themselves to the high throughput demands of genomic DNA sequencing, for example.
Terminator chemistry methods are more amenable to high throughput applications (4). Major advantages include the need for only a single extension reaction for each template and the fact that any primer may be used in the reaction. Although terminator chemistry methods have also benefited from the development of F667Y versions of Taq Pol I plus the use of improved dye-labeled terminators (5), the peak height patterns in terminator chemistry profiles are still far less uniform compared to the profiles for primer chemistry methods. In this report, we investigate the mechanistic basis for the non-uniform peak height patterns observed for dye-labeled terminators.
Historically, fluorescent dye-labeled terminator chemistry was made possible by the observation that a propargylamino linker arm attached to the 5 position of pyrimidines and to the 7-deaza position of purines still allowed for relatively high dye terminator incorporation (6). While this linker arm was generally successful, only certain dye types could be used for certain enzymes. One of the first dye sets developed for T7 DNA polymerase consisted of propargylamino linkages to fluorescein dyes (7). The dye terminators developed for Taq Pol I, however, were based on rhodamines because fluorescein terminators were only poorly incorporated (8). Consequently, rhodamine terminators have been the focus for a wide variety of thermocycle sequencing applications; however, the requirement for this particular dye structure causes problems for reaction ‘clean-up’ and for electrophoretic separations. The exocyclic amino groups on rhodamines generate zwitterionic structures which cause the fluorescent by-products produced during thermocycling to migrate in the sequence ladders unless they are removed either by very careful ethanol precipitation or by the use of expensive size exclusion columns. Rhodamine dyes also show a tendency to form or stabilize DNA secondary structures that result in electrophoretic migration artifacts. Therefore, dGTP must be replaced with dITP to minimize gel compression (9). The presence of dITP results in inefficient use of the polymerase because this nucleotide behaves as a competitive inhibitor and its incorporation requires suboptimal reaction temperatures (10).
Fluorescein-labeled terminators would actually be more desirable for thermocycle sequencing. Fluorescein terminators yield very predictable migration patterns in acrylamide gels even when dGTP is used and any extension by-products migrate faster than the primer bands in sequence ladders. However, as described above, Taq Pol I only poorly incorporates fluorescein terminators. To understand how these dyes interact with the Taq Pol I, we have measured elementary rate and binding constants for a representative rhodamine terminator (6-TAMRA-ddTTP) and a representative fluorescein terminator (6-FAM-ddTTP). This investigation also led to the development of a new assay for quickly screening dye structural effects on polymerase selectivity. Our results suggest that it may be possible to develop fluorescein-labeled terminators for specially modified forms of Taq Pol I.
MATERIALS AND METHODS
Enzymes
‘KlenTaq,’ a truncated form of Thermophilus aquaticus DNA polymerase I (11), was obtained from K. A. Johnson (Penn State University). This deletion removes the entire 5[prime]->3[prime] nuclease region. A full-length form of Taq Pol I, AmpliTaq FSR (or G46D; F667Y), was a gift from D. H. Gelfand (Roche Molecular Systems, Alameda, CA). The G46D substitution renders the enzyme 5[prime]->3[prime] nuclease deficient. The F667Y substitution improves 2[prime],3[prime]-ddNTP incorporation (1).
Figure 1. Kinetic pathway model for competing nucleotides. The figure shows part of the kinetic mechanism that is thought to be common to all Pol I-type enzymes. In the case of a DNA sequencing reaction, substrate A would be a dye-labeled terminator with concentration [A] whereas substrate B would be an extendible base or 2[prime]-dNTP represented by concentration [B]. Since both molecules compete for the same active site, the preference or ‘selectivity ratio’ that the enzyme shows for one over the other is given by a ratio of rate and binding constants as shown in the equation. All four ddNTPs were purchased as premixed solutions from Amersham Pharmacia Biotech. Dye-labeled nucleotides were synthesized and characterized by the Organic Synthesis Group at PE Applied Biosystems (Foster City, CA). The dye-labeled terminators used in this study can be obtained from the author by written request. The primer and template oligodeoxynucleotides listed in Table 1 were synthesized on an Applied Biosystems 380A DNA Synthesizer (DNA Synthesis Facility, PE Applied Biosystems) and purified using reverse phase HPLC. Each primer strand was synthesized and dye labeled using standard chemistries recommended for this instrument. Concentrations of the single-stranded oligodeoxynucleotide fragments were determined by UV absorbance. Duplex primer/template pairs were formed by annealing equimolar amounts of the dye-labeled 25mer primer with the appropriate 36mer template in a solution containing 5 mM Tris-HCl, 5 mM NaCl and 0.2 mM EDTA (pH 8.0 at 20°C) using the following temperature regimen: 5 min at 95°C, 10 min at 60°C and 15 min at room temperature. Duplex DNAs were stored at -20°C. Table 1. 
Nucleotide triphosphates and other materials
Synthetic oligodeoxyribonucleotides
Buffers
All experiments were carried out using 80 mM Tris-HCl and 2.4 mM MgCl2 (pH 9.0 at 20°C) except where noted in the text. All reaction temperatures were 60°C.
Product analysis
The dye-labeled primer and product bands were resolved on 16% polyacrylamide/8 M urea denaturing gels and ‘visualized’ on a PE Biosystems Model 373 DNA Sequencer using 672 GeneScanRfragment analysis software for peak identification and relative fluorescence measurements. The amount of DNA in each product band was calculated by normalizing the product peak areas for each time point. The resulting curves were fitted to appropriate equations depending on the experiment (as described in the text) by regression analyses using curve fitting software purchased from Synergy Software (Reading, PA).
Measuring elementary rate constants
The nucleotide concentration dependence of the rate of 26mer product formation was measured under polymerase excess (500 nM) over DNA (100 nM) reaction conditions. Reaction times for each experiment varied depending on the type of nucleotide being tested as described in the text. Plots of product formation rates versus nucleotide concentrations yielded the maximum rate of phosphodiester bond formation, kpol, and the equilibrium dissociation binding constant, Kd (12,13). The designations for the kinetic constants refer to the enzyme reaction steps represented in Figure
Measurement of the dissociation rate of the E·DNA complex
By measuring the steady-state dissociation rate of the enzyme from an E·Ddd complex with or without the next correct dye-labeled nucleotide also present in the reaction mixture (under conditions where the chemistry step has been blocked), it was possible to infer whether or not dye/linker arm structure affects the ability of the polymerase to form a tighter binding or closed conformational (E*·Ddd·dye-Nuc) complex. A preincubated solution of polymerase (1 nM) and primer/template (1000 nM) was reacted with 2.4 mM Mg2+ and 400 µM ddT1TP (the correct nucleotide for the first template position following the 3[prime]-end of the primer strand or ‘N1’) either alone or in the presence of 400 µM ddN2TP (the next correct, unlabeled nucleotide) or the presence of 400 µM dye-ddN2TP (next correct dye-labeled nucleotide) or 400 µM incorrect nucleotide (i.e. one incapable of forming a correct Watson-Crick base pair in the ‘N2’ template position). Reaction temperatures were 60°C. Timed samples were removed as indicated in the figures and quenched in ice-cold 0.48 M EDTA (final concentration). These quenched samples were processed as described below for the selectivity assay to remove excess unincorporated nucleotides and samples were loaded onto 16% denaturing gels as described above. The steady-state ddTMP incorporation rate measured in the absence of the next correct nucleotide represented the rate of dissociation of the E·Ddd complex. The ddTTMP incorporation rate measured in the presence of the next correct nucleotide corresponded to the dissociation rate of the ternary E*·Ddd·ddN2TP or ternary complex which represents the kinetic step immediately prior to the ‘chemistry’ or group transfer step in the reaction (14).
Selectivity assay format
A substrate selectivity assay was developed to measure any preference that a given enzyme showed for an unlabeled terminator over a dye-labeled terminator. This was a true selectivity assay since both types of terminators were present in the same reaction mixture and were, of course, able to compete for the same active site under the same reaction conditions. The kinetic steps involved in determining a selectivity ratio are presented in the pathway shown in Figure
Since dye-labeled terminators have higher molecular weights than their unlabeled counterparts, two different types of 26mers were produced with regard to gel mobilities: the 26mer product resulting from incorporation of the unlabeled terminator migrated as expected for a 26mer fragment; however, the other 26mer which was the product of dye-labeled terminator incorporation migrated more slowly behaving as an ‘apparent 27mer’. Consequently, there were typically three bands in each gel lane: unreacted 25mer primer; 26mer (or ddNMP product); ‘27mer’ (or dye-ddNMP product). The relative fluorescence in each of the bands was measured as described above. The ratio of normalized fluorescence in the 26mer band divided by the normalized signal in the ‘apparent 27mer’ band represented the selectivity ratio or the preference that the enzyme showed for the unlabeled terminator over the corresponding dye-labeled terminator. A selectivity ratio of unity was interpreted to mean that the enzyme could not distinguish between an unlabeled terminator and the same molecule containing a dye. A selectivity ratio greater than one was interpreted to mean that the enzyme preferred the unlabeled over the dye-labeled terminator.
During the development of this assay, our early experiments utilized a 5[prime]-32P-labeled primer. However, to avoid having to resynthesize radioactive primers, we chose to use 5[prime]-dye-labeled primers. Since a dye was present on the 5[prime]-end of the primer for detection and quantitation purposes, it was necessary to correct for any energy transfer between the detection dye and the dye on the newly incorporated terminator on the 3[prime]-end of the ‘apparent 27mers’ in the gel results (16). It was possible to empirically determine an energy transfer correction factor by using standards on the same gel. For example, to measure 6-FAM-ddNTP incorporation, it was necessary to use 5[prime]-TAMRA-labeled primers; 6-FAM ([lambda]max emission = 522 nm) transfers energy to TAMRA ([lambda]max excitation = 560 nm). Therefore, the raw TAMRA signals in each ‘apparent 27mer’ band were artificially high due to the presence of a FAM moiety on the same molecule and had to be adjusted. In this example, the correction factor was determined by loading a separate gel lane with a sample containing equal moles of 5[prime]-TAMRA-25mer (detection dye only or ‘single-dye species’) plus 5[prime]-TAMRA-26-FAM-labeled fragment (equivalent to the two-dye ‘apparent 27mer’ or ‘two-dye species’). The ratio of the relative fluorescence signals in the two-dye band over the single-dye band provided a correction factor for dye-to-dye energy transfer.
While a TAMRA/FAM combination caused an enhancement of the TAMRA signal (by 1.6-fold), other dye combinations caused quenching of the 5[prime] detection dye signal. Nevertheless, in each case, it was still possible to generate suitable correction factors in the same manner as described above.
Selectivity assays were conducted under steady-state reaction conditions, namely 1000 nM primer/template, 1 nM enzyme, 400 µM each nucleotide (except where noted in the text), 2.4 mM Mg2+ at 60°C. Timed samples were removed and quenched in ice-cold 0.48 M EDTA (final concentration). To remove excess unincorporated dye-labeled nucleotide, an appropriate aliquot of the quenched sample was transferred to 250 µl lithium/tRNA precipitation solution (0.8 M LiCl plus 0.2 µg/µl Escherichia coli tRNA carrier) and 750 µl 95% ethanol. Total nucleic acids were precipitated on ice for a minimum of 20 min and then pelleted. Pellets were washed with 70% ethanol, air dried, and dissolved in 50% formamide for loading on 16% polyacrylamide/8 M urea denaturing gels (12).
Figure 2. Dye-labeled terminator structures. The figure shows the structures for 6-TAMRA-ddTTP, 6-FAM-ddTTP, 5-TET-ddCTP and 5-HEX-deaza-ddGTTP, where R stands for 2[prime],3[prime]-dideoxyribose-5[prime]-triphosphate. Each dye consists of an ‘upper’ or xanthene triple ring structure plus a ‘bottom’ or benzoic acid ring. The propargylamino linker arm is attached to the bottom ring of the dye at one of two positions as indicated which defines the isomers examined in this study. Each ‘rung’ in a sequencing ‘ladder’ represents the incorporation of a terminator rather than an extendible base. Since both of these molecules must compete for the same active site at each template position, the probability of a termination event versus an extension event is governed by enzyme selectivity as defined by the equation in Figure Figure To be consistent with our previous kinetic studies on terminator structural effects (12), we made our initial measurements using KlenTaq. F667 versions of Taq Pol I like KlenTaq show a strong bias against 2[prime],3[prime]-ddNTP incorporation. As shown in Table 2, KlenTaq would be expected to strongly favor incorporation of 2[prime]-dTTP over unlabeled 2[prime],3[prime]-ddTTP by approximately 3100-fold as predicted by the calculated substrate/terminator selectivity ratio. 6-TAMRA attached to ddTTP via a propargylamino linker arm (Fig. Table 2. Figure 3. Pre-steady-state burst kinetics for dye-labeled terminator incorporation by AmpliTaq FS. A pre-incubated solution of AmpliTaq FS polymerase (100 nM) plus 5[prime]-dye-labeled duplex DNA (500 nM) was mixed with Mg2+ plus dye-labeled ddTTP (160 µM) as indicated in a rapid chemical quench flow apparatus (all concentrations are given as final concentrations following mixing in the instrument). (A) The plot shows the time course incorporation pattern for TAMRA-ddTTP ([closed square]). The data were fitted to a burst kinetic equation [kobs = A(1 - exp-rt) + ksst], where kobs is the observed rate of the reaction, A is the burst amplitude in nM which is proportional to the Enz·DNA concentration, r is the burst rate, t is the time in seconds and kss is the steady-state turnover rate. The burst amplitude was 100 ± 10 nM with a burst rate of 60 ± 12 s-1 and a steady-state rate of 7 ± 1 s-1. (B) The plot shows the time course incorporation for FAM-ddTTP ([closed triangle]). All reaction conditions were the same as those described above. The plot shows a fit to a single exponential curve with a burst amplitude of 300 ± 8 nM and an initial rate of 8 ± 1 s-1. To determine if the behavior exhibited by 6-FAM-ddTTP was peculiar to this dye terminator structure or if it is, in fact, a general property of fluorescein dyes, we developed a rapid gel-based selectivity assay as described in Materials and Methods using AmpliTaq FS. Because it is an F667Y version of Taq Pol I, AmpliTaq FS does not show a strong bias against terminator incorporation (1) which made it possible to investigate the kinetic effects of these dye and linker arm combinations independently from ‘terminator structural effects’. AmpliTaq FS has also become the polymerase of choice for high throughput applications and is widely used in DNA sequencing (21,22) and fragment analyses (3). Experimentally determined selectivity values for several different dye and base combinations are listed in Table 3 which shows that the actual value for unlabeled ddTTP over 6-TAMRA-ddTTP was indeed as predicted. The actual value measured for ddTTP over 6-FAM-ddTTP was 50 times which agrees well with the calculated value of 43 times for KlenTaq. The ‘fluorescein dye effect’ depended on the type of base involved. The purines 6-FAM-ddATP, 5-TET-ddGTP and 5-HEX-ddGTP showed selectivity values of 1 behaving as if these dyes and their linkers were also transparent to AmpliTaq FS. Since the dyes tested represented a range of structures as well as both isomers, these data suggest that many different fluorescein dyes may be accommodated by AmpliTaq FS without significantly affecting incorporation when a propargylamino linker arm is used to attach either isomer to a purine. The situation for the same fluorescein dyes on pyrimidines was quite different. All combinations showed selectivity values much greater than 1 indicating that AmpliTaq FS shows a very strong preference for an unlabeled pyrimidine terminator over its fluorescein-labeled counterpart. The dyes in this study are composed of two ring systems, a xanthene (or ‘upper’) triple ring structure plus a benzoic acid (or ‘bottom’) ring as shown in Figure Table 3. The fact that the nucleotide dissociation constants for unlabeled ddTTP and 6-FAM-ddTTP were the same (within experimental error) as shown in Table 2 but 6-FAM-ddTTP showed a much lower rate constant indicated that the rate limiting step or ‘fluorescein dye effect’ must occur following initial nucleotide binding. Therefore, our first mechanism measurements involved determining whether or not dye-labeled terminators showed pre-steady-state burst curves under single nucleotide incorporation conditions. Pol I-type enzymes typically show burst kinetic patterns for 2[prime]-dNTP incorporation (13) meaning that the rate limiting step during polymerization must occur after the chemistry step in the reaction. On the other hand, if the time course does not show a burst, the rate limiting step must occur before or during the chemistry step. As shown in Figure
Figure 4. Measurement of the dissociation rate for the Enz·DNA·Nuc complex. (A) A preincubated solution of AmpliTaq FS (1 nM) plus 5[prime]-(FAM)-25/36AC primer/template was mixed with 2.4 mM MgCl2 plus 400 µM ddTTP, either alone or in the presence of 400 µM ddCTP (next incorrect nucleotide) or 400 µM ddGTP (the next correct nucleotide) or 400 µM HEX(II)·ddGTP (the next correct, fluorescein-labeled nucleotide). The steady-state rates of incorporation were: 1.5 ± 0.06 s-1 for dissociation of the Enz·DNAddT binary complex (ddTTP alone; [closed circle]) and 1.2 ± 0.04 s-1 in the presence of an additional nucleotide that cannot form a Watson-Crick base pair in the next template position (ddTTP + ddCTP; [cir]). The dissociation rate of the ternary complex, Enz·DNAddT·ddGTP, was slower or 0.15 ± 0.01 s-1 suggesting the formation of a tighter binding complex when the next correct nucleotide was also present in the reaction mixture (ddTTP + ddGTP; [closed square]). The presence of a fluorescein dye attached viathe propargylamine linker arm did not affect the ability of the polymerase toform a more slowly dissociating ternary complex (ddTTP + 6-HEX·ddGTP,0.20 ± 0.1 s-1; [open square]). (B) The experiment was conducted in the same manner as the one described above except that the primer/template was 5[prime]-(FAM)-25/36AG as shown in Table 1 so that the next correct nucleotide was a pyrimidine instead of a purine. The steady-state rates of incorporation were: 0.62 ± 0.01 s-1 (ddTTP-alone; [closed circle]) and 0.59 ± 0.02 s-1 with the next incorrect terminator also present in the reaction mixture (ddTTP + ddGTP; [open circle]). The steady-state turnover rate when the next correct nucleotide was present was reduced to 0.39 ± 0.01 s-1 (ddTTP + ddCTP; [closed square]). The dissociation rate of the ternary complex was about 3-fold faster than ‘ddTTP alone’ when the next correct nucleotide had an attached fluorescein dye or 1.8 ± 0.1 s-1 (ddTTP + TET·ddCTP; [open square]). While the plot for of 6-TAMRA-ddTTP incorporation followed a burst kinetic pattern, the curve for 6-FAM-ddTTP did not. As shown in Figure 
RESULTS AND DISCUSSION
Measuring dye effects on the nucleotide rate and binding constants for KlenTaq
A
B
Measuring dye effects on AmpliTaq FS selectivity
Dye effects on the pre-steady-state burst for AmpliTaq FS
A
B
Dye effects on the conformational change step
By measuring the steady-state turnover rate of the Enz*·DNA·Nuc under conditions where the chemistry step has been blocked, it is possible to investigate the conformational change step in polymerase reactions (12,14). We have applied this assay to kinetically probe dye effects on the conformational change step of AmpliTaq FS. These experiments consisted of measuring the steady-state rate of incorporation of ddTTP in the absence or in the presence of the next correct terminator (with and without a fluorescein dye). This was accomplished by reacting a solution containing enzyme (1 nM) plus excess 5[prime]-dye-labeled primer/template (1000 nM) with ddTTP alone (400 µM) or in the presence of one of the other nucleotides (each 400 µM) as shown in Figure
As shown in Figure
The steady-state turnover behavior for fluorescein dyesattached to pyrimidines was quite different. As shown in Figure
The dissociation rate for the ddTTP + 5-TET-ddCTP case in Figure
While these results cannot rule out an additional ‘fluorescein dye effect’ on the chemistry step in the reaction, they do indicate significant interference with the ability of the polymerase to undergo a conformational change. More detailed kinetic/mechanism studies will be required to investigate any possible interference during the group transfer reaction.
CONCLUSION
Previous studies have shown that F667 versions of Taq Pol I exhibit a strong bias against 2[prime],3[prime]-ddNTP terminators (12). Results presented in this paper show that this bias was unaffected by a representative rhodamine dye but exacerbated by several different fluoresceins. 6-TAMRA-ddTTP showed equivalent nucleotide rate and binding constants to unlabeled ddTTP for KlenTaq suggesting that this may be the reason why rhodamine terminators were developed first for Taq Pol I DNA sequencing methods. A representative fluorescein terminator, 6-FAM-ddTTP, on the other hand, showed a 40-fold lower rate constant but no change for the binding constant indicating that fluorescein dyes make dye-labeled terminator incorporation worse by affecting a step following ground-state nucleotide binding. Additional kinetic measurements using an F667Y version of Taq Pol (AmpliTaq FS) indicated that fluorescein dyes interfere with the ability of this enzyme to form a closed ternary complex following initial nucleotide binding but only when these dyes are attached to pyrimidines. The nature of this interaction is unknown. However, since it is specific for fluorescein dye structures and the larger the dye, the more the bias against its incorporation, this interaction must have both chemical and steric components. Detailed molecular modeling and/or crystallography studies of an active, conformationally closed, ternary complex that includes a fluorescein-labeled pyrimidine terminator may show where and how these dyes contact the enzyme. By minimizing or eliminating these interactions, it should be possible to develop fluorescein terminator chemistries for modified versions of Taq DNA polymerase that do not discriminate between deoxy- and dideoxynucleotides.
ACKNOWLEDGEMENTS
The author wishes to thank Sandy Spurgeon for excellent DNA sequencing trace analyses and Shaheer Khan for dye-terminator syntheses. Appreciation for excellent technical discussions, suggestions and advice also go to Curtis Bloom, Elena Bolchokova, Kathy Perry, James Rozzelle and Roger O’Neill.
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 26 Mar 1999
Copyright©Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
A. Bart, M. W. J. van Passel, K. van Amsterdam, and A. van der Ende
Direct detection of methylation in genomic DNA
Nucleic Acids Res.,
August 9, 2005;
33(14):
e124 - e124.
[Abstract]
[Full Text]
[PDF]
A. F. Gardner, C. M. Joyce, and W. E. Jack
Comparative Kinetics of Nucleotide Analog Incorporation by Vent DNA Polymerase
J. Biol. Chem.,
March 19, 2004;
279(12):
11834 - 11842.
[Abstract]
[Full Text]
[PDF]
W. A. Wlassoff and G. C. King
Ferrocene conjugates of dUTP for enzymatic redox labelling of DNA
Nucleic Acids Res.,
June 15, 2002;
30(12):
e58 - e58.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (120K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (9)
Request Permissions
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Brandis, J. W.
Search for Related Content
PubMed
PubMed Citation
Articles by Brandis, J. W.
Related Collections
DNA characterisation
Social Bookmarking
What's this?