| Nucleic Acids Research | Pages |
The integral divalent cation within the intermolecular purine*purine·pyrimidine structure: a variable determinant of the potential for and characteristics of the triple helical association
Introduction
Materials And Methods
Human dhfr promoter fragment
Oligodeoxyribonucleotides
Triplex formation
DNase I protection assay
Laser densitometric analysis and evaluation
Results
Variable support for intermolecular pur*pur·pyr assembly
Stringency of triple helical alignment
Influence of the cation on affinity of the third strand for its target
Cation potency
Inhibition of triplex formation by divalent cations
Triple helical stability and cation efflux
Discussion
Support of intermolecular pur*pur·pyr triplex formation
Dependence of triplex formation on oligonucleotide(third strand) concentration
Stringency of triple helical alignment
Dependence of triplex formation on metal(II) concentration (cation potency)
Efflux of supportive divalent cations from the composite triple helical structure
Antagonism of triplex formation
Potential biological relevance
Acknowledgements
References
The integral divalent cation within the intermolecular purine*purine·pyrimidine structure: a variable determinant of the potential for and characteristics of the triple helical association
ABSTRACT
INTRODUCTION
Under certain constraints of sequence composition and environmental conditions, a segment of double-stranded DNA may be bound specifically by a third nucleic acid strand lying within the major groove (1-4). Triplex formation is analogous to Watson-Crick hybridization in its dependence upon (i) base-specific hydrogen bonding (5-7) and (ii) counterions to neutralize the charges of the deoxyribose-phosphate backbones in order to overcome the electrostatic repulsion of nucleic acid strands (8-11). Viable targets for spontaneous triplex formation are characterized by a predominance of purines on one strand and pyrimidines on the other. This purine·pyrimidine (pur·pyr) asymmetry is associated with major groove dimensions capable of accommodating a third strand (12-13), and facilitates alignment for the formation of multiple specific non-Watson-Crick hydrogen bonds with bases of the third strand. Assembly of the purine*purine·pyrimidine class of triple helix, based predominantly on antiparallel G*G·C alignments (where * represents the non-Watson-Crick interaction), generally requires the presence of magnesium(II) cations (14-17). The precise mode of coordination of the divalent cation within the triple helical structure has not yet been determined.
This laboratory has investigated the assembly of intermolecular purine*purine·pyrimidine triple helical structures at the human dihydrofolate reductase (dhfr) core promoter (18-19). This sequence contains two closely spaced and very similar regions of purine·pyrimidine asymmetry. Synthetic oligonucleotides were designed to bind in antiparallel orientation specifically to either of these target sequences, producing triple helical structures dominated by G*G·C alignments, but tolerating individual A*A·T as well as C*C·G (hydrogen bonding not necessarily inferred) alignments. Here we have employed quantitative DNase I protection titrations to investigate the relationship of the required divalent cation to the intermolecular pur*pur·pyr triplex. The cationic component was found to be continuously necessary for maintenance of triple helical integrity and to exert a considerable yet variable influence on the capacity for triplex association, third strand affinity, stringency of triple helical alignment and destabilization of the triple helical structure secondary to cation efflux. These findings may have important implications for understanding triplex formation as it might occur naturally within the cell and may be also relevant to triplex-based therapeutic antigene strategies.
MATERIALS AND METHODS
Human dhfr promoter fragment
The human dhfr promoter sequence (-112 to +56) containing the two purine·pyrimidine biased regions (-58 to -40 and -25 to -5) was obtained as described (19). This restriction fragment was excised and 3[prime]-32P-end-labeled on either the purine-rich or pyrimidine-rich strand, to optimize electrophoretic/autoradiographic visualization of the proximal or distal pur·pyr region respectively. Following isolation on a non-denaturing polyacrylamide gel, the labeled double-stranded dhfr promoter fragment was precipitated from a 1 M ammonium acetate solution, dried, then resuspended and stored in distilled, deionized water.
Oligodeoxyribonucleotides
The oligonucleotides utilized for intermolecular triplex formation were prepared as described (19). The sequences of the oligonucleotides, along with the human dhfr promoter sequence to which each is specifically targeted (plus 10 bp flanking sequences on either side) are as follows: dist-14a, 3[prime]-CGGGGCGGGGGGGC-5[prime] (target sequence underlined); distal target, 5[prime]-...CTGCGTGCGCCGGGGCGGGGGGGCGGGGC CTCGCCTGCA...-3[prime]; prox-F, 3[prime]-CGAGGGGGGCGGGGC-5[prime] (target sequence underlined); prox-G, 3[prime]-TGGGGACGAGGGGGGC-5[prime] (target sequence italic); proximal target, 5[prime]-...CCTGCACAAA TGGGGACGAGGGGGGCGGGGC GGCCACAATT...-3[prime].
For the purposes of this manuscript, triple helical `alignment' refers to the colinear antiparallel positioning of residues of the third strand along the major groove surface of the native double helix such that exclusively G*G·C, A*A·T and C*C·G triplets would result (19).
Triplex formation
The standard reaction contains the labeled dhfr promoter fragment (75-200 × 103 c.p.m./sample, ~40 nM), an oligonucleotide (32-40 µM), Tris-HCl, pH 7.2 (20 mM) and a divalent metal chloride (10 mM). When carried out for 45 min at room temperature, this incubation allows the system to approach equilibrium, which in the presence of Mg2+ results in triplex formation on a very high (>90%) proportion of specific target sites.
DNase I protection assay
Following incubation to allow for triplex formation, samples were placed on ice, then subjected to limited digestion with DNase I (2-40 U/ml for 45 s). The DNase I activity required was determined empirically under the different reaction conditions (particularly cationic composition) to achieve comparable degrees of cleavage and produce an even distribution of bands representing large and small digestion products of the dhfr promoter fragment. Labeled digestion products were separated on an 8% polyacrylamide, 8 M urea sequencing gel.
Laser densitometric analysis and evaluation
The autoradiographic results of DNase I protection assays were analyzed by laser densitometry. Scanning, peak area integration and controlled evaluation of numerical data were performed as described in detail (20). The relative intensities of (experimental) bands within the target, corrected for regional intensity of the digest as determined by vicinal reference bands, were compared from lane to lane as a measure of proportionate occupation of the target site on the population of dhfr promoter molecules by the oligonucleotide third strand.
RESULTS
Variable support for intermolecular pur*pur·pyr assembly
In the experiment shown in Figure
Figure 1. Variable support by divalent cations for assembly of an intermolecular triple helix at the proximal pur·pyr region of the human dhfr promoter. The oligonucleotide prox-G (32 µM) was co-incubated with the 204 bp restriction fragment of the human dhfr promoter in the presence of a variable divalent metal chloride (10 mM). Following the incubation, samples were subjected to limited DNase I digestion and the products analyzed on an 8% denaturing polyacrylamide gel. A Maxam-Gilbert G+A sequencing reaction is included on the left (purine-rich strand labeled). The position of the proximal (specific) pur·pyr target is marked with a solid bracket. The position of the homologous distal (mismatched) sequence is marked with a dashed bracket. For each of the divalent cations tested, a negative control digest (no oligonucleotide binding, odd numbered lanes) is included. In the presence of Mg2+, the oligonucleotide prox-G produces a clear, characteristic footprint (19) over its intended target within the proximal pur·pyr region (lane 2). Definitive endonuclease protection was also observed when Co2+ (lane 4) or Mn2+ (lane 8) replaced Mg2+. The essentially complete disappearance of bands over an area approximating the specific target sequence indicates that nearly 100% of the population of double helical dhfr promoter molecules have been bound by the oligonucleotide third strand at this site when either Mg2+, Co2+ or Mn2+ was present. Furthermore, no significant change in the position and extent of the sequence protected from DNase digestion by the oligonucleotide was seen with change in the identity of the supporting divalent metal cation, consistent with a comparable mode of triplex formation in each case. In contrast, no clear indication of triplex formation was observed in the presence of Ni2+ (lane 6). Additional assays have indicated that Ca2+ is also supportive of intermolecular triplex formation at the proximal pur·pyr region, while Cd2+ is not (data not shown). Next, these divalent cations were tested for their ability to support triplex formation at the distal pur·pyr region of the dhfr promoter (Fig. Figure 2. Variable support by divalent cations for intermolecular triplex formation at the distal pur·pyr region of the human dhfr promoter. Binding of the oligonucleotide dist-14a to the dhfr promoter in the presence of 10 mM MgCl2, CoCl2, NiCl2, MnCl2 or CdCl2 was assayed by DNase I protection as described in the legend to Figure 1. A Maxam-Gilbert G+A sequencing reaction is included on the left (pyrimidine-rich strand labeled). The position of the distal (specific) target is marked with a solid bracket. The position of the homologous proximal (mismatched) sequence is marked with a dashed bracket. The distal and proximal purine·pyrimidine regions of the human dhfr core promoter are very similar sequences, differing only by the insertion of two individual A residues. In the presence of Mg2+, in addition to its proximal (specific) target, prox-G binds weakly to the homologous distal pur·pyr region producing a minor degree of DNase protection (a light or partial footprint; Fig. Examination of the homologous (mismatched) target sites in Figures A series of experiments was carried out to further evaluate the support for intermolecular pur*pur·pyr triple helical assembly by Mg2+, Co2+ and Mn2+. In the first of these, the effect of a limiting concentration of oligonucleotide on triplex formation was titrated in the presence of a constant and sufficient concentration of each of the divalent cations. Under otherwise standard conditions, allowing 45 min for each reaction to approach equilibrium, the relative degree of facilitation of the affinity of the third strand for its double-stranded target by each cation could be ascertained (Fig. Figure 3. Titration of triple helix assembly with limiting oligonucleotide (third strand) concentration or limiting divalent cation concentration. (A) The dhfr promoter fragment (~40 nM) was incubated with variable concentrations of the oligonucleotide prox-F in the presence of one of the supportive divalent cations (10 mM). (B) The dhfr promoter fragment was incubated with the oligonucleotide dist-14a (32 µM) in the presence of variable concentrations of divalent metal chlorides. For both (A) and (B), the samples were analyzed by DNase protection assay as described in Materials and Methods and in the legends to Figures 1 and 2 and the proportions of specific target sites bound by the oligonucleotide third strand were quantitated by laser densitometric analysis. [bull], Mg2+; [cir], Co2+; *, Mn2+. Table 1. Next, the dependence of triplex formation on the concentration of the metal cations themselves was investigated. A gradient of Mg2+, Co2+ or Mn2+ was used to titrate the binding of an oligonucleotide (in constant excess) to its specific target on the human dhfr promoter (Fig. Next, the supportive effect of Mg2+, Co2+ or Mn2+ was titrated against the non-supportive tendency of Ni2+ (Fig. Figure 4. Antagonism of triplex formation by Ni2+. The relative contents of a supportive divalent cation (Mg2+, Co2+ or Mn2+) and a non-supportive divalent cation (Ni2+) were varied in otherwise standard incubations of an oligonucleotide (prox-F or dist-14a) with the dhfr promoter fragment. Maxam-Gilbert G+A sequencing reactions are included for reference. The positions of the specific targets are marked with solid brackets. The positions of the homologous (mismatched) sequences are marked with dashed brackets. The experimental protocol chosen to measure triple helical stability relies on dilution of component nucleic acid species to minimize further triple helical association and thereby allow the rate of third strand dissociation from existing triple helical structures to be assayed. Initially, the oligonucleotide in modest excess (2-4 µM) was preincubated with the dhfr promoter fragment under standard conditions to allow for accumulation of triple helical structures on a high proportion of the target DNA molecules. At that point, the reaction was diluted sufficiently to decrease the concentration of the oligonucleotide to the extent (0.04-0.08 µM; Fig. In Figure Figure 5. Stability of an intermolecular pur*pur·pyr triple helix assembled in the presence of Mg2+, and perpetual dependence upon the supportive environmental cation pool. A population of triple helical structures was preassembled during a standard 45 min incubation of the dhfr promoter fragment with an oligonucleotide (prox-F, 4 µM) in the presence of 10 mM MgCl2. The free oligonucleotide concentration was then decreased to 0.08 µM by dilution with a 50× volume of a buffer of the same cationic composition as the original binding reaction. Lane 1, negative control (no preassembly), oligonucleotide added subsequent to dilution and incubated for 45 min; lane 2, positive control, following preassembly sample was diluted (20 mM Tris-HCl, pH 7.2, 10 mM MgCl2) and immediately removed to ice for DNase I digestion; lanes 3-5, following preassembly and dilution, samples were allowed to remain at ambient temperature for varying periods of time prior to DNase I digestion; lanes 6-8, the diluent contained no MgCl2, effectively decreasing the ambient Mg2+ concentration to 0.2 mM. The target sequence is marked by a bracket. Once formed, are these triple helical structures free of dependency on supportive divalent cations? Does a population of cations remain stably associated with the three-stranded structure? Are some divalent cations continually required as an integral part of the triple helical structure, yet are susceptible to loss to the environment? For lanes 6-8 of Figure This experiment was repeated, substituting MnCl2 for MgCl2, and the results of both experiments are presented graphically in Figure We have obtained evidence for at least three distinct modes of cation coordination which critically affect several parameters relating to intermolecular pur*pur·pyr triplex assembly (Table 2). Table 2. Assembly and stabilization of the intermolecular pur*pur·pyr triple helix is sufficiently promoted by divalent magnesium cations. Since association of the alkaline earth metals with DNA is essentially limited to electrostatic interactions with the anionic oxygen atoms of the phosphodiester backbone (24-25), it must be this activity as a counterion for phosphate charge by which Mg2+ enables the triple helical structure to form. The specific mode of phosphate coordination which supports the pur*pur·pyr structure remains to be determined. (An alkaline earth metal cation may potentially coordinate with one or both of the free oxygen atoms of the phosphate, oscillate between inner and outer sphere binding, migrate from one phosphate to another or coordinate to two phosphates simultaneously; 26-27.) A subset of other divalent metal cations (Ca2+, Mn2+ or Co2+) also support intermolecular triplex formation at either of the pur·pyr target sequences of the dhfr promoter. It is apparently the capability of electrostatic binding and phosphate charge neutralization shared with Mg2+ by which Ca2+, Mn2+ and Co2+ foster a qualitatively similar association of the third strand with its target. Titrations of oligonucleotide concentration indicate that the degree to which each of the supportive divalent cations facilitates the affinity of the third strand for its double helical target correlates roughly with affinity of the cation for phosphate. Mn2+, with a greater affinity for DNA phosphate than Mg2+, Ca2+ or Co2+ (24,28-31), most efficiently utilizes a limiting third strand concentration to promote triplex formation (also 32). Figure 6. Relative rates of dissociation of triple helical structures in the presence or absence of the supportive divalent cation pool. The experimental protocol utilized in Figure 5 was repeated for characterization of the stability of the <prox-F*proximal pur·pyr> structure assembled in the presence of Mn2+ and the quantitative results of both experiments are compared. For curves A and B, 10 mM MnCl2 or MgCl2 was supplied in the diluent, thus maintaining the cation concentration of the original binding reaction. For curves C and D, the cation pool was diluted without supplementation (20 mM Tris-HCl, pH 7.2 in H2O, final divalent cation concentration 0.2 mM). [bull], Mg2+; *, Mn2+. Relative to Mg2+, a modestly relaxed stringency of triple helical alignment is observed with Co2+ and particularly Mn2+, with a greater allowance for third strand binding to a homologous (but non-identical) target. The decreased triple helical stringency observed is limited to tolerance of unbound oligonucleotide termini (also 17,21,33); no change in position of specifically aligned structures and no clear evidence of actual mispairing is seen. The mismatched triple helical structures exhibit a lower binding affinity and lower stability (data not shown) relative to the specifically aligned triplex. This tendency towards enhanced facilitation of triplex formation at the expense of specificity also correlates with measured potentials of the divalent metal cations for electrostatic binding to phosphate: Ca2+ < Mg2+ < Co2+ < Mn2+ (24,28). Augmented neutralization of anionic phosphate charge may compensate for a lessening of affinity due to misaligned residues (26,34). The titrations of triplex formation with limiting cation concentration are consistent with the notion that a certain finite number of `spots' must be filled by the supportive divalent cation in order for triple helix assembly to occur. Co2+ and to a lesser extent Mn2+ retain the ability to effectively support triplex formation at considerably lower metal(II) concentrations than Mg2+. In addition to binding phosphate, the transition metal cations Co2+ and to a lesser extent Mn2+ are capable of coordinating to nucleophilic atoms of the nucleotide bases (28,31,35). The increased potency of Co2+ and Mn2+ for support of triplex formation would appear to be a function of an augmented attraction of these cations for one or both of the nucleic acid molecules, possibly through base binding or phosphate-base chelation. The consistently accelerated rate of third strand dissociation accompanying dilution of the ambient cation pool allows us to conclude that: (i) the divalent metal cations required for assembly continue to be necessary for maintenance of the triple helical association; (ii) at least a subpopulation of these integral supportive cations are susceptible to loss via diffusion to the environment. Magnesium facilitates a relatively stable triple helical association, yet triple helical integrity is almost instantaneously compromised if the ambient cation pool is not maintained, an effect apparently accounted for by rapid efflux of essential but labile electrostatically bound magnesium cations from the triple helical structure. A substantial slowing of this cation diffusion/triplex destabilization is seen when either Mn2+ or Co2+ is utilized as the supportive divalent cation. This would apparently invoke a novel (relative to Mg2+) and longer lived metal-nucleic acid interaction, such as an intrastrand chelate, with simultaneous coordination of an individual divalent cation to a phosphate and a heterocyclic atom (25-28,36,37), or perhaps an interstrand chelate, involving for instance guanine N-7 or O-6 (as observed for Mn2+ and Co2+ in tRNA; 38) of the G-rich oligonucleotide, which would provide a temporary crosslink of the third strand to the underlying duplex. Ni2+ and Cd2+ exhibit an electrostatic affinity for DNA phosphate greater than that of Co2+ (24,28), yet are essentially non-supportive and in fact actively antagonize the intermolecular pur*pur·pyr association. This suggests the existence of an additional, detrimental mode of base coordination which may prohibit or destabilize the triple helical association through interference with intermolecular hydrogen bonding, disruption of base stacking and/or disturbance of the base triplet geometry required for compatibility with triple helical structure. These divalent transition metals Ni2+ and Cd2+ are associated with a greater propensity to bind to the nucleotide bases than Mn2+ or Co2+ (24,26) and are more likely to coordinate by the stronger, more stable inner sphere mode (27,28). Evidence that cobalt displays both beneficial and detrimental activities toward triplex formation can also be ascertained from the data. The relative ratios of Kassoc for triplex formation at 0.5 mM Mg2+, Mn2+ or Co2+ are 1.0:3.1:11.9, respectively (Fig. We have previously observed this biphasic response across the transition series, with regard to the ability of these divalent cations to counteract the K+-associated induction/stabilization of quadruplex self-association of G-rich oligonucleotides and thereby facilitate triplex assembly in the presence of K+ (20). Without inducing discernible alterations in the circular dichroism spectra of the oligonucleotides, these cations potently suppress the rapid increase in molar ellipticity at 259-261 nm which is otherwise observed upon addition of potassium. In this manner, such alterations of the cationic environment allow triplex formation to effectively compete against quadruplex formation (which otherwise, in the presence of K+, sequesters the intended third strand population). This triplex-favorable property (K+/quadruplex resistance), like cation potency and relative resistance to cation efflux, initially increases across the transition metal series; however, with higher concentrations or further progression across the series, a distinctly detrimental (to triplex formation) mode of cation coordination becomes evident and potassium-resistant triplex formation is lost. The sequences of the human dhfr promoter which we have studied are representative of a pattern (G-rich purine tracts with isolated pyrimidine interruptions) which is very prevalent in the human genome and which comprise viable targets for the formation of stable, specific triple helical structures in vitro (17-20). The particular composition of these sequences (such as the prevalence of A*A·T triplets or alignment at pyrimidine interruptions) may have important ramifications for cation coordination and triplex assembly (39-41). These studies demonstrate ways in which the potential for triplex formation and the properties of the triple helical structure might be modulated. It has been suggested that triplex formation might be utilized naturally as a means of physiological regulation of molecular biological processes within the cell (42-50). Although free intracellular concentrations of the transition metal cations are considerably lower than those utilized in these experiments, it is conceivable that some of the triplex-modulatory effects we have observed in vitro with naked transition metal cations could be accomplished naturally in vivo by a specialized accessory polypeptide domain, perhaps through presentation of a coordinated metal cation (51) or through a particular arrangement of basic (cationic) amino acid residues (52). [As precedents for these concepts, consider that formation of the triple helical intermediate involved in homologous recombination requires an accessory polypeptide factor(s) (53,54) and that possible physiological roles of cations as allosteric effectors of other types of nucleic acid-nucleic acid interactions have been proposed (55,56).] It has also been proposed that administration of an exogenous oligonucleotide to induce intermolecular triplex formation site specifically on genomic DNA might be used as a means of therapeutically modulating the expression of specific genes (17,57-62). The data presented herein support the concept that rational modification of oligonucleotide structure, to include the permanent, site-specific incorporation of a cationic moiety, might be utilized to potentiate the therapeutic efficacy of the sequence-specific triple helical interaction (also 63-67). Such investigations are currently underway in this laboratory. This work was supported by National Institutes of Health/National Cancer Institute Grants CA42664 and CA54380 (DM) and CA75467 (SB) and the VA Medical Research Service (DM).
Stringency of triple helical alignment
Influence of the cation on affinity of the third strand for its target
Divalent cation
Kassoc ± SD (per M)
Mg2+
5.8 ± 0.5 × 106
Mn2+
3.5 ± 1.0 × 107
Co2+
3.7 ± 2.0 × 106
Cation potency
Inhibition of triplex formation by divalent cations
Triple helical stability and cation efflux
DISCUSSION
Support of intermolecular pur*pur·pyr triplex formation
Dependence of triplex formation on oligonucleotide(third strand) concentration
Stringency of triple helical alignment
Dependence of triplex formation on metal(II) concentration (cation potency)
Efflux of supportive divalent cations from the composite triple helical structure
Antagonism of triplex formation
Potential biological relevance
ACKNOWLEDGEMENTS
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: 23 Dec 1998
Copyright©Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
M. Duca, P. Vekhoff, K. Oussedik, L. Halby, and P. B. Arimondo
The triple helix: 50 years later, the outcome
Nucleic Acids Res.,
September 1, 2008;
36(16):
5123 - 5138.
[Abstract]
[Full Text]
[PDF]
J. Kusic, S. Kojic, A. Divac, and D. Stefanovic
Noncanonical DNA Elements in the Lamin B2 Origin of DNA Replication
J. Biol. Chem.,
March 18, 2005;
280(11):
9848 - 9854.
[Abstract]
[Full Text]
[PDF]
A. Majumdar, N. Puri, B. Cuenoud, F. Natt, P. Martin, A. Khorlin, N. Dyatkina, A. J. George, P. S. Miller, and M. M. Seidman
Cell Cycle Modulation of Gene Targeting by a Triple Helix-forming Oligonucleotide
J. Biol. Chem.,
March 21, 2003;
278(13):
11072 - 11077.
[Abstract]
[Full Text]
[PDF]
J. Basye, J. O. Trent, D. Gao, and S. W. Ebbinghaus
Triplex formation by morpholino oligodeoxyribonucleotides in the HER-2/neu promoter requires the pyrimidine motif
Nucleic Acids Res.,
December 1, 2001;
29(23):
4873 - 4880.
[Abstract]
[Full Text]
[PDF]
F.-L. M. Lin, A. Majumdar, L. C. Klotz, A. P. Reszka, S. Neidle, and M. M. Seidman
Stability of DNA Triplexes on Shuttle Vector Plasmids in the Replication Pool in Mammalian Cells
J. Biol. Chem.,
December 8, 2000;
275(50):
39117 - 39124.
[Abstract]
[Full Text]
[PDF]
N. Puri, A. Majumdar, B. Cuenoud, F. Natt, P. Martin, A. Boyd, P. S. Miller, and M. M. Seidman
Targeted Gene Knockout by 2'-O-Aminoethyl Modified Triplex Forming Oligonucleotides
J. Biol. Chem.,
July 27, 2001;
276(31):
28991 - 28998.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (200K)
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 (20)
Request Permissions
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Blume, S. W.
Articles by Miller, D. M.
Search for Related Content
PubMed
PubMed Citation
Articles by Blume, S. W.
Articles by Miller, D. M.
Social Bookmarking
What's this?