On the viability of Escherichia colicells lacking DNA topoisomerase I
© Stockum et al; licensee BioMed Central Ltd. 2012
Received: 9 November 2011
Accepted: 28 February 2012
Published: 28 February 2012
Manipulations of the DNA double helix during replication, transcription and other nucleic acid processing cause a change of DNA topology, which results in torsional stress. This stress is relaxed by DNA topoisomerases, a class of enzymes present in all domains of life. Negatively supercoiled DNA is relaxed by type IA topoisomerases that are widespread in bacteria, archaea and eukaryotes. In Escherichia coli there is conflicting data about viability of ΔtopA cells lacking topoisomerase I.
In this study we sought to clarify whether E. coli cells lacking topoisomerase I are viable by using a plasmid-based lethality assay that allowed us to investigate the phenotype of ΔtopA cells without the presence of any compensatory mutations. Our results show that cells lacking topoisomerase I show an extreme growth defect and cannot be cultured without the accumulation of compensatory mutations. This growth defect can be partially suppressed by overexpression of topoisomerase III, the other type IA topoisomerase in E. coli, suggesting that the accumulation of torsional stress is, at least partially, responsible for the lethality of ΔtopA cells. The absence of RNase HI strongly exacerbates the phenotype of cells lacking topoisomerase I, which supports the idea that the processing of RNA:DNA hybrids is vitally important in ΔtopA cells. However, we did not observe suppression of the ΔtopA phenotype by increasing the level of R-loop processing enzymes, such as RNase HI or RecG.
Our data show unambiguously that E. coli cells are not viable in the absence of DNA topoisomerase I without the presence of compensatory mutations. Furthermore, our data suggest that the accumulation of R-loops is not the primary reason for the severe growth defect of cells lacking topoisomerase I, which is in contrast to the current literature. Potential reasons for this discrepancy are discussed.
Cellular growth and division requires unwinding of millions of base pairs to allow duplication of chromosomes or to produce the RNA transcripts needed to express genes. Unwinding of the double helix results in torsional stress, a stress solved by topoisomerases, a ubiquitous group of enzymes that are capable of managing the topological state of DNA. Topoisomerases transiently break either one (type I topoisomerases) or both strands (type II topoisomerases) of the double helix, change the topological state of the DNA and then re-ligate the break. This manipulation enables not only modification of DNA superhelicity to allow unwinding of the double helix, but allows the decatenation of circular DNAs, thereby enabling circular chromosomes or plasmids to be separated during cell division [1–3].
In Escherichia coli one of the best studied examples of a type IA topoisomerase (where the protein link is to the 5' phosphate, in contrast to type IB topoisomerases where the protein link is to the 3' phosphate) is DNA topoisomerase I, which is encoded by the topA gene. Topoisomerase I relaxes negative torsional stress and is required to prevent the chromosomal DNA from becoming extensively negatively supercoiled . Topoisomerase I requires an exposed single stranded region . In E. coli the chromosomal DNA is normally slightly negatively supercoiled due to the activity of DNA gyrase, a type IIA topoisomerase, and extensive single stranded regions are not available for topoisomerase I to act on . However, the unwinding of the double helix will result not only in single stranded regions but also in extensive changes in the local level of torsional stress. For instance, the "twin-domain" model of transcription suggests that the elongating RNA polymerase complex (RNAP) causes accumulation of positive torsional stress in front of the transcription complex, whereas negative supercoils accumulate behind . While the positive supercoils are relaxed by gyrase, the negative torsional stress leads to the formation of single stranded DNA, which is a hot-spot for relaxation by topoisomerase I .
In cells lacking the activity of topoisomerase I the chromosomal DNA becomes hypernegatively supercoiled, especially behind transcribing RNAP complexes. DNA gyrase will remove the positive torsional stress in front of RNAP, whereas the negative supercoils will persist if they cannot be relaxed by Topo I. This accumulation of negative supercoils has been thought to increase the probability that the newly generated transcript will hybridise with the template strand, thereby forming an R-loop . This idea was supported by results showing that R-loops are a substrate for topoisomerase I in vitro . Furthermore, increased levels of RNase HI, encoded by the rnhA gene, have been shown to partially suppress the growth defect of ΔtopA cells, while the deletion of rnhA exacerbated the ΔtopA phenotype .
It was initially described that ΔtopA cells can grow without apparent ill effect . However, it was later discovered that the ΔtopA mutant strains used had accumulated compensatory mutations in DNA gyrase and that ΔtopA strains without these suppressor mutations show a severe growth defect , an observation confirmed in later studies . It is not clear why growth of cells lacking topoisomerase I is so severely impeded. The genetic interaction of topA and rnhA has led to the suggestion that R-loops might be a major problem for ΔtopA cells. However, ΔtopA strains have been reported to be viable in Salmonella , a result that prompted Stupina and Wang to re-investigate the viability of E. coli cells lacking topoisomerase I and they reported that viable ΔtopA derivatives can indeed be engineered .
In this study we employed a plasmid-based lethality assay [12, 13] to investigate the viability and the phenotypes of ΔtopA cells without the presence of any compensatory mutations. Our data show that cells lacking topoisomerase I suffer from an extreme growth defect and cannot be subcultured unless they acquire compensatory mutations. This growth defect was suppressed by overexpression of topoisomerase III, the other E. coli type IA topoisomerase, as reported [4, 14]. We show that deletion of rnhA strongly exacerbates the phenotype of cells lacking Topo I, which supports the idea that processing RNA:DNA hybrids is vitally important in the absence of topoisomerase I. However, in contrast to previous results  we did not observe any suppression of the ΔtopA phenotype if the level of R-loop processing enzymes (RNase HI, RecG) was increased, suggesting that R-loops are not the primary reason for the lethality of ΔtopA single mutants.
Results and discussion
To investigate whether a ΔtopA strain can grow without compensatory mutations we employed a plasmid-based lethality assay [12, 13]. The wild type topA gene was cloned into pRC7 (pAST111), a lac + mini-F plasmid that is rapidly lost from cells. This was used to compensate for a topA::apra null mutation in the chromosome of a Δlac background. If a ΔtopA mutant is viable, plasmid-free cells will form white lac - colonies on agar plates supplemented with X-gal and IPTG. However, if a topA deletion is lethal, cells that have lost the plasmid will fail to grow, allowing only formation of blue lac + colonies. When viability is reduced but not eliminated, the colonies formed by cells retaining the plasmid are noticeably larger than those formed by plasmid-free cells [13, 15].
The ΔtopA lethality is suppressed by overexpression of topB
Many of the studies investigating the properties of ΔtopA cells have worked in a background with a conditional gyrB mutation. Mutations in gyrA or gyrB reduce the global level of supercoiling, thereby enabling ΔtopA cells to grow . In gyrB203(ts) strains the activity of gyrase is reduced at high temperature. Thus, ΔtopA gyrB203(ts) cells grow at high temperature, since the reduced activity of gyrase compensates of the absence of topoisomerase I, but are cold-sensitive . By using the plasmid-based lethality assay we were able to investigate some of the properties of ΔtopA cells without the presence of a compensatory mutation.
We repeated overexpression studies with topB, which encodes for topoisomerase III, the other member of the type IA family of topoisomerases in E. coli . DNA topoisomerase III was shown to relax transcription-induced negative supercoiling in vivo and in vitro  and high levels of expression partially suppressed the growth defect of ΔtopA strains . To investigate the effect of topB overexpression in a topA deletion background we used pECR17, a P araBAD topB expression plasmid that allows arabinose-controlled expression of topB. For these experiments cultures were grown overnight, with selection for both pECR17 and pRC7 topA. The cultures were then diluted as described in Material and Methods and parallel cultures grown with the arabinose concentration indicated, selecting only for pECR17. The cultures were then diluted as described and plated on plates with the corresponding arabinose concentration and selection for pECR17.
The ΔtopA lethality is not suppressed by overexpression of rnhA or recG
It was previously reported that the growth defect of cells lacking topoisomerase I can be suppressed by increased concentrations of RNase HI. Furthermore, ΔtopA ΔrnhA double mutants were found to be inviable even in the presence of point mutations that strongly suppress the ΔtopA phenotype . This led to the suggestion that RNA:DNA hybrids might be a major problem for ΔtopA cells . We therefore investigated whether RecG helicase suppressed the ΔtopA phenotype. RecG protein was shown to unwind the RNA from R-loops in vitro [20, 21] and overexpression of recG results in reduced yields of ColEI plasmids that initiate replication via an R-loop , suggesting that RecG can process R-loops in vivo.
To investigate whether recG overexpression suppresses the ΔtopA phenotype we used an overexpression construct as described for topB (see Material and Methods). The plasmid fully suppressed the phenotype of cells lacking RecG if expression was induced, whereas no suppression was observed under conditions where expression was repressed . As shown in Figure 2B expression of recG at high levels only marginally suppressed the topA phenotype.
To test whether increased levels of RNase HI can suppress the lethality of topA strains we integrated our proB::rnhA + expression construct into an rnhA + background. Thus, any expression from our integration construct will be in addition to the expression from the native rnhA gene. We then introduced our topA::apra allele, covering the deletion with the pRC7 topA plasmid. However, growth of this strain in medium with moderate (data not shown) or high arabinose concentrations did not lead to formation of white colonies (Figure 3B).
Since we did not directly measure the concentration of RNase HI in cells we cannot exclude the possibility that the levels in our expression constructs are not high enough for suppression of the ΔtopA phenotype. We therefore wanted to test the expression of rnhA in a system that might be more sensitive for low expression levels. It was observed before that the co-expression of both rnhA and topB resulted in a synergistic suppression of the topA phenotype . We therefore wanted to know whether the expression of rnhA from our integration construct would increase the suppression of the observed topB overexpression. To test this we transformed our ptopA/ΔtopA ΔproB::rnhA + background with the topB expression plasmid. However, co-expression did not lead to an increase in the size of the white colonies. If anything a mild reduction of viability is observed (Figure 3C).
Absence of RNase HI strongly exacerbates the phenotype of ΔtopAcells
Our results show that RecG can not compensate for the absence of RNase HI (Figure 2). However, if RecG processes some R-loops in vivo, the deletion of recG in a ΔtopA background should exacerbate the topA phenotype, as observed with rnhA. This was indeed observed. ΔrecG ΔtopA double mutants were not able to form any white colonies, neither on LB broth (data not shown) nor on minimal medium (Figure 4A panel iii). However, when we transformed the construct with the P araBAD topB overexpression plasmid we observed that overexpression of topB allowed growth of ΔrecG ΔtopA cells, with white colonies being only marginally smaller than in the topA/ptopB + control (compare Figures 4C and Figure 2A). Thus, the impact of a recG deletion is marginal in comparison to the impact of deleting rnhA, suggesting that the contribution of RecG to genome-wide processing of R-loops might be lower than anticipated.
The plasmid-based lethality assay exploited in this study provided a novel approach to investigate the phenotype of cells lacking topoisomerase I without the presence of any compensatory mutations. The results presented show that cells lacking topoisomerase I exhibit an extreme growth defect, indicating that they are under a constant selection pressure for compensatory mutations. This phenotype was partially suppressed by overexpression of topoisomerase III, suggesting that the accumulation of torsional stress is, to a certain extent, responsible for the lethality of ΔtopA cells, as reported . However, the overexpression of R-loop processing enzymes, such as RNase HI or RecG, did not result in a major suppression of the ΔtopA phenotype. This result suggests that the accumulation of R-loops does not contribute very much towards the growth defect of cells lacking topoisomerase I, which is in contrast to previous reports [4, 7]. However, the absence of RecG and especially RNase HI exacerbates the phenotype of ΔtopA cells, which suggests that the processing of RNA:DNA hybrids is vitally important in the absence of topoisomerase I. Thus, R-loops accumulate to a toxic level only in cells lacking RNase HI, while the toxicity in ΔtopA single mutants is mainly caused by an additional effect that is yet to be characterised. Further experiments will be necessary to shed light on the question as to why cells lacking Topo I have such a severe growth defect and how much R-loops contribute to this phenotype.
Escherichia coli K-12 strains
General P1 donors
F- ΔtopA75 zci2234::cat Δfnr-267? rph-1
MG1655 and derivatives
ΔlacIZYA ΔproB::rnhA + -frt > kan > frt
ΔlacIZYA ΔproB::rnhA + - frt > kan > frt pCP20
AM2283 × pCP20 to Apr
ΔlacIZYA ΔproB::rnhA + - frt
ΔlacIZYA ΔproB::rnhA + - frt > kan > frt
TB28 × P1.AM2283 to Kmr
ΔlacIZYA ΔproB::rnhA + - frt > kan > frt ΔrecG::apra
AM2290 × P1.N6052 to Aprar
TB28 × pAST111 to Apr
ΔlacIZYA ΔtopA::apra pAST111
AS1047 × P1.RCe296 to Aprar
ΔlacIZYA topA::apra ΔrecG::cat pAST111
AS1050 × P1.N4560 to Cmr
ΔlacIZYA topA::apra rnhA::cat pAST111
AS1050 × P1.N4704 to Cmr
ΔlacIZYA topA::apra pAST111 pECR17
AS1050 × pECR17 to Apr Kmr
ΔlacIZYA topA::apra ΔrecG::cat pAST111 pECR17
AS1053 × pECR17 to Apr Kmr
ΔlacIZYA topA::apra rnhA::cat pAST111 pECR17
AS1054 × pECR17 to Apr Kmr
ΔlacIZYA ΔtopA75 zci-2234::cat pAST111
AS1047 × P1.VS111 to Cmr
ΔlacIZYA ΔproB::rnhA + -frt pAST111
AM2285 × pAST111 to Apr
ΔlacIZYA ΔproB::rnhA + -frt topA::apra pAST111
AS1130 × P1.RCe296 to Aprar
ΔlacIZYA topA::apra pAST111 pAST120
AS1050 × pAST120 to Kmr (Apr)
ΔlacIZYA ΔproB::rnhA + - frt > kan > frt ΔrecG::apra pJJ100
AM2304 × pJJ100 to Apr
ΔlacIZYA ΔproB::rnhA + - frt > kan > frt ΔrecG::apra rnhA::cat pJJ100
AS1134 × P1.N4704 to Cmr
ΔlacIZYA ΔproB::rnhA + - frt topA::apra pAST111 pECR17
AS1131 × pERC17 to Kmr (Apr)
pRC7 is a low copy-number, mini-F derivative of the lac+ construct pFZY1 . pJJ100 (recG + ) and pAST111 (topA + ) are derivatives of pRC7 encoding the wild type genes indicated. The construction of pJJ100 has been described elsewhere [13, 15, 27]. For generation of pAST111 the topA gene was PCR amplified from MG1655 chromosomal DNA. To account for the complex promoter of the topA gene , 150 bp upstream of the start codon were included. Both the 5' and the 3' primer introduced ApaI sites, allowing cloning into the ApaI site within the lacI q gene of pRC7. pAST120 (recG+), pECR15 (rnhA + ) and pECR16/17 (topB + ) are all P araBAD derivatives, which allow arabinose-controlled expression of the genes indicated. For the construction of pAST120 the HindIII fragment from pDIM141 containing a kanamycin resistance marker flanked by FRT sites was cloned into the single HindIII site of pDIM104, the construction of which was described elsewhere . This allowed maintenance of the plasmid via kanamycin selection. pECR15 (rnhA) was constructed by amplifying the rnhA gene from MG1655 chromosomal DNA with the 5' primer introducing a EcoRI and the 3' primer introducing a XbaI site, allowing cloning into P ara B A D . pECR16 (topB) was generated in an analogous way. To allow maintenance of the plasmid via kanamycin the HindIII fragment from pDIM141 was cloned into the single HindIII site of pECR16, analogous as described for pAST120. pDIM141 is a derivative of pLau17 . The eCFP gene was replaced with mRFP (pDIM117). The kanamycin resistance gene was PCR amplified from EZ-Tn10 with primers introducing FRT sites either side, followed by HindIII restriction sites. This FRT-kan-FRT cassette was then cloned into the single HindIII site of pDIM117, resulting in pDIM141.
Media and general methods
Synthetic lethality assays
The rationale for synthetic lethality assays has been described [12, 13]. Essentially, a wild type gene of interest is cloned in pRC7, a lac + mini-F plasmid that is rapidly lost, and used to cover a null mutation in the chromosome, in a Δlac background. If the mutant is viable, the plasmid-free cells segregated during culture will form lac - colonies on agar plates. If, however, the deletion is lethal, they will fail to grow and only lac + colonies formed by cells retaining the plasmid will be observed. When viability is reduced but not eliminated, the colonies formed by cells retaining the plasmid are noticeably larger than those formed by plasmid-free cells. To record the phenotype, cultures of strains carrying the relevant pRC7 derivatives were grown overnight in LB broth containing ampicillin to maintain plasmid selection, diluted 80-fold in LB broth and grown without ampicillin selection to an A650 of 0.4 before spreading dilutions on LB agar or 56/2 glucose minimal salts agar supplemented with X-gal and IPTG. Plates were photographed and scored after 48 h (LB agar) or 72 h (56/2 agar) at 37°C, unless stated otherwise. Plasmid-free cells forming small white colonies were re-streaked to see if they could be subcultured, and the streak plates photographed after incubation at 37°C for 24 h to 48 h (LB agar), or 48 h to 72 h (56/2 glucose salts agar), as indicated.
We wish to thank Carol Buckman and Lynda Harris for excellent technical help, Tim Moore and Akeel Mahdi for generation of plasmids and some of the mutant alleles exploited, and Amy Upton, Ed Bolt and Peter McGlynn for critical reading of the manuscript. This work was funded by the Medical Research Council (grant G0800970). CJR was also supported by The Leverhulme Trust.
- Champoux JJ: DNA topoisomerases: structure, function, and mechanism. Annu Rev Biochem. 2001, 70: 369-413.PubMedView Article
- Deweese JE, Osheroff MA, Osheroff N: DNA Topology and Topoisomerases: Teachinga "Knotty" Subject. Biochem Mol Biol Educ. 2008, 37: 2-10.PubMedPubMed CentralView Article
- Viard T, de la Tour CB: Type IA topoisomerases: a simple puzzle?. Biochimie. 2007, 89: 456-467.PubMedView Article
- Drolet M, Broccoli S, Rallu F, Hraiky C, Fortin C, Masse E, Baaklini I: The problem of hypernegative supercoiling and R-loop formation in transcription. Front Biosci. 2003, 8: d210-d221.PubMedView Article
- Liu LF, Wang JC: Supercoiling of the DNA template during transcription. Proc Natl Acad Sci USA. 1987, 84: 7024-7027.PubMedPubMed CentralView Article
- Gowrishankar J, Harinarayanan R: Why is transcription coupled to translation in bacteria?. Mol Microbiol. 2004, 54: 598-603.PubMedView Article
- Drolet M, Phoenix P, Menzel R, Masse E, Liu LF, Crouch RJ: Overexpressionof RNase H partially complements the growth defect of an Escherichia coli delta topA mutant: R-loop formation is a major problem in the absenceof DNA topoisomerase I. Proc Natl Acad Sci USA. 1995, 92: 3526-3530.PubMedPubMed CentralView Article
- Sternglanz R, DiNardo S, Voelkel KA, Nishimura Y, Hirota Y, Becherer K, Zumstein L, Wang JC: Mutations in the gene coding for Escherichia coli DNA topoisomerase I affect transcription and transposition. Proc Natl Acad Sci USA. 1981, 78: 2747-2751.PubMedPubMed CentralView Article
- DiNardo S, Voelkel KA, Sternglanz R, Reynolds AE, Wright A: Escherichia coli DNA topoisomerase I mutants have compensatory mutations in DNA gyrase genes. Cell. 1982, 31: 43-51.PubMedView Article
- Richardson SM, Higgins CF, Lilley DM: The genetic control of DNA supercoiling in Salmonella typhimurium. EMBO J. 1984, 3: 1745-1752.PubMedPubMed Central
- Stupina VA, Wang JC: Viability of Escherichia coli topA mutants lacking DNA topoisomerase I. J Biol Chem. 2005, 280: 355-360.PubMedView Article
- Bernhardt TG, de Boer PA: Screening for synthetic lethal mutants in Escherichia coli and identification of EnvC (YibP) as a periplasmic septal ring factor with murein hydrolase activity. Mol Microbiol. 2004, 52: 1255-1269.PubMedPubMed CentralView Article
- Mahdi AA, Buckman C, Harris L, Lloyd RG: Rep and PriA helicase activities prevent RecA from provoking unnecessary recombination during replication fork repair. Genes Dev. 2006, 20: 2135-2147.PubMedPubMed CentralView Article
- Broccoli S, Phoenix P, Drolet M: Isolation of the topB gene encoding DNA topoisomerase III as a multicopy suppressor of topA null mutations in Escherichia coli. Mol Microbiol. 2000, 35: 58-68.PubMedView Article
- Rudolph CJ, Mahdi AA, Upton AL, Lloyd RG: RecG Protein and Single-strand DNA Exonucleases Avoid Cell Lethality Associated With PriA Helicase Activity in Escherichia coli. Genetics. 2010, 186: 473-792.PubMedPubMed CentralView Article
- Wang Y, Lynch AS, Chen SJ, Wang JC: On the molecular basis of the thermal sensitivity of an Escherichia coli top mutant. J Biol Chem. 2002, 277: 1203-1209.PubMedView Article
- Masse E, Drolet M: R-loop-dependent hypernegative supercoiling in Escherichia coli top mutants preferentially occurs at low temperatures and correlates with growth inhibition. J Mol Biol. 1999, 294: 321-332.PubMedView Article
- Raji A, Zabel DJ, Laufer CS, Depew RE: Genetic analysis of mutations that compensate for loss of Escherichia coli DNA topoisomerase I. J Bacteriol. 1985, 162: 1173-1179.PubMedPubMed Central
- DiGate RJ, Marians KJ: Molecular cloning and DNA sequence analysis of Escherichia coli topB the gene encoding topoisomerase III. J Biol Chem. 1989, 264: 17924-17930.PubMed
- Vincent SD, Mahdi AA, Lloyd RG: The RecG branch migration protein of Escherichia coli dissociates R-loops. J Mol Biol. 1996, 264: 713-721.PubMedView Article
- Fukuoh A, Iwasaki H, Ishioka K, Shinagawa H: ATP-dependent resolution of R-loops at the ColE1 replication origin by Escherichia coli RecG protein, a Holliday junction-specific helicase. EMBO J. 1997, 16: 203-209.PubMedPubMed CentralView Article
- Rudolph CJ, Upton AL, Harris L, Lloyd RG: Pathological replication in cells lacking RecG DNA translocase. Mol Microbiol. 2009, 73: 352-366.PubMedPubMed CentralView Article
- Datsenko KA, Wanner BL: One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA. 2000, 97: 6640-6645.PubMedPubMed CentralView Article
- Kogoma T: Stable DNA replication: Interplay between DNA replication, homologous recombination, and transcription. Microbiol Molec Biol Rev. 1997, 61: 212-238.
- Usongo V, Nolent F, Sanscartier P, Tanguay C, Broccoli S, Baaklini I, Drlica K, Drolet M: Depletion of RNase HI activity in Escherichia coli lacking DNA topoisomerase I leads to defects in DNA supercoiling and segregation. Mol Microbiol. 2008, 69: 968-981.PubMedPubMed Central
- Meddows TR, Savory AP, Lloyd RG: RecG helicase promotes DNA double-strand break repair. Mol Microbiol. 2004, 52: 119-132.PubMedView Article
- Zhang J, Mahdi AA, Briggs GS, Lloyd RG: Promoting and avoiding recombination: contrasting activities of the Escherichia coli RuvABC Holliday junction resolvase and RecG DNA translocase. Genetics. 2010, 185: 23-37.PubMedPubMed CentralView Article
- Weinstein-Fischer D, Altuvia S: Differential regulation of Escherichia coli topoisomerase I by Fis. Mol Microbiol. 2007, 63: 1131-1144.PubMedView Article
- Lau IF, Filipe SR, Soballe B, Okstad OA, Barre FX, Sherratt DJ: Spatial and temporal organization of replicating Escherichia coli chromosomes. Mol Microbiol. 2003, 49: 731-743.PubMedView Article
- Al-Deib AA, Mahdi AA, Lloyd RG: Modulation of recombination and DNA repair by the RecG and PriA helicases of Escherichia coli K-12. J Bacteriol. 1996, 178: 6782-6789.PubMedPubMed Central
- McGlynn P, Lloyd RG: Modulation of RNA polymerase by (p)ppGpp reveals a RecG-dependent mechanism for replication fork progression. Cell. 2000, 101: 35-45.PubMedView Article
- Trautinger BW, Jaktaji RP, Rusakova E, Lloyd RG: RNA polymerase modulators and DNA repair activities resolve conflicts between DNA replication and transcription. Mol Cell. 2005, 19: 247-258.PubMedView Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.