- Research article
- Open Access
A bacterial route for folic acid supplementation
© Weinkove et al. 2018
- Received: 22 April 2018
- Accepted: 16 May 2018
- Published: 15 June 2018
To prevent folate deficiencies, many countries supplement various foodstuffs with folic acid. This compound is a synthetic oxidised folate that differs from naturally occurring reduced folates in its metabolism and uptake. Notably, safety reviews of folic acid supplementation have not considered interactions with gut bacteria. Here, we use the Caenorhabditis elegans – Escherichia coli animal– microbe model to examine a possible bacterial route for folic acid uptake. It has been assumed that supplements are taken up directly by the worm, especially because E. coli is unable to take up folates. However, E. coli, like many other bacteria, can transport the folate breakdown product, para-aminobenzoate-glutamate (PABA-glu), via AbgT and use it for bacterial folate synthesis. This pathway may impact host health because inhibition of bacterial folate synthesis increases C. elegans lifespan.
Folic acid supplementation was found to rescue a C. elegans developmental folate-deficient mutant; however, a much higher concentration was required compared to folinic acid, a reduced folate. Unlike folinic acid, the effectiveness of folic acid supplementation was dependent on the E. coli gene, abgT, suggesting a bacterial route with PABA-glu uptake by E. coli as a first step. Surprisingly, we found up to 4% PABA-glu in folic acid preparations, including in a commercial supplement. Via breakdown to PABA-glu, folic acid increases E. coli folate synthesis. This pathway restores folate synthesis in a bacterial mutant defective in PABA synthesis, reversing the ability of this mutant to increase C. elegans lifespan.
Folic acid supplementation in C. elegans occurs chiefly indirectly via bacterial uptake of breakdown products via E. coli AbgT, and can impact C. elegans development and longevity. Examining how folic acid supplementation affects bacterial folate synthesis in the human gut may help us to better understand the safety of folic acid supplementation.
The folate cycle involves a series of essential biosynthetic reactions known as one-carbon metabolism . Folates are a family of molecules composed of three common elements: a central aromatic core derived from para-amino benzoic acid (PABA), a pterin ring that can be modified and a chain of one or more glutamates . At each step of the folate cycle, an enzyme mediates a modification of the pterin ring of the bound folate, allowing the transfer of a chemical group containing one carbon atom (methyl, formyl, etc.) to or from the compound being synthesised . This cofactor role results in folate molecules being recycled and thus they are only required in very small amounts.
Animals cannot synthesise folates and must acquire them from their diet or gut microbes. When these sources are insufficient, folate deficiency can lead to neural tube defects during human embryonic development . The rate of these defects can be lowered by preconception supplementation with folic acid, an oxidised form of folate not found in nature. Mandatory fortification of flour with folic acid has successfully decreased the incidence of birth defects in many countries, including the US and Canada . However, there are concerns that folic acid supplementation may have adverse effects on health, especially in older people [4–7], and there are many unknowns about the efficacy of uptake and biological utilisation of folic acid . Despite these uncertainties, recent reviews of the evidence by experts acting for government public health bodies have concluded that the risks are minimal and have recommended the fortification of flour or other food products as a beneficial intervention [3, 9, 10].
None of the above safety reviews mention potential interactions with gut bacteria. Bacteria also require folates for biosynthesis. Many make their own folates, but folates can also be taken up from the environment. Little is known about how bacterial folate biochemistry affects the host. Inhibiting Escherichia coli folate biosynthesis, either by treatment with the drug sulfamethoxazole (SMX) or mutation of the PABA synthesis pathway (e.g., a pabA or pabB mutant), extends the lifespan of the nematode Caenorhabditis elegans that feeds on it [11–13]. While these interventions decrease the folate levels in both E. coli and C. elegans, there remains sufficient folate available to support normal growth of both organisms . We have suggested that bacterial folate, at higher levels than need for growth, enables a bacterial activity that shortens the life of C. elegans and is independent of the folate status of C. elegans . Thus, the folate status of gut bacteria may be important for host health.
To understand the interaction between bacterial and animal folates, we have developed a C. elegans folate deficiency model, a mutant in gcp-2.1 and homologue of human GCPII, which is required for uptake of dietary polyglutamated folates . The only way we have found of decreasing C. elegans folate enough to slow growth is by raising the C. elegans gcp-2.1 mutant on SMX-treated E. coli. Under these conditions, worms show delayed development and infertility . This phenotype can be prevented with 1–10 μM folinic acid, a reduced folate which can be absorbed directly by C. elegans, demonstrating that this defect is due to folate deficiency. In contrast, prevention with folic acid requires much higher concentrations (100 μM) . We also discovered that, at high concentrations, folic acid can partially reverse the lifespan increase caused by inhibiting E. coli folate synthesis . A possible explanation for this result might be the restoration of E. coli folate levels, but that would be a surprising conclusion because E. coli does not possess a folate transporter [14, 15]. However, it is known that E. coli can take up the folic acid breakdown product PABA-glu through the transporter AbgT and catabolise it to PABA . PABA can also diffuse across E. coli membranes. Increases in bacterial PABA levels from either source can be used to synthesise folate. In this study, we examine whether these routes might explain the lifespan effects of folic acid.
E. coli is required for folic acid supplementation to prevent a C. elegans folate deficiency
Overexpression of abgT in the E. coli ΔpabA mutant (ΔpabA (abgT OE)) increased gcp-2.1 mutant body length at lower concentrations of folic acid, with a complete rescue achieved at 100 μM folic acid (Fig. 2ci). Consistent with the previous experiment, folinic acid rescued gcp-2.1 mutant growth at a 100-fold lower concentration and independently of abgT expression (Fig. 2cii). Analysing the experiment by two-way ANOVA, we find that there is a significant interaction effect of abgT genotype (F = 102.67, p < 0.0001) and folic acid concentration (F = 123.55, p < 0.0001) on C. elegans gcp-2.1 body length (Fig. 2ci). These results are consistent with folinic acid being taken up directly by the worm , and the major route of folic acid uptake requiring E. coli and the E. coli AbgT transporter.
Folic acid supports growth of E. coli pabA mutants via abgT-dependent uptake of PABA-glu
Folic acid increases E. coli folate levels in an AbgT-dependent mechanism
Folic acid preparations contain PABA-glu and PABA
Folic acid shortens C. elegans lifespan via AbgT-dependent uptake of PABA-glu
In this study we have found that a major route of uptake for the synthetic supplement folic acid by C. elegans is via E. coli rather than directly via the worm. Why is folic acid not taken up directly by C. elegans with the same affinity as folinic acid? One possible explanation is that FOLT-1, the only characterised C. elegans folate transporter, is a reduced folate carrier with specificity for folinic acid . In humans, folic acid is also taken up in a different manner to natural reduced folates, with the major route thought to be via the protein-coupled folate transporter in the small intestine . A C. elegans PCFT homologue has been identified, but its affinity for folic acid has not been characterized . In addition, a homologue of the human folate receptor has been characterised in C. elegans . This study does not rule out that direct routes for folic acid exist in C. elegans, but indicates that they are inferior to the E. coli route.
It is possible that this bacterial route for folic acid exists in humans. As far as we are aware, no other studies have tested folic acid supplements for the presence of PABA-glu or PABA, but several studies have reported issues with the stability and dissolution of commercial folic acid supplements [23, 24]. In light of these issues, manufacturers have adopted a policy of ‘overages’ in order to ensure sufficient folic acid is released in the small intestine following ingestion . The presence of PABA-glu and PABA in a commercially available folic acid source (Fig. 6), combined with the instability of folic acid at the low pH conditions of the stomach [26–28], makes it likely that PABA-glu and PABA will be available to the gut microbiota following supplementation. PABA has been identified as a human faecal excretory product after ingestion of folic acid . Furthermore, studies in rodents  and piglets  have demonstrated that infusion of labelled PABA into the cecum results in the incorporation of bacterially synthesized folate in host tissues. Thus, the literature suggests that the components for such a route exist in humans, but the importance of this route is yet to be determined. We detected this route in C. elegans because of the poor bioavailability of folic acid in our folate deficiency model. Folic acid is taken up well by humans and leads to increases in serum folate levels, but circulating folic acid is often found to have retained its oxidised form, suggesting that it is not always bioavailable even when absorbed by the host [3, 8, 32]. The bacterial route increases bioavailability.
Further studies are required in order to determine whether folic acid supplementation affects the folate status of human gut microbes and whether this in turn impacts host health. Interestingly, there are several diseases associated with an increased abundance of folate-synthesizing gut bacteria, such as inflammatory bowel disease  and small intestinal bacterial overgrowth , but a causal relationship between bacterial folate and disease has not been established. The abgT gene is found in the genomes of several enteric pathogens, including Enterobacter cloacae, Neisseria gonorrhoeae, Salmonella enterica, Shigella boydii and Staphylococcus aureus, in addition to E. coli. Whilst there is much consideration about the consequences of folic acid supplementation [3, 9, 10], our work here indicates that folic acid supplement instability and bacterial metabolism are previously unexplored variables that may impact human health and thus warrant consideration in future reports and studies.
The main route of uptake of the synthetic supplement folic acid by C. elegans is via E. coli, with a dependence on breakdown of folic acid. Folic acid supplements contain breakdown products, raising the possibility that this route could occur in humans. This route shortens the lifespan of C. elegans on folate-depleted E. coli, suggesting that there are circumstances in which this bacterial route could have negative consequences for human health.
Folates and related compounds
Folic acid, folinic acid, PABA-Glu, 5-formylTHF-Glu3, 5-methylTHF-Glu3 and methotrexate-Glu6 were obtained from Schircks, Switzerland. PABA, vitamin B12 and folic acid were purchased from Sigma Aldrich and folic acid supplement from Boots, UK.
DM was prepared as previously described , except that 10 nM vitamin B12 was added. Vitamin B12, folic acid and antibiotics were added post-autoclaving for agar plates. DM for liquid culture was filter sterilised. Following addition of 0.1 μM PABA to the liquid DM media, this was used to seed the plates in order to maintain bacterial growth (apart from growth experiments in Fig. 2). Then, 30 μL of 3 mL fresh overnight LB culture were used to inoculate 5 mL of DM (in 15 mL Falcon tubes). Kanamycin (25 μg/mL and 50 μg/mL of ampicillin if necessary) were added to both LB and DM pre-incubation. DM liquid cultures were incubated for 18 h at 37 °C and 220 RPM.
E. coli strains used in this study
Virk et al. 2016 
Baba et al. 2006 
Baba et al. 2006 
pUC19, pGreen 0029
WT (abgT OE)
pJ128, pGreen 0029
ΔpabA (abgT OE)
ΔabgT ΔpabA (abgT OE)
C. elegans strains used
SS104 glp-4(bn2), UF208 (WT) and UF209 gcp-2.1(ok1004) .
E. coli preparation and growth assay
E. coli was prepared as follows for all E. coli and C. elegans experiments. A 30 μL aliquot of an overnight LB culture of E. coli was transferred into 5 mL of DM and incubated for 18 h at 37 °C and 220 RPM. Then, 100 μL of the DM culture was seeded onto DM agar plates and incubated at 25 °C for 4 days. E. coli was removed by pipetting 1 mL of M9 medium onto the plate and a glass spreader was used to scrape off the bacterial lawn. The bacterial suspension was pipetted into a 1.5 mL Eppendorf and the volume was recorded (v). Tubes were vortexed vigorously to obtain a homogenised solution. Finally, 100 μL were taken and diluted with 900 μL of M9 in a cuvette. A spectrophotometer was used to read bacterial growth at 600 nm. Bacterial growth was calculated by multiplying OD600 by the volume of the sample (v).
E. coli folate extraction
Bacterial lawns were scraped from plates into microcentrifuge tubes using M9 solution and kept on ice. Volume (v) multiplied by the OD600 of the solution (diluted 1:5) gives a measure of the amount of material. Samples were concentrated in a cooled microcentrifuge and pellets were snap frozen in liquid nitrogen. Pellets were thawed and resuspended in a volume of ice-cold 90% methanol: 10% folate extraction buffer (FEB: 50 mM HEPES, 50 mM CHES, 0.5% w/v ascorbic acid, 0.2 M DTT, pH 7.85 with NaOH) in proportion to bacterial content (37.5 × OD600 × v). FEB was spiked with 10 nM methotrexate-Glu6 as an internal standard. Samples were vortexed vigorously and left on ice for 15 min before centrifugation in a cooled microcentrifuge for 15 min at full speed. Supernatants were used for analysis.
Folate LC-MS/MS analysis
Folates were detected by multiple reaction monitoring (MRM) analysis using a SCIEX QTRAP 6500 instrument. MRM conditions for folic acid, PABA, PABA-Glu, 5-Me-H4PteGlu3 (5-methylTHF-Glu3) and 5/10-CHO-H4PteGlu3 (formyl THF3) were optimised by infusion of standards into the instrument. The optimised conditions for –Glu3 folates were applied to other higher folates using MRM transitions as described by Lu et al. . Further confirmation of identity for folates of interest was achieved by performing enhanced product ion scans and comparing the fragment spectra with known standards.
The QTRAP 6500 was operated in ESI+ mode and was interfaced with a Shimadzu Nexera UHPLC system. Samples were separated using a Thermo PA2 C18 column (2.2 μm, 2.1 × 100 mm) with a gradient of 0.1% formic acid in water (mobile phase A) and acetonitrile (mobile phase B). Samples were maintained at 4 °C and 2 μL aliquots were injected. The column was maintained at 40 °C with a flow rate of 200 μL/min, starting at 2% B, held for 2 min, with a linear gradient to 100% B at 7 min, held for 1 min, before a 7 min re-equilibration step at 2% B necessary for consistent retention times. The column eluate flow to the MS was controlled via the QTRAP switching valve, allowing analysis between 4 and 8 min to minimise instrument contamination. Folates were quantified with reference to external standards and matrix effects were assessed by spiking of standards into extracted samples.
Survival analyses were performed as described . glp-4(bn2) worms were maintained at 15 °C and shifted to 25 °C at the L3 stage. At the L4/young adult stage, animals were placed on bacteria under the experimental conditions. All lifespan data is provided in Additional file 1: Table S1. Statistical significance was determined using Log Rank and Wilcoxon tests of the Kaplan–Meier survival model.
We thank the C. elegans Genetics Center, the C. elegans Knockout Consortium, and NBRP-E. coli at NIG for strains and we thank Sushmita Maitra and John Mathers for useful comments on the manuscript.
This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC) (grant number BB/J014516/1).
Availability of data and materials
Data corresponding to all lifespan analyses can be found in Additional file 1: Table S1. Data corresponding to LC-MS/MS measurements of folate levels (Fig. 4) and folic acid breakdown (Fig. 5) can be found in Additional file 3: Table S2 and Additional file 4: Table S3, respectively.
Conceptualization: CM, DW. Methodology: CM, DW, IC. Investigation: CM, IC. Data curation: CM, IC. Writing: CM, DW, IC, JG. Supervision: DW. Funding acquisition: DW. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Ducker GS, Rabinowitz JD. One-carbon metabolism in health and disease. Cell Metab. 2017;25(1):27–42.View ArticlePubMedGoogle Scholar
- Green J, Matthews R. 2007. Folate Biosynthesis, Reduction, and Polyglutamylation and the Interconversion of Folate Derivatives, EcoSal Plus. 2007; https://doi.org/10.1128/ecosalplus.18.104.22.168.
- Bailey LB, Stover PJ, McNulty H, Fenech MF, Gregory JF, Mills JL, et al. Biomarkers of nutrition for development—folate review. J Nutr. 2015;145(7):1636S–80S.View ArticlePubMedPubMed CentralGoogle Scholar
- Kim Y-I. Folate and colorectal cancer: An evidence-based critical review. Mol Nutr Food Res. 2007;51(3):267–92.View ArticlePubMedGoogle Scholar
- Marean A, Graf A, Zhang Y, Niswander L. Folic acid supplementation can adversely affect murine neural tube closure and embryonic survival. Hum Mol Genet. 2011;20(18):3678–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Pickell L, Brown K, Li D, Wang XL, Deng L, Wu Q, et al. High intake of folic acid disrupts embryonic development in mice. Birth Defects Res A Clin Mol Teratol. 2011;91(1):8–19.View ArticlePubMedGoogle Scholar
- Smith AD, Kim Y-I, Refsum H. Is folic acid good for everyone? Am J Clin Nutr. 2008;87(3):517–33.View ArticlePubMedGoogle Scholar
- Gregory JF, Quinlivan EP, Davis SR. Integrating the issues of folate bioavailability, intake and metabolism in the era of fortification. Trends Food Sci Technol. 2005;16(6):229–40.View ArticleGoogle Scholar
- Boyles AL, Yetley EA, Thayer KA, Coates PM. Safe use of high intakes of folic acid: research challenges and paths forward. Nutr Rev. 2016;74(7):469–74.View ArticlePubMedPubMed CentralGoogle Scholar
- Public Health England. Folic acid: updated SACN recommendations. 2017. Retrieved 25 May 2018. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/637111/SACN_Update_on_folic_acid.pdf.
- Han B, Sivaramakrishnan P, Lin CJ, Neve IAA, He J, Tay LWR, et al. Microbial genetic composition tunes host longevity. Cell. 2017;169(7):1249–62. e13View ArticlePubMedGoogle Scholar
- Virk B, Correia G, Dixon DP, Feyst I, Jia J, Oberleitner N, et al. Excessive folate synthesis limits lifespan in the C. elegans: E. coli aging model. BMC Biol. 2012;10:67.View ArticlePubMedPubMed CentralGoogle Scholar
- Virk B, Jia J, Maynard CA, Raimundo A, Lefebvre J, Richards SA, et al. Folate acts in E. coli to accelerate C. elegans aging independently of bacterial biosynthesis. Cell Rep. 2016;14(7):1611–20.View ArticlePubMedPubMed CentralGoogle Scholar
- Nickerson WJ, Webb M. Effect of folic acid analogues on growth and cell division of nonexacting microorganisms. J Bacteriol. 1956;71(2):129–39.PubMedPubMed CentralGoogle Scholar
- Webb M. Inactivation of analogues of folic acid by certain non-exacting bacteria. Biochim Biophys Acta. 1955;17(2):212–25.View ArticlePubMedGoogle Scholar
- Carter EL, Jager L, Gardner L, Hall CC, Willis S, Green JM. Escherichia coli abg genes enable uptake and cleavage of the folate catabolite p-aminobenzoyl-glutamate. J Bacteriol. 2007;189(9):3329–34.View ArticlePubMedPubMed CentralGoogle Scholar
- Balamurugan K, Ashokkumar B, Moussaif M, Sze JY, Said HM. Cloning and functional characterization of a folate transporter from the nematode Caenorhabditis elegans. Am J Physiol Cell Physiol. 2007;293(2):C670–81.View ArticlePubMedGoogle Scholar
- Tran PV, Nichols BP. Expression of Escherichia coli pabA. J Bacteriol. 1991;173(12):3680–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Qiu A, Jansen M, Sakaris A, Min SH, Chattopadhyay S, Tsai E, et al. Identification of an intestinal folate transporter and the molecular basis for hereditary folate malabsorption. Cell. 2006;127(5):917–28.View ArticlePubMedGoogle Scholar
- Ortbauer M, Ripper D, Fuhrmann T, Lassi M, Auernigg-Haselmaier S, Stiegler C, et al. Folate deficiency and over-supplementation causes impaired folate metabolism: Regulation and adaptation mechanisms in Caenorhabditis elegans. Mol Nutr Food Res. 2016;60(4):949–56.View ArticlePubMedGoogle Scholar
- Chaudhari SN, Mukherjee M, Vagasi AS, Bi G, Rahman MM, Nguyen CQ, et al. Bacterial folates provide an exogenous signal for C. elegans germline stem cell proliferation. Dev Cell. 2016;38(1):33–46.View ArticlePubMedPubMed CentralGoogle Scholar
- Delmar JA, Yu EW. The AbgT family: a novel class of antimetabolite transporters. Protein Sci. 2016;25(2):322–37.View ArticlePubMedGoogle Scholar
- Hoag SW, Ramachandruni H, Shangraw RF. Failure of prescription prenatal vitamin products to meet USP standards for folic acid dissolution. J Am Pharm Assoc (Wash). 1997;NS37(4):397–400.View ArticleGoogle Scholar
- Sculthorpe NF, Davies B, Ashton T, Allison S, McGuire DN, Malhi JS. Commercially available folic acid supplements and their compliance with the British Pharmacopoeia test for dissolution. J Public Health Med. 2001;23(3):195–7.View ArticlePubMedGoogle Scholar
- Andrews KW, Roseland JM, Gusev PA, Palachuvattil J, Dang PT, Savarala S, et al. Analytical ingredient content and variability of adult multivitamin/mineral products: national estimates for the Dietary Supplement Ingredient Database. Am J Clin Nutr. 2017;105(2):526–39.View ArticlePubMedGoogle Scholar
- De Brouwer V, Zhang GF, Storozhenko S, Straeten DV, Lambert WE. pH stability of individual folates during critical sample preparation steps in prevision of the analysis of plant folates. Phytochem Anal. 2007;18(6):496–508.View ArticlePubMedGoogle Scholar
- Gazzali AM, Lobry M, Colombeau L, Acherar S, Azais H, Mordon S, et al. Stability of folic acid under several parameters. Eur J Pharm Sci. 2016;93:419–30.View ArticlePubMedGoogle Scholar
- Seyoum E, Selhub J. Properties of food folates determined by stability and susceptibility to intestinal pteroylpolyglutamate hydrolase action. J Nutr. 1998;128(11):1956–60.View ArticlePubMedGoogle Scholar
- Denko CW, Grundy WE, Wheeler NC, Henderson CR, Berryman, GH, Friedemann, TE, Youmans, JB. The excretion of B-complex vitamins by normal adults on a restricted intake. Arch Biochem. 1946;11:109–17. Google Scholar
- Rong N, Selhub J, Goldin BR, Rosenberg IH. Bacterially synthesized folate in rat large intestine is incorporated into host tissue folyl polyglutamates. J Nutr. 1991;121(12):1955–9.View ArticlePubMedGoogle Scholar
- Asrar FM, O'Connor DL. Bacterially synthesized folate and supplemental folic acid are absorbed across the large intestine of piglets. J Nutr Biochem. 2005;16(10):587–93.View ArticlePubMedGoogle Scholar
- Patanwala I, King MJ, Barrett DA, Rose J, Jackson R, Hudson M, et al. Folic acid handling by the human gut: implications for food fortification and supplementation. Am J Clin Nutr. 2014;100(2):593–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Shin NR, Whon TW, Bae JW. Proteobacteria: microbial signature of dysbiosis in gut microbiota. Trends Biotechnol. 2015;33(9):496–503.View ArticlePubMedGoogle Scholar
- Camilo E, Zimmerman J, Mason JB, Golner B, Russell R, Selhub J, et al. Folate synthesized by bacteria in the human upper small intestine is assimilated by the host. Gastroenterology. 1996;110(4):991–8.View ArticlePubMedGoogle Scholar
- Baba T, Ara T, Hasegawa M, Takai Y, Okumura Y, Baba M, et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol Syst Biol. 2006;2:2006.0008.View ArticlePubMedPubMed CentralGoogle Scholar
- Moore SD. Assembling New Escherichia coli Strains by Transduction Using Phage P1. In: Williams JA, editor. Strain Engineering. Methods in Molecular Biology. New York (USA): Humana Press; 2011. p. 155–69.Google Scholar
- Yanisch-Perron C, Vieira J, Messing J. Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13mp18 and pUC19 vectors. Gene. 1985;33(1):103–19.View ArticlePubMedGoogle Scholar
- Lu W, Kwon YK, Rabinowitz JD. Isotope ratio-based profiling of microbial folates. J Am Soc Mass Spectrom. 2007;18(5):898–909.View ArticlePubMedPubMed CentralGoogle Scholar