- Open Access
CCR5 limits cortical viral loads during West Nile virus infection of the central nervous system
© Durrant et al. 2015
Received: 18 June 2015
Accepted: 25 November 2015
Published: 15 December 2015
Cell-mediated immunity is critical for clearance of central nervous system (CNS) infection with the encephalitic flavivirus, West Nile virus (WNV). Prior studies from our laboratory have shown that WNV-infected neurons express chemoattractants that mediate recruitment of antiviral leukocytes into the CNS. Although the chemokine receptor, CCR5, has been shown to play an important role in CNS host defense during WNV infection, regional effects of its activity within the infected brain have not been defined.
We used CCR5-deficient mice and an established murine model of WNV encephalitis to determine whether CCR5 activity impacts on WNV levels within the CNS in a region-specific fashion. Statistical comparisons between groups were made with one- or two-way analysis of variance; Bonferroni’s post hoc test was subsequently used to compare individual means. Survival was analyzed by the log-rank test. Analyses were conducted using Prism software (GraphPad Prism). All data were expressed as means ± SEM. Differences were considered significant if P ≤ 0.05.
As previously shown, lack of CCR5 activity led to increased symptomatic disease and mortality in mice after subcutaneous infection with WNV. Evaluation of viral burden in the footpad, draining lymph nodes, spleen, olfactory bulb, and cerebellum derived from WNV-infected wild-type, and CCR5−/− mice showed no differences between the genotypes. In contrast, WNV-infected, CCR5−/− mice exhibited significantly increased viral burden in cortical tissues, including the hippocampus, at day 8 post-infection. CNS regional studies of chemokine expression via luminex analysis revealed significantly increased expression of CCR5 ligands, CCL4 and CCL5, within the cortices of WNV-infected, CCR5−/− mice compared with those of similarly infected WT animals. Cortical elevations in viral loads and CCR5 ligands in WNV-infected, CCR5−/− mice, however, were associated with decreased numbers of infiltrating mononuclear cells and increased permeability of the blood-brain barrier.
These data indicate that regional differences in chemokine expression occur in response to WNV infection of the CNS, and that cortical neurons require CCR5 activity to limit viral burden in this brain region.
Infection with the encephalitic flavivirus, West Nile virus (WNV), is the leading cause of domestically acquired arboviral disease in the USA . Acute infectious syndromes after infection with WNV include a self-limited febrile illness, West Nile fever (WNF), or more severe neuroinvasive diseases (WNND), including meningitis, encephalitis, or flaccid paralysis. The entry of virus-specific T cells into the CNS parenchyma is essential for viral clearance and survival in both human and murine subjects with WNV encephalitis [2–6]. Indeed, the increased incidence of WNV neuroinvasive disease in patients on anti-T cell therapies [5, 7] and in mice with T cell deficiencies [4, 8–10] indicates that the clearance of WNV within the CNS relies heavily on cell-mediated immune responses that promote the CNS entry and effector functions of CD8+ T cells [11, 12]. While studies indicate that CNS regions differ in the extent of inflammatory infiltrates during viral encephalitis [13–16], regional differences in expression of guidance cues that promote T cell entry have not been established. These guidance cues, combined with the tightly controlled egress of leukocytes from the perivascular sites, coordinate the migration of leukocyte subsets for protective and pathologic purposes.
In most tissues, leukocyte recruitment is orchestrated by a series of coordinated leukocyte-endothelial interactions involving several families of molecular regulators including selectins, integrins, and chemokines [17, 18]. Chemokines are a superfamily of over 50 structurally homologous chemotactic, heparin binding, secreted proteins with their target cell specificity conferred by pertussis toxin (PTX) sensitive, Gαi-coupled seven transmembrane glycoprotein chemokine receptors. Of interest, CXCL12 and its receptors are believed to most resemble the ancestral chemokine-receptor pair  suggesting that CXC chemokines are evolutionarily older than CC chemokines. In published studies, we have determined that upregulation of proinflammatory chemokines during WNV encephalitis may occur in a region-specific fashion [2, 20]. For example, cerebellar expression of CXCL10 is required for viral clearance of this brain region by CXCR3-expressing, virus-specific CD8+ T cells . Differences in regional chemokine expression may thus determine the spatial patterns of leukocyte trafficking, leading to variability in viral clearance and immunopathology between CNS regions.
CCL3–5, chemokines that all bind the chemokine receptor CCR5, are strongly induced in the CNS after WNV infection [2, 21–23]. Monocytes, NK and T cells express CCR5 and targeted deletion of CCR5 in B6129PF2 mice is associated with depressed leukocyte trafficking, increased viral burden and enhanced mortality . Similarly, homozygosity for CCR5Δ32, a nonfunctional variant of chemokine receptor CCR5, is markedly increased among symptomatic WNV-seropositive patients [22, 24]. In the current study, we examined the role of CCR5 in C57BL/6 mice during WNV infection, focusing on CNS region-specific effects. We found that CCR5 is required for virologic control specifically within the CNS cortex. This finding was associated with a significant decrease in immune cell infiltrates, increased blood-brain barrier (BBB) permeability, and elevated levels of CCR5 ligands in WNV-infected CCR5−/− compared with WT mice. These data suggest that increased viral replication within the CNS modulates BBB function and support the notion that nonredundancy in chemokine-mediated inflammation among CNS regions may be due to evolutionary mechanisms.
Murine model of WNV encephalitis
Eight-week-old C57BL/6 wild-type mice were obtained commercially (Jackson Laboratories). Congenic ccr5 −/− mice were also commercially obtained (Jackson Laboratories, stock number 005427) and bred in the animal facilities at the Washington University School of Medicine. All animals were housed under pathogen-free conditions in the animal facilities of the Washington University School of Medicine. All experiments were performed in compliance with Washington University animal studies guidelines and comply with “Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines.” Mice were inoculated subcutaneously via footpad injection (50 μl) with 100 PFU of WNV as previously described . WNV strain 3000.0259 was isolated in New York in 2000  and passaged once in C6/36 cells. Viral titers were measured by plaque assay on BHK21-15 cells as previously described . Cortical, cerebellar, and brainstem regions of the CNS were dissected based on visual information, such as differences in color of adjacent tissues, and on the natural anatomical boundaries of certain regions present in the brain. Clinical disease was monitored and scored as previously described . The designation for the clinical scores is as follows: 1 ruffled fur/hunched, 2 paresis/difficulty walking, 3 paralysis, 4 moribund, and 5 dead.
BBB permeability assay
At day 5 and 8 after WNV infection, mice were injected intraperitoneally (IP) with sodium fluorescein dye (100 mg/ml) as previously described . After 45 min, mice were perfused, serum was collected, and CNS tissues were harvested. Both the serum and tissue homogenates were incubated overnight at 4 °C in 2 % trichloroacetic acid at 1:1 dilution to precipitate protein and then the supernatants were neutralized in equal volumes of borate buffer. Fluorescence emission at 538 nm was determined using a microplate reader Synergy™ H1 and Gen5™ software (BioTek Instruments, Inc.). Fluorescence concentration was calculated from a standard curve, and tissue fluorescence values were normalized to serum fluorescence values from identical mice.
The chemokine Bio-Plex assay was performed on tissue samples from mice at day 8 after infection from WNV-infected mice. Tissue was homogenized following extensive cardiac perfusion with PBS and was analyzed using a 6-plex Luminex assay (Millipore) followed by analysis on a Bio-Plex 200 (Bio-Rad). Concentrations of chemokines were normalized to total protein levels.
Cells were isolated from the CNS of WT and CCR5−/− mice at day 6 and 8 after WNV infection. Following cell count and viability analysis, cells were stained with fluorescently conjugated antibodies to CD4, CD8β, CD11b, and CD45 as previously described . Samples were analyzed following staining using a LSRII flow cytometer (Beckton Dickinson) to collect up to 30,000 events in a broad gate defined by forward- and side-scatter attributes. The absolute count of respective leukocyte subsets was calculated based on the percent positive cells from data analysis, which was performed using FlowJo software (Tree Star).
Brain tissues were perfused with 4 % paraformaldehyde and isolated for frozen sections. Tissue sections were permeabilized and blocked in 0.1 % Triton X-100 and 10 % goat serum, followed by incubation with primary antibodies CD-3 1:200 (Dako) and WNV antigen 1:100 (Diamond lab) or CD-3 1:200 (Dako), CD31 1:20 (BD) and endomucin 1:200 (eBioscience/Affymetrix) overnight at 4 °C. Primary antibodies were detected with goat anti-rabbit Alexa Fluor 488 (CD3), goat anti-rat Alexa Fluor 555 (WNV antigen) and goat anti-rat Alexa Fluor 555 (CD31, endomucin) followed by nuclear DAPI counterstaining. Tissues were then washed and examined by confocal microscopy.
Statistical comparisons between groups were made with one- or two-way analysis of variance; Bonferroni’s post hoc test was subsequently used to compare individual means. Survival was analyzed by the log-rank test. Analyses were conducted using Prism software (GraphPad Prism). All data were expressed as means ± SEM. Differences were considered significant if P ≤ 0.05.
CCR5-deficient mice exhibit higher cortical viral loads with increased BBB permeability
CCR5-deficient mice exhibit increased chemokines within the cortex with less inflammation
While CCR5 has been shown to be critical for survival of WNV-infected mice, region-specific roles for this receptor within the CNS has not been previously described. Here, we demonstrate that targeted deletion of CCR5 leads to loss of virologic control specifically within cortical tissues of the CNS, which also exhibit enhanced BBB permeability and elevated levels of CCR5 ligands, CCL3 and CCL5. Despite these latter findings, WNV-infected CCR5−/− mice showed decreased infiltration of all mononuclear cells into cortical regions compared with similarly infected WT animals. These data suggest region-specific roles for CCR5 ligands within the virally infected CNS.
Although mortality of WNV-infected, CCR5−/− mice were only moderately increased, symptomatic disease was significantly enhanced. This is reminiscent of findings in patients with the Δ32CCR5 mutation, which exhibit increase in symptomatic diseases after WNV infection, but no increase in mortality [22, 34]. In previous studies using CCR5−/− mice on a B6129PF2 background, infection with WNV-NY99 led to uniform mortality. The severe phenotype observed in CCR5-deficient mice may be due to strain related variations in inflammatory responses between C57BL/6 versus mixed background (B6129PF2) strains .
Increased symptoms of encephalitis in CCR5-deficient mice in our study were associated with increased cortical viral loads. This region-specific effect is consistent with the other reports of regional differences in expression of inflammatory molecules during viral infections of the CNS [33, 36]. The lack of effect of CCR5-deficiency on viral loads in the periphery or in other CNS regions suggests that CCR5+ antiviral lymphocytes are specifically required for virologic control within the cortex. Of note, adoptive transfer of CCR5-sufficient cells into WNV-infected, CCR5-deficient mice were previously shown to control viral replication and improve survival , which may have been due to improved virologic control in the forebrain. This is similar to the findings regarding the role of CXCR3, which is dispensable for the control of viral infection in the periphery and in most CNS compartments but required for CD8 T cell-mediated antiviral responses specifically within the cerebellum. That evolutionary older CXC chemokines would be important for virologic control in the hindbrain while CC chemokines, which evolved with adaptive immunity, would play a bigger antiviral role in the forebrain is quite accordant [19, 37, 38]. Recent studies also indicate that activated microglia in the setting of WNV encephalitis express proinflammatory cytokines and chemokines . Further studies evaluating regional differences in chemoattractant responses of various neural cell types during viral infections should yield important insights regarding how viruses differentially impact CNS antiviral responses.
We previously showed that clearance of virus is associated with prompt resolution of encephalitis, with decreased numbers of infiltrating T cells . Consistent with this, WT mice exhibited increased parenchymal T cells and low levels of WNV antigen at day 8 post-infection followed by decreased numbers of CNS-infiltrated leukocytes at day 10 post-infection. CNS infiltrating T cells and macrophages express cytokines, such as interferon (IFN)-γ, tumor necrosis factor (TNF)-α, and interleukin (IL)-1β, may induce disruption of interendothelial cell tight junctions (TJs) via activation of the GTPase RhoA [11, 29]. The resulting increase in BBB permeability is associated with the migration of mononuclear cells into the CNS parenchyma, with enhanced viral clearance . In our study, despite the increase in BBB permeability observed in cortical tissues of WNV-infected, CCR5−/− mice, significantly fewer leukocytes trafficked into this CNS region. The increase in BBB permeability in the setting of high MOIs of WNV is likely due to effects of inflammasome activation, which disrupts tight junction integrity via effects of IL-1β . In addition, high MOIs of WNV which induce necrotic cell death and cause cells to release immunogenic factors as they die including inflammatory cytokines like HMGB1, while low MOIs induce apoptosis, which is typically non-immunogenic . This is consistent with prior studies showing that pattern recognition receptor activation in the presence of WNV in the BBB endothelium results in cytokine-dependent modulation of Rho GTPase signaling, exerting regulatory control over BBB permeability and TJ integrity . Thus, it is possible that the high viral loads within the cortices of WNV-infected, CCR5−/− mice induce viral sensing mechanisms, such as activation of Toll-like receptor 3 [29, 42], that promote loss of BBB integrity.
In conclusion, our data identify CCR5 as critically important for cell-mediated immunity during cortical infections with neurotropic virus. Loss of CCR5 results in decreased ability to recruit antiviral mononuclear cells specifically into WNV-infected, cortical tissues, which are essential for virologic control. The data also raise new questions regarding the effects of high CNS viral loads on BBB permeability and the differential expression of chemokines within brain regions during viral infections of the CNS.
This study was supported by the National Institutes of Health/National Institute of Neurological Disorders and Stroke Grant R01NS052632 and National Institute of Allergy and Infectious Diseases Grant U19 AI083019 (both to R.S.K). B.P.D. was supported by NSF (DGE-1143954) and NIH (F31-NS07866-01) Fellowships.
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.
- Weaver SC, Reisen WK. Present and future arboviral threats. Antiviral Res. 2010;85:328–45.PubMed CentralView ArticlePubMedGoogle Scholar
- Klein RS, Lin E, Zhang B, Luster AD, Tollett J, Samuel MA, et al. Neuronal CXCL10 directs CD8+ T-cell recruitment and control of West Nile virus encephalitis. J Virol. 2005;79:11457–66.PubMed CentralView ArticlePubMedGoogle Scholar
- Shrestha B, Diamond MS. Role of CD8+ T cells in control of West Nile virus infection. J Virol. 2004;78:8312–21.PubMed CentralView ArticlePubMedGoogle Scholar
- Wang Y, Lobigs M, Lee E, Mullbacher A. CD8+ T cells mediate recovery and immunopathology in West Nile virus encephalitis. J Virol. 2003;77:13323–34.PubMed CentralView ArticlePubMedGoogle Scholar
- Kleinschmidt-DeMasters BK, Marder BA, Levi ME, Laird SP, McNutt JT, Escott EJ, et al. Naturally acquired West Nile virus encephalomyelitis in transplant recipients: clinical, laboratory, diagnostic, and neuropathological features. Arch Neurol. 2004;61:1210–20.View ArticlePubMedGoogle Scholar
- Cushing MM, Brat DJ, Mosunjac MI, Hennigar RA, Jernigan DB, Lanciotti R, et al. Fatal West Nile virus encephalitis in a renal transplant recipient. Am J Clin Pathol. 2004;121:26–31.View ArticlePubMedGoogle Scholar
- Katz LM, Bianco C. West Nile virus. N Engl J Med. 2003;349:1873–4. author reply 1873-1874.View ArticlePubMedGoogle Scholar
- Wang T, Gao Y, Scully E, Davis CT, Anderson JF, Welte T, et al. Gamma delta T cells facilitate adaptive immunity against West Nile virus infection in mice. J Immunol. 2006;177:1825–32.View ArticlePubMedGoogle Scholar
- Wang T, Scully E, Yin Z, Kim JH, Wang S, Yan J, et al. IFN-gamma-producing gamma delta T cells help control murine West Nile virus infection. J Immunol. 2003;171:2524–31.View ArticlePubMedGoogle Scholar
- Wang Y, Lobigs M, Lee E, Koskinen A, Mullbacher A. CD8(+) T cell-mediated immune responses in West Nile virus (Sarafend strain) encephalitis are independent of gamma interferon. J Gen Virol. 2006;87:3599–609.View ArticlePubMedGoogle Scholar
- Durrant DM, Daniels BP, Klein RS. IL-1R1 signaling regulates CXCL12-mediated T cell localization and fate within the central nervous system during West Nile virus encephalitis. J Immunol. 2014;193:4095–106.PubMed CentralView ArticlePubMedGoogle Scholar
- Durrant DM, Robinette ML, Klein RS. IL-1R1 is required for dendritic cell-mediated T cell reactivation within the CNS during West Nile virus encephalitis. J Exp Med. 2013;210:503–16.PubMed CentralView ArticlePubMedGoogle Scholar
- Ali M, Safriel Y, Sohi J, Llave A, Weathers S. West Nile virus infection: MR imaging findings in the nervous system. AJNR Am J Neuroradiol. 2005;26:289–97.PubMedGoogle Scholar
- Kastrup O, Wanke I, Maschke M. Neuroimaging of infections. NeuroRx. 2005;2:324–32.PubMed CentralView ArticlePubMedGoogle Scholar
- Tien RD, Felsberg GJ, Osumi AK. Herpesvirus infections of the CNS: MR findings. AJR Am J Roentgenol. 1993;161:167–76.View ArticlePubMedGoogle Scholar
- Steiner I, Budka H, Chaudhuri A, Koskiniemi M, Sainio K, Salonen O, et al. Viral encephalitis: a review of diagnostic methods and guidelines for management. Eur J Neurol. 2005;12:331–43.View ArticlePubMedGoogle Scholar
- Kim CH. Chemokine-chemokine receptor network in immune cell trafficking. Curr Drug Targets Immune Endocr Metabol Disord. 2004;4:343–61.View ArticlePubMedGoogle Scholar
- Butcher EC, Williams M, Youngman K, Rott L, Briskin M. Lymphocyte trafficking and regional immunity. Adv Immunol. 1999;72:209–53.View ArticlePubMedGoogle Scholar
- Huising MO, Stet RJ, Kruiswijk CP, Savelkoul HF, Lidy Verburg-van Kemenade BM. Molecular evolution of CXC chemokines: extant CXC chemokines originate from the CNS. Trends Immunol. 2003;24:307–13.PubMedGoogle Scholar
- Zhang B, Chan YK, Lu B, Diamond MS, Klein RS. CXCR3 mediates region-specific antiviral T cell trafficking within the central nervous system during West Nile virus encephalitis. J Immunol. 2008;180:2641–9.View ArticlePubMedGoogle Scholar
- Glass WG, Lim JK, Cholera R, Pletnev AG, Gao JL, Murphy PM. Chemokine receptor CCR5 promotes leukocyte trafficking to the brain and survival in West Nile virus infection. J Exp Med. 2005;202:1087–98.PubMed CentralView ArticlePubMedGoogle Scholar
- Glass WG, McDermott DH, Lim JK, Lekhong S, Yu SF, Frank WA, et al. CCR5 deficiency increases risk of symptomatic West Nile virus infection. J Exp Med. 2006;203:35–40.PubMed CentralView ArticlePubMedGoogle Scholar
- Shirato K, Kimura T, Mizutani T, Kariwa H, Takashima I. Different chemokine expression in lethal and non-lethal murine West Nile virus infection. J Med Virol. 2004;74:507–13.View ArticlePubMedGoogle Scholar
- Lim JK, Louie CY, Glaser C, Jean C, Johnson B, Johnson H, et al. Genetic deficiency of chemokine receptor CCR5 is a strong risk factor for symptomatic West Nile virus infection: a meta-analysis of 4 cohorts in the US epidemic. J Infect Dis. 2008;197:262–5.View ArticlePubMedGoogle Scholar
- Engle MJ, Diamond MS. Antibody prophylaxis and therapy against West Nile virus infection in wild-type and immunodeficient mice. J Virol. 2003;77:12941–9.PubMed CentralView ArticlePubMedGoogle Scholar
- Ebel GD, Dupuis 2nd AP, Ngo K, Nicholas D, Kauffman E, Jones SA, et al. Partial genetic characterization of West Nile virus strains, New York State, 2000. Emerg Infect Dis. 2001;7:650–3.PubMed CentralView ArticlePubMedGoogle Scholar
- Brien JD, Lazear HM, Diamond MS. Propagation, quantification, detection, and storage of West Nile virus. Curr Protoc Microbiol. 2013;31:15D 13 11-15D 13 18.Google Scholar
- Lanteri MC, O'Brien KM, Purtha WE, Cameron MJ, Lund JM, Owen RE, et al. Tregs control the development of symptomatic West Nile virus infection in humans and mice. J Clin Invest. 2009;119:3266–77.PubMed CentralPubMedGoogle Scholar
- Daniels BP, Holman DW, Cruz-Orengo L, Jujjavarapu H, Durrant DM, Klein RS. Viral pathogen-associated molecular patterns regulate blood-brain barrier integrity via competing innate cytokine signals. mBio. 2014;5:e01476.PubMed CentralView ArticlePubMedGoogle Scholar
- McCandless EE, Wang Q, Woerner BM, Harper JM, Klein RS. CXCL12 limits inflammation by localizing mononuclear infiltrates to the perivascular space during experimental autoimmune encephalomyelitis. J Immunol. 2006;177:8053–64.View ArticlePubMedGoogle Scholar
- Roe K, Kumar M, Lum S, Orillo B, Nerurkar VR, Verma S. West Nile virus-induced disruption of the blood-brain barrier in mice is characterized by the degradation of the junctional complex proteins and increase in multiple matrix metalloproteinases. J Gen Virol. 2012;93:1193–203.PubMed CentralView ArticlePubMedGoogle Scholar
- Wang P, Dai J, Bai F, Kong KF, Wong SJ, Montgomery RR, et al. Matrix metalloproteinase 9 facilitates West Nile virus entry into the brain. J Virol. 2008;82:8978–85.PubMed CentralView ArticlePubMedGoogle Scholar
- Glass WG, McDermott DH, Lim JK, Lekhong S, Yu SF, Frank WA, et al. CCR5 deficiency increases risk of symptomatic West Nile virus infection. J Exp Med. 2006;203(1):35-40.Google Scholar
- Lim JK, McDermott DH, Lisco A, Foster GA, Krysztof D, Follmann D, et al. CCR5 deficiency is a risk factor for early clinical manifestations of West Nile virus infection but not for viral transmission. J Infect Dis. 2010;201:178–85.PubMed CentralView ArticlePubMedGoogle Scholar
- Sellers RS, Clifford CB, Treuting PM, Brayton C. Immunological variation between inbred laboratory mouse strains: points to consider in phenotyping genetically immunomodified mice. Vet Pathol. 2012;49:32–43.View ArticlePubMedGoogle Scholar
- Phares TW, Kean RB, Mikheeva T, Hooper DC. Regional differences in blood-brain barrier permeability changes and inflammation in the apathogenic clearance of virus from the central nervous system. J Immunol. 2006;176:7666–75.View ArticlePubMedGoogle Scholar
- Magor BG, Magor KE. Evolution of effectors and receptors of innate immunity. Dev Comp Immunol. 2001;25:651–82.View ArticlePubMedGoogle Scholar
- Shields DC. Molecular evolution of CXC chemokines and receptors. Trends Immunol. 2003;24:355. author reply 356-357.View ArticlePubMedGoogle Scholar
- Quick ED, Leser JS, Clarke P, Tyler KL. Activation of intrinsic immune responses and microglial phagocytosis in an ex vivo spinal cord slice culture model of West Nile virus infection. J Virol. 2014;88:13005–14.PubMed CentralView ArticlePubMedGoogle Scholar
- McCandless EE, Zhang B, Diamond MS, Klein RS. CXCR4 antagonism increases T cell trafficking in the central nervous system and improves survival from West Nile virus encephalitis. Proc Natl Acad Sci U S A. 2008;105:11270–5.PubMed CentralView ArticlePubMedGoogle Scholar
- Chu JJ, Ng ML. The mechanism of cell death during West Nile virus infection is dependent on initial infectious dose. J Gen Virol. 2003;84:3305–14.View ArticlePubMedGoogle Scholar
- Wang T, Town T, Alexopoulou L, Anderson JF, Fikrig E, Flavell RA. Toll-like receptor 3 mediates West Nile virus entry into the brain causing lethal encephalitis. Nat Med. 2004;10:1366–73.View ArticlePubMedGoogle Scholar