Human Genetic Variation Altering Anthrax Toxin Sensitivity

Human genetic variation altering anthrax toxin sensitivity

Mikhail Martchenkoa, Sophie I. Candillea, Hua Tanga, and Stanley N. Cohena,b,1

Departments of aGenetics and bMedicine, Stanford University School of Medicine, Stanford, CA 94305

Contributed by Stanley N. Cohen, December 28, 2011 (sent for review October 31, 2011) The outcome of exposure to infectious microbes or their toxins is disease therapies (24), and genetic variation in the expression or influenced by both microbial and host genes. Some host genes sequence of these loci potentially may affect individual disease encode defense mechanisms, whereas others assist pathogen susceptibility or the efficacy of treatment. functions. Genomic analyses have associated host gene mutations Anthrax is a lethal disease caused by the toxic actions of with altered infectious disease susceptibility, but evidence for cau- proteins produced by the bacterium Bacillus anthracis: lethal sality is limited. Here we demonstrate that human genetic varia- factor (LF) and edema factor (EF) (19). Entry of each of these tion affecting capillary morphogenesis gene 2 (CMG2), which proteins into cells is mediated by interaction with a third B. encodes a host membrane protein exploited by anthrax toxin as anthracis-generated protein, protective antigen (PA). Three host a principal receptor, dramatically alters toxin sensitivity. Lympho- cell proteins [capillary morphogenesis gene 2 (CMG2), tumor blastoid cells derived from a HapMap Project cohort of 234 persons endothelial marker 8 (TEM8), and integrin β1 (ITGB1)], which of African, European, or Asian ancestry differed in sensitivity me- can interact with each other physically or functionally (25, 26), diated by the protective antigen (PA) moiety of anthrax toxin by can serve as receptors for the bipartite PA/LF and PA/EF toxins more than four orders of magnitude, with 99% of the cohort (26–28). Whereas only CMG2, TEM8, and ITGB1 are known showing a 250-fold range of sensitivity. We find that relative sen- ligands of PA, 14 additional CGEPs can assist PA binding and/or sitivity is an inherited trait that correlates strongly with CMG2 internalization (26, 29–33). mRNA abundance in cells of each ethnic/geographical group and While investigating events that affect anthrax toxin entry, we −11 in the combined population pool (P =4× 10 ). The extent of observed that toxin sensitivity varies extensively among human GENETICS CMG2 expression in transfected murine macrophages and human lymphoblastoid cell populations derived identically from differ- lymphoblastoid cells affected anthrax toxin binding, internaliza- ent persons. We show here that such variation in toxin sensitivity tion, and sensitivity. A CMG2 single-nucleotide polymorphism is inherited and that variation in cellular expression of CMG2 (SNP) occurring frequently in African and European populations is a prominent determinant of outcome of toxin exposure. We independently altered toxin uptake, but was not statistically asso- further show that a single-nucleotide polymorphism in the pro- ciated with altered sensitivity in HapMap cell populations. Our tein-coding region of CMG2 independently affects toxin uptake. results reveal extensive human diversity in cell lethality dependent Our findings, which reveal the previously unsuspected magnitude on PA-mediated toxin binding and uptake, and identify individual of genetically determined differences in toxin sensitivity among differences in CMG2 expression level as a determinant of this di- cells from different individuals, suggest a broadly applicable ap- versity. Testing of genomically characterized human cell popula- proach for investigating pathogen susceptibility in diverse human tions may offer a broadly useful strategy for elucidating effects of populations. genetic variation on infectious disease susceptibility. Results lethal factor | FP59 | cellular genes exploited by pathogens | genotype Variation in Cellular Sensitivity to PA-Mediated Toxin Lethality in Human Populations. During investigations of host mechanisms he clinical outcome of microbial infections has long been implicated in PA-mediated toxin entry, we observed that Epstein- Tknown to vary among different individuals and among ethni- Barr virus (EBV)-immortalized lymphoblastoid cell lines derived cally and geographically diverse human populations (1). A genetic from 234 individuals in geographically and ethnically diverse human populations [84 Yoruba in Ibadan, Nigeria (YRI), 63 basis for differences in infection susceptibility was first suggested Utah residents with ancestry from northern and western Europe by comparison of disease incidence in monozygotic vs. dizygotic (CEU), 44 Japanese in Tokyo, Japan (JPT), and 43 Han Chinese twins and in members of adoptee families (2). Later findings in Beijing, China (CHB)] as part of the HapMap Project (7) identified genetic traits that affect host ability to mount an im- fi showed a remarkable 30,000-fold difference in lethality to a toxin mune response (3) or protect against speci c pathogens (4). As- complex composed of PA plus FP59, a hybrid toxin that has been sociations between human genetic variation and infectious disease widely used as LF surrogate but kills cells by a different mech- incidence subsequently have been shown by linkage analysis (for anism and is lethal to lymphocytes (34, 35) (Fig. 1 A and B). The reviews, see refs. 5, 6), and by genome-wide association studies cell lines from three CEU individuals showed extraordinary re- (GWASs) using haplotype maps of the human genome (7), high- sistance to toxin lethality and removing these cells from the density microarray analysis (8, 9), and DNA sequencing methods analysis yielded a still surprisingly broad 250-fold range in LD – 50 (7, 10 13). However, GWAS normally requires examination of values among the remaining 99% of the population. The ex- fi a very large cohort to generate statistically signi cant associations traordinary toxin resistance of the three “outlier” lines was spe- (14), and inherently lacks the ability to demonstrate causality between a variant and the disease phenotype. fl Host variation has the potential to in uence infectious disease Author contributions: M.M., H.T., and S.N.C. designed research; M.M., S.I.C., and H.T. outcome not only through genes that combat pathogens (15, 16), performed research; M.M., S.I.C., H.T., and S.N.C. analyzed data; and M.M. and S.N.C. but also through host genes that are needed for pathogen prop- wrote the paper. agation (14, 17) or the detrimental effects of microbial toxins (18). The authors declare no conflict of interest. Among the pathogens and toxins that have been shown experi- Freely available online through the PNAS open access option. mentally to exploit host genes are HIV, malaria, hepatitis, lep- 1To whom correspondence should be addressed. E-mail: [email protected]. – – rosy, and anthrax toxin (11 13, 19 22). Cellular genes exploited This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. by pathogens (CGEPs) (23) may be targets for novel infectious 1073/pnas.1121006109/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1121006109 PNAS Early Edition | 1of6 Fig. 1. Human lymphoblastoid cells sensitivity to PA-mediated toxicity of FP59. (A) 234 EBV-transformed B lymphocytes (7) were treated with 500 ng/mL PA plus FP59 added at the concentration shown. Cell viability was determined by MTT assay (Materials and Methods) and is shown as the percentage of survivors relative to cells treated with

PA alone. The LD50 calculation for the most sensitive cell line is shown as an example. (B)LD50 values (ng/mL of FP59) were calculated and expressed on

an inverse log10 scale. For calculations, sensitivity is defined numerically as 1/LD50.(C) Heritability of log sensitivity in Yoruba trios. Plot of the log sensitivity of the children against the mean log sensitivity of the parents. The heritability (0.71) is estimated as the slope of the regression of the children phenotype on the midparent phenotype. (D) Population-specific distribution of sensitivities. CEU, YRI, JPT, and CHB denote European, Yoruba, Japanese, and Chinese Han, respectively. Three CEU outliers are excluded. For each population, the black bar represents the median log sensitivity; the box extends from the lower to the upper quartile and the whiskers extend to the most extreme data point that is no more than 1.5 times the inter- quartile range from the box. (E) CMG2 expression in 234 EBV-transformed B lymphocytes. Expression of CMG2 was measured by qRT-PCR and normalized for the expression of β-actin. CMG2 expression was calculated as CT ACTB − CT CMG2, where CT is the threshold cycle. cific, rather than general, as these three lines were found to show dance varied over a 100-fold range in the lymphoblastoid cell normal sensitivity to Staphylococcus aureus α-toxin. Analysis of populations we tested and is normally distributed (Fig. 1E). toxin sensitivity in parent/children trios, as measured by the dose Analysis of HapMap Project release 28 revealed no cis-acting required to kill half of the cells [log(1/LD50)], indicated that eQTL loci that might account for such variation. Additionally, relative sensitivity is a heritable trait (permutation P-value < analysis in this release of genetic variation in the 17 CGEPs C 0.05) (Fig. 1 ). Interestingly, whereas the measured range in reported to alter anthrax toxin entry into cells (26–32) revealed that toxin lethality was similar in all four human cell populations, the a SNP affecting an amino acid within a proline-rich region of the European Caucasian population on average showed only about cytosolic tail of CMG2 (P357A; rs12647691) was among the most one-third the cellular sensitivity to PA-mediated toxin killing − frequent nonsynonymous SNPs observed in the lymphoblastoid observed in other ethnic groups (P =4× 10 12) (excluding the − cells tested. Whereas this SNP was not associated with the observed three CEU highly insensitive outliers, P =2× 10 11) (Fig. 1D). variation in CMG2 expression in the human cell populations we The widespread and unimodal distribution observed for log(1/ P n P LD ) (Fig. 1B) is consistent with a polygenic inheritance model, tested [CEU, =0.98, = 62 (excluding 3 CEU outliers =0.41, 50 n P n which also is implied from evidence that multiple genes can affect = 59); YRI, = 0.62, = 84], interestingly, its allele frequency fi anthrax toxin lethality (26–32). Among these genes is CMG2,which varied signi cantly among different ethnic populations. Proline, encodes a type 1 transmembrane glycoprotein that recently has which is the ancestral amino acid found at position 357 in most been shown to be the primary anthrax toxin receptor in rodents placental mammals, was replaced by alanine at this location at (36). We found by RT-PCR analysis that CMG2 mRNA abun- a frequency of 34 and 36% in YRI and CEU, respectively, whereas

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1121006109 Martchenko et al. the ancestral allele persisted at 97% frequency in the HapMap little effect when similarly expressed (Fig. 2A, Left). However, Project’s East Asian populations CHB and JPT (Table S1). higher expression of the 357A allele (clones 357A1-5 and 357A1- 40) was associated with restoration of toxin lethality to approx- Effects of Human CMG2 Expression and P357A SNP on Anthrax Toxin imately the level observed for the parental cell line and for Internalization and Toxin Sensitivity. We designed experiments to transfected clones that express lesser amounts of hCMG2-357P determine directly whether either variation in cellular expression (Fig. 2A Center vs. 2A, Left). A clone (357P1-0.5) that produced of CMG2 or the occurrence of the P357A SNP in the cytosolic a lower amount of hCMG2-357P showed lesser sensitivity to the tail of this receptor protein modulates cellular lethality to an- PA/LF complex. Together these ﬁndings indicate that both the thrax toxin. In these experiments, cultured murine macrophages SNP affecting hCMG2 amino acid residue 357 and the extent of expressing hCMG2 were challenged by exposure to PA plus LF; cellular expression of the CMG2 gene govern sensitivity to an- as the sequence of murine CMG2 (mCMG2) differs at multiple thrax lethal toxin. They further argue that these variables act locations from the sequence of its human counterpart, it was independently in affecting toxin lethality. practical to largely inactivate mCMG2 using an shRNA construct Additional experiments elucidated the mechanisms that un- that does not target the human sequence. We selected a murine derlie the above observations: immunoﬂuorescence microscopy macrophage clone (RAW267.4–21.2) in which endogenous (Fig. 2B) and Western blotting experiments (Fig. 2C) showed that CMG2 mRNA had been reduced to <40% of normal, and stably both the extent of CMG2 expression and genetic variation af- transfected constructs expressing either the ancestral hCMG2 fecting amino acid residue 357, which is located in a hCMG2 re- protein (357P) or its variant derivative (357A), under control of gion known to be functionally important for endocytosis of toxin a cytomegalovirus promoter (28), into cells of this clone. As (29, 30), are determinants of PA internalization. The Western expression of such DNA is affected by the chromosomal in- blotting experiment also showed that the amount of PA bound sertion site (37), quantitative RT-PCR was used to identify to the macrophage cell surface was modulated by the extent of RAW267.4–21.2 subclones that express 357A or 357P mRNA to expression of either CMG2 357A or 357P. Adventitious expres- approximately the same extent, and at an abundance that when sion of either hCMG2 357P or 357A in human lymphoblastoid cell added to residual endogenous mCMG2 abundance, yielded line GM06986, which shows low endogenous CMG2 expression a total cellular CMG2 mRNA abundance that corresponds to the and is highly resistant to killing by FP59 carried into the cells by level determined for parental RAW264.7 cells (Fig. S1). the same protein (i.e., PA) that is the native carrier for anthrax 3 MTT assay of toxin lethality in six independently isolated lethal factor, generated cell populations showing an ∼2.5 × 10 GENETICS stably transfected 357A or 357P clones showed that introduction increase in toxin lethality (Table S2). However, RT-PCR analysis of the proline-357 construct restored toxin sensitivity to almost indicated that hCMG2 expression in the 357A population was 30 the pre–shRNA-treatment level, whereas the 357A allele had times higher than expression in the hCMG2 357P population.

Fig. 2. CMG2 P357A SNP and expression levels affect toxin sensitivity in RAW264.7 cells. (A) Toxin sensitivities of various RAW264.7 cells expressing different human CMG2 variants at different relative amounts (indicated as the last number in the cell line name). Each data point represents the average of results from three wells with SD bars shown. (B) Fluorescence microscopy showing Alexa Fluor 647-labeled toxin binding and toxin endocytosis in RAW264.7 cells expressing different human CMG2 variants at various levels. (C) Western blot analysis of the inﬂuence of CMG2 genetic variation and amounts on monomeric PA binding to RAW264.7 cell surface and on the internalization of heptameric PA in the cytoplasm. Numbers represent relative intensities of PA normalized for β-actin.

Martchenko et al. PNAS Early Edition | 3of6 Normalization for such differential expression indicated that We tested the possibility that differences in growth rate be- hCMG2 357P-dependent lethality of FP59 carried into cells by PA tween cell lines may be a confounder of the association between is 24 times greater than the lethality mediated by hCMG2- log sensitivity and CMG2 expression. We found that including encoded 357A. growth as a covariate has little effect on the significance of the association between log sensitivity and CMG2 expression (P = − − Association of CMG2 mRNA Abundance with Toxin Sensitivity in 1 × 10 8 compared with P =3× 10 9 not correcting for growth). Human Cells. To learn whether the effects of hCMG2 expression Therefore, the association of log sensitivity with CMG2 expression and the P357A SNP on anthrax toxin lethality demonstrated in is not confounded by differences in growth rate. the experiments described above account for any of the variation A statistically significant association between the P357A SNP, in sensitivity of human lymphoblastoid cells to PA-bound FP59 which affects toxin lethality in isogenic murine cells expressing seen in Fig. 1, we tested the association of each parameter with the variant hCMG2 protein (Fig. 2), with either the extent of toxin sensitivity. We detected a highly significant positive asso- CMG2 − expression or toxin sensitivity was not detected in the ciation in the overall population pool (P =4× 10 11, n = 234, − limited number of lymphoblastoid cell populations we studied excluding CEU outliers P =3× 10 9, n = 231) and also positive (Fig. 3B). This finding suggests that the effects we have dem- association between CMG2 expression and toxin sensitivity in − onstrated for this SNP in cell culture experiments are modulated each of the individual populations (CEU, P =4× 10 5, n = 62, in genetically complex cell populations by the actions of other excluding CEU outliers P = 0.003, n = 59); YRI, P = 0.002, n = – − genes altering toxin internalization (26 32), raising the prospect 84; CHB and JPT combined, P =8× 10 5, n = 87) (Fig. 3A). that larger cohorts may be required for detection of the effects of Moreover, a linear regression model indicated that the CMG2 this SNP in such populations. expression level alone accounts for 11–24% of the observed variation in sensitivity to toxin in various groups (CEU, 24%, Discussion excluding CEU outliers, 15%; YRI, 11%; East Asian popula- Our findings indicate that assay of toxin sensitivity in human cell tions, JPT and CHB, 17%). populations whose DNA sequences are known, together with experimental analysis of the effects of genetically variant CGEP cDNAs expressed in isogenic cells, offers a practical strategy for elucidating factors that underlie individual differences in cellular sensitivity to microbial toxins. The cell-based approach we used seems particularly applicable to lethal infectious diseases that occur at a low incidence in human populations. Application of this strategy has revealed an unexpectedly broad range in cellular toxin sensitivity among individuals in normal human populations and has shown the prominent influence of diversity in CMG2 expression on toxin sensitivity. If extended to multiple genes that are either exploited by or combat pathogens, the approach may facilitate development of bioassays useful in predicting outcome of exposure to pathogens or toxins. CMG2 was identified initially during gene profiling investigations of capillary formation in vitro (38), and inactivation of the gene results in decreased proliferation of human umbilical vein endothelial cells and diminished capillary formation (39). Like the two other known anthrax toxin receptors located at the surface of mammalian cells (TEM8 and integrin β1) (26, 27), CMG2 protein includes a von Willebrand domain (vWA) that binds to multiple extracellular matrix proteins, including collagen IV, laminin, and fibronectin (38). Mutations located within the CMG2 vWA domain, the CMG2 transmembrane region, or the CMG2 cytosolic tail can result in hyaline fibromatosis syn- drome (HFS), a rare and lethal autosomal recessive disease characterized by the accumulation of hyaline material in the skin and other organs (40, 41). Although CMG2 mutations associated with HFS have been identified in the vicinity of the polymorphic locus studied here (i.e., P357A) (40, 41) point mutations affecting the cytosolic tail do not necessarily diminish CMG2 ability to function as an anthrax toxin receptor (42, 43). Similarly, an HFS-associated mutationally truncated CMG2 protein lack- ing the terminal 132 amino acids of the cytosolic tail can mediate toxin endocytosis (43). However, other HFS-causing mutations that impair CMG2 ability to bind its PA ligand decrease CMG2 folding and protein stability, and/or decrease cellular sensitivity Fig. 3. CMG2 expression and P357A genotype associations with the sensi- to anthrax toxin (42–44). Although CMG2 HFS-causing muta- tivity to anthrax of human lymphocytes from four populations. (A)Inallfour tions in general also lead to loss of its toxin-receptor activity (42, populations, increased CMG2 expression is associated with increased sensitiv- 43), the high frequency of the polymorphorphism we observed at ity. In YRI, detected batch effects in expression were statistically removed, and amino acid 357 in European and African populations and the residual log sensitivity and expression are plotted. CMG2 expression is plotted as CT ACTB − CT CMG2. Greater numbers indicate greater CMG2 expression. (B) apparent absence of a correspondingly high HFS disease in these In CEU and YRI where the CMG2 SNP is present at high frequency, the asso- populations (45) suggests that this SNP does not cause HFS. ciation between log sensitivity and the SNP genotype under the assumption The results reported here indicate a previously unsuspected of an additive genetic model is not significant. The width of the boxplot boxes wide range of CMG2 expression and toxin sensitivity in lympho- is drawn proportionally to the square root of the sample number. blastoid cells derived from normal human populations. Whereas

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1121006109 Martchenko et al. multiple host cell genes have been shown to influence anthrax notion is consistent with the observation that P357A variation toxin endocytosis (26–32) or other cell functions required for affects PA internalization (Fig. 2). Alternatively, the single ala- toxicity (19), the ability of CMG2 expression to explain 11–24% of nine for proline replacement may simply affect the PA-induced overall variation in toxin sensitivity indicates that the level of ex- conformational change that initiates the internalization of toxin- pression of CMG2 is a major determinant of the outcome of toxin receptor complexes. exposure in human cells. By contrast, the strongest single instan- ces of genetic variation implicated thus far as determinants of Materials and Methods height or body mass index (BMI), which also is determined by the Chemicals and Reagents. PA and LF were purchased from List Biological Lab- actions of multiple genes, and similarly shows a normal distribu- oratories. FP59 was a kind gift from S. Leppla (National Institute of Allergy and tion, explain only 0.3 and 1% of these traits, respectively (46, 47). Infectious Diseases /National Institutes of Health, Bethesda, MD). Retroviral Efforts to identify cellular determinants of disease susceptibility vector pLEGFP-N1 (Clontech) encoding enhanced green fluorescent protein- by GWAS initially were based on the notion that disease pheno- tagged human CMG2 under control of a cytomegalovirus promoter (28) was types result principally from combined effects of common varia- a gift from J. Young (Salk Institute for Biological Sciences, La Jolla, CA). Hu- man CMG2 cDNA (AY233452) cloned within this vector contained proline tions (14). However, the surprisingly small fraction of phenotypic at CMG2 amino acid 357. We replaced this proline with alanine by replacing variation explained for complex traits by common variants iden- coding nucleotide G1072 by C using site-directed mutagenesis primers: tified in numerous GWASs have led some researchers to suggest TCCACCACCACCCCCCCCTGCACCAAAAGAGGAGGAAGAAGAACCTTTGC and instead that rare genetic variants may account for the bulk of CTTTTGGTGCAGGGGGGGGTGGTGGTGGAGGATCCTTAATAACCACTTTG. disease-related phenotypes (14, 48). Our study casts doubt on this notion for at least the anthrax toxin sensitivity phenotype. Cell Culture and Cell Lines. RAW264.7 mouse macrophage cells were main- Whereas the polymorphism at amino acid residue 357, which is tained in DMEM (Invitrogen) supplemented with 10% FBS (HyClone) and 100 a common variant in CEU and YRI, was demonstrated to de- μg/mL penicillin and 100 μg/mL streptomycin. Human B lymphocytes were crease toxin entry and lethality in murine cells, it was not statis- grown in Iscove’s modified Dulbecco’s medium (IMDM) (Invitrogen) sup- tically associated with altered toxin sensitivity in human cell plemented with 15% FBS (HyClone), 100 μg/mL penicillin, and 100 μg/mL populations. We speculate that the functional effects of the P357A streptomycin. CMG2 knockdown experiments in mouse macrophages were variant we observed during expression of human cDNAs in mouse done as in ref. 26. For human CMG2 expression experiments, we transfected isogenic cells may be obscured in heterogeneous human cell CMG2 knockdown RAW264.7 cells with DNA constructs using FuGENE 6 reagents according to the manufacture’s protocol, followed by single clones populations by the combined effects of multiple other genetic isolation using 1 mg/mL of G418 (Gibco). In this experiment, shRNA is specific GENETICS variants, either common or rare. to the mouse CMG2 sequence and did not affect the expression of human Our data indicate that the effects of variation in the CMG2 CMG2.ForCMG2 overexpression in human lymphocytes, infection with expression level on cellular lethality upon exposure to anthrax a retroviral-based vector expression, the 357A or 357P variant, used methods toxin are mediated at least in part through alterations in PA described in ref. 26, followed by selection using 0.6 mg/mL of G418 (Gibco). binding, suggesting that the concentration of receptor at the cell surface was limiting under the experimental conditions used. mRNA Measurements. Human CMG2 mRNA abundance was measured by real- Measured CMG2 expression in lymphoblastoid cells may or may time PCR using human CMG2-specific primers: AATTGCAGCCCTCAAGTG- not correlate with expression of this gene in other human tissues. TCTGTG and TCAGGATAGGTGCAGGACAAAGCA and normalized against Whether the human diversity in cellular sensitivity to the toxin human β-actin mRNA levels using primers AATGTGGCCGAGGACTTTGATTGC and AGGATGGCAAGGGACTTCCTGTAA in human lymphocytes or against extrapolates to analogous diversity in the outcome of clinical β infection by B. anthracis is not known. Similarly, we do not know mouse -actin in mouse macrophages using primers CTAAGGCCAACCGT- GAAAAG and ACCAGAGGCATACAGGGACA. Mouse CMG2 mRNA abundance whether the demonstrated ability of the P357A variant to alter was measured by real-time PCR using mouse CMG2-specific primers: TAGC- toxin uptake in macrophages being carried in cell culture is TACGATGGCTGCAATCC and TCCTGCACCTGTCCTGAACAA. cDNA labeling predictive of an effect of the SNP in infected persons or whether with SYBR-green dye (Bio-Rad) was measured by the MyiQ2 detection sys- evolutionarily important selection by exposure to anthrax disease tem (Bio-Rad). has influenced the strikingly different frequencies of the P357A variant observed among different populations. However, we note Toxin Treatment and Cell Viability Assays. Mouse macrophages were treated that sequencing of the genomes of B. anthracis strains has with toxin for 4 h and human B lymphocytes for 48 h. Determination of revealed differences in the geographical locations of genetically macrophage viability was done by MTT assay as described in refs. 31, 32, distinct lineages (49): the A clade of B. anthracis is globally whereas lymphocyte viability was determined by Resazurin (Invitrogen) dispersed, whereas the B lineage was found only in Africa and fluorescence, as described by the manufacturer. Each data point shown in Europe. As mortality rates from B. anthracis infections have been Figs. 1 and 2 represents the average and SD of results from three wells. Cell viability is shown as the percentage of survivors obtained relative to observed to be much higher in areas where B strains are endemic treatment by PA alone (100%). LD was calculated for each cell line and than in areas of A-strain prevalence (50–52), exposure to ge- 50 B. anthracis normalized to account for variation in toxin potency from the two dif- netically different strains of may have exerted evo- ferent lot numbers used, as determined in five and three independent lutionary pressure that has affected the incidence of the SNP in LD50 measurements of a control cell line for the first and second lots, the populations we studied. respectively. Upon binding of PA, the cytosolic tail of CMG2 undergoes a conformational change that leads to phosphorylation of cyto- Immunofluorescence Microscopy. PA protein was labeled with Alexa Fluor 647 solic tyrosines by src-like kinases (30). A consequence of such using the protein labeling kit (A20173; Molecular Probes). We determined by phosphorylation is recruitment of the adapter, β-arrestin, which MTT assay that such labeling did not affect the ability of PA/LF to kill mac- induces the binding of an E3 ubiquitin ligase that ubiquitinates rophages. Fresh serum-free IMDM media was added to cells that contained a lysine residue in the cytosolic tail (29, 30). C16 carbon acyl 1 μg/mL PA/Alexa Fluor 647. Cells were incubated for 20 min either at 4 °C chains are added to one or more cytosolic cysteine residues by for PA-binding analysis or at 37 °C for determination of PA internalization. palmitoyltransferase (53), enabling the receptor to interact with Biochemical Assay of PA Binding and Internalization. Three million macro- heterotetrameric adaptor complex-dependent multimember phage cells were exposed to 1 μg/mL of PA at 4 °C for 1 h for binding assay or at clathrin machinery and to enter cells by clathrin-dependent en- 37 °C for 20 min for internalization assay. Cells were then washed with PBS docytosis (29). Replacement of proline-357 by alanine disrupts solution three times and lysed in RIPA buffer containing a protease inhibitor a segment of proline repeats in the CMG2 cytoplasmic tail, mixture (Roche). Western blot analysis was performed using mouse mono- possibly affecting the ability of this proline-rich region to interact clonal anti-PA antibody PA (C3) sc-52129 (Santa Cruz Biotechnology), or with other proteins mediating the endocytic process (54). This mouse antiactin monoclonal antibody (Sigma-Aldrich). Chemiluminescence

Martchenko et al. PNAS Early Edition | 5of6 of bands and their relative intensities were revealed using a VersaDoc 1000 to the P/P, P/A, and A/A genotypes) and each copy of the minor allele is instrument (Bio-Rad). modeled to have the same additive effect on the phenotype. We did not detect an association between the SNP genotype and log sensitivity in CEU Statistical Analysis of the Association of the CMG2 SNP and CMG2 Gene [P value = 0.54, n = 62 (excluding three CEU outliers, P value = 0.24, n = 59)] Expression with Log Sensitivity. Statistical analyses were performed in R or YRI (P value = 0.53, n = 84), and the SNP was too rare in CHB and JPT to be (http://www.R-project.org/). In each population, we used a linear regression tested (Fig. 3B). to test for the effect of the expression measured as ΔCT = CT ACTB − CT The experimental data reported in this paper are tabulated in Table S3, CMG2 on log sensitivity [log10(1/LD50)]. Because the results are consistent and genotypic data are archived in the International HapMap database between the CHB and JPT, two East Asians populations, we pooled them (www.HapMap.org) (release 28) and in the 1000 Genomes Project (www. to obtain a more accurate estimate of effect size in the larger combined 1000genomes.org). population. Because linear regression is sensitive to outliers, we removed three outliers in log sensitivity in CEU (log sensitivity = −2.3, −2.3, and −2.6) ACKNOWLEDGMENTS. We thank Dr. Stephen Leppla for a kind gift of FP59 but we note that these outliers with low sensitivity also tended to have low toxin and Dr. John Young for CMG2 expression construct. This work was Δ − − − CMG2 expression ( CT = 15.2, 12.7, and 14.3). To test the effect of the supported by Contract HDTRA1-06-C-0039 (to S.N.C.) from the Department SNP on log sensitivity we used an additive genetic model, where genotype of Defense Chemical and Biological Defense program through the Defense is considered as zero, one, or two copies of the minor allele (corresponding Threat Reduction Agency.

1. Modiano D, et al. (1996) Different response to Plasmodium falciparum malaria in west 30. Abrami L, Kunz B, van der Goot FG (2010) Anthrax toxin triggers the activation of src- African sympatric ethnic groups. Proc Natl Acad Sci USA 93:13206–13211. like kinases to mediate its own uptake. Proc Natl Acad Sci USA 107:1420–1424. 2. Sørensen TI, Nielsen GG, Andersen PK, Teasdale TW (1988) Genetic and environ- 31. Lu Q, Wei W, Kowalski PE, Chang AC, Cohen SN (2004) EST-based genome-wide gene mental influences on premature death in adult adoptees. N Engl J Med 318:727–732. inactivation identifies ARAP3 as a host protein affecting cellular susceptibility to 3. Bruton OC, Apt L, Gitlin D, Janeway CA (1952) Absence of serum gamma globulins. anthrax toxin. Proc Natl Acad Sci USA 101:17246–17251. AMA Am J Dis Child 84:632–636. 32. Wei W, Lu Q, Chaudry GJ, Leppla SH, Cohen SN (2006) The LDL receptor-related 4. Allison AC (1954) Protection afforded by sickle-cell trait against subtertian malareal protein LRP6 mediates internalization and lethality of anthrax toxin. Cell 124: infection. BMJ 1:290–294. 1141–1154. 5. Hill AV (2001) The genomics and genetics of human infectious disease susceptibility. 33. Klimpel KR, Molloy SS, Thomas G, Leppla SH (1992) Anthrax toxin protective antigen Annu Rev Genomics Hum Genet 2:373–400. is activated by a cell surface protease with the sequence specificity and catalytic 6. Burgner D, Jamieson SE, Blackwell JM (2006) Genetic susceptibility to infectious dis- properties of furin. Proc Natl Acad Sci USA 89:10277–10281. eases: Big is beautiful, but will bigger be even better? Lancet Infect Dis 6:653–663. 34. Arora N, Klimpel KR, Singh Y, Leppla SH (1992) Fusions of anthrax toxin lethal factor 7. International HapMap Consortium (2005) A haplotype map of the human genome. to the ADP-ribosylation domain of Pseudomonas exotoxin A are potent cytotoxins – Nature 437:1299 1320. which are translocated to the cytosol of mammalian cells. J Biol Chem 267: 8. Bryant PA, Venter D, Robins-Browne R, Curtis N (2004) Chips with everything: DNA 15542–15548. – microarrays in infectious diseases. Lancet Infect Dis 4:100 111. 35. Liu S, Aaronson H, Mitola DJ, Leppla SH, Bugge TH (2003) Potent antitumor activity of 9. Storey JD, et al. (2007) Gene-expression variation within and among human pop- a urokinase-activated engineered anthrax toxin. Proc Natl Acad Sci USA 100:657–662. – ulations. Am J Hum Genet 80:502 509. 36. Liu S, et al. (2009) Capillary morphogenesis protein-2 is the major receptor mediating 10. Consortium TGP; 1,000 Genomes Project Consortium (2010) A map of human genome lethality of anthrax toxin in vivo. Proc Natl Acad Sci USA 106:12424–12429. – variation from population-scale sequencing. Nature 467:1061 1073. 37. Wilson C, Bellen HJ, Gehring WJ (1990) Position effects on eukaryotic gene expres- fi 11. Kamatani Y, et al. (2009) A genome-wide association study identi es variants in the sion. Annu Rev Cell Biol 6:679–714. – HLA-DP locus associated with chronic hepatitis B in Asians. Nat Genet 41:591 595. 38. Bell SE, et al. (2001) Differential gene expression during capillary morphogenesis in 12. Ge D, et al. (2009) Genetic variation in IL28B predicts hepatitis C treatment-induced 3D collagen matrices: Regulated expression of genes involved in basement membrane viral clearance. Nature 461:399–401. matrix assembly, cell cycle progression, cellular differentiation and G-protein signal- 13. Zhang FR, et al. (2009) Genomewide association study of leprosy. N Engl J Med 361: ing. J Cell Sci 114:2755–2773. 2609–2618. 39. Reeves CV, Dufraine J, Young JA, Kitajewski J (2010) Anthrax toxin receptor 2 is ex- 14. Vannberg FO, Chapman SJ, Hill AV (2011) Human genetic susceptibility to in- pressed in murine and tumor vasculature and functions in endothelial proliferation tracellular pathogens. Immunol Rev 240:105–116. and morphogenesis. Oncogene 29:789–801. 15. Ermini L, Wilson IJ, Goodship TH, Sheerin NS (2011) Complement polymorphisms: 40. Hanks S, et al. (2003) Mutations in the gene encoding capillary morphogenesis pro- Geographical distribution and relevance to disease. Immunobiology 217:265–271. tein 2 cause juvenile hyaline fibromatosis and infantile systemic hyalinosis. Am J Hum 16. Cooke GS, Hill AV (2001) Genetics of susceptibility to human infectious disease. Nat Genet 73:791–800. Rev Genet 2:967–977. 41. Dowling O, et al. (2003) Mutations in capillary morphogenesis gene-2 result in the 17. Tavassoli A, et al. (2008) Inhibition of HIV budding by a genetically selected cyclic allelic disorders juvenile hyaline fibromatosis and infantile systemic hyalinosis. Am J peptide targeting the Gag-TSG101 interaction. ACS Chem Biol 3:757–764. Hum Genet 73:957–966. 18. Lin AE, Guttman JA (2010) Hijacking the endocytic machinery by microbial pathogens. 42. Deuquet J, et al. (2009) Systemic hyalinosis mutations in the CMG2 ectodomain Protoplasma 244:75–90. leading to loss of function through retention in the endoplasmic reticulum. Hum 19. Moayeri M, Leppla SH (2009) Cellular and systemic effects of anthrax lethal toxin and – edema toxin. Mol Aspects Med 30:439–455. Mutat 30:583 589. 20. Fellay J, et al. (2007) A whole-genome association study of major determinants for 43. Liu S, Leung HJ, Leppla SH (2007) Characterization of the interaction between anthrax – host control of HIV-1. Science 317:944–947. toxin and its cellular receptors. Cell Microbiol 9:977 987. 21. Jallow M, et al.; Wellcome Trust Case Control Consortium; Malaria Genomic Epide- 44. Deuquet J, Lausch E, Superti-Furga A, van der Goot FG (2011) The dark sides of – miology Network (2009) Genome-wide and fine-resolution association analysis of capillary morphogenesis gene 2. EMBO J 31:3 13. fi malaria in West Africa. Nat Genet 41:657–665. 45. Rahman N, et al. (2002) The gene for juvenile hyaline bromatosis maps to chro- – 22. Kwiatkowski DP (2005) How malaria has affected the human genome and what mosome 4q21. Am J Hum Genet 71:975 980. human genetics can teach us about malaria. Am J Hum Genet 77:171–192. 46. Weedon MN, et al.; Diabetes Genetics Initiative; Wellcome Trust Case Control Con- 23. Chang AC, et al. (2006) Phenotype-based identification of host genes required for sortium (2007) A common variant of HMGA2 is associated with adult and childhood – replication of African swine fever virus. J Virol 80:8705–8717. height in the general population. Nat Genet 39:1245 1250. 24. Cohen SN (2009) in Microbial Evolution and Co-Adaptation: A Tribute to the Life and 47. Frayling TM, et al. (2007) A common variant in the FTO gene is associated with body – Scientific Legacies of Joshua Lederberg, Microbial Drug Resistance: An Old Problem in mass index and predisposes to childhood and adult obesity. Science 316:889 894. Need of New Solutions, eds Relman DA, Hamburg MA, Choffnes ER, Mack A (National 48. McCarthy MI, et al. (2008) Genome-wide association studies for complex traits: Con- Academies Press, Washington, DC), pp 173–180. sensus, uncertainty and challenges. Nat Rev Genet 9:356–369. 25. Jinnin M, et al. (2008) Suppressed NFAT-dependent VEGFR1 expression and consti- 49. Van Ert MN, et al. (2007) Global genetic population structure of Bacillus anthracis. tutive VEGFR2 signaling in infantile hemangioma. Nat Med 14:1236–1246. PLoS ONE 2:e461. 26. Martchenko M, Jeong SY, Cohen SN (2010) Heterodimeric integrin complexes con- 50. World Health Organization (2008) Anthrax in Humans and Animals (WHO, Geneva) taining beta1-integrin promote internalization and lethality of anthrax toxin. Proc 4th Ed, p 12. Natl Acad Sci USA 107:15583–15588. 51. De Vos V, Bryden H (1996) Anthrax in the Kruger National Park: Temporal and spatial 27. Bradley KA, Mogridge J, Mourez M, Collier RJ, Young JA (2001) Identification of the patterns of disease occurence. Salisbury Med Bull S87:26–31. cellular receptor for anthrax toxin. Nature 414:225–229. 52. Fouet A, et al. (2002) Diversity among French Bacillus anthracis isolates. J Clin Mi- 28. Scobie HM, Rainey GJ, Bradley KA, Young JA (2003) Human capillary morphogenesis crobiol 40:4732–4734. protein 2 functions as an anthrax toxin receptor. Proc Natl Acad Sci USA 100: 53. Abrami L, Kunz B, Iacovache I, van der Goot FG (2008) Palmitoylation and ubiquiti- 5170–5174. nation regulate exit of the Wnt signaling protein LRP6 from the endoplasmic re- 29. Abrami L, Bischofberger M, Kunz B, Groux R, van der Goot FG (2010) Endocytosis of ticulum. Proc Natl Acad Sci USA 105:5384–5389. the anthrax toxin is mediated by clathrin, actin and unconventional adaptors. PLoS 54. Williamson MP (1994) The structure and function of proline-rich regions in proteins. Pathog 6:e1000792. Biochem J 297:249–260.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1121006109 Martchenko et al.