Present addresses: MPI for Biochemistry, Department of Molecular Biology, Am Klopferspitz 18, D-82152 Martinsried, München, Germany;
Transcriptome dysregulation by anthrax lethal toxin plays a key role in induction of human endothelial cell cytotoxicity
Article first published online: 20 JAN 2010
© 2010 Blackwell Publishing Ltd
Volume 12, Issue 7, pages 891–905, July 2010
How to Cite
Rolando, M., Stefani, C., Flatau, G., Auberger, P., Mettouchi, A., Mhlanga, M., Rapp, U., Galmiche, A. and Lemichez, E. (2010), Transcriptome dysregulation by anthrax lethal toxin plays a key role in induction of human endothelial cell cytotoxicity. Cellular Microbiology, 12: 891–905. doi: 10.1111/j.1462-5822.2010.01438.x
- Issue published online: 7 JUN 2010
- Article first published online: 20 JAN 2010
- Received 7 September, 2009; revised 12 January, 2010; accepted 12 January, 2010.
- Top of page
- Experimental procedures
- Supporting Information
We have investigated how Bacillus anthracis lethal toxin (LT) triggers caspase-3 activation and the formation of thick actin cables in human endothelial cells. By DNA array analysis we show that LT has a major impact on the cell transcriptome and we identify key host genes involved in LT cytotoxic effects. Indeed, upregulation of TRAIL and downregulation of XIAP both participate in LT-induced caspase-3 activation. LT induces a downregulation of the immediate early gene and master regulator of transcription egr1. Importantly, its re-expression in LT-intoxicated cells blocks caspase-3 activation. In parallel, we found that the formation of actin cables induced by LT occurs in the absence of direct activation of RhoA/ROCK signalling. We show that knock-down of cortactin and rhophilin-2 under conditions of calponin-1 expression defines the minimal set of genes regulated by LT for actin cable formation. Together our data establish that the modulation of the cell transcriptome by LT plays a key role in triggering human endothelial cell toxicity.
- Top of page
- Experimental procedures
- Supporting Information
As a major virulence factor, Bacillus anthracis synthesizes a tripartite toxin composed of the protective antigen (PA), the oedema factor (EF) and the lethal factor (LF) (Collier and Young, 2003; Moayeri and Leppla, 2004). The toxin encoding genes are borne by the plasmid pXO1. LF and/or EF can associate with heptamers of PA to form, respectively, the lethal toxin (LT for PA+LF) or the oedema toxin (ET for PA+EF). Heptamers of PA transport LF and/or EF into host cell endosomes and promote their injection into the cytosol (Abrami et al., 2006; Sun et al., 2008a). Two receptors for PA, which contain a Willebrand factor A-like domain, have been characterized: the tumour endothelial marker 8 (TEM8) and capillary morphogenesis protein 2 (CMG2) (Bradley et al., 2001; Scobie et al., 2003). TEM8 is highly expressed in epithelial cells lining B. anthracis entry sites and tumour vasculature, whereas CMG2 is expressed in most human cells (Rmali et al., 2004; Bonuccelli et al., 2005). Upon reaching the cytosol, LF uncouples MAP kinase signalling cascades by cleavage of the amino-terminal part of MKKs (Duesbery et al., 1998). EF produces an increase in the intracellular level of cyclic-AMP through its calcium/calmodulin-dependent adenylate cyclase activity (Moayeri and Leppla, 2004). Deletion of EF or LF leads to a 10- and 1000-fold attenuation of bacterial virulence respectively (Pezard et al., 1991).
LT (PA+LF) produces vascular damages and septic shock-like manifestations, pointing to LT as a major determinant of B. anthracis-induced vascular dysfunctions (Moayeri and Leppla, 2004; Gozes et al., 2006; Stearns-Kurosawa et al., 2006; Bolcome et al., 2008). Theprogressive and belated increase in vascular permeability induced by LT is determined by its enzymatic activity (Rolando et al., 2009). Recent studies have unravelled how LT triggers a caspase-dependent cell death of both primary human umbilical vein endothelial cells (HUVECs) and primary human lung or dermal microvascular endothelial cells (HPMVECs and HDMVECs) (Kirby, 2004; Warfel et al., 2005; Gozes et al., 2006). One study also describes that LT caused the formation of actin cables in endothelial cells (Warfel et al., 2005). The re-organization of the actin cytoskeleton induced a rupture of the barrier formed by endothelial cell monolayers (Warfel et al., 2005). How MAPK pathway inhibition by LT produces a re-organization of the human endothelial cell actin cytoskeleton and caspase-3 activation remains to be defined.
Recent studies point to host cell transcription as an important target of LT. Transcriptome responses of immortalized endothelial cells 6 h after infection by either wild-type or pXO1-deficient B. anthracis Sterne were measured (van Sorge et al., 2008). Wild-type B. anthracis primarily reduces host gene expression, a phenomenon likely controlled by pXO1-borne toxin (van Sorge et al., 2008). In parallel, it has been determined that LT can block inflammatory gene expression by interfering with MKK-regulated histone modifications at their promoter (van Sorge et al., 2008; Raymond et al., 2009). Other possible consequences of corrupting the host transcriptome in LT-induced cytotoxic effects remain to be defined.
Contractile actomyosin cables have a major function in regulating endothelium permeability, owing to their ability to destabilize intercellular junctions (Wojciak-Stothard and Ridley, 2003). Activation of the Rho kinase (ROCK) by RhoA drives actin filament (F-actin) cross-linking into contractile actin cables (Jaffe and Hall, 2005). This effect results primarily from a downstream phosphorylation of the Myosin Light Chain (MLC) (Watanabe et al., 1999; Jaffe and Hall, 2005). The RhoA/ROCK pathway, as well as the Rac/Cdc42-activated kinases PAK, also stabilizes actin filaments (F-actin) via a LIM kinase-induced phosphorylation and inactivation of cofilin (Jaffe and Hall, 2005). In parallel to the RhoA/ROCK pathway, several F-actin regulators with capping, severing or cross-linking activities modulate the organization of the actin cytoskeleton (Huang et al., 2004; Le Clainche and Carlier, 2008). Importantly, MAP kinases also control the activity of actin regulators (Huang et al., 2004). Inhibition of the p38-MAPK signalling by LT can block Hsp27 phosphorylation and the induction of actin filament assembly at the leading edge of motile neutrophils, as well as actin-based motility in HeLa cells (During et al., 2005; 2007). Recent studies suggest that the control of gene expression by MAP kinases interferes with actin cytoskeleton regulation. For instance, the serum response factor transcription factor and its MAP kinase- and RhoA-regulated cofactors play a crucial role in shaping the actin cytoskeleton of skeletal and smooth muscle cells (Wang et al., 2004; Posern and Treisman, 2006; Long et al., 2007; Miano et al., 2007). How MAP kinase pathways signal to the endothelial cell actin cytoskeleton for its organization is a key question.
Here we have identified key host factors that are regulated by LT at the mRNA level and are involved in its cytotoxicity on human endothelial cells. We propose that modulation of the host cell transcriptome plays a key role in LT-induced human endothelial actin cytoskeleton re-organization and caspase-3 activation.
- Top of page
- Experimental procedures
- Supporting Information
LT triggers progressive cytotoxic effects on human endothelial cells
In order to determine the molecular basis of actin cytoskeleton remodelling and induction of human endothelial cell death by LT, we established the kinetics of appearance of these cytotoxic effects (Fig. 1 and Videos S1–S3). Control monolayers organized in stable cobblestone architecture (Fig. 1A and Video S4). Between six to eight hours after addition of the toxin, we observed that some cells started to elongate (Video S1). This phenomenon progressively affected individual cells in the monolayer to reach a maximum effect at 24 h of intoxication (Video S2). Also, following 12 h of intoxication, individual cells progressively underwent blebbing and detachment leading to the formation of large intercellular gaps in the endothelial cell monolayer (Fig. 1A and Videos S2 and S3). This correlated with a progressive increase in the activity of caspase-3, starting 8 h after toxin addition (Fig. 1B). We also measured an increase in caspase-8 and -9 activities (Fig. S1A). Moreover, we measured, after 24 h of intoxication, a threefold increase in early apoptotic cells (Fig. S1B). LT-induced elongation of cells correlated with a re-organization of the actin cytoskeleton, which forms thick actin cables parallel to each other starting around 6 h and best visualized at 24 h of intoxication (Fig. 1C). In addition, we observed a loss of the cortical F-actin network and a redistribution of the major adherens junction protein VE-cadherin (Fig. 1C). This redistribution of VE-cadherin is consistent with the formation of contractile actin cables and cell elongation resulting in cell stretching and opening of intercellular gaps (Wojciak-Stothard and Ridley, 2003). Monolayer treatment with the catalytically inactive mutant LTm (LFE687A + PA), with PA alone or LF alone, neither produced MEK2 cleavage nor cytotoxicity (Fig. S1C and D, and data not shown). LT purified from B. anthracis produced similar cytotoxic effects on the actin cytoskeleton and caspase-3 activity (Fig. S1E and F). Our data established that LT induced progressive and belated cytotoxic effects, as opposed to the early inactivation of the MAP kinase signalling (Fig. S2A). Treatment of endothelial cell monolayers with a mix of MKK/MAP kinase inhibitors (MKI) was found to phenocopy the cytotoxic effects of LT, although with a lower efficiency likely due to the lesser inhibition of MAPK phosphorylation (Fig. 1C and Fig. S1G).
Importantly, these new observations establish that LT and MAPK pathway inhibition induce a progressive and belated activation of caspase-3 and re-organization of the actin cytoskeleton.
LT reprograms gene expression of the TRAIL signalling pathway
We rationalized that a corruption of gene expression by LT might be involved in this delayed induction of cytotoxicity. Transcriptomes of endothelial cell monolayers intoxicated for 4 and 8 h by LT were compared with those of control monolayers (GEO Series Accession No. GSE12131). Most notably, at 8 h of intoxication, LT dampened levels of 2341 transcripts. In agreement with a recent study (van Sorge et al., 2008), the immediate early gene egr1 (early growth response-1) appeared as the most downregulated transcript (not shown and Fig. 2A). In contrast, a cluster of genes showed a sustained increase of expression at 4 and 8 h of intoxication (Fig. S2B). This cluster of genes comprises the TNF-related apoptosis-inducing ligand (TRAIL or APO2L). TRAIL is a type-II transmembrane protein, eventually released as a soluble cytokine, which can associate with death receptors of the TNF receptor superfamily for activation (Guicciardi and Gores, 2009). Interestingly, several studies showed that MAP kinase inhibition sensitizes cells to TRAIL-induced apoptosis and conversely that ERK activation inhibits TRAIL-induced cell death (Guicciardi and Gores, 2009). Other factors involved in the TRAIL pathway were identified: we noticed a decrease of xiap (X-linked Inhibitor of Apoptosis Protein) and an increase of xaf1 (XIAP associated factor-1) mRNA. Kinetics of modulation of egr1, trail, xiap, xaf1, as well as other LT regulated genes were established by qRT-PCR (Fig. 2A). Inhibition of MAP kinase cascades with MKI recapitulated the modulation of expression of these genes (Fig. 2B). Also LT purified from B. anthracis had a similar effect on egr1, trail, xiap and xaf1 mRNA expression (Fig. S1H). Finally, we found that addition of actinomycin (actD) blocked the induction by LT of the genes tested (data not shown).
Thus, LT triggers a sustained modulation of expression of a large number of genes. This comprises the immediate early gene egr1, as well as genes of the TRAIL signalling pathway.
LT-induced caspase-3 activation requires trail, xiap and egr-1 modulation of expression
Kinetics of TRAIL induction and down-modulation of XIAP were established at the protein level. This shows a threefold upregulation of TRAIL and a 40% downregulation of XIAP at 8 h of intoxication, concomitant with the increase in caspase-3 activity (Fig. 1B, Fig. 3A and B). The induction of TRAIL and the decrease of XIAP further progress over the first 24 h of intoxication. We next tested whether TRAIL inhibition and XIAP re-expression reverted LT-induced caspase-3 activation. Addition of TRAIL inhibitory antibodies, as well as expression of XIAP reduced the activation of caspase-3 by the toxin (Fig. 3C and D). We verified that in these conditions LT still induced MEK2 cleavage (data not shown).
We next analysed whether deregulation of an upstream transcription factor might account for LT-induced cytotoxic effects. We focused our attention on egr1 given that it is an immediate early gene acting as a master regulator of transcription (Fu et al., 2003) and that it is down-modulated by B. anthracis in endothelial cells. We observed that, consistent with the rapid decrease of egr1 mRNA induced by LT (Fig. 2A), its protein level dropped rapidly to 50% after cell intoxication (Fig. 3E). Interestingly, we measured that re-expression of Egr1 in toxin-treated cells significantly restored xiap expression (data not shown). We thus assessed whether Egr1 re-expression interferes with the activation of caspase-3 by LT. Indeed, we found that re-expression of Egr1 in LT-treated cells reduced the activation of caspase-3 (Fig. 3F).
Collectively, our data show that the modulation of TRAIL and XIAP expression downstream of Egr-1 participates to the activation of caspase-3 by LT.
LT induces an unconventional formation of actin cables
Activation of the Rho kinase ROCK by membrane-associated RhoA controls the formation of actin cables. We thus investigated whether LT might have an impact on this pathway. Surprisingly, control and LT-intoxicated monolayers showed an equivalent activity and membrane association of RhoA (Fig. 4A and B). Similar results were obtained for Rac and Cdc42 (Fig. S2C–E). We further investigated the activity of ROCK by measuring the phosphorylation of downstream targets. Control and LT-intoxicated monolayers showed an equivalent phosphorylation of both myosin light chain and cofilin (Fig. 4C and D). Complementary to these findings, we observed that direct inhibition of RhoA by ADP-ribosylation with the exotoxin EDIN of Staphylococcus aureus disrupted the actin cables produced by LT (Fig. 4E) (Wilde et al., 2003). Actin cables were also disrupted by inhibition of the RhoA effector kinase ROCK using Y27632 (Fig. 4E) (Ishizaki et al., 2000). EDIN and Y27632 did not block MEK2 cleavage by LT (Fig. S2F). Collectively, these data show that the RhoA/ROCK pathway controls the stability of these actin cables but is likely not directly modulated by LT. Our results establish that LT triggers an unconventional formation of actin cables in human endothelial cells.
Strikingly, we also made the observation that the global inhibition of transcription, using actD or alpha-amanitin, triggered formation of thin actin filaments in endothelial cells (Fig. 4F). Moreover, inhibition of transcription prevented the bundling of actin filaments into thick actin cables in the presence of LT (Fig. 4F and not shown). ActD had no effect on the activity of ERK1/2 and did not block MEK2 cleavage by LT (data not shown). These results led us to investigate whether LT-mediated re-organization of the actin cytoskeleton might be mediated at the transcriptional level.
Modulation of gene expression by LT participates to actin cable formation
Following our observation that LT induces an unconventional re-organization of the actin cytoskeleton, we searched for LT-modulated genes encoding actin regulators using gene ontology of molecular function (Table 1). We found 38 repressed genes and three induced genes. We next established the kinetics of gene expression of the two most upregulated and downregulated actin regulators and verified their MAP kinase-dependent modulation of expression (Fig. 5A). LT treatment significantly decreases the expression of cttn (cortactin) and rhpn2 (rhophilin-2) and increases cnn1 (calponin-1) and dmn (desmuslin) mRNA at 24 h of intoxication. Addition of actD blocked LT-induced expression of cnn1 (calponin-1) and dmn (desmuslin) mRNA (data not shown). Cortactin could be detected by immunoblotting, showing its downregulation in LT intoxicated cells (Fig. 5B).
|Gene symbol||RefSeq||LT 4 h/0 h||LT 8 h/4 h||Description|
|SPTBN1||NM_003128||0||1.7||Spectrin, beta, non-erythrocytic 1|
|RHPN2||NM_033103||−1.4||−5||Rhophilin, Rho GTPase binding protein 2|
|PALLD||NM_016081||−2.35||−2.6||Palladin, Cytoskeletal associated protein|
|CMYA3||NM_152381||−2.8||−2.2||Cardiomyopathy associated 3|
|FDG6||NM_018351||−0.35||−2.1||FYVE, RhoGEF and PH domain containing 6|
|CDC42EP3||NM_006449||−2.3||−1.6||CDC42 effector protein (Rho GTPase binding) 3|
|DCAMKL1||NM_004734||−0.1||−1.35||Doublecortin and CaM kinase-like 1|
|MYLK||NM_005965||−0.45||−1.25||Myosin light polypeptide kinase|
|ARHGAP30||NM_001025598||−1.4||−1.22||Rho GTPase activating protein 30|
|ARPC5L||NM_030978||−1.4||−0.9||Actin-related protein 2/3 complex. subunit 5-like|
|MRCL3||NM_006471||−1.6||−0.75||Myosin regulatory light chain MRCL3|
|LIMK2||NM_001031801||−1.35||−0.65||LIM domain kinase 2|
|ARHGAP29||NM_004815||−1.45||−0.65||Rho GTPase activating protein 29|
|FILIP1||NM_015687||−1.35||−0.6||Filamin A interacting protein 1|
|ARHGAP29||NM_004815||−1.7||−0.55||Rho GTPase activating protein 29|
|FRMD4A||NM_018027||−1.3||−0.55||FERM domain containing 4A|
|CDC42EP2||NM_006779||−1.45||−0.5||CDC42 effector protein (Rho GTPase binding) 2|
|MYO5A||NM_000259||−1.4||−0.45||Myosin VA (heavy polypeptide 12. myoxin)|
|ARHGAP22||NM_021226||−1.2||−0.4||Rho GTPase activating protein 22|
|ARHGAP18||NM_033515||−1||−0.4||Rho GTPase activating protein 18|
|ARHGAP5||NM_001030055||−1.25||−0.55||Rho GTPase activating protein 5|
|ARHGAP18||NM_033515||−1.45||−0.3||Rho GTPase activating protein 18|
|ARHGEF2||NM_004723||−1.2||−0.25||Rho/rac guanine nucleotide exchange factor (GEF) 2|
|LMCD1||NM_014583||−1.55||−0.15||LIM and cysteine-rich domains 1|
|FNBP4||NM_015308||−1.4||−0.15||Formin binding protein 4|
|STRN||NM_003162||−1.15||−0.15||Striatin, Calmodulin binding protein|
|PHACTR2||NM_014721||−1.05||−0.1||Phosphatase and actin regulator 2|
|MYO IXB||NM_004145||−1||−0.1||Myosin IXB|
|FNBP1L||NM_001024948||−1.6||−0.05||Formin binding protein 1-like|
|ROCK2||NM_004850||−1.65||0||Rho-associated coiled-coil containing protein kinase 2|
|SSH2||NM_033389||−1.25||0||Slingshot homologue 2 (Drosophila)|
We next analysed the functional implication of actin regulators identified. Cortactin integrates signals at the membrane to drive cortical F-actin polymerization (Selbach and Backert, 2005). Rhophilin-2 triggers a loss of F-actin from stress fibres by a mechanism, which remains to be established, but that is independent of Rho inactivation (Peck et al., 2002). Calponin-1 is a calcium binding protein, which inhibits myosin ATPase activity to stabilize actin cables (Danninger and Gimona, 2000). Induction of calponin-1 might thus account for the bundling of F-actin into thick actin cables. We performed a functional analysis to determine whether these three proteins contribute to the formation of actin cables induced by LT. This was assessed by measuring the percentage of transfected cells containing thick actin cables (Fig. 5C). Consistent with our hypothesis, expression of cortactin in LT-intoxicated cells produced a 20% decrease in the number of cells containing thick actin cables and restored the network of cortical actin in LT-intoxicated cells (Fig. 5C and D). Expression of rhophilin-2 reduced the number of LT-treated cells containing thick actin cables by 50% (Fig. 5C and D). Finally, we observed that calponin-1 knock-down by shRNA reduced the formation of thick actin cables in LT-intoxicated cells (Fig. 5C and D). We next assessed whether the concomitant modulation of cellular levels of these three proteins might phenocopy LT effects. At first, we observed that individual expression of rhophilin-2 shRNA, cortactin shRNA or calponin-1 had only a minor effect on actin cable formation (Fig. 5C). In contrast, the combined expression of rhophilin-2 shRNA and cortactin shRNA, together with GFP-calponin-1, produced thick actin cables in 50% of GFP positive cells (Fig. 5C and E).
Our functional analysis thus defines three key genes, whose expression is modulated by LT to contribute to the formation of thick actin cables by the toxin.
LT-induced actin cable formation is reduced upon egr1 re-expression
We rationalized that egr1 repression could also contribute to LT-induced actin cable formation. We measured that re-expression of Egr1 in toxin-treated cells significantly interferes with the modulation of expression of cortactin and rhophilin-2 (Fig. 6A). Re-expression of Egr1 in LT-treated cells also had a major inhibitory effect on the formation of actin cables and restored the formation of cortical actin (Fig. 6B). Indeed, under these conditions we measured 83.5% cells with actin cables in mock-transfected cells and only 31.2% in Egr1-transfected cells (*P < 0.05 versus control cells) (Fig. 6C).
Taken collectively, our data identify a group of host factors transcriptionally regulated by Egr1, which play key roles in LT-induced actin cable formation.
- Top of page
- Experimental procedures
- Supporting Information
It was of importance to determine the molecular basis of actin cytoskeleton remodelling and induction of caspase-3 activity by LT in human endothelial cells. Here, we identify key factors, whose modulation at the mRNA level is implicated in LT-induced cytotoxicity of human endothelial cells. We show that the inhibitory effect of LT on MAPK signalling is sufficient to account for these cytotoxic effects and that both actin re-organization and caspase-3 activation require host transcript modulations. Indeed, down-modulation of rhophilin-2 and cortactin, together with an upregulation of calponin-1, is sufficient to produce actin cables. We show that modulation of TRAIL and XIAP expression participates in caspase-3 activation. At last, we show that re-expression of the immediate early gene and master regulator of transcription Egr1 in LT-intoxicated cells blocks both actin re-organization and caspase-3 activation. Collectively, this establishes that modulation of host gene expression is crucial for the cytotoxic effects of LT in human endothelial cells.
We report that LT has a major impact on the cell transcriptome. Our findings are consistent with the previously reported observation that factors borne by the plasmid pXO1 are responsible for the modulation of 90% of host genes altered by pathogenic B. anthracis during endothelial cell infection (van Sorge et al., 2008). Interestingly, wild-type B. anthracis reduces the steady-state expression of most affected gene transcripts (van Sorge et al., 2008). Here we extend these findings by showing that LT alone reduces the steady-state expression of 84% of affected gene transcripts. Furthermore, we show that LT down-modulates the expression of the immediate-early gene egr1, a master regulator of transcription, as does wild-type B. anthracis in endothelial cells (van Sorge et al., 2008). Importantly, expression of egr1 is deregulated and participates in a large number of cardiovascular diseases (Fu et al., 2003; Khachigian, 2006). Down-modulation of egr-1 alone by RNAi knock down is not sufficient to phenocopy LT-effects on gene expression and endothelial cell cytotoxicity (not shown). This is likely due to redundancy between Egr1 family members and/or due to the involvement of other transcriptional regulators controlled by MAP kinases (Fu et al., 2003; van Sorge et al., 2008). Nevertheless, we demonstrate that re-expression of Egr-1 in LT-intoxicated cells has a protective effect on human endothelial cells as it reduces the toxin-induced cytotoxic effects on actin re-organization and caspase-3 activation. We also show that Egr1 controls the expression of several genes involved in caspase-3 activation and actin cytoskeleton re-organization.
We have identified host factors whose expression is corrupted by LT and plays key roles in LT-induced caspase-3 activation in endothelial cells. Indeed, we define that modulation of the expression of trail and the downstream factor xiap regulate caspase-3 activation. In support of our observation that LT controls apoptosis at the transcriptional level, a recent study performed on a B. anthracis infected monocytic cell line reports that pXO1 alters transcription of a number of genes encoding regulators of the extrinsic pathway of apoptosis (Bradburne et al., 2008). This delays early apoptosis of infected human macrophages. One interesting hypothesis is that LT may have cell type-specific effects as a consequence of distinct profiles of genes being regulated by MAP kinases in these cell types (Posern and Treisman, 2006). XIAP is a potent regulator of TRAIL-induced apoptosis (Rigaud et al., 2006). We found that Egr1 expression in LT-treated cells produced an upregulation of XIAP (not shown) and blocks the activation of caspase-3. Other host factors regulated by LT, such as XIAP-associated Factor-1 (XAF1), might also contribute to caspase-3 activation by the toxin. The binding of XAF1 to XIAP inhibits its anti-caspase activity (Liston et al., 2001). In conclusion, the differential regulation of signalling molecules of the TRAIL pathway, such as XIAP, and potentially XAF1, illustrates the essential role of transcriptional remodelling in LT-induced cytotoxicity on endothelial cells.
We have also identified several actin cytoskeleton regulators whose gene expression is modulated by LT in human endothelial cells (Table 1). Importantly, we unravel that LT produces an unconventional formation of actin cables requiring host gene transcription in absence of direct activation of the RhoA/ROCK pathway. Consistent with our findings, it was recently reported that LT induced formation of actin cables in absence of RhoA activation and cofilin phosphorylation in human bronchial epithelial cells (Lehmann et al., 2009). Here we provide evidence that gene expression is required for the induction of actin cables by LT. We have analysed the function of actin regulatory factors acting in parallel to RhoA/ROCK. This does not exclude that LT might also directly corrupt signalling pathways controlling actin regulators (During et al., 2005; 2007). Nevertheless, we provide functional evidence that the downregulation of cortactin and rhophilin-2 in combination with upregulation of calponin-1 is sufficient to produce actin cables. Our findings suggest that transcriptional mechanisms are essential in the remodelling of the endothelial actin cytoskeleton, and suggest that Egr1 is an important actor in this process.
In conclusion, we have identified key host factors whose gene expression is controlled by MAP kinase cascades and that are exploited by B. anthracis LT to trigger human endothelial cell cytotoxicity.
- Top of page
- Experimental procedures
- Supporting Information
Cell culture and reagents
The HUVECs (PromoCell, Heidelberg, Germany) were cultured and transfected, as previously described (Doye et al., 2006). Cells were used between passages two and five. Kindly provided protein expression vectors are pCI-HA-XIAP from Dr S. Plenchette and Dr R.G. Korneluk (Children's Hospital of Eastern Ontario Research Institute, Ottawa, Canada); pcDNA3-cortactin from Dr G. Tran Van Nhieu (Institut Pasteur, Paris, France); pEGFP-calponin1 from Dr M. Gimona (Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy); pRSV2.1-Egr1 from Dr S. Tartare-Deckert (University of Nice, France). Rhophilin-2 cDNA (Dr I. Pirson, Institute of Interdisciplinary Research, Brussels, Belgium) was cloned in pcDNA-DESTGFP vector (Invitrogen). To generate shRNA expression plasmids (shCTTN for cortactin, shRHPN2 for rhophilin-2 and shCNN1 for calponin-1), sequences corresponding to cortactin (5′-GACTGGTTTTGGAGGCAAA), to rhophilin-2 (5′-CGCCCCCGTTCAGGTTCAC) to calponin-1 (5′-ACTACTACAATTCCGCCTA) and to Egr1 (5′-GAGGCATACCAAAATCCAT) were cloned in the pSUPER RNAi system (Invitrogen), according to the manufacturer's instructions. We measured by qRT-PCR that 48 h of expression of shCTTN, shRHPN2, reduced mRNA level of cortactin and rhophilin-2 by 60% and 65% respectively (data not shown). We measured by immunoblotting that shEgr1 reduced by 40% Egr1 protein level at 48 h post transfection. The activity of shCNN1 was verified by co-transfection with pEGFP-calponin1 (data not shown).
Monoclonal antibodies used are mouse anti-Myc ([clone 9E10], Roche), anti-beta-actin ([clone AC-74], Sigma), anti-cortactin ([clone 4F11], Upstate biotechnology), anti-HA ([clone 16-B12], Covance), anti-RhoGDI ([clone A-20], Santa Cruz) and anti-transferrin receptor-1 ([clone H68.4], Zymed). From BD Biosciences: anti-RhoA [clone 26C4], anti-Rac1 [clone 102], anti-CDC42 [clone 44], anti-XIAP [clone 28] and anti-cadherin5 [clone 75]. Polyclonal antibodies used are rabbit: anti-acidic-calponin (Covance), anti-phospho-S3-cofilin courtesy of Dr Bambourg JR (Colorado State University, USA), anti-cofilin (Cytoskeleton). Rabbit anti-MEK2-N20, anti-MLC, anti-ROCK-I, anti-ROCK-II and anti-Egr1 antibodies were purchased from Santa Cruz Technologies. Rabbit immunopurified anti-TRAIL and anti-GST antibodies were purchased from Abcam and Chemicon respectively. Rabbit anti-p-ERK1/2 (pERK), anti-pT18-pS19-MLC and anti-ERK1/2 were purchased from Cell Signalling. For immunofluorescence we used Texas red-conjugated anti-mouse antibodies (Vector, Biovalley), Texas red-conjugated anti-rabbit antibodies (Jackson Immunoresearch Laboratories) or FITC-conjugated anti-rabbit antibodies (Biosys). For immunoblotting we used goat anti-mouse or anti-rabbit horseradish peroxidase-conjugated secondary antibodies (DAKO) followed by chemiluminescence using Immobilon Western (Millipore). Chemiluminescence signals were recorded on a FUJIFILM LAS-3000 and data quantified with the software MultiGauge V3.0. We used Human TRAIL/TNFSF10 Immunoassay (Quantikine, R&D Systems).
Y27632 (ROCK inhibitor, used at 20 µM), U0126 (MEK1, 2 inhibitor, used at 10 µM), SB203580 (P38α, β inhibitor, used at 20 µM) and SP600125 (JNK1, 2, 3 inhibitor, used at 10 µM) were purchased from Calbiochem. Z-Val-Ala-Asp(Ome)-FMK (zVAD) and actinomycin D were purchased from MP Biomedicals. Other biochemical reagents were purchased from Sigma Aldrich.
Recombinant toxin production
Protective antigen DNA was PCR-amplified with primers 5′-GGATCCCGAATTAAACAGGAGAACCGG and 5′-GAGCTCTGTCCTATCTCATAGCCTTTTTTAGA and cloned BamHI-SacI into pET22 (Novagen). LF gene was PCR-amplified with primers 5′-GGATCCATGGCGGGCGGTCATGGTGAT and 5′-GAGCTCTTATGAGTTAATAATGAACTTAAT and cloned BamHI-SacI into the pQE30 (Qiagen). Amplification of PA and LF was performed on genomic DNA of B. anthracis strain Sterne (a kind gift of Patrice Boquet, Nice, France). Catalytically inactive mutant LFE687A was obtained using the QuickChange Site Directed Mutagenesis kit (Stratagene). Plasmids were transformed into Escherichia coli BL21-DE3. Clarified supernatants containing LF were purified on nickel column (Amersham Bioscience). Fractions of LF, wild-type and mutant were dialysed against buffer A [25 mM Tris-HCl (pH 7.4), 50 mM NaCl] and purified on MonoQ (Amersham) using a gradient of NaCl 50–300 mM in buffer A. PA was prepared from the periplasm using standard techniques and purified on Q sepharose fast flow column (Amersham Biosciences) with a gradient of 50–300 mM NaCl in buffer B (Hepes 20 mM, pH 7). Fractions of PA were sieved on a BioSep-Sec-S column (Phenomenex) and purified on ResourceQ (Amersham Biosciences) with a gradient of 50–300 mM NaCl in buffer B. CNF1 and EDIN were purified as described (Doye et al., 2002; Boyer et al., 2006). All proteins were applied on EndoTrap Red column and absence of endotoxin was assessed using the Limulus Amebocyte Lysate QCl-1000 (Cambrex).
Transcriptome analysis and qRT-PCR
For GeneChip hybridization and statistical analysis RNA were prepared from HUVEC monolayers either control (two samples) or intoxicated by LT (PA+LF, 3 + 1 µg ml−1) for 4 and 8 h. Transcripts expression was analysed by Aros Applied Biotechnology ApS (http://www.arosab.com), using the Affymetrix Human GeneChip U133A Plus 2.0, as recommended by the manufacturer. The cRNA synthesis, hybridization and labelling, as well as statistical analysis were carried out as previously described (Munro et al., 2004). Gene expression in control condition showing variations −1 ≤ Log2 ≥ 1 were removed prior to comparison with expression at 4 and 8 h of intoxication. Microarray data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/projects/geo) and are accessible through GEO Series Accession No. GSE12131.
RNA were isolated using the RNeasy Mini kit (QIAGEN). Reverse transcriptase reactions were performed on 1 µg RNA samples using the SuperScript II reverse transcriptase (Invitrogen) and oligo(dT)15 primers (Invitrogen). Primers were defined using the primer express software and their specificity verified (Fig. S3 and data not shown). Quantitative PCR were performed in 25 µl containing 5 µl of cDNA, primers (0.4 µM each) and 12.5 µl of SYBR Green Master-mix (Eurogentec). Reactions were run on an ABI-Prism 7000 instrument (Applied Biosystems). Relative quantification of gene expression was performed using the comparative Ct method. Results were normalized using the human 36B4-housekeeping gene.
Immunofluorescence and time-lapse microscopy
Immunofluorecence studies were performed on cells fixed in 4% paraformaldehyde (Sigma). Actin cytoskeleton was labelled using 1 µg ml−1 FITC or TRITC-conjugated phalloidin (Sigma). Immunosignals were analysed with a LSM510-Meta confocal microscope (Carl Zeiss) at 63× magnification lens. Each picture represents the projection of six serial confocal sections. Cells were analysed by video microscopy on an Axiovert 200 microscope equipped with shutter-controlled illumination (Carl Zeiss) and a cooled, digital CCD camera (Roper Scientific). Images were processed using MetaMorph 2.0 image (Molecular Devices) and QuickTime pro 7 (Apple) softwares.
Caspase activity and apoptosis measurements
Caspase activity measurements were performed at various time points at 37°C using 35 µg of protein together with 0.5 mM of Ac-DEVD-AMC (caspase-3 substrate), Ac-IETD-AMC (casp-8) or Ac-LEHD-AMC (casp-9) (Alexis Biomedicals) in the presence or absence of 10 µM of Ac-DEVD-CHO (casp-3), Ac-IETD-CHO (casp-8) or Ac-LEHD-CHO (casp-9) (Alexis Biomedicals). Measures were performed using an Ascent Fluoroskan (Thermo Labsystem). For early apoptotic cell determination HUVEC monolayers were simultaneously stained with annexin-V-Alexa Fluor and propidium iodide (Roche). The cell suspensions were adjusted to 106 cells ml−1 before acquisition and analysed on a fluorescence-activated cell sorting (FACS) flow cytometer (FACSCalibur; BD Biosciences). Data acquisition was performed with the CellQuest software (BD Biosciences).
The non-parametric anova analogue test of Kruskal–Wallis was used for qRT-PCR data statistical analysis. anova with Bonferroni post hoc was used for other statistical analyses. Analyses were performed with Prism V5.0b software.
- Top of page
- Experimental procedures
- Supporting Information
We thank M. Moayeri and Stephen H. Leppla for providing us with LT toxin purified from B. anthracis, Dr J.R. Bamburg for phospho-cofilin antibodies, Dr D. Paulin and Z. Xue for alpha-synemin antibodies. We are grateful to S. Bourdoulous, P. Cossart, A. Eychène, M. Maddugoda, C. Pouponnot and T. Thibaut for fruitful discussions. We are grateful to A. Doye, F. Prodon, P. Colosetti, F. Massiera, V. Giordanengo, S. Tartare-Deckert and V. Lavallard for technical help. We thank the Conseil Régional PACAC and the Conseil Général des Alpes-Maritimes for their financial support at the microscopy facility platform MiCorBio. U.R.R., M.R. and A.G. were supported by the international DFG research training group 1141 Würzburg/Nice (GCWN) and the Franco-German University (ED-31-04). Our laboratory is supported by an institutional funding from INSERM, a grant (ARC 3800) from the Association pour la Recherche sur le Cancer, a grant and fellowship to M.R. from the Agence Nationale de la Recherche (ANR RPV07055ASA).
- Top of page
- Experimental procedures
- Supporting Information
- 2006) Receptor palmitoylation and ubiquitination regulate anthrax toxin endocytosis. J Cell Biol 172: 309–320. , , and (
- 2008) Anthrax lethal toxin induces cell death-independent permeability in zebrafish vasculature. Proc Natl Acad Sci USA 105: 2439–2444. , , , , , and (
- 2005) ATR/TEM8 is highly expressed in epithelial cells lining Bacillus anthracis' three sites of entry: implications for the pathogenesis of anthrax infection. Am J Physiol Cell Physiol 288: C1402–C1410. , , , , , , et al. (
- 2006) Induction of transient macroapertures in endothelial cells through RhoA inhibition by Staphylococcus aureus factors. J Cell Biol 173: 809–819. , , , , , , et al. (
- 2008) Transcriptional and apoptotic responses of THP-1 cells to challenge with toxigenic, and non-toxigenic Bacillus anthracis. BMC Immunol 9: 67. , , , , , , et al. (
- 2001) Identification of the cellular receptor for anthrax toxin. Nature 414: 225–229. , , , , and (
- 2003) Anthrax toxin. Annu Rev Cell Dev Biol 19: 45–70. , and (
- 2000) Live dynamics of GFP-calponin: isoform-specific modulation of the actin cytoskeleton and autoregulation by C-terminal sequences. J Cell Sci 113 (Pt 21) 3725–3736. , and (
- 2002) CNF1 exploits the ubiquitin-proteasome machinery to restrict Rho GTPase activation for bacterial host cell invasion. Cell 111: 553–564. , , , , , , et al. (
- 2006) Ubiquitin-mediated proteasomal degradation of Rho proteins by the CNF1 toxin. Methods Enzymol 406: 447–456. , , , and (
- 1998) Proteolytic inactivation of MAP-kinase-kinase by anthrax lethal factor. Science 280: 734–737. , , , , , , et al. (
- 2005) Anthrax lethal toxin paralyzes neutrophil actin-based motility. J Infect Dis 192: 837–845. , , , , , , and (
- 2007) Anthrax lethal toxin paralyzes actin-based motility by blocking Hsp27 phosphorylation. EMBO J 26: 2240–2250. , , , , , , and (
- 2003) Egr-1 target genes in human endothelial cells identified by microarray analysis. Gene 315: 33–41. , , , , , , et al. (
- 2006) Anthrax lethal toxin induces ketotifen-sensitive intradermal vascular leakage in certain inbred mice. Infect Immun 74: 1266–1272. , , , and (
- 2009) Life and death by death receptors. FASEB J 23: 1625–1637. , and (
- 2004) MAP kinases and cell migration. J Cell Sci 117: 4619–4628. , , and (
- 2007) Identification of genes selectively regulated by IFNs in endothelial cells. J Immunol 178: 1122–1135. , , , , , , et al. (
- 2000) Pharmacological properties of Y-27632, a specific inhibitor of rho-associated kinases. Mol Pharmacol 57: 976–983. , , , , , , and (
- 2005) Rho GTPases: biochemistry and biology. Annu Rev Cell Dev Biol 21: 247–269. , and (
- 2006) Early growth response-1 in cardiovascular pathobiology. Circ Res 98: 186–191. (
- 2004) Anthrax lethal toxin induces human endothelial cell apoptosis. Infect Immun 72: 430–439. (
- 2008) Regulation of actin assembly associated with protrusion and adhesion in cell migration. Physiol Rev 88: 489–513. , and (
- 2009) Lung epithelial injury by B. anthracis lethal toxin is caused by MKK-dependent loss of cytoskeletal integrity. PLoS ONE 4: e4755. , , , , and (
- 2001) Identification of XAF1 as an antagonist of XIAP anti-caspase activity. Nat Cell Biol 3: 128–133. , , , , , , et al. (
- 2007) Myocardin is a bifunctional switch for smooth versus skeletal muscle differentiation. Proc Natl Acad Sci USA 104: 16570–16575. , , , , and (
- 2007) Serum response factor: master regulator of the actin cytoskeleton and contractile apparatus. Am J Physiol Cell Physiol 292: C70–C81. , , and (
- 2004) The roles of anthrax toxin in pathogenesis. Curr Opin Microbiol 7: 19–24. , and (
- 2004) Activation and proteasomal degradation of rho GTPases by cytotoxic necrotizing factor-1 elicit a controlled inflammatory response. J Biol Chem 279: 35849–35857. , , , , , , et al. (
- 2002) The RhoA-binding protein, rhophilin-2, regulates actin cytoskeleton organization. J Biol Chem 277: 43924–43932. , , , , and (
- 1991) Contribution of individual toxin components to virulence of Bacillus anthracis. Infect Immun 59: 3472–3477. , , and (
- 2006) Actin' together: serum response factor, its cofactors and the link to signal transduction. Trends Cell Biol 16: 588–596. , and (
- 2009) Anthrax lethal toxin impairs IL-8 expression in epithelial cells through inhibition of histone H3 modification. PLoS Pathog 5: e1000359. , , , , , , et al. (
- 2006) XIAP deficiency in humans causes an X-linked lymphoproliferative syndrome. Nature 444: 110–114. , , , , , , et al. (
- 2004) Upregulation of tumour endothelial marker-8 by interleukin-1beta and its impact in IL-1beta induced angiogenesis. Int J Mol Med 14: 75–80. , , , , and (
- 2009) Injection of Staphylococcus aureus EDIN by the Bacillus anthracis protective antigen machinery induces vascular permeability. Infect Immun 77: 3596–3601. , , , , , and (
- 2003) Human capillary morphogenesis protein 2 functions as an anthrax toxin receptor. Proc Natl Acad Sci USA 100: 5170–5174. , , , and (
- 2005) Cortactin: an Achilles' heel of the actin cytoskeleton targeted by pathogens. Trends Microbiol 13: 181–189. , and (
- 2006) Sepsis and pathophysiology of anthrax in a nonhuman primate model. Am J Pathol 169: 433–444. , , , , and (
- 2008) Anthrax toxins inhibit neutrophil signaling pathways in brain endothelium and contribute to the pathogenesis of meningitis. PLoS ONE 3: e2964. , , , , , , and (
- 2008a) Phenylalanine-427 of anthrax protective antigen functions in both pore formation and protein translocation. Proc Natl Acad Sci USA 105: 4346–4351. , , , and (
- 2004) Myocardin and ternary complex factors compete for SRF to control smooth muscle gene expression. Nature 428: 185–189. , , , , , and (
- 2005) Anthrax lethal toxin induces endothelial barrier dysfunction. Am J Pathol 166: 1871–1881. , , and (
- 1999) Cooperation between mDia1 and ROCK in Rho-induced actin reorganization. Nat Cell Biol 1: 136–143. , , , , and (
- 2003) Rho-specific Bacillus cereus ADP-ribosyltransferase C3cer cloning and characterization. Biochem 42: 9694–9702. , , and (
- 2003) Rho GTPases and the regulation of endothelial permeability. Vascul Pharmacol 39: 187–199. , and (
- Top of page
- Experimental procedures
- Supporting Information
Fig. S1. Control of the cytotoxicity induced by LT. A. Measure of LT-induced activation of caspase-3, 8 and 9 in HUVEC monolayers. zVAD 20 µM was added to evaluate the specificity of caspase activation by LT. Activities are expressed as arbitrary units (A.U.) per mg of proteins (mean values of n = 3 experiments ± SEM). B. HUVECs were either left untreated or intoxicated with LT (PA+LF, 3 + 1 µg ml−1) for 24 h. Graph shows the distribution of annexin V-FITC-positive cells (x-axis) and propidium iodide positive cells (y-axis). Values correspond to the percentages of distribution of each cell population. C and D. Control immunoblots of LT activity on MEK2 cleavage. Western blot anti-actin shows equal protein loading. (C) HUVEC monolayers intoxication for 24 h with PA 3 µg ml−1 alone or together with LF at various concentrations (µg ml−1). (D) HUVEC monolayers intoxication with LT (PA+LF, 3 + 1 µg ml−1) or a catalytically inactive mutant LTm (PA+LFE687A 3 + 1 µg ml−1). E. HUVEC monolayers were intoxicated 24 h with recombinant LT purified from E. coli (PA+LF, 3 + 1 µg ml−1) or with B. anthracis purified native toxin (LT Ba: PA+LF, 3 + 1 µg ml−1). F-actin was labelled with FITC-phalloidin and VE-cadherin with anti-cadherin5. Scale bar, 10 µm. F. Caspase-3 activity in HUVEC monolayers intoxicated 24 h with recombinant LT purified from E. coli (PA+LF, 3 + 1 µg ml−1) or with B. anthracis purified native toxin (LT Ba: PA+LF, 3 + 1 µg ml−1). zVAD 20 µM was added to evaluate the specificity of caspase activation by LT. Activities are expressed as arbitrary units (A.U.) per mg of proteins (one representative experiment). G. Caspase-3 activity in HUVEC monolayers treated 24 h with a mix of MKK/MKI (10 µM U0126, 20 µM SB203580 and SP600125 10 µM) or with LT (PA+LF, 3 + 1 µg ml−1). Activities are expressed as arbitrary units (A.U.) per mg of proteins (mean values of n ≥ 2 experiments ± SEM). Immunoblots show the effect of LT and MKI on ERK phosphorylation (pERK) (inset). H. Gene transcript modulations assessed by qRT-PCR. HUVEC monolayers were intoxicated 24 h with B. anthracis purified native toxin (LT Ba: PA+LF, 3 + 1 µg ml−1). Mean values of n ≥ 3 independent biological replicates per condition ± SEM (*P < 0.05 versus control cells).
Fig. S2. Control of LT activity on gene expression and Rho signalling pathway. A. HUVEC monolayers were treated with either LT (PA+LF, 3 + 1 µg ml−1) or LTm (PA+LFE687A, 3 + 1 µg ml−1). At the indicated times the activity of LT on MEK2 cleavage and ERK phosphorylation was assessed by anti-MEK2N20 or anti-p-ERK1/2 immunoblotting. Immunoblot anti-actin and anti-ERK1/2 shows equal protein loading. B. Cluster of genes induced by LT (Log2 ≥ 1, Green) at both 4 and 8 h. Genes found twice or more are indicated in bold. C and D. Immunoblots show the absence of impact of LT on the activation of Rac and Cdc42. HUVEC monolayers were intoxicated with LT (PA+LF, 3 + 1 µg ml−1) for the indicated times. CNF1 toxin 10−9 M was used as positive control of Rho protein activation. Levels of active-Rac and active-Cdc42 were determined by GST-p21pak RBD pull down (RacGTP and Cdc42GTP). Immunoblots anti-Rac and anti-Cdc42 performed on total cell lysate show total protein levels. Immunoblot anti-actin shows equal loading. E. Immunoblots showing the levels of Rac and Cdc42 in membrane and cytosolic fractions. Cells were intoxicated 24 h with LT. Membrane fractionation was controlled by immunoblotting anti-RhoGDI (cytosol) and anti-transferrin receptor (membrane). F. Immunoblots anti-MEK2N20 showing the absence of interference of either EDIN or Y-27632 on LT-induced MEK2 cleavage. HUVEC monolayers were intoxicated 24 and 48 h with LT (PA+LF, 3 + 1 µg ml−1) together with EDIN (100 µg ml−1) or Y-27632 (20 µM).
Fig. S3. List of primers used for qRT-PCR experiments. Table shows primers used for qRT-PCR and expression vectors constructs. The asterisk ‘*’ indicates primers previously described by (Indraccolo et al., 2007).
Video S1. Endothelium monolayer intoxicated from 0 to 12 h by LT. HUVEC monolayer recorded from 0 to 12 h after intoxication with LT (PA+LF, 3 + 1 µg ml−1), using a 20× lens. Movie recorded at 1 frame per minute and played at 1 frame per 1/30 s.
Video S2. Endothelium monolayer intoxicated from 12 to 24 h by LT. HUVEC monolayer recorded from 12 to 24 h after intoxication with LT (PA+LF, 3 + 1 µg ml−1), using a 20× lens. Movie recorded at 1 frame per minute and played at 1 frame per 1/30 s.
Video S3. Endothelium monolayer intoxicated from 24 to 36 h by LT. HUVEC monolayer recorded from 24 to 36 h after intoxication with LT (PA+LF, 3 + 1 µg ml−1), using a 20× lens. Movie recorded at 1 frame per minute and played at 1 frame per 1/30 s.
Video S4. Control monolayer. HUVEC monolayer recorded from 0 to 24 h using a 20× lens. Movie recorded at 1 frame per minute and played at 1 frame per 1/30 s.
Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
|CMI_1438_sm_FigS1.tif||1827K||Supporting info item|
|CMI_1438_sm_FigS2.tif||652K||Supporting info item|
|CMI_1438_sm_FigS3.tif||356K||Supporting info item|
|CMI_1438_sm_VideoS1.avi||5330K||Supporting info item|
|CMI_1438_sm_VideoS2.avi||5166K||Supporting info item|
|CMI_1438_sm_VideoS3.avi||5338K||Supporting info item|
|CMI_1438_sm_VideoS4.avi||10028K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.