Functional interactions between anthrax toxin receptors and the WNT signalling protein LRP6

Authors


*E-mail gisou.vandergoot@epfl.ch; Tel. (+41) 21 6931 792; Fax (+41) 21 6939 538.

Summary

To exert its activity, anthrax toxin must be endocytosed and its enzymatic toxic subunits delivered to the cytoplasm. It has been proposed that, in addition to the anthrax toxin receptors (ATRs), lipoprotein-receptor-related protein 6 (LRP6), known for its role in Wnt signalling, is also required for toxin endocytosis. These findings have however been challenged. We show that LRP6 can indeed form a complex with ATRs, and that this interaction plays a role both in Wnt signalling and in anthrax toxin endocytosis. We found that ATRs control the levels of LRP6 in cells, and thus the Wnt signalling capacity. RNAi against ATRs indeed led to a drastic decrease in LRP6 levels and a subsequent drop in Wnt signalling. Conversely, LRP6 plays a role in anthrax toxin endocytosis, but is not essential. We indeed found that toxin binding triggered tyrosine phosphorylation of LRP6, induced its redistribution into detergent-resistant domains, and its subsequent endocytosis. RNAis against LRP6 strongly delayed toxin endocytosis. As the physiological role of ATRs is probably to interact with the extracellular matrix, our findings raise the interesting possibility that, through the ATR–LRP6 interaction, adhesion to the extracellular matrix could locally control Wnt signalling.

Introduction

Anthrax toxin is one of the two major virulence factors produced by Bacillus anthracis, the causative agent of anthrax (Abrami et al., 2005; Young and Collier, 2007). The toxin is composed of three polypeptide chains, the protective antigen (PA), the lethal factor (LF) and the oedema factor (EF). LF is a metalloprotease that cleaves MAP kinase kinases and EF a calmodulin-dependent adenylate cyclase (Moayeri and Leppla, 2004). PA has no enzymatic activity but is responsible for binding to target cells and delivering the enzymatic subunits to the cytoplasm. Two receptors have been identified: tumour endothelial marker 8 (TEM8) and capillary morphogenesis gene 2 (CMG2), two type I transmembrane proteins that share a high degree of similarity both in the extracellular and the cytoplasmic domains (Scobie and Young, 2005). PA is produced by the bacterium as a monomeric 83 kDa protein (PA83) that can bind to the receptors but must be proteolytically processed to a 63 kDa form (PA63) to undergo polymerization into a heptameric ring-like structure (PA637mer) (Abrami et al., 2005; Young and Collier, 2007). PA637mer acts as the receptor for EF and LF (Cunningham et al., 2002). The hetero-oligomeric PA-EF/LF complex is subsequently endocytosed via a clathrin dependent pathway (Abrami et al., 2003) and transported to endosomes where a pH-dependent conformational change leads to pore formation of PA637mer and membrane translocation of EF and LF (Abrami et al., 2005).

Using an expressed sequence tag (EST) screen to silence chromosomal genes, it was recently reported that the lipoprotein-receptor-related protein (LRP) 6 plays an essential role in endocytosis of anthrax toxin (Wei et al., 2006). LRP6 is well known for its role as a co-receptor in Wnt signalling (Clevers, 2006). Wnt signalling pathways operate during development and regulate key processes such as cell proliferation and specification of cell fate (Logan and Nusse, 2004). In canonical Wnt signalling, Wnt first binds to members of the Frizzled family which then associate with LRP6 to form a ternary complex. This interaction leads to the serine/threonine phosphorylation of the cytoplasmic tail of LRP6, which then recruits axin (Clevers, 2006). Axin is thereby retrieved from the so-called β-catenin destruction complex. β-Catenin, which is thus rescued from degradation, can migrate to the nucleus to interact with TCF/LEF factors to activate transcription (Clevers, 2006). Whereas the role of LRP6 in canonical Wnt signalling is well established, recent evidence suggests that the protein could also play a role in non-canonical Wnt signalling (Hassler et al., 2007; Tahinci et al., 2007).

The finding that LRP6 is involved in anthrax toxin endocytosis has, however, been recently challenged (Young et al., 2007; Ryan and Young, 2008). It was first found that mice with targeted deletions of the genes encoding for LRP6 or the related protein LRP5 were killed by anthrax toxin and that mouse embryonic fibroblasts from LRP6−/− mice were toxin sensitive (Young et al., 2007). It was later shown that RNAi against LRP6 in HeLa cells did not prevent endocytosis of anthrax toxin (Ryan and Young, 2008).

We here sought to clarify whether or not LRP6 interacts functionally with the anthrax toxin receptors (ATRs) and to understand what some of the consequences of this interaction are.

Results

LRP6 and ATRs can be found in a complex

To investigate whether LRP6 and CMG2 can interact, we performed immunoprecipitation experiments, using Baby Hamster Kidney cells, as both LRP6 and CMG2 can be detected by Western blotting in these cells. Immunoprecipitation of LRP6 led to the co-precipitation of CMG2, but not of the transferrin receptor, under both low (Fig. 1A) and high (not shown) salt conditions. This interaction was observed in the absence of toxin, but appeared to be stabilized by PA treatment (Fig. 1A). Due to the lack of suitable anti-TEM8 antibodies, interaction between LRP6 and TEM8 isoform 1 (TEM8/1, Fig. S1A) was investigated upon ectopic expression of the proteins in HeLa cells, which are of human origin and easy to transfect. LRP6 and TEM8/1 could be co-immunoprecipitated (Fig. S1B), as could LRP6 and TEM8 isoform 2 (not shown), which has a very short cytoplasmic tail (Fig. S1A). These observations support previous data showing that LRP6 can interact with TEM8 (Wei et al., 2006) and extend these findings to CMG2. That interactions exist between LRP6 and ATRs was further supported by the significant colocalization observed at the cell surface when LRP6 was co-overexpressed with either of the ATRs (Fig. S2), again in agreement with previous findings on TEM8 (Wei et al., 2006).

Figure 1.

LRP6 interacts with ATRs.
A. Endogenous LRP6 was immunoprecipitated from 3 × 106 BHK cells treated or not with 500 ng ml−1 PA83 for 1 h at 4°C and for 20 min at 37°C. Immunoprecipitates and TCE (total cell extracts, 1/20 of the immunoprecipitate) were analysed by SDS-PAGE and Western blotting against endogenous CMG2, LRP6 and the transferrin receptor (Tf-R).
B. HeLa cells were transfected or not for 24 h with TEM8/1-HA or CMG2/4-V5. Cell lysates (40 μg of proteins) were analysed by SDS-PAGE and Western blotting against endogenous LRP6, HA for TEM8-HA, V5 for CMG2-V5, and tubulin as an equal loading marker.
C and D. HeLa cells were transfected for 24 h with myc-LRP6 and/or TEM8/1-HA (C) or CMG2/4 (D). Cell lysates (40 μg of proteins) were analysed by SDS-PAGE and Western blotting against myc to reveal myc-LRP6, HA for TEM8-HA, V5 for CMG2-V5 and tubulin as an equal loading marker.
E. WT myc-LRP6 was expressed for 24 h in HeLa cells, with and without TEM8/1, TEM8/2. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting using anti-myc for myc-LRP6, HA for TEM8-HA and V5 for CMG2-V5.

Interestingly, when analysing total HeLa cell extracts of cells overexpressing TEM8/1-HA, we found that both endogenous (Fig. 1B) and ectopically expressed (Fig. 1C) LRP6 were strongly downregulated. The reduced amount of the myc-LRP6 protein in TEM8/1-overexpressing cells was not due to a change in the level of mRNA (Fig. S1C). This downregulation was specific to isoform 1 as overexpression of TEM8 isoform 2 – which has a short cytoplasmic tail (Fig. S1A) – had no significant effect of the LRP6 level (Fig. 1E). Downregulation of LRP6 was equally observed when overexpressing CMG2, whether tagged with V5 (Fig, 1B and D) or GFP (not shown). Thus, excess of full-length ATRs leads to the downregulation of LRP6 (Fig. 1C and D). These findings further support that LRP6 and ATRs interact.

Downregulation of LPR6 by ATRs occurs post-translationally

To investigate whether downregulation of LRP6 by overexpression of ATRs was post-translationnal, as suggested by the unaffected levels of LRP6 mRNA (Fig. S1C), we treated cells with the proteasome inhibitor MG132, and found that endogenous LRP6 was thus rescued (Fig. 2A). To analyse LRP6 degradation more directly, we performed metabolic pulse-chase experiments. The chaperone Mesd (Hsieh et al., 2003) was coexpressed under all conditions to ensure proper folding of myc-LRP6 and its proper targeting to the plasma membrane (Abrami et al., 2008). As shown in Fig. 2B and C, degradation of LRP6 was accelerated upon coexpression of either CMG2 or TEM8/1, leading a ≈2 h decrease in the half-life of the protein. LRP6 degradation was however not affected by overexpression of TEM8 isoform 2 (Fig. 2B).

Figure 2.

Influence of ATRs on the degradation of LRP6.
A. HeLa cells were transfected or not for 48 h with TEM8/1-HA or CMG2/4-V5. Cells were treated or not with 10 μm MG132 for 18 h in complete medium. Cell lysates (40 μg of proteins) were analysed by SDS-PAGE and Western blotting against endogenous LRP6, against HA for TEM8-HA and V5 for CMG2-V5.
B and C. HeLa cells (106 cells) transfected for 30 h with myc-LRP6 and MESD, and without or with TEM8/1-HA, TEM8/2 or CMG2/4, were metabolically labelled for 20 min and chased for different times at 37°C with complete medium. Proteins were immunoprecipitated and analysed by autoradiography. Quantification was performed using the Typhoon scanner. (B) represents the mean of four experiments (except for TEM8/2) and (C) shows a typical experiment. Errors represent standard deviations. Paired t-test analysis between LRP6 and LRP6 plus either of the ATRs led to **P < 0.05 at times 2 and 4 h showing that the observed differences are statistically significant.
D. HeLa cells were transfected for 24 h with the myc-LRP6 mutant C1394A-C1399A with and without TEM8/1-HA, or CMG2/4-V5. Cell lysates (40 μg of proteins) were analysed by SDS-PAGE and Western blotting against myc, HA and V5 to reveal mutant LRP6, TEM8 and CMG2 respectively.

Despite the fact that overexpressed ATRs (Abrami et al., 2006; Deuquet et al., 2008) and LRP6 (Abrami et al., 2008) (Fig. S2) localize to the cell surface, the ability of MG132 to rescue LRP6 upon overexpression of ATRs suggests that downregulation of LRP6 might occur via the ER-associated degradation pathway (ERAD) (Anelli and Sitia, 2008). ERAD is indeed mediated by the proteasome after retrotranslocation of the protein to be degraded into the cytosol, in contrast to the lysosomal route, which is generally involved in clearing proteins from the plasma membrane and is proteasome independent. To test whether downregulation of LRP6 by overexpression of ATRs could take place at the ER level, we made use of a mutant of LRP6 that is retained in the ER, namely C1394A-C1399A (Abrami et al., 2008). This mutant is palmitoylation deficient and we have shown that this leads to ER retention (Abrami et al., 2008). As the wild-type (WT) protein, LRP6 C1394A-C1399A was downregulated by overexpression of either CMG2 or TEM8/1 (Fig. 2D). Also, when WT-LRP6 was expressed in the absence of its chaperone Mesd, leading to ER retention (Abrami et al., 2008), it was still downregulated by ATR overexpression (not shown). Finally, we also found that CMG2 point mutants, such as L45P or G105D, which are found in Systemic Hyalinosis patients (Dowling et al., 2003) and are retained in the ER (Deuquet et al., 2008), can also trigger downregulation of LRP6 (not shown). Together these observations show that downregulation of LRP6 by ATRs can occur in the ER, and thus that the proteins can already interact in the early secretory pathway.

ATRs and Wnt signalling

We next investigated whether the observed interactions between LRP6 and ATRs are relevant to some of the functions of the proteins, namely Wnt signalling and anthrax toxin endocytosis. We first determine the effects of knocking down the expression of ATRs by RNAi on Wnt signalling. As an early event in this signalling pathway, we chose to follow the stabilization of β-catenin, the mediator of the signal to the nucleus (Clevers, 2006). For this, we used L cells, which are of mouse origin and are commonly used in the Wnt signalling field because of their very low level of β-catenin at steady state (Blitzer and Nusse, 2006). The amounts of mRNA of both TEM8 and CMG2 could be efficiently reduced by RNAi and there was no cross-silencing, despite the high degree of similarity between the two ATRs (Fig. 3A). Interestingly, stabilization of β-catenin upon stimulation of L cells with Wnt was strongly impaired in cells in which either of the ATRs was knocked down (Fig. 3B and C), indicating that silencing of ATRs leads to a decrease/delay in Wnt signalling.

Figure 3.

Effect of silencing TEM8 and CMG2 on β-catenin accumulation. L cells were transfected or not with control siRNAs or siRNAs against TEM8 or CMG2 for 72 h.
A. mRNA levels of TEM8 and CMG2 were analysed by quantitative real-time RT-PCR.
B and C. L cells were incubated with Wnt3a conditioned medium for various time at 37°C. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting (B) to reveal β-catenin and tubulin, as an equal loading marker. (C) β-Catenin levels were quantified using the Typhoon scanner and normalized to the level at time 0. The plot represents the mean of four independent experiments. Errors represent standard deviations.
D. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting to reveal endogenous LRP6 and tubulin, as an equal loading marker.

ATRs and the level of LRP6 protein

With the aim of understating why silencing of ATRs affected Wnt-induced β-catenin stabilization, we monitored whether ATR RNAi influenced the levels of the LRP6 protein. Remarkably, LRP6 was drastically downregulated in these cells (Fig. 3D), providing a clear explanation for the reduced stabilization of β-catenin. During Wnt signalling, the β-catenin destruction complex is indeed disassembled due to the recruitment of Axin to the phosphorylated cytoplasmic tail of LRP6 (Zeng et al., 2008), an event that cannot occur in the absence of LRP6.

To investigate whether the downregulation of LRP6 observed when silencing ATRs was a particularity or not of L cells, we performed similar experiments in HeLa cells and used three different RNAi sequences for each protein to rule out any unspecific effects of a particular RNAi duplex. RNAi against CMG2 led to a strong decrease in the protein level (Fig. 4A) and had not effect on the mRNA levels of TEM8 and LRP6 (Fig. 4B). Similarly, RNAi against TEM8 led to a decrease in the amount of TEM8 mRNA but not of CMG2 and LRP6 mRNA (Fig. 4C). Remarkably again, the level of endogenous LRP6 protein was drastically reduced by RNAi against either ATR, as observed for all RNAi duplexes (Fig. 4D). RNAis against TEM8 and CMG2 equally led to the downregulation of overexpressed myc-LRP6 (Fig. S3). This decrease in myc-LRP6 protein was not observed when silencing the genes of other transmembrane proteins such as SREBP2 (SWISSPROT entry Q12772) and SCAP (Q12770), which localizes to the ER, or Site-1 (Q14703) or Site-2 (O43462) proteases, which localize to the Golgi (Fig. S3), or an irrelevant RNAi (control RNAi in Fig. 4D).

Figure 4.

ATRs are important to stabilize LRP6. HeLa cells were transfected for 72 h with control siRNAs or siRNAs against TEM8 or CMG2. Three different duplexes (labelled 1–3) were used against each gene.
A–C. The efficiency of the RNAi duplexes was tested by Western blotting for CMG2 (A) and quantitative PCR for CMG2 (B) and TEM8 (C).
D and E. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting against endogenous LRP6, endogenous CMG2, the transferrin Receptor (Tf-R) and tubulin. In (E), the siRNA No. 2 and 1 were used to silence CMG2 (CMG22) and TEM8 (TEM81) respectively.

As HeLa cells express both TEM8 and CMG2 (L. Abrami, B. Kunz, and G. van der Goot, unpublished RT-PCR observations) it was very surprising that silencing a single of the ATRs led to an almost complete loss of LRP6, instead of reaching an intermediate level. We therefore investigated whether silencing one ATR affected the protein level of the other. As antibodies are only available for CMG2, we silenced TEM8. As shown in Fig. 4E, RNAi against TEM8 led to a strong decrease in the level of CMG2 (which sometimes migrates as a doublet), even though this effect was weaker than RNAi against CMG2 itself.

Altogether our observations indicate that ATRs are important to stabilize LRP6, and that in their absence, LRP6 is prematurely degraded. This conclusion is further supported by the observation that we were unable to detect myc-LRP6 upon expression in ATR-deficient CHO cells despite the fact that the LRP6 protein was normally synthesized as shown by a pulse of metabolic labelling (not shown). In addition, the fact that RNAi against TEM8 leads to the downregulation of endogenous CMG2 indicates that the two ATRs somehow interact. It has been previously shown that TEM8 can self-associate, through interactions in the transmembrane domain (Go et al., 2006). Possibly, TEM8 could also associate with CMG2, given the high degree of similarity between the two proteins.

LRP6 and the anthrax toxin

We next investigated whether LRP6 plays a role in anthrax toxin endocytosis. We first determined whether LRP6 is part of the toxin receptor complex. We have previously shown that ATRs initially reside in the glycerophospholipid region of the plasma membrane but redistribute to lipid rafts upon anthrax PA binding and conversion to the 63 kDa form (Abrami et al., 2003). We therefore tested whether LRP6 would undergo a similar redistribution upon PA treatment by probing for detergent insolubility. In untreated cells, endogenous LRP6 was found in detergent soluble fractions (Fig. 5A, left). After treatment with PA, LRP6 was, however, fully redistributed to detergent-resistant membrane (DRM), co-fractionating with the DRM marker caveolin-1 as well as with PA63 (Fig. 5A, right).

Figure 5.

Anthrax toxin triggers relocalization and endocytosis of LRP6.
A. HeLa cells (106 cells) were incubated or not with 500 ng ml−1 PA83 for 1 h at 4°C and for 20 min at 37°C. Cells were solubilized in 1% Triton X-100 at 4°C, loaded at the bottom of on an Optiprep gradient, and six fractions were collected from the top and analysed by SDS-PAGE and Western blotting against endogenous LRP6, caveolin-1, transferrin receptor and PA.
B. HeLa cells were co-transfected for 48 h with plasmids encoding WT myc-LRP6 and its chaperone Mesd. Cells (106 cells) were incubated or not with 500 ng ml−1 PA83 for 1 h at 4°C and different times at 37°C. Immunoprecipitates against myc-LRP6 were analysed by SDS-PAGE and Western blotting against phosphotyrosine, myc-LRP6 and PA.
C. HeLa cells were co-transfected for 48 h with plasmids encoding WT myc-LRP6 and Mesd. Cells were incubated with 500 ng ml−1 PA83 for different hours at 37°C and extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting to reveal myc-LRP6 and tubulin.

We next determined whether LRP6 and anthrax PA could be co-immunoprecipitated. Myc-LRP6 was immunoprecipitated from cell lysates of PA-treated HeLa cells. All forms of PA detectable by SDS-PAGE – PA83, PA63 and the endosomal heptameric transmembrane channel (PA637mer) that is resistant to SDS – were found in the LRP6-immunoprecipitate (Fig. 5B). Moreover, using an anti-phosphotyrosine antibody, we found that PA triggers transient tyrosine phosphorylation of LRP6 (Fig. 5B). LRP6 is well known to become serine/threonine phosphorylated upon Wnt binding (Davidson et al., 2005; Zeng et al., 2005). This is, however, the first evidence that this protein can also undergo tyrosine phosphorylation. We did not observe phosphorylation of LRP6ΔC, which contains two of the 17 tyrosines in the cytoplasmic tail of LRP6, upon toxin treatment (Fig. S4A), suggesting that more downstream residues are the sites of modification, or alternatively that the cytoplasmic tail is required to recruit the kinase. Finally, we found that PA treatment led to rapid downregulation of myc-LRP6 (Fig. 5C), presumably through toxin-induced endocytosis and degradation in lysosomes.

Altogether, these observations show that LRP6 undergoes post-translational modifications, redistribution, endocytosis and downregulation upon treatment with PA demonstrating that it is part of the toxin-receptor complex.

LRP6 modulates anthrax toxin endocytosis

Endocytosis of the anthrax toxin can be monitored by the formation of the SDS-resistant heptamer PA637mer, which only forms upon arrival in the acidic environment of the endosomes (Abrami et al., 2004). Delivery of the enzymatic subunits to the cytoplasm can in turn be monitored by following the cleavage of the N-terminus of MEK1, one of the MAP kinase kinases that is a target of the LF (Abrami et al., 2004). To address the involvement of LRP6 in toxin uptake, we first analysed the effects of its overexpression. As shown in Fig. 6A–C, overexpression of LRP6 led to an increase in the rate of formation of the SDS-resistant PA637mer and a subsequent mild acceleration of MEK1 cleavage.

Figure 6.

Effect of LRP6 overexpression on anthrax toxin endocytosis.
A. HeLa cells were transfected or not with myc-LRP6 and Mesd for 48 h and incubated with 500 ng ml−1 PA83 and 50 ng ml−1 LF for 1 h at 4°C and different times at 37°C. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting to reveal myc-LRP6, tubulin, the N-terminus of MEK1, the C-terminus of MEK1 and the different forms of PA.
B and C. Quantification of PA oligomerization and MEK1 cleavage kinetics. N-terminal MEK1 levels were normalized to the level of N-terminal MEK1 at time 0. Errors represent standard deviations (n = 3 independent experiments). Paired t-test analysis between Ctrl and LRP6 overexpression conditions were performed, **P < 0.05, showing that the observed differences are statistically significant.

We next investigated the effect of knocking down LRP6 by RNAi (Fig. 7A and Fig. S4B). In disagreement with previous observations (Wei et al., 2006), we found that the knockdown of LRP6 had no significant effect on toxin binding (Fig. 7B). In contrast, the kinetics of formation of SDS-resistant PA637mer was drastically reduced during the first 40 min (Fig. 7B and C) as were the kinetics of MEK1 cleavage (Fig. 7B and D). We have previously shown that channel formation by PA637mer occurs in early endosomes (Abrami et al., 2004), a compartment that is reached within < 15 min. The here observed decrease in formation is therefore very pronounced.

Figure 7.

LRP6 promotes anthrax toxin endocytosis.
A. HeLa cells were transfected with control siRNAs or one of three different siRNAs against LRP6 for 72 h. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting to reveal endogenous LRP6, tubulin and transferrin receptor.
B. HeLa cells were transfected with control siRNAs or one of three different siRNAs against LRP6 for 72 h and incubated with 500 ng ml−1 PA83 and 50 ng ml−1 LF for 1 h at 4°C and different times at 37°C. Cell extracts (40 μg of proteins) were analysed by SDS-PAGE and Western blotting to reveal the N-terminus of MEK1, the C-terminus of MEK1 and the different forms of PA.
C and D. Quantification of PA oligomerization and MEK1 cleavage kinetics. N-terminal MEK1 levels where normalized to the level of N-terminal MEK1 at time 0. Errors represent standard deviations (n = 4 independent experiments). Paired t-test analysis between Ctrl RNAi and LRP6 RNAi 1 were performed, **P < 0.05, showing that the observed differences are statistically significant.

Altogether these observations show that although LRP6 is not essential for toxin uptake, it accelerates the endocytic process, possibly by promoting the recruitment of the endocytic machinery. These observations are apparently in direct disagreement with recent findings that RNAi against LRP6 had no effect on anthrax toxin activity in HeLa cells (Ryan and Young, 2008). The reasons for this discrepancy might be twofold. First, Ryan and Young (2008) used concentrations of PA and LF that were 4 and 3.6 times higher, respectively, than the ones used here, and higher toxin concentrations generally make differences between two conditions less apparent. Second, Ryan and Young (2008) used a single RNAi duplex, and this duplex was different from either three of the ones used here. Thus the level of knockdown might have been different between the two studies and knockdown level of Ryan and Young (2008) even though pronounced, might have been insufficient to see a strong effect. A 30% decrease in heptamer formation was however observed upon LRP6 RNAi in the study by Ryan and Young (2008) but this was not commented upon.

Discussion

We here show that LRP6 and ATRs – TEM8 and CMG2 – can form a complex in agreement with previous findings (Wei et al., 2006). This complex, the stochiometry of which remains to be determined, can already form shortly after synthesis, when the proteins are still in the ER (Fig. 2). ATRs actually appear to be required for the stabilization of LRP6 – and thereby play a crucial role in Wnt signalling. Indeed LRP6 could not be expressed in cells lacking ATRs, and RNAis against ATRs led to the downregulation of LRP6, and thereby to an inhibition of Wnt signalling. Interactions between LRP6 and ATRs were however not restricted to the early secretory pathway and also occurred at the cell surface, as indicated by a significant colocalization of LRP6 with ATRs (Fig. S2), the stabilization of the LRP6-ATR complex by the anthrax toxin (Fig. 1A) and the presence of LRP6 in the anthrax toxin-receptor complex upon toxin endocytosis (Fig. 5).

ATRs can thus be added to the list of transmembrane proteins that can interact with LRP6, the previously documented proteins being the Wnt receptor Frizzled (Clevers, 2006) and the type I membrane protein Kremen2, which is the receptor for the Wnt signalling antagonist Dickkopf (Mao et al., 2002). As an ATR, Kremen2 was shown to regulate cellular levels of LRP6 (Hassler et al., 2007). Downregulation of LRP6 by Kremen2 was however shown to require cell surface localization, and could not occur in the ER (Hassler et al., 2007). It therefore appears that cells have evolved multiple mechanisms to control the cellular levels of LRP6, and thereby Wnt signalling: Mesd ensures proper folding of the LRP6 β-propeller domains and thus its ER exit (Hsieh et al., 2003; Abrami et al., 2008), ATRs through an interaction that can already occur in the ER affects the total cellular LRP6 levels (this study), and Kremen2 regulates the amount of LRP6 that is present at the cell surface (Hassler et al., 2007). Once LRP6 has reached the cell surface, it appears to be in a dynamic equilibrium between various complexes, assembling with Frizzleds, Kremens, ATRs and possibly itself or additional unknown partners. It will be of interest to determine the mechanisms that regulated the formation of these various complexes, their composition and stochiometry, and how they are affected by the binding of ligands such as Wnt, Dickkopf or anthrax PA.

While LRP6 appears to require ATRs for significant expression in cells, the reverse is not true. LRP6 could indeed be silenced in cells without significantly affecting the level of ATRs. Also mouse embryonic fibroblasts from LRP6−/− mice were still sensitive to anthrax toxin indicating that they expressed ATRs (Young et al., 2007).

In cells that express both ATRs and LRP6, LRP6 is however present in the toxin–receptor complex, in agreement with the initial findings (Wei et al., 2006). LRP6 even ‘responds’ to toxin treatment in that it undergoes tyrosine phosphorylation, redistribution to specific domains and endocytosis and degradation upon treatment with anthrax PA. Moreover, LRP6 accelerates toxin entry, possibly by promoting recruitment of the endocytic machinery, as RNAi against ATRs significantly delayed toxin entry. Thus, LRP6 is not absolutely required for toxin entry as initially proposed (Wei et al., 2006), but promotes its endocytosis. Thus, in the absence of LRP6, as in the study by Young and Collier (2007), toxin entry will still occur, albeit at a lower speed, a difference that would not be detectable after 24 h, the time point used in this study (Young and Collier, 2007).

One must however keep in mind that ATRs have a function, beyond their role as ATRs. The physiological roles of ATRs have not yet been elucidated; however, the ability of TEM8 to bind collagens I and VI and that of CMG2 to bind collagen IV and laminin indicated that they are involved in interacting with the extracellular matrix (Bell et al., 2001; Dowling et al., 2003; Werner et al., 2006). Moreover TEM8 was shown to serve as an adhesion receptor and mediate actin-dependent spreading of cells on collagen I (Werner et al., 2006). Finally cells from patients suffering from Systemic Hyalinosis, a disease due to mutations in CMG2, were found to have impaired binding on laminin plates (Dowling et al., 2003). The ability of ATRs to interact with the extracellular matrix in trans and with LRP6 in cis raises the possible existence of a matrix-modulated mechanism for local control of Wnt signalling as recently proposed in a hypothesis (Olsen, 2007). The here demonstrated physical and functional interaction between CMG2 and LRP6 also raises the possibility that the complex clinical manifestation of Systemic Hyalinosis might be due in part to defects in Wnt signalling.

Experimental procedures

Reagents

Antibodies against anthrax PA and CMG2 were from the Leppla laboratory; antibodies against LRP6 were previously described (Khan et al., 2007); the antibody against the N-terminal peptide of MEK1 was produced in our laboratory; anti-MEK1 C-terminus was from Santa Cruz; anti-myc monoclonals were from Covance, CA; myc conjugated to agarose beads and caveolin-1 were from Santa Cruz Biotechnology, CA; anti-HA monoclonals and protein G-agarose-conjugated beads were from Roche (Applied Science, IN); anti-V5 monoclonals were from Invitrogen; anti-phosphotyrosine (clone 4G10) was from Upstate, NY; anti-tubulin monoclonals and rabbit anti-β-catenin were from Sigma; rabbit anti-LRP6 (C5C7) was from Cell Signaling technology; anti-human Transferrin receptor was from Zymed, CA; and finally HRP secondary antibodies were from Pierce, IL. Alexa-conjugated secondary antibodies from Molecular Probes.

The proteasome inhibitor MG132 was purchased from Sigma and used at a final concentration of 10 μM for 18 h in complete medium.

Cells, plasmids, siRNA, transfections and real-time PCR

BHK and HeLa cells were grown as described (Abrami et al., 2003; 2004; Liu and Leppla, 2003). Mouse L cells (ATCC) and L cells producing Wnt3a were grown in DMEM medium complemented with 10% FCS. Human LRP6 N-terminal tagged with myc was cloned in pCS expression vector. Mutant Myc-LRP6 C1394A-C1399A was generated by Quickchange (Stratagene) (Abrami et al., 2008). Human MESD cloned in pcDNA3 expression vector was provided by B. Holdener. In order to ensure proper folding and cell surface delivery of LRP6, cells were co-transfected with the chaperone Mesd (Hsieh et al., 2003; Li et al., 2006). Human TEM8-HA/1, TEM8-HA/2 were cloned in pIREShyg2 as described (Abrami et al., 2006). The human CMG2 (isoform 4) gene tagged with a V5 epitope was cloned in pcDNA3.1/V5-HIS-TOPO expression vector. Plasmids were transfected into HeLa cells for 24 or 48 h (2 μg of myc-LRP6 WT or mutants, with 0.5 μg of Mesd or empty vector, cDNA per 9.6 cm2 plate) using Fugene (Roche Diagnostics Corporation).

siRNA target sequences were the following: human LRP6-1 and human TEM8-1 were from Santa Cruz Biotechnology, CA; human LRP6-2: ctggattgttatccgactgaa, human LRP6-3: caggtgctaaccggatagtat, human CMG2-1: taggatttagtagtgcataaa, human CMG2-2: cacgtcgacgatgccaaatta, human CMG2-3: ctggagggtcgctatcacaaa, human TEM8-2: atccgtcaaggcctagaagaa, human TEM8-3: ctcggtcacactcaatgagaa, mouse TEM8-1: cagcactagttctcaatttaa, mouse CMG2-2: cacatcgatgatgtaatgata were purchased from Qiagen. Human S1P, S2P, SCAP and SREBP1 target sequences were previously described (Gurcel et al., 2006). As control siRNA we used the following target sequence of the viral glycoprotein VSV-G: attgaacaaacgaaacaagga. To silence human LRP6, human TEM8 or human CMG2, HeLa cells were transfected for 72 h with 100 pmol per 9.2 cm2 dish of siRNA using oligofectamine (Invitrogen) transfection reagent. To silence mouse TEM8 or mouse CMG2, L cells were transfected for 72 h with 100 pmol per 9.2 cm2 dish of siRNA using oligofectamine (Invitrogen) transfection reagent.

For real-time PCR, RNA was extracted from a confluent six-well dishes of HeLa cells treated or not with siRNAs, using commercial RNA easy mini extraction kits (Quiagen). RNA was quantified by spectrometry and 1 μg was used for reverse transcription using hexanucleotides (Roche). A 1/40 dilution of the cDNA was used to perform the real-time PCR using the Cyber Green reagent (Roche).

DRMs and immunoprecipitation

Detergent-resistant membranes were prepared using Optiprep gradients as described (Abrami et al., 2003). For immunoprecipitations, cells were lysed for 30 min at 4°C in immunoprecipitation buffer (0.5% NP40, 500 mM Tris-HCl pH 7.4, 20 mM EDTA, 10 mM NaF, 2 mM benzamidine and a cocktail of protease inhibitors, Roche), centrifuged for 3 min at 2000 g and supernatants were incubated for 16 h at 4°C with antibodies and beads.

Radiolabelling experiments

Metabolic labelling was performed as described (Abrami et al., 2006). Briefly, HeLa cells were transiently transfected (30 h) with LRP6 cDNAs, with and without CMG2 or TEM8, with MESD, washed with methionine/cysteine-free medium, incubated for 20 min at 37°C with 70 μCi ml−1 35S-methionine/cysteine (Hartman Analytics), washed and further incubated for different times at 37°C in complete medium with a 10-fold excess of non-radioactive methionine and cysteine. LRP6 was immunoprecipitated. Autoradiographies were quantified using the Typhoon Imager (Image QuantTool, GE healthcare).

Immunofluorescence microscopy

LRP6-, CMG2-, TEM8-expressing HeLa cells were fixed with 3% paraformaldehyde, permeabilized with 0.1% Triton X-100 and labelled with anti-HA, anti-V5 monoclonals antibodies, rabbit anti-Lrp6 followed with Alexa 488- or 568-conjugated goat anti-rabbit, anti-mouse IgG.

Images were acquired using a 100× lens on an Axiovert 200 equipped with a VisioCam camera (Carl Zeiss Microimaging).

Acknowledgements

We thank S. Leppla for the anthrax toxin, TEM8 cDNA plasmids, anti-PA and anti-CMG2 antibodies and critical reading of the manuscript, B. Holdener for the MESD cDNA, John Martignetti for CMG2/4 cDNA and C. Niehrs for being extremely generous with all his LRP6 and Wnt signalling reagents. This work was supported by the Swiss National Science Foundation. J.D. is a recipient of a fellowship from the ENDOCYTE Marie Curie Research Training Network. G.v.d.G is an international Fellow of the Howard Hughes Medical Institute.

Ancillary