The LRP5 and LRP6 genes likely arose from duplication of a single ancestral gene approximately 500 million years ago during the emergence of the chordate phylum. Since then LRP5 and LRP6 have retained a relatively high degree structural similarity and paralogous conservation, resulting in partially redundant functions as Wnt receptors in vertebrates. However since this early gene duplication, it is likely that LRP5 and LRP6 have independently evolved to the receptors they are today in the human genome. Genetic studies suggest that LRP6 has a more dominant role in Wnt signal transduction 
. Consistent with the severity of the developmental phenotypes of Lrp5 and Lrp6 mutant mice 
, our comparison of the MEFs deficient for Lrp5 or Lrp6 indicates that a majority of Wnt/β-catenin signaling is mediated through Lrp6 (). Moreover we were able to show a dose dependent correlation of Wnt responsiveness in MEFS containing two, one or no functioning Lrp6 alleles. Our data in MEFs in general are consistent with the previous genetic analyses of Lrp5 and Lrp6 compound mutant mice with different allelic combinations 
. Further evidence of the critical importance of LRP6 is found through interspecies protein comparisons. The orthologous conservation of human LRP6 to mouse (98%), chicken (92%) and Xenopus
(85%) Lrp6 is greater than human LRP5 to mouse (94%), chicken (88%) and Xenopus
(80%) Lrp5 (Percentage numbers represent amino acid sequence identity based on BLAST2 comparisons, data not shown), indicating that LRP6 is more conserved during evolution whereas changes in LRP5 may be more tolerable to the organism. These observations may also in part explain the small number of identified human LRP6 mutations in the literature compared to those of LRP5 
. In this study we aimed to understand the molecular underpinning that accounts for different LRP5 and LRP6 signaling activities for two considerations. First, both LRP5 and LRP6 are key therapeutic targets for treatment of human diseases including cancer and/or osteoporosis; and secondly, LRP5 and LRP6 share five virtually identical orthologous PPPSPxS motifs, implying that other unknown mechanisms may operate to tune the Wnt receptor activities through these motifs.
By generating a series of reciprocal chimeric receptors between LRP5 and LRP6, we found that difference between the cytoplasmic domain of LRP5 and LRP6 underlies, to a significant degree, the different signaling activities of these two receptors (). Our previous comprehensive analysis has revealed that the five PPPSPxS motifs, from A to E (Figure S1B
), contribute to LRP6 signaling in the rank order of C>D
E>A>B, with the carboxyl C, D, and E cluster making up most of the LRP6 signaling output 
. Comparing LRP5 and LRP6, we found that motifs A plus B in LRP6 have stronger activity than the corresponding A and B pair in LRP5, while similarly the cluster of motifs C and D and E of LRP6 is more potent than that of LRP5 ( and Figure S2
). Most strikingly, our analysis identified between motifs D and E a short intervening region, termed gap4, which has significant and reciprocal effects on LRP5 and LRP6 activities. Thus LRP5 with an altered and “LRP6-like” gap4 exhibits stronger signaling comparable to LRP6, whereas LRP6 with an altered and “LRP5-like” gap4 region exhibits weaker signaling similar to LRP5 ( and ). Indeed our data suggest that the contribution by gap4 to LRP6 signaling is on par to that by motif E and is more prominent than that by motif A or B () 
It has been established that Axin binds to both LRP5 
and LRP6 
, and that phosphorylated PPPSPxS motifs provide Axin-docking sites that mediate LRP6 signaling 
. The difference between LRP5 and LRP6 signaling could therefore be a result of their different Axin-binding properties. But we found that recombinant LRP5C and LRP6C, upon in vitro phosphorylation by GSK3 and CK1, exhibit indistinguishable phosphorylation-dependent binding to Axin (), suggesting that the difference of LRP5 and LRP6 signaling in vivo is likely at a step prior to Axin-binding, i.e., at how effectively the PPPSPxS motifs are phosphorylated. Fully consistent with this possibility, we found that, using antibodies specific for phospho-A, -C, and –E, LRP6 is more readily phosphorylated at these PPPSPxS motifs than its LRP5 counterparts (). More revealingly, the gap4 region in LRP5 when altered to resemble that in LRP6 not only enhances LRP5 signaling but also LRP5 phosphorylation at these PPPSPxS motifs (). On the other hand other gap regions (gaps1, 2, and 3) have minimal effects on LRP5 signaling or phosphorylation (). Thus we suggest that the difference in LRP5 and LRP6 signaling activity is largely due to their effectiveness of phosphorylation at PPPSPxS motifs, and we have identified the gap4 region that is responsible, at least in a significant part, for the difference between LRP5 and LRP6 phosphorylation and signaling.
The gap4 region lies between motifs D and E, but it has a strong effect on phosphorylation of not only motif E (and possibly D), but also of motifs C and A that are some distance away ( and Figure S1B
). We believe that this is consistent with the “LRP6 signal amplification” model we previously proposed 
, which corroborates genetic observations in Drosophila
. This model is based on the observation of a local positive feed forward loop between Axin and LRP6 PPPSPxS motifs. Thus Axin not only binds to phosphorylated PPPSPxS motifs, but also promotes and is required for PPPSPxS phosphorylation via its recruitment of the Axin-GSK3 complex 
. Therefore phosphorylation of one PPPSPxS motif has a stimulatory effect on that of other PPPSPxS motifs. Indeed we previously demonstrated that LRP6 phosphorylation at C or E profoundly relies on the presence of other PPPSPxS motifs 
. Here we interpret that the gap4 region, by directly regulating phosphorylation at motif E (and possibly motif D) nearby, exerts significant effects on phosphorylation of other motifs such as A and C () through such a signal amplification mechanism. This may also explain why gap4 has a prominent role in the overall LRP5 and LRP6 signaling output.
The gap4 sequence SYF is conserved among vertebrate Lrp5 proteins, and is distinct from SYSH in gap4 that is invariable among vertebrate Lrp6 orthologs (Figure S1B
). Such a conserved sequence difference together with its critical modulation of LRP5 and LRP6 signaling activity are unlikely to be co-incidental. Although the serine and tyrosine residues in gap4 call for potential (and differential) phosphorylation regulation in LRP5 and LRP6, our mutational analyses do not seem to support such a scenario ( and data not shown). How does gap4 regulate phosphorylation at PPPSPxS motifs and thereby LRP5/LRP6 signaling remains to be investigated. One possibility is that gap4 in LRP5 or LRP6, perhaps in conjunction with flanking residues, serves as a binding site for an unknown protein. Alternatively the sequence difference between LRP5 and LRP6 results in conformational difference that affects relative positioning of the last two PPPSPxS motifs (D and E), impacting signal amplification. Intriguingly the predicted secondary structure of LRP6 gap4 region contains a “turn” that is lacking in LRP5, and furthermore. the LRP5+gap4S mutant gains, while the LRP6 SH>F mutant loses, this predicted turn (Figure S3
), correlating with the increase and decrease of receptor signaling strength, respectively.
We note that while other gap regions, including gap2, do not appear to influence LRP5/6 activities in Wnt/β-catenin signaling, a recent study has shown that the RMTSV region of LRP6 gap2 serves as a potential phosphorylation site (T1558) for Protein Kinase A (PKA) and mediates LRP6 and Gαs
interaction in response to parathyroid hormone (PTH) binding to LRP6 
. This PTH responsive PKA site in gap2 is absent in Lrp5 (Figure S1B
). Although our data do not favor a similar phosphorylation-dependent regulation of gap4, this and our studies together highlight important regulatory roles of different gap regions between PPPSPxS motifs in LRP5 and LRP6 in Wnt and possibly other signaling pathways.
Our study further helps to resolve a controversial issue regarding LRP5/6-Axin interaction. Earlier findings based on yeast two-hybrid assays suggested that LRP5/6-Axin association is likely direct 
and is mediated through phosphorylated PPPSPxS motifs as Axin-docking sites 
, although this model has not ruled out the caveat that GSK3-like proteins in yeast may have a role in mediating the two-hydrid interaction. This direct interaction model is complicated/challenged by the findings that GSK3 also binds to LRP6 and performs PPPSP phosphorylation 
, and that phosphorylated PPPSPxS peptides can inhibit GSK3 phosphorylation of β-catenin presumably through direct interaction with GSK3 
. Furthermore the serine/threonine-rich region upstream of the PPPSPxS motif A (Figure S1B
) may also bind to GSK3 
. Given these scenarios and the established Axin-GSK3 interaction, a recent in vitro study has argued that Axin-LRP6 interaction is indirect and requires GSK3 as an intermediate bridge 
. Using recombinant GST-LRP5C or -LRP6C plus GSK3 (and CK1) we performed in vitro reconstitution of phosphorylation-dependent LRP5C/LRP6C-Axin interaction. We found that GSK3 indeed binds to GST-LRP5C and -LRP6C, but not to control GST, this binding is however independent of LRP5C/LRP6C phosphorylation (). By contrast, Axin binds only to GST-LRP5C and -LRP6C that have been phosphorylated by GSK3 plus CK1 (). Importantly the presence of GSK3 with LRP5C/LRP6C does not result in Axin recruitment (). Furthermore, LRP6CΔS/T, which harbors a deletion of the serine/threonine-rich region that was suggested to be a GSK3-binding site 
, binds to Axin in a phosphorylation-dependent manner that is indistinguishable to the wild type LRP6C (). Therefore our results are consistent with a direct LRP5/6-Axin interaction and do not support the model that GSK3 is the intermediate bridge between LRP5/6 and Axin. We note that in experiments supporting the GSK3-bridging model 
, the authors employed an Axin fragment that lacks the so-called DIX domain, which was suggested to be required for LRP5/6-Axin interaction 
In summary we have determined that the cytoplasmic domain of LRP5 and LRP6 plays a major role in the different signaling activity of these two Wnt receptors, and identified between the last two carboxyl PPPSPxS motifs an intervening gap4 region that has a key modulatory function in LRP5/LRP6 phosphorylation and signaling output. We have also provided evidence that argues for direct LRP5/LRP6-Axin interaction. Collectively our data provide significant new insights into the molecular mechanism of LRP5/LRP6 in Wnt signal transduction in development and human diseases.