Functional Dissection of the CCBE1 ProteinNovelty and Significance
A Crucial Requirement for the Collagen Repeat Domain
Rationale: Collagen- and calcium-binding EGF domain–containing protein 1 (CCBE1) is essential for lymphangiogenesis in vertebrates and has been associated with Hennekam syndrome. Recently, CCBE1 has emerged as a crucial regulator of vascular endothelial growth factor-C (VEGFC) signaling.
Objective: CCBE1 is a secreted protein characterized by 2 EGF domains and 2 collagen repeats. The functional role of the different CCBE1 protein domains is completely unknown. Here, we analyzed the functional role of the different CCBE1 domains in vivo and in vitro.
Methods and Results: We analyzed the functionality of several CCBE1 deletion mutants by generating knock-in mice expressing these mutants, by analyzing their ability to enhance Vegfc signaling in vivo in zebrafish, and by testing their ability to induce VEGFC processing in vitro. We found that deleting the collagen domains of CCBE1 has a much stronger effect on CCBE1 activity than deleting the EGF domains. First, although CCBE1ΔCollagen mice fully phenocopy CCBE1 knock-out mice, CCBE1ΔEGF knock-in embryos still form rudimentary lymphatics. Second, Ccbe1ΔEGF, but not Ccbe1ΔCollagen, could partially substitute for Ccbe1 to enhance Vegfc signaling in zebrafish. Third, CCBE1ΔEGF, similarly to CCBE1, but not CCBE1ΔCollagen could activate VEGFC processing in vitro. Furthermore, a Hennekam syndrome mutation within the collagen domain has a stronger effect than a Hennekam syndrome mutation within the EGF domain.
Conclusions: We propose that the collagen domains of CCBE1 are crucial for the activation of VEGFC in vitro and in vivo. The EGF domains of CCBE1 are dispensable for regulation of VEGFC processing in vitro, however, they are necessary for full lymphangiogenic activity of CCBE1 in vivo.
- CCBE1 protein
- endothelium, vascular
- Hennekam lymphangiectasia-lymphedema syndrome
- vascular endothelial growth factor
The lymphatic vasculature plays a key role in tissue fluid homeostasis, fat absorption, and trafficking of immune cells.1 Dysfunction of lymphatic vessels leads to edema formation, and lymphatic vessels have also been implicated in other pathological conditions, such as inflammation and tumor metastasis.2
In vertebrates, lymphatic vessels originate from blood endothelial cells in the cardinal vein. In mice, lymphatic endothelial cells (LECs) are first specified on the dorsal side of the cardinal vein by expression of the lymphatic endothelial marker Prox1.3 Subsequently, LECs migrate away as strings of loosely connected cells and reorganize into 2 large, lumenized lymphatic vessels, the dorsolaterally situated peripheral longitudinal lymphatic vessel (PLLV) and the more ventrally localized primordial thoracic duct (pTD).4,5 From these structures, a lymphatic network emerges through sprouting lymphangiogenesis.
In zebrafish, lymphatic precursor cells arise when venous sprouts migrate dorsally from the cardinal vein at 32 hours post fertilization. Roughly half of these sprouts connect to an artery to form intersegmental veins, while the other half migrate dorsally to the horizontal myoseptum region and form a pool of lymphatic precursors cells, the parachordal lymphangioblasts.6,7 These parachordal lymphangioblasts subsequently migrate dorsally and ventrally along arteries to reorganize into the main lymphatic vessels, the TD, and the dorsal longitudinal lymphatic vessels.8
The main lymphangiogenic signaling pathway involves vascular endothelial growth factor-c/vascular endothelial growth factor receptor-3 (VEGFC/VEGFR3) signaling. Both in mouse and in fish, VEGFC/VEGFR3 signaling is crucial for the sprouting of venous cells from the cardinal vein.6,9,10 In VEGFC knock-out mice, LECs are specified, but fail to migrate away from the cardinal vein.11 Thus, the current model postulates that VEGFC acts as a morphogen to activate VEGFR3 signaling in LECs, which drives LECs out of the cardinal vein. In Vegfc mutant zebrafish, as well as in Vegfr3/Flt4 kinase dead mutants, venous sprouts fail to migrate out of the cardinal vein and subsequently, the formation of parachordal lymphangioblast and the major lymphatic vessels (TD and dorsal longitudinal lymphatic vessels) is impaired.9,12 In humans, mutations in VEGFR3 or VEGFC have been linked to Milroy and Milroy-like disease, a form of hereditary primary lymphedema.13,14
Recently, collagen- and calcium-binding EGF domains 1 (CCBE1) was identified to be essential for lymphatic development in zebrafish and mice.12,15 CCBE1 encodes a secreted protein that contains an EGF domain and a calcium-binding EGF domain (Ca-EGF) at the N-terminus and 2 collagen domains at the C-terminus. In zebrafish, Ccbe1 is required for lymphovenous sprouting from the cardinal vein. In CCBE1 knock-out mice, lymphatic progenitor cells are specified, but fail to migrate away from the cardinal vein (CV), hence resembling the VEGFC phenotype. However, in contrast to VEGFC knock-out mice, where no lymphatic sprouts are formed, CCBE1 knock-out embryos display a distinct migratory effect and exhibit abnormal, short Prox1-positive sprouts which fail to segregate from the cardinal vein.5 In both zebra fish and mice, CCBE1 interacts genetically with VEGFC.5,10 Recently, it has been shown that CCBE1 is a crucial regulator of VEGFC processing and regulates Vegfc-mediated induction of Vegfr3 signaling during embryonic lymphangiogenesis.10,17 In order for VEGFC to reach its full signaling capacity, the VEGFC protein needs to be activated by cleavage of the N- and C-terminal propeptides from the VEGF homology domain.16 In fact, CCBE1 regulates activation of VEGFC by enhancing proteolytic processing of VEGFC via the metalloprotease ADAMTS3.17 Importantly, CCBE1 is linked to Hennekam syndrome (HS), a human recessive disease characterized by lymphedema, lymphangiectasia, and mental retardation.18 To date, 9 different mutations in CCBE1 were identified to be causative19 for HS, with most mutations encoding missense mutations that affect the N-terminal part of the protein, namely the EGF domain, the Ca-EGF domain, or the cysteine-rich sequence upstream of the EGF domain. Only 2 of the known mutations affect the collagen repeat domains, probably affecting the stability of a triple helix conformation of the collagen domain by altering conserved glycines in a Gly-X-Y motif.19 One mutation introducing a frame-shift and encoding a predicted truncated protein lacking the collagen domain was found as a compound heterozygous mutation together with one of the other mutations.
Until now, it is unknown which parts of the CCBE1 protein are needed for its function in vivo. Thus, we performed a functional analysis of the different domains of CCBE1 using 3 different approaches. First, we generated knock-in mice expressing deletion mutants of CCBE1. Second, we tested whether mutant Ccbe1 molecules were able to increase the activity of Vegfc in an in vivo zebrafish assay. Third, we analyzed whether mutant CCBE1 molecules can activate VEGFC processing in vitro. We found that deletion of the collagen domain inhibited CCBE1 function in vivo phenocopying CCBE1 knock-out mice. Furthermore, a CCBE1ΔCollagen mutant was not able to activate VEGFC signaling in vivo and in vitro. Surprisingly, however, we found that deletion of the EGF domains had no effect on VEGFC activation by CCBE1 in vitro and a CCBE1ΔEGF mutant is still partially active in vivo as CCBE1ΔEGF mice develop rudimentary lymphatics. Consistent with this, Ccbe1ΔEGF is able to increase Vegfc signaling in zebrafish.
Collectively, our data suggests that the collagen domains of CCBE1 are crucial for its ability to increase VEGFC activity in vivo and in vitro. The EGF domains are necessary for its full lymphangiogenic activity in vivo; however, they are dispensable for enhancement of VEGFC processing in vitro.
Our results provide important insights into the functionality of CCBE1 protein domains, with possible implications for therapeutic lymphedema treatment.
Mouse lines were maintained at the Hubrecht Institute using standard husbandry conditions. Experiments were approved by the local Animal Experimentation Committee (DEC). Generation of knock-in mice is described in the Supplement.
VEGFC Processing In Vitro
Analyzing VEGFC processing in vitro is described in detail in the Supplement.
CCBE1 Variants Lacking Either the EGF or the Collagen Repeat Domains Cannot Substitute for WT CCBE1 In Vivo
To investigate the relevance of the different domains of CCBE1 in vivo, we generated knock-in mice expressing different deletion mutants of CCBE1, lacking either the EGF and Ca-EGF domains (CCBE1ΔEGF mice) or lacking both collagen repeat domains (CCBE1ΔCol mice; Figure 1A). The EGF domains were removed by generating mice with floxed exons 4 and 5 (Figure 1B). Cre-mediated recombination led to an in-frame deletion of exon 4 and 5 removing both EGF domains. Correct splicing from exon 3 to 6 was verified by sequencing of cDNA from homozygous CCBE1ΔEGF embryos (Online Figure IA). For CCBE1ΔCol mice, removal of the 2 collagen repeat domains was achieved by replacement of exon 7 of CCBE1 with the cDNA of exon 7 to 11 lacking amino acids 248 to 337 followed by a transcriptional stop (Figure 1C). Analogous to this knock-in, a CCBE1 full length (CCBE1FL) knock-in mouse was generated as a control in which exon 7 is replaced with the cDNA of exon 7 to 11 including the collagen domains followed by a transcriptional stop. Expression levels of CCBE1ΔCol and CCBE1FL in knock-in mice were confirmed by Western blotting (Online Figure IB).
Homozygous CCBE1FL control knock-in mice were viable and fertile and did not show any overt morphological defects (Figure 1D). Homozygous fetuses expressing CCBE1ΔEGF or CCBE1ΔCol presented with severe edema at E14.5 and die in utero, thus phenocopying the CCBE1 full knock-out in which a LacZ cassette was placed into the CCBE1 locus (CCBE1LacZ; Figure 1D).15 Heterozygous animals for both CCBE1ΔEGF and CCBE1ΔCol mutants were viable and fertile and did not show any overt edemic phenotype, but fetuses compound heterozygous for CCBE1ΔEGF and CCBE1ΔCol show the same phenotype as mice homozygous for the 2 alleles (Online Figure IC).
Thus, mutant CCBE1 proteins lacking either the EGF domains or the collagen repeat domains cannot substitute for wild-type (WT) CCBE1 in vivo.
CCBE1ΔCollagen Mice Lack All Lymphatic Structures, But Lumenized Lymphatic Structures Develop in CCBE1ΔEGF Fetuses
Previously, we have shown that CCBE1 knock-out mice lack all lymphatic vessels, whereas the patterning of blood vessels is not affected.15 To analyze lymphatic and blood vessel development in mice expressing CCBE1 lacking either the EGF domains or the collagen repeat domains, we performed skin stainings of mutant fetuses at E14.5 to E16.5 (Figure 2A, Online Figure II) using Lyve-1 and Prox1 as lymphatic markers and the pan-endothelial marker CD31 to visualize blood vessels. We observed no abnormalities in blood vessels in any of the mice. Moreover, CCBE1FL knock-in mice do not show obvious differences in lymphatic vessel development in the skin compared with WT animals. Homozygous CCBE1ΔCol fetuses phenocopy CCBE1LacZ mice lacking all lymphatic vessels. Interestingly, however, CCBE1ΔEGF mice do exhibit some lymphatic structures in the skin. Like lymphatic vessels in WT control fetuses, these structures sprouted into the direction of the dorsal midline at E14.5, but they failed to develop into a contiguous, branched plexus at both time points analyzed. At E16.5, patches of LECs had aggregated and formed large spherical structures.
Next, we performed Lyve-1 stainings of paraffin sections of E14.5 fetuses (Figure 2B) to investigate lymphatic development close to the CV. Although the pTD and large lymphatic vessels in other areas, such as the skin, could be readily identified in WT control and in CCBE1FL knock-in fetuses, lymphatic vessels were undetectable in CCBE1ΔCol or CCBE1LacZ mice. Analogous to the whole mount stainings, lumenized lymphatic structures were found in the skin of E14.5 CCBE1ΔEGF fetuses. Moreover, lymphatic structures were also identified at the position of the pTD; however, these structures were smaller than in WT control animals.
Collectively, these results reveal that although both the collagen domains and the EGF domains are necessary for lymphatic development, a CCBE1ΔEGF mutant does retain partial functionality in vivo.
Specified LECs Migrate From the Cardinal Vein in CCBE1ΔEGF Embryos, But Not in CCBE1ΔCollagen or CCBE1LacZ Embryos
To analyze lymphatic development in more detail and at an earlier stage, we visualized the formation of the first lymphatic structures by whole mount staining and optical clearing of E12.5 embryos. Recently, it had been shown that specified LECs leave the CV as streaks of cells to form the first 2 large, lumenized lymphatic vessels, the PLLV, and the pTD. In CCBE1LacZ knock-out embryos, as well as in CCBE1ΔCol embryos, Prox1+ or Lyve-1+ LECs were unable to leave the CV and were not detected outside of the CV. In CCBE1ΔEGF mice, however, some Prox1+ cells were able to leave the cardinal vein but only formed discontinuous lymphatic structures at the positions where the pTD and PLLV are normally located (Figure 3).
Thus, the CCBE1ΔCol mutant has no lymphangiogenic activity in vivo. The CCBE1ΔEGF mutant is able to induce migration of lymphatic progenitor cells out of the CV at early time points of lymphatic development, although it cannot induce the formation of the main lymphatic structures, the PLLV, and pTD.
Second Collagen Domain of Ccbe1 is Essential for Enhancing Vegfc Signaling in Zebrafish
To confirm and extend our findings in an independent system, we used an assay that assesses Vegfc signaling in zebrafish.10 A gene of choice is expressed specifically in the ventral-most region of the neural tube, the floor plate, from the shh promoter. The effect of the transgene on adjacent intersegmental vessels can then be monitored in vivo and over time.13 Previously, we have shown that Tg(shh:vegfc;shh:ccbe1) zebrafish, expressing both vegfc and ccbe1 as independent transgenes in the floorplate, exhibit aberrant ectopic turning of arteries at 32 hours post fertilization,10 whereas transgenic zebrafish carrying single transgenes do not. This phenotype can be attributed to high Vegfc/Vegfr3 signaling in arteries.9 Injection of ccbe1 mRNA in Tg(shh:vegfc) embryos that only express vegfc from the shh promoter also efficiently induced strong arterial sprouting (Figure 4B and 4H), whereas injection of ccbe1 mRNA into WT embryos had no effect (Online Figure III). First, we tested the activity of a ccbe1 mutant, fofhu3613, in this assay. fofhu3613 harbors a mutation in the EGF domain (D162E), leading to a loss of lymphatic development in zebrafish mutants.12 Unexpectedly, fofhu3613 mRNA has levels of activation that are indiscernible from wt ccbe1 (Figure 4B, 4C, and 4H). Interestingly and in accordance with the mouse data, ccbe1ΔEGF was still able to induce misturning of arteries, albeit to a lower degree than WT ccbe1 (Figure 4D and 4H). In contrast, ccbe1ΔCol mRNA injection did not induce ectopic arterial turning (Figure 4E and 4H).To further narrow down which part of the collagen domains was required for its function, we tested mutant forms of ccbe1 in which either the first or the second collagen domain was removed. Interestingly, we found that deletion of the first collagen domain, ccbe1ΔColA, only led to a partial loss of function in the ability to induce ectopic turning, whereas loss of the second collagen domain, ccbe1ΔColB, led to a complete loss of function (Figure 4F–4H).
Thus, the EGF and Ca-EGF domain, as well as the first collagen domain, are required for full function of the Ccbe1 protein, however, only removal of the second collagen domain fully renders the Ccbe1 protein dysfunctional in regulating Vegfc signaling in zebrafish. These data are in alignment with the above-described mouse mutant analysis.
Second Collagen Domain of CCBE1 Is Required for Enhancement of VEGFC Processing by CCBE1 In Vitro
CCBE1 has recently been shown to enhance proteolytic processing of VEGFC, which leads to an increase in bioavailability of the mature 21 kDa form of VEGFC.10,17 To assess how the various functional domains affect the ability of CCBE1 to increase mature VEGFC levels, we coexpressed human VEGFC, together with WT or various mutant forms of CCBE1 in HEK-293T cells. Conditioned media containing secreted VEGFC and CCBE1 were analyzed by Western blotting. As reported before, CCBE1 clearly stimulated the proteolytic processing of VEGFC as is apparent from an increased amount of the 21 kDa form of VEGFC. Surprisingly, we found that the EGF domains are not required for CCBE1 function in VEGFC processing in vitro as coexpression of CCBE1ΔEGF led to an increase in mature VEGFC comparable with that of WT CCBE1 (Figure 5B). Moreover, we tested whether CCBE1ΔEGF had different kinetics than WT CCBE1 by combining conditioned media from CCBE1 and VEGFC expressing cells for various periods of times. Interestingly, we could find no difference between CCBE1 and CCBE1ΔEGF (Online Figure IV). Also by lowering the amount of transfected CCBE1, we could not detect any difference in the activity of CCBE1 and CCBE1ΔEGF (Online Figure V). Moreover, CCBE1D170E, which harbors a point mutation that is equivalent to the fofhu3613 mutation in zebrafish, was also able to enhance processing of VEGFC (Online Figure V), further corroborating that perturbation of the EGF domain does not impair CCBE1-mediated processing of VEGFC in vitro.
In contrast, the CCBE1 mutant that lacks the collagen domains (CCBE1ΔCol) lost the ability to increase VEGFC processing (Figure 5B). Both deletion of solely the collagen domains or truncation of the C-terminal half of CCBE1 (CCBE1 1–175) abolished its activity (Online Figure VI). Furthermore, we metabolically labeled cells that were transfected with VEGFC and CCBE1 variants and precipitated VEGFC using VEGFR3-Fc, which further confirmed these results (Online Figure VII). In conclusion, the collagen domains are of crucial importance for the effect of CCBE1 on proteolytic processing of VEGFC.
To test how the different CCBE1 variants regulate VEGFC-mediated processes, such as proliferation and downstream signaling on LECs, we incubated LECs with conditioned media from HEK-293T cells.
Conditioned medium containing VEGFC and CCBE1 or CCBE1ΔEGF were able to increase proliferation even further compared with medium containing only VEGFC or VEGFC and CCBE1ΔCol (Online Figure VIII). To investigate downstream signaling, we also analyzed phosphorylation of ERK. In agreement with the proliferation assay, combinations of VEGFC with either CCBE1 or CCBE1ΔEGF induced a stronger phosphorylation of ERK than VEGFC on its own or VEGFC in the presence of CCBE1ΔCol (Figure 5C). These findings show that the collagen domains are crucial for the VEGFC-mediated activation of LECs.
Because the zebrafish experiments indicated that the 2 collagen domains are not equally important for CCBE1 function in mediating VEGFC signaling in vivo, we tested whether these domains would also have differential effects on VEGFC processing in vitro. Strikingly, deletion of the first collagen domain (CCBE1ΔColA) was dispensable for activation of VEGFC processing in this assay, however, deletion of the second collagen domain (CCBE1ΔColB) inhibited activation of VEGFC processing by CCBE1 (Figure 5D, compare lanes 12 and 13). Moreover, even deletion of amino acids 92 to 294, which deletes all parts of the protein from the EGF domains to the first collagen domain (CCBE1ΔEGFΔColA), did not have an overt effect on CCBE1 function in this assay (Figure 5D).
Because CCBE1 has been suggested to mediate processing of VEGFC via the ADAMTS3 protease,17 we also tested if ADAMTS3 cooperates with mutant forms of CCBE1. We generated a stable cell line expressing VEGFC and subsequently coexpressed ADAMTS3 and CCBE1. We found that overexpression of ADAMTS3 on its own induced a mild increase of proteolytic processing which was further increased by coexpression of CCBE1 or CCBE1ΔEGF (Figure 5D). In agreement with our other findings, lack of the collagen domains led to an almost complete abrogation of the synergism between CCBE1 and ADAMTS3 (Figure 5D). We found that under these experimental conditions CCBE1ΔCol gave a minor increase in VEGFC processing. The increase in processing was more pronounced, when we used higher amounts of CCBE1 (variant) medium (Online Figure IX). These data suggest that CCBE1ΔCol retains a weak capacity to mediate VEGFC processing in the presence of ADAMTS3, but only to a degree that is much lower than WT CCBE1 or CCBE1ΔEGF.
We further confirmed that the synergism of CCBE1 with ADAMTS3 depended more on the second collagen domain, as CCBE1ΔColA enhanced processing as effectively as WT CCBE1, whereas CCBE1ΔColB showed much less activity. However, CCBE1ΔColB expression did cause the appearance of low levels of mature VEGFC, which was not seen to that extent in response to CCBE1ΔCol (Figure 5D, compare lane 11–13). This probably reflects partial functionality of ColA, which is only apparent in presence of high levels of ADAMTS3. Next, we tested the binding capacities of the different CCBE1 variants to ADAMTS3. We consistently observed a lower binding propensity of CCBE1ΔCol to bind ADAMTS3 than WT CCBE1 (Online Figure X).
Taken together, these data suggest that the collagen domains, especially the second collagen domain, are of major importance for processing of VEGFC by CCBE1.
Hennekam Mutation in the Collagen Domain, But Not in the Ca-EGF Domain Fully Impairs CCBE1 Function
CCBE1 mutations have been shown to be causative for HS, a congenital autosomal recessive disease, which is characterized by lymphangiectasia, generalized lymphedema and mental retardation.18 Most of the mutations inducing HS have been identified affecting the N-terminal part of the protein, containing a cysteine-rich sequence N terminal of the EGF domains and the EGF and Ca-EGF domains (Figure 6A). Only 2 mutations have been found within the collagen domains, with 1 patient harboring a mutation in the second collagen domain, G327R. To interrogate the effect of the mutations responsible for the HS phenotype on the functionality of CCBE1, we tested 2 constructs that bear mutations found in patients with HS. One mutation, R158C, is located within the Ca-EGF domain and 1 mutation, G327R, within the second collagen repeat. First, we tested the functionality of these 2 mutants in vivo by injecting zebrafish ccbe1 mRNAs that bear the corresponding mutations into Tg(shh:vegfc) embryos. Whereas a zebrafish R150C mutation, equivalent to human R158C and located in the Ca-EGF domain, only marginally lowered activity compared with wt ccbe1 in this assay, the mutation in the collagen domain, G313R in zebrafish, completely abolished the ectopic Ccbe1 effect (Figure 6B and 6C). Moreover, in an in vitro processing assay using human CCBE1, the R158C mutation in the Ca-EGF domain did not affect functionality of CCBE1 compared with the WT molecule, whereas the G327R mutation in the second collagen domain led to a strong loss of VEGFC maturation (Figure 6D).
CCBE1 is one of the few genes that are indispensable for embryonic lymphatic development.1 It has been shown that CCBE1 is a crucial regulator of VEGFC activation via the protease ADAMTS3. However, which CCBE1 domains are necessary to exert its function is not known. Here, we have conducted a functional domain analysis, based on different in vitro and in vivo systems, to gain comprehensive insight into the requirement of the different conserved protein domains for CCBE1 activity.
Several lines of evidence have stressed the importance of the EGF domains in the past. First, the majority of mutations in CCBE1 in patients with HS have been found within or in close proximity to the EGF domains. Second, in the original and subsequent screens that identified zebrafish ccbe1 as an essential gene for lymphatic development, several alleles were identified in which the causative mutation was found to be located in the EGF or Ca-EGF domain.12 To gain more insight into the function of the EGF and Ca-EGF domains, we generated knock-in mice expressing a CCBE1 variant lacking these domains. Surprisingly, unlike the full CCBE1 knock-out fetuses which lack all lymphatic structures, CCBE1ΔEGF fetuses do develop rudimentary, lumenized lymphatic structures in the skin, and at the position of the pTD and PLLV. In the skin, these lumenized larger structures sprouted into the same direction to the dorsal midline as lymphatic vessels in WT fetuses. They did, however, fail to form a properly branched lymphatic network. In CCBE1ΔEGF embryos, remnant, fragmented lymphatic structures were located at the position of the pTD and PLLV indicating that specified LECs do migrate out of the cardinal vein roof to the same location as LECs in WT control embryos. In accordance with this, a Ccbe1ΔEGF mutant was partially functional in zebrafish, where the injection of ccbe1ΔEGF can still exert aberrant arterial sprouting in the Tg(shh:vegfc) zebrafish. Although fofhu3613 could increase Vegfc signaling in arteries to the same extent as wt ccbe1, it cannot substitute for wt ccbe1 in zebrafish because in contrast to wt ccbe1 mRNA, it is not able to rescue a ccbe1 mutant.12 Unexpectedly, loss of the EGF domains did not have any effect on the ability of CCBE1 to enhance processing of VEGFC in cell culture into the active form detected by the antibodies used. Thus, in vitro, the EGF and Ca-EGF domains are dispensable for the function of CCBE1 in VEGFC activation. However, as a CCBE1ΔEGF molecule cannot substitute for WT CCBE1 in mice and has reduced activity in increasing Vegfc signaling in zebrafish, the EGF domains do have a crucial function in vivo. The precise role of the EGF and Ca-EGF domains in vivo is not yet clear and needs further investigation in the future. It might be that this domain is necessary in other processes, such as binding to extracellular matrix components (ECM). Consistent with this notion, a CCBE1ΔCollagen-Fc protein has been shown to bind to vitronectin, Collagen I, Collagen IV, and Collagen V in an in vitro binding assay.15 Alternatively, the EGF domains could be involved in other aspects of VEGFC/VEGFR3 signaling, such as receptor binding, possibly also mediated through prior or simultaneous interaction of CCBE1 with ECM. Indeed, in a different assay, a cornea pocket assay, the N-terminal part of CCBE1 was able to enhance the lymphangiogenic effect of purified, fully processed VEGFC.15
In all 3 different model systems tested, we show that loss of the collagen domains has a profound effect on CCBE1 function. CCBE1ΔCol knock-in mice phenocopy the CCBE1LacZ full knock-out mice characterized by a failure of specified LECs to leave the cardinal vein. This was further confirmed by a total lack of enhancement of VEGFC signaling by ccbe1Δcol in Tg(shh:vegfc) zebrafish and the inability of CCBE1ΔCol to induce proteolytic processing of VEGFC. In particular, the second collagen domain was found to be of higher importance, as both our zebrafish and our in vitro experiments show that loss of the second collagen domain alone strongly reduce the function of the CCBE1 protein. The crucial role of the collagen domains may be because of their involvement in direct activation of ADAMTS3. Indeed, coexpression of ADAMTS3 and CCBE1ΔCol did not synergize in VEGFC processing and CCBE1ΔCol has a reduced capacity to bind ADAMTS3. Previously, it has been shown that a recombinant protein containing only amino acids 1 to 175 of CCBE1 was able to induce processing of recombinant pro-VEGFC.17 However, a comparative analysis between the N- and C-terminal parts of CCBE1 was not performed in that assay. Also in our system, we detected a slight increase in the amount of fully processed VEGFC when ADAMTS3 was overexpressed and high amounts of CCBE1ΔCol were used. Considering that CCBE1ΔCol also binds ADAMTS3, albeit with reduced affinity, ADAMTS3 may have binding epitopes both in the N- and C-terminal parts of CCBE1. Nevertheless, in vitro, even in excess, the activity of the N-terminal part of CCBE1 was markedly reduced compared with WT CCBE1 or the C-terminal part of CCBE1. Thus, these data suggest that, whereas the N-terminal part of CCBE1 can have some activity in proteolytic processing, this activity is clearly reduced compared with WT CCBE1 or the C-terminal part of CCBE1.
Our findings shed new light on the molecular defects that underlie HS and may provide a rationale that explains why so few HS mutations were found in the collagen domains. To date, most patients exhibit missense mutations affecting the N-terminal domain clustered in the EGF domain, the Ca-EGF or the cysteine-rich domain upstream of the EGF domain. We could show that a mutation within the Ca-EGF domain inducing HS, C158R, and a mutation that leads to lymphatic defects in zebrafish, fofhu3613, does not affect the ability of CCBE1 to activate VEGFC processing in vitro and can still enhance Vegfc signaling in vivo, indicating that these mutations are hypomorphic alleles. In fact, the phenotype of fofhu3613 mutant zebrafish can partially be rescued by expressing high levels of full length Vegfc in the floorplate of zebrafish.10 These results and the fact that a CCBE1ΔEGF molecule is partially active in vivo suggest that the HS mutations found within the EGF and EGF-Ca domains of CCBE1 are most likely hypomorphic alleles that retain some function.
In contrast, only 1 of 13 patients identified to date harbors a mutation within the second collagen domain. This mutation, G327R, probably affects the secondary structure of the collagen repeat.19 In our analyses, we found that the G327R mutation strongly reduced CCBE1 function in vitro and in vivo, almost completely impairing its ability to activate VEGFC processing. This result is surprising as such a strong loss of function would most likely result in a complete absence of lymphatics, and it is unlikely that many patients could survive who have mutations in the collagen-B domain. Therefore, we consider it more likely that predominantly weaker alleles of CCBE1 are viable and can be found in patients, for example, alleles that affect the EGF domain. Full loss-of-function situations are unlikely to be viable in humans. Of note, the only known patient bearing the G327R mutation is the only surviving child of 4 siblings.18 Furthermore, this patient is the only patient known to exhibit vascular defects in addition to lymphatic defects indicating that this mutation has a stronger effect than the other HS mutations found.19
Taken together our findings show that the collagen domains of CCBE1 are indispensable for its function in VEGFC activation in vitro and in vivo. The EGF domains are dispensable for this function in vitro, but are necessary for full activity of Ccbe1 in enhancing Vegfc/Vegfr3 signaling in vivo and for proper lymphangiogenesis in the mouse embryo. Contrary to the intuitive belief that the collagen domains bind to the ECM and that the EGF domain is more functionally relevant, we show here that it is the collagen repeat domain that is crucial for enhancement of proteolytic processing of VEGFC by CCBE1, presumably by activating ADAMTS3. The role of the EGF domain should be subject of future investigations but previous work has indicated their capacity to bind ECM proteins. In zebrafish, ccbe1 is expressed along the migration route of lymphatic precursor cells; thus, it is possible that binding of the EGF domains to ECM serves to direct specific locations of CCBE1 activity. At these dedicated areas, the collagen repeat domains of CCBE1 mediate enhanced processing of VEGFC by activation of ADAMTS3. Thus, CCBE1 provides positional information for VEGFC signaling, which orchestrates the migration of LECs (Figure 7). These molecular insights provide new entry points for therapeutical approaches to either stimulate or inhibit lymphangiogenesis by modulating CCBE1 function.
Imaging was performed at the Hubrecht Imaging Center. We would like to thank Thomas Clapes and Catherine Robin as well as René Hägerling and Friedemann Kiefer for protocols and suggestions. We thank Jeroen Korving for histological assistance and injection of ES cells into blastocysts.
Sources of Funding
D. Schulte was supported by a Marie Curie Intra-European Fellowship, and an EMBO Long-Term Fellowship (ALTF 1408–2011). M. G. Roukens was supported by a VENI grant (863.11.022) from the Netherlands Organization for Scientific Research (NWO). S. Schulte-Merker was supported by the KNAW and the CiM Cluster of Excellence EXC 1003/CiM. K. Alitalo, M. Jeltsch, and V.-M. Leppänen were supported by the European Research Council (ERC-2010-AdG-268804) and the Finnish Cancer Research Organizations. European patent No. 2359134 on the use of CCBE1 has been granted.
In February 2015, the average time from submission to first decision for all original research papers submitted to Circulation Research was 13.9 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.116.304949/-/DC1.
- Nonstandard Abbreviations and Acronyms
- calcium-binding EGF domain
- collagen- and calcium-binding EGF domain–containing protein 1
- CCBE1 lacking both collagen repeat domains
- CCBE1 lacking the EGF and Ca-EGF domains
- CCBE1 full length
- Hennekam syndrome
- intersomitic arteries
- lymphatic endothelial cells
- peripheral longitudinal lymphatic vessel
- primordial thoracic duct
- vascular endothelial growth factor-C
- vascular endothelial growth factor receptor-3
- Received July 31, 2014.
- Revision received March 25, 2015.
- Accepted March 26, 2015.
- © 2015 American Heart Association, Inc.
- Schulte-Merker S,
- Sabine A,
- Petrova TV.
- Wigle JT,
- Harvey N,
- Detmar M,
- Lagutina I,
- Grosveld G,
- Gunn MD,
- Jackson DG,
- Oliver G.
- Yang Y,
- García-Verdugo JM,
- Soriano-Navarro M,
- Srinivasan RS,
- Scallan JP,
- Singh MK,
- Epstein JA,
- Oliver G.
- Hägerling R,
- Pollmann C,
- Andreas M,
- Schmidt C,
- Nurmi H,
- Adams RH,
- Alitalo K,
- Andresen V,
- Schulte-Merker S,
- Kiefer F.
- Hogan BM,
- Herpers R,
- Witte M,
- Heloterä H,
- Alitalo K,
- Duckers HJ,
- Schulte-Merker S.
- Le Guen L,
- Karpanen T,
- Schulte D,
- Harris NC,
- Koltowska K,
- Roukens G,
- Bower NI,
- van Impel A,
- Stacker SA,
- Achen MG,
- Schulte-Merker S,
- Hogan BM.
- Karkkainen MJ,
- Haiko P,
- Sainio K,
- Partanen J,
- Taipale J,
- Petrova TV,
- Jeltsch M,
- Jackson DG,
- Talikka M,
- Rauvala H,
- Betsholtz C,
- Alitalo K.
- Gordon K,
- Schulte D,
- Brice G,
- Simpson MA,
- Roukens MG,
- van Impel A,
- Connell F,
- Kalidas K,
- Jeffery S,
- Mortimer PS,
- Mansour S,
- Schulte-Merker S,
- Ostergaard P.
- Bos FL,
- Caunt M,
- Peterson-Maduro J,
- Planas-Paz L,
- Kowalski J,
- Karpanen T,
- van Impel A,
- Tong R,
- Ernst JA,
- Korving J,
- van Es JH,
- Lammert E,
- Duckers HJ,
- Schulte-Merker S.
- Joukov V,
- Sorsa T,
- Kumar V,
- Jeltsch M,
- Claesson-Welsh L,
- Cao Y,
- Saksela O,
- Kalkkinen N,
- Alitalo K.
- Jeltsch M,
- Jha SK,
- Tvorogov D,
- Anisimov A,
- Leppänen VM,
- Holopainen T,
- Kivelä R,
- Ortega S,
- Kärpanen T,
- Alitalo K.
Novelty and Significance
What Is Known?
Collagen- and calcium-binding EGF domains 1 (CCBE1) protein is required for lymphatic vessel development in fish and in mice.
CCBE1 is associated with Hennekam syndrome, characterized by lymphedema, lymphangiectasia, and developmental delay.
CCBE1 regulates vascular endothelial growth factor-C (VEGFC) signaling via activation of VEGFC processing by ADAMTS3.
What New Information Does This Article Contribute?
The second collagen domain of CCBE1 is crucial for the activation of VEGFC processing via ADAMTS3 in vitro and for its lymphangiogenic activity in vivo.
The EGF domains of CCBE1 are negligible for the activation of VEGFC processing via ADAMTS3 in vitro, but are needed for its full lymphangiogenic activity in vivo.
A Hennekam syndrome mutation within the collagen domain affects the function of CCBE1 stronger than a Hennekam syndrome mutation within the EGF domains.
CCBE1 is a secreted protein, necessary for lymphangiogenesis in fish and in mice. In humans, CCBE1 is associated with the Hennekam syndrome, characterized by lymphatic abnormalities. The causative mutations in CCBE1 are mostly located in the N-terminal part of the protein. CCBE1 exerts its function via activation of proteolytic processing of the main lymphatic growth factor VEGFC via the protease ADAMTS3. To date, it is not known which parts of the protein are necessary for its function. We show here that the collagen domains of CCBE1 are crucial for its lymphangiogenic activity in vivo as well as for the activation of VEGFC via ADAMTS3. The EGF domains are necessary for its full lymphangiogenic activity in vivo, but are dispensible for its ability to induce VEGFC processing in vitro. The crucial role of the collagen domains possibly explains why few Hennekam mutations are found in the collagen domains. This study furthers our understanding of how CCBE1 orchestrates lymphangiogenesis and has implications for new pro- or anti lymphangiogenic therapeutic approaches.