Cbln1 is essential for interaction-dependent secretion of Cbln3

Abstract

Cbln1 and the orphan glutamate receptor GluRdelta2 are pre- and postsynaptic components, respectively, of a novel transneuronal signaling pathway regulating synapse structure and function. We show here that Cbln1 is secreted from cerebellar granule cells in complex with a related protein, Cbln3. However, cbln1- and cbln3-null mice have different phenotypes and cbln1 cbln3 double-null mice have deficits identical to those of cbln1 knockout mice. The basis for these discordant phenotypes is that Cbln1 and Cbln3 reciprocally regulate each other's degradation and secretion such that cbln1-null mice lack both Cbln1 and Cbln3, whereas cbln3-null mice lack Cbln3 but have an approximately sixfold increase in Cbln1. Unlike Cbln1, Cbln3 cannot form homomeric complexes and is secreted only when bound to Cbln1. Structural modeling and mutation analysis reveal that, by constituting a steric clash that is masked upon binding Cbln1 in a "hide-and-run" mechanism of endoplasmic reticulum retention, a single arginine confers the unique properties of Cbln3.

FIG. 1.

Cbln3 is a secreted glycoprotein that participates in heteromeric complexes with Cbln1. (A) Medium or cell lysate of primary cultures of cerebellar neurons was immunoblotted with an anti-Cbln3 antiserum. (B) Cerebellar extracts were treated with (+) or without (−) N-glycosidase F and immunoblotted with anti-Cbln3 antiserum. Molecular mass markers are shown to the right of the panel. (C) Cerebellar extracts were immunoblotted with anti-Cbln1 and anti-Cbln3 antiserum. The same cerebellar extracts were also immunoprecipitated (IP) with either rabbit immunoglobulin G (Rb-IgG) or an anti-Cbln3 antiserum, and both were immunoblotted with anti-Cbln1 antiserum. Molecular mass markers are shown at right of the panel. The position of Cbln1 is shown.

FIG. 2.

Generation and characterization of cbln3-null mice. (A) The gene targeting strategy. The cbln3 targeting vector used to generate cbln3^−/− embryonic stem cells is shown as well as the disrupted cbln3 allele. The positions of external probe A and internal probe B are shown. Restriction sites are shown as follows: A, ApaI; C, ClaI; N, NotI; X, XhoI; and Xb, XbaI. Technical details are provided in Materials and Methods. (B) DNA isolated from wild-type (WT) and heterozygous founder (cbln3^+/−) mouse tails was digested with XhoI and ClaI and analyzed by Southern blotting. The wild-type and targeted alleles gave rise to bands of 13.3 kb (wild type) and 4.2 kb (targeted), respectively. (C) Total RNA isolated from cerebella of wild-type, cbln3^+/−, and cbln3^−/− mice was analyzed by Northern blotting using a probe for cbln3. cbln3^−/− mice show loss of cbln3 mRNA, and cbln3^+/− mice have intermediate levels of mRNA on Northern blots. (D) Cerebellar extracts of wild-type, cbln3^+/− and cbln3^−/− mice were immunoblotted with an anti-Cbln3 antiserum. Intermediate levels and complete loss of Cbln1 were evident in cbln3^+/− and cbln3^−/− mice, respectively. (E) Neuroanatomical analysis of cbln3^−/− mice. Cresyl violet-stained sagittal sections of 1-month-old wild-type and cbln3^−/− mice show no overt neuroanatomical anomalies. PEP-19 immunostaining of Purkinje cells in wild-type and cbln3^−/− (lower panels) mice also revealed no overt differences in the positioning, number, and overall dendritic morphology between the two genotypes. Scale bars: upper panel, 1 mm; middle and lower panels, 50 μm. (F) Rota-rod test of cbln1^−/− and cbln3^−/− mice. Performance of wild-type (n = 6), cbln1^−/− (n = 7), and cbln3^−/− (n = 8) mice was assessed on a standardized accelerating Rota-rod. Motor performance was scored as the mean latency to fall (min) on the accelerating rod. In contrast to the low performance of cbln1^−/− mice (P = 0.00006), cbln3^−/− mice showed a level of performance similar to that of wild-type mice. Error bars represent standard error of the means.

FIG. 3.

Inverse relationship between the levels of Cbln1 and Cbln3. (A) Cerebellar RNA isolated from wild-type, cbln1^−/−, and cbln3^−/− mice was assessed by Northern blotting using cbln1, cbln3, and β-actin (loading control) probes. No overt compensatory responses were observed for either gene in the knockout strains. (B) Cerebellar extracts of wild-type, cbln1^−/−, and cbln3^−/− mice were immunoblotted with antisera to Cbln1, Cbln3, and β-actin (loading control). Note the marked changes in levels of Cbln1 and Cbln3. (C and D) Western blot (C) and quantification (D) of Cbln1 and Cbln3 protein levels in wild-type, cbln1^+/−, and cbln1^−/− cerebella (n = 3). Cbln3 levels are markedly reduced in cbln1^+/− (P = 0.026) and cbln1^−/− (P = 0.0009) mice. Error bars represent standard error of the means. (E and F) Western blot (E) and quantification of Cbln1 and Cbln3 protein levels (F) in wild-type, cbln3^+/−, and cbln3^−/− cerebella (n = 3). Cbln1 levels are significantly increased in cbln3^+/− (P = 0.008) and cbln3^−/− (P = 0.0001) cerebella. Error bars represent SEM.

FIG. 4.

Cbln1 is required for ER export and secretion of Cbln3. (A) Cerebellar extracts were subjected to deglycosylation in the presence (+) of endo-H or N-glycosidase F (PNGase F) for 4 h. Subsequently, cerebellar extracts from wild-type (50 μg), cbln1^−/− (100 μg), and cbln3^−/− (20 μg) mice were immunoblotted with anti-Cbln1 and anti-Cbln3 antisera, respectively. The asterisk indicates the presence of endo-H-sensitive Cbln3 exclusively in the cbln1-null mouse cerebellum. (B) After 21 days in vitro, medium and cell lysates from primary dissociated cerebellar cultures from wild-type, cbln1^−/−, and cbln3^−/− mice were immunoblotted with an anti-Cbln3 antiserum. (C) COS-7 cells were transfected with vectors encoding Cbln1 and Cbln3 singly or in combination. Culture medium and cell lysates were immunoblotted with anti-Cbln1 or anti-Cbln3 antiserum. −, absence of; +, presence of. (D) COS-7 cells were transfected with vectors encoding Flag-tagged Cbln1 (Flag-Cbln1) and V5-tagged Cbln3 (Cbln3-V5). Medium was immunoprecipitated with anti-Flag (IP Flag) or anti-V5 (IP V5) antibodies, then immunoblotted with anti-Flag (IB Flag) or anti-V5 (IB V5) antibodies. −, absence of; +, presence of. (E) COS-7 cells were transfected with vectors encoding Cbln1 and Cbln3. At 24 h after transfection, cells were incubated for 8 h in the presence of dimethyl sulfoxide as a vehicle control (Con), the lysosome inhibitor 100 μM chloroquine (CQ), or a proteasome inhibitor, 10 μM lactacystin (LA), 10 μM MG132 (MG), or ZAL. Medium and cell lysates were immunoblotted with anti-Cbln1 or anti-Cbln3 antisera.

FIG. 5.

Cbln4 rescues secretion of Cbln3. (A) LexA and VP16 fusion constructs of Cbln1, Cbln2, Cbln3, and Cbln4 were expressed in yeast. Protein-protein interactions between all possible binary combinations of proteins were tested and scored. Methodological details of yeast two-hybrid assays are provided in Materials and Methods. ++, colonies turned blue within 1 h; −, colonies were negative at 2 h. (B) COS-7 cells were singly or doubly transfected with vectors expressing Cbln1, V5-tagged Cbln2 (Cbln2-V5), Cbln3, and Cbln4 (Cbln4-V5) as shown. Medium and cell lysates were immunoblotted with anti-Cbln3 and anti-V5 antibodies. −, absence of; +, presence of.

FIG. 6.

The C1q domains of Cbln1 and Cbln3 are essential for heteromeric interactions. Schematic diagrams of the truncation mutants of Cbln1 (A) and Cbln3 (B) used in yeast two-hybrid analysis are shown. The signal sequence is shown in blue, two cysteines are red, and the C1q domain is orange. The location of cerebellin, a naturally occurring peptide fragment of Cbln1, is shown by a black box. The various Mt forms of Cbln1 and Cbln3 were assayed in combination with full-length proteins as well as with themselves. All constructs scored as negative when expressed alone (data not shown). ++, colonies turned blue within 1 h; +, colonies turned blue by 2 h; ±, marginal coloration at 2 h; −, no coloration at 2 h.

FIG. 7.

Arg119 determines the inability of Cbln3 to undergo self-assembly and secretion. (A) The amino acid sequences of the C β strands of Cbln1 to Cbln4 are aligned as shown (left panel). Residues in red are identical, those in blue are conservative substitutions, and dissimilar residues are black. Note the critical arginine residue (italicized) in Cbln3. A trimeric structural model of the C1q globular domain of Cbln3 was built using the Swiss-PdbViewer. Top and side views of predicted Cbln3 trimers are shown in the right and middle panels. The C strands are white, and the arrows point to Arg119 of Cbln3. (B) Structural models of homo- and heterotrimeric complexes of Cbln1 and Cbln3. Left panel, homomeric Cbln1 model; middle panel, homomeric Cbln3 model; right panel, heteromeric Cbln1 and Cbln3 model. Cbln1 subunits are red, and Cbln3 subunits are green. The arrows point to Asn115 in Cbln1 and Arg119 in Cbln3. (C) LexA and VP16 fusions of Cbln1, Cbln3, and Cbln3(R119A) were expressed in yeast for interaction analyses. Protein-protein interactions were scored as described for Fig. 5 and 6. Dotted circles indicate the altered properties of Cbln3R119A compared to those of Cbln3. (D) COS-7 cells were transfected with combinations of vectors expressing Cbln1, Cbln3, and Cbln3(R119A). Medium and cell lysate were immunoblotted with anti-Cbln1 and anti-Cbln3 antibodies. Note the secretion of Cbln3R119A (R119A) in the absence of Cbln1 (left panel).