Calreticulins are not all the same.

S ecretory proteins fold and assemble in the endoplasmic reticulum (ER), under the assistance of many folding helpers that also retain newly synthesized polypeptides in this compartment: only upon correct structural maturation are proteins allowed to traffic along the secretory pathway to their destinations. Structurally defective polypeptides that fail correct folding interact more extensively with folding helpers, are retained in the ER, and are often targeted for degradation. Collectively, these events are termed ER quality control (ERQC) (1). A study in this issue of PNAS (2) shows that only one of the three forms of the ER folding helper calreticulin (CRT) of the plant Arabidopsis thaliana interacts with a mutated form of BRI1, the plasma membrane leucine-rich-repeat kinase receptor for brassinosteroids, plant-specific hormones playing important roles in plant growth (3). Arabidopsis CRT1 and CRT2 are very similar and have homologs in nonplant organisms, but the BRI1-interacting CRT3 seems to be a plant-specific form, with orthologs in higher and lower plant species (4). The important observations by Jin et al. (2) point to plant-specific functional divergence in the CRT family. CRT is involved in Ca2 storage and the ERQC of N-glycosylated proteins (5). Many secretory proteins are cotranslationally N-glycosylated to specific Asn residues by the addition of the oligosaccharide Glc3Man9GlNAc2. Soon after this event, the three Glc residues, which are linearly linked to each other, are sequentially removed by ER glucosidases I (outermost residue) and II (middle and innermost residue). The processing intermediate Glc1Man9GlNAc2 is recognized and bound by the soluble lectin CRT or its close relative calnexin, a membrane protein (5). Binding by these ER residents retains intermediates of glycoprotein folding in the ER, prevents unspecific aggregation, and favors the formation of correct disulfide bonds by the action of Erp57, which is in a complex with the lectins (1, 5). Removal of the inmost Glc residue by glucosidase II liberates the newly synthesized glycoprotein, which can, however, be monoglucosylated again by another ER resident, UDPglucose-glycoproteinglucosyltransferase (UGGT). The cycle of lectin binding and release is interrupted upon correct folding, because UGGT only recognizes non-native structures, thus acting as a sensor of productive folding (5). Another ERQC mechanism, active possibly on every secretory protein and therefore more promiscuous, relies on the activity of the binding protein (BiP), an ER-located member of the heat shock 70 family of molecular chaperones (1). BiP associates to short sequences enriched in hydrophobic amino acids, which are exposed on the surface of folding intermediates and misfolded polypeptides but are masked by correct folding and assembly in most proteins. As is the case of CRT, BiP inhibits aggregation and retains proteins in the ER. The two mechanisms often work in concert. Interestingly, CRT is also able to bind unglycosylated polypeptides and prevent their in vitro aggregation, thus acting directly as a chaperone. Also CRT binds sequences enriched in hydrophobic amino acids (6). ERQC is very efficient but it is not an exception to the rule that no one is perfect. It can fail to recognize defective proteins, leading to the accumulation of harmful structures, or be overzealous in recognition when otherwise functioning proteins are misrecognized as defective (1). Both types of failure are the bases of important severe human diseases. Jin et al. (2) have made their exciting discovery during studies that led to the identification of overzealous QC in plants. Some mutated bri1 alleles encode receptors that are functional but fail to reach the cell surface, thus causing reduced growth and other defective phenotypes associated to the lack of brassinosteroid signaling (7, 8). Jin et al. (2) now show that loss-of-function mutations in CRT3 restore the wild-type phenotype in bri1–9 plants, acting as suppressors of the bri1–9 phenotype, but mutations in CRT1/2 have no effect (Fig. 1). bri1–9 is retained in the ER, but in CRT3-defective plants it correctly reaches the plasma membrane in relevant proportions (ref. 2 and Fig. 1). Coimmunoprecipitation assays detected interactions of bri1–9:GFP with CRT3, BiP, and calnexin, but not with CRT1/2 (2, 7). Mutations in CRT3 are not the