Identification of four novel DC-SIGN ligands on Mycobacterium bovis BCG

Dendritic-cell-specific intercellular adhesion molecule-3-grabbing non-integrin (DC-SIGN; CD209) has an important role in mediating adherence of Mycobacteria species, including M. tuberculosis and M. bovis BCG to human dendritic cells and macrophages, in which these bacteria can survive intracellularly. DC-SIGN is a C-type lectin, and interactions with mycobacterial cells are believed to occur via mannosylated structures on the mycobacterial surface. Recent studies suggest more varied modes of binding to multiple mycobacterial ligands. Here we identify, by affinity chromatography and mass-spectrometry, four novel ligands of M. bovis BCG that bind to DC-SIGN. The novel ligands are chaperone protein DnaK, 60 kDa chaperonin-1 (Cpn60.1), glyceraldehyde-3 phosphate dehydrogenase (GAPDH) and lipoprotein lprG. Other published work strongly suggests that these are on the cell surface. Of these ligands, lprG appears to bind DC-SIGN via typical protein-glycan interactions, but DnaK and Cpn60.1 binding do not show evidence of carbohydrate-dependent interactions. LprG was also identified as a ligand for DC-SIGNR (L-SIGN; CD299) and the M. tuberculosis orthologue of lprG has been found previously to interact with human toll-like receptor 2. Collectively, these findings offer new targets for combating mycobacterial adhesion and within-host survival, and reinforce the role of DC-SIGN as an important host ligand in mycobacterial infection.     

Lysines 128 and 132 enable lipopolysaccharide binding to MD-2, leading to Toll-like receptor-4 aggregation and signal transduction

Three cell-surface proteins have been recognized as components of the mammalian signaling receptor for bacterial lipopolysaccharide (LPS): CD14, Toll-like receptor-4 (TLR4), and MD-2. Biochemical and visual studies shown here demonstrate that the role of CD14 in signal transduction is to enhance LPS binding to MD-2, although its expression is not essential for cellular activation. These studies clarify how MD-2 functions: we found that MD-2 enables TLR4 binding to LPS and allows the formation of stable receptor complexes. MD-2 must be bound to TLR4 on the cell surface before binding can occur. Consequently, TLR4 clusters into receptosomes (many of which are massive) that recruit intracellular toll/IL-1/resistance domain-containing adapter proteins within minutes, thus initiating signal transduction. TLR4 activation correlates with the ability of MD-2 to bind LPS, as MD-2 mutants that still bind TLR4, but are impaired in the ability to bind LPS, conferred a greatly blunted LPS response. These findings help clarify the earliest events of TLR4 triggering by LPS and identify MD-2 as an attractive target for pharmacological intervention in endotoxin-mediated diseases.     

Rhizobium leguminosarum chaperonin 60.3, but not chaperonin 60.1, induces cytokine production by human monocytes: activity is dependent on interaction with cell surface CD14

As part of a program of work to understand the interaction of bacterial chaperonins with human leukocytes, we have examined 2 of the 3 chaperonin 60 (Cpn 60) gene products of the nonpathogenic plant symbiotic bacterium, Rhizobium leguminosarum, for their capacity to induce the production of pro- and antiinflammatory cytokines by human cells. Recombinant R. leguminosarum Cpn 60.1 and 60.3 proteins were added to human monocytes at a range of concentrations, and cytokine production was measured by sandwich enzyme-linked immunosorbent assay. In spite of the fact that the 2 R. leguminosarum Cpn 60 proteins share 74.5% amino acid sequence identity, it was found that Cpn 60.3 induced the production of interleukin (IL)-1beta, tumor necrosis factor alpha, IL-6, IL-8, IL-10, and IL-12, but not IL-4, interferon gamma, or GM-CSF (granulocyte-macrophage colony-stimulating factor), whereas the Cpn 60.1 protein failed to demonstrate any cytokine-inducing activity. The use of neutralizing monoclonal antibodies showed that the cytokine-inducing activity of Cpn 60.3 was dependent on its interaction with CD14. This demonstrates that CD14 mediates not only lipopolysaccharide but also R. leguminosarum Cpn 60.3 cell signaling in human monocytes.     

The unusual mycobacterial chaperonins: evidence for in vivo oligomerization and specialization of function

The pathogen Mycobacterium tuberculosis expresses two chaperonins, one (Cpn60.1) dispensable and one (Cpn60.2) essential. These proteins have been reported not to form oligomers despite the fact that oligomerization of chaperonins is regarded as essential for their function. We show here that the Cpn60.2 homologue from Mycobacterium smegmatis also fails to oligomerize under standard conditions. However, we also show that the Cpn60.2 proteins from both organisms can replace the essential groEL gene of Escherichia coli, and that they can function with E. coli GroES cochaperonin, as well as with their cognate cochaperonin proteins, strongly implying that they form oligomers in vivo. We show that the Cpn60.1 proteins, but not the Cpn60.2 proteins, can complement for loss of the M. smegmatis cpn60.1 gene. We investigated the oligomerization of the Cpn60.2 proteins using analytical ultracentrifugation and mass spectroscopy. Both form monomers under standard conditions, but they form higher order oligomers in the presence of kosmotropes and ADP or ATP. Under these conditions, their ATPase activity is significantly enhanced. We conclude that the essential mycobacterial chaperonins, while unstable compared to many other bacterial chaperonins, do act as oligomers in vivo, and that there has been specialization of function of the mycobacterial chaperonins following gene duplication.     

JNK-interacting protein 3 associates with Toll-like receptor 4 and is involved in LPS-mediated JNK activation

Lipopolysaccharide (LPS) is recognized by Toll-like receptor (TLR) 4 and activates NF-kappaB and a set of MAP kinases. Here we have investigated proteins associated with the cytoplasmic domain of mouse TLR4 by yeast two-hybrid screening and identified JNK-interacting protein 3 (JIP3), a scaffold protein for JNK, as a TLR4-associated protein. In mammalian cells, JIP3, through its N-terminal region, constitutively associates with TLR4. The association is specific to JIP3, as the two other JIPs, JIP1 and JIP2, failed to bind TLR4. In HEK 293 cells exogenously expressing TLR4, MD2 and CD14, co-expression of JIP3 significantly increased the complex formation of TLR4-JNK and LPS-mediated JNK activation. In contrast, expression of C-terminally truncated forms of JIP3 impaired LPS-induced JNK activation in a mouse macrophage cell line, RAW264.7. Moreover, RNA interference of JIP3 inhibited LPS-mediated JNK activation. In RAW264.7 cells, JIP3 associates MEKK-1, but not with TAK-1. Finally, JIP3 also associates with TLR2 and TLR9, but not with TLR1 or TLR6. Altogether, our data indicate the involvement of JIP3 in JNK activation in downstream signals of some TLRs.     

Mastoparan, a G protein agonist peptide, differentially modulates TLR4- and TLR2-mediated signaling in human endothelial cells and murine macrophages

Previous studies have implicated a role for heterotrimeric G protein-coupled signaling in B cells, monocytes, and macrophages stimulated with LPS and have shown that G proteins coimmunoprecipitate with membrane-bound CD14. In this study, we have extended these observations in human dermal microvessel endothelial cells (HMEC) that lack membrane-bound CD14 and in murine macrophages to define further the role of heterotrimeric G proteins in TLR signaling. Using the wasp venom-derived peptide, mastoparan, to disrupt G protein-coupled signaling, we identified a G protein-dependent signaling pathway in HMEC stimulated with TLR4 agonists that is necessary for the activation of p38 phosphorylation and kinase activity, NF-kappaB and IL-6 transactivation, and IL-6 secretion. In contrast, HMEC activation by TLR2 agonists, TNF-alpha, or IL-1beta was insensitive to mastoparan. In the murine macrophage cell line, RAW 264.7, and in primary murine macrophages, G protein dysregulation by mastoparan resulted in significant inhibition of LPS-induced signaling leading to both MyD88-dependent and MyD88-independent gene expression, while TLR2-mediated gene expression was not significantly inhibited. In addition to inhibition of TLR4-mediated MAPK phosphorylation in macrophages, mastoparan blunted IL-1R-associated kinase-1 kinase activity induced by LPS, but not by TLR2 agonists, yet failed to affect phosphorylation of Akt by phosphoinositol-3-kinase induced by either TLR2- or TLR4-mediated signaling. These data confirm the importance of heterotrimeric G proteins in TLR4-mediated responses in cells that use either soluble or membrane-associated CD14 and reveal a level of TLR and signaling pathway specificity not previously appreciated.     

Homologous cpn60 genes in Rhizobium leguminosarum are not functionally equivalent

Many bacteria possess 2 or more genes for the chaperonin GroEL and the cochaperonin GroES. In particular, rhizobial species often have multiple groEL and groES genes, with a high degree of amino-acid similarity, in their genomes. The Rhizobium leguminosarum strain A34 has 3 complete groE operons, which we have named cpn.1, cpn.2 and cpn.3. Previously we have shown the cpn. 1 operon to be essential for growth, but the two other cpn operons to be dispensable. Here, we have investigated the extent to which loss of the essential GroEL homologue Cpn60.1 can be compensated for by expression of the other two GroEL homologues (Cnp60.2 and Cpn60.3). Cpn60.2 could not be overexpressed to high levels in R. leguminosarum, and was unable to replace Cpn60.1. A strain that overexpressed Cpn60.3 grew in the absence of Cpn60.1, but the complemented strain displayed a temperature-sensitive phenotype. Cpn60.1 and Cpn60.3, when coexpressed in Escherichia coli, preferentially selfassembled rather than forming mixed heteroligomers. We conclude that, despite their high amino acid similarity, the GroEL homologues of R. leguminosarum are not functionally equivalent in vivo.     

The dimeric structure of the Cpn60.2 chaperonin of Mycobacterium tuberculosis at 2.8 Å reveals possible modes of function

Mycobacterium tuberculosis expresses two proteins (Cpn60.1 and Cpn60.2) that belong to the chaperonin (Cpn) family of heat shock proteins. Studies have shown that the two proteins have different functional roles in the bacterial life cycle and that Cpn60.2 is essential for cell viability and may be involved in M. tuberculosis pathogenicity. Cpn60.2 does not form a tetradecameric double ring, which is typical of other Cpns. We have determined the crystal structure of recombinant Cpn60.2 to 2.8 Å resolution by molecular replacement; the asymmetric unit (AU) contains a dimer, which is consistent with size-exclusion high-performance liquid chromatography and dynamic light-scattering measurements of the soluble recombinant protein. However, we suggest that the actual Cpn60.2 dimer may be different from that identified within the AU on the basis of surface contact stability, solvation free-energy gain, and functional aspects. Unlike the dimer found in the AU, which is formed through apical domain interactions, the dimeric form we propose here provides a free apical domain that is required for normal chaperone activity and may be involved in M. tuberculosis association with macrophages and arthrosclerosis plaque formation. Here we describe in detail the structural aspects that lead to Cpn60.2 dimer formation and prevent the formation of heptameric rings and tetradecameric double rings.     

The intercellular signaling activity of the Mycobacterium tuberculosis chaperonin 60.1 protein resides in the equatorial domain

The major heat shock protein, chaperonin 60, has been established to have intercellular signaling activity in addition to its established protein-folding function. Mycobacterium tuberculosis is one of a small proportion of bacteria to encode two chaperonin 60 proteins. We have demonstrated that chaperonin 60.1 from this bacterium is a very active stimulator of human monocytes. To determine structure/function relationships of chaperonin 60.1 we have cloned and expressed the apical, equatorial, and intermediate domains of this protein. We have found that the signaling activity of M. tuberculosis chaperonin 60.1 resides in the equatorial domain. This activity of the recombinant equatorial domain was completely blocked by treating the protein with proteinase K, ruling out lipopolysaccharide contamination as the cause of the cell activation. Blockade of the activity of the equatorial domain by anti-CD14 monoclonal antibodies reveals that this domain activates monocytes by binding to CD14. Looking at the oligomeric state of the active proteins, using native gel electrophoresis and protein cross-linking we found that recombinant M. tuberculosis chaperonin 60.1 fails to form the prototypic tetradecameric structure of chaperonin 60 proteins under the conditions tested and only forms dimers. It is therefore concluded that the monocyte-stimulating activity of M. tuberculosis Cpn60.1 resides in the monomeric subunit and within this subunit the biological activity is due to the equatorial domain.     

The receptor for heat shock protein 60 on macrophages is saturable, specific, and distinct from receptors for other heat shock proteins.

Previous studies have shown that human heat shock protein (hsp) 60 elicits a strong proinflammatory response in cells of the innate immune system with CD14, Toll-like receptor (TLR) 2, and TLR4 as mediators of signaling, but probably not of binding. In the present study, we directly demonstrate binding of hsp60 to the macrophage surface and find the binding receptor for hsp60 different from the previously described common receptor for several other heat shock proteins, including hsp70, hsp90, and gp96. Fluorescence-labeled human hsp60 bound to cell surfaces of the murine macrophage lines J774 A.1 and RAW264.7 and to mouse bone marrow-derived macrophages. By flow cytometry, we could demonstrate for the first time that hsp60 binding to macrophages occurred at submicromolar concentrations, is saturable, and can be competed by unlabeled hsp60, but not by unrelated proteins, thus confirming the classic characteristics of specific ligand-receptor interactions. Binding of hsp60 at 4 degrees C was followed by endocytosis at 37 degrees C. Hsp60 binding to macrophages could not be competed by excess hsp70, hsp90, or gp96, all of which share the alpha(2)-macroglobulin receptor as binding site. Hsp60 binding occurred in the absence of surface TLR4. However, no cytokine response was induced by hsp60 in TLR4-deficient macrophages. We conclude that hsp60 binds to a stereo-specific receptor on macrophages, and that different surface molecules are engaged in binding and signal transduction. Furthermore, the binding site for hsp60 is separate from the common receptor for hsp70, hsp90, and gp96, which suggests an independent role of hsp60 as danger Ag and in immunoregulation.     

The CD14 ligands lipoarabinomannan and lipopolysaccharide differ in their requirement for Toll-like receptors

Mammalian Toll-like receptor (TLR) proteins are new members of the IL-1 receptor family that participate in activation of cells by bacteria and bacterial products. Several recent reports indicate that TLR proteins mediate cellular activation by bacterial LPS via a signaling pathway that is largely shared by the type I IL-1 receptor. We previously showed that Chinese hamster ovary (CHO) fibroblasts engineered to express CD14 (CHO/CD14) were responsive to LPS, but not to a distinct CD14 ligand, mycobacterial lipoarabinomannan (LAM). These CHO/CD14 cells were subsequently found to possess a frame-shift mutation within the TLR2 gene which resulted in their inability to express functional TLR2 protein. Thus, we hypothesized that TLR2, but not TLR4, was necessary for LAM signaling. In this paper we show that CHO/CD14 cells engineered to express functional TLR2 protein acquired the ability to be activated by LAM. Similarly, overexpression of TLR2 in murine macrophages conferred enhanced LAM responsiveness. Together, our data demonstrate that the distinct CD14 ligands LAM and LPS utilize different TLR proteins to initiate intracellular signals. These findings suggest a novel receptor signaling paradigm in which the binding of distinct ligands is mediated by a common receptor chain, but cellular activation is initiated via distinct signal-transducing chains that confer ligand specificity. This paradigm contrasts with many cytokine receptor complexes in which receptor specificity is conferred by a unique ligand-binding chain but cellular activation is initiated via shared signal-transducing chains.     

Mapping the major interaction between binding protein and Ig light chains to sites within the variable domain.

Newly synthesized Ig chains are known to interact in vivo with the binding protein (BiP), a major peptide-binding chaperone in the endoplasmic reticulum. The predominant interactions between the light chain and BiP are observed early in the folding pathway, when the light chain is either completely reduced, or has only one disulfide bond. In this study, we describe the in vitro reconstitution of BiP binding to the variable domain of light chains (VL). Binding of deliberately unfolded VL was dramatically more avid than that of folded VL, mimicking the interaction in vivo. Furthermore, VL binding was inhibited by addition of ATP, was competed with excess unlabeled VL, and was demonstrated with several different VL proteins. Using this assay, peptides derived from the VL sequence were tested experimentally for their ability to bind BiP. Four peptides from both beta sheets of VL were shown to bind BiP specifically, two with significantly higher affinity. As few as these two peptide sites, one from each beta sheet of VL, are sufficient to explain the association of BiP with the entire light chain. These results suggest how BiP directs the folding of Ig in vivo and how it may be used in shaping the B cell repertoire.     

Identification of a 14-kDa laminin binding protein (HLBP14) in human melanoma cells that is identical to the 14-kDa galactoside binding lectin.

The carbohydrate moieties present on laminin play a crucial role in the multiple biological activities of this basement membrane glycoprotein. We report the identification of a human laminin binding protein with an apparent molecular mass of 14 kDa on sodium dodecyl sulfate-polyacrylamide gels that was found, after purification and amino acid microsequencing, to be identical to the previously described 14-kDa galactoside binding soluble L-14 lectin. We have designated this human laminin binding protein as HLBP14. HLBP14 was purified from human melanoma cells in culture by laminin affinity chromatography and gel electroelution. We demonstrate that HLBP14 binds specifically to the poly-N-acetyllactosamine residues of murine laminin and does not bind to other glycoproteins that do not contain such structures, such as fibronectin. HLBP14 was eluted from a murine laminin column by lactose, N-acetyllactosamine, and galactose but not by other control saccharides, including glucose, fucose, mannose, and melibiose. It did not bind to laminin treated with endo-beta-galactosidase. Lactose also eluted HLBP14 off a human laminin affinity column, implying that human laminin also contains poly-N-acetyllactosamine residues. On immunoblots, polyclonal antibodies raised against HLBP14 recognized HLBP14 as well as 31- and 67-kDa molecules that are also laminin binding proteins, indicating that these proteins share common epitopes. L-14, a dimeric lactose binding lectin, is expressed in a wide variety of tissues. Although the expression of this molecule has been linked to a variety of biological events, the elucidation of its specific functions has been elusive. The observation that HLBP14, a human cancer cell laminin binding protein, is identical to L-14 strongly suggests that the functions attributed to this lectin could be mediated, at least in part, through its ability to interact with the poly-N-acetyllactosamine residues of laminin. HLBP14 could potentially play a role during tumor invasion and metastasis by modulating the interactions between cancer cells and laminin.     

Fibronectin inhibits cytokine production induced by CpG DNA in macrophages without direct binding to DNA

Fibronectin (FN) is known to have four DNA-binding domains although their physiological significance is unknown. Primary murine peritoneal macrophages have been shown to exhibit markedly lower responsiveness to CpG motif-replete plasmid DNA (pDNA), Toll-like receptor-9 (TLR9) ligand, compared with murine macrophage-like cell lines. The present study was conducted to examine whether FN having DNA-binding domains is involved in this phenomenon. The expression of FN was significantly higher in primary macrophages than in a macrophage-like cell line, RAW264.7, suggesting that abundant FN might suppress the responsiveness in the primary macrophages. However, electrophoretic analysis revealed that FN did not bind to pDNA in the presence of a physiological concentration of divalent cations. Surprisingly, marked tumor necrosis factor - (TNF-)α production from murine macrophages upon CpG DNA stimulation was significantly reduced by exogenously added FN in a concentration-dependent manner but not by BSA, laminin or collagen. FN did not affect apparent pDNA uptake by the cells. Moreover, FN reduced TNF-α production induced by polyI:C (TLR3 ligand), and imiquimod (TLR7 ligand), but not by LPS (TLR4 ligand), or a non-CpG pDNA/cationic liposome complex. The confocal microscopic study showed that pDNA was co-localized with FN in the same intracellular compartment in RAW264.7, suggesting that FN inhibits cytokine signal transduction in the endosomal/lysosomal compartment. Taken together, the results of the present study has revealed, for the first time, a novel effect of FN whereby the glycoprotein modulates cytokine signal transduction via CpG-DNA/TLR9 interaction in macrophages without direct binding to DNA through its putative DNA-binding domains.     

Expression and subcellular localization of cpn60 protein family members in Leishmania donovani

We have identified two diverged members of the cpn60 gene family in Leishmania donovani, causative agent of Indian Kala Azar. One of the genes, cpn60.1, although actively transcribed, is not expressed to detectable levels of protein in cultured L. donovani. The other gene, cpn60.2, which, compared with cpn60.1, shows a higher sequence conservation with the hsp60 genes from Trypanosoma brucei and Trypanosoma cruzi is expressed constitutively in cultured promastigotes. The abundance of the gene product, Cpn60.2, increases by 2.5-fold under heat stress and in axenic amastigotes of L. donovani. Cpn60.2 is also found enriched in mitochondrial cell fractions and localizes to the mitochondrial matrix. We conclude that Cpn60.2 is the major mitochondrial chaperonin in Leishmania.     

Comprehensive proteomic analysis of interphase and mitotic 14-3-3-binding proteins

14-3-3 proteins regulate the cell division cycle and play a pivotal role in blocking cell cycle advancement after activation of the DNA replication and DNA damage checkpoints. Here we describe a global proteomics analysis to identify proteins that bind to 14-3-3s during interphase and mitosis. 14-3-3-binding proteins were purified from extracts of interphase and mitotic HeLa cells using specific peptide elution from 14-3-3 zeta affinity columns. Proteins that specifically bound and eluted from the affinity columns were identified by microcapillary high pressure liquid chromatography tandem mass spectrometry analysis. Several known and novel 14-3-3-interacting proteins were identified in this screen. Identified proteins are involved in cell cycle regulation, signaling, metabolism, protein synthesis, nucleic acid binding, chromatin structure, protein folding, proteolysis, nucleolar function, and nuclear transport as well as several other cellular processes. In some cases 14-3-3 binding was cell cycle-dependent, whereas in other cases the binding was shown to be cell cycle-independent. This study adds to the growing list of human 14-3-3-binding proteins and implicates a role for 14-3-3 proteins in a plethora of essential biological processes.     

Differential effects of a Toll-like receptor antagonist on Mycobacterium tuberculosis-induced macrophage responses

We previously showed that viable Mycobacterium tuberculosis (Mtb) bacilli contain distinct ligands that activate cells via the mammalian Toll-like receptor (TLR) proteins TLR2 and TLR4. We now demonstrate that expression of a dominant negative TLR2 or TLR4 proteins in RAW 264.7 macrophages partially blocked Mtb-induced NF-kappa B activation. Coexpression of both dominant negative proteins blocked virtually all Mtb-induced NF-kappa B activation. The role of the TLR4 coreceptor MD-2 was also examined. Unlike LPS, Mtb-induced macrophage activation was not augmented by overexpression of ectopic MD-2. Moreover, cells expressing an LPS-unresponsive MD-2 mutant responded normally to Mtb. We also observed that the lipid A-like antagonist E5531 specifically inhibited TLR4-dependent Mtb-induced cellular responses. E5531 could substantially block LPS- and Mtb-induced TNF-alpha production in both RAW 264.7 cells and primary human alveolar macrophages (AM phi). E5531 inhibited Mtb-induced AM phi apoptosis in vitro, an effect that was a consequence of the inhibition of TNF-alpha production by E5531. In contrast, E5531 did not inhibit Mtb-induced NO production in RAW 264.7 cells and AM phi. Mtb-stimulated peritoneal macrophages from TLR2- and TLR4-deficient animals produced similar amounts of NO compared with control animals, demonstrating that these TLR proteins are not required for Mtb-induced NO production. Lastly, we demonstrated that a dominant negative MyD88 mutant could block Mtb-induced activation of the TNF-alpha promoter, but not the inducible NO synthase promoter, in murine macrophages. Together, these data suggest that Mtb-induced TNF-alpha production is largely dependent on TLR signaling. In contrast, Mtb-induced NO production may be either TLR independent or mediated by TLR proteins in a MyD88-independent manner.     

Retrieval of HDEL proteins is required for growth of yeast cells

The ERD2 gene of Saccharomyces cerevisiae encodes the receptor which retrieves HDEL-containing containing ER proteins from the Golgi apparatus. Viable erd2 mutants have been isolated that show no obvious HDEL-dependent retention of the luminal ER protein BiP, suggesting that retrieval of HDEL proteins is not essential for growth. However, cells that lack Erd2p completely have a defective Golgi apparatus and cannot grow. This observation led to the suggestion that the receptor had a second function, possibly related to its ability to recycle from Golgi to ER. In this paper we investigate the requirements for Erd2p to support growth. We show that mutations that block its recycling also prevent growth. In addition, we show that all mutant receptors that can support growth have a residual ability to retrieve BiP, which is detectable when they are overexpressed. Mere recycling of an inactive form of the receptor, mediated by a cytoplasmic KKXX sequence, is not sufficient for growth. Furthermore, saturation of the receptor by expression of an HDEL-tagged version of pro-alpha factor inhibits growth, even of strains that do not show obvious BiP retention. We conclude that growth requires the HDEL-dependent retrieval of one or more proteins, and that these proteins can be recognized even under conditions where BiP is secreted. Genetic screens have failed to identify any one protein whose loss could account for the Erd2p requirement. Therefore, a growth may require the retention of multiple HDEL proteins in the ER, or alternatively the removal of such proteins from the Golgi apparatus.