ERdj3, a luminal ER DnaJ homologue, binds directly to unfolded proteins in the mammalian ER: identification of critical residues

Abstract

ERdj3 was identified as a soluble, lumenal DnaJ family member that binds to unassembled immunoglobulin heavy chains along with the BiP chaperone complex in the endoplasmic reticulum of mammalian cells. Here we demonstrated that ERdj3 binds directly to unfolded substrates. Secondary structure predictions suggested that the substrate binding domain of ERdj3 was likely to closely resemble Ydj1, a yeast cytosolic DnaJ family member, which was previously crystallized with a peptide bound to the C-terminal fragment composed of domains I, II, and III. Mutation of conserved residues in domain I, which formed the peptide binding site in Ydj1, affected ERdj3's substrate binding ability in mammalian cells and in vitro binding studies. Somewhat unexpectedly, we found that domain II, which is highly conserved among ERdj3 homologues, but very different from domain II of Ydj1, was also critical for substrate binding. In addition, we demonstrated that ERdj3 forms multimers in cells and found that the conserved carboxy-terminal residue phenylalanine 326 played a critical role in self-assembly. In vitro binding assays revealed that mutation of this residue to alanine diminished ERdj3's substrate binding ability, arguing that multimerization is important for substrate binding. Together, these studies demonstrate that the Ydj1 structure is conserved in another family member and reveal that among this group of DnaJ proteins domain II, which is not present in the closely related type II family members, also plays an essential role in substrate binding.

Figure 1. Comparison of ERdj3 with Ydj1 and Sis1

(A) Secondary structure predictions for ERdj3, Sis1 and Ydj1. Colors represent the different domains: Light grey: ER targeting sequence, grey: domain J, bold and italics: domain I, bold and underlined: domain II, black: domain III, H: α helix, and E: β sheet. (B) The schematic representation domain structures of Ydj1, Sis1 and ERdj3 showing that Sis1does not contain a cysteine-rich domain II and that ERdj3 is more similar to Ydj1, although it contains an atypical, smaller domain II. (C) Domain II sequences of ERdj3 from various species were aligned and compared to domain II of Ydj1. Clear boxes indicate amino acid identities, whereas grey shading indicates amino acid similarity. Cysteine residues (CysXXCys motif) are underlined.

Figure 2. Binding of wild-type and mutant ERdj3 to γHC

(A) Schematic representation of the domain composition of various ERdj3 mutants. (B) COS cells were transfected with the indicated ERdj3 constructs in the HA-DSL vector, metabolically labeled and immunoprecipitated with anti-ERdj3 or Protein A Sepharose alone. (C) COS cells were cotransfected with γHC and either wild-type ERdj3 or the indicated ERdj3 mutants. Cells were metabolically labeled, treated with DSP before lysing and immunoprecipitated with Protein A Sepharose to isolate γHC or with a monoclonal anti-HA antibody to isolate ERdj3 proteins. Samples were analyzed by reducing SDS-PAGE, and mutants that show diminished binding are indicated with arrows.

Figure 3. Both domain I and domain II of ERdj3 contribute to its ability to bind to denatured luciferase

(A) Model of ERdj3’s putative substrate binding domain (left) shows amino acid residues that could form a structure similar to that found on Ydj1. An overlap of the peptide binding site of Ydj1 with the model of the corresponding region of ERdj3 is shown on right. Grey: substrate peptide, green: Ydj1, pink: ERdj3 model, red: peptide interacting residues on Ydj1, and blue: corresponding residues on ERdj3. (B) Measurement of complex formation between wild-type ERdj3 and the ΔIa and ΔII +GSGG mutants and denatured luciferase was performed as described in the Materials and Methods section. The quantity of luciferase bound to wild-type ERdj3 was set to 100%, and the values for the various mutants were expressed as a percent of this value. Standard errors are indicated. (C) Measurement of complex formation between the indicated ERdj3 point mutants and luciferase.

Figure 4. ERdj3 forms dimers in cells

(A) COS cells were transfected with Cdna encoding untagged wild-type ERdj3 (lanes1–3), 3×HA-ERdj3 (lanes 4–6), untagged ERdj3 F326A (lanes 7–9) and F326D (lanes 13–15), HA-ERdj3 F326A (lanes10–12) or HA-ERdj3 F326D (lane 16–18) alone. (B) COS cells were cotransfected with ×HAERdj3 and untagged ERdj3 (lanes 4–6), with 3×HA-ERdj3 and untagged F326A (lanes 7–9), with 3×HA-ERdj3 and untagged F326D (lane 10–12), with untagged F326A and HA-F326A (lane 13–15), or with untagged F326D and HA-F326D (lane 16–18). The cell lysates were divided to three parts equally and incubated with indicated antibodies. Immune complexes were precipitated with Protein A Sepharose, and samples were analyzed by reducing SDS-PAGE.

Figure 5. Deletion of domain III or mutation of phenylalanine 326 to aspartic acid affects ERdj3’s substrate binding ability

Plates were coated with he indicated ERdj3 proteins and denatured luciferase was allowed to bind. ELISA assays were performed and quantified as described in Figure 3B.