Structural features of thyroglobulin linked to protein trafficking

Abstract

Thyroglobulin must pass endoplasmic reticulum (ER) quality control to become secreted for thyroid hormone synthesis. Defective thyroglobulin, blocked in trafficking, can cause hypothyroidism. Thyroglobulin is a large protein (~2750 residues) spanning regions I-II-III plus a C-terminal cholinesterase-like domain. The cholinesterase-like domain functions as an intramolecular chaperone for regions I-II-III, but the folding pathway leading to successful thyroglobulin trafficking remains largely unknown. Here, informed by the recent three-dimensional structure of thyroglobulin as determined by cryo-electron microscopy, we have bioengineered three novel classes of mutants yielding three entirely distinct quality control phenotypes. Specifically, upon expressing recombinant thyroglobulin, we find that first, mutations eliminating a disulfide bond enclosing a 200-amino acid loop in region I have surprisingly little impact on the ability of thyroglobulin to fold to a secretion-competent state. Next, we have identified a mutation on the surface of the cholinesterase-like domain that has no discernible effect on regional folding yet affects contact between distinct regions and thereby triggers impairment in the trafficking of full-length thyroglobulin. Finally, we have probed a conserved disulfide in the cholinesterase-like domain that interferes dramatically with local folding, and this defect then impacts on global folding, blocking the entire thyroglobulin in the ER. These data highlight variants with distinct effects on ER quality control, inhibiting domain-specific folding; folding via regional contact; neither; or both.

Keywords: contact interface; cysteine; disulfide bond; endoplasmic reticulum; hypothyroidism; protein trafficking; regional misfolding.

FIGURE 1

Location of the mutation sites examined in this study, as mapped onto the three‐dimensional structural representation of the homodimeric bovine Tg. Structural representation of bTg (see Section 5): the C408–C608 disulfide bond (blue arrow) faces an interior cleft of the Tg dimer and is near the boundary between two “halves” of region I (shown in green from one monomer of the Tg homodimer). C2444 and C2455 (red arrows) in the ChEL domain (shown in gold from the same monomer of the Tg homodimer) are localized near the surface of the Tg molecule. The K2225 residue (purple arrow) in the ChEL domain is at the contact interface with Tg region I.

FIGURE 2

bTg protein export upon disruption of the C408–C608 disulfide bridge. (a) Schematic of bTg wild‐type (WT) and mutants studied in this figure. (b) 293 T were either nontransfected (NT) or transfected to express the indicated constructs. After a 13‐h period, the media (M) were collected and cells (C) were lysed and analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, electrotransferred to nitrocellulose, and immunoblotted with polyclonal anti‐Tg. HSP90 is a loading control. (c) From three independent experiments, recovery of Tg in 293 T cell lysates (dark gray) and media (light gray) were quantified for each construct from reducing SDS‐PAGE analysis; mean ± SD. No significant differences were observed between groups.

FIGURE 3

Structural differences between the ChEL domain and AChE are concentrated at the contact interfaces between ChEL and Tg region I. (a) Alignment and superimposition of the structures of hAChE (PDB 3LII) and bTg ChEL (PDB 7n4y, Tg positions 2209–2730). The C‐alpha root‐mean‐square deviation (RMSD) values were calculated for AChE versus ChEL, with unique AChE structure colored in red. The dotted box highlights contact interface #1. (b) Local structural representation of bTg (see Section 5) enlarged from the area enclosed by the dotted box in panel (a), showing contact interface #1 between Tg region I (green) and ChEL (gold) within a Tg monomer.

FIGURE 4

Functional role of K2225 in secretory ChEL to rescue the secretory trafficking of co‐expressed Tg‐I–II–III (that lacks its own ChEL). (a) Schematic representation of wild‐type (WT) and mutant secretory ChEL: the white box represents an engineered signal peptide to make ChEL into a secretory protein. (b) 293 T were either nontransfected (NT) or transfected to express the indicated mouse Tg‐derived constructs (Tables S1 and S2). After a 13‐h period, the media (M) were collected and cells (C) were lysed and analyzed by SDS‐PAGE and Western blotting with polyclonal anti‐Tg. β‐actin is a loading control. (c) From three independent experiments like that shown in panel (b), the relative secretion of ChEL was calculated for each construct; mean ± SD; N.S = non‐significant. (d) 293 T were nontransfected (NT), transfected with Tg region I–II–III alone or co‐expressed with the indicated secretory ChEL construct. C = cells; M = media. β‐actin is a loading control. (e, f). From three independent experiments like that shown in panel (d), the relative secretion of ChEL (panel e) and I–II–III (panel f) was calculated; mean ± SD; N.S = non‐significant; *, p < 0.05.

FIGURE 5

Functional role of K2225 in the trafficking of full‐length Tg. (a) Schematic representation of wild‐type (WT) Tg and Tg‐K2225E (for amino acid numbering differences between species, see Table S2). (b) 293 T were nontransfected (NT) or transfected to express the indicated mouse Tg constructs. After a 13‐h period, the media (M) were collected and cells (C) were lysed and analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, electrotransferred to nitrocellulose, and immunoblotted with polyclonal anti‐Tg. HSP90 is a loading control. (c) From three independent experiments like that shown in panel (b) (reduced samples), the relative secretion of mouse Tg was calculated for each construct; mean ± SD; *, p < 0.05.

FIGURE 6

Role of the C2444–C2455 disulfide bond in the trafficking of secretory ChEL. (a) Schematic representation of wild‐type (WT) or mutant secretory ChEL constructs. The putative C2444–C2455 disulfide loop in the ChEL domain is circled above. For amino acid numbering differences between species, see Table S2. (b) 293 T were either nontransfected (NT) or transfected to express the indicated secretory ChEL constructs derived from mouse Tg. After a 13‐h period, the media (M) were collected and cell (C) were lysed and analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, electrotransferred to nitrocellulose, and immunoblotted with polyclonal anti‐Tg. β‐actin is a loading control. (c) From three independent experiments like that shown in panel (b) (reduced samples), the relative secretion of secretory ChEL was calculated for each construct; mean ± SD; *, p < 0.05.

FIGURE 7

Role of the C2444–C2455 disulfide bond in the trafficking of full‐length Tg. (a) Schematic representation of wild‐type (WT) or mutant Tg constructs. The putative C2444–C2455 disulfide loop in the Tg ChEL domain is circled above. For amino acid numbering differences between species, see Table S2. (b) 293 T were either nontransfected (NT) or transfected to express the indicated bTg constructs. After a 13‐h period, the media (M) were collected and cells (C) were lysed and analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, electrotransferred to nitrocellulose, and immunoblotted with polyclonal anti‐Tg. HSP90 is a loading control. (c) From three independent experiments like that shown in panel (b) (reduced samples), the relative secretion of bTg was calculated for each construct; mean ± SD; *, p < 0.05.

FIGURE 8

Absence of the C2444–C2455 disulfide bond affects folding of the ChEL domain. (a) 293 T were either nontransfected (NT) or transfected to express the indicated secretory ChEL constructs derived from mouse Tg. (For amino acid numbering differences between species, see Table S2.) After a 13 h period, the media (M) were collected and cells (C) were lysed and analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, electrotransferred to nitrocellulose, and immunoblotted with polyclonal anti‐Tg. β‐actin is a loading control. Bands detected from NT cells (and media) are nonspecific. By nonreducing SDS‐PAGE, intracellular accumulation of misfolded mutant ChEL monomers and aberrant disulfide‐linked complexes were detected. (b) Lysates from transfected 293 T cells expressing the indicated constructs were analyzed by SDS‐PAGE under nonreducing (left) and reducing (right) conditions, followed by Western blotting with polyclonal anti‐Tg. β‐actin is a loading control. Notably, the ChEL‐K225E construct does not form the aberrant disulfide‐linked complexes that are detected for the other two misfolded mutant ChEL constructs.