Portal de Periódicos da CAPES

Metabolic Instability of Type 2 Deiodinase Is Transferable To Stable Proteins Independently of Subcellular Localization

2006; Elsevier BV; Volume: 281; Issue: 42 Linguagem: Inglês

10.1016/s0021-9258(19)84067-x

1083-351X

Aniko ́ Zeo ̈ld, Li ́via Pormu ̈ller, Monica Dentice, John W. Harney, Cyntia Curcio‐Morelli, Susana M. Tente, Antônio C. Bianco, B. Gereben,

Erythrocyte Function and Pathophysiology

Thyroid hormone activation is catalyzed by two deiodinases, D1 and D2. Whereas D1 is a stable plasma membrane protein, D2 is resident in the endoplasmic reticulum (ER) and has a 20-min half-life due to selective ubiquitination and proteasomal degradation. Here we have shown that stable retention explains D2 residency in the ER, a mechanism that is nevertheless over-ridden by fusion to the long-lived plasma membrane protein, sodium-iodine symporter. Fusion to D2, but not D1, dramatically shortened sodium-iodine symporter half-life through a mechanism dependent on an 18-amino acid D2-specific instability loop. Similarly, the D2-specific loop-mediated protein destabilization was also observed after D2, but not D1, was fused to the stable ER resident protein SEC62. This indicates that the instability loop in D2, but not its subcellular localization, is the key determinant of D2 susceptibility to ubiquitination and rapid turnover rate. Our data also show that the 6 N-terminal amino acids, but not the 12 C-terminal ones, are the ones required for D2 recognition by WSB-1. Thyroid hormone activation is catalyzed by two deiodinases, D1 and D2. Whereas D1 is a stable plasma membrane protein, D2 is resident in the endoplasmic reticulum (ER) and has a 20-min half-life due to selective ubiquitination and proteasomal degradation. Here we have shown that stable retention explains D2 residency in the ER, a mechanism that is nevertheless over-ridden by fusion to the long-lived plasma membrane protein, sodium-iodine symporter. Fusion to D2, but not D1, dramatically shortened sodium-iodine symporter half-life through a mechanism dependent on an 18-amino acid D2-specific instability loop. Similarly, the D2-specific loop-mediated protein destabilization was also observed after D2, but not D1, was fused to the stable ER resident protein SEC62. This indicates that the instability loop in D2, but not its subcellular localization, is the key determinant of D2 susceptibility to ubiquitination and rapid turnover rate. Our data also show that the 6 N-terminal amino acids, but not the 12 C-terminal ones, are the ones required for D2 recognition by WSB-1. The major secretory product of the thyroid gland is the prohormone thyroxine (T4), which can be activated to 3,5,3′-triiodothyronine (T3), the hormone form that binds the thyroid hormone receptor. Type 2 iodothyronine deiodinase (D2) is the key enzyme that catalyzes T4 to T3 conversion in both the developing and adult brain. D2 is a thioredoxin fold-containing selenoenzyme that is one of a structurally and functionally related group of enzymes that also includes types 1 and 3 deiodinases (D1 and D3, respectively) (1Callebaut I. Curcio-Morelli C. Mornon J.P. Gereben B. Buettner C. Huang S. Castro B. Fonseca T.L. Harney J.W. Larsen P.R. Bianco A.C. J. Biol. Chem. 2003; 278: 36887-36896Abstract Full Text Full Text PDF PubMed Scopus (114) Google Scholar, 2Bianco A.C. Salvatore D. Gereben B. Berry M.J. Larsen P.R. Endocr. Rev. 2002; 23: 38-89Crossref PubMed Scopus (1103) Google Scholar). Deiodinases are integral membrane proteins with a single N-terminal transmembrane domain located in different cell compartments. Whereas D1 and D3 are plasma membrane proteins (3Baqui M. Botero D. Gereben B. Curcio C. Harney J.W. Salvatore D. Sorimachi K. Larsen P.R. Bianco A.C. J. Biol. Chem. 2002; 4: 4Google Scholar, 4Baqui M.M. Gereben B. Harney J.W. Larsen P.R. Bianco A.C. Endocrinology. 2000; 141: 4309-4312Crossref PubMed Scopus (106) Google Scholar) with a long half-life, D2 is a short-lived (5Leonard J.L. Kaplan M.M. Visser T.J. Silva J.E. Larsen P.R. Science. 1981; 214: 571-573Crossref PubMed Scopus (174) Google Scholar, 6St. Germain D.L. Endocrinology. 1988; 122: 1860-1868Crossref PubMed Scopus (67) Google Scholar) endoplasmic reticulum (ER) 2The abbreviations used are: ER, endoplasmic reticulum; D1, type 1 deiodinase; D2, type 2 deiodinase; D3, type 3 deiodinase; T4, thyroxine; aa, amino acid; HEK, human embryonic kidney; EndoH, Endoglycosidase H; NIS, sodium-iodide symporter; E3, ubiquitin-protein isopeptide ligase. 2The abbreviations used are: ER, endoplasmic reticulum; D1, type 1 deiodinase; D2, type 2 deiodinase; D3, type 3 deiodinase; T4, thyroxine; aa, amino acid; HEK, human embryonic kidney; EndoH, Endoglycosidase H; NIS, sodium-iodide symporter; E3, ubiquitin-protein isopeptide ligase. resident protein (4Baqui M.M. Gereben B. Harney J.W. Larsen P.R. Bianco A.C. Endocrinology. 2000; 141: 4309-4312Crossref PubMed Scopus (106) Google Scholar) that undergoes ubiquitination (7Gereben B. Goncalves C. Harney J.W. Larsen P.R. Bianco A.C. Mol. Endocrinol. 2000; 14: 1697-1708Crossref PubMed Scopus (127) Google Scholar) and proteasomal degradation (8Steinsapir J. Harney J. Larsen P.R. J. Clin. Investig. 1998; 102: 1895-1899Crossref PubMed Scopus (87) Google Scholar). D2 is ubiquitinated by a catalytic core complex, modeled as Elongin BC-Cul5-Rbx1 (ECSWSB-1), with WSB-1 functioning as an E3 ubiquitin ligase subunit (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). WSB-1 (also known as SWiP-1) is a SOCS box-containing WD-40 protein that is induced by Hedgehog signaling in embryonic structures during chicken development (10Vasiliauskas D. Hancock S. Stern C.D. Mech. Dev. 1999; 82: 79-94Crossref PubMed Scopus (46) Google Scholar). The WD-40 propeller of WSB-1 recognizes the D2 molecule, whereas the SOCS box domain mediates its interaction with a ubiquitinating catalytic core complex. Although both ubiquitin-conjugating enzymes UBC6 and UBC7 can support D2 ubiquitination (11Botero D. Gereben B. Goncalves C. De Jesus L.A. Harney J.W. Bianco A.C. Mol. Endocrinol. 2002; 16: 1999-2007Crossref PubMed Scopus (50) Google Scholar, 12Kim B.W. Zavacki A.M. Curcio-Morelli C. Dentice M. Harney J.W. Larsen P.R. Bianco A.C. Mol. Endocrinol. 2003; 17: 2603-2612Crossref PubMed Scopus (45) Google Scholar), specific deubiquitination of D2 is also possible through von Hippel-Lindau protein-interacting deubiquitinating enzymes VDU1 and VDU 2 (13Curcio-Morelli C. Zavacki A.M. Christofollete M. Gereben B. de Freitas B.C. Harney J.W. Li Z. Wu G. Bianco A.C. J. Clin. Investig. 2003; 112: 189-196Crossref PubMed Scopus (110) Google Scholar), which rescues D2 from irreversible proteolysis. However, it is not clear whether D2 is intrinsically unstable or its instability is due to co-localization in the ER with the ECSWSB-1 catalytic core complex. Here we have shown that the instability loop in D2, but not its subcellular localization, is the key determinant of its rapid turnover rate. Expression Constructs—The ER-specific NKT or the Golgi-specific YTPPP glycosylation motif was fused to the N or C terminus of D2, while the opposite terminus was tagged with FLAG using Vent PCR and standard recombinant DNA methods. The generated fragments were inserted into D10 expression vector, yielding N-FLAG-hD2-C-terminal-NKT/YTPPP and N-terminal YTPPP/NKT-hD2-C-FLAG (Fig. 1A). Lys for Arg mutations in the human D2 (hD2) were performed by site-directed mutagenesis using overlap extension Vent PCR and inserts were cloned into D10 vector (Fig. 2). The hD2 fragment encoding an hD2 protein from amino acids (aa) aa 42 to 273 (or its Δ18-aa version lacking aa 92-109 as described in Ref. 9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar) was fused to the C terminus of the plasma membrane located rat sodium-iodide symporter (rNIS) in pCI-Neo vector 3′ to a FLAG tag. The NIS plasmid was donated by Dr. Nancy Carrasco, New York, NY). A similar construct was made by fusing the rat D1 fragment (aa 34-257) to NIS (Fig. 3A). The ER resident human Sec62 (kindly donated by Dr. E. Hartmann, Lu¨beck, Germany) was FLAG tagged on its N terminus, and the previously mentioned D2, Δ18-D2, and D1 fragments were fused to the C terminus of Sec62 in D10 vector (Fig. 4A).FIGURE 2Intracellular localization of D2 arginine mutants. Specific lysine residues of the D2 protein were replaced with arginine (indicated in rows A-D) followed by transient expression in HEK-293 cells and confocal microscopy using the same conditions. Scale bar, 10 μm.View Large Image Figure ViewerDownload Hi-res image Download (PPT)FIGURE 3A, schematic diagram of the NIS-D2/D1 proteins with FLAG. The D2 fragment encoding D2 protein from aa 42 to 273 or the D1 fragment (aa 34-257) was fused to the C terminus of rat sodium-iodide symporter (NIS). The FLAG epitope was placed into the junction. B, immunofluorescence confocal analysis of HEK-293 cells transiently expressing NIS-D1 or FLAG-D2 and co-stained with anti-FLAG antibody (red) and ER tracker (green). Scale bar, 10 μm. C, half-life of NIS-D2 and NIS-D1 in the plasma membrane. HEK-293 cells expressing NIS-D2, NIS-Δ18-D2, and NIS-D1 or the empty D10 vector were treated with 100 μm cycloheximide as translational blocker and subjected to Western blot with anti-FLAG antibody.View Large Image Figure ViewerDownload Hi-res image Download (PPT)FIGURE 4A, schematic diagram of the Sec62-D2/D1 with FLAG. The D2 fragment encoding D2 protein from aa 42 to 273 or the D1 fragment (aa 34-257) was fused to the C terminus of human Sec62. The N terminus of Sec62 was tagged with a FLAG epitope. B, immunofluorescence confocal analysis of HEK-293 cells transiently expressing Sec62-D1 or Sec62-D2 and co-stained with anti-FLAG antibody (red) and ER tracker (green). Scale bar, 10 μm. C, half-lives of Sec62-fused deiodinases in the ER. HEK-293 cells expressing Sec62, Sec62-D2, Sec62-Δ18-D2, D2, Sec62-D1, and D1 were treated with 100 μm cycloheximide and subjected to Western blot with anti-FLAG antibody. For densitometry see panel D. D, densitometry analysis of panel C indicating the instability of Sec62-D2 and D2. E, pulse-chase of Sec62-D2 chimera. HEK-293 cells expressing Sec62, Sec62-D2, or the empty D10 vector were subjected to pulse-chase (h, hour).View Large Image Figure ViewerDownload Hi-res image Download (PPT) The Δ12-D2 (12-aa deletion between aa 98 and 109) and Δ6-D2 (6-aa deletion between aa 92 and 97) were generated using overlap extension PCR without additional residues. The NcoI-PstI fragment was swapped into a N-FLAG wild-type D2 construct containing a SelP SECIS element. Overlap extension PCR was used to generate the 92 Ala for Thr replacement of the Δ12-D2 (Δ12-D2-92A) and the D1 + 6 aa, a rat D1 containing the aa 92-97 hD2 region (TEGGDN) between aa 102 and 103 (Fig. 5). Glycosylation Assay and Cell Surface Biotinylation Assay—Assay for glycosylation was performed as previously described (14Pedrazzini E. Villa A. Longhi R. Bulbarelli A. Borgese N. J. Cell Biol. 2000; 148: 899-914Crossref PubMed Scopus (51) Google Scholar). To study cell surface biotinylation, HEK-293 cells transiently expressing FLAG-tagged D2 lysine mutant proteins were washed with phosphate-buffered saline and incubated with 1.0 mg/ml sulfo-NHSLC-biotin (sulfo-biotin; Pierce). Cells were quenched twice with 100 mm glycine, harvested, and lysed. The lysate was centrifuged at 12,000 × g for 10 min and the supernatant incubated with streptavidin-agarose beads (Pierce) at 4 °C. The beads were washed and finally the complexes processed for Western analysis with anti-FLAG M2 antibody (1:3000). Western Blot and Cell Fractionation—Western blot was performed as described (7Gereben B. Goncalves C. Harney J.W. Larsen P.R. Bianco A.C. Mol. Endocrinol. 2000; 14: 1697-1708Crossref PubMed Scopus (127) Google Scholar) using mouse anti-FLAG M2 (Sigma) and mouse anti-tubulin (DM1A) antibody (Sigma). Microsomal pellet and cytosol fractions were prepared as described (4Baqui M.M. Gereben B. Harney J.W. Larsen P.R. Bianco A.C. Endocrinology. 2000; 141: 4309-4312Crossref PubMed Scopus (106) Google Scholar) from HEK-293 cells expressing Sec62-D2 or Sec62-D1. Confocal Laser Microscopy—The FLAG/ER tracker co-staining was performed as described (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). Sec62-D2 or Sec62-D1 and NIS-D2 or NIS-D1 constructs were transiently transfected into HEK-293 cells. For fluorescent co-staining of ER and the FLAG-tagged proteins, living cells were incubated with ER-Tracker (Molecular Probes) at growth conditions followed by fixation with 4% paraformaldehyde and permeabilization with 0.2% Triton X-100 as described (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). The FLAG epitope was detected with mouse anti-FLAG M2 (Sigma) and anti-mouse-Cy3 (Jackson ImmunoResearch), followed by confocal laser microscopy. FLAG-tagged arginine mutants were stained with anti-FLAG M2 and anti-mouse-DTAF (Jackson ImmunoResearch). Pulse-chase and Immunoprecipitation—Pulse-chase with [35S]Met/Cys and anti-FLAG immunoprecipitation was performed as described (7Gereben B. Goncalves C. Harney J.W. Larsen P.R. Bianco A.C. Mol. Endocrinol. 2000; 14: 1697-1708Crossref PubMed Scopus (127) Google Scholar, 9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). Small Interfering RNA-based WSB-1 Knockdown and Deiodinase Assays—These methods were performed as previously described (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). Deiodinase activity transiently expressed in HEK-293 cells was denominated by milligrams of protein and Renilla used as internal control to monitor transfection efficiency. D2 Is Subjected to Static Retention in the ER—To investigate the ER residency mechanism of D2, we studied transiently expressed D2 molecules containing ER-(NKT; N-linked) or Golgi-specific (YTPPP; O-linked) glycosylation signals at the N or C termini (Fig. 1A). Fusion of NKT to the N-terminal of D2 resulted in an EndoH-sensitive glycosylated ∼40-kDa form of D2, whereas fusion to the C-terminal of D2 did not (Fig. 1B), suggesting that D2 does not reach the medial Golgi, a compartment where resistance to EndoH is acquired (14Pedrazzini E. Villa A. Longhi R. Bulbarelli A. Borgese N. J. Cell Biol. 2000; 148: 899-914Crossref PubMed Scopus (51) Google Scholar). Still, D2 could reach the cis-Golgi and be returned to the ER. To test this possibility, Golgi-specific O-linked glycosylation signal sequences were fused to N or C termini of D2, but no D2 glycosylation was detected (Fig. 1C), supporting the contention that a static retention mechanism explains ER residency. Absence of glycosylation from both of the tags fused to the C terminus confirms our previous suggestion that D2 is a type 1 protein (4Baqui M.M. Gereben B. Harney J.W. Larsen P.R. Bianco A.C. Endocrinology. 2000; 141: 4309-4312Crossref PubMed Scopus (106) Google Scholar). Next, because the best characterized ER retention signals contain dilysine motifs (15Teasdale R.D. Jackson M.R. Annu. Rev. Cell Dev. Biol. 1996; 12: 27-54Crossref PubMed Scopus (447) Google Scholar) and D2 contains numerous lysine residues at the C-terminal (Fig. 2), we tested whether lysine-to-arginine D2 mutants remain in the ER when transiently expressed in HEK-293 cells or leak to the cell surface as demonstrated for the OST48 protein (16Hardt B. Bause E. Biochem. Biophys. Res. Commun. 2002; 291: 751-757Crossref PubMed Scopus (22) Google Scholar). 13 lysines in the human D2 were replaced by arginine in different combinations, while the extreme C-terminal (aa 267/268) lysine doublet was removed by truncation. None of the mutant proteins were sorted to the plasma membrane (Fig. 2). This finding was also confirmed by the absence of D2 biotinylation after cells were exposed to the cell-impermeant sulfo-biotin probe (data not shown), confirming retention of the lysine-to-arginine D2 mutants in the ER. Reassignment of D2 to the Plasma Membrane Destabilizes NIS—The signal in D2 encoding ER retention could be located in its single transmembrane domain or its globular domain. To test the role of the transmembrane domain in this process, we prepared a D2 with a truncation in the first 42 aa corresponding to the transmembrane domain and detected it in the cytosol during transient expression. 3B. Gereben and A. C. Bianco, unpublished observations. Therefore, this D2 could not be used to study ER retention mechanisms. Next, we studied the role of the globular domain by fusing this truncated D2 to the cytosolic C terminus of the NIS (Fig. 3A), a large and stable plasma membrane protein. The resulting NIS-D2 chimera was sorted to the plasma membrane during transient expression, indicating that, if present in the globular domain, the D2 ER retention signal is easily overridden by the NIS-plasma membrane-sorting signal (Fig. 3B). In transient expression studies using cycloheximide, a general translation-blocking agent, the NIS-D2 chimeras have a much faster turnover rate when compared with NIS-D1 (Fig. 3C). NIS-D1 was used to test whether this was due to a global nonspecific destabilization mechanism associated with fusion to another protein, i.e. D1, a stable plasma membrane resident deiodinase structurally similar to D2. Although NIS-D1 was also sorted to the plasma membrane (Fig. 3B), its turnover rate remained slow, indicated by the absence of signal decrease after cells were exposed to cycloheximide. This suggests that the intrinsic metabolic instability of D2 was specifically transferred to NIS, even if this chimera were sorted to a subcellular location outside the ER, where it is normally expressed. This supposition was further strengthened by the observation that the metabolic instability of NIS-D2 depends on the presence of the destabilizing 18-aa loop in D2, located after the transmembrane domain in the cytosolic part of the protein. Its truncation results in a NIS-D2 chimera (NIS-Δ18-D2) that is as stable as NIS-D1 (Fig. 3C). Fusion to D2, but Not to D1, Destabilizes Sec62 in the ER—D1 is normally present in the plasma membrane, but because of its remarkable structural similarity to D2 it could be targeted by ubiquitination if retained in the ER. To test this hypothesis and to further test the transferability of D2 metabolic instability to other long lived proteins, we prepared a chimera between the globular domain of D1 or D2 and the stable ER resident protein, Sec62 (Fig. 4A). Subcellular fractionation followed by Western analysis showed that both Sec62-D1 and Sec62-D2 proteins were found in the microsomal pellet but not in the soluble cytosol fraction (results not shown). Using confocal microscopy we found that Sec62-D1 was now rerouted to and Sec62-D2 retained in the ER, as evidenced by co-localization with the ER tracker (Fig. 4B). Although these observations do not exclude the possibility that the molecular signal directing D1 to the plasma membrane originates from the transmembrane domain, they may indicate that, if present in the globular domain, the signal can be easily overridden by the signal in Sec62 directing it to the ER. Remarkably, using cycloheximide we found that fusion to D1 did not destabilize the Sec62 protein, confirming that deiodinase metabolic instability cannot be explained by ER residency. However, fusion to D2 once more produced an unstable chimera through a mechanism dependent on the presence of the destabilizing 18-aa loop in D2 (Fig. 4, C and D). These data have been confirmed by pulse-chase experiments (Fig. 4E) demonstrating the long half-life of Sec62 and the destabilizing potency of fusion to D2. The Instability Loop in D2 Can Be Reduced to the N-terminal 6 Amino Acids without Loss of Function but It Is Not Functional in D1—To identify critical aa in the 18-aa D2 loop, we created a truncated D2 molecule lacking the 12 C-terminal of these 18 aa (Fig. 5, Δ12-D2). In transient expression studies in HEK-293 cells, this molecule retained activity and responded similarly to wild-type D2 when exposed to 30 nm T4, 1 μm MG132, 100 μm cycloheximide, or during WSB-1 knock down (Table 1). This indicates that the remaining 6 aa in this loop (TEGGDN, aa 92-97 in human D2) should account for the destabilizing properties described above. To test this, a Δ6-D2 truncated D2 molecule was created (Fig. 5), and when transiently expressed in HEK-293 cells this molecule retained short half-life and sensitivity to T4 and MG132 but did not respond to WSB-1 knock down (Table 1). One of these 6 aa is Thr-92, in a position in which a polymorphism has been reported in humans. Therefore, we replaced Thr-92 by Ala and then prepared a T92A-D2 construct that also contained the 12-aa truncation (Δ12-D2-92A, Fig. 5). However, the resulting new molecule performed similarly to Δ12-D2 during transient expression studies under the circumstances described above (data not shown). Next, a sequence of aa corresponding to the remaining 6 residues in the shortened D2 loop was inserted in the analogous position (between aa 102-103) of the rat D1 molecule (Fig. 5, D1 + 6aa) and transiently expressed in HEK-293 cells. The resulting chimera retained D1 activity and was not affected by exposure to 30 nm T4, 1 μm MG132, or 100 μm cycloheximide, behaving identically to wild-type D1.TABLE 1Characterization of D2 loop mutantsT4MG132CXWSB-1 RNAi co-transfectionWild-type D251 ± 14*420 ± 84*48 ± 9.6*775 ± 292*(3)(3)(3)(3)Δ12-D221 ± 4.9*290 ± 60*24 ± 4*740 ± 105*(6)(6)(4)(3)Δ6-D213 ± 11*149 ± 6131 ± 0*70 ± 32(3)(9)(3)(6) Open table in a new tab Type 2 deiodinase is an ER resident protein that undergoes selective ubiquitination, which constitutes an on/off mechanism controlled by the level of substrate, T4. Ubiquitination also results in D2 proteasomal degradation, explaining its 20-min half-life. In contrast, the structurally related D1 is not ubiquitinated and has a half-life >12 h (7Gereben B. Goncalves C. Harney J.W. Larsen P.R. Bianco A.C. Mol. Endocrinol. 2000; 14: 1697-1708Crossref PubMed Scopus (127) Google Scholar). The difference depends on a D2-specific 18-aa loop that mediates interaction with the D2-specific E3 ligase adaptor, WSB-1 (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). Here we have shown that D2 is retained in the ER and the instability conferred by the 18-aa loop is transferable to otherwise stable plasma membrane or ER resident proteins, independently of their subcellular localization. ER resident proteins can be actively retained in the ER (static retention) or recycled to the ER by retrieval signals after they are sorted to the Golgi (dynamic retrieval) (15Teasdale R.D. Jackson M.R. Annu. Rev. Cell Dev. Biol. 1996; 12: 27-54Crossref PubMed Scopus (447) Google Scholar, 17Andersson H. Kappeler F. Hauri H.P. J. Biol. Chem. 1999; 274: 15080-15084Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar). In some cases both mechanisms operate for the same protein, to retrieve molecules from the Golgi complex that escaped the retention mechanism, as demonstrated for calreticulin (18Sonnichsen B. Fullekrug J. Nguyen Van P. Diekmann W. Robinson D.G. Mieskes G. J. Cell Sci. 1994; 107: 2705-2717Crossref PubMed Google Scholar). Because D2 is not glycosylated, we used targeted glycosylation (14Pedrazzini E. Villa A. Longhi R. Bulbarelli A. Borgese N. J. Cell Biol. 2000; 148: 899-914Crossref PubMed Scopus (51) Google Scholar) to study the mechanisms that regulate subcellular distribution of D2. This approach is based on the fact that different signal peptides direct specific glycosylation reactions in distinct cell compartments. The N- and O-linked glycosylation consensus sequences have been characterized as directing ER- or Golgi-specific glycosylation, respectively (14Pedrazzini E. Villa A. Longhi R. Bulbarelli A. Borgese N. J. Cell Biol. 2000; 148: 899-914Crossref PubMed Scopus (51) Google Scholar, 19Yoshida A. Suzuki M. Ikenaga H. Takeuchi M. J. Biol. Chem. 1997; 272: 16884-16888Abstract Full Text Full Text PDF PubMed Scopus (68) Google Scholar). In the present study, N-terminal, but no C-terminal, NKT-tagged D2 was found to be glycosylated (EndoH-sensitive), compatible with D2 being a type 1 transmembrane protein that does not reach the medial Golgi, the compartment where EndoH resistance is acquired (Fig. 1B). YTPPP-tagged D2 in either terminus showed no O-linked glycosylation, a process that starts in the cis-Golgi (20Roth J. J. Cell Biol. 1984; 98: 399-406Crossref PubMed Scopus (148) Google Scholar, 21Piller V. Piller F. Fukuda M. J. Biol. Chem. 1990; 265: 9264-9271Abstract Full Text PDF PubMed Google Scholar), indicating that D2 trafficking does not reach this compartment in the Golgi apparatus. The lack of Golgi-specific glycosylation in YTPPP-tagged D2 also indicates that D2 is probably not recycled from the cis-Golgi (Fig. 1C). Our current knowledge on ER retention signals is limited (22Wrzeszczynski K.O. Rost B. Cell. Mol. Life Sci. 2004; 61: 1341-1353Crossref PubMed Scopus (18) Google Scholar). Certain type 1 ER resident proteins have lysine residues that serve as signals for ER localization (dynamic retention) via COP-1-mediated retrieval or ER retention (15Teasdale R.D. Jackson M.R. Annu. Rev. Cell Dev. Biol. 1996; 12: 27-54Crossref PubMed Scopus (447) Google Scholar, 17Andersson H. Kappeler F. Hauri H.P. J. Biol. Chem. 1999; 274: 15080-15084Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar, 23Jackson M.R. Nilsson T. Peterson P.A. EMBO J. 1990; 9: 3153-3162Crossref PubMed Scopus (724) Google Scholar). In addition, substitution of C-terminal lysines for arginine in the OST48 subunit of the oligosaccharyl transferase complex resulted in cell surface expression (16Hardt B. Bause E. Biochem. Biophys. Res. Commun. 2002; 291: 751-757Crossref PubMed Scopus (22) Google Scholar). Although D2 contains cytosolic C-terminal (di)lysine motifs, their substitutions for arginine did not indicate their involvement in the ER retention mechanism (Fig. 2). In the absence of positive transport signals, such as in D2, the localization of a protein in the ER may result from signals in the transmembrane domain and its interaction with the membranes, as has been shown for UBC6 (ubiquitin-conjugating enzyme) (24Yang M. Ellenberg J. Bonifacino J.S. Weissman A.M. J. Biol. Chem. 1997; 272: 1970-1975Abstract Full Text Full Text PDF PubMed Scopus (111) Google Scholar) and cytochrome b5 (14Pedrazzini E. Villa A. Longhi R. Bulbarelli A. Borgese N. J. Cell Biol. 2000; 148: 899-914Crossref PubMed Scopus (51) Google Scholar). A limiting factor, however, is that truncating the D2 transmembrane domain yields a cytosolic protein that cannot be used to understand the molecular determinants of D2 subcellular localization (data not shown). This limitation was bypassed by fusing the truncated D2 protein to NIS, which resulted in a plasma membrane D2-NIS chimera. The fact that D2-NIS was found exclusively in the plasma membrane suggests that the ER retention mechanism in D2 is located in the transmembrane domain, although we cannot exclude the possibility that the D2 mechanism might be weak, easily overridden by the plasma determinant contained in the NIS protein. The observations that the NIS-D2 chimera is unstable due to the presence of the 18-aa loop (Fig. 3C) and that both NIS-Δ18-D2 and NIS-D1 are stable proteins suggest that NIS-D2 is subjected to ubiquitination and proteasomal degradation. This was not expected, because WSB-1, the E3 ligase adaptor that is D2 specific, has been shown to co-localize with D2 in the ER and not to be present in the plasma membrane or cell periphery (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). These findings suggest that a different E3 ligase adaptor could recognize the D2-specific instability domain in the NIS-D2 protein and mediate its ubiquitination. Sec62 is a stable ER resident protein required for the import of secretory protein precursors into the ER. It has two ER membrane-spanning domains, and its hydrophilic N and C termini are located in the cytosol (25Deshaies R.J. Schekman R. Mol. Cell. Biol. 1990; 10: 6024-6035Crossref PubMed Scopus (90) Google Scholar). Incorporation of the yeast Deg1 degradation signal of the short-lived yeast transcription factor MATα2 in the N portion of Sec62 produced an unstable ER resident chimera (26Mayer T.U. Braun T. Jentsch S. EMBO J. 1998; 17: 3251-3257Crossref PubMed Scopus (179) Google Scholar). In the present studies we used a similar strategy and generated Sec62-D1, Sec62-D2, and Sec62-Δ18-D2 chimeras. Fusion of D2, but not D1 or Δ18-D2, to the cytosolic C terminus of Sec62 resulted in Sec62 destabilization (Fig. 4, C and D). It could be argued that D1 is normally not ubiquitinated because it is sorted to the plasma membrane, a compartment that lacks WSB-1. The fact that Sec62-D1 has a long half-life, even when expressed in the ER, confirms the critical role played by the 18-aa residue in D2 for its susceptibility to ubiquitination. We have previously shown that removal of the 18-aa loop prolongs D2 half-life by impairing its interaction with WSB-1 (9Dentice M. Bandyopadhyay A. Gereben B. Callebaut I. Christoffolete M.A. Kim B.W. Nissim S. Mornon J.P. Zavacki A.M. Zeold A. Capelo L.P. Curcio-Morelli C. Ribeiro R. Harney J.W. Tabin C.J. Bianco A.C. Nat. Cell Biol. 2005; 7: 698-705Crossref PubMed Scopus (179) Google Scholar). The present truncation analysis of this loop indicates that the N-terminal 6 aa are sufficient to maintain the D2 molecule in the conformation required for WSB-1 susceptibility (Table 1), although its transfer to the D1 molecule did not produce an unstable protein. Therefore, it is possible that the capacity to destabilize proteins does not depend on the aa sequence of the loop itself but rather on a D2-specific structural conformation resulting from the presence of the loop in the D2 molecule. This is also supported by the fact that residue substitutions within the D2-specific 18-aa loop, such as the T92A polymorphism (Fig. 5), which in humans has been associated with insulin resistance (27Mentuccia D. Proietti-Pannunzi L. Tanner K. Bacci V. Pollin T.I. Poehlman E.T. Shuldiner A.R. Celi F.S. Diabetes. 2002; 51: 880-883Crossref PubMed Scopus (170) Google Scholar), played no role in protein destabilization. This is further supported by the lack of conservation of the loop sequence among D2 proteins and by the observation that blasting the aa sequence of the 18-aa loop against GenBank™ identified only the human D2 protein. The removal of the N-terminal 6 aa of the D2 loop interfered with WSB-1-mediated ubiquitination such that Δ6-D2 activity was not increased during WSB-1 knock down. However, this truncation did not eliminate the susceptibility to exposure to T4 or prevent proteasomal degradation (Table 1). This suggests that another E3 ligase could mediate the ubiquitination of the Δ6-D2. This is supported by a recent report in which yeast-expressed D2 is targeted by Doa10, a ubiquitin ligase known to ubiquitinate other ER resident proteins (28Ravid T. Kreft S.G. Hochstrasser M. EMBO J. 2006; 25: 533-543Crossref PubMed Scopus (211) Google Scholar), although it is presently unknown whether TEB4, its likely human ortholog (29Kreft S.G. Wang L. Hochstrasser M. J. Biol. Chem. 2006; 281: 4646-4653Abstract Full Text Full Text PDF PubMed Scopus (116) Google Scholar), mediates D2 degradation in vertebrates. In conclusion, the key thyroid hormone-activating enzyme D2 is subject to static retention in the ER. Its metabolic instability can be transferred to otherwise stable proteins by fusion with an 18-aa loop-containing D2 molecule, independently of subcellular localization. Although the loop can be reduced to 6 critical N-terminal aa without loss of function, its capacity to promote metabolic instability is not transferable when isolated from D2. We thank Dr. E. Hartmann (Lu¨beck, Germany) for the hSec62 encoding plasmid and Dr. N. Carrasco (New York, NY) for the rNIS construct.