INTRODUCTION

Thyroid peroxidase (TPO)¹ is the major enzyme in the biosynthesis of thyroid hormones, because it catalyzes the iodination and coupling of iodotyrosine residues on thyroglobulin to produce thyroxine (T4) and 3,3,5-triiodothyronine (1-3). TPO is also a major autoantigen in the pathogenesis of autoimmune thyroid diseases (4-6). Also, defective TPO causes congenital hypothyroidism because of an iodide organification defect linked with mutations of the TPO gene (7-9).

TPO is a membrane-bound glycoprotein of 933 amino acids containing a heme prosthetic group. Kimura et al. suggested that His-407 or His-414 is the proximal heme binding site (10). On the other hand, His-586 is conserved in human and porcine TPO and also in human myeloperoxidase (11), which suggests that His-586 may be a critical residue for the enzymatic activity of the protein by being the putative distal heme binding site. Four potential asparagine-linked glycosylation sites are also present in the extracellular region, and there is some evidence for at least one disulfide bond in this part of the molecule (12).

Distribution and delivery of TPO are apically polarized, but only ~30% TPO is detected at the thyrocyte surface (13). Mistargeting of another small fraction of TPO to the basolateral surface seems to occur (13, 14). This may account for the hypothetical presentation of TPO to the circulating immune system, resulting in its antigenicity and its part in autoimmune thyroid diseases. In the presence of H₂O₂, TPO catalyzes the iodide organification of thyroglobulin, which is processed mainly at the apical membrane of the thyrocyte (15-17). However, a quantitatively minor iodination of thyroglobulin appears to occur intracellularly (18). TPO gene codes for the full-length TPO (TPO1) and generates alternatively spliced forms (for review see Ref. 5). The main alternatively spliced TPO mRNA species described so far has a 171-base pair deletion (exon 10) and codes for a protein of 876 amino acids (TPO2) (19, 20). TPO2 is 57 amino acids shorter than the major protein and lacks one of the potential glycosylation sites and His-586. TPO2 mRNA has been detected in normal and Graves thyroids (19, 20) but also in a case of congenital hypothyroidism caused by a premature termination signal in the TPO gene, suggesting that TPO2 is enzymatically inactive (21). Note that 10 years after the discovery of TPO2-specific mRNA, nothing is known about the protein. Hence no further information is available on its enzymatic and immunological activities.

Using CHO cell lines expressing TPO2 and TPO1, we investigated the structural and functional aspects of TPO2. With reference to the major protein TPO1, we determined how changes of the primary sequence affected the conformational structure of the protein, as shown through its epitopic recognition by TPO mAbs. We determined the effects of those structural modifications on the intracellular trafficking, the localization, and the enzymatic activity of the protein.

MATERIALS AND METHODS

Full-length 3060-kilobase human TPO1 cDNA kindly provided by B. Rapoport, University of California, San Francisco, CA, was cloned into HindIII and XbaI sites of the eukaryotic transfer vector pcDNA3 (Invitrogen, Leak, The Netherlands). A 578-nucleotide cDNA fragment corresponding to a region of TPO2 (1411-1989 base pairs) around the splicing was amplified by polymerase chain reaction from the gt11 human thyroid cDNA library (22). Oligonucleotides FP, 5-TATGAATTCTATGACTCCACCGCC, and RP, 5-GTCTCTAGACTCAAAGTCTTC, were used as primers (23). This 578-base pair-amplified fragment containing at both extremities the NcoI restriction site was cleaved by this enzyme. TPO1 cDNA was was released from TPO1-pcDNA3 construction and subcloned in pBluescript (Stratagene, La Jolla, CA). TPO1-pBluescript was then cleaned of the 748-base pair fragment containing the sequence that corresponds to exon 10 by double cleavage with NcoI and dephosphorylated by alkaline phosphatase. TPO1-pBluescript and the TPO2-NcoI fragment were then ligated. TPO2-pBluescript construct was cleaned of the TPO2 cDNA, which was cloned into HindIII and XbaI sites of the eukaryotic transfer vector pcDNA3. The nucleotide sequence of the polymerase chain reaction-derived NcoI fragment was determined by dideoxynucleotide sequencing to verify that the engineered TPO2-pcDNA3 construct was correct. Escherichia coli strain XL1 Blues (Stratagene) was transformed with the TPO2-pcDNA3 construct, and pure plasmid DNA preparations were obtained with the Wizard midipreps kit (Promega, Madison, WI). Enzyme reactions and DNA manipulations were done as in Sanbrook et al. (24).

CHO cells (ECACC n°85050302) were maintained in Ham's F-12 medium supplemented with 10% FBS, penicillin (100 IU/ml) and streptomycin (0.1 mg/ml). Cells were transfected by the lipofectAMINE method (Life Technologies, Inc.) with either pcDNA3-TPO1 or pcDNA3-TPO2 or pcDNA3 alone as control. Cells were cultured in a humidified incubator (5% CO₂) at 37 °C. Stable transfectants were selected in the presence of geneticin (400 µg/ml) and were subcloned by limiting dilutions. Positive TPO-expressing cell lines were identified by Western blotting or by immunoprecipitation after [³⁵S]Met + [³⁵S]Cys labeling (Expre³⁵S³⁵S protein labeling mix, DuPont NEN Life Science Products). A significant TPO1 and TPO2 expression was obtained by culturing CHO TPO cell lines with 10 mM sodium butyrate.

Cells were incubated in cysteine- and methionine-free Dulbecco's modified Eagle's medium supplemented with 10% FBS, 10 mM sodium butyrate, and 100 µCi/ml [³⁵S]Met + [³⁵S]Cys. Incubation was done for 5 or 48 h. In pulse-chase experiments, cells were incubated 1 h in Cys- and Met-free Dulbecco's modified Eagle's medium supplemented with 10% dialyzed FBS and 10 mM sodium butyrate. Cells were then pulsed for 30 min in the presence of 100 µCi/ml [³⁵S]Met + [³⁵S]Cys. After the pulse, the labeling medium was removed, and the cell surface was washed three times with PBS and then replaced by Dulbecco's modified Eagle's medium supplemented with 10% FBS, 5 mM Met, and 5 mM Cys. Chases were performed for 1, 5, 22, 48, and 72 h.

After being metabolically labeled, cells were harvested on ice by scraping in 1 ml of PBS and centrifuging at 700 × g for 7 min. Cell pellets were resuspended in 600 µl of TPO extraction buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.3% sodium deoxycholate, protease inhibitors (Complete^TM, Boehringer Mannheim), vortexed every 2 min during 20 min, and centrifuged at 10,000 × g for 5 min. Radiolabeled supernatants were incubated for 2 h at room temperature with mAbs recognizing either a sequential region (mAb 47) or conformational epitopes (mAb 15 and mAb 53) on the surface of the TPO molecule (25). mAbs 47 and 53 were used for TPO2, and mAbs 47 and 15 were used for TPO1. The use of two TPO mAbs in immunoprecipitation experiments allowed us to recover more than 90% of both TPO1 and TPO2 expressed by CHO cells (data not shown). mAbs were chosen for their strong recognition of each isoform (see "Results") through a panel of TPO mAbs. These mAbs were previously complexed with protein A-Sepharose 4B (Zymed Laboratories Inc., San Francisco, CA) by incubation overnight at 4 °C. Immune complexes were then retrieved by a brief centrifugation (10,000 × g, 10 s) and were washed 6 times with 1 ml of TPO extraction buffer and once with 1 ml of PBS. Immunoprecipitated TPO was recovered from mAb-protein A-Sepharose 4B complexes by boiling for 5 min in 80 µl of electrophoresis buffer (62 mM Tris-HCl, pH 6.8, 2% SDS, 5% 2-mercaptoethanol, and 5% glycerol) and then analyzed by SDS-PAGE (7.5%). Protein-associated radiactivity was visualized and quantified by phosphoimager (Fudjix BAS 1000).

TPO was also immunoprecipitated with a panel of mAbs directed against various antigenic domains of the TPO1 molecule (25). In this experiment, [³⁵S]Met + [³⁵S]Cys-radiolabeled CHO TPO cell lysates were precleared with zysorbin (Zymed Laboratories Inc.) before being immunoprecipitated 4 h at 25 °C with 50 µg of each of the TPO mAbs previously complexed with protein A-Sepharose 4B.

CHO TPO2 cell lysates were also tested against TPO autoantibodies in the sera from two patients with Hashimoto's thyroiditis who exhibited titers of TPO autoantibodies higher than 10,000 IU/ml as measured by a two-site chemoluminescent assay (Lumitest anti-TPO, Brahms, Berlin, Germany). Each sera (500 µl) was precleared with zysorbin before being complexed with protein A-Sepharose 4B by overnight incubation at 4 °C. The serum of a normal individual was used as a control. Metabolically labeled TPO2 was extracted as described previously and then incubated 2 h at room temperature with complexes of autoantibodies and protein A-Sepharose 4B. These complexes were retrieved by a brief centrifugation and treated as the mAbs. TPO2 immunoprecipitated by autoantibodies was eluted from complexes and analyzed by SDS-PAGE.

[³⁵S]Met + [³⁵S]Cys-labeled TPO2 was immunoprecipitated with mAb 47 and mAb 53 as described previously. TPO2 was eluted from the immunocomplexes by incubation with 5 µl of 10% SDS, diluted in the incubation buffer with either endo--N-acetylglucosaminidase H or peptide-N-glycanase F, centrifuged (10,000 × g, 10 s), and boiled for 5 min. Treatment with 0.2 unit of endo--N-acetylglucosaminidase H (Boehringer Mannheim) was done in 0.15 M sodium acetate buffer, pH 5.0, for 16 h at 37 °C. Deglycosylation by 1 unit of peptide-N-glycanase F (Boehringer Mannheim) was done in 0.15 M phosphate buffer, pH 8.5, for 16 h at 37 °C. Samples were then analyzed by SDS-PAGE.

CHO TPO1 and CHO TPO2 confluent monolayers were metabolically labeled for 16 h with 100 µCi/ml [³⁵S]Met + [³⁵S]Cys in the presence of 10 mM sodium butyrate and cell surface-biotinylated as in Ref. 26. Cells were washed twice with PBS supplemented with 1 mM CaCl₂ and 1 mM MgCl₂ and exposed to a 0.5 mg/ml Immunopure NHS-SS-biotin (Pierce) 20 min at 4 °C. Cross-linker was removed, and the same procedure was repeated once. The biotin reagent was quenched by incubation with 50 mM NH₄Cl in PBS for 10 min at 4 °C. Cells were washed with PBS and harvested. To recover the immunoprecipitated antigens, we supplemented the complexes with 10 µl of 10% SDS, boiled them for 5 min, diluted them with 600 µl of TPO extraction buffer, and centrifuged (10,000 × g, 3 min). Supernatant containing total TPO was incubated 2 h with avidin-agarose (Pierce). Biotinylated surface TPO and intracellular TPO were separated by centrifugation (10,000 × g, 3 min). The beads were washed four times with TPO extraction buffer and once with PBS, resuspended in electrophoresis buffer, and boiled for 5 min. The supernatants were analyzed by SDS-PAGE.

Cells were incubated in Ham's F-12 medium supplemented with 10 mM sodium butyrate, 1% FBS, and 5 µCi/ml [¹⁴C]aminolevulinic acid (NEN Life Science Products) for 16 or 48 h. Cells were then harvested, and TPO was immunoprecipitated as described previously before being analyzed by SDS-PAGE.

Cells were cultured in the presence of 10 mM sodium butyrate for 48 h and washed once with PBS. Cells were harvested by scraping in PBS and centrifuged at 200 × g for 7 min. Pellets were resuspended in 15 mM Tris-HCl, pH 7.4. Cells were ruptured by a Teflon glass homogenizer, and the suspension was centrifuged for 1 h at 110,000 × g.

Microsomal fraction pellets were solubilized by resuspension in 15 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1 mM KI. Microsomes prepared from CHO cells transfected with pcDNA3 were used as a negative control. Microsomal fractions were centrifuged (10,000 × g, 2 min), and the supernatant was used for the enzymatic assay. Extracts containing the same amounts of TPO1 and TPO2 and extracts from cells transfected with pcDNA3 were added to 1 ml of 40 mM guaiacol (Fluka Chimie, St. Quentin-Fallavier, France) and 67 mM sodium phosphate buffer, pH 7.5. The reaction was done at room temperature and initiated by adding H₂O₂ to reach a final concentration of 0.25 mM. Guaiacol oxidation was measured by absorbance at 470 nm and spectrophotometrically monitored every 30 s for 3 min. Variation of absorbance for each assay was calculated on the basis of a mean of three experiments.

RESULTS

To investigate the structure-function relations and the intracellular trafficking of TPO2 isoform as compared with the major variant TPO1, we generated two CHO cell lines stably expressing TPO2 and TPO1. CHO cells were transfected with either TPO2-, TPO1-pcDNA3 constructs or with pcDNA3 alone as control. After obtaining several clones expressing TPO2 and TPO1, we did the experiments in the presence of 10 mM sodium butyrate, which made TPO2 and TPO1 expression levels higher than those obtained without butyrate addition (data not shown).

After metabolic labeling of CHO TPO cells with [³⁵S]Met + [³⁵S]Cys for 5 h, we immunoprecipitated TPO2 and TPO1 with a combination of two TPO mAbs. As expected (28), TPO1 gave a 110-kDa specific band (Fig. 1, lane A), whereas TPO2 was a 98-kDa band on SDS-PAGE analysis (Fig. 1, lane B). The difference in apparent molecular mass between TPO1 and TPO2 isoforms reflects the lack of 57 amino acids in the extracytoplasmic domain of the TPO2 molecule. This difference may also be explained by the loss of a potential asparagine N-linked glycosylation site (Asn-569). These two bands were shown to specifically represent TPO1 and TPO2 isoforms, as they correspond to a binding with TPO mAbs and were not present in CHO cells transfected with pcDNA3 alone (Fig. 1, lane C). Similar results were obtained by Western blotting analysis of TPO1 and TPO2 with mAb 47 as first antibody (data not shown).

[View Larger Version of this Image (44K GIF file)]

During the subcloning of the transfected cells, we constantly found a higher amount of TPO1 than of TPO2 in the clones. Within the selected clones, quantification of each of the TPO band intensities after either steady-state labeling with [³⁵S]Met + [³⁵S]Cys for 48 h or Western blotting revealed a strong difference between TPO1 and TPO2 levels recovered. We always recovered 25-fold less TPO2 than TPO1. We believe that post-translational modification events or differences in processing of the TPO2 isoform through the cell occur. To get more insight on these findings, we used a structural approach of the TPO2 isoform before studying the intracellular trafficking of both isoforms. In all the experiments, we took into account the quantitative difference between TPO1 and TPO2 levels in CHO cells. We thus established experimental conditions to reliably compare similar amounts of both isoforms.

To investigate some of the structural aspects of TPO2, we did glycosylation studies. TPO2 was immunoprecipitated from cells metabolically labeled with [³⁵S]Met + [³⁵S]Cys for 5 h and then treated with either endo--N-acetylglucosaminidase H or peptide-N-glycanase F before analysis by SDS-PAGE (Fig. 2). TPO2 was sensitive to endo--N-acetylglucosaminidase H digestion (lane 2) and a similar pattern was obtained after deglycosylation with peptide-N-glycanase F (lane 3), suggesting that TPO2 bears high mannose-type structures exclusively.

[View Larger Version of this Image (34K GIF file)]

We then determined TPO2 immunological reactivity against a panel of mAbs directed to TPO1 (25). After [³⁵S]Met + [³⁵S]Cys metabolic labeling for 5 h, TPO1 and TPO2 were tested by immunoprecipitation with each of the 13 TPO mAbs. TPO1 immunoreactivity was observed with all the mAbs tested (Fig. 3A). For TPO2, a strong recognition was observed with mAb 47 and mAb 53, which led us to use these mAbs in immunoprecipitation experiments on TPO2 (Fig. 3B). Four other mAbs also clearly recognized TPO2 (mAbs 1, 15, 24, and 30), although 7 out of 13 had slight or no reactivity to the protein (Fig. 3B). That TPO2 exhibited either a weaker or no immunoactivity to some of the mAbs suggests that the molecule changed conformation.

[View Larger Version of this Image (52K GIF file)]

Given these structural changes, we questioned whether the TPO2 molecule is still recognized by TPO autoantibodies. The 57-amino acid region deleted in TPO2 is upstream from the main epitope recognized by autoantibodies, but the conformational changes previously deduced may be expected to modify the exposition of some autoepitopes. We thus determined the antigenic immunoreactivity of TPO2 against sera of patients with high titers of TPO autoantibodies. Using immunoprecipitation experiments, we showed that serum TPO autoantibodies specifically bound TPO2 (Fig. 4, lanes 2 and 3), whereas the serum of normal individuals did not (Fig. 4, lane 4). This suggests that TPO2 remained recognized by autoantibodies and exhibited all or part of the autoepitopes at the surface of the molecule.

[View Larger Version of this Image (39K GIF file)]

To investigate whether the loss of exon 10 and the structural changes in TPO2 alter the stability of the glycoprotein, we did pulse-chase studies and determined the half-life of TPO1 and TPO2. Cells were pulsed for 30 min in the presence of [³⁵S]Met + [³⁵S]Cys and chased at various times. Immunoprecipitated TPO1 and TPO2 were analyzed by SDS-PAGE (Fig. 5). [³⁵S]Met + [³⁵S]Cys TPO1 was still detected after 72 h of chase (Fig. 5A). In contrast, radiolabeled TPO2 was not detected at 22 h of chase (Fig. 5B), suggesting that TPO2 is degraded more rapidly than the major variant in the cell. Phosphoimager quantification of both TPO1 and TPO2 in several experiments revealed a shorter half-life for TPO2 (3 h) than for TPO1 (11 h) (Fig. 5C). The short half-life of the TPO2 isoform may explain the lower amounts of TPO2 than TPO1 recovered from CHO TPO cell lines, as in immunoprecipitation or Western blotting experiments.

[View Larger Version of this Image (29K GIF file)]

We then investigated whether the structural changes in the TPO2 variant modify the localization and the intracellular trafficking of the molecule. We thus evaluated the cell surface expression of the TPO1 and TPO2 isoforms by surface biotinylation experiments. Cells were [³⁵S]Met + [³⁵S]Cys metabolically labeled to steady state (48 h), and surface proteins were tagged by vectorial biotinylation with NHS-SS-biotin, a reagent that reacts with the amino groups of extracytoplasmic lysine residues. TPO was recovered by immunoprecipitation, released from the mAb by denaturation in SDS, and incubated with avidin-agarose. This procedure separated biotinylated surface TPO bound to avidin-agarose from intracellular TPO, which is hence inaccessible to the tagging agent. TPO1 was found predominantly in the intracellular compartments; a small fraction was present at the cell surface (Fig. 6, lanes 2 and 6). Quantification of the bands revealed that only 15% of TPO1 reached the cell surface. On the contrary, TPO2 remained in the intracellular compartment exclusively and did not reach the cell surface (Fig. 6, lanes 4 and 8).

[View Larger Version of this Image (23K GIF file)]

Another important post-translational event for TPO is its incorporation of heme, which occurs most likely in the ER. We were interested in the consequence of the modification of the three-dimensional structure of TPO2 on such incorporation, which is required for the enzyme to function. We thus evaluated whether TPO1 and TPO2 were able to bind to the heme by adding ¹⁴C-radiolabeled aminolevulinic acid heme precursor to cell cultures. ¹⁴C-labeled immunoprecipitated TPO1 was detected after 16 h of labeling and markedly increased at 48 h (Fig. 7, lanes 1 and 3). In the same experiment, no ¹⁴C-labeled TPO2 was detected, showing its inability to bind to heme (Fig. 7, lanes 2 and 4).

[View Larger Version of this Image (79K GIF file)]

As TPO2 did not reach the cell surface, extracts from microsomes were used to detect peroxidase activity. Guaiacol oxidation was measured for the two TPO variants and the control pcDNA3. As expected, we detected a significant and reproducible peroxidase activity for TPO1 as compared with the negative control (Fig. 8). Conversely, TPO2 guaiacol oxidation curves remained at the same basal level of those for the negative control. This result allows us to conclude that TPO2 is enzymatically inactive.

[View Larger Version of this Image (16K GIF file)]

DISCUSSION

To establish the role of TPO2 in the thyroid function, we studied the intracellular trafficking and the enzymatic activity of this protein. We transfected the cDNA of TPO1 and of the alternatively spliced form (TPO2) in CHO cells. We generated cell lines stably expressing significant levels of each protein but observed a lower cell expression of the alternatively spliced isoform. When we immunoprecipitated the cells metabolically labeled with [³⁵S]Met + [³⁵S]Cys, we found that the amounts of TPO2 recovered were 25-fold lower than those of TPO1. We first investigated the stability of the glycoprotein within the cell by pulse-chase experiments. Labeled TPO2 was not detected in the cell up to 5 h. TPO2 half-life was dramatically decreased as compared with TPO1 (3 versus 11 h), suggesting TPO2 degrades faster than TPO1. This could explain the low amount of TPO2 recovered from cells.

Cell surface biotinylation experiments showed that TPO1 is mainly intracellular and only 15% of the protein reaches the cell surface. Similar data were reported for the thyrocyte by Arvan's group (13), who found no more than 30% TPO at the plasma membrane. We demonstrated that no TPO2 reaches the cell surface, hence the whole protein remains intracellular. This suggests that TPO2 does not exhibit the structural features required to come out from the intracellular compartments and that it is retained in the cell. It was reported that differences in the processing of a protein may lead to a diminished trafficking to the cell surface and that incorrectly folded proteins are retained mainly in the ER (28-30).

Glycosylation is an important step in the maturation and trafficking of the cell surface proteins (31). The N-glycosylation of proteins occurs co-translationally in the ER. In this organelle, a series of trimming reactions is catalyzed by exoglycosidases, leading to high mannose-type structures. Oligosaccharide processing optionally continues in the Golgi apparatus, leading to the generation of complex-type structures. TPO1 synthetized by CHO cell lines bears only high mannose-type structures (27), but the small part of TPO1 present at the cell surface bears complex-type structures.² Given these observations, we did glycosylation studies of TPO2. Its susceptibility to digestion by endo--N-acetylglucosaminidase H indicates that TPO2 undergoes N-glycosylation processing through the ER and bears high mannose-type structures. Indeed, as TPO2 bears only high mannose-type structures, the protein is processed through the ER and is not trafficked beyond the cis-Golgi.

Incorrectly folded proteins are retained mainly in the ER (27, 32-34). This would also be true for TPO2 since the altered folding resulting from conformational changes may decrease its half-life in the cell and abolish its intracellular transport. To detect any three-dimensional modification of TPO2 compared with TPO1, we chose a structural approach by mapping the antigenic surface through a panel of 13 mAbs directed against TPO1 (25). Only one of them (mAb 47) recognizes a sequential epitope on the molecule; the others are directed against the conformational structure. More than half of the mAbs misrecognize TPO2, suggesting that their epitopes are misexposed. These findings lead to the conclusion that some TPO1 conformational epitopes were also recovered at the surface of TPO2, whereas others were not. This conclusion is consistent with a change of the three-dimensional structure of the molecule, likely with some parts structurally unchanged. If these data are analyzed with reference to the epitopic map of TPO1 (25), two clusters of mAbs still recognize TPO2, although the other two do not. It appears that epitopes in antigenic domains C and D remain exposed, although epitopes in domains A and B are in part masked due to the folding of the molecule, which differs from that of TPO1.

The lack of 57 amino acids in the extracellular region of TPO2 probably causes changes in the three-dimensional structure of the protein. The misfolding of TPO2 could also be potentialized by the loss of the potential site of N-glycosylation (Asn-569) in the deleted region. N-Glycans are important in ensuring the proper conformation of a protein. Alteration of a single N-linked glycosylation site may impair the protein function and lead to its degradation (35). However, it is not sure that the loss of one potential site of N-glycosylation is sufficient to totally explain the changes in immunoreactivity features of TPO2. Indeed, we reported that removal of N-glycans from TPO1 did not decrease its binding to mAbs (36), but this sugar removal occurred after the formation of a correct three-dimensional structure of the molecule.

The conformational changes in TPO2 are not sufficient to abolish the exposure of all the autoepitopes at the surface of the molecule. We demonstrated that TPO2 retains its ability to bind autoantibodies of the two patients with Hashimoto's disease. The epitopes expressed by domain A and B on the epitopic map of TPO1 have been reported to react strongly with autoantibodies, whereas epitopes from domains C and D are less involved in this recognition (25, 37). On the other hand, the linear epitope specific for mAb 47 represents less than 20% of the human autoantibody repertoire (38). However, our results suggest that at least a part of autoepitopes are expressed at the surface of TPO2. The domain(s) recognized by the autoantibodies remain to be determined, but it will be difficult to purify enough quantities of materials because of the low quantities of TPO2 found in the cells. Note that our results on the recognition of TPO2 by autoantibodies appeared to contradict those of Finke et al. (39) who used a two-site immunoassay and recombinant TPO2 expressed in insect cells. This contradiction may result from a different folding of TPO2 in insect cells and mammal cells or from an incomplete or improper glycosylation of the protein (40). An alternative hypothesis is an aggregation of TPO2 related to its overexpression in this system. On the other hand, Tonacchera et al. (23) reported that a fragment from human TPO2 lacking amino acids Gly-533-Gly-590 produced by bacterial cells was still recognized by TPO autoantibodies and that the 57 spliced-out residues are probably not involved in antibody binding.

TPO is a heme-containing glycoprotein. To get more insight on the structure-function relationship of the two TPO variants, we investigated the incorporation of heme into cultured cells labeled with [¹⁴C]aminolevulinic acid. As expected, TPO1 incorporated labeled heme, and this incorporation increased with time. Heme incorporation likely occurs in the ER, as reported for myeloperoxidase precursors, which incorporated heme in the ER before acceding to final processing (41). In contrast to TPO1, TPO2 does not incorporate labeled heme despite long exposure to the isotope and an increase in TPO2 synthesized. The lack of incorporation of heme demonstrates that misfolded TPO2 does not exhibit the structural features required to bind to it. Within the 57 amino acids deleted in the TPO2 molecule, His-586 is considered as a putative proximal heme binding site. In addition to the improper folding, lack of His-586 may account for the inability of TPO2 to bind to heme.

Failure of incorporation of heme may in turn affect TPO2 trafficking in the cell. It was reported that inhibiting heme synthesis with succinyl acetone modifies the intracellular trafficking of myeloperoxidase (42). As suggested before, TPO binding to heme is expected to occur in the ER. It is not excluded that incorporation of heme is required to achieve the proper conformation of the protein that allows it to come out from the ER. That TPO2 does not bind to heme may account for the non-sorting of TPO2 from the ER.

Binding to heme is obviously required for TPO to be enzymatically active. We evaluated TPO activity by measuring peroxidase activity from cell microsomes by using a guaiacol assay. As expected, non-heme-bound TPO2 exhibits no enzymatic activity. Conversely, peroxidase activity is detected in the microsomal fraction of TPO1. This is linked with the heme incorporation in the intracellular compartment where TPO is mainly found. The binding of heme to intracellular TPO1 by achieving its proper folding may thus make the protein enzymatically active. Within the thyrocyte, these events may account for the smaller intracellular iodination of thyroglobulin (18).

In this paper, we demonstrated how normal and alternatively spliced isoforms may differ in both structure-function relation and intracellular trafficking. TPO1 enzymatic activity and immunological features have been extensively studied (1, 5, 25). Our work provides insight into the intracellular trafficking of TPO1 through the cell in relation with function and heme binding. The present data provide some advances in the knowledge of the alternatively spliced TPO2 variant. We demonstrate that TPO2 is enzymatically inactive. Although such a result was expected, the few previous studies suspected it on the basis of indirect findings (21). Furthermore, we show that the lack of enzymatic activity of TPO2 is related to the failure of incorporation of heme. This may be either the cause or the consequence of an improperly folded protein, which accounts for the changes observed in immunological reactivity to mAbs, which is related to the epitopic map of the molecule. Because of these structural changes, TPO2 is unable to reach the cell surface and remains intracellularly, where it is rapidly degraded as shown by its markedly lower half-life than that of the normal variant TPO1. Considering those structural and functional characteristics, it stands to reason that TPO2 is not implicated in thyroid hormone synthesis, but a hypothetical role for this isoform in physiology or in thyroid diseases cannot be totally discarded.