Biosynthesis and metabolism of 2-iodohexadecanal in cultured dog thyroid cells.

2-Iodohexadecanal (2-IHDA) is a major thyroid iodolipid. It mimics the main regulatory effects of iodide on thyroid metabolism: inhibition of H2O2 production and of adenylyl cyclase. The biosynthesis of 2-IHDA and its metabolism have been investigated in cultured dog thyroid cells maintained in a differentiated state by forskolin. Incubation of these cells with [9,10-3H]hexadecan-1-ol or [9,10-3H]palmitic acid labeled several phospholipids, but [9, 10-3H]hexadecan-1-ol was selectively incorporated into plasmenylethanolamine. In the presence of an exogenous H2O2 generating system (glucose oxidase), iodide induced the production of [9,10-3H]2-IHDA from [9,10-3H]hexadecan-1-ol-labeled cells but not from [9,10-3H]palmitic acid-labeled cells. 2-IHDA was also generated during the lactoperoxidase-catalyzed iodination of brain and heart plasmalogens, and of ethyl hexadec-1-enyl ether, a synthetic vinyl ether-containing compound. Taken together, these results show that thyroid 2-IHDA is derived from plasmenylethanolamine via an attack of reactive iodine on the vinyl ether group. 2-Iodohexadecan-1-ol (2-IHDO) was also detected in these studies; it was formed later than 2-IHDA, and thyroid cells converted exogenous 2-IHDA into 2-IHDO in a time-dependent way. The ratio of 2-IHDO/2-IHDA increased with H2O2 production and decreased as a function of iodide concentration. An aldehyde-reducing activity was detected in subcellular fractions of the horse thyroid. No formation of 2-iodohexadecanoic acid could be detected. Reduction into the biologically inactive 2-IHDO is thus a major metabolic pathway of 2-IHDA in dog thyrocytes.

It has been known as early as 1955 that the iodinating capacity of the thyroid gland is not restricted to tyrosyl residues in thyroglobulin. Some of the unknown iodinated compounds were characterized as lipids. The major iodolipid formed in the horse thyroid incubated in vitro with iodide was identified as 2-iodohexadecanal (2-IHDA) 1 (1). This compound was also detected in the rat, the dog, and the human thyroid (1,2). Another ␣-iodoaldehyde, 2-iodooctadecanal, was also detected in the rat thyroid and in the dog thyroid where it was even more abundant than 2-IHDA (1). The pleiotropic inhibitory actions of excess iodide on the thyroid are well known and constitute an homeostatic mechanism of protection against thyrotoxicosis in case of sudden exposure to an abundant supply of iodine (3). It was shown recently that 2-IHDA mimics several of the actions of iodide. In particular, it directly inhibited the H 2 O 2 -producing NADPH oxidase, which is the target of the Wolff-Chaikoff effect (4), in thyroid porcine membranes (5) and adenylyl cyclase in human thyroid membranes (6). This latter effect shared several features of the inhibition of cAMP formation by iodide (7)(8)(9) in intact cells or in membranes prepared from thyroid tissue exposed to iodide. Furthermore a comparison with various 2-IHDA analogues demonstrated that the effects of 2-IHDA are highly specific and identified the critical role of two structural determinants: the aldehyde function and the iodine at C2 (6,10). These data suggest that 2-IHDA is the mediator of the regulatory actions of iodide on the thyroid. So far little attention has been paid to the biosynthesis and metabolic fate of 2-IHDA. The attack of reactive iodine on the vinyl ether group of plasmalogens appears as a rather straightforward mechanism of 2-IHDA formation. The present study was started in order to test that hypothesis.

EXPERIMENTAL PROCEDURES
Materials-1 H NMR spectra were recorded in CDCl 3 on a Bruker WM 250 spectrometer and are reported in ppm from internal tetramethylsilane on the ␦ scale. Infrared spectra were taken with a Bruker IFS 25 instrument, and the samples were examined as deposited films on NaCl discs or in chloroform solution. Electron impact mass spectra were recorded on a VG Micromass 7070 or on a FISONS VG AUTO-SPEC spectrometer. In both cases, peak intensities are expressed as percentages relative to the base peak. Thin layer chromatography analyses were performed on 0.25-mm POLYGRAM silica gel SILG/UV 254 precoated plates (MACHEREY NAGEL). Unless otherwise stated, column chromatographies were performed over silica gel (MN Kieselgel 0.04 -0.063 mm) using flash technique or over florisil 0.15-0.25 mm (Merck). Glucose oxidase from Aspergillus niger (type 5), D-amino acid oxidase from porcine kidney (type 2), bovine serum albumin (fatty acid poor), thyrotropin, standard phospholipids for TLC, and phosphatidylethanolamine (P9137) or -choline (P9513) containing 60 or 30% plasmalogens, respectively, were purchased from Sigma. Lactoperoxi-dase was purchased from Boehringer Mannheim. [9, H]Palmitic acid (50 Ci/mmol) and [ 125 I]NaI were provided by Amersham Corp. Sep Pak silica gel cartridges were provided by Waters Associates. HPTLC aluminum sheets silica gel 60 (20 ϫ 20 cm) for nano-TLC were provided by Merck. The various 2-iodoalkylaldehydes, 2-iodohexadecan-1-ol, and 2-iodohexadecanoic acid were synthesized according to the procedures described previously (5,10).
Separation and Detection of Iodolipids-First the lipid extracts were purified on silica gel cartridges (Sep Pak R ) from which they were eluted with chloroform (10 ml). An aliquot of the sample was chromatographed on HPTLC silica gel plates over 55 or 120 mm, with n-hexane/diethyl ether/acetic acid (8:2:0.1). The radiolabeled lipids were detected by autoradiography of the TLC plate. In this system, the R F of the standards 2-IHDA and 2-IHDO were 0.83 Ϯ 0.05 and 0.33 Ϯ 0.04, respectively (mean Ϯ S.D. of seven elutions). The detection of 2-IHDA by RP-HPLC was performed on a octadecyl silica gel column (4.6 ϫ 250 mm, 5 m) eluted isocratically with acetonitrile (1 ml/min). The retention times of 2-IHDA, 2-IHDO and 2-iodohexadecanoic acid detected spectrophotometrically at 276 nm were, respectively, 11.0 Ϯ 0. GC-MS analyses were performed using a capillary GC (TRACOR 540) coupled to an ion trap detector mass spectrometer (FINNIGAN ITD 800) on a 1.5 m ϫ 0.25 mm fused silica gel column (Wcot) containing chemically bonded polyethylene glycol (CP-Wax 52CB, Chrompack). The following elution program was applied: initial temperature, 100°C (1 min); final temperature, 200°C; rate of rise, 30°C/min. The temperature of the transfer line was 220°C. Helium was used as carrier gas (pressure, 15 p.s.i.). Injections were performed directly on-column. Electron impact (EI) with an electron beam energy of 70 eV or chemical ionization with ammonia were both used. 2-IHDA was analyzed either as such or after derivatization into its O-(2,3,4,5,6-pentafluorobenzyl)oxime derivative, prepared as follows: a solution of 0.6 mg (2.4 mol) of O- (2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride in water (100 l) was added to 2-IHDA (15 g, 41 nmol) dissolved in ethanol (100 l). The mixture was vortexed for 5 min and allowed to stand at room temperature for 25 min. The mixture was diluted with water (0.4 ml) and extracted with hexane (3 ϫ 0.5 ml). The hexane phase was analyzed by GC-MS using the conditions described above.
Phospholipid Analysis-Phospholipids (PLE, Sigma P 9137) were dissolved in chloroform/methanol (2:1) and analyzed with a Waters Associates HPLC apparatus equipped with a Waters 510 pump, a thermostatic regulator Waters TCM, and a variable wavelength Merck Hitachi L-4000 UV detector (detection at 205 nm). Instrument control, data acquisition, and data processing were provided by a Waters Millennium 2.1 data system. Phospholipid molecular species were separated with RP-HPLC on a RP-18 Lichrospher 100 (4 ϫ 250 mm, 5 m) column (Merck) as the stationary phase and a mobile phase of methanol/acetonitrile/water (90.5:7.0:2.5) containing 20 mM choline chloride by isocratic elution at 1.5 ml/min, at 35°C (15). Liquid secondary ion mass spectrometry (LSIMS) of phospholipids was performed on a FISONS VG AUTOSPEC spectrometer. Phospholipids were dissolved in chloroform/methanol (1:1) and mixed with glycerol used as the matrix. For acid methanolysis, PLE (100 g, Sigma P 9137) was dissolved in chloroform (100 l) and treated with 2.5 N methanolic HCl (200 l) at room temperature for 30 min under magnetic stirring (16). After complete evaporation of the solvents under nitrogen, the residue was dissolved in chloroform/methanol (2:1) (500 l) and analyzed by RP-HPLC using the conditions described above.
Culture of Dog Thyroid Cells-Minced dog thyroid tissue was digested by collagenase type I (120 units/ml) and deoxyribonuclease (100 units/ml) in basal Eagle's medium for 60 min at 37°C (17). The resulting suspension of follicles was filtered through nylon mesh, separated from isolated cells by three centrifugations (2 min at 770 ϫ g followed by 2 min at 100 ϫ g twice) and seeded on 6-cm diameter Petri dishes at a density of 350,000 cells/cm 2 in a mixture of Dulbecco's modified Eagle's medium/Ham's F-12 medium/MCDB 104 medium (2:1:1, v/v/v) supplemented by 2 mM sodium pyruvate, 5 g/ml insulin, 40 g/ml ascorbic acid, 100 units/ml penicillin, 100 g/ml streptomycin, and 2.5 g/ml amphotericin B. The Petri dishes were maintained in a watersaturated incubator at 37°C in an atmosphere of 5% CO 2 . After 24 h of cell spreading, the medium was replaced by the same mixture supple-REACTION 1. Synthesis of hexadecanal diethyl acetal.
Labeling of Dog Thyroid Cell Plasmalogens-Dog thyrocytes were labeled with [9,10-3 H]hexadecan-1-ol (100 Ci/dish) for the last 24 h of culture (19) in the presence or the absence of methimazole (500 M). Cells were scraped in 2 ml of glacial phosphate-buffered saline, centrifuged at 4°C for 5 min at 770 ϫ g and homogenized in 1 ml of methanol/chloroform (2:1) at 0°C. The homogenate was centrifuged at 4°C for 5 min at 2770 ϫ g. The resulting pellet was homogenized a second time at 0°C in 1 ml of methanol/chloroform (2:1), and the homogenate was centrifuged at 4°C for 5 min at 2770 ϫ g. The supernatants of the two centrifugations were pooled, and the solvent compo-sition was adjusted (methanol/chloroform/water, 1:1:0.9) in order to allow phase separation (20). Aliquots of the total lipid extracts were resolved on HPTLC aluminum silica gel 60 plates (Merck, 200 ϫ 200 mm) by a two-step one-dimensional elution. The first elution was developed over 7 cm with chloroform/methanol/acetic acid/water (58:38: 0.9:2.7) (21) and was followed by a second one using ethyl acetate/ isooctane/acetic acid/water (58:27:11:4) over 12 cm. Standard lipids co-migrating with the samples were visualized by iodine vapor, and the corresponding zones in sample lanes were cut, eluted, and counted in a ␤-scintillation counter. To quantify plasmalogen labeling on the sn 1 position, the vinyl ether bond of the plasmalogen was cleaved by acidic treatment of the plate (15 min of HCl 37% vapor) after the first elution; the radioactivity removed by HCl corresponds to plasmalogens (22).
Measurement of Aldehyde Reductase and Alcohol Dehydrogenase Activities-Aldehyde reductase and alcohol dehydrogenase activities were determined in horse thyroid subcellular fractions by monitoring the decrease in the absorbance at 343 nm due to the oxidation of NAD(P)H in the presence of 2-IHDA. Horse thyroid membranes (6) (30,000 ϫ g for 10 min) or the supernatant of this preparation solubilized by Triton X-100 (1%) inactivated or not by incubation 3 min at 100°C were incubated at 0.5 mg protein/ml in 500 l of sodium phosphate buffer (0.1 M, pH 7.4) containing 1 mM EDTA and 0.1 mM NADH (for the assay of alcohol dehydrogenase) or 0.1 mM NADPH (for the assay of aldehyde reductase) in the presence of 20 M 2-IHDA, 2-IHDO, 2-iodododecanal, dodecanal, 2-iodooctanal, or octanal. The kinetics of NADH or NADPH disappearance were performed over 30 min at 37°C on a Uvikon 930 (Kontron Instruments) spectrophotometer. The molar extinction coefficient of NAD(P)H at 343 nm is 6200 M Ϫ1 ⅐cm Ϫ1 .

RESULTS
The incubation of a crude preparation of bovine brain PLE or of bovine heart phosphatidylcholine containing plasmenylcholine with lactoperoxidase, radioiodide, and H 2 O 2 generated an iodinated material co-migrating with synthetic 2-IHDA in HPTLC (Fig. 1a)  Lysophosphatidylethanolamine resulting from the cleavage of the alkenyl chain had a retention time less than 10 min and therefore is not apparent on the chromatogram. The LSIMS analysis of phospholipids remaining after iodination showed that the major peak (t R ϭ 46.0 min) was constituted of diacyl derivatives. C, RP-HPLC analysis of PLE after acid methanolysis. c, GC-MS characterization of the 2-IHDA generated by lactoperoxidase iodination of EHDE. 2-IHDA, purified by NP-HPLC, was analyzed by gas chromatography coupled to a Finnigan ion trap detector mass spectrometer, either as such (A) or after derivatization into its O-pentafluorobenzyloxime derivative (B and C). A capillary column (CP-Wax 52CB, 1.5 m ϫ 0.25 mm) was used with a temperature gradient from 100 to 200°C at a rate of 30°C/min, as described under "Experimental Procedures." A, the mass spectrum of 2-IHDA in the EI mode presented an ion at m/z 239 corresponding to the loss of iodine from the molecular ion and an ion at m/z 170 resulting from a Mac Lafferty rearrangement (40). B, the mass spectrum of the O-pentafluorobenzyloxime derivative of 2-IHDA in the EI mode presented an ion at m/z 434 corresponding to the loss of iodine from the molecular ion and an ion at m/z 181 corresponding to the pentafluorotropylium cation. C, the spectrum in the chemical ionization mode (NH 3 as collision gas) of this derivative presented a quasi-molecular ion at 562 Da and an intense fragment peak at m/z 434.
was not produced when synthetic PE was incubated under the same experimental conditions (Fig. 1a). The iodination of the PLE preparation by lactoperoxidase was associated with the consumption of plasmenylethanolamine and not phosphatidylethanolamine (Fig. 1b, B). To establish the specific role of the vinyl ether bond of plasmalogens in the generation of 2-IHDA, EHDE (23) was synthesized. Incubation of EHDE with lactoperoxidase and radioiodide generated an iodinated material co-eluting with synthetic 2-IHDA in HPTLC (Fig. 1a) and in RP-HPLC (not shown). The identity of this compound as 2-IHDA was confirmed by GC-MS (Fig. 1c).
When dog thyrocytes in culture were incubated with radioiodide, they produced two iodolipids co-eluting in RP-HPLC with synthetic 2-IHDA (Fig. 2) and 2-iodooctadecanal (data not shown). This production was amplified in the presence of the H 2 O 2 generating system glucose-glucose oxidase (Fig. 2): 2.4 units/ml glucose oxidase induced a 11.7 Ϯ 5.0-fold increase (mean Ϯ S.D. of five experiments) of the radioactivity incorporated into the neutral lipids fraction containing 2-IHDA. As control, no radioactivity was found in the lipid extract when nonthyroid cells (COS-7 cells) were incubated with radioiodide and glucose oxidase (data not shown). Using [9,10-3 H]hexadecan-1-ol to label plasmalogens in the sn 1 position (19), we investigated if they were the source of 2-IHDA in dog thyrocytes. When the cells were incubated with [9,10-3 H]hexadecan-1-ol or [9,10-3 H]palmitic acid, several classes of lipids were labeled (Table I). The labeling patterns were similar, except for a larger incorporation of [9,10-3 H]hexadecan-1-ol in the fraction co-migrating with the PE standard. This difference was abolished following exposure of the plates to HCl vapor, which is known to cleave the vinyl ether bond of plasmalogens (22). The RP-HPLC chromatogram of phospholipids remaining after acid methanolysis of the PLE mixture was similar to the chromatogram obtained after lactoperoxidase-catalyzed iodination of PLE mixture with disappearance of the same plasmalogen peaks (Fig. 1b, C). The results are thus consistent with a selective incorporation of [9,10-3 H]hexadecan-1-ol at the sn 1 position of plasmenylethanolamine. On the basis of the differ-   [9, H]hexadecan-l-ol] Dog thyrocytes, cultured as described under "Experimental Procedures," were labeled with [9,10-3 H]hexadecan-l-ol (100 Ci/dish) or [9,10-3 H]palmitic acid (100 Ci/dish) for the last 24 h of the culture. The cells were scraped in 2 ml of glacial phosphate-buffered saline, centrifuged, and homogenized in methanol/chloroform (2:1) as described under "Experimental Procedures." After extraction of the lipids following Bligh and Dyer (20), the total extract was analyzed on HPTLC plates by the two-step elution described under "Experimental Procedures." Before the second elution, some plates were dried and submitted to a 15-min HCl vapor treatment. After the second elution, the standards, coeluting with the sample, were revealed by iodine vapor, cut, eluted by methanol/water (2:1), and counted in a liquid scintillation counter. Results are expressed as percentages of the total radioactivity incorporated in the cell extract, before and after HCl treatment. LPC, lysophosphatidylcholine; PC, phosphatidylcholine; PS, phosphatidylserine; PI, phosphatidylinositol; PE, phosphatidylethanolamine; PA, phosphatidic acid; DAG, diacyl glycerol; TG, triglycerides; NL, neutral lipids. Incubation of [9,10-3 H]hexadecan-1-ol-labeled dog thyrocytes with both iodide (100 M) and glucose oxidase (2.4 units/ ml) led to the generation of a tritium-labeled material coeluting with synthetic 2-IHDA in RP-HPLC and representing 0.11-0.56% of the cell-associated [9,10-3 H]hexadecan-1-ol labeling (range of five experiments) (Fig. 3). Production of this material was inhibited by methimazole at a concentration (500 M) that is known to inhibit iodolipid formation in horse thyroid slices (1). That compound was not detectable when [9,10-3 H]palmitic acid-labeled dog thyrocytes were incubated with iodide and glucose oxidase (Fig. 4). In these experiments, the radioactivity recovered in the total lipid extract was 22 Ϯ 3 10 6 and 41 Ϯ 7 10 6 cpm, respectively, with hexadecan-1-ol and palmitic acid labeling, indicating that the negative result with [9,10-3 H]palmitic acid was not due to a lower incorporation into the thyrocytes.
To confirm its identity to 2-IHDA, the tritium-containing peak co-eluting with synthetic 2-IHDA in RP-HPLC was rechromatographed in NP-HPLC (Fig. 5a) and submitted to HPTLC on silica gel plates (Fig. 5c). NP-HPLC resolved two radioactive peaks, one of which co-eluted with a 2-IHDA standard (Fig. 5a). In the same way, two spots were obtained in TLC, one of which had the same R F as 2-IHDA (Fig. 5c). A similar result was obtained when analyzing directly in NP-HPLC (Fig.  5b) or TLC (Fig. 5c); the radioiodinated product formed in dog thyrocytes incubated with [ 125 I]NaI. The same amount of 2-IHDA was produced when TSH was omitted from the incubation medium (data not shown). The second peak (or spot) had a behavior identical to that of synthetic standard of 2-IHDO in three different chromatographic systems: RP-HPLC, NP-HPLC, and TLC. The 2-IHDO/2-IHDA ratio was variable from one experiment to the other (10.2 Ϯ 5 with 2.4 units/ml glucose oxidase, mean Ϯ S.D. of seven experiments) and was critically dependent on the experimental conditions (see below). In none of the experiments was it possible to detect a radioactive peak at the retention time (6.7 min) of 2-iodohexadecanoic acid (Fig.  5b); this compound was also undetectable in RP-HPLC where its retention time was 8.1 Ϯ 0.7 min (Fig. 2).
Two types of experiments were performed to determine if 2-IHDO is formed by reduction of 2-IHDA; the kinetics of appearance of the two lipids were compared, and the conversion of exogenous [9,10-3 H]2-IHDA into [9,10-3 H]2-IHDO by thyrocytes was monitored. The production of [ 125 I]2-IHDA by dog thyrocytes incubated with radioiodide in the presence of glucose oxidase was maximal at the earliest time studied (10 min) and decreased over time. In parallel to this decrease, there was a progressive increase in 2-IHDO (Fig. 6a). When   (Fig. 6b). No conversion into [9,10-3 H]2-IHDO was obtained when synthetic [9,10-3 H]2-IHDA was incubated in the same conditions but in the absence of cells (data not shown). The 2-IHDO/2-IHDA ratio decreased with iodide concentration (Fig. 6c) and increased with glucose oxidase activity (Fig. 6d). Another H 2 O 2 generating system, D-alanine ϩ D-amino acid oxidase, had the same effect on the 2-IHDO/2-IHDA ratio (data not shown). TSH, present in the incubation medium, did not affect the 2-IHDO/2-IHDA ratio (data not shown). To test the presence of aldehyde reductase activity, the consumption of NADH or NADPH by subcellular fractions of horse thyroid tissue was measured spectrophotometrically in the presence or the absence 2-IHDA. Table II shows that 2-IHDA induced a consumption of NADPH by the supernatant (30000 ϫ g for 10 min) of an homogenate of horse thyroid but not by the membranes of the same preparation, whereas with NADH an activity was found in both fractions. The consumption of NAD(P)H was strongly reduced after boiling and was not observed in the presence of 2-IHDO, indicating the role of the aldehyde function (Table II). The activity was not specific for 2-IHDA; at equimolar concentrations, 2-IHDA, 2-iodododecanal, and 2-iodooctanal induced a very similar consumption of NAD(P)H by the thyroid membranes and supernatant, whereas that consumption was about 50% lower with octanal and dodecanal (data not shown). DISCUSSION We have previously proposed the hypothesis that 2-IHDA is formed as the result of the attack of a reactive iodine species (I ϩ or I°) on the vinyl ether group of plasmalogens (1). This hypothesis was supported by preliminary data showing the for- mation of 2-IHDA when a bovine brain preparation containing both phosphatidylethanolamine and plasmenylethanolamine was incubated with lactoperoxidase, iodide, and H 2 O 2 (1). We have confirmed and extended this result in the present study. In particular, 2-IHDA could also be generated from a bovine heart preparation containing phosphatidylcholine and plasmenylcholine but was not formed from synthetic PE. The lactoperoxidase-catalyzed formation of 2-IHDA from the bovine brain PLE preparation was associated with the disappearance of plasmenylethanolamine but not of phosphatidylethanolamine. As assessed by GC-MS, 2-IHDA was obtained when synthetic EHDE, which contains a vinyl ether group, was submitted to lactoperoxidase-catalyzed iodination.
The biosynthesis of 2-IHDA was also investigated in thyroid cells. Because plasmalogens are biosynthesized from long chain fatty alcohols (24,25), [9,10-3 H]hexadecan-1-ol was used to label them in dog thyrocytes. Actually [9,10-3 H]hexadecan-1-ol labeled several classes of lipids, in a way quite similar to [9,10-3 H]palmitic acid. This is not surprising because in other experimental models a substantial proportion of administered hexadecan-1-ol was oxidized into palmitic acid (19,26,27). However, the incorporation into the TLC fraction comigrating with the PE standard was greater for [9,10-3 H]hexadecan-1-ol than for [9,10-3 H]palmitic acid, and this difference was abol-ished after acid treatment to which plasmalogens are known to be sensitive (22). This indicates that part of the added [9,10-3 H]hexadecan-1-ol was truly incorporated into plasmenylethanolamine. By comparison, there was little incorporation of the label into plasmenylcholine; this discrepancy can be explained by the known abundance of plasmenylethanolamine but not plasmenylcholine in thyroid phospholipids (28).
Because [9,10-3 H]2-IHDA was formed in dog thyroid cells labeled with [9,10-3 H]hexadecan-1-ol but not with [9,10-3 H]palmitic acid, it must derive from plasmenylethanolamine. The need to add an exogenous H 2 O 2 generating system in order to detect [9,10-3 H]2-IHDA can be explained by the limited labeling of plasmalogens and also by the low iodinating capacity of our model of cultured dog thyrocytes (29,30), probably due to an ineffective utilization of H 2 O 2 , a major limiting step in thyroid iodination. Indeed these cells are able to produce H 2 O 2 , but the extracellularly formed H 2 O 2 is diluted in the incubation medium (31). In dog thyroid slices, where endogenous H 2 O 2 is released in the follicular lumen, significant amounts of 2-IHDA are formed in the absence of an exogenous H 2 O 2 generating system (1). Taken together our results support the concept that the biosynthesis of 2-IHDA in the thyroid results from the attack of a reactive iodine species on the vinyl ether group of plasmenylethanolamine (Fig. 7). FIG. 7. Biosynthesis and catabolism of 2-iodohexadecanal. Hexadecan-1-ol is selectively incorporated at the sn 1 position of plasmenylethanolamine during its de novo biosynthesis. Iodination of the vinyl ether group of plasmenylethanolamine generates an unstable iodinated derivative, which breaks into lysophosphatidylethanolamine and 2-iodohexadecanal. 2-Iodohexadecanal is reduced into 2-iodohexadecan-1-ol by a mechanism that is down-regulated by iodide and enhanced by H 2 O 2 .

TABLE II
Aldehyde reductase activity in the horse thyroid Horse thyroid membranes or the supernatant of the membrane preparation (0.5 mg protein/ml) were incubated at 37°C for 30 min in the presence of NADH or NADPH (0.1 mM) with or without 2-IHDA or 2-IHDO (20 M). The disappearance of NAD(P)H was followed at 343 nm. The activity was calculated from the negative slope of NAD(P)H disappearance, which was linear and expressed as nmol NAD(P)H consumed/min ⅐ mg protein. As control, 2-IHDA did not interfere with the measure. ND, not determined. There are several pathways of aldehyde metabolism in mammalian cells. They can be oxidized into the corresponding carboxylic acids by various aldehyde dehydrogenases and aldehyde oxidases. Multiple forms of these enzymes have been identified that differ inter alia by their subcellular localization, cytosol versus mitochondrion (32,33). They can also be conjugated to glutathione by cytosolic glutathione transferases (34,35). Reduction into alcohol by enzymes such as aldo-keto reductases or alcohol dehydrogenase represents a third possibility. For instance, the lens aldose reductase has a lower K m for 4-hydroxynonenal, a product formed during lipid peroxidation, than for glucose (36). The relative importance of these pathways is variable from one cell type to the other. For instance, whereas glutathione transferase represents the major pathway in normal hepatocytes, aldehyde dehydrogenase plays a major role in some hepatoma cell lines; other lines have prominent NADPH-dependent aldehyde reductase activity and/or NADHdependent alcohol dehydrogenase activity (37). Reduction into the corresponding alcohol is clearly the major metabolic transformation of 2-IHDA in the thyroid, and there is no evidence of a significant oxidation into 2-iodohexadecanoic acid. This reduction seems to involve both a soluble NADPH-dependent aldehyde reductase and a NADH-dependent alcohol dehydrogenase present in membranes as well as in supernatant. These activities are not specific for ␣-iodoaldehydes; indeed octanal and dodecanal were also reduced, although at a lower rate than 2-iodooctanal, 2-iodododecanal, or 2-IHDA. Because unlike 2-IHDA, 2-IHDO is biologically inactive both on H 2 O 2 production (5, 6) and on adenylyl cyclase (10), reduction clearly represents an inactivating pathway. One intriguing finding was the increased conversion of 2-IHDA into 2-IHDO as a function of the rate of H 2 O 2 generation. This is reminiscent of the observation that the lens aldose reductase is activated by reactive oxygen species (38,39). The physiological significance of this effect in the thyroid remains unclear.