Defective Thyroglobulin Synthesis in an Experimental Rat Thyroid Tumor LACK OF MEMBRANE-BOUND SIALYLTRANSFERASE ACTIVITY

SUMMARY The incorporation of carbohydrates has been studied in vitro in the experimental rat thyroid tumor l-lC2. The uptake of iV-[aH]acetyhnannosamine, a precursor of sialic acid, is less than 5% of that in normal thyroid gland; no label is found in 19 S thyroglobulin or its precursors either in the soluble or in the solubilized proteins. N-1 3H]Acetylglucosa-mine is incorporated at a slower rate than in normal thyroid gland and is present in the membrane-bound thyroglobulin; there is no conversion into sialic acid in the soluble thyroglobulin of tumor as occurs in normal thyroid. [14C]Mannose and [14C]galactose are incorporated at an almost normal rate and p4C]fucose, at a subnormal rate; they are present mainly in the particle-bound proteins as ~18 S thyroglobulin and its precursors. Measurement of sialyltransferase activity showed that, with desialylated thyroglobulin as acceptor, the 22,000-105,000 X g pellet of tumor contains ~3% of normal sialyltransferase activity, and the soluble fraction

??0014 SUMMARY The incorporation of carbohydrates has been studied in vitro in the experimental rat thyroid tumor l-lC2. The uptake of iV-[aH]acetyhnannosamine, a precursor of sialic acid, is less than 5% of that in normal thyroid gland; no label is found in 19 S thyroglobulin or its precursors either in the soluble or in the solubilized proteins. N-1 3H]Acetylglucosamine is incorporated at a slower rate than in normal thyroid gland and is present in the membrane-bound thyroglobulin; there is no conversion into sialic acid in the soluble thyroglobulin of tumor as occurs in normal thyroid.
[14C]Mannose and [14C]galactose are incorporated at an almost normal rate and p4C]fucose, at a subnormal rate; they are present mainly in the particle-bound proteins as ~18 S thyroglobulin and its precursors.
Measurement of sialyltransferase activity showed that, with desialylated thyroglobulin as acceptor, the 22,000-105,000 X g pellet of tumor contains ~3% of normal sialyltransferase activity, and the soluble fraction contains -25 % of normal activity.
Desialylated fetuin and orosomucoid are generally better acceptors than desialylated thyroglobulin for tumor sialyltransferase.
The findings indicate that defective thyroglobulin release and absence of incorporation of sialic acid into thyroglobulin coexist in this tumor.
The very low activity of thyroglobulinspecific sialyltransferase activity in the tumor particulate fraction accounts for the failure of sialic acid incorporation into thyroglobulin and may explain the defect in thyroglobulin release.
The transplantable rat thyroid tumor (1)) Wollman line l-lC2, shows very little formation of soluble thyroglobulin; most of the protein immunochemically and structurally related to thyroglobulin remains bound to cell membranes (2, 3). Since thyroglobulin is a glycoprotein (4)  erally extracellular products of secretion (5, 6) it was considered that a defect in carbohydrate incorporation into the oligosaccharide chains of thyroglobulin might be responsible for its incomplete maturation and release. A likely candidate for such a defect is N-acetylneuraminic acid (sialic acid), a charged terminal residue of thyroglobulin oligosaccharides and known to be important in preventing tissue uptake of circulating glycoproteins (7).
We recently showed (8) that normal thyroid in vitro incorporates N-acetylmannosamine (ManNAc) into thyroglobulinbound sialic acid, and incorporates N-acetylglucosamine (Glc-NAc) into thyroglobulin-bound GlcNAc and, to a lesser extent, into sialic acid. Similar studies with tumor l-lC2, reported here showed no formation of thyroglobulin-bound sialic acid from these precursors, whereas all other carbohydrates normally present in thyroglobulin were incorporated into tumor thyroglobulin. Further iuvcstigation showed that the tumor is deficient in membrane-bound sialyltransfcrase activity.

MATERIALS ASD METHODS
The tumor line I-lC2 was kindly supplied by Dr. S. Wollman of the National Cancer Institute (1) and was carried in our laboratory as previously described (2, 3). Over the course of time the function of this tumor has changed significantly.
Radioiodine uptake decreased from w-350/;, of the dose per g at 18 hours (2) to -1% per g (3), and the proportion of labeled iodoprotein in the soluble fraction from -20% of total radioiodinc (2) to -8% ( Digitonin was from Fisher Scientific Co. and was prepared as described previously (3) ; Triton X-100 was from Rohm and Haas; neuraminidase of Vibrio chokrae was from Schwarz-Mann; fl-galactosidase from Diplococcus pneumoniae was a gift of Dr. L. Van Lenten of this institute.
Incubation of Thyroid and Tumor-Rats carrying tumors weighing 1.5 to 2 g were killed by guillotine and the tumors were removed, trimmed of connective tissue, and immersed in cold (4") 0.1 M Tris-HCl buffer, pH 7.4. The washed tumor was then cut into -200.mg slices (-5 x 2 x 3 mm), immersed in 3 ml of Earle's solution which had been pregassed for 20 min with OzCOZ (95:5), and preincubated for 20 min at 37" as described (8). (Because the tumor was very friable, thin slices could not be made uniformly and tended to disintegrate during incubation.) After the incubation, 50 &i of labeled carbohydrate, 25 $Zi of [i4C]leucine or 50 PCi of i2?, were added, and incubation at 37" in a shaking water bath was continued for 1, 3, 6, and 12 hours. Each flask was gassed every 15 min with O&02.
In some experiments the slices were directly homogenized (see below) in 3 ml of Earle's solution and the homogenate was incubated as above. In each experiment a parallel incubation of normal rat thyroid gland (pooled hemilobes, 20 mg per flask) was performed as a control. Preparation of Soluble and Solubilized Proteins-The homogenization and differential centrifugation were performed as previously described (3,8). Briefly, after incubation the tumor slices were taken, washed with cold 0.1 M Tris-HCI, pH 7.4, and homogenized with 2 ml of 0.1 M Tris-HCl, 0.25 M sucrose, pH 7.4, in a Potter-Elvehjem homogenizer with a Teflon pestle at 1,100 rpm, three strokes in ice. The homogenate was centrifuged at 105,000 x g for 1 hour and t.he supernatant, Si, was dialyzed for 2 days at 4" against Tris-HCl buffer, changed several times.
The pellet was suspended in 5 ml of Tris-HCl buffer, centrifuged at 105,000 x g for 1 hour at 4"; the washed pellet was rehomogenized in 1 ml of 2y0 digitonin (3), 10 strokes at 2,400 rpm in ice, and recentrifuged at 105,600 x g for 1 hour at 4". The supernatant, Sz, was dialyzed against Tris-HCI buffer for 2 days and the pellet was dissolved in 1 ml of NCS or Protosol.
Aliquots of Si, S2, and pellets were counted in a Nuclear Chicago Mark I liquid scintillation counter in 15 ml of Aquasol, shaken with 1 ml of distilled water.
Aliquots of the dialyzed soluble (&) and solubilized (SZ) proteins were precipitated with 95% ethanol or 10% trichloroacetic acid; the precipitate after washing was dissolved in 1 ml of distilled water added to 15 ml of Aquasol and counted in a scintillation counter (10).
For determination of sialic acid, the protein was hydrolyzed with 0.1 N H2SG4 for 1 hour at 80" in a sealed tube, and neutralized with 0.1 M Ba(OH)2; the precipitate was removed 1 The abbreviation used is : HEPES, N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid. and the supernatant was dried under vacuum. The dried material was dissolved in 0.2 ml of distilled water and applied on a strip of Whatman No. 3MM filter paper, 5 x 47 cm. For determination of GlcNAc, thyroglobulin was hydrolyzed with 4 K HCl for 6 hours at 100" in a sealed tube. The hydrolysate was then dried at 50" under vacuum, redissolved in HzO, and passed through a small column of Dowex 50-X4 (H+ form) (12,13), eluted with 2 N HCl, and the cluate dried at 50" under vacuum. The material, dissolved in water, was then acetylated and analyzed by paper electrophoresis in borate buffer on a strip of Whatman No. 3MM paper (8). The neutral sugars were hydrolyzed from thyroglobulin with 2 N HzS04 for 6 hours at 100" in a sealed tube (14). The hydrolysate was passed through a column of Dowex as above, coupled to a column of Dowex l-X8, 200 to 400 mesh (formate form), and eluted as described (8). The effluent was concentrated under vacuum and then applied to Whatman paper strips as above.
Chromatography was performed by the descending method. For sialic acid, the system was isoamyl acetate-acetic acid-water (3:3: 1) for 8 hours (15). For neutral sugars, the system was 1-butanol-ethanol-water (10 : 1: 2) for 12 hours. After chromatography the strips were dried, cut into 2-cm segments, and extracted for 24 hours with 1 ml of distilled water.
Each extract and paper segment was placed in 15 ml of Aquasol for scintillation counting.
Standard sugars were run on parallel chromatogram strips and were detected by the silver nitrate method (16).
Assay of Thyroglobulin Catabolism-This assay was performed using 19 S thyroglobulin from normal Fisher rats labeled by incubating normal thyroid hemilobes with N-[3H]acetylglucosamine (specific activity 1.5 X lo3 net cpm per mg of protein) (17). Thyroid tumor or normal thyroid gland was homogenized gently (see above) in 0.1 M Tris-HCl, 0.25 M sucrose, 0.001 M EI>TA, pH 7.4. The homogenate was centrifuged at 800 x g for 10 min at 4" and the supernatant was then centrifuged at 22,000 x g for 10 min at 4"; the pellets were suspended in the same buffer and recentrifuged at 800 x g and 22,000 x g as before.
The 800-22,000 X g pellet was suspended in 0.1 M potassium acetate buffer, pH 5.6. The equivalent of 40 mg of protein, measured by the Lowry method (18)) was incubated with 0.25 mg of labeled 19 S thyroglobulin for 1,3, and 6 hours at 37". After incubation the tubes were centrifuged at 30,000 x g for 10 min at 4' and the supernatant was precipitated with 1.4 M (NH4)&04, and then by 1.8 M (NH&Sod. The 1.4 to 1.8 M (NH&S04 fraction, the 1.8 M supernatant fraction, and the 30,000 x g pellet were counted for radioactivity. Assay of Glycosyltransjerases-The tumor, or thyroids from normal Fisher rats weighing 250 g, were taken and slices of tumor or pooled thyroid hemilobes weighing 250 mg were gently homogenized in 2 ml of 0.1 M Tris-HCl, 0.25 M sucrose, pH 7.4, as described (8), then centrifuged for 10 min at 10,000 x g at 4". The supernatant was centrifuged at 22,000 x g for 10 min at 4' and the resulting supernatant was centrifuged at 105,000 x g for 1 hour at 4". The 105,000 X g pellet was suspended with a hand-driven, all glass homogenizer in 0.1 M HEPES buffer, pH 6.5, then preincubated with 0.5 PC1 of CMP-[i4C]sialic acid at 37" with 0.1% Triton X-100.
After 15 min of preincubation 0.5 or 1 mg of desialylated thyroglobulin (with less than 20% of its normal content of sialic acid) dissolved in 0.1 M Tris-HCl, pH 7.4, was added and incubated for 0.25, 0.5, 1, 3, and 6 hours.
The number of theoretical acceptor sites was determined as the difference in the sialic acid content before and after treatment with neuraminidase (see below) and is expressed as nanomoles of sialic acid. The final pH of the incubation mixture was 6.6 in a final volume of 0.4 ml. After incubation the mixture was chilled in ice and centrifuged at 105,000 x g for 1 hour at 4"; the supernatant was dialyzed for 2 days at 4" against several changes of 0.1 nc Tris-HCl buffer, pH 7.4. An aliquot of 0.1 ml was precipitated with 10% trichloroacetic acid and the precipitate counted in a liquid scintillation counter as described (8) ; other aliquots were analyzed by density gradient centrifugation and by paper chromatography after hydrolysis (8). To assay galactosyltransferase in the 105,000 X g pellet, incubation was carried out as described for the sialyltransferase assay except for the following differences: 0.1 M Tris-HCl buffer, pH 7.4, 0.5 FCi of UDP-[14C]galactose, 5 ~1 of Triton X-100, 5 Mmoles of manganese acetate, and 45 nmoles of theoretical acceptor sites.
To prepare desialylated thyroglobulin, rat, bovine, and hog 19 8 thyroglobulin, purified by ammonium sulfate fractionation, Sephades G-200 chromatography, and sucrose gradient ultracentrifugation, were separately incubated in 0.1 M Tris-HCl, p1-I 7.4, or in 0.05 M acetate buffer, pH 5.6, at 37" for 48 hours; the incubation mixture contained at the start 5 mg of thyroglobulin per 25 units of neuraminidase per ml (19). More enzyme was added every 12 hours so that at the end of incubation a total of 100 units of enzyme had been added.
After incubation the mixture was centrifuged at 3000 x g for 10 min at 4"; the supernatant was precipitated between 1.4 and 1.8 M ammonium sulfate, dissolved in the above Tris-HCl buffer, dialyzed, and residual neuraminidase was removed by affinity chromatography (9). Desialylated thyroglobulin was eluted with 0.05 M acetate buffer, pH 5.6. At least 80% of sialic acid in thyroglobulin was released, as measured by the thiobarbituric acid assay (20). To preparc thyroglobulin from which galactose was removed, the desialylated rat thyroglobulin was dialyzed against 0.05 M citrate-phosphate buffer, pH 6.6, then incubated with P-galactosidase from D. pneumoniae (activity 2.5 pmoles per min per 100 ~1) (21). Fifteen milligrams of desialylated rat thyroglobulin were incubated with 0.1 ml of ,&galactosidase at 37" for 24 hours in citrate-phosphate buffer, pH 6.3. After incubation the mixture was passed through a DEAE-cellulose column (1.5 x 5 cm). Elution of free galactose was performed with 0.01 M phosphate buffer, pH 7.2. After a subsequent elution of P-galactosidase with 0.04 M phosphate buffer, pH 7.4, most of the thyroglobulin was eluted with 0.5 M NaCl, pH 6.9. The released galactose was measured with the galactose dehydrogenase assay (Boehringer Mannheim Corp.) ; this procedure showed that 60% of the galactose had been removed by the P-galactosidase.
Completely desialylated orosomucoid was a gift of Dr. G. Ashwell (22) ; fetuin was from Grand Island Biological Co. and was desialylated as described above for thyroglobulin.
Its sialic acid content was <20% that of the starting preparation.
The protein content of the 105,000 x g pellet was determined by Lowry assay (18). The DNA content of the tissue was measured (kindly performed by J. Toal) with diphenylamine reagent with deoxyadenosine as standard (23). Sialic acid was determined by the thiobarbituric acid assay of Warren (on samples hydrolyzed in 0.1 N I&So4 for 60 min at 80" (20) and by the resorcinol reaction (24)). Immunoprecipitation with rabbit antiserum against rat 19 S thyroglobulin was performed as described previously (2).

Incorporation
of Amino Acid and Iodine-Preliminary experiments with [14C]leucine and lZ51-were carried out to see whether the tumor could synthesize thyroglobulin in the in vitro system employed and for comparison with the carbohydrate studies. Both 14C and l*jI continued to accumulate over a B-hour period, reaching0.35ycof thedose per g of tissue for [14C]leucine incubated with tumor slices, and 0.03% in homogenate; and for ?I reaching 0.66% per g in tumor slices, and -0.2% in homogenate.
The distribution of the label in soluble and solubilized proteins is presented in Table I as well as the immunoprecipitation  results  with anti-rat thyroglobulin serum. The data show that ['"Clleucine incorporation increases with time both in slices and homogenates.
From the immunoprecipitation results it can be calculated that up to 9% of the total 14C is immunochemically related to thyroglobulin in S1 of slices and up to 187, in homogenates. From density gradient ultracentrifugation patterns it was seen that about 15% of the radioactivity in S1 was in an 18 S peak at 3 hours. In the solubilized protein (SZ) after [14C]leucine, total 14C accumulation was similar but the proportion in 18 S was about 2 times greater.
In the experiment with lz61 (Table I) the soluble proteins contained less radioactivity than in the case of [14C]leucine. However, a larger proportion was thyroglobulin-like, giving a value of 7% of total Ia51 related to thyroglobulin in S1 of slices and 9% in homogenates.
Density gradient ultracentrifugation showed that about 87, of S1 radioactivity w-as in the -19 S zone at 1 hour.
I*51 in the St fraction was low in homogenates, but in slices it increased to 39% of the total at 6 hours.
Density gradient ultracentrifugation patterns showed -30% of SB radioactivity in the -19 S zone at 1 hour.
These experiments showed that tumor slices and homogenates could form thyroglobulin from [*4C]leucine and from 12;11--, and that most of the thyroglobulin-like protein, especially after lz51 Incorporation of Carbohydrates-In Fig. 1 the uptake of labeled carbohydrates in tumor and normal thyroid is shown. Uptake of [3H]ManNAc in tumor was less than 5% that of the control in slices and about 8 y0 in homogenate and showed no increase with time.
The uptake of [3H]GlcNAc, although less than that of normal thyroid, showed a continuing increase with time up to 12 hours of incubation.
[W]Galactose uptake by tumor was no different from normal; for [i4C]manaose and [i4C]fucose the total uptake in the tumor was less than normal.
The distribution of all the carbohydrates studied in the soluble and solubilized fractions and their trichloroacetic acid and immunoprecipitates are shown in Table II. Ethanol precipitation gave results similar to trichloroacetic acid and are not reported in the table. There is in general an increase in the soluble proteins with time except for galactose in which the soluble fraction is maximal at 1 hour. The soluble fraction, however, never exceeded 367, of the total and the solubilized fraction always exceeded the soluble.
The trichloroacetic acid precipitate and the immunoprecipitate ranged from 35 to 87% of the Si and Sz fractions for all carbohydrates except ManNAc; in the latter, <15Y, was precipitable with trichloroacetic acid and none with antithyroglobulin serum. The soluble and solubilized proteins were further analyzed by density gradient ultracentrifugation as shown in Figs. 2 and 3. All carbohydrates were incorporated into thyroglobulin except [3H]ManNAc; in more than 30 separate analyses, tumor incubated with [3H]ManNAc never showed a peak of labeled thyroglobulin or its 12 S precursor.
It is to be noted that all carbohydrates showed a greater peak of 19 S in the solubilized than in the soluble proteins at all times studied.
To identify the labeled products (8, 25-27) the soluble and solubilized proteins were precipitated between 1.4 and 1.8 in ammonium sulfate, hydrolyzed with I-12SOI or I-ICI, and chromato-    proteins showed that about 75% of the radioactivity was present as galactose and the remainder as mannose.
[14C]Mannose-labeled proteins showed that 65 to 76% of the label was present as mannose and the remainder as galactose.
To investigate whether the lack of incorporation of [3H]Man-NAc was due to an excessive catabolism of thyroglobulin, normal thyroglobulin labeled with [3H]ManNAc or with lzBI were incubated with "lysosomes" of tumor or of normal thyroid. As shown in Fig. 4 the tumor "lysosomes" showed no significant difference in the catabolism of thyroglobulin from those of normal thyroid.
Furthermore, the protein in soluble (Si) and solubilized (Ss) fractions of tumor which precipitated between 1.4 and 1.8 M ammonium sulfate showed a content of sialic acid, by both the thiobarbituric acid and resorcinol methods, of less than 0.1%; that is, less than 10% of normal.
To see whether the high proportion of labeled thyroglobulin that was membrane-bound was due to a nonspecific adsorption control experiment was performed : 12jI-labeled thyroglobulin was desialized with V. cholerae neuraminidase, then homogenized with a tumor slice or with normal thyroid hemilobes, and subjected to differential centrifugation as described under "Materials and Methods." For normal thyroid, 7 y0 of the desialized [1"5I]thyroglobulin remained with the pellet, and <20'$$ for the tumor; this radioactivity was completely released with vigorous homogenization and detergent both for normal thyroid and for tumor.
Glycosyltransjerase Activity-The 105,000 X g pellet from normal thyroid of Fisher rats showed sialyltransferase activity that was dependent on time, on the amount of enzyme and on the amount of substrate (Fig. 5). A similar pellet from the tumor revealed little or no sialyltransferase activity (Fig. 5). The largest quantity of tumor represented in Fig. 5B was 250 mg. A further test was performed using the 105,000 x g pellet obtained from 2 g of tumor.
Under these conditions, sialyltransferase activity amounting to -3% of that in normal thyroid could be shown.
The sucrose gradient ultracentrifugation pattern of the reaction product with the pellet from normal thyroid is shown in Fig. 6; it can be seen that CMP-[i4C]sialic acid was transferred mainly into 19 S thyroglobulin.
Labeled 19 S thyroglobulin was also seen when the pellet from 2 g of tumor was used. Treatment of the labeled thyroglobulin with neuraminidase removed all the 14C from thyroglobulin, as determined by sucrose gradient ultracentrifugation.
The 105,000 x g supernatant was also tested by incubation with CMP-[14C]sialic acid and desialylated thyroglobulin.
Soluble sialyltransferase activity was detected in normal thyroid gland and also in tumor as shown in Figs. 7 and 8. The soluble enzyme activity in normal thyroid was much less than that in the pellet on the basis of tot,al protein content.
In the cast of the tumor, the soluble enzyme activity was slightly lower than normal when calculated on the basis of protein content of the soluble fraction. It was much lower than normal, however, when calculated on the basis of DNA content of the tissue from which the soluble fraction was prepared, since the DNA content of tumor was about 4 times that of normal thyroid.
To test the specificity of sialyltransferase in normal thyroid and tumor, desialylated fetuin and desialylatcd orosomucoid were incubated as above with the 105,000 X g pellet and the 105,000 x g supernatant.
As shown in Table III, these proteins were acceptors for the sialyltransferase activity for the particulate and soluble fractions of normal thyroid gland and tumor. Desialylated fetuin was a comparable or poorer acceptor than desialylated thyroglobulin in normal thyroid and a better acceptor than desialylated thyroglobulin in tumor, both in supernatant and pellet fractions.
Desialylated orosomucoid was a poorer acceptor than desialylated thyroglobulin in all fractions except the tumor pellet.
The tumor pellet is deficient especially in the sialyltransferase activity specific for desialylated thyroglobulin. The identification of the labeled product after digestion with 0.1 N H2S04 at 80" for 1 hour and chromatography showed that in all of the cases studied, the only labeled product was sialic acid.
Since this tumor incorporates [14C]galactose into thyroglobulin at a normal rate, the membrane-bound galactosyltransferase was presumed to be present in the tumor.
As shown in Table IV, the tumor pellet showed galactosyltransferase activity that was only slightly less than that found in the control using desialylateddegalactosylated thyroglobulin as acceptor. When desialylateddegalactosylated fetuin was used as exogenous acceptor the galactosyltransferase activity of the tumor was about 5% of that of thyroglobulin during the extracting procedure the following using thyroglobulin. The product of the incubation of desialyl-  proteins.
[14C]Galactose, on the other hand, is incorporated into tumor thyroglobulin at a normal rate suggesting that the formation of the oligosaccharide unit proceeds at an almost normal rate up to this penultimate sugar (28). The finding that N-acetylglucosamine is incorporated at a lower rate than in normal could  is not clear. A possible explanation could be that membranebound thyroglobulin is partially solubilized during the homogenization and fractionation procedures.
The relatively smaller amount of '2jI-labeled thyroglobulin compared to W-and 3Hlabeled thyroglobulin could be csplained if only the more mature molecules can be labeled with iodine, perhaps because of their location in the cell. However, a defect in the iodinating system itself has not been excluded.
Decreased sialyltransferase activity necessary for the synthesis of gangliosides of the cell membranes has been recently described in a virus-transformed cell of hamsters (30) and in a mouse cell (31). The finding of defective membrane-bound sialyltrawsl'erase activity in the thyroid tumor studied here is the first evidence of a specific defect in the oligosaccharide chain synthesis of thyroglobulin.
A search for similar defects in other thyroid diseases in which there is an increase in the "particulate" iodoproteins is warranted.

279, 118
which have a different physiological role than the membranebound enzymes and which are not involved in thyroglobulin synthesis. This is consistent with the fact that glycosyltransferases are considered specific even though they show some activity with different acceptors as has been reported here and by others