Purification and characterization of a large, tryptic fragment of human thyroid peroxidase with high catalytic activity

ARCHIVES

OF BIOCHtSMIS’I’RY

AN11 BIOPHYSICS

Vol. 278, No. 2, May 1, pp. 3X-341,

1990

Purification and Characterization of a Large, Tryptic Fragment of Human Thyroid Peroxidase with High Catalytic Activity’ Alvin Taurog,*s2 Martha

L. Dorris,*

Naokata Yokoyama,*,”

and Clive Slaughter?

*Department of Pharmacology, and l-Department of Biochemistry and Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, Texas 75235

Received September 8,1989, and in revised form December

15,1989

Thyroid peroxidase (TPO) was purified from human thyroid tissue, obtained at surgery from patients with Graves’ disease, by a procedure similar to one that we had previously used for the purification of porcine TPO. The membrane-bound enzyme was solubilized by treatment of the thyroid particulate fraction with trypsin plus detergent. After precipitation with ammonium sulfate, the enzyme was purified by a series of column treatments, including ion-exchange chromatography on DEAE-cellulose, gel filtration through Bio-Gel P100, and hydroxylapatite chromatography. Although a high degree of purification was achieved, the finally isolated product was considerably more heterogeneous than the TPO obtained from porcine thyroids. Several pools of active enzyme differing in values for A412/A280 and in specific activity were collected. Gel electrophoresis was performed under native, denaturing [sodium dodecyl sulfate (SDS)] and denaturing plus reducing conditions. Native gel electrophoresis indicated that the active enzyme (93 kDa) was heavily contaminated with an inactive 60-kDa fragment, which we were unable to remove by HPLC. The inactive fragment was highly antigenic when tested on immunoblots with an antibody to TPO. The presence of the inactive fragment greatly reduced values for A412/A280 in the finally purified human TPO. Two of the pools, with A4,2/A280 values of 0.159 and 0.273, were used for further testing. Catalytic activity was very similar in these two pools when measured on the basis of heme content by several different assays. Moreover, the specific activities of both, based on heme content, were very similar to those observed with a porcine TPO preparation with A4,2/ ’ This work was supported by LJSPHS Grant DK-03612. ’ To whom correspondence and requests for reprints should he addressed. .’ Present address: The First Department of Internal Medicine, Nagasaki Iiniversity School of Medicine, Nagasaki 852, Japan. Copyright cb1990 by Academic Press, Inc. All rights of reproduction in any form reserved.

indicate that the inactive A 280 = 0.48. These findings 60-kDa fragment most likely did not contain heme. On SDS-polyacrylamide gel electrophoresis under reducing conditions, the 60-kDa fragment completely disappeared and was replaced by a 36- and a 24-kDa component. Amino terminal sequence information obtained on these components indicated that the 24-kDa component represents the amino terminal portion of the active 93-kDa fragment, whereas the 36-kDa fragment represents the carboxyl terminal portion. A model is proposed suggesting that the 60-kDa fragment was generated by trypsin cleavage of native TPO at two internal sites within a disulfide loop (res -300 and res 564) and at one further internal site (res 280). In addition, trypsin cleavage is proposed at sites near the amino and carboxy1 ends common to both the active 93-kDa and the inactive 60-kDa fragments. According to this model, the inactive BO-kDa fragment is missing that portion of the polypeptide chain extending from res -300 to res 564. Since the 60-kDa fragment lacks heme, this region of the molecule appears to be important for heme binding and may contain the heme binding site. ,C’ 1990 Academic

Press,

Inc.

Thyroid peroxidase (TP0)4 is a membrane-bound, glycosylated, hemoprotein enzyme that plays a key role in thyroid hormone biosynthesis by catalyzing both the iodination of thyroglobulin and the coupling of iodotyrosyl residues in thyroglobulin (Tg) to form T, and T:, (1). Purification procedures have been reported for porcine (2-g), bovine (lo), and human (11, 12) TPO. Por4 Abbreviations used: TPO, thyroid peroxidase; Tg, thyroglohulin; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; TSH, thyroid-stimulating hormone; BSA, bovine serum albumin; DIT, diiodotyrosine.

333

334

TAUROG

tine thyroids have been most commonly used. Two methods of solubilization for the membrane-bound enzyme have been used prior to enzyme purification, one involving the use of trypsin plus a detergent, the other utilizing detergent alone. The former procedure yields a large tryptic fragment (13), whereas the latter has been used to isolate the native enzyme (8, 9, 11, 12). The full length primary amino acid sequence of both porcine (14) and human (15-17) TPO has been deduced from cDNA cloning experiments. We have previously described (6) the isolation of highly purified and extremely active porcine TPO from the particulate fraction of porcine thyroids after solubilization with trypsin plus detergent. More recently (13), we were able to characterize our purified trypsin fragment of porcine TPO in relation to the known sequence of the native enzyme. During the past few years it has been established by studies in many laboratories (11, 18-23) that TPO is very closely related to, if not identical with, the thyroid microsomal autoantigen that elicits the production of the serum microsomal autoantibodies associated with autoimmune thyroid disease. In our initial studies on the relationship between TPO and thyroid microsomal autoantibodies (23), we used our highly purified porcine TPO. Although some useful information was obtained, it became apparent that cross-reactivity between human autoantibodies and porcine TPO was low and that human TPO would be greatly preferable for such studies. We therefore undertook to purify human TPO from surgically removed thyroids of patients with Graves’ disease, using a trypsin-detergent solubilization procedure similar to that previously used in our purification of porcine TPO. The use of this purified human TPO in studies with thyroid microsomal antibodies is reported elsewhere (24). In the present communication we describe our purification procedure for human TPO and provide data on enzymatic activity and structural relationship to the native enzyme. Characterization of a large, inactive fragment of TPO in our enzyme preparation also made it possible to locate a region of the molecule that may be involved in binding the heme. METHODS

Purification

AND

MATERIALS

Procedure

Tissue source. Surgically removed thyroid tissue from patients with Graves’ disease was obtained from two sources: (1) the Ito Thyroid Clinic and Hospital, Tokyo, Japan, through t.he courtesy of Drs. Naofumi Ishikawa and Kunihiko Ito, and 2) the Thyroid Laboratory, University of Sao Paulo, Brazil, through the courtesy of Dr. Gerald0 Medeiros-Neto. The patients had been treated with iodide and antithyroid drugs prior to surgery, Approximately 1500 g of tissue was accumulated over a period of about 18 months. During this period the glands were stored at -70°C. All procedures were performed in the cold room or in ice baths, unless otherwise specified.

ET AL. Homogenization of tissue and collection of particulate fraction. The tissue was processed in three separate batches, each consisting of about 500 g (lo-15 individual glands per batch). The glands were allowed to thaw slightly, and then they were sliced into sections approximately 4 mm thick in a Cuisinart food processor (Model DLC-7E). The slices were washed four times, each washing with 2 liters of cold 0.15 M NaCl to remove blood from the highly vascular tissue. After the last wash and decantation of the supernate, the tissue was transferred to a l-gallon, stainless steel Waring Blendor, with a total of 1000 ml of cold 50 mM Tris-HCl, pH 7.0, containing 0.1 mM KI and 1 mM EDTA. The tissue was blended four times for 30 s on the high setting, the homogenate being stirred with a heavy glass rod between treatments. After filtration through four layers of cheese cloth, the homogenate was centrifuged at 25,000g for 70 min at 4°C in a Sorvall centrifuge. The supernates were decanted and discarded, and lipid on the sides of the centrifuge bottles was wiped up with cotton-tipped applicators. The pellets were suspended with a total 180 ml of 0.15 M KCll0.05 M Tris-HCl, pH 8.0-0.1 mM KI, stirred vigorously with a heavy glass rod, and treated twice with a large Polytron homogenizer, equipped with a 45T generator, each time for 15 s on a setting of 7. Solubilization with trypsin-deoxycholate-Z’riton. The Polytrontreated homogenate was transferred to a flask with an additional 100 ml of buffer, placed in a water bath on a magnetic stirrer, and the mixture was brought to 24°C. To the vigorously stirred homogenate at 24°C were added in succession (1) 20% sodium deoxycholate, (2) Triton X-100, and (3) crystalline trypsin, in amounts required to provide final concentrations of l%, l%, and 140 mg/l, respectively. Stirring was continued for 2.5 h, after which an amount of soybean trypsin inhibitor sufficient to neutralize the trypsin was added. The mixture was centrifuged in a Beckman ultracentrifuge at 31,500 rpm (115,OOOg) for 75 min. The supernates were drawn off with aspiration and combined, giving approximately 470 ml of turbid, reddish yellow, somewhat viscous solution. Ammonium sulfate precipitation. The acetone precipitation step previously used at this stage in the porcine TPO preparation (6) was omitted. Ammonium sulfate precipitation was performed at 0°C directly on the 115.OOOg supernate by addition of an equal volume of saturated (at 0°C) ammonium sulfate solution. After vigorous stirring, the thick slurry was centrifuged at 25,000g for 1 h on a Sorvall centrifuge at 4°C. This yielded a soft, but cohesive, Roating precipitate, which was recovered by drawing off the mother liquor. The precipitate was suspended with a total of 200 ml of 0.15 M KCI-0.025 M TrisHCl, pH 7.0/0.1 mM KI and stirred in a water bath at 24°C on a magnetic stirrer for 30 min. After centrifugation at 115,OOOgfor 75 min, the relatively clear, lightly colored supernate was dialyzed in the cold for 16 h against 5.5 liters of 0.025 M Tris-HCl, pH 7.0-0.5 mM KI. A second dialysis was performed for 7 h against 5.5 liters of fresh cold buffer. The dialyzed solution was recentrifuged at 115,OOOg for 75 min, and the clear supernate was concentrated in an Amicon ultrafiltration cell to a volume of about 150 ml. DEAE-cellulose chromatography. The Amicon-concentrated sample was pumped onto a previously equilibrated DE-52 column (5 X 23 cm) at a flow rate of 94 ml/h. Elution at the same flow rate with 0.05 M KCl-0.025 M Tris-HCl, pH 7.0-0.5 mM KI for 21 h removed about 50% of the protein, but no enzyme activity. The active material was collected in a single broad peak by gradient elution with 0.05 to 0.25 M KCl, in 0.025 M Tris-HCl, pH 7.0-0.5 mM KI. The peak fractions were pooled and concentrated in an Amicon ultrafiltration cell. This procedure was preferable to precipitation with 50% saturated ammonium sulfate, which in the case of Batch 2 did not give a cohesive precipitate. The DEAE-cellulose chromatography increased the specific activity from 1.4 to 74 (average of 3 batches, Table I). Bio-Gel P- 100 chromatography. The material from the DE-52 column was dialyzed against 0.15 M KCl-0.025 M TrissHCl, pH 7.0~0.5 mM KI, and applied to a Bio-Gel P-160 column (2.5 X 44.5 cm) by gravity flow. Elution was performed with the same buffer under a pressure head of 30 cm at a flow rate of approximately 14 ml/h. The peak

PURIFICATION

OF HUMAN

THYROID

TABLE

Activity

Purification

335

PEROXIDASE

I

of Human TPO at Various Stages of Purification”

step

1. Ammonium sulfate precipitation, sum of three separate hatches 2. DEAE-cellulose chromatography, sum of three separate hatches 3. Bio-Gel P-100 chromatography a. Sum of three separate hatches, ammonium sulfate-precipitahle fraction h. Sum of three separate batches, ammonium sulfate-soluble fraction 4. Second DEAE-cellulose chromatography, pooled ammonium sulfate-precipitable (three batches) 5. Hydroxylapatite chromatography pooled ammonium sulfate-precipitahle Pool A Pool B Pool c pooled ammonium sulfate-soluble Pool A-l Pool A-2 Pool B-l Pool B-2

Total U of activity*

Total protein’ bd

Specific activity

58,150

41,300

1.4

43,768

589

74

21,300

225

95

12,847

28.7

448

11,970

53.5

224

4,579 1,656 3,638

8.0 4.4 10.0

572 376 364

0.159 0.107 0.106

1,596 2,430 2,142 1,062

2.2 3.29 2.92 1.23

727 664 734 775

0.303 0.231 0.271 0.268

’ Three hatches, each containing approximately 500 g of tissue, were processed separately for the first three purification steps. In the third step (Bio-Gel PlOO), the eluted activity was divided into 50% ammonium sulfate-precipitahle and ammonium sulfate-soluble fractions. Beginning with the fourth purification step, the material collected in the three individual hatches was pooled. * Based on iodide oxidation assay. ’ Based on absorbance at 280 nm, assuming AEX,,= 1 .O for 1 .O mg/ml protein.

fractions were pooled and precipitated at 0°C with 50% saturated ammonium sulfate. After centrifugation at 115,OOOgfor 45 min, the prepH 7.0-0.5 cipitate was dissolved in 0.15 M KCl-0.025 M Tris-HCl, mM KI. In contrast to results previously obtained with porcine TPO, a very significant fraction of the enzyme activity remained in the supernate after centrifugation (Table I). Moreover, the material in the supernate had a much higher specific act,ivity than that in the precipitate (Table I). Therefore, the ammonium sulfate-precipitable and ammonium sulfate-soluble fractions were processed separately in subsequent purification steps. The hatch procedure was discontinued at this point, and the respective ammonium sulfate fractions from Batches 1, 2, and 3 were pooled for further treatment. 2nd DEAtk~llulose column. This treatment was applied only to the pooled ammonium sulfate-precipitahle fractions from Batches 1, 2, and 3. The pooled sample was concentrated in an Amicon cell, dialyzed against 0.02 M phosphate, pH 7.5-0.1 mM KI, and pumped onto a DE-52 column (2 X 14.5 cm) at a flow rate of 15 ml/h. The active material was eluted with a gradient of 0.03 M to 0.15 M KC1 in the same phosphate-K1 mixture. The fractions in the center of the activity peak were pooled, concentrated in an Amicon cell, and dialyzed against 0.01 M phosphate, pH 6.8-0.1 mM KI. Hydroqlapatite chromatography. This treatment was applied separately to the ammonium sulfate-precipitahle and ammonium sulfatesoluble fractions (Tahle I). The former was pumped onto a column (1.5 X 25.4 cm) that had heen equilibrated against 0.01 M phosphate, pH 6.8-0.1 mM KI at a flow rate of 14.8 ml/h. The column was eluted stepwise at pH 6.8 with 0.02 M, 0.04 M, and 0.1 M phosphate, each containing 0.1 mM KI. Almost all the activity eluted with 0.02 M phos-

phate, in two separate peaks. The later peak was broad and trailing, hut it contained material of higher specific activity and a higher value This was called Pool A (Table I). The earlier peak was for -4d&80. sharper, hut it contained material of lower specific activity. It was divided into two fractions, Pool B (center portion), and Pool C (combined ascending and descending portions). The pooled, concentrated ammonium sulfate-soluble material was dialyzed against 0.01 M phosphate, pH 6.8-0.1 mM KI and pumped onto an hydroxylapatite column (0.9 X 27 cm) that had been equilibrated against 0.01 M phosphate, pH 6.8-0.1 mM KI at a flow rate of 10 ml/h. Elution was started with 0.01 M phosphate-O.1 mM KI, and this removed a fairly sharp activity peak. The center portion was taken as Pool A-l (Table I) and the combined ascending and descending portions as Pool A-2 (Table I). Elution with 0.02 M phosphate-O.1 mM KI removed a broader, more trailing peak, which was divided into Pool B-l (center portion) and Pool B-2 (combined ascending and trailing portion) (Table I). All pools of purified enzyme were concentrated to a low volume in an Amicon ultrafiltration cell surrounded by an ice bath at 0-2°C.

Gel Electrophoresis Polyacrylamide gel electrophoresis (PAGE) was carried out under nondenaturing (native) or denaturing conditions. The procedure of Hamada et al. (25) was used for native gels. These were stained either for protein with Coomassie blue, or for enzyme activity with guaiacol plus H202. In the latter procedure the gel was transferred to a plastic tray containing 5 mM guaiacol in 0.02 M phosphate, pH 6.8, and gently shaken for lo-15 s. Hydrogen peroxide was

TAUROG added to a final concentration of 0.5 mM, and the tray was gently shaken for about 2 min. Enzyme activity was displayed by the development of a readily visible orange color. The solution in the tray was decanted and replaced with fresh 0.02 M phosphate. To preserve the color the gel was stored in the dark at 4°C. SDS-PAGE under reducing or nonreducing conditions was carried out as previously described (13).

Immunoblotting This was performed as previously described (13). Three different antibody sources were used: (i) a rabbit antiserum against porcine TPO (13), (ii) a rabbit antiserum against a peptide corresponding to residues 7809793 of porcine TPO (13), and (iii) sera from 24 patients with suspected autoimmune thyroid disease, displaying thyroid microsomal autoantibody titers ranging from 1:400 to 1:102,400 (24).

Sequencing Procedure Sequencing of components separated by gel electrophoresis formed as previously described (13).

was per-

Assays for Catalytic Activity During the purification procedure enzyme activity was determined by the iodide oxidation assay, as previously described (6). Three additional assay procedures were also used to measure the specific activity of two pools of purified human TPO (Pool A and Pool B-l) and to compare these with a previously purified porcine TPO preparation. These assays were based on (1) guaiacol oxidation, (2) iodination of BSA, and (3) coupling of ‘“‘I-diiodotyrosine in thyroglobulin to form ‘“‘I-thyroxine. The guaiacol and iodination assays were performed as previously described (6). The coupling assay differed in the following respects from the one described previously (6): (1) the reaction was performed in the presence of 1 pM diiodotyrosine, a potent stimulator of the coupling reaction, (2) the enzyme concentration was increased by a factor of about 2, (3) the glucose oxidase concentration was 20 mu/ml, instead of 39 mu/ml, and (4) the incubation time was 20 min, instead of 10 min. Since the human TPO preparation contained a large proportion of an inactive TPO fragment (see below), specific activities were calculated on a heme basis for better comparison with porcine TPO. The heme content was determined by measurement of the absorbance at 412 nm, assuming a mM extinction coefficient of 114 as previously reported for lactoperoxidase (26).

Reagents Deoxycholic acid, Na salt, was purchased from Calbiochem, trypsin (2 X crystalline) from ICN, trypsin inhibitor from Sigma, Triton X100 (scintillation grade) from Kodak, reagents for gel electrophoresis from Bio-Rad, ‘““I-Protein A for immunoblotting from ICN, and glucose oxidase (200 U/mg) from Boehringer-Mannheim. RESULTS

Activity at Various Stages of Purification of Human TPO Table I shows total units of activity (iodide oxidation assay), total protein (based on Azso ), and specific activity of human TPO at various stages of purification. The earliest entry is the ammonium sulfate-purified material obtained after solubilization of the TPO with trypsin plus detergent. At earlier stages of purification the assay

ET AL.

results were not considered reliable. Values for A412/A280 are shown for the final stage of purification. The total units of activity in the approximately 1500 g of starting tissue amounted to 58,000. This very likely represents a low value because the initial ammonium sulfate supernates were discarded before we fully realized that a portion of the TPO was soluble in 50% ammonium sulfate. Based on the value of 58,000, the human tissue contained about 39,000 U/kg compared to about 15,000 U/kg in porcine thyroid tissue (6). An increased TPO concentration in Graves’ thyroids (27) and in thyroids of TSH-treated rats (28), compared to normal, has been previously reported. The highest specific activities and values for A412/A2R0 were obtained from material that was soluble in 50% saturated ammonium sulfate (Pools A-l, A-2, B-l, B-2, Table I). However, all of these values were much lower than those previously obtained for porcine TPO (6). As demonstrated below, the human TPO preparation contained a high percentage of an inactive fragment of TPO, presumably arising through tryptic cleavage sites not seen with porcine TPO. We attempted to separate the inactive and active fragments by HPLC using a size exclusion column (Spherogel-TSK), but this yielded only a minor increase in purity with a substantial loss in total activity. Attempts at further purification were abandoned, therefore, because of the limited supply of material, and also because it was possible to obtain useful information with the partially purified product already at hand. Pool A and Pool B-l (Table I) were used for the experiments described below. Native Gels of Purified

Human TPO

Figure 1 shows results obtained with native gels of the purified human TPO preparation, tested for protein, enand immunoactivity. Results are zymatic activity, shown for Pool A (lane a) and Pool B-l (lane b). In Fig. 1A the gel was stained for protein with Coomassie blue. Two prominent bands were observed, corresponding to approximately 90 and 60 kDa as determined from simultaneously added markers. In Fig. 1B the gel was stained for enzyme activity with guaiacol + HzOz. Only the 90-kDa band displayed enzyme activity. Pool A contained a greater proportion of the inactive 60-kDa component, as would be expected from its lower specific activity and value for A412/A2R0(Table I). The immunoblots in Fig. 1C demonstrate that the protein in both the 90- and 60-kDa bands was highly antigenie. The 60-kDa component, therefore, most likely represents an inactive fragment of human TPO. Such a component was not observed in the previously reported porcine TPO purified by a similar procedure (6). It appeared, therefore, that trypsin cleavage sites were different for human and porcine TPO. Further evidence for this conclusion is presented below.

PURIFICATION

B

OF HUMAN

THYROID

TABLE

C Kda

-90 -60

Conditions of gel electrophoresis

Nonreducing

b

ab

SDS-Gel Electrophoresis Figure 2 shows results of SDS-polyacrylamide gel electrophoresis on Pool B-l, under nonreducing conditions (A), and under reducing conditions (B). Only protein staining was used, since enzyme activity is lost on treatment with SDS. Under nonreducing conditions the results were similar to those seen with native gel electrophoresis (Fig. lA), except that much better separation was obtained between the two major protein components.

A

B

FIG. 2. SDS-polyacrylamide gel electrophoresis. Results obtained with Pool B-l under (A) nom-educing and (B) reducing conditions. The gels were stained with Coomassie blue. Approximately 10 ~g of protein was applied to the gel.

M, of gel band 24 36 93 60

Amino terminal sequence

Location in human TPO

SQHPT

109 564 109 112

SQHPTD

112 -+

TQQSQH

109 -+ 564 + 280 +

TQQS LFVLS

TQQSQ

LFVLSD PAAGTA

ab

FIG. 1. Results obtained with native polyacrylamide gel electrophoresis of purified human TPO. (A) Gel stained with Coomassie blue for protein. (B) Simultaneously run gel stained with guaiacol plus H,O, for catalytic activity. (C) Immunoblot of separately run gel after transfer to nitrocellulose membrane. The membrane was incubated with rabbit anti-porcine TPO antiserum and then with ““I-protein A. Lane a in each panel represents Pool A, and lane b represents Pool B-l. Approximately 10 pg of protein was applied to the gels in A and B, and 5 jog in C.

II

Sequencing Data on Human TPO, Pool B-l, after SDS-PAGE

Reducing

a

337

PEROXIDASE

+ -+

Note. Initial yields of PTH amino acids from the peptides recovered after electrophoresis ranged from 10 to 28 pmols.

Under reducing conditions (Fig. 2B), the 60-kDa component completely disappeared, and two new components appeared at approximately 36 and 24 kDa. The high molecular weight component corresponding to the enzymatically active enzyme (93 kDa) was relatively unaffected. The disappearance of the 60-kDa protein under reducing conditions and its apparent conversion to a 36and a 24-kDa component suggested that the 60-kDa component is comprised of 36- and 24-kDa polypeptide fragments of TPO linked by one or more disulfide bridges. This was further investigated by the sequence studies described in the next section. Sequence Studies The amino terminal sequences of the 24-, 36-, and 93kDa components, excised from a reducing gel, are shown in Table II. The 24- and the 36-kDa components corresponded to polypeptide fragments starting at amino acid positions 109 and 564, respectively, in the known sequence of human TPO (15-17). The 93-kDa component showed two different amino terminal sequences. The most abundant one corresponded exactly with that of the 24-kDa component, starting at position 109. This is also homologous to the amino terminal position previously observed with purified porcine TPO (13). Another, less abundant amino terminus was observed in the 93-kDa fragment of human TPO, starting at position 112. This does not correspond to a tryptic cleavage site, and it presumably represents the action of an endogenous proteolytic enzyme. The amino terminal sequence of the 60-kDa component, excised from a nonreducing gel, was unexpectedly complex. Four separate sequences were observed. Three of these corresponded to sequences observed with the 24-, 36-, and 93-kDa components, but one was located at a different site, starting at position 280.

338

TAUROG

848-871

Aplcal Membrane + Cytoplasm

Le” 933 FIG. 3. Proposed model of native human TPO showing sites of cleavage by trypsin during the solubilization procedure. The model is patterned after that proposed for porcine TPO in a previous communication (13). Sites A, B, and D can be definitely assigned on the basis of the sequencing data shown in Table II. Site E is located on the amino side of the putative membrane spanning region (848-871) but carboxyl to position 793 (See Fig. 4). See Discussion for approximate location of site C.

Proposed Structure of 93- and 60-kDa Trypsin Fragments in Relation to Structure of the Native Enzyme In a previous communication (13) we presented a model for native porcine TPO showing the sites of cleavage in the purified, trypsin-solubilized enzyme. We propose that, at least in its general characteristics, this model also applies to native human TPO. Figure 3 represents a model for native human TPO showing the proposed trypsin cleavage sites in our TPO preparation. Two of the sites, designated A and D, correspond exactly to homologous trypsin cleavage sites in porcine TPO. Site E was not definitely located in either the porcine or the human TPO preparation, but in both cases it must be similarly situated, close to or on the lumenal side of the apical membrane. The biggest difference between our human and porcine TPO preparations was the presence in the former, but not in the latter, of a very prominent, enzymatically inactive 60-kDa fragment. This fragment was observed on native gels and on SDS gels under nonreducing conditions. However, under reducing conditions it disappeared completely and appeared to be converted to com-

ET AL.

ponents of 36 and 24 kDa. Sequence data, summarized in Table II, indicate that the amino terminus of the 36kDa fragment is Leu 564, whereas that of the 24-kDa fragment is Thr 109. The latter is also the amino terminus of the active 93-kDa fragment, indicating that the 24-kDa fragment represents the amino terminal portion of the trypsin-treated human TPO molecule. Another amino terminus, not so obviously explained, was observed at Pro 280. To explain the origin of the inactive 60-kDa fragment, consistent with the sequence data, we propose the model shown in Fig. 3. This model is similar to the one initially proposed for porcine TPO (13), except that it includes two additional trypsin cleavage sites and one additional disulfide bridge. Trypsin cleavage is shown at two sites, designated C and D, within a large disulfide loop. The closest cysteine residues that could form such a loop occur at positions 390 and 599. The trypsin cleavage site at D (564) is homologous to one formerly postulated (13) for porcine TPO (561). The two additional cleavage sites for human TPO are shown at B and C. Site B is postulated to explain the amino terminus at 280 found in the 60-kDa fragment. We assume that this site is situated between two relatively closely spaced disulfide bridges. Native human TPO molecules cleaved at sites B, C, and D could thus give rise to a 60-kDa component consisting of the following: (1) A polypeptide chain extending from site A (109) to site B (280), which we propose corresponds to the 24-kDa fragment seen after gel electrophoresis under reducing conditions. Assuming an average molecular weight of 110 per residue, the molecular weight of this fragment, consisting of 171 residues, would be 18,810, considerably lower than the 24,000 estimated from the gel. However, this region of the molecule may contain N-linked carbohydrate (17,29), and its molecular weight is most likely greater than 18,810. Also, the 24-kDa value may be an overestimate because glycoproteins tend to give high values on SDS-PAGE (30). (2) A polypeptide chain extending from D to E, comprising the 36-kDa fragment seen after gel electrophoresis under reducing conditions and connected to the 24-kDa fragment by one or more disulfide bonds. (3) A short peptide chain extending from B to C connected to the 36kDa fragment by a disulfide bridge. The existence of this fragment was postulated to explain the amino terminus at residue 280 observed in the 60-kDa polypeptide. The molecular weight of the short peptide may be estimated as the difference between the total (60,000) and the sum of the other two fragments (36,000 + -24,000). As indicated above, the 24-kDa value may be an overestimate. If we assume a value of 22 kDa for this fragment and accept the other values (60 kDa and 36 kDa) as correct, this would give an estimate of about 2000 for the molecular weight of the short peptide fragment. According to the model in Fig. 3, the inactive 60-kDa fragment lacks that portion of the active TPO molecule

PIJRIFICATION

OF HIJMAN

339

PEROXIDASE

viously reported (24). Most of the patient sera also showed a positive immunoreaction with the 36-kDa fragment. However, none of the patient sera reacted with the 24-kDa protein. This region of the TPO molecule, therefore, is much less likely to contain epitopes for autoantibody production than the region represented by the 36-kDa protein. This is in agreement with results previously reported by Vassart’s laboratory (16,31).

A Kda

36-

THYROID

#b Catalytic Activity

24ab I:100

14600

1~25,600

NORMAL SUBJECTS

FIG. 4. Immunoblots obtained with purified human TPO. Approximately 4 pg of Pool A was applied per lane. After SDS-PAGE under reducing conditions, the separated components were transferred to a nitrocellulose membrane. In A, the membrane was treated with either an antibody to peptide region 780-793 (lane a) or with an antibody to purified porcine TPO (lane b). B, results obtained with the same antigen but with serum from 24 patients with suspected autoimmune thyroid disease and from 3 normal subjects. After incubation with respective antisera, the membrane was incubated with ‘““I-protein A. Microsomal hemagglutination antibody titers are indicated. All antisera were diluted 1:400 before use.

extending from site C to site D. The lack of enzyme activity in the 60-kDa fragment could be explained if the region of the TPO molecule between site C and site D is essential for activity. Evidence for this is discussed below. Immunoblots

with Peptide Antiserum

As reported previously (13), we have prepared an antiserum to a synthetic peptide representing residues 780793 in the native sequence of porcine TPO, and we have shown that this antiserum reacts with a crude preparation of human TPO (23). This antiserum was used in the present study to confirm the structure of the 60-kDa fragment proposed in the preceding section. As shown in the immunoblots in Fig. 4A, the antipeptide antiserum reacted with the 36-kDa component, but not with the 24-kDa component, as would be predicted on the basis of the proposed structure. The antipeptide antiserum also reacted with the 93-kDa protein, providing further evidence for the model shown in Fig. 3. Fig. 4B shows immunoblot results obtained with the same human TPO antigen used in Fig. 4A, but with serum from 24 patients with suspected autoimmune thyroid disease as the antibody source. The microsomal antibody titers in the patient sera varied from 1:lOO to 1: 102,400 (24). Results obtained with serum from 3 normal subjects are also included. There was a good correlation between the antibody titer and the intensity of the immunoblot reaction with the 93-kDa protein, as pre-

Several different assay procedures were used to compare the catalytic activity of the human TPO preparation with that of porcine TPO. Measurements were made on Pool A and Pool B-l. Specific activities were calculated on the basis of heme content, as described under Methods and Materials. The results are shown in Table III. Although the values for A412/Az80 were much lower for human TPO than for porcine TPO, the catalytic activities of the two preparations based on heme content were quite similar in the iodination and coupling assays. On the basis of the iodide oxidation assay, both Pool A and Pool B-l of human TPO showed somewhat higher specific activities than porcine TPO did, but porcine TPO was more active in the guaiacol assay. The results in Table III, showing very similar specific activities for Pool A and Pool B-l based on heme content, despite very different values for AdI2 /Az80, provide evidence that the inactive 60-kDa component of both preparations does not contain heme. The same conclusion is suggested by the similar specific activities observed for the porcine and human TPO preparations. On the basis of the model proposed in Fig. 3, the region of native TPO lying between sites B and C is missing from the 60-kDa protein. It seems reasonable to suggest, therefore, that this region of TPO is important for binding the heme prosthetic group. Pools A-l and A-2. As shown in Table I, Pools A-l and A-2 displayed specific activities and values for A4,2/ Azxo comparable to those of Pools B-l and B-2. However, the gel patterns for Pools A-l and A-2 (not shown) were

TABLE

III

Comparison of Specific Activities of Human TPO and Porcine TPO on the Basis of Heme Content Assay

p-TPO

procedure A,,,/A,,, U/nmol Guaiacol Iodide oxidation Iodination Coupling

h-TPO,

= 0.479 A,,,/A,,, heme

lJ/nmol

Pool A = 0.159 heme

h-TPO,

Pool R-l

A412/A2R0= 0.271 IJ/nmol

heme

244

156

154

235

334

300 15.2 1530

13.1 1440

15.8 1620

340

TAUROG

more complicated than those in Fig. 2, and characterization of the gel components in these pools was deferred for future investigation. Of interest in connection with the discussion below, Pools A-l and A-2 displayed a band at 98 kDa in addition to the one at 93 kDa. DISCUSSION The method used to purify human TPO in the present study was similar to one that we had previously used for the purification of porcine TPO (6). The human thyroid tissue was obtained by surgical removal from patients with Graves’ disease, and even though it was much less readily available than porcine thyroid tissue, this was partially offset by a several fold higher concentration of TPO. The purification scheme involved solubilization of the membrane-bound enzyme with trypsin plus detergent, a procedure that had been shown (13) in the case of porcine TPO to remove approximately 90 amino acids from both the amino and carboxyl ends of the native molecule (molecular weight of approximately 110,000) and also to cleave an internal peptide bond after Arg residue 561. With human thyroid tissue the trypsin-detergent solubilization procedure produced cleavage sites at the amino and carboxyl ends of TPO very similar to those in porcine TPO and also at the homologous site after Arg residue 563. However, at least one additional internal site was cleaved, and this resulted in the formation of an inactive 60-kDa fragment of TPO that we were unable to separate from the active 93.kDa fragment even by HPLC. The inactive 60-kDa fragment completely disappeared after SDS-PAGE under reducing conditions, leading us to propose (Fig. 3) that the trypsin cleavage sites that produced it occurred within a disulfide loop. It is of interest that the 93-kDa active fragment of human TPO was unaffected by SDS-PAGE under reducing conditions (Fig. 2). This differs from results previously reported with porcine TPO (13), in which a fragment of similar size (88 kDa) was largely converted by SDS-PAGE under reducing conditions into fragments of 59,32, and 29 kDa. The latter observation was explained by the presence in porcine TPO of a trypsin cleavage site after Arg residue 561 within a disulfide loop. Since with human TPO we also observed cleavage at an homologous site (after Arg residue 563), which we postulate occurs within a disulfide loop (Fig. 3), there appears to be a discrepancy between the present findings with human TPO and the previous findings with porcine TPO. A possible explanation for this discrepancy is that only a portion of the native human TPO molecules were cleaved by trypsin at sites B, C, and D (Fig. 3), and that the 93-kDa fragment represents molecules that were not cleaved in this manner. The relative intensity of the 60and 90-kDa bands in Fig. 1 would suggest that about 50% of the molecules were cleaved at sites B and C. In this

ET AL.

connection, it is of interest that Kimura et al. (15) reported that there are two different forms of human TPO, most likely arising from alternately spliced mRNAs. The shorter form lacks a 57-amino acid sequence extending from residue 534 through residue 590. This form of the enzyme, therefore, would lack the tryptic cleavage site after Arg residue 563. This raises the possibility that the trypsin cleavage sites at B, C, and D in Fig. 3 occur only in the longer of the two human TPOs and that it is only the longer form that generates the inactive 60-kDa fragment. In this case, the active 93-kDa fragment, at least in Pools A and B-l, would represent primarily the shorter form of human TPO. The validity of this hypothesis requires further testing. A 98-kDa band, presumed to be derived from the longer form of TPO, was observed in Pools A-l and A-2 (See above). Values of A412/A2R,j were much lower for the human TPO prepared in this study than for porcine TPO. Pool A had a value of only 0.159, while Pool B-l had a value of 0.271. Corresponding values for previously prepared porcine TPO ranged from 0.48 to 0.54. However, when specific activities of human TPO were compared with porcine TPO on the basis of heme content (Table III), the values were quite comparable. In assays based on iodide oxidation, human TPO appeared to be somewhat more active than porcine TPO, and in assays based on iodination of BSA and on coupling of DIT to form Tq, human TPO was at least equally as active as porcine TPO. Only in the guaiacol assay was porcine TPO significantly more active than human TPO. Moreover, Pool A, which had a much lower value for A412/A2H0 than Pool B-l and a greater proportion of the inactive 60-kDa fragment (Fig. l), displayed specific activity values as high as those of B-l when compared on the basis of heme content. It can be inferred from these results that the inactive 60-kDa fragment did not contain heme. The structure of the 60-kDa fragment, on the basis of the model shown in Fig. 3, indicates that it lacks the sequence of amino acids from site C to site D. Site D was definitely located at residue 564, but site C was not defined. Its location depends on the molecular weight of the peptide fragment, BC, in Fig. 3. From the data at hand the molecular weight of this peptide can be only roughly estimated, and, as indicated above, an arbitrary value of 2000 was selected. This would place site C at about residue 300. On this basis the missing sequence in the 60-kDa inactive fragment extends from about residue 300 to residue 564. Since the 60-kDa fragment lacks heme, it appears that the heme binding site may be located in this region. This is consistent with the suggestion recently made by Kimura and Ikeda-Saito (32) concerning the location of the proximal histidine that is thought to bind the iron center of the heme prosthetic group in TPO. From evidence based on comparison with peroxidases in which the heme binding site is known, they proposed that His 407 is the most likely candidate

PURIFICATION

OF HUMAN

THYROID

341

PEROXIDASE

A., and Rapoport, B.

for the proximal histidine. This would readily explain why elimination of residues 300 through 563 in TPO would produce an inactive enzyme. Our results are also consistent with the observations of Libert et al. (16), who reported homology between a segment of subunit I of cytochrome c oxidase (residues ll-65), which has been suggested as the heme-binding domain, and the region of human TPO extending from res 510 through 567.

14. Magnusson, R. P., Gestautas, J., Taurog, (1987) J. Biol. Chem. 262, 13,885513,888.

ACKNOWLEDGMENTS

18. Portman, L., Hamada, N., Heinrich, G., and DeGroot, J. C’lin. Endocrinol. Metab. 61, 1001-1003.

15. Kimura, S., Kotani, T., McBride, 0. W., Umeki, K., Hirai, K., Nakayama, T., and Ohtaki, S. (1987) Proc. N&l. Acad. Sci. lJSA 84,5555-5559. 16. Libert, F., Ruel, J., Ludgate, M., Swillens, S., Alexander, sart, G., and Dinsart, C. (1987) E’MZ30 J. 6,4193-4196.

N., Vas-

17. Magnusson, R. P., Chazenhalk, G. D., Gestautas, J., Seto, P., Filleti, S., DeGroot, L. J., and Rapoport, B. (1987) Mol. Endocrinol.

1,856-861. We are indebted to Drs. Naofumi Ishikawa and Kunihiko Ito, Ito Thyroid Clinic and Hospital, Tokyo, Japan, and to Dr. Gerald0 Medeiros-Neto, Thyroid Laboratory, University of Sao Paulo, Brazil, for providing the Graves’ thyroid tissue used for the isolation of purified human TPO.

REFERENCES 1. Taurog, A. (1986) in Werner’s The Thyroid (Inghar, S. H., and Braverman, L. E., Eds), 5th ed., pp. 53-97, J. B. Lippincott Co., Philadelphia. 2. Hosoya, T., and Morrison, M. (1967) J. Viol. Chem. 242, 282828.35. 3. Coval, M., and Taurog, A. (1967) J. Biol. Chem. 242, 5510-5523. 4. Taurog, A., Lothrop, M. I,., and Estabrook, R. W. (1970) Arch. Biochem. Hiophys. 139,221-229. 5. Pommier, .J., DePrailaune, S., and Nunez, J. (1972) Biochimie (Paris) 54, 483-492. 6. Rawitch, A. B., Taurog, A., Chernoff, S. B., and Dorris, M. 1,. (1979) Arch. Biochem. Bioph.ys. 194, 244-257. 7. Ohtaki, S., Nakagawa, H., Nakamura, M., and Yamazaki, I. (1982) J. Bid. C’hem. 257,761-766. 8. Nakagawa, H., Kotani, T., Ohtaki, S., Nakamura, M., and Yamazaki. I. (1985) Hiochem. Biophys. Res. Commun. 127, 8-14. 9. I,ukat, G. S., Jabro, M. N., Rodgers, K. R., and Gaff, H. M. (1988) Riochim. Hiophys. Acta 954, 265-270. 10. Alexander, N. M. (1977) Endocrinoloqy 100, 1610-1620. 11. Czarnocka, B., Ruf, .J., Ferrand, M., Carayon, P., and Lissitzky, S. (1985) FEBS Lett. 190,147-151. 12. Ohtaki, S., Kotani, T., and Nakamura Y. (1986) J. Clin. Endocrinol. Metab. 63, 570~576. 13. Yokoyama, 844.

N.. and Taurog,

A. (1988) Mol. Endocrinol.

2, 838-

I,. (1985)

19. Kotani, T., Umeki, K., Matsunaga, S., Kato, E., and Ohtaki, (1986) J. Clin. Endocrinol. Metab. 62,928-933. 20. Kohno, Y., Hiyama, Y., Shimojo, N., Niimi H., Nakajima, Hosoya, T. (1986) Clin. Exp. Immunol. 65,534-541.

S.

H., and

S., Anelli, S., Ruf, J., Bechi, R., Czarnocka, B., Lom21. Mariotti, bardi, A., Carayon, P., and Pinchera, A. (1987) J. Clin. Endocrinol. Metab. 65,987-993. 22. Dohle, N. D., Banga, J. P., Pope, R., Lalor, McGregor, A. M. (1988) Immunol. 64,23-29.

E., Kilduff,

P., and

23. Yokoyama, N., Taurog, A., and Klee, G. G. (1988) J. Clin. Endocrinol. Metnb. 68, 766-773. 24. Yokoyama, N., Taurog, Metab., in press.

A., and Klee, G. G. J. Clin. Endocrinol.

25. Hamada, N., Jaeduck, N., Portmann, I,., Ito, K., and Defiroot, L. J. (1987) J. Clin. Endocrinol. Metab. 64, 230-238. 26. Morrison,

M., and Bayse, G. S. (1970) Biochemistry

9, 2995-3000.

27. Nagataki, S., IJchimura, H., Ikeda, H., Kuzuya, N., Masuyama, Y., Kumagai, L. F., and Ito, K. (1977) J. Clin. Invest. 59, 723729. 28. Nagataki, S., IJchimura, H., Masuyama, Endocrinology 92,363-37 I.

Y., and Nakao, K. (1973)

29. Rawitch, A., Pollock, G., and Taurog, A. (1989) 18th Annual Meeting of the European Thyroid Assoc. Copenhagen, June 2630, Abstract No. 160. 30. Freifelder, D. (1982) Physical Biochemistry: Application to Biochemistry and Molecular Biology, 2nd ed., p. 286, W. H. Freeman, New York. 31. Ludgate, M., Mariotti, S., Lihert, F., Dinsart, C., Piccolo, P., Santini, F., Ruf, ,J., Pinchera, A., and Vassart, G. (1989) J. Clin. Endocrinol. Metnb. 68, 1091l1096. 32. Kimura, S., and Ikeda-Saito, &met. 3.113120.

M. (1988) Proteins.. Strut.

Funct.