Proc. Natl. Acad. Sci. USAVol. 82, pp. 34-37, January 1985Biochemistry

Isolation of tyrosine-O-sulfate by Pronase hydrolysis fromfibronectin secreted by Fujinami sarcoma virus-infectedrat fibroblasts

(protein sulfation/fibronectin separation/temperature-sensitive virus)

MING-CHEH LiU AND FRITZ LIPMANNThe Rockefeller University, 1230 York Avenue, New York, NY 10021

Contributed by Fritz Lipmann, August 30, 1984

ABSTRACT In a recent paper, we reported the loss oflarge amounts of protein-bound tyrosine sulfate after infectionof rat fibroblasts by avian sarcoma viruses. The analogy to thereported loss of surface fibronectin on malignant transforma-tion, which contained sulfate of unknown location, called ourattention to this compound. In a previous paper, we brieflyreported on isolation from the supernatant fraction of rat fi-broblasts infected by Fujinami sarcoma virus fibronectin thatyielded tyrosine-O-sulfate on Pronase hydrolysis. In this pa-per, we confirm and enlarge on this observation. Highly puri-fied fibronectin was obtained from the supernatant fractionsecreted by Fujinami sarcoma virus infected rat fibroblaststhat contained 1.52 residues of sulfated tyrosine per proteinmolecule after exhaustive Pronase hydrolysis. Assuming someloss during work up, this probably indicates 2 residues of thetyrosine sulfated per fibronectin molecule.

We recently reported (1) on the remarkable loss of tyrosinesulfated proteins from avian sarcoma virus-infected rat fi-broblasts. Since synthetic tyrosine-O-sulfate, as well as thistyrosine sulfate bound in proteins is very acid labile, a com-bination of enzymatic treatment and heating with Ba(OH)2had previously been generally used. Since, however, it ap-pears that by heating with Ba(OH)2 a considerable amount oftyrosine-O-sulfate is decomposed (2), we used instead thePronase hydrolysis for 48 hr at pH 7.8, which was found inour tests to be complete.As mentioned in our previous paper, preliminary experi-

ments using an analogous procedure yielded tyrosine-O-sulfate from fibronectin secreted by Fujinami sarcoma virus-infected rat fibroblasts (ts225-3Y1) grown at 34.5°C as de-scribed in ref. 1. We were eager to carry out these experi-ments because fibronectin is known to be lost from the sur-face of malignantly transformed cells and also has beendescribed to contain sulfate in unknown positions (3, 4).

In the present paper, we report experiments using the fi-bronectin released from ts225-3Y1 cells grown at permissivetemperature on a larger scale; then we found it to contain <2mol of tyrosine-O-sulfate per mol fibronectin in the S-Slinked dimeric molecule composed of equal parts.

EXPERIMENTAL PROCEDURESMaterials. Protein molecular weight standards were ob-

tained from Bethesda Research Laboratories. Gelatin Seph-arose-4B and protein A-Sepharose-CL-4B were products ofPharmacia Fine Chemicals. Pre-coated TLC cellulose plateswere from Brinkmann, carrier-free [35S]sulfuric acid and L-[ring-2,6-3H]tyrosine were from New England Nuclear. The

goat anti-rat fibronectin antiserum and Pronase of Calbio-chem-Behring were used. Tyrosine-O-sulfate standard wassynthesized by the procedure of Jevons (2). Bovine plasmafibronectin was purified as described below. L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin was fromWorthington. All other reagents were of the highest grade.

Cell Culture and Radioactive Labeling. The ts225-3Y1 ratembryo fibroblasts infected by a temperature sensitive (ts)mutant of Fujinami sarcoma virus were a gift from H. Hana-fusa, at this university. These cells were grown at the per-missive temperature of 34.50C in Dulbecco's modified Ea-gle's medium (DME medium) containing 10% calf serum(Flow Laboratories). Confluent cells were labeled with ei-ther [35S]sulfate (0.3 mCi/ml; 1 Ci = 37 GBq), [3H]tyrosine(50 ,uCi/ml), sulfate-free, or low tyrosine content (10%)DME medium containing 10% fibronectin-free (gelatin Seph-arose-treated) calf serum. Cells were incubated for 48 hr inlabeling medium for the production of a reasonable amountof secreted fibronectin; only a trace amount of secreted fi-bronectin was obtained from the medium after 24 hr incuba-tion time. Spent culture medium was removed after 48 hr forthe purification of secreted fibronectin.

Purification of Secreted Fibronectin. It should be men-tioned that for preparing electrophoretically homogeneoussecreted fibronectin by the one-step purification proceduredescribed below, serum-free DME medium was used in thefinal incubation of confluent ts225-3Y1 cells to prevent con-tamination arising from serum proteins. In radioactive label-ing experiments, however, 10% fibronectin-free calf serumwas used to reduce the background due to newly synthesized[35S]sulfated proteoglycans that tightly bind to the secretedfibronectin during the purification process and interfere con-siderably with the analysis of [35S]sulfated secreted fibro-nectin. The procedure of Chiquet et al. (5) was adopted forthe purification of secreted fibronectin from the spent cul-ture medium; it was applied directly onto a gelatin-Sepha-rose-4B column (1 X 4 cm) equilibrated with serum-freeDME medium. After the whole fraction had passed throughthe column, 10 ml ofDME medium was applied to elute non-adsorbing molecules. Elution was preceded by washing with50 ml of phosphate-buffered saline (Pi/NaCI) followed by100 ml of 0.5 M urea in Pi/NaCl to remove less tightly boundcontaminants. Fibronectin was then eluted by 25 ml of 4 Murea in P1/NaCl. The eluate was dialyzed extensively against10 mM Tris HC1 (pH 7.5) and concentrated by Amicon Dia-flo (PM 30 membrane) to -1 ml for further experiments.

Immunoprecipitation. For immunoprecipitation, this fibro-nectin fraction was incubated in Pi/NaCl with goat anti-fi-bronectin antiserum for 1 hr at room temperature. Protein A-Sepharose was added then and the mixture was shaken at4°C for 30 min. It was brought down by centrifugation with

Abbreviation: ts, temperature sensitive.

34

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

Proc. Natl. Acad. Sci. USA 82 (1985) 35

the immune complex bound to it, washed 5 times with Pi/NaCl, and placed in Laemmli sample buffer (6) for subse-quent electrophoresis.Enzymatic Hydrolysis to Tyrosine-O-Sulfate by Pronase. Fi-

bronectin fractions, as prepared or immunoprecipitated,were subject to electrophoresis on NaDodSO4/4.59-10%polyacrylamide gradient gel using the method of Laemmli(6). Thereafter, the gel was stained with Coomassie brilliantblue in 50% methanol/10% acetic acid, and destained with25% methanol/7.5% acetic acid solution. The gel was thendried in vacuum at room temperature. An autoradiographwas taken from the dried gel. Radioactive fibronectin bandswere located by autoradiography and were excised from thegel. Gel pieces containing fibronectin were further slicedinto thin strips and placed in individual test tubes. Two milli-liters of 50 mM ammonium bicarbonate solution containing150 Ag-ml-l of Pronase was added to each tube. The prepa-ration was shaken at 370C to allow the fibronectin in the gelto be hydrolyzed and eluted. After 24 hr, 0.2 ml of 1.5 mg m-1of Pronase in 50 mM ammonium bicarbonate was added andthe mixture was incubated for another 24 hr to ensure com-plete hydrolysis. The resulting gel eluate was separated bycentrifugation and lyophilized. To the lyophilysate, 30 Al ofcold tyrosine-O-sulfate standard solution (1 mg/ml) was add-ed. After carefully redissolving, 10 pl of the solution wasspotted on a TLC-cellulose plate (20 x 20 cm). The plate wasthen subject to high voltage electrophoresis (500 V; 90 min)in 5% acetic acid/0.5% pyridine, pH 3.5 (7), and air dried. Toseparate tyrosine-O-sulfate, it was then subject in the seconddimension to ascending chromatography in n-butanol/formicacid/isopropanol/H20 (3:1:1:1) (8). Upon completion of thechromatography, the plate was dried and sprayed with nin-hydrin solution (0.5% in acetone). Either the plate was auto-radiographed, or the ninhydrin spot of tyrosine-O-sulfate onthe plate was scraped off for radioactivity determination.

Miscellaneous Methods. Tryptic peptide mapping was per-formed on the purified fibronectin using the procedure ofTuazon et al. (9). Amino acid composition was determinedby use of a Durrum amino acid analyzer model D-500 afterhydrolysis with 6 M HCO at 110°C for 24 hr.

RESULTSFibronectin purified from spent serum-free culture mediumof ts225-3Y1 cells was judged to be homogeneous by NaDod-S04/polyacrylamide gel electrophoresis. As shown in Fig. 1,purified fibronectin migrated as a single protein band underreducing conditions with 1% 2-mercaptoethanol. At nonre-ducing conditions without the 2-mercaptoethanol, a majorband with slower mobility and a minor band migrating at thesame position as that under reducing conditions were detect-ed. This was also found in the case of bovine plasma fibro-nectin when co-electrophoresed (lanes 2 and 5). Secreted fi-bronectin was, however, different from bovine plasma fibro-nectin, which yielded two closely migrating protein bands ofabout equal intensity. Thus, it seems from these results that,rather than a heterodimer (10) as in the case of plasma fibro-nectin, the secreted fibronectin is composed of two equal-sized subunits held together also by interchain disulfidebonds. Furthermore, the molecular weight of secreted fibro-nectin was, based on its electrophoretic mobility, significant-ly higher than that of the bovine plasma fibronectin. In thisrespect, our secreted fibronectin is similar to cell surface fi-bronectin, which was reported similarly to be larger than theplasma fibronectin and composed of subunits of equal size.

Fig. 2 shows the autoradiograph of [35S]sulfate-labeled fi-bronectin fractions (purified from spent culture medium con-taining 10% fibronectin-free calf serum) after NaDodSO4/polyacrylamide gel electrophoresis. Under either reducingor non-reducing conditions, the fibronectin protein bandswere found to be radioactive (lanes 1 and 2). When immuno-

precipitated by goat anti-rat fibronectin antiserum, a radio-active band was again detected at the same position as thatof the purified fibronectin. Nonimmune control using normalgoat serum, however, revealed only a weak radioactive bandthat probably resulted from the extremely adhesive proper-ties of fibronectin. This, therefore, confirms that the puri-fied fibronectin is sulfated.To investigate the site of sulfation of the fibronectin mole-

cule, the radioactive fibronectin band on the dried polyacryl-amide gel was excised; protein molecules in the gel wereeluted and hydrolyzed by Pronase. As shown in Fig. 3, two-dimensional separation of the fibronectin Pronase hydroly-sate indicated tyrosine-O-sulfate to be the only radioactiveninhydrin-positive spot. Two weak radioactive spots of un-known identity were also detected. They are likely to be dueto the contaminating carbohydrate-bound sulfate, becausefibronectin also has high affinity toward sulfated proteogly-can (11). To gain further evidence that the purified fibronec-tin actually is sulfated on tyrosine, but not on the carbohy-drate side chain(s), ts225-3Y1 cells were labeled with[35S]sulfate in the presence of tunicamycin (1 ,xg/ml). Thisnon-glycosylated fibronectin, as tunicamycin has beenshown to inhibit asparagine-linked glycosylation (12, 13),was purified from spent culture medium and found to be sul-fated on tyrosine by the same procedure. Also 3H radioactiv-ity was detected on the tyrosine-O-sulfate spot when Pron-ase hydrolysate of fibronectin labeled with [3H]tyrosine wasanalyzed. These results further confirm the presence of sul-fated tyrosine in the purified secreted fibronectin.

Attempts to quantify the number of tyrosine residues sul-

1 2 3 4 5

Mr x lo-,

200

- 92.5

66.2

- 48

- .-31

- _ ~ - 21.514.4

FIG. 1. NaDodSO4/polyacrylamide gel electrophoresis patternsof purified ts225-3Y1 secreted fibronectin and bovine plasma fibro-nectin. Different samples were electrophoresed on a 4.5%-10%polyacrylamide gradient gel. Lanes: 1, ts225-3Y1 secreted fibronec-tin under nonreducing conditions; 2, bovine plasma fibronectin un-der nonreducing conditions; 3, Mr protein standards, cytochrome c(Mr, 12,300),B3-lactoglobulin (Mr, 18,400), a-chymotrypsinogen (Mr,25,700), ovalbumin (Mr, 43,000), bovine serum albumin (Mr,68,000), phosphorylase b (Mr, 92,500), myosin (H-chain) (Mr,200,000); 4, ts225-3Y1 secreted fibronectin under reducing condi-tions; 5, bovine plasma fibronectin under reducing conditions.

Biochemistry: Liu and Lipmann

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

36 Biochemistry: Liu and Lipmann

2 3 4

Ml x 10-3

_ 200

FIG. 2. Autoradiograph of 35S-labeled purified and immunopre-cipitated ts225-3Y1 secreted fibronectin fractions. Different sampleswere electrophoresed on a 4.5%-10% polyacrylamide gradient gel.Lanes: 1, purified fibronectin under nonreducing conditions; 2, puri-fied fibronectin under reducing conditions; 3, antiserum precipitatedfibronectin; 4, nonimmune serum control of lane 3.

fated per fibronectin molecule were undertaken using the[3H]tyrosine-labeled secreted fibronectin. It was determinedthat 0.84% of the total 3H radioactivity present in fibronectinPronase hydrolysate migrated to the tyrosine-O-sulfate spotduring the previously described two-dimensional analysis(Table 1). Assuming a molecular weight of 500,000-secret-ed fibronectin was shown to have a higher molecular weightthan plasma fibronectin (450,000)-and a tyrosine content of4.2 residues per 100 amino acid residues (Table 2), the num-

ber of sulfated tyrosine residues per intact fibronectin mole-

Tyr-O-S03

,

Q.

.I

u

r_X

suto

ct

Origin 5

Thin-layer chromatography

FIG. 3. Autoradiograph of Pronase hydrolysate of 35S-labeledpurified ts225-3Y1 secreted fibronectin after two-diniensional sepa-ration. Dashed line indicates position of tyrosine-O-sulfate standard,detected by ninhydrin staining.

Table 1. Fraction of tyrosine sulfated in [3H]tyrosine-labeledsecreted fibronectin

3H radioactivity,cpm % total

Total Pronase hydrolysate 10,286 100Tyrosine-O-sulfate* 86 0.84

*Amount of 3H radioactivity present in tyrosine-O-sulfate wasdetermined after two-dimensional analysis.

cule was calculated to be 1.52. This data we interpret to indi-cate that approximately one tyrosine residue is sulfated ineach of the two subunits of secreted fibronectin molecule,because tryptic peptide mapping of reduced [35S]sulfated fi-bronectin revealed only a single major radioactive spot (Fig.4). This finding is compatible with less than one tyrosine res-

idue being sulfated per fibronectin subunit, the value of 1.52being due to loss of sulfate during the analysis or to incom-plete sulfation.The amino acid composition of purified fibronectin togeth-

er with those of three plasma fibronectins are listed in Table2. It appears that the amino acid compositions of the threeplasma fibronectins are very similar, as shown in Table 2.The secreted purified fibronectin showed significant quanti-tative differences. The contents of proline, leucine, andphenylalanine were higher and glutamic acid (plus gluta-mine) and valine were lower than in plasma fibronectins.

DISCUSSIONThe recent introduction of sulfation as a widespread occur-rence in proteins seems to introduce a new possibility formodulation of activity. It is very likely that the striking stim-ulation by sulfation of the activity of the hormones gastrin(17) and cholecystokinin (18) indicates modulatory effects.Furthermore, the presence of nearly equal amounts of thesulfated and sulfate-free gastrin found in some preparations(19) appears quite indicative of such an effect.Another interesting feature of sulfation is that in proteins

so far it seems uniquely attached to tyrosine. Phosphoryl-ation, by contrast, preferably is attached in proteins to ser-

Table 2. Amino acid composition (residues per 100) of secretedrat fibronectin as compared with those of human, porcine, andbovine plasma fibronectins

Amino ts225-3Y1 Human Porcine Bovineacid secreted plasma plasma plasma

Asx 9.2 9.8 9.1 8.4Thr 10.3 10.0 10.4 9.7Ser 7.3 6.2 8.0 7.3Glx 9.3 12.5 11.4 12.1Pro 10.8 8.0 7.7 8.4Gly 8.2 8.4 8.4 8.0Ala 4.5 4.3 4.4 4.3Cys* 2.5 2.7 1.6 3.3Val 6.4 8.3 8.4 9.0Met 1.0 1.2 1.1 1.1Ile 4.5 4.3 4.6 4.4Leu 6.2 5.3 5.6 5.7Tyr 4.2 3.8 4.0 4.1Phe 3.1 2.3 2.2 2.1Lys 3.2 3.6 3.8 3.7His 1.9 1.9 2.1 1.9Arg 5.0 5.2 5.1 5.0Trpt 2.5 2.4 2.1 1.3

Human data from ref. 14; porcine data from ref. 15; bovine datacalculated from ref. 16.*Determined as half-cystine.tCalculated assuming the ratio of tyrosine to tryptophan to be 1.6.

Proc. NatL Acad Sci. USA 82 (1985)

I

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020

Proc. NatL. Acad. Sci. USA 82 (1985) 37

0

Cdgi

0.

oU

Thin-layer chromatography

FIG. 4. Autoradiograph of tryptic digest of [35S]sulfate-labeledts225-3Y1 secreted fibronectin after two-dimensional analysis. Highvoltage electrophoresis in the first dimension was performed in 1%ammonium carbonate (500 V; 1 hr) and chromatography in the sec-ond dimension was in n-butanol/acetic acid/water/pyridine(15:3:12:10).

ine, less often to threonine, and rather sparingly to tyrosine.However, tyrosine phosphorylation is 5-fold (or more) in-creased by the Rous sarcoma virus, which carries a phos-phokinase in the translational product of the transformingsrc gene, which according to Hunter and Sefton (20) is quitespecific for tyrosine.

In contrast to the mentioned stimulation of tyrosine phos-phorylation, we recently reported on a quite rapid reductionof tyrosine-O-sulfate-containing proteins by retrovirus-in-fected rat fibroblasts (1). This actually prompted our searchfor tyrosine-O-sulfate in fibronectin, which is known to belost on malignant transformation (21). The loss was found todeprive the tumor cells of general adhesion to various sur-faces and to each other, and caused changes in cell shape.An interesting genetic disease (metachromatic leukodys-

trophy) is due to the absence of hydrolysis of sulfate fromthe so-called brain sulfatide, a galactosulfate ceramide (22).The missing enzyme is called arylsulfatase A. This enzymeappears to need to combine with an adjuvant to act on brainsulfatide (23). The absence of the latter causes the disease bythe inability of the body to desulfate the brain sulfatide as thefirst step of normal degradation. The sulfatide accumulatesin the brain and all over the body because of deregulation ofthe balance between sulfated and desulfated brain sulfatide.The adjuvant is identified as a low molecular weight heatstable protein that binds to the sulfatide in a reversible reac-tion (23). It is present normally in lysosomes but absent in atype of metachromatic leukodystrophy (24).

As mentioned, cell surface fibronectin is lost from malig-nantly transformed cells (21). It is remarkable that the addi-tion of extracted cell surface fibronectin to transformed cellscauses an extensive normalization of cell morphology as de-scribed in refs. 25 and 26. However, it does not abolish theunrestricted growth.

This work was supported by Grant PCM8315-47 from the NationalScience Foundation, and Grant GM 13972-18 from the National In-stitutes of Health.

1. Liu, M. C. & Lipmann, F. (1984) Proc. Nati. Acad. Sci. USA81, 3695-3698.

2. Jevons, F. R. (1963) Biochem. J. 89, 621-624.3. Dunham, J. S. & Hynes, R. 0. (1978) Biochim. Biophys. Acta

506, 242-255.4. Ruoslahti, E., Jalanko, H., Comings, D. E., Neville, A. M. &

Raghaven, D. (1981) Int. J.fCancer 27, 763-767.5. Chiquet, M., Puri, E. C. & Turner, D. C. (1979) J. Biol. Chem.

254, 5475-5482.6. Laemmli, U. K. (1970) Nature (London) 227, 680-685.7. Hunter, T. (1980) Cell 22, 647-648.8. Ushiro, H. & Cohen, S. (1980) J. Biol. Chem. 255, 8363-8365.9. Tuazon, P. T., Merrick, W. C. & Traugh, J. A. (1980) J. Biol.

Chem. 255, 10954-10958.10. Vartio, T. & Vaheri, A. (1983) Trends Biochem. Sci. 8, 442-

444.11. Hynes, R. 0. & Yamada, K. M. (1982) J. Cell Biol. 95, 369-

377.12. Lehle, T. & Tanner, W. (1976) FEBS Lett. 71, 167-170.13. Olden, K., Pratt, R. M. & Yamada, K. M. (1978) Cell 13, 461-

473.14. Vuento, M. & Vaheri, A. (1979) Biochem. J. 183, 331-337.15. Isemura, M., Yosizawa, Z., Takahashi, K., Kosaka, H., Ko-

jima, N. & Ono, T. (1981) J. Biochem. 90, 1-9.16. Skorstengaard, K., Thogersen, H. C., Wibe-Pedersen, K., Pe-

tersen, T. E. & Magnusson, S. (1982) Eur. J. Biochem. 128,605-623.

17. Holstein, B. (1982) Acta Physiol. Scand. 114, 453-459.18. Bodanzsky, M., Natarajan, S., Hahne, W. & Gardner, J. D.

(1977) J. Med. Chem. 20, 1047-1050.19. Rehfeld, J. F., Hansen, H. F., Larsson, L.-I., Stengaard-Pe-

dersen, K. & Thorn, N. A. (1984) Proc. Natl. Acad. Sci. USA81, 1902-1905.

20. Hunter, T. & Sefton, B. M. (1980) Proc. Natl. Acad. Sci. USA77, 1311-1315.

21. Vaheri, A. & Mosher, D. F. (1978) Biochim. Biophys. Acta516, 1-25.

22. Kolodny, E. H. & Mosher, H. W. (1983) in The Metabolic Ba-sis of Inherited Diseases, eds. Stanburry, J. B., Wyngaarden,J. B. & Frederickson, D. S. (McGraw-Hill, New York), 5thEd., pp. 881-905.

23. Fischer, G. & Jatzkewitz, H. (1978) Biochim. Biophys. Acta528, 69-76.

24. Stevens, R. L., Fluharty, A. L., Kihara, H., Kaback, M. M.,Shapiro, L. J., Marsh, B., Sandhoff, K. & Fischer, G. (1981)Am. J. Hum. Genet. 33, 900-906.

25. Yamada, K. M., Yamada, S. S. & Pastan, I. (1976) Proc. Natl.Acad. Sci. USA 73, 1217-1221.

26. Ali, I. U., Mautner, V., Lanza, R. & Hynes, R. 0. (1977) Cell11, 115-126.

Biochemistry: Liu and Lipmann

Dow

nloa

ded

by g

uest

on

Mar

ch 1

1, 2

020