Developmental expression of extra-embryonic endodermal cytockeletal proteins

THEJOURNALOF BIOLOGICAL CHEMISTRY Val. 257, No. 7, Issue of April 10, pp. 3414-3421. 1982 Printed in U.S.A.

Developmental Expression of Murine Extra-embryonic Endodermal Cytoskeletal Proteins* (Received for publication, September 29, 1981)

Robert G. OshimaS From the Department

of

Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139

Murine extra-embryonic endodermal cell lines de- cytoskeletal proteins (Endo B)’ is distinct from tubulin, virived from either teratocarcinomas or mouse embryos mentin, desmin, glial fibrillar acidic protein, neural filament contain a cytoskeletal protein (Endo A) of M, = 55,000. protein, and keratin. This study describes immunochemical Endo A was immunoprecipitated from [36S]methionine-studies on the second endodermal cytoskeletal protein (Endo labeled lysates of three parietaI endodermalcell lines, A) that indicates it is also distinct from other previously a presumptive visceral endodermal cell line, and afetal described cytoskeletal proteins and,like Endo B, is induced in hepatoma cell line, but not from fibroblasts, myoblasts, embryonal carcinoma cells treated with retinoic acid. In aderythroleukemic cells, neuroblastoma cells, keratinocytes, or embryonal carcinomacells. Embryonal carci- dition, Endo A is identified as the major antigenic protein in noma cells induced to differentiate by exposure to ret- parietal endodermal cell lines that is recognized by a monoinoic acid synthesized increased amounts of Endo A clonal antibody made by other investigators against trophecapproximately48 h after exposure to the inducer. Two- todermal cytoskeletal proteins (10). dimensional gel analysis of immunoprecipitated samMATERIALS AND METHODS ples confirmed that Endo A is distinct from vimentin All cell lines and cultureconditions have been described previously and murine keratinocyte proteins recognized by two (9). Log phase cultures were rinsed twice with methionine-free Duldifferent keratin antisera. Comparison by two-dimen- becco’s modified Eagle’s media and incubated for 4 h at 37 “C in sional gel electrophoresis of immunoprecipitated Endo methionine-free DME medium supplemented with 50 pCi/ml of [35S] A labeled with either [3sS]methionineor[32P]ortho- methionine (lo00 Ci/mmol, Amersham), 10% fetal bovine serum, phosphate indicated that the multiple forms of EndoA penicillin, streptomycin, and glutamine. The cells were then rinsed resolved by isoelectric focusing were due, at least in three times with cold PBS and dissolved in 1 or 2 ml (depending on part, to phosphorylation. Serine was identified as the the cell density) of 0.1% SDS, 10 mM Tris-HC1, pH 7.2, 3 mMMgC12, mM CaC12,0.5 mM phenylmethylsulfonyl fluoride, 1 mM N-ethylphosphorylated amino acid. EndoA was the only major 0.1 maleimide, and 0.19 trypsin inhibitor unit/ml of Aprotinin (Sigma). antigenic protein found in a parietal endodermal cell Protease-free DNase I (11)was added to 10 pg/ml and after approxline which was recognized by a monoclonal antibody imately 30 s to 1min, SDS was added to a final concentration of 0.5%. preparedby other investigators against trophoblast The lysate was transferred to a glass tube containing 0.1 volume of cytoskeletons. The results indicate that Endo A, like 0.1 M EDTA, pH7.2, and heated at 100 “C for 2 min. After cooling on the previously described Endo B protein, is distinct ice, a 0.1 volume of a solution of10% Nonidet P-40 detergent was from other cytoskeletal proteins and will be useful as added. The lysates were stored frozen a t -85 “C until use. Endo A was purified as previously described (9). Rabbit antiserum a marker of the differentiation of murine embryonal was produced by the immunization of a single adult female New carcinoma cells to extra-embryonic endoderm. Zealand white rabbit with approximately 200 pg of Endo Aemulsified in complete Freund‘s adjuvant and a booster of approximately 170 pgof Endo A. Immune sera was recovered 10 weeks later. Rabbit antiserum to Endo B was described previously (9). Rabbit antiserum The differentiation of murine embryonal carcinoma cells in to vimentin (12) was a gift from Dr. Richard Hynes. Rabbit antiserum vitro can be considered a model system of normal early to human stratum corneaum keratin (13) was a gift from Dr. Howard mammalian development (1-3).This alternative to dissected Green. Guinea pig antiserum to bovine muzzle keratin (desmosomeembryos can provide sufficient material for molecular and associated tonofdaments) (14) was a gift from Dr. Werner Franke. biochemical studies of the earliest differentiated murine cell Affinity-pursed, chicken immunoglobulin to the rat liver nuclear types. Several culturedembryonal carcinoma cell lines differ- lamina A and C proteins(15, 16) was obtained from Dr. Larry Gerace. The ratTROMA 1 monoclonal antibody (IO)was kindly provided by entiate to cells resembling parietal endoderm either sponta- Dr. Rolf Kemler. neously (4-6) or after exposure to retinoic acid (7,8). Recently, Immunoprecipitation analysis shown in Fig. 2 was performed as this laboratory has described the identification and isolation previously described (9). All other immunoprecipitations were perof two apparently unique cytoskeletal proteins from a parietal formed as previously described except for the following modifications. endodermal cell line (9). Electrophoretic and immunoprecip- The final detergent concentrations of the lysates were increased to 0.5% SDS and 1%Nonidet P-40. Incubation time with immune sera itation analyses, amino acid composition data, andpartial was increased to 2h on ice. The SaCI-immune complexes were peptide mapping indicated that one of theseendodermal washed twice with 0.5% SDS, 1%Nonidet P-40, 10 pg/ml of soybean trypsin inhibitor (Worthington), twice with 0.5 M NaCI, 50 mM TrisHCl pH 7.2, 5 mM EDTA, 0.05% Nonidet P-40, 10 pg/ml of soybean * Supported by Grant CA 27580 from the National Cancer Insti- trypsin inhibitor, and once with NET buffer (0.15 M NaCI, 50 mM tute, United States Department of Health, Education and Welfare. The costs of publication of this article were defrayed in part by the ’ The abbreviations used are: Endo A andEndo B, extra-embryonic payment of page charges. This articlemust therefore be hereby endodermal cytoskeletal proteins A and B, respectively; DME, Dulmarked “adoertisement” in accordance with 18 U.S.C. Section 1734 becco’s modification of Eagle’s media; PBS, phosphate-buffered saline; IEF, isoelectric focusing; SaCI, formalin-fixed Staphylococcal solely to indicate this fact. $Present address, La Jolla Cancer Research Foundation, 10901 aureus Cowan strain I; SDS, sodium dodecyl sulfate; dansyl, 5-diNorth Torrey Pines Rd., La Jolla, CA 92037. methylaminonaphthalene-1-sulfonyl.

34 14

A

Protein Cytoskeletal Endodermal Tris-HCI, pH 7.2, 5 mM EDTA) containing 0.05% Nonidet P-40 and 10 pg/ml of soybean trypsin inhibitor. Samples were solubilized in SDS sample buffer (18). An indirect immunoadsorbent was used for chicken immunoglobulins and rat TROMA 1 monoclonal antibody (17). An excess of rabbit anti-chicken IgG or anti-rat IgG antiserum (Miles Laboratories, Inc., Elkhart, IN) was incubated with SaCI for 1 h a t 4 "C.The bacteria were washed as described for the immunoprecipitation procedure, resuspended in 0.05% Nonidet P-40, 1 mg/ml of ovalbumin in NET buffer, and stored a t -85 "C. Lysates used for the chicken antibodies were fust cleared of nonspecific reactants with 75 ~1 of the appropriate indirect immunoadsorbent(10%w/v suspension) then incubated with 25 pl of affinity purified antibodies (AzMl= 0.5). An additional 75 pl of the indirect immunoadsorbent was used to recover the immune complexes. Fifty pI of the indirect immunoadsorbent for rat IgG was used to clear lysates of nonspecific reactants and to recover the immune complexes. Fifty p1 of the TROMA 1 hybridoma culture media was used as immune sera. The immune complexes of bothindirectimmunoprecipitationprocedures were washed as described above for the direct method. Immunoprecipitated proteins were analyzed by polyacrylamide gel electrophoresis (15% acrylamide, 0.04% bisacrylamide) in the presence of 0.1% SDS (18)as modified by Thomas and Kornberg(19). Gels 25 cm in length and1 mm in thickness were used. Flat bed isoelectric focusing was performed by the procedure of Singer et al. (20) as modified by Shackelford and Strominger (21) in a Bio-Rad apparatus (model 1405). IEF gels contained 7% acrylamide and 0.18% bisacrylamide, 8.1 M urea (ultrapure, Schwarz-Mann), 2% NP-40,2% (w/v) ampholytes (LKB Instruments, Inc., Rockville, MD) composed of 30% pH 4 to 6,308 pH5 to 8, and 40% pH 3.5 to 10. Gels were polymerized with 0.016% ammoniumpersulfateand 0.04% N,N,N,N"tetramethylethylene diamine. Gelpiecesfrom the first dimension SDS gel were excised with razor blades using dansylated bovine serum albumin and ovalbumin (22) as markers. Dansylated proteins were visualized by UV light illumination. After equilibration for 1 h in 6 M urea, 0.08% pH 9 to 11 ampholytes, and 0.08% pH 3.5 to 10 ampholytes, thegel strips were loaded directly on the IEF gel near hrs + R A 24 012 1 2 1

3

43

4

5

6

72

012 1

7 . 8

0

101211

24 48

72

1315 14

*

3415

the acid electrode. A circulating water bath (8 "C) was attached to the cooling bed to dissipate heat. The IEFgel was run a t 1OOO V for 8 h. IEF gels were fixed in 12.5% trichloroacetic acid, 4% sulfosalicylic acid, washed in 50% methanol, 10%acetic acid to remove precipitated Nonidet P-40, and fluorographed as described by Chamberlain (23). Analysis of the proteolytic fragmentsof Endo A, Endo B, and the proteins recognized by keratin antiserawere performed by the method of Cleveland et al. (24) modified as previously described (9). :"PO.,labeled Endo A and EndoB proteins were purified by immunoprecipitation and SDS-gelelectrophoresis. PF HR9cells were labeledfor 16 h in phosphate-free DME media supplemented with 0.5 mCi/ml of :"PO4 (carrier-free, New England Nuclear), 10% fetal bovine serum, penicillin, streptomycin, and fresh glutamine. The cells were lysed and EndoA or EndoB immunoprecipitated as described above. After electrophoresis in the presence of SDS,the proteins were recovered from gel slices by electroelution into dialysis bags. The protein solutions were centrifuged at 10,OOO X g for 10 min to remove a small amount of debris andlyophilized. Recovery of the :l'PO.I-labeled Endo A and B proteins ranged from 78 to 95%. The protein was redissolved in 200 pl of water, precipitated with 9 volumes of cold acetone, and recovered by centrifugation a t 10,OOO X g for 10 min. The resulting pellet was washed twice with cold 10% trichloroacetic acid to remove precipitated glycine, twice with ether, and once with acetone. The precipitate was dried under reduced pressure and resuspended in 100 pl of 10 mM NHdHCO:,, 1% 2-mercaptoethanol,and 100 pg/ml of proteinase K (Merck). After 24 h a t 37 "C, the samples were lyophilized, then redissolved in water, and lyophilized three times, consecutively. The proteins were dissolved in 100 1-11 of 6 N HCI and hydrolyzed a t 110 "C in evacuated hydrolysis tubes for 1 h. The HCI was removed under reduced pressure. The resulting hydrolysis products were dissolved in a mixture of phosphoserine. phosphotyrosine, andphosphothreonine,each at 1 mg/ml. in the first dimension electrophoresis buffer. Acid hydrolysates were analyzed on cellulose thin layer plates as described by Hunter and Sefton (25). Standards were visualized by ninhydrinstaining (0.4%in acetone).Small amounts of ['%]methionine-labeled protein were spotted and dried to permit alignment of the film and theplate. The plates were exposed to XAR-5 film for 8 days with an intensifier screen (Dupont). RESULTS

The EndoA protein was purified fromthe PFHRS parietal endodermal cells as previously described (9). An antiserum to the EndoA protein was produced by immunization of a rabbit with the SDS gel-purified Endo A protein. This antiserum immunoprecipitated a protein of the size of Endo A ( M , = 55,000) from PFHRS endodermal cell lysates. Prior incubation of the sera with purified Endo A protein completely blocked the precipitationof [''5S]methionine-labeledEndo A (data not shown). Fig. 1 shows a fluorographic exposure of an SDS gel containing the radiolabeled proteins immunoprecipitated by L... Endo A and Endo B antisera from embryonal carcinoma cells .L. after varying times of exposure to retinoic acid. The Endo A and B proteins were synthesized in increased amounts apV " A " proximately 48 h after exposure to lo-" M retinoic acid. This Bcorresponds well to the time of detection of increased numbers of differentiated cells which secrete plasminogen activator and stain strongly with laminin antiserum (26). Immunofluorescence staining of retinoic acid-induced embryonal carcinoma cells with Endo A antiserum revealed the typical filamentous staining pattern previously described for the TROMA1 antibody (10) and the Endo B antiserum ( 9 ) (data not shown). The Endo A protein appears to be a good candidate marker of a differentiated functionwhich is expressedwhen emFIG. 1. Induction of Endo A and EndoB synthesis by retinoic acid. F9.22 embryonal carcinoma cells were exposedto IO"' M retinoic bryonal carcinoma cells differentiate to parietal endodermacid for varying lengths of time, washed, labeled with ["S]methionine like cells. Fig. 2 shows the results of immunoprecipitation and gel for 4 h, and lysed. 10' cpm of acid-insoluble radioactivity wassubjected to immunoprecipitation analysis with 10 pI of antiserum to Endo A electrophoresis of 15 different murine cell lines and HeLa (lunes 2 to 8) or Endo B (lunes 9 to 15). PFHR9 endodermal cell cells. The PFHRS, PYS, and MB2 parietal endodermal lines lysates were used for the samples separated in lanes 8 and 15. Marker all contained immunoprecipitable Endo A. In addition, the proteins and immunoprecipitated samples were separated in a 15% MB4 presumptive visceral endodermal cell line (9) and the acrylamide, 0.04% bisacrylamide slab gel containing 0.1% SDS and visualized in the dried gel by fluorographic exposure of XAR-5 X-ray BWl-J fetal hepatomacell line contained immunoprecipitable film for 3 days. Lanes 1 and 16 are marker proteins. V,vimentin; A, Endo A. Two fibroblast lines, a myoblast line, Friend erthroleukemic cells, and neuroblastomacells were negative as were Endo A; B, Endo B; Ac, actin; RA, retinoic acid.

Protein Cytoskeletal Endodermal

3416

A

compare the proteins recognized by the Endo A antiserum with the previously described Endo B, vimentin and keratin proteins, [:'5S]methionine-labeledlysates of HeLa cells, BW 1L, (4 J hepatoma cells, MB4 presumptivevisceral endodermal cells, and PFHR9 parietal endodermal cells were immunoprecipitated with keratin, vimentin, Endo A, and Endo B antiserum. Fig. 3A shows a fluorographic exposure of a SDS-polyacrylamide gel used to separate the proteins recognized by the different antiserum.All four cell lines contained vimentin and proteins reactive with Endo A and Endo B antisera. HeLa Endo A and Endo B-like proteins migrate faster than their murine counterparts. Some EndoA was found in the EndoB V' immunoprecipitates from the murinecell lines as previously A noted (9). Keratin antiserum failed to recognize proteins of B the size of vimentin, Endo A, and Endo B in either PFHR9 I C cells or BW1-J cells. However, a small amount of protein whichco-migratedwith Endo Awasfoundin the keratin antiserum immunoprecipitate of MB4 cells. With extremely FIG. 2. Identification of Endo A in murine extra-embryonic long exposures,a very small amountof Endo A wassometimes seen in the keratin antiserum immunoprecipitates of PFHR9 endodermal, fetal hepatoma, and HeLa cell lines. Logphase cultures of the indicated cell line were labeled with ["'S]methionine cells (Fig. 5A). Keratin antiserum recognized two HeLa cell for 4 h. Approximately 2 X 10' cpm of acid-insoluble radioactivitywas proteins of size similar to but not identical with the HeLa cleared with 5 pl of preimmune serum and 50 pl of a 10%suspension Endo A and B proteins. of SaCI, then immunoprecipitated with 5 pI of Endo A antiserum, and A portion of a second gel loaded with aliquots of the same 50 pl of a 10%suspension of SaCI. Reactive protein were separated by SDS-acrylamide gel electrophoresis and visualized by fluorography. samples shown in Fig. 3A was excised for further analysisby isoelectric focusing. The gel strip, indicated by the brackets Lanes 1 and 20 contain marker proteins, previously identified as actin and containing proteins of M , = 43,000 to 68,000 in Fig. 3A, (Ac), Endo B ( B ) ,Endo A ( A ) ,and vimentin (V).Lanes 2 , 3 , 4 , and 5 show embryonal carcinoma cell lines: F9.22, FOT5, PCCAO,and was equilibrated and loaded directly upon a flat bed IEF gel. PSAI, respectively. Lanes 6, 7, and 8 show parietal endoderm-like A fluorographic exposure of the IEF gel is shown in Fig. 3B. celllines of PFHR9,PYS,andMB2, respectively. Lane 9, MB4 Some lateraldiffusion was apparent but each protein band of presumptive visceral endodermal cell line. Lanes 10 and 1 I , fibroblast the fmt dimension S D S gel was resolved into multiple forms. cell lines D 5 and STO, respectively. Lane 12, myoblast line984,cl 10. Lane 13, Friend erytholeukemia cellline 745.Lane 14, fetal hepatoma Vimentin focused in approximately the sameposition regardless of the cell line of origin. Similarly, all murine Endo A line BWl-J. Lane 25, neuroblastoma NB-AI. Lane 16, HeLa cells. Lanes 17, 18, and 19, PYS cells immunoprecipitated with Endo B proteins focus identically. However, the HeLa cell Endo A antiserum, Endo A antiserum, and normal rabbit serum, respectively. proteins focus ina more acidic positionthan the murine Endo A. Similarly, the HeLa cell Endo B-like protein is distinthree embryonal carcinoma cell lines. One embryonal carci- guished from the murine Endo B-focusing pattern. Finally, noma cell line was weakly positive. Very long exposures of the two protein bands immunoprecipitated from HeLa cell similar gels reveal weak reaction with other embryonal carci- lysates by keratin antiserum focus as two distinct families of noma lines as well. However, allthe embryonal carcinoma cell bands.Additional experiments wereperformedbyloading lines examined are contaminated with varying but low num- single gel segments containing the HeLa cell keratins in an bers of spontaneously appearing differentiated cells which orientation orthogonal to the direction of the SDS-gel electroappear parietalendodermal-like on thebasis of their secretion phoresis. In such an orientation, the separationof the HeLa of plasminogen activator and reaction with laminin antiserum cell keratins according to size is retained in the IEF gel and (6, 26). Endo A found in embryonal carcinoma cultures may the proteins focus as spots instead of bands. These experibe due to either the contamination with differentiated cells ments permitted the identification of the more acidic HeLa which contain EndoA or a low level of synthesis by the stem keratin proteins as the smaller molecular weightform and the cells or both. The minor bands beneath Endo A of several more basic keratins as the largermolecular weight form. The positive cell lines was due to partial proteolysis. Subsequent more basic HeLa cell keratins areof similar size to HeLacell modifications of the immunoprecipitation procedure elimi- Endo A (Fig. 3A, lanes 2 and 4) but focus differently than the nated this problem. The results shown in Fig. 2 were con- HeLa cell Endo A. Thus, HeLa cells appear to contain Endo f m e d in independent experiments and evenvery long expo- A and Endo B-like proteins in addition to vimentin and two sures did not reveal any EndoA in negative cell lines except keratin groups. for embryonal carcinoma cells. The immunoprecipitation of Fig. 4 shows the results of immunoprecipitations of ["5S] [:"S]methionine-labeled HeLa cell lysates with Endo A anti- methionine-labeled, cultured murine keratinocytes with two serum resulted in the identification of a protein band which different keratin antisera. One keratin antiserum was prepared migrated slightly faster than murine EndoA. However, prior against extracted human stratum corneum keratins (13)while incubation of the antiserum with purified Endo A abolished the otherwas prepared againstbovine desmosome-associated the recovery of the HeLa cell Endo A-like protein (data not tonofilaments (14). Bothantisera recognize three sizes of shown). This results suggests that the human and murine proteins, designatedas keratins a,b, and c, in murine keratinEndo A-like proteins are immunologically related. ocytes. These proteins appear to correspond to the M, = HeLa cells contain both keratin and vimentin as interme- 58,000,52,000, and 45,000 murine keratin bands described by diate filament cytoskeletal proteins (27). Previous studies by Sun and Green(13).It is evident that the three keratin bands indirect immunofluorescence indicated the EndoA and Endo do not co-migrate with Endo A or Endo B. Both the antiB proteins are found in the cytoskeleton of endodermal cells human keratin serum and the anti-bovine tonofilament sera in a filamentous network very similar to that seenfor inter- appear to recognize the same murine keratinocyte proteins. mediate filament proteins of other cell types (9). In order to Isoelectric focusing of the immunoprecipitated keratins in a I

2 3 4 5

6 7 8

.*

S l O l l l 2 13141516171819 20

A

Protein Cytoskeletal Endodermal

(

n

HeLa k v a 3

2 1 ,l_

)(

b

4

5

k

v

BW1-J a b

6

7

8

n

9

)(

k

v

10 11

12

. " . " " I " ~ ~ ~ . . u ~ " - - _r--

n

k 16

13 14 15

- ..

~ ....-

)(

MB4 a b

3417

PFHRQ ) v a b 19 20 -

17 18

- -

B

Ac

( n

k

HeLa v

a

1

2

3

4

)( b

k

5

6

BWI-J v a b 7

8 10 9

)( n

k

MB4 v

11

a 12

b 13

n 14

)

( k 1517 16

PFHRS a v 19

18

) b 20

acid

base FIG. 3. Two-dimensional gel analysis of Endo A. A, 2-day fluorographic exposure of a polyacrylamide gel containing SDS. Approximately 10' cpm of acid-insoluble radioactive lysateof HeLa cells, BW1-J fetal hepatoma cells, MB4 presumptive visceral endodermal cells, or PFHPS parietal endodermal cells were immunoprecipitated with 10 pl of either nonimmune rabbit serum (n),human keratin antiserum ( k ) , NiL8-hamster cell vimentinantiserum ( u ) , Endo A

antiserum (a),or Endo B antiserum ( b ) .Lane 20 contained marker proteins vimentin (V),Endo A ( A ) ,Endo B ( B ) ,and actin (Ac).B, 4day fluorographic exposure of flat bed isoelectric focusing gel of samples shown in A. A portion of a gel (indicated by the brackets in A ) was excised from a duplicate gel of that used in A, equilibrated, and loaded directly on theIEF gel. After electrophoresis thegel was fixed, processed for fluorography, dried, and exposed to x-ray film.

A

Endodermal Cytoskeletal Protein

34 18

second dimension confirmed that both keratin antisera recognized the sameset of proteins (data notshown). Also shown in Fig. 4 (lane 2) is the resulting protein immunoprecipitated from endodermal cells by antibodies preparedagainst the nuclear lamina protein described by Gerace and Blobel (15). The nuclear lamina protein has similar solubility properties to EndoA and Endo B but is clearly larger in size. Fig. 5A shows the extentof immunological cross-reaction of keratin antiserum with endodermal cells. Only after anexposure 10 times longer than necessary to detect immunoprecipitated Endo A was any radioactivity which co-migrates with Endo A visualized in the keratin antiserum immunoprecipitate of endodermal cells (Fig. 5A, lane 6). Considering the number of other proteins of similar intensities visualized a t this length of time of exposure and the strong reaction of this keratin antiserum with presumptive keratins in murine keratinocytes, it is difficult to conclude that keratin antiserum specifically recognizes Endo A. Fig. 5B shows the result of immunoprecipitation of murine keratinocytes with Endo A antiserum and keratin antiserum. No Endo A is detected in the keratinocyte 1

2

3

4

5

6

7

lysate. In addition, Endo A antiserum does not significantly cross-react with murine keratins a,b, or c. Fig. 6 shows the patternof ['5S]methionine-containingpeptides generated by the partial proteolytic digestion of immunoprecipitated Endo A, Endo B, and keratins a, b, and c. As expected, the patternsof peptides generatedfrom the proteins recognized by antiserum to bovine keratin (Sample 2) were very similar to those generated from the proteins recognized by antiserum to human keratin (Sample 4). The pattern of digestion products forEndo A was distinct from that of Endo B and the murinekeratin b or c. However, greater similarity was apparent between Endo A and keratin a. At the concentration of protease used in this experiment, a total of 12 peptides were detected for Endo A and 14 for keratin a. Five of the proteolytic fragments of Endo A appear to co-migrate withkeratina fragments. On the basis of the weak but detectable cross-reaction of keratin antiserum with Endo A (Fig. 3A, lane 11) and the modest similarity of proteolytic fragments of Endo A and keratin a, it appears that Endo A is distinct from epidermal keratin but possibly related to one of the molecular weight forms of keratin. Recently, Brulet et al. (10) described a monoclonal antibody, designated TROMA 1, which reacts with intermediate filament-like structures of trophoblast cells, PYS parietal endodermal cells, and HeLa cells. Fig. 7A compares the proteins immunoprecipitated by the TROMA 1 antibody from endodermal cells with Endo A, Endo B, and vimentin. It is apparent that the majorantigenic ["S]methionine-labeled protein recognized by the TROMA 1antibody co-migrated in an SDS gel with Endo A. Fig. 7 B shows the IEF pattern of A

a

k

B b

XB2 n n k 1

2

PFHR 0 3

4

5

6

7

8

k

-a

c r

9

.

XB2 a k 1

2

1 L -Ac -B -A ,V

FIG. 5. Immunoprecipitation of PFHRS endodermal cells with keratin antiserum and murinekeratinocytes with Endo A antiserum. A, approximately 10' cpm of either PFHRS parietal FIG. 4. Comparison of Endo A and Endo B with proteins immunoprecipitatedfrom murine keratinocytes by human keratin antiserum, and bovine keratin antiserum. Approximately 10' cpm of ['5S]methionine-labeled PFHRS lysate (lanes 2, 3, and 5)were immunoprecipitated with nuclear lamina protein antibody (lane 2 ) , Endo B antiserum (lane 3), or Endo A antiserum (lane 5). The same amount of XB-2 murine keratinocyte lysate was immunoprecipitated with either bovine keratin antiserum (lane 4 ) or human keratin antiserum (lane 6).Marker proteins are shown in lane 7: A, Endo A; B, Endo B; a, 6, and c, keratins a, b, and c, respectively. Proteins were separated by electrophoresis in the presence of SDS.

endodermal lysate or XB-2 keratinocyte lysate was immunoprecipitated with 10 pl of either nonimmune sera (n),bovine keratin antiserum ( h ) ,Endo A antiserum (a), or Endo B antiserum (6). Proteins were separated in a polyacrylamide gel containing SDS, and processed for fluorography. Lanes I and 2 contained samples immunoprecipitated from XB-2keratinocyte lysates. Lanes 4 to 8 contained samples from PFHRS parietal endodermal cell lysates. Lanes 3 and 9 contain marker proteins. Lane 4, a-fetoprotein antiserum. Exposure time, 12 days. V,vimentin; A, Endo A; B, Endo B; Ac, actin. B, shows a 2-day exposure of the result of immunoprecipitating XB-2 keratinocyte lysate with either Endo A antiserum or antiserum to human keratin.

Endodermal Cytoskeletal Protein A 1

B

2

3

a bc

A

4

a

A

34 19

1 2 3 4

bc

6

1

2

3 4 acid

V A

-0

V

B

B -

I* base

FIG. 7. Comparison of [3'S]methionine-labeled proteins of PFHRS cells immunoprecipitated by Endo A antiserum and TROMA 1 monoclonal antibody. A, fluorographic exposure of an SDS gel used to separate the proteins immunoprecipitatedwith Endo B antiserum (lane I ) , TROMA 1 monoclonal antibody (lane 2), Endo A antiserum (lane 3 ) ,or vimentin antiserum(lane 4). Exposure time, 3 days. B, fluorographic exposureof a flat bed IEF gel containing the samples shownin A . A gel slice from a duplicate gel of that shown in A was excised and loaded on an IEF gel. Exposure time, 12 days. A, Endo A B, Endo B; V, Vimentin. FIG. 6. Partial proteolytic fragmentationof ["S]methioninelabeled Endo A, Endo B, and keratins a, b, and c. PFHRS endodermal cell lysates(Samples I and 3) were immunoprecipitated with either Endo B antiserum or Endo A antiserum. The immunoprecipitated proteins were excised from an SDS gel and loaded into a second gel of 15% acrylamide, 0.4% bisacrylamide. Similargel pieces containing the proteins immunoprecipitated from XB-2 cell lysates by either antiserum to bovine keratin (Sample 2) or human keratin (Sample 4 ) wereloaded in the second gel perpendicular to the direction of electrophoresisof the first gel. Each sample was overlaid with 90 ng of staphylococcal protease V8. After electrophoresis, the gel was processed for fluorography and exposed for3 weeks. A and B indicate the immunoprecipitated Endo A and B proteins. a, b, and c indicate the immunoprecipitated keratin proteins. Arrows indicate Endo A and keratin a peptides whichmayco-migrate.Note the similarity of the peptide patterns of keratin proteins immunoprecipitated from keratinocytesby antiserum toeither bovine keratin (Sample 2 ) or human keratin (Sample 4).

Endo B. Phosphoserine was the only phosphorylated amino acid derivative detected in both proteins. DISCUSSION

The previous investigation of parietal endodermal specific proteins resulted in the identification, isolation and intracellular localization of the Endo A and Endo Bcytoskeletal proteins (9). In this study, an antiserum to Endo A, improved conditions for thespecific immunoprecipitation of intermediate filament proteins, and two-dimensional gel electrophoresis have been used to identify Endo A in a very limited number of differentiated cell lines and in embryonal carcinoma cultures induced to differentiateby exposure to retinoic acid. Recent studies by a number of investigators indicate that different differentiated cell types may contain immunologically distinct intermediate filament, cytoskeletal proteins (28). Endo A and the proteins immunoprecipitated by the TROMA The identification of Endo A in a relatively small number of 1antibody. The TROMA1 antibody recognized multiple IEF murine differentiated cell lines (Fig. 2) suggests that Endo A forms of Endo A. Longer exposures of the gel exposed in Fig. is not glial fibrillar acidic protein, neurofilament protein, despreviously described stability of 7B confirmed that the TROMA 1 antibody recognized all of min, tubulin, or keratin. The the multiple forms of Endo A. All of the IEF forms of Endo A theintermediate filament-likeimmunofluorescent staining contain the TROMA1 antigenic determinant and likely rep- pattern of Endo A to treatment with Colcemid (9, 10) reinforces the conclusion that EndoA is not tubulin orvimentin. resent modified forms of the EndoA polypeptide. Both EndoA and EndoB are phosphorylated as determinedThe identification of vimentin in endodermal cells (Fig. 3, A by immunoprecipitation fromcells labeled with :'2P04.Fig. 8A and B)confirms that Endo A is distinct from vimentin. The compares the IEF pattern of Endo A immunoprecipitated size and IEF pattern of Endo A are distinct from murine from cells labeled with either "PO, or [:'"S]methionine. It is keratinocyte keratins (Fig. 4). In addition, thelack of signifiapparent that thetwo most basic forms of Endo A were not cant cross-reaction of murine keratinocyte proteins with Endo labeled with :"PO4, while the more acidic forms were labeled. A antiserum (Fig. 5B) reinforces this conclusion. However, by partial proteolyticfragmenThe multiple IEF formsof Endo A are due, a t least in part, to the peptide patterns generated phosphorylation. Similarly, Fig. 8B shows that the more acidic tation of Endo A are similar to those derived from one murine form of Endo B was phosphorylated. Fig. 9 shows the results keratin. In addition, variable and generally weak but detectof phosphoamino acid analysis of "'P04-labeled Endo A and able cross-reaction of Endo A with keratin antiserum was

A

Protein Cytoskeletal Endodermal

3420

phosphorylated on serine residues helpsexplain the multiple forms of both proteins observed in isoelectric focusing gels. However, the functional significance of the phosphorylation 1 2 1 2 of Endo A and Endo B remains tobe determined. .. acid ,cid TheTROMA 1 monoclonal antibody wasproduced by immunization withtrophoblast cytoskeletal proteins (10). This reagent was shown to react with the cytoskeleton of trophoblast cells, the PYS endodermal cell line and HeLa cells, but not with a variety of other cell types. The demonstration that the TROMA 1 antibodyimmunoprecipitates Endo A from endodermal cell lysates suggests that trophoblast cells may also contain EndoA. The presence of Endo A inthe BW1-J fetal hepatoma cell line and in t.he MB2 embryonic cell line indicates that Endo A is not limited exclusively to parietalendoderm. A recentstudy of the reaction of the TROMA 1 antibody with mouse embryo and adult tissues indicates that until day 9 of development, trophectoderm and parietal and visceral extra embryonic endoderm are theonly cells stained by the antibody.After day 9, a variety of epithelial cells of all three germ layers are reactive including fetal liver but not adult skin (45). The identification of Endo A and Endo B-like proteins in HeLa cells (Fig. 2, lane 16, and Fig. 3) indicate that at least five distinct proteins (Endo A, Endo B, vimentin, and two forms of keratin) may be associated with the intermediate filaments found within these human tumor cells. However, further studies are necessary to determine if Endo A is a structural componentof intermediate filaments in endodermal base base cells. The time of increased synthesis of Endo A and Endo B in FIG. 8. Endo A and Endo B are phosphorylated. Approxi- embryonal carcinoma cultures treated with retinoic acid comately 10' cpm of acid-insoluble [J5S]methionine-labeled PFHRS incides with the appearance of differentiated cells which exlysate or 10" cpm of :"P04-labeled lysate was immunoprecipitated press other parietal endodermal functions. The differentiation with 10 1.11 of Endo A or Endo B antiserum. After electrophoresis in a polyacrylamide gel containing SDS, the Endo A and B proteins were of murine embryonal carcinoma cells to parietal endoderm is inmorphology (4, 7, 32, 33), a excised and the proteins were analyzed by isoelectric focusing. Gels associated with alterations were exposed to XAK X-ray film for 2 days with an intensifier screen. decrease or loss of tumorigenicity (3, 34), decreased alkaline A, Endo A antiserum immunoprecipitates; B, Endo B antiserum phosphatase activity(4,35,36),loss of a stage specific embryimmunoprecipitates. onic surface antigen (36, 37), and loss of one subtype of the

A

32p

B32p 35s

35s

"-

I

detected (Figs. 3 and 5). These data indicate that Endo A is A B distinct from murine epidermal keratin but possibly related. Comparison of the aminoacid compositions of different intermediate filament proteins (9), studies of the domain structures of bovine keratin and vimentin (29), and most recently and definitively,comparison of COOH-terminal amino acid sequences of porcine vimentin and desmin (30), indicate that immunologically distinct intermediate filament proteins can .P-t h r P-thr be closely related. Vimentin and desmin appear to be members "* of a multigene family(30) that mayinclude other intermediate filament proteins. ' P-tyr A recent report concluded that the PYS parietal endo8 .* dermal cell line contained keratin-like intermediate filament proteins on the basis of immunofluorescent staining(31).It is possible that fixation procedures used for immunofluorescence FIG. 9. Phosphoamino acid analysis of immunoprecipitated may preserve a more native protein structure than the immunoprecipitation proceduredescribed in this study. Keratin S2P04-labeled EndoA and Endo B. "*P04-labeledEndo A and antisera mayrecognize structural similarities between polym- Endo B were purified by immunoprecipitation followed by SDS-gel electrophoresis. The radiolabeled proteins were recovered by elecerized forms of keratin a and Endo A that are not present troelution, digested with protease K, and subjected to limited acid whentheproteinsare solubilized. However, sincegreater hydrolysis. The resulting hydrolysis products were mixed with nonspecificity of reaction was the primary criterion used to im- radioactive standards and separated by electrophoresis on thin layer prove this procedure, other techniques are necessary to eval- cellulose plates at pH 1.9 and pH 3.5. The phosphoamino acid uate the relatedness of different antigenic proteins. Primary standards were located by ninhydrin staining and the radioactive amino acid sequence information would clarify the relation- phosphoamino acids by exposure to x-ray film in the presence of an intensifier screen for 8 days. P-ser, P-thr, and P-tyr indicate the ship of Endo A to keratin a and other intermediate filament positions of the ninhydrin-stained standards of phosphorylated serine, proteins. threonine, and tyrosine, respectively. A, Endo A phosphoamino acids; The demonstration that both Endo A and Endo B are B, Endo B phosphoamino acids.

... ... . I

.-...."

Protein Cytoskeletal Endodermal

A

3421

10. Brulet, P., Babinet, C., Kemler, R., and Jacob, F. (1980) Proc. Natl. Acad. Sci. U. S. A . 77,4113-4117 11. Otsuku, A. S., and Price, P. A. (1974)Anal. Biochem. 62,180-187 12. Hynes, R. O., and Destree, A. T. (1978) Cell 13, 151-163 13. Sun, T.-T., and Green, H. (1978) J.Biol. Chem. 253,2053-2060 14. Franke, W. W., Weber, K., Osborn, M. , Schmid, E., and Freudenstein, C. (1978) Exp. Cell. Res. 116,429-445 15. Gerace, L., Blum, A., and Blobel, G. (1978) J.Cell. Biol.79,546566 16. Gerace, L., and Blobel, G. (1980) Cell 19,277-287 17. Shackelford, D., and Strominger, J. J.Immunol., in press 18. Laemmli, U. K. (1970) Nature (Lond.)227,680-685 19. Thomas, J. O., and Kornberg, R.(1975) Proc. Natl. Acad. Sci.U. S. A . 72,2626-2630 20. Singer, B. S., Morrissett, H., and Gold, L. (1978) Anal. Biochem. 85,224-229 21. Shackelford, D., and Strominger, J. (1980)J. Exp. Med.151,144165 22. Yphantis, D. A., and Talbot, D. N. (1971) Anal. Biochem. 44, 246-253 23. Chamberlain, J . P. (1979) Anal. Biochem. 98, 132-135 24. Cleveland, D. W., Fischer, S. G., Kirschner, M. W., and Laemmli, U. K. (1977) J. Biol. Chem. 252, 1102-1106 25. Hunter, T., and Sefton, B. M. (1980) Proc. Natl. Acad.Sci. U. S. A . 77, 1311-1315 26. Oshima, R., and Linney, E. (1980) Exp. Cell. Res. 126,485-490 27. Franke, W.W., Schmid, E., Osborn, M., and Weber, K. (1978) Proc. Natl. Acad. Sci. U. S. A . 75, 5034-5038 28. Lazarides, E. (1980) Nature 283,249-256 2 9 . Steinert, P. M., Idler, M. W., and Goldman, R. D. (1980) Proc. Natl. Acad.Sci. U. S. A . 77,4534-4538 Acknowledgments-I thank Dr. John Buchanan for laboratory 30. Geisler, N., and Weber, K. (1981) Proc. Natl. Acad.Sci. U. S. A . 78,4120-4123 space, use of equipment, and encouragement, Ms. Maria Felstinsky for technical assistance, Dr. Howard Green and Dr. Werner Franke 31. Paulin, D., Babinet, C., Weber, K., and Osborn, M. (1980) Exp. Cell. Res. 130,297-304 for keratin antisera, Dr. Richard 0. Hynes for vimentin antiserum, Dr. Larry Gerace for nuclear lamin protein antibodies, Drs. Rolf 32. Lehman, J. M., Speers, W. C., Swartzendruber, D. E., and Pierce, G. B. (1974) J. Cell. Physwl. 84, 13-28 Kemler and PhiUipe Brulet for TROMA 1monoclonal antibody, and 33. Chung, A., Estes, L., Shinozuka, H., Braginski, J., Lorz, C., and Dr. John Burrfor phosphoamino acid standards. Chung, C. A. (1977) Cancer Res. 37,2072-2081 34. Strickland, S., and Sawey, M. J. (1980) Deu. Biol. 78, 76-85 REFERENCES 35. Bernstine, E. G., Hooper, M. L., Grandchamp, S., and Ephrussi, B. (1973) Proc. Natl. Acad.Sci. U. S. A . 70, 3899-3903 1. Finch, B., and Ephrussi, B. (1967) Proc. Natl. Acad. Sci.U. S. A . 36. Strickland, S., Smith, K. K., and Marotti, K. R. (1980) Cell 21, 57,615-621 347-355 2. Hogan, B. L. M. (1977) Int. Reu. Biochem. 15,333-376 37. Solter, D., and Knowles, B. B. (1978) Proc. Natl. Acad. Sci. U. S. 3. Graham, C. F. (1977) in Concepts in Mammalian Embryogenesis A. 75, 5565-5569 (Sherman, M. I., ed) pp. 315-394, M. I . T. Press, Cambridge, 38. Oshima, R., Curiel, D., and Linney, E. (1980) J.Supramol. Struct. MA 14,85-96 4. Martin, G. R., and Evans, M. J. (1975) Proc. Natl. Acad.Sei. US. 39. Strickland, S., Reich, E., and Sherman, M. I. (1976) Cell 9, 231A. 72,1441-1445 240 5. McBurney, M. W. (1976) J.Cell. Physwl. 89,441-456 Chung, A. E., Jaffe, R., Freeman, I. L., Vergnes, J. P., Braginski, 40. 6. Oshima, R.(1978) Differentiation 11, 149-155 J . E., and Carlin, B. (1979) Cell 16,277-287 7. Strickland, S., and Mahdavi, V. (1978) Cell 15,393-403 8. Jetton, A. M., Jetten, M. E. R., and Sherman, M. I. (1979) Exp. 41. Solter, D., Shevinsky, L., Knowles, B. B., and Strickland, S. (1979) Deu. Biol. 70,515-521 Cell. Res. 124,381-391 Howe, C. C., and Solter, D. (1980) Deu. Biol. 77, 480-487 42. 9. Oshima, R. G. (1981) J.Biol. Chem. 256,8124-8133 43. Hogan, B. L. M. (1980) Deu. Biol. 76, 275-285 44. Adamson, E. D., and Ayere, S. E.(1979) Cell 16,953-965 * R. Lennox, R. G. Oshima, and L. Cohen, manuscript submitted 45. Kemler, R., Brulet, P., Schnedlen, M., Gaillard, J., and Jacob, F. for publication. (1981) J. Embryol. Exp. Morphol. 6 4 , 4 5 4 J. Tabor, and R. G . Oshima, manuscript submitted for publication. 46. Linney, E., and Levinson, B. B.(1977) Cell 10,297-304

H1 histone multigene family (38).' In addition, parietal endodermal cells synthesize increased amounts of plasminogen activator (7, 26, 39, 46), laminin (26, 40-43), type IV collagen (36, 44), and the Endo A and Endo B proteins. Endodermal functions expressed in embryonal carcinoma cultures treated with retinoic acid appear in a roughly coordinate fashion beginning approximately 48 h after exposure to the inducer (7, 26). Embryonal carcinoma cells treated with retinoic acid continue to proliferate at an average rate that would permit several cell divisions before the expression of parietal endodermal functions are detected (7,26). Strickland and co-workers have shown that theamount of differentiation observed in treated cultures increases with increasing time of exposure M (7, 34). and increasing concentration to a maximum of However, removal of retinoic acid after 24 h still results in a detectable increase in the population of cells which express endodermal functions 3 days later (34). Therefore, it appears that retinoic acid may initiate a program of events which take approximately 2 days to result in the stable, coordinate, and retinoic acid-independent expression of parietal endodermal functions. Because increased synthesis of the Endo A and Endo B proteins can be correlated with increased levels of biologically active messenger RNAs for these protein^,^ both of these gene products appear to be excellent candidates for further studies of the developmental control of endodermal differentiated functions.