Development MM, 403-413 (1988)Printed in Great Britain © The Company of Biologists Limited 1988

403

Isolation of differentially expressed human cDNA clones: similarities

between mouse and human embryonal carcinoma cell differentiation

MICHAEL V. WILES*

Laboratory of Human Molecular Genetics, Imperial Cancer Research Fund, Lincoln's Inn Fields, London, WC2A 3PX, UK

•Current address: Basel Institute for Immunology, Grenzacherstrasse 487, Postfach, CH-4005 Basel, Switzerland

Summary

The study of early human development is of greatimportance but has been limited by the lack of suitablereagents. Recently, however, the human embryonalcarcinoma (EC) cell line NT2D1 has been isolated.This cell line will differentiate upon exposure toretinoic acid (RA). A cDNA library was constructedfrom poly(A)+ RNA derived from NT2D1 cells treatedwith 10~sM-RA for 7 days (ANT2D1 cells). By differ-ential cDNA screening, it was found that 1-12% ofANT2D1 cDNA recombinants screened detected anincrease in signal with 32P-cDNAs derived fromANT2D1 as compared with NT2D1.

To compare RA-induced differentiation of mouseand human EC cells, the ANT2D1 cDNA library wasrescreened with 32P-cDNAs derived from the mouseEC cell line F9 and the result compared with 32P-cDNA derived from F9 differentiated to parietal-endoderm (F9PE)-like cells and visceral-endoderm(F9VE)-like cells. Approximately 1-2% of theANT2D1 cDNA recombinants detected a differentialincrease in signal following differentiation of mouseEC cells to F9VE and/or F9PE. Of these homologousregulated sequences, 0-3 % were common to bothmouse and human EC cell RA-induced differentiation.

Five different cDNA clones were isolated that detecta marked increase (5- to 75-fold) in mRNA abundancefollowing RA-induced differentiation of NT2D1. Ofthese five clones, three detect homologous mRNAswhich also increase in abundance following differen-tiation of the mouse EC cell line F9 to PE- and/or VE-

• like cells; the other two clones do not detect sequencesin the mouse mRNAs tested. One clone shows hom-ology to SPARC, a gene known to be regulated duringmouse embryonic development. While another clone,SO5A, has a limited range of expression, beingdetected in F9VE and in a human parietal-endoderm-like cell, but not in F9PE and a human visceral-endoderm-like cell.

This work shows that there are both similarities anddifferences in mouse and human EC cell differen-tiation, and these cDNA clones provide some of thefirst reagents for studying the molecular biology ofhuman development.

Key words: human embryonal carcinoma, differentiation,cross species, regulated genes, mouse embryonalcarcinoma, retinoic acid, SPARC, human development,visceral endoderm, parietal endoderm.

Introduction

Mouse embryonal carcinoma (EC) cells are the stemcells of teratocarcinomas (Stevens, 1970) and havebeen used extensively as models for early mouseembryonic development (Martin, 1980; Hogan et al.1983). Although analogous human cell lines do exist(reviewed Andrews et al. 1983), not until recentlyhave the conditions to induce extensive somaticdifferentiation been denned. In addition, whereasmouse EC cells are regarded to some extent asdevelopmentally similar to cells of the primitive

ectoderm (Brinster, 1974; Evans et al. 1979), thedevelopmental stage to which human EC cells corre-spond has been difficult to determine (Andrews et al.1983). NT2D1 is a human EC cell line capable ofextensive differentiation in response to retinoic acid(RA), producing several novel cell types (Andrews,1984; Gonczol etal. 1984; Skowronski & Singer, 1985;Hauser et al. 1985; La Femina & Hayward, 1986; Lee& Andrews, 1986; Fenderson et al. 1987). However,except for neurones (Lee & Andrews, 1986), theseother cells have not been characterized.

For this study, a cDNA library was constructed

404 M. V. Wiles

from NT2D1 cells that had been induced to differen-tiate in response to RA. (NT2D1 cells treated with10~5 M-RA for 7 days are referred to here as ANT2D1cells). Sequences corresponding to mRNAs that in-crease in abundance following RA-induced differen-tiation of NT2D1 cells were isolated by differentialcDNA hybridization (St. John & Davis, 1979; Wil-liams & Lloyd, 1979). The differentiation of NT2D1cells is poorly characterized as compared to mouseEC models, for example F9 (reviewed by Hogan et al.1983). It would therefore be expected that a compari-son to previously characterized mouse EC modelswould yield information, denning similarities, if any,and differences between mouse and human EC celldifferentiation. Such a comparison would apply thepremise that homologous sequences similarly regu-lated between mouse and human during differen-tiation relate to common pathways of differentiationand control in mammals (see review of Gurdon,1987). Whereas sequences found to be unique tohuman EC cell differentiation may be the result ofsequence divergence or point to developmental dif-ferences between mice and humans.

To identify cDNA recombinants that are bothconserved in sequence and similarly regulated follow-ing mouse and human EC cell differentiation, t]ieapproach of differential cDNA hybridization wasagain applied. The mouse model used to comparemouse and human EC differentiation is the well-characterized F9 system (reviewed by Hogan et al.1983). The mouse EC cell F9 can be induced todifferentiate in vitro either to cells resembling visceralendoderm (F9VE; Hogan et al. 1981) or parietalendoderm (F9PE; Strickland et al. 1980). It has beensuggested, however, on the basis of detection ofepidermal growth factor, transferring (Adamson &Hogan, 1984) and higher than expected levels ofhomeobox-containing mRNAs (Colberg-Poley et al.1985) in F9 compared with other mouse EC cell linesthat F9 may be predetermined. For this reason, themouse EC cell line PCC41 was also included in thedifferential screenings.

By comparing differentiation in both mouse andhuman EC cells, it should be possible to isolatesequences that are similar between species in bothsequence and temporal expression. Such sequencesmay be developmentally important in mammalianembryogenesis (Gurdon, 1987); also by the study ofhuman EC cells, insights, particularly into earlyhuman development, may be gained.

Materials and methods

Cell lines and culture conditionsNT2D1 was cloned by Andrews (1984) from Tera2 (Fogh &

Trempe, 1975). F9 (Bernstine et al. 1973) and PCC4azal(Jakob et al. 1973) are mouse EC cell lines originallyderived from the same transplantable teratocarcinoma,OTT6050 (Stevens, 1970). PCC41RA (Benham et al. 1983)is a ouabain-resistant, RA-adapted mouse EC cell line,subclone of PCC4azal.

All cell lines were maintained at 37°C in Dulbecco'smodified Eagle's medium (high glucose formulation), sup-plemented with 10% fetal calf serum. Cells were subcul-tured at, or near, confluence with 0-25 % (w/v) trypsin,0-2 mM-EDTA every 2 to 3 days. NT2D1 cells were subcul-tured to a minimum cell density of 7xl04cellscm2. F9 cellswere grown on tissue culture flasks treated with 0-1%gelatin.

Induction of differentiationNT2D1 cells were seeded at 7xl04cellscm2 in the presenceof 10~3M-RA (all-trans retinoic acid, dissolved in DMSO at10~2M). After 6 days, the medium was replaced, maintain-ing the same regime. On the 7th day, cultures wereharvested for RNA isolation and FACS analysis. NT2D1cultures so treated are referred to here as ANT2D1.

F9 cells were differentiated to parietal endoderm-likecells as described by Strickland et al. (1980). Monolayercultures of F9 were treated with RA at 10~7 M plus dibutyrylcyclic AMP at 10~3M. This regime was maintained for7 days, after which the cells were harvested for RNAisolation. F9 cultures so treated are referred to as F9PE.

For F9 visceral endoderm-like differentiation, the pro-cedure of Hogan et al. (1981) was used. Monolayer culturesof F9 cells were lightly 'trypsinized' so as to cause thedetachment of cells as small clumps. These clumps wereplated out onto bacteriological grade plates in the presenceof 5X10~8M-RA. The 'embryoid bodies' that formed weremaintained in this regime for 15 days before being har-vested for RNA isolation. These cells are referred to asF9VE.

GTC27 is a human EC cell line, GTC72 is a human cellline resembling visceral endoderm, GTC44 is a human cellline resembling parietal endoderm (Pera et al. 1987); cellswere provided by Dr Martin Pera.

FACS analysisCells were prepared and labelled for cell surface antigensFACS analysis as described in Andrews et al. (1981). X63(Kohler & Milstein, 1975) was used as a negative controlmonoclonal antibody. For internal antigen staining (anti-bodies TROMA1 and TROMA3), single-cell suspensionswere permeabilized and fixed in ice-cold 1:1 (v/v) ethanol:water. After washing in phosphate-buffered saline, fixedpermeabilized cells were treated as for surface antigenstaining. SSEA1 was detected using the monoclonal 480(Solter & Knowles, 1978), SSEA3 was detected using themonoclonal 631 (Shevinsky et al. 1982).

RNA isolation and analysisRNA was prepared using the guanidinium thiocyanatemethod of Chirgwin et al. (1979). Enrichment for thepoly(A)+ fraction was by a single round of oligo d(T)-cellulose chromatography (Maniatis et al. 1982).

Northern blot analysis used 1 fig of glyoxalated poly(A)+

Cell line

Mouse and human EC cell differentiation 405

Table 1. Summary of FACS analysis of NT2D1 and ANT2D1

X63 anti-SSEAl anti-SSEA3 X63* TROMA1' TROMA3*

NT2D1ANT2D1

5,25,5

2, 1023,28

71,819, 15

4 , 34 ,5

71,6543,40

3, 137,40

Percent cells labelling in FACS analysis of NT2D1 and ANT2D1 cells. Results shown are from two independent experiments.•Indicates cells fixed in ethanol:water (1:1) to permeabilize before labelling.

RNA (Thomas, 1980) electrophoresed through 1-2% aga-rose gels and transferred to Gene Screen Plus (NEN)membrane. Prehybridization and hybridization was in 50 %(v/v) formamide, lM-NaCl, 10% (w/v) dextran sulphateand 1 % (w/v) SDS at 42°C. After 18 h hybridization, filterswere washed at 50°C in 0-lxSSC, 0-1% SDS for 40min.Autoradiography was at — 70 °C with preflashed XAR5(Eastman Kodak) film between Du Pont Lightening Plusintensifying screens (Laskey & Mills, 1975) for 3-30 h.

For subsequent reprobing, Northern blot filters werestripped of previous probes by incubating at 95 °C in0-lxSSC, 1% SDS for 20min.

32P-labeUed DNA probes were prepared using the ran-dom oligo-primer method of Feinberg & Vogelstein (1984)to a specific activity of >108ctsmin" /xg"1.

Differential hybridization of ANT2D1 cDNA libraryA cDNA library of >107 recombinants was constructedfrom poly(A)+ RNA derived from ANT2D1 described as inWiles et al. (1988), using the EcoRl site of the lambdainsertion vector NM1149 (Murray, 1983). The ^P-cDNAprobes were prepared from oligo d(T) primed poly(A)+

RNA, as described previously (Wiles et al. 1988) to aspecific activity of >108ctsmin~I^g"1, with an averagelength of 400 bp. 5000 recombinants of the ANT2D1 cDNAlibrary (unamplified) were screened using four replicatefilters (22x22 cm sheets of Hybond-N, Amersham) pre-pared as described by Benton & Davis (1977). Duplicatefilters were probed with 32P-cDNA (S-exK^ctsmin"1

ml"1) derived from NT2D1 and ANT2D1 under conditionsdescribed in Wiles et al. (1988). After 18 h, filters werewashed in 0-lxSSC, 01 % SDS at 65°C for 1 h. Autoradi-ography was as described for Northerns, but with exposuresof 1, 3 and 5 days.

For subsequent rescreening, the replicate filters werestripped of previous probes by incubating in 0-4M-NaOH at45 °C for 30min. Filters were sequentially rescreened induplicate, with 32P-cDNA (3-6X106ctsmin-1 ml"1) de-rived from F9, PCC41RA, F9PE and F9VE and washed inlxSSC, 0 1 % SDSat65°Cforlh. Autoradiography was asdescribed for Northerns with exposures of 1-3 days. Adifferential signal was scored when 'filters' showed induplicate a markedly stronger signal with 32P-cDNA de-rived from differentiated EC cells compared with undiffer-entiated.

Results

Differentiation of NT2D1 in vitroAfter treating NT2D1 cells with 10~5 M-RA for 7 days

(ANT2D1 cells), cell surface and internal antigenswere examined by FACS analysis (data summarizedin Table 1). Before RA treatment, NT2D1 expresseshigh levels of the cell surface antigen SSEA3(Shevinsky et al. 1982), high levels of a cytokeratinantigen recognized by TROMA1 (Kemler et al.1981), low levels of the cytokeratin antigen recog-nized by TROMA3 (Brulet et al. 1980) and low levelsof the cell surface antigen SSEA1 (Solter & Knowles,1978). Following RA induction of NT2D1, cell sur-face antigen SSEA3 decreases, while SSEA1 ex-pression increases; these changes are consistent withAndrews (1984). Cytokeratin expression was alsofound to change, with a decrease in TROMA1antigen and an increase in TROMA3 antigen. TheTROMA3 antigen may correspond to a 40xl03Mr

keratin polypeptide that Damjanov et al. (1984)showed increased following RA treatment of humanEC NTera2 cl.4D3 cells (both NTera2 cl.4D3 andNT2D1 are derived from the same parental line,NTera2). Together with these phenotypic changes,there is also a marked change in morphology. Allthese changes are consistent with RA-induced differ-entiation of NT2D1 cells.

Differential cDNA hybridization of the ANT2D1cDNA libraryA AcDNA library was constructed from poly(A)+

RNA isolated from ANT2D1 cells (Wiles etal. 1988).5000 ANT2D1 cDNA library recombinants werescreened for clones corresponding to mRNAs that aremore abundant following RA-induced NT2D1 differ-entiation. By comparisons of matched autoradio-gTaphs, it was found that 56 (1-12 %) ANT2D1 cDNArecombinants detected a marked increase in signalwith ^P-cDNA derived from ANT2D1 as comparedwith NT2D1 (see Fig. 1 for examples). For brevityrecombinants that behave in this way are referred toas being regulated.

Recombinants homologous to mouse sequenceswere identified by repeated hybridizations using thesame filters. Each filter was rehybridized sequentiallywith ^P-cDNAs derived from PCC41RA, F9, F9VEand F9PE. Washing criterion was relaxed to allowdetection of hybrids with greater than 77 % homology(see Anderson & Young, 1985); examples of thisscreen are shown in Fig. 1. Comparison of signals

406 M. V. Wiles

ANT2D1

PCC41RAi

Mouse and human EC cell differentiation 407

obtained with 32P-cDNAs derived from mouse ECcells (PCC41RA and/or F9) with those of differen-tiated F9 cells (F9VE and/or F9PE) show 60 (1-2 %)of the ANT2D1 recombinants are homologous tomRNA species that are more abundant followingdifferentiation of F9 as compared with undifferen-tiated mouse EC cells. Of these homologous regu-lated recombinants, 15 (0-3 %) were also common tothose regulated following NT2D1 differentiation. Ofrecombinants regulated in the human system but notin the mouse, about two thirds gave a signal withmouse-derived 32P-cDNAs but were not detected asbeing regulated following F9 differentiation. A sum-mary of these results is given in Table 2.

Isolation of regulated cDNAs clonesSix recombinants giving strong differential signalswhen probed with 32P-cDNAs derived from ANT2D1as compared with NT2D1 were isolated. Theserecombinants are referred to as SO2, SO5, SO6,SO7, SO9 and SO 13. Upon EcoRI cleavage, recom-binants SO5 and SO13 each released two inserts.These double inserts are the result of either ligationof unrelated cDNAs during library construction orthe presence of intrinsic EcoRI sites within the cDNAmolecules (note, the cDNA library was constructedwith EcoRI methylase-treated cDNA). The larger ofeach insert is suffixed with an 'A', the smaller with a'B'. These data are summarized in Table 3. Insertswere cross hybridized with each other and homologywas detected between inserts SO2 and SO5A; hom-ology was also detected between inserts SO7 andSO13B.

Fig. 1. Autoradiographs showing differential screening ofANT2D1 cDNA library. Examples of signals obtainedwith 32P-cDNA synthesized from poly(A)+ enrichedRNAs. The same area (~8cm x 8 cm) of filter is shown•with all six probes used: Top panels, 32P-cDNA probederived from NT2D1 and ANT2D1; niters washed to0-1 x SSC, 65°C. Middle panels, 32P-cDNA probe derivedfrom the mouse EC cell line F9, and F9VE (F9 cellsinduced to VE-like cells); filters washed to lxSSC at65°C. Bottom panels, ^P-cDNA probe derived from themouse EC cell line PCC41RA, and F9PE (F9 cellsinduced to PE-like cells); filters washed to lxSSC at65°C. Circled areas, from left to right, show: (1) Anexample of a recombinant that detects 32P-cDNA that ismore abundant in ANT2D1 compared with NT2D1, andin F9VE compared with F9 (and PCC41RA) (this areasubsequently yielded clone SO5A); (2) An example of arecombinant that detects ^P-cDNA that is moreabundant in ANT2D1 compared with NT2D1, but showsno apparent change following F9 differentiation; (3) Anexample of a recombinant where no change in 32P-cDNAabundance is detected between ANT2D1 and NT2D1,but is more abundant following F9 to F9PEdifferentiation.

Table 2. Summary of differential cDNAhybridizations

Cell lines

Undifferentiated

NT2D1F9F9PCC41RAPCC41RAMouse EC

used

Differentiated

ANT2D1F9PEF9VEF9PEF9VE

F9VE+F9PE

Recombinantsshowing a

UlllCIdltlol

signal

56(1-12%)28(0-56%)37(0-74%)39(0-78%)28(0-56%)60(1-20%)

Similarlyregulated inNTT?r>i tn

ANT2D1

7(0-14%)7 (0-14%)9(0-18%)

10(0-20%)15(0-30%)

5000 cDNA recombinants probed with ^P-cDNAs derivedfrom NT2D1, ANT2D1, PCC41RA, F9, F9VE and F9PE. Adifferential signal was scored when an increase in hybridizationwas seen in duplicate that was equal or greater than three-foldhigher with probe derived from differentiated cells as comparedto uninduced EC cells. Sequences detected as regulated betweenF9 and F9VE+F9PE were only partly in common with thePCC41RA to F9VE+F9PE comparison.

Northern blot analysis of SO-clonesNorthern blots of poly(A)+ RNA derived fromNT2D1 and ANT2D1 were probed with each of thecDNA inserts (examples shown Fig. 2, tracks 1 and2). A summary of the Northern blot analysis togetherwith the result of densitometric scans for the NT2D1and ANT2D1 autoradiographs are given in Table 3. Itshould be noted that, because of the kinetics ofNorthern blot hybridization, increases in abundanceof mRNAs as estimated by densitometric scans ofNorthern autoradiographs are probably underesti-mates (see review of Anderson & Young, 1985). Inaddition, it is assumed that /3-actin abundance issimilar between cells, not varying significantly follow-ing differentiation. With the exception of insertSO5B, all clones detect mRNAs that increase inabundance following RA-induced NT2D1 differen-tiation. The increase in abundance ranges from 4-foldwith SO13A, to at least 75-fold with SO9 and SO5A.

Cross-species hybridizationcDNA inserts were hybridized to mouse poly(A)+

RNAs derived from the mouse cell lines PCC41RA,F9, F9VE and F9PE (Fig. 2B,C,E and F, tracks 3, 4,5 and 6; data summarized in Table 3). Three differentNT2Dl-derived clones detect homologous sequencesin mouse poly(A)+ RNAs (SO2/SO5A, SO13A andSO7/SO13B), while no signal was detected withclones SO6 and SO9.

SO clones that detect homologous mouse se-quences all detect an increase in mRNA abundancefollowing F9 differentiation. The cDNA clonesSO7/SO13B (Fig. 2F, tracks 4, 5 and 6) and SO13A(Fig. 2B, tracks 4, 5 and 6) detect mouse mRNAs thatare more abundant following F9 differentiation; the

408 M. V. Wiles

signal from SO13A being more intense in F9PEcompared with F9VE poly(A)+ RNAs. SO5A/SO2detect an mRNA only in F9VE (Fig. 2C, tracks 4, 5and 6), with no signal being detected in PPC41RA, F9or F9PE poly(A)+ RNAs.

Characterization of clone SO5ASO5A was singled out for further investigation as itdetected a large increase in steady-state mRNAabundance following NT2D1 differentiation and simi-larly with F9 differentiation to F9VE but not toF9PE. Fig. 3 panel A shows the result of probing totalRNAs derived from NT2D1 cells subjected to 10~5 M-RA over time with SO5A. The abundance of mRNAshomologous to SO5A increases markedly after 2 to 3days exposure to RA and remains constant over 7days (compared to the abundance of /3-actin andClass-I HLA mRNAs).

Direct analysis of SO5A expression in humanembryos was not possible. However, Pera etal. (1987)have isolated human cell lines that have character-istics of visceral endoderm (GTC72) and parietalendoderm (GTC44). When Northern blots of totalRNA derived from these cells were probed withSO5A a signal was detected only in the parietal-endoderm-like line, GTC44 (Fig. 3 panel B). This isin apparent contrast to the mouse F9 model whereexpression was detected only in F9VE (visceral-endoderm-like).

Putative identification of a human homologue to themouse gene SPARCDuring a screen of sequences known to be regulatedin mouse, it was noted that SPARC (cDNA clonepPE30, Mason et al. 1986) detects on Northern

analysis of NT2Dl-derived poly(A)+ RNAs a majorband at 2-2 kb plus a minor band at ~2-7kb. Thissignal was stronger in ANT2D1 than NT2D1poly(A)+ RNAs. To check if any of the SO cloneswere homologous to SPARC, inserts were hybridizedwith the SPARC cDNA probe pPE30. Homology wasdetected between pPE30 and cDNA clone SO13A.When used as a probe on Northern blots of poly(A)+

RNAs (Fig. 2B), SO 13A detects a 2-2 kb signal fol-lowing RA-induced differentiation of F9 to F9VE andF9PE; the stronger signal being seen in F9PE. This isin agreement with the result observed by Mason et al.1986 with the mouse SPARC. However, the insertSO13A is ~3 kb. A blunt-end ligation of unrelatedcDNAs during library construction may account forthis, or the cloning of a possible precursor. The latterpossibility is favoured, as following longer exposureof Northern blots of poly(A)+ RNA from NT2D1 andANT2D1, probed with SO13A, additional bands at2-7kb and ~3-3kb are detected. Swaroop et al.(1988) also detected a 2-7 kb band in human-derivedRNAs, this reflects the use of another polyadenyl-ation site.

Discussion

Differential cDNA hybridizationCommon pathways of mammalian differentiationwould be expected to elicit coordinated expression ofgenes that are homologous between species (Gurdon,1987). The identification of cDNA recombinants thatare homologous and similarly regulated in bothhuman and mouse is therefore not surprising. Thisdata is, however, in contrast to previous work where

Table 3. Summary of Northern blot hybridization data

Northern analysis

CloneInsertsize NT2D1 ANT2D1

Ratio ofANT2D1/

NT2D1 PCC41RA F9 P9VE F9PESize ofmRNA

SO2SO5ASO5BSO6SO7SO9SO 13 ASO13B/S-actinClass-I HLA

1-lkbl-7kb0-4 kb1-lkb1-4 kb1-4 kb

-30 kb1-5 kb

>2875

16

22>75

4>10

12

2-6kb2-6kb2-2 kbl-4kbl-8kb2-7 kb2-2kbl-7kb2-2 kbl-7kb

The signals detected from ANT2D1 and NT2D1 (as determined by densitometric scans) were normalized against /S-actin (see Fig. 2,panels A and D).

+ indicates just detectable with the number corresponding to signal strength.- is shown where no signal was detected. /S-actin probe pGFl; Class-I HLA probe pHLA-A (Lee et al. 1982). (Poly(A)+-enriched

RNA was loaded to five approximately equal signals when probed with 0-actin; mRNA sizes were estimated using DNA markers asstandards {Haelll cleaved 0X176 and Hindlll cleaved lambda).

1 2 3 4 5 6

Mouse and human EC cell differentiation 409

1 2 3 4 5 6 1 2 3 4 5 6

\

0-6 —

B

1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6

0-6

D E F

Fig. 2. Northern blot analysis of SO-clones. Northern blots of poly(A)+ RNAs derived from (1) NT2D1, (2) ANT2D1,(3) PCC41RA, (4) F9, (5) F9VE and (6) F9PE. Panels A and D, shows the integrity and loading of mRNA byhybridization with /3-Actin; panel A is the control hybridization for B and C; panel D is the control hybridization for Eand F. Panel B shows the hybridization pattern obtained with the insert SO13A, panel C with SO5A, panel E with SO6and panel F with SO7. Note that with the probe SO7 the exposure shown of mouse-derived poly(A)+ RNAs was 4times longer that of human poly(A)+ RNAs.

it has generally been differences between mouse andhuman EC cells, not their similarities, that have beenmost evident (e.g. Andrews et al. 1983).

It should be remembered that with the differen-tiation of F9 and NT2D1 EC cells it is not just theirEC phenotype that is alike, but also the reagent used

410 M. V. Wiles

-Days in RA-

0 0-5 1 2 3 4 5 6 7 17 HFL

CM "tf h»s ^ f ooO O OCD O O

S05A S05A

/8-Actin I — 2-3 kb— 20

HLA ^6-Actin

B ? • • *Fig. 3. Northern blot analysis of SO5A. (A) Northern blot of total RNA derived from NT2D1 cells treated with RAagainst time, the numbers refer to days grown in 10~5M-RA. The same filter was probed with (top) SO5A, (middle) */5-actin (probe pGFl), and (bottom) Qass-I HLA (probe pHLA-A; Lee et al. 1982). Two tracks were run for each timepoint, the first containing ~5 jig RNA, the second track half this amount. The last track is total RNA derived fromhuman foreskin fibroblasts (known as HFL). *The changes in signal detected with the /S-actin probe are thought to bedue to interference from 18S ribosomal RNA. (B) Northern blot of total RNAs derived from: ANT2D1; NT2D1;GTC72 (human VE-like cell line); GTC44 (human PE-like cell line); GTC27 (human EC cell line); probed with SO5A.Below same filter probed with /S-actin (probe pGFl).

to induce this differentiation; i.e. retinoic acid.Therefore, it is possible that sequences regulated incommon in these two model systems are the directresult of RA treatment and not differentiation per se.An example of this has been observed by Deschampset al. (1987) who described the 'inappropriate' ex-pression of a mouse homeobox-containing geneH24.1. They showed that in RA-treated mouse ECcells expression of H24.1 appears to be correlated tothe presence of RA, as H24.1 was not detected byNorthern blot analysis of RNA derived from culturesthat had been differentiated spontaneously. How-ever, this result does not preclude appropriate geneexpression under the influence of RA. For example,with the differentiation of F9 to PE-like and VE-likecells, changes in vitro appear to correlate with in vivochanges; laminin, collagen IV, tissue plasminogenactivator and SPARC are detected in F9PE andmouse parietal endoderm; in addition, alphafetopro-tein, transferrin, SPARC and cytokeratins have beendetected in both F9VE and normal mouse visceralendoderm (reviewed by Hogan et al. 1981; see also

Mason et al. 1986). In the case of RA-induceddifferentiation of NT2D1, inappropriate expressionof genes due to RA cannot account for all thedifferential signals detected. For instance, increasesin mRNA homologous to SO5A would have beendetected in both F9VE and F9PE, not just F9VE.Also upon differential cDNA hybridization, not allthe recombinants recognized as regulated with RA-induced NT2D1 differentiation were similarly regu-lated following F9 differentiation. These differenceswere not generally due to a lack of homology betweenmouse and human sequences, as the majority of theserecombinants gave signals but showed no detectablechange following F9 differentiation. FurthermoremRNA homologues to SO5A were detected inCTG44, a cell line that was not exposed to RA.

Regulated clones

Five different clones were isolated that detectedsequences more abundant in ANT2D1 than inNT2D1 poly(A)+ RNAs. When these cDNA cloneswere used as probes on Northern blots of poly(A)+

Mouse and human EC cell differentiation 411

RNA derived from PCC41RA, F9, F9VE and F9PE,two general patterns were found; (i) clones that donot detect homologous mouse sequences (SO6 andSO9), and (ii) those that do detect homologousmouse sequences (SO2/SO5A, SO7/SO13B andSO 13A). The presence of the former class mayindicate that some abundance changes in mRNAspecies are specific under conditions used here tohuman EC cell differentiation in vitro. This may be areflection of the heterogeneous nature of NT2D1RA-induced differentiation, or suggest that humanEC cells represent a different embryonic state thanmouse EC cells and thus exhibit different develop-mental capabilities (Andrews et al. 1983). The pres-ence of the latter class of homologous regulatedclones shows, however, that there are similarities ingene expression between mouse and human EC celldifferentiation. Further work needs to be done toestablish the relevance of these two classes of clones.

Clone SO5A detects by Northern blot analysis, anmRNA that is 75-fold more abundant in ANT2D1 ascompared to NT2D1. With the mouse F9 modelsystem, a mouse homologue to SO5A is detected inRNAs derived from F9VE but not in F9PE. Thisspecificity may give a clue as to the phenotype/sinduced following NT2D1 differentiation. It alsoprovides a further marker for F9VE differentiationand thus possibly for normal mouse visceral endo-derm. The hybridization of SO5A to RNA derivedfrom GTC44, a human PE-like cell line is surprisingin the light of the apparent specificity seen withF9VE. This result may reflect basic differences be-tween mouse and human extraembryonic tissues, orpossibly the lack of characterization of the cell linesinvolved. Interestingly, when mouse SPARC (M. F.Pera and R. Krumlauf, personal communication) orthe putative human homologue (SO13A; data notshown) are hybridized to GTC44 and GTC72 RNAs,steady-state expression is only marginally higher inthe human PE-like cell line GTC44. This is again incontrast to mouse where high levels of SPARCmRNAs are detected in PE and PE-like cells (Masonet al. 1986). Further experiments, including DNAsequencing and the comparing of expression of theseclones in mouse VE and PE with human embryonictissues may shed light on these differences.

One clone, SO13A, has been tentatively identifiedby hybridization as a human homologue of the mousegene SPARC. Mason et al. (1986) found that SPARCmRNA is induced to moderate levels following F9differentiation to F9VE and to high levels followingdifferentiation to F9PE in vitro. The induction of thisgene in vitro reflects the differentiation of mouseprimitive endoderm to visceral endoderm and par-ietal endoderm in vivo (Mason et al. 1986). Theidentification of a known mouse gene which is regu-

lated both in vitro and in vivo during the differen-tiation of primitive endoderm indicates that cross-species cDNA differential hybridization may assist inthe isolation of homologous sequences involved inearly embryogenesis. Also, it can be argued thatgenes that are conserved in sequence and in temporaland spatial expression are likely to be developmen-tally important (Gurdon, 1987). Future experimentscould make effective use of this using additionaldifferential cDNA screenings, the 32P-cDNAs beingderived from a number of different characterized cellsand tissues (of the same, or different, species); thusallowing one to survey a large number of cDNArecombinants simultaneously. The autoradiographsproduced would lead to the composition of a form of'pictorial ROT curve' for each tissue/phenotype. Bysuch repeated cDNA differential hybridizations itshould be possible to construct a molecular pheno-type of NT2D1 differentiation.

Conclusions

From the data presented, it can be surmized that RA-induced differentiation of F9 and NT2D1 share someelements in common. As F9 differentiation in vitroalso shares features with mouse development in vivo,genes that are homologous and similarly regulatedbetween NT2D1 and F9 may be developmentallyimportant. Differences, however, are also high-lighted and these may point to marked differencesbetween mouse and human early development.Further experiments are required to correlate thechanges observed with NT2D1 differentiation withthe in vivo situation; for example, using abortusmaterial and human VE-like and PE-like cell lines.However, the isolation of these clones representsome of the first reagents that can be used for thestudy of human developmental molecular biology.

I thank Dr Peter N. Goodfellow in whose laboratory thiswork was done; also Drs B. L. M. Hogan and E. F. Wagnerfor helpful discussions and assistance. Cell lines and RNApreparation for Fig. 3 was done in collaboration with DrsM. F. Pera and S. Cooper. Monoclonal antibodiesTROMA1 and TROMA3 were provided by Dr R. Kemler.I must also thank my colleagues B. Williams, B. Pym,C. Pritchard, G. Knott, P. J. Goodfellow, S. Carson andG. Banting for their patience and help.

References

ADAMSON, E. D. & HOGAN, B. L. M. (1984). Expressionof EGF receptor and transferrin by F9 and PC13teratocarcinoma cells. Differentiation 27, 152-157.

ANDERSON, M. L. M. & YOUNG, B. D. (1985).Quantitative filter hybridization. In Nucleic Acid

412 M. V. Wiles

Hybridization - a Practical Approach (ed. B. D.Hames & S. J. Higgins), pp. 73-111. Oxford: IRLPress.

ANDREWS, P. W. (1984). Retinoic acid induces neuronaldifferentiation of a cloned human embryonal carcinomacell line in vitro. Devi Biol. 103, 285-293.

ANDREWS, P. W., BRONSON, D. L., WILES, M. V. &

GOODFELLOW, P. N. (1981). The expression of MHCantigens by human teratocarcinoma derived cell lines.Tissue Antigens 17, 493-500.

ANDREWS, P. W., GOODFELLOW, P. N. & DAMJANOV, I.

(1983). Human teratocarcinoma cells in culture. CancerSurv. 2, 42-73.

BENHAM, F. J., WILES, M. V. & GOODFELLOW, P. N.

(1983). Differentiation in vitro of human-mouseteratocarcinoma hybrids. Mol. Cell Biol. 3, 2259-2270.

BENTON, W. D. & DAVIS, R. W. (1977). Screening Agtrecombinant clones by hybridization to single plaquesin situ. Science 196, 180-182.

BERNSTINE, E. G., HOOPER, M. L., GRANDCHAMP, S. &

EPHRUSSI, B. (1973). Alkaline phosphatase activity inmouse teratoma. Proc. natn. Acad. Sci U.S.A. 70,3899-3902.

BRINSTER, R. L. (1974). The effects of cells transferredinto the mouse blastocyst on subsequent development.J. exp. Med. 140, 1049-1056.

BRULET, P., BABINET, C , KEMLER, R. & JACOB, F. (1980).Monoclonal antibodies against trophoectoderm-specificmarkers during mouse blastocyst formation. Proc. natn.Acad. Sci. U.S.A. 77, 4113-4117.

CHIRGWIN, J. M., PRZYBYLA, J. E., MACDONALD, R. J. &

RUTTER, W. J. (1979). Isolation of biologically activeribonucleic acid from sources enriched in ribonuclease.Biochemistry 18, 5294-5295.

COLBERG-POLEY, A . M . , VOSS, S. D . , C H O W D H U R Y , K. &GRUSS, P. (1985). Structural analysis of murine genescontaining homoeobox box sequences and theirexpression in embryonal carcinoma cells. Nature,Lond. 314, 713-718.

DAMJANOV, I., CLARK, R. K. & ANDREWS, P. W. (1984).Cytoskeleton of human embryonal carcinoma cell. CellDiff. 15, 133-139.

DESCHAMPS, J., DE LAAF, JOOSEN, L., MEDLINK, F. &DESTREE, O. (1987). Abundant expression ofhomeobox genes in mouse embryonal carcinoma cellscorrelates with chemically induced differentiation.Proc. natn. Acad. Sci. U.S.A. 84, 1304-1308.

EVANS, M. J., LOVELL-BADGE, R. H., STERN, P. L. &STINNAKRE, M. G. (1978). Cell lineages of the mouseembryo and embryonal carcinoma cells; Forssmanantigen distribution and patterns of protein synthesis.In Cell Lineage, Stem Cells and Cell Determination.INSERM Symposium No. 10 (ed. N. Le Douarin), pp.115-129. Amsterdam: Elsevier, North HollandBiomedical Press.

FEINBERG, A. P. & VOGELSTEIN, B. (1984). A techniquefor radiolabelling DNA restriction endonucleasefragments to high specific activity. Anal. Biochem. 137,266-267.

FENDERSON, B. A., ANDREWS, P. W., NUDELMAN, E.,CLAUSEN, H. & HAKOMORI, S.-I. (1987). Glycolipid

core structure switching from Globo- to Lacto- andGanglio series during retinoic acid-induceddifferentiation of TERA2 derived human embryonalcarcinoma cells. Devi Biol. 122, 21-34.

FOGH, J. & TREMPE, G. (1975). New human tumour celllines. In Human Tumour Cell in Vitro (ed. J. Fogh) pp.115-159. New York: Plenum.

GONCZOL, E., ANDREWS, P. W. & PLOTKJN, S. A. (1984).Cytomegalovirus replicates in differentiated but notundifferentiated human embryonal carcinoma cells.Science 224, 159-161.

GURDON, J. B. (1987). Embryonic induction - molecularprospects. Development 99, 285-306.

HAUSER, C. A., JOYNER, A. L., KLEIN, R. D., LEARNED,T. K., MARTIN, G. R. & TJIAN, R. (1985). Expressionof homologous homeo-box-containing genes indifferentiated human tetratocarcinoma cells and mouseembryos. Cell 43, 19-28.

HOGAN, B. L. M., BARLOW, D. P. & TILLY, R. (1983). F9teratocarcinoma cells as a model for the differentiationof parietal and visceral endoderm in mouse embryos.Cancer Surv. 2, 115-140.

HOGAN, B. L. M., TAYLOR, A. & ADAMSON, E. D.(1981). Cell interactions modulate embryonalcarcinoma cell differentiation into parietal or visceralendoderm. Nature, Lond. 291, 235-237.

JAKOB, H., BOON, T., GAILLARD, J., NICOLAS, J.-F. &JACOB, F. (1973). Teratocarcinome de la souris:Isolement culture et proprietes de cellules apotenlialites multiple. Annls Micro. Inst. Pasteur.(Paris) 124(B), 269-283.

KEMLER, R., BRULET, P., SCHNEBELEN, M. I., GAILLARD,J. & JACOB, F. (1981). Reactivity of monoclonalantibodies against intermediate filaments duringembryonic development. J. Embryol. exp. Morph. 64,45-60.

KOHLER, G. & MJLSTEIN, C. (1975). Continuous culturesof fused cells segregating antibody of predefinedspecificity. Nature, Lond. 256, 495-497.

LA FEMINA, R. & HAYWARD, G. S. (1986). Constitutiveand retinoic acid-inducible expression ofcytomegalovirus immediate-early genes in humanteratocarcinoma cells. J. Virology 58, 434-440.

LASKEY, R. A. & MILLS, D. A. (1975). Quantitative filmdetection of 3H and 14C in polyacrylamide gels byfluorography. Eur. J. Biochem. 56, 335-341.

LEE, J., TROWSDALE, J. & BODMER, W. F. (1982). cDNAclones coding for the heavy chain of human HLA-DRantigen. Proc. natn. Acad. Sci. U.S.A. 79, 545-549.

LEE, V. M.-Y. & ANDREWS, P. W. (1986). Differentiationof NTera-2 clonal human embryonal carcinoma cellsinto neurons involves the induction of all threeneurofilament proteins. J. Neuroscience 6, 514-549.

MANIATIS, T., FRTTSCH, E. F. & SAMBROOK, J. (1982).Molecular Cloning. A Laboratory Manual. Cold SpringHarbor Laboratory, New York: Cold Spring Harbor.

MARTIN, G. R. (1980). Teratocarcinomas and mammalianembryogenesis. Science 209, 768-776.

MASON, I. J., MURPHY, D., MUNKE, M., FRANCKE, U.,ELLIOTT, R. W. & HOGAN, B. L. M. (1986).Developmental and transformation-sensitive expression

Mouse and human EC cell differentiation 413

of the SPARC gene on mouse chromosome 11. EMBOJ. 5, 1831-1837.

MURRAY, N. E. (1983). In The Bacteriophage Lambda,Vol. 2 (ed. R. Hendrix, R. Weisberg, F. Stahl & J.Roberts). Cold Spring Harbor Laboratory, New York:Cold Spring Harbor.

PERA, M. F., LAFTTA, M. J. B. & MILLS, J. (1987).

Cultured stem cells from human testicular teratomas:The nature of human embryonal carcinoma, and itscomparison with two types of yolk sac carcinoma. Int.J. Cancer 40, 334-343.

ST JOHN, T. P. & DAVIS, R. W. (1979). Isolation ofgalactose-inducible DNA sequences fromSaccharomyces Cerevisiae by differentiated plaque filterhybridization. Cell 16, 443-452.

SHEVINSKY, L. H., KNOWLES, B. B., DAMJANOV, I. &SOLTER, D. (1982). Monoclonal antibodies to murineembryos defines a stage-specific embryonic antigenexpressed on mouse embryos and humanteratocarcinoma cells. Cell 30, 697-705.

SKOWRONSKI, J. & SINGER, M. F. (1985). Expression of acytoplasmic LINE-1 transcripts is regulated in a humanteratocarcinoma cell line. Proc. natn. Acad. Sci.U.S.A. 82, 6050-6054.

SOLTER, D. & KNOWLES, B. B. (1978). Monoclonalantibody defining a stage-specific mouse antigen

(SSEA-1). Proc. natn. Acad. Sci. U.S.A. 75,5565-5596.

STEVENS, L. C. (1970). The development oftransplantable teratocarcinomas from intratesticulargrafts of pre- and postimplantation mouse embryos.Devi Biol. 21, 364-382.

STRICKLAND, S., SMITH, K. & MAROTTI, K. (1980).Hormonal induction of differentiation interatocarcinoma stem cells: generation of parietalendoderm by retinoic acid and dibutyryl cAMP. Cell21, 347-355.

SWAROOP, A., HOGAN, B. L. M. & FRANCKE, U. (1988).Genomics (in press).

THOMAS, P. N. (1980). Hybridization of denatured RNAand small DNA fragments transferred to nitrocellulose.Proc. natn. Acad. Sci. U.S.A. 77, 5201-5205.

WILES, M. V., ALEXANDER, C. M. & GOODFELLOW, P. N.(1988). Isolation of an abundantly expressed sequencefrom the human X chromosome by differentialscreening. Soma. Cell, molec. Genet. 14, 31-39.

WILLIAMS, J. G. & LLOYD, M. M. (1979). Changes in theabundance of polyadenylation RNA during slimemould development measured using cloned molecularhybridization probes. /. molec. Biol. 129, 19-35.

{Accepted 19 July 1988)