J. Cell Sd. 72, 37-64 (1984)                                                                        37
Printed in Great Britain © The Company of Biologists Limited 1984


          S. THOMPSON 1 , P. L. STERN 1 , M.WEBB2, F. S. WALSH2,
          W. ENGSTROM 1 , E. P. EVANS3, W.-K. SHI 1 , B. HOPKINS 1 AND
          C. F. GRAHAM1
            Department of Zoology, South Parks Rd, Oxford 0X1 3PS, UX.
            Institute of Neurology, Queen Square, London WC12 3BG, U.K.
            Dunn School of Pathology, South Parks Rd, Oxford 0X1 3RE, U.K.

   Single cell clones were isolated from the human teratoma line, Tera-2. The cells of three of these
clones were studied. The progressively growing cells were shown to be tumorigenic, and they were
characterized by the lack of expression of /32-microglobulin and HLA-A,B,C determinants on the
cell surface. The majority of the cells expressed Thy-1 antigen and a 90 X 103 molecular weight
protein recognized by the monoclonal antibody F10.44.2; between a third and half of the cells
expressed the sugar specificities detected by the anti-SSEA-1 monoclonal antibody. In response to
5 X 10~5 M-retinoic acid applied to cells in monolayer culture, the cells differentiated into a popula-
tion of flat static cells arrested in the G\ phase of the cell cycle. A substantial proportion of these
differentiated cells expressed /32-microglobulin and 43 X 103 molecular weight HLA-A,B,C
polypeptides, Thy-1, SSEA-1 sugar determinants, and the 90 X lC^Mr protein recognized by
F10.44.2. The apparent molecular weight of fibronectin secreted by the cells decreased by about
5 X 103Afr to 235 X 103M, after differentiation. The progressively growing cells lacked reactivity
with reagents that mark cells in the nervous system. Following aggregation and retinoic acid treat-
ment, neuron-like cells were formed. These cells reacted with reagents that also react with human
neurons in culture: they reacted with tetanus toxin, the anti-neurofilament antibodies BF10 and
RT97, the anti-ganglioside, GQlc antibody F12 A2B5, and anti-Thy-1.
   The progressively growing cells of these Tera-2 clones are therefore capable of forming at least
two types of cell: the flat cells in monolayer cultures and the neuron-like cells. None of the cell
populations reacted with the monoclonal antibody against SSEA-3 and these cloned cells are
therefore distinct from previous isolates from Tera-2.

   If all the cell types seen in human teratomas are derived from a single multipotential
stem cell, then that cell would be recognized by its ability to form those cell types.
Human testicular tumours frequently contain embryonal carcinoma, seminoma, and
cells that resemble those found in extra-embryonic foetal tissues, for example placen-
tal trophoblast and yolk sack; at a lower frequency there are groups of cells that look
like embryonic and adult tissues, such as muscle, cartilage and nerve (e.g. see Mostofi
& Price, 1973; Pugh, 1976).
   In the serum of patients with teratomas there are found secreted products that are
characteristic of extra-embryonic foetal tissues; these include the beta subunit of
Key words: teratoma, neurons, fibronectin.
38                                  5. Thompson and others
human chorionic gonadotrophin (/J-HCG), alphafoetoprotein (AFP), and the am-
niotic form of fibronectin. Some teratoma-derived cell lines have been shown to form
these products in xenografts or to secrete them into culture medium (reviewed by
Raghavan, 1981). However, the identity of a multipotential stem cell would only be
firmly established if a single cell, picked from a human testicular tumour, could be
grown up to form at least some of these products or some of the characteristics of
different embryonic tissues. So far, cells cloned from human testicular tumours have
shown only a limited ability to differentiate (reviewed by Andrews, Goodfellow &
Damjanov, 1983; Mcllhinney, 1983; Stern, 1984). At low frequency, one clone
(N2102Ep, clone 2A6) will form cells expressing /J-HCG (Damjanov & Andrews,
1983); this clone was derived from the human testicular teratocarcinoma line, 2102Ep
(Andrews et al. 1980, 1982). There is also a series of clones derived from the cell line
Tera-2; under suitable conditions these clones have been shown to differentiate into
several cell forms, including neuron-like cells, which express neurofilament antigenic
determinants and tetanus toxin receptors (Andrews, 1984; Andrews et al. 1984).
These clones were isolated from tumours formed by Tera-2 in nude mice; they are
called NTera-2 clones (Andrews, 1984; Andrews et al. 1984).
   We describe the isolation of additional clones from the original Tera-2 cell line,
which had not been passaged through a nude mouse (F0gh & Trempe, 1975). We have
characterized the progressively growing small cells in cultures of these clones by
studying their surface phenotype, their secreted products, and their ability to form
other cell types in culture.


Isolation and culture of clones
   The Tera-2 cell line was a gift from Professor J. F0gh (F0gh & Trempe, 1975). The cells were
grown in the alpha modification of minimal essential medium lacking nucleosides and deoxynucleo-
sides (Stanners, Eliceiri & Green, 1971; Gibco Europe, Paisley, Scotland), with the addition of
10% (v/v) heat-inactivated foetal calf serum (called alpha 10% FCS). Small cells with a high
nuclear-to-cytoplasmic volume ratio were abundant in cultures that continued to grow progressive-
ly, and cells with this form were subsequently called undifferentiated cells (Fig. 1). For routine
passage, the cells were exposed to trypsin for the minimum time to effect disaggregation (TVP;
Bemstine, Hooper, Grandchamp & Ephrussi, 1973). After sucking off the TVP, the cells were
briefly incubated in EGTA/PBS, which contained 0-5mM-EGTA in PBS lacking calcium and
magnesium ions (solution A of Dulbecco & Vogt, 1954, called PBS). The cells were vigorously
pipetted from the bottom of the dish after the addition of 2-4 ml of alpha 10% FCS, spun down,
and plated out at approximately 1 X 106 per 50 mm diameter culture dish containing 5 ml of alpha
10 % FCS pre-equilibrated with 5-7 % (v/v) CO2 in humidified air at 37°C (dishes from Sterilin,
Richmond, Surrey). The cells adhered firmly to tisssue culture surfaces that had been previously
treated by incubation with a 0-1 % (w/v) aqueous gelatin solution for 2 h at 5 °C, and then air dried
at 40°C for 24h (swine skin, type 1 gelatin, from Sigma, Poole, Dorset).
   The clones were established from bulk cultures of the Tera-2 cells at the 41st passage of the line.
Two days before cloning, each well of a 24-well plate was filled with 1-2 ml of alpha 10% FCS
(plates from Gibco Europe). In some experiments, the surface of each well was covered with 2 X
105 to 4 X 105 mitomycin-treated 10TJ cells (Reznikoff, Bertram, Brankow & Heidelberger, 1973;
from M. Williams, Dunn School of Pathology, Oxford; treated as described by Martin & Evans,
   Near-confluent 25 cm2 Gibco Europe flasks of the cell line were refed with medium between 1 and
                           Differentiation of human teratoma cells                                 39
2h before the flasks were shaken 20 times to dislodge cells. The detached cells were pipetted into
a 90 mm diameter bacteriological grade dish on a dissecting microscope stage (Wild M5). About 20
single cells were sucked into the tip of a mouth-controlled micropipette, which had been drawn to
an internal tip diameter of approximately 150 [an (hard glass capillary tubing, BDH, Poole, Dorset).
Single cells were blown into 50^1 drops of alpha 10% FCS, which had been placed on the surface
of another bacteriological dish and covered with liquid paraffin (Boots Pure Drug Co., Notting-
ham) . Each drop was scored for the presence of a single healthy cell using an inverted phase-contrast
microscope (Wild M40), and such cells were transferred to individual wells of the 24-well plate using
a different micropipette for each transfer.
   After 10 days of incubation, 0-5 ml of fresh medium was added to each well, and in each of the
succeeding 6 weeks half the medium was replaced with fresh medium. As each colony became
confluent in the well, so it was sucked off the bottom with a wide-mouthed micropipette, and
transferred to a 25 cm2 flask containing 10mlofalphal0% FCS. These flasks were either untreated,
covered with 10Ti mitomycin-treated feeder cells, or coated with gelatin.
   The cells in these flasks were described as passage 1 cells. Thirty two of these contained some
groups of undifferentiated cells (Fig. 1); growth of these cells was usually obvious after 3 weeks
incubation in flasks, but one undifferentiated cell colony did not appear until after 3 months
incubation and it remained slow-growing. No further undifferentiated cell colonies appeared during
the next 8 months. A variety of cell morphologies were seen in the flasks and the undifferentiated
cells continued to grow only in 25 flasks. Cells in these flasks could be passaged and their karyotypes
were scored at passages 1 and 2. In the preparation of the karyotypes, the cells were scraped from
the dishes with a rubber policeman, briefly centrifuged, and the pellet treated with hypotonic 0'56 %
(w/v) KC1 for 6 min at room temperature. They were briefly centrifuged, and the pellet was fixed
in methanol/acetic acid ( 3 : 1 , v/v), and the cells were air dried onto clean dry microscope slides.
The slides were 'aged' for between 3 and 10 days at room temperature, and then G-banded using
a modification of the combined ASG/trypsin method of Gallimore & Richardson (1973).
   Dr Peter Andrews provided NTera-2, clone B9 (Andrews et al. 1984) and N2102Ep, clone 2A6
(Andrews et al. 1982); these were tested at passages 17 and 87, respectively.

Retinoic acid treatment in monolayer
   To induce differentiation we used a high concentration of retinoic acid, as recommended by
Andrews (1984). Gelatin-coated 90 mm diameter tissue culture dishes were preincubated for 1-2 h
with 15 ml of alpha 10% FCS containing 5 X 10~ 5 M all trans retinoic acid diluted from a stock of
1 X 10"1 M-retinoic acid in dimethylsulphoxide stored at —70°C (Eastman Kodak retinoic acid).
This medium was called alpha RA. To obtain an even dispersal of cells across the dish, populations
of undifferentiated cells were disaggregated and between 8 X 105 and 1-5 X 10* of the cells were
taken up in the warm alpha RA medium from each dish and quickly pipetted back on to the dish,
which was immediately returned to the incubator.

Aggregation and retinoic acid treatment
   In an attempt to provoke more extensive differentiation, we grew the cells in free-floating lumps.
Single undifferentiated cells adhere weakly to each other once they are in suspension, and to obtain
aggregates the undifferentiated cell populations were given a brief exposure to trypsin, so that cell
clumps could be blown off the bottom of the dishes after the addition of alpha 10% FCS. These
clumps were either placed directly into 90 mm diameter bacteriological dishes, or cultured for
5-16 h at high density in 50 mm diameter bacteriological dishes to promote the formation of bigger
aggregates before transfer to the larger dish. Usually, retinoic acid at between 5 x 10~s M and 5 X
10"7M was added to the dishes 1 day after aggregation. One week to 10 days after aggregation, the
aggregates were plated out on to gelatin-coated tissue culture dishes in the presence or the absence
of retinoic acid (see Results).

   Approximately 1 X 107 undifferentiated cells were gently disaggregated as above, and incubated
for 5-16 h in the bottom of a 10 ml plastic centrifuge tube under 2-3 ml of alpha 10% FCS to
promote the formation of a coherent lump. This was transferred as a slurry beneath the left kidney
40                                  S. Thompson and others
capsule of a nude mouse using a wide-bore mouth-controlled micropipette (technique of Rayner &
Graham, 1982). The mice were palpated at weekly intervals to detect tumour growth, and the
tumours were weighed, fixed and sectioned as described previously (Rayner & Graham, 1982; lies,
Bramwell, Deussen & Graham, 1975).

Foetus samples
  Human embryonic liver and yolk sack were obtained from suction abortions. The foetal aspirates
were washed out of the collecting vessel with an equal volume of dextrose saline within 5 min of the
removal of the foetus. Within the next 2h, the material was washed free of blood by rinsing the
sample with 1-2 litres of ice-cold PBS through a plastic sieve with a mesh size of approximately
3 mm2. The liver and yolk sack could be easily recognized in the retained material when it had been
washed from the sieve into a 90 mm diameter bacteriological dish and viewed under a dissecting
microscope. The identity of the liver samples was also checked by sectioning small fragments. For
labelling, the yolk sack was torn open and the liver was cut into cubes approximately S mm . Each
sample weighed between 0-01 and 0-02g (wet weight) and within 3-4 h of foetal aspiration each was
placed in 1-2 ml of radioactive medium (see below). The age of the samples after fertilization ranged
between 5 and 7 weeks as estimated from the time of the last menstrual period.

   In suspension. Cells were taken from the dishes with either EGTA/PBS or with 0-125 % (w/v)
trypsin in PBS containing 0-5 mM-EDTA (EDTA/trypsin). The cells were resuspended at 1 X
10 cells per ml in alpha medium with 0' 1 % (w/v) sodium azide and 5 % (v/v) FCS; 25 \i 1 samples
were added to portions of various first-layer antibodies and incubated at 4°C for 45 min. Following
a single wash, the cells were resuspended in a 1/30 dilution of rabbit anti-mouse immunoglobulin
G (IgG)-fluorescein isothyocyanate (FITC) for 45 min at 4°C. After washing, the cells were moun-
ted on slides and viewed under a Zeiss IV epifluorescence condenser microscope. At least 100 cells
were counted in each sample.
   Labelling of cells on coverslips. Antibodies were diluted as described under Reagents (see below)
in complete phosphate-buffered saline (Dulbecco & Vogt, 1954) containing lOmM-sodium azide
and either 10 % (v/v) FCS, 1 mg/ml bovine serum albumin (BSA) or 1 mg/ml gelatin. For staining
of antigens exposed on the cell surface, live cells attached to coverslips were incubated in 80^*1 of
first antibody for 30-40 min at room temperature (RT). The coverslips were washed three times by
gentle immersion in PBS/azide/FCS and 80fi\ of second antibody was applied for a further
30-40 min. After washing as above, the cells were fixed in 4% (w/v) paraformaldehyde in PBS
(10min, RT), and the coverslips were mounted in 50% (v/v) glycerol/PBS, sealed with nail
varnish, and viewed under a Leitz microscope equipped with epi-illumination.
   The intracellular neurofilaments and glial fibrillary acidic proteins were stained by fixing the cells
in paraformaldehyde as above, permeabilizing the cells in 0-1 % (w/v) Triton X-100 in PBS (RT,
10 min), and then exposing them to the first antibody as described above.
   Labelling for analysis on the fluorescent-activated cell sorter (FACS). Cells were released from
the culture surface as described for cell passage. They were transferred to 10 ml plastic centrifuge
tubes (Sterilin), and washed in ice-cold PBS/azide with the addition of either FCS, gelatin or BSA
as described above. After aspirating the supematants, cell pellets of approximately 5x10* cells were
gently dispersed by tapping the tubes, and then incubated in 100 /U1 of first antibody on ice with
occasional shaking for 45 min. After two washes in 10 ml of medium, the cells were incubated for
45 min on ice in 100 ji 1 of second antibody with occasional shaking. The cell suspension was diluted
to 0 5 ml and analysed on a Becton Dickinson FACS II; 100000 cells were analysed per labelling
using a 100/im diameter nozzle, and the appropriate positive and negative control antibodies were
used to set the gates.

   Cells grown on quartz coverslips were fixed in a 10% (w/v) buffered neutral formalin for 12-24 h.
This fixation stabilizes DNA during Feulgen hydrolysis and minimizes intercellular variability in
staining. The cells were stained by the Feulgen reaction with hydrolysis at 22°C in 5 M-HC1 for 1 h.
The cytophotometric measurements were performed in a rapid scanning microspectrophotometer
                          Differentiation of human teratoma cells                                 41
equipped with a field-limiting device (Caspersson & Lomakka, 1970; Caspersson, 1979; Caspersson
& Kudynowski, 1980; Caspersson, Auer, Fallenius & Kudynowski, 1983). Imprints from human
cerebellum were prepared and used as staining controls. Such nuclei contain DNA values corres-
ponding to the normal diploid content (2-OC), with a variation in the majority of cells between 1 -9
and 2-1C. Occasionally, cerebellar nuclei with 2-3C values were recorded and this value is taken as
the upper limit of DNA values from a diploid nucleus; all measured DNA values over Tera-2 nuclei
were expressed in C units with respect to the cerebellar controls.
   In order to test if any quantitative errors were introduced when small strongly light-absorbing
objects were measured, the measurements were also performed off-peak at 610 nm, at which
wavelength the extinction was about 40 % of that obtained at the absorption peak around 546 nm.
Independent of cell type, the nuclei of control cells and of Tera-2 exhibited a constant ratio of
extinction at the two wavelengths. This shows that no error of any quantitative significance was
introduced when measuring small light-absorbing cells with compact chromatin.

Radioactive labelling and immunoprecipitation
   Near-confluent cell cultures were labelled for 20-26h in leucine-free MEM (Gibco Europe),
containing 0-5-5% (v/v) dialysed FCS, and between 40 and 200 jiCi/ml of L-[4,5-3H]leucine (sp.
act. 131 Ci/mmol, Amersham International pic, Amersham, Bucks). In 35 mm diameter dishes,
2ml of medium over 5 X 105 to 9 X 105 cells was used, and this was increased to between 3-5 and
5 ml over 1 X 106 to 2 X 106 cells in 50 mm diameter dishes.
   A variety of culture conditions was used to increase the chance of detecting the possible synthesis
and secretion of extra-embryonic foetal products. These included aggregating monolayer cultures
that had been exposed to retinoic acid for 7 days, and culturing monolayers exposed to retinoic acid
for 4 days in alpha 2 % FCS with the addition of either 0-1 % (v/v) gelatin (swine skin, type 1 from
Sigma), or 0-5 unit/ml insulin (Sigma), or 5 X 10~ 7 M 17B-oestradiol diluted from a 1 X 10~ 2 M
stock in dimethylsulphoxide (DMSO) (Sigma), or combinations of these substances for a further
4 days before labelling.
   Metabolically labelled cells and supernatants were collected and immunoprecipitated as follows.
Supernatants were collected and cleared of cell debris by micro-centrifugation at 13 000 g for 10 min
(volume 2ml or less) or centrifugation at 35 000 # for 15 min (greater than 2ml). Cells were
solubilized and removed from dishes by scraping them off the dish with a rubber policeman in PBS
containing 1 % (v/v) Nonidet P40 (NP40) and 0-1 mM-phenylmethylsulphonyl fluoride (PMSF)
as a proteolytic inhibitor. After repeated vortex mixing over a 30-min period at 4°C, the solubilized
cellular components were removed from any remaining cellular debris by micro-centrifugation.
Tissues and cell aggregates were treated as above.
   The amount of radioactivity incorporated into protein was determined by precipitation with
trichloroacetic acid of small portions of supernatants and solubilized cells. All samples were then
divided into fractions and stored at — 70°C before immunoprecipitation.
   Cell surface proteins were labelled by the lactoperoxidase/ I method exactly as previously
described (Stern et al. 1984).
   Immunoprecipitation was carried out by incubating 100 to 500/il portions of supernatants and
100^t 1 portions of solubilized cells with 10/zl of the appropriate antibody for 30min at room
temperature. To each tube was added 50/il of a 50% (w/v) suspension of protein A-Sepharose
(Pharmacia) in 1 % (v/v) NP40 in PBS, and the samples were incubated again for 30 min at RT.
Unbound material was then washed away from the Sepharose beads by five washes at 4°C under
gravity or brief centrifugation in 2-5 ml of 1 % (v/v) NP40 in 50mM-Tris HC1 (pH 7-4), contain-
ing 0-15 M-NaCl, 5mM-EDTA and 0-02% (w/v) NaN3. Finally, the antigen-antibody complexes
were removed from the beads by the addition of 50/il of 2X concentrated sodium dodecyl sul-
phate/polyacrylamide gel electrophoresis sample buffer lacking 2-mercaptoethanol (2-ME), and the
supernatants were harvested after micro-centrifugation. The samples were boiled after the addition
of 5 % (v/v) 2-ME and subjected to electrophoresis in 5'5 %, 8-0% and 13 % (w/v) polyacrylamide
gels, using the method of Laemmli (1970), as modified by Thompson & Maddy (1982). Molecular
weight standards were red blood cell membrane-protein bands 1,2,4-1,4-2, 5,6 and 7 with molecular
weights of 240, 220, 78, 72, 43, 35 and 29 (X103), respectively, and a 14C-labelled methylated protein
mixture containing myosin (200 X \C^MT), phosphorylase 6 (92-5 X KfiM^), bovine serum albumin
(BSA, 69 X \(PM,), ovalbumin (46 X UfiM,), carbonic anhydrase (30 X l(PMT), and lysozyme
42                                  S. Thompson and others
(14-3 X 10 iVfr). In comparisons between undifferentiated and differentiated cell labelling,
precipitation analysis was carried out by applying the antibody to samples that contained equal
trichloroacetic acid-precipitable counts. The samples were counted at an efficiency of approximately
40% in a LKB 1215 Beta counter.
   Preflashed Fuji X-ray film was used for autoradiography with Dupont Cronex Lighting Plus
screens (Laskey & Mills, 1977), and fluorography was conducted with scintillant-impregnated gels;
both procedures conducted at -70°C (Bonner & Laskey, 1974; Laskey & Mills, 1975).

   The specificities and originators of the reagents are described in Table 1. We thank the following
for gifts of reagents (dilutions used): Dr A. Williams, Dunn School of Pathology, Oxford, for
W6/32 (culture supernatant) and W3/25 (culture supernatant); Dr J. Fabre, Queen Victoria
Hospital, East Grinstead, Sussex, for F10.44.2 (1/100 ascites) and F15.42.1 (1/50 ascites); Dr D.
Solter, Wistar Institute, Philadelphia, for anti-SSEA-1 (1/100 culture supernatant) and anti-
SSEA-3 (culture supernatant); Dr M. Schachner, Department of Neurobiology, University of
Heidelberg, 69 Heidelberg, Im Neveheimer Feld 504, FDR, for 04 (1/1000 ascites); Professor A.
McMichael, John Radcliffe Hospital, Oxford, for PA 2.6 (1/100 ascites) and MHM 5 (1/100
ascites); Dr R.O. Thomson, Wellcome Laboratories, Beckenham, Kent, for tetanus toxin (10
mg/ml), affinity-purified horse anti-tetanus toxin antibody (1 /50), and fluorescein rabbit anti-horse
(1/20); D r J . Kemshead, ICRF Laboratories, Institute of Child Health, Guilford Street, London
WC1, for U.13A (1/500 ascites), 308 (1/50 direct FITC conjugate) and MIN 1 (1/100 ascites);
Dr B. H. Anderton, Department of Biochemistry, St George's Hospital Medical School, London
SW 17, for BF10 (1/1000 ascites) and RT97 (1/200 ascites).
   The other antibodies were prepared in the laboratory and used at the following dilutions: F12
A2B5 (1/100 ascites), 2-3 F9 (1/100 ascites), anti-GFAP (1/50 serum prepared as described by
Pruss, 1979).
   Other antibodies were obtained from the following: rabbit anti-human fibronectin (Dr D. Tur-
ner), and rabbit anti-apoprotein Al, anti-albumin, anti-transferrin, anti-AFP, anti-prealbumin,
anti-/3-HCG, from Boehringer; mouse anti-/32-microglobulin directly labelled monoclonal, from
Becton Dickinson. Second-layer antibodies were rabbit anti-mouse IgG-FITC (Miles, 1/60 to
1/100), and rhodamine-labelled sheep anti-mouse and anti-rabbit IgG (1/40).

   The results are divided into four sections. First, the growth and morphological
differentiation of the clones are described, and then we record the changes in surface
phenotype and secreted products that accompany their differentiation in retinoic acid.
Finally, we characterize the development of a neuron-like phenotype that forms after
aggregation and treatment with retinoic acid.

Clonal growth, karyotypes and morphological differentiation
   In the cloning procedure, it was apparent that isolated single cells of Tera-2
required feeder cells if they were to form colonies: none of the single cells formed
visible colonies in alpha 10% FCS alone (« = 75), while 4 1 % grew into visible
colonies during the 3 weeks after the single cells had been plated on 10T^ feeder cells
(n = 113). However, none of the clones required feeder cells once they were suf-
ficiently numerous to be passaged at high cell density.
   A number of different cell morphologies were seen at the first passage of the clones.
Cells with long branching processes were found in groups in 7/25 of the clones. We
selected clones 5, 12 and 13 for detailed study because they could be maintained with
                          Differentiation of human teratoma cells                                  43

    Fig. 1. Tera-2 clone 13 at passage 3. The piled up and monolayer forms of undifferen-
    tiated cells occupy most of the photograph. In the monolayer at the top right, the high
    nuclear-to-volume ratio is clear. At the bottom left are several differentiated cells; these
    had formed spontaneously but they resemble the cell shapes formed in response to retinoic
    acid. Bar. SO/im.

a relatively homogeneous appearance. The form of the majority of cells in untreated
populations of these clones is described as undifferentiated (Fig. 1).
   The karyotypes of clones 5, 12 and 13 were all abnormal and the cells in each clone
had a range of mitotic chromosome counts within two passages of cloning: clone 5 had
a modal chromosome number of 58 (range 57-59, n = 10); clone 12 had a modal
chromosome number of 63 (range 62-65, n = 8); clone 13 had a modal chromosome
number of 60 (range 57-63, n = 100). These chromosome numbers are similar to
those found in uncloned Tera-2 cells and in the NTera-2 clones (Andrews et al. 1984).
   The G-banded karyotype of Tera-2 clone 13 was analysed in detail and its features
showed some similarity to those previously described for uncloned Tera-2 and the
NTera-2 clones (Andrews et al 1984). There were the following normal-looking
chromosomes (Fig. 2): one copy of chromosomes Y and 1; two copies of chromosomes
2, 4, 5, 6, 8, 10, 12, 13, 14, 15, 17, 18, 19, 20 and 22; and three copies of chromosomes
3, 7, 9, 11 and 21. The presence of rearranged chromosomes resulted in further
recognizable partial trisomies and tetrasomies. For example, there were three
additional copies of chromosome 1, each with the short arm deleted in different
         5. Thompson and others






         0%                       f




V                                               l
               CD                     0)

                   Fig. 2
                         Differentiation of human teratoma cells                               45
places; the metacentric chromosome derived from the short arm of chromosome 1
translocated to the short arm of chromosome 2 (Andrews et al. 1984), and a
chromosome 16 with a deleted short arm. The single X chromosome was abnormal,
with a wide positive band in the long arm at q25. Neither the previously reported
numerous chromosome 12 involvements, the isochromosome of the long arm of 17,
nor the deleted chromosome 22 were seen (Andrews et al. 1984). A number of
characteristic marker chromosomes was observed but we could not confidently deter-
mine which chromosomes had contributed to these extra elements. To reduce the
possibility that changes in the modal chromosome number might alter the behaviour
of the clones during the course of these experiments, we made our observations on
cells that had been grown for less than 16 passages after cloning.
   Clones 12 and 13 were shown to form tumours at 2-3 months after inoculation of
the cells into nude mice. The tumour formed by clone 12 cells weighed 0-032g, and
the two tumours from clone 13 cells weighed 0-014 and 0-055 g. On sectioning these
were found to contain groups of cells with the appearance of carcinoma, loose connec-
tive tissue and cuboidal epithelia around cavities, and there were long processes deep
in the tumour that had the appearance and staining properties of nerve axons (Figs
3-6). There were also cells with a variety of other morphologies. These tumours
resembled those formed by uncloned Tera-2 in nude mice (Jewett, 1978). Unlike the
tumours formed by the NTera-2 clones, glandular epithelia were not found, and the
tumours also lacked the cells described by histopathologists as 'embryonal carcinoma
cells arranged in papillary formation', which are found in primary teratomas.
   After exposure to retinoic acid in monolayer culture, the cell numbers in the dishes
increased approximately fivefold in the next 7-10 days. Measurements of amounts of
nuclear DNA in the undifferentiated and treated populations showed that within 8
days the cells transit from a rapidly dividing state, with DNA amounts ranging from
3C to 8C, to a population in which the majority of the cells are arrested in the G\ phase
of the cell cycle displaying 3C nuclear DNA amounts (Figs 7—10). In many experi-
ments, the retinoic acid was withdrawn at this time, and the cell number did not
increase further during the following 2 months.
   A variety of cell morphologies was seen in the treated cultures. The undifferen-
tiated cells seem to have disappeared from these cultures because their characteristic
morphology was not apparent after an extensive search of several hundred retinoic-
acid-treated cultures. Further, when the cells were cultured on for several months in
the absence of retinoic acid, then in four separate experiments with both clone 12 and
clone 13 cells, the undifferentiated morphology did not recur.

Changes in surface phenotype on differentiation in monolayer culture
  We characterized the phenotype of undifferentiated cell populations and of the

    Fig. 2. G-banded karyotype of Tera-2, clone 13:61, X (abnormal) Y, - 1 , +del 1, +del
    1, +del 1, +t(l; 2), +3, +7, +9, +11, +del 16, +21, +mar, +mar, +mar, +mar, +mar,
    + mar (?12h). Although there were variations in the marker chromosomes, the other
    features described in this legend were apparent in each of the 10 karyotypes analysed in
46   5. Thompson and others


            Fig8 3-6
                          Differentiation of human teratoma cells                                 47
differentiated cells formed in monolayer cultures in response to retinoic acid. The
antigenic phenotype of the undifferentiated cells was distinct from that of the NTera-
2 clones, and there were striking changes in the expression of HLA-A,B,C common
determinants during this differentiation.
   The expression of HLA-A,B,C common antigens and ^2-microglobulin was
studied with monoclonal antibodies to these determinants (W6/32, PA 2-6, anti-jSz).
The undifferentiated cell populations rarely reacted with these monoclonal antibodies
(Table 2) and a similar result was obtained with MHM5, another monoclonal to
HLA-A,B,C common determinants (results not shown). In those experiments in
which more than 3 % of the undifferentiated cell populations appeared to react with
these antibodies, an independent observer had previously recorded that the popula-
tions contained morphologically distinct flat cells. There is therefore a close associa-
tion between the undifferentiated appearance and the lack of reactivity with these
antibodies. The majority of these observations were made by eye, but the rare reactiv-
ity of undifferentiated cell populations was also confirmed on FACS. It was found that
2-5 % more cells reacted with W6/32 than with an irrelevant antibody (W3/25) in a
sample that was scored as 3 % reactive with W6/32 by eye.
   After exposure to retinoic acid for a week, there was a considerable increase in the
proportion of cells that reacted with the monoclonal antibodies W6/32, PA 2.6, and
the anti-/S2-microglobulin antibody (Table 2). This observation was confirmed on the
FACS using the W6/32 antibody reaction to clone 12. This new phenotype was stable
following withdrawal of retinoic acid after treatment for 1 week; for example, 1 month
after withdrawal, the differentiated clone 13 cells exhibited 97% reactivity with
W6/32, 100% reactivity with PA 2.6, and 100% reactivity with the anti-/S2-micro-
globulin antibody. In another experiment, 4 months after the withdrawal of retinoic
acid, 55% of clone 12 differentiated cells reacted with W6/32. The increase in
reactivity with W6/32, PA 2.6, and anti-/J2-microglobulin antibodies was progressive
during continuous exposure to retinoic acid: for instance, in one particular experi-
ment, the proportion of clone 12 cells that reacted with W6/32 increased progressively
to give 1-3 % reactivity at 2 days, 7 % at 4 days, 26 % at 6 days and 53 % at 9 days.
The reactivity of the cells with PA 2.6 and anti-/?2-microglobulin antibodies increased
in parallel with W6/32 reactivity in this experiment.

    Figs 3-6. Histology of tumours formed by Tera-2 cells grown under the kidney capsule
    of nude mice.
       Fig. 3. Tera-2, clone 13. The tumour principally consists of nests of carcinoma cells in
    loose connective tissue. Apparently normal mouse kidney tubules occupy the bottom half
    of the figure. Bar, 50 (im.
       Fig. 4. Tera-2, clone 13. A nest of carcinoma cells with a high mitotic index beside
    kidney tubules along the right of the picture. Bar, 20 |im.
       Fig. 5. Tubule within a Tera-2, clone 12 tumour. The simple epithelium is surrounded
    by connective tissue. Bar, 20 fim.
       Fig. 6. Axon-like process within a Tera-2, clone 12 tumour. It is impossible to exclude
    the possibility that the axon is derived from the surrounding host kidney but it was found
    deep in the tumour. Holmes' silver stain. Bar, 20|tm.
                                    5. Thompson and others

                                 cl12                                               cl13


                                                                                    + RA

       2C      4C              8C        10C                               6C      8C      10C
                                        DNA content (relative units)
    Figs 7-10. Distribution of nuclear DNA content (Feulgen staining material) in undif-
    ferentiated cells and differentiated cells, formed after 8 days retinoic acid (RA) treatment.
    Each histogram is based on the cytophotometric analysis of approximately 100 cells. The
    DNA values are expressed in relation to the mean staining values of the control cerebellum
    cells: these were given the arbitrary value of 2C (denoting the diploid DNA content), and
    a broken line on each figure represents the upper limit of the staining control (2'1C).
       Fig. 7. Clone 12, undifferentiated.
       Fig. 8. Clone 12, differentiated.
       Fig. 9. Clone 13, undifferentiated.
       Fig. 10. Clone 13, differentiated.

   The monoclonal antibody W6/32 was used to immunoprecipitate leucine-labelled
material from the differentiated cells and it specifically precipitated a 43 X 1037Wr
molecule, which is slightly smaller than the expected size of HLA-A,B,C polypep-
   The majority of the undifferentiated cells of the NTera-2 clones were known to
                            Differentiation of human teratoma cells                                    49

                                Table 1. Reactivities of reagents
Monoclonal antibody
 or reagent
 (original reference)          Major determinant or molecule bearing determinant (reference)
W6/32 (1)                 HLA-A,B,C (44*) in association with human or mouse ft-microglobulin
                             (12*). Weak reaction with HLA alone, none with /32-microglobulin
                             alone (2,3,4,5). Structural gene on chromosome 6
PA 2.6 (6)                HLA-A,B,C (44#) in association with human /32-microglobulin (12*) (7)
MHM-5 (8)                 HLA-A,B,C (44*) in association with human /32-microglobulin (12*) (8)
Anti-ft-microglobulin     /32-microglobulin (12*) alone or in association with HLA-A,B,C.
  (9)                        Structural gene on chromosome 15 (7)
SSEA-3 (10)               3GalNAc/31—»3Galal—>4Gal, on glycoproteins and glycolipids of an
                             human teratoma line (2102Ep); on preimplantation mouse embryos
                             up to the 8-cell stage (10,11,12)
SSEA-1 (13)               Gal/51—+4(Fucal—»3)GlcNAc, principally found on glycolipids; on
                             preimplantation mouse embryos from morula stage onwards
F10.44.2(17)              87-89 X 103jWr sialoglycoproteins on human brain and blood
                             mononulear cells. Structural locus or expression control on
                             chromosome 11 (17,18,19)
F15.42.1 (20)             23-5 X 1037tfr, human Thy-1 (20)
RT97(21)                  210 X \02MT neurofilament on human brain (21)
BF10(21)                  155 X 103Mr neurofilament on human brain (21)
Tetanus toxin             Reacts with gangliosides with decreasing affinity, G T l b , G D l b , G D l a .
                             Reacts with neurons, astrocytes, pancreatic cells, thyroid cells and
                             present on 10th day mouse brain (22,23,24,25,26,27)
F12.A2B5 (28)             Highest affinity for GQlc gangliosides. Reacts with neurons, some
                             astrocytes, some pre-oligodendrocytes and 'APUD' cells (25,28,29)
GFAP (30)                 Glial fibrillary acidic protein. Restricted to glial cells (30)
U13A(31)                  On human foetal muscle cultures, reacting with myoblasts, myotubes
                             immature myofibres and adult regenerating myofibres. On neurons in
                             foetal brain cultures (32,33,34,35)
S.1H11 (33)               Distribution similar to U13A
MIN 1 (36)                Astrocytes of human brain cultures and granulocytes (36,37)
0 4 (39)                  Mouse oligodendrocytes and galactocerebroside positive cells on human
                            brain cultures (35,38,39)
2.3 F9 (40)               Fibronectin (40)

   The origin of the reagents and the distribution of their reactivities is described in these references.
(1) Williams et al. (1977); (2) Barnstable et al. (1978); (3) Parham et al. (1979); (4) Trowsdalee/
al. (1980); (5) Goodfellowet al. (1976); (6) Parham & Bodmer (1978); (7) Brodskyefa/. (1979);
(8) Hildreth (1982); (9) Becton Dickinson; (10) Shevinsky et al. (1982); (11) Kannagi et al.
(1983a); (12) Kannagi et al. (19836); (13) Solter & Knowles (1978); (14) Gooietal. (1981); (15)
Hakomorie* al. (1981); (16) Kannagi et al. (1982); (17) Dalchau et al. (1980); (18) Goodfellow
et al. (19S2a,b); (19) McKenzieet al. (1982); (20) McKenzie & Fabre (1981); (21) Andertonet al.
(1982); (22) Van Heynigen (1963); (23) Lebdeen&Mellanby (1977); (24) Rogers &Snyder (1981);
(25) Eisenbarth <* al. (1982); (26) Raff etal. (1983); (27) Koulakoff et al. (1983); (28) Eisenbarth
et al. (1979); (29) Kasai & Yu (1983); (30) Pruss (1979); (31) J. T. Kemshead; (32) Walsh et al.
(1983); (33) Walsh et al. (1981); (34) Walsh & Ritter (1981); (35) Hurko & Walsh (1983); (36) G.
Dickson, personal communication; (37) Kemshead et al. (1981); (38) Dickson et al. (1983); (39)
Somner & Schachner (1981); (40) Walsh et al. (1981).
   •Molecular weights are given in parenthesis (Xl0~ 3 ).
50                                      5. Thompson and others

             Table 2. Reactivity of undifferentiated and differentiated cells
                            Percentage of cells reactive: mean, (individual observations)

                         Undifferentiated populations                   Differentiated populations
Reagent       Uncloned       Clone 5      Clone 12   Clone 13       Clone 5     Clone 12     Clone 13

W6/32               0           3           2-2          01              18        40-6         20-8
                   (0)       (0-5,2,7) (0,0,2,3,6) (0,0,0,0,0,          (18)      (7,46,     (1,19,24,
                                                   0,0,0,0,9)                    54,54)       25,35)
PA2.6             nd            nd          3-2          0               nd        56-5         38
                                           (3-2)        (0)                      (40,73)       (38)
Anti-ft           nd            3           4-5          0               nd        54-4           6
                               (3)         (4-5)        (0)                       (54-4)         (6)
SSEA-3              0           0           0            0                0         0             0
                   (0)         (0,0)      (0,0)        (0,0)            (0,0)      (0,0)        (0,0)
SSEA-1              4          31          43          49                36        33           55
                   (4)      (27,28,38) (20,67) (20,22,42,               (36)     (27,39)    (0,64,100)
F10.44.2          80           92          85-5        66                35        53-5         69
                 (80)         (89,95)    (76,95)    (3,60,81,           (35)     (20,87)      (67,71)
F15.42.1          95           97          81          94-6              39        76-5        34
                 (95)        (95,100)    (67,95) (86,94,95,             (39)     (57,96)      (34)

BF10              nd            0            0           0               nd         3           nd
                               (0)          (0)         (0)                        (3)
Tetanus           nd            1-1          1           6-4             nd        17           nd
  toxin                        (1-1)       (0,2)        (6-4)                     (17)
F12A2B5           nd            0-75         0           2-8             nd         7           nd
                               (0-75)       (0)         (2-8)                      (7)
Anti-GFAP         nd            0            0           0                0         0            0
                               (0)          (0)         (0)              (0)       (0)          (0)
U13A              nd            nd           8           nd              nd        43            nd
                                            (8)                                   (43)

  The reactivities of the cells were not always tested on the same day with the same reagents;
because these are not matched comparisons we give the individual observations made on different
days with different cell populations. This establishes the reproducibility of the results. Observations
on cells in suspension are recorded in the top half of the Table. Studies on cells attached to coverslips
are in the bottom half of the Table.
  nd, not determined.

react with W6/32 (Andrews et al. 1984). We therefore compared the reactivity of cells
from NTera-2 clone B9, Tera-2 clones 5 and 12, and the cloned 2102Ep cells, which
were also known to react with W6/32 (Table 3). It is clear that the expression of HLA-
A,B,C determinants distinguishes the two sets of clones that were independently
derived from Tera-2.
                          Differentiation of human teratoma cells                                   51

    Table 3. Comparison ofNTera-2 clones, Tera-2 clones 5 and 12 and2102Ep
                          Undifferentiated cell population                  Differenti: ited cell
                                          K                                    popula tions
                Clone 5       Clone 12        N.Tera-2 2102Ep clone                   A
                                              clone B9     2A6          Clone 5           Clone 12

W6/32              <1               0            24           53             18               7
SSEA-3               0              0            87          >9S              0               0
SSEA-1              27             67            27            2             36              27
F10.44.2          >95            >95            >95          >95             35              20
F15.42.1          >95            >95            >95          >95             39              57

  Comparisons between the reactivity of these cells with various reagents was made in one experi-
merit using the same reagents for each,

    In the same experiment, Tera-2 clones 5 and 12 were further distinguished from
the NTera-2 clone by the lack of reactivity with the anti-SSEA-3 monoclonal anti-
body. This monoclonal antibody recognizes a sugar sequence that is expressed in the
earliest stages of preimplantation mouse development (Table 1), and it has been
proposed that it identifies human teratoma cells at an analogous stage of development;
it reacts with the NTera-2 cells and the 2102Ep cells (Shevinsky, Knowles, Damjanov
& Solter, 1982; Andrews et al. 1984). The SSEA-3 determinants are clearly absent
from the morphologically similar Tera-2 clones 5, 12 and 13, and they do not appear
on differentiation (Tables 2 and 3).
    The monoclonal antibody against SSEA-1 recognizes determinants that appear at
a later stage of preimplantation mouse development than those recognized by the
antibody against SSEA-3 (Table 1; Solter & Knowles, 1978). This reagent reacted
with between a third and an half of the" cells from clones 5, 12 and 13 in both the
undifferentiated and differentiated populations (Table 2 and 3). About a quarter of
NTera-2 clone B9 cells reacted, but only a few 2102Ep cells reacted.
    Most human teratoma-derived lines with an undifferentiated appearance express
Thy-1 (Andrews et al. 1984), and the undifferentiated cells of clones 5, 12 and 13
behaved similarly when tested with the monoclonal antibody F15.42.1; on dif-
ferentiation there was a consistent decrease in the proportion of cells that expressed
this determinant (Table 2 and 3).
   The monoclonal antibody F10.44.2 recognizes an approximately 90 X 103Mr
sialoglycoprotein that is present on human brain and white blood cells. This reactivity
was also present on the cells of Tera-2 clones 5, 12 and 13, as well as on NTera-2 clone
B9 and 2102Ep cells (Tables 2 and 3). The undifferentiated cells of Tera-2 clones 5,
12 and 13 were surface-labelled with lactoperoxidase/12SI and the detergent-
solubilized labelled proteins were precipitated with this antibody'. The molecule that
was immunoprecipitated ran as a broad band, as would be expected of a glycoprotein,
with an apparent molecular weight of 85 to 95 X 103Mr. This immunoprecipitate is
52                                 5. Thompson and others
shown for clone 13 cells in Fig. 11. Although the proportion of the cells that react with
F10.44.2 decreases on differentiation, there is no change in the molecular weight of
this protein (data not shown). Fig. 11 also illustrates other general changes in cell
surface phenotype that occur after retinoic acid treatment. For example, proteins of
molecular weights 260, 200, 75, 52, 50, 48 X l^Mr seem to disappear and a
prominent band appears at a molecular weight of 67 X lCPM,-.

Changes in synthesized proteins and secreted products on differentiation
  The pattern of protein synthesis changed on retinoic acid treatment, and this
change was obvious both in cell lysates and in the secreted products of the cells (data
                               1       2           3






     Fig. 11. Cell surface labelled components of undifferentiated clone 13 (lane 1), and
     differentiated clone 13 (lane 2). The material immunoprecipitated by the monoclonal
     antibody F10.44.2 from the undifferentiated cell sample is shown in lane 3. The samples
     were all run on the same 8 % polyacrylamide gel, but a longer exposure was required for
     detection of the immunoprecipitated molecule. MT values are given (X 1(T3) on the right.
                         Differentiation of human teratoma cells                                   53
not shown). The synthesis of fibronectin was examined, since this protein is often
expressed as human teratoma cells differentiate (Andrews, 1982). The anti-
fibronectin antibody was used to precipitate material from the culture medium over


                                         * * > • •



   Fig. 12. Labelled cell culture medium treated with anti-fibronectin antibody. Undifferen-
   tiated clone 13 (lane 1), differentiated clone 13 (lane 2), foetal yolk sack (lane 3), foetal
   liver (lane 4), undifferentiated clone 12 (lane 5), differentiated clone 12 (lane 6) and
   differentiated clone 12, after incubation with gelatin, insulin and /3-oestradiol (lane 7). F
   marks the position of fibronectin. Other radioactive bands were specifically and repeatedly
   immunoprecipitated by this antibody. Immunoprecipitation was carried out on samples
   with 480000 c.p.m. of trichloroacetic acid-precipitable material, except for the yolk sack
   sample, which contained 100000 c.p.m. The total immunoprecipitate was applied to the
   gel, except for differentiated clone 12 samples, where only 40 % of the immunoprecipitate
   (lane 6), or 20% of the immunoprecipitate (lane 7) was added to the gel. Fluorograph of
   a 5 5 % polyacrylamide gel.
54                             iS. Thompson and others
undifferentiated and differentiated cells; the electrophoretic analysis of the
precipitated material for Tera-2 clones 12 and 13 is presented in Fig. 12. The undif-
ferentiated cells of Tera-2 clones 5, 12 and 13 secreted fibronectin, and this had an
approximate molecular weight of 240 X 103. All the differentiated cultures apparently
secreted proportionately more of this protein compared to the undifferentiated cell
cultures. On differentiation, the apparent molecular weight of the secreted fibronec-
tin decreased by about 5 X \{9Mr (Fig. 12).
   It was possible that a multipotential stem cell from a teratoma might be induced to
secrete those products that are characteristic of the extra-embryonic membranes of the
human foetus (see Discussion).
   However, we were unable to detect the synthesis and secretion of /3-HCG, AFP,
prealbumin, albumin, apoprotein Al or transferrin from labelled undifferentiated and
differentiated cells or from the culture medium over these cells. In contrast, all these
products except HCG were secreted into the medium by the human foetal liver and
yolk sack. We were able to immunoprecipitate specifically AFP, albumin, apoprotein
Al and transferrin from these tissues and from the culture medium conditioned by
these tissues, even when the samples contained down to 30-fold less counts in total
trichloroacetic acid-precipitable material.

Appearance of cells vnth a neuron-like phenotype
   The undifferentiated and differentiated cells in monolayers rarely reacted with
reagents that react with human neurons (Table 2). A few cells with long processes
were seen amongst the variety of cell morphologies formed in response to retinoic
acid. The proportion of cells that reacted with reagents that mark neurons increased
on retinoic acid treatment; these reagents included tetanus toxin, F12 A2B5, U13A
and5.1Hll (Table 2).
    Cells with long branching processes were obvious in cultures that had been exposed
to retinoic acid as aggregates, and then grown on, after a week's treatment in either
the presence or absence of retinoic acid; continued treatment with retinoic acid had
little effect on the results (Figs 13—18). All the cells with long processes in such
cultures of clone 12 reacted with the anti-neurofilament monoclonal antibody RT97
(where number of cells observed (n = 46), with the anti-specific ganglioside reagent
tetanus toxin (n = 32), with the anti-GQlc ganglioside monoclonal antibody F12
A2B5 (n = 14), with the anti-Thy-1 monoclonal antibody F15.42.1, and with two
monoclonal antibodies that are characteristic of excitable cells, namely U13A (n = 13)
and 5.1H11 (n = 13) (Figs 13-18). None of the cells reacted with the monoclonal
antibodies MIN 1 (n = 42), 04 (n = 20), or with the glial marker anti-GFAP
(n = 100). Similarly, all cells with long processes in previously aggregated and treated
cultures of clones 5 and 13 reacted with the anti-neurofilament antibody BF10. In
each experiment, the proportion of cells that reacted with the anti-neurofilament
antibodies varied between 1 and 3 %.
   Three of the clones that spontaneously formed cells with branched processes at first
passage were tested for their ability to express neuron markers when they differen-
tiated at high density in the absence of retinoic acid. Clones 4, 25 and 29 displayed
                          Differentiation of human teratoma cells                                 55


     Figs 13-18. Tera-2 expression of antigens that are characteristic of the nervous system.
     Tera-2 clone 12 cells were aggregated in the presence of retinoic acid, and the aggregates
     grown up for a further 4 weeks (see Materials and Methods). They were stained with
     tetanus toxin (Figs 13, 14), with F12 A2B5 monoclonal antibody (Figs 15, 16), and with
     the monoclonal antibody to neurofilaments, BF10 (Figs 17, 18). Figs 13, 15 and 17,
     photographed with phase optics. Figs 14, 16 and 18 are the paired fields photographed
     under fluorescent illumination. Bars: Fig. 13, 100 fim\ Figs 15 and 17, 50/an.

cells with processes that stained with the anti-neurofilament monoclonal antibody
BF10, and when tested these cells also reacted with tetanus toxin and F12 A2B5.
56                             S. Thompson and others


Recognition of the undifferentiated cell
  The morphology of the rapidly growing small cells in our cultures resembles that
described for the NTera-2 clones and that observed in a variety of other teratoma-
derived cell lines (e.g. see Andrews et al. 1980). Fig. 1 shows that these Tera-2 cells
usually have a more flattened appearance than NTera-2 cells (e.g. see Andrews et al.
1984, fig. 1A), although they retain the high nuclear : cytoplasm profile ratio of the
growing NTera-2 cells. We have called it an undifferentiated cell (Fig. 1). The main
reason for believing that single undifferentiated cells can develop into cells with the
variety of morphologies seen in these experiments, is that most of the cells in the Tera-
2 cultures that were cloned looked undifferentiated; 41 % of single cells picked from
these cultures to feeder cells subsequently grew on, to form passage 1 colonies in
which a wide range of morphologies could be seen, including cells with long processes.
There are several reasons for believing that they are the only progressively growing
cells in cultures of Tera-2 clones 5, 12 and 13. First, cells with this morphology were
abundant in all passage 1 bottles in which cells continued to multiply, and when they
were absent or disappeared from passage 1 bottles, then the remaining cells did not
grow progressively over an eight-month period. Second, the undifferentiated cell
features disappeared on retinoic acid treatment, and the treated populations did not
grow progressively after retinoic acid had been withdrawn for many months. Such
cultures did not grow progressively, even when the cells were sub-confluent.
   We have studied clones that were relatively easy to grow as cell populations in which
the majority of cells looked undifferentiated. Nevertheless, it was common to observe
that about 5 % of the cells had a more flattened profile. The origin of this cell type is
not known. It is likely that the appearance of such cells in undifferentiated cultures
indicates that a small proportion of the cells have spontaneously started on the changes
that lead towards the phenotype of the retinoic acid-treated cells, whose form they
resemble. Similar 'spontaneous' differentiation has also been observed in other studies
with Tera-2 clone 13 cells (Engstrom, Heath & Rees, unpublished data). It is also
possible that such cells could have been produced by chromosome changes in the
population because there was considerable variation in mitotic cell chromosome num-
bers in individual clones within two passages of cloning. We may not have eliminated
this possibility by restricting these observations to the first 15 passages after cloning.

Cell surface phenotypic changes
   Despite the common ability of both these Tera-2 clones and the NTera-2 clones to
form neuron-like cells, there were considerable differences between the phenotype of
the small progressively growing cells of these two sets of clones. These differences
have been confirmed by the exchange of cells and reagents with Dr Peter Andrews.
   HLA-A,B,C and $2-microglobulin. It is agreed that uncloned Tera-2 cells have few,
if any, HLA-A,B,C determinants on their surface that are available for reaction with
antibodies against these determinants; these determinants have been searched for
                       Differentiation of human teratoma cells                          57
with monoclonal antibodies, human alloantisera, and rabbit antisera to purified HLA
(Andrews, Bronson, Wiles & Goodfellow, 1981; Avner, Bono, Berger & Fellous,
1981; Mcllhinney, 1981). The uncloned Tera-2 cells used in this work rarely reacted
with W6/32 (see Table 2). Similarly /32-microglobulin appears to be rare or absent
from the surface of the uncloned cells; its availability for reaction with antibodies has
been tested with xenoantisera and monoclonal antibodies (Holden et al. 1977; An-
drews et al. 1981; Avner et al. 1981; Andrews, 1983).
   The three monoclonal antibodies that we have used to detect the HLA-A,B,C
common determinants on the undifferentiated cells of these Tera-2 clones, would
react avidly with these determinants only in the presence of /32-microglobulin (Par-
ham, Barnstable & Bodmer, 1979; Trowsdale, Travers, Bodmer & Pattillo, 1980).
Since /32-microglobulin determinants are rare or absent from the undifferentiated
cells, we are unable to exclude the possibility that the HLA-A,B,C polypeptides exist
as free glycoprotein. It is also possible that these determinants are not inserted into
the cell membrane, and are therefore not available for reaction with antibodies in these
cell surface assays.
   We interpret the low proportion of cells in the undifferentiated cell populations that
react with W6/32 and the anti-/S2-microglobulin antibody, as either the tail of a
distribution of a population of very weakly positive cells (Andrews et al. 1981), or as
evidence that some of the cells have started to diverge from the undifferentiated cell
phenotype. We favour the latter interpretation because W6/32 reacting cells were
seen only when the undifferentiated cell population had been scored by an indepen-
dent observer as containing 5 % or more flattened cells, and because these reacting
cells became more abundant as the proportion of flattened cells increased after retinoic
acid treatment.
   Tera-2 clones 5, 12 and 13 resemble mouse embryonal carcinoma cells in the lack
of expression of major histocompatibility complex (MHC) products (reviewed by
Stern, 1983). Amongst the other human teratoma-derived lines only SuSa and LICR
LON HT53 rarely react with antibodies to HLA-A,B,C determinants; the latter cell
line forms AFP, when it is grown as a tumour (Hogan, Fellous, Avner & Jacob, 1977;
Mcllhinney & Patel, 1983).
   Sugar specificities. Few cells in uncloned Tera-2 populations react with the
monoclonal antibody against SSEA-3 (Shevinsky, Knowles, Damjanov & Solter,
1982; Andrews et al. 1984; observations, this paper). It is therefore not surprising
that Tera-2 clones 5, 12 and 13 also lack this sugar determinant and the overlapping
epitope recognized by the antibody against SSEA-4 (Peter Andrews, personal com-
munication). Although the antibody against SSEA-3 reacts with some 'embryonal
carcinoma' cells in primary teratomas (Damjanov et al. 1982), the phenotype of these
clones demonstrates that tumorigenic bipotential cells that do not react with the
antibody against SSEA-3 can be derived from human teratomas.
   Uncloned Tera-2 cells show variable levels of reaction with the antibody against
SSEA-1, and the stem cell populations of NTera-2 clones rarely react (Andrews et al.
1980, 1984; Andrews, 1984). Since this monoclonal antibody does not react with
embryonal carcinoma in primary solid human tumours, it is thought to mark more

58                              5. Thompson and others
differentiated cells, including the differentiated cells produced by retinoic acid treat-
ment of NTera-2 (Damjanov et al. 1982; Andrews et at. 1984; Andrews, 1984). A
substantial proportion of both the undifferentiated and differentiated populations of
the clones 5, 12 and 13 reacted with this antibody, which also reacts with the em-
bryonal carcinoma cells of mouse teratoma (Solter & Knowles, 1978).
   Uncloned Tera-2 cells also express a number of other sugar specificities in common
with mouse embryonal carcinoma. The monoclonal antibody 2H9 reacts with both,
as does the lectin peanut agglutinin (Bono et al. 1981; Stern et al. 1984). Further,
antibodies raised against mouse embryonal carcinoma cross-react with Tera-2 cells
(Ostrand-Rosenberg, Edidin & Jewett, 1977).

Extraembryonic secreted proteins
   The majority of primary human teratomas contain some cells that react with
antibodies against /J-HCG and other cells that react with anti-AFP antibodies;
frequently the serum of such patients contains elevated levels of these glycoproteins
(papers in Norgaard-Pedersen, 1978; Anderson, Jones & Milford Ward, 1981; New-
lands & Reynolds, 1983). A fully multipotential stem cell from a teratoma would be
expected to be able to develop into cell types that synthesize both these products.
Further, the abundance of cells in primary teratomas that resemble cells in the yolk
sack (e.g. see Beilby, Home, Milne & Parkinson, 1979), suggests that, in addition to
AFP, the other secreted serum proteins of the yolk sack would be produced by the
differentiated derivatives of a multipotential cell; these serum proteins include
prealbumin, albumin, transferrin and apoprotein Al (Gitlin & Perricelli, 1970; Git-
lin, Perricelli & Gitlin, 1972; Shi et al. 1984).
   We have been unable to detect the synthesis of any of these products under a variety
of culture conditions. This is unlikely to be due to technical difficulties because these
products could be readily detected in cultures of human yolk sack and foetal liver that
contained down to 30-fold less total trichloroacetic acid-precipitable counts. Similar-
ly, it has been impossible to detect the synthesis of HCG or AFP in cultures of the
NTera-2 clones either before or after retinoic acid treatment (Andrews et al. 1984).
   The failure to induce the synthesis of these products may either indicate that the
cells are not fully multipotential stem cells, or show that it is difficult to reproduce the
unknown conditions that lead to the synthesis of these products in a solid tumour; the
culture conditions used here included those that promote the secretion of these
products in primary hepatocyte cultures (Be"langer et al. 1978).
   Fibronectin was detectable in the cultures of both the undifferentiated cells and
their retinoic-acid-treated derivatives. Although there are some human teratoma cell
lines that do not synthesize fibronectin, our observations confirm previous studies on
both uncloned Tera-2 and the NTera-2 clones; it is also known that mouse embryonal
carcinoma cells synthesize and secrete this molecule (Wolfe, Mautner, Hogan & Tilly,
1979; Hogan, 1980; Andrews, 1982; Mcllhinney & Patel, 1983; Cossu & Warren,
1983; Andrews et al. 1984). The high molecular weight form of fibronectin is secreted
by human teratoma lines, primary human teratomas, amniotic epithelial cells and
mouse embryonal carcinoma cells; in the case of both the human and mouse teratoma
                       Differentiation of human teratoma cells                        59
cells the high molecular weight is attributed to the covalent linkage of fibronectin to
lactosaminoglycans and heparan sulphate, but it is also possible that the differential
splicing of fibronectin messenger RNA might contribute to this shift in apparent
molecular weight (Crouch et al. 1978; Alitalo et al. 1980; Ruoslahti et al. 1981;
Mcllhinney & Patel, 1983; Cossu, Andrews & Warren, 1983; Schwarzbauer, Tamku,
Lemischka & Hynes, 1983; Kornblihtt, Vibe-Pedersen & Baralle, 1984). Whatever
the explanation is for the decrease in the molecular weight of fibronectin, it is clear
that one of the secreted proteins produced by Tera-2 clones 5, 12 and 13 changes on

Markers of neurons
    A number of reagents have been shown to react with human neurons in culture.
These include tetanus toxin, anti-Thy-l(F15.42.1), anti-ganglioside GQlc (F12
A2BS) and the monoclonal antibody U13A. However, neurofilament proteins are the
only apparently unequivocal markers of neurons used in this study (see Table 1). The
presence of cells with long branched processes that react with the two anti-
neurofilament antibodies BF10 and RT97 (Anderton et al. 1982) strongly suggests
that neuron-like cells are formed in response to aggregation and retinoic acid treat-
ment. These cells are also neuron-like in their reaction with the reagents F12 A2B5,
tetanus toxin and F15.42.1. By analogy with studies on the rat optic nerve, the
absence of staining with anti-GFAP-excludes the possibility that the cells with long
processes are fibrous astrocytes (Raff et al. 1983). Further, in cell cultures derived
from human foetal brain the antibody MIN 1 stains a subpopulation of astrocytes
(Dicksone* al. 1982, 1983), and this antibody did not react with the cells bearing long
branching processes in these Tera-2 cultures. Cells with this appearance did not stain
with the 04 antibody which marks oligodendrocytes. Although this antibody has been
found to label galactocerebroside-positive cells in foetal human brain cultures (G.
Dickson, personal communication), previous studies on the mouse nervous system
exclude the possibility that the branched cells represent a type of oligodendrocyte
(Somner & Schachner, 1981).
    The undifferentiated cells did not express the morphological or antigenic charac-
teristics of neurons that we have studied. Upon aggregation and treatment with
retinoic acid, these undifferentiated cells give rise to a population in which some cells
express a neuron-like morphology and markers, which in the case of neurofilaments
are restricted to neurons. This phenotype was stable over a period of weeks. However,
it still remains to be determined whether these cells have the electrical and neurotrans-
mitter characteristics of neurons, and also whether it is possible to direct the cells
towards glial differentiation.
    The conclusions to be drawn from this study are as follows. (1) The observation
that cloned cells from the Tera-2 human teratoma line can form neuron-like cells has
been confirmed (Andrews, 1984), and we have extended the range of antigenic deter-
minants that these cells share with human neurons in culture. (2) The capacity to form
neuron-like cells can reside in an undifferentiated cell population that is characterized
by a lack of expression of HLA-A,B,C, /82-microglobulin and SSEA-3 determinants.
60                                 5. Thompson and others
The majority of the cells express Thy-1, a 90 X \$Mr protein recognized by
F10.44.2, and about a third of the cells express the SSEA-1 sugar specificities.
(3) Undifferentiated cell populations can be induced to form neuron-like cells under
controlled experimental conditions. However, there are also clones available that
'spontaneously' form these cells. (4) T h e phenotype of the undifferentiated cells of
these clones described in this paper closely resembles that of uncloned Tera-2 cul-
tures. (5) The undifferentiated cells can develop into populations of static cells in
monolayer that have a distinct, but heterogeneous morphology. They can also develop
into cell populations with neuron-like cells. They are therefore a stem cell of these
cultures and they are developmentally at least bipotential.

  These studies were generously supported by the Cancer Research Campaign, the Medical
Research Council and the Muscular Dystrophy of Great Britain Research Fund. We particularly
thank Dr Peter Andrews for introducing us to the Tera-2 system and for telling us of his unpublished
results. We are indebted to Ms L. Johnson and Dr D. W. Mason for assistance and advice with the
FACS analysis; the machine was made available by Dr A. F. Williams, MRC Cellular Immunology
Unit, Oxford. Dr Engstrom was on sabbatical leave from the Department of Tumour Pathology,
Karolinska Institutet, Stockholm, supported by travel funds from the Karolinska Institutet and
Bristol-Myers. Dr Shi was on sabbatical leave from the Shanghai Institute of Cell Biology,
Academia Sinica, China, supported by the Royal Society: Chinese Academy of Sciences exchange
scheme and the Henry Lester Fund.

ALITALO, K., KURKINEN, M., VAHERI, A., KREIG, T. & TIMPL, E. (1980). Extracellular matrix
  components synthesized by human amniotic epithelial cells in culture. Cell 19, 1053-1062.
ANDERSON, C. L., JONES, W. G. & MILFORD WARD, A. (1981). Editors of Germ Cell Tumours.
  London: Taylor & Francis.
  J., WOOD, J. N. & KAHN, J. (1982). Monoclonal antibodies show that neurofibrillary tangles and
  neurofilaments share antigenic determinants. Nature, Land. 298, 84-86.
ANDREWS, P. W. (1982). Human embryonal carcinoma cells in culture do not synthesize fibronectin
  until they differentiate. Int. J. Cancer 30, 567-571.
ANDREWS, P. W. (1983). The characteristics of cell lines derived from human germ cell tumours.
  In The Biology of Human Teratomas (ed. I. Damjanov, B. B. Knowles & D. Solter). Clifton,
  N.J.: The Human Press.
ANDREWS, P. W. (1984). Retinoic acid induces neuronal differentiation of a human embryonal
  carcinoma cell line. Devi Biol. 103, 285-293.
  comparative study of eight cell lines derived from human testicular teratocarcinoma. Int. J.
  Cancer 26, 269-280.
ANDREWS, P. W., BRONSON, D. L., WILES, M. V. & GOODFELLOW, P. N. (1981). The expression
  of MHC antigens by human teratocarcinoma-derived cell lines. Tissue Antigens 17, 493-500.
  & F0GH, J. (1984). Pluripotent embryonal carcinoma clones derived from the human teratocar-
  cinoma cell line, Tera-2; differentiation in vivo and in vitro. Lab. Invest. 50, 147-162.
ANDREWS, P. W., GOODFELLOW, P. N. & DAMJANOV, I. (1983). Human teratocarcinoma cells in
  culture. Cancer Surveys 2, 41-73.
  (1982). Cell-surface antigens of a human clonal embryonal carcinoma cell line: morphological and
  antigenic differentiation in culture. Int.jf. Cancer 29, 523-531.
AVNER, P., BONO, R., BERGER, R. & FELLOUS, M. (1981). Characterization of human teratocar-
  cinoma cell lines for their in vitro developmental properties and expression of embryonic and
  major histocompatibility locus-associated antigens. J. Immunogenet. 8, 151-162.
                          Differentiation of human teratoma cells                              61
  & ZIEGLER, A. (1978). Production of monoclonal antibodies to group A erythrocytes, HLA, and
  other human cell surface antigens: new tools for genetic analysis. Cell 14, 9-20.
BEILBY, J. O. W., HORNE, C. H. W., MILNE, G. D. & PARKINSON, C. (1979). Alpha-fetoprotein,
  alpha-1-antitrypsin, and transferrin in gonadal yolk-sac tumours. J'. din. Pathol. 32, 455—461.
BELANGER, L., HAMEL, D., DUFOUR, D., GUILLOUZO, A. & CHUI, J. F. (1978). Hormonal control
  and putative cell cycle dependency of AFP production: further observations in vivo and in vitro.
  Scand.jf. Immun. 8 (suppl. 8), 239-246.
BERNSTINE,   E. G.,   HOOPER,    H. L.,   GRANDCHAMP, S. & EPHRUSSI,         B. (1973).   Alkaline
  phosphatase activity in mouse teratoma. Proc. natn. Acad. Set. U.SA. 70, 3899-3903.
BONNER, W. M. & LASKEY, R. A. (1974). A film detection method for 3H-labelled proteins and
  nucleic acids in polyacrylamide gels. Eur.J. Biochem. 46, 83-88.
BONO, R., CARTRON, J. P., MULET, C , AVNER, P. & FELLOUS, M. (1981). Selective expression
  of blood group antigens on human teratocarcinoma cell lines. Blood Transfusion Immunohaemat.
  24, 97-106.
  Monoclonal antibodies for analysis of the HLA system. Immun. Rev. 47, 3-61.
CASPERSSON, T. (1979). Quantitative tumor cytochemistry. Cancer Res. 39, 2341-2355.
CASPERSSON, T., AUER, G., FALLENIUS, A. & KUDYNOWSKI, J. (1983). Cytochemical changes in
  the nucleus during tumor development. Histochem. jf. 15, 337-362.
CASPERSSON, T. & KUDYNOWSKI, J. (1980). Cytochemical instrumentation for cytopathological
  work. Int. Rev. exp. Path. 21, 1-54.
CASPERSSON, T . & LOMAKKA, G. (1970). Recent progress in quantitative cytochemistry: instru-
  mentation and results. In Introduction to Quantitative Cytochemistry II (ed. G. L. Weid & G. F.
  Bahr), pp. 27-56. New York: Academic Press.
Cossu, G., ANDREWS, P. W. & WARREN, L. (1983). Covalent binding of lactosaminoglycans and
  heparan sulphate to fibronectin synthesized by a human teratocarcinoma cell line. Biochem.
  biophys. Res. Commun. I l l , 952-957.
Cossu, G. & WARREN, L. (1983). Lactosaminoglycans and heparan sulphate are covalently bound
  to fibronectin synthesized by mouse teratocarcinoma stem cells. J. biol. Chem. 258, 5603-5607.
CROUCH, E., BALLIAN, G., HOLBROOK, K., DUKSIN, D. & BORNSTEIN, P. (1978). Amniotic fluid
  fibronectin. Characterisation and synthesis by cells in culture. J. Cell Biol. 78, 701-715.
DALCHAU, R., KIRKLEY, J. & FABRE, J. W. (1980). Monoclonal antibody to a human brain-
  granulocyte-T lymphocyte antigen probably homologous to the W3/13 antigen of the rat. Eur.
 J. Immun. 10, 745-749.
DAMJANOV, I. & ANDREWS, P. W. (1983). Ultrastructural differentiation of a clonal human em-
  bryonal carcinoma cell line in vitro.' Cancer Res. 43, 2190-2198.
DAMJANOV, I., Fox, N., KNOWLES, B. B., SOLTER, D., LANGE, P. H. & FRALEY, E . E . (1982).
  ImmunohistochcrrMcal localization of murine stage specific embryonic antigens in human tes-
  ticular germ cell tumors. Am.J. Path. 108, 225-230.
DICKSON, J. G., FLANIGAN, T. P. & WALSH, F. S. (1982). Cell surface antigens of human fetal
  brain and dorsal root ganglion cells in tissue culture. In Human Motor Neuron Diseases (ed. L. P.
  Rowland), pp. 435-451. New York: Raven Press.
  Identification of cell surface antigens present exclusively on a sub-population of astrocytes in
  human foetal brain cultures. .7. Neuroimmun. 5, 111-123.
DULBECCO, R. & VOGT, M. (1954). Plaque formation and isolation of pure lines with poliomyelitis
  viruses. J . exp. Med. 99, 167-182.
EISENBARTH, G. S., SHIMIZU, K., BOWRING, M. A. & WELLS, S. (1982). Expression of receptors
  for tetanus toxin and monoclonal antibody A2B5 by pancreatic islet cells. Proc. natn. Acad. Set.
  U.SA. 79, 5066-5070.
EISENBARTH, G. S., WALSHE, F. S. & NIRENBERG, M. (1979). Monoclonal antibody to a plasma
  membrane antigen of neurons. Proc. natn. Acad. sci. U.SA. 76, 4913-4917.
F0GH, J. &TREMPE, G. (1975). New human tumor cell lines. lnHuman TumorCells In Vitro (ed.
  ]. F0gh), pp. 115-159. New York: Plenum Press.
GALLIMORE, P. H. & RICHARDSON, C. R. (1973). An improved banding technique exemplified in
  the karyotype analysis of two strains of rat. Chromosoma 41, 259-263.
62                                 5. Thompson and others
GITLIN, D. & PERRICELLI, A. (1970). Synthesis of serum albumin, prealbumin, a-foetoprotein,
  ai-antitrypsin, and transferrin by human yolk sack. Nature, Lend. 228, 995-997.
GITLIN, D., PERRICELLI, A. & GITLIN, G. M. (1972). Synthesis of a-foetoprotein by liver, yolk
  sac, and gastrointestinal tract of the human conceptus. Cancer Res. 32, 979-982.
  Introduction of a human X-6 translocation chromosome into a mouse teratocarcinoma: investiga-
  tion of control of HLA-A,B,C expression. Proc. natn. Acad. Set. U.SA. 79, 1190-1194.
  E., DALCHAU, R. & FABRE, J. W. (19826). The gene, MIC4, which controls expression of the
  antigen defined by monoclonal antibody F 10-44-2, is on chromosome 11. Eur.jf. Invnun. 12,
  (1976). Expression of HLA antigens on placenta. Transplantation 22, 595-603.
Gooi, H. C , FEIZI, T., KAPADIA, A., KNOWLES, B. B., SOLTER, D. & EVANS, M . J . (1981).
  Stage specific embryonic antigen (SSEA-1) involves the 3-fucosylated type 2 precursor chain.
  Nature, bond. 292, 156-158.
HAKOMORI, S., NUDELMAN, E., LEVERY, S., SOLTER, D. & KNOWLES, B. B. (1981). The hapten
  structure of a developmentally regulated glycolipid antigen (SSEA-1) isolated from human
  erythrocytes and adenocarcinoma: a preliminary note. Biochem. biophys. Res. Commun. 100,
HILDRETH, J. E. (1982). D. Phil, thesis, Oxford.
HOGAN, B. L. M. (1980). High molecular weight extracellular proteins synthesized by endoderm
  cells derived from mouse teratocarcinoma cells and normal extraembryonic membranes. Devi
  Biol. 76, 275-285.
HOGAN, B. L. M., FELLOUS, M., AVNER, P. & JACOB, F. (1977). Isolation of a human teratocar-
  cinoma cell line which expresses F9 antigen. Nature, Land. 270, 515-518.
HOLDEN, S., BERNARD, O., ARTZT, K., WHITMORE, W. F. & BENNETT, D. (1977). Human and
  mouse embryonal carcinoma cells in culture share an embryonic antigen (F9). Nature, Land. 270,
HURKO, O. & WALSH, F. S. (1983). Human foetal muscle specific antigen is restricted to regenerat-
  ing myofibres in diseased adult muscle. Neurology 33, 737-743.
ILES, S. A., BRAMWELL, S. R., DEUSSEN, Z. A. & GRAHAM, C. F. (1975). Development of
  parthenogenetic mouse embryos in the uterus and in extra uterine sites. J'. Embryol. exp. Morph.
  34, 387-405.
JEWETT, M. A. S. (1978). Testis carcinoma: transplantation into nude mice. Natn. Cancer Inst.
  Monogr. 49, 65-66.
  B. B. & SOLTER, D. (1983a). Stage specific embryonic antigens (SSEA-3 and -4) are epitopes of
  a unique globoseries ganglioside isolated from human teratocarcinoma cells. EMBO. J. 2,
  (19836). New globoseries glycosphingolipids in human teratocarcinoma reactive with monoclonal
 antibody directed to a developmentally regulated antigen, stage specific embryonic antigen 3.
 J. biol. Chem. 258, 8934-8942.
KANNAGI, R., NUDELMAN,       E., LEVERY, R. B. & HAKOMORI, S. (1982). A series of human
  erythrocyte glycosphingolipids reacting to the monoclonal antibody directed to a developmentally
  regulated antigen, SSEA-1. J. biol. Chem. 257, 14865-14874.
KASAI, N. & Yu, R. K. (1983). The monoclonal antibody A2B5 is specific to ganglioside GQlc.
 Brain Res. 277, 155-158.
KEMSHEAD, J. T., BICKNELL, D. & GREAVES, M. F. (1981). A monoclonal antibody detecting an
  antigen shared by neural and granulocytic cells. Paedial. Res. 15, 1281—1286.
KORNBLIHTT, A. R., VIBE-PEDERSEN, K. & BARALLE, F. E. (1984). Human fibronectin: molecular
  cloning evidence for two mRNA species differing by an internal segment coding for a structural
  domain. EMBO.J. 3, 221-226.
KOULAKOFF, A., BIZZINI, B. & BERWALD-NETTER, Y. (1983). Neuronal acquisition of tetanus
  toxin binding sites: relationship with the last mitotic cycle. Devi Biol. 100, 350-357.
                          Differentiation of human teratoma cells                               63
LAEMMLI, U. K. (1970). Cleavage of structural proteins during the assembly of the bacteriophage
  T4. Nature, Land. 227, 680-685.
LASKEY, R. A. & MILLS, A. D. (197S). Quantitative film detection of 3H and H C in polyacrylamide
  gels by fluorography. Eur.jf. Biochem. 56, 335-341.
LASKEY, R. A. & MILLS, A. D. (1977). Enhanced autoradiographic detection of 32P and 12SI using
  intensifying screens and hypersensitized film. FEBS Lett. 82, 314-316.
LEBDEEN, R. W. & MELLANBY, J. (1977). Gangliosides as receptors for bacterial toxins. In Perspec-
  tives in Toxinology (ed. A. W. Bernheimer), pp. 16-42. New York: John Wiley.
MARTIN, G. R. & EVANS, M. J. (1975). Differentiation of clonal lines of teratocarcinoma cells:
  formation of embryoid bodies in vitro. Proc. natn. Acad. Sci. U.SA. 72, 1441-1447.
MCILHINNEY, R. A. J. (1981). Cell surface molecules of human teratoma cell lines. Int. J. Androl.
  suppl. 4,93-110.
MCILHINNEY, R. A. J. (1983). The biology of human germ cell tumours: experimental approaches.
  In Current Problems in Germ Cell Differentiation (ed. A. McLaren & C. C. Wylie), pp. 175-197.
  Cambridge University Press.
MCILHINNEY, R. A. J. & PATEL, S. (1983). Characterization of fibronectin synthesized by germ cell
  tumours. Cancer Res. 43, 1282-1288.
MCKENZIE, J. L., DALCHAU, R. & FABRE, J. W. (1982). Biochemical characterisation and localiza-
  tion in brain of a human brain-leucocyte membrane glycoprotein recognized by a monoclonal
  antibody. J. Neurochem. 39, 1461-1466.
MCKENZIE, J. L. & FABRE, J. W. (1981). Human Thy-1: unusual localization and possible func-
  tional significance in lymphoid tissue. J. Immun. 126, 843-850.
Mc-STOFI, F. K. & PRICE, E. B. (1973). Tumors of the Male Genital System. Washington: AFIP.
NEWLANDS, E. S. & REYNOLDS, K. W. (1983). Clinical management of malignant germ cell
  tumours. Cancer Surveys 2, 21-40.
NHRGAARD-PEDERSEN, B. (1978). Editor of Carcinoembryonic Proteins: Recent Progress. Oxford:
  Blackwell's Scientific.
OSTRAND-ROSENBERG, S., EDIDIN, M. & JEWETT, M. A. S. (1977). Human teratoma cells share
  antigens with mouse teratoma cells. Devi Biol. 61, 11-19.
PARHAM, P., BARNSTABLE, C. J. & BODMER, W. F. (1979). Use of monoclonal antibody (W6/32)
  on structural studies of HLA-A,B,C antigens. J. Immun. 123, 342-349.
PARHAM, P. & BODMER, W. F. (1978). Monoclonal antibody to a human histocompatibility allo-
  antigen, HLA-A2. Nature, Land. 276, 397-399.
PRUSS, R. M. (1979). Thy-1 antigen on astrocytes in long term culture of rat central nervous system.
  Nature, Land. 280, 688-690.
PUCH, R. C. B. (1976). Pathology of the Testis. Oxford: Blackwell's Scientific.
RAFF, M. C , ABNEY, E. R., COHEN, J., LINDSAY, R. & NOBLE, M. (1983). Two types of astrocytes
  in cultures of developing rat white matter: differences in morphology, surface gangliosides, and
  growth characteristics. 7- Neumsci. 3, 1289-1300.
RAGHAVAN, D. (1981). The application of xenografts in the study of human germ cell tumours. In
  Workshop on Early Detection    of Testicular   Cancer (ed. W. E. Skakkebaek, J . G . Berthelson,
  K. M. Grigor & J. Visfeldt), Int. J. Androl., Suppl. 4, 79-92.
RAYNER, M. J. & GRAHAM, C. F. (1982). Clonal analysis of the change ingrowth phenotype during
  embryonal carcinoma cell differentiation. J. Cell Sci. 58, 331-344.
REZNIKOFF, C. A., BERTRAM, J. S., BRANKOW, D. W. & HEIDELBERGER, C. (1973). Quantitative
  and qualitative studies of chemical transformation of cloned C3H mouse embryo cells sensitive
  to postconfluence inhibition of cell division. Cancer Res. 33, 3239-3249.
ROGERS, T. B. & SNYDER, S. H. (1981). High affinity binding of tetanus toxin to mammalian brain
  membranes. 7. biol. Chem. 256, 2402-2407.
  nectin from germ cell tumours resembles amniotic fluid fibronectin. Int.J. Cancer 17, 763-767.
SCHWARZBAUER, J . E . , TAMKUN, J.W., LEMISCHKA, I.R. & HYNES, .R.O. (1983). Three dif-
  ferent fibronectin mRNA's arise by alternative splicing within the coding region. Cell 35,
SHEVINSKY, L. H., KNOWLES, B. B., DAMJANOV, I. & SOLTER, D. (1982). Monoclonal antibody
  to murine embryos defines a stage-specific embryonic antigen expressed on mouse embryos and
  human teratocarcinoma cells. Cell 30, 697-705.
64                                 S. Thompson and others
SHI, W-K., HOPKINS, B., THOMPSON, S., HEATH, J . K . , LUKE, B.M. & GRAHAM, C. F. (1984).
  Synthesis of apolipoproteins, alphafoetoprotein, and transferrin, by human foetal yolk sack and
  other foetal organs. J. Embryol. exp. Morph. (in press).
SOLTER, D. & KNOWLES, B. B. (1978). Monoclonal antibody defining a stage-specific mouse
  embryonic antigen (SSEA-1). Proc. natn.Acad. Set. U.SA. 75, 5565-5569.
SOMNER, I. & SCHACHNER, M. (1981). Monoclonal antibodies (01 to 04) to oligodendrocyte cell
  surface: an immunocytological tool in the central nervous system. Devi Biol. 83, 311-327.
STANNERS, E. P., ELICEIRI, G. & GREEN, H. (1971). Two types of ribosome in mouse-hamster
  hybrid cells. Nature, new Biol. 230, 52-54.
STERN, P. L. (1983). Serological and cell-mediated immune recognition of teratocarcinoma. In
  Current Problems in Germ Cell Differentiation (ed. A. McLaren & C. C. Wylie), pp. 157-173.
  Cambridge University Press.
STERN, P. L. (1984). Differentiation antigens of embryonal carcinoma cells and embryos. Br. med.
  Bull. 40, 218-223.
STERN, P., GILBERT, P., STERNBERG, S., THOMPSON, S. & CHADA, K. (1984). A monoclonal
  antibody which detects a 125 kDa glycoprotein on embryonal carcinoma cells and is mitogenic for
  murine spleen cells. J . Reprod. Immun. 6, 313-328.
THOMPSON, S. & MADDY, A. H. (1982). Gel electrophoresis of erythrocyte membrane proteins.
  In Red Cell Membranes:   a Methodological Approach (ed. J . D . Young & J . C. Ellory), p p . 6 7 - 9 3 .
  New York: Academic Press.
TROWSDALE, J., TRAVERS, P., BODMER, W. F. & PATTILLO, R. F. (1980). Expression of HLA-A,
  -B, and -C and /fe-microglobulin antigens in human choriocarcinoma cell lines. J. exp. Med. 152,
VAN HEYNINGEN, W. E. (1963). The fixation of tetanus toxin, strychnine, serotonin, and other
  substances by gangliosides. J. gen. Microbiol. 31, 375-387.
WALSH, F. S., MOORE, S. E. & DHUT, S. (1981). Monoclonal antibody to human fibronectin:
  production and characterization using human muscle cultures. Devi Biol. 84, 121-132.
 SON, J. G. (1983). Characterisation of human muscle differentiation antigens. Exp. biol. Med. 9,
WALSH, F. S. & RITTER, M. A. (1981). Surface antigen differentiation during human myogenesis
 in culture. Nature, Land. 289, 60-64.
WILLIAMS, A. F., GALFRE, G. & MILSTEIN, C. (1977). Analysis of cell surfaces by xenogeneic
  myeloma-hybrid antibodies: differentiation antigens of rat lymphocytes. Cell 12, 663-673.
WILLIAMS, L. K., SULLJVAN, A., MCILHINNEY, R. A. J. & NEVILLE, A. M. (1982). A monoclonal
  antibody marker of human primitive endoderm. Int.J. Cancer 30, 731-738.
WOLFE, J., MAUTNER, V., HOGAN, B. L. M. & TILLY, R. (1979). Synthesis and retention of
  fibronectin (LETS protein) by mouse teratocarcinoma cells. Expl Cell Res. 118, 63-71.

                                          (Received 14 May 1984 -Accepted 8 June 1984)

To top