Bull. 1842, in press.
15. We thank C. Davis and E. Harrison for their
assistance with data collection and analysis. E. Harp
and J. Walder provided thoughtful critiques for the
manuscript. The NRCDP rainfall and landslide sim-
ulator was used with the help and collaboration of
T. Fukuzono, M. Tominaga, H. Moriwaki and N.
Oyagi, as part of the NRCDP project on snow
avalanche management and landslide prediction and
control. Our work was supported by a United
States-Japan cooperative program on research and
development in science and technology.
29 August 1989; accepted 26 September 1989
Germ-Line Transmission of a c-abl Mutation
Produced by Targeted Gene Disruption in ES Cells
PAMELA L. SCHWARTZBERG, STEPHEN P. GOFF,
ELIZABETH J. ROBERTSON
A substitution mutation has been introduced into the c-abl locus ofmurine embryonic
stem cells by homologous recombination between exogenously added DNA and the
endogenous gene, and these cells have been used to generate chimeric mice. It is shown
that the c-abl mutation was transmitted to progeny by several male chimeras. This
work demonstrates the feasibility ofgerm-line transmission ofa mutation introduced
into a nonselectable autosomal gene by homologous recombination.
HE INTRODUCTION OF MUTATIONS
into the germ line of an organism is
one of the most powerful genetic
methods for determining the functions of a
specific gene product (1). Recent advances
in the detection of rare homologous recom-
bination events have facilitated the modifi-
cation ofdefined chromosomal loci in mam-
malian cell lines (2-8). The use of these
techniques in combination with cultured
embryonic stem (ES) cells should now allow
the replacement of normal cellular genes in
the mouse germ line by mutant alleles with
defined sequence alterations (9). ES cells are
pluripotent cells derived from preimplanta-
tion mouse embryos (10), which can be
propagated in culture and subsequently re-
introduced into mouse blastocysts by micro-
injection to form chimeric mice. Such chi-
meras, if constructed with euploid ES cells,
have high rates oftransmission ofthe ES cell
component in the germ line (11). To date,
however, only mutations at the X-linked
locus encoding the enzyme hypoxanthine-
(HGPRT), for which there are genetic selec-
tions, have been successfully transferred into
the mouse germ line by this strategy (12,
We are interested in the function of v-abl,
the oncogene carried by the Abelson murine
leukemia virus (A-MuLV), and its cellular
homolog c-abl (14). A-MuLV causes the
P. L. Schwartzberg and S. P. Goff, Department of
Biochemistrv and Molecular Biophvsics, Columbia Uni-
versity, College of Phvsicians & Surgeons, New York,
E. J. Robertson, Department of Genetics and Develop-
ment, Columbia University, College of Physicians &
Surgeons, New York, NY 10032.
10 NOVEMBER I989
rapid induction of lymphosarcoma in sus-
ceptible mice and can transform both fibro-
blasts and lymphocytes in culture (15, 16).
The human c-abl has been implicated in at
least two forms of cancer, chronic myeloge-
nous leukemia and acute lymphocytic leuke-
mia, where the gene is activated by chromo-
somal translocation (17). Although much is
known about the oncogenic potential of
both v-abl and c-abl, little is known about the
function ofthe normal gene in development
or in the life ofthe adult organism. The c-abl
gene is transcribed in most tissues to give
rise to at least two major mRNAs found in
approximately equal abundance (18). Post-
meiotic spermatids have been shown to ex-
press very high levels of a distinctive c-abl
mRNA (19, 20) that is truncated in the 3'
untranslated region by polyadenylation at a
novel site (21). The roles ofthe c-abl protein
(Abl) in these cells and in the rest of the
Fig. 1. Scheme for the replacement
of the normal c-abl gene with a c-
abl-neo fusion by homologous re-
combination. A linear DNA con-
taining neo embedded in c-abl se-
quences, but devoid of signals for
transcription and translation, is in-
troduced by electroporation (27). A
double crossover in the flanking c-
abl sequences replaces the normal
gene with the fusion gene and acti-
vates expression of the neo. After
digestion ofDNA from drug-resist-
ant clones with Apa LI and Xba I,
probe EX (28) detects DNA frag-
ments of novel sizes from the mu-
tant allele, as well as fragments of
the normal size from the unaltered
allele. A, Apa LI; X, Xba I.
membership in the class of tyrosine-specific
kinases suggests that it may be involved in
signal transduction. The generation of de-
fined mutations in c-abl in the mammalian
germ line could provide insights into the
function of the gene product.
We have used homologous recombina-
tion to introduce a substitution mutation
into the c-abl locus of mouse ES cells. We
chose to introduce mutations affecting only
the COOH-terminal third of the Abl pro-
tein, downstream from the tyrosine kinase
domain and nuclear targeting sequences
(22). Since c-abl is expressed ubiquitously,
we were concerned that introduction of a
null mutation would have severe deleterious
effects on the development ofthe mouse, or
might even be lethal, at a very early stage.
Mutations limited to this region might not
represent null mutations and might generate
a less severe phenotype. Deletions affecting
the COOH-terminus of v-abl have shown
that this domain is not needed for tyrosine
kinase activity or for the transformation of
fibroblasts but is important for transforma-
tion oflymphocytes. These deletion mutants
also exhibit a reduction in the toxicity asso-
ciated with high-level expression of the
wild-type viral oncogene in certain cell lines
(23). The tissue specificity of the effects of
these mutations in A-MuLV suggested that
we might obtain informative tissue-specific
phenotypes from similar mutations in c-abl.
To select for the rare homologous recom-
bination ofDNA with the endogenous c-abl
locus, we designed
pAbXRl, in which a promoterless neomy-
cin-resistance gene (neor) is fused to c-abl
genomic sequences (24); this DNA would
confer resistance to the drug G418 only
after certain recombination events. Expres-
sion ofneo in the construct could be activat-
ed either when a nonhomologous integra-
tion event places the sequence next to an
arbitrary cellular promoter, or, alternatively,
when homologous recombination inserts
Select for G418R
l Test for rearrangement
on December 4, 2008
the DNA into the target locus (4, 5). A
requirement of the procedure is that the c-
abl locus be expressed in the target cell line;
we have shown by Northern (RNA) blots
that the CCE ES cell line used in this study
(25) does indeed express the two normal c-
abl mRNAs (26). The plasmid containing
the c-abl-neo fusion was digested with Xba I,
the DNAs were introduced into CCE ES
cells by electroporation, and the cells were
plated into medium containing G418 (27).
Drug-resistant colonies were isolated, and
DNA preparations from these cell lines were
examined for rearrangements by Southern
Genomic DNAs from G418r clones were
digested with Apa LI, separated by electro-
phoresis, transferred to nitrocellulose, and
hybridized with a labeled probe (EX) ho-
mologous to a region of c-abl outside the
introduced DNA fragment (28). Integration
ofthe neo sequences by homologous recom-
bination should convert one copy ofthe 6.5-
kb Apa LI wild-type fragment detected by
this probe to a novel 7.1-kb fragment (Fig.
1). We identified seven independent (single
copy) homologous integration events out of
239 colonies screened, giving a frequency of
1 in 34 G418r clones. Examination of the
genomic DNA from five of the clones by
digestion with other restriction enzymes
demonstrated that, in four of the five, the
alteration of restriction pattern was as ex-
pected for a simple substitution of the c-abl
region by the added DNA (Fig. 2). Details
of the generation of these lines will be
presented elsewhere (29).
The CCE cell line used for these experi-
ii .4. kbE
Fig. 2. Southern blot analysis of genomic DNA
from individual G418' cell lines. DNA prepara-
tions from various clones were digested and ana-
lyzed by Southern blotting. (A) DNAs were
digested with Apa LI and probed with the Abel-
son-specific EX probe (28). Left lane, a nonho-
mologous transformant; right lane, clone 2bl.
The bands derived from the parental c-abl allele
(6.5 kb) and from the rearranged allele (7.1 kb)
are indicated by arrows. (B) DNAs were digested
with the indicated enzyme and hybridized with
nick-translated pAb3sub3 probe (14). Left lanes,
CCE control line; right lanes, clone 2bl.
Table 1. Rate ofgeneration of overt chimeras with the 2bl cell line.
*Of these progeny, 91 survived to an age to pernit scoring for coat color chimerism.
ments was originally derived from a single
XY blastocyst of the 129/Sv//Ev strain (25).
The strain is homozygous at the black (b)
and agouti (a) loci. Additionally, the strain
is homozygous for the GPI-lC allele, encod-
ing a rare electrophoretic variant of the
glucose phosphate-isomerase enzyme. This
line was chosen because it has been shown
reproducibly to colonize the germ line of
male chimeras with high efficiency (13, 25).
For the generation ofchimeras, host blasto-
cysts were obtained from CD-1, MF1, and
C57B16 mouse strains. The CD-1 and MF1
outbred strains are both albino (homozy-
gous for the c allele), whereas the C57B1/6
inbred strain is black, nonagouti (BB aa).
Thus, chimeric mice could be scored by the
presence ofpigmented coat hair in the CD-1
and MFi genetic background, and by the
presence ofagouti hair in the C57B1/6 back-
Six clonally derived cell lines, each con-
taining one disrupted allele of c-abl, were
initially tested for their ability to generate
chimeric mice (30). For each line, approxi-
mately 10 to 15 cells were injected into 3.5-
day-old blastocysts derived from CD-1 mat-
ings. These were then transferred into pseu-
dopregnant females and allowed to develop
to term. A single cell line, termed 2bl, that
gave the best rate and extent ofchimerism in
the resulting live-born animals was chosen
for more extensive analysis. A G-banding
analysis ofmetaphase chromosomes showed
that 2bl was predominantly euploid XY
(25/32 spreads scored; 78%). It was also
noted on separate occasions that a consist-
ently higher proportion of cells than normal
(approximately 20%) were tetraploid. When
the line was karyotyped after ten passage
generations (approximately 35 doublings)
after transformation, a significant propor-
tion (10%) of the cells had acquired a
specific chromosomal translocation (6:14
Data on the rate ofchimera formation by
the 2bl line, in combination with host
blastocysts from different strains, are pre-
sented in Table 1. The overall frequency of
formation of chimeric animals was approxi-
mately equal in all backgrounds (32 to 52%
of live-born animals examined). The degree
of ES cell colonization to the coat was
markedly influenced by the genetic back-
ground of the host blastocyst (Fig. 3, A to
C). We consistently found that the ES cells
contributed very extensively in combination
with C57B1/6 blastocysts. In the CD-1 back-
ground coat pigmentation was less pro-
nounced. This poor contribution in CD-1
mice could not be attributed to the use of
outbred albino recipients; injections of 2bl
cells into blastocysts ofMF1 mice, another
outbred albino strain, yielded better chime-
ras, intermediate between the CD-1 and
C57B116 chimeras (Fig. 3C). We conclude
that the genetic background provided by the
CD-1 outbred strain does not favor the
incorporation of 129-derived ES cells.
Since the CCE cell line was derived from a
male and is ofXY genotype, one indicator of
good chimera formation is a distortion of
the normal ratio of male to female animals
among the chimeras born. This is because of
the high localized incorporation of Y-bear-
ing cells into the genital ridge, which in-
duces the development of a male reproduc-
tive system. There was only slight sex distor-
tion among chimeras derived from the 2bl
line in any strain background.
To screen for germ-line transmission, we
caged phenotypically male chimeras with
tester females. The CD-1- and MFl-based
chimeras were mated to albino females,
whereas the C57Bl/6-based animals were
mated to nonagouti [genotype (C57B1/6 x
DBA/2) F1] females. From these test mat-
ings, the litters were inspected for progeny
carrying ES cell-derived agouti pigmenta-
tion. Glucose phosphate isomerase analysis
of peripheral blood samples was used to
verify that the agouti progeny were derived
from the CCE cells (25).
Of the first 17 male C57Bl/6-based chi-
meric mice that were proven to be fertile, we
have obtained six which have produced
agouti progeny. These six chimeras have to
date sired a total of 17 agouti offspring in a
total of 13 litters (Table 2; examples shown
in Fig. 3). In contrast, none of the I1 CD-
i-based male chimeras that have been tested
has produced any pigmented offspring
among the 80 to 180 offspring produced per
chimera (Table 2). Similarly, none of the
MFl-based chimeras of the first nine tested
has yet produced pigmented offspring in the
SCIENCE, VOL. 246
on December 4, 2008
Fig. 3. Examples of chimeric mice and progeny.
(A to C) Chimeras derived with clone 2bl inject-
ed into host blastocysts of various genetic back-
grounds. (A) CD-1. (B) C57B1/6. (C) MFi. (D)
Members of a family in which the c-abl-neo allele
has been transmitted in the germ line. The adult
brown mouse is male chimera C, and the adult
DBA)F 1. Two oftheir offspring shown are agou-
ti, indicating transmission ofthe ES cell markers.
Southern blot analysis revealed that one of these
two inherited the c-abl-neo allele (Fig. 4).
is a tester female (C57BI/6 x
first two litters.
Southern analysis of DNA from the first
ten agouti progeny of the chimeric males
designated "C," "J," and "O" has demon-
strated that six carry the c-abl mutation
originally present in the 2bl cell line (Fig.
4). Thus, we have successfully introduced
the original mutant allele into the mouse
germ line; this allele has been named c-ablmI.
These mice, heterozygous for the mutant c-
abl allele, are phenotypically normal at 10
weeks ofage. None have shiown the appear-
ance of palpable tumors or lymphomas.
Our use of promoterless selectable mark-
ers as a means of selecting for rare homolo-
gous recombinants has several advantages.
In the case of c-abl, we found that 1 in 34
drug-resistant clones had undergone a ho-
mologous recombination event. This fre-
quency is sufficiently high to allow the direct
and rapid screening of DNAs from cell
clones by Southern blotting. If we assume
an approximately 100-fold enrichment with
this technique, the ratio of unselected ho-
mologous to nonhomologous integration
events is approximately 1/3400, comparable
to frequencies observed at other endoge-
nous chromosomal loci (2, 6, 7, 31, 32) and
at foreign, exogenously added sequences (3-
IO NOVEMBER I989
5). The various efficiencies of homologous
recombination at different target loci could
depend on the level of expression of the
target gene, but other factors may be impor-
tant (2, 6).
Another advantage to the method of se-
lection we have used is that it only relies on a
single selection in the antibiotic G418, and
does not require subsequent selection or
screening steps. Previous studies have dem-
onstrated that G418 selection alone does
not abolish the ability ofES cells to generate
germ-line chimeras (25, 33). We feel that it
is important that the procedure minimizes
the number ofpassages required before cells
are used to generate chimeras. Even with a
single selection scheme, there is considerable
variability in the quality of independent
clones. The culture history and growth char-
acteristics of the ES cells are likely to be
major factors in good chimera formation.
The 2bl cell line consistently gave chimeras
with an extremely high contribution both to
somatic and germ cell lineages. Other mu-
tant clones tested, however, produced sig-
nificantly fewer and poorer chimeras. It is
unclear whether these differences between
ES clones are due to exposure to the selec-
tive drug, or to heterogeneity in the starting
ES cell population that is resolved in the
cloning to yield cell lines with a spectrum of
different potentials to differentiate. In either
case, it may be important to test several
independent clones for good chimera forma-
Although the selection for activation of
the drug-resistance marker is a powerful
method of enrichment for mutant cells, an
obvious limitation to the strategy is that it is
likely to be applicable only to genes ex-
pressed in ES cells. Other selection methods
that have been recently described, such as
"positive-negative selection" (6, 34) and the
use ofpolymerase chain reaction for screen-
ing (7), may permit targeting of any gene
sequence for homologous recombination,
whether or not it is expressed. There may be
unknown difficulties, however, with the
generation of chimeras with cells derived
from some of these procedures. Multiple
drug selections may further compromise the
ability of the cells to colonize the embryo;
similarly, the extended passage of the cells
required while the targeted cells are being
retrieved from large pools may adversely
affect the pluripotency of the cells.
When we used the mutant ES cell clones
to generate chimeric mice, we found that the
use of recipient blastocysts from different
mouse strains had a significant effect on the
degree of contribution to the chimeras. In
particular, when CCE cells and clones de-
rived from them were introduced into blas-
tocysts derived from C57B1/6 mice, many
chimeras were obtained that showed a great-
er than 95% contribution to the coat by the
ES cells (35). In contrast, chimeras generat-
ed in the CD-1 background had much lower
levels ofcontribution to their coat hairs. We
are scoring contributions to two different
tissues in these different mouse strains (36).
In the CD-1 albino outbred mice, chime-
rism is scored by the presence ofpigmented
coat hair and eyes resulting from ES contri-
bution to melanocytes, derived from neural
crest. In the C57B1/6 chimeras, chimerism is
scored by the presence of agouti coat hairs
resulting from colonization ofthe hair folli-
cle, derived from both mesoderm and ecto-
derm. The different scoring method, howev-
er, does not account for the difference be-
tween the two strains since, in control ex-
periments, GPI analysis of the internal
organs of chimeras generated from the pa-
rental CCE cells indicated a strong correla-
tion between overt chimerism and ES cell
colonization ofthe somatic tissues (35). We
Table 2. Breeding data from fertile male chime-
on December 4, 2008
analysis of DNAs from
two of the progeny of
chimera J. DNAs were
isolated from tail biop-
sies oftwo agouti-proge-
ny mice (41) and from
control cell lnes, digest-
ed withApa LI, and ana-
lyzed on Southem blots
pSV2neo probe. Lane 1,
4. Southem blot
* 7.1 kb
f g gg
CCE control line; lane 2, clone 2bl, containing
the mutant allele; lane 3, mouse J2.1; lane 4,
therefore believe that the genotypes of the
recipient blastocysts can profoundly influ-
ence the overall incorporation of ES cells in
the developing conceptus. Similar conclu-
sions were reached in an analysis ofaggrega-
tion chimeras made between the 129 and
C57B1/6 strains; the 129 component was
found to predominate in all somatic tissues
The differences among strains in the rela-
tive efficiency of ES cell contribution to
somatic tissues is also reflected in efficiency
of contribution to the germ line. For exam-
ple, although large numbers of germ-line
chimeras have been obtained with 129-de-
rived ES cells in an MF1 background (i1,
25) and in a C57B1/6 background (33), we
and others have only been able to generate
control germ-line chimeras in the CD-1
background at a very reduced efficiency
compared to other strains (35, 38). The
dependence on host strain may be enhanced
when genetically modified ES clones are
used; the choice of host can be the crucial
transmission. This conclusion is supported
by our test-breeding experiments reported
here: whereas none of the 11 chimeras
generated in the CD-1 background and
none of 9 in the MF1 background have
transmitted ES cell markers to their off-
spring, 6 of the first 17 chimeras tested in
the C57B1/6 background have produced
The c-abl gene has been shown to express
a haploid germ cell-specific message, and
the protein product of the gene has been
demonstrated to be present in cells at late
stages of spermatogenesis (20, 39). Howev-
er, we have observed normal rates of trans-
mission ofthe mutant allele in the passage of
the gene through the germ line ofchimeras.
It should be possible to introduce more
severe, null mutations that will abolish the
kinase activity of c-abl, as well as mutations
that specifically prevent formation ofone or
the other of the alternate forms of c-abl. In
mice and humans, c-abl mRNA species with
at least two and perhaps four alternative 5'
first exons have been detected (40). Since
in obtaining successful germ-line
these exons encode alternate NH2-termini of
the protein, it would be of interest to elimi-
nate each of the two most frequently used
first exons and to determine what roles the
different c-abl proteins have in the mouse.
These experiments demonstrate that a
mutation can be introduced into an endoge-
nous nonselectable gene in ES cells by ho-
mologous recombination and that the muta-
tion can be transmitted through the germ
line of the resultant chimeric mice. The
results suggest that it will be possible to
generate mutations in mice at any locus
defined by a cloned DNA sequence. Using
these techniques, we have introduced a mu-
tation that abolishes the COOH-terminal
portion of the c-abl gene product. Mice
heterozygous for the mutation show no
growth or developmental defects, demon-
strating that this allele does not act in a
dominant fashion. Analysis of homozygous
c-ablml/c-ablm' mice should ultimately reveal
much about the function of the normal
REFERENCES AND NOTES
1. D. Botstein and G. R. Fink, Science 240, 1439
2. 0. Smithies, R. G. Gregg, S. S. Boggs, M. A.
Koralewski, R. S. Kucherlapati, Nature 317, 230
(1985); K. R. Thomas and M. R. Capecchi, Cell 51,
503 (1987); T. Doetschman et al., Nature 330, 576
3. F.-L. Lin, K. Sperle, N. Stemberg, Proc. Natl. Acad.
Sci. U.S.A. 82, 1391 (1985); K. R. Thomas, K. R.
Folger, M. R. Capecchi, Cell 44, 419 (1986).
4. M. Jasin and P. Berg, Genes Dev. 2, 1353 (1988).
5. J. M. Sedivy and P. A. Sharp, Proc. Natl. Acad. Sci.
U.S.A. 86, 227 (1989).
6. S. L. Mansour, K. R. Thomas, M. R. Capecchi,
Nature 336, 348 (1988).
7. A. L. Joyner, W. C. Skames, J. Rossant, ibid. 338,
153 (1989); H.-X. Kim and 0. Smithies, Nudeic
Acids Res. 16, 8887 (1988); A. Zimmer and P.
Gruss, Nature 338, 150 (1989).
8. M. R. Capecchi, Science 244, 1288 (1989).
10. M. J. Evans and M. H. Kaufman, Nature 292, 154
(1981); G. Martin, Proc. Natl. Acad. Sci. U.S.A. 78,
11. A. Bradley, M. Evans, M. H. Kaufman, E. Robert-
son, Nature 309, 255 (1984).
12. M. L. Hooper, K. Hardy, A. Handyside, S. Hunter,
M. Monk, Nature 326, 292 (1987); S. Thompson,
A. R. Clarke, A. M. Pow, M. L. Hooper, D. W.
Melton, Cell 56, 313 (1989).
13. M. R. Kuehn, A. Bradley, E. J. Robertson, M. J.
Evans, Nature 326, 295 (1987).
14. S. P. Goff, E. Gilboa, 0. N. Witte, D. Baltimore,
Cell 22, 777 (1980); J. Y. J. Wang et al., ibid. 36,
15. H. T. Abelson and L. S. Rabstein, Cancer Res. 30,
2208 (1970); ibid., p. 2213.
16. N. Rosenberg and 0. N. Witte, Advances in Virus
Research (Academic Press, San Diego, CA, 1988),
vol. 35, pp. 39-81.
17. A. de Klein et al., Nature 300, 765 (1982); N.
Heisterkamp et al., ibid. 306,239 (1983); J. Erikson
etal., Proc. Natl. Acad. Sci. U.S.A. 83, 1807(1986);
A. Hermans et al., Cell 51, 33 (1987).
18. J. Y. J. Wang and D. Baltimore, Mol. Cell. Biol. 3,
773 (1983); M. W. Renshaw, M. A. Capozza, J. Y.
J. Wang, ibid. 8, 4547 (1988).
19. R. Muller, D. J. Slamon, J. M. Trembley, M. J.
Cline, I. M. Verma, Nature 299, 640 (1982).
20. C. Ponzetto and D. J. Wolgemuth, Mol. Cell Biol. 5,
21. S. Meijer et al., EMBOJ. 6, 4041 (1987); C. Oppi,
, Trends Genet. 5, 70 (1989).
S. K. Shore, E. P. Reddy, Proc. Natl. Acad. Sci.
U.S.A. 84, 8200 (1987).
22. R. A. Van Etten, P. Jackson, D. Baltimore, Cell 58,
23. S. P. Goff, 0. N. Witte, E. Gilboa, N. Rosenberg,
D. Baltimore, J.
Watanabe and 0. N. Witte, J.
(1983); R. Prywes, J. G. Foulkes, N. Rosenberg, D.
Baltimore, Cell 34, 569 (1983); S. F. Zeigler, C. A.
Whitlock, S. P. Goff, D. Baltimore, Cell 27, 477
24. Cloning manipulations were carried out by standard
methods [T. Maniatis, E. F. Fritsch, J. Sambrook,
Molecular Cloning: A Laboratory Manual (Cold Spring
Harbor Laboratory, Cold Spring Harbor, NY,
1982); B. Vogelstein and D. Gillespie, Proc. Natl.
Acad. Sci. U.S.A. 76, 615 (1979)]. Plasmids used
included: p16-1, a linker insertion mutant of v-abl
plasmid pTabl [R. W. Rees-Jones and S. P. Goff, J.
Virol. 62, 978 (1988)]; pXVX, a derivative of
pRSVneo containing multiple cloning sites in front
ofthe second amino acid ofneo and a Xho I site after
the polyadenylation signal (gift of G. Gaitanaris);
Bluescript KS(-) (Stratagene); and pIB131 (IBI).
The 1-kb Xho I-Sma I neo-containing fragment of
pXVX was subcloned into pIBI-31, excised with
Xho I plus Eco RI, and joined to the large Xho I-
Eco RI fragment of p16-1, to form the v-abl-neo
fusion plasmid pVX16-lR. pAbXbl and pAbXb2
were constructed by subcloning, respectively, the
7.5-kb and the upstream 2-kb Xba I fragments from
phage Xabl2 (14) into Bluescript KS(-). Plasmid
pAbXR1 was constructed with the Xho I-Sal I piece
containing the neo insert from pVX16-lR to replace
the corresponding fragment of pAbXbl.
25. E. Robertson, A. Bradley, M. Kuehn, M. Evans,
Nature 323, 445 (1986).
26. P. L. Schwartzberg and S. P. Goff, unpublished
27. CCE ES cells were maintained on mitomycin-treat-
ed STO feeder layers, as previously described [E. J.
Robertson, in Teratocarcinomas anid Embryonic Stem
Cells: A Practical Approach, E. J. Robertson, Ed.
(IRL Press, Oxford, 1987), pp. 71-112]. Cells used
for electroporations were at passage 9 and trans-
formed [G. Chu, H. Hayakawa, P. Berg, Nudeic
Acids Res. 15, 1311 (1987)] in an electroporator
(Anderson Electronics) under the following condi-
tions. Cells were plated at 3 x 106 cells per 10-cm
dish, fed on the following 2 days, and trypsinized 2
hours after the second feeding. Trypsinized cells
were washed once with phosphate-buffered saline
(PBS) and resuspended in PBS at a concentration of
4 x 107 cells per milliliter. The cell suspension (0.5
ml) was mixed in the electroporation cuvette with
10 to 20 Lg of plasmid pAbXR1 DNA digested
with Xba I, and electroporated at 200 V and 1000
LF. Cells were held for 10 min at room temperature
after electroporation, and 5 x 106 cells were plated
onto 10-cm dishes containing feeder layers. Selec-
tions were carried out in either oftwo ways. In some
experiments cells were fed fresh media the next day
and refed with media containing 400 ,ug of G418
per milliliter 1.5 days after electroporation. In other
experiments cells were refed at both 1 and 2 days
after electroporation and then passed onto fresh
feeder layers in media containing G418. Cells were
refed with media containing G418 every 2 days.
Colonies were picked into 24-well cloning trays 9 to
12 days after plating into selective media, and the
G418 concentration was lowered to 100 Rg/mn.
Cells from the 24-well trays were platedonto two 6-
cm dishes in nonselective media. Portions of these
cells were frozen and used to make high molecular
weight DNA. After screening, selected clones were
thawed, grown for two passages, and refrozen.
28. Probe EX was prepared by isolating the approxi-
mately 160-bp Eco 0109-Xba I fragmentofplasmid
pAbXb2 and labeled by extension ofhexanucleotide
primers (Pharmacia) with Klenow DNA polymerase
in the presence of cs-P32 dCTP (3000 Ci/mmole;
Amersham) and the three other unlabeled triphos-
phates [A. P. Feinberg and B. Vogelstein, Anal.
Biochem. 132,6 (1983)].
29. P. L. Schwartzberg, E. J. Robertson, S. P. Goff,
30. C57BI/6 (Jackson Labs), C57B1/6 (Charles River),
Virol. 38, 460 (1981); S. M.
Virol. 45, 1028
on December 4, 2008
CD-1 (Charles River), and MFI (Harlan Sprague-
Dawley) females were used as the source of blasto-
cysts. (C57BI/6 x CBA) Fl, (C57BI/6 x DBA) F,
(Charles River), or CD-1 females were used for
foster mothers. Cells for injection were thawed and
maintained in culture for no more than six addition-
al passages. Approximately 12 to 15 cells were
injected into blastocysts collected 3.5 days post
coitus, as previously described (7) [A. Bradley, in
Teratocarcinomas atnd Embryonic Stetn Cells: A Practical
Approach, E. J. Robertson, Ed. (IRL Press, Oxford,
1987), pp. 113-151]. Microinjected blastocysts
were introduced into the uterine horns of pseudo-
pregnant mice 2.5 days post coitus. Chimerism was
scored by coat and eye pigmentation in the CD-1
and MFL albino background and by the presence of
agouti coat color in the C57BI/6 background.
31. T. Doetschman, N. Maeda, 0. Smithies, Proc. Natl.
Acad. Sci. U.S.A. 85, 8583 (1988); G. Gaitanaris et
al., in preparation.
32. J. Charron and L. Jeannotte, unpublished data.
33. A. Gossler, T. Doetschman, R. Korn, E. Serfling, R.
Kemler, Proc. Natl. Acad. Sci. U.S. A. 83, 9065
34. R. S. Johnson et al., Science 245, 1234 (1989).
35. E. J. Robertson, S. P. Goff, R. L. Schwartzberg, J.
Charron, L. Jeanotte, manuscript in preparation.
36. W. K. Silvers, The Coat Colors ofMice: A Model for
Mammalian Gene Action and Interaction (Springer-
Verlag, New York, 1979).
migration of cells along soluble gra-
dients of chemical substances. In
mammals, several cell types are able to mi-
grate in a chemotactic manner, especially
cells from the hematopoietic system, for
IS THE DIRECTIONAL
C. Dargemont, D. Dunon, M.-A. Deugnier, M. De-
noyelle, J.-M. Girault, J. P. Thiery, B. A. Imhof, Labora-
toire de Physiopathologie duDeveloppementCNRS and
Ecole Normale Superieure, 46, rue d Ulm, 75230 Paris
Cedex 05, France.
F. Lederer and K. H. D. Le, INSERM U25, Immunolo-
gi Clinique, Hopital Necker, 161, rue de Sevres, 75743
Paris Cedex 15, France.
F. Godeau, Laboratoire de Biologie Moleculaire du Gene
INSERM U277, Institut Pasteur, 25, rue du Docteur
Roux, 75724 Paris Cedex 15, France.
*Present address: Basel Institute for Immunology, Gren-
zacherstrasse 487, CH-4005 Basel, Switzerland.
37. A. C. Peterson, P. M. Friar, H. R. Rayburn, D.
Cross, Soc. Neurosci. Symp. 4, 258 (1979).
38. Y. Suda, M. Suzuki, Y. Ikawa, S. Aizawa, J. Cell.
Phys. 133, 197 (1987).
39. C. Ponzetto, A. G. Wadewitz, A. M. Pendergast, O.
N. Witte, D.
40. Y. Ben-Neriah, A. Bemards, M. Paskind, G. Q.
Daley, D. Baltimore, Cell 44, 577 (1986); A.
Bernards, M. Paskind, D. Baltimore, Otncogetne 2,
297 (1988); E. Shtivelman, B. Liffshitz, R. P. Gale,
B. A. Roe, E. Canaani, Cell 47, 277 (1986).
41. B. Hogan, F. Costantini, E. Lacy, Manipulating the
Mouse Embryo: A Laboratory Manual (Cold Spring
Harbor Laboratory, Cold Spring Harbor, NY, 1986).
42. Supported by PHS grant P01 CA 23767 from the
National Cancer Institute to S.P.G. and grant R01
HD 25208 from Child Health and Human Devel-
opment to E.J.R. Additional funds for the animal
facility and for equipment were provided by the
Columbia University Comprehensive Cancer Cen-
ter. We thank G. Gaitanaris and R. Rees-Jones for
plasmids and members of the laboratories of F. Alt
and F. Costantini for equipment and support. Spe-
cial thanks are due to J. Charron for guidance and
advice and to the other members of the Robertson
and Goff laboratories for helpful support, especially
J. Wolgemuth, Oncogene 4, 685
11 July 1989; accepted 6 October 1989
monocytes, and mature T lymphocytes (2).
It has been suggested (3) that chemotaxis is
involved in directing the migration ofhema-
topoietic precursors from their site of emer-
gence, the bone marrow, to the thymus.
Indeed, we showed that avian thymic pep-
tides attract T cell precursors from quail
bone marrow in vitro (4). In vivo, migration
ofhematopoietic precursors into the thymus
is a prerequisite for T cell differentiation,
which requires the influence of the thymic
epithelium and thymic accessory cells (5).
We also showed that secretion products of
rat thymic epithelial cells induce the migra-
tion of the hematopoietic precursors from
rat (6) and mouse (7) bone marrow. This
chemotactic activity was due to an 11-kD
protein called thymotaxin (6). In Boyden
migration chambers, thymotaxin selectively
attracted immature lymphoid cells, devoid
ofmature T and B cell differentiation mark-
ers. These nonreplicating cells failed to grow
in methylcellulose when stimulated with
growth factors. The selected population did
acquire T cell differentiation markers and
synthesized T cell receptor ao and 1B chain
transcripts on coculture with embryonic
thymic tissue (8); thus, this population con-
tained T cell precursors. We now report that
thymotaxin is biochemically and functional-
ly identical to rat 32-microglobulin (,B2m).
Thymotaxin, secreted in a serum-free me-
dium conditioned by IT-45 Rl rat thymic
epithelial cell line (9), was purified by gel
filtration or reversed-phase high-perform-
ance liquid chromatography (HPLC). Col-
umn fractions were tested for their ability to
induce the migration of rat bone marrow
cells in a Boyden chamber assay (10). Bone
marrow cells were partially depleted of ery-
throid and myeloid cells and enriched in
low-density hematopoietic precursors by
centrifugation on a 28% bovine serum albu-
min (BSA) gradient (6). In both gel filtra-
tion and reversed-phase HPLC, a cell migra-
tion activity was associated with the same
11-kD peptide when analyzed by SDS-
polyacrylamide gel electrophoresis (SDS-
PAGE) (Fig. 1, A and B). The active re-
onto a polyvinyl membrane and sequenced
directly from membrane strips in a gas phase
microsequencer (11, 12). The 11 NH2-ter-
minal amino acids were IQKTPQIQVYS,
identical to those of rat 132m (13). Since the
amino acid sequence of bovine P2mcan
easily be distinguished from that of ratP2m
in this region, we confirmed that thymo-
taxin was a product ofthymic epithelial cells
and not a contaminant from the fetal calf
serum used in the culture medium.
The biological activity ofthymotaxin pro-
duced by IT-45 Rl cells was then further
investigated. The thymotaxin fraction from
reversed-phase HPLC (Fig.
10"-1M and another slightly lower peak at
3 x 10-9M (Fig. 2A). The activity found at
10-l'M was completely removed by passage
over an affinity column prepared with rabbit
polyclonal antibodies to mouse ,B2m (anti-
,B2m), which cross-react with rat ,B2m (Fig.
2A). The activity found at 3 x 10-9M,
which was not retained on the anti-,B2m
affinity column, seemed to be due to a
second peptide of 8 kD, which comigrated
with thymotaxin on reversed-phase HPLC
(Fig. iB, lane c) and which could be re-
moved by running the reversed-phase frac-
tion on SDS-PAGE and electroeluting the
lB, lane c)
a peak of maximal
10 NOVEMBER I989
Thymotaxin, a Chemotactic Protein, Is Identical to
CATHERINE DARGEMONT, DOMINIQUE DUNON,
MARIE-ANGE DEUGNIER, MONIQUE DENOYELLE,
JEANNE-MARIE GIRAULT, FLORENCE LEDERER, KIM Ho DIEP Lt,
FRAN ACOIS GODEAU, JEAN PAUL THIERY, BEAT A. IMHOF*
Thymotaxin, an 11-kilodalton protein chemotactic for rat bone marrow hematopoietic
precursors, was purified from media conditioned by a rat thymic epithelial cell line.
The NH2-terminal sequence ofthymotaxin was identical to that of rat 112-microglob-
ulin (,2m). Antibodies to j2m removed thymotaxin activity from the fraction
containing the l1-kilodalton protein. Chemotactic activity was observed with rat
plasma ,B2m, human ,B2m, and mouse recombinant ,B2m, further supporting the
identity of thymotaxin with ,B2m. The directional migration, as opposed to random
movement, of the cells was also confirmed. The only rat bone marrow cells that
migrated toward (12m were Thyl+ immature lymphoid cells devoid ofT cell, B cell,
and myeloid cell differentiation markers.
on December 4, 2008