Aug 26, 1994 - in vitro. Key words: cell differentiation/ES cells/hemopoiesis/ lymphopoiesis//lymphoid precursor cells. Introduction. During the differentiation of ...
The EMBO Journal vol.13 no.22 pp.5274-5283, 1994
In vitro generation of lymphoid embryonic stem cells
Alexandre J.Potocnik, Peter J.Nielsen and Klaus Eichmann1 Max-Planck-Institute for Immunobiology, Stubeweg 51, 79108 Freiburg, Germany 'Corresponding author Communicated by K.Eichmann
Murine embryonic stem (ES) cells represent a model system for studying certain aspects of hemopoiesis because they can differentiate in vitro into several cell types, including those of the hemopoietic system. We developed cell culture conditions in which ES cells undergo hemopoietic differentiation in a low-oxygen (5% 02) atmosphere without additional exogenous factors. After 15-20 days of culture under these conditions, cells bearing surface markers found on cells of the lymphoid lineage (Thyl+, Pgp-1+, c-kit+ and B220+) were detected. After 13-15 days, transcripts for the recombinase activating genes (RAG) 1 and 2, interleukin (IL) 7, IL-7 receptor and c-kit were expressed. We also investigated rearrangements of the immunoglobulin (Ig) heavy and light chain and the T cell receptor (TCR) loci. After 15 days of differentiation, we detected DJH gene rearrangement with Nregion diversity. Productive VHDJH rearrangements are found after 20 days, paralleled by VJ, recombinations indicating a developmental stage comparable, at least, with that of pre B cells. Rearrangements of TCR y as well as 6 chain segments were also observed, but no TCR 3 chain rearrangement. These results demonstrate that ES cells reproducibly generate lymphoid cells in vitro. Key words: cell differentiation/ES cells/hemopoiesis/ lymphopoiesis//lymphoid precursor cells
Introduction During the differentiation of the hemopoietic system, progeny of undifferentiated pluripotent stem cells become committed to one of several developmental pathways and thereby give rise to precursors and mature cell types of a particular hemopoietic lineage. The molecular and cellular events leading to the generation of the lymphoid system represent an unresolved fundamental issue. The signals governing T and B cell development as well as their differentiation stages are poorly characterized, in part due to the lack of appropriate in vitro systems to assess cells early during lymphoid commitment. The approach described in this paper is based on the capacity of embryonic stem (ES) cells, originally isolated from the inner cell mass of the mouse embryo, to differentiate spontaneously in vitro into cells of the hemopoietic lineage
precursors
from
(Doetschman et al., 1985). Upon in vitro differentiation, so-called embryoid bodies (EBs) are found, which contain clustered, erythroid, nucleated cells resembling blood islets in the yolk sac (YS) on days 7-7.5 of embryogenesis (Doetschman et al., 1985; Lindenbaum and Grosveld, 1990; Schmitt et al., 1991; Wiles and Keller, 1991). Within these EBs, not only primitive erythrocytes but also myeloid cells and granulocytes have been observed (Burkert et al., 1991; Wiles and Keller, 1991). The lymphopoetic potential of ES cells has been previously shown in reconstitution experiments: intraperitoneal injection of differentiated ES cells infected with oncogenic retroviruses into adult mice (Chen et al., 1992) or intravenous injection of uninfected cells into newborn mice (Muller and Dzierzak, 1993) repopulated the lymphoid compartment of recipient mice; Gutierrez-Ramos and Palacios (1992) demonstrated lymphopoetic capacity by reconstitution with ES cells cocultured with a stromal cell line in the presence of various exogenous interleukins (ILs). These results suggested the potency of induced ES cells to develop into several lineages of the lymphohemopoietic system. We investigated to what extent lymphoid differentiation could occur in ES cells in vitro without addition of exogenous factors other than fetal calf serum (FCS). Recently it has been demonstrated that the transition from primitive ectoderm to hemopoietic cells in vitro parallels, to some extent, the sequence of events observed in the early YS of the mouse embryo (Keller et al., 1993). Since, at this developmental stage, YS cells committed to the lymphoid lineage are already present (Liu and Auerbach, 1991; Cumano et al., 1993; Palacios and Imhof, 1993), we established conditions for prolonged maintenance of this particular stage in vitro, and analyzed molecular events indicative of lymphoid differentiation. In this report we present evidence for the generation of lymphoid precursors in ES cells differentiating in vitro without addition of exogenous factors. Our results indicate that autocrine factors and/or cellular interactions that provide signals for the commitment of less-differentiated hemopoietic cells to lymphoid precursors are produced endogenously in the ES cell culture model.
Results Induction of Iymphohemopoietic differentiation In an initial set of experiments, we tested culture conditions for their capacity to generate colonies of differentiated ES cells (EBs) producing hemopoietic cells as identified by the appearance of nucleated red cells. As previously shown (Wiles and Keller, 1991), ES cells cultured in semi-solid media containing elevated concentrations of reducing agent, efficiently generated EBs containing nucleated red cells. We found that hemopoietic differentiation, as well as the plating efficiency of ES cells in semi-solid media,
5 4 Oxford University Press 5274
Lymphoid
could be further increased by reducing the 02 content from 19 to 5%, a finding in accordance with recently published observations (Wiles, 1993). When 3-4x 103 cells were seeded in 1.5 ml methyl cellulose medium with 5% 02, BL/6-III ES cells generated 60-90 EBs on day 3, representing a 5-fold augmentation of plating efficiency when compared with 19% 02. In order to facilitate refeeding of long-term differentiating ES cells, we established equivalent conditions for suspension cultures. Cells plated at 1.5-2.OX 104 cells per dish gave rise to 60-100 EBs in 5% 02 or 15-25 EBs in 19% 02. Hemoglobinization developed more rapidly in the 5% 02 cultures with 6075% of EBs containing red cells on day 8, compared to -20% in 19% 02 (Figure 1). Moreover, whereas this phase 0 *"
o
80%
+
60% + 40%
c
._
20% + 0%
-
0
m
Day 0
from ES cells
is transitory in 19% 02, we observed a prolongation of erythropoiesis in ES cells cultured in 5% 02 (Figure 1). Similar kinetics of development were obtained with the ES cell line D3 as well as with the feeder-independent ES cell line CCE/R (data not shown). In order to detect hemopoietic lineages other than erythroid cells, we analyzed the surface phenotype of differentiated ES cells. Two-color FACS analyses of BL/ 6-III ES cells differentiated for 20 days, revealed the presence of Thyl+, Pgp-l+ (CD44+), c-kit+ and F4/80+ cells (Figure 2). The majority of Thyl+ and c-kit+ cells were also positive for Pgp- 1, whereas F4/80 was not found on Pgp- I + cells (Figure 2). The Pgp- 1 +, Thy I + population comprised 11.6% (range 9.3-18.1% in five experiments) and the Pgp-l+, c-kit+ 8.9% (8.1-13.4%) of all induced ES cells. In addition, a small but significant population of Pgp-1+, B220+ cells (3.6-7.1%, mean 5.8%) was observed (Figure 2). Kinetic analysis revealed that Pgp- 1 and Thy appeared between days 13 and 15, whereas c-kit and B220 appeared on day 20 (Figure 3). However, no CD3+, CD4+, CD8+ or surface IgM+ mature lymphocytes were found at any time. Similar results were obtained in five independent differentiation experiments (data not shown) and indicated the presence of cells showing the surface phenotype of lymphocyte progenitors.
Day 20
Day 10
Days of differentiation Fig. 1. Induction of hemopoietic development in differentiating ES cell cultures. The ES cell line BLU6-III was differentiated in 19% 02 (0), or in 5% 02 (U). For each time point, 100 EBs were counted and scored for the presence of red nucleated cells.
0.0
precursors
Gene expression during in vitro differentiation Since several surface markers characteristic for lymphoid cells were found on a proportion of differentiated ES cells, we analyzed the expression of several genes known to be involved in lymphoid development by reverse transcription PCR. Figure 4 shows the analysis for several
0.0 A 1II
I-
0.5 a
Neg. control
5.3
1.2
0 0
Go ?
CM
m 32.3 *1
1
Ila'
T1
114
Pgp-1 Fig. 2. Generation of B220+, Thyl +, c-kit+ and Pgp-l + lymphoid precursor cells by BL/6-III ES cells after 20 days of differentiation in 5% 02. Expression of the surface markers was analyzed by two color flow cytometry. This figure shows data from one representative experiment. Induced ES cells stained with normal rat immunoglobulin and anti-SSEA-1 antibody MC-480 served as negative control. Undifferentiated ES cells were negative for all surface markers used in this experiment.
5275
A.J.Potocnik, P.J.Nielsen and K.Eichmann
Pgp-1 a
Thyl
2.0
'4.9
8.5
>sS1X~ .
B220
c-kit
1 .4
i,j Iid.
I-
E~
.f
b
A 4
X
.R v0.7
- 12.9 e : :342
fluoresence intensity Fig. 3. Representative kinetic analysis of ES cells BL/6-III surface phenotype cultured in 5% 02 for 11 (a), 15 (b) and 20 (c) days. Results include control samples, with isotype matched controls (gray), and test samples (black). Percentage of positive cells for every test sample is indicated.
of these genes (for primers see Table I); a complete synopsis of our analyses is given in Table II. Most significantly, the recombinase activating genes (RAG) 1 and 2 were expressed after 15 days of differentiation (Figure 4a), preceded by the expression of IL-7 and its receptor. Terminal deoxytransferase (TdT) was also found to be expressed starting on day 15 of differentiation (Table Ila). In the developing embryo, RAG I as well as RAG 2 transcripts could be observed even in YS (Figure 4b). In addition, c-kit mRNA could be detected in day 13 ES cells and in day 8.5 YS cells. The induction of these genes is highly reproducible and showed a variation of no more than 2 days in five independent time course studies using the C57/B16-derived ES cell line BL/6-III. A similar pattern of genes transcribed was observed with the feederindependent 1 29/Sv-derived ES cell line CCE/R, excluding that gene expression was dependent on feeder cells present before ES cell differentiation. Whereas RAG I expression was never observed in ES cells differentiating in 19% 02, we occasionally saw faint bands for RAG 2 (data not shown). In summary, the profile of gene expression is fully consistent with the surface analysis and further supports the case for induction of lymphoid cells during the in vitro culture of ES cells in 5% 02
Rearrangement of the B and T cell receptor loci in differentiated ES cells Immunoglobulin ,u DJ and VDJ rearrangements in differentiated ES cells. To study possible functional consequences of the expression of RAG I and RAG 2, we investigated the rearrangement status of the immunoglobulin (Ig) heavy chain locus. We first studied the status of DQ52 to JH rearrangement, known to take place in a large proportion of lymphocytes at a very early stage of development (Chang et al., 1992). DJ rearrangements were detected by PCR using primers hybridizing 5' to DQ52 and 3' to JH4. DNA of differentiated ES cells clearly showed DQ52 to JH rearrangement (Figure Sa), as was the case for fetal liver day 11.5 and adult thymus. The latter result is in agreement with previously published data (Born et al., 1988; Chang et al., 1992). No rearrangement
5276
a ES cells
I
*4
HPRT
RAG 1
RAG 2
IL-7
IL-7 R
b fetal tissues
HPRT
RAG 1
RAG 2
co C9 Co Co )e c oe be e CV
4,)
c
c
4, 4,
4,
.I
A
-Gn
4,)
, 4, 4 4, 4,
4,
)
v (:)
.2 k 4,, 4
... ..... ...
Fig. 4. Gene expression of ES cells differentiated in vitro (a) and of various embryonic tissues (b). After 30 cycles of amplification, samples were electrophoresed and ethidium bromide-stained. (a) shows the induction of various genes from day 5 to day 22 in differentiated ES cells, (b) various control samples including yolk sac (YS), fetal liver (FL), embryonic fibroblasts (EF) and undifferentiated ES cells BL/6-III. For each sample, the appropriate positive control as listed in Table I was included.
Lymphoid precursors from ES cells Table I. Oligonucleotide primers used for gene expression by RT-PCR Gene
Size/bp
5'-sequence
3'-sequence
Positive control (reference)
c-kit
765
5'TGTCTCTCCAGTTTCCCTGC
5'TTCAGGGACTCATGGGCTCA
HPRT IL-7 IL-7 receptor TdT RAG 1 RAG 2
249 355 236 313 648 316
5'CACAGGACTAGAACACCTGC 5'ACATCATCTGAGTGCCACA 5'CAAAGTCCGATCCATTCCCCATAAC 5'GAAGATGGGAACAACTCGAAGAG 5'TGTCTTCCACTCCATAACCAG 5'CCCAGAGAACCACAGAAAAAT
5'GCTGGTGAAAAGGACCTCT
Bone marrow (Keller et al., 1993) (Keller et al., 1993)
5'CTCTCAGTAGTCTCTlTTAG 5'GTTTTCTTATGATCGGGGAGACTAGG 5'CAGGTGCTGGAACATTCTGGGAG 5'CGATTCATTTCCCTCACTT 5'TAACCACCCACAATAACAAAT
Thymus S57-cells (IL-7 dependent) Liver (Li et al., 1993) Thymus Thymus
Abbreviations: HPRT = hypoxanthine phosphoribosyltransferase, TdT = terminal deoxytransferase. Table Ila. Complete summary of PCR data on ES time course, days 0-22 Gene
Signal detected after indicated days of differentiationa 0
5
8
-
-
-
-
RAG2 IL 7 IL-7 receptor c-kitb
-
-
-
-
+
+ +
-
+
+
+
RAG I
TdTb
-
11
13
15
+
+ +
-
16
18
20
22
+
+
+ + +
+ + + +
±
+ + +
+ +
+ +
+ +
+ + + + + +
+ + +
Table Ilb. Complete summary of PCR data on yolk sac and fetal liver cells in vivo Gene
YS day 8.5
YS day 9
YS day 9.5
YS day 10
FL day 11.5
FL day 12.5
Embryonic fibroblasts
RAGJ
-
-
+
+
+
+
-
RAG2
-
-
-
-
+ +
+ + +
+ + +
+ + +
-
IL-7 IL-7 receptor
+ + +
c-kitb
+
+
+
+
+
+
+
TdTb
-
-
-
-
-
-
-
aAfter 30 cycles of amplification, samples were gel electrophoresed and photographed; signal. bData not shown in Figure 4.
was found in day 8.5 YS cells, undifferentiated ES cells or embryonic fibroblasts. The clone HAFTL 4, originated from pre-B cell line HAFTL 1-14 (Alessandrini et al., 1987), has undergone a DQ52 to JH2 rearrangement (P.J.Nielsen and J.Pelkonen, unpublished result) and served as internal control for the specificity of the PCR method. To further exclude an origin of the lymphocyte precursor from the 1 29/Sv neo-transgenic cells used as feeder during culturing of non-differentiated ES cells, all DNA preparations were subjected to a PCR specific for the neo resistance gene. No signal was detected in any DNA samples of differentiated ES cells (data not shown). We extended these studies by analyzing VHDJH rearrangements involving the VHQ52 and the VH7 183 gene families. In the germline, these elements are at vast distances from JH4 and therefore cannot be amplified using PCR. Figure Sb and c shows the results for VHQ52 and VH7 183 to DJH rearrangements respectively. Members of both VH families have therefore undergone rearrangement in differentiated ES cells. A time course analysis for VHQ52 to DJH rearrangements is shown in Figure 6. No rearrangements were observed before day '15, in concordance with the expression of RAG I and RAG 2. Similar results were obtained in differentiation experi-
-
(±) corresponds to faint signal, (+) strong signal, (-) no
ments with the feeder-independent ES cell line CCE/R (Figure 6), suggesting that embryonic fibroblasts are not necessary for the generation of lymphoid progenitors. Based on the strength of the PCR product bands, the frequency of DJ rearrangement in differentiating ES cells appears to be comparable with that in fetal liver cells. In contrast, the frequency of VDJ rearrangement was significantly lower in differentiating ES cells when compared with fetal liver. This finding clearly demonstrates that ES cells can differentiate in vitro into lymphoid cells with several characteristics corresponding to early stages of B cell development. Ig K rearrangements in differentiated ES cells. To further define the developmental stage of the differentiated ES cells, we studied Ig light chain rearrangement usually observed only after productive rearrangements of VH to DJH segments (Rolink and Melchers, 1991). After 20 days of in vitro differentiation in two separate experiments, V, to J,K2 rearrangements were detectable in the progeny of ES cells (Figure 7). In one case, recombination in the Ig light chain locus was even observed on day 15. Whereas undifferentiated ES cells, embryonic fibroblasts and YS cells showed no rearrangement, the sensitivity of this method was sufficient to detect signals in brain (probably
5277
A.J.Potocnik, P.J.Nielsen and K.Eichmann
a
DQS2JH
40 z
4.) 4) Z Z. -,.i Z. g Z..,. .,'(O N,- " 4. 4.. 4.7 4. .Z, -.q 410, N
.o
C.,
VQ52DJH
co 'O, ., .- A,."
`.
e A:.-, .r Q. ,
1
k
yGermline
epa"
a
;..P .4iwl#
b
VQS2DJH
$
$
e> o
o
10
4% .
i
~N~~Z..'
~
Z
Z
'S
DJ,
DJ2 I DJ.,
St:
4
ID40
-Z~ a
BL/6-111
VDJ1 0 Ooiv
VOJ2
DJ4
VDJ3 VDJ4
"
.Z
z C C e-. -,v e,
Q
~
N~ r
4~
ai
4)
4~
SSSSSSa%eboPw~i ~-9~
4J4
N,
NvK
~
v
~41
VDJ1 -.-vDJ2 --
4,, e
V7183 DJ H -,
-,
VDJ3
CCE/R
VDJ VDJ2
VDJ4
4w
VDJ3
VDJ4
'O >
4t1e::~Zi,C
ec
e aA7,.
i~.1p
VDJ, VDJ3
Fig. 6. Rearrangement of immunoglobulin heavy chain genes during in vitro differentiation of ES cells. Using two different ES cell lines, VHQ52 to DHJ rearrangements were analyzed by PCR as described in Figure 5. (a) shows the pattern obtained in two independent differentiation experiments using the feeder-dependent ES cell line BL-6/III; and (b) the results for the feeder-independent ES cell line CCE/R.
VOJ4
Fig. 5. Rearrangement of Ig heavy chain genes during in vitro differentiation of ES cells. In two independent experiments, ES cells were harvested after 15 or 20 days of differentiation and their Ig heavy chain locus was analyzed by PCR. Products were separated by gel electrophoresis, blotted to a nylon membrane and assayed for hybridization to a JH4 probe. In DJ rearrangements (a), priming was done 5' to DQ52 and 3' to JH4. Fragments corresponding to the respective recombinations are indicated. The pre-B cell clone HAFTL 4, known to carry a DQ52 to JH2 rearrangement, served as additional control. In VDJ rearrangements, 5' priming was done for VHQ52 (b) or VH7183 (c). As expected, no germline band is obtained in (b) and (c).
due to contaminating lymphocytes) and fetal liver, consistent with a low frequency for this recombination in differentiated ES cells. Rearrangement in the T cell receptor loci of differentiated ES cells. Since DJ rearrangements of the Ig heavy chain sometimes also occur in T cells, we next studied the rearrangement status of the T cell receptor (TCR) gene loci. Vs1 to D82 recombinations have been shown to take place at very early stages during T cell development (Chien et al., 1987). PCR analyses revealed that this particular rearrangement is constantly observed in EBs after 20 days of in vitro differentiation (Figure 8a). We further extended this analysis to the Vy2, Vy3, Vy4 to J71 cluster of rearrangements in the TCR y chain locus which is also observed early during ontogeny (Havran and Allison, 1988; Raulet, 1989) and is often found in combination with V51 rearrangement (Ito et al., 1989). Using a recently described PCR assay (Goldman et al., 1993), Vw,2, Vy3 and VY4 to J,y recombinations could be demon5278
b
VVJ h
40
.,
,,
4.
Q
CQ
N
~141474~-94 Q~A~ ~
z
Fig. 7. Rearrangement of the Ig light chain genes during in vitro differentiation of ES cells. V. to J,2 rearrangements were analyzed by PCR using VK degenerate and J,,2 primer combinations and detected by hybridization as described in Materials and methods.
strated in differentiated ES cells (Figure 8b, c and d). Both TCR y and 6 rearrangements were detected in two different ES cell lines (Figure 8a-d). In contrast, no rearrangements involving TCR ,B chain D to J segments were found in differentiated ES cells. Figure 8e shows the pattern obtained in a PCR assay specific for Dpl to Jp2.1-2.6 recombinations; equivalent results were observed for Dp2 to Jp2. 1-2.6 specific rearrangements (data not shown).
Rearrangement junctions in the Ig receptor and TCR loci in differentiated ES cells N-region diversity at the junctions of D and J segments and at the junctions of V and DJ segments is rare if not totu4ly absent in fetal B cells (Feeney, 1990; Gu et al., 1990)
Lymphoid precursors from ES cells
I
(ob, co
a
V61D6I
,
,
;ll
'Adik, . .Alk
A&.II ,.!!W"
--I
b
V.-/-)J7 Ink.
--I
C
V-/,-J-7l
Aimilmadl.l.
--
__j
d V.,4J.,1 I
I
e
D03lJFl
binding upstream of DQ52, DFL16.1, Dsp2.7 and downstream of JH4 in the Ig heavy chain locus. PCR products corresponding to D to JH2 or JH3 rearrangements were cloned and sequenced. The results for these PCR-amplified sequences (Figure 9) showed heterogeneous junctions including N-nucleotide insertions. We next examined Ig ,u chain VDJ rearrangements. Figure 10 presents the sequences for rearrangements utilizing VHQ52 to DJH3 or DJH4. In virtually all cases, we observed heterogeneous junctions with N-nucleotide insertions. The majority (10/13) of VDJ sequences obtained from differentiated ES cells represent productive rearrangements which could code for a functional Ig heavy chain. In addition, the utilization of D segment reading frame (see Figure 10 legend) was similar to ratios observed in fetal and adult VDJ (Gu et al., 1991; Chang et al., 1992). These results clearly indicate the presence of cells committed to the B cell lineage. A similar approach was undertaken to characterize the TCR 6 locus junctions. V61 to D62 rearrangements have so far been reported only for T cells and are thought to be absent in heavy chain rearranging cells (Raulet, 1989). Using a nested PCR, nine cloned PCR fragments showing V61 to D82 rearrangement were isolated after 20 days of in vitro differentiation. In this set, only five different junctions were present, whereas two clones each were found three times (Figure 11). This presumably reflects the low frequency of rearrangement in the TCR 6 locus in differentiated ES cell cultures. These results may be the first indication for T cell lineage precursors generated in vitro. The majority of clones showed N-region diversity (see Figures 9-11) and therefore did not resemble B or T cell receptor repertoires usually generated during fetal development. This result points to the possibility that ES cell differentiation in vitro may not faithfully reflect all aspects of the developmental process during ontogeny.
Discussion 0A
Fig. 8. Rearrangement of the T cell receptor genes during in vitro differentiation of ES cells. (a) The Vsl to D82 rearrangement pattern for BL-6/III derived ES cells (ES IV and ES V) or CCE/R-derived cells (ES 1, ES 2) after 20 days of in vitro differentiation. Specificity of the PCR reaction was confirmed by hybridization with a Va1specific 32P-labeled probe. As a positive control for the sensitivity of this technique, a DNA preparation of a day 14.5 fetus was used. Rearrangements involving the Vy2, V13, Vy4 to Jyl elements in the TCR y chain locus of differentiated ES cells were analyzed as described by Goldman et al. (1993); results are presented in (b-d). (e) The pattern obtained in a PCR assay specific for Dpl to Jp2.1-2.6 recombinations of the T cell receptor f chain; equivalent results were observed for Dp2 to Jp2.1-2.6 specific rearrangements (data not shown).
and fetal 'y8 T cells (Elliott et al., 1988), corresponding to the lack of the expression of TdT (Gregoire et al., 1979; Gilfillan et al., 1993; Komori et al., 1993). Since we found TdT transcripts, we were interested whether this would also lead to N-region diversity. Total DNA of day 20 differentiated ES cells was amplified using PCR primers
Several groups have reported the in vitro generation of myeloid and erythroid cell lineages from ES cells in semisolid media or suspension culture supplemented with IL- 1, IL-3 or erythropoietin (Lindenbaum and Grosveld, 1990; Burkert et al., 1991; Schmitt et al., 1991; Wiles and Keller, 1991). During in vitro culture however, no evidence for the generation of mature lymphoid cells was detected. Nevertheless, several results indicate the presence of precursor cell types that are able to develop to mature lymphocytes under appropriate conditions. ES cells differentiated in vitro have been shown to develop into lymphoid cells in vivo when transferred into adequate hosts (Muller and Dzierzak, 1993). Similar transfer experiments suggested the generation of lymphoid precursors from ES cells when cultured on bone marrow stromal cell lines in the presence of several ILs (Gutierrez-Ramos and Palacios, 1992; Nisitani et al., 1994). Our experiments were meant to test the possibility that within the developing EB, a suitable microenvironment for improved hemopoiesis could be created using appropriate culture conditions. Interestingly, this was achieved by the reduction of the 02 content, which not only accelerates hemopoiesis, but also delays the subsequent down-regula-
5279
A.J.Potocnik, P.J.Nielsen and K.Eichmann
Clone
hepta DQ52
Qupl Qup2 Qup3 Qup4b Qlow4 Qlow5 Qlow6 Qlow6b Qlow7 Qlow7b Qlow8
D
N
... CACAGTG CTAACTGGGAC CACAGTG CTAACTGGGAC
of
...
"1
CACAGTG ... CACAGTG ... CACAGTG ... CACAGTG ... CACAGTG
"1
"1
"1
"1
"1
...
CTAA
CACAGTG CACAGTG ... CACAGTG ... CACAGTG
CTAACTGGGA CTAACTGGG CTAACTGGGAC CTAACTGGGAC CTAACTGGGAC CTAACTGGGAC CTAAC CTAACTGGGA
...
if
...
"1
"1
J
ATA
CG GTG
A TGG
N
D Ds f9
DFL16.1
Ds f lOb
Dsfl3
if
Dsfl2
Dsp2.7
TACTTTGACTACTGG... ACTACTTTGACTACTGG... GACTACTGG... GACTACTGG... CCTGGTTTGCTTACTGG... GGTTTGCTTACTGG... CTGGTTTGCTTACTGG... CTGGTTTGCTTACTGG... GGTTTGCTTACTGG... GGTTTGCTTACTGG... GGTTTGCTTACTGG...
CCCCTT
JH2 JH 11
11
JH3
J
...TACTGTG TTTATTACTACGGTAGTAG CCAG ... TACTGTG TTTATTAC ... TACTGTG TTTATTACTACGGTAGT GC
GACTACTGG... GACTACTGG...
JH2
GCTTACTGG...
JH3
... TACTGTG CCTACTATGGTAACTAC
GACTACTGG...
JH2
JH
Fig. 9. Immunoglobulin ,u chain DJ rearrangement junctions in differentiated ES cells. Bold sequences indicate N-region insertions. Clone 27 5 42 44
V052
TGT TGT ... TGT ... TGT ... TGT 7 42b ... TGT ... TGT 21 ... TGT 24 9 ...TGT ... TGT 3 ... TGT 43 ... TGT 26
N
...
GCC AGA GA
...
GCC AGA G GCC AGA GCC AGA GCC AGA GCC AGA GCC AGA GCC AGA GCC AGA
G
GA G GA GA
GOC A G
GCC AGA GA
V052 2
...TGT GCC AGA G
D
N
TTACTACGGTAGTAGC TAGT GAGGA TTCATTACTACGGCTA ACAGCTCG GGCTAC AATAA GCTAC AGCYAATGA TATGGTAACTA TCAA GGTAACTAC GAGAGGGAACATCTACTATGGTAACTAC AAGCCTAAAGC ATGGTAACTAC CCT CCTACTATAGTAA TAAGAGGGG GGTACGAC CGGGAAG TATAGGTACGAC TTC
N CT
CTAT
AT TAC TAT GCT ATG GAC TAC TGG...
AT G TAC TAT TCAGGA AT ACRACTA AT GCGA AT TAC TAT GTAAGG AC TAT G T GC AT GGAGAGG GTTGAG TAC TAT
D
D segment used
JH4
N
T GCT GCT GCT GCT GCT GCT GCT GCT GCT GCT
ATG ATG ATG ATG ATG ATG ATG ATG ATG ATG ATG
GAC GAC GAC GAC GAC GAC GAC GAC GAC GAC GAC
TAC TAC TAC TAC TAC TAC TAC TAC TAC TAC TAC
DFL16.1 or Dsp2.x DFL16.2 TGG..."
TGG... TGG... Dsp2.1 or 2.5,2.7,2.8 TGG... TGG... Dsp2.5 TGG... " TGG... Dsp2.x TGG... Dsp2.10 or 2.11 TGG...
JH3
TATGATTACGAC GGACGACGGCC
DFL16.1
TGG... TGG...
G TTT GCT TAC TGG...
D segment used Dsp2.2
Fig. 10. Immunoglobulin ,u chain VDJ rearrangement junctions in differentiated ES cells. Nucleotide sequences are shown utilizing VHQ52 to JH4 or J3 respectively. Bold sequences indicate N-region insertions. Ten out of 13 clones analyzed represented productive rearrangements, DH reading frame usage was 7:4:1 according to the nomenclature of Ichihara et al. (1989).
tion of hemopoietic activity described by others (Wiles and Keller, 1991). The reason for this effect might be a prolonged preservation or even rescue of a multipotent stem cell population which requires extended culturing to develop into lymphoid cells. In addition, the low oxygen atmosphere might favor a particular microenvironment inducing developmental progression via the secretion of growth and differentiation factors or the expression of certain adhesion molecules. In line with this hypothesis is the presence of cytokines like IL-7 and of cytokine receptors like IL-7 receptor or c-kit, known to contribute to lymphoid differentiation in vitro. Such a local environment could synchronize the developmental program of several distinct cell types which might have to interact in a well defined pattem. It is also possible that during progression 5280
toward the lymphoid lineage, certain stages of progenitor cells are highly sensitive to the effects of oxygen in culture and become eliminated or arrested. This might explain the fact that when ES cells were differentiated under standard cell culture conditions only a very limited in vivo repopulation by in vitro differentiated cells has been achieved (Muller and Dzierzak, 1993). Following the induction of ES cell differentiation, we observed cells expressing Pgp-1, Thyl, B220 or c-kit. The expression of these markers is consistent with the presence of lymphoid progenitors, because the same markers are found on lymphoid progenitors in bone marrow or fetal liver. Particularly the populations of Pgp-l+, B220+ cells and of Pgp-1+, Thyl+ cells are presumptive candidates for lymphoid precursors. The fact that these populations
Lymphoid precursors from ES cells Clone ESI14 ESI15 ESI9a ESI9b ESI9c ESI14a ESI16a
ESI16b ESI16c
V81
...
GGGTCAGATATC
...
GGGTCAGAT
We therefore focused on rearrangements at the TCR 6 chain locus in differentiated ES cells. Besides being one
D82
N CCAT
ATCGGAGGG
CCCAT
ATCGGAGGG
of the first
events in T cell development (Chien et al., 1987), rearrangement between V61 and D62 seems to be
...
GGGTCAGA
ACCTCCTTCTTCT
ATCGGAGGG
restricted to T cells. Our demonstration of rearrangements
...
GGGTCAGA
ACCTCCTTCTTCT
ATCGGAGGG...
both in the TCR y and
...
GGGTCAGA
ACCTCCTTCTTCT
ATCGGAGGG...
lineage
...
GGGTCA
AAT)kGCCCGT
ATCGGAGGG
DP2
...
GGGTCAGATA
GCCCGT
ATCGGAGGG
TCR
...
GGGTCAGATA
GCCCGT
ATCGGAGGG...
ment
...
GGGTCAGATA
GCCCGT
ATCGGAGGG...
et
Fig. 11. T cell receptor V81 to D62 rearrangement junct tions differentiated ES cells. Bold sequences indicate N-regio'n insertions.
made up 7-15% of all cells in the culture indicates a substantial tendency towards lymphoid dev'elopment in our cell culture system. A crucial indicator for the commitment to tthe lymphoid lineage is the expression of the two recombinavse-activating genes RAG I and RAG 2, which were dettectable at a reproducible time-point during in vitro culture Expression of these genes is consistent with the observration of the recombination of the Ig heavy and light chaiin as well as the TCR y and 6 chain loci. Rearrangementt of the DJH locus was observed using PCR after 15 dayrs in culture. Initially, we focused on the DQ52 element since this is known to be used in DJ joints at very ear^ly stages of differentiation (Chang et al., 1992). Neverthe less, we also found usage of DFL16.1 and Dsp2.7. Stril kingly, these junctions showed N-region diversity, which iis in contrast to the junctions found during fetal developrnent (Elliott et al., 1988; Feeney, 1990) or in mice 1 acking TdT (Gilfillan et al., 1993; Komori et al., 1993). 1 [hese results are in line with the fact that TdT expressiion was not detected during fetal development (Gregoire et al., 1979; Li et al., 1993) but was observed in our diffe rentiated ES cell cultures. Whether this reflects an overe: xpression of TdT in ES cell cultures or a fundamental de,viation from the in vivo situation is presently not clear. Whereas DJH recombination is not restricte to B cells (Bom et al., 1988), VHDJH rearrangement is usually only observed in cells committed to the B cell line-age (Rolink and Melchers, 1991). The fact that we observe;d rearrangements involving VHQ52 and VH7 183 s,egments in differentiating ES cells strongly suggests the presence of committed B cells. However, a comparison wiith the PCRproduct signal intensity in day 11.5 fetal livei r indicated a lower frequency for these rearrangements in ES cell cultures. Interestingly, D to JH rearrangeiments were observed at roughly the same level in both cases. This might suggest that the fetal liver promotes the progression from DJ to VDJ more efficiently than our cell culture
~d
system.
Analogous questions have to be asked re garding the commitment to the T cell lineage. Our dlata provide evidence for rearrangements in the TCR y and 6 loci. Nevertheless, recombinations within the inves,tigated Vy3, Vy4 to Jl, cluster might also be due 1to abnormal rearrangement activities in cells committecd to the B cell lineage, as has been reported for T(I'R y chain rearrangements in pre B cell lines (Traune,cker et al., 1986; Cook and Balaton, 1987; McCubrey e ,t al., 1989).
to
loci
are
first indications for T
commitment in differentiated ES
JP2.1-2.6
recombinations
locus rearrangement, which
cells.
Dp1
or
starting point in precedes the rearrange-
are
the
a locus (Raulet et al., 1985; Snodgrass 1985). These rearrangements were not detected,
of the TCR
al.,
suggesting that ES cells under
cells fail to develop in differentiated ca3 Tculture conditions. This may be due our
to the absence of a
thymic microenvironment, thought to be essential for cxi T cell development. In addition, it should be stressed that for both the Ig chains and the TCR loci, no precise quantitative statement on the frequency of a given rearrangement could be made. Nevertheless, these molecular data strongly suggest the development of immature lymphoid cells in our in vitro model, perhaps in both the T and B cell lineages. To what extent is our cell culture system similar to the developing embryo? Several genes expressed during in vitro differentiation of ES cells are also expressed in fetal YS cells, suggesting a similar regulation of expression during in vitro and in vivo differentiation (Keller et al., 1993). In addition, several reports demonstrated erythroid and myeloid precursors within the EBs (Doetschman et al., 1985; Lindenbaum and Grosveld, 1990; Burkert et al., 1991; Schmitt et al., 1991; Wiles and Keller, 1991). Our report provides evidence that lymphoid precursors with rearranged or partially rearranged Ig heavy and light chain or TCR y or 6 genes are present in differentiating ES cells, which are not observed in day 8.5 YS. Cells undergoing rearrangements of the Ig or TCR loci appear first in the fetal liver (Carding et al., 1990; Chang et al., 1992). Moreover, the surface phenotype of these differentiating ES cells shows similarities to cells found in fetal liver. However, the most striking difference between differentiating ES cells and fetal liver lymphoid progenitor cells so far, is the appearance of N-region diversity in our cell culture model. In summary, ES cells generate lymphoid precursors during in vitro differentiation which strongly resemble, but are not identical to, the fetal liver stage of lymphoid precursors. The availability of an in vitro model for lymphopoiesis provides a unique tool to manipulate the developmental process and to identify genes and their functions along this pathway. Of practical interest is the possibility of testing directly the role of cytokines and differentiation factors as well as of their respective receptors. Moreover, the model can be applied to ES cells which have been inactivated by homologous recombination on both alleles for developmentally regulated genes. Results from this model should provide new insights into the program of lymphoid differentiation and facilitate future studies of the mechanisms controlling this process.
V.2,
Materials and methods Differentiation of ES cells
C57/B16-derived ES cell lines BL/6-III (Ledermann and Burki, 1991) or 129/Sv-derived D3 were maintained in an undifferentiated state
The
5281
A.J.Potocnik, P.J.Nielsen and K.Eichmann by culture on a monolayer of mitomycin-C-inactivated embryonic fibroblasts (10 jg/ml at 37°C for 1.5 h) prepared from 14 day embryos of a 129/Sv mouse carrying a neo transgene (Gossler et al., 1986) as described (Doetschman et al., 1985). The 129/Sv-derived ES cell line CCE/R, adapted to grow in the presence of leukemia inhibitory factor (LIF) and without feeder cells, was kept undifferentiated by culturing in media containing 1000 U/mI LIF (supplied by Gibco, Eggenstein, Germany). Only ES cells cultured for
