Putative “Stemness” Gene Jam-B Is Not Required for Maintenance of Stem Cell State in Embryonic, Neural, or Hematopoietic Stem Cells

Abstract

Many genes have been identified that are specifically expressed in multiple types of stem cells in their undifferentiated state. It is generally assumed that at least some of these putative “stemness” genes are involved in maintaining properties that are common to all stem cells. We compared gene expression profiles between undifferentiated and differentiated embryonic stem cells (ESCs) using DNA microarrays. We identified several genes with much greater signal in undifferentiated ESCs than in their differentiated derivatives, among them the putative stemness gene encoding junctional adhesion molecule B (Jam-B gene). However, in spite of the specific expression in undifferentiated ESCs, Jam-B mutant ESCs had normal morphology and pluripotency. Furthermore, Jam-B homozygous mutant mice are fertile and have no overt developmental defects. Moreover, we found that neural and hematopoietic stem cells recovered from Jam-B mutant mice are not impaired in their ability to self-renew and differentiate. These results demonstrate that Jam-B is dispensable for normal mouse development and stem cell identity in embryonic, neural, and hematopoietic stem cells.

Embryonic stem cells (ESCs) are derived from mammalian preimplantation embryos and possess the remarkable ability to differentiate into all embryonic cell types. Moreover, they can grow indefinitely without losing this pluripotency if cultured under appropriate conditions. A number of key transcription factors, such as OCT-4, Nanog, SOX-2, and FOXD3 (5, 7, 12, 22, 25), have been shown to be essential for sustaining ESC properties. However, it is not known how these molecules contribute to maintenance of the pluripotent state.

Somatic stem cells, including neural stem cells (NSCs) and hematopoietic stem cells (HSCs), share some of the properties of ESCs, including multipotency and self-renewal. In the event of severe injury, numerous types of tissue-specific stem cells can give rise to cells of heterologous lineages (39, 42, 43), although in some cases, fusion of stem cells with other cells appears to be involved in transdifferentiation (21, 29, 46). Thus, it is possible that ESCs and somatic stem cells share a common genetic program that maintains stem cell identity (20, 37, 40, 42). Recently, Ivanova et al. (13) identified 283 genes or expressed sequence tags, including a gene encoding junctional adhesion molecule B (JAM-B) (nomenclature of the protein in NCBI Database is JAM2), that are expressed in three different stem cell lines, by means of DNA microarray analysis. Although it is assumed that at least some of these genes are involved in the maintenance of stem cell properties, no data confirming this have yet been reported.

Here we investigate this possibility. We have focused on Jam-B, a putative “stemness” gene identified by Ivanova et al. (13), because of the extreme difference in expression levels between undifferentiated and differentiated ESCs. Jam-B encodes an immunoglobulin superfamily protein that is specific to tight junctions and mediates cell-cell contacts between T cells and endothelial cells and many other systems (2-4, 9, 11, 16-18, 32). We first generated ESCs in which Jam-B was doubly targeted. We also generated Jam-B knockout mice by targeting disruption to examine the role of the Jam-B gene in maintenance of the stem cell state of NSCs and HSCs and in other aspects of development. These analyses revealed that Jam-B mutant ESCs are normal in morphology and retain pluripotency. Moreover, we found that Jam-B knockout mice were viable and indistinguishable from wild-type mice in appearance. Furthermore, we found that NSCs and HSCs recovered from Jam-B mutant mice are equivalent to those recovered from wild-type mice in the common properties of stem cells, such as multipotency. Unexpectedly, our analyses also revealed that Jam-B mutant male mice were also normal in spermatogenesis, although it has been assumed that the JAM-B protein present in Sertoli cells plays crucial roles in spermatogenesis by interacting with the JAM-C protein present in spermatids (11).

MATERIALS AND METHODS

DNA microarray analysis.

RNA was prepared from undifferentiated and differentiated ZHBTc4 ESCs (28), and poly(A)⁺ RNA samples were recovered using an oligo(dT) cellulose column. One microgram of poly(A)⁺ RNA was used for reverse transcription using a T7-oligo(dT) primer bearing the T7 RNA polymerase promoter (Affymetrix, Santa Clara, CA) and SuperscriptII (Invitrogen). After second-strand synthesis and purification of double-stranded cDNA, cRNA was synthesized by in vitro transcription using the Bioarray RNA transcript labeling kit (Affymetrix). Fifteen micrograms of cRNA was cleaved into 35- to 200-base fragments, according to the manufacturer's instructions (Affymetrix). The fragmented cRNA was mixed with hybridization solution containing Control Oligonucleotide and Hybridization Controls (Affymetrix) and hybridized to Affymetrix mouse U74Bv2 arrays. Hybridized arrays were scanned and analyzed by Affymetrix MAS 4.0 software.

Cell culture.

ZHBTc4 (28), E14tg2A (38), and TT2 (45) embryonic stem (ES) cells were cultured as described previously (27). Differentiation of ZHBTc4 ES cells (feeder free) was done simply by adding tetracycline (1 μg/ml) to normal ES medium containing leukemia inhibitory factor when transferred to a new tissue culture dish (5 × 10⁵ cells per 10-cm dish) and cultured as a monolayer for 48 h.

Neurosphere culture.

Forebrain cells were prepared as described previously (23) from 12.5- or 14.5-days postcoitum (dpc) embryos obtained from intercrosses of wild-type or Jam-B homozygous mutant mice. After dissociation into a single-cell suspension, cells were seeded onto noncoated 10-cm dishes at a concentration of 1 × 10⁵ to 2 × 10⁵ cells/ml. Cells were cultured for 4 to 6 days with 1× B27 supplement (Invitrogen)-containing medium plus 20 ng/ml basis fibroblast growth factor (bFGF) and epidermal growth factor (EGF). For comparing efficiency of neurosphere formation between wild-type and Jam-B mutant mice, dissociated forebrain cells (see Fig. 6B) or primary neurosphere cells (see Fig. 6D, left panel) were seeded into 96-well plates at the indicated concentrations. After 6 to 10 days of culture, neurospheres were counted under a microscope. For differentiation of neurosphere cells, primary spheres were dissociated into single cells and then cultured on poly-l-ornithine and fibronectin-coated dishes in the absence of bFGF and EGF.

6.

Loss of Jam-B function does not affect neural stem/progenitor cells. (A) Cells from the forebrain of 12.5-dpc embryos were used for neurosphere formation, and generated colonies were stained for β-galactosidase expression. (B) Numbers of neurospheres generated from forebrain cells of 12.5-dpc wild-type or Jam-B mutant embryos cultured at different densities. Data were obtained from three independent experiments with comparable results. (C) Generation of neurons and glia from Jam-B-mutant neurosphere cells. Immunostaining of neurons (MAP2 positive) or glia (GFAP positive) generated by differentiating primary neurosphere cells for the indicated number of days. MAP2-positive and GFAP-positive cells were visualized with Alexa 594 and Alexa 488 Fluor dye-conjugated secondary antibodies, respectively, to red and green. Cells were counterstained with 4′,6′-diamidino-2-phenylindole. (D) Self-renewal and differentiation abilities of neurospheres from wild-type and Jam-B mutant mice. Primary neurospheres of similar size (about 0.2 mm in diameter) generated under clonal conditions were individually dissociated and recultured for secondary neurosphere formation (left panel). The number of generated secondary neurospheres was determined under a microscope. The data were obtained from 3 independent experiments in which 16 colonies were used for each experiment. For assessing differentiation ability (right panel), the dissociated cells from primary neurospheres were induced to differentiate by culturing in the absence of bFGF and EGF. After 5 days, cells were fixed and subjected to double-immunostaining procedures using anti-GFAP and -MAP2 antibodies. Sixteen independent primary neurospheres were used for this experiment. (E) Immunostaining of 12.5-dpc wild-type and Jam-B mutant forebrains. The solid and dotted lines correspond to the outer and the surfaces of the brain. (F) Immunostaining of 14.5-dpc wild-type and Jam-B mutant spinal cords. Dorsal is on top.

RNase protection assays.

RNA preparation and RNase protection assays were performed as described previously (27). For detection of the various mRNA species, the following cDNA sequences were used: Jam-A, 266 to 556; Jam-B, 275 to 594; Jam-C, 294 to 577; Oct-4, 1 to 401; Nanog, 54 to 254; Sox-2, 480 to 905; Utf1, 226 to 544; Rex-1, 75 to 351; Gfap, 67 to 416; Nestin, 4153 to 4462; Map2, 208 to 482; Pax6, 1045 to 1311; Gata4, 1187 to 1560; Gata6, 1611 to 1882; HNF3β, 376 to 604; MyoD, 1101 to 1424; Tnni1, 153 to 403; Tnni3, 148 to 403; and β-Actin, 903 to 1023. The numbers represent the nucleotide position, with the adenine nucleotide of the initiating methionine codon set to +1.

Reverse transcription-PCR analysis.

Hematopoietic cells were recovered from bone marrow, thymus, and spleen of C57BL/6 mice and fractionated using fluorescence-activated cell sorting (FACS) with antibodies against the cell surface markers indicated in Fig. 2C. Total RNA was isolated using ISOGEN-LS (Nippon Gene, Tokyo, Japan) and reverse transcribed using an oligo(dT) primer. After normalization of cDNA content using the glyceraldehyde-3-phosphate dehydrogenase cDNA level as a standard, semiquantitative PCR was carried out. β-Actin was used as an internal control. The primer set sequences for Jam-B and β-Actin cDNA are as follows: Jam-B, 5′-ACGAAGCTTTCAATATACGAATCAAAA-3′ and 5′-CATGTTGAATTGCAGAAT TC-3′; β-actin, 5′-GGTCAGAAGGACTCCTATGT-3′ and 5′-ATGAGGTAGTCTGTC AGGTC-3′. Real-time PCR analyses shown in Table 2 were performed as follows. Total RNAs were prepared from various tissues of 8-week-old mice and embryos at variable stages and reverse transcribed using a random primer. These samples were used to examine the levels of gene expression of Jam-A, -B, -C, and β-Actin by real-time PCR using TaqMan probes from Applied Biosystems.

FIG. 2.

Expression of Jam-B transcript in ESCs, NSCs, and HSCs. (A) ESCs were induced to differentiate according to the method of Robertson (34). RNAs were recovered and used for RNase protection assay. Jam-B expression in ESCs declines rapidly upon induction of differentiation. ES, undifferentiated ESCs; d, days of differentiation. (B) RNase protection using total RNA from neurosphere-derived cells attached to poly-l-ornithine- and fibronectin-coated dishes and cultured in the presence (GF+) or absence (GF−) of EGF and bFGF for 48 h. Jam-B expression is higher in undifferentiated NSCs than in differentiated NSCs. (C) Restricted expression of Jam-B in highly purified HSCs. Hematopoietic cells recovered from bone marrow, thymus, and spleen were sorted by FACS, and semiquantitative PCR was carried out to detect JAM-B transcripts. β-Actin was used as an internal control, while total RNA from ESCs was used as a positive control.

TABLE 2.

Quantitation of mRNAs for JAM-A, -B, and -C by real-time PCR

Tissue samplea	mRNA levelb
	JAM-A		JAM-B		JAM-C
	WT	KO	WT	KO	WT	KO
Brain	1.43 ± 0.08	1.14 ± 0.06	1.30 ± 0.10	NDc	3.34 ± 0.28	3.33 ± 0.16
Lung	10.89 ± 1.04	10.41 ± 0.96	0.58 ± 0.04	ND	0.50 ± 0.04	0.48 ± 0.04
Heart	13.99 ± 1.12	15.69 ± 0.85	9.27 ± 0.42	ND	3.92 ± 0.22	3.89 ± 0.18
Liver	17.35 ± 1.44	11.97 ± 1.01	0.59 ± 0.04	ND	ND	ND
Spleen	1.19 ± 0.06	0.99 ± 0.09	0.11 ± 0.08	ND	0.70 ± 0.07	0.45 ± 0.04
Kidney	14.61 ± 1.24	14.51 ± 1.06	0.25 ± 0.01	ND	3.25 ± 0.24	2.95 ± 0.26
Skeletal muscle	4.54 ± 0.24	5.13 ± 0.44	2.10 ± 0.12	ND	2.64 ± 0.17	2.99 ± 0.25
Testis	7.05 ± 0.55	6.33 ± 0.52	0.58 ± 0.03	ND	5.21 ± 0.26	4.63 ± 0.37
9.5-dpc whole embryo	3.59 ± 0.29	4.31 ± 0.36	0.75 ± 0.05	ND	6.92 ± 0.58	6.85 ± 0.36
12.5-dpc whole embryo	1.74 ± 0.13	1.54 ± 0.13	0.50 ± 0.03	ND	3.56 ± 0.26	3.12 ± 0.30
14.5-dpc whole embryo	2.79 ± 0.22	3.07 ± 0.25	0.51 ± 0.04	ND	4.60 ± 0.22	4.93 ± 0.35
16.5-dpc whole embryo	4.46 ± 0.41	4.72 ± 0.24	0.49 ± 0.03	ND	3.25 ± 0.31	3.04 ± 0.21

^aAdult tissues were obtained from 8-week-old-male mice.

^bData are expressed as percentages of the expression level of β-actin in each sample. All data are means ± standard deviations for three independent experiments.

^cND, not detectable.

Construction of Jam-B targeting vector.

For generating targeting vectors, two genomic clones containing portions of the Jam-B locus were isolated from a lambda phage C57BL/6 genomic DNA library. These two genomic clones are overlapping, with the first covering the region from introns 1 to 3, and the second covering intron 2 to the end of the gene. A 9.2-kb HindIII fragment isolated from the first clone was used for the upstream homologous region, while a 5.6-kb EcoRI fragment from the second genomic clone was used for the downstream homologous region. An internal ribosome entry site (IRES) β-geo reporter gene (24, 41) and Diphtheria Toxin A (44) were used for positive and negative selection, respectively. The targeting vector was constructed using a pBR322-based vector so that the IRES β-geo reporter gene was flanked by upstream and downstream regions of Jam-B (see Fig. 3A). A targeting vector bearing the puromycin resistance gene under the control of the phosphoglycerate kinase (PGK) promoter was constructed in a manner similar to that described above, using the puromycin resistance cassette in place of the IRES β-geo cassette.

FIG. 3.

Targeted disruption of the Jam-B locus. (A) Restriction maps of wild-type, PGK-puromycin, and IRES-β-geo knock-in alleles of the Jam-B gene. The 5′ and 3′ probes were located outside the homology arms of the targeting vector, and the Puro and Neo internal probes were used to detect insertion of selection cassettes. Restriction enzymes indicated are as follows: RI, EcoRI; Hd, HindIII; Sh, SphI; Sp, SpeI; RV, EcoRV. (B) Southern blot analyses of ESC genomic DNA. EcoRI-, SphI-, SpeI-, and EcoRV-digested tail DNA was hybridized to 5′, 3′, Puro, and Neo probes, respectively. Expected sizes of genomic DNA fragments are as follows: 5′ probe, 20 kb for wild-type and 13 kb and 11.6 kb for PGK-puromycin and IRES-β-geo knock-in alleles, respectively; 3′ probe, 14 kb, 8.6 kb, and 9.3 kb for wild-type, PGK-puromycin, and IRES-β-geo knock-in alleles, respectively; Puro probe, 10 kb for PGK-puromycin knock-in allele; Neo probe, 9.4 kb for IRES-β-geo knock-in allele.

Generation of Jam-B mutant ESCs.

To generate Jam-B mutant ESCs, the Jam-B locus was first targeted with the puromycin resistance targeting vector using E14tg2A ESCs (38). After confirmation of homologous recombination by Southern blotting, positive clones were electroporated with the IRES β-geo targeting vector. After collecting 17 clones in which homologous recombination occurred with the second targeting vector, additional Southern blots were performed to determine whether the vector targeted the wild-type Jam-B locus or the previously targeted Jam-B locus carrying the puromycin resistance cassette (see Fig. 3B).

Teratoma formation.

Jam-B mutant as well as wild-type ESCs (1 × 10⁷ cells/0.1 ml phosphate-buffered saline) were independently injected subcutaneously in nude mice as described previously (27).

Immunohistochemistry.

For immunohistochemistry of teratoma, the tumors were recovered from nude mice and fixed with 4% paraformaldehyde. The tumors were steeped in 30% sucrose solution overnight and then embedded in optimal cutting temperature compound. The frozen sections (thickness, 10 μm) were incubated with anticytokeratin (mixture of clones C-11, PCK26, CY-90, KS-1A3, M20, and A53-B/A2; Sigma), -α-fetoprotein (clone MAB1368; R & D Systems), -cardiac troponin I (clone 19C7; Fitzgerald), or -MYOD (sc-780; SANTA CRUZ) antibody. Appropriate Alexa Fluor dye-conjugated secondary antibodies from Invitrogen were used to visualize immunostaining. For immunostaining of forebrain and spinal cord, the same procedures as described above were applied using developing brains from 12.5-dpc embryos and 14.5-dpc whole embryos, respectively. The frozen sections were incubated with anti-Nestin (clone RAT401; BD pharmingen), -MAP2 (clone HM-2; Sigma), or -phosphohistone H3 (clone 6G3; Cell Signaling Technology) antibody. Alexa-594 Fluor dye-conjugated antibodies were used as secondary antibodies. For immunocytochemistry, cells were cultured in slide chambers coated with poly-l-ornithine and fibronectin. Cells were fixed with 4% paraformaldehyde for 20 min at room temperature. Cells were then subjected to immunostaining procedures as described above. Anti-MAP2 and -anti-glial fibrillary acidic protein (GFAP) (ab929; abcam, Cambridge, United Kingdom) antibodies were used to detect neurons and astrocytes, respectively.

Generation of Jam-B homozygous mutant mice.

The IRES β-geo targeting vector (100 μg) was linearized with NotI and electroporated into 1 × 10⁷ TT2 ESCs (45) as described previously (27). Electroporated ESCs were grown under G418 selection (300 μg/ml) for 1 week. Surviving ESC colonies were picked and expanded in 24-well tissue culture dishes. Once the cells became confluent, half of each clone was used for a cell stock, while the remaining half was further expanded. Genomic DNA was recovered from the expanded cells and used for Southern blot analysis using three probes (Fig. 3A). Two independently targeted ES cell clones were injected into eight-cell-stage embryos to generate chimeric mice. Chimeras were bred to generate F₁ heterozygous animals, which were intercrossed to generate Jam-B homozygous mutant mice.

Genotyping of mice.

Genomic DNA was extracted from tail tips of mice and was used for genotyping by Southern blotting or PCR. For Southern blot analyses, 10 μg of genomic DNA was digested with an appropriate enzyme (see Fig. 3B), electrophoresed on a 1% agarose gel, and transferred onto a nylon membrane. Membranes were hybridized with radiolabeled Neo, 5′, or 3′ genomic probes. For PCR, two sets of primers that amplify the deleted region of the Jam-B locus (550 bp) and a portion of the neomycin resistant gene (700 bp) were added together in each reaction. The sequences of these primers are as follows: Jam-B locus, 5′-CAGGTGCCTGAATTGATAGCTGCAGAACCC-3′ and 5′-CAGCC AGAGC AGAAAGCTTGCTGATCAC-3′; neomycin resistant gene, 5′-GAACTGCAGGACG AGGCAGCGCGGC-3′ and 5′-TATGAATTCCGAAGCCCAACCTTTCATAG-3′.

Northern blot analysis.

Northern blot analyses were done as described previously (30) using poly(A)⁺ RNAs from wild-type and Jam-B mutant 12.5-dpc whole embryos. The filters were hybridized with either exon I-II, exon VI-X, or Neo probes. The exon I-II and exon VI-X probes encompass the sequence from −206 to +135 and that from +589 to +1103 of JAM-B mRNA, respectively, in which the adenine nucleotide of the translation initiation codon is +1.

Hematopoietic indices and FACS analysis of blood cells.

Peripheral blood and bone marrow cells from wild-type and Jam-B mutant mice were recovered from the retroorbital plexus and the tibia/femur, respectively. Approximately 200 μl peripheral blood was mixed with 10 μl 100 mM EDTA (pH 7.5) to prevent coagulation, and half of each sample was applied to an automatic counter (Nihon Kohden, Tokyo, Japan) to determine the hematopoietic index. Mononuclear cells were recovered from the remaining blood samples by lysing erythrocytes with ammonium chloride and treating with 1 μl Fc blocker (BD Biosciences) at room temperature for 10 min. Subsequently, cells were incubated with biotinylated anti-Ly5.2, followed by fluorescein isothiocyanate-conjugated streptavidin. The cells were simultaneously stained with allophycocyanin (APC)-conjugated anti-Gr-1 and phycoerythrin (PE)-conjugated anti-Mac-1 or PE-conjugated anti-CD4, APC-conjugated anti-CD8, and PE-Cy7-conjugated anti-B220 antibodies. All antibodies were purchased from BD Biosciences. Multicolor analysis was performed using a FACS Vantage machine. Mononuclear cells were also prepared from bone marrow cells, and the number of CD34-negative KSL cells was quantitated as described by Osawa et al. (31).

Colony formation assay.

Mononuclear cells from bone marrow were recovered as described above and overlaid with sodium metrizoate. Low-density cells were harvested, and lineage-positive cells were removed from these cells by utilizing biotinylated antilineage markers (Mac-1, Gr-1, B220, CD4, CD8, and TER119). These cells were applied to a FACS Vantage machine, and CD34-negative KSL cells were individually sorted to 96-well tissue culture dishes. These cells were cultured for 2 weeks with medium supplemented with stem cell factor (SCF), interleukin 3 (IL-3), thrombopoietin (TPO), and erythropoietin (EPO).

Histological staining and FACS analyses of testis.

Testes were recovered from 8-week-old wild-type and Jam-B mutant mice, fixed in Bouin's fix, and embedded in paraffin. Five-micrometer sections were prepared and were subjected to hematoxylin and eosin (H&E) staining. Staging of the seminiferous epithelium during spermatogenesis was determined as described by Russell et al. (35). For FACS analyses, testes were recovered from 8-week-old mice and treated with collagenase (1 mg/ml) for 15 min at 37°C. Subsequently, the samples were treated with a mixture of trypsin (0.2%) and DNase I (1.4 mg/ml) for 15 min at 37°C and then dissociated into a single-cell suspension with a pipette. Cells were incubated with anti-GFRα1 (sc-6156; SANTA CRUZ) and biotin-conjugated anti-CD9 (clone KMC8; BD Bioscience) antibodies. These cells were further incubated with Alexa-488 Fluor dye-conjugated secondary antibodies from Invitrogen and APC-labeled avidin and then subjected to FACS analyses using a FACSCalibur device.

RESULTS

Jam-B is highly expressed in undifferentiated ESCs, NSCs, and HSCs.

A number of genes that are common to ESCs, NSCs, and HSCs have been identified by DNA microarray analysis (13, 33). To examine the roles of such putative “stemness” genes in maintaining the stem cell properties, we first performed DNA microarray analysis using RNA from undifferentiated and differentiated ESCs. The ZHBTc4 cells used carry inactivated Oct-4 alleles as a result of gene targeting but harbor a tTA-inducible Oct-4 transgene (28), allowing these cells to maintain stem cell state in the absence of tetracycline (Tc) but not in the presence of Tc. We focused on genes with high expression levels in ESCs but low levels in differentiated cells. As shown in Fig. 1 (upper panel), we identified a number of genes whose expression scores were significantly higher in undifferentiated ESCs (Tc⁻) than in differentiated cells (Tc⁺), with some showing a greater than 30-fold increase in signals in undifferentiated cells than in differentiated cells (Fig. 1, lower panels). Jam-B, one of the stemness genes identified by Ivanova et al. (13), is included among them.

FIG. 1.

Microarray analysis of undifferentiated and differentiated ESCs. Upper panel, scatter plot of mouse Gene Chip data. The seven diagonal lines show n-fold changes of 2, 3, and 10, with the middle line corresponding to equivalent expression scores between undifferentiated and differentiated ZHBTc4 ESCs. The arrow indicates the Jam-B transcript. Red spots correspond to transcripts present in both samples, while yellow spots indicate transcripts that are absent or at low levels in both samples. Blue spots represent transcripts present in either undifferentiated or differentiated cells. Lower panels, enlargement of the portion of scatter plot indicated by the triangle and a list of genes and/or expressed sequence tags whose hybridization scores are more than 30-fold higher in undifferentiated ESCs than in differentiated cells. Acc. no., sequence accession no.

To confirm that Jam-B is specifically expressed in undifferentiated ESCs, RNase protection assays were performed using RNA from undifferentiated and differentiated ESCs. For these experiments, we used E14tg2a ESCs (38), which are wild type for the Oct-4 locus, and induced differentiation by embryoid body formation in the absence of leukemia inhibitory factor (see references 15 and 34 for details). As shown in Fig. 2A, we confirmed that Jam-B expression is high in undifferentiated ESCs. However, expression decreased abruptly during early stages of differentiation. At later stages of differentiation (day 20), expression was again elevated, likely reflecting the differentiation of cells into various tissue types, such as heart and skeletal muscle, which are known to express Jam-B (3, 8). We also examined the expression levels of markers for undifferentiated ESCs, including Oct-4 (25), Rex-1 (6), and Utf1 (26, 27, 30), as well as the extraembryonic endodermal marker Gata4 (10, 28), to monitor the differentiation state of the ESCs. From these analyses, we noted that Jam-B expression declined even more rapidly than that of pluripotent ES cell marker genes during differentiation of ESCs.

Next, we investigated Jam-B expression in neural stem/progenitor cells. Primary neurospheres generated from 12.5-dpc mouse embryonic brains were dissociated and grown on dishes coated with poly-l-ornithine and fibronectin. The cells were cultured for an additional 48 h in the presence or absence of EGF and bFGF, which are required for preserving neural stem/progenitor identity (Fig. 2B). Like Nestin (14), Jam-B was strongly expressed in neurosphere cells cultured in the presence of EGF and bFGF, while markers of differentiation, such as Gfap and Map2, showed reduced expression compared to that observed in cells cultured without EGF and bFGF.

Next, we examined Jam-B expression in hematopoietic cells. HSC-enriched CD34-negative KSL cells (31), as well as a variety of other hematopoietic cell lineages, were isolated from either mouse bone marrow, thymus, or spleen by FACS, and RNA was prepared from sorted cells. Subsequently, reverse transcription-PCR was performed using a mixture of primer sets for Jam-B and β-Actin mRNA. As shown in Fig. 2C, Jam-B expression is detected only in CD34-negative KSL cell populations. These data indicate that JAM-B is specifically expressed in HSCs but not in their differentiated derivatives, similar to the results seen for ESCs and neural stem/progenitor cells. However, the expression level of Jam-B in HSCs is likely not very high, since we were unable to detect Jam-B expression in a lineage-negative cell population from which CD34-negative KSL cells were derived.

Generation of Jam-B mutant ESCs.

Next, we characterized the function of Jam-B by targeted disruption of the locus. First, we attempted to disrupt both alleles of Jam-B in ESCs to assess whether the gene is required for the maintenance of the stem cell state in ESCs. We prepared two targeting vectors, one of which carries a puromycin resistance cassette behind the PGK promoter and the other of which contains an IRES β-geo (24) cassette (Fig. 3A). Both were designed to replace exons 3, 4, and 5 of Jam-B, which encode two immunoglobulin-like folds that are required for cell-cell contact (3, 8, 16, 19). We first introduced the puromycin targeting vector into E14tg2A ESCs by electroporation. PCR and Southern blot analysis revealed that the targeting vector correctly disrupted the Jam-B locus in 3 out of 75 independent puromycin-resistant clones. Subsequently, one of these three clones (Jam-B^Puro/+) was used for subsequent targeting using the IRES β-geo cassette. Among 21 G418-resistant colonies, we found that 17 were correctly targeted. This high frequency of recombination was likely due to the function of the IRES cassette in the vector, since it is assumed that the vector is able to support the reporter gene expression only in the case that the vector is integrated into one of the genes expressed in ESCs, including the JAM-B gene, but not into the genomic portions containing no functional gene or genes which are not expressed in ESCs (for details, see reference 24). In 13 clones, homologous recombination occurred in the previously targeted Jam-B allele, leaving the other allele intact (Jam-B^geo/+), but in four cases, the vector disrupted the remaining wild-type allele (Jam-B^Puro/geo) as desired (Fig. 4A).

FIG. 4.

Generation and characterization of Jam-B null ESCs. (A) Summary of trials of double-knockout Jam-B in ESCs. At first, one of the Jam-B loci was disrupted with the vector bearing the puromycin resistant gene (Jam-B^Puro/+). After confirmation of homologous recombination of the vector at the Jam-B locus, the second targeting vector carrying the IRES-β-geo reporter was incorporated by electroporation in such Jam-B heterozygous ESCs. Genomic DNAs were recovered from 21 independent G418-resistant ESC colonies, and the DNAs were used to determine by Southern blot analyses whether the second vector was incorporated outside of Jam-B locus (Jam-B^Puro/+), homologously integrated in the previously targeted Jam-B allele with the first targeting vector (Jam-B^geo/+) or the remaining wild-type allele (Jam-B^Puro/geo). These data were also confirmed by the sensitivities of the ESCs to puromycin and G418. (B) Morphology of wild-type (left panel) and Jam-B mutant (right panel) ESCs. (C) Expression analysis of pluripotent ESC markers (Oct-4, Nanog, and Sox-2) and Jam family members in wild-type (WT), heterozygous (Het), and homozygous (KO) ESCs by RNase protection. β-Actin was used as a control. (D) Doubling time of wild-type and Jam-B mutant ESCs. Cells (1 × 10⁴) were transferred to tissue culture dishes. Subsequently, cells were trypsinized at 48, 60, 72, 84, 96, or 108 h posttransfer, and the total number of cells was determined. The doubling time was obtained by plotting data on semilogarithmic graph paper. The data were obtained from three independent experiments with comparable results. (E) Expression of cardiac and skeletal muscle cell marker genes during differentiation of Jam-B mutant ESCs. The Jam-B mutant ESCs were differentiated in vitro by inducing embryoid body formation as for Fig. 2A. RNAs were recovered and used to examine expression levels of the indicated genes. Tnni1, slow skeletal muscle troponin I; Tnni3, cardiac troponin I. (F) H&E staining of teratomas generated by subcutaneous injection of wild-type (WT) or Jam-B mutant (KO) ESCs in nude mice. The bar corresponds to 100 μm. (G) RNase protection to analyze gene expression levels in teratomas derived from wild-type and Jam-B mutant ESCs. (H) Immunohistochemical analyses of teratomas derived from Jam-B knockout ESCs. Teartoma sections were immunostained with the indicated antibodies. Sections were counterstained with 4′,6′-diamidino-2-phenylindole.

Jam-B mutant colonies exhibited typical ESC colony morphology consisting of densely packed cells (Fig. 4B) and expressed normal levels of Oct-4, Nanog, and Sox-2 (Fig. 4C), which are known to play crucial roles in maintaining the stem cell state of ESCs (5, 7, 22, 25). Moreover, Jam-B mutant ESCs showed normal cell growth rates that were equivalent to that of wild-type ESCs (Fig. 4D) and could be passaged and expanded for at least 3 months without losing typical ESC colony morphology (data not shown). When these cells were induced to differentiate by embryoid body formation in the absence of leukemia inhibitory factor, down-regulation of Oct-4 and up-regulation of Gata4 were observed (Fig. 4E), as was seen with wild-type ESCs (Fig. 2A). Since Jam-B is known to be expressed in cardiac and skeletal muscles, we examined whether generation of such muscle cells was impaired during differentiation of Jam-B mutant ESCs. However, we found that expression of both skeletal (Tnni1) and cardiac (Tnni3) muscle-specific genes was properly up-regulated during the differentiation (Fig. 4E), and magnitude of induction of both genes was comparable to that of wild-type ESCs (data not shown). Next, to examine the teratoma-forming activity and pluripotency of Jam-B mutant ESCs, we subcutaneously injected wild-type or Jam-B mutant ESCs into nude mice. Jam-B mutant ESCs were able to efficiently produce teratomas with multiple differentiated cell types, such as epithelial tissue and muscle, that were indistinguishable from those produced by wild-type ESCs (Fig. 4F). Moreover, Jam-B mutant ESCs could give rise to endodermal, ectodermal, and mesodermal cells, as shown by RNase protection for markers of these lineages (Fig. 4G). These data were also confirmed by immunohistochemistry showing that cytokeratin (ectoderm)-positive, α-fetoprotein (endoderm)-positive, cardiac troponin I (cardiac muscle)-positive, and MYOD (skeletal muscle)-positive cells were evident in teratomas from Jam-B mutant (Fig. 4H) and wild-type (data not shown) ESCs, and these cell populations were organized. From these results, we conclude that Jam-B is not required for preserving the stem cell state of ESCs.

Jam-B knockout mice have no obvious abnormalities.

Next, to elucidate the in vivo functions of Jam-B, we disrupted the Jam-B locus using the earlier-described IRES β-geo targeting vector and used these ESCs to make chimeric mice. Germ line transmission of the targeted allele was obtained from two independent ESC clones. Southern blot analyses shown in Fig. 5B revealed that heterozygous intercrosses resulted in homozygous mutant mice in normal Mendelian ratios (Fig. 5A), with both male and female homozygous mutant mice being fertile with no apparent abnormalities (data not shown). Next, Northern blot analyses were performed using RNAs from wild-type and mutant 12.5-dpc whole embryos. The size of wild-type JAM-B mRNA is 2.9 kilobases, while the calculated size of mRNA from mutant Jam-B allele is 5.4 kilobases (exons I and II of the Jam-B gene plus IRES-β-geo). As shown in Fig. 5C, we found that both expected mRNAs were visible with a probe bearing exon I and II (left panel). Moreover, transcripts from wild-type and Jam-B mutant alleles were also detected with exon VI-X (right panel) and Neo (middle panel) probes, respectively, and no unexpected band corresponding to an abnormal isoform of the JAM-B transcript was detected in any lanes of the three panels, validating our assumption that the Jam-B gene is indeed disrupted in homozygous mutant mice. However, we cannot completely eliminate the possibility that the transcript from the Jam-B mutant locus may express certain JAM-B functions, since the transcript contains a portion of the Jam-B coding sequence.

FIG. 5.

(A) Analyses of Jam-B heterozygous intercross progeny. Jam-B heterozygotes were intercrossed, and their progeny at 6 weeks old was genotyped by Southern blotting. (B) Southern blot genotyping of mice. EcoRI-, SphI-, and EcoRV-digested tail DNA was hybridized to 5′, 3′, and Neo probes, respectively. Expected sizes of genomic DNA fragments are as follows: 5′ probe, 20 kb for wild-type and 11.6 kb for targeted locus; 3′ probe, 14 kb for wild-type and 9.3 kb for targeted locus; Neo probe, none for wild-type and 9.4 kb for targeted locus. (C) Northern blot analyses of transcripts from wild-type and mutant Jam-B loci. RNAs were prepared from wild-type and Jam-B mutant 12.5-dpc whole embryos. After fractionation by agarose gel electrophoresis, RNA (0.7 μg of poly(A)⁺ RNA in each lane) was transferred to a filter and hybridized with the exon I-II, exon VI-X, or Neo probe, in which the roman numeral corresponds to the exon number of the Jam-B gene. Open and solid arrowheads indicate the positions of 28S and 18S ribosomal RNAs, respectively.

JAM-B is not required for self-renewal and multipotency of NSCs.

In order to determine whether JAM-B plays a significant role in maintaining the stem cell state in NSCs, we generated neurospheres from 12.5-dpc wild-type and Jam-B mutant mouse brains. We found that Jam-B mutant brains contain a number of neurosphere-forming cells comparable to that of wild-type brains (Fig. 6A and B). Next, to examine the differentiation potential of Jam-B mutant neurosphere cells, differentiation was induced by withdrawing bFGF and EGF from the culture medium and cells were stained with antibodies against neural cell markers. As shown in Fig. 6C, neurospheres from both wild-type and Jam-B mutant mice were able to produce both neurons (MAP2-positive) and glia (GFAP-positive) efficiently upon differentiation. In addition, we generated neurospheres from 14.5-dpc embryos in which overt differentiation of the subventricular zone is evident. We performed these experiments clonogenically. That is, dissociated cells were cultured at clonal density (1,000 cells/well) and were permitted to develop into primary neurospheres. Under these conditions, more than 95% of the neurospheres generated were clonal, as revealed by mixing experiments using enhanced green fluorescent protein- or enhanced cyan fluorescent protein-positive cells (data not shown). To assess the self-renewal capacity of Jam-B mutant neurosphere cells, 16 independent primary neurospheres of equivalent size (about 0.2 mm in diameter) derived from wild-type and mutant mice (32 in total) were chosen from cultures and the efficiencies of secondary neurosphere formation were compared. Jam-B mutant cells displayed activity comparable to that of wild-type cells (Fig. 6D, left panel). To evaluate the differentiation potential of Jam-B null mutant cells, 32 independent primary neurospheres with equivalent sizes were chosen as above and were induced to differentiate by culturing in the absence of growth factors. In both genotypes, 13 out of 16 neurospheres produced both neurons and glia upon differentiation, while the remaining 3 produced only glial cells (Fig. 6D, right panel). Thus, Jam-B mutant neurospheres exhibit differentiation potential comparable to that of wild-type cells. Collectively, these results indicate that the loss of JAM-B does not affect production or maintenance of neural stem/progenitor cells during central nervous system development. This conclusion is supported by immunostaining of the Jam-B mutant forebrain and spinal cord, since there were no significant differences between wild-type and Jam-B mutant mice in the expression patterns or levels of Nestin, MAP2, and mitotic cell-specific phospho-histone H3 (Fig. 6E and F).

JAM-B is not required in HSCs.

Next, we examined the function of JAM-B in hematogenesis. First, we examined peripheral blood cell populations and saw no apparent difference between wild-type and Jam-B mutant mice in the numbers of white blood cells, red blood cells, and platelets. In addition, there were no differences in hemoglobin content and hematocrit levels (Table 1). Furthermore, FACS analysis with mononuclear cells from peripheral blood revealed that there was no significant difference between wild-type and mutant mice in myeloid and lymphoid cell populations (Fig. 7A). No differences were observed in CD34-negative KSL HSC cells recovered from bone marrow (Fig. 7B). Contents of CD34-negative KSL cells were 0.0043% and 0.0041% of total mononuclear cells from bone marrow of wild-type and Jam-B mutant mice, respectively.

TABLE 1.

Hematopoietic indices for wild-type and JAM-B^−/− mice

Mouse groupa	Blood contentb
	White blood cells (102/μl)	Red blood cells (104/μl)	Hemoglobin (g/dl)	HCT (%)	MCV (fl)	MCH (pg)	MCHC (g/dl)	Platelets (104/μl)
WT	52.3 ± 21	699.0 ± 139	9.4 ± 1.7	38.1 ± 6	54.7 ± 2.3	13.3 ± 0.3	24.6 ± 0.5	74.0 ± 14.3
KO	42.0 ± 3.6	683.0 ± 4	9.6 ± 0.1	38.7 ± 0.9	56.7 ± 1.7	14.0 ± 0.3	24.7 ± 0.3	65.2 ± 5.4

^aWT, wild type; KO, knockout.

^bAll data are means±standard deviations for three 8-week-old male mice.

FIG. 7.

Characterization of hematopoietic cell population for Jam-B null mice. (A) Distribution of peripheral white blood cells in Jam-B mutant mice. Mononuclear cells from peripheral blood of wild-type or Jam-B null mutant mice were analyzed by FACS. Representative FACS profiles obtained from CD45-positive, propidium iodide-negative cells are shown. Upper left panel, T cells that are positive for CD4 and/or CD8. Upper right panel, B220-positive B cells. Lower left panel, Gr-1-positive granulocytes. Lower right panel, Mac-1-positive monocytes. No difference between wild-type and Jam-B mutant cells was observed. Data were obtained from six independent mice. (B) FACS profiles of CD34-negative, Lin-negative bone marrow cells. Bone marrow cells were recovered from wild-type or Jam-B mutant mice, and CD34-negative, Lin-negative cells were examined for Sca-1 and c-Kit expression. Data were obtained from six independent mice. (C) Colony formation assay of CD34-negative KSL cells. CD34-negative KSL cells (31) were recovered from bone marrow cells from which lineage-positive cells had been depleted, and cells were individually sorted into 96-well tissue culture dishes containing medium supplemented with SCF, IL-3, TPO, and EPO using a FACS Vantage machine and were cultured for 2 weeks. The number of cells per colony was estimated based on colony size, and cells generating colonies that appear to contain more than 10⁴ were categorized as colony-forming cells with highly proliferative potential (HPP-CFC).

Finally, we compared the colony-forming abilities of CD34-negative KSL HSCs from wild-type and mutant mice. Bone marrow cells were obtained from 8-week-old wild-type or mutant male mice, and CD34-negative KSL cells were purified after depletion of cells positive for lineage marker. The purified CD34-negative KSL cells were directly sorted into a 96-well plate containing regular cell culture medium supplemented with SCF, IL-3, TPO, and EPO at a density of one cell/well using a FACS Vantage SE system (Becton Dickinson), as described by Takano et al. (36). Cells were cultured for 2 weeks. We found that CD34-negative KSL cells from mutant mice and wild-type CD34-negative KSL cells were equally multipotent (Fig. 7C), with colonies formed in 80 out of 96 wells for mutant cells, compared to 77 out of 96 wells for wild-type cells. About 70% of the colonies consisted of more than 10⁴ cells. Moreover, characterization of cell components revealed that about 40% of large colonies (>10⁵ cells) for both wild-type and mutant cells contained neutrophils, macrophages, erythroblasts, and megakaryocytes (data not shown). From these results, we assume that JAM-B does not play a prominent role in the maintenance of multipotency in HSCs.

Normal testis morphology in JAM-B null mice.

As described above, both male and female Jam-B mutant mice were viable and fertile. However, it was surprising that Jam-B homozygous mutant males were fertile, since it is generally believed that JAM-B/JAM-C interaction in the testis plays a pivotal role in spermatogenesis (for details, see Discussion). Therefore, we performed histological analysis of the seminiferous tubules of 8-week-old wild-type and Jam-B mutant mice. As shown in Fig. 8A, we did not see any significant difference between wild-type and mutant mice at stage II-III (upper panels), stage V (lower panels), or any other stages (data not shown). We also quantitated the amount of spermatogonia cells. These cells are subdivided into A _single, A_paired, and A_aligned spermatogonia according to their topographical arrangement on the basement membrane, and A_single and A_paired cell populations in total can be identified by FACS as GFRα1 and CD9 double-positive cells (for details, see reference 1). Our FACS analyses (Fig. 8B) revealed that again there was no prominent difference in the content of these spermatogonia cells between wild-type and Jam-B mutant mice. Thus, these results, together with the fertility of Jam-B mutant mice, indicate that spermatogenesis occurs normally in Jam-B mutant mice.

FIG. 8.

No obvious abnormalities in Jam-B mutant testis. (A) Normal testicular structure of JAM-B null mice. H&E staining of testes from 8-week-old wild-type and Jam-B mutant mice at stage II-III (upper panels) and stage V (lower panels). Yellow and green arrows indicate spermatogonia and spermatocyte, respectively. Seminiferous epithelium is also indicated. The bar corresponds to 50 μm. (B) Quantitation of spermatogonia cells of testes from 8-week-old wild-type or Jam-B mutant mice by FACS analyses. Dissociated testis cells were subjected to quantitation of spermatogonia cells as GFRα1 and CD9 double-positive cells by FACS. Data were obtained from three independent experiments.

Normal expression of Jam-A and -C in Jam-B mutant mice.

Next, we inquired into the possibility that lack of detectable phenotypes for Jam-B mutant mice is due to up-regulation in expression of the other Jam family members, Jam-A and Jam-C. RNAs were prepared from various tissues of 8-week-old male mice and embryos of variable embryonic stages and then reverse transcribed. Subsequently, these samples were used to examine the levels of gene expression of Jam-A, -B, and -C by real-time PCR. However, as shown in Table 2, we did not see any elevation of Jam-A or Jam-C expression in samples from Jam-B mutant mice compared to those from wild-type mice. In contrast, a slight but reproducible reduction of expression of Jam-A in the liver and Jam-C in the spleen was observed with mutant mice, although the molecular bases and physiological meanings of these reductions are not known at present.

DISCUSSION

Recent microarray analyses have led to the identification of putative “stemness” genes that are expressed in ESCs, NSCs, and HSCs but not in their differentiated derivatives (13, 33). It is generally assumed that many, if not all, of these genes are involved in promoting or maintaining stem cell properties, including multipotency and self-renewal. Here we address this question by characterizing one of these genes, Jam-B, by targeted disruption. We focused on this gene because our microarray analysis revealed a large difference in expression level between pluripotent ESCs and differentiated cells. We confirmed that a sizable change in expression level during differentiation is also evident in NSCs and HSCs. We also confirmed a high level of Jam-B expression in neurosphere cells, which are highly enriched for neural stem/progenitor cells, but a dramatic reduction in expression when cells are cultured without bFGF and EGF, which are required for maintenance of neural stem/progenitor cell identity. HSC-specific expression of Jam-B is rather noteworthy, since expression is restricted to the CD34-negative KSL HSC population. Since a major portion of the JAM-B protein is exposed at the cell surface (8, 9, 32), it is likely that using an antibody against the protein will provide a simple and efficient way to recover HSCs from bone marrow samples.

However, although Jam-B is specifically expressed in ESCs, NSCs, and HSCs and not their differentiated derivatives, disruption of JAM-B function did not produce any overt abnormalities in any of these three types of stem cells. Indeed, we found that Jam-B mutant ESCs exhibit normal morphology and maintain pluripotency, generating teratomas containing cells from all three germ layers. We also generated Jam-B mutant mice and found that they are completely viable and fertile. Moreover, NSCs and HSCs recovered from Jam-B mutant mice are equivalent to those from wild-type mice in their stem cell properties.

The lack of fertility defects in Jam-B mutant mice is rather unexpected, based on data recently reported by Gliki et al. (11). The JAM-B protein is expressed in Sertoli cells, which assist spermatids undergoing cytodifferentiation into spermatozoa, while JAM-C, a protein whose amino acid sequence shows similarity to that of JAM-B (2-4, 9), is present in spermatids. These two proteins are known to mediate cell-cell contacts through heterophilic complex formation (2, 17, 18), and Gliki et al. (11) demonstrated a JAM-B-JAM-C interaction at the interface between spermatid and Sertoli cells. Moreover, they have shown that disruption of Jam-C function renders homozygous male mice completely infertile due to impairment of the morphological conversion of round spermatids into spermatozoa (11). Based on these data, it has been generally assumed that Jam-B knockout mice would exhibit similar spermatogenesis defects. However, our data demonstrate that a complete loss of Jam-B function does not affect fertility of either male or female mice. Histological analysis and quantitation of spermatogonia cells by FACS demonstrated that the seminiferous tubules appear normal in Jam-B mutant mice. These results indicate that disruption of JAM-B-JAM-C interactions between Sertoli and spermatid cells is not a major cause of the spermatogenesis defects observed in Jam-C mutant mice.

In summary, we found that disruption of Jam-B function does not cause any overt defects in three different types of stem cells or in spermatogenesis. However, since our experiments examined only embryonic and young-adult stages, it is possible that a phenotype exists in older animals. Although our data suggest that JAM-B function is not required for stem cell maintenance or spermatogenesis, it is possible that functional redundancy involving other JAM protein family members compensates for the loss of JAM-B. The three JAM protein family members, JAM-A, JAM-B, and JAM-C, have high amino acid sequence similarity (2-4, 9), and the expression profile of Jam-C in stem cells resembles that of Jam-B. Like Jam-B, Jam-C is expressed in NSCs and HSCs, with expression restricted to the multipotent state of these cells (13). No such difference in Jam-C expression was observed between pluripotent and differentiated ESCs, however. The variable penetrance of the lethal phenotype (60%) in Jam-C mutants (11) also suggests that JAM-B may compensate for loss of JAM-C. Thus, our future studies will examine ESCs, NSCs, and HSCs in JAM-B/JAM-C double-mutant or JAM-A/JAM-B/JAM-C triple-mutant mice.