Mammalian Twisted Gastrulation Is Essential for Skeleto-Lymphogenesis

Abstract

Dorsoventral patterning depends on the local concentrations of the morphogens. Twisted gastrulation (TSG) regulates the extracellular availability of a mesoderm inducer, bone morphogenetic protein 4 (BMP-4). However, TSG function in vivo is still unclear. We isolated a TSG cDNA as a secreted molecule from the mouse aorta-gonad-mesonephros region. Here we show that TSG-deficient mice were born healthy, but more than half of the neonatal pups showed severe growth retardation shortly after birth and displayed dwarfism with delayed endochondral ossification and lymphopenia, followed by death within a month. TSG-deficient thymus was atrophic, and phosphorylation of SMAD1 was augmented in the thymocytes, suggesting enhanced BMP-4 signaling in the thymus. Since BMP-4 promotes skeletogenesis and inhibits thymus development, our findings suggest that TSG acts as both a BMP-4 agonist in skeletogenesis and a BMP-4 antagonist in T-cell development. Although lymphopenia in TSG-deficient mice would partly be ascribed to systemic effects of runtiness and wasting, our findings may also provide a clue for understanding the pathogenesis of human dwarfism with combined immunodeficiency.

Transforming growth factor β (TGF-β) superfamily members bone morphogenetic proteins (BMPs) are critical developmental regulators. Mutations in TGF-β family ligands, receptors, and signal transducers such as SMADs are associated with a number of human diseases. TSG was identified in Drosophila as one of the seven zygotic genes that govern the fate of dorsal cells in Drosophila embryos (37). TSG encodes a secreted, cysteine-rich protein that modulates the activity of the Decapentaplegic (DPP) protein, which corresponds to vertebrate BMP-4, and mutations in TSG result in defects of dorsal midline structures called amnioserosa in Drosophila (20). In searching for essential soluble factors produced from the aorta-gonad-mesonephros (AGM) region where definitive hematopoiesis arises (21), we employed the retrovirus-mediated signal sequence trap method previously developed (16), using mRNA from the AGM region of the 10.5-day-postcoitum (dpc) mouse embryo, and isolated a mouse homologue of Drosophila TSG. In 2000, Xenopus twisted gastrulation (TSG) was found to bind directly to BMP-4 to promote BMP-4 signaling by regulating the extracellular availability of BMP-4 (25). Since the dorsoventral axis is inverted between Drosophila and vertebrates and ventralizing factor BMP-4 is essential for mesoderm formation (33) and hematopoietic stem cell (HSC) survival (4), we speculated that TSG may also be involved in ventralization and mesoderm-derived organogenesis, including hematopoiesis, in mammals. Meanwhile, four groups using fine molecular analyses reported that TSG acts rather as a BMP-4 (DPP) antagonist by forming a ternary complex of TSG, BMP-4, and BMP-4 antagonist Chordin or by collaborating with a protease, Tolloid, to generate a Supersog (truncated stable form of Sog [Chordin]) in fly, fish, and frog (6, 28, 29, 35). Thus, it is controversial whether TSG acts as a BMP-4 agonist or antagonist. It is possible that TSG functions in both ways, depending on the developmental stage and topology in vivo (17, 27). Therefore, it is intriguing to evaluate the function of TSG, not only by analyses in particular spatial and temporal situations that are limited to early embryonic development, but also by in vivo analyses as a whole in higher organisms.

Here we demonstrate that TSG deficiency in mice results in various degrees of impairment in the development of multiple organs, particularly thymus, spleen, cartilage, and bone. Interestingly, thymocyte proliferation and differentiation have recently been reported to be suppressed by BMP-4, which is generally thought to function as a mesodermal morphogen (10). These findings suggest that TSG is indispensable for mammalian immuno-osseous development and that it acts as both an agonist and an antagonist for BMP-4 signaling.

MATERIALS AND METHODS

In situ hybridization.

A 0.6-kb mouse TSG cDNA fragment from the beginning to the EcoRV site, containing most of the coding sequence, was subcloned into BamHI-EcoRV sites of pBluescript II SK (Stratagene). A sense RNA probe was synthesized by T3 RNA polymerase from the EcoRV-digested DNA template, and an antisense RNA probe was made by T7 RNA polymerase from the BamHI-digested template by using ³⁵S-UTP. Hybridization was carried out as described previously (30).

Gene targeting of TSG.

The targeting vector was constructed by inserting a 1.1-kb XhoI-SalI fragment of pMC1NeoPolyA (Stratagene) into the exonic NheI site of the TSG genomic clone (derived from mouse strain 129/SvJ and cloned in λ phage [Stratagene]) in pBluescript II, which is 20 bp downstream of the first ATG in an antisense orientation, after the NheI site had been converted to the SalI site by using a SalI linker (Stratagene). The herpes simplex virus thymidine kinase gene (tk) cassette was also inserted into the NotI site of pBluescript II for negative selection (see Fig. 3A). The vector was linearized at the 5′ end of the left arm and electroporated into E14-1 embryonic stem cells. Screening by Southern blot analysis for homologous recombination was as described previously (23). The targeting efficiency was 5 to 10%. All the mice were kept under specific- pathogen-free conditions.

FIG. 3.

Targeted disruption of TSG. (A) Maps of the TSG locus (top), the targeting construct (middle), and the targeted locus (bottom). Restriction enzymes are EcoRI (E), BamHI (B), and NheI (N). The targeting vector harbors 4.8- and 1.6-kb TSG fragments, the neomycin resistance gene (Neo) cassette derived from pMC1NeoPolyA, and the herpes simplex virus thymidine kinase gene (tk) cassette containing the same promoter as that in pMC1NeoPolyA. (B) Southern blot analysis of tail DNAs from the mice. Genomic DNAs digested with EcoRI were hybridized with the 5′ flanking probe. The 8.1- and 7.4-kb bands represent the wild-type and mutated alleles, respectively. When the blot was rehybridized with a Neo probe, only the 7.4-kb band was detected, and BamHI digests probed with a 3′ flanking exonic probe further corroborated appropriate homologous recombination (data not shown). (C) Northern blot analysis of multiple tissues from the adult wild-type mouse with a 4.0-kb TSG cDNA probe. The blot was rehybridized with a mouse β-actin probe. (D) Northern blot analysis of total RNAs from livers and kidneys of adult TSG^+/+ or TSG^−/− mice with a 0.6-kb cDNA probe, followed by rehybridization with a human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe. The positions of RNA ladder markers (in kilobases; Promega) are given on the left.

Northern blot analysis.

Mouse multiple tissue blot was purchased from Clontech and hybridized with a 4.0-kb mouse TSG full-length cDNA probe followed by rehybridization with a mouse β-actin probe as described previously (24). To confirm null mutation of TSG, total RNA was isolated from mouse liver and kidney by using TRIzol reagent (Invitrogen) and Northern blot analysis was performed as described previously (24) with a 0.6-kb TSG cDNA probe which covers the region that is 3′ to the neomycin resistance gene (Neo) cassette. The probe spans the region from the SacI site, which is 120 bp downstream of the first ATG codon of the TSG cDNA, to the BspHI site, which is 34 bp downstream of the stop codon.

TUNEL assay.

A terminal deoxynucleotidyltransferase-mediated dUTP nick-end labeling (TUNEL) assay was carried out by using the ApoAlert DNA fragmentation assay kit (Clontech) according to the manufacturer's instructions. Formalin-fixed, paraffin-embedded thymic tissue sections mounted on glass slides were used after the removal of paraffin by xylene. DNA fragmentation was detected by incorporation of fluorescein-dUTP. All the cells were visualized with propidium iodide treatment in the presence of RNase and were examined by using a fluorescence microscope (IX70; Olympus) equipped with a SenSys/OL cold charge-coupled device camera (Olympus) and IP-Lab software (Signal Analytics Co.).

Proliferation assay for thymocytes and splenocytes.

Thymocytes (5 × 10⁴) from triplicate cultures were stimulated for 24 h with either concanavalin A (ConA; 20 μg/ml), ConA plus mouse interleukin-2 (IL-2; 10 U/ml; R & D Systems), ConA plus mouse IL-7 (5 ng/ml; R & D Systems), anti-CD3ɛ antibody (clone 145-2C11; 10 μg/ml) plus phorbol 12-myristate 13-acetate (PMA; 10 ng/ml), or anti-CD3ɛ plus PMA plus IL-7, in the presence of 10% knockout serum replacement (KSR; Invitrogen) instead of fetal bovine serum (FBS), and pulsed with 0.5 μCi of [³H]thymidine for 14 h, after which the incorporation of [³H]thymidine was measured. Splenocytes were analyzed similarly with KSR or KSR plus lipopolysaccharide (10 μg/ml).

Colony-forming assay.

Bone marrow cells (2 × 10⁴) were cultured in triplicate in α-minimum essential medium (Flow Laboratories), 1.2% methylcellulose (Shin-etsu Chemical), 30% FBS (HyClone), 1% deionized fraction V bovine serum albumin (Sigma), 100 μM 2-mercaptoethanol (Eastman Organic Chemicals), and various combinations of hematopoietic growth factors (100 ng of mouse stem cell factor [Kirin Brewery]/ml, 10 ng of mouse granulocyte-macrophage [GM] colony-stimulating factor [Sumitomo Pharmaceutical Co.]/ml, 20 ng of human thrombopoietin [Kirin Brewery]/ml, 10 ng of mouse IL-3 [Amgen]/ml, 2 U of human erythropoietin [Kirin Brewery]/ml, 100 ng of human IL-6 [Tosoh Co.]/ml, and 100 U of mouse IL-7 [Toray Industries]/ml) as described previously (2, 34).

Western blot analysis.

Freshly isolated thymocytes (6 × 10⁶) were stimulated for 45 min with recombinant human BMP-4 (R & D Systems) in the presence or absence of recombinant mouse TSG (R & D Systems) in RPMI 1640 medium containing 7.5% KSR. The cells stimulated in vitro as well as untreated cells were collected by centrifugation, lysed in 200 μl of the sample buffer (62.5 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate, 10% glycerol, 5% 2-mercaptoethanol, 0.02% bromophenol blue), sonicated, and centrifuged, and 5 μl out of the 200-μl supernatant was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The proteins were transferred to nitrocellulose, probed with an anti-phospho-SMAD1 (Ser 463/465) polyclonal antibody (Upstate Biotechnology), and visualized with the enhanced chemiluminescence detection system as described by the manufacturer (Amersham).

Reverse transcription-PCR analysis.

Total RNA was isolated from the fresh thymocytes by using TRIzol reagent, 10 μg of RNA was reverse transcribed with SuperscriptII (Invitrogen), and 1/30 of the reaction mixture was subjected to 25 cycles of PCR with ExTaq (Takara) at 94°C for 30 s, 63°C for 30 s, and 72°C for 30 s to amplify RUNX1 cDNA or 21 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s to amplify β₂-microglobulin (β₂-MG) cDNA. The primers used are as follows: 5′RUNX1, 5′-CCAGCAAGCTGAGGAGCGGCGA-3′; 3′RUNX1, 5′-CCGACAAACCTGAGGTCGTTGAATCTCG-3′; 5′β₂-MG, 5′-ATGGCTCGCTCGGTGACCCTA-3′; 3′β₂-MG, 5′-TCATGATGCTTGATCACATGTCTCGATCC-3′.

RESULTS

Targeted disruption of TSG.

The mouse TSG cDNA we isolated from the AGM region is 4.0 kb in length and encodes 222 amino acids of TSG protein which are identical to those published by other groups (9, 28, 29). In situ hybridization of the 11.5-dpc mouse embryo disclosed that TSG mRNA exists in the aortic wall, the gonad-mesonephros region, the area where vertebrae will develop, and interstitial tissues in general (Fig. 1). Expression of the TSG mRNA was also investigated in the 17.5-dpc sections of the lung, thymus, and kidney (Fig. 2). TSG mRNA was present on the alveolar and bronchial epithelial cells in the lung, in interstitial tissues in the thymus, particularly in the medulla, and in the cells surrounding the tubules in the kidney. To delineate the physiological roles of TSG in mammalian development, we have generated TSG-deficient mice by gene targeting. We constructed a TSG targeting vector to disrupt the first coding exon covering the initial 40 amino acids, which should create an allele that is null for expression (Fig. 3A). The construct was electroporated into E14-1 embryonic stem cells, and three independent clones were injected into C57BL/6 blastocysts to create chimeric mice. Chimeric mice were backcrossed to C57BL/6 mice to make heterozygous (+/−) mice, and +/− mice were bred to generate homozygous mutant (−/−) mice. Genotyping (Fig. 3B) of 100 progeny yielded 27 wild-type (+/+) mice, 47 +/− mice, and 26 −/− mice; this expected ratio (1:2:1) indicated no embryonic lethality. Northern blot analysis on the adult mouse tissues revealed the 4.1-kb TSG transcripts in heart, lung, liver, and kidney (Fig. 3C). The mRNA harboring the TSG coding sequence downstream (3′) of the inserted Neo gene cassette was hardly detected by Northern blot analysis in two different tissues from TSG-deficient mice, whereas 4.1-kb TSG mRNA was detected in the wild-type mice (Fig. 3D). In addition, a putative fusion gene transcript (TSG-antisense Neo-TSG), if existing in a trace amount, could harbor no open reading frames because of the presence of many stop codons. Furthermore, because the insertion of the Neo cassette disrupts the signal sequence (amino acids 1 to 24) of the TSG, any protein translated from mRNA generated by aberrant splicing or cryptic promoter lacks the signal sequence. This raises virtually no possibility for the production of secreted molecules from the mutated allele. Altogether, the targeted allele possesses a null mutation of TSG.

FIG. 1.

TSG expression in the AGM region. Sagittal sections of the 11.5-dpc wild-type mouse embryo with in situ hybridization of TSG are shown. White spots represent the hybridization signals. (A) Probed with a sense strand of TSG RNA. (B to D) Probed with an antisense strand of TSG. (C and D) Partial magnifications of panel B with hematoxylin-eosin (HE) staining. V, an area where vertebrae will emerge; A, aorta; GM, gonad-mesonephros.

FIG. 2.

TSG expression in lung, thymus, and kidney in the 17.5-dpc wild-type mouse embryo. Each tissue section was subjected to in situ hybridization with a TSG antisense or sense riboprobe and HE staining. Magnification is equal in all the panels.

Growth retardation in TSG-deficient mice.

Although 12.5% of the −/− mice died on the day of birth, the rest of the −/− mice looked healthy at birth except that they were 10 to 20% smaller than their +/+ and +/− littermates (which will be described later), and this difference became evident with age (Fig. 4). Severe growth retardation was observed in more than half of the neonatal −/− mice, followed by a sick appearance and subsequent death or sudden death in a few cases (Fig. 5). Some of the sick −/− mice had closed eyes even at 3 weeks of age, an uncoordinated gait, and whole-body tremors before death. The cause of death in TSG^−/− mice was not uniform except that, for all such mice, it included wasting. All the neonatal −/− mice had milk spots that were comparable in size to those of their +/+ or +/− littermates, indicating proper nursing. Autopsies revealed that some of the sick mice had pulmonary fibrosis or intra-abdominal or subarachnoid hemorrhages in addition to defective development of the thymuses and spleens (described later). Independent of the presence of the sick appearance, nearly half of the −/− mice had apparently kinked tails (Fig. 6A). The kinky tail, reported in several cases of gene-disrupted mice, including BMP-7 null mice (7, 15, 18), was observed as early as day 4 after birth in TSG^−/− mice. An X-ray photograph revealed the partial disappearance of intervertebral disks, with calcification in the kinky tails of the −/− mice (Fig. 6B and C).

FIG. 4.

Gross appearance of a TSG^−/− mouse and a TSG^+/+ littermate at 15 days of age.

FIG. 5.

Growth retardation and mortality in TSG^−/− mice. (A and B) Growth profiles of TSG^+/+, TSG^+/−, and TSG^−/− mice with age. TSG^−/− mice were classified into two groups, healthy and sick, by their appearances. All of the profoundly small mice died or were sacrificed within a month of birth when they looked severely sick. (A) Naso-anal lengths (+/+, n = 17; +/−, n = 25; −/−, n = 29). (B) Body weights (+/+, n = 19; +/−, n = 21; −/−, n = 27). (C) Survival rates of TSG^+/+, TSG^+/−, and TSG^−/− mice with age. Sick mice sacrificed before natural death were excluded from the data.

FIG. 6.

Kinky tail in the TSG^−/− mouse. (A) Kinky tail in 26-day-old TSG^−/− mouse. (B) X-ray image of the tail of the TSG^−/− mouse shown in panel A and that of a TSG^+/+ littermate. (C) Partial magnification of a section of panel B. Arrowheads indicate the partial disappearance of intervertebral disks with calcification.

Retarded skeletogenesis and nephrogenesis in TSG-deficient mice.

The X-ray images (a typical pattern is shown in Fig. 7) and skeletal preparations stained with Alcian blue (for cartilages) and Alizarin red (for mineralized bones) (Fig. 8A) as well as autopsies (data not shown) of the +/+, +/−, and −/− littermates revealed the short and thin structure of the limb bones and very thin cranial bones in TSG^−/− mice, but there was no difference in the mineralized areas among these mice at 27 days of age (Fig. 8A). Histological analysis of the distal metaphyseal cartilaginous growth plates in the femurs disclosed the markedly enlarged resting zones with thin proliferating and hypertrophic zones of chondrocytes, and the epiphyseal (secondary) ossification was remarkably delayed in the −/− mice compared with that in the +/+ and +/− littermates (+/+, n = 1; +/−, n = 5; −/−, n = 3) (Fig. 8B to E), indicating that endochondral ossification in the −/− mice was reduced. Furthermore, the growth of femoral trabecular and cortical bones in the −/− mice was delayed in comparison with that in the +/− littermates at 17 to 32 days of age (data not shown). Bone densities of the femurs and lumbar vertebrae of the −/− mice with severe dwarfism at 32 days of age were nearly half those of the +/+ littermates, and a tendency toward mild reduction in bone density was also observed in long-surviving −/− mice (Fig. 9). These phenotypes are similar to those of the transgenic mice expressing a truncated dominant-negative type IB BMP receptor driven by the osteoblast lineage-specific promoter (36). These findings suggest that TSG deficiency results in dwarfism and osteopenia due to delayed endochondral ossification caused by a blockade of the transition of chondrocytes from the resting to the proliferating state together with reduced intramembranous ossification resulting from the attenuated BMP signaling. Thus, it is likely that TSG acts preferentially as a BMP-4 agonist during the development of the skeletal system. In addition to the impaired osteogenesis, TSG^−/− mice also showed histologically poor development of the kidneys, with small and immature structures of the glomeruli, compared with the +/+ and +/− littermates (Fig. 10) (+/+, n = 2; +/−, n = 5; −/−, n = 4). This finding is reminiscent of the impaired kidney development in mice deficient for BMP-7, the activity of which as a mesoderm inducer is higher when it is forming a heterodimer with BMP-4 than when it is in the form of a homodimer (31), although it is possible that correct kidney development is a consequence of the accumulation of a variety of signaling pathways.

FIG. 7.

X-ray image of a TSG^−/− mouse and a TSG^+/+ littermate at the age of 22 days. Each bone is thinner in the −/− mouse than in the +/+ mouse.

FIG. 8.

Dwarfism with delayed endochondral ossification in TSG^−/− mice. (A) Skeleton of 27-day-old TSG^−/− mouse and those of littermates. The skeletons were stained with Alizarin red and Alcian blue. (B and C) HE-stained sections of distal epiphyses of femurs in 17-day-old TSG^+/− and TSG^−/− mice, respectively. (D and E) Partial magnifications of panels B and C, respectively, showing metaphyseal cartilaginous growth plates. Resting (R), proliferating (P), and hypertrophic (H) zones of the growth plates are indicated.

FIG. 9.

Bone densities of TSG^+/+, TSG^+/−, and TSG^−/− mice. Bone mineral densities of the right femurs and lumbar vertebrae L2 to L4 were measured roentgenographically by using DCS-600EX-II (Aloka). +/+, n = 7; +/−, n = 3; −/−, n = 9. Three −/− male mice (one 189-day-old mouse and two 224-day-old mice) and 16 female mice (of various genotypes) were analyzed.

FIG. 10.

Immature structure of the glomerulus of the kidney in TSG^−/− mouse. HE-stained sections of the kidney from TSG^+/+ (A) and TSG^−/− (B) mice at 32 days of age are shown.

Depletion of lymphoid cells in TSG-deficient mice.

We isolated mouse TSG cDNA from the AGM region. BMP-4 was also found to be expressed in the human AGM region (19). Although BMP-4 cannot expand numbers of pluripotent hematopoietic repopulating cells, unlike its upstream regulator Sonic hedgehog (Shh) (3), BMP-4 can maintain them (4). We were therefore interested in the effects of TSG deficiency on hematopoiesis. Peripheral blood cell counts and smears from the −/− mice showed severe reductions of lymphoid cell numbers, moderate decreases of platelets, and mild decreases of erythrocytes compared to those from +/+ mice. On the other hand, relative percentages of granulocytes and monocytes were increased (Fig. 11). Consistent with this finding, more than half of the −/− mice showed very poor development of the thymuses and spleens. These organs were surprisingly found to contain extremely small numbers of cells, representing reductions of around 3,000-fold, in some of the −/− mice; whose body weights differed only two- to fivefold from those of the +/+ or +/− littermates (Fig. 12A to D). There was a positive correlation between the levels of growth retardation and lymphoid deficiency, raising a possibility that lymphoid deficiency is a secondary event. However, numbers of thymocytes and splenocytes in the −/− mice at birth were already decreased by two- to fourfold and two- to eightfold, respectively, compared with those in the +/+ or +/− littermates (Fig. 12C and D), suggesting that lymphoid deficiency is not due simply to stress from runtiness and wasting after birth. Histologically, the white pulp areas were decreased in the −/− spleens (data not shown), and most cells in the −/− thymuses showed an apoptotic appearance while the +/+ thymocytes did not (Fig. 12E and F). TUNEL analysis disclosed that many thymocytes from the −/− sick mouse but not those from the +/+ mouse underwent DNA fragmentation to various degrees (Fig. 12G and H). Flow cytometry analysis of the thymocytes revealed increased percentages of CD4- and CD8-single-positive cells and decreased percentages of double-positive cells in the −/− mice compared with those in the +/+ littermates (Fig. 13, upper panels). Total numbers of bone marrow cells were not very much reduced in the −/− mice if the numbers are normalized according to body size, except for those in a few cases (data not shown). CD43/B220 and B220/immunoglobulin M (IgM) double staining of the bone marrow cells demonstrated the dramatic reduction of pro-B (CD43⁺ B220^low IgM⁻), pre-B (CD43⁻ B220^low IgM⁻), and immature B (CD43⁻ B220^low IgM⁺) cells, with retention of mature B cells (CD43⁻ B220^high IgM⁺), in the −/− mice relative to those in the +/+ littermates (Fig. 13, middle four panels). These findings suggest that the expansion of progenitor B and T cells is impaired and that lymphoid differentiation itself is not blocked in TSG^−/− mice. However, splenocytes and thymocytes from the −/− mice showed normal mitogenic activity in vitro; upon stimulation with polyclonal activators, such as lipopolysaccharide, ConA, and anti-CD3 antibody plus PMA, or with IL-7 in culture with KSR (namely, in the absence of fetal bovine TSG), they proliferated at levels comparable to or rather enhanced over those of splenocytes and thymocytes from the +/+ mice (Fig. 14). Thus, lymphoid deficiency in TSG^−/− mice does not seem to result from intrinsic defects in B and T cells but from abnormalities in microenvironments, such as stroma cells and cytokine production and distribution. The percentage of CD11b⁺ Gr-1⁺ myeloid cells was increased in the −/− bone marrow cells, indicating the enrichment of mature granulocytes (Fig. 13, lower panels). Clonal culture assay of bone marrow cells from 7-month-old mice (+/+, n = 3; −/−, n = 3) disclosed that the formation of GM colonies, megakaryocyte (Meg) colonies, erythroid bursts, erythrocyte-Meg (E-Meg) colonies, and mixed hematopoietic colonies (GM-Meg, GM-erythrocyte, or GM-E-Meg) was reduced in TSG-deficient mice (Table 1). Interestingly, the efficiency of B-lymphoid colony formation was not decreased in the −/− mice (Table 1). This is consistent with the findings shown in Fig. 14. The same assay with mice at ages of 25 to 28 days (+/+, n = 3; −/−, n = 3) also gave similar results (data not shown). Although we cannot exclude the possibility that bovine TSG, which may exist in the FBS that was used for the colony assay (KSR was not used because of its very low efficiency for colony formation), acted to enhance the growth and differentiation of the TSG^−/− cells in vitro, the assay indicates that TSG^−/− bone marrow cells retain the B-lymphoid progenitors at a level comparable to that of the +/+ marrow cells. On the other hand, −/− bone marrow cells have reduced numbers of progenitor cells of the other lineages.

FIG. 11.

Lymphoid deficiency in peripheral blood of TSG^−/− mice. (A) Analysis of peripheral blood in TSG^−/− and TSG^+/+ littermates. WBC, white blood cell; RBC, red blood cell; Hb, hemoglobin; PLT, platelet. Three typically ill-looking −/− mice with severe dwarfism, four healthy-looking −/− mice with mild dwarfism, and four +/+ littermates were analyzed at 12 to 28 days of age for their peripheral blood. Each bar pattern represents the same subgroup of mice in panels A and B. (B) Differential counts of WBCs. The same samples described for panel A were analyzed.

FIG. 12.

Defective lymphoid development in TSG^−/− mice. (A and B) Freshly isolated spleens and thymuses from a 27-day-old TSG^−/− mouse and a TSG^+/+ littermate. (C and D) Comparison of splenocyte and thymocyte numbers among TSG^+/+, TSG^+/−, and TSG^−/− mice at different ages. (C) +/+, n = 9; +/−, n = 10; −/−, n = 13. (D) +/+, n = 7; +/−, n = 6; −/−, n = 13. (E and F) HE-stained sections of thymuses from a 15-day-old TSG^+/+ mouse (E) and a TSG^−/− littermate (F). (G and H) TUNEL assay using fluorescein-dUTP (green) on the cells visualized with propidium iodide (red). The tissue sections shown in panels E and G derived from the same paraffin-embedded specimen. The same is true for the sections shown in panels F and H.

FIG. 13.

Flow cytometry analysis of thymocytes and bone marrow cells from a 27-day-old TSG^−/− mouse and a TSG^+/+ littermate. Thymocytes (Thy) and bone marrow cells (BM) were analyzed on a FACScalibur flow cytometer (Becton Dickinson). A pattern typical of results from four independent experiments is shown.

FIG. 14.

Intact proliferation of TSG-deficient lymphocytes in vitro. Thymocytes and splenocytes from 26-day-old TSG^+/+ and TSG^−/− littermates were stimulated for 24 h with various polyclonal activators, and [³H]thymidine incorporation was measured after pulse-labeling. LPS, lipopolysaccharide.

TABLE 1.

Colony-forming efficiencies of 2 × 10⁴ bone marrow cells from TSG^+/+ and TSG^−/− mice at the age of 7 months^a

Hematopoietic growth factors	Mouse	No. of colonies						Total
		GM	Meg	B-E	E-Meg	Mix	B-Ly
SCF + IL-3 + IL-6 + GM-CSF + EPO	+/+ no. 1	112 ± 17	8 ± 6	8 ± 1	6 ± 3	33 ± 10		167 ± 12
	+/+ no. 2	85 ± 16	9 ± 4	11 ± 5	5 ± 1	18 ± 4		127 ± 16
	+/+ no. 3	124 ± 14	8 ± 2	5 ± 2	3 ± 1	17 ± 8		157 ± 18
	−/− no. 1	62 ± 13	3 ± 1	4 ± 3	2 ± 2	10 ± 6		81 ± 15
	−/− no. 2	53 ± 8	3 ± 2	2 ± 2	1 ± 1	10 ± 2		68 ± 7
	−/− no. 3	54 ± 4	4 ± 3	2 ± 0	0	7 ± 4		67 ± 7
SCF + TPO + EPO	+/+ no. 1	83 ± 6	13 ± 5	6 ± 3	1 ± 1	4 ± 2		107 ± 3
	+/+ no. 2	50 ± 7	11 ± 1	9 ± 4	1 ± 1	6 ± 1		78 ± 4
	+/+ no. 3	50 ± 7	11 ± 4	10 ± 2	3 ± 2	8 ± 6		82 ± 5
	−/− no. 1	28 ± 5	5 ± 2	2 ± 3	0	4 ± 2		39 ± 5
	−/− no. 2	30 ± 7	2 ± 1	4 ± 2	0	4 ± 2		40 ± 7
	−/− no. 3	25 ± 6	3 ± 2	2 ± 1	0	2 ± 2		33 ± 7
SCF + IL-7	+/+ no. 1	67 ± 8					47 ± 9	114 ± 13
	+/+ no. 2	34 ± 4					17 ± 2	51 ± 5
	+/+ no. 3	73 ± 13					69 ± 3	142 ± 16
	−/− no. 1	31 ± 6					24 ± 6	55 ± 11
	−/− no. 2	41 ± 3					73 ± 6	114 ± 5
	−/− no. 3	31 ± 5					53 ± 6	84 ± 10

^aAbbreviations: SCF, stem cell factor; GM-CSF, GM colony-stimulating factor; EPO, erythropoietin; TPO, thrombopoietin; B-E, burst-forming units for erythroid; Mix, mixed hematopoietic colonies; B-Ly, B-lymphoid colonies.

Increased levels of transcripts of RUNX1 with enhanced phosphorylation of SMAD1 in TSG-deficient thymocytes.

In order to elucidate the molecular mechanisms of TSG function, downstream events of BMP-4 signaling were investigated. The status of serine phosphorylation of SMAD1 was examined in thymocytes from the +/+ and −/− littermates at the age of 10 days. As shown in Fig. 15A, SMAD1 was already phosphorylated in thymocytes in vivo, and further stimulation of these cells with BMP-4 in the absence or presence of recombinant TSG in vitro scarcely affected the phosphorylation status of SMAD1. However, the level of phosphorylation of SMAD1 in vivo in the −/− cells was significantly higher than that in the +/+ cells. This suggests that TSG acts as a BMP-4 antagonist from the perspective of phosphorylation of SMAD1 in thymocytes in vivo. Consistent with this finding, the expression of RUNX1, the product of which functionally binds to SMAD1 (13) and is predicted to stimulate transcription of RUNX1, was up-regulated in the −/− thymocytes relative to that in the +/+ thymocytes in vivo (Fig. 15B).

FIG. 15.

Enhanced phosphorylation of SMAD1 with increased levels of RUNX1 transcripts in TSG^−/− thymocytes. (A) Western blot analysis of thymocytes from +/+ and −/− mice at the age of 10 days with an anti-phospho-SMAD1 antibody. Freshly isolated thymocytes (labeled as in vivo) were cultured for 45 min in the medium containing 7.5% KSR plus 100 ng of human BMP-4/ml (labeled as BMP-4) or KSR plus 100 ng of human BMP-4/ml and 1 μg of mouse TSG/ml (labeled as BMP-4 + TSG) before harvest. The cell lysates corresponding to 1.5 × 10⁵ cells were loaded onto each lane. Results of two independent experiments (Exp 1 and Exp 2) are shown. The left panel demonstrates the result for the thymocytes in vivo only. (B) Reverse transcription-PCR analysis of thymocytes from +/+ and −/− mice at the age of 37 days. RUNX1 mRNA is detected as a PCR product of 292 bp. The integrity of the RNA was confirmed by the detection of β₂-MG transcripts as a product of 373 bp.

Variability in the severities of the phenotypes of TSG-deficient mice.

The phenotypes of the TSG-deficient mice vary among the individuals; some TSG-deficient mice can survive and generate progenies (both male and female TSG^−/− mice are fertile), but more than half of the TSG-deficient mice display severe dwarfism due to delay in ossification, lymphoid deficiency with depletion of lymphoid progenitor cells, and retarded kidney development and die within a month after birth by wasting and/or additional diseases, such as pulmonary fibrosis. The reason for the variability in the phenotypes of TSG^−/− mice may be derived from the unbalanced distribution of the soluble factors, including BMPs and Chordin, in the microenvironment in the absence of TSG and also from the modifications in the BMP signaling resulting from the difference in genetic backgrounds. One may speculate that maternal TSG would rescue the −/− embryos in utero with different local availabilities. But this possibility was excluded because the mating of surviving TSG^−/− mice with mild phenotypes produced −/− pups which were phenotypically indistinguishable at birth from −/− pups derived from +/− parents (Table 2). The −/− parents with mild phenotypes produced −/− mice with severe phenotypes that became apparent after birth, as well as pups with mild phenotypes (data not shown).

TABLE 2.

Neonatal growth profiles of TSG^−/− mice and those of controls

Genotype (no.)	Body weight (g)			Naso-anal length (cm)
	Day 0	Day 4	Day 9	Day 0	Day 4	Day 9
−/− from −/− parents (11)	1.3 ± 0.1	2.5 ± 0.2	4.4 ± 0.2	2.5 ± 0.2	3.5 ± 0.2	4.1 ± 0.1
−/− from +/− parents (7)	1.3 ± 0.1	2.6 ± 0.2	4.3 ± 0.2	2.5 ± 0.1	3.6 ± 0.1	4.0 ± 0.1
+/+ or +/− from +/− parents (21)	1.5 ± 0.1	3.5 ± 0.2	6.0 ± 0.2	2.9 ± 0.1	4.0 ± 0.1	4.9 ± 0.1
+/+ from +/+ parents (11)	1.6 ± 0.1	3.4 ± 0.2	6.0 ± 0.2	3.0 ± 0.1	4.1 ± 0.1	5.0 ± 0.1

DISCUSSION

TSG is expressed in the AGM region, and TSG-deficient mice displayed defective lymphoid development with moderate thrombocytopenia and mild anemia. BMP-4 at low concentrations is known to induce the proliferation and differentiation of CD34⁺ CD38⁻ Lin⁻ human HSCs, whereas BMP-4 at higher concentrations promotes the survival of HSCs (4). TSG is hence speculated to function as an agonist for mesoderm inducer BMP-4 at the early stage of hematopoiesis. However, BMP-4-mediated regulation of hematopoiesis is more complex. It was shown that Shh, the upstream regulator of BMP-4, is produced by the thymic stroma, and Shh arrests thymocyte differentiation at the double-negative stage through its receptors Patched and Smoothened, which are expressed in double-negative thymocytes (26). Recently, BMP-4, produced by the thymic stroma, has also been shown to inhibit the proliferation and differentiation of the thymocytes, and TSG, of which expression in thymocytes is induced by T-cell receptor signaling, has been demonstrated to synergize with Chordin to block the BMP-4-mediated inhibition of thymic development (10). Therefore, TSG as a BMP-4 antagonist is supposed to be a positive regulator of thymocyte development. Our findings on the impaired development of the thymus in TSG^−/− mice before and after birth are consistent with these results in vitro. We also found that TSG^−/− thymocytes show enhanced phosphorylation of SMAD1 and increased levels of transcripts of RUNX1. If TSG functioned as a BMP-4 antagonist, TSG would suppress the phosphorylation of SMAD1 and the absence of TSG would elevate the basal level of SMAD1 phosphorylation. Our finding thus supports the notion that TSG acts as a BMP-4 antagonist in thymocytes. Recently, RUNX1 has been demonstrated to be required for active repression of CD4 through CD4 silencers in CD4⁻ CD8⁻ thymocytes (32). In accordance with this finding, transgenic expression of RUNX1 was reported to skew thymocytes into the CD8-single-positive lineage but not to drive the maturation of CD8-single-positive cells (14). Nevertheless, it is possible that TSG functions not only by modulating BMP-4 signaling but also by acting as a growth and differentiation factor for unidentified targets that would affect the stroma function, since TSG has a weak homology to a mitogenic peptide connective tissue growth factor (5) which is a modulator of signaling by TGF-β as well as BMP-4 and is involved in a variety of processes, including angiogenesis, skeletogenesis, and wound healing (1, 12, 22). Thus, our finding of extremely small thymuses in TSG-deficient sick mice could be explained by complex effects of dysregulated expression of RUNX1 as well as modulation of the other unknown target genes and also by the systemic problems affecting the viability of the thymocytes.

TSG has recently been proposed to exert both agonistic and antagonistic functions sequentially on BMP signaling; first, TSG forms a ternary complex with BMP and the full-length Chordin to prevent the binding of BMP to its receptor, and second, once all Chordin is cleaved by Xolloid (human BMP-1, Drosophila Tolloid), TSG promotes BMP signaling by competing for the binding to BMP with Chordin fragments that still retain the anti-BMP activity (25). Thus, the turning off and on of BMP signaling can be controlled sharply by TSG (17). Our findings in TSG^−/− mice suggest that fine control of BMP activity by TSG as a molecular switch is essential for proper development of multiple organs. It is possible that dually functional TSG elicits different BMP signaling patterns during development in different species depending on the local concentrations of BMP, Chordin, TSG, Xolloid, and other related molecules. TSG-deficient mice can be used to unveil the molecular mechanism of the development of early hematopoietic progenitor cells, thymus, spleen, cartilage, bone, and kidney that is regulated by hedgehog-BMP-SMAD signaling and/or putative BMP-independent signaling.

The human TSG gene is mapped to chromosome band 18p11.3, to which no immuno-osseous diseases have so far been linked, although the holoprosencephaly-4-associated gene TGIF (TG-interacting factor, SMAD2-interacting homeodomain protein) is located within 5 Mb of the TSG locus (9, 11, 29). The phenotype of the sick TSG^−/− mice is similar to the manifestation of human early lethal short-limbed skeletal dysplasia with severe combined immunodeficiency (8). Since the pathogeneses of most human immuno-osseous diseases have not yet been molecularly defined, TSG-deficient mice will also be useful for elucidating the underlying mechanisms for the development of these diseases.

In summary, the phenotype of dwarfism and lymphoid deficiency in TSG-deficient mice suggested the dual functionality of TSG in vivo, as a BMP-4 agonist for skeletogenesis and a BMP-4 antagonist for thymocyte development. Also, another function of TSG as a family of connective tissue growth factors remains to be clarified in future studies. In either case, mammalian TSG is essential for proper development of mesodermal organs.