Identification of Integrin-α4, Rb1, and Syncytin A as Murine Placental Target Genes of the Transcription Factor GCMa/Gcm1*

Members of the GCM (glial cells missing) transcription factor family have been shown to act as master regulators in different cells during mammalian and fly development being responsible for processes including gliogenesis, hematopoiesis, placental formation, and development of the parathyroidea. In the central nervous system of flies, several target genes for GCM have been reported, namely repo, pointed, and tramtrack. In mammals, two GCM genes are known (GCMa and GCMb), but the knowledge of their target genes is very limited. Here, we present for the first time a global approach aimed to identify GCMa target genes. We found 66 genes up-regulated and 11 genes down-regulated in GCMa-deficient chorionic tissue of mice at embryonic day 9.5. Moreover, we verified by quantitative reverse transcription-PCR all 11 down-regulated genes. The two most strongly down-regulated genes, integrin-α4 and retinoblastoma (Rb1), were further analyzed by promoter studies. Additionally, we identified down-regulation of the murine syncytin A gene, which is fundamental for syncytiotrophoblast formation. Finally, we proved strong down-regulation of integrin-α4 and Rb1 transcript levels by in situ hybridization in murine GCMa-deficient placentae at embryonic day 9.5. Our data demonstrate for the first time that genes encoding key regulators of placental tissue formation and architecture are regulated by GCMa.

The GCM 2 (glial cells missing; in flies also named glide ϭ glial cell-deficient) proteins represent a small family of transcription factors characterized by a highly conserved zinc-coordinating DNA binding domain at the amino terminus composed of two subdomains each largely consisting of ␤-sheets (1,2). The interaction of the GCM DNA binding domains with their cognate octameric binding motif 5Ј-ATGCGGGT-3Ј is mediated by contacts between amino acid residues of the ␤-sheets and the major groove of double-stranded DNA (3)(4)(5)(6).
All GCM family members are involved as master regulators in key steps of differentiation processes. The prototypical GCM protein of Drosophila melanogaster turned out to be mandatory for gliogenesis (7)(8)(9). Later, GCM in flies was also shown to be responsible for the specification of the plasmatocyte/macrophage lineage during hematopoiesis (10). Apart from these main functions, fly GCM proteins are additionally important for neurogenesis in the developing visual system and the postembryonic brain and for terminal tendon cell differentiation (11)(12)(13).
Only two GCM proteins have been reported in mammals, GCMa/Gcm1 and GCMb/Gcm2. GCMa is necessary for the differentiation of trophoblasts to syncytiotrophoblasts during placental labyrinth formation (14,15). Consequently, GCMa knock-out mice die at mid-embryogenesis due to the failure of placental labyrinth formation and subsequent lack of nutrient and oxygen supply to the embryo. Accordingly, GCMa is the first transcription factor capable of initiating syncytiotrophoblast formation. The placental phenotype of the GCMa knockout mice resembles human placental pathologies such as preeclampsia or intrauterine growth restrictions associated with fetal to neonatal mortality and morbidity (14). Pre-eclampsia, clinically defined by hypertension and proteinuria (14), is still one of the main causes of maternal and perinatal morbidity. In fact, both decreased and increased placental GCMa transcript amounts have been reported in mothers with pre-eclampsia (16,17). In adult mice, additional GCMa expression sites have been reported in kidney, thymus, and brain (18,19). GCMb is required for the generation of parathyroid glands (20). Consequently, GCMb knock-out mice suffer from hypoparathyroidism characterized by affected calcium and phosphate homeostasis.
To date, several target genes of GCM have been identified in flies and in mammals. In flies, three transcription factors, encoded by the loci repo, pointed, and tramtrack, are expressed in GCM-dependent fashion and are all required for terminal differentiation of lateral glia (21)(22)(23). In addition, fly GCM is able to regulate its own expression (24,25). In flies, a number of further GCM target genes were identified by a genome-wide analysis (26). In mammals, only aromatase and syncytin are reported as placental targets of human GCMa (27,28). These targets have been of limited help for understanding placental abnormalities.
To further understand how GCMa is involved in the process of placental labyrinth formation, we conducted a comparative analysis of placental genes in the absence of GCMa in mouse placenta tissue by microarray technology. This is the first genome-wide analysis for mammalian GCMa targets. A number of genes are up-or down-regulated in GCMa-ablated placenta. We further focused on those genes that were down-regulated in the absence of GCMa in the placenta. We proved their transcriptional down-regulation by quantitative RT-PCR. The two most down-regulated genes in placenta lacking GCMa were verified by promoter and in situ hybridization studies. Our data suggest that GCMa acts at a crucial step of placental development, and its further characterization might help to understand placental pathologies in humans.

Primers Used for Quantitative RT-PCR and Plasmid
Constructs-GCMa targets derived by microarray studies were verified by quantitative RT-PCR (LightCycler, Roche Applied Science) using the 5Ј-and 3Ј-primers listed in supplemental Table 1.
Tissue Culture, Transfection, and Luciferase Assay-HEK293 cells were maintained in Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum (Invitrogen). The cells were transfected in 60-mm dishes with 1 g of luciferase reporter and 0.1 g of cytomegalovirus-driven GCMa expression vector using DNA calcium phosphate precipitates. At 48 h after transfection, cells were harvested for luciferase assays as described (30). All experiments were repeated independently at least three times.
A minimum of 40 either GCMa ϩ/ϩ or GCMa lacZ/lacZ chorions were pooled, and total RNA was extracted using TRIzol reagent (Invitrogen) (30). The total RNA was further purified by DNaseI digestion and column purification (Qiagen DNaseI-, Qiagen RNeasy kit) according to the manufacturer's instructions. After reverse transcription, first strand cDNA was used in PCR experiments. Amplification products were separated on 2% (w/v) agarose gels after 30 cycles (for ␤-actin) or 40 cycles for integrin-␣4 and Rb1. PCR products were subcloned and verified by DNA sequencing. Quantitative PCR reactions were performed using the LightCycler-FastStart DNA Master SYBR Green kit and the Light-Cycler thermal cycle system (Roche Applied Science) according to the manufacturer's instructions. Densitometrical analysis of PCR fragment band intensities using conventional PCRs was performed using the Scion Image program (Scion Corp.).
Microarray Analysis-RNA extraction and purification from a minimum of 40 placental tissues at stage E9.5 were performed as described (see above). Synthesis of double-stranded cDNA, generation of biotinylated cRNA, hybridization to Mouse Genome430_2.0 arrays (Affymetrix, Santa Clara, CA), washing, staining, and scanning in a GeneChip Scanner3000 of the arrays were done as recommended in the Affymetrix Expression_s2_manual (41). Signal intensities and expression changes were determined using the GCOS1.4 software (Affymetrix). A scaling across all probe sets of a given array to a target intensity of 1000 was included to compensate for variations in the amount and quality of the cRNA samples and other experimental variables.
In Situ Hybridization Experiments-Extra-embryonic placental tissues from mice at E9.5 were dissected and fixed in 4% paraformaldehyde, deposited in 12-well plates, and hybridized with sense or antisense RNA specific for integrin-␣4, Rb1, Dlx3, or GCMa according to protocols previously described (32).

Overview of Genes Differentially Expressed in the Absence of
GCMa in the Placenta-To identify placental GCMa target genes, we decided to analyze the pattern of genes down-regulated in the absence of GCMa. To this end, we dissected chorions of murine GCMa ϩ/ϩ or GCMa lacZ/lacZ placentae and extracted RNA for further use in microarray studies. Previous reports demonstrated normal development of GCMa-ablated chorions until embryonic stage E9.0 -9.5. Thereafter the labyrinth layer failed to form (14,15). We collected RNA from wildtype and mutant (GCMa lacZ/lacZ ) chorions at E9.5. At this embryonic stage, syncytiotrophoblast differentiation and tissue morphogenesis were not impaired in mutant chorions (14,15). We dissected Ͼ200 chorions, used the associated embryos for DNA isolation and genotyping (data not shown), and grouped the different chorions according to their genotype as GCM lacZ/lacZ , GCMa ϩ/lacZ , or GCMa ϩ/ϩ . As shown before, the genotype of the chorions followed strictly Mendelian rules of inheritance (14,15). Although no haploinsufficiency was ever observed for GCMa heterozygotes, we decided to compare for further studies only RNAs derived from GCMa lacZ/lacZ chorions with GCMa ϩ/ϩ chorions. We used Ͼ40 chorionic tissues of each genotype, extracted RNA, and performed Affymetrix microarray studies. Using stringent analysis parameters for data mining analysis (signal log ratio Ͼ1.1 or ϽϪ1.1; p value Ͻ0.001 or Ͼ0.999), 66 genes were up-regulated, and 11 genes were downregulated (Table 1; for up-regulated genes, see supplemental Table 2). Transcript amount of GCMa itself was significantly down-regulated in GCMa-deficient chorion tissue ( Table 1). The down-regulation of GCMa transcript amount is explainable by loss of GCMa mRNA parts complementary to array probes as the array probes are directed against exon 6 of GCMa, whereas exons 3-6 are deleted in GCM lacZ/lacZ tissue (15). Although integrin-␣4 mRNA level was down-regulated by only 0.76-fold on the array, we decided to further examine its level because integrin-␣4 is known to play a fundamental role during choriogenesis at E9.5 and might therefore be a target of GCMa (33). Until now, all reported GCM targets were shown to be induced by GCMs (21-23, 27, 28, 34). Therefore, we decided to focus only on the genes down-regulated in the absence of GCMa in the placenta as they might be induced as primary targets by GCMa.
Analysis of Differentially Expressed GCMa Targets by Quantitative RT-PCR-To verify the GCMa target genes identified by microarray technology, we performed quantitative RT-PCR of all 11 down-regulated candidates. As template, we generated first strand cDNA from RNAs extracted from dissected chorions of GCMa ϩ/ϩ or GCMa-deficient placentae. Most strongly down-regulated were transcript levels of integrin-␣4 and retinoblastoma (Rb1) gene (0.01-and 0.32-fold, respectively; Table  1). Previously, it was reported that human GCMa regulates the transcription of the fusogenic syncytin gene important for syncytiotrophoblast formation. By quantitative RT-PCR studies, we also observed for the first time a significant down-regulation of murine syncytin A in GCMa-deficient chorions (0.52-fold; Table 1, Fig. 1A).
Using conventional RT-PCR, we even failed to observe a PCR-amplified fragment for integrin-␣4 after gel electrophoresis (Fig. 1B). In the case of Rb1, the amount of PCR-amplified fragment was significantly reduced (Fig. 1B).
Further, we quantified the degree of down-regulation of integrin-␣4 and Rb1 transcript levels by densitometrical analysis (Fig. 1C). Note that the degree of down-regulation of integrin-␣4 and Rb1 is similar independent of using either microarray or quantitative RT-PCR experiments (compare Table 1 with Fig. 1C).

Reporter Studies and Promoter Analysis Verify Integrin-␣4 and Rb1 as Primary Targets of GCMa-Integrin-␣4 and Rb1
belong to the most strongly down-regulated genes according to our quantitative and qualitative RT-PCR data. Both genes are known to play fundamental roles during syncytiotrophoblast formation and normal labyrinth architecture in the placenta (33,35). Hence, we decided to focus on the characterization of these two genes as GCMa targets. To this end, we investigated whether the promoters of these genes are regulated by GCMa. We fused an ϳ6-kb region upstream of the ATG start codon of each gene to a luciferase reporter. After transient transfection of the reporter plasmids in cultured cells, GCMa stimulated the promoter of integrin-␣4 and Rb1 by 20.9-and 15.9-fold, respectively (Fig. 2, A and B). Assuming that integrin-␣4 and Rb1 might be primary targets of GCMa, we addressed whether the

TABLE 1 Given values represent changed transcript levels of genes down-regulated in murine GCMa-deficient placentae as detected by microarray technology and confirmed by quantitative RT-PCR
Due to the absence of syncytin A from the array, no value is given in the respective box. Murine syncytin A was examined by quantitative RT-PCR. In two cases, no PCR signals were detectable in chorionic tissue irrespective of its genotype (ND ϭ not detectable). promoters of both genes contain potential GCM binding sites. Although GCM proteins are known to interact with an octameric binding site composed of 5Ј-ATGCGGGT-3Ј, it was shown that only specific bases within the octamer are crucial for interaction (1,5). Taking this into account, we searched for the presence of less stringent octameric motifs, namely 5Ј-NTG-MKGGK-3Ј (M ϭ A, C; K ϭ G, T) in the 6-kb regions upstream of the ATGs of integrin-␣4 and Rb1 genes. In fact, several putative GCM binding sites are present within both promoters (Fig.  2, A and B).

Gene title
To find out which of these GCM binding sites are of importance for transcriptional activation, we created a series of consecutive 5Ј-deletions for each promoter, transiently transfected these reporter plasmids in cultured cells, and examined their inducibility by GCMa (Fig. 3, A-C). We detected for none of the four 5Ј-deletions of the integrin-␣4 promoter reporters the same high transcriptional activation as for the whole 6-kb fragment observed ( Figs.  2A and 3, A and B). Additionally, if we used only a 513-bp fragment upstream of the ATG of integrin-␣4 for reporter studies, we measured an 8.3-fold induction. Accordingly, our data demonstrate that the relevant GCMa-dependent transactivation sites are located within the integrin-␣4 promoter between Ϫ5022 and Ϫ2009 and downstream of Ϫ513. For the Rb1 promoter, we generated three consecutive 5Ј-deletions of the used ϳ6-kb promoter (Fig. 3A). GCMa induced all Rb1 reporter constructs except Ϫ399 to ϩ150, indicating that GCMa-dependent transactivation of Rb1 gene requires the sites at Ϫ675 and Ϫ449 (Fig. 3C).
Analysis of Spatial Expression of GCMa Targets Integrin-␣4 and Rb1 by in Situ Hybridization of Murine Placentae-So far, our data proved the impact of integrin-␣4 and Rb1 as potential placental GCMa targets by in vitro methodology. To verify both genes as GCMa targets in vivo, we decided to analyze their spatial transcription pattern by whole mount in situ hybridization of placentae. We used placentae of mice at E9.5. At this time point, no signs of impaired tissue development was observed in GCMa-deficient placentae (14,15). First, we proved the spatial transcription profile of GCMa and mammalian Distal-less homolog Dlx3, which is expressed in the labyrinthine layer in a subset of trophoblast cells (15). As reported previously, in GCMa-deficient placentae, the transcription of Dlx3 was not affected (data not shown) (15). In placental tissue from GCMa ϩ/ϩ mice, we detected GCMa in the layer of the placenta containing the labyrinthine trophoblasts as shown before (14,15). In GCMa-ablated placentae, we failed to detect any signal for GCMa transcripts (Fig. 4, A and B). In agreement with our in vitro data, we also failed to detect transcripts of integrin-␣4 and only a weak signal intensity of Rb1, arguing for rare transcription of the latter in the labyrinthine layer of placentae (Fig. 4B). We evaluated by densitometrical analysis the observed pattern and obtained less then 1% of the signal intensity found in GCMa ϩ/ϩ placentae for integrin-␣4 and 31.4% for Rb1 in GCMa-deficient placentae (Fig. 4C).

DISCUSSION
The transcription factor GCMa has been shown to play a fundamental role as master regulator conducting trophoblast  differentiation and syncytiotrophoblast formation (14,15). Accordingly, GCMa is expected to be important in different human placental pathologies. In fact, several reports correlated changed GCMa expression levels with pre-eclampsia (16,17). Additionally, GCMa was shown to induce gene expression of the human fusogenic syncytin gene, which plays a crucial role during differentiation of trophoblast to syncytiotrophoblast (27). In this respect, it fits that syncytin was also shown to be down-regulated in pre-eclampsia (36).
For better understanding of how GCMa is involved in placental pathologies, we decided to search for the first time by genome-wide analysis for mammalian placental GCMa target genes. We compared by array studies GCMa ϩ/ϩ and GCMadeficient murine chorionic tissue and identified 11 down-regulated genes in the absence of GCMa (Table 1). Human syncytin, a protein encoded by the envelope gene of a recently identified endogenous defective human retrovirus, human endogenous hetrovirus-W (HERV-W), is highly expressed in placental tissue, and as mentioned above, regulated by GCMa. Previously, a mouse syncytin homologue was not believed to exist, but recently, two murine genes of substantial homology were identified, and one of them (syncytin A) was shown to play a role in mouse placental development (37,38). In fact, we confirmed by this study that murine syncytin A is also a GCMa target (Table 1). Aromatase as another described target of GCMa in humans could be neglected in this study as, curiously, aromatase is not expressed in mouse placenta (39). Finally, we focused on the characterization of the two most strongly downregulated genes in the absence of GCMa in murine placenta, namely integrin-␣4 and Rb1.
As a matter of fact, the promoters of both integrin-␣4 and Rb1 contain a number of potential GCM binding sites. Our effort to map both promoters revealed that for integrin-␣4, regions containing GCM binding sites at positions Ϫ3252, Ϫ2458, Ϫ2355, Ϫ317, and Ϫ225 are crucial for strong induction of integrin-␣4 promoter activity (Fig. 3B). In the case of Rb1, two GCM binding sites at positions Ϫ675 and Ϫ449 are effectively regulating GCMa-dependent Rb1 gene expression as 5Ј-deletion of the region containing these two sites abolishes activation of Rb1 transcription by GCMa (Fig. 3C).
Integrin-␣4 plays a fundamental role in labyrinthine layer formation. Integrin-␣4 null mutants show two defects that result in embryonic lethality. One of them is the failure of fusion of the allantois with the chorion during placentation (33). Remarkably, the same ϳ6-kb integrin-␣4 promoter fragment used in this study was earlier fused to lacZ and shown to be expressed with the same spatial and temporal expression profile as GCMa (40). Previously, it was found that drugs being used to block integrin-␣4 function during inflammation might raise side effects such as abortions due to impaired placental integrin-␣4 expression (33). It might be worth considering whether other known drugs leading to abortions are causally associated with yet unknown impaired placental GCMa expression. Data from another study indicated that a VCAM1-integrin-␣4 interaction is necessary for the induction of GCMa expression (31). In agreement with those previous data, our data suggest that GCMa and integrin-␣4 regulate their expression in an interdependent manner.
Recently, excessive proliferation of trophoblast cells and a severe disruption of the normal labyrinth architecture in the placenta were identified as the actual cause of the embryonic lethality of Rb1-deficient mice (35). Surprisingly, our studies identified Rb1 and integrin-␣4 as targets of GCMa, placing GCMa in a very key position in development of trophoblast cells, reminiscent of its prototypical role as master regulator of gliogenesis in D. melanogaster. It appears that it is a common theme that GCM proteins regulate very important steps of developmental processes. Further studies are necessary to understand the role of GCMa in human placental pathologies.