Increased Activity of Hypoxia-Inducible Factor 1 Is Associated with Early Embryonic Lethality in Commd1 Null Mice^▿

Generation of the target vector and transgenic mice.

The genomic sequences that flank Commd1 exon 2 at the 5′ and 3′ ends were amplified with sense primer 5′-TCTAGACCCCAGATCAAGATACTGATTG-3′ and antisense primer 5′-TCTAGAGAGCCAGCTCTTCAGTGATCAG-3′ and with sense primer 5′-GGATCCTCTAGAGCTACATGTCTTTGCAGTTGTC-3′ and antisense primer 5′-CTCGAGCAGAACTATTCTTCACAATTGGAG-3′, respectively. The amplified products were separately cloned into the pCR2.1 TA cloning vector (Invitrogen, Breda, The Netherlands). The 5′ target sequence (1.8 kb) and the 3′ target sequence (2.1 kb) were subcloned into the target vector (pMCneo) using the XbaI restriction site for the 5′ target sequence and the BamHI and XhoI restriction sites for the 3′ target sequence (Fig. 1A). The target vector contained the neomycin (neo) and thymidine kinase (tk) genes for positive and negative selection, respectively. The XhoI-linearized targeting vector was electroporated into the 129Sv/Ev embryonic stem (ES) cell line (33). Selection for neomycin resistance and thymidine kinase was carried out, and two correctly targeted ES cell lines, cell lines 7 and 62, were injected into blastocysts collected from C57BL/6J mice according to standard procedures to generate two independent mouse lines. ES cells and F₁ animals were genotyped by PCR using primers localized outside the target sequences and primers localized inside the neo gene. A 5′ fragment (2.3 kb) was amplified using sense primer 5′for (5′-CTGGTATTAAATGAGAATGCATG-3′) and antisense primer 5′rev (5′-CCTGCGGCAATCCATCTTG-3′), and a 3′ fragment (2.4 kb) was amplified using sense primer 3′for (5′-CCTCGTGCTTTACGGTATCG-3′) and antisense primer 3′rev (5′-GAGTGGTCCATTTCAATTGCTG-3′) (Fig. 1A and B). Genomic DNA isolated from tails (19), embryos, or yolk sacs (30) was used for genotyping by duplex PCR using sense primers Ex2 (5′-CAG ATA CAC AGC CCT GTT GC-3′), Neo (5′-CCT CGT GCT TTA CGG TAT CG-3′), and I1 (5′-GAG ACA ACT GCA AAG ACA TGT AGC-3′), yielding a 239-bp product for the wild-type allele and a 200-bp product for the targeted allele.

RNA isolation and RT-PCR.

Total RNA was isolated from embryos at 9.5 days postcoitum (dpc) by using RNAStat60 (Tel-Test Inc., Friendswood, TX) according to the manufacturer's specifications. Total RNA was treated with DNase I (Invitrogen), and approximately 1 μg of total RNA was used to generate first-strand cDNA using oligo(dT) primers and Superscript SVII (Invitrogen, Breda, The Netherlands). DNase treatment and reverse transcriptase reactions were performed according to the manufacturer's specifications. Successful DNase treatment was verified by a PCR specific for genomic DNA. Reverse transcription-PCR (RT-PCR) was used to genotype the embryos by using sense primer Cfor (5′-CGCAGAACGCCTTTCACGG-3′) and antisense primer Crev (5′-TTTGCTTGACTTTAACTTCATC-3′), yielding a 453-bp product for the wild-type transcript and a 169-bp product for the mutant transcript. The expression of U2af1-rs1 in wild-type, heterozygous, and homozygous Commd1 null embryos was determined by semiquantitative PCR using sense primer U2afor (5′-TCTAATTCCCAACCAAGTTAC-3′) and antisense primer U2arev (5′-AAAACAACATGGGAAGCCAC-3′) (Fig. 1A). As an endogenous reference, the expression of 18S rRNA was determined using sense primer 18Sfor (5′-CCACATCCAAGGAAGGCAG-3′) and antisense primer 18Srev (5′-GCTGGAATTACCGCGGCTG-3′).

Copper supplementation.

Drinking water containing 6 mM CuCl₂ was provided to female Commd1^+/− mice starting 3 weeks before intercrossing, and copper supplementation continued through gestation and lactation. As described previously, these mice ingested approximately 50 to 100 times more copper than mice on a control diet (24).

Expression constructs, cell culture, transfections, and immunoblotting.

Sequences representing either full-length mouse Commd1 or exons 1 and 3 of mouse Commd1 were amplified from mouse liver cDNA using primers ex1for (5′-GGATCCGCCACCATGGCGGG CGATCTGGAGG-3′), ex1rev (5′-GCGGCCGCCTTGAGAAGTCCTCTCATC-3′), ex3for (5′-GGATCCGCCACCATGGAATCTGAATTTTTGTGTC-3′), and ex3rev (5′-GCGGCCGCGGCTGCCTGCATCAGCCTG-3′). PCR products were subcloned into pEBB-GST by using the BamHI and NotI restriction sites. The sequences of the constructs were verified by sequence analysis. Human embryonic kidney (HEK) 293T cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum, l-glutamine, and penicillin-streptomycin. Calcium phosphate precipitation was used to transfect HEK 293T cells as described previously (12). After transfection, cells were lysed in lysis buffer (1% Triton X-100, 25 mM HEPES, 100 mM NaCl, 1 mM EDTA, 10% glycerol) supplemented with 1 mM Na₃VO₄, 1 mM phenylmethylsulfonyl fluoride, protease inhibitors (Roche, Basle, Switzerland), and 10 mM dithiothreitol. Cell line extracts representing 60 μg protein were electrophoresed and then transferred to Hybond-P membranes. (These techniques are described in reference 22.) Immunoblots were probed with an anti-COMMD1 antiserum (1:2,000), an anti-glutathione S-transferase (anti-GST) antiserum (Santa Cruz, Santa Cruz, CA), or anti-β-actin monoclonal antibodies (clone AC-74; Sigma-Aldrich, Zweindrecht, The Netherlands). Normal and Commd1^−/− embryos at 9.5 dpc were used for the preparation of total homogenates with the lysis buffer containing 0.4 M NaCl, 0.1% NP-40, 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, and 400 nM NaCl. Equal amounts of protein from the embryos were transferred to Hybond-P membranes and were probed with anti-HIF-1α (NB 100-449; Novus, Huissen, The Netherlands) or anti-α-tubulin (clone DM 1A; Sigma). HIF-1α protein expression in HEK 293T cells was detected with anti-HIF-1α (clone 54; BD Bioscience, Alphen aan den Rijn, The Netherlands).

Generation of stable HEK 293T COMMD1 knockdown cells.

Plasmids encoding short hairpin RNAs were generated by cloning a target sequence specific for COMMD1 (GTCTATTGCGTCTGCAGAC) into pRETRO-SUPER as previously described (4, 5). To produce amphotropic retrovirus supernatants, Phoenix packaging cells were transfected with either empty pRETRO-SUPER or a pRETRO-SUPER construct targeting COMMD1 by using Fugene-6 (Roche, Woerden, The Netherlands) according to the manufacturer's instructions. Upon addition of 4 μg/ml Polybrene, HEK 293T cells were infected with the retrovirus supernatants and grown on a medium supplemented with 1 μg/ml puromycin (Sigma-Aldrich) to allow for monoclonal selection. Western blot analysis with an anti-COMMD1 antibody was used to quantitate the level of knockdown, typically ≥90% (P. de Bie et al., submitted for publication).

Histology and immunohistochemistry.

Embryos at 7.5 to 10.5 dpc were dissected from the decidua, and the yolk sac was removed for genotyping. The embryos were photographed and fixed for 12 h in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C. Embryos were dehydrated in ethanol and embedded in paraffin, and serial sections (thickness, 4 μm) were prepared. Sections were stained with hematoxylin and eosin. Apoptosis was detected by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining using the DeadEnd colorimetric TUNEL system (Promega, Leiden, The Netherlands). Incorporation of 5-bromo-2′-deoxyuridine (BrdU) was used to evaluate DNA synthesis during cell proliferation. Pregnant females received 50 mg of BrdU (Sigma)/kg of body weight by intraperitoneal injection. After 2 h, embryos were dissected, genotyped, and fixed as described above.

For the immunohistochemistry study, sections were deparaffinized, and primary antibodies against the mitosis marker phosphohistone H3 (Ser10) (06-570; Upstate, Veenendaal, The Netherlands) or against BrdU (ab6326, Abcam, Huissen, The Netherlands) were used to study cell proliferation. Immunoreactivity was visualized with the appropriate species-specific fluorescein isothiocyanate- or horseradish peroxidase-conjugated antibodies. Histochemical copper staining was performed using rhodamine.

Whole-mount in situ hybridization.

Wild-type embryos at different embryonic stages were dissected from the decidua, fixed for 12 h in 4% paraformaldehyde in PBS, washed with PBS-0.1% Tween 20, and dehydrated in methanol. Whole-mount in situ hybridization was performed as described previously (43). Commd1 antisense and sense probes were generated as described previously (44).

qRT-PCR.

Expression of BNIP3 mRNA in HEK 293T knockdown and control cell lines was determined by quantitative RT-PCR (qRT-PCR) on the ABI-7900 system (Applied Biosystems, Nieuwerkerk aan den Ijssel, The Netherlands) using SYBR PCR master mix (Applied Biosystems). Total RNA was isolated from HEK 293T cells using Trizol (Invitrogen) according to the manufacturer's specifications. The high-capacity cDNA archive kit (Applied Biosystems) was used to generate single-stranded cDNA from 1 μg of total RNA. We used primers 5′-AATATTCCCCCCAAGGAGTTCC-3′ and 5′-CTGCAGAGAATATGCCCCCTTT-3′ to amplify BNIP3. β-actin was used to normalize our data and was amplified using primers 5′-CATGTACGTTGCTATCCAGGC-3′ and 5′-CTCCTTAATGTCACGCACGAT-3′. PCR was initiated by heating for 2 min at 50°C and proceeded with a denaturing step at 94°C for 10 min, followed by 40 cycles of 94°C for 15 s and annealing and elongation for 1 min at 60°C. The results were expressed as n-fold changes in the mRNA expression level and as means ± standard errors of the means.

Luciferase reporter assay, GST-pull down assay, and coimmunoprecipitation.

For luciferase assays, HEK 293T cells were seeded in 24-well plates in triplicate for each treatment group. Cells were cotransfected with a HIF-1-dependent 5×HRE-luciferase reporter plasmid (36) and expression vectors encoding TK-Renilla luciferase, COMMD1-Flag (10), and/or HIF-1α. Twenty-four hours after transfection, cells were exposed to normoxic or hypoxic conditions (1% O₂) for 24 h. A microplate luminometer (Berthold, Regensdorf, Switzerland) was used to determine the luciferase reporter activity as described by the manufacturer (Promega dual-luciferase reporter assay system). Luciferase activity was normalized for Renilla luciferase activity to correct for transfection efficiency.

HEK 293T cells were cotransfected with HIF-1α-Flag and either COMMD-GST, GST, or COMMD1. Precipitations with glutathione-Sepharose were performed as previously described (10). HIF-1α-Flag immunoprecipitation was performed using FLAG M2 agarose beads (Sigma-Aldrich) according to the manufacturer's instructions. An anti-COMMD1 antiserum was added to HEK 293T cell lysates to immunoprecipitate COMMD1; this procedure was performed as described previously, with minor changes (11). Instead of protein G-agarose beads, protein A/G-agarose (Santa Cruz, Heidelberg, Germany) was used. Protein was detected by immunoblotting for GST, HIF-1α (clone 54; BD Bioscience), COMMD1, or Flag (A8592; Sigma).

Microarray data accession numbers.

The microarray data and the complete protocols used in this study have been deposited in the MIAME (minimal information about a microarray experiment) database (in progress) at http://www.ebi.ac.uk/arrayexpress under the following accession numbers: microarray layout, A-UMCU-7; data, E-MEXP-832; protocols for embryo dissection, P-MEXP-30783; RNA isolation, P-MEXP-30781; mRNA amplification, P-MEXP-30781; cRNA labeling, P-MEXP-28902/3; generating the reference pool, P-MEXP-28722; hybridization and washing of slides, P-MEXP-28890; scanning of slides, P-MEXP-28893; data normalization, P-MEXP-28894.

Generation and initial characterization of Commd1-deficient mice.

To study the function of Commd1 in vivo, we generated Commd1-deficient mice by replacing exon 2 with a neo gene. As in humans and dogs, the Commd1 gene in mice consists of three exons (Fig. 1A), so this strategy mimics the in-frame exon 2 deletion observed in Bedlington terriers affected with copper toxicosis. Three of 132 ES cell clones were correctly targeted, and 2 of these clones (clones 7 and 62) (Fig. 1B) were used to generate two independent transgenic mouse lines. F₁ heterozygous mice were generated by crossing the male chimeras with 129Sv/Ev females (Fig. 1B and D). Heterozygous mice were born healthy and were fertile; no apparent phenotype has been observed.

We confirmed by genomic PCR (Fig. 1B), RT-PCR, and immunoblotting (Fig. 1C and E) that exon 2 was correctly targeted. Analysis of 9.5-dpc embryos indicated that replacing exon 2 with the neo gene resulted in a truncated Commd1 transcript (Fig. 1C). After we had confirmed that Commd1 exon 2 was correctly targeted, we genotyped the mice by using a duplex genomic PCR with primers corresponding to exon 2 or the neo gene and a primer within intron 1 of the Commd1 gene (Fig. 1D). This in-frame deletion leads to a Commd1 null allele; no Commd1 protein could be detected in Commd1^−/− embryos, whereas Commd1 protein was readily detected in wild-type and heterozygous embryos (Fig. 1E). We confirmed that the anti-COMMD1 antisera (22) would recognize the putative in-frame deleted protein encoded by the truncated transcript (the exon 1-exon 3 predicted protein product of 11 kDa) by immunoblotting. As shown in Fig. 1F, anti-COMMD1 antisera detected the part of the Commd1 protein encoded by Commd1 exon 1 but failed to detect amino acids 154 to 188, encoded by Commd1 exon 3. Hence, a shortened in-frame protein product, if expressed, would have been detected. These combined analyses clearly demonstrated that replacing exon 2 with a neo gene leads to a complete loss of function of the Commd1 protein, similar to that observed for Bedlington terriers affected with copper toxicosis (22, 41).

Intron 1 of the mouse Commd1 locus contains the imprinted gene U2af1-rs1 (Fig. 1A) (29). The single-exon U2af1-rs1 gene is not located within either the human or the dog COMMD1 locus. For mice, it has been shown that U2af1-rs1 is transcribed mainly from the paternal allele in the adult brain and the liver and that it may interfere with transcription of the paternal Commd1 allele (44, 46). We addressed the possibility that our targeting strategy had interfered with U2af1-rs1 expression by analyzing the expression of the intronic gene U2af1-rs1 in 9.5-dpc embryos. This analysis indicated that U2af1-rs1 mRNA expression was not affected by targeting the Commd1 allele, since no difference in U2af1-rs1 expression was observed between Commd1^−/− and Commd1^+/+ embryos (Fig. 1G). Recently, a novel transcript, U2mu, which is transcribed from a part of U2af1-rs1 and exons 2 and 3 of Commd1, has been described (45). The transcript of U2mu was also truncated after replacing exon 2 of the Commd1 gene. The U2mu transcript is confined to the mouse transcriptome, is in frame with exons 2 and 3 of Commd1, and encodes a predicted protein of approximately 26 kDa (45). We and others (45) performed extensive RNA expression analysis in different mouse tissues during development, which revealed that the expression of U2mu is extremely low compared to that of Commd1. Finally, although our anti-COMMD1 antisera recognize the protein encoded by Commd1 exon 2 (data not shown), no protein of 26 kDa could be detected by anti-COMMD1 antisera in normal mice. These observations strongly imply that the U2mu transcript does not encode a functionally expressed protein. Together, these data therefore suggest that our targeting of Commd1 exon 2 affects only the specific function of the Commd1 gene.

The Commd1^+/− mice were intercrossed in order to produce Commd1^−/− mice. Surprisingly, loss of Commd1 results in a recessive embryonic lethal phenotype, since no homozygous null mice were detected among 259 offspring (Table 1). Embryonic lethality was also observed on a C57BL/6J genetic background, indicating that this lethal phenotype is independent of the genetic background of the mouse strain (E. Burstein, University of Michigan, personal communication). To characterize the timing of the embryonic lethality, we collected embryos from timed Commd1^+/− intercrosses at different stages of gestation. Commd1^−/−, Commd1^+/−, and Commd1^+/+ embryos with the expected Mendelian ratio were recovered up to 9.5 dpc (Table 1). Only a few 10.5-dpc embryos and one Commd1^−/− embryo at 11.5 dpc were recovered, but these were starting to be resorbed (Table 1; Fig. 2A), implying that Commd1^−/− embryos die between 9.5 and 10.5 dpc.

Phenotypes of the Commd1 null embryos.

To characterize the nature of the embryonic lethality, we studied the morphology of embryos from timed Commd1^+/− intercrosses at different stages of gestation. Commd1^−/− embryos were atypically small and were markedly delayed in development at 8.5 and 9.5 dpc (Fig. 2A). In spite of the developmental delay, no gross morphological abnormalities could be seen in Commd1^−/− embryos at 9.5 dpc, and they were comparable in size to normal embryos at 8.5 dpc (Fig. 2A). Heart beating was observed for 9.5-dpc Commd^−/− embryos as well as for normal embryos (both Commd1^+/+ and Commd1^+/−) at this embryonic stage, indicating that Commd1^−/− embryos are still alive at 9.5 dpc. At 9.5 days of embryonic development, the wild-type and heterozygous embryos had undergone axial rotation, but this was not seen in Commd1^−/− embryos (Fig. 2A).

The Commd1^−/− embryos were further characterized by histochemical staining of transverse and sagittal serial sections of 9.5-dpc embryos. Although their development was markedly delayed, no gross abnormalities of the primitive organs could be seen in the Commd1^−/− embryos compared to 8.5-dpc wild-type and heterozygous embryos. This was confirmed by studying the expression of the mesodermal marker Fgf8 and the somitic marker Meox1 by whole-mount in situ hybridization. No clear differences in mRNA expression of these genes were observed between 9.5-dpc Commd1^−/− embryos and 8.5-dpc wild-type and heterozygous embryos (data not shown). Primitive nucleated red blood cells were seen throughout the whole Commd1^−/− embryos, indicating that primitive hematopoiesis was not affected in Commd1^−/− embryos (Fig. 2B). Although allantochorion fusion was observed at 9.5 dpc, the labyrinth formation was retarded and vascularization of the placenta had not started in the Commd1^−/− embryos, whereas this was seen in both wild-type and heterozygous embryos (Fig. 2B). In addition, an increased number of maternal erythrocytes were seen in the placenta compared to both wild-type and heterozygous embryos (Fig. 2B).

Expression of Commd1 during normal mouse embryogenesis.

The essential role of Commd1 in mouse embryogenesis demanded a thorough analysis of the developmental expression of Commd1. We studied the expression of mouse Commd1 at different stages (8.5 to 13.5 dpc) of embryonic development by whole-mount RNA in situ hybridization (Fig. 3). At 8.5 dpc, Commd1 expression was detected throughout the embryo, with the highest expression in the headfold, allantois, and chorionic plate, but no clear expression was seen in the yolk sac (Fig. 3A). Diffuse expression of Commd1 continued during embryogenesis at stages 9.5 dpc, 10.5 dpc, and 11.5 dpc, but no expression was observed in the primitive heart (Fig. 3B to D). At later stages during development, Commd1 expression was relatively high in the neuroepithelium of the brain, the neural tube, and the epithelium of the otic vesicle (Fig. 3C and D). Expression was also clearly discernible in the caudal extremity of the tail, the limb buds, and the branchial arches. The same expression pattern was observed in 13.5-dpc embryos (Fig. 3E), but at this stage Commd1 expression was more pronounced in the interdigital regions of the hindlimb bud, the spinal cord, the caudal extremity of the tail, the developing brain, and the nose. Thus, Commd1 is ubiquitously expressed in the embryo and in the primitive placenta at 8.5 to 11.5 dpc but more distinctively at 13.5 dpc. In addition, these data clearly show that Commd1 expression is vital during the early embryonic stages of mouse development.

Analysis of cell proliferation and apoptosis in Commd1^−/− embryos.

Since controlled cell proliferation and apoptosis are essential for the development of multicellular organisms, we investigated whether affected cell proliferation or increased cell death could explain the developmental delay and ultimately the arrest of development of Commd1^−/− embryos. DNA replication in 9.5-dpc embryos was studied by determining BrdU incorporation in DNA. We observed no detectable difference in BrdU incorporation between Commd1-deficient embryos and either wild-type or heterozygous embryos (Fig. 4A). Cells in the G₂ phase of the cell cycle were studied by determining the phosphorylation status of histone H3. At 9.5 dpc, phosphohistone H3-positive cells were clearly discerned in Commd1^−/− embryos (Fig. 4B). The localization of the phosphohistone H3-positive cells (mesenchymal and epithelial cells), as well as the approximate ratios of phosphohistone H3-positive versus -negative cells, did not differ between heterozygous and wild-type embryos (Fig. 4B). Detection of TUNEL labeling was used to study cell death in 9.5-dpc embryos. An increase in the number of TUNEL-positive mesenchymal cells in the developing brain was seen for 9.5-dpc Commd1^−/− embryos compared to both wild-type and heterozygous embryos (Fig. 4C), whereas no clear difference in the number of TUNEL-positive cells could be identified elsewhere in the embryos. Immunohistochemical analysis revealed almost no cleaved capsase-3-positive cells in 9.5-dpc Commd1^−/− embryos or in wild-type or heterozygous embryos (data not shown). These combined data indicate that cell proliferation is not appreciably affected in 9.5-dpc Commd1^−/− embryos, but the increase in the number of apoptotic cells indicates that 9.5-dpc Commd1^−/− embryos are in poor vigor.

Copper homeostasis in Commd1^−/− embryos.

Since COMMD1 plays an essential role in copper homeostasis in dogs, we investigated whether embryonic copper overload could explain the lethal phenotype of Commd1^−/− mice. The possibility of copper overload in Commd1^−/− embryos or in the primitive placenta was analyzed using rhodamine staining of embryonic sections. This specific histochemical copper staining was used to identify copper overload in the livers of Bedlington terriers affected with copper toxicosis (18), but it did not show granular copper in 9.5-dpc Commd1^−/− embryos or in the primitive placenta (data not shown). These data imply that Commd1^−/− embryos do not have detectably elevated copper levels.

Previous studies have shown that severe copper deficiency leads to early embryonic death (23, 24, 27). In line with these observations, we hypothesized that inefficient copper transport from the mothers to their offspring due to the loss of Commd1 could explain the embryonic lethality of Commd1^−/− embryos. In an attempt to circumvent this potential copper deficiency, heterozygous females were provided with a high-copper diet (6 mM CuCl₂ supplied via the drinking water [24]) starting at 3 weeks before intercrossing and lasting through gestation and lactation. This intervention did not rescue the lethal phenotype of the Commd1^−/− embryos, suggesting that copper deficiency is not the main cause for this phenotype. Alternatively, Commd1 might not be essential for copper uptake even under conditions of excessive copper intake.

Genome-wide gene expression analysis of whole embryos.

A gene expression study using microarray analysis was performed to elucidate the underlying mechanism leading to embryonic lethality in Commd1^−/− embryos. Since 9.5-dpc Commd1^−/− embryos are morphologically more similar to 8.5-dpc normal embryos than to 9.5-dpc normal embryos, we compared the gene expression profiles of 9.5-dpc Commd1^−/− embryos with those of 8.5-dpc as well as 9.5-dpc wild-type and heterozygous embryos. Microarray analysis of variance resulted in a total of 236 genes with significantly different levels of mRNA expression (P < 0.05) in 9.5-dpc Commd1^−/− embryos compared to 8.5-dpc and 9.5-dpc wild-type and heterozygous embryos (Fig. 5). As shown in the Venn diagrams, the number of differentially expressed genes for 9.5-dpc Commd1^−/− embryos compared to 8.5-dpc wild-type and heterozygous embryos was lower than that for 9.5-dpc Commd1^−/− embryos compared to 9.5-dpc wild-type and heterozygous embryos (Fig. 5A). Many of the same genes differentially expressed for 9.5-dpc Commd1^−/− versus 9.5-dpc wild-type embryos (20%) were also differentially expressed for 8.5-dpc wild-type versus 9.5-dpc wild-type embryos (Fig. 5A). This demonstrates that the gene expression profiles of 9.5-dpc Commd1^−/− embryos reflect their delay in embryonic development.

Next, we reasoned that most information on the mechanisms of embryonic lethality would be provided by genes that were differentially expressed in Commd1^−/− embryos compared with wild-type and heterozygous embryos at both 8.5 and 9.5 dpc. We identified an overlap of only 16 upregulated and 2 downregulated genes between the comparison of 9.5-dpc Commd1^−/− with 9.5-dpc wild-type and heterozygous embryos and the comparison of 9.5-dpc Commd1^−/− with 8.5-dpc wild-type and heterozygous embryos (Fig. 5A). Most interestingly, 8 of the 16 upregulated genes appeared to be targets of HIF-1 (Fig. 5A and B) (also see http://humgen.med.uu.nl). The mRNA expression of eight additional HIF-1 target genes was significantly higher in 9.5-dpc Commd1^−/− embryos than in 9.5-dpc wild-type and heterozygous embryos, and that of two additional genes was significantly higher than in 8.5-dpc wild-type and heterozygous embryos (Fig. 5A and B) (also see http://humgen.med.uu.nl). Almost 10% of the significantly differentially expressed genes in 9.5-dpc Commd1^−/− embryos have previously been shown to be induced by the transcription factor HIF-1 (1, 16, 25, 35, 42) (Table 2). These genes are involved in different cellular processes, including angiogenesis and glycolysis (Table 2). These data indicate that transcriptional regulation by HIF-1 is markedly affected in Commd1^−/− embryos, which provides a possible mechanism for the embryonic lethality associated with Commd1 deficiency in mice.

HIF-1α protein expression, activity, and interaction with COMMD1.

HIF-1 is a heterodimeric basic helix-loop-helix transcription factor composed of HIF-1α and HIF-1β. HIF-1α is constitutively expressed but rapidly degraded under normoxic conditions, a process that involves oxygen-dependent prolyl hydroxylation and ubiquitination. Given the observation that Hif-1 target genes were transcriptionally upregulated in Commd1^−/− embryos, we investigated whether this was the result of higher Hif-1α protein levels in Commd1^−/− embryos than in wild-type and heterozygous embryos. Hif-1α protein expression in 9.5-dpc normal and Commd1^−/− embryos was determined by immunoblot analysis. As a control for distinguishing active Hif-1α, NIH 3T3 cells were treated with the iron chelator desferrioxamine (DFO). DFO mimics hypoxia by reducing the hydroxylation of HIF-1α via the prolyl-hydroxylases (2), resulting in a clear accumulation of active Hif-1α, which was visible as a smear, similarly to previous experiments using the same antiserum (Fig. 6A) (39). Increased active Hif-1α protein levels were also readily detected in 9.5-dpc Commd1^−/− embryos compared to 9.5-dpc wild-type embryos (Fig. 6A).

Next, we set out to investigate the relation between COMMD1 and HIF-1 activity in independent in vitro studies. To investigate whether COMMD1 deficiency leads to increased HIF-1 activity in cultured cells, reporter assays were performed with normal HEK 293T cells and with HEK 293T cells in which the expression of COMMD1 was stably knocked down to undetectable levels by RNA interference (Fig. 6B). These cells were transfected with a HIF-1-dependent reporter plasmid and were grown under normoxic or hypoxic conditions. As shown in Fig. 6B, reporter gene activity was induced by hypoxia. Consistent with our observations for Commd1-deficient embryos, COMMD1 knockdown in HEK 293T cells resulted in sixfold-higher induction of HIF-1 reporter gene activity compared to that for normal HEK 293T cells. To further corroborate these findings, overexpression of COMMD1 in HEK 293T cells resulted in marked inhibition of reporter induction under both hypoxic and normoxic conditions (Fig. 6C). To study the effect of COMMD1 deficiency on endogenous HIF-1 activity, we examined the mRNA expression of the HIF-1 target gene BNIP3 following hypoxic exposure (3% O₂). BNIP3 was one of the HIF-1 target genes, which were transcriptionally induced in Commd1^−/− embryos compared to wild-type embryos (Fig. 5B). In both cell lines, hypoxic exposure resulted in time-dependent increases in BNIP3 mRNA expression (Fig. 6D). Consistent with our previous observation, the level of BNIP3 mRNA expression was induced in HEK 293T COMMD1 knockdown cells relative to HEK 293T control cells.

Although increased Hif-1α protein expression was observed in Commd1^−/− embryos, no aberrant HIF-1α expression was seen in HEK 293T COMMD1 knockdown cells under normoxic conditions. To further investigate if COMMD1 mediates HIF-1α protein levels, we studied the stability of hypoxia-stabilized HIF-1α in HEK 293T COMMD1 knockdown and control cells. These cells were cultured for 18 h under hypoxia (1% O₂) and were subsequently exposed to 20% O₂ for different periods. Under hypoxia (time zero), both cell lines showed high HIF-1α expression, but the expression level decreased as soon as the cells were exposed to 20% O_2. Interestingly, HIF-1α expression was almost absent in control cells after 10 min of exposure to 20% O₂, whereas HIF-1α protein was still clearly observed in COMMD1 knockdown cells (Fig. 7A). Remarkably, after the cells were exposed to 20% O₂ for 2 min, HIF-1α proteins of increasingly higher molecular mass were detected in both cell lines (Fig. 7A). These higher-molecular-weight proteins most probably represent ubiquitinated HIF-1α, since high oxygen levels promote the ubiquitination of HIF-1α. These data support the in vivo data showing that COMMD1 deficiency is associated with increased HIF-1α protein stability.

Next, we investigated whether the role of COMMD1 in mediating HIF-1 activity and HIF-1α stability was regulated by a physical association between COMMD1 and HIF-1α. GST-pull-down studies were performed using lysates from HEK 293T cells coexpressing COMMD1-GST and HIF-1α-Flag. HIF-1α was clearly detected in COMMD1-GST samples precipitated by glutathione-Sepharose but not in GST precipitates, indicating a specific interaction between COMMD1 and HIF-1α (Fig. 7B). Coimmunoprecipitation studies were performed using lysates from HEK 293T cells coexpressing COMMD1 and HIF-1α-Flag or untransfected cells to confirm the physical association between HIF-1α and COMMD1. COMMD1 was clearly detected in HIF-1α-Flag immunoprecipitates but not in immunoprecipitates from cells transfected with an empty vector (Fig. 7C). HIF-1α was also observed in COMMD1 immunoprecipitates but not in control immunoprecipitates, demonstrating the specificity of the interaction between COMMD1 and HIF-1α (Fig. 7D). Taken together, these in vitro data indicate that COMMD1 modulates the protein level of HIF-1α and HIF-1 activity, possibly via a COMMD1-HIF-1α protein interaction.