ARS2 Is a Conserved Eukaryotic Gene Essential for Early Mammalian Development

Determining the functions of novel genes implicated in cellsurvival is directly relevant to our understanding of mammaliandevelopment and carcinogenesis. ARS2 is an evolutionarily conservedgene that confers arsenite resistance on arsenite-sensitiveChinese hamster ovary cells. Little is known regarding the functionof ARS2 in mammals. We report that ARS2 is transcribed throughoutembryonic development and is expressed ubiquitously in mouseand human tissues. The mouse ARS2 protein is predominantly localizedto the nucleus, and this nuclear localization is ablated inARS2-null embryos, which in turn die around the time of implantation.After 24 h of culture, ARS2-null blastocysts contained a significantlygreater number of apoptotic cells than wild-type or heterozygousblastocysts. By 48 h of in vitro culture, null blastocysts invariablycollapsed and failed to proliferate. These data indicate ARS2is essential for early mammalian development and is likely involvedin an essential cellular process. The analysis of data fromseveral independent protein-protein interaction studies in mammals,combined with functional studies of its Arabidopsis ortholog,SERRATE, suggests that this essential process is related toRNA metabolism.

INTRODUCTION

Top

INTRODUCTION

Arsenic contamination, particularly of drinking water, is ofworldwide concern, and a variety of diseases and syndromes relatedto chronic exposure to arsenic are known (reviewed in reference22). Arsenic is a carcinogen, although its mode of action isstill under active investigation (27). Despite its carcinogenicity,arsenic trioxide is used to treat relapsed acute promyelocyticleukemia and can induce complete remission in patients withoutthe myeloid suppression that accompanies other treatments (reviewedin reference 38). The reasons for the increased sensitivityof leukemic cells to arsenic are still being elucidated. ARS2(for arsenite resistance gene 2; originally called ASR2) confersarsenite resistance on arsenite-sensitive Chinese hamster ovary(CHO) cells (28). ARS2 resides within a region of human chromosome7 recurrently deleted in myeloid leukemia (40). However, itis not known whether the loss of ARS2 contributes to the increasedsensitivity of leukemic cells to arsenic.

With the notable exception of Saccharomyces cerevisiae, ARS2orthologs are found in the genomes of most eukaryotic organisms.The ARS2 protein is highly conserved; within mammals, thereis $≥$ 98% amino acid identity. An "ARS2 domain" comprising thelast 200 amino acids of the protein (Interpro no. IPR007042;Pfam no. PF04959) has been identified and exclusively assignedto ARS2 orthologs. A general feature of ARS2 orthologs is thatthey encode a single C₂H₂-like zinc finger domain in which oneof the cysteines has been lost in metazoans. There are alsomultiple bipartite nuclear localization motifs (25), which suggeststhat the protein may function in the nucleus. Despite its potentialinvolvement in arsenite resistance, its location in a genomicregion commonly deleted in acute myeloid leukemia, and its conservationin eukaryotic organisms, little is known about the functionof mammalian ARS2.

In plants, ARS2 is a developmental regulator. The ArabidopsisARS2 ortholog is known as SERRATE (SE), and se mutants developpleiotropic abnormalities during shoot development, indicatingthat it is involved in plant organogenesis (5, 13, 24, 25, 32).Overexpression of SERRATE causes accelerated plant growth (increasedleaf production and shorter time to flowering) (25). ARS2 isalso essential in zebrafish. In a large-scale zebrafish insertional-mutagenesisproject, three independent retroviral insertions occurred inARS2. One of these mutations was bred to homozygosity and resultedin an embryonic-lethal phenotype with multiple abnormalities,including central nervous system and body necrosis, small headand eyes, and pericardial edema (1, 11).

Uncovering the functions of genes implicated in the cellularresponse to arsenic is of interest, as arsenic is used to treatpromyelocytic leukemia and is being considered asa treatmentfor other cancers (37). Developmental phenotypes caused by mutationsof ARS2 orthologs in zebrafish and Arabidopsis, its presencein diverse eukaryotic genomes, and its high sequence identitybetween species suggest that ARS2 has essential functions. Forthese reasons, we investigated the role of ARS2 in mammals bycharacterizing its expression in mouse and human tissues andablating its expression through targeted disruption in the mousegenome.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Northern blot analysis. A 314-bp mouse ARS2 probe was amplified by PCR from a full-lengthARS2 clone (BC066831), using the primers Ars2_1180F and Ars2_pet30_XhoR(see Table SA1 in the supplemental material). An 880-bp humanARS2 probe was amplified from human liver first-strand cDNAusing the primers hARS2_3F and hARS2_exon9R (BD Biosciences).Both probes were gel purified prior to random primer labelingwith [ ${alpha}$ ³²P]dCTP, using the RediPrime kit (Amersham). The mouseand human probes were hybridized overnight at 65°C to commerciallyprepared Northern blots of mouse and human poly(A)⁺ RNA, respectively,using the manufacturer-supplied ExpressHyb hybridization solution(mouse MTN blot no. 7762-1 and human 12-lane MTN blot no. 7780-1;BD Biosciences). After hybridization, the blots were washedtwice in 2x SSC (1x SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1%sodium dodecyl sulfate (SDS) at room temperature for 15 mineach, followed by two 15-min washes in 0.1x SSC-0.1% SDS performedat 65°C. The blots were visualized using a Storm phosphorimager(Molecular Dynamics).

cDNA expression studies. Mouse and human expression studies of ARS2 were carried outon Clontech multiple tissue panels (BD Biosciences) using HotStarTaqDNA polymerase (Qiagen). A 50-µl volume was used for allexpression studies, and the cycling parameters were as follows:95°C for 15 min for Taq activation and eight cycles of 94°Cfor 30 s, 65°C for 1 min, and 72°C for 1 min, followedby 30 to 33 cycles of 94°C for 30 s, 58°C for 1 min,and 72°C for 1 min and a final extension of 5 min at 72°C.The mouse ARS2 PCR was performed using the primers ARSupstrFand Ars_mus1R, which amplified the expected 1,890-bp product.The human ARS2 PCR was performed with ARSupstrF and hARS-9R1,which amplified an expected 1,925-bp product. The PCR productsfrom the cDNA were verified by DNA sequencing.

RNA blotting. Human RNA master blots from Clontech contain 50 human poly(A)⁺RNA samples that have been normalized to the mRNA expressionlevels of eight different genes. RNA blots were hybridized withthe same probe used for Northern blotting. Both probes wererandomly labeled with [ ${alpha}$ -³²P]dCTP using the RediPrime randomprimer labeling kit (Amersham). The ARS2 5' probe was hybridizedovernight using ExpressHyb hybridization solution at 65°C.The blot was washed according to the manufacturer's protocol.

Production and characterization of anti-ARS2 antibodies. Antibodies were generated using the longest human ARS2 clone,BC069249. This clone is missing the first 64 amino acids ofthe predicted full-length ARS2 protein (40). Monoclonal andpolyclonal antibodies were raised to the region correspondingto the N terminus of BC069249 (amino acids 65 to 370) and weredesignated LX186.3 and XL12.2, respectively.

ARS2 protein expression. The complete coding region of ARS2 (2,628 bp) was PCR amplifiedfrom the full-length mouse ARS2 clone (BC066831) using PlatinumPfx DNA polymerase (Invitrogen). All vector inserts were confirmedto be mutation free by DNA sequencing. The full-length ARS2gene was cloned in frame into a retroviral vector derived fromthe pMSCVpuro vector (BD Biosciences) but modified to containan N-terminal three-FLAG sequence (derived from the p3XFLAG-CMVvector [Sigma]). The full-length ARS2 coding sequence was amplifiedusing the restriction enzyme-modified PCR primers Ars2_MSCVpacFand Ars2_MSCVpacR. A C-terminal Myc-His fusion protein was createdby cloning the full-length ARS2 product into the mammalian expressionvector pcDNA3.1/Myc-His(–) A vector (Invitrogen) usingthe restriction site-modified primers Ars2_mycF and Ars2_mycR.

Stable cell lines expressing ARS2 were generated using the humankidney-derived cell line BOSC-23 (ATCC CRL 11270) transfectedwith the 3XFLAG-ARS2 or ARS2-Myc-His vector. The vector (10µg) was added to 1.5 ml of serum-free Dulbecco's modifiedEagle's medium (DMEM) and mixed with an additional 1.5 ml serum-freeDMEM containing 40 µl of transfectin (Bio-Rad). This transfectionmixture was then added to BOSC-23 cells. Selection for successfultransfection events was performed with puromycin (25 µg/ml).The cells were maintained until they reached near confluenceand individual colonies were visible. Colonies were pooled andlysed in PLC lysis buffer (50 mM HEPES, pH 7.5, 150 mM NaCl,10% glycerol, 1% Triton, 1 mM EGTA, 1.5 mM MgCl₂). After celllysis, the debris was pelleted (14,000 x g; 10 min) and thesupernatant was retained for Western blotting.

Western blotting. Cell lysates were separated by SDS-polyacrylamide gel electrophoresis(10%), followed by transfer to a polyvinylidene difluoride membrane.The membrane was blocked with 5% nonfat milk (NFM) in Tris-bufferedsaline-Tween 20 (0.5%) (TBST) for 2 h. The 3XFLAG-ARS2 was detectedwith the primary anti-FLAG M2 antibody (Stratagene) used at0.5 µg/ml in 3% NFM overnight at 4°C. The ARS2-Myc-Histag protein was detected with an anti-Myc antibody (Santa CruzBiotechnology) used at 1:400 dilution. After being washed withTBST, the blots were incubated with the secondary anti-mousehorseradish peroxidase-conjugated antibody diluted 1:10,000in TBST containing 3% NFM. After being washed in TBST, the blotswere developed using the ECL-plus kit (Amersham) according tothe manufacturer's protocol.

Targeted disruption of ARS2 in R1 ES cells and mouse production. The ARS2 containing mouse chromosome 5 BAC 423o3 (AF312033 [40];CITB-CJ7-B mouse library derived from the 129Sv/J mouse strain;Research Genetics, Huntsville, AL) was digested with SpeI, andthe 8,418-bp fragment was subcloned into the XbaI site of theosdupde1 vector (kindly provided by Oliver Smithies). The 1,702-bpshort arm was amplified by PCR using Asr_20580_NheI and Asr_22282_XhoIprimers and Vent polymerase (New England Biolabs). The fragmentwas digested and cloned into the NheI and XhoI sites of osdupdeIand verified by PCR and DNA sequencing of vector/insert junctions.The targeting vector was linearized using NotI and electroporatedinto 129S1/Sv-derived R1 embryonic stem (ES) cells (21) at aconcentration of $~$ 2 µg per million cells. Electroporationwas done in electroporation buffer (Specialty Media) using aBio-Rad GenePulsar set at 250 V and 500 µF. The electroporatedcells were plated on gelatinized plates as previously described(20). G418 (150 µg/ml) was added 24 h after the electroporation;48 h after the electroporation, FIAU (5-iodo-2'-fluoro-2'deoxy-1-β-D-arabino-furanosyl-uracil)was added to the ES-DMEM at a concentration of 0.2 µM.After 7 days of selection, 192 isolated colonies were pickedand grown up individually on two 96-well plates. DNA preparationswere performed in the 96-well plates as previously described(26).

Electroporation of the vector construct into R1 ES cells yielded11 positive ES cell clones that showed the predicted band sizeshifts when analyzed by Southern blotting. These 11 clones wereexpanded, and the Southern blot analysis was repeated usingBamHI and the short-arm probe. The BamHI-digested wild-typeband detected by this probe was 9,579 bp in length, and themutant band was 5,658 bp. Further analysis was performed byEcoRI digestion in combination with the long-arm probe to ensurea correct long arm. EcoRI-digested genomic DNA produced a 14,786-bpband (wild type) or a mutant 11,997-bp band. The BamHI Southernblots were also probed with the neomycin cassette to confirmthat a single integrant was obtained. From this analysis, twoclones were chosen to make aggregation chimeras (A4-E1 and A4-G6).

Aggregation chimeras. Cleavage stage Crl:CD1(ICR) and129S2/SvPasCrl embryos (CharlesRiver, Canada) were aggregated with correctly targeted ES clonesusing the established procedure (16). Germ line transmissionwas obtained for ES cell clones A4-E1 and A4-G6. Chimera micewere bred to both 129S2/SvPasCrl and Crl:CD1(ICR) backgrounds(Charles River, Canada). Initial characterization of the knockoutmice was carried out on both strains using both cell lines.Cre-mediated excision of the neo cassette from the Ars-E1 line(ARS2^neo) was achieved by crossing the Ars-E1 line with thepCX-NLS-Cre line (19). Southern blot analysis of ARS2^neo/+ micewas performed as described above with BamHI digestion that produceda 4.4-kb mutant band. We cryopreserved two ARS2^+/neo lines designated129S2/SvPasCrl-Ars2^tm1BFK and 129S2/SvPas/CD1Crl-Ars2^tm1BFK.

PCR genotyping. PCR genotyping was performed on embryos and ear punches thathad been digested at 55°C (4 h to overnight) in 10 to 100µl ear punch buffer (50 mM KCl, 10 mM Tris HCl, pH 8.3,2.0 mM MgCl₂ 0.1 mg/ml gelatin, 0.45% Nonidet P-40, 0.45% Tween20) containing 0.5 mg/ml proteinase K. The genotyping PCR wasperformed in 25-µl reaction mixtures using the manufacturer-supplied10x reaction buffer containing 10 mM MgCl₂. Each reaction mixturecontained 0.48 µM (final concentration) of each primer,10x deoxynucleoside triphosphate stock consisting of 2 mM ofeach deoxynucleoside triphosphate and 0.625 unit of HotStarTaq(Qiagen). Primers Ars2_27244F and Ars2_27576R amplify a wild-type333-bp band. Ars2_21942F and the vector-derived primer 5066Ramplify a 563-bp band that detects the ARS2 locus mutated withthe neo cassette. Both primer sets were used simultaneouslyin the same reaction. The cycling parameters for all PCR genotypingassays were as follows: 95°C for 15 min for Taq activationand 40 cycles of 94°C for 20 s, 58°C for 1 min, 72°Cfor 1 min, and a final extension of 5 min at 72°C.

Isolation and culture of mouse embryos. Timed natural matings between ARS2^+/– mice were used toobtain embryos. The time of detection of the vaginal plug wasdesignated embryonic day 0.5 (E0.5). Blastocysts were collectedat E3.5 by flushing the uteri with M2 medium (Sigma) supplementedwith antibiotics. The blastocysts were transferred to gelatin-coated96-well, 4-well, or chambered glass coverslips containing ES-DMEMwith or without leukemia inhibitory factor (LIF). Treatmentwith acid Tyrode's solution was used to remove the zona pellucidain some experiments. The embryos were photographed every 24h using a Nikon Diaphot microscope and ImageJ software. Embryosbetween E6.5 and E7.5 were carefully dissected in M2 mediumor cold phosphate-buffered saline (PBS), rinsed several times,and then genotyped by PCR. Photographs were taken using an OlympusSZX9 microscope. Most intercrosses used to obtain E6.5 to E7.5embryos were performed between ARS2^+/neo and ARS2^+/neo mutants,since the detection of embryos with three alleles (ARS2^neo,ARS2^neo, and ARS2 wild type) would serve as an indicator ofmaternal contamination in the genotyping assay. Individual implantationsof E5.5 and E7.5 litters were fixed at 4°C overnight in4% paraformaldehyde (PFA), embedded in Epon, serially sectioned,and stained with Richardson's blue stain.

Immunofluorescence. Blastocysts were fixed in ice-cold 4% PFA for 2 minutes at roomtemperature. Permeabilization was performed with PBS-0.5% TritonX-100 for 30 min at room temperature. The blastocysts were rinsedextensively with PBS-0.1% Triton X-100 containing 6 mg/ml bovineserum albumin. Blocking was done for 1 h at room temperaturein 10% lamb serum in PBS-0.1% Triton X-100. The following primarycommercial antibodies were used: anti-OCT4 (polyclonal; SantaCruz Biotechnology; sc-9081), anti-CDX2 (monoclonal; BioGenex;CDX2-88), anti-histone H3 (phospho-S10) (monoclonal; Abcam;ab5176), anti-FLAG, and anti-Myc. Alexa Fluor 488 goat anti-mouseimmunoglobulin G (IgG) (1/400) and Alexa Fluor 568 goat anti-rabbitIgG (1/1,200) were used as secondary antibodies in PBS-0.1%Triton X-100. The nuclear morphology was visualized using thefar-red dye DRAQ5 (1:500 in PBS plus 0.1% Triton X-100; BioStatusLtd.). Embryos were imaged using the Eclipse T-E2000-U Nikonconfocal laser scanning microscope. Cell counting was performedusing Imaris software (Bitplane AG) in combination with manualverification of the automated results.

TUNEL assay. Terminal deoxynucleotidyltransferase-mediated dUTP fluoresceinnick end labeling (TUNEL) assays were performed on E3.5 blastocystsusing the In Situ Cell Death Detection kit (Roche). Blastocystswere cultured for 24 h in ES-DMEM without LIF and fixed in ice-cold4% PFA for 10 min at room temperature. The blastocysts wererinsed with PBS-0.1% Tween 20 containing 6 mg/ml bovine serumalbumin. Permeabilization was done for 10 min at room temperaturein PBS, 0.5% Triton X-100, 0.1% sodium citrate. The embryoswere rinsed again several times and then incubated with thefluorescein-dUTP and terminal deoxynucleotidyltransferase enzymefor 1 h at 37°C. Positive controls consisting of blastocyststreated with DNase and negative controls, in which the terminaldeoxynucleotidyltransferase enzyme was omitted, were includedfor every experiment. Nuclei were labeled with DRAQ5 for 30min (1:500 in PBS-0.1% Tween 20) and transferred to slides.The embryos were imaged using confocal laser microscopy andthen genotyped by PCR.

RESULTS

Top

RESULTS

Conservation and mRNA expression of ARS2. We and others have previously shown that ARS2 is conserved inmammals and plants (25, 40). Further analysis of eukaryoticgenome assemblies and expressed sequence tags (ESTs) suggeststhat there is a single ARS2 homolog in most eukaryotic genomes(Fig. 1). This suggests that ARS2 arose early in eukaryoticevolution and that it may serve a basic cellular function.

View larger version (32K):
[in this window]
[in a new window]

FIG. 1. Overview and evolution of ARS2. (A) Phylogenetic analysis of the ARS2 genes from representative eukaryotic species. Amino acid alignments were made using CLUSTALW and manually edited. Columns containing gaps were deleted, and 224 informative sites were used to construct the phylogeny. A neighbor-joining tree using a Poisson distance matrix was made. (B) Alignment of the mouse ARS2 locus with representative mRNAs and genomic DNAs. Representative mRNAs and ESTs were prealigned by the University of California Santa Cruz using the BLAT algorithm. Genomic conservation of the ARS2 gene between mouse and seven additional species is shown as vertical black lines. The alignments were precomputed using the Multiz align algorithm (3) and can be found at positions chr5:136,243,845 to 136,260,875 of the mouse March 2005 assembly on the UCSC genome browser website.

Conservation between mouse ARS2 and its orthologs can be observedat both the mRNA and genomic-sequence levels (Fig. 1B) (25,40). We examined the expression of ARS2 in mouse and human tissuesby PCR and Northern blotting analysis. By PCR, ARS2 was expressedin all mouse and human tissues tested (Fig. 2). Evidence ofa slightly larger mouse transcript variant ( $~$ 150 bp) can be seenin most tissues (Fig. 2A) and may represent previously describedalternative splice variants of ARS2 (40). To determine ARS2transcript size, relative abundance, and distribution in bothmouse and human tissues, Northern blot analysis was performed.ARS2 showed ubiquitous expression, with the highest levels inmouse testes, heart, and liver and human heart and skeletalmuscle (Fig. 2C and D). The predominant transcript size of $~$ 3.0kb corresponds to the full-length mouse cDNA sequence (BC066831)that is predicted to encode a 100-kDa protein product. Withthe 3' probe, smaller weak bands could be seen at $~$ 2 kb and 1kb. These products likely represent ARS2 isoforms generatedfrom internal promoters, as smaller C-terminal transcripts havebeen identified in human, mouse, Drosophila, and Chinese hamster(Fig. 1B). Interestingly, the original arsenite resistance ARS2cDNA from CHO cells was a C-terminal isoform containing theARS2 domain. This suggests that C-terminal messages may be functionallyrelevant. An expanded screen of a normalized RNA dot blot of50 human tissues confirmed that ARS2 is ubiquitously expressed,with the highest levels of expression in the testes, fetal lung,placenta, pituitary gland, bone marrow, thymus, and prostate(Fig. 2E).

View larger version (39K):
[in this window]
[in a new window]

FIG. 2. Expression analysis of mouse and human ARS2 mRNAs. (A and B) Semiquantitative PCR analysis of normalized first-strand cDNAs in mouse and human tissues. The mouse 1,890-bp ARS2 product was observed in all tissues, and a mouse G3pdh primer set was used as a positive control. (C) Mouse ARS2 Northern blot. (Top) Normalized mouse Northern blot tissues probed with the ARS2 3' probe. (Middle) Northern blot probed with the ARS2 5' probe. (Bottom) Beta actin control probe. (D) (Top) Human ARS2 Northern blot probed with the ARS2 human 3' probe. (Bottom) Northern blot probed with beta actin. (E) Normalized RNA blot of multiple human tissues using the ARS2 5' cDNA as a probe.

We also examined ARS2 expression during mouse development. ARS2is expressed at E7, E11, E15, and E17 (Fig. 2A). In addition,the distribution of ARS2 transcripts observed in the mouse ESTdatabase is biased toward prenatal stages. For example, 165ARS2 transcripts that were derived directly from mouse tissues(not including cancer cell lines) are found in the EST database.Ninety of the 165 ESTs can be assigned to a prenatal developmentalstage (preimplantation through late gestation). When these ESTcounts were normalized to reflect the total number of transcriptsderived from each tissue-specific library, 480 ARS2 transcriptsper million were found to be derived from prenatal tissue comparedto 275 transcripts per million derived from postnatal tissue(39). Within the prenatal tissues, preimplantation expressionis most dominant and includes ES cells and trophoblast cells.

Protein expression of mouse and human ARS2. The full-length mouse ARS2 cDNA produced the predicted 100-kDafull-length protein when expressed as an N terminus three-FLAGfusion protein (Fig. 3A) and as a C terminus Myc-His fusionprotein (Fig. 3B). Polyclonal and monoclonal antibodies raisedagainst ARS2 gave a predominant 100-kDa band and a less intense90-kDa band in human H1299 cell line lysates (Fig. 3C). Boththe N- and C-terminal full-length ARS2 fusion proteins showedsimilarly sized smaller bands when expressed in BOSC cells,suggesting this band may be a product of proteolytic degradation.ARS2 contains multiple nuclear localization signals and a zincfinger-like domain, which suggests ARS2 may function in thenucleus. Strong nuclear localization, with weak cytoplasmicstaining, was detected using anti-FLAG antibodies in FLAG-ARS2-expressingcell lines (Fig. 3D).

View larger version (39K):
[in this window]
[in a new window]

FIG. 3. ARS2 is expressed predominantly as a 100-kDa protein. (A to C) Western blot analysis of N terminus FLAG tag full-length mouse ARS2 protein stably transfected into the BOSC kidney cell line (A), C terminus MYC full-length mouse ARS2 fusion protein (transient transfection of BOSC cells) (B), and H1299 cell lysate using polyclonal and monoclonal anti-ARS2 antibodies raised against human ARS2 (C). (D) Anti-FLAG antibody staining of three-FLAG N-terminally labeled mouse ARS2 in a stably transfected BOSC cell line and a negative control in which the primary FLAG antibody was omitted (–ve primary). FITC, fluorescein isothiocyanate; DAPI, 4',6'-diamidino-2-phenylindole.

ARS2 is essential for early development. To elucidate the role of ARS2 in mammals, a gene-targeting vectorwas designed to remove the mouse ARS2 putative promoter to,and including part of, the third exon (Fig. 4A and B). In additionto the conserved proximal promoter region, this deletion removedthe first 49 amino acids of ARS2, which are conserved throughoutthe Eukaryota. Germ line transmission was obtained for the independentES cell clones A4-E1 and A4-G6 (Fig. 4C). Initial characterizationof the knockout mice was carried out on both 129S2/SvPasCrland Crl:CD1(ICR) background strains using mice derived fromboth cell lines. Cre-mediated excision of the neo cassette fromthe Ars-E1 line was achieved by crossing the Ars-E1 line withthe pCX-NLS-Cre line (19). Southern blot analysis of ARS2^+/neomice produced the expected 4.4-kb mutant band (Fig. 4D). Analysiswas carried out on the ARS2^+/neo and ARS2^+/neo lines. As therewere no phenotypic differences between the ARS2^+/neo and ARS2^+/neolines, the heterozygous ARS2 mutant allele is referred to asARS2^+/– hereafter.

View larger version (45K):
[in this window]
[in a new window]

FIG. 4. Gene targeting of mouse ARS2. (A) Mouse ARS2 genomic locus and surrounding genes. Mouse exon positions, repetitive elements, neighboring genes, and thelocations of the arms of homology for gene targeting are shown to scale on the x axis. Percent identity shared with the orthologous human locus is shown on the y axis. The coordinates are based on the sequence AF312033. (B) Removal of the proximal promoter and exons 1 to 3 of mouse ARS2 using homologous recombination. Southern blot probes for BamHI and EcoRI digests are labeled P1 and P2, respectively. The targeting vector with a neomycin cassette (neo) and herpes simplex virus thymidine kinase (HSV-tk) are indicated, along with the arms of homology. (C) Southern blot screening of R1 ES cells: BamHI digestion and P1 produced a wild-type band of 9.6 kb and a mutant band of 5.6 kb, and EcoRI digestion and P2 produced a wild-type band of 14.7 kb and a mutant band of 12.0 kb. (D) Southern blot analysis of the ARS2 locus using BamHI and P1 after Cre-mediated removal of the neo cassette. (E) Southern blot analysis of E12.5 yolk sacs from two intercrosses of ARS2^+/– mice using BamHI and P1. (F) PCR genotyping of blastocysts (E3.5) from ARS2^+/– intercrosses. The wild-type band is 333 bp, and the mutant band is 563 bp.

Breeding of ARS2^+/– mice produced no homozygous ARS2^–/–offspring for either of the stem cell-derived lines (Ars-E1or Ars-G6) after 28 and 16 intercrosses, respectively (Table1). No significant (P $≤$ 0.05) deviation from the expected 1:2Mendelian wild-type-to-heterozygous ratio was detected ( ${chi}$ ² [1,n = 129] = 1.71; P = 0.25). ARS2^+/– mice were healthyand showed no obvious differences from their wild-type littermates(data not shown).

View this table:
[in this window]
[in a new window]

TABLE 1. Breeding analysis of ARS2^+/– intercrosses

Yolk sacs from E10.5 and E12.5 embryos obtained from ARS2^+/–intercrosses were genotyped by Southern blotting, and no ARS2^–/–embryos were detected (Fig. 4E). PCR genotyping of E6.5 andE7.5 embryos failed to identify any ARS2^–/– embryos(Table 1). However, approximately 18% of deciduas from embryosdissected at E6.5 and E7.5 were small and devoid of embryos(Fig. 5A). Serial sectioning of E7.5 deciduas revealed no evidenceof the embryo proper in these smaller deciduas (Fig. 5B). Serialsectioning of litters from E5.5 identified small botryoid (grape-like)clusters of abnormal cells with pycnotic nuclei (Fig. 5C). Thissuggests that a majority of ARS2^–/– embryos implantbut die around E5.5 (Table 1).

View larger version (120K):
[in this window]
[in a new window]

FIG. 5. Dissections and histological analysis of wild-type and presumptive ARS2^–/– embryos at E7.5 and E5.5. (A) Representative example of relatively small deciduas routinely obtained from ARS2^+/– intercrosses at E7.5 (arrows). Embryos were never recovered from within these deciduas or small deciduas from E6.5. (B) Serial sectioning of random large and small deciduas at E7.5 revealed no discernible embryonic structure in presumptive mutants (n = 2). (C) Transverse, serial sectioning of E5.5 implantations from ARS2^+/– intercrosses revealed a clear outer boundary of the outer zone of the decidual reaction (zd) in both presumptive ARS2^–/– and ARS2^+/? embryos. A small mass of abnormal cells with pycnotic nuclei was characteristic of presumptive ARS2^–/– embryos (n = 3). Wild-type morphology containing a visceral endoderm (ve), parietal endoderm (pe), epiblast (epi), and proamniotic cavity was observed for all ARS2^+/? embryos.

ARS2-null embryos are unable to expand in vitro. ARS2^–/– blastocysts were detected at day E3.5 byPCR analysis in Mendelian ratios ( ${chi}$ ² [2, n = 85] = 2.86; P =0.24) (Fig. 4F). To further investigate this early embryoniclethality, we cultured E3.5 blastocysts under conditions usedto derive ES cells. To determine the expression of ARS2 in blastocysts,we performed immunofluorescence analysis on wild-type, heterozygous,and homozygous cultured embryos. Both monoclonal (LX186.3) andpolyclonal (XL12.2) anti-ARS2 antibodies showed strong nuclearlocalization and diffuse cytoplasmic staining in all cells ofcultured ARS2^+/– and ARS2^+/+ E3.5 embryos and a lack ofnuclear localization in ARS2^–/– E3.5 embryos (Fig.6; data not shown for XL12.2). The localization of ARS2 in embryoswas consistent with FLAG-ARS2 localization in BOSC cells andconfirms that ARS2 is enriched within the nucleus. ARS2-nullblastocysts did not adhere well to the gelatinized-medium dishand did not normally hatch from their zona pellucida. ARS2-nullblastocysts were noticeably smaller after 24 h in culture andinvariably collapsed by 48 h of culture (Fig. 7). Even withthe zona pellucida removed, ARS2-null embryos did not expand.After 120 h, a small mass of cells remained that showed severemembrane blebbing and abnormal morphology (Fig. 7). The mutantphenotype was the same regardless of the addition of LIF tothe media. ARS2 showed nuclear localization in ARS2^+/+ and ARS2^+/–E3.5 embryos before and after 24 and 48 h of in vitro culture(Fig. 6). A lesser amount of nuclear-localized ARS2 was observedin ARS2^–/– embryos at E3.5. This nuclear localizationwas lost after 24 h of culture and coincided with the onsetof the mutant phenotype (Fig. 6). Both serial sectioning ofE5.5 embryos and culturing of E3.5 embryos suggested that theembryonic-lethal phenotype begins as early as E4.5 and is fullypenetrant by E5.5.

View larger version (45K):
[in this window]
[in a new window]

FIG. 6. ARS2 distribution in ARS2^+/+ ARS2^+/– and ARS2^–/– E3.5 embryos before and after 24 and 48 h of culture. Before culture (t = 0), the nuclear localization of ARS2 was detected with anti-ARS2 polyclonal antibody and Alexa Fluor 568 goat anti-rabbit IgG (red) in the wild type and was present in smaller amounts or even absent in null embryos. Nuclear DNA was stained using DRAQ5 (blue); the merge of red and blue and the differential interference microscopy (DIC) are indicated. Embryos were observed by confocal microscopy, and a representative single optical section is shown. At time zero, the arrow indicates the nuclear localization of ARS2 in a cell undergoing anaphase. After 24 h of culture (t = 24), both ARS2^+/– and ARS2^–/– embryos had formed expanded blastocysts and ARS2^–/– embryos had completely lost their nuclear ARS2 localization. After 48 h of culture (t = 48), the null blastocyst had collapsed and several abnormal nuclei could be observed. The scale bars correspond to 50 µm.

View larger version (45K):
[in this window]
[in a new window]

FIG. 7. Absence of blastocyst outgrowths in ARS2^–/– embryos. Representative examples of cultured E3.5 blastocysts from ARS2^+/– intercrosses are shown. The embryos were photographed every 24 h using phase-contrast microscopy. All embryos were genotyped after being cultured. At E3.5, no obvious differences between ARS2^+/+ ARS2^+/– and ARS2^–/– embryos were observed. By E3.5 plus 24 h (E3.5 + 1), some ARS2^–/– embryos had collapsed and were generally not attached to the culture dish. ARS2^+/+ and ARS2^+/– embryos had both expanded and attached by this time. By 48 h (E3.5 + 2), ARS2^+/+ and ARS2^+/– embryos hatched from their zona pellucida (ZP), whereas ARS2^–/– embryos invariably failed to hatch and were normally found floating in the medium or loosely attached to the gelatin-coated plates. The removal of the ZP in null blastocysts did not affect the phenotype (column 4). After 120 h (E3.5 + 5) of culture, ARS2^+/+ and ARS2^+/– embryos were composed of outgrowths of ICM and were surrounded by a single layer of trophectoderm giant cells (TGC), whereas null embryos remained a small cluster of abnormal cells showing severe membrane blebbing (arrows).

ARS2-null embryos are able to initiate the decidual reactioneven though they generally do not hatch from their zona pellucidain vitro. To further investigate the ability of ARS2 mutantsto undergo trophectoderm differentiation, we looked at the trophectoderm-specificexpression marker Cdx2 and the inner cell mass (ICM)-restrictedmarker Oct4. Embryos from ARS2^+/– intercrosses were culturedfor 48 h and analyzed using immunofluorescence and confocalmicroscopy. ARS2^–/– embryos maintain an outer layerof Cdx2-positive cells and an ICM that is Oct4 positive (n =2). Some polar and mural trophectoderm cells did show eitherOct4 or Cdx2 expression, and in those cases, condensed nucleiindicative of apoptosis were observed (Fig. 8). Overall, thissuggests that the embryonic-lethal phenotype is not due to alack of differentiation in early embryonic lineages.

View larger version (15K):
[in this window]
[in a new window]

FIG. 8. Localization of the lineage-specific markers Cdx2 and Oct4 in ARS2^+/+ and ARS2^–/– embryos. The trophectoderm marker Cdx2 was labeled with anti-CDX2 and Alexa Fluor 488 goat anti-mouse IgG (green). The ICM marker Oct4 was labeled with anti-OCT4 and Alexa Fluor 568 goat anti-rabbit IgG (red). Cells without either label showing condensed nuclei are indicated by arrows. The nuclei were labeled with DRAQ5 (blue). The embryos were observed by confocal microscopy, and a representative single optical section is shown.

To investigate the possibility of apoptosis in ARS2^–/–embryos, we cultured E3.5 embryos for 24 h and identified celldeath using the fluorescent TUNEL assay. Excessive apoptosiswas observed in ARS2^–/– embryos (n = 7; mean [M]= 6.7; standard deviation [SD] = 2.5) compared to ARS2^+/+ andARS2^+/– embryos (n = 12; M = 1.5; SD = 1.4) (Fig. 9A and B).This difference was statistically significant by the Studentt test (t[28] = 7.4; P < 0.0001). Cell numbers were significantlydecreased (t[38] = 2.3; P = 0.0283) in ARS2^–/– embryos(n = 17; M = 79.6; SD = 15) compared to ARS2^+/+ and ARS2^+/–embryos (n = 23; M = 91.9; SD = 18) (Fig. 9). The mitotic indicesbetween ARS2^+/+, ARS2^+/–, and ARS2^–/– embryoswere not significantly different (t[10] = 1.45; P = 0.177) inARS2^–/– embryos (n = 6; M = 3.7; SD = 1.7) and ARS2^+/+and ARS2^+/– embryos (n = 6; M = 2.6; SD = 1.0) (Fig. 9D).

View larger version (64K):
[in this window]
[in a new window]

FIG. 9. Cultured ARS2-null embryos show excessive apoptosis. E3.5 blastocysts were cultured for 24 h, apoptotic cells were identified using a fluorescent TUNEL assay (green), and morphology was assessed with DRAQ5 nuclear stain (blue). The embryos were genotyped by PCR (ARS2^+/+ ARS2^+/–, n = 12; ARS2^–/–, n = 7). (A) A representative ARS2^–/– embryo showing excessive TUNEL labeling predominantly in the trophectoderm and an ARS2^+/– littermate. (B) There was a significant increase in TUNEL-positive nuclei in ARS2^–/– over ARS2^+/+ ARS2^+/– embryos (t[28] = 7.4; P < 0.0001). (C) Cell numbers in E3.5 embryos cultured for 24 h showed a statistically significant decrease in ARS2^–/– cell numbers over ARS2^+/+ ARS2^+/– E3.5 blastocysts cultured for 24 h (t[38] = 2.3; P = 0.0283). (D) No statistically significant difference in the mitotic index was observed between ARS2^+/+ ARS2^+/– (n = 12) and ARS2^–/– embryos (t[10] = 1.45; P = 0.177).

DISCUSSION

Top

DISCUSSION

Understanding the functions of genes involved in arsenic-relatedtoxicity and chemotherapy is an ongoing challenge. To elucidatethe effect of arsenic on cells, Rossman and Wang (28) lookedfor genes that conferred arsenite resistance on arsenite-sensitiveCHO cells by screening a cDNA library made from arsenite-resistantcells (28). From this screen, Asr2 (ARS2), a gene of unknownfunction, was identified. Here, we show that ARS2 orthologsare found in almost all eukaryotes; that in mammals, ARS2 isa ubiquitously expressed gene that encodes a nuclear-localizedprotein; that ARS2 is essential for early embryonic development;and that the ablation of the ARS2 protein leads to excessiveapoptosis in the early embryo.

While we obtained Mendelian ratios of ARS2^–/– blastocystsat E3.5, the lack of expansion of presumptive ARS2^–/–embryos in vitro, the presence of empty deciduas at E7.5, andthe small degenerate embryos detected at E5.5 indicate thatARS2^–/– embryos die around the time of implantation.We conclude that the removal of the conserved upstream regionand the first three exons was sufficient to cause the earlyembryonic lethality. A relatively small amount of nuclear-localizedARS2 could be observed in uncultured ARS2^–/– E3.5embryos compared to heterozygous or wild-type embryos. Thisnuclear localization was reduced to background levels after24 h of culture, which corresponds to the onset of the mutantphenotype. As observed after the disruption of many other "essential"genes, the onset of the ARS2 mutant phenotype may have beendelayed by maternal mRNA or protein stores (2, 31, 41).

A major challenge in cases of early embryonic lethality is todistinguish genes that are essential for cellular function fromgenes that are required for a developmental program. At E3.5,ARS2^–/– embryos are defective and generally do nothatch from the zona pellucida and do not adhere well in culture,nor do they expand. ARS2^–/– embryos contained bothtrophectoderm-positive (Cdx2) and ICM positive (Oct4) cells,suggesting that early cell fate determination was not affected.In contrast, ARS2^–/– embryos displayed an increasedlevel of apoptosis, suggesting ARS2 has an essential cellularfunction. The later, yet still developmentally related, pleiotropicphenotypes observed in zebrafish (11) and Arabidopsis ARS2 mutants(25) indicate that these previously described random mutationswere hypomorphic.

Several high-throughput screens have identified potential ARS2protein partners. In a yeast two-hybrid screen using the transcriptionfactor Rqcd1 (CG9573) as bait, the Drosophila melanogaster ARS2ortholog (CG7843) was identified as a Rqcd1 binding partner(10). Furthermore, chromatin immunoprecipitation studies revealedthat HNF4-alpha transcription factor bound to human ARS2 andRQCD1 promoters in the human adult liver (23), suggesting thatARS2 and RQCD1 are coregulated. In mammals, Rqcd1 has been shownto mediate retinoic acid-induced cell differentiation in cellculture and mouse lung explants (15), and all trans-retinoicacid is used in conjunction with arsenic trioxide to treat acutepromyelocytic leukemia (33). Based on these observations, wecloned the full-length mouse Rcqd1 gene and performed coimmunoprecipitationby transfecting stable three-FLAG-ARS2 or three-FLAG-Rcqd1 celllines with the Myc-His-tagged ARS2 and Rcqd1 genes. However,we did not detect any interaction between ARS2 and Rqcd1 inthis system (data not shown).

More recently, human ARS2 has also been shown to interact withRNA binding protein S1 (RNPS1) in two separate high-throughputprotein interaction studies, one in human kidney cells (7) andthe other in a study directed at purified exon junction complexesfrom HeLa cells (36). RNPS1, which is responsible for regulatingpre-mRNA splicing in vivo (29), was also shown to interact withSMN1 in a high-throughput protein interaction study by Ewinget al. (7). Importantly, Gemin4, an essential component of theSMN complex (4), also interacts with ARS2 (6). The SMN1 proteinis the core of the SMN complex, and mutations in SMN1 accountfor more than 98% of all spinal muscular atrophy cases (reviewedin reference 14). The SMN complex assembles Sm proteins ontothe small nuclear RNAs, which in turn mediate pre-mRNA splicing.Gemin4 has been mapped to a minimum loss of heterozygosity regionfor hepatocellular carcinoma (43), and the ablation of Gemin4is sufficient to disrupt Sm core assembly (34). The independentlyconfirmed protein-protein interaction between ARS2 and RNPS1,combined with their individual associations with componentsof the SMN complex, suggests that ARS2 may play a role in someaspect of RNA metabolism. Importantly the inactivation of themouse Smn1 gene resulted in a phenotype similar to that of ourARS2-null mice; most notably, the formation of disorganizedmulticystic structures, extensive cellular degeneration, andmassive cell death due to apoptosis are shared between Smn1and ARS2 mutants (31). In addition, the ablation of the Smn1-interactingzinc finger protein Zpr1 also resulted in a similar early embryonic-lethalphenotype with increased apoptosis and lack of proliferationin embryonic tissues (9). Clearly, determining whether ARS2is involved in RNA metabolism will be important in the future.In addition to RNPS1 and Gemin4, five human ARS2-interactingproteins have been reported (chromodomain helicase-DNA-bindingprotein 3 [CHD3], ARF GTPase-activating protein [GIT1], vimentin[VIM], mitochondrial inner membrane protein [IMMT], and smallubiquitin-related modifier 2 precursor [SUMO2]) (35). The significanceof these interactions remains to be determined.

Consistent with a putative role in RNA metabolism, recent studiesof the Arabidopsis ARS2 ortholog SERRATE (SE) demonstrated thatSE plays a general role in processing primary microRNA transcripts(12, 17, 42). SE has also been shown to colocalize to nucleardicing bodies containing proteins essential for microRNA processingin Arabidopsis (8). While studies in plants convincingly showthat SE is involved in microRNA processing, this role in animalshas yet to be demonstrated. In the context of arsenite resistance,it has recently been shown that treatment of the human lymphoblastline TK-6 with sodium arsenite led to a global increase in microRNAexpression due to a lack of proper precursor processing (18).This finding is particularly intriguing in light of SE's essentialrole in processing Arabidopsis microRNAs. Whether the role ofARS2 in arsenic resistance is related to some aspect of posttranscriptionalregulation remains to be seen.

The loss of single-copy proteins conserved throughout the Eukaryotawould be predicted to result in early and severe developmentalphenotypes. Here, we performed an in vivo analysis of the ARS2gene, and we show that abolishing the nuclear expression ofARS2 leads to peri-implantation lethality with an increase inapoptosis and a decrease in cell numbers. These data show thatARS2 is involved in an essential cellular process required forthe development of diverse eukaryotic organisms. ARS2 residesin one of the regions within 7q22 that is recurrently deletedin acute myeloid leukemia (reviewed in reference 30). However,we observed no obvious phenotype in ARS2^+/– blastocystsand adult mice (data not shown), suggesting that the haploinsufficiencyof ARS2 alone is not enough to cause disease. Because of theearly lethality, we could not examine the role of ARS2-nullmutations in adult tissues. SERRATE's role in microRNA processingin Arabidopsis and ARS2's association with mammalian proteinsinvolved in pre-mRNA splicing are intriguing, and future studiesof ARS2 may be relevant not only to arsenic resistance and cancer,but also to other human diseases.

ACKNOWLEDGMENTS

We thank Stacy Lew and Raymond Wong for making the aggregationchimeras, Daniel Morgado for animal care assistance, Jill Brandonfor BOSC cell transfection, Bob Chow and Haiquan Liu for useof and advice on confocal microscopy, Brent Gowen for serialembryo sectioning, and Yojiro Yamanaka and Claudia Kutter forhelpful discussions.

This work was supported by an NSERC grant to B.F.K. and a MichaelSmith Foundation for Health Research trainee award to M.D.W.

FOOTNOTES

* Corresponding author. Mailing address: Centre for Biomedical Research, University of Victoria, P.O. Box 3020 Station CSC, Victoria, BC, Canada V8W 3N5. Phone for P. Howard: (250) 472-4074. Fax: (250) 472-4075. E-mail: phoward{at}uvic.ca. Phone for B. F. Koop: (250) 472-4067. Fax: (250) 472-4075. E-mail: bkoop{at}uvic.ca

Published ahead of print on 17 December 2007.

Supplemental material for this article may be found at http://mcb.asm.org/.

Present address: Cancer Research UK, Cambridge Research Institute,Cambridge, United Kingdom.

REFERENCES

Top