Mice Lacking Alkbh1 Display Sex-Ratio Distortion and Unilateral Eye Defects

Background Eschericia coli AlkB is a 2-oxoglutarate- and iron-dependent dioxygenase that reverses alkylated DNA damage by oxidative demethylation. Mouse AlkB homolog 1 (Alkbh1) is one of eight members of the newly discovered family of mammalian dioxygenases. Methods and Findings In the present study we show non-Mendelian inheritance of the Alkbh1 targeted allele in mice. Both Alkbh1−/− and heterozygous Alkbh1+/− offspring are born at a greatly reduced frequency. Additionally, the sex-ratio is considerably skewed against female offspring, with one female born for every three to four males. Most mechanisms that cause segregation distortion, act in the male gametes and affect male fertility. The skewing of the sexes appears to be of paternal origin, and might be set in the pachythene stage of meiosis during spermatogenesis, in which Alkbh1 is upregulated more than 10-fold. In testes, apoptotic spermatids were revealed in 5–10% of the tubules in Alkbh1−/− adults. The deficiency of Alkbh1 also causes misexpression of Bmp2, 4 and 7 at E11.5 during embryonic development. This is consistent with the incompletely penetrant phenotypes observed, particularly recurrent unilateral eye defects and craniofacial malformations. Conclusions Genetic and phenotypic assessment suggests that Alkbh1 mediates gene regulation in spermatogenesis, and that Alkbh1 is essential for normal sex-ratio distribution and embryonic development in mice.


Introduction
The Eschericia coli (E. coli) DNA repair enzyme AlkB demethylates e.g. 1-methyladenine (1-meA) to adenine -generating succinate and formaldehyde -in the presence of iron as cofactor and 2oxoglutarate as cosubstrate [1,2]. To date, eight AlkB homologs have been identified in the mammalian genome [3]. Except for Alkbh5, all the remaining proteins have been identified throughout the animal kingdom, suggesting fundamental roles in biological processes [4]. Two of these homologs, ALKBH2 and ALKBH3 in humans (Alkbh2 and Alkbh3 in mice), are similar to E. coli AlkB in that they efficiently repair damaged nucleic acids in the presence of iron and 2-oxoglutarate in vitro [5][6][7][8][9]. In mice, Alkbh2 is the major, probably only, dioxygenase that repairs 1-meA DNA in vivo and mice lacking Alkbh2 accumulate 1-meA in the genome during ageing [10]. This year, two groups reported that Alkbh8 is a tRNA methyltransferase required for the final step in the biogenesis of mcm 5 U [11,12]. ALKBH8 plays important roles in the survival and progression of human bladder cancer both in vitro and in vivo [13]. A likely ninth AlkB homolog, the obesity-associated Fto protein, was shown to have potential to demethylate 3-methylthymine (3-meT) [14,15]. Crystal structure of the FTO protein recently confirmed this, and indicated that single-stranded RNA is the primary substrate of FTO [16]. Similarly, recombinant truncated Alkbh1 enzyme may demethylate 3-methylcytosine in vitro [17], but it remains unclear whether this activity is physiologically relevant.
All eight mammalian AlkB homologs contain the conserved iron-and 2-oxoglutarate dioxygenase domain. However, the region of E. coli AlkB that interacts with the nucleic acid substrate, the N-terminal nucleotide recognition lid, does not share sequence similarity with the mammalian homologs. Therefore, one cannot exclude the possibility that the targets of such proteins are not nucleic acids, but other macromolecules such as proteins. Since JmjC histone demethylases remove methyl groups from histones using the same mechanism as E. coli AlkB, it has been suggested that Alkbh1, 4 and 7 might be involved in histone/protein demethylation [18,19]. However, for Alkbh1 we, and others, have been unable to identify DNA/histone demethylation activity [6,7,20,21]. In 2008 a paper on Alkbh1 was published by Pan et al, where a gene-targeting study in mice showed that Alkbh1 localizes to nuclear euchromatin and functions in epigenetic regulation of gene expression [20]. Their study demonstrated impaired placental trophoblast lineage differentiation in Alkbh1 2/2 mice, and a strong interaction of Alkbh1 with Mrj, an essential placental gene that mediates gene repression by recruitment of class II histone deacetylases (HDAC) [20].
In the present study we attempt to elucidate the role of Alkbh1 by targeted deletion in C57/BL6 mice. We demonstrate that Alkbh1 deficiency in mice results in apoptosis in adult testes and sex-ratio distortion of offspring, most likely caused by defects in the pachytene stage during spermatogenesis. An incompletely penetrant phenotype apparent during embryonic development is consistent with Bmp2, 4 and 7 misexpression. Although many mechanistic aspects of Alkbh1 function remain to be revealed, these results show that Alkbh1 is crucial for normal embryonic development and viability in mice, and plays an important role during spermatogenesis.

Generation of Alkbh1 Targeted Mice
A specific 360-bp murine probe of exon 6 in the Alkbh1 gene was amplified from mouse genomic DNA by polymerase chain reaction (PCR) and used to screen a 129 SvJ mouse genomic library (Stratagene). To generate the targeting construct, we subcloned fragments from a ,14-kb genomic clone on both sides of neomycin (neo) in the pGT-N38 vector (New England Biolabs). Homologous arms consisting of a 3.0-kb MfeI/HindIII fragment and a 3.7-kb BsrGI fragment facilitated removal of a 3.8-kb HindIII/BsrGI fragment including exon 6 and replacement with the neo cassette. The targeting construct was electroporated into 129 SvJ embryonic stem (ES) cells, and transfectants were selected in geneticin (G418) and expanded for further analysis. Chimaeric mice were produced by microinjection of one targeted ES cell clone with normal karyotype into C57/BL6 blastocysts at embryonic day 3.5 (E3.5). We verified germline transmission of the targeted allele by Southern-blot analysis of ScaI-digested genomic DNA on the 59 end and PCR analysis on the 39 end. 59 and 39 homologous recombination in the F 1 generation were confirmed by PCR analysis. Heterozygous males were backcrossed for three generations onto C57/BL6 females. All mouse experiments were approved by the Norwegian Animal Research Authority (Ref. nr. 08/9940) and done in accordance with institutional guidelines at the Centre for Comparative Medicine at Oslo University Hospital. Animal work was conducted in accordance with the rules and regulations of the Federation of European Laboratory Animal Science Association's (FELASA).

Genotyping
For Alkbh1 genotyping, ear-clip samples were degraded by incubation in PBND buffer (50 mM KCl, 10 mM Tris-HCl pH 8.3, 2.5 mM MgCl 2 -6H 2 O, 0.1 mg/ml gelatin, 0.45% v/v NP40, 0.45% v/v Tween 20) and 0.5 mg/ml proteinase K at 55uC over night. Samples were heated to 95uC for 10 min to inactivate proteinase K, and PCR amplified for 35 cycles with an annealing temperature of 60uC (see primers below). For sex genotyping of embryos, a small piece of tissue was obtained from the embryosac or -tail and washed three times in PBS to eliminate maternal contamination. The tissue was degraded by a 3-hour incubation, and subsequently treated as above. PCR analysis of Sry (Y-linked gene) was performed to determine maleness and Rapsn was used as an autosomal, internal control as described (Mouse Phenotypes, a Handbook of Mutatation Analysis, Cold Spring Harbor laboratory press, Chapter 3, page 40,2005).

Whole-Mount In Situ Hybridization
We carried out whole-mount in situ hybridization on E9.5 to E12.5 embryos fixed in paraformaldehyde as described (Henrique et al. 1995). Mouse antisense and sense (control) RNA probes were prepared using DIG RNA labeling mix (Roche) together with T3 or Sp6 and T7 RNA polymerases (Roche). Templates for the labeling reaction were PCR products amplified from full-length mouse cDNA with T3, Sp6 or T7 promoters added to the PCR primers. For Alkbh1 the template contained 465-bp of exon 6, for Bmp2 519-bp of exon 2-3 and for Bmp7 559-bp of exon 2-5. For Bmp4, linearized pSP72 plasmid with a 1550-bp insert was used as template. Embryos were examined on a SMZ1500 microscope (Nikon).

Quantitative Real-Time PCR (qPCR) Analysis
Total RNA was isolated from embryos, organs and germ cells using the Fast RNA Pro Green Kit (MP Biomedicals) according to the manufacturers protocol. Any DNA remnants were removed using TURBO DNase (Ambion) and cDNA was made using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). The quantitative PCR reactions were carried out on a StepOne-Plus or 7500 Fast instrument using 50 ng cDNA, TaqManH Fast Universal PCR Master Mix and appropriate TaqMan primers and probes (all from Applied Biosystems). Pre-designed primers and probes were used both for the target genes (Alkbh1, Vav2, Mapk8, Ccdc80, Rest, and Hif1a) and endogenous controls (Gapdh, 18s and bactin). All samples were run in triplicates and with one technical parallel (2 runs per sample). The relative quantity was calculated using the equation RQ = 2 2DDCT , where RQ is the relative quantity of the target gene. DDCT is the difference in CT-value between the target gene and the endogenous control minus the difference in CT-values between the reference gene and the endogenous control.

STAPUT Isolation of Testicular Cells
Male germ cells were isolated from testes using an adapted version of the STAPUT method [22]. Pachytene cells and round spermatids were isolated from six 12-week old males, while a total of sixty 10-day old males were sacrificed for the isolation of type A and type B spermatogonia. The testes were put in ice cold DMEM medium containing antibiotics and then carefully detunicated. The tubules were treated with DNaseI, collagenase, trypsin and hyaluronidase (all from Sigma-Aldrich) at 34uC to remove connective tissue and somatic cells, yielding a cell suspension of germinal cells in DMEM containing 0.5% BSA. The cell suspension was loaded into the cell loading chamber of the STAPUT apparatus and separated by sedimentation velocity at unit gravity in a 2-4% w/v BSA gradient in DMEM medium at 4uC for 2.5 hours. After sedimentation, 10 ml fractions were collected and checked under the microscope. Fractions containing pure germ cells were pooled and the cell number counted in a CountessH Automated Cell Counter (Invitrogen). Cells were spun down and the pellet was snap frozen in liquid nitrogen before placed in 270uC. An aliquot of isolated cells was fixed on SuperFrost Plus slides (VWR) using Cell Adherence Solution (Crystalgen, Lot no 425081) for microscopic analysis of purity.

TUNEL Assay of Testes
We fixed testes from 3-and 9-month old animals in neutralbuffered formalin, progressively dehydrated them in a graded ethanol series, and embedded them in paraffin. Sections (4-mm) were deparaffinized and treated with proteinase K for 15 min and quenched in 3% hydrogen peroxide in PBS for 5 min at room temperature. Subsequently, nuclear staining in apoptotic cells was detected using ApopTag kit (Chemicon, http://www.chemicon. com) according to the manufacturers instruction. Sections were analysed on an Axioplan 2 microscope (Zeiss).

Immunofluorescent Staining of Testicular Cells
Testicular cells from 12-month old males were spread on SuperFrost Plus slides (VWR), progressively dehydrated in a graded ethanol series and dried completely. Slides were washed in 16PBS and fixated in 4% PFA in PBS. Slides were blocked in 5% serum in PBS for 1 hour at room temperature and incubated with primary antibodies overnight at 4uC prior to detection with secondary antibodies. Primary antibodies used were rabbit anti-MacroH2A (1:500, Upstate) and mouse anti-FK2 (1:5000, Biomol). Secondary antibodies used were goat anti-rabbit Alexa Fluor 488 (green dye) (Invitrogen) and goat anti-mouse Alexa Fluor 594 (red dye) (Invitrogen), respectively. Single Z-sections were captured by Axioplan 2 microscope (Zeiss).

DNA Microarray Analysis
High quality of total RNA extracted from adult testes was verified on Agilent Bioanalyzer 2100 (RIN value between 9.8 and 10.0). 15 mg of biotinylated and fragmented cRNA was then hybridized onto the GeneChip Mouse Genome 430 2.0 Array (Affymetrix) according to manufacturers protocols (Affymetrix). QCs including scale factor, background, noise, spikes and RNA degradation were checked and validated using the yaqcaffy library (http://www.bioconductor.org/packages/2.3/%20bioc/vignettes/ yaqcaffy/inst/doc/yaqcaffy.pdf).
Affymetrix raw data was generated with GCOS 1.4 (GeneChip Operating Software, Affymetrix), and the signal intensities of each probe set were normalized with the RMA (Robust Microarray Anlaysis) algorithm. To find differentially expressed genes, t-test with randomized variance was used as statistical test and the cutoff (p-value) was set to 0.05 with a FDR correction. Class comparison analysis was used to identify interesting genes. The signal in one group was always (i.e. for all the triplicate) higher or lower compared to the other group. Fold change for all the genes that passed the above criteria was computed and only the genes with $2-fold change were studied. The heatmap was generated using the GeneSpring GX 10 demoversion (Agilent). All data is MIAME compliant and the raw data has been deposited in a MIAME compliant database, the accession number is GSE22073.

Skeletal Staining
For skeletal analysis, skin and internal viscera of E18.5 embryos and newborn mice were removed. We then fixed the animals overnight in 95% ethanol and carried out Alcian blue 8GX

Histological Analysis of Eyes
We fixed adult eyes in neutral-buffered formalin or paraformaldehyde added 20% absolute alcohol for 24 hours, progressively dehydrated them in a graded ethanol series, and embedded them in paraffin. Sections (4-mm) were deparaffinized, rehydrated and stained with hematoxylin and eosin. Sections were analysed on an AxioCam HRc microscope (Zeiss).

Deletion of Alkbh1 in Embryonic Stem Cells and Mice
To gain more insight into the role of the Alkbh1 dioxygenase we have generated mice lacking Alkbh1. Alkbh1 was the first mammalian AlkB homolog to be identified [23], and is the AlkB homolog most similar in sequence to Eschericia coli (E. coli) AlkB. The region of greatest similarity includes 107 amino acids, 37% of which are identical between the E. coli and mouse Alkbh1. The conserved RvNmTvR and HvD…H motifs of the 2-oxoglutarate and iron binding sites, respectively, are also present in both proteins. The conserved domains of Alkbh1 are encoded by exon 5 and 6 at the 39 end of the mouse Alkbh1 gene. To fully eliminate the activity of Alkbh1 and keep the overlapping Nrp gene intact, we substituted exon 6 with a neomycin-resistance gene cassette by homologous recombination in mouse embryonic stem cells (Figs. 1A-D). The expression of the Nrp gene was confirmed by qPCR (data not shown).

Expression Analysis of Alkbh1 in Embryos, Organs and Male Germ Cells
The expression pattern of Alkbh1 was analysed in embryos at different stages by whole-mount in situ hybridization ( Fig. 2A) and by qPCR (Fig. 2B). Weak expression of Alkbh1 was observed throughout the embryo at E8.5 (data not shown). As the cells migrate and differentiate during organogenesis the expression becomes more specific, and Alkbh1 was detected in the spinal cord, forebrain and branchial arches at E9.5, and also in limb buds at E10.5 ( Fig. 2A). Peak expression was detected at E11.5 in the frontonasal process including telencephalon (tc), maxillary, mandibular and hyoid arches (ba), upper and lower limb buds (lb), and midbrain and rhombomere 1 (r1) roof plates (rp) ( Fig. 2A). Alkbh1 expression decreased considerably from E11.5 to E12.5 ( Fig. 2A-B). In adult organs, Alkbh1 was highly expressed in testis (RQ = 44.0), with slightly lower expression in eye, brain and kidney (RQ = 16.0, 15.4, 14.4) (Fig. 2C). Moreover, the expression of Alkbh1 was studied at different stages during spermatogenesis, and was found to be significantly elevated in the pachytene spermatocytes (PS) (RQ = 11.3) compared with spermatogonia A and B (Sg A, Sg B) and round spermatids (RSd) (RQ = 1.7) (Fig. 2D). This is the third stage of the prophase of meiosis I, in which synapsis is completed and homologous recombination occurs. Thus, Alkbh1 may have considerable potential for gene-function in embryonic development and in the pachytene stage during spermatogenesis.

Non-Mendelian Inheritance and Sex-Ratio Distortion in Alkbh1 Targeted Mice
Mendelian inheritance, in which each parent contributes one of two possible alleles for a given trait, has a characteristic ratio of 1:2:1 after heterozygous crosses. Initial crosses of mice carrying either one or two targeted Alkbh1 loci revealed non-Mendelian distribution. Therefore, we carried out extensive breeding analysis and genotyped more than 1400 Alkbh1 mutant mice and embryos (Fig. 3). Following heterozygous breedings, the survival of Alkbh1 2/2 pups after 1 month was only 20% compared with wild-type littermates (Fig. 3A). In addition, the frequency of viable Alkbh1 +/2 mice was only 60% of the expected rate (Fig. 3A). The non-Mendelian distribution was clearly significant with a p-value of 3.8610 27 (x 2 -test). A similar pattern was observed in Alkbh1 +/2 male x Alkbh1 2/2 female crosses, p = 5610 24 (x 2 -test) and Alkbh1 2/2 male x Alkbh1 +/2 female crosses, p = 3.7610 25 (x 2 -test) (Fig. 3C). In general, the average litter size decreased as the number of targeted alleles in the parental generation increased (Fig. 3B). The mean litter size was 9.2 for wild-type crosses, 6.2 for heterozygous crosses and 3.2 for homozygous crosses (Fig. 3B). Notably, paternal inheritance of the targeted allele seemed to be more critical than maternal transmission for the survival of offspring. Another evident phenotype was the growth retardation observed in viable Alkbh1 2/2 mice compared with wild-type littermates (Fig. S1).
One process of non-Mendelian inheritance is segregation distortion. There are a number of mechanisms that can cause segregation distortion, and both autosomal sex-ratio distortion as well as segregation distortion of the sex chromosomes exist [24]. In Alkbh1 +/2 male x Alkbh1 2/2 female crosses, the ratio of female to male homozygous offspring at 1 month was approximately 1:1 (Fig. 3C, left panel). In Alkbh1 2/2 male x Alkbh1 +/2 female crosses, the ratio of homozygous Alkbh1 2/2 pups was significantly skewed against females, with one female born for every three to four males (Fig. 3C, right panel). The survival of Alkbh1 2/2 male pups was 60% compared with Alkbh1 +/2 pups, whereas the proportion of viable Alkbh1 2/2 female pups was only 18%, p = 7.1610 25 (x 2test) (Fig. 3C, right panel). Following heterozygous crosses, the survival of Alkbh1 2/2 offspring was significantly reduced, 30% of the males and just 10% of the females survived compared with wild-type littermates, p = 1.4610 26 (x 2 -test) (data not shown). A sex-ratio distortion was also seen in mid-stage Alkbh1 2/2 embryos (E10-E12.5) after heterozygous breedings (17 litters), with 89% male and 60% female embryos present compared with wild-type embryos (data not shown).

Spermatogenic Defects in Alkbh1 Deficient Testis
Reduced testis weight was observed in Alkbh1 2/2 males at 12week and 12-month of age, constituting three-quarters and twothirds the mean weight of testis from wild-type littermates, respectively (Fig. 4A). TUNEL staining of testes from 12-week old wild-type and Alkbh1 2/2 males were histologically indistinguishable and showed no apoptotic cells (data not shown). On the other hand, extensive apoptosis and reduced number of germ cells were revealed in 5-10% of the seminiferous tubules in 9-month old Alkbh1 2/2 males (Fig. 4B, Fig. S2). In Alkbh1 2/2 testes, no apoptosis was detected in the spermatogonia (Sg) located at the edges of the tubules and in the meiotic spermatocytes (Sc) residing mostly in the two to three subbasal layers (Fig. 4B, Fig. S2). However, numerous apoptotic and degraded cells were seen in the subbasal regions corresponding to spermatocytes and spermatids, as well as in degenerating round and elongating spermatids (Sd) in the more luminal layers of the tubules (Fig. 4B, Fig. S2). In wild-type, a few apoptotic cells were occasionally located mainly at the basal layer of the seminiferous tubules (Fig. 4B, Fig. S2). To better define the basis for arrest in germ cells and the sex-ratio distortion, we focused on the XY-body in the pachytene stage of meiosis. The XY-body is a condensed chromatin structure containing the sex chromosomes, which is thought to be essential for meiotic progression. In midpachynema the XY-body forms a spherical structure near the nuclear periphery [25]. Two different markers against XY-bodies were used, macroH2A and FK2, however visible sex-bodies were readily identified in pachytene spermatocytes from 12-month old wild-type and Alkbh1 2/2 testes (Fig. 4C). MacroH2A recognizes the sex chromatin, and FK2 detects the abundant ubiquitination of H2A in the XY-body. We also did antibody staining against several specific stages throughout spermatogenesis, but no significant differences between wild-type and Alkbh1 2/2 mice were revealed (Fig. S3). The fact that sex-body formation is not impaired in Alkbh1null males does not exclude the hypothesis of an epigenetic and silencing defect of the paternal X chromosome in those mice. Another possibility is that the skewing of the sexes in Alkbh1 2/2 mice is related to autosomal sex-ratio distortion. It is well known that most mechanisms that affect segregation distortion act in the male gametes and affect male fertility [24].
Expression Profiling in Wild-Type and Alkbh1 2/2 Testis Due to the pivotal role of Alkbh1 in mouse survival and potentially in germ cells, we searched for Alkbh1-regulated genes in adult testes. Microarray analysis of whole testes from 12-week old males identified 25 genes that were differentially expressed in Alkbh1 2/2 versus wild-type, using the class comparison strategy (Fig. 5A). Ptpro were also statistically significantly upregulated in Alkbh1 2/2 testes (Table S1; All data is deposited in GEO, accession number GSE22073). The function of PTPRO in adult testis has not been explored, but Avraham et al found expression of PTPRO in testis in humans [26]. Ptpro is suggested to be involved in the differentiation and axonogenesis of central and peripheral nervous system neurons, where it is in position to regulate phosphotyrosine levels in intracellular signaling cascades [27]. qPCR was performed on selected genes, to verify the class comparison analysis (Fig. 5B). Upregulation of Vav2 and Ccdc80 was confirmed in Alkbh1 2/2 versus wild-type whole testes. Vav2 is a guanine nucleotide exchange factor important for the formatin of adherens junctions between Sertoli cells and spermatids in testis, as well as in the formation of synapses in neurons [28]. The function of Ccdc80, also known as steroid sensitive gene 1, has not been studied in testis, but is supposed to be expressed in this organ according to its EST profile in the Unigene database (http://www. ncbi.nlm.nih.gov/UniGene/ESTProfileViewer.cgi?uglist = Mm. 181074). Ccdc80 is expressed in human mesenchymal stem cells and mouse embryo cartilage, suggesting a role in skeletogenesis [29]. Together, these findings point towards a role in regulating the expression of genes having diverse functions -in spermatogenesis, in the nervous system and in skeletogenesis, although the genes affected in the microarray analysis are merely indirect targets of the Alkbh1 protein.
To identify any abnormalities in addition to small or missing eyes, E18.5 embryos and newborn mice were analysed by skeletal  (Fig. 6E). Skeletal staining also showed assymetric shortening of the nasal bones, curving unilaterally in Alkbh1 2/2 mice causing maldeveloped teeth (Fig. S4A), as well as reduced ossification in the phalanges (P) and the metatarsals (M) of the autopod of Alkbh1 2/2 newborns (Fig. S4B). The most crucial step in skeletal morphogenesis is the formation of mesenchymal condensations at E9.5 to E11.5 in mouse development [34]. The Alkbh1 variable phenotype indicates incomplete condensation of mesenchymal cells during skeletogenesis.

Incomplete Penetrance of Unilateral Eye Defects
Penetrance is described as incomplete when a trait associated with a specific allele is expressed in a proportion of the population carrying the allele variant [35]. The eye phenotype associated with lack of the Alkbh1 allele is characterized by incomplete penetrance (Fig. 7A-B). The Alkbh1 2/2 mouse in Fig. 7B has developed normally except for the deficiency of one eye. In contrast, the Alkbh1 2/2 embryo in Fig. 7A has gross developmental abnormalities, in addition to one small eye with only a residual mass of retinal cells, and one eye missing. The excessive brain tissue outside the skull is characteristic of a condition in which the neural tube fails to close, called exencephaly. Exencephaly is a neural tube defect (NTD), together with spina bifida (open spine) and anencephaly (open skull) [36]. At E10.5-E11.5, NTDs were observed in 23% of Alkbh1 +/2 embryos (n = 12/52) and 10% of Alkbh1 2/2 embryos (n = 3/31). The defects originated primarily from disrupted closure in the midbrain-hindbrain region (Fig. 7A) and upper spinal region, and were frequently associated with head and facial malformations (Fig. S4C). Around 50% of embryos with NTDs simultaneously displayed eye malformations (n = 14/27). The eye-and NTD-defects observed in Alkbh1 mutants correspond with the expression pattern of Alkbh1 seen in embryos and adult mice.
Gross morphological and histological analysis of adult Alkbh1 2/2 eyes revealed a range of serious deformities and size variations (Fig. 7C-D). Hematoxylin and eosin (HE) staining of paraffinembedded sections showed that the lens was either completely missing or clearly smaller and displaced in the eye field (Fig. 7D). Furthermore, the lens fiber cells had lost their ordered lamination pattern, and swollen and liquefied fibers as well as vacuoles were seen throughout the lens (Fig. S4D). In retinal cells, there was a severe loss of organization even though all the retinal cells were present (Fig. 7D). In some areas, the neural retina (NR) was dysplastic with inclusions of rods and cones surrounded by outer nuclear layer cells (ONL), forming rosettes (Fig. 7E). In others, regions of thick layers of retinal pigment epithelium (RPE) cells were observed, with RPE cells appearing inside the NR layers in direct contact with the lens (Fig. 7E). Hence, Alkbh1 is important for growth and appropriate positioning and survival of lens and retinal cells.

Altered Expression of Bmps in Alkbh1 Deficient Embryos
Embryonic development and tissue regeneration are regulated by four major families of signaling molecules. One of the largest families is the bone morphogenetic proteins (Bmps) [37]. In skeletogenesis, Bmp signaling plays an important role in regulating chondrocyte differentiation and establishment of joint boundaries [38]. Current evidence indicates that Bmp2, Bmp4 and Bmp7 are the main source of Bmp signaling in vertebrate limb buds [39]. Similar signaling mechanisms are suggested for growth and regional specification of the forebrain, branchial arches and eye during development [40][41][42]. This prompted us to examine the expression of Bmp2, Bmp4 and Bmp7 in apparently normal Alkbh1 2/2 embryos at E11.5 (Fig. 8A). Bmp2 and Bmp7 were induced in the lateral telencephalon (tc) of Alkbh1 2/2 embryos, and expression of Bmp2 also increased in the frontonasal process (Fig. 8A). Moreover, Bmp4 and Bmp7 became upregulated specifically in the maxillary and mandibular cleft, while Bmp2 was upregulated throughout the maxillary, mandibular and hyoid Figure 5. Expression profiling in wild-type and Alkbh1 2/2 adult testis. (A) Microarray analysis of whole testes from three wild-type and three Alkbh1 2/2 12-week old males. Results are presented following class comparison analysis and visualized by GeneSpring v 6.0. (B) Verification of differentially expressed genes from the microarray analysis of wild-type and Alkbh1 2/2 testes. A selection of genes identified in the class comparison analysis (Vav2, Mapk8, Ccdc80, Rest, Hif1a) were checked for significance by qPCR. Upregulation of Vav2 and Ccdc80 were confirmed, while the differential expression of Mapk8, Rest and Hif1a were not found to be significant. On the RNA used for the microarray study. Reference sample, wildtype Ccdc80 (RQ = 1.00); endogenous control, 18s. doi:10.1371/journal.pone.0013827.g005 mesenchyme (Fig. 8A). In limb buds, Bmp4 and Bmp7 were highly upregulated in the apical ectodermal ridge (AER) and in two broader domains anteriorly and posteriorly (Fig. 8B). Bmp2 expression disappeared from the posterior domain in hindlimb, and expression in AER of forelimb diffused proximally into the mesenchyme (lm) (Fig. 8B). The disrupted expression of Bmp2, Bmp4 and Bmp7 might be the cause of the somewhat smaller limb buds in Alkbh1 2/2 embryos. Regulation of these Bmp genes is important for AER formation, which is the major signaling center for limb outgrowth [37]. In general, both increased and decreased Bmp signaling can result in skeletal phenotypes [38].

Discussion
Our data point towards an important role of Alkbh1 in spermatogenesis and embryonic development. Several genes involved in spermatogenesis, in the nervous system and in skeletogenesis were found to be differentially expressed in Alkbh1 2/2 whole testes. Adult males deficient in Alkbh1 exhibited dramatically increased levels of apoptosis in 5-10% of the seminiferous tubules of testes; in spermatids and in degenerated germ cells in the subbasal regions corresponding to spermatocytes and spermatids. The reduced number of all spermatogenic cells in the apoptotic tubules, might reflect an indirect effect of prolonged arrest in spermatids in the affected tubules. Similar nonspecific defects have been seen in miwi-null mice [43] and TRF2 mutants [44,45]. Most genes involved in spermatogenesis display pleiotropic and leaky mutant phenotypes, as presented in this paper. Targeted disruptions of genes resulting in a variable range of defects and incomplete penetrance of spermatogenesis is even the case for regulatory genes, such as those encoding RNA binding proteins DAZLA [46] and MVH [47], and cell cycle regulators HSP-70.2 [48,49] and cyclin A1 [50].
The sex-ratio distortion lead us to study the XY-bodies in pachytene spermatocytes from Alkbh1 2/2 testes, however visible XY-bodies were detected showing that X and Y chromosomes paired normally during male meiosis. This does not exclude the hypothesis of an epigenetic and silencing defect of the paternal X chromosome in those mice, which could explain the sex-ratio distortion observed. Moreover, embryonic and postnathal lethality seen in Alkbh1 2/2 mice seem to be of paternal origin and Alkbh1 2/2 males exhibit subfertility compared to wild-type males. Several characteristics of the Alkbh1 2/2 mice are comparable with those described for the Jmjd1a histone lysine demethylase and the G9a histone lysine methyltransferase mutant mice, although to a milder extent than demonstrated in the histone disrupted mouse models [51,52]. Jmjd1a deficiency caused extensive germ cell apoptosis and blocked spermatid elongation, resulting in small testes and infertility in male mice [51]. Inactivation of G9a in the germ-lineage resulted in sterility due to a drastic loss of mature gametes [52]. The specific upregulation of Alkbh1 in the pachytene stage, together with the sexratio distortion, suggests a potential to regulate the expression of genes during meiosis in the germline. Future investigations will focus on the regulation of specific genes in pachytene spermatocytes isolated from Alkbh1 2/2 and wild-type testes.
Alkbh1 mutant mice displayed phenotypes of incomplete penetrance, including unilateral eye malformations, neural tube defects, and craniofacial and skeleton associated abnormalities. Around 10% of the Alkbh1 2/2 mice appeared relatively normal, whereas the most affected mice died early during embryogenesis. The phenotypes are similar to published results on the bone morphogenetic proteins (Bmps), such as haploinsufficiency of Bmp2 causing exencephaly comparable to Fig. 7A [53], and compound heterozygous mutants for Bmp2 and Bmp4 showing unilateral microphthalmia similar to Fig. 6-7 [54]. In addition, postnatal lethality and sex-ratio distortion against females have been shown in Bmp4 tm1/+ heterozygous at weaning [55]. Altogether, this led us to investigate the effect on Bmps, and the  misexpression of Bmp2, Bmp4 and Bmp7 in Alkbh1 2/2 embryos at E11.5 might explain the inconsistent phenotypes presented. This is due to the critical dependence of gene dosage for proper Bmp function together with the expression-and function-overlap of the Bmps in different tissues [39,40]. Mouse models of Bmp4 and Bmp7 have shown that redundancy between Bmp4 and Bmp7 is not sufficient to prevent the eye phenotype to occur [40,41,56]. In the skull, signaling pathways involving Bmp2, Bmp4 and Bmp7 regulate mesenchymal condensation size, and intense expression of these signaling genes is necessary for closure of sutures [34]. In addition to modifier genes such as Bmps, genetic and epigenetic components can cause variable phenotypic outcomes from specific genes [57], leading to irregular patterns of inheritance as seen for the Alkbh1 deficient mice. A recent paper has shown that the osteoblast-specific transcription factor Osterix is regulated by the JmjC histone demethylase NO66 [58]. Experiments in the chick embryo have revealed that epigenetic factors are required for the establishment of left-right asymmetries, together with the action of well-studied genetic and signaling mechanisms [59,60].
The reduced viability and developmental phenotypes apparent in our mouse model, was not reported in the Alkbh1-null mice generated by Pan et al [20]. However, they showed severe growth defects in Alkbh1 2/2 embryos and newborns in addition to placentas [20], and the growth retardation demonstrated in pups at four weeks of age are comparable with our data (Pan et al. Suppl. Fig. 2 and this paper Fig. S1). No obvious color variation (from red/pink to pale brown/bluish) or growth retardation was observed in Alkbh1 2/2 placentas compared to wild-type placentas. Our results are based on extensive breeding studies of Alkbh1 targeted mice, which revealed a dramatic effect on lethality and sex-ratio in adult mice. We therefore sought to characterize testes and embryos in more detail, as well as the prominent abnormalities in eye development. The different mouse background chosen as well as the dissimilar targeting strategies deleting different parts of the Alkbh1 gene (Exon 6 in our strain, Exon 3 in Pan et al) could be a possible explanation for the discrepancies in the penetrance of phenotypes in the two knockout mice models. Even so, together with the findings on Alkbh1 by Pan et al, these data suggest that the effect of Alkbh1 deficiency is pleiotropic and dependent on cell type and/or stage of development.
Recent studies have recognized roles for 2-oxoglutarate dependent dioxygenases in histone and nucleic acid demethylation, as well as in signaling protein hydroxylation [19]. For the demethylating enzymes, several have been shown to carry out its reaction in a manner similar to the potential Alkbh1 mediated, iron-and 2-oxoglutarate dependent, hydroxylation [1,2,61]. Previously, mouse models for histone methyl transferases and histone demethylases have been characterized with multiple developmental defects [31][32][33]62]. Our working hypothesis, based on the variable developmental phenotype of Alkbh1 deficient mice together with the localization of Alkbh1 to nuclear euchromatin [20], is that Alkbh1 possibly works as a histone demethylase during embryogenesis and spermatogenesis. We believe that the hydroxylation activity of Alkbh1 is dependent on yet undefined partners specific for the different stages/tissues where it has an important role, and this will be addressed in future studies for the pachytene stage of meiosis in male germ cells -when homologues chromosomes pair and crossing over can occur. Figure S1 Average body weight of Alkbh1 targeted males and females. (A) 1-month old wild-type (19.062.0 g, n = 45) and Alkbh1 2/2 (14.663.8 g, n = 48) males, and 1-month old wild-type (17.761.7 g, n = 43) and Alkbh1 2/2 (14.862.2 g, n = 33) females. The average weight was 25% lower for Alkbh1 2/2 males than for wild-type males and 15% lower for Alkbh1 2/2 females than for wild- type females. About one out of five Alkbh1 2/2 males showed more than 40% lower weight compared to wild-type males. (B) 9-month old wild-type (40.564.2 g, n = 23) and Alkbh1 2/2 (32.562.9 g, n = 31) males, and 9-month old wild-type (31.563.4 g, n = 28) and Alkbh1 2/2 (29.363.8 g, n = 41) females. The average weight of Alkbh1 2/2 males was 20% below that of wild-type males, and the average weight of Alkbh1 2/2 females was 7% below that of wildtype females. No weight difference was demonstrated between the Alkbh1 +/2 and wild-type (data not shown). +/+ (wild-type), black bars; 2/2 (Alkbh1 2/2 ), grey bars. Found at: doi:10.1371/journal.pone.0013827.s001 (0.10 MB TIF) Figure S2 Closer view of the DAPI and TUNEL staining of testis sections shown in Fig. 4. (A, B) Sections from 9-month old wild-type (left panel) and Alkbh1 2/2 (right panel) mice are presented. Apoptosis was detected in degenerating spermatids (Sd) in the luminal layers of Alkbh1 2/2 tubules, as well as in severely degraded cells in the subbasal regions corresponding to spermatocytes and spermatids. No apoptotic cells were seen in spermatogonia (Sg) and spermatocytes (Sc) in Alkbh1 2/2 mice, although the amount of all spermatogenic cells are reduced in the apoptotic tubules. (Magnification: 620). Found at: doi:10.1371/journal.pone.0013827.s002 (3.93 MB TIF) Figure S3 Immunostaining with stage-specific antibodies against spermatogenic cells in Alkbh1 deficient testes. (A) Testis sections from 12-month old wild-type and Alkbh1 2/2 males stained with TRA98 antibody specific for spermatogonia, which were present both in wild-type and mutant. Although several tubules showed spermatogonia not only in the first basal layer, but also in the subbasal layers in the Alkbh1 2/2 mice, no significant differences were detected when compared to wild-type. (B) Testis sections from 12-month old wild-type and Alkbh1 2/2 males stained with TRA369 specific for pachytene spermatocytes through elongating spermatids, which were present both in wild-type and mutant. (Magnification: 620).  2/2 mice causing mal-developed teeth (n = 4 Alkbh1 2/2 ; n = 1 Alkbh1 +/2 ). Ossified areas are shown in red and cartilage in blue. (B) Limb defects. Dorsal view of the autopod limb skeleton revealing reduced ossification in the phalanges (P) and the metatarsals (M) of the autopod of Alkbh1 2/2 newborns (n = 4/4 Alkbh1 2/2 ). Ossified areas are shown in black and cartilage in blue. (C) Eye defects and NTDs. Side view of embryos at E12.5. The Alkbh1 2/2 embryo has a bilateral microphthalmic eye phenotype in combination with a neural tube defect (NTD). The NTD is originating from disrupted closure in the upper spinal region, and is associated with head and facial malformations leading to a shortened, broad snout. In addition, a severe intracranial hemorrhage is visible. (D) Lens defects. Histological analysis of paraffin-embedded eye sections from adult mice. In Alkbh1 2/2 eyes the lens fiber cells have lost their ordered lamination pattern, and swollen and liquefied fibers as well as vacuoles are seen throughout the lens. (Magnification: 610). Found at: doi:10.1371/journal.pone.0013827.s004 (6.28 MB TIF)

Supporting Information
Table S1 Statistically upregulated genes in Alkbh1 2/2 versus wild-type testes identified in the microarray analysis. Microarray analysis of RNA extracted from whole testes from three wild-type and three Alkbh1 2/2 12-week old males identified 6 genes that were statistically upregulated in Alkbh1 2/2 versus wild-type. To find differentially expressed genes, t-test with randomized variance was used as statistical test and the cut-off (p-value) was set to 0.05 with a FDR correction. Found at: doi:10.1371/journal.pone.0013827.s005 (0.22 MB TIF)