Broad, ectopic expression of the sperm protein PLCZ1 induces parthenogenesis and ovarian tumours in mice

Mammalian oocytes are typically arrested at meiotic metaphase II (mII)until a fertilising sperm induces cell cycle resumption and the initiation of embryogenesis. Multiple candidates for the causative sperm entity have been proposed (reviewed by Runft et al.,2002) and database searching presumptive of the involvement of phospholipase C (PLC) identified an erstwhile uncharacterised mammalian isoform, PLC-zeta (PLCZ1) (Saunders et al., 2002). PLCZ1 expression was reportedly restricted to the testis (Saunders et al., 2002)and shown by protein correlation profiling to be present in demembranated sperm (Fujimoto et al., 2004). PLCZ1 is unusual among PLCs because it lacks pleckstrin homology (PH) and Src homology (SH) domains; proceeding from the N- to the C-terminus, it comprises four EF hand domains, X and Y domains and a C-terminal C2 region(Saunders et al., 2002). PLCZ1 deletion analyses suggest that EF hand domains 1 and 2 (EF1 and EF2) are primarily responsible for phosphatidylinositol 4,5-bisphosphate(PIP₂) hydrolysis, whereas EF3 mediates its exquisite Ca²⁺ sensitivity (Kouchi et al., 2005). PIP, but not PIP₂, binding is mediated in vitro by the C2 domain, which may coordinate with Ca²⁺ sensing to regulate pulsatile Ca²⁺ release from oocyte stores(Kouchi et al., 2005).

Oscillations in the concentration of intracellular free Ca²⁺,referred to as [Ca²⁺]_i, initiate 1-3 minutes after mouse gamete membrane fusion (Jones et al.,1995; Lawrence et al.,1997) and are thought to modulate the activity of calmodulin kinase II (CAMK2) towards the cytostatic factor FBXO43 (also known as EMI2),potentiating secondary FBXO43 phosphorylation by PLK1 and targeting it for proteolytic degradation to induce meiotic exit(Lorca et al., 1993; Rauh et al., 2005; Shoji et al., 2006).

This model of PLCZ1 activity leaves several issues open. The PH domains of other PLC family members mediate their interactions with phosphoinositides,but PLCZ1 does not possess a PH domain and the PLCZ1 C2 domain does not bind to PIP₂ in vitro (Kouchi et al., 2005). PLCZ1 phospholipid targeting remains poorly understood and a simple interaction might not account for the generation of IP₃ - a prelude to Ca²⁺ release. Moreover, although∼1.25-2.5 fg of native PLCZ1 (corresponding to ∼3-6% of the amount in a single sperm) efficiently induces oocyte activation, a ∼120-240-fold excess of baculovirus-expressed PLCZ1 (300 fg) is required to induce[Ca²⁺]_i oscillations resembling those of fertilisation(Fujimoto et al., 2004; Kouchi et al., 2004). Premature attenuation or hyperstimulation of [Ca²⁺]_ioscillations does not prevent development to term(Ozil et al., 2006), showing that embryogenesis in the mouse is tolerant of a range of[Ca²⁺]_i dynamics during oocyte activation. Finally,distinct sperm-borne entities reduce the PLCZ1 signalling threshold(Perry et al., 1999b; Perry et al., 2000; Fujimoto et al., 2004) and play unresolved roles in meiotic exit(Manandhar and Toshimori,2003; Sutovsky et al.,2003; Wu et al.,2007). These findings raise the possibility that multiple factors play a role in sperm-dependent meiotic resumption.

Sperm-independent meiotic resumption (parthenogenetic activation) can be induced by exposing mature mII oocytes to one of a multiplicity of exogenous non-physiological challenges in vitro, including electrical stimulation(Tarkowski et al., 1970),ethanol (Cuthbertson, 1983)and strontium chloride (Whittingham and Siracusa, 1978). In vivo, the two best-known models of sperm-independent meiotic interference are LT/Sv(Stevens and Varnum, 1974) and gene-targeted Mos-null mouse strains(Colledge et al., 1994). Independently targeted MOS-deficient oocytes exhibit aberrant spindle migration to produce an abnormally large and persistent first polar body(Pb₁) (Choi et al.,1996), deregulation of MAPK activity during oocyte maturation(Araki et al., 1996) and pronucleus formation following Pb₁ extrusion(Colledge et al., 1994); all reflect dysfunctional meiosis I, in contrast to parthenogenetic activation in vitro, which acts upon mature mII oocytes. Mice lacking MOS develop ovarian teratomas, which are accordingly likely to be a consequence of first meiotic deregulation. Teratomas occur in ∼30% of MOS-deficient females when they are 4-8 months old, but not outside this age range; occasionally, the tumours are malignant and metastatic (Furuta et al., 1995).

Teratomas also occur in the other well-known model of maternal meiotic dysfunction, LT/Sv, although their incidence is slightly higher (37-52%) in females of 3-4 months, with tumours appearing as early as 2 months(Stevens and Varnum, 1974). LT/Sv oocytes frequently undergo mI arrest and/or fail to establish mII arrest(Hampl and Eppig, 1995), and it has been concluded that parthenogenetic activation potentiates teratoma formation (Stevens and Varnum,1974). However, experiments aimed at rederiving LT/Sv from its progenitor strains (BALB and C58) revealed that mI arrest is necessary but not sufficient to elicit parthenogenetic activation(Eppig et al., 1996). Furthermore, some LT/Sv-related strains undergo mI arrest and parthenogenesis without developing teratomas (Eppig et al., 1996), showing that meiotic failure does not necessarily result in tumour formation. LT/Sv thus possesses a complex and polygenic phenotype and predisposing loci have been mapped to chromosomes 1, 6 and 9(Lee et al., 1997; Everett et al., 2004). The single mouse Plcz gene, Plcz1, lies on chromosome 6.

We here evaluated PLCZ1 specificity by forcing its ectopic expression in a broad range of tissues, enabling us to determine whether this interfered with maternal meiosis. These experiments reveal that endogenous PLCZ1 induces activation of mature mII oocytes with a high degree of specificity in vivo,linking fertilisation to tumourigenesis and representing a unique model of parthenogenesis.

B6D2F₁ testis-derived Plcz1 cDNA was amplified by PCR with primers (all are shown 5′ to 3′): GACAAGCGGCCCAGATCATG and GTCTAGATTACTCTCTGAAGTACCAAACATAAATAAAC. For the CS series(PLCZ1ctFLAG-i-Venus), a single FLAG epitope was introduced to generate a C-terminal fusion with PLCZ1 by PCR using ACATGCATGCACTAGTATGGAAAGCCAACTTCATGAGCTC and GGAATTCCATATGTCACTTGTCGTCATCGTCTTTGTAGTCCTCTCTGAAGTACCAAACAT. An SpeI-NdeI PLCZ1-FLAG fragment downstream of the hybrid cytomegalovirus IE enhancer-chicken Actb promoter, pCAG,which is active in many tissues (Takada et al., 1997), and an IRES-Venus (IRES, internal ribosome entry site) cassette (our unpublished data), were introduced downstream of this. For the CV series (PLCZ1ctVenus), a Plcz1amplimer generated with the primers ACATGCATGCACTAGTATGGAAAGCCAACTTCATGAGCTC and GGAATTCCATATCCCCGGGCCCTCTCTGAAGTACCAAACATA was used to generate a fusion protein with the junction sequence YVWYFREARGSTMVS; the linker between the C-terminus of PLCZ1 (...FRE) and the start of Venus (MVS...) is underlined. For the generation of promoter-mapping constructs (see Fig. S1B in the supplementary material), a 4.5 kb Plcz1 putative promoter fragment (pPlcz1) was amplified with TCAGAGGTCACCCAACACGG and TCCCCCGGGATTTCATGATCTGGGCCGC and used to generate pPlcz1→Cre. A 4.1 kb Plcz1 fragment was removed from pPlcz1→Cre to produce pPlcz1→Plcz1-FLAG and pPlcz1→Plcz1-FLAG-IRES-Venus. All transgene (tg)constructs were introduced by mII transgenesis(Perry et al., 1999a) to produce founder (F₀) B6C3F₂ lines that were subsequently crossed with C57BL/6 unless stated otherwise. Mice were maintained according to local institutional guidelines.

Oocyte and embryo retrieval, manipulation and culture were essentially as described previously (Shoji et al.,2006; Yoshida and Perry,2007). For movies, oocytes were transferred to a chamber(37°C, 5% CO₂) on the stage of a Zeiss Axiovert 200 microscope and collected as described (Shoji et al.,2006). Nuclear transfer was into B6D2F₁ mII oocytes essentially as described (Yoshida et al., 2007).

Analysis of mRNA was by PCR as described(Shoji et al., 2006; Amanai et al., 2006b). Plcz1 transcripts in testis and brain samples(Fig. 1A) were amplified for 25 and 30 cycles, respectively. Amplification of Cre or recombinant Plcz1 (rPlcz1) mRNAs in pPlcz1 transgenic lines was with 30 or 35 cycles, respectively. RNA from clinical paraffin sections was isolated using a High Pure RNA Paraffin Kit (Roche) according to the manufacturer's instructions. First-strand cDNA was synthesized from 1 μg of isolated RNA using a Transcriptor First Strand cDNA Synthesis Kit (Roche)primed with 1 μl of 50 pmol/μl oligo(dT)₁₈ and 2 μl of 600 pmol/μl random hexamer in a final volume of 20 μl. One microlitre of each cDNA reaction was used per standard 25 μl PCR reaction. PCR reactions were typically accompanied by RT-minus (lacking reverse transcriptase)controls, performed on at least two independent preparations and where appropriate were examined following 2% (w/v) agarose gel electrophoresis. Ratiometric quantification of mRNAs (qPCR) was as described(Shoji et al., 2006; Amanai et al., 2006b). Amplification of Venus and Plcz1 tgs (with Plcz1intron-flanking primers to avoid native gene amplification) in 8 ng of tumour or control genomic DNA was performed to estimate their relative genomic complements. The ratio of tg signal to endogenous control Actb or H2afz genomic DNA levels in (diploid) somatic tissue from hemizygotes was normalised to 1.0 and used as a calibrator for tumours. PCR primer sequences are given in Table 1.

Following standard trans-cardial perfusion with PBS followed by buffered 4%(w/v) paraformaldehyde (PFA, Wako), ovaries for frozen sectioning were fixed in 4% PFA overnight at 4°C, and subsequently subjected to cryoprotection in 30% (w/v) sucrose (Wako) in PBS for 3 days. The samples were then embedded in OCT compound (Sakura, Tokyo) and sections (14 μm) prepared on a Microm HM560 cryostat (Microm, Walldorf). Sections were air-dried, washed in PBS and stained with Hoechst 33258 before being mounted in fluorescent mounting medium(DakoCytomation, CA) and analysed by confocal microscopy.

For histopathology, resected tumours were acutely fixed in 4% buffered PFA(pH 7.0) and embedded in paraffin by standard methods. Sections (1-2 μm)were deparaffinised, dried and stained with Haematoxylin and Eosin. Bright-field visualisation and image capture to enable classification (WHO)was on a Leica Microsystems Application Suite (Leica). Tumours were scored according to established parameters (World Health Organization, 2003).

Relative levels of ooplasmic [Ca²⁺]_i were determined for immature germinal vesicle (GV) and mI oocytes or mature mII oocytes. Immature oocytes were collected 48 hours after equine chorionic gonadotropin injection and analysed within 2 hours (GV oocytes) or 8-10 hours (mI oocytes)after collection. Mature, mII oocytes were analysed ∼14.5 hours after human chorionic gonadotropin (hCG) injection. Prior to recording, oocytes were loaded for 30 minutes with 5 mM Fura 2 acetoxymethyl ester (Fura 2-AM;Molecular Probes) in a humidified atmosphere of 5% (v/v) CO₂ in air at 37°C in KSOM. Oocytes were then placed on the 37°C-heated stage of an Eclipse TE2000-U inverted fluorescence microscope (Nikon). Fluorescence images were obtained at 10 second intervals following exposure at 340 and 380 nm (0.2 seconds apart), collected through a 490-530 nm emission filter and processed with AQUA COSMOS ratio imaging application software (Hamamatsu Photonics).

Rabbit polyclonal antibodies against a C-terminal fragment of mouse PLCZ1(PLCZ1ct, residues 111-648) were affinity-purified from immune serum using protein G-sepharose beads (Amersham).

For western blotting, tissues were extracted in HBS lysis buffer,containing 0.1% (w/v) sodium dodecyl sulphate (SDS), 0.5% (w/v) deoxycholate,1.0% (v/v) Nonidet P40, 150 mM NaCl, 10% (w/v) glycerol and 50 mM HEPES (pH 8.0) and 20 μg protein analysed per sample. For immunoprecipitation, 1 mg/ml protein preparations were mixed with anti-FLAG-conjugated beads (Sigma)for CS, or anti-PLCZ1ct-conjugated protein G-sepharose beads (Amersham) for CV, at 4°C overnight. Beads were washed with HBS (CS) or a wash buffer containing 150 mM NaCl, 2 mM CaCl₂, 5 mM MgCl₂, 0.05%(v/v) Nonidet P40, protease inhibitor cocktail and 10 mM Tris-Cl, pH 7.5 (CV). Bound material was removed by boiling in sample buffer for 5 minutes. Immunoblotting was typically using enhancer with a LumiGLO Reserve Chemiluminescent Substrate Kit (Kirkegard and Perry Laboratories) as described previously (Shoji et al.,2006).

For immunofluorescence imaging, oocytes were fixed in 4% PFA for 15 minutes at room temperature following brief (1-2 minute) serial washes in 1% and then 2% PFA. Fixed oocytes were labelled with mouse anti-TUBA (also known asα-tubulin) antibodies (Sigma; 1:9000) followed by anti-mouse IgG Alexa488 conjugate (Molecular Probes; 1:500) and stained with propidium iodide(Sigma). Fluorescence was visualised on a Nikon Eclipse E600 microscope equipped with a Radiance 2100 laser scanning confocal system (BioRad).

RNA was recovered from acutely isolated samples using the mirVana miRNA isolation kit (Ambion) and analysed essentially as described previously(Amanai et al., 2006a). Data sets were deposited at the NCBI Gene Expression Omnibus(http://www.ncbi.nlm.nih.gov/projects/geo/)with the series accession number GSE4822.

Determining the specificity of PLCZ1 should illuminate its role and mechanism. Mouse Plcz1 transcripts were present in the testes of post-pubertal males (4 weeks and older) and at low levels in pre- and post-pubertal brains of both sexes (Fig. 1A; data not shown). The identity of brain Plcz1 was confirmed by sequencing. A narrow expression profile for PLCZ1 (protein) has previously been noted but did not include the brain(Saunders et al., 2002).

To map the promoter elements responsible for this restricted expression,transgenic mice were generated in which genomic DNA fragments 4.5 and 4.1 kb upstream of the putative Plcz1 translational start codon respectively directed transcription of either a Cre or Plcz1 cassette(Fig. 1B). Each fragment included the Plcz1 transcriptional start mapped by 5′-RACE(Fig. 1B)(Fujimoto et al., 2004). Transgene (tg) constructs were expressed in the brains of males and females and the testes of males, with a high degree of tissue but not developmental specificity in three out of the five independent lines analysed (see Table S1 in the supplementary material). This suggests that the 4.1 kb Plcz1promoter fragment directs spatial, but not temporal, restriction of Plcz1 expression. In addition to a 67 nucleotide region(5′-CATGTG...ACACAG-3′) of strong Z-DNA potential(Champ et al., 2004), the fragment harbours canonical recognition motifs for sex-determining region Y(SRY) protein (Harley et al.,1992) and a perfect repeat of the palindromic cAMP-responsive element binding protein (CREBP; also known as CREB5) cognate sequence(Fig. 1B)(Maekawa et al., 1989). A similar configuration directs thimet oligopeptidase (Thop1) gene expression in spermatid-derived cell lines(Morrison and Pierotti,2003).

Having verified the native Plcz1 expression profile, we were able to evaluate the specificity of PLCZ1 by forcing its expression ectopically. A promiscuous promoter was selected to direct broad expression in transgenic lines encoding PLCZ1 either fused to Venus (CV) or whose coding sequence was separated from that of Venus by an IRES (CS)(Fig. 1C).

Founder ovarian Plcz1 transcript levels were quantified (not shown). Adult males and females of lines CV3 and CS16 expressed tg transcripts and recombinant PLCZ1 (rPLCZ1) protein in multiple tissues(Fig. 1D). Protein expression in the CV3 F₁, CV3-13, was higher than that of subsequent outbred generations.

CV3 and CS16 individuals of either sex appeared healthy for the first∼3 months, indicating that PLCZ1 is largely inert at the levels found in these lines. Male hemizygotes outcrossed with C57BL/6 produced litter sizes within the control range (7.91±0.29, n=105, P>0.05), but female CV3 or CS16 hemizygotes crossed with C57BL/6 males produced litter sizes significantly smaller than those of controls(0.85±0.348, n=26, P<0.0001). Such disruption of female reproductive function by rPLCZ1 was consistent with maternal meiotic abnormalities.

Immature oocytes collected from rPlcz1 transgenic females underwent GV breakdown and entered mI before extruding a Pb₁ and forming an mII spindle (Fig. 2A). Time-lapse imaging (see Movie 1 in the supplementary material) showed normal kinetics of Pb₁ extrusion in hemizygotes(P=0.81) en route to mII (Fig. 2A-D); the mean Pb₁ extrusion time for hemizygotes was 12.80±0.42 hours post-collection (n=34 oocytes), and for non-transgenic littermates it was 12.65±0.47 hours (n=24 oocytes). Spindle and chromosome behaviour during maturation were indistinguishable in oocytes from hemizygotes and controls(Fig. 2A). Injection of wild-type GV oocytes with sperm-derived active PLCZ1(Fujimoto et al., 2004) did not interfere with meiotic progression (n=38). Superovulated oocytes from hemizygotes analysed ∼13 hours after human chorionic gonadotropin(hCG) administration possessed an mII plate(Fig. 2A,C,D), a clear Pb₁, condensed metaphase chromosomes attached to spindles and Fbxo43 mRNA at 99.7±9.8% of control levels (n=6)(Shoji et al., 2006). During maturation, oocytes from transgenic (GV, n=18; mI, n=18) and non-transgenic (GV, n=17; mI, n=18) females exhibited a similar pattern of [Ca²⁺]_i oscillations(Fig. 2E) reminiscent of that previously described for wild-type oocytes(Carroll et al., 1994). Oscillation amplitudes in GV oocytes from transgenic females were greater than those of controls (Fig. 2E). Although PLCZ1 may boost [Ca²⁺]_i oscillation amplitude at the GV stage, these findings indicate in different ways that PLCZ1 does not functionally interfere with meiotic maturation and that most, or all, oocytes establish mII in the presence of rPLCZ1.

During or soon after oocyte collection (∼13 hours post-hCG), metaphase arrays were observed to rotate and/or separate in preparation for cytokinesis(Sun and Schatten, 2006),indicating meiotic exit (Fig. 2D). Collected oocytes exhibited spontaneous[Ca²⁺]_i rises of variable (and often very large)amplitude with first-to-second recorded peak intervals of 12.48±0.82 minutes (n=16) initiated by a basal pacemaker rise characteristic of fertilisation (Jones et al.,1995; Perry et al.,2000) (Fig. 2E). Parthenogenotes emitted a Pb₂, formed a single maternal pronucleus of normal size and developed efficiently to the morula/blastocyst stage in vitro (Fig. 2F,G).

It is possible that rPLCZ1 expression during meiotic maturation abrogated the installation of one or more activities needed to sustain mII arrest,rather than by inducing mII exit as at fertilisation. We evaluated this by transferring the nuclei of transgenic cumulus cells (which contain rPlcz1 mRNA, not shown) into wild-type mII oocytes. Nuclear transfer induced pronuclear activation in ∼60% of cases; this level approximates to the proportion of oocytes activated by 1.25 fg of sperm-derived PLCZ1 (50%)(Fujimoto et al., 2004),suggesting that a similar amount of rPLCZ1 was transferred with each cumulus cell nucleus. Nuclear transfer zygotes cleaved efficiently and developed in vitro (Fig. 2H,I). Thus, rPLCZ1 produced in vivo directly stimulates the parthenogenetic activation of mII oocytes.

The level of rPlcz1 mRNA in mature mII oocytes from hemizygotes was only 12.3±11.3% of that of GV oocytes (n=8, P=0.0001). Consistent with this, maturing ovarian (but not mII)oocytes of the line CS16 exhibited clear Venus epifluorescence(Fig. 3A); Venus is encoded by the same bicistronic mRNA as rPLCZ1 in the line CS16(Fig. 1C). This suggests that meiotic exit was induced by PLCZ1 that had been produced prior to mII.

Oviductal parthenogenotes at later cleavage stages, including blastocysts,were recovered from transgenic females 16 hours post-hCG(Fig. 3B), showing that activation occurred independently of superovulation.

Hemizygous CS16 and CV3 ovaries were typically of healthy appearance at 3-4 months. However, histochemical sectioning of one intact transgenic ovary revealed an early-stage follicular choriocarcinoma or yolk sac tumour(Fig. 3C).

Most young rPLCZ1-expressing individuals were overtly asymptomatic, but by 5-6 months, many females had developed abdominal swellings caused by ovarian tumours (Fig. 4A). Tumourigenesis exhibited a typical latency of ≥3 months, although onset was apparent macroscopically as early as 61 days. Age-matched rPLCZ1-expressing males remained largely asymptomatic. Tumour formation in females was highly penetrant (Fig. 4B) and occurred bilaterally or unilaterally (Fig. 4A). Hemizygous females occasionally (14.7%, n=116)contained abortive implantation fossa (Fig. 4C) without evidence of uterine tumourigenesis. Development of ovarian tumours was also highly penetrant in ICR outcrosses; 92.9% of females from ICR×CS16 crosses developed tumours (n=14). Tumour development was therefore not highly restricted by genetic background.

The rPlcz1 tg integrant dosage in most (69%, n=13)tumours was generally ∼1.0-1.5 per diploid genome complement as determined by qPCR (Fig. 4D). Apparent tg dosages were generally conserved even when genomic DNA samples were taken from multiple (up to six) sites in the same growth(Fig. 4D) and thus did not reflect an averaging of 0.0 and 2.0 integrations per genome across the entirety of a given tumour.

Tumours contained elevated levels of rPlcz1 mRNA relative to controls (P=0.0046) and decreased levels of Mos and Fbxo43 transcripts (P≤0.00023), which are downregulated post-activation (Fig. 5A)(Shoji et al., 2006). Whereas miRNA profiles are the signatures of some solid tumours(Lu et al., 2005; Volinia et al., 2006), the profiles of teratomas were diverse, reflecting teratoma heterogeneity (see Fig. S1 in the supplementary material).

Imprinted gene expression in tumours was found to be segregated, with paternally-expressed transcripts present at significantly lower levels(P<0.05 for seven out of the ten genes examined) than in controls,whereas steady-state levels of seven out of nine maternally-expressed mRNAs were ∼1.0 or >1.0 compared with controls(Fig. 5B). Transcripts for placental markers Plac1, Gcm1, Zfp36l3, Plib and Tpbpa(Fig. 5C) were not detected. These profiles are expected for parthenogenetic tumours in which paternally imprinted alleles and placental tissue are depleted or absent.

Histopathology of tumours at 4-6 months revealed mature cystic teratoma(epidermoid cyst), cystadenoma with borderline malignancy, and mixed germ cell tumours (Fig. 6). Histological heterogeneity reflects the extensive capacity for multipotent differentiation of parthenogenetic lineages (Stevens,1978). The presence of mature cystic teratoma may represent remnants of incompletely regressed Wolfian duct, and cystadenomas are also found in aging mice, either of which might be due to somatic cell differentiation. However, we never observed ovarian tumours in age-matched,non-transgenic littermates.

The appearance of dysgerminoma, yolk sac tumourigenesis and choriocarcinoma characterise the differentiation of pluripotent embryonic or germ cells. The predominance and frequency of such tumours provide a clear causal link between PLCZ1-induced parthenogenesis and tumourigenesis.

We addressed the possibility that LT/Sv oocytes contain active PLCZ1. With the exception of the brain, a site of wild-type Plcz1 expression(Fig. 1A), Plcz1transcript levels in other female LT/Sv tissues (including ovaries) at 8 weeks were undetectable (Fig. 5D). We verified this result by sequencing the 4.5 kb Plcz1 promoter region(Fig. 1B) of LT/Sv and that of its presumptive relative, the non-parthenogenetic strain BALB/c(Eppig et al., 1996); the sequences were identical (not shown). Promoter sequence conservation and the lack of ovarian Plcz1 mRNA indicate that LT/Sv phenotypes are not due to anomalous Plcz1 expression.

rPlcz1 transgenic mice exhibited occasional (n=7) hind limb ataxia (see Movie 2 in the supplementary material). The high-level expresser, CV3-13, was ataxic and although no CS16 hemizygotes exhibited the phenotype, two homozygotes did. Of the remaining four affected members of the CV3 line, two were male and two female. These data show that the phenotype was not sex-specific and may correlate with PLCZ1 expression levels.

Some cases of human ovarian cancer also present with ataxia(Geomini et al., 2001). We investigated whether human tumours contained elevated levels of PLCZ1mRNA in common with the tumours of CV3 and CS16 lines (not shown). However, we found no evidence for genetically predisposed PLCZ1 expression in human breast epithelial (n=15), or ovarian epithelial (n=15)or benign ovarian germline (n=22) tumours(Fig. 5E and not shown).

The developmental specificity of PLCZ1 is highlighted by its narrow wild-type expression profile and the restriction of its activity to oocytes(and possibly brain) when low-level expression is forced ectopically in multiple tissues.

The principal phenotypes induced by rPlcz1 tg expression -parthenogenesis, tumourigenesis and, to a lesser extent, ataxia in both sexes- are thus apparently associated with those tissues in which Plcz1 is normally expressed (Fig. 1A). This argues that the cellular machinery required to transduce PLCZ1 signalling is restricted to the same tissues: oocytes and the brain. Our preliminary data suggest that within the brain, Plcz1 mRNA is predominantly localised to the telencephalon (not shown). In one model, the presence of telencephalic PLCZ1 signal-transducing machinery would render the telencephalon susceptible to PLCZ1 overexpression, resulting in motor function defects that account for sporadic ataxia. A plausible role for this machinery (and its counterpart in the oocyte) would be to facilitate the targeting of PLCZ1 to PIP₂(as its C2 domain is insufficient to do so) in a manner similar to the interaction of PLCB1 with GNAQ (Wang et al., 1999; Kouchi et al.,2005). The possibility remains open that higher levels of ectopic PLCZ1 expression overcome the requirement for an adaptor to induce broader(lethal) embryonic phenotypes not investigated here.

Although PLCZ1 is normally expressed in the testis, we did not find evidence of abnormal cellular proliferation in the testes of transgenic males. This could be because tg expression did not significantly augment the total level of PLCZ1 (relative to native PLCZ1) and/or because the testis also lacks PLCZ1 signal-transducing machinery. A downstream target of PLCZ1 signalling in oocytes, FBXO43, is abundant in, and exclusive to, the testis in adult males(Shoji et al., 2006). Teratomas in males are more likely to be malignant than those in females(Stevens, 1967), so although a male meiotic role for FBXO43 is presumptive, PLCZ1-FBXO43 signalling in the testis could perturb the cell cycle and result in catastrophic neoplastic germ cell transformation, implying the stringent need to avoid it.

In summary, the data presented here suggest that PLCZ1 activity in vivo requires tissue-specific accessory factors. These might include adaptors that bind to PLCZ1, thereby compensating for its inherent lack of PH or SH domains.

Several lines of evidence suggest that oocytes in rPlcz1transgenic lines CV3 and CS16 complete meiotic maturation before undergoing activation. Ectopic rPLCZ1 expression during oogenesis thus represents the first in vivo model in which the normal program of coordinated maternal cytoplasmic and nuclear maturation precedes autonomous parthenogenetic activation.

Endogenous expression from rPlcz1 tgs induced meiotic exit in a manner characteristic of normal fertilisation. Demonstration of this faculty in vivo circumvents some drawbacks of injection experiments, which do not completely eliminate the possibility that non-physiological RNA or exogenous impurities act as co-factors; enzymes such as telomerase have an RNA component(Greider and Blackburn, 1987)and non-DNA-metabolising signalling enzymes such as DNA protein kinase require DNA (Carter et al., 1990).

Parthenogenesis occurs in both Mos-deficient and LT/Sv oocytes. In the absence of MOS, MAPK signalling is not established during mI(Araki et al., 1996; Choi et al., 1996). LT/Sv oocytes frequently arrest at mI (Hampl and Eppig, 1995); this failure is associated with precocious cell cycle progression but is not sufficient to induce it(Eppig et al., 1996). LT/Sv females heterozygous for the polymorphic marker Gpi1 produce homozygous tumours and, although it was inferred from this that the teratomas arose from oocytes that completed meiosis I(Eppig et al., 1977),reductive division of tetraploid cells at any stage in early tumourigenesis,followed by clonal selection (or other pathways), could produce the same result. Moreover, tumour cells of rPlcz1 transgenic mice are apparently hemizygous. Finally, LT/Sv oocytes efficiently induce activation when fused to wild-type mII oocytes(Ciemerych and Kubiak, 1998),yet this is not owing to expression of PLCZ1(Fig. 5D). Previous in vivo models of parthenogenesis reflect aberrant mI and differ from the one described here.

In one mechanistic model linking parthenogenesis in vivo to tumourigenesis,PLCZ1 activates mII oocytes that fail to be released from their ovarian environment, resulting in quasi-embryonic growth to form a teratoma. Histopathology and the patterns of imprinted and placental gene expression corroborate the parthenogenetic provenance of the tumours, but this model is simplistic.

The model does not explain the apparent hemizygosity of most tumours within hemizygous females, as the clonal generation of diploid cells from haploid parthenogenotes (Kaufman et al.,1983) would generally result in homozygosity [as it does in LT/Sv,where tumours arise from meiosis I failure and are homozygous in ∼90% of cases (Eppig et al., 1977)]. This objection is addressed if a primary tumour in rPLCZ1-expressing females impeded subsequent ovulation, thereby increasing the likelihood of supernumerary tumours in the same ovary and enabling the fusion of the cells of different early teratomas. Tumour cells generally (but not always)contained the rPlcz1 tg, implying a post-activation selective advantage of PLCZ1 expression.

Tumourigenesis could have followed failure of meiosis I and subsequent cytoplasmic maturation, allowing activation by rPLCZ1 of an oocyte containing four genomic complements (Mehlmann and Kline, 1994; Carroll et al.,1996). We found no evidence to support this model and several observations argue against it. PLCZ1 expression did not interfere with oocyte maturation (Fig. 2A-E and see Movie 1 in the supplementary material), and even where parthenogenesis due to the failure of meiosis I occurs at high frequency [∼100% in the case of Mos-null oocytes (Colledge et al., 1994)], the frequency of tumour formation is markedly lower:30% for Mos-null mice versus ≥60% for ectopic rPLCZ1 expression(Furuta et al., 1995)(Fig. 4B). Parthenogenesis in LT/Sv is not sufficient to induce tumour formation(Eppig et al., 1996) and the genotype predisposing to the LT/Sv phenotype maps to at least three unascribed loci (Lee et al., 1997; Everett et al., 2004), none of which is Plcz1 (Fig. 5D).

Although wild-type parthenogenotes exhibit retarded extra-embryonic development, they are able to develop for ∼10 days in vivo(Kaufman et al., 1977) and, in keeping with this, we observed uterine implantation in transgenic virgins(Fig. 4C). Implantation can occur ectopically (McLaren and Tarkowski,1963) and embryonal carcinoma cells generate tumours in vivo(Stevens, 1970). The ovary is clearly not a unique niche for teratoma development and the failure of uterine tumours to develop suggests either that growth was too slow or that uterine mechanisms exist to prevent parthenogenic tumour development.

We found no evidence of metastasis in rPLCZ1-expressing mice, although the pronounced growth of ovarian tumours indicated angiogenesis(Folkman and Klagsbrun, 1987)and at least one tumour from the CS line(Fig. 5C) expressed all four signature genes (Ereg, Ptgs2, Mmp1a and Mmp2) for lung tumourigenesis and metastasis (Gupta et al., 2007). We were unable to detect expression of the cancer stem cell marker protein PROM1 (CD133) (Hemmati et al., 2003) immunohistochemically in teratomas (not shown). Although this suggests that in general, cell fate commitment was an early event in ovarian tumour establishment(Avilion et al., 2003; Kania et al., 2005), the presence of NM_026894 and Pou5f1 transcripts in tumours(Fig. 5A) is consistent with a small population of relatively undifferentiated cells(Monk and Holding, 2001; Tai et al., 2005).

These data collectively demonstrate that oocytes exposed to endogenous PLCZ1 mature normally and establish mII. PLCZ1 is sufficient to induce meiotic exit, parthenogenetic development and teratoma formation with exquisite specificity. The studies establish a novel relationship between fertilisation and tumourigenesis and imply a tractable model with which to study the dysregulation (and thereby the productive orchestration) of embryogenesis.

We are grateful to Dr Joe Xhou of LC Sciences for miRNA analysis, to Satoko Fujimoto for assistance with western blotting, to the Laboratory of Animal Resources and Genetic Engineering for their unsung heroism and to Dr Kazuhiro Kitada for the generous provision of LT/Sv mice.