Genetic depletion of Polo‐like kinase 1 leads to embryonic lethality due to mitotic aberrancies

Polo‐like kinase 1 (PLK1) is a serine/threonine kinase that plays multiple and essential roles during the cell division cycle. Its inhibition in cultured cells leads to severe mitotic aberrancies and cell death. Whereas previous reports suggested that Plk1 depletion in mice leads to a non‐mitotic arrest in early embryos, we show here that the bi‐allelic Plk1 depletion in mice certainly results in embryonic lethality due to extensive mitotic aberrations at the morula stage, including multi‐ and mono‐polar spindles, impaired chromosome segregation and cytokinesis failure. In addition, the conditional depletion of Plk1 during mid‐gestation leads also to severe mitotic aberrancies. Our data also confirms that Plk1 is completely dispensable for mitotic entry in vivo. On the other hand, Plk1 haploinsufficient mice are viable, and Plk1‐heterozygous fibroblasts do not harbor any cell cycle alterations. Plk1 is overexpressed in many human tumors, suggesting a therapeutic benefit of inhibiting Plk1, and specific small‐molecule inhibitors for this kinase are now being evaluated in clinical trials. Therefore, the different Plk1 mouse models here presented are a valuable tool to reexamine the relevance of the mitotic kinase Plk1 during mammalian development and animal physiology.

Polo-like kinase 1 has a canonical kinase domain at the Nterminus. The regulatory domain, so-called the "polo-box domain" (PBD), is located at the C -terminus, and it confers the distinctiveness of the Polo-like kinase family [12,32,33]. The PBD is a unique structure that allows the recognition of proteins by a "pincer-like" 3D structure [34][35][36]. The mechanism by which Plk1 binds to its substrates is mainly based on the recognition of residues previously phosphorylated by other kinases [35,37]. Once Plk1 and the substrate are bound together, Plk1 can phosphorylate the substrate at neighboring residues. The recognition site is mostly based on the consensus sequence Ser-[pSer/ pThr]-[Pro/X] [35]. Although this "phospho-priming" mechanism is the most common way for Plk1 to find substrates, there are also formal demonstrations that the Plk1 can bind substrates regardless of their phosphorylation status [36,38].
Polo-like kinase 1 inactivation in somatic cells, either by means of small-interference RNA or by chemical inhibition, leads to spindle aberrations and problems in chromosome segregation and eventually to cell death due to a prolonged mitotic arrest [39][40][41]. In addition, Plk1 genetic depletion during animal development also promotes similar mitotic phenotypes and consequently animal death. Mutant alleles of polo, the Drosophila orthologue, leads to monopolar spindles in larval neuroblasts [2], and a hypomorphic polo allele shows cytokinesis failure in spermatocytes [42,43]. Similarly, depletion of the Plk1 yeast orthologues (Cdc5 in budding yeast, or plo1 in fision yeast) leads to proliferation arrest due to division alterations [44].
In mammals, Plk1 is expressed mainly in proliferating tissues [45,46]. In addition, this kinase is frequently found overexpressed in many tumors, and this feature often correlates with poor prognosis [47,48]. Consequently, Plk1 is currently considered a bona fide cancer therapeutic target. There is a growing collection of small compounds that are able to inhibit the kinase activity of Plk1 with high specificity. Several of these inhibitors are currently tested in chemotherapy clinical trials for different cancer types [49][50][51][52]. Therefore, any data resulting from Plk1 depletion in a mammalian model are of high value in the evaluation of Plk1 kinase as an anticancer therapeutic target.
In recent years, two mouse strains with genetically altered expression of Plk1 were reported. In 2008, Junjie Chen and colleagues described that embryos homozygous for a Plk1 genetrap allele did not show any mitotic alterations. Surprisingly, cells appear to be arrested during interphase [53]. The other mouse model, reported in 2011 [54], is an inducible knock-down of Plk1 by inserting a short-hairpin RNA (shRNA) in the murine Rosa26 locus. This depletion strategy is conditional, allowing depletion in the adult animal. However, because depletion is driven by shRNA, it is difficult to evaluate the effects of full Plk1 depletion in this model, and embryonic development was not studied.
To clarify the effects of Plk1 depletion during embryonic development, we have generated a gene-trap allele, targeting intron 2 of the mouse Plk1 locus. Our data confirms that Plk1 is an essential gene in mammals and Plk1-null mice die at the morula stage of embryo development due to severe mitotic aberrancies. Interestingly, Plk1 haploinsufficiency is fully compatible with animal life and Plk1(+/À) animals develop normally, as shown in previous reports [53]. We have also generated a conditional knock-out mice, to allow Plk1 depletion after embryonic mid-gestation. Conditional depletion of Plk1 at embryonic day E12.5 also leads to mitotic aberrations. Concomitantly, mouse embryonic fibroblasts (MEFs) derived from the Plk1(+/À) animals do proliferate with similar kinetics as their wild-type littermates and do not show any alteration in the cell cycle, whereas MEFs fully depleted of Plk1 show all the classical mitotic alterations already described. Thus, here we show that Plk1 is essential for the mammalian embryonic development, and its depletion leads to mitotic alterations and lethality at different stages during mammalian development.

Plk1-deficient embryos show high levels of mitotic aberrancies
We have generated a Plk1(+/À) mutant mouse by using homologous recombination in Embryonic Stem (ES) cells with a specific insertion of a β-geo cassette into intron 2 of the mouse Plk1 locus (Fig. 1A). No viable Plk1(À/À) mice were born from crosses between Plk1(+/À) heterozygous mice, and homozygous embryos were not found at any stage after implantation (E8.5) (Fig. 1B). We then tested the viability of the Plk1 null embryos at day E1.5 by isolating them immediately after fertilization and tracking their development in an in vitro culture. As depicted in Fig. 1C, both Plk1(+/+) and Plk1(+/À) embryos develop well in vitro during 4 days, reaching a morula like stage. The insertion of the β-geo cassette was verified by PCR (Fig. 1D), and the Plk1 expression during these initial stages was monitored by β-galactosidase staining activity (Fig. 1E), as well as immunofluorescence with Plk1-specific antibodies ( Fig. 2B and C). Plk1(À/À) mutant embryos arrest at the morula stage by E3. A detailed immunofluorescence microscopic analysis shows that mitotic cells in the Plk1(+/+) embryos show symmetric bipolar spindles, with chromosomes properly arranged in the metaphase plate (Figs. 2B and 3A). In anaphase and telophase cells, sister chromatids seem to be equally segregated to each daughter cell (Fig. 2B). In contrast, Plk1 null embryos displayed a substantial arrest of cells in a prometaphase-like stage (42.2% in average), with incomplete metaphase plates and misaligned chromosomes (Fig. 3B). Additionally, no anaphase and telophase figures were observed, confirming the impairment of chromosome segregation. Most of the mitotic arrested cells, in the Plk1 null embryos, displayed monopolar (34.7% in average) or multipolar spindles (13.0% in average) with poorly focused spindle poles (52.3% in average). Consequently, cells also showed misaligned chromosomes ( Fig. 3B arrow heads), thus confirming the role for Plk1 in the establishment of bipolar spindles and proper chromosome alignment also during the embryonic development. Interestingly, few Plk1 null cells showed enlarged nucleus with an increased number of centromeres (as depicted by anti-centromere antibody (ACA) staining) when compared with Plk1(+/+) embryos or even the neighboring cells in the same blastocyst ( Fig. 3Carrow). These cells most likely progress through mitosis and, as a consequence of cytokinesis defects, form polyploid cells. These cytokinesis alterations might also explain the higher incidence of multipolar spindles we observed, as polyploid cells probably enter in the following round of division with an extra number of centrosomes. All these cellular defects eventually can lead to cell death, thus explaining the Plk1(À/À) embryo lethality by E3.5-E5.5.

Plk1 haploinsufficient MEFs do not show any cell cycle alteration
Because Plk1 haploinsufficient animals were born and develop normally, we were able to obtain MEFs from Plk1(+/À) E13.5 embryos. When compared with littermates Plk1(+/+) MEFs, Plk1(+/À) expresses half the levels of Plk1 protein in both proliferating and arresting cultures (Fig. 4A). Plk1(+/À) cells did not display any alteration in their cell cycle profile, showing similar percentages within all cell cycle phases as depicted by DNA content via flow cytometry (Fig.4B). Additionally, Plk1(+/+) and Plk1(+/À) MEFs showed similar proliferation rates (Fig. 4C). This is consistent with the in vivo data where E16.5 heterozygous embryos showed similar levels of BrdU incorporation rates as their wild-type littermates (Fig. 4D). Finally, despite the reduction on Plk1 levels in the heterozygous MEFs, these cells did not show any significant change in ploidy when compared with Plk1(+/+) MEFs (Fig. 4E). Altogether, these data demonstrate that Plk1 haploinsufficiency is compatible with a normal cell cycle progression and proliferation rate.

Conditional depletion of Plk1 leads to mitotic arrest and cell death
In order to confirm that full depletion of Plk1 leads to mitotic arrest and cell death, we have also generated a conditional depletion model for Plk1, by using the CAG-Flpe transgenic strategy [55]. Two lox-P sites were inserted flanking the Plk1 exon 2 by homologous recombination in ES cells, generating the Plk1(lox) Early embryonic lethality of the Polo-like kinase 1 (Plk1) null mice. A: Scheme of the Plk1 gene-trap AS0407. A splicing acceptor sequence (SA) fused to a beta-geo (b-geo) cassette, and followed by a poly-adenilation signal (pA), are inserted into intron 2 of the Plk1 murine locus. B: Table showing the offspring Mendelian statistics of the Plk1 AS0407 strain. There are no Plk1(À/À) born animals, neither mid-gestation embryos (E8.5-10.5). When embryos are genotyped at E1.5 post-coitum, we find the adequate Mendelian ratio of each genotype. C: Postcoitum embryos are extracted from crosses between Plk1(+/À) animals. E1.5 embryos are cultured in vitro and followed in a daily time course under the microscope. Whereas Plk1(+/+) and Plk1(+/À) progress normally to the blastocyst stage, Plk1(À/À) embryos arrest at the morula stage by E3.5 and undergo apoptosis. D: After 4 days of culture, DNA is extracted from cultured embryos and subjected to PCR using specific oligonucleotides, in order to evaluate the insertion of the b-GEO cassette into the Plk1 locus, and determine the genotype regarding the gene trapping. E: Plk1(+/À) blastocysts express the beta-gal cassette and can be stained for beta-galactosidase in order to verify the expression of the transgene, whereas the Plk1(+/+) that do not harbor the b-GAL cassette remain stainless.
To analyze cell cycle progression in the absence of Plk1, Plk1(lox/lox) MEFs were synchronized in G0 by serum starvation, infected with Adeno-Cre viruses and then released in high-serum media and followed in a time course (Fig. 5D). Plk1(Δ/Δ) cells enter into the cell cycle and go through G1, S and G2 phases with no evident restrictions (Fig. 5E). However, 36 hours post-release in high-serum media, Plk1(Δ/Δ) MEFs showed an arrest in the mitotic phase, as measured by MPM2 staining, in contrast to Plk1(lox/lox) cells. Concomitantly, the cell cycle DNA content profile shifts towards the G2/M peak (Fig. 5F). Microscope examination of arrested cells revealed all the typical mitotic aberrancies due to Plk1 inhibition, such as monopolar and multipolar spindles, non-aligned chromosomes and lagging chromosomes ( Fig. 5G and H).
We then wanted to extrapolate these in vitro data into the mouse embryo, to verify that Plk1 is also essential during post-implantation embryonic development. Plk1 is ubiquitously expressed in the E14.5 mouse embryo as detected by immunohistochemical analysis (Fig. 6A). There are specific areas where Plk1 expression is higher such as the developing neuroepithelia, the fetal liver, intestines and other epithelial structures. Plk1 is highly expressed in the mitotic cells along the embryo, nicely decorating the spindle poles and the mid-body of the mitotic cells (Fig. 6B).
To test the effects of Plk1 depletion during mid-gestation, Plk1(+/lox) female mice were crossed with Plk1(+/lox) male mice. At E12.5 day post-coitum, tamoxifen citrate salt (0.3 mg per gram of animal body weight) was intraperitoneally injected into the pregnant females and the embryos were collected for histology 2 days after tamoxifen injection, at day E14.5. Although changes in the embryo size were not detected, a detailed inspection of the proliferative areas such as the neuroepithelia showed that Plk1 depletion impairs mitosis. Whereas the Plk1(+/Δ) embryos displayed a normal distribution of mitotic cells in the neuroepithelia (Fig. 6C open arrows), the Plk1(Δ/Δ) embryos showed an increase in aberrant mitotic figures, with evident alterations in chromosome alignment (Fig. 6Cclosed arrows). There was an associated increase in the mitotic cell population, as depicted by phospho-Ser10 Histone H3 immunostaining (Fig. 6Dupper  panel). The Plk1 depletion was confirmed in the aberrant mitotic cells by the absence of Plk1 staining (Fig. 6D lower panel, closed arrows). Interestingly, similar phenotypes could be observed in other proliferating tissues such as the fetal liver (data not shown). Altogether, our data confirm that Plk1 is essential during embryonic development at several stages, and Plk1 depletion during the mouse embryonic development leads to severe mitotic aberrancies such as multipolar and monopolar spindles, lagging chromosomes and cytokinesis failures.

Discussion
From yeast to mammals, Plk1 is one of the crucial mitotic regulators. In addition, Plk1 is a relevant molecule in the cancer clinics because it is highly expressed in many tumors, it is a poor prognostic marker and, more importantly, it is considered as a relevant therapeutic target. There are several smallmolecule inhibitors for Plk1 already in clinical trials for cancer therapy [50]. In this context, any relevant information of the physiological changes generated by Plk1 inhibition in mammals is important as a validation strategy for these inhibitors.  Complete genetic depletion of Plk1 leads to embryonic lethality at the morula stage. In the present study, the Plk1(À/À) morula cells recapitulate all the established phenotypes associated to Plk1 inhibition, from metaphases with misaligned chromosomes, to monopolar or multipolar spindles and even cytokinesis failure, as there are some interphase cells with double amount of centromeres. A previous report by J. Chen and colleagues also showed that the constitutive genetic depletion of Plk1 leads to embryonic lethality at the morula stage [53]. Surprisingly, these Plk1-depleted embryos had no mitotic alterations, and cells are somehow arrested at an interphase stage. The possible discrepancies between the results observed in both reports might result from using different gene-trap strategies. J. Chen Plk1 knock-out mice were generated by using a gene-trap in intron 9 of the Plk1 locus. Thus, there is still the chance of having a Plk1 protein expression in a truncated form only lacking the very C -terminal part of the protein and, therefore, still having some functionality. Another possible explanation for the non-mitotic phenotype is the fact that Plk1 was also described as an important driver for mitotic entry [56], thus, cells depleted in Plk1 would not be able to progress through G2 and stop before reaching mitosis. Although Plk1 participates in mitotic entry through phosphorylation of CDC25C, Myt1, Cyclin B and FoxM1 [57] and promotes recovery from DNA damage [58], it is not strictly required for mitotic entry [20,25,41]. The most probable explanation is that many of the proteins involved in the molecular network for mitotic entry are redundant in their function, as they play into several signaling feedback loops with no exclusive dependency in any of them. We validate that in our models all the cells are arrested at mitosis upon Plk1 depletion, either in the mice embryos or in vitro cell cultures, and there is no evidence of cell cycle stoppage before mitosis. All together, these data argue against an essential role for Plk1 in regulating mitotic entry and confirms an essential requirement for this protein during the formation of a functional bipolar spindle and chromosome segregation. The predominant phenotype described when Plk1 is acutely chemically inhibited is the generation of monopolar spindles due to impartment of centrosome maturation and separation [41]. Interestingly, Plk1-depleted embryos not only display monopolar spindles but also multipolar spindles with unfocused poles. This circumstance probably comes from the fact that Plk1 depletion does not happen as efficiently as the chemical inhibition of the catalytic activity of Plk1. Thus, cells might be able to go through mitosis with a minimal residual of Plk1 but not have enough Plk1 levels to accomplish cytokinesis properly, as this seems to be one of the more demanding activities of Plk1 [24,30,59]. Consequently, cells would exit mitosis being polyploid, and they probably enter in the subsequent mitotic round with an extra number of centrosomes. Another explanation is the fact that Plk1 interplays with centrosomal proteins such as Kizuna, Aurora A or TPX2, and this might lead to multipolar spindles as well [60,61].
Polo-like kinase 1 null embryos are able to reach the morula stage with 12 to 16 cells. Thus, several rounds of cell division happened despite the Plk1 depletion. This is most probably due to maternal contribution, because Plk1 activity inhibition by specific drugs stops the very first mitotic division in the mouse embryo [62]. Similarly, Plk1 is essential for the mouse oocyte maturation, as it is critical for the oocyte meiotic resumption [63].
Polo-like kinase 1 needs to be entirely depleted or acutely inhibited in order to provoke aberrant mitosis. Meanwhile full depletion of Plk1 leads to severe mitotic aberrancies in embryo development and in MEFs in culture, Plk1(+/À) mice are viable and they follow the correct Mendelian ratio. In vitro, Plk1(+/À) MEFs do not show any alteration in their cell cycle profile and have identical proliferation index as the wild-type littermates derived cells (Fig. 4B and C). Thus, Plk1 haploinsufficiency does neither compromise cell viability nor cell cycle progression. Concomitantly, Klaus Strebhardt and colleagues engineered a mouse with an inducible Plk1-shRNA [54]. The shRNA used in this report is only able to reduce Plk1 levels up to 70% in mouse fibroblasts. These mice, even though they are able to reduce Plk1 levels very efficiently in certain tissues (86% in testis, 72% in bone marrow and 60% in spleen), silencing is not complete in some others (stomach or colon), and mice live with no major drawbacks with the remnant Plk1. Indeed, Plk1 knocked-down animals do not differ significantly from the wild-type counter littermates in terms of histology and metabolism. Accordingly, MEFs derived from these inducible knock-down mice tolerate Plk1 reduction up to 90%, with no alteration in cell proliferation. Plk1 being an essential kinase for cell proliferation and animal life, a minimal threshold of Plk1 expression or function is sufficient for cell progression and thereby an interesting observation given the current relevance of Plk1 as a putative cancer target.
The Plk1 conditional targeting construct was assembled following the same strategy as previously reported in [64]. We flanked exon 2 of the murine Plk1 locus with loxP sequences, thus generating the Plk1(lox) locus. A neomycin phosphotransferase (neo) cassette was used for positive selection of ES cell clones. Recombinant ES cells and clones were selected by southern blot. To conditionally generate a null allele Plk1(Δ), we crossed the Plk1(+/lox) mice with transgenic mice that express the Cre recombinase fused to the estrogen receptor (Cre-ERT2) inserted in the collagenase locus (Fig. 5A). All animals were maintained in a mixed 129/Sv (25%) times CD1 (25%) times C57BL/6J (50%) background. The following oligonucleotides for the Plk1 cKO genotyping were used: EX1a_F (5 -ACAGCGACTTTGTATTTGTAGTTTTG-3 ) and IN1b_R (5 -CACTTTATGAATCCATTTCCTGTACC-3 ) for detecting the wild-type and lox alleles and IN2_R (5 -TTTCAGCTTAGTAAAGAGACA-3 ) for the depleted allele.

Histological and pathological analysis
Mice were housed at the pathogen-free animal facility of the Centro Nacional de Investigaciones Oncológicas (CNIO, Madrid) following the animal care standards of the institution. These animals were observed on a daily basis, and sick mice were humanely euthanized in accordance with the Guidelines for Humane End Points for Animals used in biomedical research. For histological observation, dissected organs were fixed in 10% buffered formalin (SIGMA-Aldrich St. Luis, MO, USA) and embedded in paraffin wax. Three-or five-micrometer-thick sections were stained with hematoxylin and eosin. Additional immunohistochemical examination of the tissues and pathologies analyzed was performed using specific antibodies against Plk1 (rat monoclonallaboratory made), phospho-histone H3 Ser10 (Merck-Millipore, Billerica, MA, USA. 06-570) and anti-BrdU (GE Healthcare Buckinghamshire, UK RPN202). To detect β-galactosidase activity, embryos were fixed for 5 minutes in PBS containing 1% formaldehyde, 0.2% glutaraldehyde and 1% serum. After fixation, embryos were rinsed with 1% serum in PBS and then transferred to a β-galactosidase reaction mixture (4 mM K 3 Fe(CN) 6 , 4 mM K 4 Fe(CN) 6 , 2 mM MgCl 2 and 1 mg/ml X-gal in PBS) at 37°C overnight. Embryos were washed once in PBS and kept at 4°C. Positive embryos were scored 48 hours after the reaction was initiated.

MEFs extraction, cell cycle profile and immunofluorescence
Mouse embryonic fibroblasts were prepared from E13.5 embryos and cultured using standard protocols [64]. Cell cycle profiling analysis was performed by detecting DNA content and EdU incorporation using the FACSCanto flow cytometry device (BD Biosciences Franklin Lakes, NJ, USA). EdU was added to exponential growing MEFs for 20 hours, and then cells were trypsinized and fixed in cold 70% ethanol. EdU staining protocol was performed following manufacture instructions (Click-iT, Invitrogen, Eugene, OR, USA), and DNA was stained with propidium iodide for 30 minutes. Mitotic index was determined by immunostaining with anti-MPM2 antibody (Millipore 05-368).
Immunofluorescence was performed by fixing the cells in 4% paraformaldehyde in PBS. After permeabilization with cold methanol, cells are blocked with 10% fetal bovine serum (FBS) in PBS and probed with specific antibodies against alphatubulin (DM1a, Sigma) and phospho-histone H3-Ser10 (Millipore 06-570). The secondary antibodies coupled to either Alexa488 or Alexa594 dyes are from Molecular Probes (Invitrogen). DNA is counterstained with DAPI, and cells are finally mounted in glass slides using Mowiol. Pictures were obtained using a confocal ultra-spectral microscope (Leica TCS-SP5-AOBS-UV).

Statistical analysis
Statistical analysis was performed using Student's t-test or analysis of variance (GraphPad Prism 5). All data are shown as mean ± SEM; probabilities of p < 0.05 were considered significant.