Casein kinase II is required for proper cell division and acts as a negative regulator of centrosome duplication in Caenorhabditis elegans embryos

ABSTRACT Centrosomes are the primary microtubule-organizing centers that orchestrate microtubule dynamics during the cell cycle. The correct number of centrosomes is pivotal for establishing bipolar mitotic spindles that ensure accurate segregation of chromosomes. Thus, centrioles must duplicate once per cell cycle, one daughter per mother centriole, the process of which requires highly coordinated actions among core factors and modulators. Protein phosphorylation is shown to regulate the stability, localization and activity of centrosome proteins. Here, we report the function of Casein kinase II (CK2) in early Caenorhabditis elegans embryos. The catalytic subunit (KIN-3/CK2α) of CK2 localizes to nuclei, centrosomes and midbodies. Inactivating CK2 leads to cell division defects, including chromosome missegregation, cytokinesis failure and aberrant centrosome behavior. Furthermore, depletion or inhibiting kinase activity of CK2 results in elevated ZYG-1 levels at centrosomes, restoring centrosome duplication and embryonic viability to zyg-1 mutants. Our data suggest that CK2 functions in cell division and negatively regulates centrosome duplication in a kinase-dependent manner.


INTRODUCTION
Control of proper centrosome number is crucial for the fidelity of cell division (Gönczy, 2015). In animal cells, centrosomes organize microtubules to direct the formation of bipolar mitotic spindles that contribute to accurate segregation of genomic content. Centrosomes comprise two orthogonally arranged centrioles surrounded by a dense network of proteins termed pericentriolar material (PCM). Centrioles must duplicate exactly once per cell cycle to provide daughter cells with the correct number of centrosomes. Cells with abnormal centrosome number are prone to errors in DNA segregation and cytokinesis, leading to genomic instability and tumorigenesis (Godinho and Pellman, 2014).
Conversely, Protein phosphatase 2A (PP2A) plays a key role in centrosome duplication, and such role appears to be conserved in humans, flies and nematodes (Brownlee et al., 2011;Kitagawa et al., 2011;Song et al., 2011), underlying the coordinated kinase/ phosphatase action in regulating centrosome duplication. The C. elegans catalytic subunit of PP2A, LET-92, was identified by proteomic analysis of the RNA-binding protein SZY-20 that negatively regulates a core centriole factor ZYG-1 (Song et al., 2008(Song et al., , 2011. It has been shown that PP2A positively regulates centrosome duplication by promoting the stability and/or localization of ZYG-1 and SAS-5 in the C. elegans embryo (Kitagawa et al., 2011;Song et al., 2011). However, the counteracting kinase(s) to PP2A in centrosome duplication has not yet been identified. Interestingly, the proteomic approach that identified SZY-20 associating factors has listed both positive and negative regulators of centrosome duplication (Song et al., 2011;Stubenvoll et al., 2016). Thus, it is possible that one or more kinases interacting with SZY-20 might counteract PP2A in centrosome assembly.
One of the SZY-20 interacting factors we identified is the C. elegans Casein kinase 2 holoenzyme (CK2), an evolutionarily conserved serine/threonine protein kinase (Hu and Rubin, 1990). In general, CK2 acts as a tetrameric holoenzyme comprising two catalytic (CK2α) and two regulatory (CK2β) subunits (Niefind et al., 2009). CK2 targets a large number of substrates that are involved in cellular proliferation (Meggio and Pinna, 2003). Aberrant CK2α activity has been shown to result in centrosome amplification in mammalian cells (St-Denis et al., 2009). Consistently, elevated CK2 activities are frequently observed in many human cancers (Guerra and Issinger, 2008;Trembley et al., 2010). The C. elegans genome contains a single catalytic and regulatory subunit encoded by kin-3 and kin-10, respectively Rubin, 1990, 1991). In C. elegans, CK2 is also shown to function in a broad range of cellular processes, including cell proliferation in the germ line (Wang et al., 2014), Wnt signaling (de Groot et al., 2014), male sensory cilia (Hu et al., 2006) and microRNA targeting (Alessi et al., 2015). However, while genomewide RNAi screening indicated that CK2 is required for the viability of C. elegans embryos (Fraser et al., 2000;Sönnichsen et al., 2005), the function of CK2 during embryonic development remains largely undefined. In this study, we investigated the role of C. elegans CK2 during cell division in early embryos. We show that protein kinase CK2 is required for proper cell division and cytokinesis, and negatively regulates centrosome duplication.
As zyg-1 is essential for centrosome duplication, it is likely that kin-3(RNAi) restores embryonic viability to zyg-1(it25) by rescuing centrosome duplication. To test this, we examined zyg-1(it25); kin-3 (RNAi) embryos that formed bipolar mitotic spindles at the second mitosis to quantify the event of successful centriole duplication during the first cell cycle. Compared to control RNAi, kin-3(RNAi) led to a significant fold increase in bipolar spindle formation to zyg-1(it25) embryos at all temperatures examined (Fig. 1B,C). However, we observed no signs of monopolar spindle formation in kin-3 (RNAi) treated wild-type embryos. Therefore, kin-3(RNAi) restores embryonic viability to zyg-1(it25) through restoration in centriole duplication. We then questioned if KIN-3 regulates centrosome duplication as part of the CK2 holoenzyme or as a free subunit, independently of the holoenzyme. If the holoenzyme CK2 functions in centrosome regulation, we should observe similar effects on zyg-1 (it25) mutants by depleting KIN-10, the sole regulatory subunit (CK2β) of CK2 in C. elegans (Fig. 1B,C). Compared to control, kin-10(RNAi) resulted in significantly increased levels of embryonic viability and bipolar spindle formation to zyg-1(it25) embryos at the semi-restrictive temperature, 22.5°C. Thus, it is likely that the protein kinase CK2 holoenzyme has a role in centrosome duplication as a negative regulator. Fig. 1. Knocking down CK2 partially restores embryonic viability and bipolar spindle formation to zyg-1(it25). Depleting CK2 subunits by either kin-3 (RNAi) or kin-10(RNAi) leads to an increase in both (A) embryonic viability and (B) bipolar spindle formation to zyg-1(it25) mutants, which were examined at restrictive (24°C) and semi-restrictive temperatures (22.5°C). (A,B) Mean values are presented. Error bars are standard deviation (s.d). n is given as the number of embryos (A) or the number of blastomeres (B) at second mitosis. ***P<0.001, **P<0.01, *P<0.05 (two-tailed t-test). (C) Immunofluorescence of zyg-1(it25) embryos raised at 22.5°C illustrates mitotic spindles at second mitosis. kin-3(RNAi) or kin-10(RNAi) restores bipolar spindles to zyg-1(it25) embryos, but control embryos display monopoles. SAS-4 was used as a centriole marker. Scale bar: 10 µm.
Together, our results suggest that CK2 specifically regulates ZYG-1 levels at the centrosome. We then asked if increased ZYG-1 levels at first anaphase centrosomes resulted from a cell cycle shift by loss of KIN-3. To address cell cycle dependence, we recorded 4D time-lapse movies using embryos expressing GFP::ZYG-1-C-term that contains a C-terminal fragment (217-706 aa) of ZYG-1 and localizes to centrosomes (Peters et al., 2010;Shimanovskaya et al., 2014), starting from pronuclear migration through visual separation of centriole pairs at anaphase during the first cell cycle (Fig. 2D, Movie 1). Throughout the first cell cycle in kin-3(RNAi) embryos, we observed a 1.5-fold increase in PCM-associated GFP::ZYG-1 levels relative to control (Fig. 2E), but no significant difference in cytoplasmic levels between kin-3(RNAi) and control embryos ( Fig. 2E), suggesting that KIN-3 influences centrosomeassociated ZYG-1 levels throughout the cell cycle. However, we cannot exclude the possibility that KIN-3 might regulate overall ZYG-1 levels, and thereby influence ZYG-1 levels at centrosomes. While direct measurement of endogenous levels of ZYG-1 will address this question, we were unable to assess overall levels of ZYG-1 due to technical limits to the detection of endogenous levels of ZYG-1, largely owing to low abundance of ZYG-1 in C. elegans embryos.

CK2 is required for proper cell division in C. elegans embryos
Prior studies in C. elegans have shown that CK2 is required for embryonic viability (Fraser et al., 2000;Sönnichsen et al., 2005) and that CK2 functions in stem cell proliferation in germ line development (Wang et al., 2014). However, the specific role of CK2 in C. elegans embryos has not been examined in detail. As homozygous mutant animals for kin-3 or kin-10 arrest at late larval stages, we treated L4 stage worms by RNAi-feeding to knockdown the catalytic subunit (KIN-3/CK2α) of CK2, and examined the knockdown effect in early embryos. As reported previously (Alessi et al., 2015), animals exposed to kin-3(RNAi) for 24 h produced no significant embryonic lethality, although these animals exhibited other phenotypes such as sterility and protruding vulva (wormbase. org; Wang et al., 2014). Strikingly, when L4 worms were exposed for an extended period (36-48 h) to kin-3(RNAi), these animals produced significant and reproducible embryonic lethality (Fig. 4A, Table S2) and a reduced number of progeny produced for a 24 h period (Fig. 4B). We also observed that kin-10(RNAi) led to a similar level of embryonic lethality (Fig. 4A, Table S2), suggesting that protein kinase holoenzyme CK2 is required for C. elegans embryogenesis.
KIN-3, the catalytic subunit of CK2 exhibits dynamic changes in subcellular localization during the cell cycle To determine how CK2 might function in the early cell cycle, we examined the subcellular localization of KIN-3 in early embryos. Live imaging of embryos expressing KIN-3::GFP revealed dynamic localization patterns during cell cycle progression (Fig. 5, Movie 4). During interphase and early mitosis, KIN-3 is enriched at nuclei. Upon nuclear envelope breakdown (NEBD), KIN-3 appears to localize to the metaphase spindle and remain associated with spindle microtubules and centrosomes throughout mitosis (Fig. S6). During later stages of cell division, KIN-3 is highly enriched in the cytokinetic midbody (Fig. 5A,B), which is evidenced by colocalization of KIN-3::GFP and mCherry-tagged plasma membrane (Green et al., 2013;Kachur et al., 2008) (Fig. 5B). KIN-3::GFP localizes within a spherical area associated with the cytokinetic furrow that is surrounded by the plasma membrane supporting that KIN-3 localizes to the midbody. kin-3(RNAi) nearly abolished KIN-3::GFP expression (Fig. 5A), suggesting that KIN-3::GFP expression represent localization patterns specific to KIN-3.
Given our finding on KIN-3 localization at the midbody and its related role in cytokinesis, we asked if KIN-3 functions at the midbody to regulate cell division. The Aurora/Ipl1p-related kinase AIR-2 regulates multiple steps of cell division including chromosome segregation and cytokinesis (Schumacher et al., 1998). In particular AIR-2 plays a key role in cytokinesis through ZEN-4, both functioning at the midbody (Kaitna et al., 2000;Severson et al., 2000). To address the possible role of CK2 at the midbody, we tested a genetic interaction between kin-3 and air-2 or zen-4 (Fig. S7, Table S2). kin-3(RNAi) led to an increase in embryonic lethality of air-2(or207) or zen-4(or153) (Severson et al., 2000), suggesting that kin-3 has a positive genetic interaction with air-2 and zen-4, two genes associated with the midbody. Thus, it seems likely that KIN-3 function in early cell cycle as a component of the midbody structure.

CK2 dependent phosphorylation likely functions in centrosome duplication and the cell cycle
To further support the CK2 holoenzyme-dependent regulation of zyg-1, we tested whether chemical inhibition of CK2 could restore embryonic viability to zyg-1(it25). We used the highly selective chemical inhibitor of CK2, 4,5,6,7-tetrabromobenzotriazole (TBB) that competes for binding at the ATP-binding site of CK2 (Sarno et al., 2001;Szyszka et al., 1995). It has been reported that TBB specifically abolishes CK2-dependent phosphorylation through in vitro kinase assay without affecting the expression levels of CK2 subunits (Alessi et al., 2015;Pagano et al., 2008;Sarno et al., 2001;Szyszka et al., 1995;Wang et al., 2014;Yde et al., 2008). Compared to DMSO controls, TBB treatment partially restores embryonic viability and bipolar spindle formation to zyg-1(it25) animals ( Fig. 6A,C,D). Consistent with CK2 depletion by RNAi, we also observed that TBB-treated embryos possess increased levels of ZYG-1 at centrosomes (1.48±0.57-fold; P<0.001) compared to the DMSO control (Fig. 6E,F). While TBB produced no significant effect in embryonic viability, TBB treated animals produced a significantly reduced number of progeny (Fig. 6B) that was also observed by RNAi knockdown (Fig. 4B). Cytological analysis further confirmed that embryos treated with TBB exhibit cell cycle defects including detached centrosomes, DNA missegregation and abnormal PCM morphology (Fig. 6G). Taken together, our data suggest that the protein kinase CK2 holoenzyme functions in centrosome duplication and cell division, likely through CK2 holoenzyme-dependent phosphorylation.
based depletion of CK2 results in abnormal cell divisions, including chromosome missegregation, a delay in cell cycle progression, cytokinesis failure, as well as aberrant centrosome behaviors. Furthermore, treating embryos with TBB, a chemical inhibitor of CK2, produces similar effects on early cell divisions to those by RNAi knockdown. TBB inhibits CK2 kinase activity via competitive binding to the ATP-binding site of CK2 (Sarno et al., 2001) and it has been reliably utilized for the specific inhibition of CK2 function in the C. elegans system (Alessi et al., 2015;Wang et al., 2014). Thus, it seems likely that protein kinase activity of the CK2 holoenzyme is responsible for proper cell division.
The pleiotropic phenotypes caused by CK2 depletion are likely a combined outcome of multiple substrates targeted by CK2 (Meggio and Pinna, 2003). The most prevalent phenotype we observed is abnormal chromosome segregation (Fig. 4C, 68%, n=145), including chromosome misalignment at metaphase, lagging DNA at anaphase, and extra DNA during early cell divisions in C. elegans embryos. Such roles appear to be evolutionarily conserved in yeast and human cells (Peng et al., 2011;St-Denis et al., 2009). For example, CK2 is known to phosphorylate kinetochore factors (Ndc10, Mif2/CENP-C) in budding yeast (Peng et al., 2011), and the microtubule plus-end tracking protein CLIP-170 that ensures kinetochores attachment to mitotic spindles in human cells (Li et al., 2010). In addition, CK2 targets the chromosome passenger complex (survivin/BIR-1) that functions in chromosome segregation and cytokinesis in mammalian cells (Barrett et al., 2011;Kitagawa and Lee, 2015) and phosphorylates Mad2 to regulate the spindle assembly checkpoint in yeast (Shimada et al., 2009). While we do not know specific cell cycle regulators phosphorylated by C. elegans CK2 in early cell division, it appears that C. elegans CK2 functions in chromosome segregation and cell cycle progression, possibly through multiple factors targeted by protein kinase CK2.
The subcellular localization of KIN-3, the catalytic subunit of C. elegans CK2, appears to correlate well with the function of CK2 in cell division. Our confocal live imaging reveal that KIN-3::GFP from first metaphase to second metaphase in AB or P1 cell. Each dot represents an embryo. Bottom: wild-type embryo representing cell cycle stages used for quantification. Note that second metaphase of the anterior blastomere (b, arrow) initiates before second metaphase of the posterior blastomere (c, arrow). (A,B,E) Box ranges from the first through third quartile of the data, and thick bar represents the median. Dashed line extends 1.5 times the inter-quartile range or to the minimum and maximum data point. ***P<0.001 (two-tailed t-test). Scale bar: 10 µm. exhibits highly dynamic patterns of subcellular localization during the cell cycle. KIN-3 localizes to nuclei during interphase and early mitosis, and associates with mitotic spindles and centrosomes at mitosis. At later steps of cell division, KIN-3 becomes highly enriched as a distinct focus at the midbody. Our observations are consistent with subcellular localization of the catalytic subunit CK2α described in human cells, where CK2α localizes to nuclei (Penner et al., 1997), midbodies (Salvi et al., 2014), microtubules (Lim et al., 2004) and centrosomes (Faust et al., 2002). In the C. elegans germ line, the regulatory subunit KIN-10 is also shown to localize at centrosomes in mitotic cells (Wang et al., 2014). Many known midbody components are known to be required for proper cytokinesis (Green et al., 2012). A role for CK2 at the midbody is further supported by a proteomic survey of the mammalian midbody, where the regulatory subunit CK2β was identified (Skop et al., 2004). Our mass spectrometry analysis also suggests that KIN-3 physically associates with several known midbody proteins including ZEN-4 and CYK-4 (Table S1) Raich et al., 1998). In addition, we show that kin-3 exhibits a positive genetic interaction with air-2 and zen-4 that encode midbody proteins regulating cytokinesis (Kaitna et al., 2000;Severson et al., 2000), suggesting that CK2 functions as a part of the midbody structure. Therefore, the holoenzyme CK2 is required for early cell division in C. elegans embryos, which appears to be conserved between nematodes and mammals.
CK2 negatively regulates centrosome duplication in the C. elegans embryo Protein kinase CK2 was identified as part of the SZY-20 immunocomplex. Given that szy-20 is a genetic suppressor of zyg-1 (Kemp et al., 2007;Song et al., 2008), we speculated that CK2 might have a role in centrosome assembly. Our results show that inhibiting either subunit of CK2 restores centrosome duplication and embryonic viability to zyg-1(it25) mutants at the semi-restrictive temperature, suggesting the C. elegans CK2 holoenzyme functions in centrosome assembly as a negative regulator. In contrast, CK2 depletion did not restore embryonic viability to zyg-1(it25) mutants at the restrictive temperature, suggesting that CK2 is not a bypass suppressor but requires ZYG-1 activity. Furthermore, kin-3 appears to exhibit a positive genetic interaction with szy-20 in regulating centrosome duplication, consistent with their physical association. Inhibiting CK2 kinase activity with TBB leads to restoration of centrosome duplication and embryonic viability to zyg-1(it25) mutants, indicating that CK2-dependent phosphorylation plays a critical role in centrosome duplication. While it has been shown that aberrant CK2α activity leads to centrosome amplification in mammalian cells (St-Denis et al., 2009), it remains unclear how CK2 function is linked to centrosome assembly. Given that KIN-3 localizes at centrosomes, CK2 might influence centrosome assembly via phosphorylation of centrosome-associated factors, although it is also possible that CK2 targets centrosome regulators in the cytoplasm before they are recruited to centrosomes.
Our work suggests that C. elegans CK2 might function in centrosome duplication by targeting ZYG-1. Both RNAi and TBB mediated inhibition of CK2 function led to elevated levels of ZYG-1 at centrosomes, suggesting that CK2-dependent phosphorylation regulates ZYG-1 by controlling either localization or stability. ZYG-1 phosphorylation by CK2 might interfere with ZYG-1 recruitment to centrosomes. Alternatively, CK2-dependent phosphorylation of ZYG-1 may be a targeting signal for proteasomal degradation. Thus, inhibiting CK2 activity prevents ZYG-1 from degradation, increasing overall ZYG-1 abundance and thereby centrosomal levels. The latter is consistent with previous studies showing that phosphorylation is required for proteasomal degradation of the ZYG-1 homolog Plk4 in the mammalian system (Cunha-Ferreira et al., 2013;Guderian et al., 2010;Holland et al., 2010;Klebba et al., 2013). In either case, increased ZYG-1 levels at centrosomes by inhibiting CK2 activity, at least partially, explains how reducing CK2 activity restores centrosome duplication to zyg-1(it25) embryos. It has been known that CK2 is a constitutively active Ser/Thr kinase that favors a conserved target motif including acidic amino acid residues near the phosphorylated residue (Salvi et al., 2009). Although it is beyond the scope of our current study, identifying substrates and specific amino acid residues targeted by CK2 will help in understanding how CK2 regulates centrosome duplication, in particular, how protein kinase CK2 influences ZYG-1 levels at centrosomes in C. elegans embryos. In any event, our data suggest that the holoenzyme CK2 functions to influence ZYG-1 levels at centrosomes through its kinase activity and thus, we report the protein kinase CK2 as a negative regulator of centrosome duplication.
In this study, we investigated the role of the conserved protein kinase CK2 in early C. elegans embryos, and show that CK2 acts as a negative regulator of centriole duplication and is required for proper cell cycle progression and cytokinesis.

C. elegans strains and genetic analysis
The C. elegans strains used in this study were obtained from were obtained from the Caenorhabditis Genetics Center (CGC) (and indicated in Table S3) and maintained on MYOB plates seeded with E. coli OP50. All strains were derived from the wild-type Bristol N2 strain using standard genetics (Brenner, 1974;Church et al., 1995). Strains were maintained at 16 or 19°C unless otherwise indicated. A full list of strains used in this study is listed in Table S3. The KIN-3::GFP::3xFLAG strain (MTU5) was generated by standard particle bombardment (Praitis et al., 2001). The KIN-3::GFP::3XFLAG construct was acquired from TransgenOme (Sarov et al., 2012, construct number: 6236103120536928 D08), which contains 22kbp of the kin-3 5′UTR and 9kbp of the kin-3 3′UTR (Sarov et al., 2012). RNAi feeding was performed as previously described (Kamath et al., 2003), and the L4440 empty feeding vector was used as a negative control (Kamath et al., 2003).
For embryonic viability and progeny number assays, individual L4 animals were transferred to new plates and allowed to self-fertilize for 24 h at the temperatures indicated. For extended RNAi treatments (36-48 h), animals were transferred to a new plate in 24 h, and allowed to self-fertilize for an additional 24 h before removal. Progeny were allowed at least 24 h to complete embryogenesis before counting the number of hatched larvae and unhatched (dead) eggs.
Confocal microscopy was performed as described in (Stubenvoll et al., 2016) using a Nikon Eclipse Ti-U microscope equipped with a Plan Apo 60×1.4 NA lens, a Spinning Disk Confocal (CSU X1) and a Photometrics Evolve 512 camera. Images were acquired using MetaMorph software (Molecular Devices, Sunnyvale, CA, USA). MetaMorph was used to draw and quantify regions of fluorescence intensity and Adobe Photoshop CS6 was used for image processing. To quantify centrosomal signals (SPD-2:: GFP, TBG-1), the average intensity within a 25-pixel (1 pixel=0.151 µm) diameter region was measured within an area centered on the centrosome and the focal plane with the highest average intensity (corresponding to the centrosome) was recorded. The average fluorescence intensity within a 25pixel diameter region drawn outside of the embryo was used for background subtraction. Centriolar signals (ZYG-1, SAS-5, SAS-4) were quantified in the same manner, except that 8-pixel diameter regions were used.

Immunoprecipitation
Embryos were extracted by bleaching gravid worms in a hypochlorite solution [1:2:1 ratio of M9 buffer, bleach (5.25% sodium hypochlorite) and 5 M NaCl], washed with M9 buffer, flash-frozen in liquid nitrogen and stored at −80°C until use. For α-GFP immunoprecipitation experiments, 20 μl of Mouse-α-GFP magnetic beads (MBL, Naka-ku, Nagoya, Japan) were used per reaction. Beads were prepared by washing twice for 15 min in PBST (PBS; 0.1% Triton-X), followed by a third wash in 1× lysis buffer [50 mM HEPES ( pH 7.4), 1 mM EDTA, 1 mM MgCl 2 , 200 mM KCl, and 10% glycerol (v/v)] (Cheeseman et al., 2004). Embryos were ground in microcentrifuge tubes containing an equal amount of 1× lysis buffer supplemented with complete protease inhibitor cocktail (Roche, Indianapolis, IN, USA) and MG132 (Tocris, Avonmouth, Bristol, UK) and briefly sonicated prior to centrifugation. Samples were spun in a desktop centrifuge at 12,000 g twice for 20 min, collecting the supernatant after each spin. Protein quantification was then determined using a NanoDrop spectrophotometer (Thermo-Fisher, Hanover Park, IL, USA) and adjusted such that the same amount of total protein was used for each reaction. Beads were then added to the microcentrifuge tubes containing embryonic lysates. Samples were incubated and rotated for one hour at 4°C, and subsequently washed three times with PBST. Samples were then resuspended in 20 µl of a solution containing 2× Laemmli Sample Buffer (Sigma, St-Louis, MO, USA) and 10% β-mercaptoethanol (v/v) and boiled for five minutes. Mass spectrometry analysis was performed as described previously (Song et al., 2011).

TBB treatment
MYOB plates were first seeded with OP50 bacteria and allowed to dry overnight. The media was then supplemented with 0.5 mM TBB (Tocris, Avonmouth, Bristol, UK) dissolved in a solution of 50% DMSO. TBB was added to the surface of plates, such that the final concentration of TBB was 15 µM based on the volume of media, and allowed to soak and diffuse through media overnight. Final TBB concentrations were derived from (Wang et al., 2014). An equal volume of solution containing 50% DMSO was added to plates and used as a control.

Statistical analysis
All P-values were calculated using two-tailed t-tests assuming equal variance among sample groups. Statistics are presented as mean±s.d. unless otherwise specified. Data were independently replicated at least three times for all experiments and subsequently analyzed for statistical significance.