Identification of cell invasion regulators among differentially expressed genes in invasive melanoma cells

In order to identify genes that are functionally involved in regulating cell invasion, we investigated candidate genes whose orthologs were up-regulated in human melanoma cells that had been stimulated with TGF-β, exposed to hypoxia, or which overexpressed the transcription factor SOX9, and included genes up-regulated in migrating neural crest cells that had undergone EMT [16–18]. This resulted in a list of 104 differentially expressed human genes that were up-regulated under at least 2 conditions and for which a C. elegans ortholog could be identified (S1 Table).

RNAi-mediated knock-down of 32 of the 104 C. elegans orthologs resulted in a protruding vulva (Pvl) phenotype, which is characteristic of defects in vulval morphogenesis or in AC invasion. In order to specifically identify regulators of AC invasion, we scored BM breaching by the AC after RNAi-mediated knock-down. AC invasion was analyzed in L3 larvae after the 1° VPCs had divided twice, at the VPC 4-cell (Pn.pxx) stage of vulval development. If viability and fertility permitted, we scored the F1 progeny of RNAi-treated mothers, or else we scored the P0 animals that had been exposed to the dsRNA-producing bacteria from the L1 stage on. To visualize BM breaching by the invading AC, we used a LAM-1::GFP reporter that labels the 2 BMs separating the AC from the underlying vulval cells (Fig 1A) [5]. Among the 104 candidates tested, RNAi knock-down of 18 genes perturbed AC invasion (S1 Table). Notably, 3 genes required for normal AC invasion encode cell cycle regulators; cdc-6 encodes an essential component of the DNA pre-RC [12,15], cyd-1 encodes the only Cyclin D homolog in C. elegans [19], and cdk-12 encodes the Cyclin-dependent kinase required for activation of RNA polymerase II [20]. In particular, Cyclin D was shown to be a driver for invasion in a study investigating the development of thin melanoma [21]. After RNAi knock-down of cdc-6, the AC failed to invade in 32% of the cases (Fig 1A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Several pre-RC components are required for AC invasion.

(A) DIC images (left panels), corresponding green fluorescence image of the LAM-1::GFP reporter (middle panels), and merged images including the AC marker mCherry::moeABD (qyIs50) in magenta (right panels). The arrowheads in the DIC and merged panels point at the AC nucleus, and the arrows indicate the sites of BM breaching. In control animals fed with the empty RNAi vector (top panels), the BM was breached at the Pn.pxx stage. RNAi knock-down of cdc-6, cdt-1, mcm-7, orc-2, and orc-5 prevented BM breaching (see Table 1 for the numbers of animals scored). A lf mutation in the pre-IC component psf-1 did not prevent BM breaching despite the VPC proliferation arrest (bottom panels; see S2 Table for the number of animals scored). (B) AC invasion and VPC proliferation defects occur independently. The 4 different combinations of phenotypes observed after empty vector control and mcm-7 RNAi are shown; (from top to bottom) normal BM breaching and VPC proliferation, no BM breaching but VPC proliferation, BM breaching but VPC proliferation arrest, no BM breaching and VPC proliferation arrest. (C) Correlation between the AC invasion and VPC proliferation defects after mcm-7 RNAi. The frequencies of the 4 combinations of phenotypes among 110 animals scored are shown inside the boxes. (D) Expression of CKI-1 in P6.p using the zhEx334[egl-17>cki-1] transgene blocked the division of P6.p as reported [24] (left panels) but did not prevent BM breaching (middle panels). BM breaching was visualized with the LAM-1::mCherry marker (qyIs127) and the 1° VPCs were labeled with the ayIs4[egl-17>gfp] reporter. An L3 larva at the Pn.pxx stage and an L4 larva during vulval morphogenesis are shown. The asterisks in the bottom panel of (A) and in (B) and (D) label the nuclei of the undivided P6.p(x) cells. The right panels show the merged DIC and LAM-1::mCherry images including the egl-17>GFP signal in green. The scale bars are 5 μm. AC, anchor cell; BM, basement membrane; DIC, differential interference contrast; pre-IC, pre-initiation complex; pre-RC, pre-replication complex; VPC, vulval precursor cell.

https://doi.org/10.1371/journal.pbio.3001317.g001

In the following, we focused our experiments on studying the role of the pre-RC genes during cell invasion.

Several components of the pre-replication complex regulate AC invasion

Since the AC remains arrested in the G1 phase of the cell cycle while it breaches the BMs [8], the identification of the pre-RC component CDC-6 as a regulator of AC invasion was unexpected. We therefore tested if other pre-RC components are also required for BM breaching by the AC. Besides cdc-6, RNAi-mediated knock-down of several orc genes, as well as of cdt-1 and mcm-7 led to BM breaching defects (Fig 1A and Table 1). The penetrance of the invasion defects ranged from 15% for cdt-1 to 51% upon orc-2 RNAi (not taking into account the single cases observed after orc-3 and orc-4 RNAi). Among the mcm genes, only mcm-7 RNAi resulted in an invasion defect in 34% of the animals (Fig 1A and Table 1). However, the absence of AC invasion defects for other mcm genes tested could be due to low efficiencies of the RNAi knock-down.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Knock-down of several pre-RC components prevents BM breaching.

The genes encoding different pre-RC components were analyzed by RNAi for their potential to breach the BM at the P6.p.xx stage as described in Fig 2. The observed BM breaching defects are indicated in absolute numbers. Pvl indicates a protruding vulva and Ste a sterile phenotype at the adult stage.

https://doi.org/10.1371/journal.pbio.3001317.t001

Thus, several components of the pre-RC promote BM breaching by the AC. In addition to the AC invasion defects, RNAi of pre-RC complex components induced a proliferation arrest in different cells of the larvae, notably in the dividing VPCs adjacent to the AC (Fig 1B).

The DNA replication pre-initiation complex is not required for AC invasion

At the onset of the S-phase, the pre-RC transitions into the pre-initiation complex (pre-IC) through the recruitment of accessory proteins of the GINS complex (PSF-1, PSF-2, PSF-3, SLD-5) and additional factors, such as CDC-45, MCM-10, CDC-7, and PCN-1, which are necessary to initiate DNA replication [12,15,22]. In order to test if the initiation of DNA replication is required for AC invasion, we examined the AC invasion phenotype of a psf-1 null (lf) allele and performed RNAi against different pre-IC components. Except for a single case in cdc-45 RNAi-treated animals, we did not observe any invasion defects after RNAi knock-down or in mutants of pre-IC genes (Fig 1A bottom row and S2 Table). In some animals, the gap the AC created in the BMs of psf-1(lf) mutants was not expanded as the underlying VPCs had ceased to proliferate (Fig 1A bottom row), though the average size of the BM gap was not significantly changed in psf-1(lf) mutants (6.8 ± 1.8 μm in the wild type versus 7.7 ± 4.5 μm in psf-1(lf)) [23].

We conclude that components of the pre-IC are not necessary for AC invasion.

VPC proliferation is not required for AC invasion

Invasion by the postmitotic AC is linked to the differentiation and proliferation of the underlying 1° VPCs that produce guidance signals, which attract the AC ventrally [7]. It has previously been shown that arresting VPCs in S-phase by hydroxyurea (HU) treatment does not prevent AC invasion [7]. However, in that experiment, the timing of invasion could not be assessed relative to individual VPC fates, since HU treatment arrested the division of all VPCs. We therefore tested whether the AC invasion defects observed after down-regulation of pre-RC components might be caused by a proliferation arrest of the VPCs, and if preventing the 1° VPCs from entering into S-phase could inhibit AC invasion. We first examined on an animal per animal basis whether the AC invasion defects caused by mcm-7 RNAi correlated with the proliferation arrest of the VPCs. An analysis of 110 animals revealed no correlation, as mcm-7i invasion defects occurred independently of whether the VPCs had divided or not (Fig 1B and 1C). Next, we blocked the cell cycle of the 1° VPC (P6.p) in the G1 phase by overexpressing the CDK inhibitor cki-1 under control of the 1° lineage-specific egl-17 promoter (zhEx334[egl-17>cki-1::gfp; myo-2::mCherry] [24] and used a LAM-1::mCherry reporter to score BM breaching. In all egl-17>cki-1::gfp; lam-1>lam-1::mCherry animals with an undivided P6.p cell at the P5/7.p 4-cell (n = 16) or 8-cell stage (n = 16) the BMs were breached, indicating that AC invasion does not depend on S-phase entry of the 1° VPC (Fig 1D).

We conclude that entry of the 1° VPCs into S-phase is not required for AC invasion and, therefore, the AC invasion defects caused by mcm-7 RNAi are not due to a VPC cell-cycle arrest.

mcm-7 is expressed in the AC prior to invasion

In order to determine the expression pattern of the pre-RC components during AC invasion, we focused on mcm-7, because it encodes a core subunit of the MCM complex that is the last component recruited to the pre-RC. We used CRISPR/Cas9 engineering to insert a gfp tag upstream of the translational start codon in the mcm-7 locus along with 2 flippase (FLP) recognition target (FRT) sites within the gfp cassette. This permitted the tissue-specific inactivation of mcm-7 by FLP-induced recombination (zh118[frt::gfp::mcm-7], S1 Fig) [25,26]. zh118[frt::gfp::mcm-7] animals exhibited GFP::MCM-7 expression in the nuclei of proliferating cells, including many uterine cells and the VPCs (Fig 2A). In mid-L2 larvae, GFP::MCM-7 was expressed at comparably lower levels in the newly specified AC. During the late L2 and early L3 stage, GFP::MCM-7 expression in the AC further declined but was still detectable in mid-L3 larvae just prior to invasion. In late L3 larvae, after the BM had been breached, no GFP::MCM-7 expression was detectable in the AC (Fig 2A). Besides MCM-7, also CDT-1::GFP expression was consistently observed in the AC nucleus prior to and during invasion (S2 Fig). In contrast to GFP::MCM-7, CTD-1::GFP continued to be expressed in the AC after BM breaching in late L3 larvae (Pn.pxx stage). The adjacent uterine cells showed predominantly cytoplasmic CDT-1::GFP expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. mcm-7 acts in the AC to induce BM breaching.

(A) GFP::MCM-7 (zh118) and LAM-1::GFP expression beginning at the mid-L2 stage, shortly after AC specification, until the late-L3 (P6.pxx) stage after BM breaching are shown. DIC images are shown in the left panels; GFP::MCM-7 and LAM-1::GFP expression in green are shown in the middle panels. The dashed boxes indicate the areas around the AC shown at 3-fold magnification in the center panels. The right panels show the GFP images merged with the AC marker ACEL>mCherry (zhIs127) in magenta. The arrowheads point at the AC nuclei. (B) Pn.p and AC-specific RNAi of mcm-7 was performed as described in Materials and methods. The top row shows DIC and LAM-1::GFP expression in empty vector–fed control animals and the bottom row mcm-7 RNAi-treated animals at the Pn.p xx stage. BM breaching is indicated by the arrows. The table to the right shows the occurrence of BM breaching defects in absolute numbers. (C) FLP/FRT-mediated mcm-7 mosaic analysis using the zh118(frt:gfp::mcm-7) allele with heat shock induced flp expression to excise mcm-7, as described in Materials and methods and S1 Fig. DIC images are shown in the left panels, LAM-1::GFP expression used to score BM breaching together with GFP::MCM-7 in the middle panels, and the merged images including the AC marker cdh-3> mCherry::PLCδPH (qyIs23) (magenta) in the right panels. The top row shows a control animal at the Pn.pxx stage that did not show any GFP::MCM-7 excision. Examples for the 3 major classes of mosaic animals observed are shown underneath; excision in most uterine cells, excision in the 1° VPCs, and excision in both tissues. The arrows indicate the sites of BM breaching and the arrowheads point at the AC nuclei. (D) Quantification of the AC invasion defects in the different classes of FLP-induced zh118(frt::gfp::mcm-7) mosaic animals carrying no transgene, the wild-type, or the mutant rescue transgenes. The 3 major classes of mosaic animals and control animals without mcm-7 excision were scored separately for each genotype. The data shown are the combined results obtained in 3 biological replicates. n refers to the number of mosaic animals scored, and the error bars indicate the SEM. p-Values were determined with a nonparametric Kruskal–Wallis test with Dunn’s multiple comparison correction and are indicated as ** for p < 0.01 and n.s. for p > 0.5. See S1 Data for the numerical values used to generate the graph. The scale bars are 5 μm. AC, anchor cell; BM, basement membrane; DIC, differential interference contrast; FLP, flippase; FRT, FLP recognition target sequence; VPC, vulval precursor cell.

https://doi.org/10.1371/journal.pbio.3001317.g002

Since GFP::MCM-7 expression levels were highest in the newly formed AC, we tested whether mcm-7 is required for the specification or function of the AC during vulval induction. We first examined the expression of the b-HLH transcription factor HLH-2, whose expression onset in early L2 larvae instructs the AC fate [27]. mcm-7 RNAi had no significant effect on HLH-2::GFP expression levels in the AC of mid-L3 larvae (S3A and S3B Fig). The differentiated AC then expresses the proto-cadherin cdh-3 [28] and the EGF homolog lin-3, which induces the differentiation of the adjacent VPCs [4,29]. A transcriptional cdh-3>gfp reporter was not changed after mcm-7 RNAi, while lin-3>gfp expression was reduced (S3C–S3F Fig). However, the lin-3 levels in the AC after mcm-7 RNAi were sufficient to induce vulval differentiation, as the 1° vulval cell fate marker egl-17>gfp [30] was expressed in the 1° VPCs of all mcm-7 RNAi animals (S3G Fig). Thus, MCM-7 is not necessary for the specification of the AC fate or for the normal function of the AC during the induction of vulval differentiation.

mcm-7 acts in the AC to promote BM breaching

The observation that a proliferation arrest of the VPCs did not impair BM breaching raised the possibility that mcm-7 acts cell-autonomously in the AC to promote invasion, besides its established role in replication origin licensing in proliferating cells. To test whether the low GFP::MCM-7 expression levels observed in the AC of L3 larvae promote invasion, we first performed Pn.p cell and AC-specific RNAi of mcm-7 [31–33]. Notably, 16% of the animals treated with AC-specific mcm-7 RNAi showed a BM breaching defect, whereas none of the control RNAi-treated or Pn.p cell-specific mcm-7 RNAi animals exhibited BM breaching defects (Fig 2B). This analysis suggested a cell-autonomous role of mcm-7 in the postmitotic AC. To independently confirm these findings, we performed FLP/FRT-induced mosaic analysis in the zh118[frt::gfp::mcm-7] strain [26]. The FLPase was expressed under control of the heat shock–inducible hs-16-48 promoter (hs>flp) and induced by a single heat shock. Two cells with the potential to become the AC (Z1.ppp and Z4.aaa) are born during the third round of gonadal cell division in the mid to late L1 stage [27]. During the L2 stage, one of these 2 precursor cells adopts the AC fate, while the other cell becomes the ventral uterine (VU) precursor cell. To test if MCM-7 functions in the AC or in its mother cell (Z1.pp or Z4.aa), freshly hatched L1 worms were grown until the late L1/early L2 stage before heat shock induction of the FLPase (see Materials and methods). This protocol resulted in the development of mosaic animals that had lost GFP::MCM-7 expression either in most uterine cells, in the VPCs, or in both lineages (Fig 2C). Among the VPC mosaics, we focused on animals showing loss of GFP::MCM-7 expression in the 1° lineage because the 1° VPCs are necessary to induce AC invasion [7]. Around 29% of animals with a loss of GFP::MCM-7 expression in most uterine cells and 38% of animals lacking expression in both the uterus and the 1° VPCs showed a BM breaching defect (Fig 2C and white bars in 2D). Except for the AC, other uterine cells are not required for BM breaching [34]. By contrast, only 3.1% of the 1° VPC mosaics and 2.7% of the cases without an apparent loss of GFP::MCM-7 expression in any of the 2 tissues exhibited a BM breaching defect.

Taken together, the tissue-specific RNAi experiments and the mosaic analysis indicated that mcm-7 acts cell-autonomously in the postmitotic AC to positively regulate BM breaching. Though, the mosaic analysis could not distinguish if mcm-7 functions while the AC is adopting its fate or afterwards. It should also be noted that the highest penetrance of BM breaching defect after FLP/FRT-mediated inactivation of mcm-7 was 38%, suggesting that pre-RC activity may not be absolutely required for AC invasion.

The DNA helicase activity of the pre-RC is required for AC invasion

The MCM protein complex provides the helicase activity of the pre-RC that is necessary for DNA unwinding during the formation of a replication fork. To test if the MCM helicase activity is necessary for AC invasion, we introduced 4 point mutations (K₃₉₆S₃₉₇→AA and D₄₅₄E₄₅₅→AA) in the strongly conserved ATPase motif of MCM-7 (S4A Fig). These mutations have previously been shown to eliminate the helicase activity of the human MCM-4/6/7 complex [35]. By expressing the wild-type or mutant MCM-7 proteins tagged with CFP under control of the AC-specific lin-3 enhancer element (ACEL) [36] and the pes-10 minimal promoter (zhIs119[ACEL>mcm-7(wt)::cfp] and zhIs142[ACEL>mcm-7(mut)::cfp]) and simultaneously eliminating endogenous MCM-7 activity using the zh118[frt::gfp::mcm-7] allele to create hs>flp-induced mosaics, we could selectively assay MCM-7 function in the AC. MCM-7::CFP expression was first observed in late L1/early L2 larvae in the differentiating AC, and expression became more robust during the mid to late L2 stage after the AC had adopted its fate (S4B and S4C Fig). Remarkably, both wild-type and mutant rescue transgenes that were integrated as single copies at the same chromosomal site by MosSci [37] showed relatively faint MCM-7::CFP expression despite the use of the strong lin-3 ACEL enhancer element (S4B Fig). Therefore, MCM-7 expression appears to be down-regulated in the AC at the posttranscriptional level. Nevertheless, the wild-type ACEL>mcm-7(wt)::cfp transgene partially rescued the AC invasion defect of mcm-7 mosaic animals (Fig 2D, black bars). Since no ACEL>MCM-7::CFP expression was observed before the birth of the AC during the L1 stage, these data further support the conclusion of the mosaic analysis and tissue-specific RNAi experiments, which indicated that MCM-7 acts cell autonomously in the postmitotic AC. By contrast, the helicase-deficient transgene ACEL>mcm-7(mut)::cfp failed to restore the BM breaching activity to the AC (Fig 2D, gray bars).

Therefore, the MCM helicase activity is not only required for the initiation of DNA replication, but also for BM breaching by the AC. Alternatively, the ATP-binding motif in MCM-7 may perform another MCM helicase-independent function during AC invasion.

mcm-7 controls BM breaching in the G1-arrested AC

It was previously reported that the AC must remain arrested in the G1 phase of the cell cycle in order to adopt the invasive cell fate [8]. We therefore examined whether the pre-RC controls the cell cycle state of the AC by examining the expression of cell cycle reporters. A transcriptional reporter for the G1 phase marker cyd-1 cyclinD (zhIs131[cyd-1>gfp]) was expressed in the AC of mcm-7 RNAi-treated animals at similar levels as in control animals (Fig 3A). Conversely, the expression of the S phase marker rnr-1::gfp was undetectable in the AC of control and of mcm-7 RNAi-treated animals (Fig 3B) [38]. Consistent with these findings, the histone deacetylase hda-1 that is induced in the AC upon G1 arrest and is required for the invasive fate remained expressed after mcm-7 knock-down (S5 Fig) [8].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. mcm-7 controls invasion in the G1-arrested AC.

(A) Expression of the G1-phase marker cyd-1>gfp in an empty vector–treated control (top row, n = 19) and after mcm-7 RNAi (bottom row, n = 25). (B) Expression of the S-phase marker RNR-1::GFP in an empty vector–treated control (top row, n = 30) and after mcm-7 RNAi (bottom row, n = 22). (C) AC-specific expression of the CDK activity sensor in an empty vector–treated control (top row), after mcm-7 RNAi (middle row) and after nhr-67 RNAi (bottom row). (D) Quantification of the CDK sensor activity as the cytoplasmic to nuclear GFP intensity ratios, as described in Materials and methods. The numbers of animals scored are indicated in brackets, and the horizontal lines indicate the median values. Statistical significance was determined with a two-tailed t test for independent samples and is indicated as *** for p < 0.001. See S1 Data for the numerical values used to generate the graph. (E) Strong GFP::MCM-7 in the 3 noninvasive ACs formed after nhr-67 RNAi (bottom row) compared to the weak expression in the single AC of an empty vector–treated control animal (top row). The DIC images are shown in the left panels, the reporter expression in the middle panels, and in (C) and (E) the merged images together with the AC markers ACEL>mCherry (zhIs127) and cdh-3>mCherry::PLCδPH (qyIs23), respectively, in magenta are shown in the right panels. The arrowheads point at the AC nuclei. The scale bars are 5 μm. AC, anchor cell; DIC, differential interference contrast.

https://doi.org/10.1371/journal.pbio.3001317.g003

In addition to the analysis of cell cycle markers, we made use of a CDK activity sensor to directly quantify CDK activity in the AC [39]. This sensor consists of a portion of the DNA Helicase B containing 4 CDK phosphorylation sites, a nuclear export, and a nuclear import signal, N-terminally fused to an eGFP tag. Upon phosphorylation by CDKs, the sensor translocates from the nucleus to the cytoplasm. Thus, a low cytoplasmic to nuclear ratio of the eGFP signal indicates weak CDK activity. For our purpose, we expressed the CDK sensor specifically in the AC under control of the lin-3 ACEL enhancer/pes-10 minimal promoter fragment (zhIs130[ACEL>cdk sensor::egfp]). In control animals, the average cytoplasmic to nuclear fluorescence ratio in the AC was 0.51 ± 0.12 (Fig 3C and 3D). mcm-7 RNAi did not cause a change in sensor activity (average fluorescence ratio 0.54 ± 0.12), indicating that the pre-RC is not required for the G1 arrest of the AC. By contrast, RNAi-mediated knock-down of the nuclear hormone receptor nhr-67, which is required for the G1 phase arrest of the AC [8], resulted in a significant increase in CDK activity (average ratio 0.84 ± 0.14) (Fig 3C and 3D). Moreover, reducing nhr-67 levels resulted in the formation of multiple ACs that failed to invade, as reported previously [8], and expressed high levels of GFP::MCM-7 characteristic of proliferating cells (Fig 3E).

Hence, the down-regulation of mcm-7 did not affect the G1 arrest of the AC. Low expression levels of MCM-7 are associated with G1 arrest and AC invasion, whereas high MCM-7 expression levels accompany the loss of the invasive phenotype in proliferating ACs.

mcm-7 regulates invadopodia formation and extracellular matrix gene expression in the AC

Prior to BM breaching, the AC is polarized ventrally and extends actin-rich protrusions (invadopodia) toward the 1° VPCs, which are aligned at the ventral midline [5]. To assess whether the pre-RC controls invadopodia formation, we quantified the polarity of an F-actin reporter after RNAi knock-down of mcm-7. The probe consisted of the F-actin-binding domain of moesin fused to mCherry and was expressed under control of the AC-specific mk62-63 enhancer element of cdh-3 (mCherry::moeABD) [6]. In control animals, the mCherry::moeABD signal was polarized toward the ventral side and enriched at the tip of the invasive protrusion, where the AC breaches the BM (Fig 4A and 4B). Upon mcm-7 RNAi treatment, the F-actin reporter was depolarized and no properly organized invasive membrane domain was formed by the AC (Fig 4A and 4B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The pre-RC regulates invadopodia formation and extracellular matrix gene expression in the AC.

(A) F-actin localization in the AC visualized with the mCherry::moeABD reporter (qyIs50) in empty vector–treated controls (top row) and after mcm-7 RNAi (bottom row). The middle panels show single frames of 3D heat-maps obtained by reconstruction of the mCherry::moeABD signal as shown in S1 Movie. (B) Quantification of mCherry::moeABD polarity. Intensity plots along the dorso-ventral axis of the AC were generated as described in Materials and methods and in [74]. The numbers of animals analyzed are indicated in brackets. (C) him-4>ΔSP::GFP expression in the AC in control RNAi (top row) and in mcm-7 RNAi-treated animals (bottom row). (D) Quantification of him-4>ΔSP::GFP expression levels in the AC. (E) zmp-1>CFP expression in the AC in control RNAi (top row) and in mcm-7 RNAi-treated animals (bottom row). (F) Quantification of zmp-1>CFP expression levels in the AC. (G) zmp-1>CFP expression in control RNAi (top row) and in cdc-6 RNAi-treated animals (bottom row). (H) Quantification of zmp-1>CFP expression levels in the AC. (I) fos-1a>FOS-1a::YFP expression in control RNAi (top row) and in mcm-7 RNAi-treated animals (bottom row). (J) Quantification of fos-1a>FOS-1a::YFP expression levels in the AC. (K) egl-43L>GFP expression in control RNAi (top row) and mcm-7 RNAi-treated animals (bottom row). (L) Quantification of egl-43L>GFP expression levels in the AC. For each reporter analyzed, the left panels show the fluorescence signal and the right panels the merged DIC/fluorescence images. The arrowheads point at the AC nuclei. The numbers in brackets in the graphs refer to the numbers of animals analyzed, and the horizontal lines indicate the median values. Statistical significance was determined with a Mann–Whitney test and is indicated as n.s. for p > 0.05, * for p < 0.05, ** for p < 0.01, and *** for p < 0.001. See S1 Data for the numerical values used to generate the graphs in (B), (D), (F), (H), (J), and (L). The scale bars are 5 μm. AC, anchor cell; DIC, differential interference contrast; n.s., not significant; pre-RC, pre-replication complex.

https://doi.org/10.1371/journal.pbio.3001317.g004

The invasive AC phenotype is characterized by the expression of several proinvasive genes, such as the matrix MMP zmp-1 and the hemicentin him-4, which is secreted by the AC to remodel the BM [7]. RNAi knock-down of mcm-7 decreased expression of the him-4>ΔSP::gfp and zmp-1>cfp reporters in the AC (Fig 4C–4F). The reduction in proinvasive gene expression was not limited to the pre-RC component mcm-7 but was also observed after RNAi knock-down of cdc-6 (Fig 4G and 4H). Since zmp-1 and him-4 expression are regulated by the AP-1 transcription factor fos-1 and the zinc finger transcription factor egl-43 [28,40], we analyzed the expression patterns of these 2 transcription factors. Neither fos-1 nor egl-43 expression was changed after mcm-7 RNAi (Fig 4I–4L). These data suggested that the pre-RC acts downstream of or in parallel with fos-1 and egl-43 to activate zmp-1 and him-4 expression in the AC. In order to test whether mcm-7 might be regulated by fos-1, we examined GFP::MCM-7 expression after fos-1 RNAi. Even though fos-1 RNAi led to a penetrant invasion defect, GFP::MCM-7 expression was not affected by fos-1 RNAi, indicating that MCM-7 expression in the AC is independent of FOS-1 (S6 Fig).

In summary, the analysis of AC-specific gene expression indicated that the pre-RC positively regulates the expression of specific extracellular matrix genes that facilitate AC invasion.

Expression profiling of human cancer cells identifies several PI3K pathway genes as candidate pre-RC targets

To systematically identify the changes in gene expression caused by inhibiting the pre-RC, we turned to human cancer cell lines. Since the pre-RC genes were originally selected based on their differential expression in human melanoma cells, we examined the changes in gene expression caused by MCM7 depletion in the A375 melanoma cell line, and to extend our study to another cancer type, included the A549 lung cancer cell line in the expression analysis. Moreover, to exclude possible effects caused by changes in cell cycle progression and focus on the replication-independent functions of MCM7, we analyzed gene expression in cells arrested at the G1/S boundary. For this purpose, we generated A375 and A549 cells stably expressing an MCM7 shRNA or a scrambled shRNA as negative control, each under control of the doxycycline (Dox)-inducible enhancer/promoter. Addition of 1 μg/ml Dox to induce shMCM7 expression resulted in a 90% to 80% reduction in MCM7 protein levels in A375 and A549 cells, respectively (S7C–S7E Fig). Cell cycle analysis revealed that MCM7 depletion caused a slight enrichment of cells in the G1 phase (S7A and S7B Fig). Using a double thymidine (Thy) synchronization protocol to arrest cells at the G1/S boundary [41], we obtained cell populations, in which over 95% of the cells were arrested in the G1 phase of the cell cycle (S7A and S7B Fig). From these synchronized cell populations, mRNA was extracted to obtain global expression profiles by RNAseq for both cell lines.

RNAseq analysis revealed widespread changes in the mRNA expression profile of MCM7-depleted and G1/S phase-arrested cells (S8A and S8B Fig and S2 Data). Clustering of the MCM7-regulated genes revealed an enrichment of genes encoding extracellular matrix components and cell junction and plasma membrane proteins among the up-regulated genes (S8A and S8B Fig). To identify a core set of MCM7 target genes, we determined the overlap between the significantly up- and down-regulated genes in A375 and A549 cells (≥1.5-fold change with p < 0.01 and FDR <0.05). This analysis identified 259 genes up-regulated and 327 genes down-regulated after MCM7 depletion in both cell lines (S8C and S8D Fig). GO enrichment analysis of the molecular functions and biological processes of these core MCM7 target genes indicated that many of the up-regulated genes encode extracellular matrix, integrin, actin, and ß-catenin interacting proteins, while many of the down-regulated genes encode enzymes required for mitochondrial respiration and ribosomal RNA-binding proteins (S8E and S8F Fig).

In particular, we observed several components of the 1-phosphatidylinositol-3-kinase (PI3K) signaling pathway among the most strongly down-regulated genes (Fig 5A and S2 Data). Expression of the regulatory PIK3R2 (p85β) and PIK3R3 (p55γ) subunits as well as the AKT3 kinase that acts downstream of PI3K was reduced upon MCM7 knock-down in G1/S-arrested A375 and A459 cells, and the PIK3CD (p110δ) catalytic PI3K subunit was also significantly reduced in A549, but not in A375 cells (Fig 5A). To examine whether the decreased expression of PI3K pathway components manifests in a diminished activity of the PI3K pathway, we quantified the levels of phosphorylated, activated AKT kinase using a phospho-specific AKT antibody for western blot analysis. Insulin or EGF stimulation of serum-starved A375 and A549 cells that had previously been depleted of MCM7 resulted in an attenuated activation of the PI3K pathway (Fig 5B and S7C–S7E Fig; note that A375 cells did not respond to EGF stimulation).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The pre-RC positively regulates the PI3K pathway to promote invasion.

(A) Reduced expression of PI3K pathway genes and MCM7 after MCM7 knock-down in G1/S-arrested A375 cells (left) and A549 cells (right). For each gene, the fold change in expression between Dox-treated and untreated shMCM7 cells (red bars) and Dox-treated and untreated shScramble cells (blue bars) is shown (average of 3 biological replicates; see S2 Data). (B) Phospho-AKT protein levels in serum-starved and G1/S-arrested A375 (left) and A549 cells (right) stimulated with EGF or insulin. For each condition, pAKT levels in shMCM7 and shScramble cells grown in the presence and absence of Dox were quantified on western blots. The pAKT levels were normalized to the highest value in each experiment, and the averaged values of 3 independent biological replicates are shown (see S7C and S7D Fig). (C) Endogenous AAP-1::GFP and (E) GFP::AGE-1 expression in control (top rows) and mcm-7 RNAi-treated C. elegans larvae (bottom rows) at the Pn.pxx stage. The arrowheads point to the AC labeled with the mCherry::moeABD reporter (qyIs50). For each reporter, the left panels show the GFP signal in green and the right panels the merged GFP (green) and mCherry (magenta) signals. (D) Quantification of AAP-1::GFP and (F) GFP::AGE-1 levels in the AC. (G) Fraction of animals exhibiting polarized AAP-1::GFP and GFP::AGE-1 localization. (H) BM breaching defects after control, mcm-7 or egl-43 RNAi in the indicated genetic backgrounds. The numbers in brackets in each graph refer to the numbers of animals analyzed, the horizontal lines in (D) and (F) indicate the median values and error bars the standard deviation. Statistical significance was determined by two-way ANOVA followed by two-tailed t tests for independent samples and is indicated as n.s. for p > 0.5, * for p < 0.05, ** for p < 0.01, *** for p < 0.001 and as **** for p < 0.0001. In (D), the result of an F-test for variance is indicated by #### for p < 0.0001. See S1 Data for the numerical values used to generate the graphs in (A), (B), (D), and (F-H). The scale bars are 5 μm. AC, anchor cell; BM, basement membrane; Dox, doxycycline; n.s., not significant; pre-RC, pre-replication complex.

https://doi.org/10.1371/journal.pbio.3001317.g005

Moreover, we found in both cell lines significantly increased expression of TIMP2, which encodes an inhibitor of different MMPs, suggesting an involvement of MCM7 in the enzymatic BM remodeling (S2 Data). Mammalian cells express various MMPs, and C. elegans zmp-1 is orthologous to several of them [7]. mRNA levels of several MMPs were significantly changed after MCM7 knock-down, but there was no overlap between the 2 cell lines. MMP9 and MMP24 expression was down-regulated and MMP3 was up-regulated in A375 cells, while MMP10 expression was up-regulated only in A549 cells (S2 Data).

In summary, expression profiling of 2 human cancer cell lines identified several components of the PI3K/AKT pathway and the MMP inhibitor TIMP2 as candidate MCM7 targets. Since the analysis was performed with cells arrested at the G1/S boundary, the regulation of the PI3K pathway by the pre-RC appears to occur independently of cell cycle progression.

PI3K signaling in C. elegans promotes BM breaching by the AC

To examine the significance of the regulatory effect of MCM-7 on PI3K signaling, we returned to the C. elegans AC invasion model. In contrast to mammalians, the C. elegans genome encodes single catalytic (age-1) and regulatory (aap-1) PI3K subunits, respectively [42]. To observe age-1 and aap-1 expression, we created endogenous reporters by inserting gfp tags at the N-terminus of age-1 and the C-terminus of aap-1 using CRISPR/Cas9 genome engineering (zh161[gfp::age-1] and zh178[aap-1::gfp]). In untreated or empty vector–treated control RNAi animals, GFP::AGE-1 and AAP-1::GFP were both expressed in the AC and the surrounding cells at the Pn.pxx stage before and during invasion (Fig 5C and 5E). Interestingly, the GFP::AGE-1 and AAP-1::GFP proteins were asymmetrically localized in the AC and appeared polarized toward the invasive membrane. The levels of the catalytic PI3K subunit AGE-1 decreased in the AC upon mcm-7 RNAi (Fig 5C–5F). By contrast, we did not detect a significant change in the mean expression levels of the regulatory subunit AAP-1, but expression was more variable after mcm-7 RNAi (Fig 5F, p < 0.0001 in an F-test for variance). Moreover, the asymmetric distribution of AAP-1::GFP and GFP::AGE-1 in the AC was lost in most mcm-7 RNAi-treated animals (Fig 5G). By contrast, expression and localization of both PI3K subunits remained unchanged after fos-1 RNAi (S9 Fig).

Next, we tested whether overactivation of the PI3K pathway could rescue the invasion defects caused by down-regulation of mcm-7. For this purpose, we performed mcm-7 RNAi in animals carrying a gain-of-function mutation in the AKT homolog akt-1 [43] or a deletion allele of daf-18, which encodes the homolog of the PTEN tumor suppressor that antagonizes PI3K activity [42], and scored BM breaching using the LAM-1::GFP marker. The akt-1(mg144gf) and daf-18(ok480lf) mutations partially suppressed the invasion defects caused by mcm-7 RNAi, while the egl-43 RNAi invasion defects were not affected by akt-1(gf) or daf-18(lf) (Fig 5H).

Lastly, we examined whether PI3K signaling is necessary for AC invasion by observing BM breaching in age-1(mg44) null mutants. The first generation (F1) of homozygous age-1(mg44) animals segregated by heterozygous parents developed into healthy and fertile adults, presumably due to maternal rescue of larval AGE-1 function [44]. However, their offspring (F2) exhibited a severe developmental delay and formed sterile adults with a Pvl phenotype. In 23% (n = 61) of the F2 generation age-1(mg44) mutants carrying the lam-1::gfp marker, no BM breaching or only a very small BM gap was observed at the L4 (Pn.pxxx) stage (S10 Fig), suggesting that PI3K signaling is partially required for AC invasion.

Taken together, these data suggest that the positive regulation of the PI3K pathway by the pre-RC contributes to the invasive phenotype of the AC. Since hyperactivation of PI3K signaling only partially suppressed the mcm-7 invasion defects, the PI3K pathway is likely only one of several critical processes regulated by the pre-RC in the AC.

C. elegans culture and maintenance

C. elegans strains were maintained at 20°C on standard nematode growth plates as described [70]. The wild-type strain was C. elegans Bristol, variety N2. We refer to translational protein fusions with a:: symbol between the gene name and the tag, while transcriptional fusions are indicated with a > symbol between the promoter/enhancer and the tag. The genotypes of the strains used in this study are listed in S3 Table. Except for strain PS4444, which was provided by CGC, all strains were generated in this study.

Scoring AC invasion

AC invasion was scored at the Pn.pxx (4-cell) stage as described [8]. We monitored the continuity of the BM by fluorescence microscopy using the qyIs10[lam-1>lam-1::gfp] or qyIs127[lam-1>lam-1::mCherry] transgenes as markers.

Global and tissue-specific RNA interference

RNAi by feeding dsRNA-producing E. coli was performed in strains carrying the indicated markers in the rrf-3(pk1426) RNAi hypersensitive background as described in [71,72]. Synchronized populations of L1 larvae were obtained by hypochlorite treatment of gravid adults. If the RNAi treatment did not cause lethality or sterility, then the F1 generation was scored after 5 days, or else the P0 animals were analyzed after 30 to 36 hours of treatment. RNAi clones targeting genes of interest were obtained from the C. elegans genome-wide RNAi library [71] or the C. elegans open reading frame (ORFeome) RNAi library (both from Source BioScience). Genes not available in either of the 2 libraries were cloned into the L4440 vector by Gibson assembly using the oligonucleotide primers indicated in S4 Table. For AC- and Pn.p cell-specific RNAi, we used an rde-1(ne219); rrf-3(pk1426); qyIs10[lam-1>lam-1::gfp] strain carrying the qyIs102[fos-1a>rde-1] or the zhEx418[lin-31>rde-1] transgene, respectively [31,32]. The empty L4440 vector was used as negative control and an RNAi clone targeting egl-43 [40] was used as positive control.

Microscopy of C. elegans

Fluorescent and Nomarski images of C. elegans larvae were acquired with a LEICA DM6000B microscope equipped with a Leica DFC360 FX camera and a 63x (N.A. 1.32) oil-immersion lens, or with an Olympus BX61 wide-field microscope equipped with a X-light spinning disc confocal system using a 70-μm pinhole, a lumencor solid-state light source using a 60x Plan Apo (N.A. 1.4) lens and an iXon Ultra 888 EMCCD camera.

AC polarity and fluorescence intensity measurements

Fluorescence intensities of the different reporters were quantified using Fiji software with the built-in z-projection and measurement tools [73]. To measure the intensity of the different reporters, background subtracted summed z-projections spanning the width of the AC labeled with an mCherry marker were quantified. Quantification of mCherry::moeABD polarity in the AC was done using deconvolved wide-field or confocal image stacks spanning the width of the AC. Polarity plots were calculated from the summed z-projections of the AC as described in [74]. Quantification of the CKD sensor was done according to [39]; a mid-sagital section of the AC was filtered with Gaussian blur (sigma = 50) to reduce camera noise, and the average intensities in three randomly selected areas, each in the AC nucleus and cytoplasm, were used to calculate the cytoplasmic to nuclear intensity ratios.

Generation of reporter and rescue transgenes

For all constructs, fragments were PCRs amplified with Phusion DNA Polymerase (New England Biolabs) and were assembled by T4 ligase (New England Biolabs) or if Gibson assembly was used with Gibson assembly kit #E2611 (New England Biolabs). Details on the construction of the plasmids used to create the different transgenes are shown in S5 Table. The sequences of the oligonucleotide primers used can be found in S4 Table. All plasmids were verified by DNA sequencing. For each reporter, single-copy transgene insertions were created according to the MosSCI protocol [37] by microinjection of 50 ng/μl of the respective reporter plasmid together with 50 ng/μl pJL43.1 (mos-1 transposase) and 2.5 ng/μl pCFJ90 (myo-2>mCherry), 5 ng/μl pcFJ104 (myo-3>mCherry), and 10 ng/μl pGH8 (rap-3>mCherry) as coinjection markers.

Generation of the zh118[frt::gfp::mcm-7], zh161[gfp::age-1], and zh178[aap-1::gfp] alleles

The streamlined CRISPR/CAS9 method described [25] was used to generate endogenously tagged reporters. The 5′ and 3′ homology arm were amplified from N2 genomic DNA with primers indicated in S5 Table. These fragments were cloned into pMW75 digested with ClaI to create the repair templates pTD16, pTD60, and pTD72. pMW75 is a derivate of pDD282, in which the gfp cassette and the ccdB negative selections markers were replaced with a gfp cassette from pDD95.75 containing FRT-sites in introns 1 and 2. About 20 ng/μl of the resulting plasmids were injected together with 50 ng/μl of the respective single-guide RNA expression plasmids pTD17 (mcm-7 sgRNA#2 (CCACCACTTACAACACAGAC), pTD61 (age-1 sgRNA#1 TTA AGA TTT TAA GAT GTC TA) and pTD62 (age-1 sgRNA#2 TAA GAT TTT AAG ATG TCT AT) and pTD63 (age-1 sgRNA#3 GGT TCG ACT TCG AAA TGT CG) or pTD73 (aap-1 sgRNA#1 AAC TAT TCC GGG ATG GAT TT) and pTD74 (aap-1 sgRNA#2 TTT AAA TAT ATA ACT ATT CC) and 2.5 ng/μl pCFJ90 (myo-2>mCherry), 5 ng/μl pcFJ104 (myo-3>mCherry), and 10 ng/μl pGH8 (rap-3>mCherry) coinjection markers. The selection of the integrated CRISPR lines was carried out according to [25].

FLP/FRT-mediated mcm-7 mosaic analysis and AC-specific rescue

Mosaic analysis in the qyIs10[lam-1>lam-1::gfp]; qyIs23[cdh-3>mCherry::PLCδPH]; zh118[frt::gfp::mcm-7] strain was performed by FLPase expression from a heat shock–inducible promoter, using the zhEx621[hs-16-48>flp] transgene as described [26]. Late L1/early L2 larvae were heat shocked for 1 hour at 34°C and allowed to recover at 20°C. GFP::MCM-7 expression and AC invasion was scored 40 hours later. For AC-specific rescue, the wild-type zhIs119 and mutant zhIs142 transgenes were introduced into the zhEx621[hs-16-48>flp]; qyIs10[lam-1>lam-1::gfp]; qyIs23[cdh-3>mCherry::PLCδPH]; zh118 [frt::gfp::mcm-7] background and BM breaching was scored after heat shock induced flp expression, as described above.

Generation of Dox-inducible shMCM7 and shControl cells

The human cell lines A375 and A549 were tested to be free of mycoplasm and cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco 41966–029) supplemented with 10% FCS (Gibco 10500–064) and 1% Pen-Strep (Gibco 15140–122) according to standard mammalian tissue culture protocols and sterile technique.

Double stranded oligonucleotides for an shRNA targeting MCM7 (CAG AGG AGG ATT TCT ACG AAA), or a scrambled shRNA (TAA TGT ATT GGA ACG CAT ATT) were cloned into the Dox-inducible shRNA expression vector pRSIT17 (Cellecta # SVSHU6T17-L), which was then transfected in combination with pVSV-G, pMDL and pREV into HEK293T cells to produce lentiviral particles. Four days following transfection, the media from cells was collected, clarified by centrifugation, and filtered through a 0.45-μM filter to collect lentiviral particles. Subsequently, the particles were concentrated in Amicon Ultra tubes (Ultracell 100k, Millipore). A375 or A549 cells were transduced overnight with lentiviral particles (MOI of 1.4) in DMEM medium supplemented with 10 μg/ml polybrene. Three days after transduction, cells were grown in the presence of 1.2 μg/ml puromycin. One week after puromycin selection, the puromycin-resistant populations were frozen and kept as stocks used for the subsequent experiments.

Cell synchronization and cell cycle analysis

Cell synchronization with double Thy block [41] was performed by growing cells in medium supplemented with 2 mM Thy for 17 hours, followed by an 8-hour release in medium without Thy, after which cells were kept arrested in medium containing 2 mM Thy during the analysis. Cell cycle analysis by DAPI and EdU double labeling was done using the Click-iT assay (Thermo Fischer C10420) as described [75] and analyzed by flow cytometry (BD LSR II Fortessa).

RNAseq analysis of Dox-inducible shMCM7cells

Total RNA from Dox-inducible shMCM7 or parental A375 and A549 cells, each arrested in G1 by double Thy block and simultaneously treated for 72 hours with 1 μg/ml Dox, where indicated, was isolated using the RNeasy Mini Kit (#74104 Qiagen). Three biological replicates for each condition were analyzed. Quality control by Agilent Tape Station/Bioanalyzer, RNA library preparation, and whole exome sequencing were performed in triplicates by the transcriptomics service of the Functional Genomics Center Zürich (FGCZ). Using the HiSeq 2500 System (Illumina), around 200 million reads/lane were obtained. Bioinformatic analysis was done by the FGCZ using an in-house pipeline utilizing the open source tools STAR for read alignment, DESeq2 and EdgeR for differential gene expression, and GOseq for GO enrichment analysis. Bootstrap analysis with a resampling size of 1,000 was used to determine the significance of the overlap between the MCM7-regulated genes in A549 and A375 cells. The RNAseq data generated in this study are available at the NCBI Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE149523.

Western blot analysis

Cells were lysed in lysis buffer on ice (100 mM Tris/HCl, 150 mM NaCl, 1% Triton X, 1 mM EDTA, 1 mM DTT, 10 ml lysis buffer + 1 tablet protease inhibitor), scraped with a cell scraper, and snap frozen in liquid nitrogen. A volume of 100 μl of this mix was sonicated in a Bioruptor sonicator device (Diagenode), before 100 μl of 2x SDS loading dye were added. About 10 mg of protein extract were resolved by SDS-PAGE and transferred to nitrocellulose membrane. The membrane was blocked in 5% dried milk in 1x PBS plus 0.2% Tween 20 and incubated with the diluted primary antibodies overnight at 4°C. Secondary anti-rabbit or anti-mouse IgG antibodies conjugated to horseradish peroxidase (HRP) were used as the secondary antibodies. The HRP was detected by incubating the membrane with the SuperSignal West Pico or Dura Chemiluminescent Substrate (Thermo Scientific) for 4 minutes, before the signals were acquired on a digital chemiluminescence imaging system. The antibodies used for western blot analysis were anti-MCM7 (141.2 Santa Cruz), anti-Tubulin (ab18251 abcam), and anti-pAKT (44-621G Thermo Scientific). Quantification of western blots was done by measuring the band intensities in Fiji.

EGF and insulin stimulation

About 28,000 cells were seeded into 12-well plates and treated after 24 hours throughout the experiment with 1 μg/ml Dox where indicated. Around 48 hours after seeding, cells were additionally arrested at the G1/S boundary by growing cells in medium supplemented with 2 mM Thy for 17 hours, followed by an 8-hour release in medium without Thy [41]. Thereafter, growth medium was replaced with DMEM containing 2 mM Thy and lacking FCS. After 15 hours of starvation, cells were stimulated for 10 minutes with 100 ng/ml human EGF (E9644 Sigma) or with 10 nM human insulin (I9278 Sigma) before they were lysed and prepared for western blot analysis.