Figure 1.
Loss-of-function selection for invasion-suppressing genes in LNCaP cells.
A. Schematic summary of the screen. A high-throughput shRNA screening approach was used to identify genes whose knockdown induced tumor invasion, a process essential for metastasis. Highly invasive variants of LNCaP were selected using Matrigel-coated Boyden chamber invasion assays. B. Increased invasiveness induced by pools of shRNA clones over 3 rounds of selection. Upper panel: representative images of invasive GFP-expressing cells from control (pGIPZ vector) or shRNA (aliquot #2) after three selection rounds. Lower panel: invasive potential of pooled cells in each of 7 infection modules over three selection rounds, compared to LNCaP[vector] cells.
Figure 2.
FOXO4 expression in human CaP cell lines and metastatic tissues.
A. IB analysis of FOXO4 expression in human CaP cell lines (upper panel), quantified and compared to relative Matrigel invasiveness in the lower panel. Error bars, SE of three independent IB analyses quantified by densitometry (for FOXO4 levels) or Matrigel invasion assays. The relative FOXO4 level in DU145 was set at a value = 1 (dotted line). B. Oncomine studies FOXO4 RNA expression levels in BPH, primary-site (1°) CaP or metastases (mets) from LaTulippe et al. [45] and Yu et al. [46]. C. Kaplan-Meier plot analysis (http://www.cbioportal.org/public-portal/) of metastasis occurrence vs. time-to-onset in 37 CaP metastasis cases from Taylor et al. [8] in which 12 cases (32%) displayed FOXO4 downregulation and correlated with a more rapid appearance of metastases compared with the 25 cases that showed no changes in FOXO4 expression levels.
Figure 3.
FOXO4 regulates invasiveness in vitro and metastasis in vivo.
A. LNCaP cells stably transduced with shRNA or transiently transfected with siRNA against FOXO4 (vs. scrambled control) were tested for invasiveness. FOXO4 knockdown was assessed by IB (upper panel) and its effect on Matrigel invasiveness was quantified (lower panel). Error bars, S.E. of triplicate experiments. *, p<0.01. B. Ectopic expression of WT or constitutively-active (TM) FOXO4 decreases CWR22Rv1 invasiveness. Ectopic FOXO4 was assessed by IB (upper panel) and its effect on Matrigel invasiveness was quantified (lower panel). Error bars, S.E. of triplicate experiments. *, p<0.01. C. The local invasiveness of LNCaP[vector] (upper row) vs. LNCaP[shFOXO4] (lower row) was assessed for cells seeded onto Oregon Green 488-labeled gelatin, with cells labeled by DAPI. D. The same analysis as in C except comparing CWR22Rv1 stably expressing vector or Myc-FOXO4. E. Lung, liver, kidney and LN from individual mice tumored with LNCaP[vector] (control) or LNCaP[shFOXO4] were imaged using visible light (upper panel) or fluorescent light (lower panel). Arrows, LN and kidney metastases. F. Metastatic LN lesions from a LNCaP[shFOXO4] tumored mouse stained for H&E or GFP (by IHC). Triangles, examples of GFP-positive metastatic cells.
Table 1.
Incidence of macro-metastasis.
Figure 4.
Identification of candidate pro-metastasis genes regulated by FOXO4.
A. Venn diagram showing unique and commonly genes differentially expressed after FOXO4 knockdown in LNCaP cells (“cellLine”), primary-site tumors (“priTumor”) and KN metastases (“mets”). B. Heat map of common up- and downregulated genes differentially expressed after FOXO4 knockdown. C. Eight genes commonly upregulated in expression microarrays (upper panel) after FOXO4 knockdown (PIP, CAMK2N1, PLA2G16, ALDH1L1, VCX, VCX3A, APP, and PGC) were analyzed by qRT-PCR (lower panel). D. LNCaP co-transfected with siFOXO4 or scrambled siRNA (control) plus shRNAs specific for the 4 upregulated genes validated in C, were subjected to Matrigel invasion assays (upper panel). Error bars, S.E. of triplicate experiments. *, P<0.05; **, P<0.01. The relative knock of each gene was confirmed by qRT-PCR relative to non-specific shRNA controls (lower panel).
Figure 5.
FOXO4 regulates metastasis by binding to and suppressing RUNX2 transactivation ability.
A. Promoter regions of RUNX2, PIP, PGC, PLA2G16 and CAMK2N1 showing potential FOXO (DBE) and RUNX2 (RBS) binding sites relative to first exons. B. Matrigel invasion assay of LNCaP cells expressing control, FOXO4 or FOXO4 plus RUNX2 shRNAs. Error bars, S.E. of triplicate experiments. **, P<0.02. C. Relative RUNX2 RNA levels, as assessed by qRT-PCR in control shRNA vs. shFOXO4 LNCaP cells, primary tumors or LN metastases. RNA levels in each control condition were set to 1. Error bars, S.E. of triplicate experiments. n.s., not significant. Lysates of HEK293T cells transfected with HA-RUNX2 and Myc-FOXO4 were either analyzed by IB for HA, Myc or GAPDH, or immunoprecipitated with anti-myc and blotted with anti-HA (D), or immunoprecipitated with anti-HA and blotted with anti-Myc (E). F. Chromatin from LNCaP[vector] (control) or LNCaP[shFOXO4] cells were immunoprecipitated with control IgG or RUNX2 Ab, and the precipitated DNA subjected to qPCR using PIP promoter primers (Table S2). Error bars, S.E. of triplicates. **, P<0.01. G. Chromatin from HEK293T cells transfected with expression plasmids for RUNX2, RUNX2+FOXO4 or empty vector were immunoprecipitated with control IgG or HA Ab, then analyzed for PIP DNA by qPCR as in F.
Figure 6.
Control of invasiveness and pro-invasion RUNX2-regulated genes by AKT.
A. Ectopic expression of activated AKT (myr-AKT) increases LNCaP cell invasion. Error bars, S.E. of triplicate experiments. **, P<0.01. B. Relative mRNA levels of RUNX2, and RUNX2-regulated genes, PIP, PGC, MMP9, MMP13 and OP, assessed by qRT-PCR, in LNCaP cells stably expressing myr-AKT or an empty vector control. C. Model for PI3K/AKT negative regulation of FOXO4 and RUNX2 in the context of expression control of pro-invasion target genes.