Fig 2: CircZNF609 interacts with ELAVL1 protein promoting ELAVL1 binding to some of its mRNA interactors(A) Integrative Genomics Viewer (IGV) tracks for ELAVL1 PAR-CLIP crosslinked reads (red) from Kishore et al. (2011) on ZNF609 exon 2.(B) qRT-PCR showing circZNF609 and GAPDH mRNA enrichment in ELAVL1 CLIP assay in RD cells. n = 1. Data are shown as means of enrichment versus input ± SD.(C) Venn diagram showing the intersection between ELAVL1 mRNA interactors from Lebedeva et al. (2011) and circZNF609 mRNA interactors.(D) qRT-PCR showing the enrichment of several RNAs after ELAVL1 RIP in RD cells in control condition (si-SCR) or upon circZNF609 knockdown (si-Circ). n = 3. p referred to mRNA enrichment between si-Circ versus si-SCR = 0.0044 (CKAP5 mRNA); and 0.0173 (UPF2 mRNA). Data are shown as means of enrichment versus input ± SD. The ratio of each sample versus control was tested by two-tailed paired Student’s t test.(E) qRT-PCR showing the enrichment relative to control condition (ELAVL1 LNA SCR) of several RNAs after ELAVL1 RIP assay in RD cells upon transfection of LNAs targeting circZNF609 on ELAVL1 binding sites (LNAs on ELAVL1 sites). n = 4. p referred to RNA enrichment between LNAs on ELAVL1 sites versus LNA CTR = 0.006 (circZNF609); 0.029(CKAP5 mRNA); 0.003(UPF2 mRNA); and 0.027(CCNB1 mRNA). Data are shown as means of enrichment versus input relative to control condition ± SD. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.See also Figure S2.

Fig 3: CircZNF609 knockdown destabilizes MT cytoskeleton, affecting mitotic progression and chromosome segregation(A) Left: examples of MT (in red; TUBA4A staining) progressive depolymerization phenotypes observed in mitotic RD cells after incubation on ice (20 min). Chromosomes are stained with DAPI (blue). Bar, 5 µm. Right: histograms represent the frequency of MT phenotypes after cold incubation in control cultures (si-SCR) and in cultures silenced for either circZNF609 (si-Circ) or CKAP5 (si-CKAP5). n = 3. Differences in phenotype distributions were statistically analyzed using the multiple ?2 test for trend (****p < 0.0001).(B) Left: exemplifying panels of progressive MT regrowth. RD cells were stained as for (A) after incubation on ice (40 min) and rewarming to 37°C. Right: histograms representing the profile of MT regrowth phenotypes 4 min after temperature up-shift to 37°C. n = 4. Multiple ?2 test for trend compare the phenotype distribution (***p < 0.001, ****p < 0.0001).(C) Left: immunofluorescence (IF) panels exemplify frequently occurring mitotic abnormalities; RD cells were stained as for (B). Right: histograms representing the frequency of multipolar spindles (top), chromosome mis-alignment (middle; arrowed in left) and mis-segregating or lagging chromosomes (bottom; arrowed in left) in control (si-SCR) and circZNF609-downregulated (si-Circ) RD cells. n = 4. The ?2 test compares each class in si-SCR versus si-Circ cultures (*p < 0.05).See also Figure S5.

Fig 4: Loss of circZNF609/CKAP5 mRNA interaction strengthens the anti-tumor effects of MT-targeting drugs(A) Left: qRT-PCR representing circZNF609 RNA levels in xenograft mice (n = 3 in circZNF-high group, black; n = 4 in circZNF-low group, red); data are shown as individual dots representing ??Ct values relative to GAPDH mRNA; horizontal bars represent the mean. Right: graph showing tumor volume (mm3) over time (days after mice injection) in circZNF-high (black) and in circZNF-low (red) mice.(B) Boxplots representing the percentage of survival of RD cells either in control condition (si-SCR, gray) or in circZNF609 knockdown (si-Circ, red), upon treatment with either vincristine (VCR; n = 5; p = 0.0051), Taxol (TX; n = 5; p = 0.0189), or nocodazole (NOC; n = 2; p = 0.0394). The ratio of each sample versus control was tested by two-tailed paired (VCR and TX) or unpaired (NOC) Student’s t test.(C) Left: representative western blot showing CKAP5 protein levels in SK-N-BE cells in si-SCR and si-Circ, and protein quantifications relative to ACTN1 (n = 3; p = 0.0023). Right: as in (B), in SK-N-BE cells; VCR: n = 4, p = 0.0200; TX: n = 4, p = 0.0030. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.(D) As in (C), in K562 cells. Left: ACTB hybridization used as loading control; n = 8; p = 0.0025. Right: VCR: n = 5, p = 0.0078; TX: n = 5. The ratio of each sample versus control was tested by two-tailed unpaired (protein analysis) or paired (cell survival analysis) Student’s t test.(E) Boxplots representing the percentage of survival of RD cells transfected with either the control LNA SCR (gray), the LNAs BSJ (red), or the LNA Ex2 (violet) upon treatment with either VCR (n = 4; p LNAs BSJ versus LNA SCR = 0.0057) or TX (n = 3; p LNAs BSJ versus LNA SCR = 0.0262). The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.(F) Boxplots representing the percentage of survival of RD cells upon transfection of either control LNA SCR (black) or LNAs targeting circZNF609 on ELAVL1 sites (LNAs on ELAVL1 sites, purple) upon treatment with VCR (n = 4; p = 0.0326). The ratio of the sample versus control was tested by two-tailed paired Student’s t test.(G) Cartoon depicting the molecular circuit involving circZNF609, CKAP5, and other mRNAs and the ELAVL1 protein.See also Figure S6.

Fig 5: CircZNF609 sustains translation and stability of some of its mRNA interactors via ELAVL1(A) Representative western blots and protein quantifications (relative to ACTB) and RNA levels (relative to GAPDH mRNA) of CKAP5 (from protein, p < 0.0001; from RNA, p = 0.0073), SRRM1 (from protein, p = 0.0178), UPF2 (from protein, p = 0.0002), and DDX27 in RD cells in control condition (si-SCR) or upon circZNF609 knockdown (si-Circ). n = 3. Data are shown as means of relative protein levels (or mean of fold changes) ± SD. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.(B) qRT-PCRs showing mRNA stability upon actinomycin D (ActD) treatment in RD cells in control condition (si-SCR, black) or upon circZNF609 knockdown (si-Circ, red). n = 3. Referred to CKAP5 mRNA stability, p = 0.0116 (2.5 h); 0.0385 (5 h). Referred to SRRM1 mRNA stability, p = 0.0482 (2.5 h). Data are shown as means of fold changes ± SD. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.(C) qRT-PCR showing the enrichment of circZNF609 interactors CKAP5, UPF2, SRRM1, and DDX27 mRNAs, and control GAPDH mRNA in a representative RPL22-FLAG RIP assay in RD cells in control condition (si-SCR) or upon circZNF609 knockdown (si-Circ). Data are shown as means of enrichment versus input ± SD.(D) Representative western blots and protein quantifications (relative to ACTB) of CKAP5 (p = 0.0018); UPF2 (p = 0.0168); SRRM1 (p < 0.0001); DDX27, GAPDH, and CTNNB1 (p = 0.0008); and ELAVL1 (p < 0.0001) in RD cells in control condition (DMSO) or upon treatment with CMLD-2. n = 3. Data are shown as means of relative protein levels ± SD. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.(E) qRT-PCR showing the enrichment of circZNF609 interactors CKAP5, UPF2, and DDX27 mRNAs, and control GAPDH mRNA in a representative RPL22-FLAG RIP assay in RD cells in control condition (DMSO) or upon treatment with CMLD-2. Data are shown as means of enrichment versus input ± SD.(F) Representative western blot and protein quantification (relative to ACTB) of CKAP5 in RD cells in control condition (LNA SCR) or upon transfection of two different sets of LNAs targeting circZNF609 on ELAVL1 binding sites (LNAs 1+2, p = 0.0060; LNAs 1+2+3, p = 0.0405). n = 5. Data are shown as means of relative protein levels ± SD. The ratio of each sample versus control was tested by two-tailed unpaired Student’s t test.See also Figure S3.