Quaking isoforms cooperate to promote the mesenchymal phenotype

Abstract

The RNA-binding protein Quaking (QKI) has widespread effects on mRNA regulation including alternative splicing, stability, translation, and localization of target mRNAs. Recently, QKI was found to be induced during epithelial-mesenchymal transition (EMT), where it promotes a mesenchymal alternative splicing signature that contributes to the mesenchymal phenotype. QKI is itself alternatively spliced to produce three major isoforms, QKI-5, QKI-6, and QKI-7. While QKI-5 is primarily localized to the nucleus where it controls mesenchymal splicing during EMT, the functions of the two predominantly cytoplasmic isoforms, QKI-6 and QKI-7, in this context remain uncharacterized. Here we used CRISPR-mediated depletion of QKI in a human mammary epithelial cell model of EMT and studied the effects of expressing the QKI isoforms in isolation and in combination. QKI-5 was required to induce mesenchymal morphology, while combined expression of QKI-5 with either QKI-6 or QKI-7 further enhanced mesenchymal morphology and cell migration. In addition, we found that QKI-6 and QKI-7 can partially localize to the nucleus and contribute to alternative splicing of QKI target genes. These findings indicate that the QKI isoforms function in a dynamic and cooperative manner to promote the mesenchymal phenotype.

FIGURE 1:

A relationship between the balance of the QKI isoforms and total QKI expression across tissues. (A) Schematic of the QKI locus including positions of the QKI gRNA, QKI-5 siRNAs, QKI isoform specific primers (colored along with isoform specific exon) and endogenous QKI-5 specific primer pair. (B–D) Quantitation of QKI isoforms following transfection of PC-3, MDA-MB-231, or mesHMLE cells with nontargeting (Neg si #1 and #2), panQKI targeting (siQKIpan), or QKI-5 specific (siQKI-5 #1, siQKI-5 #2) siRNAs. (E) Quantitation of QKI isoforms in untreated or 1.00 μg/ml doxycycline treated LNCaP-iQKI5 cells. (F) Ratios of QKI-5, QKI-6, and QKI-7 to total QKI mRNA plotted against total QKI mRNA of qRT-PCRs performed on RNA samples harvested from breast cancer cell lines. Expression of QKI-5, QKI-6, or QKI-7 expressed as a percentage of total QKI mRNA plotted against total QKI mRNA across (G) breast tissue (GTEx project) and (H) breast cancer cell line (CCLE) samples.

FIGURE 2:

Loss of QKI evokes an epithelial-like morphology in mesenchymal breast cells. (A) Western blot for panQKI in mesHMLE control cells and mesHMLEs transfected with a QKI gRNA (pool and resulting single cell clones). This western blot was performed three times and the bar chart represents the mean –/+ SD. (B and D) Quantitative RT-PCR for the QKI isoforms and ADD3 isoforms. For D, the bar chart shows relative percentage of the ADD3 transcripts with exon 14 included or excluded. Data for qRT-PCRs is represented as the mean of three technical replicates ± SD. (C) Phase images of mesHMLE wild-type (control) and QKI KO (clone #2) cells. Scale bars represent 50 µm.

FIGURE 3:

All QKI isoforms can participate in alternative splicing. (A) Western blot of total QKI and individual QKI isoform protein following 1.00 μg/ml doxycycline induction of each isoform in QKI knockout mesHMLE cells. The panQKI western blot is quantitated in the upper panel. (B) Quantitative RT-PCR of QKI isoform mRNA levels in the same samples as (A). Data is represented as the mean of three technical replicates ± SD. Quantitation by qRT-PCR of percent spliced in of (C) ADD3 exon 14 and (D) NFYA exon 3 in the same samples as (A) from three biological replicates ± SD. P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as **p < 0.01, ***p < 0.001, and ****P < 0.0001.

FIGURE 4:

The cytoplasmic QKI isoforms can partially localize to the nucleus when overexpressed. Fluorescent microscopy images of overexpressed individual QKI isoforms in mesHMLE-QKI-KO-iQKI-5 (A), -QKI-6 (B) or iQKI-7 (C) cells cultured for 72 h without or with 0.05, 0.20, or 1.00 μg/ml doxycycline. Scale bars represent 50 µm. Quantitation of percentage nuclear and cytoplasmic isoform fluorescence per cell (middle panel) and total nuclear isoform fluorescence per cell for each cell line are plotted (right panel). All fluorescence is normalized to wild-type cells. Error bars represent ± SD of the mean of five separate images with cell numbers for iQKI-5 (300, 294, 301, 397, and 364), iQKI-6 (388, 370, 395, 307, and 296), and iQKI-7 (322, 414, 346, 395, and 432) experiments indicated in brackets (values are for mesHMLE, 0, 0.05, 0.20, and 1.00 μg/ml doxycycline, respectively). P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as **p < 0.01 and ****p < 0.0001.

FIGURE 5:

Nuclear localization of QKI-6 and -7 is enhanced by QKI-5 overexpression. Fluorescent microscopy images of the mesHMLE-QKI-KO-cQKI-5-iQKI-6 (A) and iQKI-7 (D) cell lines after culturing for 72 h with 1.0 µg/ml doxycycline. Cells were stained for QKI-5 (red) and either QKI-6 or QKI-7 (green), and counterstained with DAPI (blue). Merged images are the combination of QKI-5 and QKI-6 or QKI-7 alone. Large scale bars represent 50 µm and small scale bars represent 5 µm. Violin plots indicate total QKI-6 (B) or QKI-7 (E) fluorescence and percent nuclear QKI-6 (C) or QKI-7 (F) fluorescence with cell numbers for iQKI-6 (100, 77) and iQKI-7 (66, 91) experiments indicated in brackets (values are for 0 Dox and 1.00 μg/ml doxycycline, respectively). The interquartile ranges of QKI-5 fluorescent intensity are labeled from lowest to highest intensity as Q1, Q2, Q3, and Q4. P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as *p < 0.05, **p < 0.01, *** p < 0.001, and ****p < 0.0001.

FIGURE 6:

Coexpression of QKI isoforms collectively enhance mesenchymal cell migration. (A) Migration assay comparing wild type mesHMLE cells with derivatives of mesHMLE-QKI- treated with or without 1.00 µg/ml doxycycline, and with coexpression of constitutive QKI-5. Data are plotted as mean migrated cells per field. All experiments were performed with at least three replicates counting migrated cells in at least 10 fields per treatment. P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as ns- not significant, **p < 0.01, and ****p < 0.0001. (B) Western blots showing QKI levels following expression of different combination of QKI isoforms in mesHMLE-QKI-KO cells used in migration assays. The panQKI western blot is quantitated in the upper panel.

FIGURE 7:

QKI-5 is required for mesenchymal morphology of mesHMLE cells. (A) Fluorescence microscopy of F-actin and nuclei (DAPI) in wild-type mesHMLE, mesHMLE-QKI-KO cells, and mesHMLE-QKI-KO cells with constitutive expression of QKI-5 or doxycycline induction of either QKI-5, QKI-6, or QKI-7. Merged images are a combination of F-actin and DAPI. Large scale bars represent 50 µm and small scale bars represent 5 µm. Violin plots show quantitated changes in (B) cellular area, (C) cellular aspect (length to width) ratio and (D) nuclear area, calculated on a minimum of 32 cells per condition. P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as*p < 0.05, **p < 0.01, and ****p < 0.0001.

FIGURE 8:

QKI-6 and -7 both affect mesenchymal morphology, when coexpressed with QKI-5. (A–C) Fluorescence microscopy of F-actin and nuclei (DAPI) in mesHMLE-QKI-KO cells with constitutive expression of QKI-5 and doxycycline inducible expression of QKI-6 and/or QKI-7. Merged images are a combination of F-actin and the stained QKI isoform. Large scale bars represent 50 µm and small scale bars represent 5 μm. Violin plots show quantitated changes in (D) cellular area and (E) cellular aspect ratio, calculated on a minimum of 32 cells per condition. For comparison, data for mesHMLE, mesHMLE-QKI-KO, and mesHMLE-QKI-KO-cQKI-5 are also shown (taken from Figure 7). P values are calculated using one-way ANOVA (with Bonferroni multiple testing) and indicated as **p < 0.01 and ****p < 0.0001.