Proteasome-mediated proteolysis of SRSF5 splicing factor intriguingly co-occurs with SRSF5 mRNA upregulation during late erythroid differentiation

Abstract

SR proteins exhibit diverse functions ranging from their role in constitutive and alternative splicing, to virtually all aspects of mRNA metabolism. These findings have attracted growing interest in deciphering the regulatory mechanisms that control the tissue-specific expression of these SR proteins. In this study, we show that SRSF5 protein decreases drastically during erythroid cell differentiation, contrasting with a concomitant upregulation of SRSF5 mRNA level. Proteasome chemical inhibition provided strong evidence that endogenous SRSF5 protein, as well as protein deriving from stably transfected SRSF5 cDNA, are both targeted to proteolysis as the cells undergo terminal differentiation. Consistently, functional experiments show that overexpression of SRSF5 enhances a specific endogenous pre-mRNA splicing event in proliferating cells, but not in differentiating cells, due to proteasome-mediated targeting of both endogenous and transfection-derived SRSF5. Further investigation of the relationship between SRSF5 structure and its post-translation regulation and function, suggested that the RNA recognition motifs of SRSF5 are sufficient to activate pre-mRNA splicing, whereas proteasome-mediated proteolysis of SRSF5 requires the presence of the C-terminal RS domain of the protein. Phosphorylation of SR proteins is a key post-translation regulation that promotes their activity and subcellular availability. We here show that inhibition of the CDC2-like kinase (CLK) family and mutation of the AKT phosphorylation site Ser86 on SRSF5, have no effect on SRSF5 stability. We reasoned that at least AKT and CLK signaling pathways are not involved in proteasome-induced turnover of SRSF5 during late erythroid development.

Figure 1. Opposite patterns of expression of SRSF5 mature mRNA and protein during late erythroid differentiation.

A. Steady-state mRNA analysis. Cells were cultured in the absence (−) or presence (+) of DMSO for 4 days to trigger erythroid differentiation. SRSF5 and SRSF3 mRNAs were amplified using specific appropriate forward and reverse primers (see text). Mk: size markers.B. Real-time RT-PCR analysis of SRSF5 and SRSF3 mRNAs. Steady-state mRNA levels were normalized with respect to actin mRNA, used as internal control. Cells were cultured in the absence (untreated) or presence (DMSO) of DMSO for 4 days. Note that cell induction to erythroid differentiation led to a more than threefold increase of steady-state levels of SRSF5 mRNA, when compared with uninduced cells, whereas SRSF3 mRNA remained roughly unchanged. C. Immunoblot analysis of SRSF5 expression in MEL cells. Cells were left untreated (−), or treated (+) with 1.8% DMSO for 4 days. SRSF5 was revealed using either mAb104, an antibody that immunoreacts with all the prototypical SR proteins (left panel), or pAb-SRSF5, a specific anti-SRSF5 antibody (right panel). MEL cell induction using DMSO resulted in a decrease of SRSF5 protein signal.Actin and Grb2 were used as controls.

Figure 2. SRSF5 expression in primary ex-vivo erythroid precursors.

A. Flow cytometric analysis of proliferating (black line) and differentiating (grey line) cells. In each graph, the intensity of scatter (FSC) or fluorescence (kit (CD117), CD71, Ter119) is plotted on the Y-axis. The number of events (number of cells) is plotted on the X-axis; it is expressed as 10³-fold (K) in FSC histogram. Proliferating erythroblasts are large cells (high FSC). They express moderate to high levels of Kit and CD71, and low levels of Ter119 on their cell surface. After 2 days of maturation, these cells display a cell size decrease (low FSC), and cell surface phenotype changes, including Kit decrease and higher levels of CD71 and Ter119.B. Immunoblot analysis of SRSF5 expression in fetal liver-derived erythroblasts. Proteins were collected from proliferating cultured erythroblasts (Prolif.) and 2 days after ex-vivo differentiation (Diff.). Western blot analysis was performed using anti-SRSF5 antibody "pAb-SRSF5". Note that SRSF5 virtually vanishes as the cells differentiate.

Figure 3. Downregulation of stably expressed SRSF5 protein during erythroid differentiation.

A. Recombinant EGFP-SRSF5 mRNA expression. SRSF5 mRNA expressed from the fusion construct EGFP-SRSF5 was amplified from transfected cells using forward primer F8, and reverse primer R8 (Table S1). RNA was extracted from untreated cells or from cells exposed to DMSO-induction for 4 days. Actin mRNA was amplified with primers F6 and R6 (Table S1), and used as control. B. SRSF5 expression during erythroid differentiation. A time course DMSO-induction experiment was performed on MEL cells, transfected with EGFP-SRSF5 construct. Expression of endogenous SRSF5 and fusion EGFP-SRSF5 protein was assessed using anti-SRSF5 or anti-EGFP antibodies, respectively. Expression of control EGFP-containing proteins was also assessed by immunoblotting. These control proteins were obtained from cells transfected with the mock construct EGFP (EGFP cells), or a construct expressing EGFP-hnRNPA2 fusion (EGFP-hnRNPA2 cells), and cultured in the absence (−) or presence (+) of DMSO. C. Downregulation of EGFP-SRSF5 fusion is not clonal. Clones 4 and 7 of MEL cells stably transfected with EGFP-SRSF5 construct were analyzed before (−) or 96 h after (+) DMSO induction. Fusion protein was revealed using anti-GFP antibody.Actin served as an internal control.

Figure 4. Subcellular localization of SRSF5 during erythroid differentiation.

Cells stably expressing EGFP alone (EGFP cells) or the fusion protein EGFP-SRSF5 (EGFP-SRSF5 cells) were stained with DAPI and viewed by fluorescence microscopy. Acquired fluorescence from DAPI-stained nuclei (blue) and emitted from the expression of EGFP (green) shows a strict nuclear localization of SRSF5-containing protein, whereas EGFP protein redistributes to both the nucleus and the cytosol. Note that the fluorescence generated from the fusion protein EGFP-SRSF5 fades away as the cells differentiate, while that emitted from the control EGFP protein remains steady after DMSO exposure.

Figure 5. Proteasome-mediated proteolysis of SRSF5 in late erythroid differentiation.

A. Proteasome inhibition with MG132. Immunoblot analysis of SRSF5 expression in MEL cells treated with DMSO to induce erythroid differentiation, then with increasing concentrations of MG132 to inhibit degradation by the proteasome. Actin immunoblot was used as a loading control. Note that proteasome inhibition stabilizes SRSF5. B. Cycloheximide chase and proteasome inhibition with epoxomicin in MEL cells. Cells stably transfected with EGFP-SRSF5 construct (EGFP-SRSF5 cells) were induced to erythroid differentiation, and then exposed to cycloheximide (CHX) in a time-course experiment (0 to 12 h exposure). The cycloheximide chase assay showed a decrease of fusion protein, detectable after 4 h of exposure, as revealed by immunoblotting using anti-GFP antibody (− lanes). Degradation fragments (*) appeared at the expense of the full-length protein. This latter completely disappeared 8 h after cycloheximide administration. + lanes correspond to the same cycloheximide experiment performed on cells pre-treated with epoxomicin (Epox.) for 4 h. Cycloheximide was added after epoxomicin removal. In these cells, the fusion protein remained stable over time, providing further support that epoxomicin-mediated inhibition is irreversible, and that SRS5 proteolysis is proteasome-dependent.

Figure 6. Proteasome-mediated proteolysis of SRSF5 requires the RS domain.

A. SRSF5 deprived of the RS domain resists proteasome-mediated degradation. Cells were transfected with the full-length EGFP-SRSF5 fusion or a truncated form missing the RS domain (EGFP-SRSF5-ΔRS). Fusion proteins were analyzed by immunoblotting using anti-GFP antibody. Cells were cultured in the absence (−) or presence (+) of DMSO for 4 days. Actin served to trace loading discrepancies or protein degradation. B. Cycloheximide chase and proteasome inhibition with epoxomicin. The experiment was performed on EGFP-SRSF5-ΔRS cells, as indicated in Materials and Methods and in Figure 5 legend. Data are to be compared with cycloheximide chase and proteasome inhibition experiments on EGFP-SRSF5 cells, shown in Figure 5B. Note that RS domain-lacking proteins in EGFP-SRSF5-ΔRS cells are not intercepted by the proteasome-induced proteolysis (see also Text). Antibodies are indicated between parentheses.

Figure 7. Phosphorylation by the CLKs or by AKT is not required for proteasome-mediated degradation of SRSF5.

A. Inhibition of PI3K/AKT signaling. MEL cells were treated for 4 days with either DMSO or LY294002 (LY29), a PI3K/AKT inhibitor. Immunoblot analysis of SRSF5 expression was revealed by anti-SRSF5 antibody (pAb-SRSF5). The decrease in SRSF5 accumulation is most likely secondary to cell differentiation triggered by PI3K/AKT inhibition . Grb2 immunoblot served as control. B. Mutation of AKT phosphorylation site Ser86. Cells stably expressing the recombinant SRSF5 protein mutated at position 86 (EGFP-SRSF5-S86A cells) were treated with DMSO for 4 days, and the fusion protein assessed by immunoblot analysis using anti-GFP antibody. Alpha-tubulin immunoblot was used as control. Abolished AKT phosphorylation site at Ser86 did not affect the regulated post-translation downregulation of SRSF5. C. Inhibition of CLKs. Cells expressing either EGFP alone or the fusion protein EGFP-SRSF5 were analyzed by immunoblotting using anti-GFP, in a time-course DMSO induction experiment (0 to 72 h of exposure), combined with a 6 h exposure of TG003, a CLK inhibitor. Absence (−) or presence (+) of the inhibitor are indicated. Downregulation of SRSF5-containing protein correlated with exposure to DMSO, rather than with TG003 treatment. Expression of CLK1 was assessed in MEL cells untreated (−) or treated for 4 days (+) with DMSO, using anti-CLK1 antibody (insert).

Figure 8. Functional analysis of SRSF5 on endogenous pre-mRNA splicing.

To examine the impact of SRSF5 expression on splicing, we used 4.1R exon 16 as a model for endogenous erythroid splicing event.A. 4.1R exon 16 sequence in vertebrates. Nucleotide sequence alignment in vertebrates displays a very conserved exon 16 sequence. ESEfinder revealed two distinct conserved motifs (underlined bold sequences), potential recognition sites for 2 members of the SR family: SRSF5 and SRSF1 (see also [37]). B. Impact of ESE alteration on exon 16 splicing. The motif identified as SRSF5 ESE was altered by targeted mutagenesis. The mutated exon 16 and its flanking intronic sequences were inserted in a splicing cassette , the resulting mutated minigene (mut) was stably transfected in MEL cells, and exon 16 splicing was analyzed. Data were compared to unaltered exon 16 splicing from wildtype minigene construct (WT). C. Exon 16 splicing pattern in cells overexpressing the full-length protein or the RRM domains of SRSF5. Exon 16 splicing was analyzed in single tests (left panel) or in semi-quantitative experiments (right panel) in cells overexpressing either the full-length SRSF5 (SRSF5) or a shorter form missing the RS domain SRSF5 (SRSF5ΔRS). Data are to be compared with untransfected cells or cells overexpressing EGFP only (Mock).

Figure 9. Impact of SRSF5 overexpression on endogenous pre-mRNA splicing during erythroid differentiation.

A. Electrophoretic analysis of exon 16 splicing in cells overexpressing SRSF5. Cells were cultured either in the absence (uninduced) or presence of DMSO (+DMSO) or PI3K inhibitor LY294002 (+LY29) for 4 days. Both agents trigger cell erythroid differentiation . Exon 16 splicing was analyzed in untransfected cells (1), cells transfected with mock vector containing only EGFP (2), and in cells transfected with a recombinant vector overexpressing the fusion protein EGFP-SRSF5 (3). B. Semi-quantitative analysis of exon 16 splicing in cells overexpressing SRSF5 before (-DMSO) or after (+DMSO) DMSO-treatment. Exon inclusion is to be appraised comparing to control Mock cells (Ctr).