CAP1, a target of miR‐144/451, negatively regulates erythroid differentiation and enucleation

Abstract The exact molecular mechanism underlying erythroblast enucleation has been a fundamental biological question for decades. In this study, we found that miR‐144/451 critically regulated erythroid differentiation and enucleation. We further identified CAP1, a G‐actin‐binding protein, as a direct target of miR‐144/451 in these processes. During terminal erythropoiesis, CAP1 expression declines along with gradually increased miR‐144/451 levels. Enforced CAP1 up‐regulation inhibits the formation of contractile actin rings in erythroblasts and prevents their terminal differentiation and enucleation. Our findings reveal a negative regulatory role of CAP1 in miR‐144/451‐mediated erythropoiesis and thus shed light on how microRNAs fine‐tune terminal erythroid development through regulating actin dynamics.

components and cytoskeleton. 3,[8][9][10] Enucleation starts in orthochromatic erythroblasts, followed by an expulsion of organelles in reticulocytes. 8 These events are strictly regulated to ensure the final proper production of functional erythroid cells. For example, Rac GTPases regulate the formation of contractile actin rings (CAR) on the plasma membrane of erythroblasts, which in turn contribute to their enucleation. 11 MicroRNAs (miRNAs) are critical regulators of erythropoiesis. [12][13][14] Specially, high expression of miR-144 and miR-451 from a bicistronic miRNA locus has been observed in erythroid lineages of various species, including human 15,16 and mouse. 15,17,18 The miR-144/451 appears to be a positive regulator of erythroid maturation. 19 Severe anaemia develops in miR-144/451 knockout mice with erythropoiesis deficiency, including apoptosis of erythroblasts, 20 reduction in erythroid number and increased distribution width of red blood cells. 19 In addition, the anaemia upon miR-144/451 deletion deteriorates under oxidative stress. 21,22 Consistent with this observation, several downstream targets of miR-144/451 have been identified, including NRF2, 23 Ywhaz 21,22 and Cab39. 20 All these miR-144/451 targets participate in cellular reductive/oxidative processes and protect erythrocytes from oxidant damage. 20 Cyclase-associated proteins (CAPs) are highly conserved actin-binding proteins in eukaryotic organisms. [24][25][26][27] Srv2, the CAP homologue, first identified from yeast, is a binding partner of adenylyl cyclase and an effector of Ras during nutritional signalling. 25,28 Mammals have two CAP paralogues, CAP1 and CAP2. 29 In mice, CAP1 is expressed in most non-muscle cell types, whereas CAP2 is primarily restricted to certain brain regions and striated muscles. 30 CAP1 was reported to be a binding protein of globular actin (G-actin, monomeric). 31 In a head-to-tail fashion, G-actin polymerizes to form a helical F-actin filament with a defined polarity, in which F-actin filaments elongate at one end whereas simultaneously shrink at the other end by releasing monomeric G-actin. 32 In general, proteins interacting with F-actin contribute to the assembly of actin filaments and the construction of F-actin network, whereas G-actin-binding proteins serve to sequester monomeric actin and modulate the availability of un-polymerized G-actin. 24 The actin cytoskeletal dynamics are crucial for many cellular processes including erythroblast enucleation and proper assembly of red blood cell membrane, two key steps of terminal erythropoiesis in mammals. 33,34 As a G-actin-binding protein, CAP1 was reported to sequester actin monomers to prevent their polymerization 31 and to stimulate nucleotide exchange of ATP onto ADP-bound G-actin, a rate-limiting step in regenerating polymerizable G-actin. 35 In addition, CAP1 was found to accelerate the depolymerization of F-actin by coordinating with an actin-cofilin complex. 36,37 It however remained undefined what the functions of CAP1 are in erythroid enucleation and membrane assembly. Nor is clear how CAP1 is regulated in definitive erythropoiesis. In the current study, we revealed critical roles of CAP1 in maintaining cellular actin dynamics during erythroid development. We identified CAP1 as a direct target of miR-451 and found that CAP1 down-regulation was essential for erythroid differentiation and enucleation.
Erythroid differentiation of MEL cells was induced by adding dimethyl sulfoxide (DMSO; Sigma-Aldrich) to 2% as the final concentration in culture media. 39

| Benzidine staining
Benzidine staining was performed following the published protocols. 40 Briefly, a 0.2% solution of benzidine (Sigma-Aldrich) in 3% acetic acid was freshly prepared. Prior to use, hydrogen peroxide was added to 0.3% as the final concentration. MEL cells were first smeared on glass slides, air-dried, stained in the above described solution for 10 minutes (shielded from light) and finally fixed in 100% Methanol for 10 minutes. Images were taken with a Leica microscope.

| Contractile actin ring (CAR) staining
For CAR staining, MEL cells were harvested in PBS. Cell pellets were fixed in 100 µL PBS with 0.5% acrolein for 5 minutes, and cell concentration was adjusted to approximately 5 × 10 6 cells/mL. Cells (~100 µL) were applied to poly-l-lysine-coated slides and dried at room temperature. The slides were rinsed three times in PBS to remove unbound cells. Cells were permeabilized in PBS containing 0.05% Triton X-100 for 10 seconds, followed by three washes in PBS, then incubated in blocking buffer (PBS containing 0.5 mmol/L glycine, 0.2% fish skin gelatin and 0.05% sodium azide) for 1 hour and incubated with 1 U/mL Alexa Fluor 568-phalloidin (Thermo Fisher Scientific) for 1 hour. Slides were washed three times in blocking buffer followed by DAPI staining for 15 minutes before images were taken with a Leica fluorescence microscope.

and then inserted into a MSCV-MCS-IRES-hygro vector between
EcoRI/XhoI sites. Three Cap1-specific shRNAs and a scramble shRNA control were designed and cloned into a Plko.1-U6-shRNA-puro vector, respectively. Sequences of oligonucleotides and primers in this study are provided in the Table S1.

| Quantitative real-time PCR
Total RNAs were extracted with Trizol (Thermo Fisher Scientific), and cDNAs were synthesized using a PrimeScript ® RT reagent Kit (TaKaRa) following manufacturer's protocols, as previously described. 43,44 The primers used in this assay are provided in the Table S1.

| Luciferase reporter assay
The 3′-UTR fragment of mouse Cap1 was cloned into a pGL4-basic firefly luciferase reporter vector. Mutant 3′-UTR fragment of mouse

| Western blotting
Total proteins were isolated using RIPA lysis buffer, subjected to SDS-PAGE, transferred to PVDF membranes and incubated with primary antibodies. Quantification of target protein levels was performed using the ECL detection system and Quantity One software (Bio-Rad). Antibodies used in this study: CAP1 (NBP1-58320; Novus), β-ACTIN (sc-10731; Santa Cruz Biotech) and GAPDH (sc-25778; Santa Cruz Biotech). The relative protein expression levels were determined by the density (grey mean value) of the protein bands with the ImageJ software and normalized to the respective loading control GAPDH/β-ACTIN.

| Statistical analysis
Data were presented as mean ± standard error (SEM). All experiments were performed independently for more than three times unless otherwise stated. Statistical analysis between group differences was performed with two-tail unpaired Student's t test using Graph Prism software (version 5.0; GraphPad). A P value <.05 was considered significant (*P < .05, **P < .01, ***P < .001).

| CAP1 is a direct target of miR-451
To understand the molecular mechanisms of how miR-144/451 impacts erythropoiesis, we first predicted their targets by conducting bioinformatics analyses through Targetscan, an online software.
We identified CAP1 as a potential target of miR-451 ( Figure 1A). To further investigate whether miR-451 directly targets Cap1 mRNA, a dual-luciferase reporter assay was conducted in 293T cells. In this assay, we constructed a firefly luciferase reporter, downstream of which was inserted with a wild-type or mutant Cap1 3′-UTR fragment containing mutations within the miR-451 seed targeting sequences ( Figure 1D,E). We found that pri-miR-144/451 significantly inhibited the activity of firefly luciferases linked with the wild-type Cap1 3′-UTR. By contrast, the inhibitory effect of pri-miR-144/451 was lost in the vector containing Cap1 3′-UTR with a mutated miR-451 seed target sequence ( Figure 1F). Taken together, these results reveal that Cap1 is a direct target gene of miR-451.

| CAP1 is down-regulated during foetal erythroid differentiation
To study the role of CAP1 in erythroid development, we first examined its expression level at different stages of erythroid development. We utilized a foetal liver erythropoietic model because more than 90% of cell population from mouse foetal liver are of erythroid lineage. 11 Five distinct erythroblast populations from mouse foetal livers at represents late orthochromatophilic erythroblasts and reticulocytes. 11 As shown in Figure 2B,C, both β-globin (Hbb: Hbb-β1&β2) and pri-miR-144/451 were significantly up-regulated all the way through R4 stage. By contrast, Cap1 mRNA was promptly down-regulated from R1 to R4 ( Figure 2D). CAP1 protein showed a similar expression pattern ( Figure 2E), supporting a negative reciprocal expression of CAP1 and miR-144/451 during erythroid differentiation.
We next examined Cap1 expression levels during in vitro erythroid differentiation of CD71+/Ter119-foetal liver cells isolated from E12.5 mice (D0). 11 After induced differentiation for 2 days (D2), we clearly observed red coloured cell pellets, indicating erythroid maturation with increased levels of iron-containing haemoglobin ( Figure 2F). In addition, double staining for Ter119 and a DNA dye Hoechst 33342 demonstrated a significant increase from 0.4% to 36.7% of the Ter119 high Hoechst low population ( Figure 2G), which represents enucleated reticulocytes. 45 Compared with D0 progenitors, Cap1 expression dramatically decreased in D2 differentiated erythroid cells. This is in sharp contrast to the remarkable increased transcript levels of β-globin (Hbb: Hbb-β1&β2) and pri-miR-144/451 ( Figure 2H). Taken together, these results illustrate that CAP1 is down-regulated both in vitro and in vivo along definitive erythroid differentiation, suggesting a vital role of CAP1 in this process.

| CAP1 inhibits terminal erythroid differentiation of MEL cells
To define the function of CAP1 in erythropoiesis, three Cap1-specific shRNAs and a scramble shRNA control were introduced by lentiviral that forms a dark blue precipitate upon oxidation of the haem group in haemoglobin by hydrogen peroxide. It thus serves as a dye for the histochemical detection of differentiated red blood cells with high haemoglobin expression. We found that blue MEL cells were significantly increased from 52% to 87% upon Cap1 knock-down ( Figure 3C and S1A). Consistent with these data, real-time RT-PCR assays showed that Cap1 down-regulation significantly promoted the expression of both α-globin (Hba: Hba-α1 & α2) and β-globin (Hbb: Figure 3D).

Because erythroid cells become smaller during differentiation
and enucleation, we next examined the size distribution of MEL cells upon DMSO induction. We were able to detect three MEL popula- middle-sized cells that were likely in the process of differentiation (blue area with FSC-A peak spanning between 20K and 50K). We found significantly more small-sized erythroid cells increasing from 46.2% to 90.3% upon CAP1 inhibition (red plus blue populations in Figure 3E), whereas the percentage of erythroblasts with large-sized erythroblasts dropped from 50.6% to 8.5% ( Figure 3E), supporting that CAP1 inhibition favours erythroid differentiation. We further performed May-Grünwald Giemsa staining (MGG) staining to examine cell morphology. 41 In this assay, acidic cytoplasm is stained light blue by methylene blue, whereas nuclei are generally stained purple/ bluish violet due to interactions between eosin Y and Azure B-DNA complex. 41 We found that during DMSO-induced differentiation, were correspondingly repressed by enforced CAP1 expression ( Figure 3I). Interestingly, upon CAP1 overexpression, although the proportion of large-sized cells (FSC-A > 50K) decreased, middle-sized MEL cells (blue population) were significantly increased from 14.4% to 57.2%, suggesting these cells were arrested in the middle of differentiation ( Figure 3J). These data were further confirmed by MGG staining. We found that compared with the empty vector controls containing large erythroblasts, CAP1 overexpressing group was mainly comprised of smaller (middle-sized) erythroid cells with purple nuclei ( Figure S1C). Taken together, these results demonstrate that CAP1 inhibition favours terminal erythroid differentiation of MEL cells and CAP1 overexpression represses erythroid enucleation.

| CAP1 inhibits terminal erythroid differentiation of foetal liver erythroblasts
To confirm the negative role of CAP1 in regulating erythroid differentiation, we performed siRNA interfering experiments using in vivo developed erythroblasts from foetal livers. Three siRNAs targeting against Cap1 mRNA and one scramble control siRNA were designed and transiently transfected into E12.5 foetal liver cells. As shown in Figure 4A, all three Cap1 siRNAs efficiently blocked Cap1 expression, with siRNA-2# showing the highest knock-down efficiency.
Two days after in vitro culture following siRNA-2# transfection, we observed a significant increase of late differentiated cells at the R5 stage from 8.3% to 15.6% ( Figure 4B). Cell population at R4 stage was also slightly elevated. By contrast, significant declines were detected in the cells at the R3 stage upon CAP1 down-regulation ( Figure 4B). Concomitantly, the proportion of large-sized cells (above FSC-A > 50K) decreased from 72.0% to 55.9% ( Figure 4C), indicating increased erythroid differentiation upon CAP1 inhibition. In addition, MGG staining showed that CAP1 inhibition speeded up the process of enucleation and cells without nuclei significantly increased ( Figure 4D). Consistently, double staining for Ter119 and Hoechst 33342 demonstrated a significant increase of Ter119 high Hoechst low enucleated reticulocytes from 6.7% to 11.5% upon CAP1 knock-down ( Figure 4E). Notably, we did not observe any obvious alterations in the death of foetal liver erythroblasts in the CAP1 down-regulated group ( Figure S2) rather found a modest decrease in percentage of cells at S-phase, suggesting a reduced cell proliferation upon CAP1 knock-down ( Figure S3). In conclusion, our results reveal that CAP1 inhibition promotes terminal erythroid differentiation and enucleation of foetal liver cells.

| CAP1 inhibits miR-144/451-mediated terminal erythroid differentiation
To further determine whether CAP1 inhibition is required for miR-144/451-mediated erythroid differentiation, we examined erythroid maturation of MEL cells upon co-expressing CAP1 and pri-miR-144/451 ( Figure 5A, and S4). At day 4 after DMSO induction, the pellets from differentiated MEL cells exhibited a much brighter redness in the miR-144/451 overexpressing group than that in mock control ( Figure 5A), confirming that miR-144/451 enhances MEL differentiation. By contrast, the cell pellets displayed a light yellow colour when CAP1 was overexpressed or co-expressed with both CAP1 and pri-miR-144/451 ( Figure 5A and S4). Concomitantly, as shown in the real-time RT-PCR assay, miR-144/451 significantly promoted β-globin expression which was substantially blocked by CAP1 ( Figure 5B). The percentage of benzidine-positive/dark bluestained MEL cells was significantly elevated from 54% to 79% upon F I G U R E 4 CAP1 inhibits terminal erythroid differentiation of foetal liver cells. A, Cap1 mRNA and protein levels at day 2 following siRNA knock-down in primarily isolated Ter119-erythroblasts from E12.5 foetal livers. B, Flow cytometry analysis of primary foetal liver erythroblasts at day 2 during differentiation following Cap1 siRNA knockdown, compared with a scrambled siRNA control. C, The size distribution (FSC-A) of foetal liver erythroid cells was analysed by flow cytometry at day 2 during differentiation following CAP1 knock-down. D, MGG staining assays of foetal liver cells at day 2 during differentiation following Cap1 siRNA knockdown, compared with scrambled siRNA controls. Red arrow indicates erythroblasts, black arrows for enucleating cells and white arrows for reticulocytes without a nucleus. E, Foetal liver cells were stained with APC-Ter119 and Hoechst 33342 at day 2 during differentiation following Cap1 siRNA knock-down. Representative plots of flow cytometry analyses were shown. A, B, C, E, The data are represented as the mean ± SEM (n ≥ 3; **P < .01; NS, no significance) pri-miR-144/451 introduction ( Figure 5C). This phenomenon disappeared in the presence of CAP1, either alone or co-overexpressed with pri-miR-144/451 (subgroup b & d, Figure 5C). We thus conclude  Figure 5D).
By contrast, exogenous pri-miR-144/451 up-regulated the formation of Ter119 high Hoechst low cells, but this phenomenon largely diminished when CAP1 transgene was co-introduced ( Figure 5D).
Consistent with these findings, we found that miR-144/451 facil- ( Figure S5A). We further confirmed with MGG staining assays that the formation of middle-sized MEL cells with purple nuclei was significantly increased when CAP1 was up-regulated, whereas exog-  Late stage erythroblasts undergo cell cycle exit, chromatin condensation and extrusion of the condensed nuclei via an asymmetric cell division. 8 CAR formation is required for this enucleation process.

| D ISCUSS I ON
In immature erythroblasts, F-actin distributes patchily at cell surface. However, F-actin bundles become detectable as erythroblasts mature, and finally F-actin is concentrated to form CAR between the extruding nuclei of erythroblasts and incipient reticulocytes to enable proper enucleation. 48 We found that, as a direct target of Gelsolin 52 and mDIA2. 45 It remains to be determined whether CAP1 interplays with these proteins to precisely regulate the actin dynamics of erythroblasts.
Our study also suggests that CAP1 participates in other aspects of erythroid development. We observed a modest decrease of erythroblasts at S-phase upon CAP1 inhibition ( Figure S3). As terminally differentiated erythrocytes shuts down cell cycles by up-regulating of p27 and p18 CDK inhibitors, 53 the reduced proliferation of CAP1 knock-down cells could be due to increased percentage of mature erythrocytes. CAP1 was previously reported to be a binding partner of adenylyl cyclase and act as a Ras effector during nutritional changes in yeasts. 25,28,30 Interestingly, it is known that RAS signalling plays an important role in terminal erythroid proliferation and differentiation. 11 We thus could not exclude the possibility that CAP1 is directly involved in cell cycle regulation of erythroblasts.
Nevertheless, our study undoubtedly supports a critical role of CAP1 in miR-144/451-mediated terminal erythroid differentiation and enucleation, thereby contributing to potential clinical therapy in the field of blood transfusion.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest. Writing-original draft (equal); Writing-review & editing (equal).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the Supporting information of this article.