NRP1 and MMP9 are dual targets of RNA‐binding protein QKI5 to alter VEGF‐R/ NRP1 signalling in trophoblasts in preeclampsia

Abstract Preeclampsia (PE) is characterized by placental ischemia and hypoxia, resulting in abnormal casting of the uterine spiral artery, which is mainly caused by insufficient trophoblastic cell infiltration. A reduction in levels of growth factor‐based signalling via Neuropilin‐1 (NRP1) has been shown to contribute to dysfunctional trophoblast development. In this study, we showed that the RNA‐binding protein, QKI5, regulated NRP1 expression and significantly improved trophoblast proliferation in vitro and in vivo. QKI5 and NRP1 expressions were significantly reduced in human PE placentas and in trophoblasts during hypoxia. Overexpression of these factors significantly improved cell proliferation and migration in vitro, in contrast to a decrease upon siRNA knockdown of QKI5 and NRP1 in HTR‐8/SVneo cells. Using RIP and RNA pull‐down assays, we further showed that QKI5 directly interacted with the 3'‐UTR region of NRP1, to mediate cell proliferation and migration via matrix metalloprotease‐9. Further, similar to NRP1, QKI5 also targets matrix metalloproteinase 9 (MMP9) involved in secretion of growth factors and its effects can be counteracted by NRP1 overexpression. In vivo studies using a PE mouse model revealed that QKI5 overexpression alleviated PE‐related symptoms such as elevated blood pressure and proteinuria. Taken together, we found that QKI5 was a novel regulator, of VEGF‐R/NRP1 signalling pathway functioning in trophoblast proliferation and migration, resulting in major contributors to the pathogenesis of PE. While careful evaluation of the broad implications of QKI5 expression is still necessary, this study identified QKI5 as a promising target for treatment strategies in acute PE patients.

treatment due to severe neonatal side effects. 4,5 Progression of PE involves oxidative stress, induction of systemic inflammation and release of angiogenic factors into the circulatory system. [6][7][8] This results in placental damage due to prolonged hypoxic conditions that trigger apoptosis-induced cell death in trophoblasts. [9][10][11] Prolonged hypoxia causes irreversible changes in cells including transcriptional reprogramming and regulation of gene expression that in turn affects intrinsic cell properties towards development and migration.
Understanding these mechanisms in the context of PE remains key towards identifying alternative treatment strategies.
Several factors regulate gene expression including microRNAs, long non-coding (lnc) RNAs and RNA-binding proteins. Quaking homolog KH domain RNA-binding (QKI) protein, belonging to the Signal Transduction and Activation of RNA (STAR) family of proteins, has been known to play a role in altering cellular differentiation processes in myeloid cells. 12,13 Under hypoxia, QKI directly regulated hypoxia-induced factor-1a (HIF-1α) expression to support tumorigenesis. 14 QKI isoforms have also been implicated in altering vascular endothelial growth factor receptor (VEGF-R)/ Neuropilin-1 (NRP1) co-receptor signalling pathway in endothelial cell differentiation and functions. 15,16 They can compete with microRNAs to inhibit gene expression of NRP-1/2 that are essential for early cell development. 13 More specifically, QKI isoform 5 (QKI5) has been shown to play a role in altering stem cell differentiation and RNA-splicing events in cancer. 17,18 It regulates key differentiation pathways during erythropoiesis, osteoclastogenesis and angiogenesis. 15 However, QKI proteins and their possible regulatory roles in trophoblast differentiation nor dysregulation in the context of PE have not been elucidated so far.
Placental trophoblasts ensure differentiation via an interplay between different growth factor signalling pathways including VEGF-R/NRP1 and are strictly controlled by regulators of gene expression. 19,20 Analysis of placental tissue samples from PE patients revealed that both NRP1 expression and VEGF-R-binding protein VEGF expression were significantly decreased; 21 however, the mechanism behind its regulation remains elusive. VEGF signalling requires its secretion from extracellular matrices that are mediated by a specific class of proteinases called matrix metalloproteinases (MMPs). MMPs are capable of degrading the extracellular matrix proteins including collagen and fibronectin to facilitate release of growth factors required for cell growth and differentiation. 22 MMPs have also been studied in the context of PE for their role in altering tissue architecture and promoting trophoblast invasion. 23 MMP2 and MMP9 are implicated in the release of VEGF that is required for placental growth and uterine remodelling during the early stages of pregnancy. 24 MMPs are up-regulated during tumorigenesis to promote cell proliferation and invasion in a VEGF-NRP1-dependent manner. 25,26 In contrast, lower expression of MMPs is associated with various pregnancy-related complications, including PE, by inducing apoptosis of trophoblasts and maternal intolerance. [27][28][29] Although the role of MMPs has been broadly described in the context of trophoblast development and invasion, identifying their mechanism of regulation will lead to the development of efficient treatment alternatives.
In the present study, we showed that QKI5 was a novel regulator of NRP1 and MMP9 expression in trophoblasts during hypoxia. In addition to studying its effects in vitro in the HTR-8/SVneo trophoblast cell line, we used an in vivo PE mouse model to evaluate the effects of QKI5 expression in regulating PE symptoms during pregnancy. Our findings identified QKI5-mediated regulation of NRP1 expression as an important pathway in trophoblast development.

| Cell culture and transfection
In vitro experiments were conducted using the HTR- 8 with Lipofectamine in a 1:3 ratio. Two siRNAs (si-QKI5#1 and si-QKI5#2) specifically for QKI5 and the control siRNAs (si-NC) were synthesized by Dharmacon as previously described. 30 The DNA-Lipofectamine complex was added to cells, and a fresh medium exchange was conducted after 6 h. The cells were then used 24 h post-transfection. Immunodetection of proteins using chemiluminescence was performed with Image-Pro Plus 6.0 (Media Cybernetics).

| RNA purification and qPCR
The mRNA levels of QKI5, NRP1 and MMP9 were quantified from total RNA using qPCR. Cells or tissues were harvested, and total RNA was isolated using TRIzol reagent (Invitrogen).

| MTT assay
Cell proliferation was measured using the MTT assay (Invitrogen).
Nuclei were visualized by 4′,6-diamidino-2-phenylindole staining for 10 min at room temperature. Cell proliferation was determined by staining for the Ki-67 nuclear antigen, which is constitutively expressed throughout cell cycle progression. Images were taken using an Axio Observer microscope (Carl ZEISS) with image processing software.

| Immunohistochemistry
Sections from the placenta were immediately fixed with for- For the wound healing assay, a wound scratch was made in cells seeded in a 6-well plate and images from different fields taken at the time of wound scratching and 72 h post-wound scratching were compared to calculate the area of the wound closure. All bars show the means ± SD from at least three independent experiments. Statistical significance was calculated using Student's t test for comparing two groups and one-way analysis of variance for more than two groups. *** P < .001; **** P < .0001 compared with the si-NC group, ## and #### denote P < .01, P < .0001, respectively, compared with the hypoxia +pcDNA3.1 group [Colour figure can be viewed at wileyonlinelibrary. com]

| Cell migration assay
HTR-8/SVneo cells were starved prior to invasion assays in media containing 1% FBS. A total of 5 × 10 5 starved cells were seeded into the upper well of a Transwell chamber (8 μm pore size; Millipore) and the lower chamber containing medium supplemented with 10% FBS served as the chemoattractant. The chambers were incubated at 37°C/5% CO 2 for 24 h. Post-migration cells in the lower chambers were stained with 0.1% Crystal Violet (in 20% methanol), and the number of migrated cells was quantified using five random fields per sample.

| Wound healing assay
Pre-seeded HTR8/SVneo cells were transfected with expression constructs or siRNA for knockdown, followed by the scratch wound heal-  Luciferase activity as relative luciferase units (RLU) was measured using a luminometer in triplicate for each sample.

| RNA-protein pull-down assay
Standard RNA pull-down assays were conducted as follows. Cell

| RNA immunoprecipitation (RIP) assay
The RIP assay was performed with the EZMagna RIP kit (Millipore)

| The PE mouse model
The PE mouse model was established using a combination of intra-

| Statistical analysis
Statistical analyses between data sets from at least three independent experiments were conducted using paired or unpaired Student's t tests to compare two groups, and one-way analysis of variance was used to compare more than two groups. Statistics were performed with SPSS statistical software for Windows (SPSS), and a value of P < .5 was considered statistically significant.

| Human PE placentas show reduced QKI5 expression
To examine the role of the QKI5 RNA-binding protein in PE, we ex- in comparison with healthy placentas, indicating lower expression ( Figure 1A). To evaluate this quantitatively, total RNA was isolated from different tissues from both groups, and mRNA levels of QKI5 were determined by qPCR. QKI5 expression was significantly reduced in PE placentas at the mRNA level when compared to healthy donors, which further confirmed the immunohistochemistry results ( Figure 1B). Analysing protein expression in tissues by Western blot revealed that QKI5 expression was significantly reduced in PE tissues ( Figure 1C). These initial characterizations show that QKI5 was downregulated in PE.

| QKI5 expression is influenced by hypoxia and promotes cell proliferation and migration in vitro
To further study the role of QKI5 in PE, we performed knockdown and overexpression studies in vitro using trophoblast HTR-8/SVneo cells. Immunofluorescence assays revealed that the subcellular distribution of QKI5 was restricted to the cytoplasm (Figure 2A a FBS gradient when compared to over 80% in the NC ( Figure 2H).
In addition, hypoxia also reduced cell migration to approximately 30%, which was restored to 70% when QKI5 was ectopically overexpressed ( Figure 2H). Furthermore, wound healing assays also revealed that reduced QKI5 expression reduced cell migration and that this effect was reversed upon overexpression ( Figure 2I). Taken together, these results showed that QKI5 was influenced by hypoxic microenvironments and was a mediator of cell proliferation and migration.

| NRP1 is also influenced by hypoxia and is required for cell proliferation and migration
Neuropilin-1 (NRP1) has been previously reported to be expressed at reduced levels in women with PE, along with reduced vascular endothelial growth factor (VEGF) expression. 21 showed more efficient knockdown, we used it for further analysis.
Similar to QKI5, knockdown of NRP1 also reduced cell proliferation as measured by the MTT assay ( Figure 4C), and overexpression under hypoxic conditions significantly improved proliferation by at least twofold at 72 h ( Figure 4D). Ki-67 staining also indicated that cell proliferation was reduced to 40% upon knockdown of NRP1 or under hypoxic conditions, which was significantly improved (to 80%) when NRP1 was overexpressed ( Figure 4E). Transwell migration and wound healing assays showed that cell migration was reduced to 40%-50% upon NRP1 knockdown when compared to 80% in the NC. This effect was restored upon ectopic expression of NRP1 ( Figure 4F,G). Collectively, these results suggested that NRP1 and QKI5 showed similar effects in HTR-8/SVneo cells under hypoxic conditions.

| NRP1 expression is directly regulated by QKI5
Similar to its role in myeloid cell differentiation, 13  precipitates compared with the IgG control in NC cells, whereas this increase was completely lost upon QKI5 knockdown ( Figure 5F). In addition, overexpression of QKI5 showed an increase in NRP1 RNA levels F I G U R E 4 NRP1 promotes cell migration and proliferation in trophoblasts. A, Relative NRP1 mRNA levels as detected by qPCR under normal and hypoxia conditions, ** P < .01. B, Representative Western blotting and densitometric quantification of NRP1 protein expression in HTR-8/SVneo in non-targeting control (NC) or NRP1 siRNA-targeted cells. Statistics were calculated using one-way analysis of variance. ** P < .01; *** P < .001 (compared with the si-NC group); ### P < .001 (compared with the pcDNA3.1 group). C-D, Proliferation efficiency using the MTT assay measured as absorbance at 450 nm of trophoblast cells under normal and hypoxia conditions at 24 h post-transfection using NRP1 siRNA or the overexpression plasmid. Statistical significance was calculated using Student's t test for C and one-way analysis of variance for D. * P < .05; ** P < .01; *** P < .001. E, Representative merged micrographs and quantification of HTR-8/SVneo NC or NRP1 siRNAtargeted cells under normal conditions and HTR-8/SVneo cells overexpressing NRP1 during hypoxia stained with the Ki-67 proliferation marker (green). Nuclei were stained with DAPI (blue). At least 100 cells were counted per condition, per experiment. Scale bar: 50 μm F-G, Cell migration efficiency. HTR-8/SVneo NC or NRP1 siRNA-targeted cells under normal conditions or HTR-8/SVneo cells overexpressing NRP1 during hypoxia were assessed for their abilities to migrate in a Transwell system or by wound healing assays as mentioned in Figure 2. Shown are the means ± SD from at least three independent experiments. Statistics for panel E to G were calculated by one-way analysis of variance. *** P < .001; **** P < .0001 (compared with the si-NC group); ## P < .01; ### P < .001 (compared with the hypoxia +pcDNA3.1 group) [Colour figure can be viewed at wileyonlinelibrary.com] when precipitated with anti-QKI5 antibody, when compared to the expression of the control vector ( Figure 5F). Western blot analysis from the RNA-protein pull-down assay showed that QKI5 was bound only to the 3'UTR fraction of NRP1 and not to the 5'UTR or the coding sequence (CDS) ( Figure 5G). In summary, these results showed that QKI5 directly regulated NRP1 expression by binding to the 3'UTR region.

| NRP1 regulates cell migration and proliferation by MMP9 in PE
In addition to the effects of QKI5 on NRP1 expression, we evaluated its role in targeting other factors involved in this signalling pathway. Matrix metalloproteinase-9 (MMP9) functions in releasing growth factors to trigger VEGF-R/NRP1 pathway. We found that MMP9 mRNA expression was reduced more than twofold under hypoxic conditions in HTR-8/SVneo cells ( Figure 6A). Protein levels of MMP9 were also decreased upon QKI5 and NRP1 knockdown, and this effect was overcome upon NRP1 overexpression ( Figure 6B). In a similar manner, QKI5 and NRP1 overexpression increased levels of MMP9 expression, and this effect was reduced after knockdown of NRP1 knockdown ( Figure 6B). Relative mRNA levels were also similar to protein levels, whereupon knockdown of QKI5 and NRP1, MMP9 mRNA levels were significantly reduced, while this effect was rescued by NRP1 overexpression (Figure 6C).
Similarly, QKI5 and NRP1 overexpression significantly increased MMP9 mRNA expression, and this effect was reduced upon NRP1 knockdown ( Figure 6C). The MTT assay, which measures cell proliferation, showed that QKI5 knockdown significantly reduced cell proliferation as previously observed, and this effect was improved 48 h after overexpression of NRP1 ( Figure 6D). In addition, overexpression of MMP9 in NRP1 knockdown cells also significantly improved cell proliferation 48 h after knockdown ( Figure 6D

| Increased QKI5 expression reduces the severity of PE symptoms in an in vivo mouse model
We next sought to test these effects in an in vivo mouse model for PE. 36 Normal mice with no treatment (Normal), PE mice with or without adoptive transfer of NC HTR-8/SVneo cells (PE+NC or PE) and PE mice adoptively transferred with QKI5 overexpression cells (PE+QKI5) were used. Placental tissues from PE mice harvested towards the end of gestation showed reduced NRP1 and MMP9 mRNA expressions, whereas overexpression of QKI5 restored NRP1 levels to that of normal mice ( Figure 7A). Similarly, protein expression analysis revealed that in PE mice, both NRP1 and MMP9 levels were significantly reduced, and this effect was reversed when QKI5 was overexpressed ( Figure 7B). Immunohistochemistry analyses of tissue samples stained with cell proliferation dyes, Ki-67, NRP1 or MMP9, showed that cell proliferation was significantly reduced in PE and PE+NC tissues, whereas upon QKI5 overexpression, the levels were increased to levels comparable with normal mice ( Figure 7C). In addition, continuous monitoring of blood pressure and urine protein concentration showed a steady increase in blood pressure over time in the PE mice groups, and this effect was significantly reduced upon QKI5 overexpression ( Figure 7D). A threefold increase in total urine protein concentration was also observed in PE mice, and this effect was reduced to levels comparable with normal mice upon QKI5 overexpression ( Figure 7E). Taken together, these results showed that QKI5 promoted cell proliferation, possibly via a NRP1/MMP9dependent manner in an in vivo PE mouse model.

| D ISCUSS I ON
During early stages of pregnancy, invasion of trophoblasts through the uterine wall is an important first step in placentation.
These trophoblasts lining the uterus are responsible for nourishing the uterine environment with oxygen and sufficient nutrients.
Defects in these initial processes due to hypoxia and insufficient F I G U R E 5 QKI5 interacts with NRP1 at the 3'UTR region. A, Scatter plot of mRNA expression levels of QKI5 and NRP1 in preeclampsia placentas. Correlations were determined using Spearman's rank correlation test (N = 20, P < .0001, R 2 = 0.667). B, Relative NRP1 mRNA levels as detected by qPCR in HTR-8/SVneo cells transfected with siRNAs against QKI5 or the QKI5 overexpression vector during hypoxia. The values of mock transfection were arbitrarily set to 1. C, Representative Western blotting and densitometric quantification of NRP1 protein expressions from cells shown in panel B. D, The half-life of NRP1 mRNA was evaluated in QKI5 knockdown or overexpression HTR-8/SVneo cells at 0, 3, 6, 9 and 12 h post-treatment using a transcription inhibitor. The mRNA levels at time 0 were arbitrarily set to 100%. E, The luciferase reporter assay to evaluate 3'UTR activity. NRP1 3'UTR-mediated expression of firefly luciferase upon binding of QKI5 was determined by harvesting QKI5 knockdown or overexpression HTR-8/SVneo cells co-expressing the reporter construct. Cell lysates were incubated with luciferin and RLU (relative luciferase units) indicated the luciferase activity as measured using a luminometer. (F) RNA immunoprecipitation assay. Cells transfected with siRNA against QKI5 or QKI5 overexpression vector were immunoprecipitated with IgG (control) or anti-QKI5 antibody. Bound NRP1 RNA was detected by qPCR. Values of si-NC or the pcDNA3.1 control were arbitrarily set to 1. G, Representative Western blotting showing immunodetection of QKI5 by RNA pull-down against different regions of NRP1 (5'UTR, CDS and 3'UTR). Glyceraldehyde 3-phosphate dehydrogenase was used as the input control. Shown are the means ± SD from at least three independent experiments. Statistical significance was calculated using one-way analysis of variance for all panels except D, where the two groups were compared using Student's t test. * P < .05; ** P < .01; *** P < .001; **** P < .0001 (compared with the si-NC group) and #### P < via binding with regulatory microRNAs to affect gene expression. 34 However, they can also directly affect expression of growth factors, receptors and downstream signalling molecules.
One of the major signalling pathways required for trophoblast migration and invasion is the VEGF-R/NRP1 pathway. 35 It is activated upon binding of soluble VEGF released from the extracellular matrix by matrix metalloproteinases (MMPs). Among the different forms of VEGF-R, VEGF-A interacts with the NRP1 co-receptor to promote cell migration and angiogenesis. 36 A previous study showed that reduced trophoblast development observed in PE placentas correlated with down-regulation of VEGF and NRP1 expression. 21 We also observed a reduction in NRP1 expression at the mRNA and protein levels in placental tissues from PE patients, which further correlated with QKI5 expression. Furthermore, knockdown and overexpression of NRP1 in HTR-8/SVneo cells showed that NRP1 was also essential for cell proliferation and migration during hypoxia.
RIP assays revealed that QKI5 directly interacted with NRP1 mRNA, and RNA pull-down assays using different domains of NRP1 showed that QKI5 specifically bound to the 3'UTR region. This indicates that QKI5 does not interfere with microRNAs to regulate expression of NRP1 but directly associates with it to affect mRNA stability and reduce translation efficiency. 37 We further identified that QKI5 did not exclusively target NRP1 to interfere with the VEGF-R/NRP1 pathway but also affected expression of MMP9 in HTR-8/SVneo cells under hypoxia. MMPs regulate VEGF-R/NRP1 signalling by promoting the release of matrix-bound VEGF, and their role in trophoblast invasion has been described. 23,38,39 In cancers, increased levels of MMP9 has been associated with tumorigenesis and metastasis; 40,41 however, in PE their expression is down-regulated. 29 Although we did not elucidate the mechanism of reduced MMP9 expression in the presence of QKI5, we observed that its functional effect on cell proliferation and invasion was overcome upon expression of NRP1. A previous report studying pregnancy-related hypertensive disorders using placental tissues showed that MMP9 expression was significantly reduced in a severe hypertensive group when compared to healthy patients. 42 This further suggests that alteration in MMP9 levels as a result of comorbidities can be independent of other failures in growth factor signalling observed in PE. QKI5 emerges as a common determinant that can regulate expression of both proteins that are key promoters of cellular development.
The implications of QKI5-mediated effects in vivo during the progression of PE was studied using an adoptive transfer mouse model where received cells overexpressing QKI5. This model was chosen over the STOX-1 model that recapitulates early PE maternal syndrome and other available PE models as it proved relevant to study both placental pathology and symptoms related to PE such as rise in blood pressure and increased urine/creatine levels. 43,44 Placental tissues harvested towards the end of gestation revealed a significant increase in cell proliferation in the QKI5 treatment group. We also observed a significant decrease in PE symptoms such as blood pressure elevation and total protein concentration in the urine. Future studies using other mouse models including endocrine gland-derived VEGF (EG-VEGF) or MMP9-null mice model would provide more insight into the relevance of QKI5 in PE pathogenesis. [45][46][47] These results together identified a novel mechanism responsible for trophoblast dysregulation observed in PE, where QKI5-mediated regulation of NRP1/MMP9 expression significantly altered cell proliferation and migration. F I G U R E 6 QKI5 promotes cell proliferation and migration in trophoblasts via NRP1/MMP9. A, Relative mRNA MMP9 expression levels as determined by qPCR under normal and hypoxic conditions. B, C, Representative Western blotting and quantification of MMP9 protein expressions under different conditions in HTR-8/SVneo cells. D-E, Cell proliferation was determined using the MTT assay. At 24 h posttransfection, the cells were stained with the MTT dye, and the absorbance was measured at 450 nm. F, Representative merged micrographs and quantitative analysis of cell proliferation by Ki-67 staining. Nuclei were stained using DAPI. At least 100 cells were counted per condition, per experiment (G-H) Cell migration and invasion assays. The relative percentage migration and invasion were calculated using the Transwell migration or wound healing assays in HTR-8/SVneo cells under different conditions. Statistical significance was calculated using one-way analysis of variance for all panels except A, where the two groups were compared using Student's t test. ** P < .01; *** P < .001; **** P < .0001 (compared with the non-targeting group); ## P < .01; ### P < .001; #### P < .0001 (compared with the NRP1 or si-NRP1 group); $$$ P < .001 (compared with the pcDNA3.1 group); in panel C, && P < .01 (compared with the NRP1 group). Shown are the means ± SD from at least three independent experiments [Colour figure can be viewed at wileyonlinelibrary.com] | 5669 YANG et Al.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.