KLF4 inhibited the senescence-associated secretory phenotype in ox-LDL-treated endothelial cells via PDGFRA/NAMPT/mitochondrial ROS

KLF4 inhibited the ox-LDL-induced EC SASP in vivo and in vitro

After feeding C57BL/6J mice a high-fat diet for 8 weeks, we used histopathology to observe the distribution of KLF4-positive cells. An obvious decrease in KLF4 expression in endothelial cells and adventitial fibroblasts was observed in high-fat diet-fed mice compared to normal diet-fed mice, especially in endothelial cells (Figure 1A). Interestingly, KLF4 expression was lacking in smooth muscle cells. The results were also verified using in vitro experiments. Western blotting results showed significantly downregulated KLF4 protein expression in ox-LDL-treated HUVECs (Figure 1B). To verify the possible role of KLF4 in endothelial cells, we analyzed the mRNA-seq in NC HUVECs, KLF4 knock-in HUVECs and KLF4 knockdown KUVECs. GO and KEGG analysis revealed that multiple pathways related to endothelial cells function and aging-related were significantly differentially regulated along with KLF4 expression changes (Figure 1C). These results indicated that KLF4 may play an important role in ox-LDL intervention in endothelial cells in vivo and in vitro.

Figure 1. Analysis of KLF4 protein expression and EC dysfunction. (A) Immunohistochemical detection of vascular wall KLF4 expression. Scale bar, 20 μm. Representative images (n=5) are shown. Red arrow, KLF4 expression in the intima of mice fed a normal diet. Yellow arrow, intima lacking KLF4 expression in mice fed a high-fat diet. Blue arrow, KLF4 expression in adventitial fibroblasts. (B) Western blotting analysis of KLF4 protein expression in cultured HUVECs (100 μg/ml ox-LDL, 24 h) (n=5). (C) GO and KEGG analysis of the pathways after KLF4 knock in and knock out in HUVECs.

In high-fat diet-fed C57BL/6J mice, in addition to decreased KLF4 levels in the intima, the senescent intima area also exhibited increased SA-β-gal staining (Figure 2A). To investigate the role of KLF4 in mediating intima senescence in high-fat diet-fed mice, we designed specific conditional EC KLF4 knockout (EC KLF4^-/-) mice, and the knockout efficiency was confirmed by histopathology (Figure 2B) and Western blotting in purified vascular ECs (Figure 2F). Next, we found that EC KLF4^-/- mice fed a normal diet exhibited a larger SA-β-gal-positive area (Figure 2C) and increased expression of the senescence protein p21 (Figure 2D) in the intima. To further verify the role of KLF4 in EC senescence, we designed Adv-KLF4 and shRNA-KLF4 to regulate KLF4 expression in vitro. Then, we found that transfection of ox-LDL-treated HUVECs with Adv-KLF4 led to a lower percentage of SA-β-gal-positive cells and lower p21 protein expression of p21, whereas KLF4 knockdown led to the opposite results (Figure 2E, 2G).

Figure 2. KLF4 inhibits the EC SASP. (A) Histochemical detection of the SA-β-gal-positive area in the vascular wall of mice fed a high-fat diet. Representative images (n=5) are shown. Red arrow, SA-β-gal-positive intima. (B) Immunohistochemical detection demonstrates the efficiency of specific conditional KLF4 knockout in mouse endothelial cells (EC KLF4^-/-). Scale bar, 200 μm. Representative images (n=5) are shown. Red arrow, KLF4 expression in normal mouse intima. Yellow arrow, intima lacking KLF4 expression in mice fed a high-fat diet. Blue arrow, KLF4 expression in adventitial fibroblasts. (C) Histochemical detection of SA-β-gal-positive areas in the vascular wall of EC KLF4^-/- mice. Scale bar, 1 cm. Representative images (n=5) are shown. Red arrow, SA-β-gal-positive intima. (D) Immunohistochemical detection of p21 protein expression in the intima of EC KLF4^-/- mice. Scale bar, 200 μm. Representative images (n=5) are shown. Red arrow, p21-positive endothelial cells. (E) Histochemical detection of SA-β-gal-positive ECs in HUVECs. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (F) Western blotting analysis of KLF4 protein expression in the intima of EC KLF4^-/- mice. (G) Western blotting analysis of p21 protein expression in HUVECs after KLF4 expression is altered (n=5). (H) Immunofluorescence detection of typical SAHF formation in HUVECs (n=5). Scale bar, 5 μm. (I) Protein chip analysis in HUVECs (n=3). (J) qPCR analysis of the mRNA levels of cytokines in HUVECs (n=5). *P < 0.05. (K) Western blotting analysis of PAI-2 and uPA protein expression in HUVECs (n=5).

Cytokine release, which is also referred to as the SASP, is the most important role of senescent cells, and the specific chromatin rearrangement, called senescence-associated heterochromatin foci (SAHF), may represent its structural basis. Next, immunofluorescence results showed that KLF4 shRNA or ox-LDL treatment induced typical heterochromatin foci (Figure 2H). Protein chip and qPCR were used to observe cytokine release, and different cytokine protein production (Figure 2I and Table 1) and mRNA levels (Figure 2J) were observed in Adv-KLF4-transduced HUVECs and shRNA-KLF4-transfected HUVECs. In our experiments, we also observed that protein levels of the SASP biomarker type 2 plasminogen activation inhibitor (PAI-2) were decreased in ox-LDL-treated HUVECs transduced with Adv-KLF4 but increased in shRNA-KLF4-transfected HUVECs, whereas the opposite results were obtained for the SASP inhibition biomarker uridylyl phosphate adenosine (uPA) (Figure 2K). All of these in vivo and in vitro results indicated that KLF4 has an inhibitory effect on the SASP in ECs under ox-LDL-treatment conditions.

PDGFRA plays an essential role in KLF4-mediated regulation of the SASP

In our experiments, we analyzed the protein-protein interaction (PPI) network in NC HUVECs, KLF4 knock-in HUVECs and KLF4 knockdown KUVECs. The PDGF/PDGFR pathway was increased in KLF4 knock-in HUVECs, whereas KLF4 shRNA-transfected HUVECs exhibited the opposite result (Figure 3A). To verify communication between KLF4 and the PDGF/PDGFR pathway, we observed the expression of proteins in the PDGF/PDGFR pathway and found that only PDGFRA protein expression was increased in Adv-KLF4-transduced HUVECs. Moreover, PDGFRA protein expression was decreased in shRNA-KLF4-transfected HUVECs (Figure 3B). To further verify whether KLF4 directly promoted PDGFRA transcription, we searched the seed region (Figure 3C) and designed pGL4 probes and biotin probes for in vitro experiments. We found that transfection of the pGL4-KLF4 vector with PDGFRA, but not platelet-derived growth factor-BB (PDGF-BB) or platelet-derived growth factor receptor β (PDGFRB), into HUVECs effectively induced luciferase activity (Figure 3D). The EMSA results also showed that KLF4 tightly bound to biotin-labeled PDGFRA but not PDGF-BB or PDGFRB (Figure 3E). Next, we designed a PDGFRA mutant in the KLF4 binding seed region with a poly-A sequence and then transfected it with the pGL4-KLF4 vector; no luciferase activity was observed (Figure 3F). These findings indicated that KLF4 promotes PDGFRA transcription. We wanted to assess whether communication occurs between KLF4 and PDGFRA in the KLF4-regulated SASP. To confirm this hypothesis, we observed PDGFRA protein expression in ECs and found that PDGFRA protein expression decreased in ox-LDL-treated HUVECs (Figure 4A) and high-fat diet-fed mice (Figure 4B). Histopathology showed that PDGFRA protein expression was also reduced in EC KLF4^-/- intima in mice fed a normal diet (Figure 4B). Next, we used Adv-PDGFRA and shRNA-PDGFRA to upregulate and downregulate PDGFRA expression, respectively, in non-treated and ox-LDL-treated HUVECs. PDGFRA overexpression in ox-LDL-treated HUVECs led to an apparent decrease in p21 protein expression (Figure 4C), a decrease in the percentage of SA-β-gal-positive HUVECs (Figure 4D), and a reduced level of cytokine secretion (Figure 4F). Reduced PDGFRA expression had the opposite effect (Figure 4C–4F). Then, we downregulated PDGFRA expression in KLF4 knock-in HUVECs by transfecting PDGFRA shRNA and restored the HUVEC SASP. In the context of reduced PDGFRA expression, the percentage of SA-β-gal-positive HUVECs (Figure 4G), p21 protein expression (Figure 4H), typical SAHF structure formation (Figure 4I) and cytokine release (Figure 4J) exhibited sharp increases in KLF4 knock-in HUVECs. Then, we overexpressed PDGFRA in KLF4 knockdown HUVECs by transfecting Adv-PDGFRA. Here, the increase in the SASP in KLF4-knockdown HUVECs was attenuated by the overexpression of PDGFRA (Figure 4G–4I, 4K). These results confirmed that the regulatory role of KLF4 on the SASP is at least partly mediated by PDGFRA.

Figure 3. KLF4 promotes PDGFRA transcription. (A) PPI network analysis in cultured HUVECs. (B) Western blotting analysis of the PDGF pathway in HUVECs. (C) The amounts of PDGFRA, PDGF-BB and PDGFRB in (B) quantified using actin as control. (n=5). *P < 0.05. (D) Seed region for the KLF4 binding site. (E) Luciferase activity analysis of the PDGF pathway in HUVECs (n=5). *P < 0.05. (F) EMSA analysis of the PDGF pathway in HUVECs (n=3). *P < 0.05. (G) Luciferase activity of mutant PDGFRA in HUVECs (n=5). *P < 0.05.

Figure 4. KLF4 inhibited the HUVEC SASP through PDGFRA. (A) Western blotting analysis of PDGFRA protein expression in ox-LDL-treated HUVECs (n=5). (B) Immunohistochemical detection of PDGFRA protein expression in the intima of ND-fed WT mice, HFD-fed WT mice and HFD-fed EC KLF4^-/- mice. Scale bar, 200 μm. Representative images (n=5) are shown. Red arrow, PDGFRA-positive endothelial cells. (C) Western blotting analysis of p21 protein expression in ox-LDL-treated HUVECs after altered PDGFRA expression (n=5). (D) Histochemical detection of SA-β-gal-positive ECs in ox-LDL-treated HUVECs after altered PDGFRA expression. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (E) SA-β-gal staining positive cells were counted and presented as percentage of total cells. (F) Immunofluorescence detection of typical SAHF formation in cultured HUVECs (n=5). Scale bar, 20 μm. (G) qPCR analysis of the mRNA levels of cytokines in ox-LDL-treated HUVECs after regulating PDGFRA (n=5). *P < 0.05. (H) Histochemical detection of SA-β-gal-positive ECs in KLF4-treated HUVECs after regulating PDGFRA. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (I) Western blotting analysis of p21 protein expression in KLF4-treated HUVECs after regulating PDGFRA (n=5). (J) qPCR analysis of the mRNA levels of cytokines in KLF4-knock-in HUVECs after PDGFRA knockdown (n=5). *P < 0.05. (K) qPCR analysis of cytokine mRNA levels in KLF4-knockdown HUVECs after PDGFRA knock-in (n=5). *P < 0.05.

The NAMPT/mitochondrial ROS pathway participates in KLF4/PDGFRA-mediated inhibition of the SASP in ox-LDL-treated HUVECs

NAMPT-mediated mitochondrial ROS (MitoROS) production is a regulator of chronic inflammation regulator as documented in numerous studies. In our experiments, we observed NAMPT protein expression changes under different conditions. We found that NAMPT protein expression decreased in ox-LDL-treated HUVECs (Figure 5A) and high-fat diet-fed mice (Figure 5B), whereas KLF4 or PDGFRA overexpression increased NAMPT protein expression (Figure 5A). Moreover, histopathological assessment revealed that intima NAMPT protein expression was decreased in KLF4^-/- EC mice (Figure 5B). Then, we downregulated PDGFRA in KLF4 knock-in HUVECs by transfecting PDGFRA shRNA and assessed NAMPT protein expression. Here, reduced PDGFRA expression resulted in reduced NAMPT protein expression (Figure 5A). Next, we overexpressed PDGFRA in KLF4 knockdown HUVECs by adding Adv-PDGFRA and obtained the opposite results (Figure 5A). Then, we used MitoSOX to detect MitoROS and found that ox-LDL-treated HUVECs had higher MitoROS levels, whereas Adv-KLF4- or Adv-PDGFRA-transduced HUVECs had lower MitoROS levels (Figure 5C). In addition, we separately downregulated NAMPT in KLF4 knock-in or PDGFRA knock-in HUVECs and overexpressed NAMPT in KLF4 knockdown or PDGFRA knockdown HUVECs and measured MitoROS levels. We observed lower MitoROS levels along with higher NAMPT expression in KLF4 knockdown or PDGFRA knockdown HUVECs, whereas lower NAMPT expression and higher MitoROS levels were noted in KLF4 knock-in or PDGFRA knock-in HUVECs (Figure 5D, 5E). These results confirmed that KLF4/PDGFRA regulated MitoROS production partially through NAMPT.

Figure 5. KLF4/PDGFRA regulated NAMPT/MitoROS expression. (A) Western blotting analysis of NAMPT protein expression in cultured HUVECs (n=5). (B) Immunohistochemical detection of NAMPT protein expression in the intima of high-fat diet-fed mice or EC KLF4^-/- mice. Scale bar, 200 μm. Representative images (n=5) are shown. Red arrow, NAMPT-positive endothelial cells. (C) Immunofluorescence detection of MitoROS in cultured HUVECs. Scale bar, 20 μm. Representative images (n=5) are shown. Red, ROS. Green, Mitochondria. Yellow, MitoROS. (D) Flow cytometry analysis for MitoROS quantification in KLF4-treated HUVECs after altering NAMPT expression (n=5). *P < 0.05. (E) Flow cytometry analysis of MitoROS quantification in PDGFRA-treated HUVECs after altering NAMPT expression (n=5). *P < 0.05.

To determine the potential role of NAMPT/MitoROS in the HUVEC SASP, we performed SA-β-gal staining and found that NAMPT knockdown increased the SA-β-gal-positive staining percentage in Adv-KLF4-treated or Adv-PDGFRA-treated HUVECs (Figure 6A). In addition, NAMPT overexpression yielded the opposite result in KLF4 shRNA-treated HUVECs or PDGFRA shRNA-treated HUVECs (Figure 6B). Next, we used Mitoquinone Q (MitoQ, 1 μM) treatment to scavenge MitoROS and assessed SA-β-gal-positive HUVECs. We found that MitoQ treatment decreased the SA-β-gal-positive staining percentage in KLF4 shRNA-treated HUVECs and PDGFRA shRNA-treated HUVECs (Figure 6C). Then, we assessed p21 protein expression and found that NAMPT knock-in or MitoQ treatment reduced p21 protein expression in KLF4 shRNA-transfected HUVECs or PDGFRA shRNA-transfected HUVECs (Figure 6D). Next, the immunofluorescence results showed that NAMPT knock-in or MitoQ treatment prevented typical SAHF formation in ox-LDL-treated HUVECs, whereas knockdown of NAMPT mediated typical SAHF formation (Figure 6E). Then, we assessed cytokine mRNA levels. NAMPT knock-in or MitoQ treatment decreased cytokine mRNA levels in ox-LDL-treated HUVECs, whereas NAMPT knockdown increased cytokine mRNA levels (Figure 6F). These results demonstrated that NAMPT/MitoROS are involved in KLF4/PDGFRA-mediated SASP inhibition in ox-LDL-treated HUVECs.

Figure 6. NAMPT/MitoROS regulated the SASP in HUVECs. (A) Histochemical detection of SA-β-gal-positive ECs in NAMPT knockdown HUVECs after knock-in of KLF4 or PDGFRA. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (B) Histochemical detection of SA-β-gal-positive ECs in NAMPT knock-in HUVECs after KLF4 or PDGFRA knockdown. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (C) Histochemical detection of SA-β-gal-positive ECs in MitoQ-treated HUVECs after KLF4 or PDGFRA knockdown. Scale bar, 50 μm. Representative images (n=5) are shown. Blue, SA-β-gal-positive ECs. (D) Western blotting analysis of p21 protein expression in NAMPT knock-in- or MitoQ-treated HUVECs (n=5). (E) Immunofluorescence detection of typical SAHF formation in cultured HUVECs after altering NAMPT expression or treatment with MitoQ (n=5). Scale bar, 20 μm. (F) qPCR analysis of cytokine mRNA levels in cultured HUVECs after altering NAMPT expression or treatment with MitoQ (n=5). *P < 0.05.

Animals

All animal testing procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH revised 2011) and approved by the Institutional Animal Care and Use Committee of Tongji University. Eight-week-old C57BL/6J male mice were purchased from the Experimental Animal Centre, Chinese Academy of Sciences (Shanghai, China). Specific conditional EC KLF4 knockout mice were established by Saiye Corp. (Suzhou, Jiangsu, China). Briefly, we crossbred Tek-Cre: KLF4 and Floxp/Floxp mice. Genotyping was performed at 2 weeks after birth. The Tek-Cre: KLF4/Floxp mice have specific KLF4 deletion ECs (called KLF4^-/- ECs). After being fed a normal diet or a high-fat diet for 8 weeks, the mice were anesthetized using isoflurane and euthanized using cervical dislocation.

Reagents

Adv-vectors (KLF4, PDGFRA and NAMPT) were used to overexpress target genes, and sh-RNAs (KLF4, PDGFRA and NAMPT) were used to knock down target genes. All were purchased from Viogene Bio. (Jinan, Shandong, China). The KLF4 luciferase reporter plasmid (pGL4-KLF4) was purchased from Yeasen Bio. (Shanghai, China). A fluorescent dual luciferase reporting system and murine leukemia virus reverse transcriptase were obtained from Promega Co. (Mannheim, Germany). The SA-β-gal-kit was obtained from Beyotime (Haimen, China). The MitoSOX Mitochondrial Superoxide Indicator was obtained from Yeasen (Shanghai, China). The anti-p21 antibody was obtained from HUABIO (Hangzhou, China). The anti-PDGF-BB, anti-PDGFRA, anti-PDGFRB, anti-KLF4, anti-NAMPT, anti-PAI-2, anti-uPA and anti-β-actin antibodies and secondary antibodies were obtained from Cell Signaling Technology (Danvers, MA, USA). Other reagents were purchased from Sigma (St. Louis, MO, USA).

Cell culture

The human umbilical vein endothelial cell (HUVEC) line and endothelial cell media (ECM) were obtained from ScienceCell (Carlsbad, CA, USA). HUVECs were grown in ECM (glucose: 5.5 mM). After starvation by FBS deprivation for 12 h, HUVECs were exposed to ox-LDL (100 μg/ml) for 24 h to induce senescence.

Real-time reverse transcription PCR (RT-PCR)

Total RNA was converted to cDNA using the murine leukemia virus reverse transcriptase system. RT-PCR was performed using an ABI 7500 system and SYBR Premix chemistry (Takara, Shiga, Japan) following the manufacturer’s instructions. β-Actin was chosen as an endogenous expression standard. The cycle number represents the relative quantity of the specific template when the fluorescence of the amplified gene product first reached a preset threshold (Ct). Gene expression levels were calculated using the double delta Ct method. All amplifications were performed independently three different times. The following primers were used (Sangon Bio., Shanghai, China): IL-1β: Forward: 5’-ATGATGGCTTATTACAGTGGCAA-3’, Reverse: 5’-GTCGGAGATTCGTAGCTGGA-3’; IL-6: Forward: 5’-ACTCACCTCTTCAGAACGAATTG-3’, Reverse: 5’-CCATCTTTGGAAGGTTCAGGTTG-3’; IL-18: Forward: 5’-TTGGCCCAGGAACAATGGCTGC-3’, Reverse: 5’-TGCGGTTGTACAGTGAAGTCGG-3’; MMP9: Forward: 5’-CAGTACCGAGAGAAAGCCTATT-3’, Reverse: 5’-CAGGATGTCATAGGTCACGTAG-3’; TNF-a: Forward: 5’-TGCACTTTGGAGTGATCGGC-3’, Reverse: 5’-AGCTTGAGGGTTTGCTACAAC-3’; TGF-β: Forward: 5’-CCGTCTCCTACCAGACCAAGG-3’, Reverse: 5’-CCGTCTCCTACCAGACCAAGG-3’; CXCR-2: Forward: 5’-AAGGTGAATGGCTGGATTTTTG-3’, Reverse: 5’-CCCAGATGCTGAGACATATGAA-3’.

Protein extraction and Western blot

Cultured HUVECs or mouse aortic walls were lysed in RIPA buffer (EpiZyme, Shanghai, China) and sonicated briefly before protein extraction. Proteins were quantified using a bicinchoninic acid (BCA) protein assay kit (EpiZyme, Shanghai, China). Proteins (40 μg) were separated on SDS-PAGE gels and transferred onto 0.45-μm PVDF membranes (Millipore, Burlington, MA, USA), which were then incubated with primary antibodies at 4° C overnight after blocking with 5% fat-free milk in TBST buffer. On the following day after incubation with HRP-conjugated secondary antibodies for 1 h at room temperature, the membranes were visualized using enhanced ECL reagent (Share-Bio, Shanghai, China). The relative intensities of the protein bands were analyzed using ImageJ software.

SA-β-gal analysis

SA-β-gal analysis was performed according the SA-β-gal kit manufacturer’s instructions. Briefly, after fixation for 15 min, fresh aortic walls or cells were incubated with 5-bromo-4-chloro-3-inolyl-b-D-galactoside (β-gal) substrate for 24 h at 37° C and then imaged.

SAHF detection

After fixation with 90% ice-cold methanol for 2 min, the cells were stained with 0.1 μg/mL anti-fade DAPI for 2 min and then analyzed using a laser confocal microscope.

EMSA

Nucleoproteins were obtained from cultured HUVECs. After incubation with specific probes with biotin, the samples were resolved by TBE-PAGE and then visualized using a streptavidin-HRP conjugate.

The following probes (1000 fmol) were used:

PDGF-BB, bio5’-gcttgttaccacacccagctccag-3’, 3’-cgaacaatggtgtgggtcgaggtc-5’-bio;

PDGFRA, bio5’-caggcgcaaccaggcccaggtggc-3’, 3’-gtccgcgttggtccgggtccaccg-5’-bio;

PDGFRB, bio5’-gttgaggctgggtgcggtggctcaag-3’, 3’-caactccgacccacgccaccgagttc-5’-bio.

Fluorescent dual luciferase reporting analysis

For luciferase analysis, the pGL4-KLF4 vector and PDGF-BB, PDGFRA and PDGFRB vectors were cotransfected, and cells were incubated for 48 h. The samples were collected and subjected to 3 freeze-thaw cycles. The results were obtained using a Lumat LB9507 luminometer (EG and GBerthold). Transfection efficiency was normalized to Renilla luciferase.

mRNA-seq array

An mRNA-seq array of total RNA obtained from HUVECs was performed by Riobio Tech, Inc. (Guangzhou, China).

Inflammatory factor spectrum array

An inflammatory factor spectrum array from total proteins obtained from lysing HUVECs was performed by Riobio Tech, Inc. (Guangzhou, China).

Histopathology

For histological analysis, freshly dissected aortas were fixed in 10% (v/v) formalin for 24 h, embedded in paraffin, sliced into 4-mm thick sections, stained with β-gal and specific primary antibodies and observed using a ZEISS AXIO scanning microscope.

MitoROS detection

MitoROS analysis was performed according to the MitoSOX kit manufacturer’s instructions. Briefly, freshly cultured HUVECs were first stained with a mitochondria staining kit, and then the MitoSOX stain was added. Samples were incubated at 37° C for 15 min. The samples were analyzed using a laser confocal microscope or flow cytometry.

Statistical analysis

Statistical analysis was performed using SPSS 22.0 software. All data are presented as the mean ± SEM and were compared using one-way ANOVA followed by Tukey’s HSD test. A value of P < 0.05 was considered to indicate statistical significance.