The CDK9-cyclin T1 complex mediates saturated fatty acid-induced vascular calcification by inducing expression of the transcription factor CHOP

Abstract

Vascular calcification (or mineralization) is a common complication of chronic kidney disease (CKD) and is closely associated with increased mortality and morbidity rates. We recently reported that activation of the activating transcription factor 4 (ATF4) pathway through the saturated fatty acid (SFA)-induced endoplasmic reticulum (ER) stress response plays a causative role in CKD-associated vascular calcification. Here, using mouse models of CKD, we 1) studied the contribution of the proapoptotic transcription factor CCAAT enhancer-binding protein homologous protein (CHOP) to CKD-dependent medial calcification, and 2) we identified an additional regulator of ER stress-mediated CHOP expression. Transgenic mice having smooth muscle cell (SMC)-specific CHOP expression developed severe vascular apoptosis and medial calcification under CKD. Screening of a protein kinase inhibitor library identified 16 compounds, including seven cyclin-dependent kinase (CDK) inhibitors, that significantly suppressed CHOP induction during ER stress. Moreover, selective CDK9 inhibitors and CRISPR/Cas9-mediated CDK9 reduction blocked SFA-mediated induction of CHOP expression, whereas inhibitors of other CDK isoforms did not. Cyclin T1 knockout inhibited SFA-mediated induction of CHOP and mineralization, whereas deletion of cyclin T2 and cyclin K promoted CHOP expression levels and mineralization. Of note, the CDK9-cyclin T1 complex directly phosphorylated and activated ATF4. These results demonstrate that the CDK9-cyclin T1 and CDK9-cyclin T2/K complexes have opposing roles in CHOP expression and CKD-induced vascular calcification. They further reveal that the CDK9-cyclin T1 complex mediates vascular calcification through CHOP induction and phosphorylation-mediated ATF4 activation.

Keywords: CDK9; CHOP; ER stress; atherosclerosis; cardiovascular disease; cyclin-dependent kinase (CDK); endoplasmic reticulum stress (ER stress); fatty acid; kidney disease; renal dysfunction; saturated fatty acids; vascular biology; vascular calcification; vascular smooth muscle cells.

Figure 1.

SMC-specific CHOP overexpression induces medial calcification. A, scheme of construct design for targeting the CHOP transgene in the Rosa26 locus. Triangles indicate the loxP site. CHOP conditional TG mice were backcrossed 10 times with C57Bl6 mice. CHOP conditional TG mice were crossed with SMMHC–Cre^(ER)T2 TG mice to generate SMC–CHOP TG mice. 5-Week-old male mice under NKD were injected with vehicle or 1 mg of tamoxifen for 5 consecutive days to generate control mice and SMC–CHOP TG mice, respectively. The 8-week-old mice were subjected to either sham operation (NKD) or 5/6 nephrectomy (CKD) and were sacrificed at 20 weeks of age. B, immunoblot analysis of CHOP and its targets (BiP, GADD34, Bax, and BcL2) in the aortic media. The medial layer of aortas was isolated from control and SMC–CHOP TG mice. C, immunofluorescence microscopic analysis of CHOP (green) in the aortic arch. DAPI (blue). D–F, levels of BUN, serum creatinine, and phosphorus in SMC–CHOP TG mice. Serum creatinine levels were analyzed by LC-MS/MS. Other parameters were analyzed with colorimetric assays. G, representative photograph (×10) of the lesions in aortic arches stained with von Kossa. Arrows (black lesions) indicate calcification. H, quantitative analysis of calcified lesions in the aortic arches. I, aortic calcium content in SMC–CHOP TG mice. J, quantitative analysis of apoptotic lesions (terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL)–positive nuclei) in the aortic arches of SMC–CHOP TG mice. Two-way ANOVA was used for comparison between NKD and CKD. n = 8; *, p < 0.05; ***, p < 0.001; and ****, p < 0.0001; Cont, control.

Figure 2.

CDK9 regulates CHOP expression. A, immunoblot analysis of CHOP protein to screen for inhibition of SFA-induced CHOP induction with a kinase inhibitor library containing >140 compounds. VSMCs were pretreated with 10 μm of each inhibitor for 2 h prior to co-treatment with 500 μm C18:0 and 10 μm of each inhibitor for 6 h. CHOP expression is expressed as a fold value relative to that of vehicle-treated (DMSO-treated, fold value is 1) cells with GAPDH correction by band densitometry. The experiments were repeated three times. B, immunoblot analysis of CHOP protein to screen for inhibition of SFA-induced CHOP induction. VSMCs were pretreated with 10 μm of each inhibitor for 2 h prior to co-treatment with 500 μm C18:0 and 10 μm of each inhibitor for 6 h. Total cell lysates of VSMCs were then prepared. BSA (Veh) with DMSO (lane 1), C18:0 with DMSO (lane 2), C18:0 with triacin C (acyl-CoA–specific synthetase inhibitor, lane 3), C18:0 with H89 (PKA inhibitor, lane 4), C18:0 with Gö6983 (PKC inhibitor, lane 5), C18:0 with Akti-1/2 (Akt inhibitor, lane 6), C18:0 with U0126 (MEK inhibitor, lane 7), C18:0 with rapamycin (mTOR inhibitor, lane 8), C18:0 with SB2003580 (p38 MAPK inhibitor, lane 9), C18:0 with SP600125 (JNK inhibitor, lane 10), C18:0 with AT7519 (pan-CDK inhibitor, lane 11), C18:0 with SL0101–1 (RSK inhibitor, lane 12), and C18:0 with CHIR99021 (GSK-3 inhibitor, lane 13) are as indicated. C, immunoblot analysis of CHOP protein to screen for CDK-specific inhibition of SFA-induced CHOP induction. VSMCs were pretreated with 10 μm of each inhibitor for 2 h prior to co-treatment with 500 μm C18:0 and 10 μm of each inhibitor for 6 h. Total cell lysates of VSMCs were then prepared. BSA (Veh) with DMSO (lane 1), C18:0 with DMSO (lane 2), AT7519 (lane 3), flavopiridol (lane 4), dinaciclib (lane 5), RO-3306 (lane 6), AZD5438 (lane 7), roscovitine (lane 8), palbociclib (lane 9), CAY10574 (lane 10), and LDC000067 (lane 11) in the presence of 500 μm C18:0 are as shown. D, immunoblot analysis of ATF4, CHOP, and GAPDH proteins in VSMCs co-treated with SFA and CDK9 inhibitor. VSMCs were pretreated with DMSO (−) or 30 μm CAY10574 (CDK9 inhibitor) for 2 h prior to co-treatment with 250 μm C18:0 and DMSO or 30 μm CAY10574 for 6 h. Total cell lysates of VSMCs were then prepared. qRT-PCR analyses of CHOP (E), sXBP-1 (F), and BiP (G) in VSMCs co-treated with SFA and CDK9 inhibitor are as shown. VSMCs were pretreated with DMSO or 30 μm CAY10574 (CDK9 inhibitor) for 2 h prior to co-treatment with 250 μm C18:0 and 30 μm CAY10574 for 6 h. Total RNA of VSMCs was then prepared (n = 4). One-way ANOVA with a Student-Newman post hoc test was used for statistical analysis. *, p < 0.05.

Figure 3.

CDK9 inhibitor or heterozygous knockout (+/−) of CDK9 blocked ER stress-mediated CHOP induction in VSMCs. A, immunoblot analysis of ATF4, CHOP, and GAPDH in VSMCs co-treated with thapsigargin (TG) or tunicamycin (TM) and CDK9 inhibitor. VSMCs were pretreated with DMSO or 30 μm CAY10574 for 2 h prior to co-treatment with DMSO (Veh), 0.1 μm thapsigargin, or 0.1 μg/ml tunicamycin and DMSO (−) or 30 μm CAY10574 (+) for 6 h. B, qRT-PCR analysis of CHOP in VSMCs co-treated with thapsigargin or tunicamycin and CDK9 inhibitor. VSMCs were pretreated with DMSO or 30 μm CAY10574 for 2 h prior to treatment with DMSO (Veh) or co-treatment with 0.1 μm thapsigargin, or 0.1 μg/ml tunicamycin and DMSO (−) or 30 μm CAY10574 (+) for 6 h. Total RNA of VSMCs was then prepared (n = 4). C, immunoblot analysis of CDK9, CHOP, ATF3, BiP, and GAPDH proteins in BSA (Veh) or 500 μm C18:0 treated scrambled (Scramb.) sg or CDK9 heterozygous KO VSMCs for 6 h. D, levels of CHOP; E, BiP; and F, sXBP-1 by qRT-PCR in BSA or 250 μm C18:0 treated Scramb. sg or CDK9 hetero-KO VSMCs (n = 4). One-way ANOVA with a Student-Newman post hoc test was used for statistical analysis. *, p < 0.05.

Figure 4.

Flavopiridol blocks vascular calcification in vitro and in vivo. A, mineralization of VSMCs treated with C18:0 and CDK9 inhibitor. VSMCs were incubated with 2.0 mm phosphate with BSA and DMSO (Veh), 250 μm C18:0 and DMSO, or 50 nm flavopiridol (CDKi) for 7 days (n = 6). *, p < 0.05. One-way ANOVA with a Student-Newman post hoc test was used for comparison between DMSO and flavopiridol-treated VSMCs. B, aortic CHOP expression in DBA/2J mice treated with flavopiridol. 5/6 nephrectomized DBA/2J mice (n = 8) were treated by daily i.p. injection of flavopiridol (CDKi, 0.5 or 2.5 mg/kg) for 8 weeks. C, representative photograph (×10) of the lesions in aortic arches stained with von Kossa. Arrows (black lesions) indicate calcification. D, quantitative analysis of calcified lesions in the aortic arches; E, aortic calcium content of DBA/2J mice treated with flavopiridol. ****, p < 0.0001; ***, p < 0.001; **, p < 0.01; and *, p < 0.05 (one-way ANOVA with a Student-Newman post hoc test).

Figure 5.

Cyclin T1 (Ccnt1) gene is required for ER stress-mediated CHOP induction and vascular calcification. A, immunoblot analysis of CDK9, cyclin T1, cyclin T2, cyclin K, ATF4, CHOP, and GAPDH proteins in BSA (Veh) or 500 μm C18:0 treated scrambled (Scramb.) sg or Ccnt1 KO VSMCs for 6 h. B, qRT-PCR analysis of CHOP in BSA or 250 μm C18:0 treated Scramb. sg or Ccnt1 KO VSMCs (n = 4). C, qRT-PCR analysis of CHOP in DMSO, 0.1 μm thapsigargin (TG), or 0.1 μg/ml tunicamycin (TM)-treated Scramb. sg or Ccnt1 KO VSMCs (n = 4). D, mineralization of Ccnt1 KO VSMCs. VSMCs were incubated with 2.0 mm phosphate with or without 250 μm C18:0 for 7 days (n = 6). One-way ANOVA with a Student-Newman post hoc test was used for statistical analysis. *, p < 0.05.

Figure 6.

Cyclin T2 (Ccnt2) and cyclin K (Ccnk) deficiency induces CHOP expression and calcification, opposite of cyclin T1 deficiency. A, immunoblot analysis of CDK9, cyclin T1, cyclin T2, cyclin K, ATF4, CHOP, and GAPDH proteins in BSA (Veh) or 500 μm C18:0 treated scrambled (Scramb.) sg or Ccnt2 KO VSMCs for 6 h. B, qRT-PCR analysis of CHOP in BSA (Veh) or 250 μm C18:0 treated Scramb. sg or Ccnt2 KO VSMCs (n = 4). C, immunoblot analysis of CDK9, cyclin T1, cyclin T2, cyclin K, ATF4, CHOP, and GAPDH proteins in BSA (Veh) or 500 μm C18:0 treated Scramb. sg or Ccnk heterozygous KO VSMCs for 6 h. D, qRT-PCR analysis of CHOP in BSA (Veh) or 250 μm C18:0 treated Scramb. sg or Ccnk heterozygous KO VSMCs (n = 4). E, mineralization of Ccnt2 KO VSMCs. VSMCs were incubated with 2.0 mm phosphate with or without 250 μm C18:0 for 7 days (n = 6). F, mineralization of Ccnk hetero-KO VSMCs. VSMCs were incubated with 2.0 mm phosphate with or without 250 μm C18:0 for 7 days (n = 6). One-way ANOVA with a Student-Newman post hoc test was used for statistical analysis. *, p < 0.05.

Figure 7.

CDK9–cyclin T1 complex regulates the recruitment of ATF4 to the CHOP promoter under ER stress. A, HEK293T cells were co-transfected with firefly luciferase reporter plasmid containing three repeats of the C/EBP–ATF composite (−303/−292) of the mouse CHOP gene, pCMV-LacZ, and control or human ATF4 expression plasmid. 4 h after transfection; cells were incubated in DMEM containing 10% fetal bovine serum with DMSO (Veh) or CDK9 inhibitors (CDK9i) such as 300 nm flavopiridol or 30 μm CAY10574 for 24 h. The results are expressed as luciferase (Luc)/β-gal units of induction (n-fold) over the control value for each construct. One-way ANOVA with a Student-Newman post hoc test was used for comparison between vehicle, flavopiridol, and CAY10574-treated cells (n = 6). B, ChIP analysis of SFA-induced ATF4 recruitment to the C/EBP-ATF composite of the CHOP promoter. VSMCs were pretreated with 10 μm flavopiridol (CDK9i) for 2 h, and then treated with BSA (Veh) or 250 μm C18:0 with or without 10 μm CDK9i. After 6 h, chromatin fractions of VSMCs were prepared. ChIP was performed with rabbit control IgG or ATF4 antibodies. Purified immunoprecipitated DNA was analyzed by qRT-PCR with primers for the C/EBP–ATF composite of the CHOP promoter. The results are expressed as the percentage of antibody binding versus the amount of PCR product obtained using a standardized aliquot of input chromatin (% input) (n = 3). One-way ANOVA with a Student-Newman post hoc test was used for statistical analysis. *, p < 0.05. C, phosphorylation of ATF4 by the CDK9–cyclin T1 complex was analyzed using an in vitro kinase assay. Recombinant human ATF4 and human CDK9–cyclin T1 complex were incubated with or without 100 nm flavopiridol (CDK9i) in a kinase buffer containing [γ-³²P]ATP. ATF4 was phosphorylated by both CDK9–cyclin T1 and CDK9–cyclin T2 complexes, and flavopiridol blocked phosphorylation of ATF4. The CDK9–cyclin T1 complex more potently phosphorylated ATF4. Additionally, CDK9 itself was autophosphorylated by the CDK9–cyclin T1 complex. D, human ATF4 phosphorylation site by CDK9. E, luciferase assay using ATF4 with alanine point mutations on each predicted CDK9-specific phosphorylation site. HEK293T cells were co-transfected with firefly luciferase reporter plasmid containing three repeats of the ATF4RE (−303/−292) of the mouse CHOP gene, pCMV-LacZ, and control or human ATF4 WT or serine/threonine mutation expression plasmid (n = 3). One-way ANOVA with a Student-Newman post hoc test were used for statistical analysis. **, p < 0.01 versus ATF4 WT.

Figure 8.

CDK9–cyclin T1 complex activates ATF4 activity through multiple phosphorylations under ER stress, leading to vascular calcification in CKD. The proposed mechanism of CHOP induction by ATF4 with CDK9–cyclin complexes in vascular calcification is shown. CKD increases levels of C18:0. C18:0 activates the PERK–eIF2α–ATF4–CHOP axis of ER stress signaling, resulting in calcification. The CDK9–cyclin T1 complex phosphorylates and mediates translocation of ATF4 to the ATF4RE of the CHOP promoter, and cyclins T2/K compete with cyclin T1 for binding with CDK9.