Zinc finger protein 91 loss induces cardiac hypertrophy through adenosine A1 receptor down‐regulation under pressure overload status

Abstract The function of zfp91 is mainly studied in vitro, but there is no study in vivo. Accumulative data suggest that zfp91 may be an important gene to regulate all aspects of human response. However, there are no data to date about the function of zfp91 on cardiac homeostasis. Thus, we aimed to observe the role of zfp91 gene in mouse cardiomyocytes on myocardial homeostasis and related mechanisms under pressure overload. In the study, zfp91 mRNA and protein levels were significantly reduced in TAC‐operated WT mice as compared with controls. Genetic ablation of zfp91 dramatically led to pathological cardiac dysfunction and hypertrophy after transverse aortic constriction (TAC). Adenosine A1 receptor (Adora1) mRNA and protein expressions were significantly down‐regulated in the heart of zfp91‐deletion mice with TAC. Zfp91 overexpression reversed isoproterenol‐induced cardiomyocyte hypertrophy, which was abolished by selective Adora1 antagonist. Dual‐luciferase reporter and ChIP‐qPCR assays indicated that zfp91 acted on Adora1 promoter through its binding site. Last, Adora1 agonist rescued heart dysfunction and cardiac hypertrophy in zfp91 loss mice after TAC. Zfp91 may transcriptionally regulate Adora1 expression in the heart, which mainly maintained cardiac homeostasis under pressure overload status. It will provide a new approach to treat cardiac hypertrophy.

lead to heart dysfunction, the obstruction and disorder of electrocardiographic conduction, and eventually lead to heart failure and sudden death. [2][3][4][5] Although there are extensive and in-depth studies on the mechanism of myocardial hypertrophy, the concrete mechanism is still unclear. It has been shown that the activation of some transcription factors, including GATA binding protein 4 (GATA4) and myocyte enhancer factor 2 (MEF2), leads to heart remodelling and hypertrophy. 6,7 Zinc finger transcription factors can be used as transcriptional activators or inhibitors to regulate the expression of downstream target genes, so as to produce a wide range of molecular effects in maintaining cell homeostasis, regulating the expression of a variety of anti-inflammatory and/or pro-inflammatory factors, as well as cell proliferation, migration and apoptosis and other physiological or pathological processes. 8,9 The transcription factor zinc finger protein 91 (zfp91) belongs to zinc finger protein family. It contains the classical domain C2H2 which binds to nucleic acid, and regulates the expression of many downstream target genes. Until now, the function of zfp91 is mainly studied in vitro, but there is no study in vivo.
Accumulative data suggest that zfp91 may be an important gene to regulate all aspects of human response. 10 Once it mutates, it may induce a variety of diseases.
However, there are no data to date about the function of zfp91 on regulating cardiac homeostasis. Therefore, in the present study we knock out zfp91 gene in mouse cardiomyocytes to observe its role on myocardial homeostasis and related mechanisms under pressure overload status.

| Mice, transverse aortic constriction (TAC) model and drug therapy
Mice with a C57BL/6 genetic background were housed in groups with 12-hour dark/light cycles and with free access to food in accordance with the regulations on mouse welfare and ethics of Nanjing Hospital affiliated to Nanjing Medical University (Nanjing, China). The animal protocol was reviewed and approved by the Ethics Committee of Nanjing Hospital Affiliated to Nanjing Medical University. Cervical dislocation was a method to provide the mouse with a fast and painless death. Zinc finger protein 91 (zfp91)-floxed mice were maintained on a C57BL/6 genetic background. To delete zfp91 in cardiomyocytes, zfp91-floxed mice were crossed with a-myosin heavy chain (αMHC)-Cre mice, and the progenies were genotyped by PCR. DNA primers for zfp91 genotyping were as follows: 5′-AGTTAGGCTTTGCTTTGA-3′ and 5′-ATAATCAGGCTTAGGTGC-3′ to amplify wild-type and 5′ floxed alleles, respectively. PCR products for wild-type and floxed alleles were 223 bp and 260 bp, respectively.
TAC was generated following a method previously reported in mice with slight modifications. 11 Briefly, 2-month-old mice were anaesthetized intraperitoneally with pentobarbital sodium (30-50 mg/ kg). Before TAC, we measured mice the aortic arch diameter by echocardiography. Furthermore, TAC procedure was only done by an experienced operator. An endotracheal tube was introduced into the trachea, and a volume-cycled rodent respirator (model 683; Harvard Apparatus) provided positive pressure ventilation at 2-3 mL per cycle with a respiratory rate of 120 cycles per minute. After the left parasternal skin incision, the transverse aorta was exposed between the thymus gland, and a 7-0 silk suture was placed around a 27-gauge blunt-ended needle on the transverse aorta, which was removed immediately to yield a narrowing 0.4 mm in diameter. The sham-operation mice were performed using the same procedure without the transverse aortic ligation. The chest wound was closed with a 7-0 silk suture.

| Cardiac function assessment by echocardiography
On day 42 following the TAC induction, an echocardiographic examination was performed using a Vevo 770 UBM system or a Vevo 2100 UBM system (Visual Sonics), equipped with a 30-MHz transducer, which was used for non-invasive transthoracic echocardiography.
Two-dimensional guided M-mode tracings were recorded. The internal diameter of the LV in the short-axis plane was measured at end diastole and end systole from M-mode recordings just below the tips of the mitral valve leaflets. After measurement, left ventricular enddiastolic volume (LV Vol;d) as indicative of heart diastolic function were determined. Three indexes of heart systolic function including left ventricular end-systolic volume (LV Vol;s), LV fractional shortening (LVFS) and LV ejection fraction (LVEF) were calculated. LV mass and LV mass corrected were two indexes of heart weight. 12

| Western blot analysis and quantitative realtime PCR
Western blot analyses were performed as previously reported. 13 Left ventricles of mice were dissected and snap-frozen in liq-

| Histology and immunofluorescence staining
The protocols for Masson's staining, haematoxylin-eosin staining (HE) and immunofluorescence (IF) were performed as reported previously. 13 Briefly, heart samples were first washed with ice-cold Immunofluorescence (IF) staining was performed by using anti-wheat germ agglutinin (WGA) (Abcam) antibody at 4°C room overnight. Goat anti-rabbit IgG (Abcam) diluted in PBS was then incubated for 2 hours at room temperature. Fluorescence microscopy images were obtained with a Research Fluorescence Microscope (Olympus America Inc) equipped with a digital camera. Images were collected and recorded by using Adobe PhotoshopR 5.0 (Adobe Systems Inc) on an IBM R52 computer (IBM). Fibrotic area and cardiomyocyte area were measured and averaged after calculating in 5 high-power fields.

| Microarray analysis
Three hearts (left ventriculars) were dissected from each group of control and zfp91-deletion mice with TAC for 1 week, and total RNA was isolated from these hearts. In total, six mRNA samples were applied for microarray analysis (Affymetrix Mouse 430 2.0 Array, Shanghai Biotechnology). The data were analysed with the eBioService System (Shanghai Biotechnology).

| Construction of plasmids and recombinant virus vectors
To construct expression plasmid, zfp91 cDNA was amplified by RT-PCR and its C-terminal fusion Flag was constructed on pcDNA3.1 eukaryotic expression vector, which was cloned into the pcDNA3.1-zfp91-FLAG plasmid. Additionally, 200 bp near the predicted binding site of Adora1 gene was constructed into pGL3 basic vector. At the same time, the mutant plasmid of Adora1 gene was constructed: Adora1 promoter-wild type (sense strand): TAGGGGAGGAGA

CCGCCGCCCCCGCCCCCGCCCCC TCCCCC TCCCGGTGTGC
GGAGCCCGATTGTC and Adora1 promoter-mutant (sense strand): TACTGCCGGTGTGCGGAGCCCGATTGTC. Also, recombinant adenovirus that expressed full-length mouse zfp91 cDNA with Flag tag (Ad-ZFP91) was constructed. Viruses were amplified and titrated in 293 cells according to the manufacturer's instruction. Adenovirus containing empty plasmid (Ad-con) served as control. 14

| Cardiomyocyte culture and adenovirus transfection
Ventricular myocytes were prepared from Sprague-Dawley rats (2-3 days after birth) and cultured on Permanox chamber slides (Thermo Scientific Nunc) coated with 1% gelatin and 0.0015% laminin solution, as previously described. 15 First, these cells were

| Dual-luciferase reporter assay and chromatin immunoprecipitation (ChIP)-qPCR analysis
293 cell lines were inoculated in 24-well culture plate the day before transfection, and the cell density was 70%-80% at the time of transfection. 2 μL lipo2000 was diluted to 50 μL serum-free DMEM, plasmids were diluted to 50 μl serum-free DMEM and incubated at room temperature for 20 minutes, and serum-free DMEM was supplemented to 100 μL. The culture medium from the dish was removed; 200 μL of the transfection complex prepared in the previous step was added and cultured at 37°C for 5 hours. The medium was removed and changed to 0.5 mL complete medium, and cultured at 37°C for 48 hours and transfected repeatedly for 5 times. Transfection plasmid combination was as follows: ① pCDNA3.1 + pGL3-Basic + pRL-TK; ② zfp91 + pGL3-Basic + pRL-TK; ③ pCDNA3.1 + Adora1-WT + pRL-TK; ④ zfp91 + Adora1-WT + pRL-TK; ⑤ pCDNA3.1 + Adora1-MT + pRL-TK; and ⑥ zfp91 + Adora1-MT + pRL-TK. Lysate cells were collected 10 000 g of lysate and centrifuged for 5 minutes, and supernatant was taken as test solution and then operated according to the instructions of double luciferase test kit. In the case of taking the marine luciferase as internal reference, the RLU value determined by the firefly luciferase was divided by the RLU value determined by the marine luciferase. According to the ratio obtained, the activa- Data are given as means ± SEM. *P < .05 or **P < .01 vs WT or ZFP91 F/F mice F I G U R E 2 Zfp91 loss in cardiomyocytes had no effect on heart weight to bodyweight ratio, heart function and histology in mice after 10 months. A, Heart weight to bodyweight ratio (n = 7); B-G, echocardiography measurement (n = 7). H, Heart histology. LV: left ventricular; LV Vol;d: left ventricular end-diastolic volume; LV Vol;s: left ventricular end-systolic volume; EF: eject fraction; FS: fractional shortening; HE: haematoxylin and eosin. Data are given as means ± SEM F I G U R E 3 Loss of Zfp91 in cardiomyocytes accelerated cardiac dysfunction and aberrant heart remodelling after TAC operation. A-G, Echocardiography measurement (n = 8); H, heart weight to bodyweight ratio (n = 8); I, gross heart morphology; J, HE staining; K, Masson staining; L, quantitative analysis (n = 5); M, WGA staining; N, quantitative analysis (n = 5). WGA: wheat germ agglutinin. Data are given as means ± SEM. **P < .01 vs ZFP91 F/F or ZFP91 F/F ; αMHC-Cre mice; # P < .05 or ## P < .01 vs ZFP91 F/F mice with TAC 95°C for 2 minutes, 72°C for 1 minute, 95°C for 30 seconds, and 55°C for 10 seconds (repeat 80 times), 30°C for 1 minutes. After PCR, relative data quantification was performed using the 2 −ΔΔCt method, and the result was calculated in the form of %Input, which was given by the following formula: %Input = 2 (Ctinput−CtChIP) × input dilution factor × 100. A segment of the Adora1 promoter containing the Zfp91-binding sites was amplified using the primers 5′-CTCACACTGAATCACTTCCTTTGTTTAG-3′(forward) and 5′-GGTAGCCAGCCGAGACTCC-3′(reverse). 16

| Statistical analysis
Data are presented as means ± SEM values. Statistical analyses were performed using SPSS version 20 (SPSS Inc). Tukey's post hoc test was used between two groups following a P < .05 of one-way ANOVA. A P < .05 was considered statistically significant.

| Generation of cardiomyocyte-specific Zfp91deletion mice
Compared with the control group, we found that the expression levels of zfp91 mRNA and protein in the heart of mice with TAC after 1 month were significantly reduced (P < .05) ( Figure 1A-C).
In order to further clarify whether zfp91 plays an important role in cardiac structure and functional homeostasis, we specifically knock out zfp91 gene in mouse cardiomyocytes. Female Zfp91floxed mice (Zfp91 F/F ) were crossed with male aMHC-Cre mice to obtain Zfp91 F/+ ; αMHC-Cre mice. Mice were genotyped by PCR ( Figure 1D), and loss of Zfp91 in heart tissue was confirmed by RT-PCR and Western blot analysis. The results revealed that Zfp91 mRNA and protein levels were significantly reduced in Zfp91 F/F -deletion mice as compared to controls (P < .01) ( Figure 1E-G).

| Loss of Zfp91 in cardiomyocytes accelerated cardiac dysfunction and aberrant heart remodelling after TAC
After 10 months, there were not any differences about heart weight to bodyweight ratio, as well as cardiac systolic and diastolic function between Zfp91 F/F and Zfp91 F/F ; αMHC-Cre mice (P > .05) (Figure 2A-G)

. Additionally, HE and MASSON staining results
showed that the heart histology in Zfp91 F/F mice did not change significantly as compared to Zfp91 F/F ; αMHC-Cre mice ( Figure 2H). Therefore, we used aortic constriction to increase pressure load of the mouse heart, and observed whether the absence of zfp91 in cardiomyocyte aggravates myocardial injury under the pathological stimulus.
Six weeks after TAC surgery, echocardiography measurement indicated that heart systolic function (EF and FS) was markedly impaired in control TAC mice and in Zfp91-deletion TAC mice as compared to control mice without TAC surgery (P < .01) ( Figure 3A-C).

Furthermore, loss of zfp91 notably decreased EF and FS values, and
greatly increased the left ventricular internal diastolic and systolic volume, LV mass and LV mass (corrected) compared with control TAC mice (P < .01) (Figure 3A,D-G). Additionally, the heart weight to bodyweight ratio was raised significantly in Zfp91-deletion TAC mice as compared to control mice with or without TAC surgery (P < .01) ( Figure 3H). Last, histological analysis displayed markedly increased general heart size, fibrosis area and cardiomyocyte size in Zfp91deletion TAC mice as compared to control mice with or without TAC surgery (P < .01) ( Figure 3I-N).
Taken together, these results indicated that genetic ablation of Zfp91 dramatically led to pathological heart remodelling and heart dysfunction after TAC surgery. Thus, Zfp91 may function to protect against myocardial damage in the heart under pressure overload condition.

| mRNA sequencing analysis of heart tissue 1 week after TAC in mice
According to heat map shown in Figure 4A, there were 201 differential mRNAs in heart tissues of Zfp91-deletion TAC mice as compared to the control TAC mice (P < .05 or P < .01). Among them, 133 downregulated mRNAs were found in heart tissue of Zfp91-deletion mice with TAC, and 68 mRNAs of heart tissues were up-regulated (P < .05 or P < .01).
Next, we analysed Gene Ontology (GO) information from biological process (BP), cellular component (CC) and molecular function (MF) categories. In the BP aspect, muscle system process, regulations of heart contraction and cardiac muscle contraction may be the key BPs involving in the development of Zfp91 lossinduced cardiac hypertrophy (P < .01). In the CC aspect, protein binding, channel activity and substrate-specific channel activity may take an important role in the pathology of Zfp91 loss-induced cardiac hypertrophy (P < .01). In the MF aspect, intracellular, cytoplasm and cell projection may play a great role in development of Zfp91 loss-induced cardiac hypertrophy (P < .01) ( Figure 4B).

Additionally, Kyoto Encyclopedia of Genes and Genomes (KEGG)
pathway was also under consideration in our study. We found mRNAs related to hypertrophic cardiomyopathy were changed significantly (P < .01) ( Figure 4C).
From analysis of heat map, Adora1 mRNA was significantly down-regulated in heart of zfp91-deletion mice with TAC as compared to control TAC mice (P < .01). RT-PCR confirmed that zfp91-deletion mice with TAC displayed significant reduction of zfp91 and adora1 mRNAs in heart of zfp91-deletion mice with TAC (P < .01) ( Figure 4D,E). From bioinformatics analysis, Adora1 belonged to cell projection, and muscle skeletal and cardiovascular systems. Last, Western blot analysis also showed that Adora1 was notably decreased at protein level in heart of zfp91-deletion mice with TAC as compared to control TAC mice (P < .01) ( Figure 4F,G).
In summary, these results indicated that adora1 may be the major target gene of ZFP91.
F I G U R E 4 mRNA sequencing analysis of heart tissue 1 week after TAC surgery in mice. A, heat map; B, GO analysis; C, KEGG pathway analysis; D, ZFP91 mRNA level measured by real-time PCR (n = 5); E, A1R mRNA level measured by real-time PCR (n = 5); F, Western blot analysis of cardiac A1R protein expression. G, quantitative analysis (n = 5). A1R: adenosine A1 receptor; GO: Gene Ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes. Data are given as means ± SEM. ** P < .01 vs ZFP91 F/F mice with TAC

| Selective Adora1 antagonist abolished reversion of ISO-induced cardiomyocyte hypertrophy by Zfp91 overexpression
Next, we further investigated the role of overexpressing Zfp 91 gene in primary neonatal rat cardiomyocytes (PNRCs) interfered with ISO. After 24 hours, the sizes of PNRCs in ISO group were significantly large as compared to those in control group (P < .01). However, after simultaneous transfection with Ad-ZFP91-Flag, the sizes of PNRCs were obviously small as compared to those of ISO group, which was abolished by Adora1 antagonist 8-Cyclopentyl-1,3-dimethylxanthine (CPT) (P < .01) ( Figure 5A,B).
Meanwhile, success overexpression of ZFP91 protein in PNRCs was confirmed by Western blot shown in Figure 5C,D. Also, expression of Adora1 protein was markedly increased after transfection with Ad-ZFP91-Flag in PNRCs ( Figure 5C,D).
In short, overexpression of ZFP91 alleviated cardiomyocyte hypertrophy induced by ISO, which may be closely associated with Adora1 up-regulation.

| Interaction between transcription factor ZFP91 and adora1 promoter
According to the information of ZFP-related transcription factors, the possible binding site of zfp91 and adora1 promoter was predicted on the website (http://jaspar.gener eg.net/). When zfp91 and its possible binding site in adora1 promoter were cotransfected, the F I G U R E 5 Selective Adora1 antagonist abolished reversion of ISO-induced cardiomyocyte hypertrophy by ZFP91 overexpression. A, cTnT staining (red) of cardiomyocytes and the cell nucleus (blue). B, Quantitative analysis (n = 5); C, Western blot analysis of ZFP91-FLAG and A1R protein expression; D, quantitative analysis (n = 5). ISO: isoproterenol; cTnT: cardiac troponin T; Ad: adenovirus; CPT: 8-Cyclopentyl-1,3-dimethylxanthine. Data are given as means ± SEM. ** P < .01 vs control group; # P < .05 or ## P < .01 vs ISO group dual-luciferase activity was significantly enhanced as compared to its control group (P < .01). Then, the possible binding site of zfp91 was mutated, and the result showed that the dual-luciferase activity was not changed after cotransfection with zfp91 as compared to its control group (P > .05) ( Figure 6A).
In order to further test interaction between zfp91 and its possible binding site in adora1 promoter, we did ChIP-qPCR after H9C2 cell line successfully transfected with Ad-zfp91-Flag, which was confirmed by Western blot ( Figure 6B). Figure 6C displays that DNA products were fragmented by ultrasound for 10 minutes and their fragment size is between 100 and 500 bp.
After qPCR with the primer of the Adora1 promoter segment, we found that as compared to Ig group, input value of anti-Flag group was markedly increased, indicating that Adora1 promoter segment was enriched in zfp91-Flag group (P < .01) ( Figure 6D).
Last, we sequenced the PCR products and found that they were part of Adora1 promoter sequences, including zfp91 binding site ( Figure 6E).
Totally, these results revealed that transcription factor zfp91 can act on Adora1 promoter through its binding site.

| Adora1 agonist reversed cardiac dysfunction and aberrant heart remodelling in zfp91 loss mice after TAC surgery
Next, we used potent selective Adora1 agonist N6-Cyclopentyladenosine (CPA) to treat zfp91 loss mice after TAC for 6 weeks. After zfp91-deletion TAC mice and control TAC mice treated with CPA, heart systolic and diastolic dysfunction, LV mass and LV mass (corrected) were greatly improved (P < .05 or P < .01).
In summary, these results suggested that selective Adora1 agonist can rescue the phenotype of heart dysfunction and adverse heart remodelling in zfp91 loss mice under pressure overload.

| D ISCUSS I ON
In the present study, we provided the first genetic evidence that ablation of zfp91 significantly accelerated heart dysfunction and adverse pathological heart remodelling after TAC surgery. Zfp91 overexpression reversed ISO-induced cardiomyocyte hypertrophy, which was abolished by Adora1 antagonist CPT. Then, dual-luciferase reporter and ChIP-qPCR assays demonstrated that transcription factor zfp91 can directly act on Adora1 promoter through its binding site. Last, selective Adora1 agonist can greatly improve the phenotype of adverse heart remodelling in zfp91-deletion mice after TAC operation. These results suggested that zfp91-mediated Adora1 mRNA and protein expressions were involved in the development of cardiac hypertrophy under the pressure overload status.
Unoki and his colleagues have reported that the new zfp91 gene for the first time was found in these patients with acute myeloid leukaemia and it expressed highly in most patients' leukaemia cells. Its gene products were located in the nucleus and showed the characteristics of transcription factors (including five zinc finger domains, leucine zipper and several nuclear localization signals). In vitro, they found that zfp91 may play an important role in cell proliferation and anti-apoptosis. When zfp91 expression was inhibited by antisense oligonucleotide, the cell proliferation was decreased and cell apoptosis was increased. 17 Jin and his co-workers found that zfp91 protein was overexpressed in these patients with colon cancer. By means of ChIP, EMSA gel migration and luciferase reporter assays, it was found that ZFP91 could cooperate with NF-κB/p65 to transcribe hypoxia inducible factor-1 to regulate development of colon cancer. 18 In our study, we found that transcriptional factor zfp91 can directly regulate Adora1 mRNA and protein expressions by dual-luciferase reporter and ChIP-qPCR methods. Conclusionally, Zfp91 as a transcription factor may regulate the expressions of multiple genes.
Cardiac myocytes can produce adenosine, and there are adenosine receptors (including A1, A2 and A3 receptor subtypes) in the heart. Adenosine has a high affinity for A1 receptor. 19,20 Matherne and his team overexpressed Adora1 in mice, which significantly reduced infarct area and improved heart systolic and diastolic dysfunction. 21 Kitakaze et al used phenylephrine, angiotensin II and isoproterenol to stimulate cardiomyocytes hypertrophy in neonatal rat in vitro, and found that using adenylate analogue to activate Adora1 reduced cardiomyocyte size. Then, through using adenylate analogue to treat TAC mice, it was found that it can improve cardiac hypertrophy and dysfunction. 22 Recently, Devaux's team found that Adora1 agonist can reduce myocardial hypertrophy, interstitial fibrosis and oxidative stress in mice induced by phenylephrine stimulation. 23 Totally, long-term sustained activation of β-adrenergic receptor played a key role in development of pathological ventricular remodelling and heart failure, and activation of adenylate signalling pathway exerted antagonistic effect. Thus, in our present study significant reduction of Adora1 mRNA and protein expressions in zfp91 loss mice was closely associated with heart dysfunction and adverse cardiac remodelling under pressure overload condition, which were reversed by selective Adora1 agonist intervention. Furthermore, it was also validated that in vitro test Adora1 antagonist CPT abolished reversion of ISO-induced cardiomyocyte hypertrophy by ZFP91

overexpression.
In summary, we demonstrated that zfp91 may transcriptionally regulate Adora1 mRNA expression in heart, which mainly regulated myocardial homeostasis under the status of pressure overload. It will provide a new way for treating myocardial hypertrophy.

ACK N OWLED G EM ENTS
Thanks to Zhongzhou Yang for supplying zfp91 F/F and αMHC-Cre mice.

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

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.