Transient induction of cell cycle promoter Fam64a improves cardiac function through regulating Klf15-dependent cardiomyocyte differentiation in mice

The introduction of fetal or neonatal signatures such as cell cycle promoting genes into damaged adult hearts has been vigorously pursued as a promising strategy for stimulating proliferation and regeneration of adult cardiomyocytes, which normally cannot divide. However, cell division of cardiomyocytes requires preceding dedifferentiation with sarcomere disassembly and calcium dysregulation, which, in principle, compromises contractile function. To overcome this intrinsic dilemma, we explored the feasibility of optimizing the induction protocol of the cell cycle promoter in mice. As a model of this approach, we used Fam64a, a fetal-specific cardiomyocyte cell cycle promoter that we have recently identified. We first analyzed transgenic mice maintaining long-term cardiomyocyte-specific expression of Fam64a after birth, when endogenous expression was abolished. Despite having an enhanced proliferation of postnatal cardiomyocytes, these mice showed age-related cardiac dysfunction characterized by sustained cardiomyocyte dedifferentiation, which was reminiscent of the dilemma. Mechanistically, Fam64a inhibited glucocorticoid receptor-mediated transcriptional activation of Klf15, a key regulator that drives cardiomyocyte differentiation, thereby directing cardiomyocytes toward immature undifferentiated states. In contrast, transient induction of Fam64a in cryoinjured wildtype adult mice hearts improved functional recovery with augmented cell cycle activation of cardiomyocytes. These data indicate that optimizing the intensity and duration of the stimulant to avoid excessive cardiomyocyte dedifferentiation could pave the way toward developing efficient strategy for successful heart regeneration.

of Fam64a after birth, when endogenous expression was abolished. Despite having an enhanced 23 proliferation of postnatal cardiomyocytes, these mice showed age-related cardiac dysfunction 24 characterized by sustained cardiomyocyte dedifferentiation, which was reminiscent of the dilemma. 25 Mechanistically, Fam64a inhibited glucocorticoid receptor-mediated transcriptional activation of Klf15,26 a key regulator that drives cardiomyocyte differentiation, thereby directing cardiomyocytes toward 27 immature undifferentiated states. In contrast, transient induction of Fam64a in cryoinjured wildtype 28 adult mice hearts improved functional recovery with augmented cell cycle activation of cardiomyocytes. 29 These data indicate that optimizing the intensity and duration of the stimulant to avoid excessive 30 cardiomyocyte dedifferentiation could pave the way toward developing efficient strategy for successful 31 heart regeneration. 32

Introduction 36
The limited proliferation potential of adult cardiomyocytes (CMs) is a major obstacle hindering 37 regeneration of myocardium lost following injury. The introduction of fetal-specific signatures such as 38 cell cycle promoting genes into damaged adult hearts has recently emerged as a promising strategy for 39 stimulating CM proliferation (Borden et  understanding the mechanisms of these phenomena is crucial to overcome the intrinsic dilemma. 55 Here we tested the feasibility of optimizing the induction protocol of cell cycle promoting 56 genes to find an effective way to circumvent the dilemma. As a model of this approach, we utilized 57 Fam64a (family with sequence similarity 64, member A; also known as Pimreg, Cats, or Rcs1), a fetal-58 specific CM cell cycle promoter that we have recently identified (Hashimoto et al., 2017). The strong 59 Fam64a expression in fetal CM nuclei was almost completely lost in postnatal CMs from mice 60 (Hashimoto et al., 2017) and sheep (Locatelli et al., 2020). Fam64a knockdown inhibited and its 61 overexpression enhanced fetal CM proliferation (Hashimoto et al., 2017). High expression of Fam64a 62 has been noted in various types of tumors and this expression was well correlated with poor prognosis,

property in CMs 120
Calcium homeostasis in CMs is essential for the maintenance of normal cardiac function. The Ca 2+ 121 transient measurements conducted in isolated CMs from aged mice (29-32 wks) revealed a reduction in 122 the peak amplitude and a delay in the time to peak, indicating impaired Ca 2+ mobilization in TG mice 123 as compared to WT mice ( Figure 3A-C). Although not statistically significant, a tendency was observed 124 toward an increased time constant during the decay phase ( Figure 3D) and a decreased sarcoplasmic 125 reticulum (SR) Ca 2+ content ( Figure 3E), suggesting impaired Ca 2+ re-uptake into the SR in TG mice. 126 Cell shortening in response to electrical stimuli was decreased in TG mice at all the frequencies tested, 127 indicating impairment of the CM contractile properties ( Figure 3F). These findings were corroborated 128 by qPCR analysis showing that many of the principal genes involved in Ca 2+  Surprisingly, genome-wide RNA-seq analysis of heart samples identified a pathway called circadian 155 rhythm as the most differentially regulated pathway in TG mice, as indicated by an enrichment score far 156 greater than other pathways like cardiac muscle contraction, hypertrophic cardiomyopathy, and dilated 157 cardiomyopathy ( Figure 5A; RNA-seq data have been deposited in DDBJ sequencing read archive 158 (DRA) under the accession number DRA009818, http://trace.ddbj.nig.ac.jp/DRASearch). Some of the 159 principal genes relating to circadian rhythm, including Arntl (known as Bmal1), Cry1, Per2, Npas2, and 160 Dbp, were dysregulated in TG hearts ( Figure 5B). Telemetric measurements using freely moving 161 conscious mice revealed perturbed heart rate regulation in the TG mice: (1) Heart rate was consistently 162 low in TG mice, irrespective of daytime or nighttime, throughout the course of measurements of up to 163 8 days ( Figure 5C). (2) Nighttime-to-daytime ratios of heart rate in TG mice were slightly, but 164 significantly, lower than in WT mice, and had values of less than 1, indicating an abnormal daytime 165 (inactive phase)-dominant heart rate regulation ( Figure 5D). In addition, TG mice frequently developed 166 premature ventricular contraction, either as a single form or more hazardous serial forms, in sharp 167 contrast to WT mice that displayed virtually no such arrhythmias ( Figure 5E, F). Decreased expression 168 of connexin 43 could partially explain these aberrant phenotypes ( Figure 4C). 169 Locomotor activity analysis of mice using the infrared motion detector also demonstrated 170 perturbation of rhythmic behavior in TG mice: Whereas the nighttime (active phase)-dominant activity 171 was observed in both WT and TG mice, the activity in TG mice was decreased during nighttime and 172 increased during daytime when compared to WT mice ( Figure 5-figure supplement 1). The abnormal 173 daytime-dominant heart rate regulation ( Figure 5D) might be responsible for this phenotype. These data 174 indicate that rhythmic CM electrical activity and locomotor activity were perturbed in Fam64a TG mice.

Fam64a inhibits Klf15 activity by glucocorticoid receptor-mediated transcriptional regulation, 177
thereby inducing the FGP-mediated CM dedifferentiation and the rhythm disturbance in TG mice 178 We explored the mechanisms underlying the FGP-mediated dedifferentiation and the rhythm 179 disturbance induced in TG mice by focusing on Klf15, a key regulator that drives CM differentiation by 180 repressing the FGP and establishing CM rhythmic activity (Fisch et al., 2007;Jeyaraj et al., 2012a;181 Leenders et al., 2010, 2012Zhang et al., 2015). We found that mRNA expression of Klf15 and its 182 downstream target Kcnip2 (Kv channel-interacting protein 2, also known as KChIP2) (Jeyaraj et al.,183 2012a) was strongly upregulated during the course of differentiation in WT hearts but was severely 184 depressed in Fam64a-overexpressing TG mice hearts, suggesting that Fam64a inhibits Klf15 activity at 185 the transcriptional level ( Figure 6A). We conducted a comprehensive search for interacting partners of 186 Because Fam64a could be a putative transcriptional repressor (Archangelo et al., 2006(Archangelo et al., , 2013 we tested whether Fam64a inhibits GR-mediated transcriptional activation of Klf15 using luciferase 196 reporter assay in HEK293T/17 cells. Three reporter constructs on the human KLF15 locus was used 197 ( Figure 6C). Construct (ⅰ) contained a common promoter sequence upstream of the first exon. Construct 198 (ⅱ) and construct (ⅲ) contained, respectively, three of the four and the remaining GR binding sites 199 previously reported (Asada et al., 2011;Sasse et al., 2013). At baseline, in the absence of exogenous 200 induction of GR signaling, a tendency toward a repressed activity was observed in all three constructs 201 by Fam64a overexpression ( Figure 6D). The repression was most strongly and significantly observed in 202 the construct (ⅱ), which contains the majority of the GR binding sites ( Figure 6C, D). We observed a 9 similar repression in construct (ⅱ) in the presence of exogenous induction of GR signaling by GR 204 overexpression and dexamethasone treatment ( Figure 6E). We corroborated these findings using primary 205 cultures of isolated CMs to show that Fam64a repressed Klf15 mRNA expression in the absence or 206 presence of exogenous GR induction by dexamethasone ( Figure 6F). Dexamethasone-induced activation 207 of Klf15 was completely blocked by Fam64a. These data indicate that Fam64a inhibits Klf15 activity at 208 least in part by GR-mediated transcriptional regulation through action on the previously described GR 209 binding sites. evoked an FGP-driven dedifferentiation through the Klf15-KChIP2 axis, thereby disrupting CM 257 rhythmic activity and contributing to cardiac dysfunction. Thus, we propose that Fam64a is not merely 258 a cell cycle promoter; rather, it has a previously unknown function in directing CMs toward immature 11 undifferentiated states through inhibition of Klf15. These data also show that abnormal CM rhythmic 260 activity is a useful indicator of dedifferentiation-mediated cardiac dysfunction, as well as sarcomere 261 disassembly and calcium dysregulation. 262 We demonstrated that GR could, at least in part, mediate the inhibitory effect of Fam64a on 263 Klf15 at the transcriptional level ( Figure 6B data; YUjihara performed experiments using isolated CMs and analyzed data; AH contributed to 313 molecular cloning, plasmid construction, and baculovirus production; SM supervised the study and 314 contributed to manuscript preparation. All authors critically read and approved the manuscript.

Declaration of interests 317
The authors declare no competing interests. adult (E, 6-7 wks), and aged (F, > 25 wks) stages. Data were shown as normalized to WT. n = 3-12 331 mice per group. * p < 0.05, ** p < 0.01, *** p < 0.001 as compared to WT by Student's two-tailed 332 unpaired t-test. Error bar = SEM.  In WT mice, no significant change was observed in LVDd, LVDs, and FS over the course of the 341 study, with the only exception of LVDd at 31-35wks significantly larger as compared to 5-10 wks 342 (One-way ANOVA with Tukey's post hoc test). By contrast in TG mice, LVDd and LVDs were 343 significantly increased, and FS was significantly decreased at 21-25 wks and afterwards as 344 compared to 5-10 wks (One-way ANOVA with Tukey's post hoc test). Consequently, in TG mice, 345 LVDd and LVDs were significantly larger and FS was significantly smaller at all stage as compared 346 to WT mice of the same age (* p < 0.05, ** p < 0.01, and *** p < 0.001 as compared to WT by Student's two-tailed unpaired t-test.). n = 17-83 mice per group at each age which partially includes 348 repetitive measurements of the same animal at different age. Error bar = SEM. CMs from 3 WT mice and 23-50 CMs from 2 TG mice. In B-E, * p < 0.05 as compared to WT by 364 Student's two-tailed unpaired t-test. In F, * p < 0.05 and ** p < 0.01 as compared to WT under the 365 same stimulating frequency by Student's two-tailed unpaired t-test. Error bar = SEM. 366 G-I. qPCR analysis of Ca 2+ handling genes in WT and TG mice hearts at neonatal (G, P12-P15), adult 367 (H, 6-7 wks), and aged (I, > 25 wks) stages. Data were shown as normalized to WT. n = 3-6 mice 368 per group. * p < 0.05, ** p < 0.01, *** p < 0.001 as compared to WT by Student's two-tailed unpaired 369 t-test. Error bar = SEM. A. Immunofluorescence for sarcomeric α-actinin (red) and DAPI (blue) in WT and TG mice heart 373 sections at > 25 wks aged stages. In WT mice, highly organized sarcomere structure with small, rod-374 shaped nuclei (arrows) was observed. In contrast, disorganization of sarcomeres with enlarged, irregular-shaped nuclei were frequently observed in TG mice (arrows). Scale bar = 20 µm. 376 B. Representative image of freshly isolated CMs from WT and TG mice at > 25 wks aged stages, 377 obtained by differential interference contrast optics. CM cell size was evaluated as a two-dimensional 378 projected area. n = 75 CMs from 3 WT mice and 113 CMs from 2 TG mice. ** p < 0.01 as compared 379 to WT by Student's two-tailed unpaired t-test. Error bar = SEM. Scale bar = 50 µm.  A. qPCR analysis of Klf15 and Kcnip2 (KChIP2) at fetal, neonatal, adult, and aged stages from WT 408 (circle) and TG (triangle) mice hearts. Data were shown as normalized to WT at fetal stage set at 1. 409 In WT mice, Klf15 expression was significantly increased at 6W-13W and afterwards as compared 410 to fetal stage (E15-E18) (One-way ANOVA with Tukey's post hoc test). Likewise, Kcnip2 411 expression was significantly increased at P12-P24 and afterwards as compared to fetal stage (E15-412 E18) (One-way ANOVA with Tukey's post hoc test). By contrast in TG mice, the expressions of both 413 genes were significantly attenuated at all stage as compared to WT mice of the same age (* p < 0.05, 414 ** p < 0.01, and *** p < 0.001 as compared to WT by Student's two-tailed unpaired t-test.). n = 3-8 415 mice per group. Error bar = SEM. 416 B. Immunoprecipitation (IP) against FLAG peptide that was expressed as a C-terminal tag of 417 overexpressing Fam64a protein in TG mice hearts, followed by western blotting (WB) using 418 glucocorticoid receptor (GR) antibody, which detected GR protein in TG, but not in WT mice heart 419 lysates. This indicates that Fam64a forms complex with GR in CMs. Western blotting using FLAG 420 antibody correctly detected Fam64a-FLAG fusion protein (*) in TG, but not in WT mice heart 421 lysates, validating the immunoprecipitation procedure.

Immunoprecipitation and mass spectrometry 517
We identified the interacting partners of Fam64a using immunoprecipitation against the FLAG peptide 518 that was expressed as a C-terminal tag of the overexpressed Fam64a protein in TG mice hearts, followed 519 by mass spectrometry analysis (n = 2 biological replicates). Immunoprecipitates from WT mice hearts 520 were used as a negative control. Heart tissues were freshly isolated from WT and TG mice, minced, and 521 homogenized using a Kinematica™ Polytron™ homogenizer (Fisher Scientific) in IP lysis buffer 522 (Thermo-Fisher) or cytoplasmic extraction reagent I & II (Thermo-Fisher) in the presence of a protease 523 inhibitor cocktail (Thermo-Fisher). After centrifugation and protein quantification, the lysates were 524 subjected to immunoprecipitation using the EZview Red Anti-FLAG M2 affinity gel system (F2426, 525 performed with 3× FLAG peptide (F4799, Sigma-Aldrich). The immunoprecipitation procedure was 527 validated by western blotting using FLAG and Fam64a antibodies, which correctly detected the 528 Fam64a-FLAG fusion protein in TG, but not in WT, mouse heart lysates ( and Min # peptides = 2. Proteins detected only in TG samples, but not in WT samples, were considered 559 as candidate interacting partners of Fam64a, and the interaction of those proteins with Fam64a in heart 560 tissues was subsequently confirmed by immunoprecipitation and western blotting using specific 561 antibodies. All mass spectrometry data have been deposited in ProteomeXchange Consortium via jPOST, 562 with the dataset identifiers PXD020570 and JPST000921 (Preview code for reviewers 563 https://repository.jpostdb.org/preview/497638005f1cd5ace27c1, Access key: 6340). 564 565

Quantitative PCR (qPCR) 566
Heart tissues were collected from mice, cut into small pieces, and immediately immersed in RNAlater® 567 Stabilization Reagent (Qiagen, Germany). The stabilized tissues were homogenized with a 568 Kinematica™ Polytron™ homogenizer (Fisher Scientific), and total RNA was isolated using the 569 ISOGEN or ISOGEN-II systems (Nippon Gene, Japan). For cultured CMs, harvested cell pellets were 570 processed similarly to heart tissues but without the use of the homogenizer. After assessing RNA yield 571 and quality using a NanoDrop One spectrophotometer (Thermo-Fisher), the RNA samples were reverse-572 transcribed with PrimeScrip RT Master Mix (TaKaRa Bio, Japan), and quantitative real-time PCR was 573 performed using TaqMan® Fast Advanced Master Mix in a StepOnePlus™ real-time PCR system 574 (Applied Biosystems, USA). Quantification of each mRNA was carried out with Actb or Ubc as 575 reference genes, using the ΔΔCT method, as previously described (Hashimoto et al., 2018). 576 577

Luciferase reporter assay 578
Three reporter constructs spanning the promoter region of human KLF15 locus were used ( Figure 6C Cells were treated with dexamethasone (Dex) at 1 µM for 24 h. Luciferase activity was measured on the 591 next day using the LightSwitch™ luciferase assay system (SwitchGear Genomics) as per the 592 manufacturer's protocol, as previously described (Hashimoto et al., 2017). The luciferase activity of 593 each reporter construct was normalized to that of the control reporter construct (pLightSwitch_Prom) 594 and was expressed as the activity of the control empty vector set at 1.

Locomotor activity measurement 611
The locomotor activity of mice was monitored using an infrared motion detector (Actimo-100, 612 Shinfactory, Japan), which consists of a free moving space (30 × 20 cm 2 ) with a side wall equipped with 613 photosensors at 2 cm intervals to scan animal movement as described (Kurokawa et al., 2011). Activity 614 counts accumulated over a 1 h period were measured for a total of 4 days in a 12 h light:12 h dark cycle 615 (lights on at 8 am). Total activity counts during the daytime (8 am to 8 pm) and nighttime (8 pm to 8 616 am) were considered to reflect the locomotor activity in each phase. During the nighttime, we found that 617 the most mice showed characteristic biphasic patterns of locomotor activity, i.e., the first peak during 618 the time period from 8 pm to 2 am, and the second peak during the time period from 2 am to 8 am 619 (typical example shown in Figure 5-figure supplement 1). Thus, the peak activity counts in each phase 620 were used as a measure of the locomotor activity during nighttime. Modified mRNA for EGFP was used as a negative control. The chest and skin were closed and mice 662 were allowed to recover on a heating pad until normal respiration was obtained. In this procedure, the 663 delivery of modified mRNA at the time of cryoinjury allowed to avoid a second surgery later, which 664 contributed to reduce mortality rate. Echocardiography was performed before and after cryoinjury to 665 assess cardiac function over the course of experiments as described in the aforementioned section. At the end of the experiments (5 weeks after cryoinjury), mice were sacrificed, and frozen tissue sections 667 containing a cryoinjured region were assessed for the expression of cell cycle marker Ki67 by 668 immunofluorescent analysis, as described in the aforementioned section. 669

RNA-seq 671
Heart tissues were collected from mice, cut into small pieces, and immediately immersed in RNAlater® 672 Stabilization Reagent (Qiagen). The stabilized tissues were homogenized with a Kinematica™ 673 Polytron™ homogenizer (Fisher Scientific), and total RNA was isolated using ISOGEN or ISOGEN-II 674 system (Nippon Gene). After assessing RNA yield and quality using a 2100 Bioanalyzer (Agilent Primary CMs were isolated from the ventricles of mice at aged stages (29-32 wks), as previously 695 described (Ujihara et al., 2019). Briefly, the heart was excised and a cannula was inserted into the aorta. 696 Coronary perfusion was initiated with cell-isolation buffer (CIB; 130 mM NaCl, 5.4 mM KCl, 0.5 mM 697 MgCl2, 0.33 mM NaH2PO4, 22 mM glucose, 50 nM/ml bovine insulin, and 25 HEPES-NaOH (pH 7.4)) 698 containing 0.4 mM EGTA. The perfusate was changed to the enzyme solution in CIB containing 0.3 699 mM CaCl2, 1 mg/ml collagenase type II (Worthington Biochemical, USA), 0.06 mg/ml protease (Sigma-700 Aldrich), and 0.06 mg/ml trypsin (Sigma-Aldrich). The left ventricles were cut into small pieces and 701 further digested in the enzyme solution for 10-15 min at 37ºC by gentle agitation. In this enzyme 702 solution, the CaCl2 level was increased to 0.7 mM, and 2 mg/ml BSA was supplemented. After 703 centrifugation at 14 × g for 5 min, the pellet was resuspended in CIB containing 1.2 mM CaCl2 and 2 704 mg/ml BSA, and then incubated for 10 min at 37ºC. After a further centrifugation, the cells were 705 resuspended in Tyrode's solution (140 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 0.5 mM MgCl2, 0.33 706 mM NaH2PO4, 11 mM glucose, 2.0 5 mM HEPES-NaOH (pH=7.4)) containing 2 mg/ml BSA. Photonics, Japan) mounted on the side port of an inverted microscope (IX73, Olympus) with a 20× 714 objective lens (UCplanFLN, Olympus). The Ca 2+ transients were measured by loading isolated CMs 715 with 5 μM Fura-2 AM (Dojindo, Japan) for 30 min, as previously described (Honda et al., 2018). The 716 Fura-2-loaded cells were alternately excited at 340 and 380 nm using an LED illuminator (pE-340 fura , 717 CoolLED). The Ca 2+ content of the sarcoplasmic reticulum (SR) was evaluated by rapidly applying 10 718 mM caffeine and measuring the resulting Ca 2+ transients in isolated CMs. Data were analyzed using 719 MetaMorph version 7.8.0.0 software (Molecular Devices, USA). 720 721 Primary CMs were isolated from the ventricles of fetal mice at embryonic day E17, essentially as 723 described previously (Hashimoto et al., 2018). Briefly, pregnant mice were euthanized with Sevofrane, 724 and fetal heart ventricles were rapidly excised, cut into small pieces, and digested four times with 0.06% 725 trypsin and 0.24 mM EDTA in PBS for 10 min at 37°C. After a 20 min culture to exclude non-CMs, the 726 CMs were plated onto fibronectin-coated culture vessels in DMEM with 5% FBS and cultured under 727 standard conditions at 37 °C with 5% CO2. In each isolation procedure, 5-10 fetal hearts were pooled 728 and used for the isolation.

Figure 4
WT TG    Representative genotyping results using genomic tail DNA from 3 WT and 3 TG mice. Positions for left (L) and right (R) primers were marked in A. C. Representative western blots (WB) of heart homogenates from 9 mice at F1 generation (6 wks of age) derived from 7 founder TG lines using anti-FLAG or anti-Fam64a antibody, both of which detect overexpressing Fam64a-FLAG fusion protein. Based on the amount of the protein detected, we classified 7 founder lines into 3 categories; TG-strong (2 lines, for example #1, #8, and #9), TG-medium (2 lines, for example #7), and TG-weak (3 lines, for example #2, #4, and #5). Descendants of TG-strong lines were used for subsequent experiments.

Kcna5
Figure 4-figure supplement 1. Genes for K + channel subunits were consistently repressed in Fam64a TG mice qPCR analysis of genes encoding several K + channel subunits in WT and TG mice hearts. Data were shown as normalized to WT. n = 5-7 mice per group. * p < 0.05, *** p < 0.001 as compared to WT by Student's two-tailed unpaired t-test. Error bar = SEM.

Kcnj11
Figure 4-figure supplement 1 ***p=1 x 10 -5 ***p=2 x 10 -7 ***p=0.0003 ***p=0.0006 *p=0.026 *p=0.012 The locomotor activity of mice was monitored using the infrared motion detector. Activity counts accumulated over a 1 h period were measured for a total of 4 days in a 12 h light:12 h dark cycle (lights on at 8 am). A: Representative tracings of the locomotor activity from WT (top) and TG (bottom) mice. During the nighttime (8 pm to 8 am), the most mice showed characteristic biphasic patterns of locomotor activity, i.e., the 1 st peak during the time period from 8 pm to 2 am, and the 2 nd peak during the time period from 2 am to 8 am (arrows). B: The 1 st and the 2 nd peak activity during nighttime in WT (filled bar) and TG (open bar) mice. C: Total activity counts during daytime (8 am to 8 pm) in WT and TG mice. Data pooled for 4 days were shown. Data were analyzed from 7 WT mice and 8 TG mice at adult stages (9-16 wks of age). *** p < 0.001 as compared to WT by Student's two-tailed unpaired t-test. Error bar = SEM.

. Comprehensive search for interacting partners of Fam64a
A: Immunoprecipitation (IP) against FLAG peptide that was expressed as a C-terminal tag of overexpressing Fam64a protein in TG mice hearts, followed by western blotting (WB) using FLAG and Fam64a antibody, which correctly detected Fam64a-FLAG fusion protein (*) in TG, but not in WT mice heart lysates, validating the immunoprecipitation procedure. B: The same IP/WB procedure was applied for Trim28, indicating that Fam64a forms complex with Trim28 in CMs.