Alternative Splicing of Pericentrin Contributes to Cell Cycle Control in Cardiomyocytes

Induction of cardiomyocyte proliferation is a promising option to regenerate the heart. Thus, it is important to elucidate mechanisms that contribute to the cell cycle arrest of mammalian cardiomyocytes. Here, we assessed the contribution of the pericentrin (Pcnt) S isoform to cell cycle arrest in postnatal cardiomyocytes. Immunofluorescence staining of Pcnt isoforms combined with SiRNA-mediated depletion indicates that Pcnt S preferentially localizes to the nuclear envelope, while the Pcnt B isoform is enriched at centrosomes. This is further supported by the localization of ectopically expressed FLAG-tagged Pcnt S and Pcnt B in postnatal cardiomyocytes. Analysis of centriole configuration upon Pcnt depletion revealed that Pcnt B but not Pcnt S is required for centriole cohesion. Importantly, ectopic expression of Pcnt S induced centriole splitting in a heterologous system, ARPE-19 cells, and was sufficient to impair DNA synthesis in C2C12 myoblasts. Moreover, Pcnt S depletion enhanced serum-induced cell cycle re-entry in postnatal cardiomyocytes. Analysis of mitosis, binucleation rate, and cell number suggests that Pcnt S depletion enhances serum-induced progression of postnatal cardiomyocytes through the cell cycle resulting in cell division. Collectively, our data indicate that alternative splicing of Pcnt contributes to the establishment of cardiomyocyte cell cycle arrest shortly after birth.


Introduction
Ischemic heart disease (IHD) is the leading cause of death worldwide [1]. Currently, IHD is treated by reopening occluded vessels (acute percutaneous coronary intervention) to reestablish perfusion and pharmacologically (e.g., beta blockers) to minimize cardiac remodeling and further deterioration of heart function. Except heart transplantation, which is limited due to a shortage in organ donors, there are no therapies available to effectively reverse heart injury. One promising approach to regenerate lost heart muscle tissue is the modulation of developmental pathways to induce endogenous proliferation of differentiated cardiomyocytes [2], which would also be beneficial for pediatric patients with congenital heart disease.
Mammalian cardiomyocytes stop proliferating shortly after birth. The cells progress through a last cell cycle, which results in polyploidization or binucleation and subsequently cell cycle arrest [3][4][5]. Whether this cell cycle arrest is permanent or can be reversed has been long debated. In 2005, it was demonstrated that adult mammalian cardiomyocytes can be induced to undergo cell division [6]. Subsequently, a number of studies have been published supporting the idea that cardiac regeneration can be achieved by induction of cardiomyocyte proliferation [7,8]. Yet, it has also been suggested that induction of postnatal cardiomyocyte cell division is associated with improper distribution of microtubules and

Isolation and Cell Culture of Postnatal Cardiomyocytes
The present investigation conforms to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. Extraction of organs and preparation of primary cell cultures were approved by the local Animal Ethics Committee in accordance with governmental and international guidelines on animal experimentation (protocol TS-9/2016 Nephropatho). Rat ventricular cardiomyocytes were isolated from Sprague Dawley rats on day 3 after birth (P3) as previously described [6], seeded on 1 mg/mL fibronectin (Sigma-Aldrich, St. Louis, MO, USA)-coated glass coverslips, and cultured in cardiomyocyte medium (DMEM-F12, Glutamax TM-I, 3 mM Na-pyruvate, 0.2% bovine serum albumin (BSA), 0.1 mM ascorbic acid, 0.5% Insulin-Transferrin-Selenium (100×, Life Technologies, Carlsbad, CA, USA), and penicillin/streptomycin (100 U/mg/mL)) containing 5% horse serum, if not stated otherwise.

SiRNA Knockdown
Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific, Waltham, MA, USA) was utilized to transfect cardiomyocytes 24 h post-seeding with an SiRNA targeting the Pcnt Sspecific 5 UTR (siPcntS, 50 nM or 100 nM as indicated, 5 -CAUAUGUUCUUGUAUAAAAtt-3 ) or a single SiRNA targeting both Pcnt B and Pcnt S (siPcntB + S, 200 nM, 5 -CAGGAACUC ACCAGAGACGAA-3 ). Control cells were mock-transfected. After transfection, cells were incubated for 72 h in medium containing 5% HS for testing depletion efficiency and effects on centriole cohesion. For cell cycle studies, cells were seeded in medium with 1% HS, transfected 24 h post-seeding, and subsequently cultured for 72 h in 10% FBS. SiRNAmediated depletion efficiency was assessed by determining the median intensity of total Pcnt at the nucleus in P3 cardiomyocytes. Pcnt signal from the centrosome was considered noise. The baseline was set as the median nuclear Pcnt signal of non-myocytes, which contains overlapping centriolar signal as well-and subtracted from all measurements. To determine the effect of siPcntS on Pcnt B expression, the maximum intensity of the Pcnt B signal per cardiomyocyte was measured as an approximation of the centrosomal Pcnt B signal.

Analysis of DNA Incorporation
To detect entry into S phase, 30 µM EdU was added to the medium for 24 h (cardiomyocytes) or 3 h (C2C12 cells) before fixation. EdU incorporation was determined utilizing the kit Click-iT ® EdU Alexa Fluor ® 488 (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol.

Image Preparation and Analysis
Images were arranged with Fiji [23], a distribution of ImageJ (Public Domain), custom Python scripts (Python Software Foundation, https://www.python.org/, accessed on 5 October 2020), the R package ggplot2 [24], and Adobe Illustrator (Adobe, San Jose, CA, USA). For image analysis, Fiji was used for preprocessing, custom Cellprofiler [25] pipelines for the extraction of features and measurements in an SQLite database, and the Cellprofiler Analyst [26] for manually labeling the training dataset. Random Forest classifier were trained in Python on the training set and performance was evaluated via 5-fold crossvalidation. The classifier was utilized to classify all cells automatically for the proceeding analysis and the classification results were written in the SQLite database. Centrioles were identified by visualizing the log normalized intensity values while classifying the images. All code, including Cellprofiler pipelines for image analysis, classification, and analysis/figure scripts, have been deposited in a public repository (https://osf.io/nckw3/, accessed 15 June 2021).

Statistical Analysis
Image intensities were normalized per experiment. Binucleated cardiomyocytes were excluded from the quantification of centriole cohesion, Ki67 ratio, EdU ratio, and cardiomyocyte count, to control for false positive proliferation. We consider one distinct primary cardiomyocyte isolation as one biological replicate. Per isolation, hearts from typically 10 to 20 pups were obtained and pooled. H0 was defined as the group distributions being equal, and H1 was defined as them being different. Statistical analysis was performed using R 3.6.1 [27] and reading the measurements and classifications from an SQLite database. Data of at least three biological independent experiments with at least 10 technical replicates each were either expressed as a bar plot with mean ± SD or a box and whiskers plot. To test for differences in the mean between groups and to take into account the biological and technical replications, we used a nested ANOVA design by fitting a linear mixed effects models with restricted maximum likelihood using the R package nlme [28]. Bonferroni-corrected p values were obtained by comparing the linear hypothesis of the mixed effects model. p values were reported as significant, when they were lower than the alpha level of 0.05 and the figures were annotated accordingly with n.s.: p > 0.05, *: p < 0.05, **: p < 0.01, ***: p < 0.001. Additional Bayesian statistical analysis was performed by using the BayesFactor package [29]. To compute the posterior distribution of the group wise treatment effect, we used Marcov Chain Monte Carlo sampling and a noninformative Jeffreys prior for the variance and a standard Cauchy prior for the mean.

Pcnt B Preferentially Localizes at the Centrosome and Pcnt S at the Nuclear Envelope
Previously, it was reported that "Pericentrin S, and not Pericentrin B, is the predominant Pericentrin isoform at the nuclear envelope in P3-isolated cardiomyocytes" [17]. This conclusion was drawn based on an antibody specific for Pcnt B and an antibody detecting both Pcnt B and S. To specifically address the location and function of Pcnt S, we designed an SiRNA specific for the 5 UTR of Pcnt S (siPcntS) and a single SiRNA targeting both Pcnt B and Pcnt S (siPcntB + S targeting the C-terminal part of Pcnt) (Supplementary Figure S1A). Knockdown efficiency in ventricular cardiomyocytes isolated from threeday-old rats (P3 cardiomyocytes) was validated by semi-quantitative immunofluorescence analysis of median intensity of nuclear Pcnt B + S signal ( Figure 1A,B) and the Pcnt B signal at the centrosome ( Figure 1C,D). Only cells positive to troponin I and with PCM1 localization at the nuclear envelope were considered cardiomyocytes and included in quantitative analyses.
The nuclear signal of Pcnt contained some noise from overlapping centriolar signals. To overcome this noise, the median Pcnt signal from the nucleus of non-myocytes, which do not have Pcnt at the nuclear envelope, was subtracted from all measurements. In three independent experiments, a total of 7325 cardiomyocytes (troponin I-positive) were analyzed. In control experiments, we detected for both Pcnt B + S as well as Pcnt B a positive signal at the nuclear envelope. Upon SiRNA-mediated knockdown of Pcnt S or Pcnt B + S, a significant reduction in nuclear Pcnt B + S signal was observed in all treatment groups (50 nM siPcntS: 56.8%, 100 nM siPcntS: 47.4%, 200 nM of siPcntB + S: 18.2%, ANOVA: p < 0.001, Figure 1A,B). Notably, depletion of Pcnt S had no marked effect on the weak but constant Pcnt B expression at the nuclear envelope ( Figure 1C). The specificity of the SiRNA against Pcnt S was confirmed by RT-PCR (Supplementary Figure S1B,C). These data indicate that Pcnt B is, like Pcnt S, present at the nuclear envelope.
To determine whether siPcntB + S is sufficient to deplete Pcnt B from the centrosome, we compared the maximum intensity of the Pcnt B signal per cardiomyocyte as an approximation of the centrosomal Pcnt B signal, as Pcnt B at the centrosome represents by far the strongest signal in a cell. siPcntB + S-mediated depletion of both Pcnt isoforms resulted in a clear reduction in the maximum Pcnt intensity to 65.7% (p < 0.001). In contrast, siPcntS-mediated depletion resulted only in a slight reduction to 94.5% (50 nM, p < 0.05) or 91.2% (100 nM, p < 0.05), respectively ( Figure 1C,D). Similarly, siPcntS had no obvious effect on the Pcnt B + S signal at the centrosomes ( Figure 1C, lowest row). These data suggest that if at all, only small amounts of Pcnt S are localized at the centrosome, and that Pcnt S depletion does not affect the expression of Pcnt B at the centrosome. resulted in a clear reduction in the maximum Pcnt intensity to 65.7% (p < 0.001). In contrast, siPcntS-mediated depletion resulted only in a slight reduction to 94.5% (50 nM, p < 0.05) or 91.2% (100 nM, p < 0.05), respectively ( Figure 1C,D). Similarly, siPcntS had no obvious effect on the Pcnt B + S signal at the centrosomes ( Figure 1C, lowest row). These data suggest that if at all, only small amounts of Pcnt S are localized at the centrosome, and that Pcnt S depletion does not affect the expression of Pcnt B at the centrosome.
Taken together, the SiRNA-related data indicate that siPcntS efficiently depletes Pcnt S and does not interfere with Pcnt B localization. In contrast, siPcntB + S is not efficient in depleting centrosomal Pcnt, even though it resulted in a notable reduction in centrosomal Pcnt.  Taken together, the SiRNA-related data indicate that siPcntS efficiently depletes Pcnt S and does not interfere with Pcnt B localization. In contrast, siPcntB + S is not efficient in depleting centrosomal Pcnt, even though it resulted in a notable reduction in centrosomal Pcnt.
To validate our initial conclusions that Pcnt S and Pcnt B can localize to the centrosome as well as the nuclear envelope, FLAG-tagged Pcnt B or Pcnt S were ectopically expressed in P3 cardiomyocytes. Anti-FLAG staining revealed that Pcnt B-FLAG and Pcnt S-FLAG localized to the centrioles/centrosome (γ-tubulin-positive) as well as the nuclear envelope ( Figure 2). Notably, the signal for Pcnt S-FLAG was more pronounced at the nuclear envelope than that for Pcnt B-FLAG.
To validate our initial conclusions that Pcnt S and Pcnt B can localize to the centrosome as well as the nuclear envelope, FLAG-tagged Pcnt B or Pcnt S were ectopically expressed in P3 cardiomyocytes. Anti-FLAG staining revealed that Pcnt B-FLAG and Pcnt S-FLAG localized to the centrioles/centrosome (γ-tubulin-positive) as well as the nuclear envelope ( Figure 2). Notably, the signal for Pcnt S-FLAG was more pronounced at the nuclear envelope than that for Pcnt B-FLAG. Collectively, our data indicate that in postnatal cardiomyocytes, the dominant Pcnt isoform at the nuclear envelope is Pcnt S and at the centrosome Pcnt B. Yet, our data suggest that both isoforms can localize to the nuclear envelope as well as the centrosome.

Pcnt S Is Not Required for Centriole Cohesion
It is well known that Pcnt is required for centriole cohesion [30,31]. Yet, the role of Pcnt S in this process is unclear. Notably, loss of centriole cohesion in cardiomyocytes and upregulation of Pcnt S occur at around the same time in development [17]. To determine whether depletion of Pcnt S affects centriole cohesion, P3 cardiomyocytes were transfected with the indicated SiRNAs, stained for Pcnt B + S, and analyzed for centriole configuration (paired or split) whereby centrioles were considered split if the distance between them exceeded 2 µm ( Figure 3A). Since neither siPcntS nor siPcntB + S completely eliminate the Pcnt signal from the centrosome (Figure 1), we utilized Pcnt B + S staining to distinguish between paired and split centrioles ( Figure 3A). Analysis of three independent experiments and a total of 3725 cardiomyocytes revealed that Pcnt S depletion had no significant effect on centriole splitting compared to control (mock transfected) cells (paired centrioles: control: 10.5%, 50 nM siPcntS: 12.1%, 100 nM siPcntS: 10.9%, Figure 3B,C). In contrast, depletion of both isoforms significantly decreased the number of cardiomyocytes with paired centrioles (200 nM siPcntB + S: 5.0%, p < 0.01, Figure 3B,C), These data are in agreement with the known role of Pcnt B in centriole cohesion and suggest that Pcnt S is not required for centriole cohesion. However, splitting of centrioles has already occurred in P3 cardiomyocytes when Pcnt S is depleted, and thus, we cannot rule out the possibility that Pcnt S induces the splitting of centrioles. Collectively, our data indicate that in postnatal cardiomyocytes, the dominant Pcnt isoform at the nuclear envelope is Pcnt S and at the centrosome Pcnt B. Yet, our data suggest that both isoforms can localize to the nuclear envelope as well as the centrosome.

Pcnt S Is Not Required for Centriole Cohesion
It is well known that Pcnt is required for centriole cohesion [30,31]. Yet, the role of Pcnt S in this process is unclear. Notably, loss of centriole cohesion in cardiomyocytes and upregulation of Pcnt S occur at around the same time in development [17]. To determine whether depletion of Pcnt S affects centriole cohesion, P3 cardiomyocytes were transfected with the indicated SiRNAs, stained for Pcnt B + S, and analyzed for centriole configuration (paired or split) whereby centrioles were considered split if the distance between them exceeded 2 µm ( Figure 3A). Since neither siPcntS nor siPcntB + S completely eliminate the Pcnt signal from the centrosome (Figure 1), we utilized Pcnt B + S staining to distinguish between paired and split centrioles ( Figure 3A). Analysis of three independent experiments and a total of 3725 cardiomyocytes revealed that Pcnt S depletion had no significant effect on centriole splitting compared to control (mock transfected) cells (paired centrioles: control: 10.5%, 50 nM siPcntS: 12.1%, 100 nM siPcntS: 10.9%, Figure 3B,C). In contrast, depletion of both isoforms significantly decreased the number of cardiomyocytes with paired centrioles (200 nM siPcntB + S: 5.0%, p < 0.01, Figure 3B,C), These data are in agreement with the known role of Pcnt B in centriole cohesion and suggest that Pcnt S is not required for centriole cohesion. However, splitting of centrioles has already occurred in P3 cardiomyocytes when Pcnt S is depleted, and thus, we cannot rule out the possibility that Pcnt S induces the splitting of centrioles.

Ectopic Expression of Pcnt S Induces Centriole Splitting and Centriolar γ-Tubulin Reduction
Expression of Pcnt S is specific to muscle tissue (cardiac and skeletal) (Supplementary Figure S2A,B). To determine whether Pcnt S is sufficient to induce centriole splitting, we expressed Pcnt S-T2A-eGFP or Pcnt B-T2A-eGFP and analyzed centriole configuration in adult retinal pigment epithelial cells (ARPE-19), which do not exhibit centriole splitting ( Figure 4A). The introduction of the T2A sequence should result in the expression of the Pcnt isoforms and GFP as separate proteins. Consequently, GFP localized in Pcnt B-T2A-eGFP-transfected cells in the cytoplasm (Supplementary Figure S3A). Yet, in PCNT S-T2A-eGFP-transfected cells, GFP accumulated in a centrosome-like manner, indicating that the T2A sequence is not functional in PCNT S-T2A-eGFP (sequence has been verified by Sanger sequencing) (Supplementary Figure S3A). Immunostaining of Pcnt in Pcnt B-T2A-eGFP-and PCNT S-T2A-eGFP-transfected cells demonstrated that in both cases, Pcnt is overexpressed (Supplementary Figure S3A). In order to determine whether any of the two Pcnt isoforms induces centriole splitting, centriolar γ-tubulin signal was analyzed ( Figure 4A). As previously introduced, cells were classified by their centriole status and centrioles were considered split if the distance between them exceeded 2 µm. Analysis of 2469 cells from three independent experiments showed increased centriole splitting after overexpression of both Pcnt isoforms ( Figure 4B,C). While overexpression of Pcnt B-T2A-eGFP increased centriole splitting in the GFP + over the GFP − cells from 6.3% to 10.8% (p = 0.055), the difference was much more pronounced after ectopic expression of Pcnt S-T2A-eGFP with 5.0% to 15.8% (p > 0.001). Thus, ectopic expression of Pcnt S-T2A-eGFP induced ~46% more centriole splitting than overexpression of Pcnt B-T2A-eGFP.

Ectopic Expression of Pcnt S Induces Centriole Splitting and Centriolar γ-Tubulin Reduction
Expression of Pcnt S is specific to muscle tissue (cardiac and skeletal) (Supplementary Figure S2A,B). To determine whether Pcnt S is sufficient to induce centriole splitting, we expressed Pcnt S-T2A-eGFP or Pcnt B-T2A-eGFP and analyzed centriole configuration in adult retinal pigment epithelial cells (ARPE-19), which do not exhibit centriole splitting ( Figure 4A). The introduction of the T2A sequence should result in the expression of the Pcnt isoforms and GFP as separate proteins. Consequently, GFP localized in Pcnt B-T2A-eGFP-transfected cells in the cytoplasm (Supplementary Figure S3A). Yet, in PCNT S-T2A-eGFP-transfected cells, GFP accumulated in a centrosome-like manner, indicating that the T2A sequence is not functional in PCNT S-T2A-eGFP (sequence has been verified by Sanger sequencing) (Supplementary Figure S3A). Immunostaining of Pcnt in Pcnt B-T2A-eGFP-and PCNT S-T2A-eGFP-transfected cells demonstrated that in both cases, Pcnt is overexpressed (Supplementary Figure S3A). In order to determine whether any of the two Pcnt isoforms induces centriole splitting, centriolar γ-tubulin signal was analyzed ( Figure 4A). As previously introduced, cells were classified by their centriole status and centrioles were considered split if the distance between them exceeded 2 µm. Analysis of 2469 cells from three independent experiments showed increased centriole splitting after overexpression of both Pcnt isoforms ( Figure 4B,C). While overexpression of Pcnt B-T2A-eGFP increased centriole splitting in the GFP + over the GFP − cells from 6.3% to 10.8% (p = 0.055), the difference was much more pronounced after ectopic expression of Pcnt S-T2A-eGFP with 5.0% to 15.8% (p > 0.001). Thus, ectopic expression of Pcnt S-T2A-eGFP induced~46% more centriole splitting than overexpression of Pcnt B-T2A-eGFP.
While examining centriole splitting, we noticed that γ-tubulin signal appeared to be reduced in Pcnt S-T2A-eGFP-transfected ARPE19 cells ( Figure 4A). Semi-quantitative analysis confirmed that centriolar γ-tubulin signal intensity in GFP + cells was notably lower compared to the GFP − cells (100% to 76.7%, p < 0.001) in cultures transfected with Pcnt S-T2A-eGFP ( Figure 4D,E). In contrast, γ-tubulin intensity was only mildly reduced in GFP + cells after overexpressing Pcnt B-T2A-eGFP from 100% to 92.5% (p = 0.73). This indicates that ectopic Pcnt S expression interferes with γ-tubulin localization to centrioles/centrosomes. Taken together, our data suggest that ectopic expression of Pcnt S can induce centriole splitting in ARPE-19 cells. This further indicates that the developmental upregulation of Pcnt S might contribute to the postnatal cell cycle arrest in mammalian cardiomyocytes. While examining centriole splitting, we noticed that γ-tubulin signal appeared to be reduced in Pcnt S-T2A-eGFP-transfected ARPE19 cells ( Figure 4A). Semi-quantitative analysis confirmed that centriolar γ-tubulin signal intensity in GFP + cells was notably lower compared to the GFP − cells (100% to 76.7%, p < 0.001) in cultures transfected with Pcnt S-T2A-eGFP ( Figure 4D,E). In contrast, γ-tubulin intensity was only mildly reduced in GFP + cells after overexpressing Pcnt B-T2A-eGFP from 100% to 92.5% (p = 0.73). This indicates that ectopic Pcnt S expression interferes with γ-tubulin localization to centrioles/centrosomes. Taken together, our data suggest that ectopic expression of Pcnt S can induce centriole splitting in ARPE-19 cells. This further indicates that the developmental upregulation of Pcnt S might contribute to the postnatal cell cycle arrest in mammalian cardiomyocytes.

Ectopic Pcnt S Expression Impairs DNA Synthesis
Considering that Pcnt S is upregulated when cardiomyocyte cell cycle progression is arrested, we wondered whether Pcnt S overexpression inhibits cell cycle progression. As cardiomyocyte proliferation cannot be maintained in fetal cardiomyocytes and postnatal

Ectopic Pcnt S Expression Impairs DNA Synthesis
Considering that Pcnt S is upregulated when cardiomyocyte cell cycle progression is arrested, we wondered whether Pcnt S overexpression inhibits cell cycle progression. As cardiomyocyte proliferation cannot be maintained in fetal cardiomyocytes and postnatal cardiomyocytes quickly establish a nuclear MTOC and cell cycle arrest [17,32], we overexpressed Pcnt S in proliferating C2C12 myoblasts. C2C12 myoblasts, like cardiomyocytes, establish a nuclear MTOC during differentiation [33]. In addition, Pcnt S, while not expressed in the myoblast stage, is upregulated during differentiation (Supplementary Figure S2C) [15,16]. C2C12 cells were transfected with plasmids encoding Pcnt S-T2A-eGFP, Pcnt B-T2A-eGFP, or eGFP and cell cycle activity was analyzed based on nucleotide analog incorporation (5-ethynyl-2 -deoxyuridin (EdU)). Similarly to the experiment in ARPE-19 cells, we performed immunostaining of Pcnt in Pcnt B-T2A-eGFP-and PCNT S-T2A-eGFPtransfected C2C12 to confirm overexpression of Pcnt isoforms (Supplementary Figure S3B). The analysis of over 27,000 GFP-positive myoblasts from three independent experiments revealed that 33.2% of control-transfected myoblasts incorporated EdU ( Figure 5A-C).
Taken together, these results suggest that Pcnt S and thus alternative splicing of Pcnt might contribute to the establishment of cell cycle arrest in cardiomyocytes.  Taken together, these results suggest that Pcnt S and thus alternative splicing of Pcnt might contribute to the establishment of cell cycle arrest in cardiomyocytes.

Pcnt S Depletion Enhances Serum-Induced Cell Cycle Re-Entry in Cardiomyocytes
In order to determine whether depletion of Pcnt S increases the potential of cardiomyocytes to proliferate, P3 cardiomyocytes were stimulated with 10% FBS upon Pcnt S or Pcnt B + S depletion and progression into different cell cycle phases was assessed. Analysis of Ki67, a widely used marker to estimate the number of proliferating cells [34], revealed that SiRNA-mediated depletion of Pcnt S significantly increased the number of Ki67-positive cardiomyocytes (control: 16.2%, 50 nM siPcntS: 26.4%, p < 0.001, 100 nM siPcntS: 28.2%, p < 0.001, 200 nM siPcntB + S: 26.2%, p < 0.001, Figure 6A-C). We analyzed 28,496 cardiomyocytes from three independent experiments. cardiomyocytes to proliferate, P3 cardiomyocytes were stimulated with 10% FBS upon Pcnt S or Pcnt B + S depletion and progression into different cell cycle phases was assessed. Analysis of Ki67, a widely used marker to estimate the number of proliferating cells [34], revealed that SiRNA-mediated depletion of Pcnt S significantly increased the number of Ki67-positive cardiomyocytes (control: 16.2%, 50 nM siPcntS: 26.4%, p < 0.001, 100 nM siPcntS: 28.2%, p < 0.001, 200 nM siPcntB + S: 26.2%, p < 0.001, Figure 6A-C). We analyzed 28,496 cardiomyocytes from three independent experiments. As it has previously been indicated that Ki67 is not a very reliable marker to determine cardiomyocyte proliferation [11,35], we next determined the effect of Pcnt S depletion on 10% FBS-induced S phase entry by assessing EdU incorporation. The analysis of 56,081 cardiomyocytes from three independent experiments showed that SiRNAmediated depletion of Pcnt S significantly increased the number of EdU-positive cardiomyocytes (troponin I-positive) (control: 19.9%, 50 nM siPcntS: 34.4%, p < 0.001, 100 nM siPcntS: 40.3%, p < 0.001, 200 nM siPcntB + S: 24.9%, p < 0.01, Figure 6D-F). Taken together, these data indicate that Pcnt S depletion enhances serum-induced cell cycle reentry in cardiomyocytes into S phase. As it has previously been indicated that Ki67 is not a very reliable marker to determine cardiomyocyte proliferation [11,35], we next determined the effect of Pcnt S depletion on 10% FBS-induced S phase entry by assessing EdU incorporation. The analysis of 56,081 cardiomyocytes from three independent experiments showed that SiRNA-mediated depletion of Pcnt S significantly increased the number of EdU-positive cardiomyocytes (troponin I-positive) (control: 19.9%, 50 nM siPcntS: 34.4%, p < 0.001, 100 nM siPcntS: 40.3%, p < 0.001, 200 nM siPcntB + S: 24.9%, p < 0.01, Figure 6D-F). Taken together, these data indicate that Pcnt S depletion enhances serum-induced cell cycle re-entry in cardiomyocytes into S phase.

Discussion
We conclude that expression of Pcnt S contributes to the cell cycle arrest in postnatal mammalian ventricular cardiomyocytes and Pcnt S depletion promotes cardiomyocyte proliferation. Several lines of evidence support these conclusions. First, ectopic expression of Pcnt S increases centriole splitting in ARPE-19 cells and inhibits DNA synthesis in C2C12 myoblasts. Second, the depletion of Pcnt S enhances serum-induced cell cycle activity, facilitates DNA synthesis, and promotes mitotic entry as well as anaphase transition in cardiomyocytes and results in a higher density of mononucleated cardiomyocytes.
Pcnt B is an anchoring protein of the MTOC that is located via its PACT domain at the centrosome [30,31]. In contrast, Pcnt S is found predominantly at the nuclear envelope [17]. The difference in localization is surprising, as both isoforms contain the C-terminal PACT domain [36]. Why Pcnt S is found predominantly at the nuclear envelope is unknown. However, it has been shown that "the partially translated PCNT nascent polypeptide starts to interact with the dynein motor complex once the dynein light intermediate chain 1 (LIC1)-interacting domain in the N-terminal half of PCNT is synthesized and folded" [37]. Pcnt S lacks this N-terminal LIC1 domain. Consequently, Pcnt B is the predominant Pcnt isoform at the centrosome due to an active transport mechanism enriching Pcnt B at the centrosome. This would explain our finding that both isoforms can bind to both centrosome and nuclear envelope (via their PACT domain) with different efficiencies due to an active transport of Pcnt B to the centrosome. As the primary structure and protein domains of Pcnt S are shared with the C-terminal part of Pcnt B, a transport mechanism to specifically enhance Pcnt S at the nuclear envelope is unlikely (Supplementary Figure S1). Yet, it might be that the different 5 UTR is utilized to transport the mRNA to the nuclear envelope followed by local translation.
The cell cycle arrest in cardiomyocytes might be a requirement for Pcnt S expression or might be the result of Pcnt S expression. Based on our data that ectopic Pcnt S expression induces centriole splitting and impairs cell cycle progression, whereas Pcnt S depletion enhances serum-induced cell cycle progression, it appears that the cell cycle arrest in cardiomyocytes is a consequence of Pcnt S expression.
Yet, how Pcnt S affects cell cycle progression is unclear. Loss of centriole cohesion has been proposed to promote cardiomyocyte cell cycle exit and is associated with the upregulation of Pcnt S [17]. Importantly, ectopic expression of Pcnt S resulted in ARPE-19 cells in centriole splitting, further suggesting that upregulation of Pcnt S during cardiomyocyte differentiation actively induces cell cycle arrest by inactivating the centrosomal MTOC. Finally, our data indicate that depletion of Pcnt S enhances the effect of factors stimulating cardiomyocyte proliferation. Notably, depletion of total Pcnt limited this stimulating effect. The most likely explanation of this is that depletion of total Pcnt induces a loss of centrosome integrity (Figure 3) [30,31,38], which is known to result in a G1/S arrest [38]. This would be in agreement with a reduced number of EdU-positive cells (G1/S arrest) and no change in the number of Ki67-positive cells (cells are still in the cell cycle). It is well known that serum stimulation induces in postnatal cardiomyocytes binucleation instead of cell division. Yet, here we show that serum stimulation upon depletion of Pcnt S resulted in increased cardiomyocyte cell division. Thus, it will be of interest to test in the future whether Pcnt S depletion can improve current strategies to regenerate the heart. Pcnt S is upregulated during the time in which cardiomyocytes reorganize their cytoskeleton, potentially to meet increased functional demands. Thus, increased expression of Pcnt S might be a requirement for improved muscle function or a consequence of enhanced muscle function. While there is no obvious reason why Pcnt S should have a different function than Pcnt B, there are arguments for the need for a nuclear envelope MTOC which is facilitated by Pcnt S expression. An increasing work load of the heart during development requires an increasing contractile force generation by cardiomyocytes, which causes more force on the nucleus. Thus, a nuclear envelope MTOC might be required to generate a microtubule-based cage around the nucleus to protect it from intracellular shear and compression forces to prevent DNA damage and maintain genome organization for a proper transcriptional program. On the other hand, the postnatal heart needs to respond to environmental changes (e.g., increased work load upon exercise) and this might require the rearrangement of the cytoskeleton. Similar is true for skeletal muscle. In addition, cardiomyocytes increase significantly in size during postnatal development and the formation of microtubules from a point source to a larger central radial source might be required to ensure the highly intracellular organization (e.g., sarcomeres, mitochondria, and t-tubules, invaginations of the muscle cell membrane) as well as efficient microtubulebased intracellular transport [39]. Notably, it has recently been demonstrated that the nuclear MTOC is required for cardiomyocyte hypertrophy [19]. In the future, it will be important to determine whether Pcnt S has any specific role at the nuclear MTOC and to establish systems in which the nuclear envelope MTOC can be modulated and cellular organization and function can be assessed.

Conclusions
Taken together, this study suggests that the expression of Pcnt S contributes to cell cycle arrest in postnatal cardiomyocytes which, in turn, promotes a post-mitotic state. Given the increasing interest in cardiac regeneration and the role of non-centrosomal MTOCs in cell differentiation and function, understanding the establishment and function of non-centrosomal MTOCs and cell cycle arrest during development may reveal new mechanisms to regulate cell proliferation and function, with implications not only for the treatment of striated muscle disease but in general for regeneration and cancer.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/jcdd8080087/s1. Figure S1: Overview of alternative splicing of PCNT and challenges for specific depletion of Pcnt S, and expression of Pcnt S and Pcnt B by RT-PCR. Figure S2: Pcnt B and Pcnt S expression in different tissue types. Figure S3: Verification of Pcnt expression after ectopic expression in ARPE-19 cells and C2C12 myoblasts. Figure S4: Representative examples of different cell cycle stages of cardiomyocytes.