Protective effects of Gαi3 deficiency in a murine heart-failure model of β1-adrenoceptor overexpression

We have shown that in murine cardiomyopathy caused by overexpression of the β1-adrenoceptor, Gαi2-deficiency is detrimental. Given the growing evidence for isoform-specific Gαi-functions, we now examined the consequences of Gαi3 deficiency in the same heart-failure model. Mice overexpressing cardiac β1-adrenoceptors with (β1-tg) or without Gαi3-expression (β1-tg/Gαi3−/−) were compared to C57BL/6 wildtypes and global Gαi3-knockouts (Gαi3−/−). The life span of β1-tg mice was significantly shortened but improved when Gαi3 was lacking (95% CI: 592–655 vs. 644–747 days). At 300 days of age, left-ventricular function and survival rate were similar in all groups. At 550 days of age, β1-tg but not β1-tg/Gαi3−/− mice displayed impaired ejection fraction (35 ± 18% vs. 52 ± 16%) compared to wildtype (59 ± 4%) and Gαi3−/− mice (60 ± 5%). Diastolic dysfunction of β1-tg mice was prevented by Gαi3 deficiency, too. The increase of ANP mRNA levels and ventricular fibrosis observed in β1-tg hearts was significantly attenuated in β1-tg/Gαi3−/− mice. Transcript levels of phospholamban, ryanodine receptor 2, and cardiac troponin I were similar in all groups. However, Western blots and phospho-proteomic analyses showed that in β1-tg, but not β1-tg/Gαi3−/− ventricles, phospholamban protein was reduced while its phosphorylation increased. Here, we show that in mice overexpressing the cardiac β1-adrenoceptor, Gαi3 deficiency slows or even prevents cardiomyopathy and increases shortened life span. Previously, we found Gαi2 deficiency to aggravate cardiac dysfunction and mortality in the same heart-failure model. Our findings indicate isoform-specific interventions into Gi-dependent signaling to be promising cardio-protective strategies. Supplementary information The online version contains supplementary material available at 10.1007/s00210-023-02751-8.


Introduction
Heart failure is a major cause of cardiovascular diseases affecting at least 26 million people worldwide (Savarese and Lund 2017).β-Adrenoceptor antagonists are a cornerstone in the therapy of chronic heart failure because some have been proven to reduce mortality independent of age and gender of the patients (Kotecha et al. 2016(Kotecha et al. , 2017)).Guarding the heart from (excessive) β-adrenergic stimulation seems to be cardio-protective mainly by preventing G s -protein mediated signalling (Baker 2014).G s -proteins are the cognate interaction partners of β 1 -adrenoceptors (Xiao et al. 1999;Seyedabadi et al. 2019).Overexpression of β 1 -adrenoceptors (β 1 -AR) in murine hearts has been shown to cause dilative cardiomyopathy leading to severe heart failure (Engelhardt et al. 1999(Engelhardt et al. , 2001a)).Although cardiac overexpression of β 2 -adrenoceptors (β 2 -AR) also leads to cardiac failure, a significantly higher level of overexpression is required (Liggett et al. 2000).β 2 -Adrenoceptors couple to both G s and G i proteins (Xiao et al. 1999).G i proteins are thought to be involved in the protection against excessive β-adrenergic stimulation in heart failure (Brown and Harding 1992;El-Armouche et al. 2003), and G i -protein-mediated signaling downstream from β 2 -AR has been shown to be anti-apoptotic (Chesley et al. 2000).Despite the differences between β-adrenoceptor isoforms regarding G-protein coupling, it should be considered that G i proteins seem to modulate both β 1 -and β 2 -adrenergic signalling (Li et al. 2004;Martin et al. 2004;Melsom et al. 2014).It has to be mentioned that not all studies support the idea of G i proteins mediating the cardio-protective effects of β 2 -adrenoceptor stimulation (Xiao et al. 2003;Ahmet et al. 2005) or of G i -protein signaling being cardioprotective in general (Hussain et al. 2013).At least two Gα i -isoforms, Gα i2 and Gα i3 , are expressed in the cardiovascular system, which have been shown to interplay (Thompson et al. 2007) and to exhibit redundant but also distinct functions (Gohla et al. 2007;Dizayee et al. 2011;Plummer et al. 2012;Wiege et al. 2012Wiege et al. , 2013;;Köhler et al. 2014;Wang et al. 2014;Devanathan et al. 2015;Mauriac et al. 2017;Beer-Hammer et al. 2018).Of particular interest, in a murine ischemia-reperfusion model, Köhler et al. showed lack of Gα i2 to worsen cardiac damage while lack of Gα i3 was beneficial (Köhler et al. 2014).Thus, the increased Gα i2 -expression observed in failing myocardium might be interpreted as compensatory, while the role of Gα i3 remains unclear (Eschenhagen et al. 1992b;Kompa et al. 1999).
In a previous study, we reported that lack of Gα i2 (Gα i2 −/− ) had detrimental effects in β 1 -transgenic (β 1 -tg) mice (Keller et al. 2015): survival of β 1 -tg/Gα i2 −/− mice was drastically shortened, and these animals showed a significantly impaired cardiac function.This occurred already at an age of about 300 days, i.e., when β 1 -tg or Gα i2 −/− mice were unaffected in this regard.Considering the unknown consequences of functional isoform redundancy between the closely related Gα i2 and Gα i3 proteins on the one hand and isoform-specific, distinct functions on the other hand, we now examined the impact of Gα i3 deficiency on cardiac function of β 1 -tg mice.In particular, we asked whether the lack of Gα i3 impairs heart function of β 1 -tg mice, is not detrimental, or may even rescue from β 1 -AR-induced cardiomyopathy.
We find Gα i3 deficiency to be cardio-protective in terms of slowing down or even preventing the development of β 1 -AR-induced cardiomyopathy.Together with previous findings, our study indicates isoform-specific targeting of Gα iprotein-mediated signaling to be a promising novel strategy to treat cardiovascular diseases.Parts of the data have already been published as a conference abstract (Schröper et al. 2020).

Mouse models
Mice with cardiac overexpression of the human β 1 -AR (β 1tg) have been described earlier (Engelhardt et al. 1999).We had backcrossed these FVB/N-based transgenic mice to a C57BL/6 J background (Keller et al. 2015).In the current study, β 1 -tg mice were crossbred with mice globally lacking Gα i3 (Gohla et al. 2007), to produce β 1 -tg Gα i3 -deficient mice (β 1 -tg/Gα i3 −/− ).Age-matched wildtype and Gα i3deficient (Gα i3 −/− ) littermates served as controls.Animals of both sexes were used for our study (sex distribution given in table S1).We kept mice in individually ventilated cages with a 12 h/12 h dark/light cycle and food and water ad libitum.For genotyping, tail or ear clips from 3-week-old mice were processed.Genomic DNA was prepared and genotyping PCR for Gα i3 and the β 1 -AR was performed as described previously (Dizayee et al. 2011;Keller et al. 2015).Animals were killed by cervical dislocation.Since, in a previous study, cardiac β 1 -AR overexpression on a C57BL/6 J background by itself had no effect on cardiac function or survival at the age of 300 days (Keller et al. 2015), we chose a second target age to address putative effects of Gα i3 deficiency in β 1 -tg mice.Based on our own data and the report of another group, we thus additionally analyzed animals at the age of 550 days (Lee et al. 2015;Keller et al. 2015).The responsible federal state authority approved animal breeding, maintenance and experiments (Landesamt fuer Natur-, Umwelt-und Verbraucherschutz Nordrhein-Westfalen;references: 84-02.05.20.12.294, 84-02.05.20.13.060, and 84-02.04.2016.A422).All animal experiments complied with the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

Ventricle-to-body-weight ratio
Non-fasting mice were weighed directly before being killed.Immediately after cervical dislocation, we removed the heart, cut the atria and eliminated remaining intraventricular blood.We analyzed mice at an age of 304 ± 7 days and at the second target age of 553 ± 6 days, including mice just examined by echocardiography.

Histology and histomorphometrical analysis of fibrotic area
Only mice at the advanced age (553 ± 3 days) were used for this analysis.Cryo-Sects (6 μm thickness) were obtained from excised hearts frozen in liquid nitrogen, fixed in ice-cold acetone, subsequently immersed in Roti®-Histol for 10 min at room temperature, and transferred to water through descending concentrations of ethanol (100%, 96%, 75%).Staining was performed using a 0.1% solution of Sirius Red F3BA in saturated aqueous solution of picric acid for 45 min at 25 °C.Subsequently, slices were rinsed in 1% acetic acid for 2 min.Sections were dehydrated in ascending concentrations of ethanol (75%, 96%, and 100%, each 1 min) and cleared in two stages in Roti®-Histol, 10 min each.Sections were covered with Roti®-Histokitt mounting medium (Carl Roth, Karlsruhe, Germany) and a glass cover slip.After scanning the Picro Sirius Red sections with the Keyence BZ-9000E microscope, images were taken at mid-ventricular level (× 20 magnification), and interstitial fibrosis was quantified as percentage of total tissue area in the field of view.Planimetry was performed using a Keyence BZ2-Analyser software using hybrid cell count algorithm (Keyence, Osaka, Japan).

Echocardiography
Echocardiography was performed using the high-frequency VisualSonics Vevo® 3100 Imaging System (Fujifilm) with a MX550D transducer (22-55 MHz; axial resolution: 40 µm).Mice were prepared and examined under light inhalation anesthesia with oxygen and 1.5% isoflurane through a nose cap.Chest and upper abdominal hair was shaved, and the mice were placed on a warmed platform to maintain physiological conditions.We monitored ECG, heart rate, core temperature, and respiratory frequency.Systolic parameters were obtained by using the B-and M-Mode in parasternal long and short axis views of the left ventricle.Doppler flow profiles were acquired to estimate the isovolumic relaxation time (IVRT), an indicator of diastolic ventricular function.
We evaluated the ultrasound imaging data by working with the software Vevo LAB (Fujifilm).Strain analyses via Speckle Tracking were performed by using the Vevo Strain Software (Fujifilm).Younger mice used for echocardiographic investigation were 303 ± 4, older 551 ± 13 days of age.

Survival analysis
Survival was analyzed by Kaplan-Meier estimation and logrank test.We defined a priori "spontaneous" death as the event of interest, while being killed for any reason (e.g., organ removal) and survival at the end of the study were considered censored events.Mice used for breading were not included into the survival analysis.Total numbers of mice included in our analysis were 408, 262, 157, and 82 for wildtype, β 1 -tg, Gα i3 −/− , and β 1 -tg/Gα i3 −/− , respectively.Numbers of events of interest during the period of observation were 17, 52, 11, and 13, respectively.

Quantitative real-time PCR
Quantitative real-time PCR (qPCR) was used to reveal the relative ventricular mRNA-expression levels of the G i isoforms Gα i2 (Gnai2) and Gα i3 (Gnai3), the cardiomyopathy markers atrial natriuretic peptide ANP (Nppa) and brain natriuretic peptide BNP (Nppb), and the phosphorylation targets of protein kinase A (PKA) ryanodine receptor 2 (Ryr2), phospholamban (Pln), and troponin I (Tnni3).Ventricles were stored at − 80 °C until mRNA-isolation.All procedures were performed according to the manufacturer's protocol (QIAGEN, Hilden, Germany).The RNeasy® Fibrous Tissue Kit (QIAGEN) was used to isolate the mRNA.Quality and quantity of the purified mRNA were controlled by NanoDrop 8000 Spectrophotometer (Thermo Scientific, Waltham, MA, USA).Reverse transcription was done by using the QuantiTect® Reverse Transcription Kit (QIA-GEN).All qPCRs were run in triplicates with the ORA™ qPCR Green ROX L Mix, 2X Kit (highQu).Primer pairs for Gnai2, Gnai3, Nppa, Nppb, Ryr2, Pln, and Tnni3 have been reported before (Dizayee et al. 2011;Wiege et al. 2012;Bai et al. 2013;Keller et al. 2015) and are listed in Table S2.The gene encoding 40S ribosomal protein S29 (Rps29) served as a housekeeping gene (Figure S1).The qPCR was initiated with incubation at 95 °C for 15 min.Next, 45 cycles of denaturation were conducted at 95 °C for 15 s.Subsequently, annealing at 60 °C for 25 s, and elongation at 72 °C for 10 s were applied with a transition rate of 20 °C per second.A melting curve analysis was performed at the end to control the product purity at 64 °C for 1 min with a transition rate of 0.1 °C per second.Younger mice used for mRNA-analyses were 301 ± 4, older 554 ± 5 days of age.

Western blot analysis
Liquid-frozen ventricles were homogenized in 500 μl protein lysis buffer (20 mmol/l Tris, pH 8.3; 0.67% SDS; 238 mmol/l 2-mercaptoethanol; 0.2 mmol/l PMSF).Electrophoretic separation of Gα i isoforms was performed in gels containing 6 M urea (Gohla et al. 2007).The proteins were visualized by immunodetection using the following primary antibodies described elsewhere (Beer-Hammer et al. 2018): rabbit anti-Gα i1/i2 (7.2 ng/ml) (Leiss et al. 2020), rabbit anti-Gα i3 (50 ng/ml) (Vega et al. 2020).The protein levels of Gα i2 and Gα i3 were quantified using densitometric analysis software (Image Lab; Bio-Rad, Gräfelfing, Germany) and were normalized to the levels of GAPDH (#2118; Cell Signalling Technology, Frankfurt, Germany) of the same samples.Twenty micrograms protein per lane were loaded.The membranes were first stained with the Gα i2 or Gα i3 antibody, respectively.Membranes were than stripped and stained with the Akt antibody, stripped again, and subsequently stained with the GAPDH antibody to control for equal loading.Ventricles from three animals per genotype were analyzed in three independent experiments.For size orientation, protein standards were loaded (BioRad Precision Plus Protein Standard Dual Colour, and Nippon Genetics BlueStar PLUS Prestained Protein Standard).For the analysis of Akt phosphorylation, we used rabbit antibodies recognizing either total Akt or pAkt only when phosphorylated at Ser473 (#9272 and #9271; Cell Signalling Technology).Phospholamban expression was determined using a mouse monoclonal antibody provided by Badrilla Ltd. (#A010-14).Younger mice used for Western blot analyses were 307 ± 8, older mice 552 ± 8 days of age.

Myocyte preparation for proteomics analyses
We isolated ventricular myocytes from wildtype, β 1tg, Gα i3 −/− , and β 1 -tg/Gα i3 −/− mice (n = 3 each; age: 200 ± 56 days).Isolation followed a modified procedure according to Ackers-Johnson et al. (Ackers-Johnson et al. 2016).The chest of anesthetized mice was opened to expose the heart.Descending aorta was cut, and the heart was immediately flushed by injection of 7 ml EDTA buffer into the right ventricle.The heart was removed, and ascending aorta was retrogradely cannulated.Digestion was achieved by sequential injection of 10 ml EDTA buffer, 3 ml perfusion buffer, and 20 ml Liberase buffer (Roche, Liberase TM 0.05 mg/ml) via coronary circulation.Ventricles were then gently pulled into 1-mm pieces using forceps.Cellular dissociation was completed by gentle trituration, followed by addition of 5 ml stop buffer (perfusion buffer containing 5% FBS).Cell suspension was passed through a 100-μm filter, and cells underwent 4 sequential rounds of gravity settling, using perfusion buffer.The supernatant was discarded.The cell pellet in each round was enriched with myocytes and ultimately formed a highly pure myocyte fraction.Cardiomyocyte yields and percentage of viable rod-shaped cells were controlled under an inverse microscope.The final pellet was lysed in buffer (4% SDS in 100 mM Tris/HCl, pH 7.6).Lysates were homogenized, heated at 70 °C for 10 min, and clarified by centrifugation and protein concentrations were determined using the Bio-Rad DC assay.Proteins (1 mg) were precipitated with acetone for 1 h at -20 °C.The pellet was washed with 80% acetone once and resuspended in 8 M urea buffer (6 M thiourea, 2 M urea in 10 mM HEPES pH 7.5).Proteins were reduced with DTT (10 mM), alkylated with IAA (55 mM) and digested for 3 h with LysC (1:50 enzyme:substrate ratio, Wako chemicals).For further digestion with Trypsin (1:100 enzyme:substrate ratio, Promega), samples were diluted with ammonium bicarbonate buffer (50 mM).For whole proteomics analysis, aliquots of 50 µg were taken and desalted on stage tips.For phosphoproteomic analysis, the remaining 950 µg peptide solution was desalted using SepPak C18 cartridges and dried, and phospho-peptides were enriched using the High-Select TiO2 Phosphopeptide Enrichment Kit (Thermo scientific) following the manufacturer's instruction.

Proteomics analyses: sample measurement and data processing
Samples were measured on a Q Exactive Plus Hybrid Quadrupol-Orbitrap mass spectrometer (MS) coupled to an EASY-nLC 1000 UHPLC (Thermo Fisher Scientific) and analyzed with a 240 min gradient using Top 10 DDA method.Phospho-proteomic samples were analyzed with a 90 min gradient using Top 10 DDA method.Raw MS data files were analyzed using MaxQuant software (Max Planck Institute of Biochemistry, Martinsried, Germany) (Cox and Mann 2008).We used the Uniprot Mouse database (release November 2019) extended by the human β 1 -AR (hADRB1) sequence for spectral matching.Default settings were used, and peptides and proteins were identified using a false discovery rate (FDR) of 1%.As variable modification, p(STY) was enabled.To appreciate the biological significance of the differentially phosphorylated proteins, the ingenuity pathway analysis (IPA, QIAGEN, Germany) was used to predict regulated ontology lists, "tox lists" and networks related to cardiovascular function and disease.

Parameters analyzed
With respect to our previous study (Keller et al. 2015), we defined ventricle-to body-weight ratio, conventional echocardiographic parameters of systolic left-ventricular (LV) function (ejection fraction, LV end-systolic volume, LV end-diastolic volume, LV end-systolic length), survival time, mRNA-expression levels of Gnai2, Gnai3, Nppa, Nppb, Ryr2, Pln, and Tnni3, as well as protein expression levels of Gα i2 and Gα i3 as primary parameters.We furthermore included left-ventricular global longitudinal strain (GLS), isovolumic relaxation time (IVRT), and the ratio of E' and A' (early and late ventricular relaxation velocity) as additional echocardiographic parameters.GLS, derived from speckle tracking-based echocardiography, is a sensitive parameter for early detection of LV systolic and diastolic dysfunction and has been shown to be an independent predictor of all-cause mortality in (human) heart failure with reduced ejection fraction (Sengeløv et al. 2015;de Lucia et al. 2019).IVRT and the E' to A' ratio are sensitive indicators of diastolic function (Alex et al. 2018;Schnelle et al. 2018).Furthermore, fibrotic alterations were quantified as percentage of tissue area in ventricular slices.Western blots were performed to reveal the level of phospholamban expression and the ratio of phosphorylated to total Akt.Levels of protein expression and phosphorylation were furthermore obtained by mass spectrometry done with cardiomyocyte homogenates.Non-primary parameters were obtained and analyzed with an exploratory intention.

Performance of experiments and data analysis
Animals have not explicitly been chosen for a specific experiment or a specific date in a prospective manner.Thus, sequence of investigation was by chance due to availability of an animal at an appropriate age.Sequence of analysis was by chance, too.We did not apply specific methods for randomization of the sequence of experiments or analyses or for blinding of the experimenters.Thus, the analyses were neither specifically blinded nor actively unblinded.Knowledge of the genotype to the investigator may therefore have occurred by chance in individual cases.

Data presentation and statistical analysis
Data are depicted as scatter plots and reported as mean ± standard deviation (SD) in the text.Scatter plots were created using GraphPad Prism and show median and interquartile range or mean ± SD.We performed ANOVAs that (if significant) were followed by Bonferroni-corrected post-tests comparing all groups with respect to most primary parameters ("Performance of experiments and data analysis" section).For mRNA as well as PLN protein expression, we applied Holm-Šídák post-tests referring to agematched wildtype littermates.In addition, β 1 -tg and β 1 -tg/ Gα i3 −/− mice were compared here.In case of Gnai3 mRNA, wildtype and β 1 -tg mice were compared by non-parametric Mann-Whitney test.Due to data distribution, log10 values of 2 −ΔΔCt were analyzed with respect to ANP and BNP mRNA expression.Due to the small number of samples, we applied the pair wise fixed reallocation randomization test © using the REST-2009® software for mRNA expression in mice at 300 days of age (Pfaffl 2002).The distribution of EF and IVRT values within two groups was compared using a two-sided Fisher's exact test.Sample size estimation for echocardiography was performed a priori using G*Power 3.1.9.2 software (Heinrich Heine Universität Düsseldorf, Germany).In general, sample sizes were not increased after reviewing the corresponding data, except for qPCR, where the sample size was increased from n = 4 to n = 6-12 per group during a process of manuscript revision.Survival times are reported as mean and 95% confidence interval (CI).Survival analysis was performed by Kaplan-Meier estimation followed by the log-rank test.Throughout, we considered p values < 0.05 as indicating statistically significant differences in confirmatory analyses, i.e., analyses of the primary parameters according to our main scientific questions (see "Parameters analyzed" and "Introduction").In figures, asterisks then indicate p values below 0.05 (*) and 0.01 (**), respectively.If statistical tests were applied with an exploratory intent (see "Parameters analyzed"), p values are given as numbers down to 0.001, but not indicated by asterisks.Statistical approaches have been specified a priori, except for the analysis of distribution of EF and IVRT values ("Gα i3 deficiency reduces risk of ventricular dysfunction in β 1 -tg mice" section) that was applied due to the apparent partial overlap of data from β 1 -tg and β 1 -tg/Gα i3 −/− mice, respectively, and the calculation of Cohen's d affects sizes that was done using the Psychometrica online effect size calculators (Table S3) (Lenhard and Lenhard 2016).From some animals/tissue probes, we obtained more than one parameter.No particular statistical approach was taken to account for this.

Gα i3 deficiency prolongs survival time in β 1 -tg animals
In order to get insights into the individual role of the Gα i3 protein in murine cardiomyopathy, mice overexpressing the cardiac β 1 -adrenoceptor but globally lacking the Gα i3 protein (β 1 -tg/Gα i3 −/− ) were compared to β 1 -adrenoceptor overexpressing (β 1 -tg), Gα i3 -deficient (Gα i3 −/− ), and wildtype mice.In all mouse lines used, the distribution of genotypes followed Mendel's rule and that of sex was almost equal (48% male and 52% female).Mice showed no obvious physical phenotype and normal behavior.Deaths occurred suddenly in all groups, at best preceded sporadically by (unspecific) symptoms, e.g., reduced ingestion, striking behavior, or impaired movement.

3
Survival is a major outcome parameter of heart-failure studies, and in our previous study, we found that Gα i2 deficiency caused a significantly shortened lifetime of β 1 -tg mice (Keller et al. 2015).In the current study, we followed the survival of the mice up to a maximum age of 880 days.Kaplan-Meyer estimation and log-rank tests revealed that the mean survival time of β 1 -tg mice was 624 days (95% CI: 592-655), significantly shorter than that of all other genotypes (Fig. 1).Importantly, the concomitant absence of Gα i3 increased the mean life span of β 1 -tg mice to 696 days (644-747), which was no longer statistically different from the survival time of 789 days for wildtype (739-840) and 742 days for Gα i3 −/− mice (693-790), respectively.In summary, the life span of β 1 -tg/Gα i3 −/− mice was significantly longer than that of β 1 -tg mice and was not statistically different from that of wildtype littermates, whereas in a previous study, the absence of Gα i2 shortened the life expectancy of β 1 -tg mice.

Study of animals at an age of 300 days
Given that Gα i2 deficiency already showed adverse effects in 300-day-old β 1 -tg mice (Keller et al. 2015), we first focused on effects of Gα i3 deficiency at this age.
In summary, neither ventricular hypertrophy nor dysfunction was observed in any of the investigated groups at the age of 300 days.In contrast to lack of Gα i2 (Keller et al. 2015), a detrimental effect of Gα i3 deficiency in β 1 -tg mice at this age is unlikely.

Study of animals at an age of 550 days
In line with increased mortality of β 1 -tg mice at more advanced ages, we next examined animals at 550 days of age.Effect sizes obtained by comparing wildtype with β 1 -tg, Gα i3 −/− and β 1 -tg/Gα i3 −/− and β 1 -tg with β 1 -tg/ Gα i3 −/− mice at this age can be taken from supplemental Table 3 (Table S3).
These findings of hypertrophy and fibrosis may be related to ventricular dysfunction in β 1 -tg mice on the one hand and to protective effects of Gα i3 deficiency on the other.Therefore, we describe below our findings on left ventricular function obtained by echocardiography.
Taken together, Gα i3 deficiency reduced the risk of both systolic and diastolic LV dysfunction in β 1 -tg mice.
Akt activation has been linked to cardiomyopathy, and studies from other tissues indicated isoform-specific modulation by G i proteins.We thus analyzed expression of phosphorylated Akt protein.Western blots, however, revealed no obvious differences when comparing pAkt/ Akt ratios in ventricles of wildtype, β 1 -tg, Gα i3 −/− and β 1tg/Gα i3 −/− mice (Fig. 6).Consistent with this, quantitative proteomics analyses demonstrated no changes in Akt expression or Akt phosphorylation (see the "Proteomics and pathway analyses" section).
In summary, both Gnai2 and Gnai3 mRNA levels were increased in 550-day-old β 1 -tg mice while we found no change at the protein level.Akt phosphorylation was not obviously affected by either β 1 -AR overexpression or Gα i3 deficiency.
In summary, Western blots indicated reduced PLN expression in ventricles of β 1 -tg, but not β 1 -tg/Gα i3 −/− mice at the age of 550 days.Ryr2, Pln, and Tnni3 mRNA levels appeared to be unaffected.

Proteomics and pathway analyses
Protein and protein phosphorylation levels in ventricular myocytes from wildtype, β 1 -tg, Gα i3 −/− , and β 1 -tg/ Gα i3 −/− mice (n = 3 animals each, age: 200 ± 56 days) were determined by mass spectrometry.We did not detect any change in Akt expression or phosphorylation in agreement with the data from Western blot analysis (cp.Fig. 6).For phospholamban, however, we found a statistically significant increase in phosphorylation at Ser16 (p = 0.007) and Ser17 (p = 0.013) in β 1 -tg compared to wildtype ventricular myocytes that was not seen in β 1 -tg/Gα i3 −/− myocytes.For further analysis, we fed the protein phosphorylation data into a so-called ingenuity pathway analysis (IPA).IPA is an algorithm-based analysis that uses the QIAGEN knowledge base to identify differences in signaling pathways.When comparing our β 1 -tg and β 1 -tg/Gα i3 −/− target genotypes, analyses revealed statistically significant differences associated with multiple cardiac disorders and diseases (Fig. 8A).IPA suggested activation of the predefined protein ontology list "cardiac fibrosis" in β 1 -tg compared to wildtype ventricles, while activity was reduced in β 1 -tg/ Gα i3 −/− compared with data from β 1 -tg mice (Fig. 8B).To provide further evidence for possible mechanisms, the proteins detected in our probes were mapped to the networks available in the underlying QIAGEN database and then scored using a network score based on p values obtained in Fisher's exact test.For the sample-specific network that achieved the highest score, β 1 -tg and β 1 -tg/Gα i3 −/− myocytes showed some differences in phosphorylation of interacting proteins linked to "tox lists" such as "cardiac fibrosis" and "cardiac hypertrophy" (Fig. 8C, D).For example, PLN, RYR2, and calmodulin kinase II (CaMK II) phosphorylation is seen to be increased in β 1 -tg compared to wildtype myocytes (color-coded red in Fig. 8C), whereas it is lower in β 1 -tg/Gα i3 −/− compared with β 1 -tg mice (color-coded green in Fig. 8D).
Taken together, proteomics analyses showed that in β 1 -tg ventricular myocyte phospholamban phosphorylation levels were significantly increased.Pathway and network analyses based upon protein phosphorylation indicated opposite patterns in β 1 -tg compared to β 1 -tg/Gα i3 −/− mice with respect to cardiac dysfunction and disease.These results are in good agreement with our in vivo data of ventricular dysfunction and our in vitro data such as ventricular fibrosis or increased expression of hypertrophy markers.

Discussion
Given the previously shown detrimental effects of a Gα i2 deficiency in mice with a cardiac overexpression of β 1 -AR (Keller et al. 2015), we now asked for the role of the , 12 (GLS), 11 (E'/A') and 13 (IVRT) for β 1 -tg, and 9 (GLS) and 10 (E'/A' and IVRT) for β 1 -tg/Gα i3 −/− mice, respectively.Scatter plots with median and interquartile range are shown.If the ANOVA indicated statistically significant differences, it was followed by Bonferronicorrected post-tests between all groups.Analysis was done with an exploratory intention, and thus, exact p values are given if < 0.05 closely related Gα i3 isoform in this murine heart-failure model.Since β 1 -transgenic (β 1 -tg) mice develop progressively impaired cardiac functions accompanied by a significantly shortened life span, this heart-failure model is suitable to test for effects of an additional Gα i3 deficiency.We wondered how Gα i3 deficiency affects cardiac function and outcome of β 1 -tg mice, i.e., whether it is detrimental, protective, or has no effect.

G i proteins in β-AR-mediated heart failure
Our current study revealed that the absence of Gα i3 in β 1 -AR-overexpressing mice was protective, slowing or even preventing the development of heart failure.In contrast, we previously found that the absence of Gα i2 in β 1 -tg mice resulted in a distinct heart-failure phenotype even before it was evident in mice overexpressing only the β 1 -AR (Keller et al. 2015).Thus, the possibility that Gα i3 deficiency mimics the Gα i2 -knockout phenotype in the β 1 -tg model of dilative cardiomyopathy can be excluded.One may hypothesize that the remaining Gα i isoform functionally replaces the missing isoform.Indeed, the absence of one Gα i isoform is often accompanied by upregulation of the remaining one (Wiege et al. 2012;Köhler et al. 2014;Devanathan et al. 2015;Beer-Hammer et al. 2018), although Western blot analyses have been inconsistent regarding an increase of cardiac Gα i2 expression in Gα i3 -deficient mice at the protein level (Gohla et al. 2007;Dizayee et al. 2011;Hippe et al. 2013;Köhler et al. 2014).
In the current study, we did not see an upregulation of Gα i2 in Gα i3 -deficient hearts.However, one should keep in mind that cardiac Gα i2 expression exceeds that of Gα i3 per se.Therefore, we cannot exclude the possibility that Gα i2 contributes by functional substitution even in the presence of unchanged (i.e., "normal") expression levels.On the other hand, (cardiac) Gα i3 levels might be generally too low to compensate for Gα i2 deficiency.This could explain why we observed adverse effects of Gα i2 deficiency in the previous study, although there was a statistically significant increase in Gα i3 -protein expression (Keller et al. 2015).Unfortunately, due to its embryonic lethality, the Gα i2/i3 double knockout mouse model cannot be used to test the assumption that the Gα i isoforms can substitute for each other in the β 1 -tg mouse model (Gohla et al. 2007).At the mRNA level, expression of both Gα i2 (Gnai2) and ).Mean values ± SD are depicted Gα i3 (Gnai3) transcripts appeared to be increased in β 1 -tg ventricles while there was no obvious change at the protein level.In rats treated with isoproterenol, the increase in Gnai mRNA transcript levels was significantly more pronounced than the increase in Gα i protein expression (Mende et al. 1992;Eschenhagen et al. 1992a).Thus, we cannot exclude that we have missed an only slight increase of Gα i expression at the protein level.
Although survival is reduced by cardiac overexpression of β 1 -adrenoceptors alone, it has been even worse in β 1 -tg mice lacking Gα i2 (Keller et al. 2015).In contrast, we now find that the life span of β 1 -tg mice is significantly increased if they lack Gα i3 .Although 550-day-old β 1 -tg mice lacking Gα i3 showed increased cardiac ANP and BNP mRNA levels compared to wildtype littermates, the ANP increase was significantly lower compared to mice only overexpressing the β 1 -AR.In addition, it appeared to be clearly lower than in Gα i2 -deficient β 1 -tg mice at an age of 300 days as analyzed in our previous study (Keller et al. 2015).Cardiac overexpression of β 2 -adrenoceptors also leads to cardiac failure, although a significantly higher level of overexpression is required (Liggett et al. 2000).Similar to our recent findings with β 1 -tg mice, lack of Gα i2 drastically shortened the lifespan of mice with a cardiac overexpression of the β 2 -AR subtype in another study (Foerster et al. 2003;Keller et al. 2015).Of note, β 2 -tg mice with a homozygous Gα i2 knockout were virtually non-viable and already heterozygous Gα i2 deficiency reduced life span to a similar extent as did the complete absence of Gα i2 on a background of cardiac β 1 -AR overexpression.These findings may reflect the role of G-proteins for either β 1 -AR-or β 2 -AR-mediated signaling: while G s proteins are the cognate interaction partners of β 1 -AR (Xiao et al. 1999;Seyedabadi et al. 2019), it is widely accepted that β 2 -AR couple to both G s and G i proteins (Xiao et al. 1999(Xiao et al. , 2003)).With respect to putative isoform-specific effects of G i proteins, it should be noted that in a mouse model of ischemia-reperfusion-induced cardiac damage, Köhler et al. also found detrimental effects of Gα i2 deficiency on the one hand while Gα i3 deficiency appeared to be cardioprotective on the other hand (Köhler et al. 2014).
In conclusion, we show that in a mouse model of dilative cardiomyopathy, Gα i3 deficiency is beneficial.In contrast, lack of Gα i2 was clearly detrimental in previous studies, either in the same heart-failure model of β 1 -AR overexpression, a model of β 2 -AR overexpression or with ischemia-reperfusion as pathophysiological stimulus (Foerster et al. 2003;Köhler et al. 2014;Keller et al. 2015).

Differences of Gα i2 -and Gα i3 -dependent effects at the cellular and subcellular level
Data from neutrophils suggest an interesting difference in Gα i2 -versus Gα i3 -mediated signaling: in a study of Kuwano et al.,Gα i2 deficiency led to an increase, but Gα i3 deficiency to a decrease of Akt phosphorylation (Kuwano et al. 2016).Akt has been described to be involved in cardio-protective signaling, while on the other hand, chronic activation of the PI3K/Akt cascade is related to cardiac hypertrophy, and Akt activity was increased in human failing hearts (Haq et al. 2001;Nagoshi et al. 2005).In a previous study, we found no difference between Akt phosphorylation in Gα i2 -or Gα i3deficient mice, neither under basal conditions nor after treating mice with carbachol (Dizayee et al. 2011).The opposing results of our previous study and that of Kuwano et al. may be explained not only by the various tissues analyzed but the different genetic backgrounds (C57/BL6 and 129/Sv, respectively) which have been associated with phenotypic differences in G i -knockout models (Offermanns 1999;Kuwano et al. 2016).Regarding our previous findings on Akt in mice lacking Gα i2 or Gα i3 (Dizayee et al. 2011), one should bear in mind that carbachol is rather considered a non-pathologic stimulus, and stimulation of muscarinic receptors might be beneficial under pathological conditions, e.g., heart failure (Communal et al. 1999;Olshansky et al. 2008;Lorenz et al. 2009).Although we cannot eventually rule out a change, no obvious differences in ventricular pAkt expression were found in the current study.Given the otherwise pronounced effects on Akt phosphorylation in human heart failure and in heart failure models (Haq et al. 2001;Baba et al. 2003;Miyamoto et al. 2004), it seems at least unlikely that the marked differences in cardiac function and survival in our study can be explained by changes in Akt phosphorylation.
Western blots suggested a reduced PLN expression in β 1 -tg compared to both wildtype and β 1 -tg/Gα i3 −/− mice.We furthermore used protein expression and phosphorylation data obtained from ventricular myocytes for ingenuity pathway analysis (IPA).IPA has the advantage of a lower risk of bias than manual analysis of the results would have.Our data indicate significant differences between β 1 -tg and β 1 -tg/Gα i3 −/− mice with respect to intracellular signaling relevant to several cardiac diseases including arrhythmia, heart failure or cardiac fibrosis.Of note, data obtained with ventricular myocytes from Gα i2 -deficient mice indicate significant differences to mice lacking Gα i3 in signaling related to cardiac diseases, too (not shown).Proteomics analyses indicated increased phosphorylation of PLN in ventricular myocytes of β 1 -tg but not β 1 -tg/Gα i3 −/− mice.Our findings on PLN expression and phosphorylation suggest reduced SERCA inhibition in β 1 -tg hearts.This may be considered compensatory, as SERCA expression and activity are reduced in the setting of heart failure (del Monte and Hajjar 2008).In agreement with this, Engelhardt et al. found that genetic PLN ablation rescued β 1 -tg mice from heart failure (Engelhardt et al. 2004).It is tempting to speculate that the absence of compensatory PLN changes in β 1 -tg/Gα i3 −/− mice is indicative of cardioprotection by Gα i3 deficiency, as 1 3 reduced PLN activity is not needed here.In Gα i2 -deficient ventricles, our proteomics analysis revealed an increase of PLN phosphorylation similar to that in β 1 -tg specimens (not shown).This is interesting because in our previous study, Gα i2 deficiency alone already led to reduced life expectancy, but this effect was dramatically more pronounced when these animals also overexpressed the cardiac β 1 -AR (Keller et al. 2015).
Previously, we found a decreased density of ventricular L-type calcium currents (LTCC) in ventricular cardiomyocytes from Gα i2 -deficient mice, while it was increased in Gα i3 -deficient cardiomyocytes (Dizayee et al. 2011).Though in other models an increase of ventricular calcium currents led to cardiac damage and dysfunction in the long run (Muth et al. 1999;Nakayama et al. 2007;Beetz et al. 2009), Gα i3 deficiency does not impair cardiac function ( (Jain et al. 2001) and this study).β 2 -Adrenoceptors couple to both G s and G i proteins, while G s proteins are considered the cognate interaction partners of β 1 -adrenoceptors (Xiao et al. 1999;Seyedabadi et al. 2019).However, β 1 -and β 2adrenergic signaling seems to be modulated by G i proteins including mechanisms independent of direct receptor coupling (Li et al. 2004;Martin et al. 2004;Melsom et al. 2014).Thus, it cannot be excluded that the above-mentioned differences between ventricular calcium currents in either Gα i2 -or Gα i3 -deficient mice also have a role in the development of cardiomyopathy in the β 1 -tg mouse model.
The data discussed so far do not explain our findings regarding the opposing effects of Gα i2 and Gα i3 , but PLN expression and activity as well as ventricular L-type calcium currents should be the subject of further investigations into possible molecular mechanisms underlying the differential effects of Gα i isoforms in cardiomyopathy.Figure 9 and Table 3 summarize results on mechanisms that might contribute to isoform-specific signaling via inhibitory G-proteins in the heart.

Limitations of the study
The focus of our study centered on the hypothesis that Gα i3 and Gα i2 have different isoform-specific effects in a mouse model of dilated cardiomyopathy, despite sharing very high amino acid identity.In fact, we found significant functional differences between the two Gα i isoforms.When evaluating these results, however, some methodological peculiarities must be considered that have an impact on the interpretation of the results.
Firstly, the data on qualitatively distinct differences in the effects of Gα i2 (Keller et al. 2015) and Gα i3 deficiency are based on two separate studies in which the gene-deficient mice were each tested against wildtype controls, but not directly against each other.Not least for animal welfare reasons, we were not able to retest a Gα i2 -deficient cohort in our current study.
Furthermore, we used global Gα i knockouts in the current and our previous study (Keller et al. 2015).Thus, we cannot exclude the possibility that extra-cardiac effects had an impact on the cardiac phenotype.However, previous studies comparing Gα i2 -and Gα i3 -deficient mice with their respective wildtype controls showed that, for example, basal heart rate or blood pressure was unchanged (Jain et al. 2001;Albarrán-Juárez et al. 2009).Furthermore, unaltered hypotensive effects following systemic α 2 -AR stimulation indicated normal circulatory regulation in Gα i2 -and Gα i3deficient mice, respectively (Albarrán-Juárez et al. 2009).We did not obtain catecholamine levels in our study.In a previous study, however, norepinephrine release from atria or brain cortex slices was not altered in Gα i2 -or Gα i3 -deficient mice (Albarrán-Juárez et al. 2009).Furthermore, given the unchanged basal values of heart rate and blood pressure in the absence of Gα i2 or Gα i3 , significant changes in catecholamine levels seem rather unlikely (Jain et al. 2001;Albarrán-Juárez et al. 2009;Keller et al. 2015).Another study revealed the contribution of endogenous catecholamines to the phenotype of β 1 -tg mice to be negligible, thus arguing against a significant increase in catecholamine levels in this model, too (Engelhardt et al. 2001b).
The mouse model of β 1 -AR overexpression is a wellestablished and thoroughly characterized murine heartfailure model, but differs in some features from human heart failure.For example, there is up-instead of downregulation of β 1 -AR (Bristow et al. 1986;Engelhardt et al. 1999).However, β 1 -AR overexpression can be considered as mimicking the chronically increased sympathetic stimulation observed in human heart failure (Engelhardt et al. 1999;Baker 2014).Although the transgenic approach displays a "non-physiologically" high β 1 -AR expression level, Fig. 8 Protein phosphorylation levels in ventricular myocytes (three per genotype) were obtained by mass spectrometry and fed into a socalled ingenuity pathway analysis (IPA), an algorithm-based analysis that uses the QIAGEN knowledge base to identify differences in signaling pathways.A As indicated by − (log p) values, IPA revealed significant differences between β 1 -tg and β 1 -tg/Gα i3 −/− mice in activation of disease-associated pathways.B Heatmap representing differential phosphorylation of proteins assigned to the "cardiac fibrosis pathway."Activation Z-scores referring to the respective genotypes compared are color-coded (top row) from blue ("lower activity") to orange ("higher activity").An increase in the phosphorylation of a particular protein is shown in red, a decrease in green (lower rows).Gray color indicates comparisons that did not reveal statistical significance (p > 0.05).When comparing β 1 -tg with wildtype (C), and β 1 -tg/ Gα i3 −/− with β 1 -tg mice (D), the top scoring IPA network indicated differences related to cardiac dysfunction and cardiovascular disease.Red indicates that in the genotype mentioned first, a protein is more phosphorylated relative to the comparator; green stands for reduced phosphorylation.For example, phospholamban (PLN) phosphorylation is increased in β 1 -tg compared to wildtype myocytes, while it is reduced in β 1 -tg/Gα i3 −/− compared to β 1 -tg mice.Relationships between proteins (nodes) and heart diseases (cardiac fibrosis, cardiac hypertrophy) are indicated (Tx, toxicity-related ("tox") lists) ◂ the (over-)expression levels in the C57BL/6-based mice we used here are significantly lower than on the FVB/N background on which the model was originally generated (Keller et al. 2015).
Here, we analyzed cardiac function only under basal conditions.A future study using stressors (e.g., dobutamine), could potentially reveal further differences between Gα i isoforms, e.g., a possibly increased functional reserve in β 1 -tg mice lacking Gα i3 .Some of our results were supported by proteomics analysis (e.g., with respect to PLN, fibrosis, or hypertrophy).However, it should be noted that this was mainly a screening approach.Nevertheless, the results obtained may point to future studies on the role and molecular mechanisms of Gα i2 -and Gα i3 -mediated signaling in heart failure.

Conclusion
Gα i3 deficiency has no detrimental effects in a mouse model of dilative cardiomyopathy and even appears to be cardio-protective.Our current and previous results indicate a β 1 -AR-mediated impairment whose development is oppositely associated with the expression of either Gα i2 or Gα i3 .Although the underlying molecular mechanisms remain to be elucidated in further studies, our findings indicate isoform-specific interventions into G i -dependent signaling pathways (e.g., inhibiting Gα i3 ) to be promising novel strategies for cardio-protective therapies.
Fig. 9 Isoform-specific Gα i functions possibly involved in heart disease.1: Gα i2 , but not Gα i3 , mediates signal transduction upon M-AChR stimulation and thereby may protect against β 1 -AR-mediated overstimulation, e.g., with respect to Ca 2+ influx via L-type Ca 2+ channels (Ca V 1.2) (Nagata et al. 2000).2: With increased signal transduction via β 2 -AR, Gα i2 increases the activity of individual Ca V 1.2, while Gα i3 inhibits channel activity (Foerster et al. 2003;Klein 2009).3: The coupling of β 2 -AR to Gα i3 may be stronger than to Gα i2 , e.g., depending on the local membrane charge (Strohman et al. 2019).4: Under basal conditions, Gα i2 appears to increase Ca V 1.2-mediatedI CaL or to compensate for presumed inhibitory effects of Gα i3 and vice versa ( (Dizayee et al. 2011), but: (Nagata et al. 2000)).5: Gα i2 , but not Gα i3 , mediates phosphorylation of ERK and may thereby be involved in the stimulation of Ca V 1.2 (Dizayee et al. 2011).The effect of ERK phosphorylation on heart disease has been described to be either harmful or protective, probably depending on the stimulus (Lorenz et al. 2009;Ruppert et al. 2013).Overall, G i proteins differentially modu-late Ca V 1.2 and thus I CaL via several isoform-specific mechanisms that depend among other things on the (level of) activity of β 1 -AR, β 2 -AR, and/or M-AChR.Alterations in Ca V 1.2 activity and/or I CaL have been associated with cardiomyopathy and heart failure (6).7: Gα i2 deficiency led to an increase, Gα i3 deficiency to a decrease of Kir3/GIRK mediated currents (Nobles et al. 2018).Lack of Gα i2 thus might be pro-arrhythmic.Arrhythmia is a major reason of death in heart failure, and rhythm disturbances might cause or aggravate cardiomyopathy (8)." + " means stimulation/increase, " − " inhibition/decrease. "?" indicates that an interaction, a contribution or a consequence is not clear or fully understood.AC: adenylyl cyclase; β 1 -AR: β 1 -adrenoceptor; β 2 -AR: β 2adrenoceptor; ERK: extracellular signal-regulated kinase; Kir3/GIRK: inward-rectifier potassium channel/G protein-coupled inward-rectifier potassium channel; M-AChR: muscarinic acetylcholine receptor

Table 3
Possible isoform-specific effects of Gα i2 or Gα i3 in cardiac signaling observed in Gnai-deficient mice (italics: conclusions on the role of the respective Gα

Table 1
Ventricle-to body-weight ratios and echocardiographic data at an age of 300 days (mean ± SD, number of animals in brackets)For each parameter, a one-way ANOVA was performed (*p < 0.05).# p < 0.05 vs. wildtype and § p < 0.05 vs. β 1 -tg in Bonferroni-corrected posttests between all groups

Table 2
qPCR data on ventricular mRNA expression at an age of 300 days (% of wildtype; mean ± SD) p < 0.05 vs. wildtype in a pair wise fixed reallocation randomization test © using the REST-2009® software *