Cardiac inflammatory CD11b/c cells exert a protective role in hypertrophied cardiomyocyte by promoting TNFR2- and Orai3- dependent signaling

Early adaptive cardiac hypertrophy (EACH) is initially a compensatory process to optimize pump function. We reported the emergence of Orai3 activity during EACH. This study aimed to characterize how inflammation regulates store-independent activation of Orai3-calcium influx and to evaluate the functional role of this influx. Isoproterenol infusion or abdominal aortic banding triggered EACH. TNFα or conditioned medium from cardiac CD11b/c cells activated either in vivo [isolated from rats displaying EACH], or in vitro [isolated from normal rats and activated with lipopolysaccharide], were added to adult cardiomyocytes before measuring calcium entry, cell hypertrophy and cell injury. Using intramyocardial injection of siRNA, Orai3 was in vivo knockdown during EACH to evaluate its protective activity in heart failure. Inflammatory CD11b/c cells trigger a store-independent calcium influx in hypertrophied cardiomyocytes, that is mimicked by TNFα. Pharmacological or molecular (siRNA) approaches demonstrate that this calcium influx, depends on TNFR2, is Orai3-driven, and elicits cardiomyocyte hypertrophy and resistance to oxidative stress. Neutralization of Orai3 inhibits protective GSK3β phosphorylation, impairs EACH and accelerates heart failure. Orai3 exerts a pathophysiological protective impact in EACH promoting hypertrophy and resistance to oxidative stress. We highlight inflammation arising from CD11b/c cells as a potential trigger of TNFR2- and Orai3-dependent signaling pathways.

described in the non-excitable cells, TRPCs, STIM1 and Orai1 molecules drive store-operated Ca 2+ entry 10,11,[13][14][15] . An alternative Ca 2+ entry pathway, independent of store-depletion, involves the key participation of the Orai3 molecule [14][15][16][17][18][19][20] . Orai3-driven store-independent Ca 2+ entry relies on initial arachidonic acid (AA) production, and is selectively activated by AA itself (ARC channels) or its leukotriene C4 (LTC4) metabolite (LRC channels), in all cell types examined to date. Knowledge regarding Orai3 contribution to cardiac remodeling remains scarce. We recently demonstrated the emergence of an Orai3-dependent pathway that drives an AA-dependent Ca 2+ influx in hypertrophied cardiomyocytes from rats subjected to abdominal aortic banding 12 . This study documented the essential role of constitutive Orai3-dependent activity to initiate and maintain early adaptive hypertrophy in response to pressure overload. But pathophysiological triggers and mechanisms leading to Orai3 activation during EACH remained unknown, as well as its direct impact on cardiomyocytes and its functional relevance in HF.
Cardiac remodeling is a complex inflammatory syndrome 5 , and beneficial or detrimental role of inflammatory signaling during EACH is not fully understood. Growing evidence indicates that inflammatory responses emerging in EACH and HF are different, displaying divergent cytokine profiling 21 . The pro-inflammatory cytokine TNFα is upregulated in CH and HF. In the 1990's, the "cytokine hypothesis" argued for the detrimental contribution of an excessive production of TNFα to the pathogenesis of HF 22 , via binding to the TNFR 1 receptor subtype, suggesting that TNFα neutralization would be beneficial. Surprisingly, large clinical trials failed to demonstrate a benefit of anti-TNFα strategies 23,24 . There is now evidence that TNFα can also improve remodeling and hypertrophy and alleviate inflammation and fibrosis upon binding to the TNFR 2 receptor subtype or regulation of TNFR 1 signaling, in cardiomyocytes, or indirectly after induction of GM-CSF secretion by endothelial renal cells, or influencing cardiac immune cell phenotypes 2,[25][26][27][28][29] . In this context, we have previously shown that AA mediates dual effect of TNFα on Ca 2+ transients and contraction of adult rat myocytes 30 and identified TNFR 2 -dependent activation of the cytosolic phospholipase A 2 (cPLA 2 ) activity as a pathway leading to AA production and conferring resistance of adult cardiomyocytes to H 2 O 2 26 . Recent studies suggested the potential adaptive role of TNFα in early cardiac remodeling showing that myocardial gene expression of TNFα is significantly higher in patients with well compensated aortic stenosis than in patients with decompensated stenosis 31 and the association of circulating TNFα with concentric left ventricular remodeling 32 . The present study aimed to investigate the potential regulation of the AA-dependent Orai3 influx by TNFα in early adaptive cardiac remodeling, identify the potential cellular source of such an inflammatory signal, evaluate the impact of TNFα-induced Orai3 regulation on cardiomyocyte hypertrophy and resistance to H 2 O 2, and assess the functional relevance of Orai3 activity in HF.
Our study points out a novel TNFR 2 -dependent signaling pathway in cardiomyocytes that triggers Orai3-driven Ca 2+ influx enhancing hypertrophy and promoting an increased resistance to oxidative stress. Cardiac CD11b/c cells arise as a potential source of this protective inflammatory stimulus. Neutralization of Orai3 during EACH fosters evolution towards HF.

Results
TNFα triggers activation of Orai3-Ca 2+ influx in hypertrophied cardiomyocytes. To investigate the regulation of the Orai3-Ca 2+ influx by TNFα, we first used adult cardiomyocytes isolated from normal rats and incubated or not with isoproterenol (iso) (100 nM overnight) to elicit in vitro hypertrophy, as demonstrated by an increased cell area (2256 ± 37 vs. 2541 ± 53 µm 2 , n = 308 control vs. 313 Iso cells, p < 0.0001, Mann Whitney U test). We performed Ca 2+ -imaging experiments in Fura 2 -loaded cardiomyocytes to directly evaluate the impact of inflammation on Orai-dependent Ca 2+ influx. After electrical stimulation as a quality test, cells were placed in a medium appropriate for the measurement of voltage-and store-independent Ca 2+ influx containing diltiazem and ryanodine where Na + was replaced by the large organic ion N-methyl-D-glucamine, as previously reported 12 (Fig. S1). After equilibration in the absence of Ca 2+ , 1 mmol/L Ca 2+ was added into the extracellular medium and the resultant initial increase in Fura 2 fluorescence (1 st slope) was taken as an index of initial rate of Ca 2+ influx. This protocol was routinely applied a second time (2 nd slope) to allow paired comparison, either between two identical perfusion conditions and to ascertain reproducibility of measurements (Fig. 1A,C), or between two different perfusion conditions (Fig. 1B,D). Two successive applications of the same "basal" medium gave rise to similar rates of Ca 2+ influx, either in control or hypertrophied cardiomyocytes (Fig. 1A,C). In contrast, addition of TNFα to the second incubation medium induced a significant increase in the rate of Ca 2+ influx (2 nd slope) as compared to basal (1 st slope). Importantly, this effect was selectively detected in hypertrophied cardiomyocytes ( Fig. 1B) but not in control ones (Fig. 1D). These data show that TNFα selectively induces Ca 2+ influx in hypertrophied cardiomyocytes.
We then asked whether Orai3 drives this TNFα-activated store-and voltage-independent Ca 2+ influx. In in vitro iso-treated hypertrophied cells, two successive applications of identical TNFα containing medium gave rise to similar rates of Ca 2+ influx ( Fig. 2A left). However, preincubation for 10 min with Orai pharmacological inhibitors, YM58483 or Synta66, reduced the rate of Ca 2+ influx observed in response to a second challenge with TNFα, to a value similar to basal ( Fig. 2A middle and right). Moreover, in vivo molecular knockdown of Orai3, via intramyocardial injection with Cy3-tagged Orai3 siRNA, also blunted the TNFα effect on Ca 2+ influx that was still observed in cells isolated from scramble siRNA-injected hearts (Fig. 2B). Orai3 neutralization was performed as previously reported 12 and demonstrated by quantitative RT-PCR and Western-blot (Fig. 2C) and by detection of Cy3 positive cells (Fig. 2B). Of note, knockdown of Orai3 by siRNA injection in normal rats did not modify the cardiomyocyte size (Table S1). In contrast, both siScramble and siOrai3 cardiomyocytes had a tendency to be bigger after in vitro post-treatment with iso and displayed similar sizes. This suggested that the in vitro iso-hypertrophic response was not altered in siOrai3 cardiomyocyte, in contrast to TNFα/Orai3 signaling.
Complementary experiments were performed in hearts from rats subjected to chronic iso-infusion for 14 days (1.5 mg/kg/day) to elicit in vivo hypertrophy (Table 1). Iso-induced EACH remodeling was confirmed by an increase in end-diastolic and end-systolic interventricular septum (IVSd and IVSs), posterior wall thicknesses (PWd and PWs) and concentric hypertrophy (h/r, diastolic wall thickness to radius ratio) in iso-treated rats, Activation of Orai3-Ca 2+ influx by TNFα relies on binding to TNFR 2 , stimulation of cPLA 2 and potential production of AA metabolites via the lipoxygenase pathway. Next we aimed to evaluate the role of TNFα receptors 1 and 2 and of the cPLA 2 pathways in TNFα signaling. In in vitro iso-treated hypertrophied cells, stimulation of Ca 2+ influx by TNFα was unaffected by the preincubation for 1 hour either with control IGg2A or anti-TNFR 1 -antibodies (Ab) but was impaired in the presence of neutralizing TNFR 2 -Ab (Fig. 3A). Preincubation with the cPLA 2 inhibitor, methyl arachidonyl fluorophosphonate (MAFP), suppressed TNFα signaling whereas addition of the phospholipase A 2 activating peptide (PLAP) mimicked TNFα effect, and stimulated Ca 2+ influx (Fig. 3B). TNFα-induced activation was unaffected by the pretreatment with the cyclo-oxygenase inhibitor indomethacin but impaired in the presence of a lipoxygenase inhibitor nordihydroguaiaretic acid (NDGA), suggesting the potential requirement of AA metabolism into leukotrienes in this signal transduction (Fig. 3C). TNFα signaling persisted in the presence of the antagonist of leukotriene receptors, montelukast, suggesting an effect independent of binding to external receptors (Fig. 3C). These results indicate that TNFα signals Figure 1. TNFα activates a store-and voltage-independent Ca 2+ influx in hypertrophied but not in control cardiomyocytes. Representative recordings of Fura 2 fluorescence ratio (F340/F380) in iso-treated hypertrophied (A,B) or control (C,D) cardiomyocytes subjected to 2 successive measurements of the rate of voltage-and storeindependent Ca 2+ entry and quantification of rates of Ca 2+ entry. Two successive applications of the same 1 mM Ca 2+ basal perfusion medium gave rise to similar rates of Ca 2+ entry, both in hypertrophied (A) and normal cells (C). Addition of TNFα to the second 1 mM Ca 2+ perfusion medium enhanced the rate of Ca 2+ entry (2 nd slope as compared to 1 st slope) in hypertrophied cardiomyocytes (B) but not in normal cardiomyocytes (D). Number of cells analyzed and number of cell isolations (rats) as indicated. Mean ± SEM of cells, Wilcoxon matched-paired tests to examine if the mean of the 2 nd rate differs from the 1 st one, arbitrarily set as 1, ****p < 0.0001.

Figure 2.
TNFα activates a store-and voltage-independent Ca 2+ influx in hypertrophied cardiomyocytes further identified as Orai3 dependent. Hypertrophied cardiomyocytes were loaded with Fura 2 before measurement of voltage-and store-independent Ca 2+ influx. (A) Two successive applications of the same TNFα perfusion medium on iso-treated hypertrophied cardiomyocytes gave rise to similar rates of Ca 2+ entry. The first TNFα-induced Ca 2+ influx was arbitrarily set to 2.2 to allow comparisons with the control conditions (see Fig. 1B). Preincubation with Orai inhibitors, YM58483 or Synta66, prior the second application of TNFα perfusion medium blunted activation of Ca 2+ entry by TNFα. (B) Hypertrophied cardiomyocytes isolated from hearts injected with Cy3-tagged scramble or Orai3 siRNAs three days before isolation and iso-treatment. Typical images show Cy3-fluorescence in cardiomyocytes. TNFα activates Ca 2+ influx in scramble siRNAtransfected cells but not in siOrai3-transfected cardiomyocytes. Number of cells analyzed and number of cell isolations (rats) as indicated, mean ± SEM of cells, Wilcoxon matched-paired tests to examine if the mean of the 2 nd rate differs from the 1 st one, arbitrarily set as 2.2 (A) or 1 (B), **p < 0.01, ***p < 0.001, ****p < 0.0001. www.nature.com/scientificreports www.nature.com/scientificreports/ in iso-hypertrophied cardiomyocytes through TNFR 2 to activate cPLA 2 and produce AA potentially leading to increased leukotriene levels which in turn activate Orai3.
Of note, we recently demonstrated the emergence of an Orai3-dependent pathway that drives an AA-dependent Ca 2+ influx in hypertrophied cardiomyocytes from rats subjected to abdominal aortic banding (AAB) for 28 days 12 . Complementary experiments were performed to examine the regulation of Orai3 by TNFα in this model (protocol in Fig. 4A). EACH remodeling was confirmed in AAB rats, relative to Sham rats, by an increase in IVSd, IVSs, PWd and PWs and concentric hypertrophy (h/r) ( Table 2). AAB-induced hypertrophy was also attested by an increased cell area (4718 ± 128 vs. 3311 ± 124 µm 2 , n = 129 vs. 117 cells, p < 0.0001, Mann Whitney U test). A slight decrease of FS was observed in this model ( Table 2).
TNFα also induced activation of Orai3-dependent Ca 2+ influx in AAB-induced hypertrophied cardiomyocytes ( Fig. 4B,C) that relied on binding to TNFR 2 ( Fig. 4D) and potential production of AA metabolites via activation of the lipoxygenase pathway (Fig. 4E).
Thus, emergence of a TNFR 2 -dependent Orai3-driven Ca 2+ influx characterized cardiomyocyte hypertrophy triggered by either iso-treatment or pressure overload.

Inflammatory CD11b/c cells trigger TNFR 2 -dependent activation of Orai3-Ca 2+ influx in hypertrophied cardiomyocytes.
Next experiments aimed to evaluate the potential cellular source of inflammation in EACH hearts. Immunohistological examination of cardiac sections from iso-induced EACH rats indicated an increased number of TNFα-positive cells as compared to control and 66 ± 4% of TNFα-positive cells were identified as myeloid CD11b/c-positive cells (Fig. S2).
Rat hearts (obtained from rats implanted with iso-pump for 14 days, Table 1) were used to isolate cardiomyocytes, fibroblasts and myeloid CD11b/c cells (denominated as in vivo cell activation) (Fig. 5A,B). Conditioned medium (Cmed) obtained from cardiomyocytes or cardiac fibroblasts were without effect on voltage-and store-independent Ca 2+ influx in in vitro hypertrophied cardiomyocytes in contrast to the Cmed obtained from their cardiac CD11b/c counterparts (Fig. 5C). This pointed out the CD11b/c cells as the potential source of the Orai3 inflammatory trigger in EACH hearts. Of note, levels of TNFα detected in Cmed from CD11b/c cells (1.15 ± 0.3 pg/ml, n = 6) were 10 to 20 fold higher than levels measured in Cmed from their cardiac counterparts. CD11b/c-Cmed-induced activation was sensitive to neutralizing TNFR 2 antibodies and Orai inhibitor YM58483 (Fig. 5D).
A similar effect was triggered using in vitro CD11b/c-Cmed (obtained from CD11b/c cells isolated from normal rat hearts and in vitro stimulated with lipopolysaccharide (LPS) (10 ng/ml for 2 hours) to induce pro-inflammatory activation) (Fig. 5B,E). When CD11b/c cells were pretreated with the anti-inflammatory drug semapimod (Sema) before LPS stimulation, in vitro CD11b/c-Cmed Sema was without effect on Ca 2+ influx ( Fig. 5B,E). Anti-inflammatory impact of semapimod was indicated by a reduction in TNFα-positive staining of CD11b/c cells (Fig. S3) and a limited TNFα content in Cmed Sema as compared to Cmed LPS (0.01 ± 0.004 vs. 0.24 ± 0.06 pg/ml, n = 5 vs. 8, p < 0.01, Mann Whitney U test).
These results highlighted the cardiac inflammatory CD11b/c cells as the potential triggers of Orai3-dependent Ca 2+ influx in hypertrophied cardiomyocytes.
These results indicate that inflammation (TNFα or Cmed from cardiac inflammatory CD11b/c cells) enhances cardiomyocyte hypertrophy via TNFR 2 -and Orai-dependent signaling pathways.
We also evaluated the potential impact of inflammation on the resistance to oxidative stress using in vitro hypertrophied cardiomyocytes and demonstrated the role of TNFR 2 and Orai signaling pathways. Iso-hypertrophied (C) Efficient knockdown of Orai3 mRNA and protein in cardiomyocytes isolated from Wistar rats at day 3 following injection with Cy3-tagged scramble or Orai3 siRNAs. Histograms representing relative transcript levels normalized to the RPL32 mRNA or relative protein levels normalized to Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) protein. Mean ± SEM of cardiomyocytes from 2-6 rats/group, Mann-Whitney U test, *p < 0.05. Full length blots were included in SI. (D) Hypertrophied cardiomyocytes isolated from rats after fourteen days of chronic iso-infusion injected with Cy3-tagged scramble or Orai3 siRNAs three days before isolation. TNFα activates Ca 2+ influx in hypertrophied cardiomyocytes isolated from hearts injected or not with Cy3-tagged scramble siRNA, but not in hypertrophied cardiomyocytes isolated from hearts injected with Cy3tagged Orai3 siRNA, three days before isolation. Number of cells analyzed and number of cell isolations (rats) as indicated, mean ± SEM of cells, Wilcoxon matched-paired tests to examine if the mean of the 2 nd rate differs from the 1 st one, arbitrarily set as 1, **p < 0.01, ****p < 0.0001. www.nature.com/scientificreports www.nature.com/scientificreports/ d14 control rats (n = 7) d14 iso-treated rats (n = 21)  www.nature.com/scientificreports www.nature.com/scientificreports/ cardiomyocytes (100 nM for 18 hours) were preincubated with TNFα or in vitro CD11b/c-Cmed, and cell resistance was estimated by the number of rod-shaped cells after H 2 O 2 -oxidative stress (100 µM H 2 O 2 for 2.5 hours) (Fig. 7A). Counting of resistant cells indicated 21 ± 2% vs. 100% in the presence and in the absence of H 2 O 2 , respectively (Fig. 7B). Treatment with TNFα increased resistance of hypertrophied cardiomyocytes from 21 ± 2% to 36 ± 3%, in a TNFR 2 -and YM58483-sensitive manner (rod-shaped cells reduced to 23 ± 1% and 26 ± 4%, after preincubation with the TNFR 2 -Ab or YM58483, respectively) (Fig. 7B). CD11b/c-Cmed alleviated the deleterious impact of H 2 O 2 , improving the yield of resistant cells to 50 ± 6% (Fig. 7B). Pretreatment of CD11b/c cells with anti-inflammatory semapimod blunted the beneficial impact of Cmed resulting in only 21 ± 4% resistant cells, as well as pretreatment with the TNFR 2 -Ab or YM58483 (36 ± 3% and 38 ± 2% resistant cells, respectively) (Fig. 7B).
These results demonstrate that inflammation (TNFα or Cmed from cardiac inflammatory CD11b/c cells) improves resistance to H 2 O 2 in hypertrophied cardiomyocytes via TNFR 2 -and Orai-dependent protective signaling pathways.
Taken together our in vitro results argue for a protective Orai3-driven signal emerging during the phase of EACH and promoting adaptive cardiomyocyte hypertrophy and beneficial resistance to oxidative stress. In order . TNFα activates Orai-Ca 2+ influx after binding to TNFR 2 and potential production of AA metabolites via the lipoxygenase pathway in hypertrophied cardiomyocytes from rats with AAB-induced EACH. (A) Schematic representation of the protocol where rats with AAB-induced EACH for 28 days were subjected to echographic analyses. (B-E) Hypertrophied cardiomyocytes isolated from AAB rats were loaded with Fura 2 before measurement of voltage-and store-independent Ca 2+ influx. TNFα activates Ca 2+ entry (B) in a manner sensitive to Orai inhibitor YM58483 (C), neutralizing TNFR 2 -Ab (D) or lipoxygenase inhibitor NDGA (E). Number of cells analyzed and number of cell isolations (rats) as indicated, mean ± SEM of cells, Wilcoxon matched-paired tests to examine if the mean of the 2 nd rate differs from the 1 st one, arbitrarily set as 1 (B,D,E) or 2.2 (C) ****p < 0.0001. www.nature.com/scientificreports www.nature.com/scientificreports/ to evaluate their relevance in the pathophysiology of HF, we performed a kinetic analysis of echocardiographic parameters and evaluated tissue remodeling in response to a unique intramyocardial Orai3 siRNA injection applied at the onset of EACH.

Orai3 neutralization during EACH impairs cardiac hypertrophy, fosters alteration of function and dilation and deleterious tissue remodeling.
To validate in vivo the protective role of Orai3 activity on EACH, we studied the effect of cardiac Orai3 knockdown during iso infusion in mice. First, as shown in Fig. S4, as compared to previous results obtained in rats (Table 1 and 12 ), we checked that this model displayed similar iso-induced EACH remodeling after a 14 days infusion period in control mice. EACH was characterized by evolution of echocardiographic parameters, increase in heart weight to body weight ratio and cardiomyocyte size, and associated with an increase in Nppa and Tnfα mRNA expressions, but no change in Orai3 mRNA level ( Fig. S4 and Table S2). This model was chosen to develop a novel approach of percutaneous intramyocardial injection of siRNA under echographic guidance validated in mice by other groups (i.e. 33 ) in order to avoid potential artefactual consequences of the surgical thoracotomy on tissue remodeling that are susceptible to affect the post-operative echographic surveillance and evolution of cardiac parameters.
Iso-infused mice were subjected to a unique intramyocardial injection of either Scramble or Orai3 siRNA at day 8 following pump implantation (Fig. 8A). Efficient knockdown of Orai3 during EACH was attested at the mRNA and protein levels (Fig. 8D,E). At day 15, hearts from siOrai3-injected mice displayed a decreased hypertrophy (lower PWs), an increased dilation (higher LVs) and an altered cardiac function (lower FS) as compared to hearts from siScramble-injected mice ( Fig. 8C and Table 3). At day 28, siOrai3-injected mice still presented with a lower heart weight to body weight ratio, a smaller cardiomyocyte area, a decrease in Myh7 mRNA level, as compared to siScramble-injected mice (Fig. 8F). Orai3 knockdown during EACH also fostered fibrosis attested by histological analysis and increased Col1a1, Col3a1 and Tgfβ mRNA levels (Fig. 8G). Mechanistically, a decrease in the ratio phospho-GSK3β/GSK3β was correlated with the reduction of Orai3 expression (Fig. 8H).
Importantly, the intramyocardial injection of Orai3 siRNA in control mice showed that knockdown of Orai3 was without functional impact ( Fig. S5 and Table S3).
Taken together, these results demonstrated the emergence of a functional protective role of Orai3 during EACH.

Discussion
Our results highlight the emergence of a protective role for Orai3 in the hypertrophied cardiomyocytes. We identified TNFα as a mechanistic trigger of the TNFR 2 -dependent activation of Orai3-Ca 2+ influx and showed that CD11b/c cells are a potential driving source of this signaling in EACH hearts. This paracrine signaling enhances hypertrophy and promotes resistance to oxidative stress of hypertrophied cardiomyocytes. Furthermore, we show that Orai3 knockdown during EACH fosters HF (Fig. 9).
Cardiac Orai3-dependent Ca 2+ influx was previously identified as a prohypertrophic stimulus in AAB-induced CH 12 . We confirm these results in a model of iso-induced EACH, a model of reproducible progressive concentric hypertrophy. In our study, we found that in vivo reduction in Orai3 expression in iso-infused rats decreased the mean cardiomyocyte area. In keeping with this atrophic impact of Orai3 siRNA, neutralization of Orai3 during EACH in iso-infused mice triggers a rapid and significant decrease in hypertrophic parameters still detectable at day 20 post injection. Furthermore, our in vitro experiments document a direct prohypertrophic effect of Orai3 in isolated iso-treated cardiomyocytes.
Interestingly, our results highlight novel protective properties of Orai3-dependent Ca 2+ influx. In line with the reported resistance of Orai3 channel to redox regulation 34 , we show that Orai3 activation confers resistance to oxidative stress in isolated hypertrophied cardiomyocytes. Furthermore, our in vivo results indicate that efficient cardiac knockdown of Orai3 during EACH inhibits adaptive hypertrophy, alters cardiac function and promotes fibrosis. Mechanistically, recent results from our laboratory indicate that the Orai3-interacting protein STIM1 is essential to tune the Akt/GSK3β prosurvival signaling 8 www.nature.com/scientificreports www.nature.com/scientificreports/ Figure 5. Conditioned medium from CD11b/c cells isolated from EACH heart activates Ca 2+ influx in a manner sensitive to TNFR 2 -Ab and Orai inhibitor YM58483. (A) Schematic representation of the protocol where rats implanted with an iso-pump for fourteen days were subjected to echographic analyses and developed EACH. (B) CD11b/c cells, cardiomyocytes and fibroblasts were isolated from EACH hearts and cultured for 18 hours before recovery and concentration of conditioned media. CD11b/c cells pre-incubated with/without the anti-inflammatory drug, semapimod, prior in vitro LPS application to induce pro-inflammatory activation, were isolated from normal hearts and cultured for 18 hours before recovery and concentration of conditioned media. Iso-treated hypertrophied cardiomyocytes were loaded with Fura 2 before measurement of voltage-and store-independent Ca 2+ influx, 1 st in the absence and 2 nd in the presence of these conditioned media (Cmed). (C) Only Cmed from CD11b/c cells isolated from EACH hearts activates a voltage-and store-independent Ca 2+ influx (D) in a manner sensitive to TNFR 2 -Ab and Orai inhibitor YM58483. (E) Cmed from CD11b/c cells isolated from normal hearts and in vitro stimulated with LPS activates a voltage-and store-independent Ca 2+ influx in a manner sensitive to neutralizing TNFR 2 -Ab, Orai inhibitor YM58483 and anti-inflammatory pretreatment with semapimod. Number of cells analyzed, number of cell isolations (rats) and number of Cmed tested as indicated, mean ± SEM of cells, Wilcoxon matched-paired tests to examine if the mean of the 2 nd rate differs from the 1 st one, arbitrarily set as 1, ***p < 0.001, ****p < 0.0001.    Table 3, ANOVA for repeated measures followed by Dunn-Sidak post-hoc tests). (D,E) Efficient knockdown of Orai3 mRNA and protein levels in cardiac homogenates at d12, d15 and d28 following injection (mean ± SEM of mice, n = 7 mice/group, Kruskal-Wallis followed by Dunn's post-hoc tests). Full length blots were included in SI. (F,G) SiOrai3 injection induced a decrease in Heart Weight/Body Weight ratio, myocyte area and Myh7 mRNA level, an increase in fibrosis, and an elevation of Col1a1, Col3a1 and Tgfβ mRNA levels (mean ± SEM of mice, n = 7 mice/group, Mann-Whitney (2019) 9:6047 | https://doi.org/10.1038/s41598-019-42452-y www.nature.com/scientificreports www.nature.com/scientificreports/ Orai3 during EACH is associated with a decrease in phospho-GSK3β/GSK3 ratio. In the current absence of published knockout data for Orai3, our study is the first to address its pathophysiological relevance and indicate its essential protective role during EACH.
We have identified the first pathophysiological trigger of Orai3-driven store-independent Ca 2+ influx in cardiomyocytes, namely a TNFα/TNFR 2 -dependent signaling. Interestingly, the regulation of Orai3 by TNFα is detected in hypertrophied cardiomyocytes (both in response to iso treatment or in the AAB-model), but not in normal cardiomyocytes. Of note, iso-hypertrophied and normal cardiomyocytes display similar TNFR 2 expression (2.4 ± 0.4 vs. 2.4 ± 0.3 pg TNFR 2 /mg, respectively, n = 4). However, and as previously reported in AAB-induced CH hearts 12 , co-immunoprecipitation experiments using STIM1 antibodies indicate enhanced Orai3 recruitment to STIM1 in the iso-induced EACH (Fig. S6). Since Orai3-STIM1 interaction is a prerequisite for Orai3-dependent Ca 2+ influx activation 14,17 , its absence in normal cells and its enhancement in hypertrophied cells constitute a major difference likely to explain the lack of impact of Orai3-dependent Ca 2+ -influx as well as the absence of TNFα regulation in normal cells. Accordingly, our in vivo and in vitro experiments in normal mice show that the knockdown of Orai3 is without impact on echocardiographic parameters as well as heart or cardiomyocyte size. They argue for a pathophysiological impact of Orai3 exerted during EACH but not under control conditions. Dominant impact of inflammation is currently reported as deleterious in normal hearts, as it is associated with reactive oxygen species production, proapoptotic signaling and a dominant role of TNFR 1 pathways overwhelming TNFR 2 signaling 26 . Using TNFR 1 and TNFR 2 knockout mice implanted with iso-pumps, Prabhu et al. demonstrated that TNFR 2 but not TNFR 1 signaling prevents the detrimental long-term effects of β-adrenergic receptor stimulation in the heart 37 . In another study, the group of Prasad proposed that the beneficial effects of TNFR 2 signaling in presence of sympathetic overdrive could act through the preferential TNFR 2 -mediated recruitment of GRK2 to mediate βAR desensitization thus reducing deleterious cardiac signaling and remodeling. In agreement with these studies, our study further suggests the emergence of a protective TNFR 2 pathway during EACH development via the stimulation of a novel downstream effector Orai3.
We show here that cardiac CD11b/c cells are a potential source of inflammatory TNFα/TNFR 2 -dependent signaling leading to Orai3-dependent Ca 2+ channel activation. Both Cmed from in vitro and in vivo activated CD11b/c cells stimulate a store-independent Ca 2+ influx in hypertrophied cardiomyocytes in a TNFR 2 -Ab and YM58483 sensitive manner. This suggests the presence of activated inflammatory CD11b/c cells in EACH hearts. We demonstrated that LPS-activated CD11b/c cells enhance hypertrophy and promote resistance of hypertrophied cardiomyocytes to oxidative stress, in a TNFR 2 -Ab and YM58483 sensitive manner. This protective impact is blunted by preincubation of CD11b/c cells with the anti-inflammatory drug semapimod. Our data strengthen the concept that inflammation arising from cardiac myeloid cells may exert paracrine beneficial impact on cardiomyocytes 38,39 . Of note, cardiac macrophages are an emerging focus for therapeutic strategies aimed at minimizing cardiomyocyte death, ameliorating pathological cardiac remodeling and for treating HF 40 .
We observed that TNFα-dependent activation of Orai3-Ca 2+ influx relies on cPLA 2 activation and is mimicked by a cPLA 2 activator. This is in accordance with the reported presence of the lipid interaction site located in the NH 2 terminal intracytosolic sequence of Orai3 20 . TNFα-dependent effect is sensitive to NDGA treatment which suggests the potential requirement of a lipoxygenase-dependent AA metabolism to activate Orai3-Ca 2+ influx. Accordingly, Trebak et al. previously described LTC4, a lipoxygenase AA-metabolite, as an activator of Orai3 in the vascular smooth muscle cells 41 . However, this proposal needs to be tempered since NDGA, in addition to inhibit lipoxygenase activity may also exert several off target effects (i.e., PKC inhibition, overall anti-oxidant, ER-Golgi protein shuttling inhibition). Our results further illustrate the dual role of cPLA 2 -AA signaling in mediating TNFα effects in adult cardiac myocytes. We have identified Orai3 as a novel protective TNFα-cPLA 2 -AA pathway. Accordingly, the beneficial impact of TNFα-cPLA 2 -AA pathways has been previously reported on the cardiomyocyte calcium transients and contraction (i.e. by the group of Oceandy 42 and our previous results 26,30 ) and on the survival to oxidative stress 26 . In contrast, the group of Gugiyama reported the deleterious impact of a TNFα-cPLA 2 cardiac signaling in a model of ischemia-reperfusion 43 .
In conclusion, our in vitro and in vivo studies characterize the Orai3 signaling pathway that exerts a direct protective role against HF in hypertrophied cardiomyocytes. Mechanistically, we have identified Orai3 as a novel driver of TNFR 2 -dependent inflammation instrumental in this protection. Furthermore, we showed a protective Orai3-dependent paracrine role of cardiac myeloid cells leading to adaptive hypertrophy and improved resistance to oxidative stress. In summary, our results are the first to address the functional role of Orai3 signaling in HF that may open new perspectives for patients' treatments.

Methods
Ethics. Care of the animals and surgical procedures were performed according to the Directive 2010/63/ EU of the European Parliament, which had been approved by the Ministry of Agriculture, France, (authorization for surgery C-75-665-R). The project was submitted to the French Ethic Committee CEEA (Comité d'Ethique en Expérimentation Animale) and obtained the authorization Ce5/2012/050 and APAFIS#1729-2015-083114195840v8. All experiments were performed in accordance with relevant named guidelines and regulations.
In vivo intramyocardial ultrasound-guided transthoracic siRNA delivery in mice. On-target plus Scramble and Orai3 siRNA (Dharmacon GE healthcare) were injected by ultrasound-guided transthoracic intramyocardial injection, as described in 33 , at day 8 after iso-pump implantation. Echocardiographic parameters were measured regularly as stated.
Cardiac cell isolation and culture. Rat cardiac myocytes were isolated either from chronically iso-infused hearts or from normal hearts, after injection or not with Cy3-tagged siRNA, three days before, when stated. Rats were administered a sodium pentobarbital (Ceva Sante Animale, France) intra-peritoneal injection (200 mg/ kg). Hearts were harvested and kept in ice-cold Krebs-Henseleit (KH) solution supplemented with 10 mmol/L taurine and 0.5 mmol/L EGTA, then rapidly canulated and mounted on the Langendorff apparatus. The hearts were retrogradely perfused through the aorta, first with a Krebs-Henseleit (KH) solution supplemented with 10 mmol/L taurine for 5 minutes, then with enzymatic solution for 20 minutes. The KH solution contained (in mmol/L): 100 NaCl, 4 KCl, 5.5 NaHCO 3 , 1 KH 2 PO 4 , 1.7 MgCl 2 , 10 D-glucose, 15 2,3-butanedione monoxime, 22 Hepes, pH = 7.4 with NaOH. The enzyme solution was supplemented with 1 mg/ml collagenase A (Roche Applied Science, Meylan, France) and 5 mg/ml bovine serum albumin BSA (Sigma, Lyon, France). All chemicals were from Sigma (Lyon, France. Cell suspension was then used to differentially isolate ventricular cardiomyocytes, cardiac fibroblasts and cardiac CD11b/c cells. Ventricular cardiomyocytes from iso-pump rats (in vivo hypertrophy) were plated onto laminin-coated glasses and maintained overnight in M199 medium (Life technologies). Cells isolated from normal hearts or from saline-pump rats were added with 100 nM isoproterenol plus 100 µM ascorbic acid (Sigma), when stated, to induce in vitro hypertrophy, three hours after plating, and let overnight in M199 medium.
Cardiac fibroblasts were isolated by centrifugation, plated onto 12-well plates and maintained in DMEM medium (Life technologies).
Cardiac CD11b/c cells were isolated by centrifugation, enriched using an anti-CD11b/c antibody coupled to magnetic beads (MiltenyiBiotec) and maintained overnight in RPMI medium (Life technologies) supplemented with 10 mmol/L Hepes. When stated, CD11b/c cells were in vitro polarized towards a pro-inflammatory phenotype upon incubation for 2 hours with 10 ng/ml lipopolysaccharide (LPS) before overnight incubation with a new LPS-free medium.
All cardiac cells were cultured for 18 hours following plating. Conditioned media from cardiac cells were concentrated on Amicon 10 kDa centrifugal filter. Fura-2 AM calcium imaging. Isolated rat ventricular myocytes were loaded with Fura 2 -AM (Molecular Probes, Life Technologies) as reported 26 . Transfected cells (detected by fluorescence imaging as cells positive for Cy3-tagged siRNA) or non-transfected cells, rhythmically beating in response to electrical stimulation (square waves, 0.5 Hz, as previously described 26 ), were analyzed. Measurements were recorded on a Zeiss Platform equipped with an Axio Observer Z1 microscope, a DGA plus illuminator and a camera Coolsnap HQ2 (workstation Carl Zeiss).
Cells were first paced for few cycles and Ca 2+ transients were recorded to ensure viability and functionality of the cell. Cells were incubated in tyrode buffer (1.8 mmol/L Ca 2+ ) to check the stability of basal cytosolic calcium level and then switched to appropriate store-and voltage-independent Ca 2+ -free buffer. Store-independent Ca 2+ entry was then measured upon readdition of 1 mmol/L Ca 2+ . Ca 2+ off/Ca 2+ on protocols repeated twice allowed paired comparison between two similar (for reproducibility assessment) or distinct perfusion conditions. Tyrode buffer contained (in mmol/L): 135 NaCl, 4 KCl, 1 MgCl 2 , 10 D-glucose, 20 Hepes, pH = 7.4 with NaOH. Store-and voltage-independent buffer contained 1 µM ryanodine, 20 µM diltiazem and 135 mmol/L N-methyl D-glucamine (NMDG) instead of NaCl. All chemicals, excepted ryanodine (Tocris), were from Sigma.
Data analysis was performed using the Zen Software (2012, blue edition). The rates of Ca 2+ entry were estimated by the slope of the first minute of initial increase in Fura-2 fluorescence ratios in response to the re-addition of Ca 2+ . 10-100 myocytes isolated from 2-16 animals were analyzed per experimental condition (as stated in Figs).
Measurement of cell hypertrophy. Cardiomyocyte hypertrophy was estimated after 18 hours in culture in the presence of isoproterenol. After an initial incubation with iso alone (100 nM) for 1.5 hours, cardiomyocytes were then treated or not for 1 hour with TNFR 2 -Ab or YM58483 before addition of control medium, TNFα or CD11b/c Cmed. Cardiomyocytes were visualized using brightfield at x20 magnification and cell area was measured in at least 300 cells per condition per experiment. Results were the mean of at least three different experiments performed on two cell isolations (using at least two different in vitro CD11b/c-Cmed).

Measurement of cell resistance to H 2 o 2 .
Cell resistance experiments were performed as previously described 26 . In vitro hypertrophied cardiomyocytes were preincubated for 1 hour with or without TNFR 2 -Ab or for 10 minutes with or without YM58483. Then, TNFα or Cmed from in vitro activated CD11b/c cells, or control medium were added for 10 minutes before subsequent treatment or not with H 2 O 2 (100 µM, (Sigma)) for  www.nature.com/scientificreports www.nature.com/scientificreports/ Quantification of cardiomyocyte area and tissue fibrosis. Frozen sections fixed in paraformaldehyde were labeled with Wheat Germ Agglutinin (WGA)-Alexa 647 (1/500 dilution, Molecular Probes). Tissue sections were analyzed with a Zeiss Axio Observer Z1 microscope using ImageJ software. A low vs. high threshold allowed quantification of cardiomyocyte area or tissue fibrosis, respectively, as previously reported 47 . Results were quantified from 7 mice/group (12 images/animal).