Differential changes in cyclic adenosine 3′‐5′ monophosphate (cAMP) effectors and major Ca2+ handling proteins during diabetic cardiomyopathy

Abstract Diabetic cardiomyopathy (DCM) is associated with differential and time‐specific regulation of β‐adrenergic receptors and cardiac cyclic nucleotide phosphodiesterases with consequences for total cyclic adenosine 3′‐5′ monophosphate (cAMP) levels. We aimed to investigate whether these changes are associated with downstream impairments in cAMP and Ca2+ signalling in a type 1 diabetes (T1D)‐induced DCM model. T1D was induced in adult male rats by streptozotocin (65 mg/kg) injection. DCM was assessed by cardiac structural and molecular remodelling. We delineated sequential changes affecting the exchange protein (Epac1/2), cAMP‐dependent protein kinase A (PKA) and Ca2+/Calmodulin‐dependent kinase II (CaMKII) at 4, 8 and 12 weeks following diabetes, by real‐time quantitative PCR and western blot. Expression of Ca2+ ATPase pump (SERCA2a), phospholamban (PLB) and Troponin I (TnI) was also examined. Early upregulation of Epac1 transcripts was noted in diabetic hearts at Week 4, followed by increases in Epac2 mRNA, but not protein levels, at Week 12. Expression of PKA subunits (RI, RIIα and Cα) remained unchanged regardless of the disease stage, whereas CaMKII increased at Week 12 in DCM. Moreover, PLB transcripts were upregulated in diabetic hearts, whereas SERCA2a and TnI gene expression was unchanged irrespective of the disease evolution. PLB phosphorylation at threonine‐17 was increased in DCM, whereas phosphorylation of both PLB at serine‐16 and TnI at serine‐23/24 was unchanged. We show for the first time differential and time‐specific regulations in cardiac cAMP effectors and Ca2+ handling proteins, data that may prove useful in proposing new therapeutic approaches in T1D‐induced DCM.


| INTRODUC TI ON
Cyclic adenosine 3′-5′ monophosphate (cAMP) is a pivotal regulator of cardiac contractility, relaxation and automaticity. The intracellular levels of this second messenger are finely tuned by their rate of synthesis by adenylyl cyclases (AC), and degradation by cyclic nucleotide phosphodiesterases (PDEs). 1 In normal myocardium, cAMP elevation, particularly through β-adrenergic receptor (β-AR) stimulation, exerts inotropic and lusitropic effects by activating cAMP-dependent protein kinase (PKA). In its inactive form, PKA holoenzyme is a heterotetramer composed of two regulatory (R) and two catalytic (C) subunits. 2-4 Four types of R subunits (RIα, RIβ, RIIα, and RIIβ) and three C subunits (Cα, Cβ, and Cγ) have been described 5

with RIα,
RIIα and Cα being the major isoforms expressed in the heart 6 and encoded by distinct genes. Two classes of PKA holoenzymes have been identified (type I and II), which differ in their R subunits (RI and RII) 7,8 and cellular localization. 9 Binding of cAMP to PKA-R subunits causes the dissociation of the tetrameric PKA holoenzyme and thus the release of the free PKA-C subunits, 10,11 hence promoting the phosphorylation and activation of key components of the cardiac excitation-contraction coupling (ECC). These include L-type Ca 2+ channels (LTCCs) and their constitutive inhibitor Rad, 12 ryanodine receptors (RyR2) of the sarcoplasmic reticulum (SR), phospholamban [PLB, a constitutive inhibitor of the SR Ca 2+ pump (SERCA2a)], Troponin I (TnI) and cardiac myosin-binding protein C (MyBP-C). 13 Activation of the main actors of the ECC enhances Ca 2+ cycling and consequently increases heart rate (HR), contraction amplitude and relaxation. 14 Along with PKA, Ca 2+ /Calmodulin-dependent kinase II (CaMKII) contributes to β-AR regulation of cardiac function 15,16 ; LTCCs, RyR2, MyBP-C and PLB are also substrates for CaMKII.
Emerging evidence supports a relevant role for Epac, a guanine nucleotide-exchange factor for the small GTPases Rap1 and Rap2, 17-20 as a mediator of cAMP signalling in the heart and a regulator of Ca 2+ signalling/contractility. 21 Two cardiac Epac isoforms have been identified. Epac1 is the major neonatal isoform whereas Epac2 expression is predominant in adults. 22 Upon cAMP binding, Epac enhances the phosphorylation of several Ca 2+ handling proteins including RyR2 and PLB, thereby facilitating Ca 2+ release from SR and reuptake. Epac also increases cardiac myofilament Ca 2+ sensitivity through phosphorylation of TnI and MyBP-C. 23 These effects have been attributed to the activation of Epac/Rap/Phospholipase C (PLC)/protein kinase C (PKC)/ CaMKII-mediated signalling. 21,24,25 Diabetes mellitus (DM) represents a major global health problem 26 and contributes to the development of diabetic cardiomyopathy (DCM), which finally culminates in heart failure (HF) 27,28 in the absence of hypertension and structural heart diseases. DCM is characterized by cardiac remodelling, myocardial fibrosis, diastolic and systolic dysfunction. 29 The underlying pathophysiology of the prolonged process that culminates in DCM and HF is complex. However, it is well accepted that cardiac dysfunction and HF caused by DM are associated with metabolic abnormalities, sympathetic nervous system (SNS) overactivity and Ca 2+ mishandling. 30,31 Nevertheless, contrasted results were obtained regarding altered β-AR-mediated response and PKA signalling in various animal models of DM. 32,33 Moreover, there is little information concerning Epac modification in DCM despite its established role in mediating pro-hypertrophic effects of β-AR stimulation. 34 Although disturbances in Ca 2+ signalling and CaMKII have been investigated in DCM, 30,35,36 data on actual changes affecting the expression and activity of the proteins involved in Ca 2+ homeostasis during the evolution of DCM are controversial. 30,31,33 In a recent study, we characterized the evolution of cardiac structure and function in a rat model of DCM at 4, 8 and 12 weeks after streptozotocin (STZ)-induced type 1 diabetes (T1D). 37 We showed that sustained hyperglycaemia in diabetic rats was associated with cardiac remodelling including steatosis and fibrosis as well as bradycardia and an early (4 weeks) increase in cardiac systolic function as evaluated by ejection fraction (EF) and fraction shortening (FS) of the left ventricle (LV). 37 This correlated with upregulation of β 1 -ARs and total cAMP levels, whereas normalization of cardiac function and cAMP occurred at later time points and correlated with upregulation of some of the major cardiac PDEs. 37 However, whether these changes were associated with downstream modifications of cAMP effectors and Ca 2+ handling proteins in DCM was not investigated. Thus, this study was designed to delineate the alterations affecting Epac1/2 isoforms and PKA (RI, RIIα and Cα subunits), as well as the expression of CaMKII and the major actors of the ECC process in the same rat model of STZ-induced T1D.
We show for the first time differential and time-specific regulations in cardiac Epac1/2, PKA subunits, CaMKII, as well as the impact of these alterations on the phosphorylation status of PLB and TnI.
These results may prove useful in proposing new therapeutic approaches in T1D-induced DCM. Eighty-nine adult male Wistar rats were used in this study. T1D was induced at 5 weeks of age in rats weighing between 80-130 g as previously described. 37 Briefly, 12 h after fasting, animals received one intraperitoneal injection of streptozotocin (STZ, Sigma-Aldrich: 65 mg/kg in 0.1 M citrate buffer, pH = 4.5). Age-matched control rats (CON) were injected with vehicle (0.1 M citrate buffer, pH = 4.5) via the same route. After 72 h, fasting glucose levels were measured in blood droplets withdrawn from the tails of animals, using Accu-Check® Performa glucometer (Roche). Eighty-five per cent of the rats injected with STZ exhibited fasting blood glucose (FBG) levels >200 mg/dL in addition to polyuria and polydipsia and were considered diabetic. All experiments were performed at 4, 8 and 12 weeks after STZ or vehicle injection ( Figure S1).

| Anatomical study
Body weights (BW) were measured under anaesthesia, and the hearts were rapidly excised, rinsed with a fresh cold physiological saline solution, weighed and stored in liquid nitrogen for real-time PCR for the genes of interest [Atrial natriuretic factor (ANF), Epac1/2, SERCA2a, PLB and TnI] and western blot analysis for Epac2, the main PKA subunits (RI, RIIα and the Cα subunit), CaMKII, SERCA2a, p-TnI (Ser 23/24 ), total TnI, p-PLB (Ser 16 ), p-PLB (Thr 17 ) and total PLB. Lungs, liver and kidneys were also removed from all animals and weighed.

| Real-time quantitative PCR
Total RNA was isolated from the frozen cardiac tissue of all control and diabetic rats using TRIZOL reagent (Ambion; life technologies).
RNA (1 μg) was reverse-transcribed to single-stranded cDNA using GAPDH was used as a housekeeping gene. Expression of the following genes was analysed: ANF, Epac1/2, SERCA2a, PLB and TnI. Table S1 shows the sequence of forward and reverse primers that were used. The relative expression level of each gene was determined using the comparative cycle threshold (Ct) method (2 −∆∆Ct ) normalized to control GAPDH gene. control ratio] was first quantified for each sample and then normalized to its corresponding total protein/loading control ratio: R2/R1.

| Statistical analysis
All quantitative data are presented as mean ± SEM and analysed by Prism (version 8.0). A two-way anova was performed in order to study the effect of STZ treatment, time and their interaction on the different outcomes. When the interaction was significant, two-way anova was followed by post hoc Tukey's multiple comparison test. Significance was set to p < 0.05.

| Induction and characterization of DCM in adult rats
In order to delineate the alterations affecting cAMP effectors and the major actors of the cardiac ECC, we used a T1D-induced DCM model, which we characterized in detail in a previous study. 37 T1D was induced by STZ injection in 5-week-old rats ( Figure S1), and glycaemia was measured in both vehicle-treated (CON) and STZ-treated (STZ) rats at 4, 8 and 12 weeks after injection. FBG levels were increased by ~4.4-fold in STZ-treated rats compared with their agematched CON ( Figure 1A). Anatomical data of control and diabetic rats monitored at 4, 8 and 12 weeks following injection indicate that all STZ-treated rats had a significantly lower BW ( Figure 1B) as well as lungs, liver and kidneys weights compared with their age-matched CON (Table 1). Moreover, a significant decrease in the heart weight (HW) by ~50% was noted in all STZ-treated rats compared with their age-matched CON ( Figure 1C and Table 1) along with increases in the expression of ANF ( Figure 1D), a biomarker gene for DCM. These data attest a structural remodelling in the diabetic hearts.

| Expression of Epac1/2 isoforms in control and diabetic hearts
To determine whether DCM impacts cAMP downstream signalling, we first measured mRNA expression of Epac1 and Epac2 in hearts from control and diabetic rats at 4, 8 and 12 weeks following either vehicle or STZ injection. Each cardiac sample was then normalized to GAPDH levels, which were similar between STZtreated rats and their age-matched CON ( Figure S2). Statistical analysis showed no significant interaction between STZ treatment and the time on the mRNA expression of Epac1; however, STZ treatment induced significant increases in its expression, similarly across time (p < 0.001; Figure 2A). In contrast, the upregulation in Epac2 transcripts by ~2-fold was not apparent in STZ-treated rats compared with their age-matched CON until Week 12 (p < 0.05; Figure 2B). We then determined whether the strong increase in Epac2 mRNA observed in the diabetic hearts translates into similar changes in protein expression. Equal amounts of proteins prepared from heart extracts from CON or STZ rats were separated on SDS/ PAGE, and Epac2 protein was detected by western blotting using Epac2 selective antibody. A single band migrating at approximately 110 kDa was detected in rat hearts for Epac2 ( Figure 2C; Figure S3) and its expression did not change in diabetic rats compared with their age-matched CON regardless of the disease stage ( Figure 2D).

| Expression of PKA subunits and CaMKII in control and diabetic hearts
We then examined whether the expression of PKA, another cAMP effector playing a pivotal role in the regulation of cardiac contraction, 13 is altered in DCM. Different PKA subunits were investigated in hearts from control and diabetic rats at 4, 8 and 12 weeks following either vehicle or STZ injection. We detected a single band F I G U R E 1 Hyperglycaemia is associated with cardiac remodelling in DCM. (A) Comparison of the fasting blood glucose levels (mg/dL) in CON (black diamonds; n = 14/13/14 rats) and STZ rats (white diamonds; n = 15/14/19 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. (B) BW (g) in CON (black diamonds; n = 14/13/14 rats) and STZ rats (white diamonds; n = 15/14/19 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. Statistical analysis was performed with a two-way anova followed by a post hoc Tukey's multiple comparison test. Statistically significant differences between CON and STZ rats of the same age are indicated as ***, p < 0.001. Statistically significant differences between CON-4 weeks and CON-8 weeks or between STZ-4 weeks and STZ-8 weeks are indicated as ^^, p < 0.01; ^^^, p < 0.001. Statistically significant differences CON-4 weeks and CON-12 weeks or between STZ-4 weeks and STZ-12 weeks are indicated as & , p < 0.05 and &&& , p < 0.001. (C) HW (g) in CON (black diamonds; n = 14/13/14 rats) and STZ rats (white diamonds; n = 15/14/19 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. (D) mRNA expression of ANF normalized to GAPDH measured in CON (black diamonds; n = 9/9/9 rats) and STZ rats (white diamonds; n = 9/10/9 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. Two-way anova analysis showed no interaction between STZ treatment and the time; however, STZ treatment alone had a statistically significant effect on HW and ANF expression (p < 0.0001). All data represent the mean ± S.E.M. HW, heart weight; ANF, atrial natriuretic factor; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase. CaMKII is another cardiac kinase exhibiting a key role in tuning ECC. 15 One band migrating at approximately 50 kDa was identified for CaMKII in rat hearts ( Figure 3A; Figure S7).
Each cardiac sample was then normalized to calsequestrin and/ or actin, since the expression of either protein was similar between STZ-treated rats and their age-matched CON ( Figure 3A). As shown in Figure 3B,C,D, no statistical difference was observed in the cardiac expression of PKA RI, PKA RIIα and PKA Cα between control and diabetic rats, regardless of the disease stage. In contrast, CaMKII expression was strongly increased in diabetic hearts at 12 weeks compared with their age-matched controls (p < 0.001; Figure 3E). In healthy myocardium, PLB is phosphorylated at Ser 16 by PKA and at Thr 17 by CaMKII. 38,39 Phosphorylation of TnI by PKA and PKC occurs at both Ser 23 and Ser 24 and leads to the depression of myofilament Ca 2+ sensitivity. [40][41][42][43][44][45][46] We tested whether these phosphorylation events were altered in the diabetic myocardium. Proteins were extracted from hearts of CON and STZ-treated rats at 4, 8 and 12 weeks, and the expression of p-TnI (Ser 23 / 24 ), p-PLB (Ser 16 ) and p-PLB (Thr 17 ) was determined by western blotting as shown in Figure 5A. No significant differences were noted in the p-TnI  Figure S11).

| DISCUSS ION
The major goal of the present study was to determine whether the differential and time-specific changes that we have previously TA B L E 1 Anatomical data of control (CON) and diabetic (STZ) rats at 4, 8 and 12 weeks after injection of either streptozotocin or vehicle.  . Epac1 mRNA expression in CON (black diamonds; n = 6/6/6 rats) and STZ rats (white diamonds; n = 6/6/6 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. (B) Epac2 mRNA expression in CON (black diamonds; CON: 6/9/6 rats) and STZ rats (white diamonds; STZ: 6/7/6 rats) at 4, 8 and 12 weeks after STZ or vehicle injection. mRNA expression of both isoforms was normalized to GAPDH in both STZ and aged-matched CON rats. Two-way anova test showed no significant interaction between STZ treatment and time; however, STZ treatment alone had a statistically significant effect on Epac1 mRNA expression (p < 0.0001). A two-way anova followed by a post hoc Tukey's multiple comparison test was performed to assess the changes in the expression of Epac2 mRNA. Statistically significant differences between CON and STZ rats of the same age are indicated as *, p < 0.05. Statistically significant differences between STZ-4 weeks and STZ-12 weeks are indicated as && , p < 0.01. Equal amounts of cardiac proteins from control (CON) and diabetic rats (STZ) were separated on SDS/PAGE and revealed with Epac2-specific antibody. Calsequestrin was used as a loading control. (C) Shown is a representative blot for Epac2 in CON and STZ at 4, 8 and 12 weeks. (D) Quantification of all data obtained in several immunoblots from hearts of CON (black diamonds; n = 4/6/4) and STZ (white diamonds; n = 3/5/4) at 4, 8 and 12 weeks, respectively, and represented as mean ± S.E.M. Statistical analysis was performed with two-way anova tests that showed no significant differences in Epac2 mRNA expression between CON and STZ rats. respectively) between CON and STZ rats. In (E), two-way anova followed by a post hoc Tukey's multiple comparison test was performed to assess the changes in the expression of CaMKII (CON: n = 9/7/7; STZ: n = 5/6/6 at 4, 8 and 12 weeks, respectively) that were dependent on both the time and STZ treatment as well as on the interaction of both factors. Statistically significant differences between CON and STZ rats of the same age are indicated as ***, p < 0.001. Statistically significant differences between STZ-4 weeks and STZ-12 weeks are indicated as &&& , p < 0.001. Statistically significant differences between STZ-8 weeks and STZ-12 weeks are indicated as # , p < 0.05.

F I G U R E 4
Expression of the major actors of the cardiac ECC in hearts from control and diabetic rats at 4, 8, and 12 weeks. Total RNA was extracted from hearts of 5 to 8 control (CON, black diamonds) and diabetic rats (STZ, white diamonds) at each time point and analysed by Real-time PCR for SERCA2a (A), PLB (B) and TnI (C). mRNA expression was normalized to GAPDH in both STZ and aged-matched CON rats. Data represent the mean ± S.E.M. Statistical analysis was performed with two-way anova that showed no significant interaction between STZ treatment and time for all proteins. However, STZ treatment had the same statistically significant effect on PLB (p < 0.05) across time.
progressed, cardiac function along with β 1 -AR transcripts and total basal cAMP levels were normalized, whereas PDE3A protein expres- were not explored, our results suggest that the genes encoding for Epac1 and Epac2 are subjected to time-specific and differential regulation of their transcription and translation. Furthermore, one cannot exclude the fact that the twelve-week timeline followed in this study to assess the molecular changes underlying the progression of DCM lies most probably within the pathophysiological window that precedes any alterations occurring in Epac2 protein levels.
Moreover, potential specific changes in the expression of Epac1/2 proteins may occur in subcellular cardiac compartments that would require more precise investigations. 47,48 Indeed, Epac1 was shown previously to mediate the pro-hypertrophic effects of β-AR stimulation 49 through either its recruitment to β 1 -ARs via β-arrestin2 47 or its association with PDE4D3, RyR2, calcineurin and the extracellular signal-regulated kinase 5 (ERK5) in a mAKAP (muscle A kinaseanchoring protein)-coordinated signal transduction complex at the nuclear envelope. 50 Moreover, Epac2 has been suggested to induce SR Ca 2+ leak and arrhythmia, an effect that seems to be mediated by This compartment-specific loss of PKA, which was reflected by reduced phosphorylation of discrete substrates and was attributed to contractile impairment, points at the potential remodelling of cAMP/ PKA pools in the diabetic myocardium. Although we did not measure PKA activity in this study, the unchanged PKA-dependent phosphorylation of PLB and TnI that we report herein could speak for a preserved PKA activity in diabetic hearts at least in the vicinity of these two targets. Indeed, similar basal PKA activity was reported in the heart of rats at 6 weeks after induction of T1D. 52 Nevertheless, controversial findings were also reported concerning both basal and stimulated cardiac PKA activity in STZ-induced T1D, which might be due to species differences and the duration of diabetes. 33,53 CaMKII, a critical transducer of Ca 2+ signalling, is a multifunctional protein kinase that phosphorylates a wide range of substrates and regulates numerous cellular functions, including ECC. 35,54 CaMKII has been proposed as a key contributor to the deleterious effects of chronic β-AR activation in DCM, primarily by exacerbating RyR2-mediated diastolic Ca 2+ leak. 55,56 Therefore, we determined whether the expression of CaMKII is altered in the diabetic myocardium during the progression of DCM. A single immunoreactive band migrating at ~50 kDa was detected in rat hearts for CaMKII. CaMKII δ isoform is the predominant isoform in the heart, including in human myocardium whereas the ɣ subunit is only expressed at low levels in cardiomyocytes. 54 Interestingly, 12 weeks after STZ injection, DCM was associated with increases in CaMKII protein levels and CaMKIIdependent phosphorylation of PLB. The rate of PLB phosphorylation may be the result of a balance between an upregulation in the kinase activity, which has been reported in DCM through phosphorylation, oxidation and O-GlcNAcylation in diabetic hearts from rodents and human patients [57][58][59][60] and changes in phosphatases activity. 61,62 In DM, the process of cardiac Ca 2+ cycling is modified in both humans and animal models, contributing to impaired cardiac contraction and relaxation. 30  In the present study, we report similar transcripts and protein levels of SERCA2a in controls and diabetic hearts at 4, 8 and F I G U R E 6 Summary of cAMP and Ca 2+ signalling pathways remodelling during DCM. We have previously shown that 4 weeks after STZ injection, a modest increase in cAMP content, although not significant, resulted from a balance between β 1 -AR receptors, PDE4B and PDE4D whose expression was upregulated in the diabetic heart. 37 At Weeks 8 and 12, β 1 -AR receptors and cAMP content were restored while the expression of PDE3A was increased. 37 In the present study, Epac2 protein levels as well as the expression of PKA subunits (RI, RIIα and Cα) remained unchanged with the progression of DCM, whereas CaMKII increased at Week 12 in the diabetic myocardium. Moreover, no changes were observed in the protein levels of PLB, SERCA2a and TnI irrespective of the disease evolution. Importantly, PLB phosphorylation at threonine-17 was increased in DCM, whereas phosphorylation of both PLB at serine- 16 (Figure 6).
To the best of our knowledge, this is the first study that delineates the intricate regulations of cAMP effectors and the major actors of ECC in the myocardium during the pathophysiological progression of T1D-induced DCM. These novel results are to be considered with respect to the previously documented changes affecting β-AR/cAMP/PDEs signalling in the same animal model 37 although they present some limitations. Importantly, the differential and time-specific changes in cardiac Epac1/2, PKA subunits, CaMKII and the phosphorylation status of PLB and TnI in diabetic myocardium that we delineate herein at the level of the whole heart, suggest potential modifications in the regulation of specific subcellular cAMP and Ca 2+ pools, which would require further characterization for a better understanding of how cAMP and Ca 2+ compartmentalization is remodelled in DCM. Moreover, more subtle and specific changes in the phosphorylation status of some ECC actors may be unmasked upon ß-AR stimulation with respect to basal conditions. One should note as well the potential confounding effects when assessing gene and protein expression in the whole heart versus cardiac myocytes. Indeed, the heart is comprised of a syncytium of cardiac myocytes and surrounding nonmyocytes, the majority of which are cardiac fibroblasts. In response to stress, cardiac myocytes become hypertrophic and can change their electrical properties, whereas fibroblasts convert into 'activated' myofibroblasts, proliferate and enhance ECM deposition, which leads to cardiac fibrosis. 72 The latter affects cardiac myocyte metabolism and performance and ultimately ventricular function. 73 We have previously documented the emergence of fibrosis in the diabetic myocardium 12 weeks following STZ injection. 37 Accumulating evidence now suggests that downstream cAMP effectors such as PKA and Epac as well as Ca 2+ signalling including CaMKII modulate a variety of fundamental cellular processes involved in fibroblasts and fibrosis-associated cardiac diseases. 74-76

| CON CLUS ION
In light of the novel data we provide herein, we conclude that abnormalities affecting cAMP effectors, CaMKII and Ca 2+ proteins handling in DCM are complex and involve differential and time-specific regulations. Our results suggest that more subtle and fine functional changes may occur in specific compartments within the cell. Thus, a fine characterization of these specific defects affecting both cardiac cAMP and Ca 2+ signalling is crucial for a better understanding of the pathophysiology of DCM, its progression and management through the identification of new therapeutic targets. Aniella Abi-Gerges: Conceptualization (lead); data curation (lead); funding acquisition (lead); methodology (lead); project administration (lead); resources (lead); supervision (lead); validation (lead); writing -original draft (lead); writing -review and editing (lead).

ACK N O WLE D G E M ENTS
This study was jointly funded with the support of the National Jean Karam for their skilful technical assistance.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that there are no conflicts of interest. Magali Samia EL HAYEK is currently an employee of Eli Lilly and Company.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are available upon request.