Gestational diabetes triggers postpartum cardiac hypertrophy via activation of calcineurin/NFAT signaling

Population-based studies identified an association between a prior pregnancy complicated by gestational diabetes mellitus (GDM) and cardiac hypertrophy and dysfunction later in life. It is however unclear whether GDM initiates this phenotype and what are the underlying mechanisms. We addressed these questions by using female rats that express human amylin (HIP rats) as a GDM model and their wild-type (WT) littermates as the normal pregnancy model. Pregnant and two months postpartum HIP females had increased left-ventricular mass and wall thickness compared to non-pregnant HIP females, which indicates the presence of concentric hypertrophy. These parameters were unchanged in WT females during both pregnancy and postpartum periods. Hypertrophic Ca2+-dependent calcineurin/NFAT signaling was stimulated two months after giving birth in HIP females but not in the WT. In contrast, the CaMKII/HDAC hypertrophy pathway was active immediately after giving birth and returned to the baseline by two months postpartum in both WT and HIP females. Myocytes from two months postpartum HIP females exhibited slower Ca2+ transient relaxation and higher diastolic Ca2+ levels, which may explain calcineurin activation. No such effects occurred in the WT. These results suggest that a GDM-complicated pregnancy accelerates the development of pathological cardiac remodeling likely through activation of calcineurin/NFAT signaling.


Results
Gestational diabetes in HIP rat females. HIP rats 28,30,31 are obese Sprague-Dawley rats with pancreatic β-cell specific expression of the human variant of amylin, a pancreatic hormone that is processed in the same secretory vesicles as insulin and co-secreted with it 32 . At 5-6 months of age, amylin level is ~ fourfold higher in plasma from heterozygous HIP versus WT females ( Supplementary Fig. 1A). Amylin acts as a satiation agent by activating receptors in the brain and slows down the gastric fluxes 33,34 . In agreement with this function, HIP females had lower body weight compared to their WT littermates ( Supplementary Fig. 1B). Moreover, amylin was previously shown to reduce insulin-stimulated glucose uptake in muscle 35,36 . Indeed, HIP females (5-6 months of age) showed impaired glucose tolerance compared to age-matched WT females (Fig. 1A) while fasting blood glucose level was similar (see time 0 in Fig. 1A). To avoid confounding factors introduced by these differences between WT and HIP females at baseline, we designed the study to compare longitudinally females that went through pregnancy with age-and genotype-matched females that were never pregnant.
At this stage, HIP and WT females were randomly assigned to pregnancy and control study groups and further randomized for euthanasia and heart collection at 1 day or 2 months postpartum. Glucose tolerance decreased during pregnancy in both WT and HIP females, with HIP females remaining significantly glucose intolerant compared to the WT ( Supplementary Fig. 2, top: time course of blood glucose level in glucose tolerance tests; Fig. 1B,C: area under the curve in these glucose tolerance tests). By the time of weaning, glucose tolerance returned to the pre-pregnancy level in HIP females ( Supplementary Fig. 2, middle; Fig. 1B) and even improved compared to the control group in WT females ( Supplementary Fig. 2, middle; Fig. 1C). In HIP females, glucose tolerance tended to worsen again by two months postpartum ( Supplementary Fig. 2, bottom; Fig. 1B). The fasting blood glucose level in two months postpartum HIP females was similar to that measured at baseline (Fig. 1D), indicating that they were still in a pre-diabetic state. In fact, fasting blood glucose was similar in all HIP and WT groups throughout the study (see time 0 in Supplementary Fig. 2). Glucose tolerance did not change significantly over this time in either WT or HIP females from the control, non-pregnant group (Fig. 1B,C). Thus, heterozygous HIP females display the typical glucose intolerance of GDM. WT females were used a model of normal, uncomplicated pregnancy.
Cardiac hypertrophy after a GDM-complicated pregnancy in rat females. Transthoracic echocardiography was used to monitor heart size and function in WT and HIP females from the control and pregnancy groups at baseline, late pregnancy and two months postpartum (Tables 1, 2, Fig. 2A-F, Supplemental Fig. 3). In HIP females, the left-ventricular mass (Fig. 2B), thickness of interventricular septum (Fig. 2D) and thickness of the left-ventricular posterior wall (Fig. 2F) were significantly increased in the pregnancy versus control group in late pregnancy and remained elevated two months after giving birth. In contrast, heart size was not significantly changed during pregnancy and postpartum periods in WT females from the pregnancy group compared to the WT control group ( Fig. 2A,C,E).
In agreement with the echocardiography data, the heart weight-to-body weight ratio was elevated two months after giving birth in HIP females from the pregnancy group compared to HIP females from the control group, while no significant differences occurred in WT females (Fig. 2G,H). Moreover, the level of the hypertrophic marker ANP was increased in hearts from postpartum HIP females but not in postpartum WT females compared www.nature.com/scientificreports/  www.nature.com/scientificreports/ to their respective non-pregnant controls (Fig. 2I). The BNP level did not change with pregnancy in either HIP or WT females (Fig. 2J). Together, these data indicate the presence of cardiac hypertrophy in the postpartum period following a GDMcomplicated pregnancy (HIP rats) but not a normal pregnancy (WT rats).
Enhanced activity of calcineurin/NFAT hypertrophy signaling in postpartum HIP rat females following a GDM-complicated pregnancy. Since the calcineurin/NFAT hypertrophy pathway is activated in both type-2 diabetes and early stages of normal pregnancies, we investigated whether enhanced calcineurin/NFAT signaling contributes to the cardiac hypertrophy following a GDM-complicated pregnancy. In this pathway, Ca 2+ /calmodulin-dependent activation of the phosphatase calcineurin leads to de-phosphorylation of NFAT, causing its translocation into the nucleus where it activates gene transcription 37 . Thus, the activity of calcineurin/ NFAT pathway was assessed from the nuclear versus cytosolic localization of NFATc4 as measured by immunofluorescence in isolated cardiac myocytes ( Fig. 3A-C). There was a tendency towards lower nuclearto-cytosolic NFATc4 ratio immediately (within 1 day) postpartum in hearts from both WT and HIP females compared to the respective control groups ( Fig. 3A-C), in agreement with data indicating downregulation of this signaling pathway in late pregnancy 18,29 . Two months after giving birth, this pathway returned to its baseline activation level in WT females (Fig. 3B). However, the ratio of nuclear-to-cytosolic NFATc4 was significantly larger in myocytes from two months postpartum HIP females compared to control (Fig. 3A,C), indicating reactivation of this hypertrophy pathway following a GDM-complicated pregnancy.
To further evaluate the activity of calcineurin/NFAT pathway, we used immunoblot to measure the expression of calcipressin-1 (also known as RCAN1 or MCIP1), an endogenous calcineurin inhibitor whose expression is under the control of calcineurin/NFAT [38][39][40] . Calcipressin-1 expression was comparable in hearts from control, 1 day postpartum and two months postpartum WT females (Fig. 3D,E). In contrast, calcipressin expression was mildly but significantly elevated in hearts from two months postpartum HIP females (Fig. 3D,F).
Since calcineurin is activated by an increase in cytosolic Ca 2+ concentration, we measured Ca 2+ transients triggered by field stimulation at various frequencies between 0.2 and 2 Hz in myocytes from control, one day postpartum and two months postpartum WT and HIP females (Fig. 4A,B). Ca 2+ transient amplitude was not significantly affected by the pregnancy and postpartum stages in either HIP or WT females (Fig. 4C,D). However, Ca 2+ declined more slowly in myocytes from 2 months postpartum versus control, not-pregnant HIP females (Fig. 4E), which resulted in a more pronounced rise in diastolic Ca 2+ levels upon an increase in the stimulation frequency (Fig. 4G). In contrast, no such differences occurred in myocytes from postpartum and control WT females (Fig. 4F,H).

Activation of CaMKII/HDAC hypertrophy signaling in pregnant rat females. CaMKII/HDAC
signaling is another Ca 2+ -activated hypertrophy pathway in which, upon phosphorylation by CaMKII, HDAC moves out of the nucleus, which facilitates gene expression 41 . Using immunofluorescent staining of isolated myocytes, we found that HDAC4 is exported from the nucleus in hearts from WT and HIP females immediately after giving birth (Fig. 5), which suggests that the CaMKII/HDAC pathway is activated in both normal pregnancy and GDM. However, this pathway returned to its baseline activation level by two months postpartum in both WT and HIP females (Fig. 5).

Discussion
Several population-based retrospective studies in large human cohorts established a strong association between GDM and the development of cardiovascular disease later in life 7-13 , including cardiac remodeling and dysfunction 12,14,15 . Using female rats that express the human variant of the pancreatic hormone amylin specifically in the β-cells (HIP rats) as a GDM model and their WT littermates as controls, we found here that pathological cardiac hypertrophy is also present two months post-delivery in female rats with GDM-complicated pregnancies Table 2. Echocardiography parameters in HIP females from control and pregnancy groups. Measurements were performed at baseline, during late pregnancy and two months after giving birth. * P < 0.05, ** P < 0.01, and *** P < 0.001 vs. control group. www.nature.com/scientificreports/ www.nature.com/scientificreports/ but not in females with normal pregnancies (Fig. 2). Human studies cannot discriminate between GDM as a cause of the ulterior cardiovascular disease or a condition that develops in, and thus identifies, women that already are at higher risk. Since we compared postpartum HIP females with HIP females that did not go through pregnancy, our data suggest that GDM triggers, or at minimum accelerates, the postpartum development of pathological cardiac remodeling/ hypertrophy. During a normal pregnancy, the heart generally undergoes physiological, mild eccentric hypertrophy, characterized by a proportional increase in chamber size and wall thickness 18,42-44 , as expected for volume www.nature.com/scientificreports/ overload-induced heart growth. However, one study reported increased thickness of the septal and posterior walls with no change in diastolic diameter in pregnant women 16 , while an investigation in pregnant mice found hypertrophy characterized by decreased left-ventricular wall thickness and larger chamber size 19 . Somewhat surprisingly, pregnant WT females did not show overt cardiac remodeling or hypertrophy in our study. The reasons for this are unclear but might reflect reduced ability of the heart to undergo physiological hypertrophy due to a more advanced age (~ 5.5 months of age). This is in line with studies reporting lack of exerciseinduced hypertrophy in aged animals 45,46 . In contrast, our data suggest that GDM is associated with concentric  www.nature.com/scientificreports/ remodeling/hypertrophy, both at term and postpartum, as HIP females showed increased diastolic thickness of both interventricular septum (Fig. 2D) and posterior wall (Fig. 2F) with no significant change in left-ventricular diameter ( Table 2) compared to HIP females that did not experience pregnancy. This is consistent with human data showing larger left-ventricular wall index at term in females with GDM compared to females with normal pregnancies 14,15 . Cardiac hypertrophy observed two months postpartum in female rats with GDM was associated with activation of calcineurin/NFAT signaling, as evidenced by the nuclear translocation of NFATc4 (Fig. 3A,C) and increased expression of calcipressin-1, a protein whose expression is often used as an indicator of NFAT activation by calcineurin (Fig. 3F). Calcineurin is activated by an increase in cytosolic Ca 2+ through binding of Ca 2+ / calmodulin. We found that Ca 2+ transient decay is slower and consequently diastolic Ca 2+ is higher in myocytes from two months postpartum versus control, not-pregnant HIP females (Fig. 4E). Thus, following each heartbeat Ca 2+ stays in the cytosol longer, which may underlie calcineurin activation.
The calcineurin/NFAT pathway is usually involved in pathological hypertrophy, caused for example by pressure overload. However, this pathway is also activated in the early stage of normal pregnancies and is required for the pregnancy-induced heart growth 18,29 . Activation of calcineurin in early pregnancy was attributed to higher progesterone levels 29 , but the underlying mechanisms are not fully elucidated. We previously reported that calcineurin/NFAT signaling is activated in pre-diabetic, insulin-resistant male HIP rats through a mechanism related to cardiac deposition of aggregated amylin 28 , a hormone that is hypersecreted along with insulin by the pancreatic β-cells in pre-diabetes. HIP rats express the human isoform of amylin in the pancreas. In contrast to www.nature.com/scientificreports/ rodent amylin, the human variant is amyloidogenic 32 , which favors amylin aggregation and deposition in the pancreas and peripheral organs, including the heart 28,47,48 . In pre-diabetic HIP rats, amylin deposition leads to an increase in the sarcolemmal permeability to Ca 2+ and thus raises cytosolic Ca 2+ level, which activates calcineurin 28 . Females with GDM exhibit glucose intolerance, which is usually associated with an increase in insulin and amylin secretion. Thus, the cardiac stress caused by higher levels of circulating amyloid-forming amylin may also contribute to calcineurin/NFAT activation in females with GDM. CaMKII/HDAC signaling is another Ca 2+ -dependent pathway that leads to expression of pro-hypertrophic genes. In contrast to calcineurin signaling, this pathway was active immediately after giving birth and returned to baseline by two months postpartum in both HIP and WT females (Fig. 5). This result suggests that the CaM-KII/HDAC pathway may contribute to cardiac hypertrophy during pregnancy but not to the GDM-induced programing of heart growth later in life.
One limitation of the current study is that similar experiments cannot be performed in humans, which leaves open the question whether the mechanism identified here does account for the postpartum cardiac hypertrophy in women with prior GDM. Partial mechanistic validation could come however from studies in additional animal models. Moreover, the mechanism needs to be further validated by experiments where the calcineurin/NFAT pathway is inhibited during pregnancy.
In summary, we found that GDM results in concentric cardiac hypertrophy, likely due to activation of calcineurin/NFAT pathway, two months after giving birth in female rats. The comparison of postpartum HIP females with non-pregnant HIP females suggests a causative role for GDM in the postpartum development of pathological remodeling of the heart. Understanding the mechanisms through which GDM predisposes the mother to heart dysfunction, coupled with early detection of GDM, will allow clinicians to design effective lifestyle and/or pharmacological interventions during the pregnancy and postpartum periods to reduce the risk.

Methods
The study is reported in accordance with ARRIVE guidelines.

Experimental animals.
All animal experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the University of Kentucky. N = 53 obese Sprague-Dawley female rats that are heterozygous for expression of the human isoform of the pancreatic hormone amylin specifically in the β-cells (HIP rats 28,30,31 ) and N = 40 wild-type (WT) littermates were used in this study. HIP (25.4 ± 0.3 weeks of age) and WT (24.6 ± 0.5 weeks of age) females were randomly assigned to pregnancy and control study groups. Females in the pregnancy groups were paired for breeding with Sprague-Dawley males. N = 6 females assigned to pregnancy groups did not become pregnant and were excluded from the study. Females in both pregnancy and control groups were further randomized for euthanasia and heart collection at one day or two months postpartum. Rats had ad libitum access to food and water. At the end of the study, rats were anesthetized with 3-5% isoflurane (100% O 2 ) and deep anesthesia was verified by lack of reflex upon toe pinch. Rats were then euthanized by exsanguination following excision of the heart.
Glucose tolerance test. Standard glucose tolerance tests (GTT) were performed at baseline and when the pregnancy groups were at 18-20 days of pregnancy, 21 days postpartum and 2 months postpartum. Rats were fasted for 8 h, injected with glucose (2 g/kg of body weight, IP) and blood glucose was monitored every 15-30 min for two hours using a glucometer (OneTouch Ultra).
Echocardiography. Transthoracic echocardiography was performed under anesthesia (0.5-3% isoflurane) at baseline and when the pregnancy groups were at 18-20 days of pregnancy and two months postpartum using a Vevo 2100 high-frequency ultrasound (VisualSonics, Toronto, ON, Canada). Two-dimensional imaging was used to identify the short-axis position. Five consecutive M-mode images in the short-axis view were then used for analysis of chamber size and heart function.
Immunoblot. Hearts were homogenized in homogenization buffer containing 150 mM NaCl, 50 mM Tris-HCl, 50 mM NaF, 2% Triton X-100, 0.1% SDS, supplemented with phosphatase and proteases inhibitor cocktail set III (Calbiochem). Homogenates were then loaded onto polyacrylamide gels for SDS-PAGE electrophoresis, transferred to PVDF membranes, blocked with 5% milk and probed with primary antibodies against atrial natriuretic factor (ANP; Millipore AB2232, 1:1000), brain natriuretic peptide (BNP; Abcam ab239510, 1:1000), and calcipressin-1 (ThermoFisher 14869-1-AP, 1:1000). Equal protein loading was verified by re-probing for GAPDH (Abcam ab8245). Bands were detected by chemiluminescent signals using the enhanced chemiluminescence method (SuperSignal West Dura Extended Duration Substrate, ThermoScientific, USA) and visualized with a G:BOX gel imaging system (SynGene, Cambridge, United Kingdom). Band intensity was measured using ImageJ software (NIH, Bethesda, USA). For each gel, the signal intensity was averaged over the control samples. Then, the signal intensity in all lanes was normalized to this average. This procedure was repeated on all technical replicates and the normalized signal intensity was averaged for each sample, followed by averaging over experimental groups. Ventricular myocyte isolation. Rats were anesthetized with 3-5% isoflurane (100% O 2 ) and hearts were excised by cutting the aorta, mounted on a gravity-driven Langendorff perfusion system and perfused with a nominally Ca 2+ free medium containing 11.2 g/L Minimum Essential Medium (MEM; Sigma M0518), 40 www.nature.com/scientificreports/ Units/L human insulin, 10 mL/L Penicillin-Streptomycin, 4.8 mM NaHCO 3, 2 mM sodium pyruvate, 10 mM Na-HEPES, 10 mM HEPES and 3.5 μL/mL heparin (pH = 7.4) for 5 min to clear the blood. The heart was then perfused with a similar medium except that heparin was excluded and 0.1 mg/mL Liberase TH (Roche-Sigma), 30 μM CaCl 2 and 8 mM taurine were added. When the heart became flaccid (~ 20 min), the tissue was cut into small pieces, dispersed, and filtered and the myocyte suspension was rinsed several times. Myocytes were kept at low Ca 2+ (30 μM) until ready to use.
Immunofluorescence. Freshly isolated myocytes were plated onto laminin-coated 8-well glass coverslips and fixed with 4% paraformaldehyde. Cells were then permeabilized with 50 µg/mL saponin (15 min www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.