The impact of empagliflozin on cardiac physiology and fibrosis early after myocardial infarction in non-diabetic rats

All animals underwent baseline cardiac function evaluation by echocardiography, blood pressure (BP) and kidney function. Subsequently, the animals underwent MI procedure. Then, we divided the animals into 2 experimental groups. The first group of untreated rats served as control; the second group received empagliflozin 30 mg/kg/day. Therapy was administered for four consecutive weeks.

We monitored BP once a week and cardiac function was re-evaluated every 2 weeks (week 2 and 4), and kidney function was re-evaluated at the end of the experiment (week 4). Right before sacrifice, we measured cardiac hemodynamics directly by means of the Millar pressure–volume (P–V) system. Finally, after euthanasia, the heart was excised and preserved for additional analyses.

We performed echocardiographic scans as previously described [18]. Briefly, under light sedation with 29 mg/kg ketamine and 4.3 mg/kg xylazine, the rats were placed in a left decubitus position and scanned via a commercial echo-scanner (Vivid i). Echocardiography at frequency of 9 MHz, depth 2.5 cm and frame rate 315 frames/sec, was used to scan two parasternal short axis sections at the apical (AP) and papillary muscle (PM) levels. Two-dimensional (2D) echocardiography scans took place at baseline before MI, 2 weeks after surgery and at the end of the experiment (4 weeks post-MI). The echocardiographic analysis included diastolic and systolic structural parameters. For ventricular function assessment, we calculated three parameters: (1) Fractional shortening (FS) which takes into consideration a 2D cross-section of the heart. Specifically, in the M-mode cines, the left ventricle (LV) intra-ventricular diameters were measured at systole and diastole (LVIDs and LVIDd, respectively), and FS was calculated as follows: FS = (LVIDd-LVIDs)/LVIDd; (2) Fractional area change (FAC) was calculated as the change in LV area during cardiac cycle. Particularly, LV end diastolic area (LVEDA) and LV end systolic area (LVESA) are measured in 2D cines and FAC were calculated as follows: FAC = (LVEDA-LVESA)/ LVEDA; (3) EF was determined by the Vivid i LV function software. The echocardiographic analysis was performed for each level (AP and PM) separately, since the extent of cardiac damage is level specific.

We monitored BP in conscious rats by a validated tail-cuff plethysmography method using CODA non-invasive BP system (Kent Scientific Corporation, Torrington CT, USA).

Each rat was placed for 24 h in a metabolic chamber (Techniplast S.p.A., Buguggiate, Italy) in which urine was collected for the determination of its volume, and glucose concentration. Subsequentially, blood sample was withdrawn from the tail vein to asses blood glucose and creatinine levels.

Rats underwent MI procedure, as previously described by Bhindi [19]. Briefly, we intubated the rats under deep anesthesia with a mixture of 87 mg/kg ketamine and 13 mg/kg xylazine and ventilated them at a rate of 80–90 breaths per minute, and 1 to 2 ml/100 gr tidal volume. Using intercostal space left thoracotomy, the chest was opened, and the pericardial sac was dissected. A stich was placed through myocardium at a slightly greater depth than the perceived level of the left anterior descending (LAD) artery. Next, we tightened the suture to ensure complete LAD occlusion, the chest was closed, the skin stitched, and the rat was placed in its cage for recovery.

Empagliflozin powder was generously provided by Boehringer-Ingelheim GmbH, Germany. Drug dosage of 30 mg/kg/day was adopted from Zhou and Wu, as they showed that this rather high dosage affected cardiac physiology as well as cellular biochemistry [20]. Accordingly, rats of group #2 were weighted weekly to adjust their dosage. First drug bolus was given immediately (~ 10–15 min) after MI by subcutaneous injection of the drug dissolved in 0.5 ml DDW. Then, for consecutive 4 weeks, drug was given by 0.5 ml gavage of empagliflozin dissolved in drinking water.

Direct cardiac hemodynamic parameters measurements were obtained by the Millar P–V system (MPVS-300, Millar Instruments, Houston, TX, USA). The Millar P–V System simultaneously and continuously measures LV pressure and volume from the intact beating heart, producing characteristic P–V loops readings. From the P–V loops the following cardiovascular parameters were derived, systolic and diastolic BPs, heart rate (HR), stroke volume (SV), cardiac output (CO), EF, stroke work (SW), dP/dt_max and dP/dt_min. Briefly, rats were anesthetized with a combination of 87 mg/kg ketamine and 13 mg/kg xylazine and placed on controlled heating pads. Next, the rats were tracheotomized and intubated (0.5 cm long 50 PE tube) to facilitate breathing. The right carotid was exposed and ligated distally, the artery was clamped and incised, a 2-Fr Mikro-Tip® catheter (SPR-838, Millar Instruments, Houston, TX, USA) was advanced through the artery into the LV under pressure control; a ligature was then tightened around the catheter to avoid blood leakage and loss [21]. After stabilization for 5 min, signals were continuously sampled at a sampling rate of 1000 samples/sec by the MPVS-300 system, recorded for 15–20 min, and displayed on a personal computer by the PowerLab System and Chart5 software (AD Instruments, Colorado Springs, CO, USA). At the end of each experiment, 4–6 boluses of 100 μL of hypertonic (30%) saline were injected intravenously, and from the averaged shift of P–V relations, parallel conductance volume (Vp) was calculated by the software and used for the correction of the cardiac mass volume. Thereafter, the catheter was withdrawn, and the animal was sacrificed by overdose anesthetic.

After sacrifice, the heart was excised, and dissected transversally at the PM level and its upper section, i.e., AP was fixed in 4% buffered formaldehyde and then embedded in paraffin for histology.

Five μm sections were stained with hematoxylin and eosin, photographed and analyzed for structural measurements: Cross-sectional area, LV cross-sectional area, LV cavity area, and septal and anterior wall widths using the ImageJ freeware (NIH, Bethesda, MD, USA).

Picro-Sirius red was used to distinct collagen from healthy muscle in the histological sections thus evaluate collagen deposition. Briefly, after de-paraffinization and re-hydration, we rinsed the slides in distilled water and stained nuclei with Weigert’s hematoxylin solution for 5 min. Then, we placed the slides in Picro-Sirius red stain for 60 min. Next, we rinsed in 2 changes of 0.5% acetic acid water followed by dehydration quickly through 3 changes of absolute alcohol and xylene. Finally, slides covered with coverslip using a permanent mounting medium. Collagen area was measured in scar and remote regions under microscope (Nikon Eclipse Ci-L, Nikon Corporation, Tokyo, Japan) using the NIS Elements software (NIS Elements 4.0, Nikon Corporation, Tokyo, Japan).

We assessed the mean collagen area of 6 fields from each slide, for each photograph, the red color (collagen), the total tissue area and total picture size were measured. Subsequently, we calculated the percentage of the collagen volume fraction (CVF) according to the formula:

Sections of myocardial tissue were dewaxed and rehydrated in a graded alcohol series. Antigen retrieval was achieved by heating the specimens for 10 min at 94 °C in citric acid buffer (0.01 mol/l; pH 6.0). Sections were then incubated with 3% hydrogen peroxide at room temperature for 20 min to inactivate endogenous peroxidase activity. After rinsing with PBS buffer (pH 7.2), primary antibodies against TGFβ1 (1:200; cat. no. Ab92486; abcam, Cambridge, MA, USA) or Smad3 (1:200; cat. no. Ab40854; abcam, Cambridge, MA, USA), were added and incubated at room temperature for 1 h. Following further rinsing, anti-rabbit horseradish peroxidase-conjugated secondary antibody (1:1,000; cat. no. D13-6; GBI Labs, Bothell, WA, USA) was added and incubated at room temperature for 15 min. Samples were subsequently incubated with DAB solution at room temperature for 6 min until color development. A total of 5 different fields of view were randomly selected under a light microscope with magnification of × 400. The average absorbance of TGF-β1 was analyzed using the NIS-Elements Software BR analysis system version 4.10.00 (NIS Elements 4.0, Nikon Corporation, Tokyo, Japan).

Collagen concentration was estimated in remote MI LV muscle samples by measuring the hydroxyproline (HOP) concentration [22]. Briefly, frozen cardiac samples were minced and then hydrolyzed at 120℃ for 20 min in 1N HCl, 450 μl of Chloramine-T was added to the hydrolysate, oxidation continued for 25 min in RT. We added 500 μl of Ehrlich’s aldehyde reagent to the samples and incubated at 65℃ for 20 min. Optical density was measured at 550 nm. All samples were measured in duplicates.

Data are presented as mean ± SD. Comparisons between groups were performed by 2-way analysis of variance (ANOVA) with repeated measures, in which the treatment and the time point were the independent variables. Whenever the ANOVA was significant, a multiple comparison was performed using the Holm-Sidak as post-hoc test. Single measured data i.e., final measurements, P–V loop derived parameters, and histological analysis were compared between groups using student's t-test. P value of < 0.05 was considered significant.

Tables 2, 3 and 4 summarize the echocardiographic parameters obtained at baseline, 2 weeks, and 4 weeks (at animals' sacrifice) post-MI, respectively. At baseline, cardiac structure and function were typical of healthy young rats, i.e., somewhat smaller areas of the AP level vs. the PM level attributing for the conical shape of the LV (Table 1). The AP level is the cardiac territory mostly affected by MI, and therefore herein we describe relevant changes of echocardiography at both levels, distinctively.

Two weeks post-MI, at both cardiac levels (AP and PM) and in both experimental groups (control and empagliflozin treated), LV size increased (P < 0.001 vs. baseline. Table 3), and the functional parameters, i.e., FS, FAC and EF decreased (P < 0.001 vs. baseline, Table 3). Collectively, these results indicate that we created an extensive MI.

Four weeks post-MI, at both cardiac levels (AP and PM) and in both experimental groups (control and empagliflozin treated), LV structure remained enlarged in comparison to the basal measurement; however, it is comparable to the two weeks measurement, i.e., no additional significant enlargement or functional deterioration was found (Table 4).

Weekly systolic and diastolic BPs and HR are presented in (Fig. 1a–c), respectively, all parameters were comparable between groups throughout the 4 weeks experiment duration (Table 4). Final systolic and diastolic BPs, as well as the parameters derived from the P–V loops recorded by the direct Millar system, are presented in Table 5. All parameters were indistinguishable different between groups, showing no effect of the treatment on cardiac function or BP.

Urine volume, glucose concentration and blood creatinine data, as indicators of empagliflozin effect 4 weeks post-MI, are presented in (Fig. 2). As expected, empagliflozin treatment increased daily urine output, (P = 0.06), decreased blood glucose and increased urine glucose output, without affecting creatinine clearance (Fig. 2).

Gross histological analysis at the PM level reveals a comparable size of the cardiac muscle in both groups (Fig. 3). Specifically, the average total cardiac muscle cross sectional area was 95.2 ± 24.7 mm² and 89.2 ± 9.3 mm², of which LV occupied 57.4 ± 13.6 mm² and 55.9 ± 11.3 mm² in the control and empagliflozin group, respectively (P > 0.05). The LV cavity cross sectional area, on the other hand, was 5.9 ± 2.4 mm² and 11.5 ± 3.8 mm² in the control and the empagliflozin group, respectively (P < 0.05). Thus, while most cardiac structural parameters were comparable, the LV cavity increased in the empagliflozin group.

Figure 4 depicts sections of cardiac muscles stained for collagen content using picro-sirius red. In these representative Figs., collagen occupies large areas of the scar in the control group, seen as red background with yellow patches of non-collagenous tissue (Fig. 4a). Scars are less collagenous in the treated group, seen as less red color in the empagliflozin representative sections, and more yellow, non-collagenous, patches of tissue (Fig. 4b). In remote areas, collagen is sporadic in both groups, with more collagen in the control group and rare red stain in the empagliflozin group (Fig. 4c, d). CVF quantification in the two regions of the heart, i.e. the scar itself and at remote areas, is presented in (Fig. 4e, f), respectively. The results indicate that CVF decreased by ~ 30% in the scar (P < 0.001), and ~ 45% in the remote area (P < 0.05).

Summarizes the HOP concentrations measured at sections of the cardiac muscle. HOP concentrations were 4.1 ± 0.4 µg/µl and 3.6 ± 0.2 µg/µl in the control and empagliflozin treated group, respectively (P = 0.07). These results indicate comparable collagen concentrations in the cardiac muscles of the different experimental groups.

As TGF-β1 and Smad3 are key proteins in the fibrosis pathway, we measured their expressions in remote areas of the cardiac muscle by immunohistochemical staining. The representative specimens in (Fig. 5) show reduced expression of both TGF-β1 and Smad3 following empagliflozin treatment (Fig. 5b, d, respectively). Quantification of 5 specimen of each group show that empagliflozin reduced TGF-β1 expression by ~ 49% (P < 0.05, Fig. 5e), and Smad3 expression by ~ 47% (P < 0.05, Fig. 5f).

A summary of the possible pathways involving the SGLT2is efficacy, must include references to the direct physiological aspect of the drugs, i.e. reduction of blood glucose level, increasing urine output, and to possible biochemical pathways such as the STAT3, NHE-1 inhibition and the TGFβ1/Smad3 pathways [27‐29]. Clinically, in diabetic patients, six months of empagliflozin treatment did not significantly improve left ventricular (LV) function, structure, adiposity, or diffuse fibrosis [29]. The DAPA-HF trial [30], and EMPEROR-REDUCED trial [31], showed a significant reduction of cardiovascular adverse events in diabetic as well as non-diabetic patients with reduced LV function. Further, it was established that empagliflozin causes reverse remodeling [32, 33] and improves LV systolic function in HFrEF patients [33].

Thus, it is still not clear whether and how SGLT2is affect the cardiovascular system. However, empagliflozin and other SGLT2is exerts a physiological effect at the renal level, i.e., it increases daily urine output and glucose excretion, lowers blood glucose and body weight gain. The former, reduces hyperglycemia, and attenuate its related oxidant production, i.e. exerts antioxidant efficacy. The latter, affects volume overload reducing cardiovascular general and particularly cardiac stresses, thus attenuate their malfunction remodeling.

Other biochemical pathways were established both in vitro and in vivo, among them the STAT3, cardiac and renal NHE-1 inhibition, and the TGFβ1/Smad3 pathways [34] The STAT3 is known for its anti-inflammation efficacy. Lee et al. [25] have shown in vivo, that the SGLT2i, dapagliflozin, preserved cardiac function and attenuate remodeling in non-diabetic rats, attributing these effect to the inhibition of the above-mentioned mechanism. NHE-1 inhibition, is another mechanism attributed to the SGLT2is, specifically it was shown that those agents inhibit the NHE-1 function, which its function is associated with cardiac function deterioration and remodeling in the diseased heart [27, 28].

We tested the in vivo effect of empagliflozin on the TGFβ1/Smad3 pathway, which is involved in the collagen formation, as was shown by Kang et al. [14]. However, our results failed to show physiological efficacy in our MI non-diabetic rats. Finally, those biochemical pathways were not yet established in diabetic or non-diabetic patients, yet it is reasonable to believe that they do affect patients who receive SGLT2i.