Dapagliflozin alleviates cardiac fibrosis through suppressing EndMT and fibroblast activation via AMPKα/TGF‐β/Smad signalling in type 2 diabetic rats

Abstract Diabetic cardiomyopathy (DCM) is one of the leading causes of heart failure in patients with diabetes mellitus, with limited effective treatments. The cardioprotective effects of sodium‐glucose cotransporter 2(SGLT2) inhibitors have been supported by amounts of clinical trials, which largely fills the gap. However, the underlying mechanism still needs to be further explored, especially in terms of its protection against cardiac fibrosis, a crucial pathophysiological process during the development of DCM. Besides, endothelial‐to‐mesenchymal transition (EndMT) has been reported to play a pivotal role in fibroblast multiplication and cardiac fibrosis. This study aimed to evaluate the effect of SGLT2 inhibitor dapagliflozin (DAPA) on DCM especially for cardiac fibrosis and explore the underlying mechanism. In vivo, the model of type 2 diabetic rats was built with high‐fat feeding and streptozotocin injection. Untreated diabetic rats showed cardiac dysfunction, increased myocardial fibrosis and EndMT, which was attenuated after treatment with DAPA and metformin. In vitro, HUVECs and primary cardiac fibroblasts were treated with DAPA and exposed to high glucose (HG). HG‐induced EndMT in HUVECs and collagen secretion of fibroblasts were markedly inhibited by DAPA. Up‐regulation of TGF‐β/Smad signalling and activity inhibition of AMPKα were also reversed by DAPA treatment. Then, AMPKα siRNA and compound C abrogated the anti‐EndMT effects of DAPA in HUVECs. From above all, our study implied that DAPA can protect against DCM and myocardial fibrosis through suppressing fibroblast activation and EndMT via AMPKα‐mediated inhibition of TGF‐β/Smad signalling.


| INTRODUC TI ON
Diabetes mellitus (DM) describes a group of metabolic disorders featured by persistent hyperglycaemia, which has constituted one of the leading causes of death globally according to the report by WHO. The increasingly rising prevalence has been predicted by the International Diabetes Federation (IDF) to reach 10.4% (642 million) by 2040. 1 Cardiovascular diseases are by far recognized as the major cause of morbidity and mortality among people with diabetes mellitus. 2 Despite that macrovascular events play a leading role in diabetic cardiac complications, heart failure in the absence of coronary artery disease, hypertension and significant valvular disease is common as well in diabetes, which is referred to as diabetic cardiomyopathy (DCM). 3,4 In terms of the complicated and interrelated pathogenesis of DCM, studies have reported that hyperglycaemia and metabolic disturbance in diabetes can give rise to cardiomyocyte apoptosis, deposition of extracellular matrix (ECM), endothelial dysfunction and oxidative stress, thus leading to the development of DCM. 5,6 Moreover, prolonged exposure to oxidative stress results in chronic inflammation and cardiac fibrosis. 7 In particular, ECM deposition and cross-linking lead to cardiac fibrosis and stiffness, which are closely associated with the primary pathophysiological features of DCM, namely ventricular remodelling, diastolic dysfunction and contractility impairment. 4,8 During the process of cardiac fibrosis, cardiac fibroblasts (CFs) are the main source of ECM. 6 Besides, in the pathological course of DCM, endothelial cells are the headmost targets of hyperglycaemia, and endothelial-to-mesenchymal transition (EndMT) is an important cellular phenotype shift that multiplies CFs and aggravates cardiac fibrosis. 9,10 Hence, fibroblast-like cells derived from endothelial cells via EndMT play a crucial role in diabetic cardiac fibrosis. 11 In the process of transdifferentiation, endothelial cells lose endothelial markers such as CD31 and vascular endothelial cadherin (VE-cadherin) and gain mesenchymal features such as α-smooth muscle actin (α-SMA), Vimentin and fibroblast-specific protein-1(FSP-1) accompanied by the up-regulated transcription level of EndMT markers (snial1, snial2, twist1 and twist2). [12][13][14] Concerning molecular mechanisms, transforming growth factorβ (TGFβ)/Smad signalling plays a vital role in the pathogenesis of cardiac fibrosis and is the utmost pathway to mediate EndMT. 15,16 The heavy financial burden and serious health threat posed by diabetic cardiac complications lead to an urgent need for effective treatment. However, classical glucose-lowering therapies generally have a neutral effect on cardiovascular mortality in diabetes and at times even aggravate heart failure. 17 Sodium-glucose cotransporter-2 (SGLT2) inhibitors, a new class of antidiabetic drugs approved by the FDA in 2013, act directly on the SGLT2 receptor of proximal tubule of the kidney and decrease renal glucose reabsorption. 18 Based on extensive clinical trials and 2019 ESC guidelines, SGLT2 inhibitors have been given priority in treating diabetic patients with cardiovascular diseases. [19][20][21] Importantly, the protective effect of SGLT2 inhibitors against heart failure has been found to act independently of glucose control, 22 whereas the underlying mechanism is not yet fully elucidated. Previous studies generally focused on inflammation and oxidative stress involved in the cardioprotective effect of SGLT2 inhibitors in diabetes, [23][24][25] with limited mechanism exploration into cardiac fibrosis alleviation.
The present study aimed to evaluate the effect of SGLT2 inhibitor dapagliflozin (DAPA) on DCM, especially cardiac fibrosis and EndMT in a rat model of type 2 diabetes. Cellular phenotype and molecule mechanisms were further explored in HG-stimulated HUVECs and CFs.

| Animals
We purchased sixty 4-week-old male Sprague Dawley (SD) rats from Beijing Weitong Lihua Experimental Animal Technology and place them in the environment accessible to feed and water with an alternate 12-hour day-night cycle at 22 ± 2°C. After 1-week adaptive feeding, all rats were randomized to four groups (n = 15 for each group).
Rats in the control group were fed with normal chow and the other three groups with high-fat feed (16% fat and 0.30% cholesterol). Four weeks later, intraperitoneal insulin tolerance test (IPITT) and intraperitoneal glucose tolerance test (IPGTT) were performed to identify rats with insulin resistance, which were then injected intraperitoneally once with streptozotocin (STZ; 35 mg/kg, Solarbio) to produce the type 2 diabetic models. One week after the injection, we took caudal vein blood samples to measure fasting blood glucose (FBG). Only rats with FBG ≥11.1 mmol/L were regarded as a successful type 2 diabetic model and selected for further investigation. After another 8 weeks of the high-fat diet, two groups of diabetic rats were respectively given dapagliflozin (DAPA, 1 mg/kg·day) and metformin (MET, 200 mg/kg·day) in the drinking water. Adjustment of the concentration of DAPA and metformin was made accordingly every 4 days based on the water intake. Eight weeks later, IPITT and IPGTT were performed and all rats were killed with bodyweight and FGB measured. In brief, rats were divided into four groups: Control, DCM, DCM +DAPA and DCM +MET. We took the animal care and experimental procedures following the Guidelines for the Care and Use of Laboratory Animals approved by Shandong University Animal Care Committee and Institutional Ethics Committee of Shandong University.

| Cardiac function
Rats were initially anaesthetized through inhalation of isoflurane (2.5%) and kept under general anaesthesia with persistent 1.5% isoflurane provided by a nasal tube connected to the anaesthesia machine. The cardiac function of rats was evaluated by echocardiography using the Vevo 770 imaging system with RMB710 transducer (VisualSonics). The left ventricle ejection fraction (LVEF), fractional shortening (FS), early-to-late mitral diastolic flow ratio (E/A) and ratio of diastolic mitral annulus velocities (E′/A′) were measured.

| Histology and immunohistochemistry
Rat heart tissues were fixed with formalin, embedded in paraffin and subsequently cut into 5 μm sections. Heart sections were stained with haematoxylin and eosin (HE) to exhibit cardiac structure. Masson's trichrome and Picrosirius red staining were carried out to detect collagen deposition. The quantitative analysis of the cardiomyocyte diameter and collagen volume was conducted with IPP (Image-Pro Plus) software in randomly chosen areas (200×).

| Cell treatment
Human umbilical vein endothelial cells (HUVECs; ATCC) were cultured in the six-well plate (Corning) with endothelial cell medium (5.5 mM glucose, ScienCell) supplemented with 5% foetal bovine serum, 1% penicillinstreptomycin and 1% endothelial cell growth supplement (ECGS) in the 5% CO 2 thermostatic (37°C) incubator for more than 12 hours. When the cell density reached 70%, HUVECs were subject to 6-hour serum starvation and pre-treatment of DAPA (dissolved in DMSO) before exposure to high glucose (HG, 33.3 mM). Thereafter, cells were cultured under HG for at least 48h in the presence or absence of different concentrations of DAPA (0.1, 0.5, 1 μM) with DMSO treatment as a control (vehicle group). Cells under the normal glucose (NG, 5.5 mM) group were exposed to 27.8 mM mannitol to balance the osmotic pressure. To figure out the role of AMPKα, compound C (2 μM, Sigma-171260) and AMPKα siRNA (Santa Cruz-sc45312) were used for inhibition, with AICAR (2 mM, Sigma-A9978) and A-769662(10 μM, MCE-HY50662) for activation. As also, HUVECs were pre-treated with the reagents mentioned above before HG treatment. RNA interference was performed in Opti-

| Immunofluorescence microscopy
The expression and the localization of different target proteins were observed using immunofluorescence methods. HUVECs and CFs were fixed with 4% paraformaldehyde, permeabilized and then blocked with 2% BSA. Thereafter, cells were incubated with primary antibodies against CD31 (Abcam, ab28364) and α-SMA (Abcam, ab5694) in HUVECs, and α-SMA (Abcam, ab5694) in CFs overnight at 4°C. After incubation with two different fluorescent secondary antibodies for 30 minutes at 37°C and staining with DAPI, cell slides were sealed using antifade mounting medium. Finally, images were obtained with immunofluorescence microscopy (Nikon).

| Determination of cell viability and proliferation rate by CCK-8 assay
The cell viability of HUVECs and proliferation rate of CFs were detected using the Cell Counting Kit-8 (CCK-8; Dojin Laboratories).
Cells were seeded into 96-well plates at an initial density of 2.5-3 × 10 4 cells/well with a group of blank control wells (without cells) and a group of untreated control wells (cells only treated with the medium). Each group was made in eight replicates simultaneously. After treatment with different concentrations of DAPA under HG condition for 24 hours, 10 μl kit reagent was added into 100 μl medium per well in the plate which was thereafter incubated at 37°C for 3 hours. OD values were obtained using a multimode microplate reader for the absorbance reading at 450 nm.

| Measurement of intracellular ROS
Oxidative stress was evaluated via detecting the production of intracellular ROS in cultured HUVECs after different treatments.
The fluorescence intensity was observed under an inverted fluorescence microscope (Nikon) with the excitation and emission wavelength respectively at 488 and 525 nm.

| Western blot
Heart tissue or cells were lysed with radioimmunoprecipitation (RIPA) lysis buffer. The prepared protein sample was separated through 10% SDS-PAGE and then transferred to PVDF membrane (Millipore). 5% fat-free milk was used for 1-hour blocking at room temperature. The membrane was subsequently incubated with primary antibodies overnight at 4°C.

| RNA extraction and RT-PCR
We extracted the total RNA from frozen rat cardiac tissue and cells with TRIZOL ™ (Roche Diagnostics), which was later reverse-transcribed into cDNA using oligo(dT) primers and the PrimeScript™RT reagent Kit with gDNA Eraser (Takara). Quantitative Real-time PCR was performed using the SYBR Premix Ex Taq Kit (TaKaRa) according to the manufacturer's protocol to determine the expression of mRNA of collagen I, collagen III, TGFβ, CTGF, snial1, snial2, twist1 and twist2. Each reaction was conducted in triplicate and the results were normalized against GAPDH. The sequences of the oligonucleotide primers are shown in Table 1.

| Statistical analysis
All data analysis was performed with Prism 8.0 (GraphPad) and SPSS 20.0. The difference comparison among groups was conducted using one-way ANOVA with subsequent Tukey's test, Scheffe's test, Bonferroni's post hoc test or Dunnett's multiple-to-one comparison test. Each experiment was repeated at least 3 times, and data were shown as means ± standard deviation (SD). P < .05 was considered statistically significant.

| Dapagliflozin attenuates left ventricular remodelling and improves cardiac function in diabetic rats
Insulin resistance combined with STZ injection contributed to type 2 diabetes in rats, which were then randomized to an untreated group and groups treated with DAPA or metformin (used as a positive antidiabetic control given its confirmed cardiac protective benefits in experimental animals). Compared with the control group, HE staining of diabetic heart tissue revealed pathological hypertrophy which was restored by the administration of DAPA and metformin ( Figure 1A). The significantly elevated ratio of heart weight to bodyweight and the increased diameter of cardiomyocytes were observed in diabetic hearts, which showed reduction after DAPA treatment (Table 2; Figure 1B).
Echocardiography results showed the systolic and diastolic dysfunction in untreated diabetic rats, as evidenced by significantly reduced LVEF, FS, E/A ratio and E′/A′ ratio. In contrast, DAPA and metformin treatment reversed this reduction equally, suggesting that DAPA has a comparable beneficial effect on cardiac dysfunction with metformin ( Figure 1C,D).

| Dapagliflozin alleviates myocardial interstitial fibrosis and perivascular fibrosis in type 2 diabetic rats
Compared with the control group, Masson and Picrosirius red staining of the diabetic heart showed an increased level of collagen deposition both in the interstitial and perivascular region, whereas DAPA and metformin effectively reduced the increment ( Figure 2C).

TA B L E 1 All primers used in our study
detected with immunohistochemical staining, Western blot and mRNA quantification analyses in diabetic hearts, which was rescued after the treatment with DAPA and metformin (Figure 2A,B,E).
Increased LV collagen volume and transcription level of fibrotic markers including TGFβ and CTGF were observed in the untreated diabetic group and then reduced after DAPA treatment ( Figure 2D,E).

| Dapagliflozin inhibits cardiac EndMT in diabetic rats
In contrast with control rats, an up-regulated expression of mesen-

| Dapagliflozin attenuates HG-induced EndMT in HUVECs
To determine the effect of DAPA on the viability of endothelial cells,  Figure 4B).

| AMPKα inhibition abrogates the anti-EndMT effects of dapagliflozin
To explore whether the protective effect of DAPA is reliant on AMPKα activation, cells were treated with compound C and AMPKα siRNA with the efficiency of inactivation and RNA interference shown respectively in Figures 5E and 7A

| Dapagliflozin augments the anti-EndMT effects of AMPKα agonist
We then used AMPKα agonist (AICAR and A-769662) to explore the ad-

| Dapagliflozin inhibits TGFβ/Smad signalling and promotes AMPKα activation in the diabetic heart and HG-stimulated HUVECs
The reduced phosphorylation level of AMPKα was detected in both diabetic hearts ( Figure 3E) and HG-stimulated HUVECs ( Figure 4E), while DAPA restored the phosphorylation level of AMPKα as effectively as metformin which is recognized as a LKB1-AMPK activator.
We then measured the TGFβ/Smad signalling which is the most important pathway during the process of EndMT. As expected, the expression of TGFβ and smad4 were markedly up-regulated in both diabetic rat hearts and HG-stimulated HUVECs, which were blocked by DAPA (1 μM) ( Figures 3E, 4E).

| Dapagliflozin rescues HG-induced EndMT through AMPKα-mediated inhibition of TGFβ/Smad signalling in HUVECs
To evaluate whether TGFβ/Smad signalling was inhibited by  Statistical analysis was carried out using one-way ANOVA.

| Dapagliflozin inhibits the proliferation, activation and collagen production of HG-stimulated cardiac fibroblasts
CFs were isolated from neonatal rats and were cultured under HG in the presence of DAPA (0, 0.1, 0.5, 1, 5 μM). CCK8 results revealed that DAPA treatment repressed the proliferation of CFs stimulated by HG in a dose-dependent manner ( Figure 8B). In addition, HG-induced up-regulated Smad4, α-SMA and Vimentin expression, which was also abolished by the administration of DAPA ( Figure 8A,C). As also, activated collagen synthesis function of the primary CFs was remarkably blocked by DAPA ( Figure 8C). dapagliflozin. (*P < .05, **P < .01, ***P < .001; data = means ± SD; n = 5 in each group; statistical analysis was carried out using one-way ANOVA)

| Dapagliflozin ameliorates oxidative stress in the diabetic cardiac tissue and HG-stimulated HUVECs via AMPKα activation
To determine the effect of DAPA on oxidative stress in diabetic rat hearts, we detected the expression of NOX4 through Western blot and immunohistochemistry. Markedly higher levels of NOX4 expression in diabetic hearts compared with the control group were reversed after DAPA and metformin treatment ( Figure 9A,B). DCFH-DA assay was then performed to detect reactive oxygen species (ROS) levels in HUVECs, showing that HG led to increased cellular ROS production, which was reduced after DAPA treatment ( Figure 9C,D). However, AMPKα siRNA abolished the antioxidant effect of DAPA, indicating the AMPKαdependence ( Figure 9C-E).

| D ISCUSS I ON
Diabetic patients are two to three times more likely to suffer from cardiovascular disorders such as myocardial infarction, stroke and heart failure, resulting in a significantly worsening prognosis. 26,27 Due to the lack of a cardioprotective effect seen with classical antidiabetics, many studies associated with the SGLT2 inhibitor have emerged, since its approval by FDA. Prior to that time, metformin had been widely used for the treatment of type 2 diabetes and was regarded EndMT, endothelial-to-mesenchymal transition; DAPA: dapagliflozin. (*P < .05, **P < .01, ***P < .001; data = means ± SD; n = 5 in each group; statistical analysis was carried out using one-way ANOVA) as one of the very few antidiabetics which exhibited cardioprotective effects independent of its antihyperglycaemic properties. [28][29][30][31] Therefore, in this study, we used metformin as a positive control in our animal experiments and compared its effect to dapagliflozin, an SGLT2 inhibitor. We found that treatment with dapagliflozin could alleviate left ventricular remodelling ( Figure 1A,B) as well as improve both diastolic and systolic cardiac function ( Figure 1C,D), in rats with type 2 diabetes. This is in line with clinical trials showing that various SGLT2 inhibitors (empagliflozin, canagliflozin and dapagliflozin) exhibited cardiovascular safety and benefits. 20,21,32 However, a statistically significant difference between dapagliflozin and metformin regarding improvement to cardiac structure and function was not seen in this study.
Myocardial fibrosis is associated with decreased microvasculature and disruption of normal myocardial structures and is an important pathophysiological process contributing to heart failure, by increasing myocardial stiffness and reducing stroke volume. 12 Therefore, it is of great importance to determine the impact of dapagliflozin on excessive collagen deposition in the diabetic heart and attempt to reveal its underlying mechanism of action. Previous stud- ies have found that metformin shows antifibrotic actions [32][33][34] and was therefore used in this study as a positive control. We have used (*P < .05, **P < .01, ***P < .001; data = means ± SD; n = 3 in each group; statistical analysis was carried out using one-way ANOVA) wide-ranging methodologies to detect the accumulation of extracellular matrix proteins around both the interstitial and perivascular areas, which are indicators of excessive collagen distribution in the hearts of type 2 diabetic rats ( Figure 2). Dapagliflozin significantly reduced collagen volume in these diabetic hearts, demonstrating a comparable antifibrotic effect to metformin (Figure 2). In view of the glucose-lowering effect of dapagliflozin (Table. 2), we followed up this finding with in vitro cell culture experiments to determine whether the antifibrotic action of dapagliflozin was independent of its control of glucose levels.
Cardiac fibrosis is a process mediated by the recruitment and activation of fibroblasts, approximately 27-35% of which are of endothelial origin, derived from the process of EndMT. 12 The fibrosis mediated by fibroblasts originating from endothelial cells contributes to the pathological process of DCM, as has been reported previously. 11 The inhibition for EndMT/EMT by metformin has been reported in several studies, 35,36 while the action by SGLT2 inhibitors is still unclear. Our study represents the first to investigate how dapagliflozin affects EndMT and fibroblasts and to thoroughly investigate its antifibrotic mechanism of action. As shown in Figure 3A, showed a dose dependency (Figure 8). When we assessed cell viability, we found that under HG, there was a suppression in HUVECs, but in contrast, a stimulation in CFs; whereas dapagliflozin exerted a bidirectional impact which was favourable to HUVECs but inhibitory to CFs (Figures 4B, 8B). From the above data, it is clear that dapagliflozin exerts its antifibrotic effect by blocking the fibroblast origin and directly suppressing the activation of CFs. Therefore, we feel that we have elucidated clearly the antifibrotic mechanism of action of dapagliflozin in this study.
AMP-activated protein kinase (AMPK) is a member of the serine/ threonine (Ser/Thr) kinase family, which acts as a 'fuel gauge' under cell stress conditions to maintain energy balance, using an α catalytic subunit as its principal functional domain. 37 Reduced AMPKα activity has been observed in failing human and animal hearts and is closely related to cardiac fibrosis. 38,39 Additionally, previous studies have linked SGLT2 inhibitors to AMPKα activation. [40][41][42] However, whether dapagliflozin attenuates cardiac fibrosis and EndMT in an AMPKα-dependent manner remains unknown. Metformin has been widely recognized as an AMPKα agonist 33,43 and served as a positive control in our study. We observed that the phosphorylation levels of AMPKα were reduced in both diabetic hearts ( Figure 3E) and HG-stimulated HUVECs ( Figure 4E) and that the restoration of AMPKα activity after dapagliflozin treatment was comparable to that of metformin, along with inhibition of cardiac fibrosis and EndMT. However, the anti-EndMT effect of dapagliflozin was abol- It has been reported that TGFβ plays a crucial role in cardiac fibrosis by repressing cardiac fibroblast activation and EndMT progression. 15 However, few studies have offered evidence for the regulation of the TGFβ/Smad pathway by SGLT2 inhibitors in cardiac tissues. 44 In addition, studies have reported that AMPKα activation can inhibit the TGFβ/Smad pathway. 39 Interestingly, recent studies have found an important antioxidant effect of SGLT2 inhibitors in cardiomyocytes. 25 In this study, we found that dapagliflozin could significantly inhibit the increased expression of NOX4 in our diabetic rat tissues and HG-stimulated HUVECs in an AMPKα-dependent manner ( Figure 9). Therefore, dapagliflozin may be capable of alleviating DCM through the AMPKα-mediated inhibition of oxidative stress, generated by NADPH oxidases. In addition, NOX4-mediated ROS production appears to serve as a permissive step towards EndMT and tissue fibrosis; 52,53 thus, dapagliflozin may exert its antifibrotic and anti-EndMT effects at least in part by the inhibition of oxidative stress and reduction of excess ROS.
SGLT2 inhibitors can act directly on the proximal tubule of the kidney, and the primary isoform of this transporter family expressed in the heart is SGLT1, with little evidence for SGLT2. 18 In addition, a previous study indicated that SGLT2 mRNA was undetectable in cultured human pulmonary and coronary artery EC lines. 54 However, a different study reported that SGLT2 was expressed in cultured and native ECs under pathological conditions such as hyperglycaemia and oxidative stress. 55 In conclusion, further experiments are required to explore the expression profile of SGLT2 under pathological conditions in the heart and ECs, and to determine a role for SGLT2 inhibitors as potential cardioprotective agents by direct inhibition of its up-regulated receptor.

| CON CLUS ION
The graphical abstract of the mechanisms under the cardioprotective effect of DAPA revealed in our study is illustrated in Figure 10.
Our study showed that DAPA can attenuate cardiac remodelling and F I G U R E 1 0 Graphical abstract. Dapagliflozin alleviates cardiac fibrosis through suppression of endothelial-to-mesenchymal transition and fibroblast activation via AMPKα-mediated inhibition of TGFβ/Smad signalling in rats with type 2 Diabetes. DAPA: dapagliflozin; ROS, reactive oxygen species; EndMT, endothelial-to-mesenchymal transition; ECM, extracellular matrix; DCM, diabetic cardiomyopathy ameliorate cardiac dysfunction through inhibition of fibrosis and oxidative stress in diabetic rats. As for the underlying mechanism, experiments in vitro demonstrate that DAPA can rescue HG-induced EndMT through inhibition of TGFβ/Smad signalling and oxidative stress in an AMPKα-dependent manner in HUVECs. Moreover, DAPA can directly inhibit the proliferation, activation and collagen production of HG-stimulated CFs. Therefore, our pre-clinical observations have offered new insights into the possible mechanisms concerning cardiovascular mortality reduction by DAPA in humans, thereby providing more reliant evidence for the application of SGLT2 inhibitors. The comprehensive mechanism elucidation of the antifibrotic effect of DAPA also offers a probability for its application to other fibrotic diseases, which remains to be further explored.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.