Acute stress induces long-term metabolic, functional, and structural remodeling of the heart

Takotsubo cardiomyopathy is a stress-induced cardiovascular disease with symptoms comparable to those of an acute coronary syndrome but without coronary obstruction. Takotsubo was initially considered spontaneously reversible, but epidemiological studies revealed significant long-term morbidity and mortality, the reason for which is unknown. Here, we show in a female rodent model that a single pharmacological challenge creates a stress-induced cardiomyopathy similar to Takotsubo. The acute response involves changes in blood and tissue biomarkers and in cardiac in vivo imaging acquired with ultrasound, magnetic resonance and positron emission tomography. Longitudinal follow up using in vivo imaging, histochemistry, protein and proteomics analyses evidences a continued metabolic reprogramming of the heart towards metabolic malfunction, eventually leading to irreversible damage in cardiac function and structure. The results combat the supposed reversibility of Takotsubo, point to dysregulation of glucose metabolic pathways as a main cause of long-term cardiac disease and support early therapeutic management of Takotsubo.

I read with interest the work by Yoganathan and colleagues on the aspects of adverse cardiac remodeling following acute stress in animal model, with the overarching aim to mimic the mechanisms of Takotsubo cardiomyopathy in human. The authors studied the metabolic, functional and structural changes in rodents at baseline and at 2 hours, 7 days, 1-month and 3-month intervals following a single administration of isoproterenol (ISO), which leads to catecholamine surge. The study is ambitious. It comprehensively touched upon multiple aspects of physiological and structural changes of the rodent heart using multi-modality imaging techniques, namely FDG positron emission tomography, high resolution echocardiography and cardiac magnetic resonance imaging, as well as electrocardiogram, clinical parameters such as blood pressure measurement, serum biochemistry, histology and proteomic analysis. The study was systematically carried out with the results clearly presented. The findings were intriguing and challenges our current understanding of Takotsubo cardiomyopathy being a reversible and rather benign entity. Most acute stress changes observed at the 2nd hour had recovered at the 7th day. However, there was high FDG uptake, reduced glucose breakdown and alternative pathways of glucose utilisation in the apex. This is coupled with tissue and vascular remodeling at the apex. In addition, fibrosis was already present in the apex at the 7th day post-ISO and persisted at 3-month, suggesting permanent structural damages to the heart following acute stress. I have the following questions and some suggestions: 1. The study protocol was well designed and provided new insight into the pathophysiology of TTS. Basically, I agree with the conclusion. Nevertheless, methods of assessment of cardiac function must be addressed.
a. Please describe the conditions in CMR imaging of the RV, RA, and LA. b. Number of frames of LV was 16 per heart cycle in CMR. When the left and right atria are imaged under the same conditions, the number of frames is insufficient to calculate early and late diastolic strain rates. In fact, the LA late diastolic strain rate (pump function) in Figure 1C increased nearly 50fold over the control during the chronic period, and the late diastolic strain rate of the RA was negative relative to the control. It is generally not possible for atrial function to increase 50-fold or become negative (reverse pumping function?) relative to controls, either in humans or in rats. All cited papers 68-70 are studies on humans and not on rats having heart rate greater than 300 beats/min. c. Line 105-107 "Furthermore, end-diastolic atrial contraction wave (A) was predominant over the early LV filling wave (E), resulting in a reduced E/A ratio, revealing an abnormal LV relaxation and a compensatory left atrial (LA) contraction." When rats with high resting heart rates further accelerate their heart rate during acute stress, it is difficult to separate the E and A waves in the mitral inflow. In this case, the apparent high A wave may reflect the high E wave ("Am J Physiol Heart Circ Physiol 283: H346-H352, 2002: "Because the heart rate was greater than 300 beats/min in most rats, it was difficult to clearly identify the A wave on transmitral flow velocity spectra. For this reason, measurement of the peak velocity of the A wave and DT was much less feasible than E wave measurement in all three groups").
d. In addition, in figure 1C, there was no evidence of compensatory LA contraction (LA booster pump function), while it is possible that the strain rate measurements were incorrect. e. During the chronic phase of the disease, atrial volume is an excellent indicator of atrial load. Please display left and right atrial volume data as atrial volume can be measured by CMR.
f. "Eur Heart J Cardiovasc Imaging 2020;21:1184-1207". Multimodality imaging in Takotsubo Syndrome: a joint consensus document of the European Association of Cardiovascular Imaging (EACVI) and the Japanese Society of Echocardiography (JSE) "RV free wall longitudinal strain is therefore recommended to detect even small areas of myocardial dysfunction hardly recognizable by visual assessment and/or by other traditional echocardiographic parameters. Conversely, RV global longitudinal strain should not be assessed because involvement of the interventricular septum can be misleading." In this study (figure 1A), the RV longitudinal strain was decreased in the acute phase and the LV longitudinal strain was increased. Since both ventricles share the ventricular septum, these results suggest a reduction in RV free wall strain. "Recent studies have further demonstrated RV involvement in TTS as an important finding associated with worse outcome." (Eur Heart J Cardiovasc Imaging 2020;21:1184-1207). The finding that LA remodeling occurs in the chronic phase of TTS with RV damage in the acute phase has important implications for the pathophysiology of TTS. g. Based on a human study, "TTS patients demonstrated a significantly decreased LA function during the acute/subacute phase of the disease. However, impairment of LA performance seems to be transient in TTS with recovery during follow-up." (Journal of Cardiovascular Magnetic Resonance (2017) 19:15). As observed in the current study, LA function in the chronic phase was reduced and complicated by LA fibrosis, suggesting that this is an animal model which recapitulates a slightly more severe form of TTS.
h. Although the term GLS (global longitudinal strain) was used in this study, it is preferable to use the simple term "longitudinal strain", since this study did not use the average value of longitudinal strain in multiple cross sections (2-, 3-, and 4-chamber view). If the average of longitudinal strain of multiple cross sections was used, please state how many cross-sectional averages were used for the LV, RV, LA, and RA.
i. Line 149 "Parameters derived from cardiac ultrasound images were normal at 7d post-ISO (figure 6; extended data 2 and 4)." There was no data of ultrasound images in figure 6.
2. The plasma concentrations of AST, ALT and CK were measured. These are non-specific and could be raised in various conditions, which are non-cardiac. If the aim was to measure myocardial damage, would the authors explain why cardiac-specific markers, e.g., CK-MB, troponins (T or I) were not measured?
3. Overall, in this study, a rat model of stress-induced cardiomyopathy is used to observe acute and long-term myocardial damage via high-end imaging and microscopic imaging. The article is well written and highly interesting to read. However, despite a huge amount of work, the article eventually offers as only claim to provide early treatment to patients with Takotsubo, which is intuitive. Please expand to what other of the current patient management the authors may refer… 4. The whole study is rationalized by the fact that long-term cardiovascular annual death rates for Takotsubo Syndrome and acute coronary syndrome are the same. Supportive references are: 1). a review article with expert panel from Lyon et al. that did not report survival rates from Takotsubo syndrome; and 2). an observational study that reported a large heterogeneity in long-term outcomes from Takotsubo syndrome, depending on the identified cause. Whilst those caused by physical activities or neurological disorders have poorer or equal outcomes than patients with acute coronary syndrome, those with emotional trigger (by far the most!) had a much better outcomes than the patients with acute coronary syndrome. This points to a significant heterogeneity of all sub-syndromes included in the umbrella of Takotsubo, and probably of the subsequent long-term damage. As such, it would be reductive to consider that a single model would help decipher the whole spectrum. It is thus important to refocus the work and the writing on the subtype which is addressed by the proposed model.
5. The claim that overactivation of the hexosamine and polyol biosynthetic pathways explain tissue and vascular remodeling and induction of reactive oxygen species is proven by the current work. However, ill my opinion, at least another hypothesis has been neglected based on the study findings: the cell-cell junction alteration (that is presumably one of the results of hyperdynamism). In a different, but close disease model, extreme athletes may end having acute myocardial damage and even longterm fibrosis via hyper dynamism without any alteration of the glucose metabolism. In general, the hypothesis-generating part if this work should be enriched to include other potential pathways for myocardial damage.
6. The demonstration of the absence of coronary obstruction part seems superfluous and could be moved to the additional data. 7. Page 2, line 37. Could the authors mean 'reversibility', rather than 'irreversibility'? 8. Please only capitalise proper nouns. As such, 'Humans', 'Ultrasound' etc. should be in small letters.
Reviewer #4 (Remarks to the Author): The overall goal of this study was to examine the effects of an acute pharmacological stress on the heart designed to mimic Takotsubo syndrome and to determine the contributions of acute and chronic metabolic remodeling to the long term damage to the myocardium.
There are however some serious limitations their evaluation and interpretation of the data. 1) In the text the authors state that 2h post-ISO PET imaging showed a 50% increase in the uptake of FDG into the heart (lines 169-170). In Figure 2A the PET images demonstrate marked increase in FDG uptake at both 2 hours and 7days; however, this is not reflected in any of the quantified data presented in Fig 2B. 2) While the data presented in Fig 2B do not appear to show any substantial changes at 2h post-ISO, it is stated that FDG uptake was increased by 70% whereas the index of glucose phosphorylation was reduced. They refer to the apparent disconnect between changes in glucose uptake and phosphorylation as "metabolic stunning". They suggest that this low capacity to use glucose as an energy source is contributes to cardiac dysfunction; however, this conclusion is flawed. It is widely recognized that glucose is not the predominant fuel for mitochondrial ATP production, rather fatty acids are the largest source with contributions from lactate, ketone bodies amino acids (For a recent review see Circulation Research. 2021; 128:1487-1513). Thus, in the absence of any other measures of substrate utilization it is not possible to conclude that potential changes in glucose metabolism have any meaningful effects. In addition, in Extended Data 3C, the circulating plasma concentrations of lactate and fatty acids increase at least 2-fold 2h post-ISO both of which can be readily oxidized by the heart and likely contribute to the apparent decrease in glucose metabolism. Indeed, high circulating lactate levels have been associated with reduced cardiac glucose uptake (e.g., J Physiol. 2002;542: 403-12).
3) Lines 180-181: "During the recovery stage (7d post -ISO), GLUT1 staining was high in the apex but returned to baseline in the basal region, while an increase of GLUT4 staining was observed in both regions". However, in Fig 3 there are no changes in GLUT4 levels in the basal region at any time point. Moreover, measurements of total GLUT1 and GLUT4 protein levels provide no information regarding their contributions to glucose uptake, as only membrane levels are of relevance to glucose metabolism. 4) Lines 193-194 "A high myocardial uptake of glucose increases collagen synthesis, myocardial myofibroblast proliferation, and the expression of fibronectin and transforming growth factor (TGF)-;$:# %,-4 -4 & /-4-05)323)5&5-10 1* 5,) 456(-)4 '-5)(" 8,-', 4,18 5,&5 ,-+, )953&')..6.&3 +.6'14) .)7).4 stimulate fibrotic changes in cultured cardiac myoblasts. I am unaware of any evidence that increased myocardial uptake of glucose induces fibrotic like changes. 5) Understanding the proteomic data presented in Fig. 4 is complicated by the fact that the legend states that "green, overexpression higher than 1.3-fold, red: under-expression lower than -1.3-fold". However, in the figure the green numbers are all preceded by a negative sign and the red numbers have no sign. In the text, which is mostly focused on the HBP and Polyol pathways it is stated that proteins in these pathways are overexpressed, so despite the legend I have assumed that the red colors indicate overexpressed proteins. The conclusion for this specific section is "notion that HBP hyperactivation induced by an acute stress is a driver of cardiac fibrosis and angiogenesis" which as stated in the Discussion this is due to increased O-GlcNAcylation. However, it is not possible to reach this conclusion based on the data presented. It is true that key elements of the HBP and O-GlcNAc regulation are increased at 7 days post-ISO. While the authors have highlighted changes in GFAT expression they have overlooked the 8-fold increase in O-GlcNAcase (OGA) expression, which removes O-GlcNAc from proteins. Therefore, assuming that protein expression equates to activity, then OGA activity is increased more 5-times that of OGT, which if true would result in reduced O-GlcNAc levels not an increase. It is not possible to draw any conclusions regarding changes in O-GlcNAcylation, without a measurement of O-GlcNAc levels at 0 and 7d post-ISO. Immunoblots of some of these key proteins to support the changes observed in the proteomic studies would also be helpful. 6) Lines 304-306: "Finally, overactivation of O-GlcNAcylation increases synthesis of extracellular matrix proteins, as observed here at 7d post-ISO in the apex." Apart from the fact that it cannot be concluded that there are any changes in O-GlcNAc levels, the references cited to support this statement indicate that increased HBP flux in mesangial cells is linked to increased synthesis of extracellular matrix proteins. Neither study includes measurements of O-GlcNAc levels.
Important changes have been made to address the comments of the four reviewers. In particular, 1 we have performed a hierarchical clustering analysis of the proteins involved in glucose pathways 2 ( Figure 4A), an analysis of GFAT2 and of O-GlcNAcylated proteins ( Figure 4B), and added an 3 analysis of the canonical metabolic, vascular and tissue pathways (Extended data 8). We have also 4 followed Reviewers' suggestion to remove the graph representing the E/A ratio from the 5 ultrasound analysis in extended data Figure 4, as well as those of the strain rate measurements 6 from figure 1. 7 8 The point-by-point answers to reviewers' comments are color-coded: 9 10 -Comments/Answers to reviewer #1 and corresponding changes in the revised manuscript are 11 highlighted in yellow. 12 -Comments/Answers to reviewer #2 and corresponding changes in the revised manuscript are 13 highlighted in green. 14 -Comments/Answers to reviewer #3 and corresponding changes in the revised manuscript are 15 highlighted in gray. In this study, Yoganathan and colleagues investigate the mechanism of Takotsubo syndrome using a 25 rodent (rat) model. Given the severity of the acute and long-term effects of TTS on heart function, this 26 is worthy of investigation. 27

28
The authors characterize the model according to detection of known biomarkers, ECG 29 measurements and myocardial deformation according to altered strain rates in myocardium. 30 Further, the animal model shared key characteristics such as reversibility and lack of coronary 31 obstruction. Key markers for glucose uptake (Glut1 and Glut4) were analysed by 32 immunohistochemistry, suggesting (subtle) differences in the response in apical and basal regions, 33 in particular the return of Glut1 to baseline levels in the LV basal region, but sustained higher 34 level in the apical region. Is there an explanation for the higher baseline level of Glut1 in the basal 35 region? 36 37 In fact, the baseline level of GLUT1 expression is slightly lower in the basal than in the apical region, 38 although the difference is not statistically significant. The use of the same scales for the graphs of panels 39 off of the p-values and the on/off criteria. So we considered as proteins of interest (POI), on/off proteins 78 and proteins with p-value less than 0.05 and fold change greater than 1.3 in absolute value.      (2 hours post ISO), and late phases (1-and 3-months post ISO, corresponding to several years of "human 205 life). Secondly, they used an ISO dose twice that of our study, which we avoided because of a significant 206 mortality rate. Thirdly, they attribute higher FDG uptake to an activation of macrophages, while our 207 histochemical data shown in Figure 3 clearly demonstrates that the increase in CD68+ stained 208 macrophages is limited and, more importantly, that GLUT 1 and GLUT4 are essentially expressed in 209 cardiomyocytes. Of note, we performed GLUT1, GLUT4 and CD68 staining on the same sections, and 210 show a complete absence of colocalization between the GLUTs and CD68. Fourthly, we showed that 211 the diversion of glucose metabolism towards alternative anabolic pathways responsible for tissue and 212 vascular remodeling, mainly the HBP and polyol pathways. We totally agree with the reviewer that the notion of reversibility in TTC was questioned by clinical 227 studies, including some cited by the reviewer that have supported long term heart failure phenotypes 228 based on in vivo cardiac imaging 1,2 . In fact, our study was inspired by clinical data suggesting an 229 unexplained irreversibility of TTC. The major advantages of using an animal model are to strengthen 230 hypotheses and explore in much deeper detail the underlying mechanisms. In particular, for obvious 231 reasons, clinical studies do not compare pre-stress and post-stress conditions in the same individuals, 232 and rely on matched "control" subjects, and neither the TTC or the control subjects are exempt of co-233 morbidities in addition to cardiac and other treatments. Thus, even though our study is reductionist and 234 calls for confirmation in human patients, it stands on an objective cause-effect paradigm. Regarding 235 Schwarz et al. 2 , their interesting observations did not report on the metabolic defects that our study 236 designates as a prominent cause of long-term cardiac disability. Scally et al. 1 showed changes in cardiac 237 31 P-spectroscopy, typical of impaired cardiac energy metabolism associated with increased native 238 myocardial T1 without change in extracellular volume. They did not investigate further the metabolic 239 pathways that were involved. Their study did not report signs of fibrosis, while Schwarz et al. 2 suspected 240 fibrosis based on increased cardiac extracellular volume, which is an indirect measurement in contrast 241 to the direct and quantitative tissue analysis that we report here. 242

243
In short, our original findings concern the mechanisms leading to long term post-stress cardiac 244 dysfunction. We show changes in glucose uptake and metabolism preceding a whole cascade of tissue 245 and vascular events, durably weakening the TTC heart. This remodeling corresponds first to a form of 246 glucose intolerance in the heart followed by inhibition of the energy metabolic pathway, glycolysis, and 247 overactivation of the non-energy anabolic pathways of glucose, namely the hexosamine biosynthetic 248 and polyol pathways. The latter two pathways are known to be responsible for myocardial abnormalities 249 and dysfunction 3,4 . Therefore, our extensive exploration of a Takotsubo Because isoprenaline increases heart rate to 450 bpm (baseline 300 bpm) and respiratory rates were 260 around 60 cycles per min, conventional echocardiography and cardiac MRI imaging were limited by 261 temporal resolution and unable to directly observe the ballooning of the LV apex. This may have been 262 possible with superfast CT or MRI, which we didn't have access to. In some cases, we observed transient 263 apical ballooning using ultrafast ultrasound, however this was incidental, restricted to two-dimensional 264 observations and largely operator dependent. For this reason, we relied on MRI-based myocardial strain 265 measurements, that showed a basal hyperkinesia and an apical akinesia at 2h post-ISO, which 266 corresponds to the apical ballooning observed in patients with Takotsubo 5,6 (page 5, lines 119-124). suggest, it would be interesting to counter the overactivation of the hexosamine biosynthetic pathway, 291 for example by blocking the activity of its limiting enzyme, GFAT. The compound 6-diazo-5-oxo-L-292 norleucine (DON) is a synthetic analog of glutamine that inhibits GFAT and has already been used 293 experimentally. Studies have shown that muscle cells chronically exposed to high levels of glucose 294 develop insulin resistance which may be prevented when incubated with DON 9 . DON inhibition of the 295 hexosamine biosynthetic pathway results in a decrease in the expression of genes related to 296 cardiomyocyte hypertrophy induced by 48 h of incubation with phenylephrine, protein synthesis and 297 cardiomyocyte growth, thus preventing cardiac hypertrophy 3 . 298 Secondly, reducing the activity of the polyol pathway is also an interesting avenue to explore as it may 299 prevent cellular glucose from being diverted to non-oxidative pathways and redirected to glycolysis 10 . 300 Several authors have proposed the inhibition of AR by sorbinil or ranirestat 11,12 . Third, antifibrotic 301 therapy may perhaps slow the progression of fibrosis in the LV and left atrium, although there is no 302 clear evidence as far as we know. The connective tissue growth factor CCN2 is a matricellular protein 303 that alters cell signaling pathways responsible for myofibroblast activation leading to fibrosis 13 . CCN2 304 antagonists may represent a novel strategy to limit and reverse cardiac fibrosis. In a preclinical study, 305 mice with pressure-overload-induced heart failure treated with anti-CCN2 monoclonal antibodies (such 306 as FG-3019) showed significant improvement in LV function compared with controls 14 . Another 307 candidate, Gal-3 is a soluble beta-galactoside-binding lectin that is considered a promising therapeutic 308 target in heart failure because it is involved in the proliferation of cardiac fibroblasts that cause fibrosis 15 . 309 Numerous studies have shown that Gal-3 expression is elevated in hypertrophied hearts 16-18 in interstitial 310 lung disease 19 and in the plasma of heart failure patients 20 . One study showed that LV hypertrophy was 311 prevented in Gal-3 knockout mice and LV function was improved 21 . Also, among atypical potential ani-312 fibrosing agents, some microRNAs (miRNAs) expressed by cardiac myocytes regulate signaling 313 pathways in fibroblasts. Silencing of miRNAs has been shown to prevent interstitial fibrosis and inhibit 314 cardiac function in mouse hearts under pressure overload 22  The study is ambitious. It comprehensively touched upon multiple aspects of physiological and 342 structural changes of the rodent heart using multi-modality imaging techniques, namely FDG positron 343 emission tomography, high resolution echocardiography and cardiac magnetic resonance imaging, as 344 well as electrocardiogram, clinical parameters such as blood pressure measurement, serum 345 biochemistry, histology and proteomic analysis. 346

347
The study was systematically carried out with the results clearly presented. The findings were intriguing 348 and challenges our current understanding of Takotsubo cardiomyopathy being a reversible and rather 349 benign entity. Most acute stress changes observed at the 2nd hour had recovered at the 7th day. However, 350 there was high FDG uptake, reduced glucose breakdown and alternative pathways of glucose utilisation 351 in the apex. This is coupled with tissue and vascular remodeling at the apex. In addition, fibrosis was 352 already present in the apex at the 7th day post-ISO and persisted at 3-month, suggesting permanent 353 structural damages to the heart following acute stress.  than 300 beats/min in most rats, it was difficult to clearly identify the A wave on transmitral flow 392 velocity spectra. For this reason, measurement of the peak velocity of the A wave and DT was 393 much less feasible than E wave measurement in all three groups"). 394 395 We thank you for this very pertinent comment. In contrast to the article that you cite, we used here a 396 high-resolution ultrasound scanner (prf: 20kHz to 30kHz, frame rate: 100 frames/s) under general 397 anesthesia of the rats, in order to allow a good visualization of the E/A curves (see Figure Curve 1 398 hereunder). However, after verification of all E/A curves, we admit that a partial or total fusion of these 399 waves is present in some rats independently of the pre-and post-ISO times (see Figure Curve 2). This 400 is probably a consequence of the high heart rate in spite of the anesthesia, as observed in the study of 401 The strain measurements for the atria were preferably performed on the 4 chamber views for reliability 421 and repeatability reasons. We were accordingly able to determine the minimum and maximum atrial 422 surfaces on this same view as a surrogate of the atrial size. We now show LA surfaces normalized to the 423 baseline value for each animal in Figure 1C. It shows a decrease of the atrial area at 2h post-ISO that 424 was coupled to ISO-mediated left ventricular hypercontractility. The atrial area tended to increase at the 425 late phases, at 7d and 3mo post-ISO, suggesting a long-term enlargement of the left atrium. This trend 426 is consistent with the significant decrease in atrial strain and the presence of fibrosis. It also shows that 427 strain measurements have a better sensitivity than atrial area/volume measurements. 428 We have added the atrial surfaces measurements in Figure 1C  We thank you for this suggestion. We did not measure the longitudinal strain in all incidences. 471 Therefore, we have made the corrections throughout the revised version of the manuscript. 472 473 i. Line 149 "Parameters derived from cardiac ultrasound images were normal at 7d post-ISO 474 ( figure 6; extended data 2 and 4)." There was no data of ultrasound images in figure 6. 475 476 Thank you, we have now corrected this error (page 6, lines 146-147). InterTAK registry showed that survival was comparable between patients receiving or not beta-blockers 509 at discharge 6 . Approximately 30% of TTC patients had a recurrence during beta-blocker treatment, 510 which raises questions about their cardioprotective effect in TTC 6 . Anticoagulation is recommended in 511 TTC patients with atrial fibrillation and thromboembolic complications, whereas prophylactic 512 administration of anticoagulants could be considered in patients with low LVEF or LV dysfunction 513 involving the apex until LV function improves. Thus, as far as we know, there are presently no 514 randomized trials of specific treatments for TTC, and we do not know what benefits could be expected 515 from any treatment. 516 In addition, the diagnosis of TTC can be difficult because of the similarity of clinical signs to those 517 encountered in AMI or acute coronary syndrome (ACS), particularly ECG abnormalities and myocardial 518 biomarkers 5 . Finally, there is a real lack of noninvasive approaches for rapid and reliable differential 519 diagnosis of TTC, as only LV coronography is considered a reliable diagnostic tool to distinguish 520 between these pathologies and make the diagnosis of TTC. 521 Our study shows that an overdose of catecholamines, in this case isoprenaline, leads to a metabolic 522 remodeling that persists over time. We show that there is a switch in myocardial metabolism with an 523 inhibition of oxidative phosphorylation and an overconsumption of glucose that is diverted to anabolic 524 pathways, in particular to the polyol and biosynthetic hexosamine pathways. We strongly suggest that 525 it is this bypass and overactivation of these two pathways that would be largely responsible for the 526 significant presence of fibrosis first in the apex and then throughout the LV and left atrium and that 527 would also be responsible for remodeling of the cardiac cytoskeleton and pathological angiogenesis. It 528 thus would be interesting to consider treatments that modulate these anabolic pathways of glucose, 529 especially if it can be demonstrated in parallel, that they attenuate fibrosis. 530 In our study, we highlighted the importance of myocardial strain measurements and FDG PET imaging 531 (or cardiac PETRUS) to assess and detect both regional and global functional, tissue, and metabolic 532 remodeling of the heart. We plead that the combination of these in vivo non-invasive imaging methods 533 will be crucial to improve the management of TTC patients, the diagnosis and ultimately the prognosis 534 of TTC. Since TTC is multifactorial and complex and often associates with comorbidity, the use of a 535 reductionist animal model has allowed us to detect imaging biomarkers characteristic of TTC that 536 deserve to be translated and explored in the clinic. 537 We have added part of the present comments in the Discussion section of the revised manuscript (page 538 13, lines 328-333). 539 540

The whole study is rationalized by the fact that long-term cardiovascular annual death rates 541
for Takotsubo Syndrome and acute coronary syndrome are the same. Supportive references are: 542

1). a review article with expert panel from Lyon et al. that did not report survival rates from 543
Takotsubo syndrome; and 2). an observational study that reported a large heterogeneity in long-544 term outcomes from Takotsubo syndrome, depending on the identified cause. Whilst those caused 545 by physical activities or neurological disorders have poorer or equal outcomes than patients with 546 acute coronary syndrome, those with emotional trigger (by far the most!) had a much better 547 outcomes than the patients with acute coronary syndrome. This points to a significant 548 heterogeneity of all sub-syndromes included in the umbrella of Takotsubo, and probably of the 549 subsequent long-term damage. As such, it would be reductive to consider that a single model 550 would help decipher the whole spectrum. It is thus important to refocus the work and the writing 551 on the subtype which is addressed by the proposed model. 552 553 It is absolutely and entirely right to underline the complexity of this multifactorial and multifaceted 554 cardiomyopathy. Takotsubo is a pathology discovered and most frequently found in menopausal 555 women, but that also concerns men and younger women, and that has multi-cause apart from ISO stress 556 (genetic, physical, neurological disorders, pheochromocytoma etc). We have indeed limited our study 557 to a reductionist, reproducible and homogeneous animal model with a single acute onset of isoprenaline, 558 independently from aging, comorbidity, repetitive stress and differences in the stressors triggering the 559 disease. The interest of this approach was above all to understand the mechanisms involved and their 560 chronology during an excessive administration of beta-adrenergic agonists as it can happen in clinic 42 . 561 We have shown that a single injection of isoprenaline resulted in severe long-term deleterious effects in 562 young adult rats. We assume that these observed effects will be much more consequential in elderly 563 subjects and with comorbidity or with another stressor triggering the disease such as neurological 564 disorder 43-45 . 565 The choice of the reductionist model is a strength but also a limitation of our study. It would be reductive 566 to consider that a single model would help decipher the whole TTC spectrum. Following you comment, 567 we have added a paragraph " study limitations" to put our study in the global context of TTC (pages 13- of this mechanical stress (hyperdynamism), we should have seen the appearance of fibrosis primarily in 587 the basal area, which was not the case. Therefore, we believe that the appearance of fibrosis is strongly 588 related to apical akinesia coinciding with an over-activation of glucose anabolic pathways. 589 590 6. The demonstration of the absence of coronary obstruction part seems superfluous and could be 591 moved to the additional data. 592

593
We report the results of coronary flow reserve in the "supplementary data" section (extended data 2), 594 as the increase in coronary flow reserve at 2 hours indicates normal coronary dilator function and is 595 therefore a good indicator of the absence of left coronary artery obstruction. We believe it is important 596 to mention this point since it discriminates TTC from acute coronary syndrome. 597 The overall goal of this study was to examine the effects of an acute pharmacological stress on the heart 611 designed to mimic Takotsubo syndrome and to determine the contributions of acute and chronic 612 metabolic remodeling to the long-term damage to the myocardium.  We thank the reviewer for pointing to the unsuitable representation of the data in Figure 2B, in which 622 boxplots do not highlight the mean and standard deviation of FDG uptake and are misleading. The 623 revised version of the manuscript presents the same data using histograms based on the mean and 624 standard deviation values, which is all the more consistent with the statistical analyses. The new Figure  625 2B makes clear the 47% increase over baseline of the mean SUV in the myocardium at 2h post-ISO, 626 and the 36% increase over baseline at 7d post-ISO. It is true that, in this case as often, quantification of 627 PET imaging from small animal series yields statistically non-significant results, although studies cannot 628 be extended to large cohorts for practical and ethical reasons. This is the reason why we have performed 629 additional extensive histological and proteomic analyses (Figures 2 and 4 and extended data 8A) that 630 support the patterns of FDG uptake and kinetics in the myocardium. Amendments have now been made 631 accordingly in the revised manuscript (page 7, lines 167-169). 632 633 2) While the data presented in Fig 2B do not appear to show any substantial changes at 2h post-634 ISO, it is stated that FDG uptake was increased by 70% whereas the index of glucose 635 phosphorylation was reduced. They refer to the apparent disconnect between changes in glucose 636 uptake and phosphorylation as "metabolic stunning". They suggest that this low capacity to use 637 glucose as an energy source is contributes to cardiac dysfunction; however, this conclusion is 638 We observed at 2h post-ISO an average increase of 70% of the rate constant K1compared to its baseline 649 value. As mentioned in the text "K1, the rate constant for FDG passage from blood to tissue, increased 650 by 70%" (page 7, line 170), this rate constant reflects the passage of glucose from plasma to tissue and 651 is calculated using two-compartment model analysis equations derived from dynamic PET imaging. In 652 contrast, SUV values do not take into account the kinetics of FDG uptake. Thus, the entry of glucose 653 in the myocardium at 2h post-ISO is increased and coincides with the increase in GLUT1 expression in 654 cardiomyocytes (Figure 3). 655

flawed. It is widely recognized that glucose is not the predominant fuel for mitochondrial ATP
The term "metabolic stunning" 41 (page 7, lines 173-175) has been used to define the lack of metabolic 656 flexibility of the myocardium in conditions of increased energy demand, e.g., hypercontractility and 657 systolic dysfunction. Although high levels of glucose, fatty acids, or lactate, are available in the plasma, 658 the myocardial energy intake (the actual consumption of these substrates by the heart) does not adapt to 659 the demand. 660 In our study, we now used the term "glucose metabolic stunning" to define the mismatch between 661 hyperglycemia, hypercontractility of the LV, directly mediated by isoprenaline (extended data 3C), 662 and the low or unchanged use of glucose as an energetic substrate by the myocardium, (Figure 1A, 663 Figure 4, extended data 4). Fatty acids are indeed the major energy source in the heart, but glucose 664 remains the second preferred energy substrate, especially during rapid augmentations of contraction 665 because, in contrast to other energy sources, glycolysis can rapidly produce ATP without the activation 666 of mitochondrial oxidative phosphorylation. This is the reason why we have named "metabolic 667 stunning" the energetic situation of the myocardium at 2 h post-ISO 46 (Figure 3). We reckon with the 668 reviewer that the term may not be well chosen here because of the increased plasmatic concentrations 669 of the other energy substrates. However, we did show using oil-red staining an accumulation of lipid 670 droplets at 2 hours, indicating that compensation of energetic needs by fatty acid beta-oxidation, 671 essentially a mitochondrial process, is highly unlikely (extended data 5B). The energetic status of the 672 heart after ISO resembles the myocardial glucose intolerance, as shown in the diabetic heart by 673  (Table below and extended data 8A). 699 The accumulation of lipid droplets at 2h post-ISO confirms that fatty acids do not act as supplementary 700 energy source in the post-ISO heart. 701 Regarding lactate, it also requires the activation of oxidative phosphorylation to be used as a source of 702 energy, which we have not observed. Myocardial lactate metabolism is favored over glucose metabolism 703 in the case of insufficient exogenous glucose supply 49 , while the post-ISO heart is offered plenty of 704 glucose by the plasma and the increased transport rate constants. 705 Consequently, all of these data and observations indicate a deregulation of myocardial glucose 706 metabolism and intolerance at 2h post-ISO, a phenomenon that most certainly has consequences on the 707 progression of the pathology. Moreover, as shown by Scally C et al. 1 using cardiac 31 P-spectroscopy, 708 there is a global deficit in ATP production that is typical of impaired cardiac energy metabolism. 709

710
We have completed the tables in extended data 8A (supplementary information, page 8). We have now corrected the revised manuscript appropriately (page 7, lines 179-180). 725 We completely agree with the reviewer that the translocation of GLUTs to the plasma membrane of 726 cardiomyocytes is the key indicator of the level of glucose entry into cells. Immunofluorescence studies 727 of cardiac sections allow to localize the distribution of GLUTs at the cellular level, which is more 728 meaningful than their expression level. AKT. The synthesis of phosphorinositides is also overactivated at 7d post-ISO. Another insulin-742 independent pathway known to promote GLUT4 translocation in the heart is the AMPK (adenine 743 monophosphate-activated protein kinase) pathway 54,55 . The AMPK pathway is also activated at 7 days 744 post-ISO in the apex (supplementary information, page 8, extended data 8A). Thus, there is both 745 direct (immunofluorescence staining of heart sections) and indirect (overactivation of the pathways 746 mentioned above) evidence for an increased translocation of GLUT4 to the cardiomyocytic plasma 747 membrane, and this evidence is corroborated by the increased entry of FDG in the apical cardiomyocytes 748 at 7d post-ISO. Taken together, these results demonstrate a significantly augmented glucose entry in the 749 cardiac apex at 7d post-ISO. 750 We have added these signaling pathways involved in the activation of GLUT4 translocation to the 751 membrane surface in extended data 8A (supplementary information, page 8). We thank the reviewer for his comment pointing out to an ambiguity due to text abridging. We meant 760 to say that (i) excess glucose entering the myocardium is diverted to alternative non-energetic pathways 761 and (ii) that a consequence of the overactivity of these pathways is extensive tissue and vascular proteins indicated that they were significantly increased at 7d post-ISO in the apex (see Figure 4B), 781 reflecting a high level of O-GlcNAcylation, consistent with the overactivation of the HBP observed by 782 proteomic and western blot analyses. It has been shown in mice that overactivation of GFAT1 increases 783 O-GlcNAcylation and fibrosis 66 . In rats, GFAT1 is the cardiomyocyte-specific isoform of GFAT, while 784 GFAT2 is the isoform specifically expressed in rat myofibroblasts 66   The additional analysis of the proteomic data is very informative, although representation of the data by PCA and hierarchical clustering does highlight the limitation of using low numbers of animals. As the authors point out, statistically significant proteins could only be identified using raw p-value, and the reason for this is evident from the PCA plot and cluster analysis.
In particular, the day-7 LV apex samples, while clearly separated from the 0 and 2hr samples in Principal Component 2 (which accounts for 11% of the variation), split into two subgroups in Principal Component 1 (which accounts for ~43% of the variation. These sub-groups are evident on the HM_apex figure, where R402, R403 and R404 are clearly distinct from R405 and R406.
I am therefore left wondering which of these subgroups is the 'representative' one, and how well the results can be extrapolated to a larger number of animals were the experiment repeated?
The authors need to explicitly comment in the manuscript on the level of biological/experimental variation revealed by this analysis and acknowledge the limitation of using n=5 animals where potentially two are 'outliers'.
Apart from these issues The authors should be commended for a very thorough revision of the manuscript and the provision of new data.
Reviewer #2 (Remarks to the Author): Dear Authors, In this very interesting study the authors highlighted the long term structural and metabolic changes in patients with TTS. As it is increasingly recognized that TTS involves long-term functional changes, this study contributes to the understanding of the possible changes and their origin. The evidence of fibrosis (even though never histologically demonstrated in patients with TTS), mid to long-term change in glucose metabolism and also engagement in the HBP signaling are interesting findings, especially when we consider the long term functional and structural changes that have been reported by Scally and Colleagues in their 2018 and 2019 published papers in Circulation. While their findings and conclusions are sound, and a great deal of effort can be seen, the N is a little disappointing in some of the analyses (probably due to the number of experiments, but still).

Comments: 1. Strain analysis:
The authors did great effort to quantify LV Dysfunction due to Isoproterenol, and instead of being satisfied with the subjective diagnose of "apical ballooning" they analyzed the strain parameters, which is a very elegant proof, however, it should be mentioned (if already happened, please indicate where) if TTS like changes could be observed in all rodents, and if not, if the rodents without changes in their strain parameters have been excluded. And since the numbers are differing: why could you obtain different numbers of measurements at equal timepoints? (Figure 1 A,B,C at 3 months: n=4, but n=5 in the circumferential strain analysis). And in the atriums we have n=6?
2. Animal count: Please indicate how many mice were treated? In how many rodents was the ventricular function quantified with strain (if not all: how many and why not in all)? -In figure 2 we have 9 measurements at every timepoints, are those the same animals was stresscardiomyopathy even induced in all of them (as defined by strain?)? -In figure 3 we have different numbers of stainings for each antigen, why is that? the same amount of tissue sections of different rats should be available?

Western Blots:
The first Blot has a substantially different look. Please show a picture of the uncropped Western Blot (GAPDH missing?). Why is there a different number of samples in every Plot? (First plot says n=4,n=5,n=5) and there is a picture of 2/3 bands? In the second Blot we see 4 different samples at 3 timepoints but shouldn't there be 3+3+3+3+3 since n=5? In the third Blot the full n is depicted, shouldn't the other pictures look alike? 4. Histology: Figure 5. Why is there a substantial bigger number that was stained for Plot G & H compared to E & F, why were the other rodents excluded? Figure 6. Why did you only stain 4 samples here?
So while the findings are very interesting and a great addition in the field, these substantial comments should be addressed.
Reviewer #3 (Remarks to the Author): My ccomments have been well addressed.

Reviewer #4 (Remarks to the Author):
This revision is improved over the initial submission; however, there remain concerns regarding how the authors over interpret their metabolic data. While not uncommon in studies involving proteomics the authors frequently extrapolate changes in the abundance of proteins with changes in activity of metabolic pathways and then link the changes in the activity of these pathways to pathology. However, the measurement of an abundance of one or more proteins in a pathway contains no inherent information regarding the activity of that pathway. Specific examples are given below.
Another concern is that the results section contains discussion of interpretation and speculation of the meaning of much of the metabolism data that should be limited to the Discussion or Introduction. For example, the information contained in lines 156-166 is much more suited to the Introduction, whereas lines 173-175 should be moved to the Discussion. Such changes would greatly streamline the Results section, while at the same time separating and thereby clarifying descriptions of the data from interpretation of these data.
Overall, the metabolic data broadly demonstrates metabolic remodeling in TTC, which could be a contributing factor to the observed cardiac dysfunction. However, attributing changes in activity of any specific metabolic pathway(s) to such remodeling and dysfunction is beyond the scope of the data presented in the manuscript. 1) Lines 173-175: In the Results at the end of the first section describing the FDG PET data shortly following ISO administration it is stated that "An increase of the entry of glucose into myocardial cells combined to a decrease in its rate of phosphorylation and lipidotoxicity observed in oil-red staining correspond to "glucose metabolic stunning".
The FDG data certainly indicate increased glucose transport and decreased glucose phosphorylation.
The conclusion regarding lipotoxocity is based on Oil Red O staining shown in Extended Data Fig 5. An increase in Oil Red staining suggests an increase in neutral triglycerides and lipids in the heart at the 2hr time point, which could occur for a variety of reasons. However, these data are insufficient on their own to substantiate the development of lipotoxicity. In the pathway analysis in Extended data 8 there does not appear to be indications of impaired fatty acid oxidation at the 2hr time point, which could be one factor contributing to increased lipid accumulation. It is noted that Shao et al. (Eur J Heart Fail 2013 Jan;15(1):9-22. doi: 10.1093/eurjhf/hfs161), surprisingly not cited here, reported very similar results following acute ISO treatment, which they also described as liptoxicity, although in that study a much more detailed analysis of lipids and lipid metabolic pathways was provided.
Nevertheless, it is unclear how such discordant metabolic changes can be used to describe a phenomenon of "glucose metabolic stunning". Metabolic stunning has more typically been used in PET studies to describe delayed recovery of metabolism compared to recovery of cardiac function or perfusion. Here there is evidence of both cardiac stress/dysfunction and metabolic dysfunction at 2hrs following ISO, consequently it is difficult to understand why this is described as "metabolic stunning".
2) In Section 4 of the results it is concluded that the proteomic data "indicate inactivation of glycolysis and oxidative phosphorylation and over-activation of glucose alternative pathways, namely the hexosamines biosynthetic (HBP) and polyol pathways" . It is a fundamental characteristic of proteomic data that it cannot alone indicate changes in the activation of specific metabolic pathways. There is no doubt that the pathway analysis data in Extended Data 8 is valuable and can provide supporting results, but it should be clearly stated that such data are "consistent with" or "supportive of changes" in pathway activities. The statement on line 195 that "large amounts of myocardial G6P were diverted to alternative pathways" cannot be supported by the pathway analyses.
3) Figure 4 and lines 195-204: Since a major theme is a decrease in glucose metabolism via glycolysis, immunoblots confirming the proteomic analysis would help support this conclusion. The same is true for OGT, OGA, aldose reductase and sorbitol dehydrogenase, since increased glucose metabolism via these pathways are proposed as contributing to the adverse remodeling observed in the TTC model.
As GAPDH changes in response to ISO treatment it cannot be used as a loading control for GFAT2 immunoblots. Also loading controls should be provided for GFAT1 and O-GlcNAc immunoblots.
It is surprising given the proposed importance of the HBP and PPP in TTC remodeling that neither pathway is identified in the pathway analyses. 4) Line 264: it is stated that "myocardial glycolysis was reduced or remained unchanged". Based on the methods used and results presented, it is unclear how the rate of glycolysis was determined. If glycolytic rate was not measure directly this needs to be rephrased to reflect the results more accurately. 5) Line 266-267: As discussed earlier the concept of glucose metabolic stunning is inherently flawed as there does not appear to be a mismatch between function and metabolism. In addition no data are provided to indicate that G6P is diverted into alternative pathways of glycolysis. This needs to be rephrased to reflect the results more accurately. 6) Lines 328-343: Please change "hyperactivation" or "overactivation" to reflect the results in this study, and those cited, more accurately. For example, "significant increase in protein expression" would be one approach. It is unfortunately a limitation in studies of the HBP and protein O-GlcNAcylation that enzyme activities and metabolic fluxes through these pathways are rarely measured. As a result, steady state changes in protein expression or metabolite levels are often, misleadingly, characterized as changes in activity or flux.
Overall, the metabolic data broadly demonstrates metabolic remodeling in TTC, which could be a 1 contributing factor to the observed cardiac dysfunction. However, attributing changes in activity of any 2 specific metabolic pathway(s) to such remodeling and dysfunction is beyond the scope of the data presented 3 in the manuscript. 4 We agree with the reviewer: metabolic remodeling is a contributing factor to cardiac dysfunction. 5 Considering the complexity of cardiac metabolism, it is most likely that a combination of mechanisms 6 concurs to long-term tissue remodeling and cardiac malfunction. Our intention was to pinpoint glucose 7 alternative metabolism as an important contributor, not to eliminate other possibilities such as acute 8 inflammatory signaling, calcium signaling, or others. 9 10 1) Lines 173-175: In the Results at the end of the first section describing the FDG PET data shortly 11 following ISO administration it is stated that "An increase of the entry of glucose into myocardial cells 12 combined to a decrease in its rate of phosphorylation and lipidotoxicity observed in oil-red staining 13 correspond to "glucose metabolic stunning". 14 The FDG data certainly indicate increased glucose transport and decreased glucose phosphorylation. The Nevertheless, it is unclear how such discordant metabolic changes can be used to describe a phenomenon 19 of "glucose metabolic stunning". Metabolic stunning has more typically been used in PET studies to 20 describe delayed recovery of metabolism compared to recovery of cardiac function or perfusion. Here there 21 is evidence of both cardiac stress/dysfunction and metabolic dysfunction at 2hrs following ISO, 22 consequently it is difficult to understand why this is described as "metabolic stunning". 23 journal of heart failure, 15(1), 9-22.) used the term "metabolic stunning" to qualify the mismatch between 30 the energetic needs of the heart to counter ventricular dysfunction and the availability of energy metabolites 31 and their dysregulation in the acute phase following stress. The term has previously been used in the context 32