Speckle-tracking echocardiography combined with imaging mass spectrometry assesses region-dependent alterations

Left ventricular (LV) contraction is characterized by shortening and thickening of longitudinal and circumferential fibres. To date, it is poorly understood how LV deformation is altered in the pathogenesis of streptozotocin (STZ)-induced type 1 diabetes mellitus-associated diabetic cardiomyopathy and how this is associated with changes in cardiac structural composition. To gain further insights in these LV alterations, eight-week-old C57BL6/j mice were intraperitoneally injected with 50 mg/kg body weight STZ during 5 consecutive days. Six, 9, and 12 weeks (w) post injections, echocardiographic analysis was performed using a Vevo 3100 device coupled to a 30-MHz linear-frequency transducer. Speckle-tracking echocardiography (STE) demonstrated impaired global longitudinal peak strain (GLS) in STZ versus control mice at all time points. 9w STZ animals displayed an impaired global circumferential peak strain (GCS) versus 6w and 12w STZ mice. They further exhibited decreased myocardial deformation behaviour of the anterior and posterior base versus controls, which was paralleled with an elevated collagen I/III protein ratio. Additionally, hypothesis-free proteome analysis by imaging mass spectrometry (IMS) identified regional- and time-dependent changes of proteins affecting sarcomere mechanics between STZ and control mice. In conclusion, STZ-induced diabetic cardiomyopathy changes global cardiac deformation associated with alterations in cardiac sarcomere proteins.


STZ-induced type 1 diabetes mellitus is associated with impaired morphological parameters.
In accordance with previous studies 23 , mice suffering from type 1 diabetes mellitus displayed increases in blood glucose levels at 6w, 9w, and 12w after STZ application versus corresponding controls, respectively (Fig. 1a). In parallel, the body weight of STZ-treated mice was lower at 6w, 9w, and 12w compared to respective control mice (Fig. 1b). Additionally, we detected a decline in the LV mass of STZ mice at all three investigated time points versus the respective controls (Fig. 1c). LV mass/body weight ratios revealed a lower LV mass/body weight ratio only at 6w after STZ application versus the corresponding control mice (Fig. 1d). Surprisingly, 9w after type 1 diabetes mellitus induction, LV mass and LV mass/body weight were higher versus 6w and 12w STZ mice, respectively (Fig. 1c,d). To further resolve the question why 9w STZ differed from 6w and 12w STZ mice, subsequent experiments and analysis were designed. Fig. 2, the mean diastolic LV anterior wall and LV posterior wall thickness were significantly reduced in STZ-treated mice after 6w and 12w, compared to corresponding control mice (Fig. 2a,b). Consistent with Shepherd and colleagues 24 , we detected a reduction in mean diastolic LV internal diameter in 6w, 9w, and 12w STZ mice compared to control mice (Fig. 2c). This was paralleled by a decline in end-diastolic LV volume at 6w, 9w, and 12w post-STZ versus controls (Fig. 2d). With respect to the LV mass and LV mass/body weight ratio, 9w STZ mice differed from 6w and 12w STZ mice, as indicated by increased diastolic LV anterior wall and LV posterior wall values. Specifically, LV anterior wall thickness of 9w STZ mice was higher compared to 6w STZ and 12w STZ mice, respectively. Analogously, LV posterior wall thickness was increased at STZ 9w versus STZ 6w and 12w, respectively. As reported previously [25][26][27] , diastolic wall strain (DWS), which displays the differences of posterior LV wall thickness during systole and diastole, is proposed to be a marker for diastolic function and correlates with LV stiffness. Compared to the respective controls, 6w STZ mice displayed increased DWS, whereas DWS was reduced in 9w STZ mice (Fig. 2e). Similar to LV wall thicknesses, differences between 9w STZ mice versus 6w STZ and 12w STZ mice were detected, respectively.

STZ-induced type 1 diabetes mellitus impairs diastolic and systolic function. As shown in
In agreement with existing data 3,24,28 , we further observed reduced cardiac ejection fraction, and fractional shortening in 9w STZ versus control mice. In addition, stroke volume and cardiac output were decreased in STZ animals compared to the corresponding control mice (Supplemental Fig. 1). Comparison within the STZ groups revealed only a reduced ejection fraction in 9w STZ mice versus 12w STZ animals. STZ-induced type 1 diabetes mellitus reduces global peak strain. Next, we examined global peak strain parameters as well as segmental deformation behaviour, indicated by regional peak strain values, to evaluate myocardial function. In comparison to the corresponding controls, GLS was reduced in mice at 6w, 9w, and 12w STZ, respectively (Fig. 3a). Furthermore, the STZ 9w group displayed a decrease in GLS compared to the STZ 6w mice. Additionally performed quantitative segmental strain analysis indicated decreased myocardial deformation behaviour of the anterior and posterior base at 9w after STZ treatment (Fig. 3c). In parallel, we observed an increase in GCS in 6w STZ and 12w STZ mice versus corresponding controls, respectively (Fig. 3b). In contrast, the 9w STZ animals displayed a lower GCS relative to 6w and 12w STZ mice, respectively. Segmental analysis of the respective short axis B-mode images showed improved contraction of the anterior free wall and posterior septum in STZ mice at 6w compared to controls (Fig. 3d). Analogous, myocardial contraction of the posterior septum was increased in 12w STZ mice versus controls. These data indicate that STZ-induced diabetic cardiomyopathy is associated with an impairment in GLS, whereas GCS is partially improved. Additionally, the GLS and GCS rates, which reflect the change in strain over time, exhibited similar regulations in STZ versus control mice as the respective GLS and GCS (Supplemental Fig. 2). STZ-induced type 1 diabetes mellitus modulates collagen I deposition and extracellular matrix turnover in a time-dependent manner. After we observed time-dependent changes in STE-assessed strain parameters following STZ treatment, we investigated possible underlying mechanisms. In comparison to the respective controls, we detected no differences in Col1a1 mRNA expression in STZ 9w animals (Fig. 4a). In contrast, collagen I protein expression was increased in 9w STZ mice versus controls, which was accompanied by an increase in the collagen I/collagen III protein ratio (Fig. 4b,c). Twelve w after STZ application, Col1a1 gene expression was reduced in mice, and this finding was associated with a reduction in collagen I protein expression (Fig. 4b). In parallel to the decline in collagen mRNA and protein levels, 12w STZ mice displayed a decrease in lysyl oxidase (Lox) and lysyl oxidase-like (LoxL)-2 mRNA levels compared to controls, respectively (Fig. 4d,e). Beside collagen cross-linking, collagen homeostasis is mainly regulated by MMPs and TIMPs. Therefore, we examined the expression of MMPs and TIMPs in the LV (Fig. 4f-h). Especially, MMP-8 and MMP-13 are responsible for collagen I degradation, indicating their role in cardiac remodeling 29 . Compared to control animals, MMP-8 mRNA levels were elevated in 9w STZ mice, whereas MMP-13 mRNA levels were reduced in 6w and 12w STZ mice, respectively (Fig. 4f,g). Similar to collagen I, mice at 9w STZ displayed an increase in MMP-8 and MMP-13 gene expression compared to the 6w STZ group. Furthermore, MMP-8 and MMP-13 mRNA levels were lower in 12w STZ mice versus 9w STZ mice. TIMP-1 expression showed a decline in mRNA levels at 9w and 12w post-STZ, respectively, compared to 6w STZ (Fig. 4h).

Discussion
In the present study, we reported time-and (region)-dependent changes in the expression of cardiac collagen, MMP/TIMP, and proteins of the cardiac sarcomere during the development of STZ-induced type 1 diabetes mellitus-associated diabetic cardiomyopathy. These alterations are associated with and likely contribute to changes in myocardial deformation behaviour as indicated by altered GLS and GCS.
Patients suffering from diabetic cardiomyopathy exhibit distinct cardiac structural and functional alterations. STZ application is a suitable method to induce experimental type 1 diabetes mellitus-associated diabetic cardiomyopathy characterized by enhanced cardiac inflammation and fibrosis 3,30-32 . There is increasing evidence that diastolic function is impaired already at early stages of diabetic cardiomyopathy, in the absence of prominent cardiac inflammation and fibrosis 4 .
However, the detection and mechanistic understanding of early cardiac dysfunction associated with pathological remodelling, has been a major challenge for several decades of research 33,34 . STE-based strain analysis provides an attractive tool for the assessment of global cardiac function and allows the detection of subtle changes in tissue deformation. As such, recent guidelines for diagnosis of heart failure include the assessment of GLS 9 . However, the link between myocardial deformation and underlying changes in structural proteins in the context of diabetic cardiomyopathy is still little explored. Only a few studies 5,24,35 have determined global LV function via strain parameters in the context of experimental diabetic cardiomyopathy so far. Therefore, the aim of our study was to evaluate myocardial deformation in combination with the underlying changes in the expression of structural components during the pathogenesis of STZ-induced type 1 diabetes mellitus-associated diabetic cardiomyopathy.
Here, we observed an impairment of GLS and GLS rate after STZ application, which is consistent with findings in patients suffering from heart failure with preserved ejection fraction 13 and diabetes mellitus 14 . Additionally performed regional peak strain analysis implicated that these changes in GLS, which were the most prominent at 9w, were probably caused by less myocardial contraction of the anterior and/or posterior base. Supporting the recently proposed hypothesis that a loss in longitudinal deformation behaviour is compensated by improved circumferential deformation behaviour 15 , we detected an enhancement in GCS and GCS rate in 6w and 12w STZ Wall. Bar graphs represent the mean ± SEM. Data were analysed with One-way ANOVA or Kruskal-Wallis test (*p < 0.05; **p < 0.01, ***p < 0.001, ***p < 0.0001 versus corresponding control; † p < 0.05, † † p < 0.01, † † † p < 0.001, † † † † p < 0.0001 versus the 6w STZ; ‡ p < 0.05, ‡ ‡ p < 0.01, ‡ ‡ ‡ p < 0.001, ‡ ‡ ‡ ‡ p < 0.0001 versus the 9w STZ; n = 12/controls and n = 14/STZ). (2020) 10:3629 | https://doi.org/10.1038/s41598-020-60594-2 www.nature.com/scientificreports www.nature.com/scientificreports/ mice. In contrast, GCS and GCS rate of 9w STZ mice were reduced relative to 6w and 12w STZ mice, and this effect was paralleled by a lack of changes in segmental contraction. Interestingly, Shepherd and colleagues, who also used a STE-based approach, detected no change in GLS in 6w STZ mice, whereas GCS, global radial peak strain and subsequently strain rates were reduced 24 . In contrast to our study, in which we performed quantitative analysis on each of the 6 single segments, Shepherd and colleagues grouped the long-and short axis images only into 2 segments, which disables a direct comparison between the studies. However, we believe that by the grouping of segments, the added value of the regional analysis, which is actually the advantage of this method, goes lost. Therefore, we are not surprised that the authors only reported regional changes of the global radial peak strain, but not of the GLS. In addition to GLS and GCS, we also investigated DWS as a marker of diastolic function. Non-invasive assessment of LV wall stiffness in experimental models was first established by Takeda et al. 25 who reported an inverse correlation of DWS with myocardial stiffness. Despite the lack of correlation in larger patient cohorts 26 , reduced DWS may be associated with a decreased diastolic function 27 , which has been confirmed by our findings.

Scientific RepoRtS |
(2020) 10:3629 | https://doi.org/10.1038/s41598-020-60594-2 www.nature.com/scientificreports www.nature.com/scientificreports/ detected a reduction in GLS, which is mainly determined by longitudinal fibres. In parallel, collagen I deposition was increased at 9w STZ. Interestingly, we observed less collagen I at 12w post-STZ compared to 9w STZ, paralleled by lower Lox and LoxL2 levels, as markers for collagen cross-linking 36 . The reduction in collagen at 12w STZ www.nature.com/scientificreports www.nature.com/scientificreports/ per se is contradictory to already published studies, reporting higher collagen levels 3,23,31 . Only one study, investigating the effect of thioredoxin-interacting protein, reported less myocardial collagen at 16w STZ compared to controls 37 . But alterations in extracellular matrix, especially collagen, are not only determined by production and cross-linking, but also by degradation. Extracellular matrix degradation is predominantly regulated by MMPs, which are counteracted by TIMPs. Several studies have previously connected the dysregulation in MMPs/TIMPs and subsequent collagen deposition in STZ-induced type 1 diabetes mellitus-associated diabetic cardiomyopathy with an impairment of myocardial function 31,38 . With respect to our study, changes in cardiac MMP-8 and MMP-13 gene expression levels were observed after STZ application, suggesting a time-dependent modulation in collagen degradation.
Recently, we have demonstrated layer-specific changes in extracellular matrix proteins, which resulted in layer-specific changes of strain parameters in an experimental model of isolated subendocardial fibrosis 20 . Based on these findings, IMS was applied to further investigate global and regional-dependent changes in cardiac protein signature. In accordance to previous proteome analysis in experimental type 1 39 and type 2 diabetes 40 , we detected alterations in intensity distribution of proteins, which are associated with cardiac energy metabolism and contractile function and play a crucial role in the progression of diabetic cardiomyopathy 41 . The higher distribution of α-actin, myosin light chain 3, ATP synthase, and titin in 6w STZ mice compared to control animals, might be a compensatory mechanism to counteract the reduced GLS and could hypothetically be reflected in the increased GCS. Similar observations were made in cardiac contractility modulation patients, where improved cardiac function was associated with an increased expression of contractile proteins, including myosin light chain and titin 42 . In contrast to that, the intensity distribution of all mentioned protein signatures was reduced in 9w STZ mice versus controls, which was paralleled by an increased collagen I/III ratio. These changes probably underlie the impaired GCS and GLS, indicative for reduced myocardial deformation behaviour. At 12w post-STZ, mice exhibited decreased myocardial α-actin, myosin light chain 3, and ATP synthase levels compared to controls. Nevertheless, titin was unaffected in 12w STZ mice, which might be an explanation for the lack of significant improvement in GCS compared to the respective controls, whereas GCS at 12w post-STZ was increased relative to 9w post-STZ.

Future directions and limitations.
Given the high sensitivity of STE-derived myocardial deformation data (global peak strain and global peak strain rate), this technique is recommended as a non-invasive diagnostic method for extensive clinical use 43,44 . Both data allow the early detection of myocardial dysfunction of different aetiologies and provide prognostic information. They are easy to acquire for a trained expert in echocardiography and helpful for therapeutic decision making and follow-up examinations. Despite these valuable advantages, clinical application of STE is still challenging 45 . Only standardization of measurements, continued training, and quality assurance ensure less measurement differences between various vendors/observers and allow an accurate interpretation. To date, STE-derived myocardial deformation data only in conjunction with other parameters, such as ejection fraction, are used for clinical decision making. The combination of STE with IMS allows timeand region-dependent alterations on molecular and functional outcome in the pathogenesis of STZ-induced type 1 diabetic cardiomyopathy. Especially, since it was not expected to see this specific cardiac phenotype found at 12w post STZ, including a decrease in collagen content, in comparison to the 9w and 6w post STZ time points. With respect to the existing literature, our findings are contradictory and were only confirmed by Myers et al. 37 . This discrepancy in findings might be explained by the STZ model. Even if the STZ-induced type 1 diabetes mellitus model is considered as the standard model for experimental type 1 diabetic cardiomyopathy, it should be taken into account that STZ is a chemical cytotoxic glucose analogue 46 . In this sense, the STZ-induced type 1 diabetes mellitus model can be regarded as a model of secondary (toxic) diabetes mellitus, which differs from  Table 1. ROC analysis of the intensity distribution from MALDI-IMS-derived identified proteins. Peptides with lowest mass difference to the LC-MS/MS reference list value and with similar discrimination characteristics were assumed as a match.
www.nature.com/scientificreports www.nature.com/scientificreports/ human type 1 diabetes mellitus, which is an autoimmune disorder 46,47 . Furthermore, since no uniform protocol to induce type 1 diabetes mellitus in rodents exists 48,49 , this consequently results in different severities of experimental diabetic cardiomyopathy (e.g. rats display a more severe phenotype compared to mice), which ultimately leads to a poor comparability of experimental type 1 diabetes mellitus models in rodents. This highlights the need for comprehensive cardiac phenotyping in the context of experimental diabetic cardiomyopathy and supports the value of our study.

conclusions
This experimental study in an early stage model of type 1 diabetes mellitus-associated diabetic cardiomyopathy illustrates the strengths of two-dimensional STE to detect subtle, segmental changes in LV deformation important for the early diagnosis of diabetic cardiomyopathy and, in general, to assess ongoing cardiac remodelling processes. The combination of STE with IMS, allowing the respective assessment of region-dependent functionality and protein expression, is unique in its kind. We are the first to our knowledge combining both imaging techniques. The present study illustrates that during the development of STZ-induced type 1 diabetes mellitus-associated diabetic cardiomyopathy the expression of structural and contractile proteins shifts in a timeand region-dependent manner.

Methods
Animals. Type 1 diabetes mellitus was induced in eight-week-old male C57BL6/j mice (Charles Rivers, Sulzfeld, Germany) via intraperitoneal injection of STZ (50 mg/kg body weight; Sigma-Aldrich, Taufkirchen, Germany, n = 14) on 5 consecutive days 4 . In parallel, controls received the appropriate volume of 0.1 mol/L sodium citrate (n = 12). Blood glucose measurements using the Accu-Chek Aviva ® (Roche Diabetes Care Deutschland GmbH, Mannheim, Germany) were performed one week before STZ application and afterwards once a w to confirm hyperglycemia after 4 hours (h) fasting. Mice were group housed under standard housing conditions (12 h light/dark cycle, 50-70% humidity, 19-21 °C) within a specific pathogen free facility with food and water ad libitum. Cages were enriched with houses, bedding material, and gnawing sticks. All investigations were performed in accordance with to the European legislation for animal welfare (Directive 2010/63/EU) and approved by the local ethics committee (Landesamt für Gesundheit und Soziales, Berlin, G0254/13).
Echocardiography. Six, 9, and 12w after type 1 diabetes mellitus induction, echocardiographic measurements were performed using a Vevo 3100 Imaging System coupled to a 30-MHz linear-frequency transducer (FUJIFILM VisualSonics, Toronto, ON, Canada). According to our previously published standard operating protocol 20 , mice were anesthetized with isoflurane (Abbott Laboratories, Abbott Park, IL, USA) and fixed in a supine position. To obtain constant heart rates, isoflurane concentrations were reduced to 1-2%. Next, mice were depilated using hair-removal crème, and pre-warmed ultrasound gel (Parker Laboratories, Fairfield, NJ, USA) was placed on the chest. B-and M-Mode images were acquired in parasternal long axis and short axis view. A trained expert in small animal echocardiography performed STE analysis of the B-Mode images in a semiautomatic manner using the VevoStrain software package (FUJIFILM VisualSonics), as reported previously 20 . Determination of cardiac systolic function was performed in a semiautomatic manner from the respective B-mode images (long axis view) using the monoplane Simpson method of disks. To calculate LV dimensions and LV mass, the acquired M-Mode images of the parasternal long axis view were used. In detail, three independent M-Mode images and 3 cardiac cycles were analysed. To determine diastolic function, DWS was calculated as previously described 27 . Tissue preparation for molecular investigations. To examine gene and protein expression, an additional set of experiments was performed. To this end, type 1 diabetes mellitus was induced as mentioned above. At the respective time points (6w, 9w, and 12w after STZ application or control treatment), 5-6 mice/group were euthanized via cervical dislocation and the LV was excised, immediately snap frozen in liquid nitrogen and stored at −80 °C until further processing. To confirm characteristics of the echocardiographic set, blood glucose levels, body weight, LV weight and LV weight/body weight ratio derived from the second set of animals were depicted (Supplemental Fig. 3).
RNA isolation and gene expression analyses from heart tissue. LV RNA was isolated using the TRIzol ® method, as described in detail elsewhere 50 . According to manufacturer's protocol, the RNA pellet was purified with the RNeasy Mini Kit (Qiagen, Hilden, Germany) and subsequently transcribed into cDNA using the High Capacity Kit (life Technologies, Darmstadt, Germany). Real-time PCR was performed using the following gene expression assays (all provided by Applied Biosystems, Darmstadt, Germany): Col1a1 (Mm01302043_g1), Lox (Mm00495386_m1), LoxL-2 (Mm00804740_m1), MMP-8 (Mm00439509_m1), MMP-13 (Mm00439491_m1), and TIMP-1 (Mm00441818_m1). Gene expression levels were normalized towards CDKN1b (Mm00438167_g1) and expressed as the n-fold change to the respective control group, set as 1.
Immunohistology. Frozen LV were embedded in Tissue-Tek (Sakura, Staufen, Germany) for cryosectioning and subsequent staining. As described previously 51 , specific antibodies directed against Collagen I (Chemicon, Limburg/Lahn, Germany) and Collagen III (Merck Millipore, Darmstadt, Germany) were used. Quantitative analysis of the positive area (%)/heart area (HA; mm 2 ) was performed by digital image analysis on a Leica DM2000 light microscope (Leica Microsystems, Wetzlar, Germany) at 100× magnification.
Left ventricular proteome analysis by MALDI-Imaging mass spectrometry. MALDI-IMS was performed as previously described 52 . In brief, formalin-fixed paraffin embedded (FFPE) LV sections of 6 µm were transferred onto Indium-Tin-Oxide slides (Bruker Daltonik, Bremen, Germany), dewaxed and passed through decreasing concentrations of ethanol 53 . Next, trypsin solution was applied onto the sections via an automated spraying device (ImagePrep; Bruker Daltonik) for 3 h incubation (37 °C; moist chamber). Finally, matrix solution (α-cyano-4-hydroxycinnamic acid) was applied using ImagePrep. For MALDI-IMS data acquisition, the positive ion reflector mode on an MALDI-TOF/MS (Autoflex III; Bruker Daltonik) was used. The corresponding detection range and raster width were set as 800-3500 Da and 100 µm, respectively. After MALDI-IMS experiments, tissue sections were stained with haematoxylin and eosin (H/E) to define the region of interest. As previously reported 53 , m/z values were identified via complementary protein identification using the "bottom-up"-nano liquid chromatography (Dionex Ultimate 3000, Thermo fischer) -MS/MS (ESI-QTOF; Bruker Daltonik) approach. Mass spectra were analysed using the mascot search engine (version 2.4, MatrixScience, UK) searching the UniProt database. To this end, the following parameters were set: (i) taxonomy: mus musculus (ii) proteolytic enzyme: trypsin, (iii) peptide tolerance: 10 ppm, (iv) maximum of accepted missed cleavages: 1, (v) peptide charge: 2+, 3+, 4+; (vi) variable modification: oxidation (M); (vii) MS/MS tolerance: 0.05 Da, and (viii) MOWSE score > 25. Identification of MALDI-IMS m/z values by using an LC-MS/MS reference list requires the accordance of more than one peptide (mass differences <0.9 Da) to subsequently correctly assign the corresponding protein 54 .
Statistical analysis. All Data are expressed as mean ± SEM and statistical analysis was performed using the GraphPad Prism 8 Software (GraphPad Software, La Jolla, CA, USA). Differences between the groups were assessed using the One-way ANOVA (Fisher´s LSD post hoc test) or Kruskal-Wallis Test (Dunn´s post hoc test) and considered significant at a value of p < 0.05. For IMS-proteome data, statistical analysis was performed using the SCiLS Lab software (Version 2015b, SCiLS GmbH, Bremen, Germany), as described before 52 . To define common molecular features among the sample sets, unsupervised multivariate principal component analysis for mass spectra was applied for each time point separately. The mesoderm was defined at the corresponding H/E staining as region of interest from STZ as well as controls. Additionally, principal component analysis was performed in the STZ group at the different time points. A receiver operating characteristic (ROC) analysis was performed pairwise (STZ vs control) to identify discriminating properties (peptides) in the mesoderm of LV tissue specimens at 6w, 9w, and 12w after type 1 diabetes mellitus induction. For those peaks with a ROC score of AUC > 0.6 or <0.4, a univariate hypothesis test (Wilcoxon rank sum test) was used to ensure the statistical significance of m/z values. Results with significant differences (p-value ≤ 0.01) in the Wilcoxon test were assumed to be potential markers.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.