Imaging transparent intact cardiac tissue with single-cell resolution.

We developed a new method, SUT (Scheme Update on tissue Transparency), to view cardiac microstructures and unveil the molecular changes underlying cardiac diseases. SUT is an effective method to clear whole-hearts from different species. Over the course of 4 - 6 days we obtained transparent whole-layer left ventricular tissues from mice with only an approximate 1% protein loss. In addition, EAL (Electrophoretic Antibody Labeling) was used to achieve fast antibody labeling by electric force, which significantly reduced antibody incubation time from days to hours. SUT, together with EAL and modern imaging techniques, were successfully used to visualize three-dimensional spatial distribution of various molecules in cardiac tissue. We also observed changes in the number and phenotypes of fibroblasts during post-myocardial infarction in a stereoscopic pattern. We believe that our technique opens a new avenue to explore the mechanisms underlying cardiac diseases.


Introduction
Imaging intact cardiac tissue will increase our understanding of the anatomical and molecular structure of the heart, and will also allow us to monitor the progression of cardiac diseases, such as cardiac remodeling after myocardial infarction (MI) and hypertrophic cardiomyopathy. However, the main obstacle of clearly imaging intact cardiac tissue is the whole-layer light transmittance through thick tissues. The two major factors that affect light transmittance through thick tissues are: (1) the light scattering that occurs between different tissue layers, in which lipids play a primary role, and (2) the light absorbance by endogenous chromophores, mainly heme in hemoglobin and myoglobin [1,2]. There are two corresponding ways to increase tissue transparency: (1) reduce lipids as much as possible, and (2) eliminate heme to the greatest extent. The heart is a blood-rich muscular organ, in which the left ventricle is the thickest. In addition, the heart contains cardiac muscle, extracellular matrix, and fibroblasts, which are the predominant interstitial cells in the mature mammal heart. These components are critical in both reparative responses and pathogenic remodeling post MI [3,4]. To obtain high transparency and view three-dimensional (3D) molecular changes during the progression of cardiac diseases, we need to eliminate most, if not all, lipids and heme in the heart.
As early as 1914, scientists like Werner Spalteholz used chemicals to acquire transparent tissue for anatomical and biomedical studies; however, their methods severely damaged tissues [5]. In the subsequent century, many chemical combinations have been used, including BABB, THF-DBE, Scale, SeeDB, CLARITY, 3DISCO, CUBIC, and PACT-PARS [1,[6][7][8][9][10][11][12][13][14][15]. However, most of these methods were utilized for brain imaging, and some methods, like CUBIC and PARS, were non-specifically applied for whole-body imaging [14,15]. Passive clearing methods are typically most often used for tissue clearing, but the main weakness is their slow clearing speed [15], which has prevented them from efficiently clearing large tissue volumes or whole-organisms. Electrophoretic tissue clearing (ETC) was introduced with CLARITY to accelerate extraction of lipids from large samples, but ETC damages tissues and causes variability in tissue quality [8,16]. Both CUBIC and PARS have been successfully used with perfusion pressure to replace electrophoretic force to clear whole-organisms, but CUBIC seems to be used more widely. Recently, a modified CUBIC protocol was used for mouse heart tissue clearing and imaging, and CUBIC has been shown to enable visualization of the 3D network of cardiac innervation [17,18]. However, EDTP, one of the components of CUBIC, is an ultra-sticky solution and copper-chelator, which makes it difficult to evaluate protein loss after clearing [13]. Based on a previous report, CUBIC caused 24% to 41% protein loss [2]. In contrast, 8% SDS, the clearing solution of PACT-PARS, was not able to effectively clear the heart and also caused heavy tissue swelling [15]. Thus, clearing conditions for different organs are distinct due to variability among tissue structures. Currently, there is no specific and effective clearing method for cardiac tissue.
In the present study, we developed a rapid and effective clearing agent, called SUT (Scheme Update on tissue Transparency), for transparent imaging of whole-layer left ventricular tissue, as well as whole-heart. Through passive clearing, we achieved high transparency of whole-layer left ventricular tissue from 10 week old CD1/ICR mice in 4 to 6 days. We were also successful in applying SUT for whole-heart clearing of CD1/ICR mice (40 g) in 2 days, Sprague Dawley rats (256 g) in 4 days, and pigs (Chinese mini-pig, 35 kg) in 29 days, by infusing SUT through the left ventricular chamber and coronary vascular system. Furthermore, we used light-sheet and confocal microscopy to display the 3D relative spatial distributions of targeted molecules, either in a general mode or with single-cell resolution.

Material and methods
We used 10 week old healthy male CD1/ICR mice purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd., which are listed in the Key Resources Table. These mice were also used for our MI model. We used 8 week old healthy male Sprague Dawley rats purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. Rats used for the final whole-heart clearing weighed 256 g. We also used adult, healthy male Chinese mini-pigs weighing 35 kg supplied by the Animal Experimental Center of Fuwai Hospital, Chinese Academy of Medical Sciences.

Development of SUT
SUT was used as a new cardiac tissue clearing agent. Before clearing, all cardiac tissues were crosslinked with hydrogels to reduce protein loss and preserve the original state of the tissue as much as possible. SUT was created by dissolving 250 g urea, 150 ml TritonX-100 and 80 g SDS in hot phosphate buffered saline (PBS) (0.1M, pH = 7.40) to a total volume of 1 L.

Tissue preparation
10 week old male CD1/ICR mice were deeply anesthetized with 4% tribromoethanol via intraperitoneal injection at a volume of 0.4 ml. The abdomen was then opened, the inferior vena cava (IVC) was punctured with an I.V. catheter (BD, 22GA, 0.9 × 25mm), and the aorta abdominalis was severed. 40 ml of hPBS (PBS with 10 U/ml heparin) was infused through the IVC until all blood was removed. Thereafter, the whole heart with the ascending aorta was removed, perfused with 5 ml hPBS again, and slowly infused with 10 ml 4% paraformaldehyde (PFA) for fixation. The following specific protocols were used in this study: (1) For mouse whole-heart clearing, the heart was directly connected to the perfusion apparatus to continuously perfuse 4% PFA.
(2) For perfusion of whole-layer left ventricles (D = 4.5 mm), the left ventricles were cut off at both roots of the anterior and posterior papillary muscles with a puncher (D = 4.5 mm).
(3) To generate cardiac 1.5 mm cross-sections, pectinatory knives were used to cut the heart perpendicular to the long axis (3 sections per heart).
(4) For rat whole-heart clearing, the rat was anesthetized with pentobarbital sodium via intraperitoneal injection at a dose of 30 mg/kg through an I.V. catheter (BD, 20GA, 1.1 × 48 mm) via IVC puncture, and infused with hPBS (at least 200 ml). The heart was harvested and a mouse lavage needle was inserted into the aorta. The aorta was then ligated, perfused with 20 ml hPBS again, and slowly infused with 40 ml 4% PFA. Thereafter, the heart was directly connected to the perfusion apparatus to continuously perfuse 4% PFA for 2 hours.
(5) For pig whole-heart clearing, Chinese mini-pigs (35 kg) were anesthetized via an intramuscular injection of pentobarbital sodium at a total dose of 30 mg, followed by tracheal intubation. After disinfecting and draping, the thoracic cavity was opened by median sternotomy, followed by pericardium opening and heparinization at a dose of 400 U/kg. After the activated clotting time reached 480 s, an infusion needle was inserted into the root of the aorta, and the heart was perfused with cold cardioplegia for 3 minutes (the infusion pressure was maintained steady at 150 mmHg). When the heart stopped beating at diastole, it was cut out and ligated at the superior vena cava and four pulmonary veins. A catheter was inserted into the aorta, which was ligated. Next, the heart was connected to our handmade perfusion device, on which the heart was perfused with 2 L hPBS, followed by continuous perfusion of 4% PFA for 4 hours.

MI model
10 week-old male CD1/ICR mice were used for the MI model. Briefly, mice were anesthetized via an injection of 0.4 ml 4% tribromoethanol into the abdominal cavity, followed by tracheal intubation with mechanical ventilation (tidal volume: 2.0 ml, respiratory rate: 120 times per minute). Chest wall muscles were dissected, and some intercostal muscles between the fourth and the fifth ribs were removed. A retractor was use to open the thoracic cavity, followed by ligation of the left anterior descending branch of the coronary artery to induce MI. Hearts were harvested at days MI-1, MI-3, MI-7, and MI-14, respectively. Those mice whose MI areas were not located in the field of circular ventricles (D = 4.5 mm) were excluded from our experiments.

Tissue-hydrogel matrix formation and tissue clearing
(1) Circular and cross-sections: After acquiring parameter measurements, specimens were fixed in 4% PFA at 4°C overnight, followed by incubation with 4% acrylamide (including 0.25% VA044) at 4°C overnight. The samples were placed in a 37°C water bath for 3 hours to form a tissue-hydrogel matrix. Each tube was degassed with nitrogen for 2 minutes [15] then transferred into the SUT solution with continued shaking and incubation at 37°C for 4 to 6 days.
(2) Whole heart: After 4 hours perfusion with 4% PFA at room temperature, the hearts were washed for 2 hours with 0.1 M PBS, followed by another perfusion with 4% acrylamide overnight at room temperature. Thereafter, the hearts were washed for 2 hours with 0.1 M PBS again, followed by perfusion with PBS (including 0.25% VA044) and degassing for 1 hour at room temperature. Then, the hearts were perfused with PBS (including 0.25% VA044) and degassed for 3 hours in a 37°C water bath to form a tissue-hydrogel matrix. Finally, PBS (including 0.25% VA044) was replaced with SUT solution, which was replenished every day in a 37.5°C water bath until the heart developed satisfied transparency. After washing in PBS, the cleared tissues were frozen directly at −80°C for long-term storage if they were not to be instantly used.

Immunostaining and imaging of SUT-cleared mouse cardiac tissue
Before immunostaining, cleared cardiac tissue sections were washed with 0.1 M PBS (pH = 7.40) for more than 12 hours with at least 8 changes of PBS until no foam residue was present. All antibodies used in the present study (Table 1) were diluted in PBS T (0.1 M PBS containing 5% goat serum, 0.1% TritonX-100, and 0.01% sodium azide). Tissue sections were incubated with the primary antibody of interest (1:400 -1:200) for 2 to 4 days at 4°C, followed by PBS washes for more than 12 hours. The sections were then incubated with the appropriate secondary antibody (1:400 -1:200) for 2 to 4 days at room temperature. After another 12 hours of washing in PBS, sections were submerged in sRIMS for 2 days. After incubation in sRIMS, tissue sections were hooked and immerged in the sample chamber filled with sRIMS for light-sheet imaging using an EC Plan-NEOFLUAR 5 × /0. 16

Parameters measurement
LENS Transmission Meter (LS108D, light aperture 0.5 mm) was used for light transmittance measurements of the 1.5 mm heart cross-sections (6 sections per group). After 5 days of clearing, six different points that were randomly selected from each section were scored and averaged. Spectrum Transmission Meter (LS108H, light aperture 3.0mm) was used to measure light transmittance of circular whole-layer left ventricular tissue (8 ventricles per group). After 5 days of clearing, each sample was measured three times and averaged. Protein loss was assessed by the BCA (bicinchoninic acid) protein assay. All tissues were cleared for 5 days using respective solutions in an equal volume of 20 ml. Each solution before tissue clearing served as the blank, and was then normalized to the weight of the sections before clearing.
To measure increases in tissue volume, every PFA-fixed adult mouse heart was cut into one circular whole-layer left ventricle (D = 4.5 mm) through both roots of the anterior and posterior papillary muscles with a puncher (8 sections per group), and cleared for 5 days. All sections were imaged with a conventional camera (CANON EOS 700D) before and after clearing. The size of each specimen was outlined and calculated using Image J and normalized to the pre-SUT measurements. All sections were weighed before and after SUT clearing and normalized to the weight before clearing.

Electrophoretic antibody labeling
We utilized our new EAL method to achieve fast primary antibody labeling by electric force. We constructed the equipment ourselves (Fig. 1). The EAL protocol is detailed here: (1) We prepared a 15% non-degenerated gel (a combination of 6.9 ml dH 2 O, 15ml 30% acrylamide, 7.5 ml 1.5M Tris pH8.8, 300μl 10% ammonium persulfate, and 20 μl TEMED), and injected the liquid gel into the assembled gel container, piston and gel pillar and gel cap container to form a solid gel and gel cap; (2) After gel cooling, the gel pillar was removed, washed with 0.1M PBS, and then the hole was filled with PBS T ; (3) We next set the tissue in the sampling tube, put the tube into the gel hole carefully, removed excess PBS T , filled the tube with antibody dilutions, and then sealed the opening with a gel cap, and made sure no bubble residues remained; (4) Lastly, we assembled all individual parts to form a whole device, added electrophoretic buffer (0.1M PBS: dH 2 O = 1:30), connected the device to the electrophoresis apparatus, adjusted the parameters and started electrophoresis.

Statistical analysis
For multiple comparisons of light transmittance, protein loss ratio, and the percentage of weight and volume increases, one-way ANOVA followed by Bonferroni post hoc was used to define statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001) with SPSS 19.0.

SUT clearing efficacy compared to established clearing methods
Before clearing, we referred to both CLARITY and PACT to process cardiac tissues [8,15]. Tissues were hybridized with paraformaldehyde and acrylamide to form a tissue-hydrogel complex for in situ protein fixation, to insure that the small molecules present in the cardiac tissues were retained. We used mouse circular whole-layer left ventricular tissue (D = 4.5 mm) to test the clearing effects of some existing clearing methods, such as ScaleA2, SeeDB, CUBIC-1 reagent, and PACT ( Fig. 2(A)). CUBIC-1 reagent achieved effective transparency in a few days, while the other three failed to clear cardiac tissues effectively. EDTP, one of the components of CUBIC-1 reagents, was used to make the clearing solution, but the solution was too sticky to prepare, and it was difficult to precisely measure protein loss via Nanodrop spectrometry. Also, due to its copper-chelating property, we were unable to use the BCA Protein Assay Kit to measure protein loss. Thus, we proceeded with a chemical screening to develop an effective, rapid and convenient clearing chemical formula.
We used 1.5 mm mouse cardiac cross-sections for chemical screening. We began our rescreening using SDS, the clearing solution of CLARITY and PACT-PARS. We set concentration gradients and used light transmittance, protein loss ratio, and the percentage of weight increase as indicators to evaluate SDS clearing effects. After 5 days of clearing, all groups had deficiencies in light transmittance. During the preparation of solutions, we found that at SDS concentrations greater than 12%, it was difficult to dissolve SDS in PBS and it was too thick to be added into other clearing solutions as well. We eventually obtained the optimal concentration range of SDS for the clearing solution, which was 4% -12% ( Fig.  2

(B)-2(D)).
Through screening of many other clearing chemicals, we found that urea and TritonX-100 could be added to the SDS solution to increase transparency of cardiac tissue. After testing different combinations, we eventually determined that a combination of 25% (wt/vol) urea and 15% (vol/vol) TritonX-100 with 8% (wt/vol) SDS was the optimal clearing solution. We named this solution SUT (Fig. 2(E)-2(G)). In our experiments, we found that 25% urea and 15% TritonX-100 were critical for increasing light transmittance and for preserving protein (Fig. 2(H) and 2(I)), while 15% TritonX-100 was the key factor that influenced tissue morphology (Fig. 2(J)-2(K)). In addition, the original pH of this compound was ~8, and it increased to a range of 8 -9 after tissue-clearing, which is optimal for fluorescent proteins [14].
We found that SUT was better than 8% SDS for increasing tissue volume (Fig. 2(J)), which was suboptimal for the single use of 8% SDS. Therefore, our data indicate that SUT is a highly effective and convenient cardiac tissue clearing solution.

SUT is compatible with immunohistochemistry
To test the compatibility of SUT with immunohistochemistry for identifying cardiac structures, we incubated SUT-treated cardiac tissues with antibodies and imaged them using light-sheet microscopy and single-photon confocal microscopy. To display the 3D local microscopic morphology of the normal left ventricles of mice and show the relative spatial distributions of those common molecules, we used two-antibody combinations together with DAPI to reveal their relative positions. As shown in Fig. 3 and Visualization 1, Visualization 2, and Visualization 3, SUT was compatible with immunohistochemistry, and small molecules were well retained after SUT clearing. sRIMS was used to balance the refractive index (RI) of different tissue layers and for long-term storage [15]. However, to obtain intact tissue structure, we advise complete imaging within one month after antibody-labeling to prevent fluorescence decay. Also, it is better to freeze cleared tissues directly at −80°C or lower for long-term storage, and thaw when needed.
We also found that SUT, with an RI around 1.400, is an effective RI matching solution that can be used for cardiac tissue clearing after antibody incubation to acquire more transparency for imaging. After antibody staining, the whole-heart will lose its high transparency, which cannot be corrected by sRIMS even after 7 days of perfusion. However, SUT can achieve high transparency in only several hours, but imaging should be completed as soon as possible because a longer incubation of cardiac tissues in SUT will reduce fluorescence.
Although the discussion of other-organ clearing is beyond the focus of this text, we have some evidence of SUT clearing and stereoscopic imaging of brain, lung, liver, kidney, and testis.

Whole-heart clearing through coronary circulation
For whole-heart clearing, we designed and constructed a perfusion apparatus, then connected the system to a mouse aorta using a 1 ml syringe needle; for rat heart, we used a mouse feeding needle. After the formation of tissue-hydrogel complexes, we perfused SUT through the left ventricle with a peristaltic pump, and obtained a highly transparent mouse heart in 2 days (Fig. 4(A)) and rat heart in 4 days ( Fig. 4(B)).
We harvested the pig heart through a common surgical procedure, and perfused it with cold cardioplegia to protect cardiac muscle cells. We then connected the aorta to a Double Head Rollerpump and our homemade container for perfusion with various solutions through the coronary vessel system. After tissue-hydrogel complex formation and 29 days of SUT perfusion, we successfully obtained a transparent pig heart ( Fig. 4(C)).

Molecular changes during cardiac remodeling post MI
MI is a global health issue, and fibroblasts play an important role in cardiac remodeling after MI [19]. Hence, it is important to reveal fibroblast behavior during MI to understand the underlying mechanisms of post-MI cardiac fibrosis. Vimentin has been used as a marker of fibroblasts in both normal and injured myocardium [19,20]. We ligated the left anterior descending branches of CD1/ICR mouse hearts to create a MI model, cut the left ventricles of the normal and infarcted hearts at days MI-1, MI-3, MI-7, and MI-14. We then stained these tissues with anti-actin and anti-vimentin antibodies and DAPI after clearing with SUT. The samples were imaged with single-photon confocal microscopy. We observed that the number of fibroblasts increased by day MI-1; the fibroblasts started to crosslink, and their quantity peaked rapidly at day MI-3. The fibroblasts also squeezed together and were highly coupled. On day MI-7, some of the squeezed fibroblasts began to change phenotypes, and the corresponding intercellular spaces were dilated. On day MI-14, almost all of the fibroblasts underwent phenotypic changes, showed less cytoplasmic processes, and sustained dilated intercellular spaces. Our data indicate that these fibroblasts differentiated into myofibroblasts (Fig. 5). Previous studies proposed that cardiac post-infarction healing is divided into three phases: the inflammatory phase, the proliferative phase, and the maturation phase [21][22][23].
Fibroblasts are the dominant cell type during the proliferative phase and undergo dramatic phenotypic changes [24]. According to our results, the transition from the inflammatory phase to the proliferative phase occurred at approximately day MI-7. Thus, this window of time is important for regulating fibroblast phenotype-changing and manipulating post-MI cardiac fibrosis.

Electrophoretic antibody labeling
Our SUT clearing method is satisfactory for easily achieving tissue transparency, but followup deep antibody labeling is difficult due to the dense distribution of cardiac muscles. Based on our experience, antibodies can stain the endocardium and superficial myocardium easily, but passive incubation of the first and secondary antibodies can span days to weeks, which influences the effectiveness of this method.
Antibodies are proteins, and nearly every protein has an isoelectric point. At the isoelectric point, no total charge will be detected at minimal solubility. Although different proteins have different isoelectric points, most are close to 5.0. In an environment in which pH is around 7.40, proteins will dissociate into anions, which carry negative charges. Thus, we speculated if charged antibodies can be forced to pass the poly-porous tissue-hydrogel matrix slowly, antibody incubation will obtain higher efficiency.
For EAL, we designed and performed a set of experiments, in which we adjusted the components of acrylamide gel to support antibody dilutions and electric conduction.
We used different means of labeling the cleared interventricular septum with primary antibodies, and shaking incubation was done for secondary antibodies at room temperature for 2 days. We found that 2 hours of EAL plus 1 hour of still incubation was better than 3 hours of shaking incubation (Fig. 6). Fig. 6. Comparison of regular shaking incubations in EAL labeled cardiac tissues. All tissues were labeled with the antibody against cardiac Troponin I followed by incubation with the second fluorescent antibody, then imaged by Leica SP8 single-photon confocal microscopy with the HC PL APO 10 × /0.40 CS and HC PL APO 40 × /1.10 W CORR CS2 objectives. The excitation light intensity of B was three times of A and C. Scale bars (40 × objective) = 50 μm; scale bars (10 × objective) = 200 μm. (A) Primary antibody: EAL was labeled for 2 hours with constant voltage at 80 V, and the start and end currents were 30 mA and 7 mA, respectively. (B) Primary antibody: shaking incubation for 3 hours at room temperature; secondary antibody: shaking incubation for 2 days at room temperature. (C) Primary antibody: shaking incubation for 2 days at room temperature; secondary antibody: shaking incubation for 2 days at room temperature.

Conclusions
In the present study, we used 3D models to present the spatial distribution of common molecules in the mouse heart, and we are the first to demonstrate fibroblast phenotypic modulation during post-MI remodeling using a 3D model. We also investigated fast antibody labeling of cleared tissues by using electric force to reduce the primary antibody incubation time from days to hours. The methods we established open new avenues of research to further understand the molecular basis of cardiac disease progression, including MI.
SUT, a new tissue-clearing agent, can achieve fast and effective passive clearing of cardiac tissues with little protein loss. Via coronary perfusion, SUT can clear whole intact pig hearts, and has the potential to also clear other organs. Compared to the PACT reagent, SUT can achieve higher transparency and less protein loss.
Tissue-clearing techniques open a door for improving 3D imaging of molecular structures in intact tissues. However, there were some limitations in the present study. For example, we did not clear aged mouse hearts, which have more extracellular matrix deposition. Thus, aged hearts may need more time to clear and result in less transparency compared to young hearts. Furthermore, coronary artery system perfusion may damage hearts more than passive methods due to the faster perfusion speed. Although our system is still imperfect, it allows us to explore the pathological mechanisms of cardiac diseases in a stereoscopic pattern. In the future, a bank containing 3D microscopic information of human tissues could be built, and, together with macroscopic anatomy, this may present secrets of life directly and vividly before our eyes.