Chapter 3 Three dimensional characterisation of cardiac cellular organisation and structure
Marked changes in structural organisation and electrical activity are a hallmark of HF (Schaper et al. 2002). However, the exact relationships between these changes are not fully understood. Although in the previous chapter, fibrosis was associated with altered electrical activity, its effect on the cellular organisation was unclear. By capturing the changes in cellular arrangement associated with fibrosis, the mechanisms behind its influence on abnormal electrical function can be better understood. To investigate cellular organisation in the presence of fibrosis, both non-fibrotic and fibrotic tissue needs to be characterised and compared over a multicellular volume and at a high enough resolution to capture the 3D cellular arrangement and architecture. As mentioned in the previous chapter, the three SHR cohorts (6, 12 and 18 months) provide the ability to examine the cellular organization in tissue with variable levels of fibrosis. Furthermore, by also capturing the electrical function of the SHR cohorts (the methods for which are presented in Chapter 4), the structural and electrophysiological relationships can be compared in the progression towards HF. This chapter covers the methods for structural characterisation. A background on three dimensional imaging is provided, followed by the development of an imaging protocol to capture the cellular arrangements. Finally, tissue level analysis of the extent of fibrosis in the SHR cohorts is then presented.
Three dimensional tissue imaging
In the past, the majority of imaging studies have made use of thin sections to provide information about cellular organisation within two-dimensional views of biological tissue (Richardson & Lichtman 2015). Tissue cellular organisation however is inherently 3D, requiring modern day researchers to contend with imaging of volumes. However it is difficult to image large volumes of tissue, as imaging tissue thicker than a few microns is not possible. This is primarily due to the collection of out-of-focus light from throughout the section which is further exacerbated by light scattering.
To acquire 3D volumes researchers have created methods of imaging planes via optical or physical sectioning. Physical sectioning techniques include microtome serial sectioning and high volume extended (block face) imaging. Both these methods require embedding of tissue and subsequent sectioning and imaging at regular intervals to obtain imaging volumes (Anderson et al. 2003). The application of serial imaging is limited to small volumes and is time consuming, while also being prone to alignment issues (Manova-Todorova et al. 2015). Although the method of block face imaging removes alignment issues of serial sectioning, the method itself is destructive, as once each section is imaged it is milled away to reveal the next block surface. Furthermore, physical sectioning methods are, in general, time consuming due to the requirement of embedding, staining and sectioning.
Non-sectioning methods for imaging volumes are less time consuming and avoid alignment issues, and the same tissue sample can, in principle, be imaged multiple times. These techniques encompass optical sectioning techniques such as laser-scanning confocal microscopy, laser scanning two-photon microscopy, parallelized confocal microscopy (i.e., spinning disk), computational image deconvolution methods, and lightsheet microscopy (Lichtman & Conchello 2005; Reynaud et al. 2008; Mertz 2011). Optical sectioning techniques function by distinguishing in-focus and out-of-focus background light. Essentially they provide information about a single plane by reducing the contributions from other portions of the volume. In turn these methods allow access to image data from a thin section in a thick sample.
Although optical sectioning techniques reduce the problem of out of focus light, they remain limited in their imaging depth by a number of different factors. In many tissue types, tissue pigment and fluorescent molecules may be present or are introduced during processing, leading to autofluorescence and cloaking of labelled structures of interest. Furthermore in most biological tissues there is an intrinsic translucence. The lack of clarity reduces the ability to capture sharp images and becomes progressively more of an impediment at greater depths. Overall, tissue pigmentation and autofluorescence can be resolved by simple measures such as bleaching and appropriate fluorescent labelling (Croce & Bottiroli 2014). On the other hand the causes of the tissue translucency must be considered when trying to reduce its effects on imaging depth. Extended volume imaging (EVI) has sought to account for these issues by utilising both optical sectioning and block face imaging to acquire serial images in pre-labelled embedded tissue specimens. Confocal microscopy is used to image the maximum depth possible, the imaged layer is then subsequently removed and the process is repeated.
However the process of EVI is also time consuming and the sample is lost following data acquisition.
More recent methods of volumetric imaging have made use of clearing techniques (Chamberlain & Tang 2007; Chung & Deisseroth 2013; Renier et al. 2014). First invented in 1914 (Spalteholz 1914), clearing techniques rely on refractive index matching or removal of components to improve tissue transparency and hence, imaging depth. Tissue translucency is produced by light scattering; the light rays are deviated numerous times as light is reflected off molecules, membranes (predominantly lipids), organelles, and cells. Therefore, tissue translucency limits the ability of deeper imaging. Light scattering of the tissue can be removed by eliminating the majority of the light-impeding components and replacing them with a solution that matches the refractive index of the tissue. By equilibrating the refractive index via the removal of inhomogenous components (commonly lipids and water) that scatter light in the tissue, transparency can be increased and tissue appears to be cleared.
Tissue clearing research is thriving, with new improved protocols being released at a rapid rate (Silvestri et al. 2016). Various adaptions of clearing such as solvent or aqueous based approaches have been developed (Richardson & Lichtman 2015). Aqueous based methods are emerging as the favoured clearing technique and many new protocols have been released in the past few years. These techniques will be discussed further in Section 3.10 of this chapter. Along with advances in clearing, imaging of 3D tissues has become possible due to developments in microscopy over the last few decades. Notably, improvements in microscope objectives have dramatically increased the resolution and imaging depth capacity for acquiring greater tissue volumes (Chamberlain & Tang 2007).
In addition, the reliability and specificity of markers for the structural components of tissue and cells have also improved (Berlier et al. 2003). Overall, the combination of clearing and imaging advancements described above, have led to the procurement of tissue image volumes at greater resolutions and sizes than previously possible (Kuwajima et al. 2013; Epp et al. 2015). Furthermore, where collection of these volumes was previously limited by data storage capacities, technological advances have made storage of large files much more achievable. Developments of novel clearing and imaging techniques have enabled modern scientists to explore detailed 3D structure, and thereby further the understanding of the interrelationship between cellular structure and function. The application of these techniques has become a focus in heart research for understanding the arrangement of cardiac structural features in diseased tissue and their effect on electrophysiological functions (Nehrhoff et al. 2016).
Utilisation of 3D imaging techniques such as clearing may provide a means to correlate structure and function. It is anticipated that the derived structural information will provide insights into the differences in cell level structure of diseased and healthy cardiac tissue. The robust quantification of structural features will help enhance our understanding of structure function relationships and are aimed toward supporting biophysical-based modelling of electrical activation. The following sections present trial staining trials and optimization conducted to develop the final 3D imaging protocol used in this thesis.
Three dimensional structural characterisation criteria for myocardial tissue
Cardiomyocytes and the ECM form the majority of the myocardium and in diseased states such as heart failure, both cardiomyocyte and ECM volumes and their distributions are significantly altered (Bing et al. 1995; A. M. Gerdes et al. 1996; Drazner 2011; LeGrice et al. 2012; Nguyen et al. 2016).
The changes observed in the 3D organization are believed to alter the electrical impulse propagation across of cardiac tissue (Smaill et al. 2013). It is thought that fibrosis delays conduction as activation follows tortuous pathways around collagen barriers (Engelman et al. 2010). There is also evidence that the volume occupied by fibrosis influences the passive distribution of extracellular potential and local currents (Rutherford et al. 2012). Moreover, it is thought that fibrosis displaces cells, which reduces the cell-to-cell coupling and in turn reduces the electrical pathways.
To be able to characterize these changes, in particular, the cellular coupling variations associated with ECM remodelling, 3D non fibrotic and fibrotic cardiac volumes need to be obtained and compared. Within these volumes three main components require identification:
Individual cells (through whole cell or cell membrane staining).
ECM components such as collagen.
Cell-to-cell connections, such as intercalated discs (ICD) or connexins.
In addition, the labelled components have to be captured over a volume large enough to encompass numerous cells and provide a 3D representation of the tissue. As cardiac cells are approximately 10-25 µm in diameter and ~130 µm in length (Gerdes et al. 1986), volumes greater than 130 µm depth are required.
By determining the cellular arrangement, coupling and collagen levels the relationship between altered cellular level myocardial structure and abnormal impulse propagation can be better understood.
Picro-sirius red staining and extended volume imaging
In consideration of all the factors mentioned above picro-sirius red (PSR), a versatile stain, previously utilized to highlight both collagen and cellular content (Pope et al. 2008), was trialed. PSR staining provides a simple method to identify fibrillar collagen networks which serves as an indicator of the level of fibrosis. In Chapter 2, the combination of PSR staining and extended volume imaging (EVI) provided a suitable method of 3D volume characterization of collagen content. Although EVI is time consuming and the sample is milled away during imaging, past studies have utilized this method to characterize the alterations in cardiac tissue structure and collagen organisation due to heart failure (LeGrice et al. 2012) and myocardial infarct (Rutherford et al. 2012). These results and analysis indicated PSR was a suitable method for characterization of 3D structure.
Picro-sirius red staining and EVI analysis
To explore the difference in structure of healthy and fibrotic tissue, high-resolution (~0.6 µm) extended-volume datasets were compared between PSR-stained 12-month WKY and SHR hearts. These volumes were readily available and the process for obtaining them has been described in detail in Rutherford et al (Rutherford et al. 2012).
Figure 3-1 highlights the presence of collagen (shown by the brighter intensity) across a slice from a large volume of SHR (A) and WKY (B) hearts. The SHR slice shows characteristically high levels of endomysial collagen and interstitial fibrosis compared to the WKY sample.
A 200 µm3 subset of this volume was extracted and analyzed with the aim of determining cell-to-cell coupling of healthy WKY and unhealthy fibrotic SHR tissue. Using ITK-SNAP, cellular boundaries highlighted by the PSR were manually segmented. Figure 3-2 shows a transverse slice view of the cell segmentation in the SHR heart volume with an initial indication of the number of cells in the plane. The large amount of endomysial fibrosis enabled the identification and segmentation of the cellular boundaries.
Both transverse (Figure 3-3A) and longitudinal planes (Figure 3-3B and C) were segmented, while connections between cells (i.e. ICDs) were highlighted using a green marker. The cells were then displayed together in a three-dimensional form (Figure 3-3D) to better visualize the number of cells and cell-to-cell connections. Subsequently, five random cells were chosen and the numbers of connected cells were determined. It was found that, on average, each cell was coupled to 9 other cells, based on the one SHR sample.
This process was then applied to healthy WKY tissue. However, the PSR stain was unable to identify the cell boundaries due to a lack of discernible collagen or fibrosis (Figure 3-4A). Cell boundaries were particularly difficult to distinguish in the longitudinal plane (Figure 3-4B).
Overall, the low amount of endomysial collagen in healthy tissue indicated PSR was an unsuitable cell marker. Furthermore, due to the harsh acidic environment of PSR, co-staining of the cell membranes was not possible. In the presence of extensive fibrosis, penetration is reduced by the heavy PSR staining. The combination of these factors meant that although PSR was able to capture fibrotic tissue structure, it was not appropriate for non-fibrotic and fibrotic tissue comparisons. As a result, a new staining and imaging protocol for structural characterization needed to be established.
The following sections of this chapter present the details of the subsequent trials conducted to establish the staining and imaging protocol for 3D cellular structural characterization.
General experimental procedures
All chemicals and reagents were purchased from Sigma (Sigma Pty Ltd, Aus) unless otherwise specified. Solutions and buffers were prepared in our laboratory.
Cardiac sample excision
Cardiac samples for development of the 3D structural imaging protocol were obtained from culled Wistar rats or from the spontaneous hypertensive rats (SHR), University of Auckland. All procedures were performed according to the guidelines of the University of Auckland Animal Ethics Committee AEC (001454).
The basic procedures for harvesting the cardiac samples were:
Rats 250-350 g were anesthetized (5% isoflurane in O2) until loss of paw withdrawal reflex. Heparin (50 I.U) was injected into the LV and allowed to circulate for 1 minute. Hearts were then quickly excised and immersed in cold (4°C) phosphate buffered saline (PBS, in mM: NaCl 137, 2.7 KCl, 10 Na2HPO4, 1.8 KH2PO4).
The coronary system was Langendorff perfused at a constant flow rate (13ml/min) with modified Krebs-Hensleit solution (mM: NaCl 118, KCl 4.75, MgSO4 1.18, KH2PO4 1.18, NaHCO3 24.8, Glucose 10, CaCl2 2.5 (bubbled with carbogen). The hearts were allowed to beat spontaneously for ~2 minutes to flush the blood from the tissue. St Thomas’ solution (in mM: NaCl 120, KCl 16, MgCl2 16.6, CaCl2 1.2, NaHCO3 10, pH = 7.8) was then perfused into the heart to ensure a relaxed state. Following St Thomas’ perfusion, 4% paraformaldehyde in PBS (PFA) was circulated to fix the tissue. Finally, the tissue was cut transversely into ~2-3 mm thick rings and placed overnight in 4% PFA in a rotator at 4°C to complete fixation.
Note: all perfusion solutions excluding the PFA were warmed to 37°C in a water bath prior to use.
Samples were either perfusion-stained in the Krebs solution or diffusion-stained following fixation (Note: fixation methods were varied according to the stain being trialed and are specified for each stain). Unless stated otherwise, all diffusion trials were conducted using 20 µM thick tissue samples sectioned using a cryomicrotome. Staining was conducted in a 100 µL solution of PBS with 0.3% Triton X-100 placed on a shaker at room temperature. A circle of wax was drawn around the tissue sections to contain the reagent liquid. In between each diffusion step, the samples were rinsed in PBS three times for 5-min each.
For antibody staining, samples were permeabilized with 0.3% Triton X-100 in PBS and blocked in PBS containing 10% normal goat serum (GS) over 2 hours. Following blocking, the samples were incubated with the primary antibodies with 10% GS and 0.3% Triton, and then placed into their respective secondaries. Preparations of ventricular tissue in which combinations of primary antibodies were omitted served as negative controls. Non-specific binding of secondary antibodies did not occur in these preparations.
Concentrations, durations and temperatures of the staining steps were varied according to the trial and are described in detail for each respective trial in the following sections.
Unless specified otherwise, all confocal imaging for the staining trials were conducted using 20 µm tissue sections. Imaging was performed using a Nikon TE2000 inverted confocal system equipped with a combination of a 4X (NA 0.13, WD 16.4mm), 20x (NA 0.45, 7.6mm) and 60X oil immersion lens (NA 1.25, WD 180 µm) were used for imaging trials. Appropriate beam splitters and filters were used to filter excitation light and emission fluorescence. Voxel dwell time was set at a constant 1.4 µs for all image acquisitions.
Volumes were acquired with the 60× oil immersion lens typically, 512 × 512 × 180 voxels and with a voxel dimensions 0.41 × 0.41 × 1 µm; 212 × 212 × 180 µm. Larger 3D images (1024 × 1024 × 500 voxels with voxel dimensions 0.6 × 0.6 × 1 µm; 368 × 368 × 500 µm) were obtained with a Zeiss LSM510 META confocal microscope with a 25× objective (LCI Plan-Apochromat 25x/0.8 Imm Corr DIC M27 600µm WD).
Preface and Motivation
List of Published Work
List of figures
List of tables
List of abbreviations
Chapter 1. Introduction
1.1 Cardiac function
1.2 Cardiac Structure
1.3 Electrical activation in the heart
1.4 Heart failure
1.5 Electrical and structural remodelling in heart failure
1.6 Optical mapping
1.7 Computational models.
1.8 Specific aims of thesis
Chapter 2. Quantifying Structural and Functional Differences between Normal and Fibrotic Ventricles
2.2 Materials and Methods
2.3 Tissue modelling
2.5 Discussion & conclusions
2.6 Chapter summary & development of research
Chapter 3. Three dimensional characterisation of cardiac cellular organisation and structure
3.1 Three dimensional tissue imaging
3.2 Three dimensional structural characterisation criteria for myocardial tissue
3.3 Picro-sirius red staining and extended volume imaging
3.5 Cell membrane labels
3.6 Cell coupling labels
3.7 Collagen labels
3.8 Optical mapping label compatibility
3.9 Tissue volume labelling
3.10 Tissue clearing
3.11 Confocal microscopy and analysis
3.12 Chapter Summary
Chapter 4. Optical mapping and electrophysiological characterisation
4.1 Optical mapping system and procedures
4.2 Electrophysiological characterisation
4.3 Experimental protocol
4.4 Data analysis
4.5 Chapter Summary
Chapter 5. Image-based motion correction for optical mapping of cardiac electrical activity
Chapter 6. Structural remodelling and electrophysiological dysfunction in the progression of hypertensive heart disease
Chapter 7. Conclusions and Future work
7.2 Future work
GET THE COMPLETE PROJECT
Cardiac Fibrosis and Ventricular Arrhythmogenesis