Atrial wall to epi-pericardial collagen expression in β2-AR mice: evidence for atrial, pericardial-dominant fibrotic remodelling
In both right and left atrial tissue sections from the mice, the epi-pericardial layer was an ‘enlarged’ layer partly irreversibly separated from adjoining atrial myo-endocardium in the transgenic mice (Tg) compared to the wildtype health control (Ntg) (Fig. 1A, B). As shown, to characterize fibrotic remodelling and collagen expression from atrial wall to the epi-pericardium, we did Masson’s trichrome and immunohistochemical stainings. Figure 1A, B reports the atrial tissue section of Tg and Ntg mice on Masson’s trichrome staining, with the epi-pericardial areas of Tg mice highly fibrotic with lesions and loss of normal parallel tissue structural alignment (Fig. 1C right ) compare with that of the Ntg (Fig. 1D right). In contrast to the epi-pericardial areas of Tg mice, the atrial walls of both the Tg and Ntg mice, and epi-pericardial areas of Ntg had no significant fibrotic remodelling (Fig. 1E). We then reveal the expression of respective collagen types by immunohistochemical staining to understand the distribution and abundance of various collagen types from the atrial wall to the epi-pericardium (Fig. 2). Epi-pericardial collagen I, III and IV increased profoundly in Tg compared with both epi-pericardial and myocardial collagen I, III, IV respectively, in Ntg mice (Fig. 2A, C–E). In fact, as per normal atrium, the pericardial area in Ntg tightly encased the atrial wall, without any signs of separation between the layer and the myo-endocardium. Indeed, there was no remodelling between the pericardium and the myocardium in Ntg. Furthermore while Tg atrial wall collagen I had a trend to increase compared with that of Ntg, Tg atrial wall collagen IV was significantly reduced compared with that of Ntg (Fig. 2C, E). There was no observational difference in the epi-pericardial and myocardial collagen in Ntg mice (Fig. 2C–E). In sharp contrast, epi-pericardial collagen I, III and IV were significantly higher compared with atrial wall collagen in Tg (Fig. 2C–E).
Atrial wall to pericardial fibrosis. Representative images of Masson’s trichrome staining of atrial fibrosis in transgenic (Tg) fibrotic (A) and non-transgenic (Ntg) non-fibrotic (B) animals at 11 months of age. ×10 magnification images of atrial wall and epi-pericardial images of Tg (C) and (D) Ntg. Ishak fibrosis score of atrial wall to pericardium in Tg and Ntg (E). Mean ± SEM. **p < 0.0001 vs. Ntg by 2-way ANOVA followed by post Tukey, Tg; n = 4, Ntg; n = 2; Scale bars for (A) and (B) = 2000 µm; Scale bars for (C) and (D) = 100 µm.
Collagen types in Tg and Ntg mice. Representative atrial wall to epi-pericardial images of immunohistochemical analysis of various collagen types in fibrotic Tg (A) and Ntg (B) at 11 months of age. Mean fluorescence intensities of collagen I (C), collagen III (D) and collagen IV (E). Mean ± SEM, **p < 0.0001 vs. Ntg. 2-way ANOVA followed by post Tukey, Tg; n = 3, Ntg; n = 2 for collagen I; Tg; n = 4, Ntg; n = 2 for collagen III; Tg; n = 3, Ntg; n = 2 for collagen IV. Scale bars for (A) and (B) left 1000 µm; Scale bars for (A) and (B) right = 100 µm.
Atrial epi-pericardial-dominant fibrotic mice displayed biatrial thrombi
A cornerstone of therapy for patients with heart failure is β-adrenergic receptor blockers13, of which arrhythmia is the major cause of death in patients with heart failure. To gain better insight into transgene dose-dependent response of β2-AR expression in heart failure in mouse heart, Liggette et al., generated a series of cardiac-specific β2-AR transgenic mice with different doses14. The authors found a spectrum of phenotypes from the expression as a critical consequence of level of the doses. While 25-week-old β2-60 (a dose of 60-fold overexpression of β2-AR) mouse had normal heart size and no phenotypic abnormalities, 25-week-old β2-350 (another dose of 350-fold overexpression of β2-AR) mouse had 4-chamber enlargement with severe left ventricular dilatation, atrial mural thrombi and aggressive or delayed cardiomyopathy14. Histological examination of the hearts revealed severe fibrotic replacement of left ventricular myocardium, left atrium enlargement with extensive laminar thrombus and left ventricle enlargement and hypertrophy14. Consistently, 11 months old β2-AR mice grossly had severe 4-chamber enlargement, and biatrial thrombi compared with its littermate control (Fig. 3A). While these recapitulate the cardiomyopathy features reported in Liggett et al., atrial epi-pericardial fibrotic lesion and biatrial thrombi has not been previously reported. Furthermore, given that the epi-pericardial layer apparently appears distended from the atrial wall (Fig. 1A compared with Fig. 1B), which has not been reported in normal atria, we report irreversible dilation of the atria in β2-AR mice at 11 months of age (Fig. 3). These myopathy features are important and open up a new research outlook for AF pathogenesis.
Biatrial thrombi and associated mechanism in Tg. Enlarged four chambers and biatrial thrombi in Tg but not in Ntg (A). Type I collagen and fibrin colocalisation (B). Type I collagen in association with platelet. Scale bars for (A) = 2000 µm; Scale bars for (B) and (C) = 100 µm.
To explain the biatrial thrombi we thought that the epi-pericardial dominant fibrotic remodelling can effect myo-endothelial lining and expose sub-endothelial collagen to provoke thrombosis, independent of atrial rate. Build-up of type I fibrillar collagen increases fibrin turnover and accumulation, and glycoprotein IIb activation and platelet aggregation as hallmarks of thrombus generation. Therefore in an attempt to characterise thrombotic propagation as a result of collagen accumulation, we determined the relative contribution and local distribution of thrombus contents: fibrin and platelets, in relation to type I collagen deposition by immunohistochemical co-staining in the transgenic mice. Images obtained from the staining showed deposition of fibrin onto type I collagen (Fig. 3B), with areas of platelet aggregation enveloped by type I collagen (Fig. 3C). Overlay images of fibrin and type I collagen demonstrate clear colocalisation, while the pattern for platelets was an association; clearly demonstrating thrombogenic potential. Though the result was not shown in real-time imaging of thrombogenic propagation, in line with previous studies15,16, it hints that atrial platelet–fibrin-dependent thrombus generation requires collagen exposure, as a determinative triggering process. Whether this is a high atrial rate clot or tissue-driven architectural process has not been established. Further studies will use novel in vivo methods to demonstrate this in atrial disease.
To demonstrate the emboli on the basis of tissue-driven architectural process, the elucidation of some of the signals that drive endothelial cell clustering and remodelling leading to the thrombogenic potential of platelet–fibrin-dependent thrombus generation upon collagen exposure will enable better understanding of how the heart reorganises their cellular and extracellular components in response to embolic events. The endocardial endothelial cell surface is exposed to constant blood flow, and as such can provide a sensory stimulus to clot formation during a focal atrial epi-pericardial remodelling. Based on the vital obligatory role of platelet endothelial cell adhesion molecule-1 (PECAM-1), as an endothelial cell surface marker, in deciphering such signals in endothelial cells17, we evaluated the clustered immunohistochemical expression of endocardial PECAM-1, as an indication of endothelial cell damage or rather pro-thrombotic state for clot development in the β2-AR mice. The immunohistochemical staining revealing the expression of PECAM-1 expectedly showed dramatic areas of PECAM-1 clusters (Fig. 4), indicating degenerative inner lining of the heart and emboli, as demonstrated in extravasated emboli in mice brain pathology using confocal imaging18. The clusters were overlaid by platelet (Fig. 4A) and fibrin (Fig. 4B). Importantly, the overlay was only present in the atria, but not in the ventricles (top, indicated by dotted lines (Fig. 4A, B).
Endothelial Cell Clustering and Embolic events in Tg. Overlay of PECAM-1 and platelet (A) and overlay of PECAM-1 and fibrin (B), associated with atrial emboli. Scale bars = 100 µm.
Aside from the theory of sub-endothelial collagen exposure to provoke thrombosis independent of atrial rate, inflammation is a known feature of cardiomyopathies. From the initial characterization of Elster, Braunwald and Wood in the 1950s, HF is inundated with pro-inflammatory systemic factors, such as C-reactive protein19. It is now appreciated that these factors also include angiotensin II, aldosterone, noradrenaline, adrenaline and atrial natriuretic peptide20, due to high neurohormonal stimulation. Therefore, in the cardiac-specific β2-AR mice, it is possible that systemic levels of these factors triggered a systemic response to the failing heart, including the atria, which can consequently trigger clot formation. Inflammation and left atrial endothelial dysfunction have been shown in a rat model after a distinct generation of early right and left insular stroke21. Furthermore, for better classification and in recognition of AF detection after stroke or transient ischemic attack, the diagnosis of atrial fibrillation post-stroke has been recently used22. This said, in the present study, it is important to note that this second theory was not investigated, suggesting the presence of an already existing ventricular myocardial neurohormonal response in the atrial pericardial pathology.
Ex vivo peptide enhancement and traditional Masson’s trichrome staining of atrial pericardial collagen
From the above epi-pericardial remodelling investigations, focal atrial pericardial fibrotic calcified lesion was gleaned (Fig. 2). Therefore, we next employed collagen I and MMP-cleaved collagen IV tracers (Cyclic peptide and T-peptide) that target these collagens with better avidity12,23,24. The aim of which is to detect the focal atrial pericardial fibrotic lesion, and quantify interstitial fibrosis, as arrhythmogenic substrate during cardiomyopathy by fluorescence-based approach of the tracers.
Fluorescence reflectance imaging as well as non-nuclear imaging proves to be important clinical approaches in cardiology arena to non-invasively detect atrial disease6,25,26. Ex vivo sulfo-Cy5.5)-peptide pericardial collagen staining was performed by incubating atrial tissue sections in the optical materials at room temperature. This was done to compare areas of collagen enhancement on the sections to that of the traditional Masson’s trichrome staining (Fig. 5A). In 4 h incubation of the sections in sulfo-Cy5.5 labelled T-peptide, enhanced pericardial collagen, blue arrows, dense red line (Fig. 3B), corresponded with the areas of the Masson’s trichrome staining (Fig. 5A), blue arrows. Similarly, incubation of Cy5.5 labelled cyclic peptide enhanced collagen areas of the pericardium, blue arrows, dense red line (Fig. 5C), and yielded a corresponding staining pattern to Fig. 5A, as annotated in blue arrows, but to a less degree, in terms of intensity, compared with that of T-peptide. Because T-peptide and cyclic peptide target type IV and I collagen, respectively, we did immunofluorescence staining of type IV and I collagen to confirm, their targeting by the peptide probes. Consistently, the stained areas on the immunofluorescence (blue arrows and dense purple lines; Fig. 5D, E) corresponded to the areas of probes staining (blue arrows and dense red lines; Fig. 5A, B). Notably, endomyocardial fibrosis by Masson’s trichrome staining better correlates with extracellular volume analysis by CMRI-LGE in cardiac amyloid patients27. Indeed, our ex vivo results also correlated with the traditional Masson’s trichrome staining. Together, our findings suggest that these peptide probes can target and identify pericardial collagen. Therefore, the development of in vivo peptide imaging tools is required, as shown by the summary illustration (Fig. 6) for roles in clinical application. These modalities can also include radioactive imaging by positron emission tomography (PET).
Traditional Mason’s trichrome staining (A) correlated qualitatively with T-peptide probe (B), cyclic peptide probe (C), type IV collagen immunofluorescence (D) and type collagen I immunofluorescence (E). T-peptide targets collagen IV while cyclic peptide probe targets collagen I. It is important to note that every normal atrium has intact pericardium tightly fastened to the wall of the working atrium. Tg; n = 2, Ntg; n = 2; Scale bar for (A) = 2000 µm; Scale bars for (B) and (C) = 1.5 mm. Blue arrow indicates disassociating pericardium.
Schematics illustration of near-infrared (NIR) fluorescence imaging of atrial fibrosis for in vivo and ex vivo clinical applications. (A) An in vivo method would require the injection of a targeted collagen peptide tracer into disease-affected and healthy individuals for NIR multiplex spectral imaging detection and analysis by a dedicated/customised analysis system. (B) An ex vivo approach would involve preparation of excised myocardial specimens from cardiac patients for investigation by high throughput screening, coupled with quantitative fluorescence microscopy.
High-throughput preclinical assessment of atrial collagen enhancement by the peptides
To assess the retention of the peptides by collagen I and IV in the pericardium, and atrial wall, and together quantify interstitial collagens, we conducted high-throughput screening and quantification of type I and IV collagen signals in mouse atrial lysate (MAL). 4 h after intravenous injection of the tracers, β2-AR mice upper chamber of the heart was obtained, homogenised and clarified. Up clarification by centrifugation, the MAL for T-peptide probe and S-peptide probe (mutant control version of T-peptide) were sampled on a near-infrared NIR 2-D scanner (Li-Cor) for odyssey imaging (Fig. 7A). Apparently, the peptides enhanced odyssey imaging demonstrated greater uptake of the peptides significantly in the Tg MAL compared with Ntg MAL (Fig. 7B).
Ex vivo Odyssey imaging of peptide enhancement of collagen in clarified MAL. Representative well-plates of the 2-D Li-Cor scan showing T-peptide enhancement in the MAL (A). Measurement of the T-peptide collagen enhancement in the MAL (B). Measurement of the T-peptide specificity in the enhancement at the disease level (C). Measurement of the T-peptide specificity in the enhancement at the basal level (D). Representative well-plates of the 2-D Li-Cor scan showing cyclic peptide enhancement in the MAL (E). Measurement of the cyclic peptide collagen enhancement in the MAL (F). Mean ± SEM, *p < 0.05 vs. Tg by unpaired t-test; n = 5.
The fluorescence reflectance in specificity of the uptake of T-peptide at disease (Fig. 7C) and basal (Fig. 7D) levels was also significant. For cyclic peptide, we adopted the T-peptide’s study design and procedure, and compared mean fluorescence uptake of the cyclic peptide, which was high in Tg MAL compared with Ntg MAL (Fig. 7E, F). Taken together, given the traditional collagen staining pattern, which showed that the atrial wall of transgenic animals had little or no collagen IV, unlike collagen I, this result shows that the tracer can be suitable for reporting pericardial collagen in the atrial wall. Our data exemplify the possibility of fluorescence imaging to visualize atrial fibrosis by means of molecular tracers. In particular, the findings hold promising implication for right ventricular imaging (as we have previously shown12), which is especially challenging given the thin nature of structures in the right heart. Notably, the right ventricle is larger in volume than its left counterpart, but has smaller mass and thinner wall, which impedes CMRI-LGE imaging.
