Ex Vivo and in Vivo Administration of Fluorescent CNA35 Specifically Marks Cardiac Fibrosis

Abstract

Cardiac fibrosis is a major hallmark of cardiac diseases. For evaluation of cardiac fibrosis, the development of highly specific and preferably noninvasive methods is desired. Our aim was to evaluate CNA35, a protein known to specifically bind to collagen, as a specific marker of cardiac fibrosis. Fluorescently labeled CNA35 was applied ex vivo on tissue sections of fibrotic rat, mouse, and canine myocardium. After quantification of CNA35, sections were examined with picrosirius red (PSR) and compared to CNA35. Furthermore, fluorescently labeled CNA35 was administered in vivo in mice. Hearts were isolated, and CNA35 labeling was examined in tissue sections. Serial sections were histologically examined with PSR. Ex vivo application of CNA35 showed specific binding to collagen and a high correlation with PSR (Pearson r = .86 for mice/rats and r = .98 for canine; both p < .001). After in vivo administration, CNA35 labeling was observed around individual cardiomyocytes, indicating its ability to penetrate cardiac endothelium. High correlation was observed between CNA35 and PSR (r = .91, p < .001). CNA35 specifically binds to cardiac collagen and can cross the endothelial barrier. Therefore, labeled CNA35 is useful to specifically detect collagen both ex vivo and in vivo and potentially can be converted to a noninvasive method to detect cardiac fibrosis.

CARDIOMYOCYTES are embedded in a network that is largely built from fibrillar collagens. Twenty-nine different types of collagen have been discovered to date, but the heart primarily expresses type I (85%) and type III (11%). ¹ The collagen network provides tensile strength, preventing excessive dilatation, contributes to diastolic ventricular suction, and plays a role in intercellular communication by the connection with the intracellular cytoskeleton. However, disproportionately increased collagen deposition (fibrosis) is a major hallmark of several cardiac diseases, from scar formation after myocardial infarction to interstitial and patchy fibrosis observed in patients with various cardiomyopathies. ² The initially compensatory fibrotic response will, when not stopped, eventually lead to increased wall stiffness and diminish diastolic function. Additionally, increased collagen deposition is a potential substrate for arrhythmias.^3–5

Currently, areas of fibrosis are frequently visualized noninvasively by magnetic resonance imaging (MRI). This technique provides high-resolution images in any desired plane without radiation ⁶ and offers a comprehensive overview of anatomy, function, and viability. The current gold standard, late gadolinium enhancement magnetic resonance imaging (LGE-MRI), detects larger patches of scarred tissue. Especially in patients with ischemic heart disease, scar formation is clearly visible with LGE-MRI. Also in other cardiomyopathies, such as hypertrophic or dilated cardiomyopathy, patchy fibrosis can be picked up by LGE-MRI. Given that detection of interstitial fibrosis by LGE-MRI is limited, new MRI techniques are currently being developed to detect individual collagen strands. Currently, T₁ mapping is the most extensively studied MRI technique to detect diffuse fibrosis in patients with cardiac diseases,^7–9 but this technique is not yet applicable in small-animal models. Moreover, spatial resolution with current contrast agents is too low to visualize specific interstitial collagen strands. Thus, the amount of cardiac fibrosis could be underdiagnosed with LGE-MRI as well as with T₁ mapping. ¹⁰ Furthermore, both techniques do not selectively and specifically image myocardial collagen and are merely a reflection of extracellular matrix expansion.

Cardiac tissue can be histologically examined ex vivo for fibrosis with dye stainings such as picrosirius red (PSR) or with specific fluorescently labeled antibodies. ¹¹ Due to the highly noncentrosymmetrical characteristics of fibrillar collagen, collagen can also be examined with second harmonic generation (SHG) microscopy. ¹² However, the acquirement of ventricular cardiac tissue in patients is not without risk; therefore, histology or SHG imaging of biopsies is not preferred in the clinical setting to examine fibrosis. Nonetheless, in animal studies, histology is the most commonly used technique for collagen detection in cardiac tissue.

Due to the high clinical relevance of fibrosis formation and the limitations of the current techniques to detect collagen deposition, the development of new methods to specifically and unequivocally detect cardiac fibrosis (preferably noninvasive) is desired. Recently, Boerboom and colleagues developed a collagen-binding molecule, named CNA35. ¹³ CNA35 is a 35 kDa bacterial binding protein domain of the Staphylococcus aureus bacterium that is easy to tag. CNA35 binds specifically to fibrillar collagen (K_d 10^-7 to 10^-6 M), with the strongest affinity for collagen type I and a relatively high affinity for collagen types II to IV. ¹⁴ Previous studies have shown that fluorescently labeled CNA35 binds specifically to newly synthesized collagen in cultured fibroblasts ¹³ but also in the murine arterial wall after both ex vivo and in vivo administration of CNA35.^13,15 Ex vivo and in vitro imaging of small arterioles has shown that CNA35 is more specific and has a higher spatial resolution than other collagen-visualizing techniques currently available. ¹³ After intravenous administration of CNA35, collagen labeling is observed in the kidneys and liver. ¹⁵

Based on these results, CNA35 is a promising candidate for highly specific collagen detection and is of particular interest for translational research because CNA35 is easy to tag. However, the capability of CNA35 staining in the heart has not been properly addressed. The aim of the current study was to quantitatively and qualitatively assess myocardial collagen content after ex vivo and in vivo administration of fluorescently labeled CNA35.

Methods

CNA35

CNA35 (Maastricht University, Maastricht, the Netherlands) was obtained, purified, and labeled with the fluorescent dye fluorescein isothiocyanate (FITC) (Sigma-Aldrich, St. Louis, MO; λ_excitation = 495 nm, λ_em = 515 nm) or Alexa Fluor 568 (Invitrogen, Eugene, OR; λ_excitation = 568 nm, λ_em = 580 nm), as described previously. ¹⁴ Both CNA35-FITC and CNA35–Alexa Fluor 568 were diluted in saline-based buffer solutions.

Animal Models

Animal handling was performed according to the Dutch Law on Animal Experimentation and the European Directive for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (86/609/EU). Mouse and rat experiments were approved by the Animal Experimental Committee of the University of Utrecht. Canine experiments were approved by the Animal Experimental Committee of Maastricht University.

Ex vivo experiments were performed on slices of heart tissue from mice, rats, and canines. For murine tissue sections, mice were subjected to transverse aortic constriction (TAC; hereafter referred to as TAC mice), a model that leads to cardiac failure and marked ventricular interstitial fibrosis. ¹⁶ Cardiac tissue sections were taken from rats with severe cardiorenal syndrome (CRS; hereafter referred to as CRS rats) because they are known to develop a combination of patchy and interstitial fibrosis. ¹⁷ Cardiac biopsies were derived from canines with a left bundle branch block (LBBB) and mitral regurgitation (MR). MR was induced with a customized ablation catheter with a hook at the distal tip to grasp and ablate the chordae tendinae suspending the mitral leaflet. After 4 weeks of recovery, LBBB was induced as has been previously described. ¹⁸ MR induces volume and pressure overload of the left atrium (LA), which induces severe fibrosis, initially of the LA, but when the disease progresses, also in the right atrium (RA). ¹⁹ The different animal models provide a variety of collagen deposition patterns and expression levels.

In vivo experiments were carried out using calcineurin transgenic (MHC-CnA) mice and their wild-type (WT) littermates. The MHC-CnA mouse model is a model in which a constitutively active form of calcineurin is overexpressed, specifically in the heart. This model has been characterized extensively.^20,21 In short, these mice develop cardiac hypertrophy already 18 days postnatally, which further progresses into heart failure and eventually can lead to sudden cardiac death. Moreover, this model shows high levels of interstitial myocardial fibrosis.

All the different animal models used for the ex vivo and in vivo experiments provide a wide range of fibrosis burden and distribution.

Ex Vivo Labeling, Imaging, and Quantification

Frozen sections (10 μm) of biopsies of both ventricles and atria of two canines (n = 12 and n = 8 sections, respectively) with MR and LBBB and both atria of one control canine (n = i6 sections) were used to cover a wide variety of collagen expression levels. Frozen tissue sections (10 μm; four-chamber view) from both CRS rats (n = 9 sections) and TAC mice (n = 4 sections) were used. Hoechst 33342 (16.2 mM; Invitrogen) and 2.5 mM CNA35-FITC were used as specific fluorescent markers for DNA/RNA and collagen, respectively. CNA35-FITC in Hank's Balanced Salt Solution (HBSS) buffer (Lonza, Verviers, Belgium) was combined in a ratio of 1:1 with a 250 times diluted stock concentration of Hoechst in HBSS buffer. Stained sections were imaged with an Olympus BX51WI spinning disk confocal fluorescence microscope coupled to a Hamamatsu EM-CCD C9100 digital camera. Stereo Investigator (MBF Bioscience, Williston, VT) was used to make high-resolution overview images of the tissue section. After imaging of CNA35-FITC, the same tissue sections were stained with PSR. PSR served as a reference for fibrosis staining and was used for a head-to-head comparison of specificity of CNA35. PSR staining was performed as previously described. ²² Briefly, slices were incubated in xylene for 30 minutes and dehydrated in an ethanol series. Subsequently, slices were stained with 0.1% sirius red (Polysciences Inc., Warrington, PA) in picric acid (Sigma-Aldrich Chemie GmbH, Steinheim, Germany) for 1 hour. PSR-stained tissue was imaged using the same microscope but using a QIcam color camera (QImaging, Surrey, BC).

ImageJ (Research Services Branch, National Institute of Mental Health, Bethesda, MD) was used to calculate the percentage of collagen by manually selecting the region of interest and adjusting the threshold. To compensate for bias in thresholding, the measurements were repeated three times per image and averaged to determine the collagen content per section.

In Vivo Injection, Ex Vivo Imaging, and Quantification

Male MHC-CnA mice (> 8 weeks old) ²⁰ and their male WT littermates (> 8 weeks old) were used to determine the perfusion capability of CNA35 after in vivo administration of CNA35–Alexa Fluor 568. They were injected with two boluses of 200 μL 38 μM CNA35–Alexa Fluor 568 (in phosphate-buffered saline). To prevent fluid overload, a 15-minute interval between the two boluses was preserved. Just prior to sacrifice, 15 minutes after the second dose, the mice were anesthetized with 4% isoflurane in 40% oxygen. Caprofen, 5 mg/kg (Pfizer Inc., Capelle a/d Ijssel, the Netherlands), was administered subcutaneously as an analgesic. Heparin (500 IU) was administered intraperitoneally as an anticoagulant to prevent clotting and accumulation of blood in the myocardium after sacrifice. Finally, the heart was excised, rinsed immediately, and directly preserved in ice-cold saline.

Qualitative examination of CNA35–Alexa Fluor 568 labeling in MHC-CnA (n = 5) and WT hearts (n = 5) was performed within 5 hours using two-photon laser scanning microscopy (TPLSM). A Nikon E600FN microscope (Nikon, Tokyo, Japan), coupled to a standard Bio-Rad 2100 multiphoton system (Bio-Rad, Hemel Hempstead, UK) was applied. A 150 fs-pulsed Ti:Sapphire laser (Spectra Physics Tsunami, Santa Clara, CA), tuned and mode-locked at 800 nm, was used as an excitation source. A 60 × magnification waterdipping lens was used to visualize individual myocytes embedded in the extracellular matrix. An HQ560LP filter was used to detect Alexa Fluor 568.

Wide-field fluorescence and transillumination microscopy on snap-frozen tissue material was performed for quantitative analysis of the CNA35–Alexa Fluor 568 labeling. MHC-CnA (n = 4) and WT hearts (n = 4) were snap frozen and cut in four-chamber view sections (10 μm slice thickness). CNA35 labeling was examined within 24 hours. Serial sections were used for PSR staining. CNA35 labeling and PSR staining were examined and quantified as described in the ex vivo section.

Statistical Analysis

Data are presented as mean (± standard deviation). Statistical analysis was performed using SPSS for Windows version 20.0 (IBM Corp., Armonk, NY). Pearson correlation linear regression analysis between CNA35 and PSR collagen percentages was performed. Bland-Altman analysis was performed to evaluate the accuracy of CNA35. A Student t-test was performed to determine differences in fibrosis levels between the two groups. A two-sided probability value < .05 was considered a statistically significant difference.

Results

Ex Vivo Labeling, Imaging, and Quantification

Ex vivo analysis of collagen labeling by CNA35-FITC showed that whole heart sections (mouse and rat) labeled with CNA35 corresponded to the PSR staining of serial sections (Figure 1, A–D). The average collagen concentration in mouse and rat heart sections was 9.6 ± 1.8% and 20.2 ± 5.3%, respectively, when determined with CNA35. When collagen concentration was determined with PSR, comparable results were observed: 9.8 ± 1.6% and 21.5 ± 4.2% in mouse and rat sections, respectively. Paired analysis of CNA35-FITC and PSR in the mouse and rat heart sections resulted in a good correlation (r = .86, p < .001).

Figure 1.

Tissue sections stained with CNA35-FITC and picrosirius red (PSR). Left panels: CNA35-FITC staining (green) and cell nuclei (blue) showing a clear demarcation of fibrous strands. Right panels: Serial sections stained with PSR (B and D) or the same section stained with PSR (F). A and B, Transverse aortic constriction mouse myocardium. C and D, Cardiorenal syndrome rat myocardium. E and F, Atrial tissue from a canine with mitral regurgitation and left bundle branch block. The dashed lines encompass the area of magnification, showing that even the smallest collagen fibers are highlighted.

Ex vivo staining of canine biopsies collected from the atria and ventricles provided a wide range of fibrosis burden (2.8–24.0%). Therefore, CNA35 staining could be properly evaluated at low and high concentrations of collagen. Overall, there was a strong agreement between CNA35 and PSR labeling of collagen, indicating the selective capacity of CNA35 to bind to myocardial collagen in canines. In the ventricular tissue, a collagen volume fraction of 4.7% ± 1.7% and 3.8 ± 1.1% was observed when determined with CNA35 and with PSR, respectively. Atrial tissue sections showed a collagen volume fraction of 7.9 ± 4.5% when determined with CNA35, which is comparable to the collagen volume fraction determined with PSR, 7.4 ± 4.5%. The LA stained with CNA35 of canines with MR and LBBB had a significant higher collagen concentration than the RA (13.1 ± 5.4% vs 8.2 ± 1.2%, respectively; p < .05). A representative picture of both CNA35 labeling and PSR staining of the canine atrium is shown in Figure 1, E and F. Quantification of paired CNA35 and PSR images showed a high correlation (r = .98, p < .001; Figure 2). Bland-Altman analysis showed a minor but significant variability in collagen percentages measured by CNA35-FITC and PSR. CNA35-FITC estimated collagen concentration 0.67% higher than PSR (upper boundary 2.5% to lower boundary –1.2%; p < .05).

Figure 2.

Correlation between collagen examined with picrosirius red (PSR) and with CNA35 after ex vivo application in canine tissue. Left panel: Correlation between the CNA35 and by PSR in canine atrial (n = 28) and ventricular tissue sections (n = 8). The amount of collagen detected by CNA35-FITC is highly correlated with PSR (r = .98, p < .001). Right panel: Bland-Altman plot of collagen detection by CNA35 with respect to PSR. CNA35 detected a significant higher percentage of collagen than PSR staining (+0.67%; p < .001).

In Vivo Labeling, Ex Vivo Imaging, and Quantification

Two-Photon Laser Scanning Microscopy

After in vivo administration of CNA35–Alexa Fluor 568, hearts from WT (n = 5) and MHC-CnA (n = 5) mice were isolated and visualized with TPLSM to observe CNA35-Alexa Fluor 568 labeling in the viable tissue. The high tissue penetration depth and spatial resolution achieved using TPLSM makes it possible to visualize individual myocytes embedded in the extracellular matrix. Figure 3 clearly shows that CNA35–Alexa Fluor 568 is able to pass the endothelial layer, penetrate the myocardial extracellular space, and bind to individual collagen strands. Both WT and MHC-CnA hearts showed CNA35 labeling surrounding the individual myocytes. Although quantification of the TPLSM images is not possible, the CNA35 labeling in the MHC-CnA hearts seems more intense than in the WT hearts.

Figure 3.

Two-photon laser scanning microscopy (TPLSM) images of CNA35-Alexa Fluor 568 after in vivo administration. A, Representative TPLSM image of CNA35–Alexa Fluor 568 labeling (red) surrounding cardiomyocytes in a wild-type (WT) heart. B, Representative TPLSM image of a calcineurin transgenic (MHC-CnA) heart with a similar view of cardiomyocytes surrounded with CNA35–Alexa Fluor 568 (red). Although the CNA35 labeling has not been quantified, the MHC-CnA heart shows a more intense CNA35–Alexa Fluor 568 signal compared to the WT heart. Both hearts are additionally stained with Syto4l (blue) to stain cell nuclei.

Quantitative Analysis of CNA35–Alexa Fluor 568 Using Histologic Slices

After in vivo administration of CNA35–Alexa Fluor 568, hearts from WT (n = 4) and MHC-CnA (n = 4) mice were isolated and sectioned. Collagen was determined with CNA35–Alexa Fluor 568 labeling and compared to PSR. A representative picture from CNA35–Alexa Fluor 568 and PSR is shown in Figure 4. Cardiac collagen concentration determined with CNA35 in WT and MHC-CnA mice was 3.8 ± 2.1% and 9.3 ± 3.6%, respectively (p < .001). A strong correlation was observed between CNA35 and PSR (r = .91, p < .001; Figure 5). This result confirms the capability of CNA35–Alexa Fluor 568 to pass the endothelial layer of the coronary arteries in both healthy and diseased mice and that CNA35 can be used to specifically visualize collagen in the myocardium. Bland-Altman analysis showed a minor but significant variability in collagen percentages measured after in vivo administration of CNA35–Alexa Fluor 568 and PSR. CNA35–Alexa Fluor 568 significantly estimated the collagen concentration 0.15% higher than PSR (upper boundary 3.5% to lower boundary –3.2%; p < .001) (see Figure 5).

Figure 4.

Calcineurin transgenic mice (MHC-CnA) and wild-type (WT) heart stained with CNA35–Alexa Fluor 568 after in vivo administration and with picrosirius red (PSR). Top panels: Whole tissue sections of a MHC-CnA heart stained with CNA35–Alexa Fluor 568 (A; red) and the serial section stained with picrosirius red (PSR) (B). Lower panels: Whole tissue sections of a WT heart stained with CNA35–Alexa Fluor 568 (C) and a serial section stained with PSR (D). Note the darker and less stained areas in the WT heart compared to the MHC-CnA heart. Magnification shows that even the smallest collagen fibers are highlighted, even in the WT mouse. The gray dashed lines show the areas of magnification.

Figure 5.

Correlation between collagen examined with picrosirius red (PSR) and with CNA35 after in vivo administration of CNA35. Left panel: Correlation plot of CNA35 against PSR in wild-type and calcineurin transgenic (MHC-CnA) cardiac tissue sections (n = 36). In particular, hearts from MHC-CnA mice show high levels of fibrosis. The amount of collagen detected by CNA35 is highly correlated with the amount of collagen stained with PSR (r = .91, p < .001). Right panel: Bland-Altman plot of collagen detection by CNA35 with respect to PSR. CNA35 detected a significantly higher percentage of collagen than PSR staining (0.15%; p < .001).

Altogether, these data show that there were no interspecies differences. Furthermore, the overall correlation between CNA35 and PSR was strong even though there were large differences in the degree and distribution of fibrosis between the animal models.

Discussion

This study shows that (1) CNA35 specifically marks cardiac collagen in different species; (2) CNA35 is able to enter the myocardium after intravenous administration; and (3) CNA35 is able to specifically mark cardiac collagen after in vivo administration. The amount of collagen detected by the fluorescently labeled CNA35 is highly comparable with the amount of collagen determined by PSR after both ex vivo and in vivo application.

It has already been shown that CNA35 specifically binds to mouse, rat, bovine, and human fibrillar collagen types I to IV.^7–9 The present study confirms the CNA35 binding to mouse and rat cardiac collagen, and in addition, we show specific collagen binding in canine myocardium. This supports the hypothesis that the detection of collagen by CNA35 is species independent. Furthermore, given that the heart mainly expresses collagen types I and III, CNA35 is a suitable candidate to evaluate cardiac fibrosis but may be less efficient for other pathologies throughout the body that are characterized by upregulation of other types of collagen.

To confirm that CNA35 specifically labeled the cardiac collagen, PSR, a dye that consists of sirius red dissolved in picric acid that strongly binds to collagen, was used as a reference. ²² In our ex vivo experiments, we compared CNA35 labeling to PSR staining and observed a slightly higher CNA35 labeling compared to the collagen stained by PSR. Although the difference in the quantitative collagen detection was relatively low, it is statistically significant. The slightly but significantly higher collagen detection by CNA35 might be caused by a difference in signal strength between fluorescent CNA35 confocal and bright-field PSR imaging. Indeed, small collagen fibers of PSR images generate a weaker signal that might be more difficult to quantify compared to fluorescent CNA35. Moreover, the different animal models used indicate that CNA35 is capable of assessing collagen content in a wide variety of fibrosis burden and distribution.

The amount of collagen marked with CNA35 in the different models in this study is comparable to the range of fibrosis that is measured in other studies. In LA tissue derived from MR dogs, we observed a 1.8-fold increase in CNA35 labeling compared to the labeling in RA tissue. This result is in line with the 1.8-fold increase in LA fibrosis observed in patients with mitral valve disease compared to patients without mitral valve disease. ¹⁹ Regarding the results after in vivo administration of CNA35–Alexa Fluor 568 in mice, we observed a 2.4-fold increase in hearts from MHC-CnA mice compared to their WT littermates, which coincides with the study of Fontes and colleagues, who observed a 2.8-fold increase in fibrosis in 4-week-old MHC-CnA mice compared to WT littermates. ²³

To our knowledge, this is the first study that shows CNA35 labeling of individual cardiac collagen strands. Megens and colleagues showed the binding of CNA35 labeled with Oregon Green 488 (OG488) in the collagenous part of the atherosclerotic plaques in mouse arteries but no labeling in the healthy arteries after in vivo administration. ¹⁵ In addition, they observed CNA35 labeling in the kidney, liver, and spleen after administration but no CNA35 labeling in the heart. Based on these results, the authors suggested that CNA35-OG488 enters tissue with a highly permeable endothelial phenotype. However, in our study, we showed that CNA35–Alexa Fluor 568 can enter the myocardium in both MHC-CnA and WT mice, despite the tight endothelial barrier. This discrepancy might be caused by the different fluorophore that has been used. Although the molecular weight of OG488 is lower than that of Alexa Fluor 568 (509 Da and 792 Da, respectively), other fluorophore characteristics, such as its chemical structure or charge, may influence the penetration of the endothelial barrier. Another explanation might be that cardiac collagen labeling by CNA35 was below their detection limit because Megens and colleagues studied mice with no known cardiac diseases and the collagen strands in the healthy myocardium are small. Furthermore, the concentration of CNA35 used by Megens and colleagues was half the concentration we used. Mees and colleagues labeled CNA35 with the radioactive agent ^99mTc tricarbonyl and investigated the labeling of CNA35 to subendothelial collagen IV in tumor neovasculature. ²⁴ Also in this study, no CNA35 labeling in the healthy murine heart was observed. However, the mice were imaged with a gamma camera to detect tumors > i0 mm in diameter. Therefore, the spatial resolution of the camera used in this study was too low to visualize collagen strands labeled with CNA35 in the healthy mouse heart. In a recent study by Danila and colleagues, CNA35 labeled with gold nanoparticles was studied as a new computed tomography (CT) contrast agent to detect myocardial scarring in mice ²⁵ . They showed that labeled CNA35 in CT imaging detected large myocardial infarctions in the nonbeating mouse heart. Although small patches of fibrosis were undetectable, these results, together with those of the current study, are promising to use labeled CNA35 for in vivo imaging of cardiac fibrosis.

Altogether, the results support the hypothesis that fluorescently labeled CNA35 specifically marks cardiac fibrosis after both ex vivo and in vivo administration. Although several techniques are already available to detect cardiac fibrosis in the experimental setting, the detection of cardiac collagen with CNA35 might have advantages compared to the current techniques. LGE-MRI is noninvasive but only detects larger patches of fibrosis and is not capable of identifying diffuse fibrosis. T₁ mapping, using gadolinium as a contrast agent, can detect diffuse expansion of the extracellular matrix in the experimental setting. Still, gadolinium is, as an extracellular contrast agent, not a specific marker. Not only fibrosis but also inflammation and edema can be picked up by T₁-weighted mapping, and an increase in signal is thus related to an increase in the extracellular volume and not an increase in fibrosis per se. ²⁶

In contrast to dye stainings, CNA35 specifically binds to fibrillar collagen, whereas stainings such as PSR and Masson's trichrome are based on base-acid interactions and therefore may not be entirely collagen specific. In addition, in this study, it was shown that CNA35 can detect cardiac collagen after in vivo administration, which to our knowledge is not possible with other dyes. Regarding the collagen antibodies, CNA35 has the advantage of being able to detect the various fibrillar types of collagen, whereas antibodies are only specific for one collagen type. In addition, antibodies are relatively large compared to CNA35, which makes it more difficult for them to penetrate the tissue. ¹⁴ Examining collagen in viable tissue with CNA35 is therefore preferred over using antibodies. An additional advantage of CNA35 is that the binding to collagen is reversible. Unfortunately, details about the unbinding of CNA35, such as the half-life, the elimination pathway, and the toxicology, are not yet known. More information about this unbinding process is a prerequisite to use CNA35 as an in vivo contrast agent in humans to evaluate the process of fibrosis with either high-spatial resolution CT scanning or MRI. Given that CNA35 binding to collagen is not limited to the heart, the determination of collagen by CNA35 might also be of interest to detect the process of collagen formation in (viable) tissue in other diseases or in regenerative medical research.

Study Limitations

Although this study shows the possibility of CNA35 binding to cardiac collagen after in vivo administration, the technique used hampers direct translation of the current study to noninvasive imaging in the clinical setting, such as MRI or single-photon emission computed tomography (SPECT). Therefore, future experiments should further focus on CNA35 labeled with agents that can be used in these noninvasive imaging techniques that are suitable for the clinical setting. However, labeling of CNA35 with different agents, both fluorescent and other, might influence the permeability and binding affinity of CNA35. When experiments are conducted with a new agent, the binding capacity and permeability need to be reassessed. This phenomenon should also be kept in mind when studies with different labeling agents are compared (i.e., OG488 versus Alexa Fluor 568 labeling, as discussed in the section above).

Given that the mice were sacrificed shortly after in vivo administration, information about the pharmacokinetic profile or any possible side effects of CNA35 is missing. At a first glance, the mice did not show any discomfort after CNA35 administration, but further pharmacologic and pharmacokinetic investigation is necessary before it is safe to use it in the clinical setting.

Unfortunately, images from different hearts cannot be quantitatively compared to TPLSM. Indeed, imaging individual myocytes using TPLSM at high magnification forbids exact localization of the obtained images. Furthermore, laser light intensity differs depending on tissue depth and fluorescent binding. Nevertheless, the MHC-CnA hearts had a higher signal of CNA35 labeling than in the WT hearts. For exact quantification, we therefore used histologic slices.

Conclusion

This study shows for the first time that CNA35 specifically binds to cardiac collagen strands and that it crosses the cardiac endothelial barrier in both healthy and diseased myocardium. Therefore, labeled CNA35 can be used to specifically detect collagen in an ex vivo setup and after in vivo administration.