Open Access
Add to BibSonomy
Abstract
The zebrafish has emerged as a useful model for human hematological disorders. Transgenic zebrafish that express green fluorescence protein (GFP) in red blood cells (RBCs) visualized by fluorescence microscopy (FLM) is a fundamental approach in such studies to understand the cellular processes and biological functions. However, additional and cumbersome efforts are required to breed a transgenic zebrafish line with reliable GFP expression. Further, the yolk autofluorescence and finite GFP fluorescence lifetimes also have an adverse impact on the observation of target signals. Here, we investigate the identification of intracerebral hemorrhage (ICH) and hemolytic anemia (HA) in zebrafish embryos using label-free photoacoustic microscopy (PAM) for imaging. First, ICH and HA in transgenic LCR-EGFP zebrafish are mainly studied by PAM and FLM. The results show that PAM is comparable to FLM in good identification of ICH and HA. Besides, PAM is more advantageous in circumventing the issue of autofluorescence. Secondly, ICH and HA in the transparent casper zebrafish without fluorescent labeling are imaged by PAM and bright-field microscopy (BFM). Because of the high contrast to reveal RBCs, PAM obviously outperforms BFM in the identification of both ICH and HA. Note that FLM cannot observe casper zebrafish due to its lack of fluorescent labeling. Our work proves that PAM can be a useful tool to study blood disorders in zebrafish, which has advantages: (i) Reliable results enabled by intrinsic absorption of RBCs; (ii) wide applicability to zebrafish strains (no requirement of a transgene); (iii) high sensitivity in identification of ICH and HA compared with BFM.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Zebrafish, as a vertebrate model, has advantages including fecundity, small body size, ex utero development, and relatively good optical transparency in the embryonic stage. The development and function of zebrafish organs are strikingly similar to those of humans. Zebrafish has been widely used to study developmental processes and the underlying cellular and molecular mechanisms [1–3]. In recent years, it has also been utilized to investigate biology of human diseases and explore new therapies. Zebrafish models of a variety of human diseases have been established [4–8]. For example, zebrafish is very powerful to recapitulate human hematological pathologies and has been used to unravel many molecular mechanisms involved in these diseases [8].
Optical microscopy is one of the major techniques used in zebrafish research. The optical accessibility of zebrafish embryos allows live imaging of the dynamic structure and behavior of the cells or organelles under normal and pathological conditions. To reveal the detailed processes at a cellular or subcellular level in vivo, fluorescence microscopy (FLM) imaging relying on fluorescent protein is a widely used approach [9,10]. To do that, it often requires the generation of transgenic zebrafish in which the fluorescent protein, e.g., green fluorescence protein (GFP), is used under the control of a tissue-specific promoter to label the cells of interest [10,11]. However, several issues and challenges remain associated with this approach. First, it might be difficult to find a promoter suitable for a given cell or molecule [12]. For example, the hemoglobin molecules have different hemoglobin species during the early and later stages of development. This change (known as hemoglobin switching) is regulated by a complex molecular process and mechanism [13,14]. It is thus difficult to create a transgene that reliably labels all the hemoglobin molecules and correspondent red blood cells (RBCs) at different stages. Secondly, after a transgene is successfully made, additional and often cumbersome efforts are further required to create the transgene in different genetic backgrounds for studying the labeled cells under different conditions. Thirdly, the long half-life of fluorescent protein (e.g., 26 hours for GFP) may prevent monitoring fast dynamic processes such as gene expression [15,16], although the long half-life time would be good to provide stable labels for fluorescence imaging. Specifically, for example, when GFP-labeled RBCs become disordered, GFP fluorescence signals will not disappear immediately until after the half-life time, and thus, the RBCs may be mistakenly considered to be normal within the half-life time, which hinders the longitudinal observation of fast changes. Fourthly, the background autofluorescence from inherent components or nearby tissues may directly interfere with the real fluorescence signal, which inevitably deteriorates the image quality and leads to less trustworthy results. Therefore, developing a label-free and noninvasive alternative imaging method is indeed necessary.
Recently, label-free photoacoustic (PA) microscopy (PAM), as a hybrid imaging modality, has been extensively used to acquire optical absorption contrast in tissue with high resolution and moderate penetration [17]. PAM can be used to image intrinsic absorption contrast such as pigment and RBCs, also called erythrocytes, which are from endogenous chromophores of melanin and hemoglobin, respectively. With increasing attention on the use of zebrafish models for biomedical research, PAM has also been widely explored to study zebrafish models in the past decade [18–26]. For example, great performance of improved PAM systems has been verified by in vivo imaging of zebrafish samples [19,22,23]. In addition, high-resolution PAM systems were employed to study the early cardio-cerebrovascular development in 1-phenyl-2-thiourea (PTU)-treated zebrafish embryos and larvae [18,20]. More recently, multi-modality imaging systems were used to extract both morphological and functional information of zebrafish larvae [21,24–26]. Although RBC-related disease models in zebrafish, such as intracerebral hemorrhage (ICH) [27–30] and hemolytic anemia (HA) [6,31–33], have been available and are valuable models to investigate disease mechanisms and identify novel therapies, the conventional approach using the transgenic zebrafish line expressing GFP in RBCs visualized by FLM still encounters several challenges, as detailed above. As a matter of fact, relying on intrinsic and high absorption contrast of RBCs, PAM could be a great tool to study RBC-related disease models in zebrafish compared with conventional FLM.
Hence, in this work, for the first time, we demonstrate label-free PAM as a useful tool to study the ICH and HA disease models in zebrafish embryos. First, PAM and FLM are mainly used to image the transgenic zebrafish strain with ICH and HA. PAM achieves comparable imaging results to FLM in high-resolution visualization of RBCs and good identification of ICH and HA. Besides, PAM is more advantageous in circumventing the issue of background autofluorescence. More importantly, relying on intrinsic absorption contrast from RBCs themselves, PAM is expected to offer more reliable images. Secondly, PAM and bright-field microscopy (BFM) are used for live imaging analysis of ICH and HA in the optically clear casper zebrafish. Because of the high contrast to reveal RBCs, PAM obviously outperforms BFM in identification of both ICH and HA, which are conducted quantitatively (i.e., a statistical study) and qualitatively, respectively. Note that FLM cannot observe casper zebrafish due to its lack of fluorescent labeling [34]. The results show that PAM can be a great tool to study zebrafish disease models with the following advantages: (i) Reliable results enabled by intrinsic absorption directly from RBCs with high contrast; (ii) wide applicability to zebrafish strains; (iii) high sensitivity in identification of ICH and HA compared with BFM. PAM may also be useful to investigate other zebrafish blood disorders besides ICH and HA, and it has potential for studying the corresponding disease mechanisms.
2. Methods
2.1 PAM system
The schematic of the PAM imaging setup is shown in Fig. 1(a). A 532 nm pulsed laser (AO-S-532, cnilaser) with a pulse duration of <10 ns and a pulse repetition rate of 2 kHz was used as the PAM excitation source. The laser beam emitted from the laser head was first split into two beams by a 1:9 beamsplitter (BS025, Thorlabs). One beam (10% power) was detected by a photodiode (DET10A2, Thorlabs) to provide triggers. The main beam (90% power) was first attenuated by neutral density filters (GCC-3011A, Daheng Optics) and then passed through a beam shaping set, consisting of two plano-convex lenses (GCL-0108, Daheng Optics) and a pinhole (ID8, Thorlabs), to produce a collimated beam with enlarged beam size. The beam was then coupled into a single-mode fiber (460-HP, Nufern) by a doublet lens (AC127-030-A-ML, Thorlabs), and the exited beam was focused by an objective (43397, Edmund) for PA excitation. A home-made probe, mainly containing the single-mode fiber and the objective, was used to facilitate laser scanning during PA image acquisition. The excited PA signals were detected by an obliquely placed custom-made needle hydrophone with center frequency of 35 MHz and bandwidth of ∼40 MHz (bandwidth determined from the frequency spectrum of the PA A-line signal obtained by imaging a 6-µm carbon fiber). The sensing element size of the hydrophone was ∼1.3 mm in diameter. The measured receiving angle of the hydrophone was ∼10° (full angle) at −6 dB. During image acquisition, the home-made probe and the hydrophone were mounted on a two-dimensional (2D) linear motorized stage (M-404, Physik Instrumente) to perform 2D scanning while the zebrafish sample was fixed and kept stationary. PA signals were amplified by a preamplifier (ZFL-500LN-BNC+, Mini Circuits) and a pulse receiver (5073PR, Olympus), recorded by a digitizer (CSE1422, GaGe), and finally stored and processed using a computer. Spatial resolution of the PAM system was experimentally verified. Lateral resolution was measured by imaging the sharp edge of a razor blade without signal averaging. Figure 1(b) shows the imaged one-dimensional profile (i.e., “Raw data” in Fig. 1(b)), which is fitted by an edge spread function (ESF). A line spread function (LSF) can be obtained by taking the spatial derivative of the ESF. Then, the lateral resolution was determined to be 3.2 µm. Axial resolution was measured by imaging a 6-µm carbon fiber and was estimated to be ∼28.5 µm (Fig. 1(c)), which is mainly limited by the bandwidth of the hydrophone. Note that during image acquisition, scanning step size was set to 4 µm (larger than the lateral resolution of 3.2 µm), which is expected to be sufficient to reveal RBCs. Considering the same region of interest to scan, not too small scanning step size also helps to save image acquisition time. For better visualization of vasculature in PAM images, a matched filter and a median filter in the post image processing were applied.
Fig. 1. (a) The schematic of PAM imaging setup. BS, beamsplitter; PD, photodiode; NDF, neutral density filter; PCL, plano-convex lens; PH, pinhole; M, mirror; DL, doublet lens; SMF, single-mode fiber; NH, needle hydrophone; WT, water tank; DAC, data acquisition card. (b) Measured lateral resolution of 3.2 µm. (c) Measured axial resolution of 28.5 µm.
For comparison, FLM images were acquired by using a fluorescence microscope (BX53, Olympus). For BFM images from ICH zebrafish disease models, an optical microscope (SZX2-ILLB, Olympus) working in reflection mode with a white reflective substrate mounted, on which samples were placed, was used for clearly identifying ICH. For other BFM images (except ICH), the same microscope (BX53) used for FLM was used. To facilitate FLM image acquisition, the abovementioned fluorescence microscope (BX53) was equipped with the extended depth-of-focus mode utilizing dynamic imaging depth. Thus, certain image acquisition time is required to achieve high imaging performance. Key parameters of the FLM system and the PAM system are provided in Table 1 for comparison.
Table 1. Comparison of FLM and PAM systems used in this work
2.2 Zebrafish sample preparation
Adult zebrafish were maintained at 28.5°C on a 14-hours light/10-hours dark cycle in a recirculating water system. The transgenic Tg (LCR-EGFP) line (with enhanced GFP-labeled RBCs [13]) and the mutant transparent casper line (devoid of melanophores and iridophores [35]) were used. Note that for the casper line, Tg (LCR-EGFP) is not introduced, and thus, RBCs are not labeled. Adult zebrafish were allowed to spawn in groups naturally. Fertilized embryos were collected, rinsed, and placed into Petri dishes in E3 medium, and then stored in an incubator at 28.5°C until treatment. Embryos were staged by hours post-fertilization (hpf). Note that the transgenic zebrafish embryos were treated with 30.4 µg/ml PTU (CAS: 103-85-5, Sigma-Aldrich) after 10 hpf to prevent excess pigment formation. In this way, PA excitation light will not be blocked by pigment of zebrafish, and thus, intrinsic contrast from RBCs can be acquired.
Zebrafish blood disorders were induced as follows. Atorvastatin (ATV, CAS: 134523-00-5, Aladdin) and Phenylhydrazine hydrochloride (PHZ, CAS: 59-88-1, Sigma-Aldrich) were prepared in dimethyl sulfoxide (DMSO) and diluted to 0.84 µg/ml and 2 µg/ml, respectively, in E3 medium. DMSO was used as a vehicle control (0.1%). For the ICH disease model, zebrafish embryos were incubated with ATV at 24 hpf and imaged at 34 hpf and 48 hpf. Further, for the statistical study of ICH identification by BFM and PAM (Section 3.3), 60 casper zebrafish embryos in total were prepared. Note that the 60 zebrafish embryos were prepared in three batches for better arrangement of imaging experiments. As for the HA disease model, zebrafish embryos were incubated with PHZ at 48 hpf and investigated at 72 hpf by BFM, FLM, and PAM.
Before imaging experiments, zebrafish embryos were anesthetized with 0.64 µg/mL Tricaine (CAS: 886-86-2, Sigma-Aldrich). Then, they were transported to a glass slide and mounted with 1% low melting point agarose (CAS: 9012-36-6, Sigma-Aldrich). All experiments using zebrafish were conducted under the auspices of animal ethics. The general breeding and handling of zebrafish were conducted following standard protocols from the Institutional Animal Care Committee of Shanghai Jiao Tong University.
3. Results
3.1 In vivo imaging of zebrafish embryos
To verify the performance of our PAM imaging system for studying zebrafish models, we first imaged a 72 hpf transgenic Tg (LCR-EGFP) zebrafish embryo using PAM (displayed in maximum amplitude projection) as well as BFM and FLM for side-by-side comparison. To improve signal detection, PTU was used to inhibit melanization in embryos, as used in other studies [36]. Lateral view of the same zebrafish embryo imaged by BFM, FLM, and PAM is shown in Figs. 2(a) − 2(c), respectively. The high-performance fluorescence microscope used to acquire the BFM image enables detailed cardiovascular structure (Fig. 2(a)). In Fig. 2(b), the main vessels, the dorsal aorta (DA) and the cardinal vein (CV) running horizontally in the trunk, are clearly seen. However, the fluorescence signals of the vertical intersegmental vessels (ISVs) and the dorsal longitudinal anastomotic vessel (DLAV) are rather weak. What is worse, the autofluorescence and light scattering near the brain and yolk parts smeared partial fluorescence signals from GFP-labeled RBCs, which deteriorates the image quality in those parts.
Fig. 2. (a)−(c) Imaging of the same 72 hpf transgenic Tg (LCR-EGFP) zebrafish embryo (with PTU treatment): BFM image (a), FLM image (b), and PAM image (c). (d)−(f) Imaging of the same 72 hpf casper zebrafish embryo: BFM image (d), FLM image (e), and PAM image (f). Scale bar: 100 µm. All images share the same scale bar.
The PAM image acquisition was detailed in Section 2.1. At a glance, good agreement in the vessel structure between FLM and PAM images is observed. For example, the PAM image (Fig. 2(c)) is also able to clearly reveal major blood vessels like the DA and CV along the zebrafish body. On the other hand, at a close look, PAM outperforms FLM in two aspects: (i) PAM image contrast from the ISVs and DLAV is higher than the FLM counterpart. (ii) No obvious light scattering is observed in PAM images. Particularly, the blood vessels in the brain and heart can be better visualized. Both (i) and (ii) benefit by the intrinsic and high absorption contrast of RBCs in PAM imaging. The great PAM image quality would facilitate the study of the cardiovascular system of zebrafish. We also noticed that the vessels in the PAM image present a granular texture, particularly in the ISVs, which is a result of the motion of the circulating RBCs.
Due to its optical transparency, the mutant casper zebrafish that lacks skin pigmentation has been widely used for in vivo imaging in different disease models for better detection. We also imaged the 72 hpf casper zebrafish embryos using BFM, FLM, and PAM. The images are shown in Figs. 2(d)–2(f). Overall, the PAM images (Fig. 2(c) and 2(f)) from different types of zebrafish embryos are comparable, showing that PAM is applicable to the two zebrafish strains. By contrast, in Fig. 2(e), due to lacking fluorescent contrast in the casper zebrafish embryo, only a dark image was obtained by FLM without further information, not to mention the details of blood vessels. Therefore, in Section 3.3, FLM images of the casper zebrafish strain were not conducted.
3.2 In vivo imaging of disease models in transgenic zebrafish
In vivo imaging of ICH and HA disease models in transgenic Tg (LCR-EGFP) zebrafish embryos was then conducted. As mentioned in Section 2.2 and Section 3.1, PTU treatment was applied to the transgenic zebrafish embryos. The zebrafish embryos with ATV treatment to induce ICH were used as the experimental group (n = 10), while those without ATV treatment were used as the control group. For the experimental group, ICH was identified (i.e., confirmed and counted) by observing evident bleeding in the fore-, mid-, and hind-brain of the ATV-treated zebrafish embryos. On the other hand, the case of no evident bleeding in the brain region was considered as no ICH identified. The same samples were imaged by BFM, FLM, and PAM, respectively, for side-by-side comparison. To optimize the BFM imaging results, the samples were placed on a white reflective substrate and imaged in reflection mode, as mentioned in Section 2.2. First, for the experimental group of the 10 zebrafish embryos, “identified ICH” and “no ICH identified” are highly consistent between PAM and FLM images. Therefore, no further quantitative analysis was conducted to compare FLM and PAM images for ICH identification. Secondly, for the same experimental group, not all “identified ICH” in PAM (and FLM) images can be successfully recognized in the corresponding BFM images. In other words, PAM can better identify ICH than BFM. A quantitative study was more comprehensively conducted to show how sensitive PAM is in identifying ICH compared with BFM in Section 3.3 (described later). Figure 3 shows the imaging results of the control and experimental (representative images) groups. Note that the images in each column are from the same zebrafish embryo. The confirmed bleeding in the brain region is marked with the triangles in Fig. 3. For the control group, as expected, no unusual patterns are identified in the brain region in the BFM, FLM, and PAM images (Figs. 3(a), 3(d), and 3(g), respectively). For the experimental group, as mentioned above, ICH is consistently found in both FLM images and PAM images (representative images: Figs. 3(e) and 3(h); Figs. 3(f) and 3(i)), while some identified ICH in FLM and PAM images cannot be recognized in the corresponding BFM image (e.g., Figs. 3(f) and 3(i) vs. Figure 3(c)). Overall, the results suggest that: (i) PAM is comparable to FLM, both performing well and consistently in ICH identification in the transgenic zebrafish strain; (ii) PAM is more sensitive in ICH identification than BFM. It is worth noting that although both PAM and FLM can identify ICH, PAM images are based on the intrinsic absorption of RBCs, which could be more stable and reliable compared with FLM images based on GFP labeling.
Fig. 3. Imaging of the transgenic Tg (LCR-EGFP) zebrafish embryos without and with ATV treatment (ICH model): representative images. (a)−(c) BFM images. (d)−(f) FLM images. (g)−(i) PAM images. For each column, the same region of the same zebrafish sample was imaged. The identified ICH regions are marked by the triangles. Scale bar: 100 µm. All images share the same scale bar.
For HA disease models, the zebrafish embryos with PHZ treatment (for 24 hours) to induce HA were used as the experimental group (n = 10), while those without PHZ treatment were used as the control group. BFM, FLM, and PAM images of the same samples were obtained, respectively, for side-by-side comparison. First, for the experimental group, HA can be easily identified by checking defective patterns in both FLM and PAM images, which is obviously different from the corresponding control group. Similarly, since PAM and FLM images show consistent results, no further quantitative analysis is necessary. Secondly, interestingly, no obvious difference is observed between the control and experimental groups of the BFM images. That is, it is almost not possible to identify HA using BFM imaging. Figure 4 shows the imaging results of the control and experimental (representative images) groups. As mentioned above, for the images in the experimental group, defective patterns can be easily observed in both FLM and PAM images (Figs. 4(d) and 4(f)), while no obvious difference is observed in the BFM image (Fig. 4(b)). For the control group (Figs. 4(a), 4(c), and 4(e)), similar to Figs. 2(a)–2(c), compared with FLM, PAM has advantages in better image quality in ISVs and DLAV in the trunk and no obvious light scattering. It is worth noting that in Fig. 4(f) of the PAM image, we found that clumps of RBCs in irregular morphology accumulated near the caudal hematopoietic tissue (CHT) region, which are probably the dying cells and the aggregated cytoplasmic hemoglobin due to the PHZ treatment, as reported previously [33]. By contrast, this is less obvious in Fig. 4(d) of the FLM image. Similarly, overall, the results suggest that: (i) Both PAM and FLM can perform well and consistently in HA identification in the transgenic zebrafish strain; (ii) PAM is predominantly sensitive in HA identification than BFM; (iii) Compared with FLM, PAM may be more sensitive to reveal ISVs, DLAV, and the clumps of RBCs in the CHT region.
Fig. 4. Imaging of the transgenic Tg (LCR-EGFP) zebrafish embryos without and with PHZ treatment (HA model): representative images. (a,b) BFM images. (c,d) FLM images. (e,f) PAM images. For each column, the same region of the same zebrafish sample was imaged. Scale bar: 100 µm. All images share the same scale bar.
3.3 In vivo imaging of disease models in casper zebrafish
We then performed in vivo imaging of ICH and HA in casper zebrafish embryos. As mentioned in Section 2.2 and as shown in Fig. 2(e), FLM is not able to visualize RBCs in the casper zebrafish due to the absence of the transgene Tg (LCR-EGFP). This is evidence that PAM can be more flexible in studying different zebrafish strains by relying on the intrinsic absorption of RBCs. As mentioned in Section 3.2, to quantitatively demonstrate that PAM can be applied to non-transgenic zebrafish disease models and is more sensitive in ICH identification than BFM, we conducted the statistical study of identifying ICH in casper zebrafish embryos by BFM and PAM. 60 casper zebrafish embryos were used, where 48 embryos were used as the experimental group (with ATV treatment) and 12 as the control group (without ATV treatment). For the experimental group, similar ATV treatment was applied to casper zebrafish embryos at 24 hpf and was sustained until imaging experiments, which were conducted at 34 hpf (n = 24) and 48 hpf (n = 24) by BFM and PAM imaging. The results are shown in Figs. 5 and 6 for ATV treatment for 10 hours and 24 hours, respectively.
Fig. 5. Imaging of the casper zebrafish embryos without and with ATV treatment (ICH model): representative images and statistical study. (a)−(c) BFM images. (d)−(f) PAM images. For each column, the same region of the same zebrafish sample was imaged. The identified ICH regions are marked by the triangles. Scale bar: 100 µm. All images share the same scale bar. (g) Statistical study results for ICH identification. SD is denoted by the error bars. *P = 0.0267 (by Tukey's test).
Fig. 6. Imaging of the casper zebrafish embryos without and with ATV treatment (ICH model): representative images and statistical study. (a)−(c) BFM images. (d)−(f) PAM images. For each column, the same region of the same zebrafish sample was imaged. The identified ICH regions are marked by the triangles. Scale bar: 100 µm. All images share the same scale bar. (g) Statistical study results for ICH identification. SD is denoted by the error bars. *P = 0.0339 (by Tukey's test).
For the control group in Fig. 5 (the first column), as expected, no obvious ICH is recognized. For the experimental group in Fig. 5 (the second and third columns), the representative results from two zebrafish embryos (out of 24 samples) are provided. In Figs. 5(b) and 5(e), ICH is observed in both BFM and PAM images, and the positions of ICH in the two images are consistent. By contrast, in Figs. 5(c) and 5(f)), PAM can detect ICH in the brain region more easily than BFM, which is because of the high contrast enabled by PAM imaging. Note that in PAM images, PA signals from the eyes are much stronger than PA signals from RBCs. This is due to the strong absorption of abundant pigment in the eyes of the casper zebrafish embryos. Fortunately, PA signals from RBCs still provide sufficient SNR for ICH identification, as shown in Figs. 5(e) and 5(f).
For the statistical analysis, all imaged results for ATV treatment for 10 hours (n = 24) were taken into account. The standard of ICH identification was detailed previously (Section 3.2). The identification rate of ICH, or ICH positive rate (denoted as ICH+), is calculated as: ICH+ = (the number of successful ICH identification)/(the total number of samples with ATV treatment). Note that in our case, the ICH negative rate (denoted as ICH-) is simply calculated as: ICH- = 1 − ICH+. We compared the ICH+ between BFM and PAM imaging methods. As mentioned in Section 2.2, three batches of zebrafish embryos were used. Therefore, for the case of ATV treatment for 10 hours (experimental group), 8 [= 24/3] zebrafish embryos were used for each batch. The “batch” ICH+ is calculated from the results of each batch (n = 8), denoted as ICH+1, ICH+2, and ICH+3, respectively. The final ICH+ is calculated from the results of the whole 24 zebrafish embryos. On the other hand, the standard deviation (SD) value of ICH+ is calculated from the three batch ICH+ (ICH+1, ICH+2, and ICH+3), in part to verify the variation among the three batches. The results of the statistical analysis are shown in Fig. 5(g) and Table 2. As can be seen, the ICH+ by BFM and PAM imaging is 17% and 47%, respectively. Analyzed by Turkey’s test, there is a significant difference (*P = 0.0267) for identifying ICH between the PAM and BFM imaging methods. The analysis suggests that PAM is more sensitive for ICH identification than BFM.
Table 2. Statistical analysis for ICH in casper zebrafish embryos
For the case of ATV treatment for 24 hours, the results are shown in Fig. 6, and their statistical analysis is also listed in Table 2. Similarly, for the experimental group in Fig. 6 (the second and third columns), the representative results from two zebrafish embryos (out of 24 samples) are provided. Figures 6(b) and 6(e) show the results in one embryo that ICH is identified in both BFM and PAM images, whereas Figs. 6(c) and 6(f) show the results from another zebrafish embryo that only PAM enables the identification of ICH. Overall, the results and trends in Fig. 6 are similar to those in Fig. 5. The difference is that compared with Fig. 5(g), ICH+ is increased to 41% and 64% for BFM and PAM, respectively, in Fig. 6(g). This is reasonable considering longer ATV treatment in Fig. 6. Besides, compared with Fig. 5(g), the reduced difference in ICH+ between PAM and BFM in Fig. 6(g) (the ratio of (ICH+ of PAM)/(ICH+ of BFM) from ∼2.8 times in Fig. 5(g) to ∼1.5 times in Fig. 6(g)) indicates that PAM has more advantage in early identification of ICH. The increased significant difference *P shows the conformance.
Finally, HA disease models in casper zebrafish embryos were studied. Similarly, the zebrafish embryos with PHZ treatment (for 24 hours) were used as the experimental group, while those without PHZ treatment were used as the control group. The BFM and PAM images are shown in Fig. 7. For better visualization, zoom PAM images are also shown in Figs. 7(e) − 7(h). For the control group (Figs. 7(a), 7(c), 7(e), and 7(g)), normal patterns of blood vessels are observed in BFM and PAM images. For the experimental group of PAM images (Figs. 7(d), 7(f), and 7(h)), PA signals from the majority of vessels became much weaker (e.g., DA and CV in Fig. 7(h)), and some vessels became disordered or disappeared (e.g., near the heart and yolk sac in Fig. 7(f)), indicating the damage of RBCs. This also confirms that HA is an RBC-related disease. By contrast, for the experimental group of BFM image (Fig. 7(b)), no unusual patterns are observed. The results show that PAM imaging can be very sensitive for HA identification. HA disease in zebrafish embryos can be easily recognized by PAM and is almost indistinguishable by BFM.
Fig. 7. Imaging of the casper zebrafish embryos without and with PHZ treatment (HA model). (a,b) BFM images. (c,d) PAM images. (e,f) Zoom PAM images of the trunk. (g,h) Zoom PAM images of the tail. For each column in BFM and PAM images ((a)−(d) only), the same region of the same zebrafish sample was imaged. Scale bars: 100 µm. (e) and (g) share the same scale bar in (g). (f) and (h) share the same scale bar in (h).
For all experiments of PAM imaging of zebrafish embryos in Section 3, we confirmed that the zebrafish embryos were still alive without obvious damage after conducting PAM imaging.
4. Discussion
Zebrafish has become an attractive animal to model and study human diseases. Due to its amenability to optical imaging, fluorescence live imaging is a fundamental tool to study cellular processes and biological functions in zebrafish. To do that, fluorescent labeling is usually required (e.g., using the transgenic zebrafish line Tg (LCR-EGFP) to label RBCs). However, this approach suffers from extra and laborious efforts in order to achieve stable GFP expression in the transgenic zebrafish line. Moreover, this method has several issues and challenges that potentially affect the image accuracy and quality. For example, the fluorescent imaging from transgene may not faithfully represent the actual structure or biology of the label cells, especially during real-time monitoring.
Alternatively, label-free PAM offers a solution to address the above issues for the study of RBC-related diseases in zebrafish models. First, PAM is based on the intrinsic absorption of RBCs to produce images, and thus, labeling is not required, which may greatly facilitate the sample preparation. Secondly, the results of PAM images can be regarded as relatively reliable and accurate because, in principle, PA signals will directly and stably come from RBCs with high contrast (or little background noise). In this work, for ICH and HA disease models in the transgenic zebrafish strain mainly studied by PAM and FLM, we experimentally proved that: (i) PAM performs well in ICH and HA identification in both the transgenic strain and the casper strain without transgene, which suggests that PAM imaging of ICH and HA are more amenable to various genetic backgrounds. By contrast, FLM can only be used for imaging the transgenic zebrafish strain. (ii) Compared with FLM images, by virtue of the intrinsic and high absorption contrast of RBCs, PAM images better reveal ISVs, DLAV, and the clumps of RBCs in the CHT region (HA models). Besides, PAM circumvents the issue of the background autofluorescence in FLM. As for ICH and HA disease models in the casper zebrafish strain imaged by PAM and BFM, the results show that PAM is prominently preferable to BFM for the identification of ICH and HA.
Note that in this work, BFM images were taken using two different microscopes: the microscope (SZX2-ILLB) in reflection mode for ICH zebrafish disease models and the fluorescence microscope (BX53) in transmission mode for HA zebrafish disease models. For ICH zebrafish embryos, the red bleeding in the brain region can be more easily captured by the SZX2-ILLB microscope in reflection mode when the sample was placed on a white reflective substrate. By contrast, we also imaged the HA zebrafish embryos by the SZX2-ILLB microscope with the same setting. We found that blood vessels look less obvious using the SZX2-ILLB microscope (results not shown) than using the BX53 microscope. This is why the BX53 microscope was used to image HA zebrafish embryos throughout this study.
For zebrafish ICH models (either in the transgenic or casper zebrafish strains), PA signals produced by the bleeding in the brain are relatively strong, while for HA zebrafish disease models (similarly, in the transgenic and casper strains), PA signals generated from the small blood vessels (e.g., ISVs and DLAV) are much weaker than those from the large blood vessels (e.g., DA and CV). Therefore, to clearly visualize the vessels for the HA study, laser energy of 160 nJ per pulse and signal averaging of 8 times were used, while for the ICH study, laser energy of 80 nJ per pulse and signal averaging of 4 times were employed. Note that laser energy of 160 nJ per pulse used in this work did not cause obvious damage to zebrafish embryos, as mentioned in Section 3. To further enhance PA signals by using higher laser energy, laser safety ultimately imposes a limit to the maximum allowable laser energy.
Currently, our PAM system employs the unfocused hydrophone for PA signal collection, which is because such a system is readily available in our laboratory. Besides, the moderate sensitivity of our PAM system can suffice PAM imaging of zebrafish in this study. Adopting focused ultrasound detection in PAM (e.g., 2nd-generation PAM [37]) indeed enables more sensitive PA signal collection. In this case, lower laser energy can be used, and thus, the sample is less prone to be damaged. Currently, the imaging speed of our PAM system is mainly limited by the laser pulse repetition rate (2 kHz) and the motorized scanning method. Adopting a higher laser pulse repetition rate and a faster scanning method (e.g., a galvanometer scanner together with a motorized stage [38]) will greatly facilitate the PAM imaging experiments in this work.
As a preliminary experiment, we tested ATV with different concentrations to induce ICH and found that 0.84 µg/ml ATV is sufficient to induce ICH without causing obvious damage to zebrafish embryos, which is consistent with the ATV concentration used in previous studies [29,39]. Therefore, 0.84 µg/ml ATV was used in this work. On the other hand, similarly, we found that 2 µg/ml PHZ can induce obvious symptoms of anemia. Compared with the previous study [40], 2 µg/ml PHZ can be considered as relatively low concentration and is supposed not to cause obvious toxicity to zebrafish embryos. Note that too high PHZ concentration can induce serious anemia, which severely damages zebrafish embryos [40,41]. Thus, 2 µg/ml PHZ was used in this work.
The common practice for ICH identification in transgenic zebrafish (i.e., with fluorescent labels) is fluorescence imaging [39,42,43]. To our knowledge, ICH identification in the zebrafish line without fluorescent labeling is investigated for the first time. Since fluorescence imaging is not applicable to such zebrafish, to our knowledge, there is no gold standard for ICH identification in casper zebrafish. Therefore, in our quantitative study of ICH identification in casper zebrafish in Section 3.3, there is no ground-truth percentile of ICH+ in ATV-treated casper zebrafish embryos. Instead, we simply counted “identified ICH” and “no ICH identified” based on the standard detailed in Section 3.2 and then calculated ICH+.
5. Conclusions
We demonstrated the ability of PAM as a potential tool for live imaging studies of RBC-related disorders in zebrafish. To demonstrate the advantage of PAM over FLM, the transgenic zebrafish strain was used. The results show that although both achieved successful identification of ICH and HA, PAM can provide better images in several aspects because of the intrinsic and high contrast of RBCs in PAM imaging. Further, the disease models in the casper zebrafish strain without fluorescent labeling were studied by PAM and BFM. We experimentally demonstrated that PAM can much better identify ICH and HA than BFM. Although mainly morphology and structural information of blood cells and vessels were acquired and analyzed in this work, PAM can acquire functional information such as angiogenesis and oxygen saturation in zebrafish models, which is of interest for future work. Besides, label-free PAM for studying other zebrafish disease models such as cardiovascular disorders is worth exploring. Our work could motivate live PAM imaging of human diseases in zebrafish models to study the disease pathogenesis and discover new therapies.
Funding
National Natural Science Foundation of China (61775134); Shanghai Pujiang Talent Program (16PJ1405500); The Open Platform on Target Identification of Innovative Drugs of Shanghai Jiao Tong University (WH101117001).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. J. T. Shin and M. C. Fishman, “From zebrafish to human: Modular medical models,” Annu. Rev. Genomics Hum. Genet. 3(1), 311–340 (2002). [CrossRef]
2. D. Beis, T. Bartman, S. W. Jin, I. C. Scott, L. A. D’Amico, E. A. Ober, H. Verkade, J. Frantsve, H. A. Field, A. Wehman, H. Baier, A. Tallafuss, L. Bally-Cuif, J. N. Chen, D. Y. R. Stainier, and B. Jungblut, “Genetic and cellular analyses of zebrafish atrioventricular cushion and valve development,” Development 132(18), 4193–4204 (2005). [CrossRef]
3. S. A. Carney, A. L. Prasch, W. Heideman, and R. E. Peterson, “Understanding dioxin developmental toxicity using the zebrafish model,” Birth Defects Res., Part A 76(1), 7–18 (2006). [CrossRef]
4. J. F. Amatruda and E. E. Patton, “Genetic models of cancer in zebrafish,” Int. Rev. Cel. Mol. Bio. 271, 1–34 (2008). [CrossRef]
5. M. W. Hung, Z. J. Zhang, S. Li, B. Lei, S. Yuan, G. Z. Cui, P. Man Hoi, K. Chan, and S. M. Lee, “From omics to drug metabolism and high content screen of natural product in zebrafish: a new model for discovery of neuroactive compound,” Evid. Based Complement. Alternat. Med. 2012, 1–20 (2012). [CrossRef]
6. K. Kulkeaw and D. Sugiyama, “Zebrafish erythropoiesis and the utility of fish as models of anemia,” Stem Cell Res. Ther. 3(6), 55 (2012). [CrossRef]
7. L. Marjoram and M. Bagnat, “Infection, inflammation and healing in zebrafish: Intestinal inflammation,” Curr. Pathobiol. Rep. 3(2), 147–153 (2015). [CrossRef]
8. J. T. Baeten and J. L. O. de Jong, “Genetic models of leukemia in zebrafish,” Front. Cell Dev. Biol. 6, 115 (2018). [CrossRef]
9. S. Isogai, M. Horiguchi, and B. M. Weinstein, “The vascular anatomy of the developing zebrafish: An atlas of embryonic and early larval development,” Dev. Biol. 230(2), 278–301 (2001). [CrossRef]
10. O. Renaud, P. Herbomel, and K. Kissa, “Studying cell behavior in whole zebrafish embryos by confocal live imaging: application to hematopoietic stem cells,” Nat. Protoc. 6(12), 1897–1904 (2011). [CrossRef]
11. E. Linney, N. L. Hardison, B. E. Lonze, S. Lyons, and L. DiNapoli, “Transgene expression in zebrafish: A comparison of retroviral-vector and DNA-injection approaches,” Dev. Biol. 213(1), 207–216 (1999). [CrossRef]
12. N. Hinfray, F. Sohm, M. Caulier, E. Chadili, B. Piccini, C. Torchy, J. M. Porcher, Y. Guiguen, and F. Brion, “Dynamic and differential expression of the gonadal aromatase during the process of sexual differentiation in a novel transgenic cyp19a1a-eGFP zebrafish line,” Gen. Comp. Endocrinol. 261, 179–189 (2018). [CrossRef]
13. J. J. Ganis, N. Hsia, E. Trompouki, J. L. de Jong, A. DiBiase, J. S. Lambert, Z. Jia, P. J. Sabo, M. Weaver, R. Sandstrom, J. A. Stamatoyannopoulos, Y. Zhou, and L. I. Zon, “Zebrafish globin switching occurs in two developmental stages and is controlled by the LCR,” Dev. Biol. 366(2), 185–194 (2012). [CrossRef]
14. G. Stamatoyannopoulos, “Control of globin gene expression during development and erythroid differentiation,” Exp. Hematol. 33(3), 259–271 (2005). [CrossRef]
15. P. Corish and C. Tyler-Smith, “Attenuation of green fluorescent protein half-life in mammalian cells,” Protein Eng. 12(12), 1035–1040 (1999). [CrossRef]
16. M. S. Allen, J. R. Wilgus, C. S. Chewning, G. S. Sayler, and M. L. Simpson, “A destabilized bacterial luciferase for dynamic gene expression studies,” Syst. Synth. Biol. 1(1), 3–9 (2007). [CrossRef]
17. J. Yao and L. V. Wang, “Photoacoustic Microscopy,” Laser Photon. Rev. 7(5), 758–778 (2013). [CrossRef]
18. S. Q. Ye, R. Yang, J. W. Xiong, K. K. Shung, Q. F. Zhou, C. H. Li, and Q. S. Ren, “Label-free imaging of zebrafish larvae in vivo by photoacoustic microscopy,” Biomed. Opt. Express 3(2), 360–365 (2012). [CrossRef]
19. L. Zhu, L. Li, L. Gao, and L. V. Wang, “Multi-view optical resolution photoacoustic microscopy,” Optica 1(4), 217–222 (2014). [CrossRef]
20. Q. Chen, T. Jin, W. Qi, X. Mo, and L. Xi, “Label-free photoacoustic imaging of the cardio-cerebrovascular development in the embryonic zebrafish,” Biomed. Opt. Express 8(4), 2359–2367 (2017). [CrossRef]
21. R. Haindl, S. Preisser, M. Andreana, W. Rohringer, C. Sturtzel, M. Distel, Z. Chen, E. Rank, B. Fischer, W. Drexler, and M. Liu, “Dual modality reflection mode optical coherence and photoacoustic microscopy using an akinetic sensor,” Opt. Lett. 42(21), 4319–4322 (2017). [CrossRef]
22. J. Yang, L. Gong, X. Xu, P. Hai, Y. Shen, Y. Suzuki, and L. V. Wang, “Motionless volumetric photoacoustic microscopy with spatially invariant resolution,” Nat. Commun. 8(1), 780 (2017). [CrossRef]
23. X. Yang, B. Jiang, X. Song, J. Wei, and Q. Luo, “Fast axial-scanning photoacoustic microscopy using tunable acoustic gradient lens,” Opt. Express 25(7), 7349–7357 (2017). [CrossRef]
24. M. J. Moore, S. El-Rass, Y. Xiao, Y. Wang, X. Y. Wen, and M. C. Kolios, “Simultaneous ultra-high frequency photoacoustic microscopy and photoacoustic radiometry of zebrafish larvae in vivo,” Photoacoustics 12, 14–21 (2018). [CrossRef]
25. C. Liu, J. Liao, L. Chen, J. Chen, R. Ding, X. Gong, C. Cui, Z. Pang, W. Zheng, and L. Song, “The integrated high-resolution reflection-mode photoacoustic and fluorescence confocal microscopy,” Photoacoustics 14, 12–18 (2019). [CrossRef]
26. R. Haindl, A. J. Deloria, C. Sturtzel, H. Sattmann, W. Rohringer, B. Fischer, M. Andreana, A. Unterhuber, T. Schwerte, M. Distel, W. Drexler, R. Leitgeb, and M. Liu, “Functional optical coherence tomography and photoacoustic microscopy imaging for zebrafish larvae,” Biomed. Opt. Express 11(4), 2137–2151 (2020). [CrossRef]
27. J. Liu, S. D. Fraser, P. W. Faloont, E. L. Rollins, J. V. Berg, O. Starovic-Subota, A. L. Laliberte, J. N. Chen, F. C. Serlucat, and S. J. Childs, “A beta Pix-Pak2a signaling pathway regulates cerebral vascular stability in zebrafish,” Proc. Natl. Acad. Sci. USA 104(35), 13990–13995 (2007). [CrossRef]
28. S. Eisa-Beygi and M. Rezaei, “Etiology of intracerebral hemorrhage (ICH): novel insights from Zebrafish embryos,” Int. J. Dev. Biol. 60(4-5-6), 119–126 (2016). [CrossRef]
29. S. Crilly, A. Njegic, S. E. Laurie, E. Fotiou, G. Hudson, J. Barrington, K. Webb, H. L. Young, A. P. Badrock, A. Hurlstone, J. Rivers-Auty, A. R. Parry-Jones, S. M. Allan, and P. R. Kasher, “Using zebrafish larval models to study brain injury, locomotor and neuroinflammatory outcomes following intracerebral haemorrhage,” F1000Res. 7, 1617 (2018). [CrossRef]
30. Z. Y. Zhou, B. Huang, S. Li, X. H. Huang, J. Y. Tang, Y. W. Kwan, P. M. Hoi, and S. M. Lee, “Sodium tanshinone IIA sulfonate promotes endothelial integrity via regulating VE-cadherin dynamics and RhoA/ROCK-mediated cellular contractility and prevents atorvastatin-induced intracerebral hemorrhage in zebrafish,” Toxicol. Appl. Pharmacol. 350, 32–42 (2018). [CrossRef]
31. E. Shafizadeh, R. T. Peterson, and S. Lin, “Induction of reversible hemolytic anemia in living zebrafish using a novel small molecule,” Comp. Biochem. Physiol. C Toxicol. Pharmacol. 138(3), 245–249 (2004). [CrossRef]
32. K. F. Ferri-Lagneau, K. S. Moshal, M. Grimes, B. Zahora, L. Lv, S. Sang, and T. Leung, “Ginger stimulates hematopoiesis via Bmp pathway in zebrafish,” PLoS One 7(6), e39327 (2012). [CrossRef]
33. A. Lenard, E. Alghisi, H. Daff, M. Donzelli, C. McGinnis, and C. Lengerke, “Using zebrafish to model erythroid lineage toxicity and regeneration,” Haematologica 101(5), e164–e167 (2016). [CrossRef]
34. M. O. Parker, A. J. Brock, M. E. Millington, and C. H. Brennan, “Behavioural phenotyping of casper mutant and 1-Pheny-2-Thiourea treated adult zebrafish,” Zebrafish 10(4), 466–471 (2013). [CrossRef]
35. R. M. White, A. Sessa, C. Burke, T. Bowman, J. LeBlanc, C. Ceol, C. Bourque, M. Dovey, W. Goessling, C. E. Burns, and L. I. Zon, “Transparent adult zebrafish as a tool for in vivo transplantation analysis,” Cell Stem Cell 2(2), 183–189 (2008). [CrossRef]
36. A. Agalou, M. Thrapsianiotis, A. Angelis, A. Papakyriakou, A. L. Skaltsounis, N. Aligiannis, and D. Beis, “Identification of novel melanin synthesis inhibitors from crataegus pycnoloba using an in vivo zebrafish phenotypic assay,” Front. Pharmacol. 9, 265 (2018). [CrossRef]
37. S. Hu, K. Maslov, and L. V. Wang, “Second-generation optical-resolution photoacoustic microscopy with improved sensitivity and speed,” Opt. Lett. 36(7), 1134–1136 (2011). [CrossRef]
38. J. Kim, J. Y. Kim, S. Jeon, J. W. Baik, S. H. Cho, and C. Kim, “Super-resolution localization photoacoustic microscopy using intrinsic red blood cells as contrast absorbers,” Light: Sci. Appl. 8(1), 103 (2019). [CrossRef]
39. S. Li, N. N. Ai, M. Y. Shen, Y. Y. Dang, C. M. Chong, P. C. Pan, Y. W. Kwan, S. W. Chan, G. P. H. Leung, M. P. M. Hoi, T. J. Hou, and S. M. Y. Lee, “Discovery of a ROCK inhibitor, FPND, which prevents cerebral hemorrhage through maintaining vascular integrity by interference with VE-cadherin,” Cell Death Discov 3(1), 17051 (2017). [CrossRef]
40. R. E. Rajagopal, M. Balasubramanian, and S. Kalyanaraman, “Phenylhydrazine hydrochloride induced dosedependent embryo cytotoxicity in zebrafish,” Bioinformation 15(4), 255–260 (2019). [CrossRef]
41. Y. Fang, Y. Sun, C. Luo, J. Gu, Z. Shi, G. Lu, J.-S. Silvestre, and Z. Chen, “Evaluation of cardiac dysfunction in adult zebrafish using high frequency echocardiography,” Life Sci. 253, 117732 (2020). [CrossRef]
42. D. A. Buchner, F. Su, J. S. Yamaoka, M. Kamei, J. A. Shavit, L. K. Barthel, B. Mcgee, J. D. Amigo, S. Kim, A. W. Hanosh, P. Jagadeeswaran, D. Goldman, N. D. Lawson, P. A. Raymond, B. M. Weinstein, D. Ginsburg, and S. E. Lyons, “pak2a mutations cause cerebral hemorrhage in redhead zebrafish,” Proc. Natl. Acad. Sci. USA 104(35), 13996–14001 (2007). [CrossRef]
43. S. Crilly, A. Njegic, A. R. Parry-Jones, S. M. Allan, and P. R. Kasher, “Using zebrafish larvae to study the pathological consequences of hemorrhagic stroke,” J. Visualized Exp. 148, e59716 (2019). [CrossRef]
