Replacement in angiogenesis research: Studying mechanisms of blood vessel development by animal-free in vitro, in vivo and in silico approaches.

A major prerequisite for the adequate use and interpretation of angiogenesis assays is the basic understanding of the dynamic process of blood vessel formation, which is characterized by the coordinated interaction of humoral factors and different cell types. It is initiated by a local imbalance of pro-angiogenic factors, such as vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF), and anti-angiogenic factors, such as endostatin and thrombospondin ( Naumov et al., 2006 ; Hu et al., 2014 ). Potent triggers for this so-called angiogenic switch are tissue hypoxia and inflammation ( Hashimoto and Shibasaki, 2015 ; Ridiandries et al., 2016 ). This results in the detachment of stabilizing pericytes from the wall of pre-existing (parent) microvessels and the angiogenic activation of microvascular endothelial cells, which start to degrade their basement membrane by the release of proteolytic matrix metalloproteinases (MMPs) ( Carmeliet and Jain, 2011 ). In a next step, the endothelial cells migrate out of the vessel wall. Tightly regulated by Notch/DLL4 signaling, they differentiate into filopodia-forming tip cells, which spearhead new vessel sprouts towards an angiogenic stimulus and are followed by proliferating stalk cells ( Potente et al., 2011 ). The elongating sprouts develop a lumen and interconnect with each other to blood-perfused microvascular networks. These networks are finally stabilized by the formation of a new basement membrane and the recruitment of perivascular cells ( Sun et al., 2015 ). This brief description indicates that the process of blood vessel development can be subdivided in several well-characterized steps, which involve angiogenic activation, matrix degradation, endothelial cell migration and proliferation as well as sprouting, network formation and vessel maturation ( Figure 1 ). Comparably, most animal-free angiogenesis assays can also be subdivided according to their main mechanism(s) of action to be studied ( Figure 1 ).

At first glance, this section seems not to fit in the present review article focusing on animal-free approaches. However, the in vivo models described below share the common feature that they are exclusively based on primitive organisms, i.e., fertilized chicken eggs and zebrafish larvae. Due to the stage of their development, these organisms lack pain perception. This is also a major reason why by law in many countries these models do not have to be registered as animal experiments and are even recommended as suitable approaches for the reduction and replacement of such. The chick chorioallantois membrane (CAM) assay is a well-established in vivo approach, which has been widely used in many different modifications and various areas of research for decades ( Ribatti, 2017 ; Chu et al., 2022 ). It uses the CAM as target tissue for the analysis of angiogenic processes. The CAM forms by the fusion of the allantois and chorion in fertilized chicken eggs and serves as a transient gas exchange surface for the embryo. It rapidly develops between day 3 and 9 into a highly vascularized tissue layer containing a dense network of arterioles, capillaries and venules ( Laschke et al., 2006 ). In the classic in ovo assay, a circular, sealable observation window is prepared into the eggshell for the repeated analysis of the CAM ( Schmitd et al., 2019 ). Alternatively, ex ovo assays with shell-less embryo cultures have been introduced, which enable an easier access to larger areas of the CAM ( Vargas et al., 2009 ; Merlos Rodrigo et al., 2021 ). On the other hand, these assays require a more demanding preparation and complex incubation environment and, thus, also bear a higher risk of infection and embryo mortality. In angiogenesis research, the CAM assay has been broadly used to investigate basic mechanisms of blood vessel development, including the migration, proliferation and differentiation of endothelial cells ( Ausprunk et al., 1974 ; Kurz et al., 1995 ) as well as ECM remodeling ( Ausprunk, 1986 ). Moreover, it is well suited to study the pro- or anti-angiogenic effects of natural factors and pharmacological compounds ( Jacoby et al., 2010 ; Farina et al., 2011 ). These can either be topically applied to the CAM ( Liu et al., 2022 ) or administered systemically by intravascular injection ( Storgard et al., 2005 ). In addition, the CAM frequently serves as host tissue to study the vascularization of implanted biomaterials as well as benign and malignant tissue grafts ( Gescher et al., 2005 ; Ademi et al., 2021 ; Ara et al., 2022 ). The latter ones can be analyzed independently of their species origin without the risk of immunological rejection, because the early chicken embryo lacks a functional immune system ( Baiguera et al., 2012 ). However, the analysis of their vascularization is only possible during a relatively short time period of ∼10 days, because chicken embryos already hatch on developmental day 21. Correspondingly, in most countries CAM-based studies are restricted to developmental day 14–15 to fulfil the criterion of animal-free experiments ( Nowak-Sliwinska et al., 2018 ). There are many possibilities to visualize and analyze blood vessel formation and patterning in the CAM assay. These range from simple macroscopic inspection, light and fluorescence microscopy to sophisticated imaging technologies, including ultrasonography, optical Doppler tomography, microcomputed tomography and magnetic resonance imaging ( Nowak-Sliwinska et al., 2014 ; Moreno-Jiménez et al., 2017 ; Eckrich et al., 2020 ). Obviously, they markedly differ in many aspects, such as resolution, required equipment and expense. Hence, the choice of the right approach is crucially dependent on the research question and the present laboratory conditions. Furthermore, it should be considered that quantitative analyses in the CAM assay (e.g., the measurement of microvessel densities or branching points) are quite challenging, because of the high variability of the angiogenic response and embryo-induced movements of the CAM. Therefore, multiple measurements need to be performed to generate statistically valid data sets. This, however, is not a major problem considering the fact that chicken eggs are quite cheap and the technical preparation of the CAM is easily feasible. Thus, this versatile assay is also suitable for large-scale in vivo screenings. The zebrafish ( Danio rerio ) is a versatile and widely used in vivo model in angiogenesis research, which combines several essential advantages. This fish is easy to keep under laboratory conditions and produces hundreds of larvae per week through mating, which enables large-scale screenings ( Lieschke and Currie, 2007 ). In early developmental stages younger than 120 h, these larvae lack the legal status of experimental animals, although they already exhibit a rudimentary yet functional cardiovascular system 24 h after fertilization ( Isogai et al., 2001 ). Moreover, they are transparent, which provides easy access to their microcirculation and internal organs for microscopic in vivo imaging. Since ∼70% of the human genes have an orthologue in the zebrafish genome ( Howe et al., 2013 ), the prediction quality of pharmacological testings in zebrafish larvae for human applications is good. In addition, the zebrafish is suitable for genetic manipulation. Accordingly, numerous transgenic zebrafish lines are available by now. They allow the visualization of endothelial cells and their precursors, perivascular cells and blood cells during vascular network development by means of cell-specific expression of fluorescent reporter proteins ( Chávez et al., 2016 ). On the other hand, gene silencing and editing, which have originally been widely performed by the application of morpholino antisense oligonucleotides ( Wyatt et al., 2015 ) and more recently by means of TALEN ( Bedell et al., 2012 ) and CRISPR/Cas9 ( Cornet et al., 2018 ), offer the opportunity to study the function of individual genes during vasculogenesis, sprouting angiogenesis and vascular remodeling ( Eberlein et al., 2021 ). In this context, it should be mentioned that zebrafish larvae are able to survive without a functional vascular system by passive oxygen diffusion up to 5 days ( Stainier et al., 1996 ). This also enables the investigation of late phenotypes of vascular malformations, which would be otherwise lethal in living mammals ( Isogai et al., 2001 ; Peterson et al., 2004 ; Chávez et al., 2016 ). In addition, it is possible to inject human cells into the larvae without immunological rejection. This allows the analysis of angiogenesis in a humanized tissue-specific environment. For instance, Wu et al. (2017) demonstrated a potent anti-angiogenic effect of VRI, a pyridinyl-anthranilamide compound inhibiting the kinase activities of both VEGF receptor-1 and 2, on the xenografted fluorescently labeled gastric cancer cell lines AGS and SGC-7901 in transgenic fli-eGFP zebrafish embryos ( Figure 4 ). Inhibition of tumor angiogenesis in zebrafish larvae according to Wu et al. (2017) . (A) Typical confocal microscopic images of subintestinal vessels of an uninjected transgenic fli-eGFP zebrafish larva at 3 days post fertilization. (B,C) Fluorescently labeled (CM-DiI) gastric cancer AGS cells (B) and SGC-7901 cells (C) were injected to zebrafish larvae and induced angiogenesis at day 1 post injection. (D,E) 50 nM VRI inhibited angiogenesis of the subintestinal vessels caused by the cell lines AGS (D) and SGC-7901 (E) . The white boxes at lower right corner show the higher magnification of the upper left white boxes. The arrows indicate the tumor cell-induced angiogenesis. Reproduced with permission from BioMed Central under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ). Despite all these advantages, it has to be considered that zebrafish larvae rapidly develop with major changes in their organ architecture and cardiovascular system over time. Accordingly, they are not suitable for long-term studies. Furthermore, they are aquatic organisms and, thus, markedly differ in many physiological aspects, such as their respiration, from mammals ( Chávez et al., 2016 ). Therefore, promising results achieved by means of zebrafish larvae may need to be further validated in mammalian models, which represent closer substitutes for humans. Nonetheless, there is no doubt that the number of complex experiments in mammals can be markedly reduced by means of pharmacological and genetic screening studies in the zebrafish model.

The development of new blood vessels is a dynamic process, which is not only dependent on the coordinated interaction of endothelial and perivascular cells of the microvasculature, but also crucially determined by hemodynamic forces, the local tissue environment and systemic factors. In vivo angiogenesis studies in animal models are considered to reflect these complex conditions and, thus, to provide data of high physiological relevance for human applications. On the other hand, they are not suitable for large-scale screenings, because they are laborious and expensive. Most particularly, however, they confront researchers with serious ethical concerns. Indeed, in line with the 3R principle of Russell and Burch, it is the obligation of the scientific community to continuously establish and refine approaches, which enable an animal-free research of highest quality standards and relevance for basic science and clinical practice. In the present narrative review, we discuss such approaches without any claim to completeness, because we did not perform a systematic literature search. Accordingly, we are aware that we may have missed the one or other interesting assay or model. Nonetheless, we feel that the herein selected and discussed approaches provide an excellent blue print how an animal-free research can be achieved. In fact, they demonstrate that even complex processes, such as angiogenesis, can be analyzed in detail from various viewpoints to gain a valid overall picture. Currently, this usually implies the reasonable combination of different assays in pre-screening studies assessing the pro- or anti-angiogenic activity of test compounds. Such pre-screening studies already contribute to a drastic reduction of animal experiments in early stages of modern drug development. However, the future goal should be to analyze angiogenesis in all its complexity by means of physiologically relevant animal-free approaches without the additional need for a final validation of the results in an animal model. Rapid progress in the generation of tissue- and organ-mimicking microfluidic systems and mathematical modeling may pave the way to turn this fiction into reality.

Angiogenesis is a fundamental biological process defined as the development of new blood vessels from pre-existing ones ( Ribatti and Pezzella, 2021 ). Because cell survival and proliferation are crucially dependent on a sufficient oxygen and nutrient supply, angiogenesis is a major prerequisite for tissue formation and growth. Accordingly, blood vessel formation plays an essential role during embryogenesis and wound healing ( Fajersztajn and Veras, 2017 ; Sorg et al., 2018 ). Moreover, it is important for the physiological reproductive function of the placenta, ovary and uterus ( Reynolds et al., 1992 ; Vollmar et al., 2001 ; Laschke et al., 2008 ). On the other hand, many pathological conditions are typically driven by angiogenesis, such as tumor growth and metastasis ( Folkman, 2002 ), endometriosis ( Laschke and Menger, 2018 ), rheumatoid arthritis ( Wang Y. et al., 2021 ), ocular neovascular diseases ( Plastino et al., 2021 ) and chronic inflammatory skin disorders ( Lee et al., 2021 ). Hence, there is a strong interest in uncovering molecular and cellular angiogenic mechanisms and in assessing the pro- and anti-angiogenic effects of various agents to provide the basis for the establishment of novel therapeutic approaches. For this purpose, a broad spectrum of in vitro , in vivo and in silico assays and models has been introduced in angiogenesis research during the last decades ( Nowak-Sliwinska et al., 2018 ). There is no doubt that animal studies markedly contribute to a better understanding of angiogenesis under different physiological and pathological conditions. However, they also raise major ethical concerns, because they face the fundamental conflict of interest between the claim of advancing scientific knowledge and the protection of animals. To address this critical issue, William Russell and Rex Burch introduced the so-called 3R principle in 1959, which defines central criteria to perform animal experiments in a more humane way ( Russell and Burch, 1959 ). This principle is based on the 3Rs “replacement, reduction and refinement”. Replacement means that animal models should be replaced by animal-free approaches whenever possible. If this is not completely achievable, researchers should at least reduce the number of individual animals required to generate statistically valid and reproducible data to an absolute minimum. Refinement, in turn, focuses on any decrease in the incidence or severity of inhumane procedures applied to those animals, which are still to be used ( Russell and Burch, 1959 ). As from an ethical point of view replacement is the most desirable aim, the present review article selectively provides an overview of common animal-free approaches in angiogenesis research. These approaches allow for the analysis of multiple biological mechanisms that are of utmost importance for the development of new blood vessels ( Figure 1 ). Hence, they do not only represent essential tools in basic science but also in early stages of drug development. The process of angiogenesis and animal-free approaches for its investigation. The process of angiogenesis can be subdivided in several well-characterized steps, which involve i) the angiogenic activation of microvessels by growth factors, ii) the detachment of stabilizing pericytes and the degradation of the basal membrane by MMPs, iii) the migration of endothelial tip cells towards an angiogenic stimulus and iv) the proliferation of following endothelial stalk cells, which results in v) the formation of angiogenic sprouts. These sprouts develop a lumen and vi) finally interconnect with each other to new blood-perfused microvascular networks, which are stabilized by the formation of a new basement membrane and the recruitment of perivascular cells. To study this process in vitro , in vivo or in silico , multiple animal-free angiogenesis assays and models are available focusing on different steps of blood vessel formation.

Mathematical modeling combines experimental data and physical laws to simulate angiogenesis and tissue vascularization in silico . Particularly in cancer research, this approach has been continuously developed further since its initiation by Anderson and Chaplain (1998) to gain detailed information about tumor-driven blood vessel formation and remodeling as well as intra-tumoral oxygen, nutrient and drug distribution ( Shirinifard et al., 2009 ; Welter and Rieger, 2016 ; Suzuki et al., 2018 ) ( Figure 5 ). In this context, it allows to identify general biological principles of angiogenesis and to set up predictive models for the testing of anti-angiogenic therapeutic regimens ( Venkatraman et al., 2016 ; Lai and Friedman, 2019 ; Akbarpour Ghazani et al., 2020 ; Mousavi et al., 2022 ). For this purpose, mathematical modeling of the tumor vasculature can be performed at the cell or the tissue scale by means of discrete (i.e., endothelial cells are treated as individual objects), continuous (i.e., endothelial cells are treated as concentrations) or hybrid (i.e., a combination of discrete and continuous approaches) models, as recently reviewed in detail by Hormuth et al. (2021) . However, it should be noted that many parameters in these models are often assumed values ( Pamuk et al., 2018 ), which may limit the biological relevance and predictive power of the generated results. Hence, it is necessary to continuously improve the calibration and validation of mathematical angiogenesis models by means of biologically based data. This can be achieved by time-resolved imaging and quantification of vascular dynamics under experimental in vivo conditions ( Perfahl et al., 2011 ). Although animal models may provide such conditions, they are laborious and not suitable for high-throughput experiments. Accordingly, in the future they may be gradually replaced by microfluidic approaches, which allow the isolated analysis of specific mechanisms during blood vessel formation and drug testing in highly controlled, repeatable but complex experimental settings ( Hormuth et al., 2021 ). This may also open the door for a broad implementation of computer simulations in combination with artificial intelligence into clinical practice and, thus, for the establishment of personalized therapeutic regimens with improved efficacy and less side effects ( Bodzioch et al., 2021 ). Time-series of 3D tumor growth and angiogenesis according to a mathematical model of Shirinifard et al. (2009) . (A) Day 0: The pre-existing vasculature and the initial normal tumor cell. (B) Day 15: The tumor grows into a sphere with a maximum diameter of about 300 µm. The purple cells are active neovascular cells. (C) Day 30: The tumor grows into a cylinder with a length of about 350 µm and a diameter of about 300 µm. The vasculature is about to rupture. (D) Day 75: The developed vascularized tumor. The white arrowhead shows neovascular cells organized into 2D sheets. Cell types: Green: normal; yellow: hypoxic; red: vascular; purple: neovascular. Axes are labeled in µm. Reproduced with permission from PloS ONE under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ).