Annexin Animal Models—From Fundamental Principles to Translational Research

Routine manipulation of the mouse genome has become a landmark in biomedical research. Traits that are only associated with advanced developmental stages can now be investigated within a living organism, and the in vivo analysis of corresponding phenotypes and functions advances the translation into the clinical setting. The annexins, a family of closely related calcium (Ca2+)- and lipid-binding proteins, are found at various intra- and extracellular locations, and interact with a broad range of membrane lipids and proteins. Their impacts on cellular functions has been extensively assessed in vitro, yet annexin-deficient mouse models generally develop normally and do not display obvious phenotypes. Only in recent years, studies examining genetically modified annexin mouse models which were exposed to stress conditions mimicking human disease often revealed striking phenotypes. This review is the first comprehensive overview of annexin-related research using animal models and their exciting future use for relevant issues in biology and experimental medicine.


Introduction
The assembly of the human genome, followed by DNA deep sequencing, changed our approach to analyse gene functions. Genetic tools, which helped to develop a major part of our understanding of modern biology, started off with phenotypes and aimed to identify the underlying gene or its allelic variation to infer function. The completion of the human genome assembly marked the onset of methods that build on reverse genetics. Genes and their genomic sequence became available and the generation of mouse models with gene deletions or targeted mutations allowed the in vivo analysis of corresponding phenotypes and functions.
Mouse models lacking individual members of the annexin gene family have been established and unravelled some of their in vivo functions. There are many excellent reviews on the annexin family covering the common and specific characteristics of individual annexin family members [1][2][3][4][5][6][7][8][9][10][11][12][13]. Hence, in the following, we provide only a brief overview necessary to appreciate the generation and functional implications from the in vivo analysis of the annexin mouse models.
The annexin family consists of twelve evolutionary conserved and structurally related Ca 2+ -and phospholipid-binding proteins [1,14]. All annexins are structurally related, The annexin family consists of twelve evolutionary conserved and structurally related Ca 2+ -and phospholipid-binding proteins [1,14]. All annexins are structurally related, they encode a variable N-terminal domain, which is responsible for different, annexin member-specific functions, and a highly conserved C-terminal core. The C-terminal domain is composed of four highly related repeats (eight in case of AnxA6) containing Ca 2+ binding sites that enable the interaction with negatively charged phospholipids ( Figure  1). Extensive in vitro analysis solved the three-dimensional structures of annexins ( [15][16][17][18], reviewed in [2,19,20]), their Ca 2+ -dependent affinities to various phospholipids and other interaction partners [1,2]. A large number of regulatory roles associated with individual annexins can be explained by their Ca 2+ -inducible and rapid translocation from the cytosol to the plasma membrane or intracellular membranes [1][2][3][4]. At these locations, their unique N-termini interact with dissimilar binding partners, which in part explains the functional diversity of annexins. In addition, the composition and distribution of phospholipids and other lipids, such as phosphatidylinositol-4,5-bisphosphate, cholesterol and ceramide within specific membrane microdomains, but also posttranslational modifications and pH, contribute to different spatiotemporal and Ca 2+ -sensitive membrane binding kinetics leading to further functional diversification of individual family members [1][2][3][4][5][6][7][8][9]11]. 1AIN [17]) and (C) bovine AnxA6 (PDB: 1AVC [18]) cores (light to dark green, spacer regions in grey, AnxA6 linker region in blue), with coordinated calcium ions (red). 3D-structures are visualized with the iCn3D software vs. 2.24.6; (https://www.ncbi.nlm.nih.gov/Structure/icn3d/icn3d.html; accessed 12 March 2021).
These processes are thought to control basic cellular functions that are pivotal to the regulation of growth, development, programmed cell death, but also cell motility, membrane repair and inflammatory response [1][2][3][4][5][6][7][8][9][10][11][12][13]. Ectopic overexpression or siRNAmediated knockdown delivered great insights in the annexin biochemistry and broadened our knowledge substantially. More advanced technologies of reverse genetics made knockout (KO) mouse models the gold standard to analyse annexin functions. During the last two decades KO-mice strains with targeted deletion of several annexins (i.e., AnxA1, A2, A4, A5, A6, A7 and A8) have been successfully generated and provide an important research tool to the community ( Figure 1). All but one of these mice strains did not display any obvious phenotype and developed entirely normal. Given the high degree of conservation this outcome was not anticipated and suggested that potentially individual members of the annexin gene family could functionally substitute for each other and hence might be redundant. However, annexin KO-mice or wildtype mice with implanted annexin KO-cells developed strong phenotypes when analysed in stress conditions that mimicked certain disease [10]. Additionally, the observed phenotypes were specific to individual annexins; a finding which strongly argued against functional redundancy of individual gene family members. In fact, as outlined further below, annexins in general might provide a toolbox to cope with cell stress and pathogen attack [12,13]. Of late, a substantial number of new studies exploring human disorders in annexin KO-models further emphasize their great therapeutic value and the suitability of whole-body annexin KO-mice or animals implanted with annexin-deficient cell lines to investigate a wide range of human diseases.
The following sections address individual annexins and their corresponding functions in rodent and other animal models in vivo.

AnxA1 and the Anti-Inflammatory Response
An extensive body of work has linked AnxA1 with the regulation of inflammatory processes. Examining glucocorticoid-dependent activation of innate immune cells in the AnxA1 KO-mice (Anxa1 −/− ) provided the first in vivo evidence that AnxA1 limits pro-inflammatory response [24] (Table 1a). Complementing earlier in vivo studies administering recombinant AnxA1 [23,25], Anxa1 −/− mice displayed a boosted immune response characterized by an increased leucocyte migratory behaviour and a substantial resistance to the anti-inflammatory action of glucocorticoids in several acute inflammation models [24]. Mechanistically, in carrageenan-induced paw oedema and zymosan-induced peritonitis, the prominent ability of AnxA1 to counteract inflammatory events occured through binding of extracellular AnxA1 to formyl peptide receptors (FPRs), a family of G-protein coupled receptors found on many cells of the innate immune system, including neutrophils and monocytes [26,27]. The AnxA1/FPR2 axis inhibits neutrophil recruitment to sites of inflammation, a critical step in the regulation of inflammation [28]. In addition, AnxA1 increases neutrophil apoptosis and recruitment of monocytes, which upon differentiation [29], allow clearance of apoptotic cells. The latter might be mediated by locally produced steroids, including estrogens, which promote production and release of AnxA1 [30]. Furthermore, the ability of AnxA1 to reduce production of pro-inflammatory mediators, such as eicosanoids, nitric oxide, and interleukins (ILs) also contributes to resolve inflammation.
Several recent studies have provided additional mechanistic insights. Mast cells critically mediate early lipopolysaccharide-induced neutrophil recruitment, a FPR2-dependent process that can be blocked by AnxA1-derived peptides [37]. Along these lines, in ovalbumin-induced atopic dermatitis-like skin lesions, AnxA1 is responsible for the production of allergen-induced immunoglobubin E, cytokines as well as the recruitment of inflammatory cells, in particular mast cells, to the lesion site [38]. Other studies identified mast cell stabilizers to treat ocular allergy, such as cromoglycate and nedocromil, to promote cellular AnxA1 release, which limits mast cell degranulation and the extend of the allergen-mediated allergic reactions [39,40]. Interestingly, AnxA1 not only impacts on neutrophil homeostasis in the presence of glucocorticoids, but also seems to modulate steady-state neutrophil maturation [31]. In efforts to clarify the role of exogenous AnxA1 for the hematopoietic system, administration of AnxA1, most likely via FPR-induced Ca 2+ and MAPK signalling, promotes myeloid and granulocytic differentiation [41].
Adding further complexity, antigen presentation and T cell activation are also modulated by the AnxA1/FPR2 axis, with consequences for the adaptive immune response [21,33,42,43]. Of note, transgenic mice overexpressing AnxA1 exclusively in T lymphocytes led to an unexpected increase in anxiety, possibly due to an anxiogenic factor released by AnxA1 overexpressing T cells, which may contribute to increased susceptibility for mental disorders in patients with autoimmune diseases [44].

AnxA1 and Progression of Chronic Diseases
In order to return local sites of inflammation to normal homeostasis, AnxA1 is now considered a pro-resolving mediator that can counteract pro-inflammatory response. In fact, administration of recombinant AnxA1 can delay or even regress the progression of chronic diseases, such as cardiovascular disease [22,[45][46][47] (Table 1b). For instance, in several murine stroke models, AnxA1 administration mediated protective effects via FPR2 after cerebral ischemia/reperfusion (I/R) injury [48] or more recently, after spontaneous intracerebral hemorrhage [49], limiting further cerebral microvascular dysfunction and tissue damage. In addition, binding of AnxA1-derived peptides to FPR2 on neutrophils was demonstrated to regulate neutrophil-platelet aggregation, which contributed to attenuate cerebral inflammation [50]. AnxA1 administration to low density lipoprotein receptor (LDLR)-deficient mice that were fed a western-type diet attenuated progression of atherosclerotic plaques and was accompanied by reduced FPR2-dependent neutrophil rolling and adhesion to endothelial cells [51]. Conversely, enhanced atherosclerotic lesion formation and arterial myeloid cell adhesion was observed in apolipoprotein E (apoE) −/− mice that lacked either AnxA1 or FPR2. Administration of the AnxA1-derived peptide Ac2-26 ameliorated this phenotype in the AnxA1 KO-mice but not in the FPR2 KO-mice, as judged by reduced atherosclerotic lesion size, lessened macrophage accumulation in lesions and a decreased FPR2-dependent recruitment of myeloid cells [32]. The protective capacity of AnxA1 was also documented in a model of restenosis, examining mechanical wire injury in high-fat diet (HFD)-fed AnxA1-deficient apoE −/− mice. In these animals, aggravated neointima development due to the accumulation and proliferation of macrophages was observed, suggesting that AnxA1 negatively regulates macrophage proliferation and its administration could serve to prevent restenosis after vascular damage [52].
Relevant for cardiac repair after myocardial infarction (MI), AnxA1 deficiency increased cardiac necrosis, inflammation, hypertrophy and fibrosis following MI and was accompanied by an impaired macrophage phenotype [53]. The reduced cardiac functionality in Anxa1 −/− mice after MI could be restored by AnxA1 administration, overcoming compromised release of proangiogenic mediator vascular endothelial growth factor (VEGF) from cardiac macrophages, and markedly improving neovascularization and car-diac repair [53]. AnxA1-based therapies improved cardiac outcomes after MI also in other studies [54] (Table 1b).

AnxA1-Derived Peptides and Anti-Inflammatory Response
As mentioned above, administration of full length AnxA1 greatly advanced efforts to explore the therapeutic potential of targeting FRP2 for anti-inflammatory strategies. Peptides of the N-terminal AnxA1 region, the most widely used being the Ac2-26 peptide, also mimic anti-inflammatory actions of full length AnxA1 [21][22][23]33,55,56]. The review articles listed here summarize a substantial amount of work demonstrating that exogenous administration of AnxA1 or Ac2-26 effectively limited or resolved inflammation in mouse models of stroke, myocardial ischemia, non-alcoholic steatohepatitis, rheumatoid arthritis, multiple sclerosis, colitis and asthma. AnxA1 deficiency was also detrimental in epithelial wound repair [11,57], lung fibrosis [58], obesity and insulin resistance [59], skin grafting [60], bacterial infection [61,62], as well as adrenal steroidogenesis in sepsis [63], or allergic conjunctivitis [64], and seems to have regulatory functions in early pregnancy [65] (Table 1a,b).
Recently, resolving AnxA1 functions were documented in a murine model of cardiotoxininduced tibialis anterior injury. In this model, which is characterized by necrotic tissue damage and extensive macrophage activity, extracellular AnxA1 was delivered to the injured tissue through neutrophil recruitment and overexpressing macrophages, triggering FPR2-mediated macrophage skewing towards a pro-reparative phenotype, dampening of inflammation and ultimately, regeneration of skeletal muscle fibers [66]. It has yet to be determined how this might relate to notexin-induced injury, which implicated AnxA1 deficiency to interfere with cell-cell fusion during myofiber regeneration [67]. A protective function for AnxA1 was also reported for mechanical injury-induced corneal scarring [68], acute colitis [69], and gout inflammation [70] (Table 1c).
Ac2-26 also showed anti-inflammatory and neuroprotective effects in diseases related to the central nervous system, such as pilocarpine-induced status epilepticus [71], kainic acidinduced temporal lobe epilepsy [72], or cerebral inflammation in sepsis [73]. However, Ac2-26 administration in a mouse model of Alzheimer disease did not exert any beneficial effects [74], possibly due to limitations in its ability to cross the blood-brain barrier (Table 1d).

AnxA1 and Diabetes
In the context of diabetic complications, initial studies reported that female Anxa1 −/− mice on the Balb/c background were more susceptible to weight gain and diet-induced insulin resistance, without showing significant changes in inflammation [59] (Table 1e). More recently, type 2 diabetic patients were found to exhibit increased AnxA1 plasma levels, which correlated with fatty liver index and elevated plasma cholesterol [34]. In these studies, a HFD in AnxA1 −/− mice on the C57BL/6 background led to a more diabetic phenotype. Treatment of HFD-fed wildtype mice with recombinant AnxA1 attenuated the development of insulin resistance [34]. AnxA1 plasma levels in human type 1 diabetics were similarly found to be elevated [75]. This coincided with a worse diabetic phenotype and severe cardiac and renal dysfunction of streptozotocin (STZ)-induced type 1 diabetes in AnxA1 −/− mice. Strikingly, AnxA1 administration attenuated cardiac and renal complications in diabetic AnxA1 −/− animals, and was associated with reduced activity of FPR2 signaling pathways [75]. In other studies, AnxA1 deficiency worsened pathological remodelling of the mesenteric vasculature in insulin-resistant, but not insulin-deficient animals [76], suggesting that AnxA1-based therapies could provide benefits for reducing vascular injury in diabetes. In support of this, a novel AnxA1-mimetic (compound 17b) provided vasoprotective effects in STZ-induced diabetic mice through the upregulation of vasodilating prostanoids [77].
Several other functions of AnxA1 may contribute to improve diabetic complications (Table 1e). For instance, exogenous AnxA1 enhanced glucose-stimulated insulin secretion of islet cells, which could be beneficial in islet transplantation strategies as a therapy for type 1 diabetes [47,78]. Furthermore, Ac2-26 application in diabetic wounds reduced neutrophil accumulation and facilitated M2 macrophage development, both anti-inflammatory and pro-repair effects required for wound closure [79]. On the other hand, it should be noted that when analyzing skin wound healing in AnxA1-deficient mice, wound inflammation, closure and formation of granulation tissue were not altered, and AnxA1 was proposed to possibly be dispensible for wound hemostasis and repair [80].

Nanotechnology-Based Approaches to Deliver AnxA1-Derived Peptides
As outlined above, the Ac2-26 peptide is the most widely used AnxA1-derived peptide to resolve inflammation. However, this peptide not only binds FPR2, but also FPR1 [23], and a novel peptide covering a larger region of the AnxA1 N-terminus (2-48) specifically bound FPR2 and reduced neutrophil recruitment during dermal inflammation [81]. Additionally, as oral delivery of peptides for therapeutic use are known to suffer from low bioavailability and metabolic liability, efforts to develop a more efficient and/or targeted delivery of AnxA1-derived peptides have been made (Table 1f). For example, Ac2-26 was packaged onto nanoparticles that target collagen type IV, which is highly enriched in atherosclerotic plaque. Advanced atheroslerotic lesions from mice receiving these Ac2-26 peptide-containing nanoparticles showed reduced lesion instability in a FPR2-dependent manner [82], indicating an efficient and localized delivery of the Ac2-26 peptide.
Similarly, delivery of oxidation-responsive and Ac2-26 containing nanoparticles to the inflamed colons of mice enabled site-specific release of the pro-resolving AnxA1mimetic peptide for the treatment of inflammatory bowel disease [83]. Oral delivery of these oxidation-labile Ac2-26 containing nanoparticles effectively reduced inflammation in the gastrointestinal tract, and improved intestinal wound healing [83]. The fact that neutrophil-derived microvesicles contain AnxA1, which bind to FPR2 and contribute to resolve cartilage protection in arthritic mice [84], also support nano-based therapeutic strategies as a promising tool to deliver AnxA1 or AnxA1-mimetic peptides.
Other approaches include the fusion of AnxA1-derived peptides with a cell-penetrating peptide to protect against neuronal apoptosis after cerebral ischemia [85], or the grafting of a small bioactive AnxA1 peptide into a stable cyclic peptide scaffold for the reduction of inflammation in a mouse model for acute colitis [86]. Alternatively, the development of a doxorubicin-DNA aptamer conjugate that specifically recognizes AnxA1 on the surface of cancer cells significantly enhanced targeted therapy against tumours in vivo [87]. Likewise, coupling anticancer drugs to a peptide that recognizes AnxA1 on the tumour vasculature surface enabled drug delivery across the blood-brain barrier to suppress growth of brain tumours [88]. Finally, one report described the utilization of a monoclonal AnxA1 antibody to suppress the autoimmune syndrome associated with an abnormal AnxA1 expression on activated B and T cells in a mouse model of systemic lupus erythematosus (SLE), thus underscoring the need for a balanced expression of this immune modulator [89].

AnxA1 and Cancer
Over the last few years, a lot of studies have explored a variety of therapeutic roles for AnxA1 in cancer (Table 1g). For instance, chemotherapeutic agents such as anthracyclines and cyclophosphamide can trigger an anticancer immune response, with long-term beneficial effects beyond the initial chemotherapy. The underlying mechanism appears to be based on the release of AnxA1 from cancer cells upon chemotherapy. AnxA1 then binds to FPR1 on dendritic cells, which phagocytose and kill malignant cells, but also present tumour-associated antigens, which are then recognized and eliminated by cytotoxic T lymphocytes [90,91]. In support of this model, anticancer immunity after chemotherapy with anthracyclines was compromised in mice with tumours lacking AnxA1, or when the dendritic host cells lacked FPR1. This regulatory circuit may be tumour-type specific and relevant in breast, colorectal, lung, and kidney, but not gastric cancers. Other limitations to boost anticancer immunity in cancer patients via AnxA1 administration include the high prevalence of a loss-of-function FPR1 allele common in all ethnic groups, which is associated with poor prognosis, at least in breast cancer patients, after anthracycline-based chemotherapy [90,92].
In contrast, scenarios driven by the binding of AnxA1 to FPRs have associated high AnxA1 levels with poor patient survival and progression in metastatic and triple-negative breast cancers (TNBC) [93]. Along these lines, AnxA1-deficient mice displayed reduced breast cancer growth and enhanced survival, and it was proposed that AnxA1 binding to FPR2 on macrophages, influencing their polarization and infiltration of solid tumours, would be the underlying cause for enhanced tumour progression [94]. Alternatively, increased binding of AnxA1 released from cancer cells to FPR2 on regulatory T cells may impede anti-tumour immune responses [95]. Adding further complexity, secretion of AnxA1 from TNBC cells might induce autocrine signalling via FPR1 to further increase breast cancer cell aggressiveness and survival and could be blocked with cyclosporin A, a FPR1 inhibitor [96].
The de-regulated signaling pathways in cancers triggered by extra-or intracellular AnxA1 activities have yet to be fully elucidated [90][91][92][93][94][95][96]. Phosphoproteomics from AnxA1deficient mammary gland epithelial cells identified up-and downregulation of several signaling pathways that modulate cell motility and could contribute to breast cancer initiation [97]. Mechanistically, AnxA1 was shown to control epidermal growth factor receptor (EGFR) receptor signaling and trafficking [98], which is probably modified through AnxA1 phosphorylation [99]. In pancreatic cancer, AnxA1 depletion conferred a less aggressive phenotype and reduced liver metastasis independent of FPR pathways [100]. In a mouse model for multiple myeloma, AnxA1 depletion improved bortezomib treatment [101], but in nasopharyngeal carcinoma xenografts, loss of AnxA1 enhanced radiotherapy resistance [102] (Table 1g).

Intracellular AnxA1 Activities
Besides the AnxA1/FPR axis, the AnxA1 KO-strain also revealed intracellular AnxA1 activities to modulate inflammation and other disease-related activities (Table 1h). For instance, Anxa1 −/− macrophages displayed an altered activity of signaling pathways and the NOD-, LRR-and pyrin domain-containing protein 3 (NLRP3) inflammasome, both affecting inflammatory cytokine production, such as interferon-β and tumour necrosis factor α (TNFα), but also release of lipid mediators [103][104][105][106]. Defects in the secretory pathway of mesenchymal stromal cells from the AnxA1 KO-mice might explain the inability of the secretome from these cells to stimulate glucose-induced insulin secretion of pancreatic β cells [47,78].
In addition, nuclear translocation of AnxA1 [107], as well as AnxA1 affecting transcriptional activation, mRNA transport and stability has been described [103,104]. Intracellular AnxA1 activities may also explain the upregulation of cyclooxygenase (COX) and cytoplasmic phospholipase A2 (cPLA 2 ) observed in Anxa1 −/− fibroblasts, probably driving enhanced and glucocorticoid-insensitive eicosanoid production in the Anxa1 −/− animals [24,108]. Anxa1 −/− fibroblasts also revealed the critical role of AnxA1 in the biogenesis of a subpopulation of internal vesicles in multivesicular bodies (MVBs) [109]. In these studies, AnxA1 was transported on to intraluminal MVB vesicles containing EGFR and was tyrosinephosphorylated upon EGFR activation. EGF-stimulated inward vesiculation was abolished in Anxa1 −/− fibroblasts, whereas basal inward vesiculation remained intact. Hence, physical association and tyrosine phosphorylation of AnxA1 for EGF-stimulated and EGFR-containing intraluminal vesicle formation within MVBs appears critical [109]. In addition, AnxA1 contributes to the establishment of membrane contact sites, which is likely relevant for trafficking of receptors and transfer of ions and lipids such as cholesterol, between endosomal and other compartments [110]. Additionally, Anxa1 −/− mice are partially protected against influenza A virus infection, involving reduced uptake and exit of internalized virus from late endosomes/MVBs [111]. Yet, this protective effect of AnxA1 against viral infection also requires the AnxA1/FPR2 signaling axis to increase alveolar macrophages, which are decisive factors in the host defence against pathogens [112] (Table 1h).

AnxA2
In mice, AnxA2 is highly expressed in the lung, pancreas, colon, ileum and adrenal tissues, while low AnxA2 levels are found in spleen, testis, kidney and liver [114]. In endoand epithelial cells, monocytes, macrophages, and many other cell types, AnxA2 regulates microdomain formation and membrane repair at the plasma membrane, trafficking along endo-and exocytic pathways, as well as RNA export from the nucleus [115,116]. Multiple membrane-related AnxA2 activities involve interactions with the actin cytoskeleton and phosphatidylinositol-4,5-bisphosphate, contributing to cell growth, differentiation, apoptosis and migration [1,5,6,116].
Notably, AnxA2 forms a heterotetramer with p11 (S100A10), a member of the S100 protein family, and upon Src kinase-mediated phosphorylation of AnxA2 at tyrosine 23, the intracellular AnxA2/p11 complex is translocated to the outer membrane leaflet. At the cell surface, the heterotetrameric AnxA2/p11 complex acts as a scaffold that modulates cell surface presentation of receptors and ion channels [117,118]. Indeed, the critical role of the AnxA2/p11 complex for the cell surface presentation of serotonine receptors has been validated in AnxA1-, p11-KO and other mouse models of depression [115,119,120].

AnxA2 and Vascular Homeostasis and Angiogenesis
On endothelial cells, extracellular AnxA2/p11 serves as a docking site for plasminogen and tissue plasminogen activator (tPA), which promote plasmin generation and vascular fibrinolysis. The AnxA2/p11-mediated plasminogen activation has been extensively discussed and we recommend several reviews for further reading [115,116,[119][120][121].
AnxA2-deficient mice [122] (Table 2a) provided the first in vivo evidence that AnxA2 mediates fibrin clearance, as the AnxA2 −/− mice showed substantial deposition of fibrin and after induction of acute carotic artery thrombosis, displayed increased thrombosis. Mechanistically, Src kinase and protein kinase C (PKC) were required for AnxA2/p11 translocation to the cell surface in the carotid artery injury model [123]. Further supporting a connected role for AnxA2 and p11 in fibrinolysis, de-regulated fibrinolysis was also observed in the p11 KO-mice [124]. This coincides with reduced p11 levels in the Anxa2 −/− mice, most likely due to increased p11 ubiquitination and degradation in these animals [125].
In other studies, Anxa2 −/− mice revealed AnxA2-mediated functions in vascular homeostasis and angiogenesis, such as the formation of new blood vessels [115,122,126]. In a mouse model for diet-induced hyperhomocysteinemia, which is implicated in thrombotic and atherosclerotic disease, fibrin accumulation and defects in neoangiogenesis were similar to those observed in the Anxa2 −/− mice [127]. Interestingly, AnxA2 isolated from mice fed a high methionine diet failed to bind tPA and activate plasminogen and could only be restored upon administration of exogenous AnxA2 [127]. In ischemic cerebral disease, recombinant AnxA2 amplified tPA-mediated thrombolysis to prevent stroke [128]. Yet, al-though inhibition of plasmin generation has therapeutic potential in atherosclerosis, AnxA2 deficiency in apoE −/− mice did not reduce lesion development [115]. The latter observation is challenged by AnxA2 deficiency suppressing atherogenic integrin α5 signalling caused by oscillary shear stress, which reduced the development of atherosclerosis in apoE −/− mice at regions with disturbed blood flow [129]. Recently, recombinant AnxA2 ameliorated fibrinolytic dysfunction in a mouse model lacking the exchange protein directly activated by cAMP (EPAC), which is required for fibrinolysis on the surface of vascular endothelial cells [130] (Table 2a).

AnxA2 and Pulmonary Vasculature
In the pulmonary vasculature, endothelial cells from Anxa2 −/− mice display reduced barrier function and develop pulmonary edema and neutrophil infiltration in the lung parenchyma in response to alveolar hypoxia. Pulmonary vascular integrity does not seem to require the AnxA1/p11 tetramer, but AnxA2 complex formation with vascular endothelial cadherin and its tyrosine phosphatases, to prevent vascular leak during alveolar hypoxia [131] (Table 2b). In addition, AnxA2 can contribute to lung injury and fibrosis by augmenting factor Xa fibrogenic activity [132]. Interestingly, within the pulmonary vasculature, AnxA2 has now been found responsible for the severe side effects caused by bleomycin, a clinically potent anticancer drug for a large variety of cancers. Bleomycin binds AnxA2, which impedes transcription factor EB mediated autophagy, leading to the induction of pulmonary fibrosis. Underlying mechanism involve bleomycin-induced NF-kappaB (NFκB) inflammatory pathways leading to AnxA2 secretion. Strikingly, these toxic side effects can be overcome by an AnxA2 peptide inhibitor (TM601) or AnxA2 deficiency in mice, alleviating bleomycin-induced pulmonary inflammation and fibrosis [132][133][134].
In addition, a critical in vivo function of AnxA2 in the secretory pathway affects pulmonary function (Table 2b). Anxa2 −/− displayed reduced exercise tolerance and impaired lung tissue elasticity that resembles the phenotype of collagen VI (COL6) deficiency, which anchors basement membranes to other collagen fibers [135]. Indeed, the Anxa2 −/− lung basement membrane lacked COL6, which accumulated and mislocalized in the Golgi apparatus [135]. This phenotype of the Anxa2 −/− mice correlated with the association of AnxA2 with COL6, but also two SNARE (soluble NSF attachment protein receptor) proteins, synaptosomal-associated protein 23 (SNAP23) and vesicle-associated membrane protein 2 (VAMP2), in secretory vesicles from bronchial epithelial cells (Table 2b). This is in line with previous reports identifying AnxA2 to regulate exocytosis via several SNAREs, including SNAP23 and VAMP2 [136,137]. This secretory function of AnxA2 may extend to other disorders associated with COL6 deficiency [135] or exocytic events driving cancer progression [138][139][140].

Other AnxA2-Related Aspects in Vascular Homeostasis
Other AnxA2-related aspects in vascular homeostasis and angiogenesis include the high AnxA2 antibody titers associated with thrombotic complications in cerebral venous thrombosis [115,119,120] as well as the antiphospholipid syndrome [141,142]. AnxA2 upregulation also correlated with hyperfibrinolysis in acute promyelocytic leukemia [119,120]. Administration of recombinant AnxA2 increased neovascularization and inflammation in mice with collagen-induced arthritis via binding to an AnxA2 receptor described previously [143] and subsequent activation of the Hedgehog pathway [144]. On the other hand, lentiviral AnxA2 receptor overexpression suppressed neovascularization in the mouse aortic ring and matrigel plug assays [145]. The role of AnxA2 in vascular homeostasis extends to endothelial permeability and integrity at the blood-brain barrier after cerebrovascular injury, with AnxA2-deficient mice displaying increased inflammation and reduced expression of cerebral endothelial junctional proteins [146,147]. For further reading of AnxA2-related aspects in vascular homeostasis and angiogenesis we recommend several excellent reviews [115,116,[119][120][121]142] (Table 2c).

AnxA2 and Lipid-Related Disorders
Several studies investigated AnxA2 in mouse models for lipid-related disorders (Table 2d). Possibly relevant for hypercholesterolemic settings, AnxA2 or the AnxA2/p11 complex inhibited proprotein convertase subtilisin/kexin-9 (PCSK9)-mediated LDLR downregulation, leading to elevated LDLR levels at the cell surface [114,148]. Anxa2 −/− mice displayed elevated LDL-cholesterol and circulating PCSK9 levels, while LDLR levels were reduced in extrahepatic tissues. Vice versa, adenoviral AnxA2 overexpression elevated hepatic LDLR levels in vivo [114]. The association of LDL-cholesterol levels with Anxa2 gene variants [149] support modulation of AnxA2 levels as a potential cholesterol-lowering strategy. Applicable for obesity and the metabolic syndrome, AnxA2 depletion attenuated obesity-induced insulin resistance through suppression of NFκB signaling [150]. Furthermore, in vascular endothelial cells of white adipose tissue (WAT), AnxA2 forms a complex with prohibitin and the fatty acid transporter CD36. Anxa2 −/− mice exhibited defective fatty acid uptake by the WAT endothelium and adipocytes, making this regulatory circuit a new target for metabolic intervention in obesity [151] (Table 2d).

AnxA2 and Cancer
A substantial number of studies in recent years examined AnxA2 in cancer growth and progression in vivo, identifying diverse AnxA2 functions in tumourigenic settings (Table 2e).
In mouse models for pancreatic ductal adenocarcinoma (PDAC), AnxA2 KO-animals displayed a substantial reduction in PDAC tumour invasion and metastasis. Src-dependent AnxA2 tyrosine 23 phosphorylation and cell surface translocation, as well as AnxA2 regulating the secretion and autocrine signaling of semaphorin 3D (SEMA3D) contribute to PDAC invasion and metastasis [138]. The reduced SEMA3D secretion upon loss of AnxA2 also lowered the incidence of perineural invasion and metastasis of orthotopic pancreatic tumours in mice [152]. PDAC metastatic behaviour is strongly influenced by stromal cells in the tumour microenvironment and pharmacological inhibition of signaling events initiated by factors released from stroma cells, including hepatocyte growth factor, insulin-like growth factor 1 or tenascin C, reduced AnxA2 phosphorylation and metastatic potential of PDAC tumours [139,140,153]. In fact, stromal AnxA2 expression has potential as a predictive biomarker for survival outcomes in PDAC [154]. The therapeutic potential of targeting AnxA2 in PDAC was further emphasized with the recent development of an AnxA2-targeting vaccine approach, which in combination with immune checkpoint inhibitors, led to survival benefits in PDAC mouse models [155]. Additional evidence for roles of AnxA2 in the tumour microenvironment came from the inoculation of prostate cancer cells with bone marrow stromal cells from the AnxA2 KO-mice. This approach revealed that AnxA2 binding to stromal-derived factor 1 strongly influenced recruitment, growth and survival of prostate cancer cells in the bone marrow microenvironment [156] ( Table 2e).

Therapeutic Potential of AnxA2 in Cancer
Several animal studies proposed multiple roles for AnxA2 in brain cancers. In pediatric neuroblastoma, AnxA2 depletion attenuated NFκB activity and enhanced anticancer drug sensitivity, which improved chemotherapy in a xenograft model for neuroblastoma [157]. Likewise, AnxA2-mediated NFκB signaling contributed to tumourigenicity in xenograft models for glioblastoma multiforma (GBM) [158], with AnxA2 depletion strongly inhibiting GBM tumour growth and improving survival [159]. In the latter studies, AnxA2-mediated phosphorylation of signal transducer and activator of transcription 3 (STAT3) increased expression of microRNA miR155 to promote epithelial-to-mesenchymal transition (EMT) in glioma [159]. Other downstream targets of the AnxA2-STAT3 axis to promote growth and phenotypic transition in GBM include the oncostatin M receptor [160] and the cyclin D1 pathway [161] (Table 2e).
In breast cancer, intra-and extracellular AnxA2 activities and locations, including exosomes [162], have become promising therapeutic targets. AnxA2 antibody-conjugated and curcumin-loaded nanoparticles effectively accumulated in tumours, providing sustained release of curcumin with potential to reduce metastatic breast cancer progression [163]. A monoclonal AnxA2 antibody not only allowed monitoring EMT in breast and ovarian cancers, but also provided antibody-dependent cell toxicity and efficient killing when conjugated to cytotoxic drugs and expressed as a chimeric immunoglobulin G1 [164][165][166]. Other beneficial effects of AnxA2 antibodies include the inhibition of neoangiogenesis in TNBC xenograft models, which involves reduced plasmin generation as well as loss of AnxA2 tyrosine 23 phosphorylation [167,168]. The latter requires interaction with Src and Rack1 kinases, and could be responsible for the acquirement of drug resistance in aggressive breast cancers [169]. In support of this, AnxA2 depletion improved the suppressive xenograft growth effect of Src kinase and VEGF inhibitors in esophageal cancers [170].
In fact, AnxA2 contributes to the development of resistance to cisplatin and tyrosine kinase inhibitors targeting EGFR in non-small cell lung cancers [171,172]. Vice versa, disruption of AnxA2/p11 interaction with small molecule inhibitors or AnxA2 antibodies reduced acute lymphoblastic leukemia proliferation and sensitized tumour cells to chemotherapy [173].
Although data from p11 KO-mice often supports findings that AnxA2/p11-dependent plasmin generation is involved in tumour progression [115,116,121], it should be noted that p11 has plenty of other tumourigenic functions that appear rather unrelated to AnxA2 [174,175]. It would also go beyond the scope of this review to discuss in detail cancer-related animal studies that identified upstream regulators of AnxA2 expression or activity [176][177][178]. However, other studies, for example in glioma, suggest that not only tetrameric, but also monomeric AnxA2 might serve as a strategy to develop it as a vaccine adjuvant to increase anti-tumour immunity [179] (Table 2e). Alternatively, a novel RNA nanoparticle harboring an AnxA2 aptamer was developed to target ovarian cancer for doxorubicin delivery [180]. Similary, a DNA aptamer that binds AnxA2 on cancer cells could become a tool to diagnose and treat multiple myeloma [181]. Along these lines, a cyclic octapeptide labeled with a near-infrared dye selectively binding phosphorylated AnxA2, which is highly expressed in a wide range of solid tumours, provides opportunity for tumour imaging and localized drug delivery [182]. Alternatively, treatment of tumour cells with a cytosol-targeting small peptide that binds to intracellular AnxA2 showed a reduced capacity to colonize lungs in several mouse models [183] ( Table 2e).

AnxA2 and Infection
Several recent studies addressed roles for AnxA2 in response to bacterial and viral infections, some of those identifying roles for AnxA2 in membrane organization and endocytic transport (Table 2f). For instance, pathogen recognition and activation of innate immunity in macrophages requires Toll-like receptor 4 (TLR4) activation at the plasma membrane and on endosomes. Remarkably, in Anxa2 −/− mice enhanced TLR4 signaling and reduced TLR4 endocytosis was associated with an increased susceptibility to bacteriainduced pulmonary inflammation [184].
During fungal infection (Cryptococcus) of macrophages, AnxA2 controls phagocytosis and non-lytic exocytosis of the pathogen. Compromised fungal uptake, propagation and release, together with an enhanced inflammatory response, probably explains that Anxa2 −/− mice died more rapidly than control animals after fungal infection [185].
In studies addressing autophagy, vesicles isolated from dendritic cells of Anxa2 −/− mice contained reduced amounts of phosphatidylserine (PS) and phosphatidylinositides. These vesicles generate phagosomes for encapsulation of organelles and biomolecules, followed by their degradation in endo-/lysosomes. Phagosome biogenesis and maturation in dendritic cells from the Anxa2 −/− mice was strongly reduced [186]. Autophagy can also serve as a defence mechanisms against bacterial infection, and was de-regulated in a mouse infection model using Anxa2 −/− animals [187].
In a mouse model for Alzheimer's disease, AnxA2 interacts with phosphorylated presenilin 1 and the SNARE protein VAMP8, the latter binding to the autophagosomal SNARE syntaxin 17. This ultimately facilitates fusion of autophagosomes with lysosomes, thereby increasing b-amyloid degradation, which has therapeutic potential for Alzheimer's disease [188].
Tick saliva contains effector molecules that inhibit host immunity and facilitate pathogen transmission. One of the tick proteins, sialostatin L2, binds AnxA2, which then impairs the formation of the NLRC4 inflammasome, responsible for the maturation and secretion of pro-inflammatory cytokines. Accordingly, AnxA2-deficient mice were more succeptible to Anaplasma phagocytophilum infection, leading to splenomegaly, thrombo-and monocytopenia [189]. On the other hand, during the pathophysiology of sepsis, Anxa2 −/− mice showed increased production of pro-inflammatory IL-17 and reactive oxygen species, increased neutrophil infiltration, but reduced bacterial clearance and animal survival [190]. While these studies rather implicate intracellular AnxA2 functions, during the adhesion and invasion of uropathogenic Escherichia coli to bladder epithelial cells, AnxA2 was identified as the receptor of the bacterial protein YadC, which participates in binding to bladder epithelial cells [191]. Inhibition of AnxA2 using the Ca 2+ -chelator BAPTA-AM or compound A-05, which inhibits the AnxA2/p11 complex [192], attenuated E.coli infection in vivo [191]. Likewise, AnxA2 on the surface of vascular endothelial cells served as a docking site for Rickettsia, with AnxA2 deficiency blocking Rickettsia adherence to the luminal surface of blood vessels in vivo [193] (Table 2f).     Poor sarcolemma repair, progressive and age-dependent decline in muscle function, lack of adipogenic replacement of myofibers Similar to dysferlin deficiency but without chronic inflammation; AnxA2 target for dysferlinopathy [194,195] Limb girdle muscular dystrophy (LGMD) 2B AnxA2 administration in mouse LGMD 2B model Enhanced muscle loss [195] (h) AnxA2 and Other Biological Activities AnxA2 KO-mice Impaired embryo implantation and placentation Impaired decidualization of endometrial stromal cells and uterine environment; AnxA2 acts as adhesion molecule during implantation [196,197] Chronic pain (tissue damage, harmful chemicals) AnxA2 KO-mice Enhanced TRPA1-dependent pain AnxA2 is a ligand for TRPA1 responsible for ion channel regulation [198] Abbreviations

AnxA2 and Muscle Repair
Two recent studies linked AnxA2 with myofiber repair and inflammation in muscle [194,195] (Table 2g). Muscle repair after sarcolemmal damage requires dysferlin, and similar to dysferlin deficiency, lack of AnxA2 in myofibers was associated with poor sarcolemma repair and a progressive age-dependent decline in muscle function. Yet, AnxA2 deficiency did not cause chronic inflammation that is commonly observed upon loss of dysferlin. Moreover, AnxA2-depleted mice showed extensive myofiber regeneration after laser injury, but not degeneration or adipogenic replacement, making AnxA2 a novel target to treat dysferlinopathy [194]. Indeed, lack of AnxA2 prevented adipogenic replacement of myofibers, while exogenous AnxA2 enhanced muscle loss in mouse models for limb girdle muscular dystrophy 2B, which is caused by dysferlin mutations [195].

Other AnxA2-Related Biological Activities
Although the underlying mechanism remain to be fully clarified, roles for AnxA2 in actin polymerization and secretion might be involved in preeclampsia, a pregnancy-specific disorder, as AnxA2 deficiency impaired decidualization of endometrial stromal cells and the uterine microenvironment that promotes embryo implantation and placentation [196]. Alternatively, AnxA2 might act as an adhesion molecule at the endometrial epithelium during embryo implantation [197] (Table 2h).
Finally, in vivo evidence that links AnxA2 with ion channel regulation [1,2] includes AnxA2 being a ligand for the transient receptor potential ion channel A1 (TRPA1), which is responsible for the detection of harmful chemicals, tissue damage and chronic pain. Anxa2 −/− mice displayed increased surface levels of TRPA1 on sensory neurons, which correlated with enhanced TRPA1-dependent pain in these animals [198].

AnxA3 and Cancer
Several studies examined AnxA3 loss-or gain-of-function in xenograft models, implicating AnxA3 depletion as a tool to improve anticancer drug sensitivity (Table 3). For instance, downregulated AnxA3 levels improved cisplatin sensitivity in ovarian cancer [205]. Similarly, AnxA3 silencing in MDA-MB-231 breast cancer cells strongly reduced lung metastasis with increased sensitivity towards doxorubicin, possibly through NFκB inhibition [200]. Others reported AnxA3 depletion in MDA-MB-231 cells to strongly reduce xenograft growth and angiogenesis in the tumour microenvironment [206]. AnxA3 silencing using miR382, which was accompanied by phosphoinositide-3-kinase/(PI3K)/protein kinase B (Akt) downregulation, reduced EMT and lymph node metastasis of pancreatic cancer cells [207]. In A549 lung adenocarcinoma cells, AnxA3 deficiency reduced MAPK signaling and decreased expression of metalloproteases and E-/N-cadherin, thereby reducing xenograft growth, as well as metastasis to lung, liver and brain after tail vein injection [208]. In lung cancer, cancer associated fibroblasts (CAFs) increased AnxA3 levels in the neighbouring cancer cells, which correlated with increased cisplatin resistance. In contrast, AnxA3 knockdown in this model increased cisplatin sensitivity, an observation that may add to overcome drug resistance in lung cancer patients [209]. In other studies, elevated AnxA3 levels suppressed PKCδ/p38 MAPK-dependent onset of apoptosis and autophagy [210] and might be a target for immunotherapy in hepatocellular carcinoma [201]. Along these lines, AnxA3 monoclonal antibodies sensitized hepatocellular carcinoma to the tyrosine kinase inhibitor sorafenib [210]. Finally, AnxA3 depletion in microglia alleviated bone cancer-induced pain through inhibition of hypoxia-inducible factor-1a/VEGF-dependent signaling [211] (Table 3).

AnxA4
AnxA4 is expressed prominently in secretory epithelia of the lung, intestine, stomach and kidney [212]. Out of the three AnxA4 mRNAs (AnxA4a-c), AnxA4a is broadly distributed, while AnxA4b and AnxA4c are only found in the digestive track and solitary chemosensory cells, respectively [213]. After Ca 2+ elevation, AnxA4 translocates to the plasma membrane or the nuclear membrane, but also has the ability to oligomerize and participate in Ca 2+ -dependent aggregation of vesicles or modulate the mobility of membrane-associated proteins [214][215][216]. While the latter may implicate a role in repair mechanisms, administration of pharmacological AnxA4 concentrations caused bleeding in skin wound repair [80]. AnxA4a-deficient urothelium did not reveal alterations in bladder function [217], but airway epithelial cell migration was impaired upon AnxA4 depletion, with consequences for airway branching morphogenesis during lung development [218].

AnxA4 in Cancer
AnxA4 is upregulated in various epithelial cancers, and AnxA4 knockdown reduced xenograft growth of gallbladder cancer cells, possibly by downregulating oncogenic NFκB signaling pathways [221]. Likewise, AnxA4 depletion decelerated the in vivo growth of tumours derived from basal-like breast cancer cell lines. In this model, AnxA4 interacted with another annexin, AnxA1, to activate Janus kinase-STAT3 signaling [222].
Several studies examined AnxA4 in the context of anticancer drug resistance (Table 4). In xenografts from endometrial carcinoma cells, AnxA4 overexpression conferred cisplatin resistance, possibly by interacting with the copper transporter ATP7A, and promoting its ability to efflux platinum drugs [223]. Vice versa, AnxA4 depletion increased sensitivity to platinum-based drugs in vivo in a Ca 2+ -dependent manner [224]. Finally, chemoresistance of lung cancer cells correlated with AnxA4 translocation to the plasma membrane. Blocking AnxA4 translocation with a heptapeptide derived from the tumour suppressor FHIT restored the chemosensitivity to paclitaxel [225].

AnxA5
AnxA5 is the most abundant annexin in almost all cells and tissues, except neurons [1,226] ( Table 5). In cells, AnxA5 is associated with the plasma membrane, nucleus, Golgi, endoplasmic reticulum, late endosomes/lysosomes, phagosomes, and mitochondria, often in a Ca 2+ -dependent manner and linked to the regulation of membrane transport, Ca 2+ signaling, ion channels, Ca 2+ -influx, cell cycle, apoptosis and phagocytosis [227][228][229][230][231][232][233]. In addition, substantial amounts of AnxA5 are found extracellularly, and due to its strong binding affinity towards PS in the outer leaflet of the plasma membrane, AnxA5 is now a widely used diagnostic tool to detect apoptotic cells in various disease settings [226] (Table 5b-e and details below).

AnxA5 and Differentiation
Given the abundant AnxA5 levels in many tissues, several studies utilized the AnxA5 KO-mice to address possible roles in development. Indeed, the targeted disruption and replacement with an Anxa5-lacZ fusion gene enabled detailed AnxA5 expression analysis during embryonal mouse development, identifying Anxa5-lacZ expression specifically in populations of perivascular cells of non-skeletal mouse tissues, with critical roles in initial steps of angiogenesis differentiation and maturation of endothelial cells by pericytes [234,235]. However, and despite high AnxA5 levels in cartilaginous tissues and bone, the AnxA5 KO-mouse lacked any defects in the calcification process during skeletal development [236] (Table 5). This led to speculations that other annexins, in particular AnxA6, could compensate for AnxA5 deficiency [236,237]. Yet, double KO-mice deficient of AnxA5 and AnxA6 (Anxa5 −/− /Anxa6 −/− ) or mice lacking both AnxA5 and AnxA6 as well as collagen X showed normal skeletal development [238,239], indicating these annexins not being essential for the calcification of the growth plate cartilage (Table 5a). Nevertheless, more recently, AnxA5 deficiency caused increased bone overgrowth at the enthesis, and it was concluded that AnxA5 prevented bone overgrowth triggered by mechanical forces and modulated expression of mineralization regulators [240].
Likewise, although being the most abundant membrane-associated protein in stereocilia, AnxA5 deficiency was dispensible for hair-bundle development and function [241] (Table 5a).

AnxA5 and Anticoagulation
AnxA5 is most abundant in the placenta and associated with thrombophilia-related complications in pregnancy [242]. Although AnxA5 deficiency was initially not considered to interfere with embryonal development, in later studies reduced litter size, increased foetal loss and placental thrombosis of maternally homozygous progeny were reported [243]. Yet, Anxa5 −/− mice exhibited regular estrous cycle and ovulation numbers [243], although altered pituitary hormone expression patterns were observed [244,245]. Interestingly, administration of heparin to prevent thrombi formation in the placental circulation reduced pregnancy loss in the AnxA5 KO-mice (Table 5b). Alternatively, administration of Zinc 2+ restored anticoagulation, upregulated AnxA5 levels in the placenta, and showed a trend to increase litter size in Anxa5 −/− mice [246]. Moreover, fusion of AnxA5, with strong affinity for outer leaflet PS, with echistatin, which binds to the integrin α IIb β 3 receptor on platelets, showed potential as antithrombotic reagent that specifically targets platelets in thrombogenesis sites [247]. Based on the property of AnxA5 to recognize PS on activated platelets, AnxA5-conjugated micelles have been developed to enable targeted delivery of drugs that promote thrombolysis [248] to alleviate the risk of hemorrhage that occurs after administration of thrombolytic agents.
In addition, the anticoagulant AnxA5 activity has implications for other diseases, as administration of wildtype AnxA5, but not an AnxA5 mutant version unable to bind PS, could overcome dysfunctional coagulation and inhibited thrombin formation in septic animals [249]. Mechanistically, the ability of AnxA5 to form two-dimensional arrays on PS-rich membranes in a Ca 2+ -dependent manner appears critical [242,250]. However, skin wound inflammation and closure were not altered in Anxa5 −/− mice [80]. In fact, increased levels of AnxA5 in the alveolus may play a detrimental role in alveolar epithelial wound healing, promoting lung inflammation and fibrosis [251]. Nevertheless, AnxA5-based approaches could become a suitable tool to overcome poor obstetric outcomes caused by antiphospholipid syndrome, preeclampsia and SLE in humans [252] (Table 5b,c).

AnxA5 and Cardiovascular Disease
The ability of AnxA5 to bind cell surface PS on apoptotic cells has proven valuable for the diagnosis of atherosclerosis [226] (Table 5d). However, a comparison of radiolabeled ( 99m TC) AnxA5 with duramycin in HFD-fed apoE −/− KO-mice identified 99m TC-duramycin to be more suitable for the quantification of vulnerable atherosclerotic plaque [253]. These technical limitations might have contributed to the inability to detect any apoptotic changes in the aortic arch of apoE −/− mice treated with low dose radiation, which is considered to protect against atherosclerosis, when using 99m TC-labeled AnxA5 for autoradiographic analysis [254]. On the other hand, AnxA5-conjugated gold nanoparticles helped to better target localized apoptotic macrophages and diagnose vulnerable atherosclerotic plaque in the HFD-fed apoE −/− mouse model [255].
The therapeutic potential of recombinant AnxA5 binding to cell surface PS on apoptotic cells and thereby providing anticoagulant, but also anti-apoptotic and anti-inflammatory activities was shown in several studies (Table 5d). In apoE −/− mice fed a western style diet, subsequent AnxA5 administration reduced recruitment and activation of monocytes to the inflamed lesion site, thereby reducing plaque inflammation [256]. Moreover, when AnxA5 was administered simultaneously with the western style diet, reduced apoptotic rates contributed to the significantly reduced plaque size [257]. Using hypercholesterolemic ApoE*3 Leiden mice as a model for MI-reperfusion injury, AnxA5 treatment strongly reduced infarct size and improved cardiac function, most likely via the attenuation of the post-ischaemic inflammatory response [258]. Moreover, administration of AnxA5 fused to stromal-derived factor 1, a cytokine that protects the heart from ischaemic injury, accumulated at the myocardium, probably via binding to PS on apoptotic cells, and provided cardioprotection, as judged by attenuated apoptosis, enhanced angiogenesis, reduced infarcted size and improved cardiac function after MI [259]. Finally, AnxA5 administration improved steatosis, inflammation and fibrosis in nonalcoholic steatohepatitis [260]. Interestingly, the underlying mechanism in the latter studies might not involve AnxA5 binding to PS, but AnxA5 interacting with pyruvate kinase 2, leading to metabolic reprogramming from glycolysis to oxidative phosphorylation in macrophages [260] (Table 5d).

AnxA5 and Cancer
The ability of AnxA5 to recognize PS exposed by apoptotic cells has been very valuable in cancer-related studies (Table 5e). It would go beyond the scope of this review to list all studies that utilized AnxA5 as a read-out to monitor apoptosis, and only research that specifically examined the contribution of AnxA5 to oncogenesis is summarized here.
In hepatocarcinoma, AnxA5 overexpression promoted in vivo tumour malignancy and lymphatic metastasis [261], most likely through MAPK and Rac1 signaling path-ways [262]. Likewise, AnxA5 upregulation promoted tumourgenicity of glioma cells in nude mice [263]. On the other hand, AnxA5 knockdown impeded tumourigenesis in subcutaneous xenografts from renal cell carcinoma [264], with the latter two studies suggesting an involvement of PI3K/Akt signaling and implicating AnxA5 as a potential biomarker in glioma and renal cell carcinoma. In non-small cell lung cancer, drug resistance to first-line EGFR tyrosine kinase inhibitors is common and AnxA5 depletion restored gefitinib sensitivity by promoting G2/M cell cycle arrest, providing opportunity to overcome this major clinical challenge [265].
In contradiction to the abovementioned tumour models, AnxA5 high-affinity binding to PS also inhibited engulfment of apoptotic cells by macrophages, which increased the immunogenicity of tumour cells undergoing apoptosis. Indeed, treatment of melanoma xenografts with AnxA5 reduced tumour growth, increased necrosis in tumour tissues and most strikingly, inhibited angiogenesis by downregulating VEGF production [266]. Hence, targeted delivery of AnxA5 could be utilized to selectively inhibit tumour growth and progression. Indeed, intra-tumoural release of AnxA5 from intravenously injected nanoparticles effectively blocked phagocytosis of apoptotic cells, which then triggered a strong cytotoxic T cell response and immunological memory that led to complete tumour regression in 50% of mice with orthotopic breast tumours [267]. Likewise, AnxA5 administration lessened the immunosuppressive properties of the tumour microenvironment generated by cisplatin treatment and enhanced the immunogenicity and anti-tumour efficacy in a preclinical tumour model with a defined tumour-specific antigen (TC-1 tumour model) [268,269]. Along these lines, fusion of AnxA5 with the peptide-major histocompatibility complex strongly augmented lymphocyte response, which could serve to improve immune modulation for cancer therapy [270]. In a mouse model for ovarian cancer, the delivery of AnxA5 fused to an enzyme that converts selenomethionine to toxic methylselenol, in combination with anti-CD73 and anti-Ox40 as immunostimulants, strongly decreased tumour burden in mice with orthotopic metastatic ovarian cancers [271] (Table 5f).

AnxA5 and Imaging
As molecular imaging techniques using radiolabeled AnxA5 provided opportunity to more rapidly assess apoptosis in disease or upon therapy, efforts over the last few years improved the imaging methodology based on AnxA5-derived tools to monitor apoptosis in vivo (Table 5d-f). As mentioned above, 99m TC radiolabeled AnxA5 and AnxA5-conjugated gold nanoparticles were valuable diagnostic imaging tools in mouse models for atherosclerosis [253][254][255] (Table 5d). AnxA5 dual-labeled with 99m TC and Alexa Fluor 568 allowed detection of ethanol-induced cell death in the femur muscle but failed to visualize cell death in a murine model for stroke [272]. Imaging of apoptosis using 111 In-and DOTA-peptide -labeled AnxA5 monitored efficacy of suicide gene therapy after treatment with ganciclovir in NG4TL4 sarcoma expressing herpes simplex virus thymidine kinase [273]. Drug-induced apoptosis in tumour-bearing mice was also detectable using AnxA5-conjugated ultrasmall superparamagnetic iron oxide [274]. A novel AnxA5-conjugated nano-sized ultrasound contrast agent was able to detect apoptotic cells inside tumours after chemotherapy [275] (Table 5e). Magnetic beads coated with AnxA5 were recently used to isolate tumour-derived extracellular vesicles from mouse models xenografted with human cancer cells, an approach with diagnostic and therapeutic potential [276]. Another alternative to 99m Tc-AnxA5, which has limitations for non-invasive molecular imaging in the clinic [277], is 68 Ga-labeled AnxA5, which efficiently visualized hepatic apoptosis in an anti-Fas antibody mouse model and efficacy of established chemoand radiotherapies in animals with Burkitt's lymphoma [278]. Alternatively, fusion of AnxA5 with a mutant of Renilla luciferase created a new and sensitive biosensor that effectively detected vascular apoptosis in ethanol-induced corneal apoptosis in mice [279]. Interestingly, labeling of AnxA5 with a pH-sensitive probe that emits fluorescence upon pH decrease provided opportunity to monitor efferocytosis, the clearing of apoptotic cells by phagocytosis [280] (Table 5f).

Other PS-Related Aspects of AnxA5 Biology
Other AnxA5-related biological activities reviewed in more detail elsewhere include viral infection, membrane invagination and membrane repair [11,226,[281][282][283]. In addition, AnxA5-KO mice showed a strongly reduced allogeneic reaction against primary and secondary necrotic cells as well as carcinoma cells [284,285]. This could also be relevant for preventing viral infections, including human immunodeficiency virus, which together with infected monocytes/macrophages, exposes increased amounts of PS on the surface [284] ( Table 5f). Possibly relevant in this context is the ability of the core domain of AnxA5 and several other annexins to interact with Dectin-1, which facilitates uptake of apoptotic cells by dendritic cells and is an important mechanism for establishing peripheral tolerance to self-antigens [286].
Alternative splicing generates two isoforms, AnxA6-1 and AnxA6-2, the latter lacking six amino acids at the C-terminus (pos. 524-529) [288,299]. The AnxA6-1 isoform is abundant in most cells and tissues, while AnxA6-2 is upregulated in some transformed cell lines [288,300,301] and isoform-specific functions related to Ca 2+ homeostasis and endo-/exocytosis may exist [288,[302][303][304], but in vivo evidence supporting these observations is still lacking. Interestingly, computational models suggested limited overlap, but rather a diversity of functions associated with the two AnxA6 isoforms [305].

AnxA6 and Cardiac Function
Initial studies with mice overexpressing AnxA6 in the heart led to enlarged dilated hearts, acute diffuse myocarditis, lymphocyte infiltration and fibrosis throughout the heart and pulmonary veins, ultimately causing heart failure [306] (Table 6a). Furthermore, cardiomyocytes from Anxa6 −/− mice showed higher contractility and accelerated removal of diastolic Ca 2+ from the cytoplasm [307]. Hence, AnxA6 scaffold and membrane organizer functions probably regulate Ca 2+ -dependent ion pumps and/or exchangers. In addition, Ca 2+ signaling and stimulus-response coupling in the heart were altered in mice overexpressing a dominant-negative N-terminal AnxA6 mutant [308]. As other N-terminal AnxA6 deletion mutants compromised endocytosis and signaling [309][310][311][312][313], these mutants probably lack Ca 2+ -dependent membrane-binding properties and remain cytosolic, trapping their interaction partners in the cytosol (Table 6a).
Nevertheless, AnxA6 deficiency in mice was not associated with alterations in the heart rate, blood pressure, cardiovascular response to septic shock or changes in B or T cell development [314] (Table 6a), challenging proposed AnxA6 functions in the heart and in lymphocyte differentiation [306][307][308]315].

AnxA6 and Skeletal Development
AnxA6 is highly expressed in bone and skeletal muscle, yet Anxa6 −/− animals and even double KO-mice lacking AnxA6 and AnxA5 did not reveal defects in skeletal development [238,239] (see also Section 6.1). However, in later studies, a reduction in the growth plate length and number of chondrocytes in newborn AnxA6 KO-mice was observed, which correlated with reduced mineralization of growth plate cartilage [316]. In addition, Anxa6 −/− chondrocytes exhibited delayed terminal differentiation, markedly decreased intracellular Ca 2+ levels upon ascorbic acid stimulation and reduced PKCα membrane translocation and activity, the latter possibly decreasing MAPK signaling required for chondrocyte differentiation [316] (Table 6b).
In studies mimicking osteoarthritis, knee cartilage destruction after IL-1β injection or partial meniscectomy was strongly reduced in AnxA6 KO-mice [317]. Mechanistically, the loss of AnxA6 interacting and promoting nuclear translocation of NFκB, a major driver in the pathogenesis of osteoarthritis, could be responsible for this phenotype. Furthermore, in articular chondrocytes from control animals, AnxA6 associates with the plasma membrane to interfere with Wnt/b-catenin signaling, affecting crosstalk with NFκB signaling, catabolic and inflammatory events, all of which contributing to cartilage degradation in osteoarthritis [318].
AnxA6 KO-mice displayed increased sensitivity to mechanical stimuli due to elevated activity of the cation channel Piezo2 in sensory neurons, which mediates mechano-gated current linked to body awareness and touch [320]. Moreover, AnxA6 overexpression attenuated mechanical pain, indicating therapeutic potential to inhibit chronic mechanical pain commonly associated with osteoarthritis (Table 6b).
Along these lines, membrane order and protein distribution in rafts and non-rafts was also altered in mesenchymal mouse embryo Anxa6 −/− fibroblasts [319] (Table 6d). These membrane-organizing functions of AnxA6 might also affect other immune responses, as AnxA6-deficient mice showed reduced bacterial clearance after oral challenge with Citrobacter rodentium, indicating a higher susceptibility to infection (The Wellcome Trust Sanger Institute Mouse Genetics Project, Database Release 2011; http://www.informatics. jax.org/reference/J:175295; accessed 12 March 2021).

AnxA6 and Lipid and Glucose Homeostasis
AnxA6 has been linked to lipoprotein and cholesterol uptake and transport, cytokine secretion as well as signaling events related to lipid and glucose homeostasis [4,9,288,292], which appear relevant in liver and adipose physiology (Table 6e).
AnxA6 was found associated with lipid droplets in adipocytes [323,324] and AnxA6 upregulation in 3T3 adipocytes lowered triglyceride storage and secretion of adiponectin, which exerts beneficial effects on glucose and lipid homeostasis [323,324]. Vice versa, serum adiponectin levels were increased in AnxA6-deficient mice. As AnxA6 levels increased in WAT of obese mice, this may contribute to impaired lipid storage and adiponectin release associated with obesity [324]. AnxA6 is also associated with hepatic lipid droplets [290,325], with primary Anxa6 −/− hepatocytes and liver sections of AnxA6 KO-mice showing reduced triglyceride accumulation and lipid droplet numbers [325]. As AnxA6 up-and downregulation occurs in hepatic lipid disorders, this may implicate a novel role for AnxA6 in fatty acid and triglyceride metabolism in the liver [325].
During the course of HFD-feeding, AnxA6 KO-mice gained less weight, in particular in WAT, compared to controls [326]. HFD-feeding caused elevated total plasma triglycerides and cholesterol levels, yet an overall lipoprotein profile commonly associated with increased insulin sensitivity was observed [326][327][328]. Indeed, AnxA6 deficiency resulted in increased hepatic insulin signaling and improved glucose clearance [326]. However, HFD-fed AnxA6 KO-mice failed to properly downregulate hepatic gluconeogenesis and revealed increased glycogen storage in the liver, implicating AnxA6 in the fine-tuning of hepatic glucose metabolism and systemic control of glucose [326] (Table 6e).
Strikingly, when the ability to regenerate the liver was challenged after partial hepatectomy (PHx), AnxA6-deficient mice exhibited a significantly lower survival rate after PHx [329]. This striking phenotype was not due to defects in hepatocyte cell cycle progression, or the delayed onset of steatosis in these animals, but a prolonged hypoglycaemia, which could not accommodate the heavy energy requirements in AnxA6 −/− animals after PHx. Mechanistically, shuttling of the Na + -coupled neutral amino acid transporter SNAT4 to the hepatocyte plasma membrane, which provides alanine as fuel to drive gluconeogenesis and fatty acid synthesis during liver regeneration, was compromised in Anxa6 −/− mice [329]. Such a decisive role in the post-hepatectomy phase has yet been documented only for a rare number of genes [330][331][332].
The liver is central for cholesterol homeostasis and contains abundant amounts of AnxA6 [333], with potential roles in the uptake and internalization of LDL-derived cholesterol [309,310,334,335]. Moreover, AnxA6 overexpression induced cholesterol accumulation in late endosomes, a phenotype reminiscent of the Niemann-Pick type C1 (NPC1) mutation [336]. Outstandingly, AnxA6 depletion in NPC1 mutant cells alleviated cholesterol accummulation by restoring cholesterol export to the endoplasmic reticulum via Rab7 and StAR-related lipid transfer domain-3 (StARD3)-dependent formation of membrane contact sites [337]. In line with these findings, Rab7-GTP levels were elevated in mouse embryo fibroblasts from Anxa6 −/− mice [337], indicating that the gatekeeper function of AnxA6 to control cholesterol transport from late endosomes could be relevant in vivo and the progression of NPC disease [338] (Table 6e).
In NPC disease, late endosomal cholesterol accumulation due to the loss of NPC1 primarily affects the brain, causing neuronal damage and affecting motor coordination. In addition, considerable liver malfunction is also common. Strikingly, lack of AnxA6 in NPC1-deficient animals (Npc1 −/− /Anxa6 −/− ) did not prevent cerebellar degeneration [339]. Furthermore, livers of the AnxA6-deficient NPC1 KO-animals contained a significantly elevated number of foam cells congesting the sinusoidal space, which is commonly associated with inflammation, further deteriorating their compromised hepatic functions and reduced lifespan [339].

AnxA6 and Cancer
Numerous studies identified AnxA6 to exhibit tumour suppressing as well as tumour promoting activities, depending on the cancer cell type and degree of malignancy [311,342,343]. These opposing activities probably relate to the multiple and diverse AnxA6 scaffolding functions, which are likely to differ in the various cancer cells due to the different repertoire and availability of interaction partners. Besides data from cell culture models or expression analysis of patient cohorts, several studies in recent years examined roles for AnxA6 in tumour growth and progression in vivo (Table 6f).
In EGFR overexpressing A431 epithelial carcinoma lacking endogenous AnxA6 [311,312], restoration of AnxA6 expression promoted PKCα-mediated threonine 654-EGFR phosphorylation, which inhibited EGFR tyrosine kinase activity, cell growth, migration and invasion [313,344]. AnxA6 overexpression in these A431 cells also reduced growth and EGFR activity in A431 xenografts in a PKCα-dependent manner [345,346]. Moreover, AnxA6 upor downregulation impacted on the efficacy of tyrosine kinase inhibitors to inhibit growth, migration and invasion of EGFR-related cancer cells [346].
In a TNBC model using BT-549 xenografts, AnxA6 depletion was associated with early onset and rapid growth of tumours, which correlated with poor overall survival of basallike TNBC patients [347]. These studies identified RasGRF2, a Ras protein specific guanine nucleotide exchange factor (RasGEF), as an additional effector besides PKCa and p120 GTPase activating protein [3,8,311,312], to confer AnxA6-dependent roles in cell growth and motility [347]. In fact, the inverse relationship of AnxA6 and RasGRF2 expression levels discriminated rapidly growing from invasive TNBC subsets [348]. This could be relevant for subset-specific therapeutic interventions, as AnxA6 upregulation was associated with the response of TNBC cells to cytotoxic and/or EGFR-targeted therapies and the development of drug resistance [343,349] (Table 6f).
Significant amounts of AnxA6 are found in late endosomes/(pre-) lysosomes [9,288,292,[335][336][337]. Inner membranes from this compartment can fuse with the plasma membrane to be released as extracellular vesicles (EVs) [4,292,350]. These EVs, including exosomes, are highly enriched in annexins, including AnxA6 (ExoCarta exosome database: www.exocarta.org; accessed 12 March 2021) and are believed to facilitate cell-cell communications and influence recipient cell behaviour.   In two mouse models of breast cancer (MMTV-PyMT, 4T1), taxane and anthracycline therapy induced secretion of EVs from tumours with enhanced metastatic capacity. These chemotherapy-elicited EVs were enriched with AnxA6 and promoted NFκB-dependent endothelial activation, cytokine secretion and monocyte subtype expansion in the pulmonary pre-metastatic niche to establish lung metastasis [351]. Hence, AnxA6 enrichment on EVs might be useful to develop a biomarker to predict the risk for metastasis in patients with poor response to chemotherapy. Similarly, AnxA6-positive EVs from CAFs supported PDAC aggressiveness, and was restricted to PDAC patients, making it a potential biomarker for PDAC grade [352]. Along these lines, AnxA6-containing EVs from CAFs also contributed to drug resistance in a peritoneal metastasis mouse model for gastric cancer, possibly via activation of integrin β1, focal adhesion kinase (FAK) and yes-associated protein 1 (YAP) signaling [353]. The identification of a monoclonal AnxA6 antibody with anti-invasive properties on aggressive pancreatic, lung squamous and breast cancer cells further supports pro-metastatic activity of extracellular AnxA6 [113] (Table 6f).

AnxA6 and Membrane Repair
Studies in human and rodent cell-based models implicated several annexins, including AnxA6, in membrane repair [11,287,294]. Likewise, in zebrafish, AnxA6 was rapidly recruited to damaged sarcolemma and cooperated with dysferlin to repair membrane lesions [293] (Table 6g). In this transparent and powerful model organism for the study of vertebrate biology, AnxA6 and other annexins accumulated at the site of muscle damage, with AnxA6 deficiency leading to myopathy [293]. Loss of dysferlin was also accompanied by impaired AnxA6 translocation and resealing after sarcolemmal disruption in mouse models [354]. Strikingly, in a mouse model of muscular dystrophy, a naturally occurring AnxA6 splice variant created a truncated AnxA6 protein lacking four of the eight annexin repeats. This mutant acted in a dominant-negative manner and interfered with membrane recruitment of full-length AnxA6 to substantially decrease efficiency of membrane repair [295]. Other genetic modifiers of muscular dystrophy also interfered with AnxA6 recruitment for sarcolemma resealing [355]. High resolution microscopy of muscle after laser injury revealed rapid recruitment of AnxA6 and other annexins (AnxA1, A2, A5) to damaged sarcolemma, contributing to the formation of a tight cap in a Ca 2+ -and actin-dependent manner over a vesicle-rich repair zone [11,287,295,356]. Interestingly, intermittent glucocorticoid treatment enhanced muscle repair, through increased expression of AnxA6 and AnxA1, without eliciting the well-known effect of steroids to trigger muscle atrophy [357]. Alternatively, administration of recombinant AnxA6 may protect against acute muscle injury, as it was capable to reseal injured membrane in mouse models [358] (Table 6g).

AnxA7
In contrast to all other annexins, AnxA7 contains a long (100 aa) hydrophobic Nterminus ( Figure 1), with alternative splicing generating a 47 kD isoform found in all tissues except skeletal muscle, and a 51 kD isoform expressed in the heart, brain and myotubes [359]. AnxA7 is predominantly associated with secretory vesicles, the plasma membrane and the nuclear envelope in a Ca 2+ -dependent manner [360]. Many reports support AnxA7 to have GTPase activity and contribute to the regulation of Ca 2+ homeostasis, exocytic pathways, and prostaglandin production, with consequences for pancreatic and cardiac functioning and inflammatory myopathies [361][362][363][364][365][366][367], but also cell survival and tumour growth [368,369].

AnxA7 and Pancreatic β-Cell Function
The first study examining the targeted disruption of the AnxA7 gene reported an embryonic lethality due to cerebral haemorrhage in homozygous AnxA7 −/− mice [361]. Yet, heterozygous Anxa7 +/− mice of this strain were viable and the pancreas of these animals exhibited islet hyperplasia, β-cell hypertrophy, aberrant metabolic gene expression patterns and reduced inositol 1,4,5-trisphosphate receptor levels. The latter was considered responsible for defects in the Ca 2+ release from intracellular stores, compromising Ca 2+dependent insulin secretion. In follow-up studies, altered ryanodine receptor (RyR)mediated Ca 2+ release was also related to abnormal insulin secretion from β-cells of these Anxa7 +/− mice [370]. In contrast, an independently generated second AnxA7 KO-strain was viable and lacked any defect in Ca 2+ -induced and cAMP-mediated insulin secretion [371]. The different genetic background, design and integration sites of the targeting construct, orientation of the inserted neo gene and possible consequences for neighbouring genes could possibly explain the strikingly different phenotypes of the two Anxa7 −/− strains [371] ( Table 7a).

AnxA7 and Cardiac Function
AnxA7 is highly expressed in the heart and based on its possible role in Ca 2+ homeostasis, adult cardiomyocytes from the viable AnxA7 KO-strain were analyzed [371]. These studies identified the misfunctioning of the apparatus responsible for cardiac muscle contraction, most likely due to defects in Ca 2+ homeostasis [371]. In fact, AnxA7 interacts with Sorcin and RyR, which both couple Ca 2+ channels to the contractile machinery in cardiac muscle [372], indicating that the loss of these interactions might contribute to the cardiac phenotype in Anxa7 −/− animals [371,373].
In addition, the viable Anxa7 −/− strain showed increased susceptibility to atrial fibrillation and ventricular tachycardia, both electrophysiological properties linked to the development of heart arrhythmia [374]. Furthermore, in atrial and ventral tissues of the Anxa7 −/− mice, impaired integrity of the basement membrane of cardiomyocytes and lack of collagen fibers in intracellular spaces was observed, possibly contributing to the de-regulated conductive properties of these animals [374]. Moreover, when examining the consequences of increased resistance to blood flow out of the heart upon transverse aortic constriction, a well-established surgical procedure to induce cardiac hypertrophy, the increase in the ratio of heart to body weight was more pronounced in AnxA7 KO-mice. AnxA7-dependent cardiac activity of the transcription factor NFAT (nuclear factor of activated T cells), regulating genes involved in the hypertrophic response following cardiac stress conditions, could contribute to the observed phenotype [367] (Table 7a).

AnxA7 and Eryptosis
Besides cardiac complications, red blood cells from Anxa7 −/− mice were characterized by an altered cell shape and increased resistance to osmotic shock [375]. Inspection of eryptosis, the suicidal death of erythrocytes, uncovered hyperosmotic shock, Clremoval or energy depletion to trigger enhanced COX-mediated prostaglandin E 2 (PGE 2 ) production in Anxa7 −/− erythrocytes [376]. This elicited cytosolic Ca 2+ elevation, subsequent PS exposure at the cell surface, and ultimately, apoptotic death. Eryptosis was also accelerated in Plasmodium-infected Anxa7 −/− erythrocytes, indicating that AnxA7 downregulation may partially protect against malaria in vivo [377]. Although several annexins have previously been proposed to inhibit PLA 2 [1,2,10], which acts immediately upstream of COX in the prostaglandin synthesis pathway, only pharmacological inhibition of COX, but not PLA 2 , abolished the differences in parasitemia and erythrocyte survival between Anxa7 −/− and wildtype animals [377] (Table 7a).
Additionally, plasma prostaglandin levels and hepatic COX activity were significantly elevated in AnxA7 KO-mice [366]. As PGE 2 upregulation can induce insulin resistance in hepatocytes, this might contribute to the decreased glucose tolerance observed in these animals. Pharmacological COX inhibition, using aspirin, improved glucose tolerance in the AnxA7-deficient strain, but also abrogated differences in insulin levels in wildtype and AnxA7 KO-animals (Table 7a).
In other studies, the increased PS exposure and eryptosis of Anxa7 −/− erythrocytes were associated with an increased adhesion to endothelial cells, an observation that might be relevant for the adhesion of erythrocytes to the vascular wall and contribute to impaired microcirculation during acute ischemic renal failure [379].
Enhanced PGE 2 formation upon AnxA7 depletion appears relevant also for implantation and fertility [380]. In women suffering from recurrent pregnancy loss, decreased AnxA7 levels in endometrial biopsies during the midluteal window of implantation were observed. This correlated with the viable Anxa7 −/− strain showing significantly increased implantation sites and litter sizes, indicating that ANXA7-mediated regulation of COXdependent PGE 2 production contributes to regulating endometrial receptivity and implantation [380].

AnxA7 and Brain Injury
Several studies addressed roles for AnxA7 in the brain (Table 7a). Possibly linked to AnxA7-related impacts on the cellular handling of Ca 2+ (Sections 8.1-8.3), lack of AnxA7 affected Ca 2+ homeostasis and proliferation of primary Anxa7 −/− astrocytes [381].
Other research suggested PKC-mediated phosphorylation of ANXA7 to promote interaction with the SNARE proteins SNAP23 and SNAP25, facilitating membrane fusion of exocytic vesicles with the plasma membrane [363,382]. These regulatory circuits may be relevant for presynaptic glutamate release and postsynaptic N-methyl-D-aspartate receptor trafficking, as proposed in studies examining brain tissue around hematoma after intracerebral hemorrhage-induced brain injury [382]. In this model, ANXA7 knockdown in vivo showed rescue effects on neuronal death, blood-brain barrier damage, brain edema, neurobehavioral deficient, and inflammatory response. Similar results were also obtained after subarachnoid hemorrhage-induced brain injury [383] and correlate with elevated AnxA7 levels in traumatic brain injury [384], making AnxA7-based therapies a potentially novel and alternative approach in the treatment of cerebrovascular disease (Table 7a).

Small Molecules Targeting AnxA7
Given the therapeutic potential of AnxA7 in a variety of disease settings, efforts to screen small molecules identified 6-amino-2, 3-dihydro-3-hydroxymethyl-1, 4-benzoxazine (ABO) to selectively bind and interfere with AnxA7 phosphorylation, altering AnxA7 localization, expression levels and its GTPase activity [385]. ABO was initially found to induce AnxA7-dependent autophagy in cell-based studies [385]. In apoE −/− mice fed a Western diet and treated with ABO, elevated AnxA7 levels in the aortic endothelium correlated with increased autophagy and reduced apoptosis in this location. Moreover, this was associated with reduced atherosclerotic plaque size and a more stable plaque phenotype, characterized by reduced lipid deposition, improved ratio of pro-/anti-inflammatory macrophages, increased collagen and smooth muscle cell content, and less apoptosis [386]. The underlying mechanisms how ABO-mediated inhibition of AnxA7-GTPase activity could reduce the size and stability of atherosclerotic plaques are complex, and include altered expression of proteins involved in survival, metabolism, and autophagy [387,388], respectively (Table 7a).

AnxA7 and Cancer
In human cancers, deletions at the Anxa7 gene locus on chromosome 10q21 are common, and a substantial number of studies identified AnxA7 and its GTPase activity as a tumour suppressor with roles in cell death, motility and invasion in a variety of cancers [368,369]. For instance, ANXA7 expression is commonly lost in hormone-refractory and locally recurrent metastatic prostate cancer, glioblastoma, and melanoma. On the other hand, AnxA7 upregulation and tumour-promoting roles in the initiation and progression of liver, colorectal, gastric, nasopharyngeal and breast cancer, have been reported. It would go beyond the scope of this review to recapitulate these opposing findings in more detail, but often the underlying mechanisms can be associated with alterations in Ca 2+ homeostasis, Ca 2+ /GTP-dependent exocytic events, expression patterns of downstream effectors, as well as AnxA7-dependent protein interactions [368,369].
Several in vivo studies support tumour suppressor roles for AnxA7 (Table 7b). For instance, the initially generated heterozygous Anxa7 +/− strain [361] revealed a cancerprone phenotype, with more than 20% of these mice developing spontaneous neoplasms, with lymphosarcoma being most frequent, especially in females, indicating gender-and possibly sex hormone -related differences [389]. This was accompanied by reduced expression of several tumour suppressors, DNA repair-and apoptosis-related genes, indicating the development of genomic instability driving disease progression upon partial loss of AnxA7 [389].

AnxA8
AnxA8 is a less well-characterized member of the annexin family, and expressed at low levels in the lung, liver, kidney, skin, placenta, and the cornea [395,396]. AnxA8 levels are upregulated by leptin, during mammary gland involution and in acute promyelocytic leukemia (APL) cells [396][397][398]. In these APL cells, all-trans retinoic acid-induced differentiation was associated with AnxA8 downregulation, indicating a role in proliferation and differentiation [398]. Indeed, over the years, elevated AnxA8 levels were reported in a variety of cancers [397][398][399], and as shown upon in vivo knockdown in endometrial cells of the uterine horn of mice, possibly promoting cell proliferation via Akt signaling pathways [400] (Table 8).
AnxA8 controls the sorting and transport of late endosomal cargo [401], a function that appears critical for leucocyte adhesion to activated endothelium, which requires stabilization of the leukocyte receptor P-selectin on the endothelial cell surface by CD63. CD63 is located in late endosomes and in order to fulfill this function, needs to be transported to the cell surface. Strikingly, cell surface delivery of CD63 in AnxA8-deficient human umbilical vein endothelial cells was strongly reduced and correlated with compromised leukocyte adhesiveness in inflammatory-activated endothelial venules of AnxA8-deficient mice [402] (Table 8).

AnxA9
AnxA9 was initially identified in expression libraries from fetal liver/spleen [403] and shares a high sequence homology with AnxA2. Yet, AnxA9 is an atypical annexin, as inactivated type II-Ca 2+ binding sites within its core domain confered Ca 2+ -independent membrane and phospholipid-binding properties [404]. Together with its low abundance during mouse development and in adult organs, in particular spleen and liver [405], this suggests a Ca 2+ -independent and rather specific role unrelated to membrane scaffolding functions proposed for the other annexins. Interestingly, autoantibodies isolated from patients with pemphigus vulgaris, a rare skin autoimmune disease, cross-reacted with AnxA9 [406]. However, the exact role of AnxA9 in this disease has yet to be clarified and an AnxA9 KO-mouse model does not exist.

AnxA10
Annexin A10 was identified only two decades ago [407] and is mainly expressed in epithelia of the gastrointestinal tract [408]. Its localization and function, alike AnxA8, remains to be fully understood, but unlike other annexins, AnxA10 displays functions independent of Ca 2+ and/or membrane-binding and was found in nuclear bodies interacting with mRNA-binding proteins, possibly involved in mRNA regulation or processing [409]. Homozygous deletion or downregulation of AnxA10 in several cancers, including blad-der, gastric, pancreas and liver, correlated with progression and poor survival, indicating potential as a diagnostic marker and tumour suppressor [408,[410][411][412].
To our knowledge, up to date, only two studies utilized gene depletion approaches to examine AnxA10 functions in vivo, both addressing neuropathic pain. In mice, spinal nerve ligation (SNL) -induced neuropathic pain was associated with AnxA10 upregulation in the spinal cord. AnxA10 colocalized with neuronal and astrocyte markers and upon AnxA10 depletion, SNL-induced pain was attenuated [413]. Similarly, AnxA10 knockdown in the spinal cord of rats suppressed SNL-induced hyperalgesia, possibly by blocking the activation of pro-inflammatory mediators, including NFκB, TNFα and IL-1β and metalloproteases [414] (Table 9).

AnxA11
AnxA11 is ubiquitously expressed in a wide variety of tissues with diverse functions in cytoplasmic and nuclear locations. Earlier studies identified AnxA11 to associate with the nuclear envelope in a Ca 2+ -and cell cycle-dependent manner, binding to S100A6 (calcyclin) [415], an interaction possibly relevant for cell growth and differentiation [415][416][417]. During cell cycle progression, AnxA11 translocates from the nucleus to the spindle poles in metaphase and to the spindle midzone in anaphase, enabling midbody formation and completion of cytokinesis [417]. In the cytosol, AnxA11 binds to secretory vesicles in a Ca 2+ -dependent manner, regulating Ca 2+ -dependent exocytic events, such as Ca 2+ -inducible insulin secretion in pancreatic β-cells [418]. In the early secretory pathway, interaction of AnxA11 with apoptosis-linked gene-2 (Alg-2) and Sec31A contributes to the exit of cargo from the endoplasmic reticulum for transport to the Golgi apparatus [419].

AnxA11 and Cancer
Although confirmation from in vivo studies in mouse models are still limited, in certain cancers AnxA11 up-or downregulation appears to be linked to disease progression and development of drug resistance [93] (Table 10). For instance, in ovarian cancer AnxA11 downregulation correlated with tumour recurrence and was associated with cisplatin resistance [420]. In a mouse model for human hepatocellular carcinoma, ANXA11 upregulation was associated with increased Akt activation and contributed to tumour growth and progression [421]. Species and/or cancer-subtype specific differences may exist, as ANXA11 downregulation in the murine Hca-P hepatocarcinoma cell line promoted in vivo tumour growth, lymph node metastatic potential and chemoresistance towards 5-fluorouracil [422].
Strikingly, a single nucleotide polymorphism (SNP) in the Anxa11 gene, causing a mutation at position 230 (R230C), was associated with an increase in the overall response rate of metastatic colorectal cancer patients to bevacizumab, an angiogenesis inhibitor [423]. In support of these findings, the R230C variant was more susceptible to bevacizumabinduced tumour suppression when examining xenograft models [424]. The underlying mechanism remains to be clarified, but this Anxa11 SNP (rs1049550) may serve to improve the identification of metastatic colorectal cancer patients sensitive to bevacizumab regimens [425].  [424] Abbreviations: 5-fluorouracil, 5-FU; Akt, protein kinase B; Anx, annexin; HCC, hepatocellular carcinoma; KO, knockout; miR, micro RNA.

AnxA11 and Autoimmune Diseases
Besides the link of several annexins, including AnxA11, to the autoimmune disease SLE, Anxa11 polymorphism were found in other autoimmune disorders, such as granulomatous disease [426]. Furthermore, several genome-wide studies in recent years identified a genetic association of the R230C variant with sarcoidosis [427,428].
Finally, Anxa11 mutations were found in the rare and related neurodegenerative diseases amyotrophic lateral sclerosis (ALS) and frontotemporal dementia. Numerous mutations in both the N-terminal tail and the C-terminal membrane binding region of Anxa11 have now been described and may account for up to 6% of familial ALS in Chinese populations [429][430][431][432].
Recent studies provided mechanistic insight, proposing disease-associated Anxa11 mutations to contribute to ALS pathogenesis through dysregulated Ca 2+ homeostasis and abnormal protein aggregation [433]. Furthermore, Anxa11 mutations in ALS altered several fundamental AnxA11 properties. In primary rat neurons, but also in an in vivo zebrafish model, ANXA11 acted as a molecular tether, coupling RNA granules to lysosomes, enabling long-distance transport of RNA along axons [432]. This novel relationship between lysosomes and RNA was disrupted in ALS-associated Anxa11 mutations, as RNA granules could not dock onto lysosomes for axonal transport. Hence, Anxa11 mutations may reduce delivery of essential mRNAs to distal regions of the neuron, which may dysregulate neuronal homeostasis and synaptic activity [432].

Conclusions
Exposure of annexin KO-mice or animals harbouring implanted annexin-deficient cell lines to a variety of disease-related stress conditions revealed important biological functions of individual annexins (Tables 1-10). These animal models enabled the characterization of prominent extracellular activities for several annexins (i.e., AnxA1, A2, A5). This group of annexins might in the future serve as diagnostic or therapeutic tools-maybe in combination with nanoparticles or radio-and fluorescence-labeling. Peptide or protein-based approaches could be suitable to facilitate targeted delivery in vivo. AnxA1 and its interaction with FPR receptors has potentially numerous therapeutic applications in the context of acute and chronic inflammation and cancer (Sections 2.1-2.4). The extracellular functions of AnxA2 are therapeutically relevant to control fibrinolysis and neoangiogenesis in several diseases (Sections 3.1 and 3.3). The strong affinity of AnxA5 for PS on apoptotic cells can be applied in a large range of diagnostic and therapeutic approaches (Sections 6.2-6.6).
Phenotypes associated with the annexins AnxA3, A4, A6, A7, A8, A10 and A11 are often linked to Ca 2+ homeostasis, as well as their membrane organizing and scaffolding properties, affecting receptors, transporters, or ion channels and their communication with cellular protein networks. However, the multifunctionality of these annexins render the development of targeted approaches difficult and fine-tuned local modulation of specific protein-protein or protein-lipid interactions remain hard to achieve. Nevertheless, drug targeting the GTPase activity of AnxA7 via small molecules hold promise for treatment of certain cancers and cardiovascular diseases (Sections 8.5 and 8.6). In addition, drug-induced AnxA6 upregulation, delivery of recombinant AnxA6 or AnxA6-targeting antibodies might lead to the development of novel treatment options in lipid disorders, liver dysfunction, several cancers and muscular damage (Sections 7.4-7.6). Finally, the controlled up-or downregulation of annexins using viral gene delivery, small pharmacological molecules or CRISPR/Cas technology could provide opportunities for therapy in certain organs, such as liver and muscle, but also in the context of many cancer-related settings. Their intraand extracellular expression levels could have diagnostic value, possibly linked to cancer progression and/or anticancer drug resistance (Sections 3.3, 4.1, 5.1, 6.4 and 7.5).
On the other hand, the actual genetic evidence to support the relevance of annexins in human disease is still sparse. In earlier studies, annexin anomalities, also called 'annexinopathies', were initially reported in two disease states, identifying AnxA2 overexpression in patients with a haemorrhagic form of acute promyelocytic leukaemia, and AnxA5 downregulation on placental trophoblasts in the antiphospholipid syndrome [434,435]. Over recent years, a substantial number of additional studies identified correlations between disease progression and expression patterns of individual annexins; as is the case in many cancer and other acute and chronic diseases [1][2][3][21][22][23]90,116,226,242,342,343,368,369]. However, aided by deep sequencing data, various SNPs were identified, suggesting a potential involvement of AnxA1 in Alzheimer's disease [436], and revealed roles for AnxA2 in sickle cell disease [119,120], but also cardiovascular risk due to elevated LDL cholesterol [114,149]. Similarly, the identification of SNPs in the Anxa3 gene locus may imply roles in gastric cancer [437]. SNPs in the Anxa5 gene are associated with the risk of pregnancyrelated venous thrombosis, preeclampsia [438,439] and malignant melanoma [440]. Certain SNPs in the Anxa6 locus have been linked to psoriasis [441,442], while another polymorphism in the Anxa6 coding region causes a dominant-negative deletion mutant in a mouse model for muscle dystrophy that interferes with membrane repair [295].
Finally, and most convincingly, the R230C variant and several other SNPs in the Nand C-terminal region of Anxa11 have been associated with neurodegenerative and autoimmune disorders, such as ALS [429][430][431][432][433] and sarcoidosis [427][428][429]. These polymorphism could be explored further and might serve as a marker for anticancer drug performance in metastatic colorectal cancer [423][424][425]. These observations, together with advanced intravital imaging techniques, -omics approaches, molecular modelling and small molecule development, in combination with tissue specific knock-in/KO-mouse models could open exciting perspectives to further pursue annexin-based therapeutic strategies.
Funding: TG is supported by the University of Sydney (RY253, U3367). CE is supported by BFU2015-66785-P from the Spanish Ministerio de Economía y Competitividad and AR0RM005 from the University of Barcelona (Spain). UR is supported by the Interdisciplinary Center for Clinical Research (IZKF, Re2/022/20) of the Münster Medical School (Germany). We acknowledge support from the Open Access Publication Fund of the University of Muenster.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.