Animal models of bleeding and tissue repair
The authors stated that they had no interests which might be perceived as posing a conflict or bias.
Abstract
Summary. While a number of animal models have been developed for human haemophilia, it has been difficult to develop reproducible measures of bleeding in these models. They have also not been extensively utilized to study the complications of haemophilia beyond blood loss. Poor haemostatic function also leads to local haematomas, joint damage and poor wound healing. Some of the abnormalities related to bleeding are because of the deleterious effects of iron deposition in the tissues. Evidence from mouse skin wound and joint haemorrhage models suggests that bleeding and iron deposition initiate a vicious cycle of inflammation, angiogenesis and renewed bleeding. However, there is much yet to be learned about the effects of bleeding on tissue responses, including validating the results of animal studies in clinical trials.
Introduction
While the most feared complication of a bleeding tendency is exsanguinating haemorrhage, local complications of excessive bleeding are much more common in the modern era. Patients with haemophilia experience local effects from haematomas (including compartment syndromes and pseudotumors), arthropathy secondary to joint haemorrhage and impaired wound healing. Surprisingly, very little research has been conducted into such non-lethal complications of haemorrhage. This review is devoted to examine the research on the effects of impaired haemostasis on the subsequent tissue response to injury.
There are a number of animal models of impaired haemostasis in a variety of species. Fundamentally, they are of three types: (i) a naturally occurring condition that mimics a human bleeding disorder; (ii) a deficiency or defect of a component of coagulation induced by pharmacologic treatment; and (iii) a deficiency or defect induced by targeted genetic modification (knockout mice).
The first well-characterized animal models of haemophilia were naturally occurring mutations in dogs [1–3] that are similar to human haemophilia A and B. There are also canine and porcine versions of von Willebrand’s disease (VWD) [3]. These models are extremely useful, because dosing of replacement and bypassing therapies in dogs and pigs are much more similar to humans than in small animal models such as mice, rats and rabbits [4,5].
However, it is easier to find an animal model of a bleeding tendency than it is to characterize bleeding in that animal and relate the findings to the clinical picture in humans. While larger animal models have a similar phenotype to human haemophilia, it can be difficult to reproducibly measure their bleeding tendency. A straightforward template bleeding time works well to demonstrate excessive bleeding in platelet disorders and vWD, but is normal in haemophilia and many other factor deficiencies [6]. This is because primary haemostasis is established by the formation of a platelet plug at a site of injury – and platelet function is normal in haemophilia. To maintain haemostasis, the initial platelet plug must be stabilized by a fibrin meshwork. This ‘secondary haemostasis’ is abnormal in haemophilia and is clinically manifested as delayed rebleeding. Therefore, challenges in which fibrin formation is essential for haemostasis are best suited to demonstrate the bleeding tendency in haemophilia. A cuticle bleeding time has primarily been used for this purpose in dogs [6,7], largely supplanting the earlier gingival resection bleeding time [1].
In spite of the advantages of dog models of haemophilia, it is very expensive and labour-intensive to maintain colonies of larger animals with bleeding tendencies and most research on haemophilia has been conducted in small animal models.
Animals from a variety of species can be rendered temporarily deficient in a coagulation factor by administration of an antibody against the factor of interest. This technique has been used to model haemophilia in mice, rats and rabbits [8,9].
The use of knockout mouse models is becoming a much more common means of modelling haemophilia and related disorders. Mice have been developed in which each of the common coagulation factor deficiencies have been reproduced. These colonies are easier and cheaper to maintain than larger animals. However, it must be remembered that, as with any animal model, mouse models do not necessarily accurately reflect all aspects of haemostasis in humans.
In mice, it is not trivially easy to produce bleeding in a reproducible and quantifiable manner. While a tail-clip bleeding time is certainly abnormal in haemophilic mice, both the bleeding time and volume bled are highly variable. Thus, it can be difficult to produce statistically significant data in this model. The haemophilic mice also do not experience the spontaneous joint and soft tissue bleeding characteristic of human haemophilia [10]. The available animal models have been used primarily to assess the haemostatic efficacy of replacement and bypassing therapies. However, they have only been used to a limited extent to study other consequences of the bleeding tendency in haemophilia.
Discussion
What do we know about how bleeding influences tissue healing?
Joint haemorrhage Bleeding into joints is a phenomenon of great clinical interest. However, it has been hard to develop reproducible models that reflect the orthopaedic complications of human haemophilia. Most of the animal models do not develop spontaneous joint haemorrhage in a high proportion of individuals. Therefore, it has been necessary to develop a reproducible means of inducing joint trauma to produce haemorrhage. Valentino’s group has developed a controlled trauma model of joint haemorrhage in haemophilic mice [11] that leads to the development of haemophilic synovitis and joint destruction [12].
It is recognized that the repeated extravasation of blood into the joint cavity is the inciting factor for synovial and cartilage changes in haemophilic arthropathy [13,14]. Synovial changes are thought to precede cartilage changes. The progressive accumulation of iron from red blood cells during successive intra-articular haemorrhages triggers synovial inflammation and may be a direct stimulus for the proliferation of synovial cells [15,16]. Free iron clearly promotes inflammation and cellular proliferation and is thought to play a role in the synovitis and excess angiogenesis observed in patients with haemophilia following repeated haemorrhage into a ‘target joint’ [15–17]. Such a joint that has suffered repeated haemorrhage not only develops synovial proliferation but also florid angiogenesis that serves as a source of repeated bleeding.
Synovial iron deposition as a result of intra-articular haemorrhage is also found in other joint disorders such as pigmented villonodular synovitis, haemangiomas of the synovial membrane and haemosiderotic synovitis. These disorders are all characterized by joint damage resembling haemophilic arthropathy [18,19]. Thus, it appears that deposition of iron in the joint space is a critical factor in bleeding-related arthropathy. In a sense, the joint damage following haemorrhage is a perversion of a normal tissue response to injury, but bleeding into a closed joint space is not a good general model for tissue responses to impaired haemostasis at other sites of injury.
Normal wound healing There are good theoretical reasons to expect that impaired haemostasis will lead to impaired wound healing. Normal wound healing has been divided into four overlapping phases: (i) haemostasis, (ii) inflammation, (iii) proliferation and (iv) remodelling or resolution. The haemostatic phase commences as soon as vessels are damaged during wounding. Platelets adhere to the extracellular matrix and provide primary haemostasis as well as releasing growth factors during degranulation. Activated platelets provide the surface on which coagulation enzymes are activated, leading to a burst of thrombin generation and formation of a stable fibrin clot. Thrombin and the fibrin clot formed during the process of haemostasis both play important roles in healing. Thrombin has cytokine and growth factor-like activities that promote chemotaxis and proliferation of monocyte/macrophages, keratinocytes, fibroblasts and endothelial cells [20–23]. While many other factors can influence fibrin clot stability, variations in the rate of thrombin generation can impact the effectiveness of healing through their effect on fibrin structure [24]. The fibrin clot forms the scaffold on which cell influx and tissue repair takes place, and its degradation products also have chemotactic activity [25]. Thrombin not only clots fibrinogen, but also activates factor XIII (FXIII) and the thrombin-activated fibrinolysis inhibitor (TAFI) to enhance clot stability. Thus, the haemostatic process not only stops blood loss, but also delivers biologically active molecules to the wound site and sets the stage for the subsequent phases of healing.
The inflammatory phase begins with the influx of neutrophils that starts within minutes after injury. While neutrophils play an important first line of defence against bacterial invasion, a lack of neutrophils does not lead to a defect in wound closure or connective tissue repair [26]. The neutrophil influx is followed in the first day after injury by the influx of monocytes from the blood into the tissues where they rapidly mature into macrophages. The macrophages degrade debris and also coordinate healing and immune responses by production of cytokines [27]. Collagen synthesis and angiogenesis are strongly influenced by macrophage-derived cytokines.
The proliferative phase is characterized by epithelial proliferation, angiogenesis and fibroblast proliferation as the fibrin clot and debris filling the wound site are replaced by granulation tissue. The ‘granulation tissue’ is a highly cellular reparative tissue. It is very delicate and bleeds easily because it contains large numbers of new vessels but little stabilizing connective tissue. During development of granulation tissue, capillary sprouts invade the wound clot from adjacent vessels and organize into a microvascular network. At the same time, fibroblasts proliferate within the granulation tissue and begin to lay down collagen and other connective tissue components to reinforce the tensile strength of the repaired tissue.
The remodelling phase can continue for weeks or months as proliferation stops, the inflammatory infiltrate resolves, many of the vascular sprouts regress, and the structure of the wound site collagen reorganizes. The bulging contour of the wound site returns to normal or even retracts as the cells are replaced by acellular connective tissue (scar).
Impaired healing in haemophilia A limited number of studies have directly addressed the question of how haemostatic clot formation influences subsequent healing. Cutaneous wound healing has been reported to be histologically abnormal in knockout mice deficient in fibrinogen [28] or the TAFI [29]. These reports suggest that the presence of a stable fibrin framework is necessary for normal restoration of tissue structure by providing a scaffold for the migration of stromal cells. However, in these models, the time to healing was not significantly delayed. By contrast, healing after tooth extraction is delayed in anticoagulated rabbits [30]. Thus, it appears that an intact thrombin-generating system drives the tempo of wound closure, but the fibrin clot structure plays a role in directing the morphology of tissue restoration. Results such as these suggest that wound healing wound be delayed and morphologically abnormal in haemophilia. However, only clinical experience has suggested poor wound healing in haemophilia until recently.
Our group has recently demonstrated that wound healing is, in fact, impaired in a mouse model of haemophilia B (HB) [31]. The haemophilic mice exhibit delayed cutaneous wound closure with abnormal histology, including: (i) subcutaneous haematoma formation near the wound site; (ii) delayed macrophage influx; (iii) delayed reepithelialization; and (iv) a surprising increase in wound site angiogenesis [31].
Studies in rabbits that were depleted of plasma fibrinogen either before or after the formation of a haemostatic clot suggested that impairing fibrin formation after initial haemostasis did not impair wound healing [32–34]. By analogy, we hypothesized that restoring the initial haemostatic burst of thrombin generation in FIX-deficient mice would normalize the subsequent phases of wound healing. Instead, we found that temporarily restoring thrombin generation with a single dose of FIX replacement or FVIIa bypassing therapy did not correct the delayed epithelial closure time [35]. However, it did lead to an earlier macrophage (MP) influx in both FIX- and FVIIa-treated HB animals compared with untreated HB mice. Thus, it appears that two major mechanisms operate to promote MP influx to a wound site: (i) thrombin plays an important role in promoting rapid MP influx in wild type (WT) mice; and (ii) haemorrhage promotes MP influx into the wound area in HB mice [31]. The pattern of MP influx in the treated HB mice is a composite of the pattern seen in WT and HB mice: an early influx of MP is related to thrombin generation during haemostasis and a late influx of MP occurs in response to recurrent haemorrhage.
The pattern of MP influx was linked to the clearance of iron from the sites of wounding. MPs ingests and degrades red blood cells and other debris at the site of injury. Wound site MP show enhanced expression of heme oxygenase that degrades the heme prosthetic group on haemoglobin and releases iron. Free iron has a great potential to promote tissue damage by transferring electrons and participating in the generation of highly reactive free radicals. The body has evolved mechanisms to limit the toxicity of iron by binding it to ferritin, which traps it in the ferric oxidation state and prevents it from transferring electrons. The common tissue iron stain only detects iron in the ferric state. Thus, the iron stain is negative in the presence of intact red blood cells. However, iron staining becomes apparent as the MPs degrades red blood cells. In our experiments, the level of tissue iron staining in WT animals reached peak intensity in the wound bed at 4 days after wounding and was no longer visible by 12 days as iron was cleared out of the wound site by MP migration. The iron staining began to appear in the deeper tissues in WT mice by 6 days because of the presence of iron-laden MPs. Iron clearance from the deeper tissues was complete by 16 days after wounding as MPs carried the iron out of the area to draining lymph nodes.
By contrast, untreated HB mice had a slow rise in the intensity of iron staining within the wound bed, peaking at 10 days after wounding. The peak corresponded to the peak in MP influx. Speeding the initial MP influx by treatment with FIX or FVIIa at the time of wounding correlated with an earlier onset of heme degradation by macrophages. Both treated groups showed a faster appearance of iron staining within the wound bed. However, as both treated HB groups experience rebleeding in the tissues during healing, iron persists in the wound beds in spite of more effective MP clearance. Iron staining also persisted and even accumulated in the deeper tissues throughout the 16-day course of our studies.
Quite likely as a consequence of the tissue iron load, the HB wounds had greater numbers of vessel profiles than did their WT counterparts. Restoring initial thrombin generation did not normalize the angiogenic response in HB mice. A single dose of FIX treatment actually further increased angiogenesis. This increased angiogenic response might be due to the combination of enhanced thrombin generation at the time of initial haemostasis with the effects of bleeding that develops after FIX has been cleared. These results suggest the possibility that inadequate haemostatic therapy might actually worsen the subsequent risk of bleeding by boosting angiogenesis.
We conclude that normal wound healing requires adequate haemostatic function for an extended period of time and not only at the time of initial haemostatic clot formation.
Delayed bleeding during healing
In the initial stages of angiogenesis, the mature vessels that provide sites for new vessel sprouting exhibit vasodilatation, increased permeability to plasma proteins, detachment of supporting cells and loosening of the adjacent extravascular matrix [36,37]. The endothelial sprouts migrate through the developing granulation tissue, acquire lumens and fuse with other sprouts to form functional capillary loops [36]. Several factors may predispose to bleeding during this process including destabilization of existing vessels as the sprouts develop, the delicate state of the neovessels and the activity of proteolytic enzymes within the granulation tissue [38]. Even normal mice show extravasation of red blood cells in areas of angiogenesis. This predisposition to bleeding during angiogenesis could partially explain ongoing haemorrhage in the HB mice even after closure of the epithelial defect.
However, at least one additional mechanism contributes to bleeding during angiogenesis: altered expression of tissue factor (TF) after wounding [39]. TF is normally expressed outside the blood vessels [40,41]. A body of in vitro data suggests that products of the coagulation cascade [42] and inflammatory mediators released following injury [43–45] promote TF expression. TF messenger RNA in skin samples has also been shown to be upregulated following cutaneous wounding in a mouse model [46]. Therefore, we expected that TF would be highly expressed in the tissue around a wound site. Instead, we found that TF disappeared from around uninjured vessels near wounds in both WT and HB mice by 1 day after wounding [39]. The TF coat was not re-expressed until 8–10 days after wounding. The angiogenic vessels within the granulation tissue also did not express TF in either WT or HB mice during healing.
Tissue factor around vessels is primarily expressed by pericytes; myofibroblast-like cells that surround the endothelium of capillaries and the muscular layer of arterioles and venules. The normal complement of pericytes is present around dermal vessels and angiogenic vessels, even while TF is absent. Thus, TF appears to be actively downregulated during angiogenesis.
We propose that a lack of TF around angiogenic vessels may reflect a mechanism to prevent thrombosis of newly formed, delicate and leaky vessels. This leads to an increased risk of bleeding from granulation tissue. Such a bleeding tendency is not a significant liability in normal individuals. However, in haemophilia, the absence of TF near wound sites could lead to repeated haemorrhage in and around the wound site, even after the surface defect has been healed.
The phenomenon of late (re)bleeding at sites of angiogenesis has implications for the treatment of haemophilic manifestations. For example, haemophilic arthropathy is characterized by repeated haemarthroses, subsequent synovial hyperplasia, angiogenesis and ultimate destruction of the joint space [47]. The initial insult of bleeding into a joint is similar to the punch biopsy as it sets into motion the cycle of inflammation and angiogenesis that promotes further bleeding. Bleeding during the process of angiogenesis could then lead to a vicious cycle of inflammation, increased angiogenesis and more bleeding.
The most efficient way of preventing inflammation, angiogenesis and recurrent bleeding is to prevent haemorrhage from occurring in the first place. Whereas extended factor therapy may prevent additional bleeding related to angiogenesis, only up-front prophylactic therapy can prevent the initial bleeding episode that triggers the subsequent bleeding cycle. It remains to be seen whether anti-inflammatory or anti-angiogenic therapies might prevent recurrent bleeding and subsequent arthropathy.
