Dr. Carlson is currently a professor of cell and developmental biology as well as a research professor at the Institute of Gerontology at the University of Michigan. He has been conducting research on the regeneration of amphibian limbs and mammalian muscle since 1960.
Since the first scientific publications on regeneration in the early 1700s (Dinsmore, 1991), researchers have studied regenerative phenomena in a wide variety of species, ranging from protozoa to humans. The number of species that are capable of some form of regeneration is immense, and the spectrum of structures that can be regenerated is vast. Some of the champions of regeneration, such as hydra and planaria, can regenerate an entire body from a small fragment. Among the vertebrates, urodele amphibians (newts and salamanders) are renowned for their ability to regenerate almost any structure that can be cut off, including their limbs, tail, lower jaw, ocular lens, retina, ventricle of the heart, small intestine, and many specific tissues.
Although mammals are not endowed with as great an ability to regenerate complex structures as salamanders, they have the potential to regenerate a surprisingly large array of injured tissues. Unlike that in salamanders, in which regeneration, once started, typically results in the formation of an almost perfect replica of the structure that was lost, mammalian regeneration proceeds with varying degrees of success. One of the major challenges in the scientific study of regeneration in mammals and its clinical application in humans is to understand why regeneration proceeds very well under some circumstances and very poorly under others.
At this point, a historical digression might be instructive. When I entered the field of regeneration in 1960, research in the United States and much of Europe was strongly focused on regeneration in amphibians and invertebrates. The neurotrophic theory of Marcus Singer (1959, 1965) dominated the field, and many researchers felt that supplying adequate innervation was the key to stimulating a structure, such as a limb, to regenerate. Shortly after entering the field, I stumbled on some Russian regeneration literature from the 1950s, when Russian biology was still heavily under the influence of Lysenko (Anonymous, 1948; Soyfer, 1994) and some even more unusual theories (Lepeshinskaya, 1945, 1951), which essentially relegated genetics to a dark corner and stressed the importance of the environment in many biological processes. Partially as a result of Lysenkoist theory, many Russian biologists began studying regeneration in mammals during the 1950s (Liosner, 1960, 1963, 1972).
One of the dominant modes of Soviet thought at that time was that mammalian regeneration depended on uslovie, a Russian word that is usually translated by the English word conditions. As a beginning student of the Russian language, that word, as applied to regeneration research, was very confusing to me, but when I became immersed in Russian regeneration biology during the mid-1960s, I realized that it really referred to local environmental conditions. Because of the taint of Lysenkoism, I was initially skeptical of the great emphasis on environment in regeneration. Certainly it did not seem to apply to most regenerative phenomena in amphibians, in which once a certain threshold was crossed, an almost unalterable program leading to nearly perfect replication of the amputated structure was unlocked. Over the years, as I concentrated more of my research efforts on mammalian regeneration and many others have done the same, it has become clear that the local environment is a critical determinant of the success of regeneration in many mammalian systems.
EPIMORPHIC LIMB REGENERATION: THE HOLY GRAIL
One of the most perfect forms of regeneration in vertebrates is that of the salamander limb, and it is the standard against which regeneration in several mammalian systems (limbs, antlers, earhole punches) has been compared. Epimorphic regeneration of an amputated limb proceeds in a well-defined sequence of stages (Table 1 and Fig. 1). One of the issues in assigning a mammalian regenerative process to the epimorphic category is whether or not the process is really parallel to that seen in amphibians. Much of this centers on what constitutes a blastema and whether it has both an appearance and properties that correspond to that of a regenerating amphibian limb.
Table I. Stages in the epimorphic regeneration of an amphibian limb
Epidermal migration to cover the amputation surface.
Demolition and Phagocytosis
Mild inflammatory response and the enzymatic removal of many matrix components just beneath the wound epithelium.
The loss of differentiated tissue types for 1 or 2 mm beneath the wound epithelium and the appearance of embryonic-looking (dedifferentiated) cells in the same area.
The aggregation of the dedifferentiated cells into a structure (blastema) reminiscent of the embryonic limb bud.
Morphogenesis and Growth
Outgrowth of the blastema, and the shaping and differentiation of the mass of blastemal cells into the cells and tissues comprising the normal limb. In larger limbs, after morphogenesis is completed, growth of the miniature regenerate continues until it is the same size as the original limb that was lost.
A number of important principles underlie epimorphic limb regeneration (for comprehensive reviews, see Carlson, 1974a; Wallace, 1981; Stocum, 1995; Tsonis, 1996). The first is that the process requires the active participation of a wound epidermis that is not underlain by dermis or a basement membrane. Exactly how the wound epidermis is important remains under discussion, but some of its important properties are production of enzymes that assist in the histolysis of underlying tissues (Polezhaev, 1936); direct phagocytic removal of tissue debris to the outside (Singer and Salpeter, 1961); production of growth factors (fibroblast growth factor); and possibly serving as a key morphogenetic boundary (Maden, 1977). Later in regeneration, the wound epidermis thickens into an apical cap, possibly homologous to the apical ectodermal ridge in the embryonic limb bud, and it is important in stimulating outgrowth of the blastema.
Dedifferentiation has been one of the most important principles in limb regeneration. Epimorphic regeneration does not proceed in the absence of dedifferentiation, but what is meant by dedifferentiation? At the tissue level, there is no doubt that mature muscle and skeletal tissues lose their differentiated properties and are replaced by a population of embryonic-appearing cells. What happens at the cellular level has been the source of great controversy over the years. According to one school of thought, the differentiated tissues or cells, such as multinucleated muscle fibers, dissociate into mononucleated cells that form the blastema and reacquire the capacities to differentiate into other cell types within the regenerating limb. Another school of thought denied this possibility and suggested that the source of the regenerate was local or distant reserve cells that become activated to participate in blastema formation (for extensive discussion, see Mauro et al., 1970). Although the concept of cellular dedifferentiation was very unpopular for many years, research performed principally on skeletal muscle over the past decade has shown conclusively that mature skeletal muscle fibers can break up into individual cellular units and that the nuclei regain the ability to enter the mitotic cycle (Brockes and Kumar, 2002; Echeverri and Tanaka, 2002; Odelberg, 2002). Dedifferentiation in amphibians appears to be closely tied to overall morphogenetic controls.
Epimorphic regeneration of a limb is strongly dependent on its nerve supply (Singer, 1952), and the important principle is that an adequate supply of nerves, regardless of type, is required for the initiation of the process. In the absence of a quantitatively sufficient nerve supply, a blastema will not form, or if a regenerate in the early blastemal stages is denervated, regeneration ceases. On the other hand, once the regenerate has progressed to more advanced stages, where some morphogenesis is apparent, limb regeneration is nerve-independent and can continue if the nerve supply is interrupted.
Interestingly, the earliest stages of amphibian limb regeneration occur in a largely avascular environment (Peadon and Singer, 1966; Rageh et al., 2002). Only as the blastema has begun to form and grow out does a network of capillaries begin to appear within the regenerate. How the complex interactions that occur during dedifferentiation and the earliest stages of blastema formation can take place in a poorly vascularized environment is not understood.
One of the hallmarks of amphibian limb regeneration is the near-perfect replication of pattern. Despite a number of good models of pattern formation in epimorphic regeneration (French et al., 1976; Maden, 1977), this remains one of the least well understood aspects of limb regeneration. Nevertheless, experimentation over the years has delineated several important principles. One could be called the principle of opposites. In animals from planaria to salamanders, the approximation of cells with opposite morphogenetic qualities (e.g., dorsal to ventral or anterior to posterior) is important even for the initiation of regeneration (Bryant et al., 1981; Agata et al., 2002). Another important principle is that morphological integrity of the stump is not necessary for integrity of the regenerate. If entire bones or most of the muscles of the limb stump are removed, the regenerate nevertheless produces a completely normal limb, even though the original defect in the stump remains (Thornton, 1938; Carlson, 1972).
Many epimorphic systems, including the regenerating amphibian limb, operate on the basis of intercalation (Bohn, 1970; French et al., 1976; Maden, 1977). As a model, it is commonly assumed that a graded array of morphogenetic information exists along each of the limb axes and that if a discontinuity is created, the limb responds by regenerating new cells until the discontinuity is filled in. This is very well illustrated by experiments conducted on cockroach limbs (Fig. 2). The phenomenon of intercalation is based on positional memory (Carlson, 1983), a very poorly understood property that in essence means that cells of the amphibian limb are somehow marked in accordance with their position in the limb. When a discontinuity is created or they are moved, these cells “remember” their original positions and influence morphogenesis accordingly.
What has often been called “the law of distal transformation” states that the regenerate will produce only structures distal to the level of amputation. If a limb is reversed in a proximodistal direction (Butler, 1951; Carlson et al., 1974), the regenerate produces only structures distal to the amputation surface even though the regenerated structures duplicate those of the reversed stump (Fig. 3). This longstanding rule is violated when amputated limbs are exposed to retinoids (Niazi and Saxena, 1978; Maden, 1982). After treatment with vitamin A, a limb amputated at the wrist can produce a regenerate with complete upper arm and forearm segments before forming a new wrist and hand. It is now known that retinoids can reset morphogenetic boundaries from distal to proximal, from anterior to posterior, and from dorsal to ventral. This may be accomplished by changing the surface properties of the blastemal cells (Crawford and Stocum, 1988) or by altering the expression pattern of a newly described gene, Prod 1, which is normally more strongly expressed in proximal than in distal locations in the limb (Morais da Silva et al., 2002).
These principles outlined above represent some of the properties of epimorphic systems that must be met if limb regeneration in mammals is to be stimulated. One of the biggest questions is whether the tissues of a mammalian limb have lost their intrinsic capacity to regenerate. Conversely, does the potential for dedifferentiation, blastema formation, and morphogenesis by an embryonic pattern-forming mechanism still exist but cannot take place because of the loss of some coordinating mechanism or repression by a too efficient wound-healing capability that chokes up the end of the limb stump with scar tissue?
REGENERATION OF MAMMALIAN TISSUES
Many mammalian tissues possess a remarkable capacity to regenerate as individual components of the body. These include bone, skeletal muscle, peripheral nerve, and urinary bladder. One of the most remarkable examples of mammalian regeneration is that of the antlers of deer and moose, which can grow at a rate of over an inch per day at the peak. Blood vessels also exhibit tremendous regenerative growth into a former area of necrosis. Some internal organs, such as the liver, have remarkable powers of regeneration, but their mode of restoration involves mainly hypertrophic mechanisms and is outside the scope of this review.
How well a damaged mammalian tissue regenerates depends on many factors. One of the most important principles is that for almost all tissue regeneration, the presence of a blood supply is crucial for initiating the process. In addition to providing oxygen and nutrients, this brings in macrophages and other phagocytic cells, which not only remove cellular and tissue debris, but also stimulate division of precursor cells through the secretion of a variety of cytokines and growth factors. The presence of an angiogenic hypoxia-inducible factor (HIF) complex that is activated by damaged and ischemic tissues (Maxwell and Ratcliffe, 2002) is an important adaptation that allows damaged tissues to initiate some components of their own repair process. HIF increases the transcription of vascular endothelial growth factor (VEGF), which is one of the most important early signals for the budding of endothelial cells from existing vessels at the edge of a wound.
Most regenerating mammalian tissues take origin from progenitor cells associated with that tissue. For example, the principal source of myoblasts in regenerating skeletal muscle is the satellite cell, a mononuclear cell located between the muscle fiber and its basal lamina (Schultz and McCormick, 1994; Chargé and Rudnicki, 2003). The dedifferentiation of multinucleated muscle fibers into mononuclear cells capable of proliferating has not been documented in normal mammalian skeletal muscle regeneration, although this has been obtained in the laboratory (Odelberg, 2002). Marrow-derived stem cells appear to play a relatively small role in mammalian muscle regeneration; their contribution appears to be less than 1% of the nuclei in the regenerated myofibers. In bone regeneration, the periosteum is the source of many of the osteogenic cells, but inside a long bone, the proximity of the endosteum to the marrow cavity makes it difficult to sort out the relative contributions of endosteal vs. marrow-derived cells. Certainly in the formation of ectopic bone in soft tissues, either local or marrow-derived stem cells represent the most logical sources of the new bone. Although structures sometimes labeled blastemas are formed in areas of healing and regenerating bone, they bear no resemblance to the blastemas that are the precursors of amphibian limb regenerates. This is simply the case of a good term (e.g., metanephrogenic blastema in kidney development) being used in connection with a different process.
A distinct difference between the regeneration of mammalian tissues and epimorphic regeneration is that the initiation of regeneration does not depend on innervation. However, in the case of skeletal muscle, final structural and functional differentiation are incomplete in the absence of motor innervation. In some species, e.g., frogs and mice, regenerating muscle does not proceed past the myotube stage, because these immature multinucleated precursors of muscle fibers undergo atrophy, probably based on apoptotic mechanisms, and disappear, leaving only a mass of connective tissue (Mufti, 1977).
In mammalian tissue regeneration, morphogenesis of the regenerated tissue is accomplished principally through the action of physical factors on the regenerate, rather than through the positional and morphogenetic field influences that operate in the regenerating and developing limb. Healing and regenerating bone is exquisitely sensitive to mechanical influences. In long bones, these influences appear to be transmitted through a complex network of interconnecting osteocytic cell processes within the canaliculi of the Haversian systems (Cowin and Moss, 2000). These signals may be mediated by the extracellular fluid in the canaliculi and/or by changes in electrical potential in the bone. New bone is deposited in areas of compression (the concave side of a nonaligned fracture), and net removal occurs on the convex side, where tensile forces are more prominent. Areas subjected to pressure have a net negative charge, whereas those subjected to tension are electropositive. New bone is deposited at the negative pole (cathode) of an applied electric field on bone (Yasuda, 1974). Experimental studies on regenerating muscle have shown that both external form and internal architecture can be accounted for on the basis of tension transmitted through the regenerating tendons and lateral pressures from the surrounding tissues (Carlson, 1972: chap. 6).
One important characteristic of mammalian tissue regeneration is that it occurs relatively more quickly than does epimorphic regeneration. Even in the amphibian limb, tissues that participate in epimorphic regeneration, such as muscle and skeleton, regenerate by the tissue mode more quickly than they would during an epimorphic regenerative process (Carlson, 1970, 1979). Some researchers have theorized that mammals do not have the luxury of spending the time required to regenerate structures epimorphically and that the more rapid tissue regenerative response is a more effective substitute even though the final regenerate may not be as perfectly formed as it would have been had an epimorphic process been operative.
ROLE OF LOCAL ENVIRONMENT IN MAMMALIAN TISSUE REGENERATION
Almost every stage of a mammalian regenerative process can be strongly influenced by the local environment. The recruitment of tissue-specific progenitor cells can be stifled by an inadequate local microcirculation and low oxygen. Growth factors released from the surrounding matrix can play an important role in supporting or not supporting the proliferation of the progenitor cells. The local extracellular matrix serves as a critical scaffolding for the early stages of regeneration (Badylak, 2002, 2004). It can provide an environment conducive to seeding by hematogenous stem cells; it can provide early orientation for the regenerating tissue; and in the case of skeletal muscle, the persisting basal lamina surrounding the damaged muscle fibers serves as a selective filter that keeps fibroblasts out, keeps most satellite cells in, and allows the relatively free penetration of macrophages.
As a regenerating tissue is taking shape, the mechanical environment can serve as a major stimulus for growth. In bone, the technique of distraction osteogenesis (Samchukov et al., 2001) involves creating a discontinuity in an existing bone and separating the two ends with a special apparatus that allows the two ends of the bone to be gradually moved apart. This causes the forming callus to increase in size to accommodate the increasing gap. When the bone ends are distracted (moved apart) to the desired distance, the callus fills in the gap and forms new bone. By this technique, shortened femurs of dwarves can be lengthened and hypoplastic bones of the face can be expanded to a more normal form. The soft tissue expansion techniques commonly used in plastic surgery involve the same principle, although in the case of skin, for instance, the expansion is often done on undamaged issue. Tension can also be used to elongate regenerating mammalian muscle. After most of the gastrocnemius muscle of the rat is excised, the proximal stump remains the same size if it attaches to nearby local tissues. On the other hand, if the Achilles tendon regenerates proximally and attaches to the proximal stump, the muscle elongates to 4–5 times its original length (Carlson, 1974a). Similarly, ongoing tension applied to a regenerating minced muscle pulls it into an elongated tongue-like structure (Fig. 4). These are examples of what could be called distraction myogenesis, but the technique has not yet been clinically applied to muscle.
The influence of the local environment is readily apparent in axonal regeneration. In an injured peripheral nerve, the axon degenerates a certain distance proximal to the site of damage. The basal lamina surrounding the Schwann cells persists and serves as the major guide for the regenerating axons. The Schwann cells themselves constitute a significant part of the local environment, because they proliferate and secrete growth factors that stimulate axonal elongation. In addition, the Schwann cells express cell surface molecules (e.g., N-CAM and N- and E-cadherin) that interact with receptors on the regenerating axons to promote elongation (Ide, 1996).
At the termination of axonal regeneration on a muscle fiber, local environmental factors again play an important role. The growth cones of the axons are strongly attracted to the original neuromuscular junction sites on the muscle fiber basal lamina even if the muscle fiber is not present (Sanes et al., 1978). In the regeneration of axons over long distances and over long periods of time, the nerve sheaths become filled with densely packed collagen fibers, which strongly impede or prevent further regeneration. Peripheral nerve surgeons are very aware of the importance of the local environment as they perform nerve repairs. Typically, when a nerve is severed, the two ends retract, and some type of bridge must be introduced to provide a channel to guide the regenerating axons from the end of the proximal stump into the cut end of the distal stump. Many different materials, ranging from pieces of muscle to cuffs of artificial material, have been used in attempts to find the most suitable substrate for the regenerating nerve fibers.
One of the most striking demonstrations of the importance of the local environment on the course of a mammalian regenerative process was the experimental approach of Aguayo to the problem of regeneration in the central nervous system. In most parts of the brain and spinal cord, lesions of tracts result in virtually no regeneration. In a breakthrough experiment, David and Aguayo (1981) inserted segments of peripheral nerve sheaths at the end of lesioned tracts in the central nervous system and demonstrated that the central axons regenerated well as long as they were traversing the peripheral graft (Fig. 5), but once they reached the end of the graft, they were again confronted with the normal environment of the CNS, and regeneration effectively ceased. Subsequent research showed that breakdown products of central myelin are strongly inhibitory to axonal extension (Berry, 1982). Later the inhibitory protein, Nogo, along with other proteins were identified as specific inhibitory molecular components resulting from the breakdown of myelin (Filbin, 2003; Spencer et al., 2003; Schwab, 2004).
The formation of local scar tissue can dramatically inhibit most mammalian tissue regenerative processes. In the case of skin, the scar is derived from the activities of myofibroblasts that are derived from fibroblasts that move into the wound area. In badly damaged muscle, local fibroblasts produce massive deposits of collagen fibers that strongly inhibit the formation of new muscle or, at a minimum, interfere with functional restoration. Within the central nervous system, glial scars, mainly of astrocytic origin, strongly inhibit the regeneration of central axons.
One of the more pervasive environmental influences on regeneration in mammals is age. Many physiological functions slow down with aging, and regeneration is one of them. Just remember how quickly a simple skin wound on your own body healed when you were a child, and compare your response today. For those over 50, the difference is very apparent. Laboratory studies have verified that both the rate and overall success of wound healing and regeneration decrease with increasing age. They have also shown that this trend is not irreversible. Because it is relatively easy to quantify changes in regenerating skeletal muscle, this tissue will be used as an example.
During the normal aging process, muscles become smaller and weaker (Larsson, 1982; Booth et al., 1994; Brooks and Faulkner, 1994). In fact, in very old age the musculature seems to melt away, a phenomenon called sarcopenia (Doherty, 2003; Kamel, 2003; Marcell, 2003). A number of specific factors underlie these gross aging changes. They include loss of muscle fibers, reduction in innervation, changes in the local microcirculation, and changes in the overall humeral environment in aged individuals.
There is virtual unanimity in the literature that old muscle in rodents regenerates significantly more poorly in old than in young individuals (Carlson and Faulkner, 1989; Grounds, 1998; Marsh et al., 1998; Conboy et al., 2003). This decline has been attributed to several factors that affect the local environment. They include the quality and quantity of innervation, the extracellular matrix (increase in interstitial collagen), a reduction in the microvasculature and local or systemic humoral factors. One of the potential targets of some of these environmental influences is the satellite cells, which decline in both number (Allbrook et al., 1971; Gibson and Schultz, 1983; Renault et al., 2002; Kadi et al., 2004) and proliferative potential (Schultz and Lipton, 1982; Conboy et al., 2003).
Several studies have shown that the age-related decline in the success of muscle regeneration is not fixed, but can be strongly influenced by environmental factors. Carlson and Faulkner (1989) used a cross-age transplantation model between young adult and old rats as a test of the role of environmental factors as determinants of regenerative success. Same-age control muscle grafts in young rats generated almost three times the maximum tetanic force of autografts in old rats. When old muscles were transplanted into young hosts, they regenerated as well as young muscle autografts, whereas when young muscles were grafted into old hosts, they regenerated no better than old muscle autografts. This experiment showed that old muscle can regenerate well if provided with a favorable environment. A more recent experiment (Conboy et al., 2005), involving the creation of heterochronous parabiotic mice, showed that muscle regeneration in the old member of the parabiotic pair was improved over that seen in control old mice. This suggests that something in the blood or overall humoral environment can exert a positive influence on regeneration in old animals and possibly a negative influence on regeneration in the younger member of the pair. A major question is what constitutes a favorable environment, and what is missing in the old body.
Experiments by several different groups have indicated that multiple environmental factors operating at different levels may be involved in the decline in regenerative success with increasing age. At the level of satellite cells, the reversal of the age-related decline in the Delta-Notch signaling system has been shown to improve regeneration significantly (Conboy et al., 2003, 2005). Similarly, reversal of an age-related decline in levels of insulin-like growth factor I (IGF-I) results in improved muscle regeneration in old animals (Barton-Davis et al., 1998; Musaro et al., 2001). Among the late-acting influences on muscle regeneration, innervation has been identified as a significant factor accounting for some of the deficit in regeneration with increasing age (Carlson, 1995; Carlson and Faulkner, 1996, 1998).
The examples outlined above provide abundant evidence in a number of different systems that regeneration is strongly influenced by both the systemic and the local environment. One of the most important messages is that the regenerative potential of a tissue is frequently higher than would be indicated by its actual degree of regeneration under a given set of circumstances. One of the major challenges for those investigating regeneration in mammals is to determine what factors might be limiting the degree of regeneration of a tissue and how to alter these factors so that the inherent regenerative potential of the tissue is released.
PROSPECTS FOR APPLICATION OF PRINCIPLES TO REGENERATION IN HUMANS
A surprisingly large number of general principles of the type referred to above are already being applied to the treatment of trauma and guided regeneration in humans. One of the first principles is that a cellular source of new tissue is required. With a greater understanding of the natural progenitor cells in many tissues, greater effort is taken to preserve these or to expand them in vitro before implanting them into the desired areas. A notable example of the latter has been the attempts to strengthen dystrophic muscles by implants of dystrophin-competent myoblasts. The tremendous current emphasis on stem cell biology is likely to yield viable techniques for reconstituting deficient organs as diverse as pancreatic islets and specific nuclei in the brain.
Any implanted cellular material must soon be connected to a vascular supply in order to retain viability, and the years of research on angiogenesis factors have yielded results that are presently being applied clinically on an experimental basis. Another major requirement for successful tissue regeneration is an appropriate substrate. Already many patients have had implants of natural (often xenogeneic) or artificial matrices either to stimulate or facilitate regeneration of various complex structures. Attempts to reduce scar tissue formation also fall into the category of providing an appropriate matrix.
The field of orthopedics has been a leader in attempts to stimulate the formation of tissues to replace damaged ones. Methods of inducing bone have evolved from grafts of whole bone tissue to fragmented bone to the present practice of inducing bone formation with localized application of bone morphogenetic protein (BMP)-2 or the application of electrical fields. With the technique of distraction osteogenesis, bone regeneration can be accurately guided along predetermined courses. Application of a similar principle in distraction myogenesis could be used to lengthen individual injured muscles.
Many new techniques are being devised to apply basic principles to traumatized or pathological regions in the central nervous system. These involve altering the natural postinjury environment with the application of antibodies to neutralize inhibitors of regeneration, the local application of growth factors, the insertion of new substrate materials, and the introduction of neural stem cells.
In the case of regeneration of extremities, clinical experience in the 1970s demonstrated that for fingertip amputations, the best thing to do is to leave the finger alone, rather than surgically covering it with a skin flap. This preserves the wound epidermis and in children results in almost perfect fingertip regeneration. To date, there has been little progress in stimulating the regeneration of any portion of an extremity amputated proximal to the terminal interphalangeal joint. This remains one of the grand challenges for the future.