Section 2.1. The origins of mammalian ESC

First attempts at preparing ESC were made by Cole et al. (1965, 1966) and since then, several books have been devoted to their genetics and developmental properties. Derived from cells of inner cell mass (ICM) isolated from rabbit blastocysts, ESC were found to be multipotent, i.e. capable of repopulating literally every body tissue including muscle, blood islands, neurons, phagocytes, among many others. Similar work on human ESC required research on human blastocysts, a step deemed as immoral by various ethicists (e.g. Edwards, 1982, 2007). Their properties have since been covered in numerous books and articles (e.g. Edwards and Brody, 1995; Marshak et al., 2000; Lanza et al., 2004) so a few selected papers will be chosen for discussion in this review.

Initial attempts to produce ESC involved the excision of the ICM from rabbit blastocysts. Cells in these ICM were then disaggregated and cultured in Petri dishes coated with collagen, sometimes with a feeder layer of HeLa cells (Table 1) (Cole et al., 1965, 1966). Collagen substrates were added, since they sustained dividing stem cells in vitro, whereas the addition of erythropoietin to culture media was ineffective. ICM cells grew well in vitro, retaining their characteristics throughout endless cell divisions. Some formed a large central mass of cells in the culture dish, whereas others migrated from the ICM to form outgrowths of differentiating tissues. Numerous ESC lines originated from these outgrowths, although many soon perished. Those that survived grew through >200 generations, a hint of immortality, and formed two distinct subgroups, one type being fibroblastic and the other epithelioid (Figure 1).

Fig. 1 Early attempts at producing cell lines and differentiating tissues from disaggregates of rabbit and mouse ICM (Cole et al., 1965, 1966; Cole and Paul, 1965). Cell strains were grown from disaggregated rabbit ICM: (a) a primary rabbit epithelioid cell strain; (b) a primary rabbit fibroblastic cell strain; (c) diploid chromosomes in the fibroblastic rabbit cell strain; (d) epithelioid rabbit cells transformed to a characteristic narrow shape when exposed to mesodermalizing inducer. These cells resumed their fibroblastic shape when placed in normal culture medium. Structures formed in vitro after mouse ICM were placed in culture and (e) spherical colonies formed from isolated ICM cells of mouse blastocysts 10 days after their attachment to the culture dish – note that these do not resemble the mouse embryoid bodies shown in Figure 3; (f) a section was taken through the cell masses shown in (e) – note that development was disorganized and that more blastocysts developed in this fashion if ATP concentrations were raised.

Rare markers were identified in long-term stem cell lines, some producing alkaline phosphatase and others producing arginase. Each type retained its characteristics throughout the culture period, other genetic markers being unavailable in the 1960s. Individual stem cell lines remained mostly diploid over numerous generations, even after they had been cryopreserved over many years (Cole and Paul, 1965; Paul and Hunter, 1969). Human embryonic stem cells were later cryopreserved by improved methods including vitrification (Reubinoff et al., 2001).

Zona-free rabbit blastocysts were also cultured in vitro. Outgrowths of trophectoderm rapidly attached to their culture dishes and formed a thin cell layer. This enabled ICM cells to migrate over them and differentiate into various tissues. Their multipotency was apparent since, after several days, these outgrowths contained cells typical of the three germ layers including blood islands, muscle cells, neurons, phagocytes and other tissues, although germ cells were not identified (Figure 2) (Cole et al., 1965, 1966). The fact that so many different cell types could be produced from ESC in vitro led to first thoughts on the therapeutic grafting of human ESC to sick recipients. The one exception was germ cells, which were not identified in the outgrowths, although Gardner (1968) later identified them in chimaeras formed by injecting ESC into mouse blastocysts.

Fig. 2 Cell colonies derived from intact zona-free rabbit blastocysts after 20 days in culture (Cole et al., 1966). (a) Cell masses derived from a zona-free rabbit blastocyst. (b) Muscle differentiating after continued culture. (c) A single blood island that developed among cell outgrowths from rabbit ICM. (d) A group of neurons in the same outgrowths. (e) A complex group of differentiating cells in which muscle cells are mixed with various types of differentiating cells; a pathologist discerned several cell types in this mass of cells.

Similar cells were later prepared from outgrowths of mouse ICM (Cole and Paul, 1965). They developed in a somewhat similar manner as in rabbits, as they attached to the culture dish. They differed from rabbit ES cells in forming large rounded lumps of tissue, which were later named embryoid bodies (). This work was repeated when Evans and Kaufman (1981) obtained ES cells in outgrowths of mouse blastocysts although they failed to establish cell lines. Surprisingly, these workers failed to refer to the earlier work on rabbits by Cole et al. (1965, 1965) even though their studies were virtually identical. Wider research had enabled Gardner and Edwards (1968) to introduce preimplantation genetic diagnosis by excising a few trophectodermal cells from rabbit blastocysts and correctly diagnosing them as precursors of male or female offspring. This research had opened the field of preimplantation genetic diagnosis, which made it possible to check inherited characteristics in blastocysts, and discard those carrying anomalies.

Prospects of preparing human ESC for clinical use became reality in 1978, when endless research on human IVF had finally opened a reliable source of human blastocysts (Table 1, Figure 4). The healthy birth of the world's first test-tube baby later confirmed that human embryos grown in vitro could be safely used for preparing stem cells (Steptoe et al., 1971, 1980; Edwards et al., 1980; Findliki et al., 2006). Immense amounts of work now led to the introduction of human IVF, and the growth of healthy human blastocysts in vitro. It was these which led to the birth of Louise Brown, the first ‘test-tube’ baby. Her normality confirmed the fact that embryos grown in vitro were fully capable of normal growth and could be used to make ESC.

Fig. 4 Immense opportunities were opened by the availability of human blastocysts at a time when marker genes were first identified (Steptoe et al., 1971; Edwards, 2004a,b). Several fundamental studies on blastocysts were now possible as depicted in this figure: (i) human blastocysts were available for IVF or for studies on early embryonic differentiation; (ii) ICM in mouse blastocysts could be excised and replaced with replacement ICM or disaggregates of ICM cells, their marker genes enabling many tissues to be traced in resulting chimaeras. This technique provided a reliable technique to estimate the potency of the grafted cells; (iii) stem cells isolated from ICM or from various tissues and injected into tail veins of lethally X‐irradiated mice saved them from death (Hollands, 1987); (iv) ESC were injected singly or in groups directly into the blastocoelic cavity to form chimaeras with the host blastocyst. Many chimaeras developed to full term, and the multipotency of their grafted cells could be assessed by their degree of colonization of various tissues in the chimaeras (Gardner, 1968). Mouse and human ICM cells were now available to prepare ESC, the cells mostly being used for grafting in the early days of regenerative medicine. Small amounts of tissue excised from mouse and human blastocysts could be used for sexing embryos or for genetic analyses aimed at discovering if the host blastocyst carried disease genes or abnormal numbers of chromosomes.

An attempt at preparing human ESC in vitro was made by Fishel et al. (1984). As in rabbits and mice, trophectoderm attached to the plastic culture dish and ICM outgrowths migrated over it. These cells released human chorionic gonadotrophin (HCG) into culture media, which indicated that trophectoderm and perhaps some ICM cells had differentiated into ESC. Most cultures were unfortunately short-lived, and this work had to be prematurely abandoned due to ethicolegal difficulties over the use of patients' embryos.

Several years later, stem cells were derived from human blastocysts (Bongso et al., 1994). Thomson et al. (1995) then grew ESC derived from primate blastocysts. Thomson et al. (1998) also prepared human ESC from IVF blastocysts. By today, numerous lines of human stem cells have been prepared, many of them held by the International Stem Cell Initiative (2007).

Section 2.2. Properties of ESC

Human ESC are known to differentiate in vitro to various tissues under the influence of various inducers (Schuldiner et al., 2001; Edwards, 2004a,b). Early research on human ESC revealed that they differed from mouse ESC, which needed a gelatin substrate and leukaemia inhibitory factor (LIF), whereas human ESC required fibroblastic feeder layers and basic fibroblast growth factor (bFGF). Other genetic factors were also involved in the development of ESC in vitro (). The properties of ESC were also reviewed by Trounson (2002) who considered them to be the ‘mother of all cell and tissue types’.

Mouse ESC have also been grafted to recipients for therapeutic purposes, sometimes across species barriers. For example, Hollands (1987) saved the lives of lethally irradiated mice by grafting mouse or rat ESC (Figure 4). Other recent investigations include Ménard et al. (2005) who grafted cardiac-committed mouse ESC to sheep showing symptoms of myocardial infarction, and which soon recovered after grafting. ESC also proliferated as they migrated to the brain and other organs (Srivastava et al., 2006). A trial by Barberi et al. (2007) involved deriving multipotent mesenchymal precursor cells and skeletal myoblasts from human ESC by means of selective culture conditions and cell sorting. Myoblasts were among the differentiating cells that maturated in vitro, an observation confirmed by the formation of twitching myotubes. Their viability long after being grafted to the tibialis anterior muscle in mice revealed a considerable potential for future therapies (Barberi et al., 2007).

ESC have proved to be valuable in other ways. One example is their beneficial treatment for age-related macular degeneration, which affects many older people and often leads to blindness. It is caused by the failure of pigmented epithelial cells in a support layer under the retina, which, if damaged, results in this disorder. Ocular surface disease is another problem, and Schwab et al. (2000) successfully transplanted bioengineered tissue replacements prepared from stem cells taken from the healthy periphery of the eye to alleviate symptoms in affected areas. Eye stem cells selected for this purpose may soon be replaced by ESC.

The work of Gardner (1968), described above, revealed how mouse ESC could form germ cells. This work was confirmed in later studies when Hübner et al. (2003) reported that mouse ESC cultured in vitro could form oocytes after 50 days in culture. These authors also demonstrated that mouse ESC could differentiate in vitro to form embryoid bodies that produced cells resembling primordial germ cells (PGC) (Figure 3). These cells differentiated into spermatozoa and expressed characteristic genetic markers of PGC in vitro. They also formed yolk sac sperm cells capable of fertilization and differentiation to blastocysts (Geijsen et al., 2004).

Section 2.3. Embryonal carcinoma cells

Another group of stem cells, termed embryonal carcinoma (EC) cells, were propagated from mouse teratocarcinomas. They expressed stem cell markers including Oct‐4, alkaline phosphatase, the glycolipids SSEA‐3 and 4, and the keratin sulfate antigens TRA‐1–60 and TRA‐1–81 (Table 1). These were later found to be reliable markers of ESC (reviewed by Solter, 2006). Mouse cell lines could also be prepared from PGC since eight of them have formed by the end of gastrulation, and 50–80 were located in mesoderm of the primitive streak. Expressing alkaline phosphatase, they migrated via the allantois to the genital ridge at day 8.5 post-coitum where they could be extracted (Mintz and Russell, 1957; see review by McLaren, 2002).

Numerous spermatogonial stem cells isolated from mouse testes also responded to factors in culture medium by acquiring properties of ESC (Guan et al., 2006). Named multipotent adult germline stem cells (maGSC), they proved to be multipotent, i.e. capable of colonizing the three germ layers in mouse embryos, and contributing to various organs including germline when injected into recipient mouse blastocysts. Human PGC extracted from the genital ridge or mesentery at 5–9 weeks of gestation, and cultured for 7–21 days on mouse fibroblasts with added LIF, basic FGF and forskolin, also produced cell colonies which were reminiscent of multipotent ESC and expressed the markers described above (Shamblott et al., 1998). The cell lines remained diploid and displayed a multipotency equivalent to that of ESC. When injected into blastocysts, they formed chimaeras, produced derivatives of the three germ layers and contributed to various organs including germline. Guan et al. (2006) stressed that maGSC could also be used to prepare personal stem cells derived from testicular biopsies, which avoided any need for human embryos.

Stembrid technology, which involves the hybridization of adult somatic cells with embryonic stem cytoblasts, can also be used to establish stem cell lines with normal and abnormal genotypes. Currently, 166 human ESC lines have been established of which 127 had a normal phenotype and 39 with genetic and chromosomal disorders (Verlinsky et al., 2005, 2006). This approach also cures genetic and acquired disorders (e.g. Strelchenko et al., 2006), an approach that is valuable ethically since it avoids any need for human oocytes.

Section 2.4. Genetics of ESC

Various aspects of the genetics of ESC have been studied and published in detail in a book edited by Lanza et al. (2004) and a selection of the numerous genes expressed in preimplantation mouse embryos will also be discussed here. One is the well-known totipotency gene Oct‐4 that was described earlier in this review. It is expressed in cleaving embryos, inner cell mass, and finally in PGC, and interacts with various genes including Fgf4 to establish extra-embryonic endoderm and polar trophectoderm (Table 1a, b, c). Sox2 shares similar functions, and also acts via Fgf4 as a factor controlling the differentiation of extra-embryonic ectoderm from polar trophectoderm (Avilion et al., 2003). A similar situation involves the nuclear reprogramming of teratocarcinoma cells using various cell extracts which involves a total of ∼1800 genes that are up-regulated and include Oct‐4, Sox2, and alkaline phosphatase (APL) (Collas et al., 2006).

Cell differentiation and morphogenesis in ESC are induced by retinoic acid, especially the formation of endoderm. It binds to its nuclear receptor, i.e. retinoic acid receptor (RAR) that recognizes and activates or represses control regions for particular genes. It also regulates Oct‐4, as shown by a 90% reduction in its mRNA as exposed ESC are committed in various ways. Once Oct‐4 is repressed, retinoic acid evokes the differentiation of ESC into endoderm-like cells, a modification resembling initial stages of embryogenesis. Nestin is among the other genes that are common in neural and pancreatic lineages, and which are widely expressed until Oct‐4 is down-regulated.

Among other important genes, polypeptide differentiation inhibitory ability (DIA), which resembles LIF, suppresses differentiation in mouse but not in human ESC. Related to interleukin, it induces differentiation in macrophages and prevents proliferation in myeloid leukaemic cells (Gearing et al., 1987). It also regulates ESC proliferation without differentiation in the absence of feeder cells (Schoonjans et al., 2003). Effects similar to those of LIF are invoked by interleukin‐6 receptor, interleukin-11, oncostatin M, and ciliary neurotrophic factor. They each invoke similar effects as LIF as a result of their binding to high-affinity receptors on haemopoietic and other stem cells.

Another significant gene, embryonal stem cell-specific gene 1 (Esg1), produces an RNA-binding protein that associates with 902 target transcripts including Cdc25a, Cdc42, Ezh2, Nfyc and NrSA2 (Tanaka et al., 2006). Its distribution resembles that of Oct‐4 and Nanog, and it is down-regulated if Oct‐4 is repressed. Normally expressed in mouse embryos between the 1‐cell stage and ICM and in polar trophectoderm, its protein levels are higher in ICM than in trophectoderm by day 3.

Takahashi and Yamanaka (2006) tested a different approach to preparing ESC for therapeutic purposes. They initially identified 24 genes that were expressed in adult mouse ESC, and selected four of them, namely Oct‐4, Sox2, c‐Myc, and Klf4. These were inserted into skin fibroblasts in an attempt to transform them into ESC (Figure 5). Relatively few recipient cells, i.e. one per thousand, were transformed. They resembled ESC in being multipotent and able to form many tissues in vitro. Named ‘induced multipotent stem’ (iPS) cells, they formed chimaeras when injected into recipient blastocysts and colonized most tissues including germline in the resulting chimaeras. When injected into tetraploid blastocysts, which cannot form ICM, the resulting offspring were solely derived from iPS cells, showing that their work was successful even if only a few cells had been transformed (Holden, 2006, 2007). Sathananthan (2007) also stressed the importance of iPS cells, their close resemblance to ESC and the fact that that human cloning was no longer needed.

Fig. 5 Transforming somatic cells to ESC (iPS cells) in vitro (Takahashi and Yamanaka, 2006). The four ESC genes that were inserted into a somatic cell are indicated. Reprinted from Cell, Takahashi K and Yamanaka S, volume 126, Induction of stem cells from mouse embryonic and adult fibroblast cultures by defined factors, pages 663–676, 2006, with permission from Elsevier.

Although Takahashi and Yamanaka (2006) had clearly shown how gene injections into somatic cells were superior to other methods of preparing multipotent human stem cells, they still faced a persisting problem, i.e. producing larger numbers of iPS cells. This involved raising numbers of reprogrammed cells, ensuring that human skin cells were transformed by the same four genes, confirming their iPS cells expressed the same genes as standard ESC, and identifying which recipient cells were easiest to transform. Deciding to make a second attempt at improving the rate of conversion to ESC, they replaced the four original ‘multipotency’ genes with Oct‐4 and Nanog. These genes were also known to induce multipotency in adult fibroblasts, and required a viral vector to weed out the non-responding somatic cells. Takahashi and Yamanaka (2006) succeeded initially, but received a setback when tumours were found in one-fifth of the 121 offspring. This was apparently due to their use of retroviruses, which activate cancer cells. Improvements are clearly needed and could include, for example, the use of electroporation to insert genes into somatic cells as practised by Hojman et al. (2007) when inserting the erythropoietin gene into murine muscle tissue.

Commenting on the work of Takahashi and Yamanaka (2006), Edwards and Ahuja (2006) suggested that further opportunities could emerge from their work since blood cells may convert more easily to neurons or cardiomyocytes, and pointed out that thousands were available for conversion.

Section 3.1. Properties of haemopoietic and mesenchymal precursors

Small numbers of multipotent tissue stem cells are now known to exist in mice, especially in bone marrow cells and their precursors. These haemopoietic precursors differentiate in embryonic mesoderm, enter yolk sac as aggregates of extra-embryonic mesodermal cells and form blood islands in yolk sac at day ∼7.5 (Edwards, 2005). This period compares with day ∼12 in humans.

Tissue stem cells are the sole source of haemopoietic stem cells (HSC). After they exit the yolk sac, they migrate successively to developing aorta, genital ridge and mesonephros and then to fetal liver after 10–11 days. They synthesize haemoglobin and bone marrow just before birth.

The yolk sac is thus the source of HSC that seed the liver and form circulating cells, including both myeloid descendants which produce erythrocytes, granulocytes and megakaryocytes, and lymphoid tissues which produce lymphocytes and platelets (Moore and Metcalfe, 1970). These cells possess migratory properties, a factor that may explain why multipotent stem cells are distributed among numerous tissues. HSC and mesenchymal stem cells (MSC) seemingly differentiate from a single stem cell, the haemangioblast or the blast-forming cell, that can be isolated from primitive streak (Edwards, 2006a,b). Their angioblast potential is queried by some investigators since these cells may also form vascular smooth muscle.

Vodyanik et al. (2006) revealed that early HSC precursor cells were also multipotent by examining their expression of CD markers. Haemopoietic progenitors were confirmed to be multipotent just as other stem cells. They produced haemopoietic, mesenchymal, endothelial, and lymphohaemopoietic cells, and again like other stem cells, they also produced muscle cells, adipocytes, osteocytes and chondrocytes and could colonize bone, muscle, lung and epithelium (Pittenger et al., 1999; Chan and Yoder, 2004; Wang et al., 2004). Forbes et al. (2002ab) added further evidence of their multipotency by confirming that HSC stem cells also colonized liver, brain and kidney, and that similar responses occurred in humans after they had been grafted with bone marrow, given organ transplants, or if women had been given male transplants.

Transplants of genetically marked adult mouse bone marrow also revealed a remarkable plasticity as they generated neuronal phenotypes in adult brain by 1 to 6 months after grafting into lethally irradiated hosts (Brazelton et al., 2000). Donor cells in the brain displayed characteristics that differed from bone marrow since hundreds of these cells found in sections of the brain expressed gene products such as NeuN, neurofilaments and class III β‐tubulin, which are typical of neurons. They also activated the transcription factor cAMP response element-binding protein (CREB), which led these investigators to conclude that bone marrow displayed a remarkable plasticity and possibly possessed clinical applications.

Matsuura et al. (2004) reported that cells expressing Sca-1⁺ in mice possess features similar to those of ESC. When isolated, some of them attached firmly to gelatin-coated culture dishes and then divided slowly. Treatment with oxytocin stimulated these cells to express cardiac transcription factors plus contractile proteins with a sarcomeric structure. Isoproterenol induced their spontaneous beating and the formation of intercellular Ca⁺⁺ transients. Cardiac Sca-1⁺ cells expressed oxytocin receptor mRNA, a process that was up-regulated by exposure to oxytocin, while some cells expressed alkaline phosphatase after osteogenic induction. Other cells expressed oxytocin receptor mRNA and displayed a raised output of oxytocin which led these investigators to conclude that Sca-1⁺ cells in the heart of adult mice have a potential reminiscent of ESC, and that they are able to contribute to regeneration in damaged hearts.

Myocardial infarction can also induce the division of human cardiac myocytes. It also induces scarring of the heart, a finding previously interpreted as evidence that the heart possesses myocytes that are unable to divide. Recent observations have stressed how myocytes can proliferate, an observation that led Beltrami et al. (2001) to assess the extent of mitosis among human myocytes after myocardial infarction. They recorded the percentage of myocyte nuclei and their mitotic index, together with estimates of the frequencies of mitotic spindles, contractile rings, cytokinesis, and karyokinesis. In infarcted hearts, 4% of myocyte nuclei located in regions adjacent to these infarcts expressed Ki-67 as compared with 1% in regions that were widely separated from them. The re-entry of myocytes into the mitotic cycle raised mitotic indices by 0.08% and 0.03% respectively in zones adjacent to and distant from infarcts. More evidence on the frequency of cell division was obtained by scoring the formation of mitotic spindles, contractile rings, karyokinesis and cytokinesis. The results were taken as evidence of myocyte proliferation and myocardial infarction, and the authors summarized their work to conclude that their evidence challenged the conception that the heart was a post-mitotic organ (Beltrami et al., 2003).

Evidence of cardioblast progenitors expressing isl1⁺ in postnatal hearts of rats, mice and humans was recently reported by Laugwitz et al. (2005). Isolated progenitor cells were grown on a cardiac mesenchymal feeder layer and were capable of adopting a cardiomyocyte phenotype. They were selectively marked and purified, and shown to be cardioblasts with the property of converting to mature cardiac cells. Many of them expressed myocytic markers at a time when cell fusion, Ca²⁺-cycling and the generation of action potentials were absent. This discovery of native cardiomyocytes revealed the existence of a genetically controlled system that sustained successive stages in the formation and maturation of cardiac lineages.

Next, Johnson et al. (2004) indicated that germline markers existed in bone marrow and that putative stem cells could colonize ovaries in sterilized mice. If this is confirmed, hundreds of oocytes could be generated, with morphologies typical of ordinary oocytes. If true, this discovery could help to provide numerous oocytes to infertile women. Whether a similar system involving spermatocytes could operate for infertile men remains unknown. Caution is clearly necessary about these claims since weaknesses exist in the data presented by Johnson et al. (2004). For example, despite receiving chemotherapy, young women given transplants of blood stem cells still display ovarian failure, and infertile men remain sterile after stem cell transplantation.

Precursors of MSC have different origins. They were isolated by Freidenstein et al. (1976) who utilized their ability to attach to plastic surfaces. Expressing cell-surface CD markers, their expression of CD14⁻CD34⁺CD45⁻ confirmed them as forerunners of endothelium, osteoblasts and adipocytes, among other tissues. Their expressed markers were typical of ESC, which tended to confirm the existence of a close relationship between them and ESC (Pittenger et al., 1999; Edwards, 2004b). They are apparently sustained by the Wnt pathway, and their descendants can be traced by various CD markers including CD34. MSC derived from bone marrow and some connective tissues may also be capable of producing HSC, stroma and some connective tissues, and future research may determine their role in abnormal tissues and bone marrow.

Section 3.2. Amniotic fluid-derived cells

According to Macek et al. (1973), tissue stem cells are distributed among many tissues including the upper respiratory tract, gastrointestinal and genitourinary tracts, skin, eyes and umbilical cord. Similar cells named amniotic fluid-derived stem cells (AFC) were isolated from cord blood, amnion and amniotic fluid. When cultured for long periods in Eagle's medium containing calf serum, AFC in early stages of culture were predominantly epithelioid and fibroblastic, whereas in later stages of growth, cell lines became uniformly fibroblastic and synthesized collagen. Their ratios of soluble to insoluble collagen altered in later stages of culture, when levels of total collagen in centrifuged preparations corresponded with the sum of hydroxyproline in isolated collagen fractions and of collagen present in supernatants containing soluble and insoluble collagen. Levels of nitrogen included amounts present in supernates, isolated collagen fractions and cell residues (Macek et al., 1973).

In long-tem cultures, epithelioid colonies became increasingly fibro-epithelioid and then uniformly fibroblastic as shown by their production of collagen. Variations between cell lines were assessed by means of hydroxyproline⁄nitrogen ratios, which varied between 2.61 and 5.52 and averaged at 3.83 except for the colony named E8 which reached 9.13 (Table 4). Similar high values were found in fibroblast cultures derived from fetal blood, and average hydroxyproline⁄nitrogen ratios of 3.3 ± 0.3 characterized levels human fibroblasts in long-term cultures. Macek et al. (1973) concluded that collagen synthesis in AFC cultures were typical of fibroblasts and proved the cells had a mesenchymal origin and could be used to investigate hereditary disorders of human connective tissue.

Later, De Coppi et al. (2007) decided to examine AFC including those recovered after amniocentesis. They reported that AFC were not ESC, were encoded at the Sl locus in mice, resembled ESC in being multipotent, and expressed c‐kit (CD117), the receptor for stem cell factor (Zsebo et al., 1990). Though few in numbers, purified AFC had produced tissues characteristic of the three primary embryonic germ cell layers including adipogeneic, osteogenic, myogenic, endothelial, neurogenic and hepatic lineages (De Coppi et al., 2007). Human AFC collected after amniocentesis expressed c‐kit, divided every 36 h and did not require feeder layers (Bryan et al., 1998). They expressed class I major histocompatibility antigens and mesenchymal and neural markers but not ESC markers such as CD29, CD44, CD73, CD90 and CD105. Unlike ESC, which can produce teratocarcinomas when implanted into recipients, they remained unchanged and diploid after 250 generations (Bryan et al., 1998). Details of their karyotypes and telomeres are shown in Figure 6. AFC became positive for nestin, formed cells resembling dopaminergic neurons and expressed the gene GIRK2. When exposed to nerve growth factor, primitive forms of AFC produced neural lineages secreting the neurotransmitter L‐glutamate. Some of those that were injected into the lateral cerebral ventricles of newborn mice participated seamlessly in developing the central nervous system after 1 month and survived in periventricular areas and in olfactory bulb for 2 months. When cultured in osteogenic medium, AFC produced functional osteoblasts, those cultured for a week being seeded on scaffolds previously implanted subcutaneously into immunodeficient mice. They also produced blocks of material resembling bone, and formed hepatic lineages lasting >45 days that produced urea. Among other studies, Johnstone and Sara (2003) and Lima (2003) independently transplanted olfactory tissue to alleviate spinal cord injury.

Fig. 6 A normal karyotype and long telomeres characterize clonal human AFC (De Coppi et al., 2007). (a) A giemsa band karyogram showing chromosomes of late passage (>250 pd) cells. (b) Flow cytometry of late-passage cells showing DNA stained with propidium iodide. G1 and G2/M respectively indicate a 2n and 4n DNA content in the cells, while S indicates those cells which are undergoing DNA synthesis and have a DNA content between 2n and 4n. (c) The conservation of telomere length in AFC in early (20 pd, lane 3) and late passage (250 pd, lane 4). Short-length (lane 1), and high-length (lane 2) telomere standards are shown. pd = population doubling; marker lengths indicated. Reprinted by permission from Macmillan Publishers Ltd: Nature Biotechnology, volume 25, pages 100–106, 2007.

AFC are virtually identical with those typical of rabbit ESC as described by Cole et al. (1966). This fact led De Coppi et al. (2007) to conclude that AFC (these authors relabelled them as AFS cells) resembled multipotent ESC, provided further evidence that many if not all stem cells share a common origin and that they could have been derived from the same foundation population. They could, if necessary, also be recovered from cord blood together with other forms of stem cells. De Coppi et al. (2007) also pointed out that current strategies in cell therapy and tissue engineering required stem cells biopsied from the patients, whereas AFC offered an alternative and equivalent source of stem cells, could be obtained during routine amniocentesis and offered a convenient source of stem cells for autologous therapy in later life.

Some of the findings of De Coppi et al. (2007) were queried by Toselli et al. (2008) who presented data indicating that AFC could not differentiate specifically into neurons or muscle cells but could yield differentiated cells expressing lineage-specific markers. They claim that nestin is expressed in neuroepithelial stem cells and in mesonephric mesenchymal endothelial cells that were developing into blood vessels and kidney cells. They stressed that in their view, GIRK2 gene expression is not a specific neuronal marker, and brought forward various lines of evidence to support their viewpoint. They also criticized evidence claiming that AFC can regenerate neurons, and demanded evidence that they could be substantiated by the formation of functional neurons.

Lima (2003) summarizes the therapeutic properties of olfactory tissue, which covers ∼2 mm in each nasal cavity. Numerous multipotent cells with regenerative potential including bipolar olfactory neurons arise from progenitor cells that reside at its base. Olfactory ensheathing cells (OEC) can treat neurological disorders, promote axonal growth and remyelinate injured spinal cord in rats. They can also repair and regenerate spinal cord if implanted at 4 weeks after the spinal cord has been transected and when OEC reside together with regenerating neurons (Lima, 2003). Stem cells aspirated from human nostrils are also reported to repair spinal cord injury when minced and implanted into spinal cord when working in tandem with other types of regenerative cells. When OEC were isolated from fetal tissue and grafted to 150 patients, they functioned without any need for immunosuppressants.

Judged from this evidence, AFC and OEC cells offer valuable alternatives to ESC. Their easy availability should find them a place among other new approaches to tissue repair.

Section 4.1. Origin

The umbilical cord is composed of two arteries and one vein and is covered by an epithelium formed from surrounding amnion as described by Meyer et al. (1983). Blood vessels in the umbilical cord are protected by a matrix known as Wharton's Jelly, a gelatinous substance. It is rich in hyaluronic acid and contains glycoprotein microfibrils and collagen fibrils that couple with primitive and multipotent mesenchymal cells. Cells in the jelly express markers for oligodendrocytes and astrocytes and other characteristics of stem cells. Hence, they may be a source of primitive cells capable of differentiating into neural cells (Mitchell et al., 2003).

Some investigators scrape Wharton's jelly off cord blood stem cells as a standard practice. It is recommended that dissection is followed by enzyme digestion, which provides a high yield when compared with the numbers of MSC present in cord blood (Peter Hollands, personal communication). These cells are particularly efficient at osteogenesis, and can be removed by overnight treatments with basic FGF followed by their transfer to medium containing low amounts of serum, butylated hydroxyanisole and dimethylsulphoxide, Developing a neural phenotype and round bodies with multiple neurite-like extensions, cord blood cells express the neuronal proteins neuron-specific class III β‐tubulin, neurofilament M, axonal growth-associated protein and tyrosine hydroxylase, plus neuron-specific enolase (NSE). They also increasingly resemble bipolar or multipolar neurons by producing networks reminiscent of cultured neurons, and develop characteristics of a neural phenotype (Mitchell et al., 2003).

Wharton's jelly stem cells (WJSC) differ from bone marrow stem cells in requiring less-stringent tissue matching. They also express ESC markers including SSEA‐1, SSEA‐4, Tra‐1–60, Tra‐1–81, alkaline phosphate and Oct‐4, plus the mesenchymal markers CD44, CD90 and CD105 (Fong et al., 2007a). Cord blood cells are turning out to be of great value in caring for sickle cell anaemia, haemophilia, haemoglobinopathies and immunodeficiencies, although bone marrow cells have a greater efficacy for repair (Mitchell et al., 2003; Ghen et al., 2006; Takahashi et al., 2007).

Cord blood contains haemopoietic and endothelial stem cells, neuronal precursors and CD 133⁺/CD34⁻ MSC. Some of its characteristics are listed in Table 5. Its mesenchymal cells can undergo more than 80 population doublings and express stem cell markers including Oct‐4, c‐kit, telomerase, the matrix receptors CD44 and CD105 and the integrin markers CD29 and CD51. They do not express haemopoietic markers (Mitchell et al., 2003; Wang et al., 2004). Human MSC isolated from Wharton's jelly can be transformed with a success rate of ∼13% into dopaminergic neurons that are characterized by the expression of dopamine and tyrosine hydroxylase (Fu et al., 2006). Non-haemopoietic, human umbilical cord perivascular (HUCPV) cells are a source of mesenchymal progenitors characterized as CD45⁻CD34⁻SH2⁺SH3⁺Thy-1⁺CD44⁺ (Sarugaser et al., 2005). They produce a morphologically homogeneous fibroblastic cell population that expresses α‐actin, desmin, vimentin and 3G5 (a marker for pericytes). Their doubling time decreased from 60 h to 20 h to result in cell expansion within 30 days, while those cultured in non-osteogenic conditions possessed a subpopulation with a functional osteogenic phenotype that elaborated bone nodules, especially when exposed to osteogenic supplements in culture media. These investigators concluded that their rapid doubling time indicated that HUCPV cells were a source of therapies based on mesenchymal cells.

Tuan et al. (2003) reviewed the multipotency of human MSC derived from bone marrow stroma and connective tissues. These cells offer exciting prospects for cell-based tissue engineering and regeneration, based on their differentiation potentials in vitro and in vivo. They stressed current belief that the use of MSC lags behind other types of stem cells, including HSC. They also suggest that future research should define cellular and molecular fingerprints and elucidate their roles in normal and abnormal tissue functions.

When comparing the properties of MSC derived from umbilical cord with those of bone marrow cells, Baksh et al. (2007) discovered that HUCPV cells proliferated in vitro and were in essence an extra-embryonic source of MSC. These cells also had a higher proliferative potential than bone marrow mesenchymal stromal cells and proved to be capable of osteogenic, chondrogenic and adipogenic differentiation. Their osteogenic differentiation proceeded more rapidly than with bone marrow stem cells, and their expression of CD146 marked mesenchymal cells and genes associated with Wnt signaling. These regulate extra-embryonic MSC, this evidence again indicating that HUPVCs and MSC could be valuable for cell-based therapies (Baksh et al., 2007).

A very recent case report described the successful use of intravenous allogeneic MSC and expanded umbilical cord blood CD34 cells to treat a patient with dilated cardiomyopathy (Ichim et al., 2008). This condition reduces the levels of oxygen reaching heart muscle and causes inflammation. Both conditions were ameliorated 3 weeks after treatment with millions of these cells, the patient exhibiting neither side effects nor graft versus host disease 11 months after treatment.

Large quantities of tent stem cells from cord blood can be cryostored just after birth and cryopreserved in case they are needed in later life. Many are stored in stem cell banks. Their value was best displayed when a couple with a child suffering from Fanconi anemia requested that preimplantation genetic diagnosis was applied during their next IVF cycle in order to avoid the birth of a second afflicted child. The procedures were successful, so the parents were advised that human leukocyte antigen (HLA)-matched cord blood stem cells taken from this newborn child could alleviate symptoms of Fanconi anaemia in their existing child. Moreover, these cells would not be rejected. Cord blood cells from the second child, nowadays called a ‘saviour baby’, were accordingly transferred to the first child who made a rapid and complete recovery (Verlinsky et al., 2000). So far, cord blood cells taken from saviour babies have been grafted to 230 recipients.

Many other children afflicted with serious diseases are now treated in a similar fashion (Ghen et al., 2006). Discussing the characteristics of human umbilical cord blood, its potential clinical applications and rates of improvement in treated patients (Figure 7), Ghen stressed that cord blood is easily available, has low risks from infectious disease and graft-versus-host disease, and can be used despite disparities in HLA types.It is valuable for patients with previous problems and confers benefits on very sick patients, including increased muscle strength and pulmonary function; two-thirds were alive at 1 year postgrafting, the others having been lost to follow‐up.

Fig. 7 Frequency of combined bulbar and limb onset amyotrophic lateral sclerosis on a Karnofsky performance scale in a group of patients (Ghen et al., 2006).

Section 4.2. Do ESC, tissue stem cells and cord blood cells share a common origin?

The sharing of totipotency and other characteristics between various types of stem cells raise queries about their having a common origin in the preimplantation embryo. Fujimori et al. (2003) accordingly traced cell lineages in preimplantation mouse blastocysts to determine the fate of individual blastomeres. Differing lineages were traced through 2‐ and 4‐cell stages, and then to ICM or trophectoderm. Opinions differ as to whether cell fate had been partitioned from 2‐ and 4‐cell stages since some investigators believe that individual blastomeres differentiate along their own specific pathways, even though evidence produced by Edwards and Hansis (2005) had revealed that the differentiation of individual blastomeres in human embryos was well in progress by the 4‐cell stage (Figure 8). One blastomere was found to secrete βHCG and two others were apparently multipotent. The fourth blastomere was suggested as being germ line precursor, although this remains to be confirmed (Hansis, 2006).

Fig. 8 Expression of Oct‐4, β‐HCG (all variants), and β‐actin mRNA in individual blastomeres of two human embryos (Hansis, 2006). Note that each embryo possesses a single blastomere expressing β‐HCG. Embryo A had five cells, and embryo B had four cells. H₂O, water control; Media, media control.

Further evidence emerged when Kurimoto et al. (2006) discovered that at least two individual cell lines had already formed in mouse ICM by day 3.5. These investigators applied several methods based on the global amplification of mRNA. These included high-density oligonucleotide microarrays, the polymerase chain reaction and linear amplification. This combination of methods confirmed the existence of two distinct cell types in disaggregated ICM cells (Table 6). One type expressed genes typical of early epiblast including nanog and other genes associated with the transcription factor c‐Myc. The second type expressed genes typical of primitive endoderm including Gata4 together with the transcription factors Sox2 and Kif2. Kurimoto et al. (2006) also showed how Oct‐4 and Cdx2 belonged to different groups of cells, their conclusion being confirmed when one group of cells in a cluster of nine was found to express nanog and fgf4.

Further evidence was found of individual cells lines in extra-embryonic lineages when Chazaud and Rossant (2006) revealed how axial patterning is initiated by restricting gene expression to dorsal visceral endoderm. This proximodistal axis is later transformed under the regulation of Nodal and Wnt signalling to an anteroposterior axis.

If mammalian stem cells are allocated to various tissues during early embryogenesis, their origin and wide distribution into numerous tissues may resemble events occurring during the widespread distribution of blastemal precursor cells in urodeles. This possibility is assessed in detail below.

Section 5.1. Origins and methods

Wider attention is now being given to tissue engineering. Original methods were introduced by Langer and Vacanti (1993), and recent approaches have led to effective methods of constructing new tissues and organs. A succession of problems have been solved, including the design of various scaffolds shaped to match the particular organ under repair. Scaffolds must also provide physical support and a natural environment to encourage cells to proliferate and undergo morphogenesis (Banerjee et al., 2006).

New technologies being developed include the introduction of polymeric growth factors for tissue engineering (Chen and Mooney, 2003). These authors describe therapies based on drug delivery technologies (DDT) that ensure cells have a satisfactory local environment of their own, including a combination of growth factors. Gene therapy has also developed in various directions including conventional therapy in which plasmid DNA adenoviruses are injected directly into tissues. These early methods required several days to induce and activate cell-based tissue regeneration, and although very promising, tissue engineering has had to face limitations from the beginning since it could not substitute for natural biological functions. Moreover, grafts could be rejected, a process that must be avoided since immunosuppressives invoked serious side-effects in some patients, which delayed the time to recovery required by grafted cells.

Discussing the nature of regenerative medical therapy, Tabata (2008) described it there as providing cells with an in-vivo local environment which enables them to proliferate and differentiate into a ‘natural scaffold’. He stressed how the body consists of three factors. These are cells, a natural scaffold consisting of extracellular matrix (ECM) that supports cell proliferation and differentiation into a natural scaffold, and signalling molecules such as growth factors and gene products. Biomaterials were soon found to also have an important role by permitting cell infiltration and providing cells with constant supplies of oxygen and nutrients. It was also clear that physical hindrance to developing organs must be avoided to ensure that cytokines, chemokines and genes were activated normally.

Artificial scaffolds demand close attention to detail (Ghen et al., 2006; Table 7). One approach is to prepare a scaffold of ECM to encourage cell attachment, proliferation and differentiation. It must be made of biomaterials that are porous and degradable, which encourages cells to produce their own ECM (Larsen et al., 2006). Scaffolds are best used in combination with cells and signalling molecules such as growth factors and cytokines that can induce tissue regeneration although direct injections of growth factors may not necessarily be effective since they may diffuse from the scaffold and be deactivated (Tabata, 2008).

Other problems included the need to be aware that proteins can be released from a protein carrier. An effective drug delivery system must prevent growth factors from diffusing away, and to provide protection from immunological attack. This was restricted by technical improvements including new drugs and growing top-quality cells on artificial extracellular matrices. Any infiltration of fibroblasts might reduce the available space so barrier membranes had to be available to maintain regenerating tissues.

Lastly, it is clearly obvious that ESC, HSC and MSC are essential for tissue engineering since they are multipotent, as described earlier. If possible, the patient's own MSC should be utilized in order to avoid immune rejection and demonstrate their therapeutic properties (Docheva et al., 2007). Neural stem cells and fatty cells are isolated from fetuses and adults and also are highly effective in differentiation, although demands for these various types of cell are so great that they are in short supply. Cells grown in vitro can be used as replacements and can be grown on ECM.

Section 5.2. Recent advances of tissue engineering

By today, numerous reports have revealed the value of tissue engineering, although experience has revealed that effective therapies are lacking for lung fibrosis, cirrhosis, dilated cardiomyopathy and chronic nephritis. This is due to the site of injury being occupied by fibrous tissue, excessive collagen fibres and fibroblasts can impair natural healing. This situation may be corrected by injecting a virus encoding a matrix metalloprotease, which can suppress tissue fibrosis and lead to improved symptoms of the disease (Iimuro et al., 2003).

Improved therapies also include the utilization of some hydrogels, which enhance the in-vivo biological activity of growth factors to induce tissue regeneration (Lutolf and Hubbell, 2005). Another biodegradable hydrogel has been prepared from gelatine. It induces tissue and organ regeneration, which results in the controlled release of biologically active growth factors. This system of controlled release can also be applied to a plasmid DNA to enhance the level and prolong the time period of gene expression (Kushibiki et al., 2003).

Some hydrogel systems have been prepared that simultaneously release two or more growth factors in differing concentrations and release profiles, and others prepared with low doses of either basic FGF or transforming growth factor β1 (TGF‐β1) (Tabata, 2008). Applying a hydrogel incorporating either of them may result in the preparation of therapies for bone defects, e.g. in rabbit skulls.

New technologies are also overcoming constraints arising in Petri dishes such as the anomalous growth of human breast epithelial cells in two-dimensional monolayers (Table 8) (Cushing and Anseth, 2007). Three-dimensional rather than two-dimensional scaffolding is preferred since ESC can differentiate into blood-forming stem cells, and incorporate cellular microenvironments with hydrogel networks of interacting, highly hydrated polymer chains with an elasticity resembling that of natural tissues. Oxygen must also be available in both two- and three-dimensional scaffolds and functional tissue equivalents must be engineered so that no cells reside >100 µm from the oxygen source.

Standardized techniques for analysing proteins and protein distributions have also been found to be essential (Cushing and Anseth, 2007). They must be isolated from matrix cells, and new fluorescent probes, non-invasive live-imaging and real-time analyses are helping to assess the effects of stratification on three-dimensional cellular functions. Matrigel is being explored for this purpose, and polyethylene glycol helps to prepare hydogels with a consistent composition although it lacks functional sites for soluble or cell-surface proteins. Synthetic hydrogels help to cope with Click reactions that link with small molecular subunits, degradation can be controlled by manipulating reactant stoichiometry and peptide concentrations, and modular networks establish interactions between hydrogen bonding, protein folding and protein–protein interactions. Conformational changes can also arise in some gel networks, e.g. those made from calcium-sensitive building blocks or single-stranded DNA components that expand or contract in response to various stimuli.

De Coppi et al. (2007) recently reported an example of tissue engineering. They differentiated AFC into various tissues including osteoblasts, which are utilized for bone formation in vivo (Figure 9). They embedded human AFC in an alginate⁄collagen scaffold in vitro, and nodules were found to be consistent with bone formation by day 45. Human AFC were also embedded into scaffolds implanted subcutaneously into immunodeficient mice, and highly mineralized tissue was identified in the scaffold 8 weeks later. By 18 weeks, material resembling bones was identified as being typical of human bone, bone formation being noted after several months. High-density tissue-engineered bone was detected by scanning a few weeks later, whereas bony tissues were absent in unseeded scaffolds. Similar results were obtained when AFC exposed to nerve growth factor differentiated to brain cells.

Fig. 9 Tissue engineered bone from human amniotic fluid-derived stem cells (AFC) cells (De Coppi et al., 2007). (a) Calcium deposition by human AFC, maintained in culture medium sustaining that, was quantified by measuring the complex levels of calcium cresolphthalein (closed line) as compared with that of undifferentiated AFC (broken line). (b) Staining by van Kossa's method of control printed, unseeded alginate⁄collagen scaffold of AFC that were recovered 8 weeks after implantation in nu/nu mice. (c) Staining by van Kossa's method of AFC in an alginate⁄collagen scaffold recovered 8 weeks after implantation: the black staining indicates strong mineralization. (d) Micro-CT scan of mice 18 weeks after implanting the printed constructs: the dark region indicates strong mineralization and the symbols indicate scaffolds seeded with AFC. (e,f) Close‐up views of scaffolds seeded by ESC, indicated by the symbols. Reprinted by permission from Macmillan Publishers Ltd.: Nature Biotechnology, volume 25, pages 100–106, 2007.

Collaborative research between material, pharmaceutical, biological and clinical scientists has been essential from the outcome of tissue engineering to cope with new advances in gene therapy. Examples include the injection of plasmid DNA and adenovirus into recipient cells, and the use of gene transfection to genetically activate stem cells and other cell types (Tabata, 2008). These examples reveal that new advances in tissue engineering, including the role of the blastema, are clearly inevitable.

Section 6.1. Differentiation and regeneration

Recent research has concentrated on tissue repair in urodeles since they are the only vertebrates able to regenerate limbs, as first discovered by the Italian physiologist Lazzaro Spallanzani in 1768. Numerous papers have clarified the nature of limb and tail regeneration in salamanders, axolotls and newts. Their astonishing feats of regeneration involve complex three-dimensional aspects of restoration, the dedifferentiation of some tissues and the regeneration of others. Salamanders are an advanced species with regard to regeneration, and are the only vertebrates capable of regenerating entire organ systems by the formation of a blastema (Voss et al., 2001). They also depend on a wide diffusion of stem cells around the body, a situation that is not unlike the wide distribution of tissue stem cells in mammals. Other urodeles, including axolotls, are also able to regenerate missing tails, limbs and eyes, especially in their larval stages (Zhang et al., 2003).

When the limbs or tails of salamanders are amputated, a layer of epidermal cells grows over the stump, succeeded by the formation of a mass of blastemal cells, just beneath the upper cell layer that consists of ∼10,000 cells before morphological differentiation is advanced (Stocum, 1996). Blastemal stem cells are ectodermal yet are able to redifferentiate into mesoderm (Echeverri and Tanaka, 2002). The red-spotted newt has similar attributes as salamanders and can also regenerate complex structures that enable it to create a pool of dedifferentiated and heterogeneous cells during limb regeneration (Vascotto et al., 2005).

The successive stages of regeneration in salamanders are well understood. It relies on tent cells and on stem cells from spinal cord that re-enter the regrowing tail to produce muscle and cartilage. Muscle cells from the stump then add to the blastema as they initiate their cell cycle and produce thousands of dedifferentiating cells including muscle cells that have lost their characteristic proteins (Stocum, 1996; Echeverri et al., 2001). The final pattern of the tail is established after the blastema has organized nerve, muscle, skin, and cartilage to help with its regeneration.

Extracellular matrix, the ‘natural scaffold’, is produced in early blastema. It resembles that found in an embryonic limb bud with its high levels of hyaluronate, fibronectin and tenascin and low amounts of collagen and chondroitin sulphate. Forming between incoming cells, it is degraded 2 days after larval limbs have been amputated, its controlled and rapid degradation by metalloproteases and acid hydrolases liberating cells that are undergoing dedifferentiation. These cells then form mesenchymal-like blastemal cells that multiply and accumulate under the wound epidermis. Migrating through the clot from the cut edges of the skin, they close the wound within 6–9 h. Neutrophils and macrophages also invade the wound to kill bacteria and phagocytose debris (Holtzer, 1956; Stocum, 1996).

New capillaries sprout into mesenchyme as the avascular blastema enlarges between 6–9 days post-amputation (Stocum, 1996). Nerves innervating the limb initially degenerate over a short distance proximal to the amputation plane, then rapidly regenerate to reinnervate wound epidermis and mesenchymal cells. Dermis may now contribute most cells to blastema, as numbers of muscle and cartilage cells decline. Blastema thus increasingly resemble a population of tent progenitors in a complex mixture of predetermined cells. Their central core of stem cells has high rates of cell division, expresses numerous genes, and is surrounded by a layer of cells consisting of the three germ layers, i.e. endoderm, mesoderm and ectoderm.

Regenerating spinal cord has an essential role in the proliferation and condensation of mesenchymal cells during the formation of a regeneration blastema (). Accumulating cells form the apical epidermal cap at the apex of the blastema, while accumulating epididymoglial cells in the stump of the spinal cord form the ependymal tube. This tube extends into the blastema as blastemal cells condense in a precise pattern to initially form the cartilage rod and then produce masses of skeletal muscle. In the regenerating tail in axolotls, the neuronal population is derived from a 500‐µm region of the mature spinal cord, which abuts the amputated plane that generates neuronal progenitors needed for regeneration (Tanaka, 2003). Most progeny remain close to the dorsoventral (D/V) distribution of parent cells whereas others migrate to multiple D/V positions where they acquire different fates and profiles in regenerating spinal cord. Those located most distally in the regenerating tail differ in their molecular activity from others located in more proximal regions, which suggests that progenitor cells in this region are destabilized. Cells in ventral regions migrate dorsally, leave the spinal cord and enter the blastema.

Fig. 10 Early blastema formation in salamanders. Note how it forms a mass of cells (A) that lie adjacent to epidermis (B) and are invaded by nerve cells (C). Photograph kindly supplied by David Stocum.

Fig. 11 Models of proximal–distal patterning during limb regeneration in newts (Tanaka, 2003). A Blastemal cells form and retain a memory of their origins then outgrow the amputated portion. Graded information along the proximal–distal axis of mature limbs may then inform neighbouring cells of their position. A limb blastema reaches 10,000 cells before morphogenesis begins, when it harbours the necessary information to autonomously reproduce the correct structure in ectopic sites (Stocum, 1996). After injury, it is essential to produce blastemal cells possessing fingertip identity, which helps to establish two boundaries for the regenerating structure. Multiple HoxA genes may co-express in early limb blastema, as a morphogen gradient specifies positional information. B An alternative model may link with the formation of gradients among cell-surface markers expressed along the proximal–distal axis. Blastemal markers are distalized as surface molecules are lowered, which stimulates proximal cells to produce new blastemal cells undergoing intercalary regeneration, i.e. missing positions are filled in. C Thirdly, it is possible that injured cells produce a randomized group of cells with different positional values, and these may be sorted later. All three models of limb patterning imply that blastema become self-patterning and autonomous. Reprinted from Cell, Tanaka EM, volume 113, Regeneration: If they can do it, why can't we?, pages 559–562, 2003, with permission from Elsevier.

Epimorphic regeneration proceeds along a gradient of rostrocaudal differentiation, with numbers of new vertebrae matching those excised by amputation (Stocum, 1996; Nicolas et al., 2003). The ability of tail blastema transplanted to a more proximal level in the stump to develop according to their level of origin led Stocum (1996) to suggest that blastema are self-organizing and inherit level-specific memories of their axial positions from stump cells.

As growth continues, blastemal mesenchyme cells redifferentiate in a proximal⁄distal sequence to form skeletal elements, connective tissue and muscle. Ectodermal stem cells involved in these processes transform to mesoderm as muscle cells in the stump resume their cell cycle, undergo dedifferentiation and lose their characteristics. Regrowth also includes the formation of nerve, muscle, cartilage, notochord, and spinal cord, with Hox genes taking a role in local cell interactions as part of an autonomous patterning mechanism. Effector molecules determining positional identity locate in or on the cell surface and can be modified by exposure to retinoic acid.

Stocum (1996) summarized these events as a ‘post-embryonic form of epimorphic pattern regulation by which a blastema forms from the controlled degradation of ECM and the dedifferentiation of the liberated cells to mesenchyme-like blastema cells that accumulate and proliferate under the wound epidermis’. In effect, a limb blastema is equivalent to a developing limb bud, capable of differentiating and organizing itself to become a perfect replica of the damaged limb including its missing portion. Its growth is sustained by numerous cell divisions through early and medium bud stages as mesenchyme redifferentiates into missing skeletal elements, soft connective tissue and muscle in amputated segments. Redifferentiation occurs in a proximal-to-distal sequence (P/D) typical of limb buds in vertebrate embryos, and regeneration blastema are unique in differentiating along an anterior–posterior (A/P) axis instead of the posterior–anterior (P/A) sequence typical of other vertebrates. Cells in regenerating blastema become organized into tissue patterns typical of the amputated limbs and display characteristics such as diffusible morphogens which function as signalling molecules.

Section 6.2. Gradients in limb regeneration in urodeles

During normal limb generation in urodeles, adjacent differentiated tissues send limb-type and level-specific instructions to blastema that determine its progression into the missing segments. Level-specific memories of axial positions in parental limb cells may have been inherited, so that grafts to a limb stump with a different proximodistal level results in these cells developing in relation to limb type, level of origin and transverse axial polarity. Such ‘positional memories’ carried by blastema cells are derived from dermis that specifies transverse axial poles in regenerates, and its constraints ensure that blastema only replace cells that were lost (Carlson, 1983).

Positional information boundaries are established in blastema in case discontinuities should emerge in its internal structure. Stocum (1996) suggested that the boundary region might be a source of diffusible morphogens, which probably impose patterns on body axes although contributing little to the structure of the regenerate (). Alternatively, patterns may be specified by local interactions between cells that are constantly aware of the positional identity of their nearest neighbours and sense their discontinuities. Re-establishing missing P/D positional identities might also involve intercalary regeneration, and a common mechanism could control the transverse and P/D axes.

Epimorphic regeneration in regenerating tail follows a rostrocaudal differentiation gradient, with numbers of new vertebrae in regenerated tail being directly proportional to numbers removed during amputation (Nicolas et al., 2003). This form of repair does not apply to mesodermal tissues, such as muscle and cartilage, which are formed from radial glial cells in the spinal cord. This evidence implies that some cells entering the blastema dedifferentiate and redifferentiate while others remain differentiated as they enter the blastema and their cells remain committed to specific tissues (Tanaka, 2003). If committed blastema cells are grafted elsewhere, they still differentiate into nerve, cartilage, skin and muscle.

Atkinson et al. (2006) investigated how dedifferentiation can involve a general histolysis of internal tissues, the re-entry of normally quiescent cells into the cell cycle, the down-regulation of markers of cell differentiation and the up-regulation of blastemal markers. Its regulatory determinants could include morphogenetic gradients and cell surface molecules. Stem cells in salamanders, together with some blastema cells, might therefore retain a memory of their position before tissue injury, whereas others are respecified and essential information of their position passes along the proximal–distal axis. Redifferentiating blastemal cells form internal tissues in regenerating structures except for nerve axons, so severe injury induces regeneration and the formation of a supernumerary limb.

This work has led to other novel forms of tissue regeneration in mammals. Some molecules stimulate regeneration in salamanders, and this knowledge has led to studies on stimulating regeneration in mammals by invoking dedifferentiation and preventing scarring. The key difference between these species is the blastema which includes a mass of totipotent cells in various amphibians as described above, and which is usually absent in adult mammals (Zazaian, 2004). Curiously, mammalian fetuses can regenerate limbs, and amphibians can generate the body parts as described above, even though it seems that blastemas are absent from the human genome. This reliance on a blastema is paraphrased from a comment by Badylak who states it is ‘a mass of cells that can turn into any number of body parts in amphibians but not in most mammals’. Perhaps it might be possible to direct blastema formation in mammals by the use of specific drugs, growth factors or cellular intervention.

Masaki and Ide (2007) recently assessed the capacity for regeneration in mouse limbs. They revealed that when limbs of neonatal mice are amputated, stump bone at the amputation site undergoes hypertrophy after cells have proliferated and cartilage has formed, although certain elements of bone and cartilage do not form. They also assessed the resulting changes by grafting mouse embryo limb buds into amputated limbs of neonatal mice at the level of the digit. The buds now grew and differentiated into a segmented pattern of bone, cartilage and soft tissues, a pattern that became more complicated if the limb buds were in more advanced stages. Masaki and Ide (2007) also assessed grafted limb buds to determine whether they developed autonomously or interacted with stump tissue when forming a limb. They also investigated if limb buds could be disaggregated and then reaggregated before they were grafted to a recipient. Reaggregated limb bud cells formed digit-like structures composed of bone and segmented cartilage and, when grafted to the stump, formed segmented cartilage. These investigators concluded that stump tissues partially permitted cartilage and bone to form in a procedure that enhanced limb formation by providing both endogenous and exogenous supplies of competent cells.

Section 6.3. The genetic regulation of regeneration in urodeles

Several geneticists have contributed to understanding the complexities of regenerating damaged tissues in urodeles (e.g. Zhang et al., 2003). Voss et al. (2001) commented that salamanders possess few chromosomes, although the size of their genome is larger than the human genome. A mapping panel was constructed by Atkinson et al. (2006) who crossed Ambyostoma mexicanum and Ambyostoma tigrinum tigrinum to construct a lineage map. This revealed that gene markers segregated in a Mendelian manner. Gene loci coding for proteins were also mapped in an attempt to clarify the structure of the amphibian genome and identify orthologous loci between salamanders and other vertebrates including humans.

These investigators also applied advanced molecular techniques to assess the significance of genetic regulation in blastema (Atkinson et al., 2006). One approach was to apply electroporation, i.e. applying an electric field to modify electrical conductivity and permeability in cell membranes. This method offers an experimental form of dedifferentiation, although care is needed when interpreting the resulting data since it can induce appendage cells to re-enter their cell cycle, internal cells to undergo histolysis, and raises the expression of differentiation markers in blastema (Hojman et al., 2007). While being helpful for gene transfer, high doses might also invoke necrosis or apoptosis, and up-regulate similar genes.

Identical forms of gene expression occurred in newt limbs that had been either electroporated or amputated (Table 9). The consequences of electroporation were similar at histological and molecular levels, although only amputation led to the regrowth of new appendages (Atkinson et al., 2006). Four members of the matrix metalloprotease MMP family were involved, including nCol and MMP3/10b which exhibited similar responses to amputation or electroporation, and MMP9 and MMP3/10 which were up-regulated after both types of injury. Most proteolytic activity in the limb arose as a consequence of the high ubiquitous expression of MMP3/10b, since this and other matrix metalloproteases prevent scarring during regeneration (Vinarsky et al., 2005).

As mentioned above, salamanders possess a large genome of 22–48 billion base pairs. This led Habermann et al. (2004) to utilize expressed sequence tag (EST) sequencing to obtain sequence information on embryonic and regenerating blastema cDNA libraries of A. mexicanum. Libraries were prepared from embryos and from regenerating tail blastemas at day 6, and sequenced EST revealed 6377 contigs, which possibly represented 25% of genes in this species (Table 10). Sequence comparisons also revealed homologies with entries in the non-redundant NCB database. Some developmental genes were involved in cell proliferation and differentiation, whereas others regulated communication between adjacent cells. Evolutionary relationships were also found to involve cell-cycle proteins of the cyclin B family, whereas the cyclin-dependent kinase inhibitor 1 gene family exhibited unusual evolutionary behaviour.

EST utilize molecular probes to study development, provide clones for constructing DNA microchips, identify candidate genes for mutant phenotypes, and assist in studies on genomic structure and evolution. EST from two ambystomatid salamanders, namely Ambyostoma mexicanum, the Mexican axolotl, and Ambyostoma tigrinum tigrinum, both of them members of the tiger salamander complex, were assessed by Putta et al. (2007). Differing in their life history, the former retains many larval features and is waterborne throughout life, whereas the latter undergoes a metamorphosis typical of many amphibians. Two cDNA libraries were therefore constructed in order to obtain transcripts derived mostly from regenerating larvae in Ambyostoma mexicanum, and from non-regenerating tissues in Ambyostoma tigrinum tigrinum. Approximately 46,000 quality cDNA sequences in regenerating limbs and tails of these salamanders were combined, together with an existing set of 16,030 cDNA sequences for A. mexicanum. This resulted in the availability of 35,413 and 20,599 high quality EST for A. mexicanum and A. t. tigrinum respectively. A total of 21,091 unique contigs were obtained overall, of which 10,592 displayed close similarities with sequences identified from the human RefSeq database which reflects diverse arrays of molecular functions and biological processes. These contigs closely resembled sequences identified in injured rat spinal cord and in regenerating fins in zebrafish.

Such resources help to target novel characteristics in organisms by isolating EST from specific tissues, and are useful, in the authors' own words, in searches of databases to identify probes for regeneration research. They characterized intra- and interspecific nucleotide polymorphisms, saturated the human–Ambystoma synteny group with marker loci and extended PCR primer sets designed for A. mexicanum⁄A. t. tigrinum orthologues to a related tiger salamander species. Comparative insights crucial for understanding diversity and variability in biological systems will be needed when establishing the genomic resources for closely related species.

Monaghan et al. (2007) applied oligonuclotide microarrays to help identify genes that control the regenerative ability of the spinal cord in A. mexicanum. Regenerating tissues were sampled at five time points after tail amputation, which led to the identification of genes producing increased amounts of RNA over the first week of regeneration. More than 1000 genes were involved, with 360 of them having a dominant role in predominant expression patterns and gene functions. Monaghan et al. (2007) therefore concluded that diverse injuries are related to mechanisms involved in extracellular matrix remodelling during the acute phases of spinal cord regeneration. They also found similarities and differences in gene expression between their results and the genes expressed following spinal cord injury in rats, and commented that their studies might enhance the discovery of genes and their functions in mammals.

Section 6.4. Positional information in regenerating urodeles

As mentioned earlier, organ regeneration in urodeles seemingly involves cells that recognize their positional situation along three axes of a morphogenetic field (Mercader et al., 2005). These investigators also interpret their proximal position in the remaining stump as an aid to full repair. Retinoic acid and the Meis-homeodomain proteins specify the proximal position for regenerating cells, a situation clarified by isolating the Meis homeobox family and assessing its roles in specifying proximal regeneration. Mercader et al. (2005) also discovered that retinoic acid regulates the Meis genes in axolotls during limb regeneration and development, and that distal blastemal cells relocate to more proximal sites during limb generation. Meis genes were proved to be significant targets for the proximalizing activity of retinoic acid on blastema cells, a situation inhibited by Meis knockdown.

Restoring a functional central nervous system including sensory, inter- and motor neurons is a unique feature of tail regeneration in axolotls. It is nevertheless unclear how spinal cord regenerates after tail amputation or how neural progenitor cells are recruited. McHedlishvili et al. (2007) assessed the size of a mature spinal cord that is able to produce regenerating spinal cord. They tracked single and groups of cells in an attempt to assess their spatial distribution along the A/P and D/V axes (Table 11). Their maps of cell fates revealed a 500 µm zone of spinal cord adjacent to the amputation plane, where neural progenitors numbering ∼800 were generated for an expanding ependymal tube. They also tracked proliferating single cells during regeneration by assessing their spatial distribution and their expression of the progenitor markers PAX7 and PAX6. Descendants were generated and expanded along the A/P axis but remained close to the D/V location of the parent. A minority of cells had assumed different molecular identities spanning multiple D/V domains, which indicated that they could become multipotent.

McHedlishvili et al. (2007) extended their studies by bulk-labelling dorsally and ventrally restricted progenitor cells. They revealed that ventral cells located at the distal end of the regenerating spinal cord switched to cells with dorsal fates, and that PAX7 and PAX6 expression revealed these markers were distributed all along the regenerating cord except at the distal terminus. There, neural progenitor cells seemed to be destabilized or altered in the region of the terminal vesicle from where cells migrate into surrounding blastema. McHedlishvili et al. (2007) concluded that they had observed a trail of cells leaving the dorsal surface of the terminal vesicle and entering the blastema during tail regeneration. They also stressed that the terminal vesicle is where cell transitions occur from epithelial to mesenchymal and where cells may delaminate from the ependymal tube to join surrounding tissues of the blastema.

The only vertebrates capable of fully regenerating limbs that have been amputated are anurans such as frogs and urodeles including newts and salamanders. Wnt/β‐catenin is essential in sustaining early stages of limb generation but not after the blastema has formed or if this pathway is suppressed. Wnt-3a is expressed in the apical epithelium of regenerating limb buds in Xenopus laevis and has a role in blastema formation (Yokoyama et al., 2007). Suppressive agents interfere with this pathway and block regeneration in salamanders, while activating agents turn the pathway on, just as in chicks that regenerate a wing.

Section 6.5. Hox genes and regeneration

The numerous genes identified in blastemas and regenerating tissues in urodeles include the closely related homeobox genes Meis1 and Meis2. They encode transcription factors of the TALE class of homeodomain proteins which share a three-amino-acid loop between the homeodome helices 1 and 2 (Bürglin, 1997). Meis homeodomain proteins consist of Meis1 and Meis2 which are closely regulated by homeobox genes that are activated by retinoic acid as the limb axis is being patterned. The limb bud is subdivided into Meis-positive and Meis-negative domains that are each essential to form the proximodistal axis.

This evidence led Mercader et al. (2005) to conclude that the Meis pathway might exclusively associate with a patterning network that is reactivated after amputation by the positional memory system after proximal amputation. The status of Prod1, a cell-surface molecule that is anchored to the cell surface by GP1, may determine activation in newts, since it is abundant in proximal blastemas and is up-regulated by retinoic acid. It is expressed in higher levels in upper regions of the arm, where its levels are regulated by retinoic acid. This situation offers an explanation as to how a blastema can transform a wrist to an upper arm and recognize differences between proximal and distal cells.

Hox genes modify the positional memory of blastema cells in all three limb axes in regenerating larval and adult urodeles. They can proximalize the PD level of origin of the blastema, so that blastema cells derived from the wrist assume the positional identity of the shoulder girdle. Retinoic acid simultaneously posteriorizes and ventralizes blastema cells, and mediates the activation of nuclear receptors belonging to the steroid receptor superfamily of transcription factors. The amounts of retinoic acid that gain access to nuclear receptors are regulated by binding cytoplasmic proteins and by nuclear receptors that bind retinoic acid via retinoic acid receptors (RAR) and via retinoid X receptors (RXR) that bind to retinoic acid response elements.

Hox and shh genes may also regulate spatial patterns in limb buds and in regenerating limbs in urodeles. 5′Hox A, C and D genes are active in regenerating limbs, in spatial patterns in limb buds and in regenerating limbs. Genes of the HoxD cluster (i.e. 10, 11, 13) are also expressed asymmetrically in developing chick wing buds, HoxD-10 and HoxA-13 being the only genes known to respond to retinoic acid (Stocum, 1996).

Nicolas et al. (2003) studied the distributions of Hoxa9 transcripts and three 5′HoxC genes along the axis of the nervous system in newts to discover the levels of gene expression in adult and regenerating tail tissues. They also assessed the expression of these genes in spinal cord of the adult newt Pleurodeles waltl. This was done after their limbs and tail had been amputated, epithelium had formed from cut edges of skin, epidermis had migrated over the amputation surface, and undifferentiating and dividing mesenchymal cells arising from the stump had accumulated under wound epithelium and were forming the regeneration blastema. As the tail regenerated, the ependymal tube was initiated and extended into the blastema, and regenerating spinal cord exerted significant effects on the proliferation and condensation of mesenchymal blastemal cells into a tight pattern.

Nicolas et al. (2003) also assessed the distributions of Hoxa9 and three 5′HoxC genes along the A/P axis of the central nervous system in adult newts. They used a reverse transcription polymerase adult and regenerating tissue chain reaction (Figure 12). A Pleurodeles regenerating tail DNA library was also screened with degenerated probes for helices α1, α3, and α4 of the homeodomain proteins. Twelve different homeobox-containing genes were isolated, the most abundant including partial sequences for PwHoxa9 and full sequences for two members of the HoxC complex, namely PxHoxc12 and PwHoxc13 (Table 12). Unique sequences isolated for PxHoxa10 did not contain the whole homeobox region, and were homologues of two previously identified amphibian Hoxc10 genes. The Pleurodeles genes Hoxc12 and Hoxc13 shared 100% and 95% amino acid identity within the homeodomain sequences reported for human homologues, and contained residues outside the homeodomain that were characteristic among Hox12 and Hox13 paralogs.

Fig. 12 Expression of Hox genes along the anterior–posterior (A/P) axis of the adult nervous system in newts Pleurodeles waltl (Nicolas et al., 2003). The central nervous system is dissected into 11 parts, ranging from the brain to the most caudal spinal cord. The graph shows the relative distribution of HoxA9, HoxC10, HoxC12 and HoxC13. A phospho-imager was used to quantify the data and the relative amount of Hox DNA was normalized to the amount of GAPDH cDNA. A correction factor was applied to superimpose the graphs in such a manner that total expression levels among the 11 samples was equivalent for each gene. Differential restrictions along the A/P axis arose in the spatial expression domain of each Hox gene, and two independent studies revealed identical variations of Hox gene expression in adult central nervous system. The lower panel shows Southern blot analysis utilizing a specific Hox cDNA on RT-PCR products obtained from different sections of the adult CNS. The various sections included: 1, brain; 2, cervical and brachial; 3–5, trunk; 6, lumbar; 7–11, tail. Reproduced with permission from Nicolas S, Papillon D, Perez Y et al., 2003, Biology of the Cell, volume 95, 589–594. © Portland Press Ltd.

Putative variations of PwHox expression levels ranging from brain to the most caudal spinal cord in the central nervous system of adult newts were also investigated. PwHoxc10 was expressed in a sharp domain from the trunk to the distal part of the tail, its expression in the trunk increasing gradually until the level of the hind limb. When placed in a more posterior position, its levels decreased slightly towards the distal part of the tail. Maximal levels of expression for each PwHoxC gene occurred at specific axial positions in adult spinal cord, with PwHoxc13 being restricted to the level of forelimb. Four PwHox genes reached significantly higher levels in regenerating tissue than in the tail of adult newts which led to Nicolas et al. (2003) arguing that Hox genes have a role in maintaining positional information in the amphibian central nervous system, and that the ability of urodeles to regenerate spinal cord is due to the relatively primitive organization of their central nervous system (CNS). They also suggested that the 5′Hox genes specified positional memory in adult spinal cord and could be involved in axial patterning in the regenerating tail. They concluded that Hox genes are associated with tissue remodelling and the renewal of differentiated cells although information on mammals is scant. They also stressed that molecular models helping to maintain positional memory are poorly understood, although recent work revealed their activity in mammary glands, adrenal glands and muscles.

As described above, the precise regeneration of the amputated structures depends on cells at the level of the amputation plane having a positional memory, and a continuum of positional value along appendages able to regenerate. Nicolas et al. (2003) cloned four AbdB-like genes found in the newt Pleurodeles wat1. They included Hoxa9, Hoxc10, Hoxc12 and Hoxc13, which are genes that are expressed along the antero-posterior axis according to the colinearity rule (Figure 12). Expression levels of PwHoxc13, PwHoxc12, and PwHoxc10 were respectively 20‐, 7‐ and 2‐fold higher in tail regenerates than in adult tail, thereby specifying axial patterning occurred in regenerating tail and positional memory in the adult tail (Nicolas et al., 2003). In adult animals, PwHox genes are expressed along overlapping domains along the A/P axis of the spinal cord, and are related to spatial restriction of 5′HoxC in spinal cord. Maintaining the order of 5′HoxC in agreement with the laws of colinearity implied these genes helped to maintain a continuum of positional values along the A/P axis of the tail in adult amphibians). The coordinated expression of 5′HoxC may maintain cellular proliferation and delay cell regeneration, its expression boundaries specifying the territory for regenerating tails and axis formation.

To conclude this section on urodeles, Stocum (1996) along with other investigators opined that limb and tail blastema inherit level-specific memories of their axial positions from cells of the stump and that mature tail cells seemingly retain a memory of their identity along the A/P axis. Nicolas et al. (2003) suggested that their co-linear Hox gene expression pattern (i.e. lying on the same straight line) indicated that genes may specify regional identity along the main body axis. This was confirmed in mice with gain or loss of Hox gene function, where a combination of Hox transcripts expressed at particular levels were good candidates for carrying positional information.

Section 7.1. Ear repair in MRL mice

MRL (Murphy Roths Large) mice were being used to study immunology and large size in mice when unexpected findings led to an astonishing study on tissue regeneration. Heber-Katz (1999) was studying the effects of body weight on immune systems in MRL mice that had to be ear-clipped as a means of identification. Punching small holes in the ear has long been the classical method of identifying individual mice. These ear holes have been widely used by numerous scientists and last for a lifetime. Heber-Katz and her colleagues were therefore surprised to discover some days later that the holes they had punched in the ears had disappeared. Each hole had been quickly repopulated, plugged and repaired by innate stem cells, and new hair follicles had covered the wounds within 30 days (Figure 13). Such a rapid and complete wound closure, resembling regeneration rather than wound repair, must have surprised these investigators. Notable features about the repair was that it had not involved scarring for up to 4 weeks and had resulted in a form of healing that involved chondrogenesis, was inheritable and under genetic control. A second notable observation was that other strains of mice had unblocked their ear holes.

Fig. 13 Regenerative systems in the MRL mouse (Clark et al., 1998). (a) An MRL mouse, its large size being due to selection for fatness. (b) The repair of holes in the ear in MRL mice is shown on the right, although it is lacking in C57BL/6 mice as shown on the left. (c) Ear hole punch closure in MRL mice between 0 and 30 days after cryoinjury shown on the right. Reprinted from Clinical Immunology and Immunopathology, Clark LD, Clark RK, Heber-Katz E, volume 88, A new murine model for mammalian wound repair and regeneration, pages 35–45, 1998, with permission from Elsevier.

It did not take a long time for investigators to realize they had made an unexpected discovery, so they decided to assess the nature of the genetic control of repairing holes in the ear (Clark et al., 1998). McBrearty et al. (1998) performed a genome-wide scan on F1 crosses between MRL/MpJ-Fas^lpr × C57BL/6) F2 mice and on backcross populations to identify the genes partaking in wound healing. If the nature of inheritance in F1 hybrids was to be decided, the two parental phenotypes had to differ, a method that revealed quantitative trait loci (QTL) on chromosomes 8, 12 and 15, at two distinct locations on chromosome 13, and possibly on chromosome 7. All alleles contributing to full wound closure were derived from the MRL/MpJ-Fas^lpr parent except for a quantitative trait locus on chromosome 8 which was derived from C57BL/6 mice.

Researchers began to study the roles of single or multiple genes in regulating this new-found ability of ear regeneration (Table 13). Quantitative genetic factors and particular gene loci were found to have roles in ear-healing in MRL mice. Several genes were identified and studied in various backcrosses, together with more than 11 QTL involved in wound healing (McBrearty et al., 1998; Heber-Katz et al., 2004). Multigenic traits located on various chromosomes, especially chromosome 17, were therefore studied in intercrosses between different strains of mice. Several were found to be sexually dimorphic, i.e. different genes controlled wound healing in the two sexes and female mice healed more rapidly than males (Blankenhorn et al., 2003).

Examples of QTL linked to the healing phenotype were identified again in later studies. Masinde et al. (2001) decided to study crosses between MRL mice and the non-healing parental strain SJL/J. A total of 633 MRL/MPJ × SJL/J F2 mice were assessed in order to identify QTL, using high-density genome-wide scans applied on days 15, 21 and 25. Ten were identified on chromosomes 1, 4, 6, 7, 9, and 13, which, when taken together, explained 70% of the degree of variance in the F2 mice. Further studies at various time points revealed eight more QTL on chromosomes 1, 4, 6 and 9 that were involved in wound healing. Backcrossed mice shared several QTL and helped to identify smad3 as a candidate gene involved in healing, while MRL/MPJ × SJL/J F2 mice displayed epimorphic regenerative responses to ear hole punching. Masinde et al. (2005) applied restriction fragment differential display PCR and isolated genes that were differently expressed identified in MRL and C57/BL mice. They also identified 36 genes that were candidates for scarless healing at various stages of wound healing. Gourevitch et al. (2003) concluded by restating that 20 known gene loci were involved in the closure of the ear holes (e.g. Blankenhorn et al., 2003), that MMP‐2 was located on chromosome 8 and that tissue inhibitor of metalloprotease (TIMP)‐2 was located on chromosome 11.

Some time later, a study on MRL x C57BL/6J mice revealed how the genetic control of healing holes in the ear was controlled by 11 QTL (Heber-Katz et al., 2004). Considering it desirable to clarify the underlying genetics, these investigators decided to compare results with a poorly healing strain of mice and to investigate mice of the inbred mouse subspecies Mus musculus castaneus. Their study confirmed a number of previously-known loci including three with quantitative traits, and a strong sexual dimorphism in MRL × C57BL/6 mice. This combination of strains had revealed strong sexual dimorphism in both healing and quantitative traits (Table 13) (Leferovich et al., 2001; Heber-Katz et al., 2004).

Differences between MRL and C57BL/6 mice in their responses to ear wounding were examined in detail by Gourevitch et al. (2003). They compared the post-injury remodelling of extracellular matrix and the responses of matrix metalloproteases and tissue inhibitors in the two mouse strains. They discovered a correlation between the ability of MRL mice to break down the basement membrane, to form a blastema and to close ear wounds. An inflammatory response was found to involve neutrophils and macrophages expressing MMP‐2, MMP‐9, TIMP‐2 and TIMP‐3 after ear injury. MMP‐2 and MMP‐9 proteins extracted from healing tissues in the ear of MRL mice were examined on days 0, 1, 3, 5, and 7 post ear-punching (Table 14). At all these stages, the pro- and active forms of MMP9 were consistently higher in MRL mice, which also expressed higher levels of the active form of MMP‐2. MRL mice also expressed higher levels of other proteins. Antibodies specific for MMP‐2 and MMP‐9, TIMP‐2 and TIMP‐3 revealed these proteins accumulated near the wound site. A strong increase in cell numbers occurred between days 1–3 post-wounding, four-fifths of them being small inflammatory cells. Examining large cells revealed that 30% were mast cells and 70% were macrophages that were expressed in larger numbers in MRL mice as compared with C57BL/6 mice. Levels of RNA for MMP9 followed those identified for proteins and were consistently higher in MRL mice, whereas MMP‐2 levels were higher in unwounded ears, some of these proteins being expressed before the cells arrived at the wound site.

These results led Gourevitch et al. (2003) to conclude that metalloproteases have an important role in the provisional matrix formed soon after wounding by breaking down type IV collagen, fibronectin, entactin and elastin, which are components of extracellular matrix. Cellular and fibrillar debris surrounding the wound bed were clarified and helped keratinocytes and fibroblasts to migrate through the basement membrane. Such events occur in newts and axolotls where they are regulated by thyroid hormone. They stress the importance of the inflammatory responses and the expression of TIMP‐2 and TIMP‐3 in both MRL and C57BL/6 mice after wounding, followed by more MMP cells in MRL mice.

Section 7.2. Regeneration of injured heart in MRL mice

After discovering the nature of regenerating ear holes in MRL mice, Leferovich et al. (2001) and Heber-Katz et al. (2002) decided to assess the consequences of cryoinjury to the right ventricle of the heart (Table 15). Tissue repair depends on differentiating stem cells and related factors such as the need for myocardial infarction to induce cardiac myocytes to proliferate and simultaneously provide D‐ribose in order to sustain new cells. ADP will also be needed to sustain ATP production by mitochondria and so provide ample energy until levels of ATP return to pre-infarct levels.

Myocardium was thought to lack stem cells and be unable to regenerate spontaneously, yet injuries inflicted by the cryoprobe were regenerated. For 1 week after injury, cells at site of the injury in MRL mice formed ‘fingers’ of cardiomyocytes that extended into the site of the wound while fibroblastic cells lined up ‘in a superstructural configuration’ although with a loose appearance (Table 15). This unusual ability in MRL mice to heal their myocardial injuries also involved replacing wounded tissue without fibrosis and with little scarring. This characteristic trait in MRL mice was regulated by seven or more genetic loci, and sexual differences were observed (Leferovich et al., 2001).

By 15 days after wounding, the heart of MRL mice appeared to be normal with cardiomycytes undergoing regular mitoses at levels 10 times higher than in controls. Cardiomyocytes in MRL mice moved early to the wound site as they divided, proliferated and restored blood supply at the site of injury. They restored normal cardiomyocardial architecture, and their high mitotic index began to fill wound sites. By day 60, mitotically active cardiomyocytes had filled the wound site without forming scar tissue, and their proliferation was marked by their uptake of BrdU. All this was accompanied by the rapid resolution of granulation tissue and a reduced scarring as myocardium recovered from injury. There was a marked contrast between the regenerating heart in MRL mice where little scarring had occurred and the situation in C57BL/6 mice where scar formation was heightened. Scarring that occurs after responding to an injury in control mice and human hearts can even contribute to heart disease and death (Oliff, 2006).

The preadipocyte factor 1 gene (Pref‐1) is expressed in proliferating cells where it maintains their undifferentiated state. The application of microarrays, reverse transcriptase-polymerase chain reaction, in-situ hybridization and immunohistochemistry identified Pref‐1 in the healing ears of MRL and C57BL/6 mice (Samulewicz et al., 2002). It was expressed more highly in the wounds of MRL mice and was found uniquely in a condensation of cells within regenerating blastemal tissue where it apparently contributes to the powers of regeneration of ear wounds of MRL mice. It was up-regulated some time after wounding and just as the mature cells in the blastema were about to redifferentiate into mature cells.

The breakdown of extracellular matrix is important in the regeneration of both the tail and the heart in MRL mice (Leferovich et al., 2001; Leferovich and Heber-Katz, 2002). He commented that the key to the response of MRL mice to injury also resembled events occurring in regenerating newt limbs where levels of collagen type I protein declined continuously as the blastema formed and matrix metalloproteases degraded existing proteins (see Johnson and Schmidt, 1974).

Mammalian myocardium does not contain reserve cells, and terminally differentiated adult cardiomyocytes do not usually proliferate or regenerate. Scar formation is the major response to injury. In contrast, amphibians display an ability for cardiac regeneration and the division of cardiomyocytes, which is seen very rarely and to a minimal degree in humans and other mammals (e.g. Beltrami et al., 2001).

Following severe cryoinjuries to the right ventricle of the heart, adult MRL mice display recoveries in structure and function, and the replacement of their myocardial tissue resembles events in amphibians. In contrast, the response in non-regenerating C57BL/6 mice is a predominant scar. When radiation chimaeras were reconstituted with fetal liver cells from MRL or C57BL/6 mice, the healing response to cardiac cryoinjury was predominantly donor, which differs from the closure of ear holes in being regulated by fetal liver stem cells (Bedelbaeva et al., 2004). All these studies on the heart of MRL mice had clarified the various aspects involved in the repair of injury, the roles of stem cells, the dedifferentiation of mature adult cardiomyocytes, angiogenesis and apoptosis in MRL mice whereas these factors had not been found in control C57BL/6 mice as discussed by Leferovich et al. (2001) and Heber-Katz et al. (2002).

Section 7.3. Healing the central nervous system in MRL mice

Regenerative responses to injuries inflicted on the spinal cord of MRL/MpJ mice were also assessed in detail to discover if similar findings arose in relation to stab wounds to the heart (Table 16). Their responses resembled damage to ear punch holes, and again displayed enhanced healing responses when compared with those found in other mouse strains. When damage was inflicted, Seitz et al. (2002) discovered that scar tissue blocked axonal regrowth. Its functions recovered when fibroblastic infiltrates were maintained at a minimum after transecting the spinal cord, to result in coordinated walking within 3 weeks. Apolipoprotein E and its receptors seemed to be up-regulated during these regenerative responses, and several molecules capable of blocking scar function were tested for their effect on healing and function (Seitz et al., 2003).

Two differing forms of injury to the CNS were applied to assess the regenerative properties of the MRL/MpJ mouse. A cortical stab was used to assess glial scarring and a scouten knife lesion since this method has been well tested. Axonal regeneration in the CNS was assessed by lesioning the dopaminergic neurons that project in the median forebrain bundle to the striatum so that regeneration could be seen easily and quantified (Hampton et al., 2004). Regeneration can be seen easily activated, and also functions longer by interacting with its enhanced receptors in MRL/MpG mice. Perhaps increased glial reactions and raised inflammatory responses in the CNS of MRL/MoJ mice increases their healing ability at the periphery (Hampton et al., 2004).

Adult female MRL/MpJ mice and adult female Swiss Webster mice were given identical treatment and were autopsied on days 2, 4, 7 and 14 after the lesion, and micrometer sections were made for examination (Table 16). Immunohistochemistry was used to classify cell types, to count cells in alternating 0.01-mm squares and to measure leakage of immunoglobulins from circulation to the brain. Hampton et al. (2004) also isolated RNA by homogenizing these tissues and then assessing them using real-time reverse transcription (RT-PCR). These authors concluded that healing ability was apparently greater in MRL/MpJ mice at the cost of raising inflammatory responses in the periphery as seen by increased glial reactions in the central nervous system.

Motor functions are lost following spinal cord transections in adult mice, and the recovery of function after complete transection can occur without intervention provided that dural injury, the displacement of the ends of cut cords, and the infiltration of fibroblasts are minimized. Underlying this function, the expression of GAP-43 in axonal growth cones, axonal extension and bridging the injury site together with the neuronal remodelling of both white and grey matter can occur without intervention (Seitz et al., 2002).

Comparisons between MRL/MpJ and Swiss Webster mice after the lesion was made included overall cell numbers, which were lower in the former, especially around the edge of the lesion. After a cortical stab, rates of cell division were greater in both strains as compared with unlesioned brain. Between days 2–7, live cells were present at the edge of the lesion as compared with uninjured cortex and with untreated Swiss Webster mice (Table 16) (Seitz et al., 2002, 2003). Cell division ceased in both mouse strains by day 14. Astrocytes labelled with anti-glial fibrillary acidic protein identified the beginning of the target region (the striatum) rostral to the lesion. Numbers on various days to day 14 post lesion were higher close to the lesion as compared with normal unlesioned brain. Labelled MRL/MpJ astrocytes also extended further into the lesion in MRL/MpJ mice as compared with Swiss Webster mice.

Section 7.4. The blood–brain barrier in MRL/MpJ and Swiss Webster mice

Numbers of microglia and their reactivity increased after lesioning as an influx occurred involving cells resembling blood-borne macrophages⁄monocytes (Hampton et al., 2004). The investigators summarized the situation as resembling more reactive changes in microglia after injury, which remained for longer and had a morphology resembling macrophages. Breakdown of the blood–brain barrier due to the cortical stab injury led to leakages that were more widespread in MRL/MpJ mice than in Swiss Webster mice, whereas differences did not rise between the two strains of mice regarding the fibroblastic invasion into the lesion following the cortical stab. Neither strain displayed any signs of axonal regeneration.

Numbers of MMP and thrombin receptors differed between MRL/MpJ mice, while levels of MMP‐2 RNA rose after injury with the greater response occurring in MRL/MpJ mice than in Swiss Webster mice. PAR1 thrombin receptor RNA levels remained constant in Swiss Webster mice but varied transiently in both strains, especially in response to trauma. Hampton et al. (2004) concluded that significant differences in the glial reaction to injury in MRL/Mpg mice had emerged to show there was a widespread and intense set of changes following the lesion. Such differences included an increased microglial reaction as seen in the expression of complement receptor and morphological changes in MRL/MpJ mice, a higher and more widespread cell division, and a greater breakdown and loss of cells surrounding the lesion at earlier timepoints.

Such differences disappeared at later time points, so that by day 14 post-lesion there were no differences between these mice and normal controls. The basis of such differences included the nature of factors responsible for the increased injuries in MRL/MpJ mice. Hampton et al. (2004) conclude that thrombin receptors are up-regulated at the sites of injury. Thrombin therefore seems important in regenerative responses since it regulates the re-entry of cell cycles in newt myotubes and in pigmented epithelial cells in adult newt iris (Brockes, 1997; Tanaka and Brockes, 1997; Imokawa and Brockes, 2003). Perhaps the selective activation of thrombin acts on a particular serum protein and that, once activated, it can function longer during regeneration by interacting with its enhanced receptors in MRL/MpG mice. Perhaps increased glial reactions and raised inflammatory responses in the CNS in these mice increases their healing ability at the periphery (Hampton et al., 2004).

Tail regeneration involves lineage-shifting events in which radial glial-like cells (so-called ependymoglial cells) produce not only neurons and glia but also integrate into newly formed muscle and cartilage. Adult newt brain can reactivate quiescent cells on injury and contains sufficient extracellular cues to direct activated neural progenitors towards a specific neurotype within an existing brain structure, so the activation or manipulation of its cellular and molecular programmes may contribute to regeneration in animal models (Minelli et al., 1990).

Section 7.5. Summary

Results gained in the laboratory were established and summarized by Ellen Heber-Katz (2007). Topics covered included basic research on the genetics and molecular biology of autoimmunity, wound healing and regeneration, and its application to numerous human diseases including heart disease and damage to tissues in the spinal cord. Many of these topics are relevant to wound healing and regeneration, and were pursued in force when the unusual amphibian-like properties of MRL mice were identified. They have since identified the genes involved in wound healing and the mechanisms permitting such events to occur. These include the up-regulation of matrix metalloproteases soon after the wounds were inflicted and prior to the formation of a blastema, just as in urodeles. Her most recent studies indicate that damage to liver, kidneys and toes is also mended in MRL mice, and that these and other examples of regeneration emerged mostly as a result of complex gene recombinations rather than by transformations involving single genes.

Section 8.1. The therapeutic value of contrasting forms of tissue repair

Wide therapeutic opportunities have been opened by new approaches to tissue repair and organ regeneration. It is not easy to foresee their relative benefits although a selected few, including ESC and tissue stem cells, may be in clinical use within a few years. They are preferred at present in view of their many advantages while much remains to be learnt about tissue regeneration in urodeles and MRL mice. ESC may nevertheless be gradually replaced by tissue stem cells and cord blood stem cells. This is shown by claims that ESC are ineffective in human patients and create tumours in animal recipients (e.g. Wakitani et al., 2004). Other investigators including Hughes (2007) have no doubt that tissue stem cells should be preferred. They also stress that stem cells can be extracted from numerous tissues including fat, bone marrow, placentae, neurons, olfactory tissue, eyes and teeth among others.

Traynor et al. (2000) and Johnstone and Sara (2003) also prefer tissue stem cells and stress their ability to form heart tissue, neural matter, skin cells, and other tissues. They are already being used to treat autoimmune diseases, multiple sclerosis, Crohn's disease, rheumatoid arthritis, cardiac misfunction, heart attacks and corneal degeneration. These methods also improve movement in patients with spinal injury and restore vision to patients who are blind or suffer from age-related macular degeneration (Traynor et al., 2000; Coffey, 2007). Moreover, fetal tissue stem cells can repair multiple sclerosis, and have opened novel approaches to gene therapy. In this context, the availability of effective vectors and the use of artificial chromosomes may have improved prospects of new forms of cell-based tissue engineering via the use of artificial chromosomes (Oshimura and Katoh, 2008).

Umbilical cord blood stem cells that have been introduced recently and offer more than either ESC or tissue stem cells. They are multipotent, easy to collect, and can be stored for long periods. They are thus available for parents wishing to repair inherited deficiencies in their children and in older family members. They are already used therapeutically for conditions such as Fanconi anaemia, leukaemia, and blood disorders and may soon be available for patients with diabetes, heart disease and nerve damage for both related and unrelated recipients (Laughlin et al., 2001; Ghen et al., 2006). Cord blood cells also avoid the ethical issue of using human blastocysts from IVF clinics (Edwards, 2007), although they have a limitation in amplifying and scaling up subpopulations without signs of ageing and/or phenotypic changes.

Research on urodeles and MRL mice has opened further opportunities for tissue repair and regeneration, although available evidence on their efficiency in mammals is still inadequate. Tanaka (2003) may well have been correct in remarking that ‘the ultimate goal of regenerating a fully functional tissue or organ is still distant’, whereas Heber-Katz et al. (2006) presents a more positive viewpoint when she predicts that it may soon be applied to humans since tissue regeneration has now been achieved in her MRL mice (Heber-Katz et al., 2006). She states ‘I believe the day is not far off when we will be able to prescribe drugs that cause severed spinal cords to heal, hearts to regenerate and lost limbs to regrow. People will come to expect that injured or diseased organs are meant to be “repaired from within”, and that in 50 years whole-body replacements will be routine’ (Heber-Katz et al., 2006). That may be so, but would highly complex human characteristics such as memory, musical ability, emotions and happiness survive repairing damaged tissues from within?

Section 8.2. The genetic regulation of regeneration in urodeles and MRL mice

Studies on the genetics of human fetuses, urodeles and MRL mice should soon identify their similarities and homologues. The underlying genetics that sustain human fetuses in this manner should clarify why human genes apparently switch off soon after finger regeneration. Nor should there be any difficulty in discovering how these genes are first activated in fetuses and then inactivated at particular stages of growth during early adulthood. Applying this knowledge clinically could well take a considerable time, so it seems inevitable that most clinicians will continue with ESC and tissue stem cells for many years to come. Caution will also be needed about applying regenerating systems involving a blastema in order to repair damaged human tissues. They could well lead to unknown anomalies in treated patients, and such reservations will be relieved only when increased knowledge on regeneration in MRL mice clarifies its risks when applied to a mammal.

Organ regeneration occurs in several species, including human fetuses, rabbits, deer, urodeles and MRL mice, which indicates that they all share numerous homologous genes. For example, those active in early human fetuses can be inactivated in later stages of growth, presumably due to modifications in chromatin, RNAi (RNA interference) and other forms of gene silencing. Perhaps novel genes were activated coincidentally when MRL mice were being selected for large body weight together with genes for blastema formation. Numerous genes are involved in these processes, although single mutations seem to be inadequate to explain this conversion. Nor have any attempts been made to discover if the ‘totipotency’ genes described in mice by Takahashi and Yamanaka (2006) are expressed in blastemal cells. There is nevertheless no doubt that numerous genes become active and interact with each other in salamanders and undergo fortuitous recombinations involving new phenotypes. It is already known that, in MRL mice, at least 20 regenerative genes are involved, including MMP‐2 and MMP‐9 that, when up-regulated, inhibit metalloproteases. These mice also express TIMP‐2 and TIMP‐3 that are later down-regulated. Widening techniques are clearly needed such as lineage tracing and gene knockdown in a model organized to provide information on regeneration.

Adult mammalian tissues must also retain positional information if they are to restore normal growth, differentiation and organization in damaged tissues. An example in Cdx2^+/− mice involves the transformation of intestinal regions to tissues resembling oesophageal epithelium, and the modification of cells between oesophagus and intestine (Tosh and Slack, 2002). Adult tissues may thus retain a latent activity to regenerate cells belonging to specific tissues, colonize different adult organs and form cell types with differing identities as if they were being pushed backwards (Tanaka, 2003). Receptive organs are weakly colonized initially by a small number of cells and do not rebuild their entire structure in recipients. Queries therefore arise as to whether cells that colonize have to retrace their previous developmental patterns. Among vertebrates, salamanders are the best regenerators and can repair limbs, tails, eyes, jaws and heart. Since regeneration is not confined to vertebrates, it is possibly an ancestral trait in salamanders that was suppressed in many lineages during evolution.

Assuming that ESC and tissue stem cells currently offer better approaches to tissue repair for the time being, it is essential to understand how they originate and differentiate in the preimplantation embryo. As described earlier in this text, two totipotent blastomeres in the 4‐cell stage seemed to be the originators of ICM, while one of the remaining blastomeres was trophectoderm precursor and the fate of the fourth was uncertain (Figure 8) (Hansis and Edwards, 2003). One of the tent cells may have been the precursor of ESC and tissue stem cells, and are apparently distributed to different sites throughout the body. The work of Kurimoto et al. (2006) confirmed that ICM possessed at least two distinct cell lines as described above, although numerous investigators still question this theory on the differentiation of early blastomeres.

Section 8.3. The future courses of various forms of regeneration

The ability of MRL mice to regenerate tissues could open a clinical lifeline to humans since their capacity to ‘repair from within’ offers a model of continuous generation and life extension (Heber-Katz et al., 2006; Heber-Katz, 2007). Ear holes in rabbits also heal spontaneously, just as in MRL mice (Hampton et al., 2004), and while promising, such research in animals will be merely a single step on a long journey. When enough knowledge on their properties has been gained, novel forms of tissue repair to humans could well replace the need for ESC, cord blood and tissue stem cells and avert the need to prepare and cryostore millions of them (Lindblad, 2004).

Regenerating organs might offer further advantages since preparing ESC and other stem cells may involve unexpected risks. When cultured in vitro, they become sensitive to various epigenetic modifications. These include preparing culture media with a high osmotic pressure when levels of ∼250 mOsm/kg are optimal. Should these levels rise to ∼300 mOsm/kg, embryo growth is impaired and p38 mitogen-activated protein kinase (MAPK) is activated, such modifications of culture media harming embryos in vitro (Purdy, 1982; Fong et al., 2007b). In another example, preimplantation bovine embryos cultured in media containing fetal calf serum gave birth to very large calves (i.e. large calf syndrome) that impaired parturition. Cells in culture can also be damaged by neighbouring cells or if placed in a new environment (Tanaka, 2003). Cell fusion is a risk arising when rare populations of tent bone marrow stem cells are cultured with feeder layers. In one study, they became tetraploid after fusing with neighbouring cells and were discarded (Jiang et al., 2002). The composition of culture media may also be harmful, as shown in various ways, and the damaging effects of fetal calf serum have been known to include cultured mesenchymal cells being induced to express cardiac troponin‐1 and N‐cadherin (Wang et al., 2004). Errors also emerged involving retroviruses used when converting mouse somatic cells to stem cells since tumours were found in many offspring (Takahashi and Yamanaka, 2006).

Correct methods of culturing stem cells are being provided by the International Stem Cell Initiative (see International Stem Cell Initiative, 2007). They comment that culture techniques vary from laboratory to laboratory which results in stem cell lines with differing properties. By now, a large central bank of ESC has been organized and well-characterized human ESC lines have now been checked and stored. Cells with diverse genotypes or those that had been exposed to differing culture methods were assessed for their purity, and each was found to be normal. Cells of the 59 human embryonic stem cell lines in store at the stem cell bank correctly expressed the keratin sulphate antigens SSEA3, SSEA4, TRA‐1–60. TRA‐1–81, GCTM2 and GCT343, the protein antigens CD9 and Thy1, alkaline phosphatase and class I HLA. Developmentally regulated genes including nanog, Oct‐4 (now named PousF1), Tdgf1, Dnmt3B, GABRB3 and CGF3 characterize cultured ESC. As expected, several lineage markers were also expressed, and female but not male cell lines expressed Xist. Checks are also made to ensure that no stem cell lines are contaminated with mycoplasma, bacteria or cytopathic viruses, confirming the methods used when handling stem cells had clearly been unsurpassable.

Urodelan methods of organ regeneration will doubtless damage cells and tissues and thwart ideas of applying them to humans. Little is known about their characteristics and their genetic regulation, and Stocum (1996) stresses that a void must be filled to gain a comprehensive understanding on gene systems and patterning mechanisms in urodeles. Digits and legs are not regenerated in higher vertebrates because wound epidermis is non-functional, and the viewpoint that innervation is inadequate for limb regeneration in higher vertebrates remains unproven. Tassava and Olsen (1982) postulated that, in urodeles, wounded epidermis is non-functional and maintains dedifferentiated cells during cell cycles in the amputated limb stump so that differentiated or undifferentiated cells do not maintain in their cell cycle and blastema fail to form. Instead, tissue regeneration occurs precociously due to the absence of a cycling stimulus. Another consequence of a non-functional wound epidermis is that scar tissue forms at the limb tips of non-regenerating vertebrates whereas, in urodeles, wound epidermis maintains dedifferentiated cells of the amputated limb stump in the cell cycle. Blastema can then be formed in view of the sufficient numbers of cell divisions in the amputated limb stump.

Tabata (2008) has stated his own position when he wrote that: ‘in the case of reconstruction surgery, biomedical devices cannot completely substitute the biological functions even for a single tissue or organ’, whereas Steven Badylak, Director of the Centre for Pre-clinical Tissue Engineering at the Department of Surgery, University of Pittsburgh is more optimistic. He states that it is essential to show that mammals can form a blastema, and that human digits can be regrown after 48 months. In other words, he wishes to have a human blastema constructed within 48 months in order to regrow digits, understand the nature of genes controlling regeneration in human fetuses, and identify genes that are turned on or off when regeneration is undertaken. He also wishes to succeed with human regeneration modelled on that existing in salamanders, and the Pittsburgh team also wishes to prove that mammals can provide essential needs for progenitor cells just as in salamanders. They will also have to find a mouse model that is incapable of tissue regeneration, and so orchestrate the formation of a blastema as opposed to causing a scar after tissue injury. The MRL mouse may show him the way to that target, although there are reports, as yet unconfirmed, that genes regulating blastemas are absent in the human genome.

Section 8.4. Applying these techniques to humans

An obvious difficulty is that risks will have to be taken when applying these methods to humans. Several researchers have nevertheless stressed that various mammals could soon be treated with forms of regeneration resembling those functioning in urodeles and MRL mice. Fetal and newborn mice can spontaneously repair the tips of foretoes in a form of regeneration related to the expression of Hox7 (Borgens, 1982; Neufield, 1989; Reginelli et al., 1995); children are able to repair amputated finger tips (Douglas et al., 1972; Illingworth, 1974); and rabbits can repair holes in the ear (Goss and Grimes, 1975). Deer antlers also offer a form of regeneration in mammals that resembles that occurring in urodeles, since they can grow blastema to form new antlers when the original ones have been lost as a result of infighting between male deers. These events were described by Price and Allen (2004), who concluded that the underlying methods of repair were poorly understood, yet it is clear that human fetuses, rabbits, deer, urodeles and MRL mice each possess similar genes regulating tissue repair, and that those that are active in human fetuses are inactivated in adults.

These forms of repair are mediated by extracellular matrices that help to avoid scarring and immune responses so that damage is soon repaired. An aspect of this method of repairing human fingers involves the use of a powdered extract of extracellular matrix largely composed of collagen that is applied to wounds (Layton, 2007). She pointed out that similar extracts prepared from pig bladder have been applied to injured soldiers whose hands were wounded in Iraq. They have also been used to repair reopened wounds, mend torn ligaments in horses, treat human ulcers by closing the hole in tissues lining the stomach, and may repair spinal injuries, amputated limbs and damaged organs in humans. Fibrin is another natural biomaterial that helps to avoid immune rejection and can be prepared from the patient (Lutolf and Hubbell, 2005).

Layton (2007) stressed how an adult who accidentally severed off a finger tip had it initially covered with a skin graft and then with powdered extract of pig's bladder (which contains no cells). It then regrew the severed tip together with nerves, skin, fingernail and other tissues. Similar events occurred when a boy of 12 years who burnt his hand was diagnosed as having a second or third degree burn. His burnt hands were mended by removing dead tissue and applying mesenchymal cells in a sterile gel and with a porous gauze to create a multilevel scaffold. Pain had gone within 24 h, new tissue had healed the wound in 21 days and infection and inflammation were absent. More evidence is clearly needed to confirm this success was gained by the patient, confirm reports that other organs could also be repaired in this manner, and demonstrate that research on damaged human fingers or finger tips could be a driving force leading to human treatments similar to methods of regeneration in urodeles and MRL mice. When a human finger is severed, cells in the region of excision normally die, and their leftovers seep into neighbouring tissues and invoke localized immune responses. The immune system is accordingly warned about the onset of inflammation and the formation of scar tissue, which prevent any further cell division.

Removal of the damaged tissues, and placing MSC in a sterile gel with porous gauze created a multilevel scaffold for these cells. The boy now had no pain after 24 h, had repaired the wound after 21 days with tissues resembling those originally there, and had suffered neither pain nor inflammation (M Ghen, personal communication). Wounds in a series of children due to amputated or trapped finger tips have also been repaired. They were covered with a thick layer of Tulle-gras which comprises a gauze cloth cut into squares and, when impregnated with paraffin, is used as a wound dressing. When left alone for 2 weeks, the children's wounds were totally repaired both functionally and cosmetically within 12 weeks without any need for antibiotics (Illingworth, 1974).

Endless questions emerge about the factors regulating the formation and activation of a blastema. They include the following queries:

• (i) Are there any similarities between the organization of stem cells into a blastema in salamanders and those formed in MRL mice?
• (ii) Are blastemal cells in urodeles distributed in numerous tissues in a manner resembling the wide distribution of tissue stem cells in mammals?
• (iii) Are internal organs including brain in urodeles generated by means of blastemas?
• (iv) Can MRL mice regenerate an amputated limb?
• (v) Did genetic selection for obesity in MRL mice have a role in triggering genes involved in tissue repair?
• (vi) Are the ‘totipotency’ genes Oct‐4, Sox2, c-Myc and Klf that were identified by Takahashi and Yamanaka (2006) active in salamanders and MRL mice and in human fetuses as they repair their fingers, liver, kidneys and toes?
• (vii) Can several blastemas function simultaneously in different parts of the body, and can various types of stem cells be administered simultaneously to patients to renovate several tissues at the same time?
• (viii) Are blastemal genes present or absent from the human genome, which form of gene silencing suppresses them, and are the regenerative genes active in human fetuses also active in blastema?
• (ix) Do blastemas renew a shortage of germ cells?
• (x) Is a degree of longevity granted by grafts of ESC, tissue stem cells, cord blood stem cells and other forms of stem cells?
• (xi) Are genetic defects mended in any way by tissue regeneration in MRL mice?
• (xii) Can disaggregated blastemal cells grown in vitro repair organ and tissue damage in normal and histoincompatible recipients?

We look forward to receiving answers to these questions.