Dedifferentiation and the role of sall4 in reprogramming and patterning during amphibian limb regeneration


  • Anton W. Neff,

    Corresponding author
    1. Indiana University School of Medicine-Bloomington, Indiana University Center for Regenerative Biology and Medicine, Bloomington, Indiana
    • IUSM-Bloomington, Jordan Hall, Bloomington, IN 47405
    Search for more papers by this author
  • Michael W. King,

    1. Indiana University School of Medicine-Terre Haute, Indiana University Center for Regenerative Biology and Medicine, Terre Haute, Indiana
    Search for more papers by this author
  • Anthony L. Mescher

    1. Indiana University School of Medicine-Bloomington, Indiana University Center for Regenerative Biology and Medicine, Bloomington, Indiana
    Search for more papers by this author


A central feature of epimorphic regeneration during amphibian limb regeneration is cellular dedifferentiation. Two questions are discussed. First, what is the origin and nature of the soluble factors involved in triggering local cellular and tissue dedifferentiation? Secondly, what role does the key stem cell transcription factor Sall4 play in reprogramming gene expression during dedifferentiation? The pattern of Sall4 expression during Xenopus hindlimb regeneration is consistent with the hypothesis that Sall4 plays a role in dedifferentiation (reprogramming) and in maintaining limb blastema cells in an undifferentiated state. Sall4 is involved in maintenance of ESC pluripotency, is a major repressor of differentiation, plays a major role in reprogramming differentiated cells into iPSCs, and is a component of the stemness regulatory circuit of pluripotent ESCs and iPSCs. These functions suggest Sall4 as an excellent candidate to regulate reprogramming events that produce and maintain dedifferentiated blastema cells required for epimorphic regeneration. Developmental Dynamics 240:979–989, 2011. © 2011 Wiley-Liss, Inc.


The importance of cell and tissue dedifferentiation for organ regeneration was clear from the earliest histological studies of limb, tail, and lens regeneration in amphibia (Morgan,1901). That cells of salamander limb regeneration blastema could be derived solely from local tissues of the amputated limb stump was demonstrated early in the 20th century (Stocum,1995; Carlson,2007). In urodeles, extensive histolysis of internal limb tissues after epidermal closure of the wound releases proliferating mesenchymal cells to form a distal population of undifferentiated cells called the blastema, which undergoes patterning and produces new tissues integrated with those more proximal in the limb stump. Myogenic cells within the blastema originate both from reserve satellite cells of dedifferentiating muscle (Morrison et al.,2006,2010) and by fragmentation of the multinucleated muscle fibers with cell cycle re-entry of individual mononucleated cells (Lo et al.,1993; Kumar et al.,2000). Other limb tissues in urodeles may also contain similar reserve cell populations (Young,2004).

Complete regeneration in limbs of anuran amphibians only occurs, following amputation in hindlimbs, at early developmental stages in premetamorphic larvae, before significant cytological differentiation of mesenchymal cells of the limb buds. As studied in larval Xenopus laevis, once differentiation begins in cell of developing hindlimbs, regeneration after amputation is normally incomplete and the ontogenetic decline in limb regenerative capacity correlates with the degree of differentiation in limb tissues at the time of amputation (Dent,1962; Korneluk and Liversage,1984; Khan and Liversage,1990). Histological analysis of postmetamorphic Xenopus limbs indicates that amputation produces very limited histolysis locally and negligible dedifferentiation of stump tissues, with most distal muscle fibers remaining intact (Korneluk and Liversage,1984).

Dedifferentiation involves reprogramming of differentiated or committed cells of the limb or limb bud into less differentiated, multipotent, proliferating mesenchymal cells, which accumulate to form the blastema from which the amputated portion of the limb is reformed. The mechanisms that allow dedifferentiation after tissue injury and the plasticity or reprogramming potential of dedifferentiated limb cells are of key importance in the field of vertebrate regeneration and recent work in this field has been reviewed by several investigators (Brockes and Kumar,2002; Odelberg,2005; Straube and Tanaka,2006; Stocum and Zupanc,2008; Tsonis,2008; Christen et al.,2010; Tamura et al.,2010). The present review addresses two questions about dedifferentiation in regenerating amphibian limbs, which are discussed in light of new studies in other models of regeneration or cell reprogramming. First, what is the origin and nature of the soluble factors involved in triggering local cellular and tissue dedifferentiation? Secondly, what role does the key stem cell transcription factor Sall4 play in reprogramming gene expression during dedifferentiation?


The term “dedifferentiation” was used historically in studies of limb regeneration to refer to the demolition of the differentiated distal tissues of the limb stump and the release of cells that will form the regeneration blastema. In this sense, dedifferentiation of tissues in amputated urodele limbs includes the simultaneous enzymatic degradation of most existing extracellular matrix components, formation and release from tissues of mesenchymal cells resembling those of the embryonic limb bud, and concomitant re-entry of those cells into the cell cycle. In the blastema of an amputated early Xenopus limb bud, the “dedifferentiated” cells are likely derived from the population of mesenchymal cells already present at the time of amputation. As a descriptor of this key early regeneration phase, the concept of dedifferentiation was based mainly on light and electron microscopic observations. Except for studies of certain major hydrolytic enzymes that act on the extracellular matrix (ECM), the molecular basis of the events that trigger dedifferentiation in amphibian regeneration remains poorly understood (Straube and Tanaka,2006; Carlson,2007).

It has been known for some time, however, that the initial stimuli for dedifferentiation clearly emerge in the aftermath of traumatic injury to the limb and are not dependent on the other tissue interactions required for blastema formation and regeneration (Straube and Tanaka,2006). A denervated salamander limb, in which there is little cell proliferation and no formation of a blastema, nevertheless undergoes tissue dedifferentiation and entry of the mesenchymal cells into the cell cycle (Mescher and Tassava,1975; Barger and Tassava,1985). Dedifferentiation and cell cycling are also seen in an amputated limb that is unable to form distal wound epithelium, the major signaling center required for blastema formation (Mescher,1976; Loyd and Tassava,1980). The dedifferentiated histological nature of limb stump tissues becomes more apparent in the presence of nerve axons and the wound epithelium as cells continue proliferating and give rise to a blastema under the mitogenic influence of factors released specifically from these two sources. However, initial dedifferentiative events such as histolysis by matrix metalloproteinases and cell cycle re-entry are triggered by the tissue injury alone (Tassava and Mescher,1975; Yang and Bryant,1994).

The injury-dependent stimuli for dedifferentiation in the amputated amphibian limb are likely to be similar to those for cell activation and ECM breakdown during the initial, inflammatory phase of wound healing and tissue repair in other animals. Study of inflammation in many different mammalian systems has revealed various major sources of matrix hydrolases, mitogens, and other signaling factors that stimulate the coordinated events of cell proliferation, migration, and ECM remodeling, which culminate in repair of the injured tissues (Gurtner et al.,2008; Medzhitov,2008). Three such sources will be discussed briefly here, along with relevant studies of epimorphic regeneration.


A highly complex array of proteases and bioactive molecules begins to appear immediately at the time of injury from extravasated blood platelets and plasma proteins. Contact of platelets with collagen causes their immediate exocytosis of various hydrolytic enzymes, specific protease inhibitors, growth factors, thrombospondin, and many vasoactive compounds such as histamine and serotonin. Tissue factor, a membrane-bound glycoprotein present in platelets, on leukocyte surfaces, and in subendothelial vascular tissues, interacts with specific polypeptide factors present in plasma to catalyze the cleavage of plasma prothrombin to thrombin, which in turn cleaves plasma fibrinogen into the clot-forming protein fibrin. The extravascular fibrin clot not only serves to prevent excessive blood loss and as a substrate for leukocytes entering the damaged tissue from the blood, it also functions as a depot for the slow release of numerous bioactive peptides formed during the many proteolytic steps of the coagulation and complement cascades (Markiewski and Lambris,2007). Serum, produced in vivo only at sites of injury, contains the myriad mitogens, enzymes, and other regulatory substances released from platelets, leukocytes, and the proteolytic cascades.

Tanaka and colleagues have clearly shown that proteins activated by thrombin and plasmin proteolysis induce dedifferentiation and cell cycle re-entry in vitro of newt myogenic cells that differentiated as multinucleated myotubes when levels of fetal bovine serum were lowered to 0.5% (Tanaka et al.,1999). Straube and Tanaka (2006) have reviewed the evidence that the still unidentified S-phase re-entry factor assayed in this cultured myotube system is also produced by serum treatment with plasmin, but not with various more specific proteases; is not abolished by delipidation or dialysis of other small compounds from serum; and acts via phosphorylation of the retinoblastoma protein. Moreover, additional factors are required for nuclei of the myotubes to undergo mitotic activity and form mononucleated myoblasts. During lens regeneration in newt eyes, thrombin activity is also required for dedifferentiation of epithelial cells of the dorsal iris (Imokawa and Brockes,2003). Also in this system the thrombin activity is generated by localized expression of tissue factor in the dorsal iris (Godwin et al.,2010).

Other active serum factors that have been implicated in dedifferentiation, reprogramming, and cell activation during regeneration are two key components of the complement activation cascade, C3 and C5. Besides being generated in serum, both of these complement components are newly expressed by cells in both regenerating limbs and lens of newts. Although neither factor is found or expressed in normal limb or lens cells, regenerating limbs express C3 in mesenchymal blastema cells and C5 in the wound epithelium (Del Rio-Tsonis et al.,1998; Kimura et al.,2003). During newt lens regeneration, C3 is expressed in the stroma and dedifferentiating pigmented epithelial cells of the iris and C5 in new lens vesicle (Kimura et al.,2003). C3, as well as other components of the complement pathway, is seen to be differentially expressed when comparing Xenopus regeneration-complete and regeneration-incomplete limb blastemas and pseudoblastemas (Grow et al.,2006). C3 and C5 mediate a wide spectrum of functions during inflammation and their expression in these regenerating systems suggests their involvement in sustaining important aspects of regeneration after the initial inflammatory response is resolved.


Neutrophils, macrophages, lymphocytes, and other white blood cells enter tissues of the amputated limb immediately from the cut blood vessels and continue to cross the endothelium of intact vasculature by diapedesis during the period of dedifferentiation. Activated leukocytes release a large number of growth factors, cytokines, and other regulatory factors, as well as matrix metalloproteinases and other proteases specific for various ECM components. Expression studies of the collagenase Mmp-9 during dedifferentiation of amputated amphibian limbs suggest that neutrophils and macrophages, whose major roles are removal of microorganisms and debris, respectively, may also be important sources of this protease. In situ hybridization of larval Xenopus (Carinato et al.,2000), adult newt (Vinarsky et al.,2005), and axolotl limbs (Yang et al.,1999), especially with sectioned material, reveals Mmp-9 -expressing cells as scattered cells located throughout the forming mesenchymal cells and the wound epithelium. In larval Xenopus limb stumps explanted to organ culture immediately after amputation, expression of various markers of dedifferentiation occurs normally, but not that of Mmp-9 (our unpublished observations). In newt limbs, Mmp-9 and related proteases are also highly expressed in bone marrow cells of the humerus (Vinarsky et al.,2005).

Whatever the source of the MMPs during regeneration, their activity is required for normal growth and patterning of the new limb (Vinarsky et al.,2005) and work in many systems indicates that this activity includes many important effects related to cellular plasticity besides remodeling of the ECM (Odelberg,2005; Heissig et al.,2010). For example, besides degrading collagen, MMP-9 also specifically cleaves and activates precursors for the regulatory factors TGF-β, Interleukin-1, and TNF-α, as well as several chemokines important for cell migration during regeneration and repair (Stamenkovic,2003).

The importance of leukocytes has been examined in various regenerating systems after interference with expression of the gene PU.1, which is required for myelopoiesis. Despite the absence of neutrophils and macrophages, regeneration occurred normally in tails of PU.1 morphant zebrafish larvae (Mathew et al.,2007) and skin wounds in adult PU.1 null mice healed in a scar-free manner like that normally found only with fetal mouse skin (Martin et al.,2003). Tail regeneration in Xenopus larvae was also found to be potentiated during its refractory period (stages 45–47) in animals injected with PU.1 morpholinos at the two-cell stage (Fukazawa et al.,2009). These studies not only indicate that most proteolytic enzymes and mitogenic factors released from neutrophils and macrophages in wounds and amputated appendages can be replaced with similar activities from other sources, they also emphasize the possibility that these cells may mediate processes inhibitory to regeneration or scarless healing (Mori et al.,2008).

Damaged Cells and ECM

The ECM of vertebrate tissues sequester large amounts of polypeptide growth factors, including the fibroblast growth factors, bound mainly to heparan sulfate and other components of proteoglycans. Upon tissue injury, these mitogens are released and activated by the hydrolases acting throughout the ECM locally (Stamenkovic,2003; Mott and Werb,2004). Moreover, fragments released from various collagens themselves also directly regulate cellular activities, for example fine-tuning the rate of new capillary outgrowth (Mott and Werb,2004).

Immunological studies indicate that several ECM component fragments and substances released from damaged cells, such as ATP and the chromatin protein HMGB1, act as endogenous “danger signals” causing macrophages and other leukocytes to release their stores of mitogens and hydrolases and activating dendritic cells and other antigen-presenting cells to initiate site-specific T- and B-cell responses within the site of injury (Matzinger,2007; Medzhitov,2008; Manfredi et al.,2009). Together with signals triggered by microbes introduced with the injury, the endogenous danger signals released specifically during cell necrosis (but not apoptosis) not only elicit a local immune reaction appropriate for the situation, but also help produce a microenvironment favorable for tissue repair.

Cellular Reprogramming and Plasticity During Dedifferentiation

Classic descriptive studies of regenerating amphibian limbs strongly suggested that each tissue of the limb stump contributes most of the cells for the corresponding tissue in the regenerate and that the dedifferentiated mesenchymal cells of the early blastema normally remain lineage-restricted (Goss,1969; Mescher,1996; Brockes and Kumar,2005). Proliferating blastema cells appear as a homogeneous population although they emerge from diverse tissues and make up a heterogeneous population of progenitor cells, heterogeneity that has been demonstrated in newt limbs by clonal analysis and monoclonal antibodies (Brockes and Kumar,2005). Recent work using axolotl limb tissues labeled with an integrated green fluorescent protein (GFP) transgene have again shown that blastemal cell types are largely restricted to their own tissue identity during the dedifferentiative phase of regeneration, without extensive metaplasia or transdifferentiation (Kragl et al.,2009). Like previous studies, that of Kragl et al. (2009) found evidence of plasticity only among “fibroblasts,” with labeled cells of dermis contributing to other connective tissue types in the regenerate, including tendons and cartilage. Overall, the evidence from regeneration studies in amphibians indicates that blastema cells are not pluripotent but at best only multipotent (Christen et al.,2010).

Dedifferentiation of tissue in the amputated amphibian limb involves reprogramming of differentiated cells into a less differentiated, multipotent mesenchymal state. In recent years, many of the molecular regulatory pathways involved in reprogramming differentiated mammalian cells into a more undifferentiated state and the stemness genes responsible for maintaining this state have been uncovered. The observation that forced expression in differentiated cells of a limited set of transcription factors associated with pluripotency in embryonic stem cells (ESC) resulted in induced pluripotent stem cells (iPSC) (Takahashi and Yamanaka,2006; Takahashi et al.,2007; Yu et al.,2007) provides insights into the dedifferentiation process during amphibian limb regeneration (Christen et al,2010). Reversion of differentiated cells to a stem cell–like state (reprogramming) involves epigenetic phenomena such as changing chromatin structure (Shafa et al.,2010), expressing new specific miRNAs (Mallanna and Rizzino,2010; Melton et al.,2010; Nakamura et al.,2010), and establishing defined transcriptional networks with a limited number of pluripotency factors, including Oct4, Sox2, c-Myc, Klf4, Nanog, Lin28, Tbx3, and Sall4 (Yang et al.,2008,2010; Yu et al.,2008; Han et al.,2010; Li et al.,2010; van den Berg et al.,2010). These reprogramming factors are involved in both positive autoregulatory loops and repressive activities that prevent differentiation (Yang et al.,2008).

Expression of such factors, including Sall4 (Neff et al.,2005), Tbx3 (Grow et al.,2006), Sox2 and c-Myc (Maki et al.,2009; Christen et al.,2010), has been shown in cells of regenerating amphibian limb blastemas. Unlike many genes related to limb patterning, such as Shh and Lmx-1, which are not expressed uniformly throughout the blastema, stemness genes such as Sall4 and c-Myc show pan-blastema expression (Neff et al.,2005; Christen et al.,2010). Consistent with the conclusions from the cell lineage studies of regenerating limbs, the proteomics profile of dedifferentiated cells in the early blastema of regenerating larval Xenopus limbs is more similar to that of multipotent “adult” tissue stem cells rather than embryonic stem cells (ESC) (King et al.,2009).

Takahashi (2010) states that in ESC, Sall4 acts as a “tower of pluripotency” in association with Oct4, Sox2, and Nanog. Other recent work in cell reprogramming emphasizes that Sall4 is one of the few genes, if not the only one, that is not only involved in the stem cell properties of ESC, but also those of adult tissue stem cells (Yang et al.,2010). We will discuss here evidence suggesting that Sall4 may play a central role in cell dedifferentiation and reprogramming during amphibian limb regeneration. Sall4 genes are present in Ambystoma (Putta et al.,2004; Smith et al.,2005), but its expression during urodele limb regeneration has not yet been investigated. Our discussion will focus on studies with the anuran Xenopus laevis where more molecular analyses and data from studies of limb regeneration are available.


SALL4 (Sal-like protein 4) is a multiple C2H2 zinc finger-containing transcription factor that belongs to the family of evolutionarily conserved Spalt/Sal transcription factors that can act both as activators or repressors of gene expression involved in tissue and organ formation (Sweetman and Munsterberg,2006; de Celis and Barrio,2009). Like the Spalt genes involved in both cell identity (homeotic) and patterning during Drosophila development, the vertebrate Sall gene family members, including Sall4, control similar functions related to cell identity and patterning during early vertebrate development (Uez et al.,2008). Human Sall4 mutations lead to Okihiro Syndrome (also known as Duane-Radial Ray Syndrome), most likely caused by loss of function and haploinsufficiency, which produces hypomorphic thumbs among other defects (Kohlhase et al.,2002; de Celis and Barrio,2009). In general, Sall4 is a key co-regulator of pluripotency in ESC, participating in positive autoregulatory loops (Yang et al.,2010). Sall4 plays a role in the transcriptional regulation of the key ESC genes Oct4, Nanog, and Sox2 (Zhang et al.,2006; Lim et al.,2008; Yang et al.,2008) and decreased Sall4 expression inhibits expression of these genes as well as c-Myc and Klf4 in ESC (Yang et al.,2008). Murine ESC contain two isoforms, Sall4a and Sall4b, which can make homodimers and heterodimers with different DNA binding sites and exhibit different roles during early embryogenesis, including differences in proximal-distal and anterior-posterior patterning (Uez et al.,2008).

Onuma and colleagues (1999) found alternative spliced variants of Sall4 expressed during early Xenopus embryogenesis, which they referred to as Xsal-3 isoforms. The long form (Xsal-3long) was detected in unfertilized eggs and embryos, where it gradually decreased after the neurula stage. The short form (Xsal-3short) was absent maternally and did not appear until the early gastrula stage. The Sall4 probe used for whole-mount in situ hybridization (WMISH) and the Sall4-specific PCR primers used by Neff et al. (2005) did not discriminate between these isoforms, so it is unknown if they play differential roles during limb regeneration. Expression analysis and functional studies of the short and long forms of Xenopus Sall4 during limb regeneration at the regeneration-complete and -incomplete state has the potential to provide insight into dedifferentiation/reprogramming.

SALL4 is an enhancer (positive regulator) of differentiated cell reprogramming toward an embryonic stem cell–like state and the reprogramming activity of SALL4 is associated with its transcription repressor activity located in the N-terminal domain of the protein (Wong et al.,2008). Reprogramming of somatic cells is a very inefficient process (Hockemeyer et al.,2008), and Sall4 has been shown to enhance the efficiency of iPSC generation from fibroblasts (Tsubooka et al.,2009). SALL4 can form a complex with OCT4 and NANOG, which then controls the expression of many genes in ESC (Yang et al.,2008). Sall4 functions both as an activator or repressor of gene transcription depending on the context and at high doses Sall4 can inhibit the expression of Oct4 in ESC as well. These activities position Sall4 as a key regulator of ESC pluripotency, self-renewal, and inhibition of differentiation. SALL4 binds to about twice as many gene promoters as NANOG in ESC and binds about four times more genes than OCT4 in ESC (Yang et al.,2008). SALL4 modulates chromatin/histone remodelers such as NurD, SWI/SNF, and BAF complexes and thereby regulates ESC gene expression and many ESC genes bound by SALL4 are marked by such bivalent chromatin modifications (Takahashi,2010).

Sall4 may be unique among the ESC pluripotency transcription factors in being expressed in non-ESC stem cells that do not express Oct4 (Yang et al.,2010) or Sox2 (Forte et al.,2009). Sall4 is expressed in extraembryonic endoderm cells, which are stem cells that do not express Oct4 (Lim et al.,2008), fetal liver stem/progenitor cells but not adult liver stem cells (Oikawa et al.,2009), neural crest stem cells (Barembaum and Bronner-Fraser,2010), mesoderm progenitor cells (Pacini et al.,2010), hematopoietic stem cells (Yang et al.,2007), endometrium cells (Forte et al.,2009), parathyroid gland stem cells (Shih et al.,2009), and multipotent adult progenitor cells (Ulloa-Montoya et al.,2007), but is not expressed in fully differentiated cells. As expected, Sall4 is expressed in cells of adult mouse ovaries and testes (Kohlhase et al.,2002). As ESC are induced to differentiate, both splicing isoforms of Sall4 are gradually down-regulated and by the time full differentiation has occurred they are no longer expressed (Rao et al.,2010).

As mentioned earlier, limb regeneration in anurans such as Xenopus, unlike urodele limb regeneration or zebrafish fin regeneration, which occurs in adults, is biphasic. At prometamorphic stages (stages 48 through 52/53) when cells of the developing limb are largely undifferentiated, amputation is followed by complete regeneration with the recapitulation of the normal anterior-posterior (A/P) patterning. Limb amputation at the premetamorphic stages (stages 57–59) and postmetamorphic stages, when limb tissues are functional and almost fully developed, results in regeneration that is incomplete, producing at the later stages a single patternless (hypomorphic) spike consisting mainly of skin-covered cartilage (Neff et al.,2005; Ohgo et al.,2010). We have exploited this difference in Xenopus regenerative abilities in differential gene and protein expression assays (King et al.,2003,2009; Grow et al.,2006). Xenopus Sall4 (originally identified as XSal-3 and reclassified as XSALL4 but hereafter referred to as Sall4) was initially cloned as a gene expressed during early embryogenesis and up-regulated in animal caps exposed to activin and retinoic acid (Onuma et al.,1999). Sall4 was identified in regeneration blastemas in a subtraction hybridization screen of regenerating stage-53 hindlimbs versus whole limb (King et al.,2003). Sall4 was subsequently found to be differentially expressed in a microarray screen comparing gene expression during limb regeneration at regeneration-complete and regeneration-incomplete stages (Grow et al.,2006). Further studies of Sall4 expression (Neff et al.,2005) suggested a role in dedifferentiation (reprogramming) and in maintaining regenerating limb blastema cells in an undifferentiated state. This was based upon WMISH-determined patterns of expression as well as qPCR analysis of Sall4 expression levels during hindlimb regeneration at the regeneration-complete stage and the regeneration-incomplete stage. These studies, with representative data shown in Figure 1A, form the basis for the following conclusions:

Figure 1.

Expression of Xenopus Sall1, Sall3, and Sall4 during hindlimb regeneration at the regeneration-complete stage and the regeneration-incomplete stage. A: Top; whole-mount in situ hybridization (WMISH) of Sall4 in stage-53 and -57 limbs and in stage-53 and -57 regenerating hindlimbs at 0, 1, 3, and 5dPA. Bottom; RT-PCR comparison of Sall1, Sall3, and Sall4 expression patterns in stage-53 and -57 regenerating hindlimb blastemas and pseudoblastemas at 0, 1, 3, and 5dPA. Amputation was through the mid-zeugopodia and blastemas (from stage-53 limbs) or pseudoblastemas (from stage-57 limbs) were collected for analysis at 0, 1, 3, and 5dPA. At 0dPA about 1 to 2 mm of tissue from the distal end of the amputated limb was used as the internal tissue control. Odc (ornithine decarboxylase) was used as an internal control. WMISH data was taken from Neff et al. (2005) with the authors' permission. B: QPCR comparison of Sall1, Sall3, and Sall4 expression patterns in stage-53 and -57 regenerating hindlimb blastemas and pseudoblastemas at 0, 1, 3, and 5dPA. Sall gene expression levels are normalized to odc and then shown relative to stage-53 0dPA levels of expression, which is arbitrarily set to 1.0. RT-PCR and QPCR were performed as described in King et al. (2009).

  • 1Sall4 expression is restricted to the mesenchyme of early limb buds and as the limb develops, Sall4 expression becomes progressively more restricted, with a P/D and A/P patterned loss of expression. As limb structures differentiate, they no longer express Sall4. As soon as cartilage condensation occurs, Sall4 is no longer expressed. The interdigital spaces are the last regions of developing limbs to express Sall4; by stage 57 when limb differentiation is well underway and A/P patterning is complete, Sall4 expression can no longer be detected. These observations are consistent with Sall4 expression patterns in developing chick (Neff et al.,2005; Barembaum and Bronner-Fraser, 2010) and mouse limbs (Kohlhase et al.,2002; Koshiba-Takeuchi et al.,2006) and with Sall4 gene expression in stage-52 hindlimbs (Christen et al.,2010).
  • 2In response to limb amputation through the mid zeugopodia at a regeneration-complete stage (st 53), Sall4 is expressed within one day post-amputation (1dPA) in the dedifferentiation zone and by 3dPA it is expressed throughout the blastema. Expression does not occur in cells of the epidermis or wound epidermis. These observations were recently confirmed by WMISH localization of Sall4 expression in stage-52 0d, 1dPA, and 3dPA blastemas (Christen et al.,2010). By contrast, at the regeneration-incomplete stage57, Sall4 is not detectable by WMISH in 0d, 1dPA, and 3dPA pseudoblastemas, but is expressed at 5dPA and 7dPA in the pseudoblastemas (Neff et al.,2005). These observations suggest that Sall4 functions differently from Shh in influencing A/P patterning because it is expressed in early limb buds before Shh is expressed, it is expressed in amputated limb blastemas at the regeneration-competent stage before Shh is expressed, and it is expressed in pseudoblastemas where there is no Shh expression and A/P patterning does not occur (Endo et al.,1997). Christen et al. (2010) further investigated regeneration and reprogramming during Xenopus hindlimb regeneration at regeneration-complete (stage 52) and regeneration-incomplete stages (stage 57) by assaying (qPCR) the expression of pluripotency factors associated with iPSC reprogramming, including Oct4, Sox2, c-Myc, Lin28, and Sall4 in 0d, 1dPA, 3dPA, and 5dPA blastemas and pseudoblastemas. The developing and regenerating limb expression profile of these genes differed from the ESC-like blastomere gene expression profile. Stage-52 limb regeneration blastemas did not express Oct4 or Lin28 and other genes such as Sox2 and Sall4 were up-regulated at low, statistically insignificant levels. Despite the low level of Sall4 expression measured by qPCR after hindlimb amputation at stage 52, the WMISH experiments presented by Christen et al. (2010) showed expression of Sall4 in stage-52 1dPA and 3dPA blastemas. These investigators amputated stage-52 limbs through the presumed zeugopodia and at this stage the expression of pluripotency markers is seen more broadly along the P/D axis than at stage 53 or later. The amputation plane through the stage-52 hindlimb indicated by Christen et al. (2010) is through the middle of the Sall4 expression domain, which would give the 0dPA control high levels of expression. Neff et al. (2005) used stage-53 hindlimbs, which are larger and can be more precisely amputated through the mid zeugopodia, which is proximal to the Sall4 expression domain (Neff et al.,2005). Surprisingly, Christen et al. (2010) found a significant increase in c-Myc, Sox2, and Sall4 expression in 1dPA and 3dPA pseudo-blastemas at stage 57. However, the overall level of expression was lower than that observed at stage 53. They concluded that some but not all iPSC reprogramming factors are expressed during epimorphic regeneration and that blastema cells are not like either ESC or reprogrammed pluripotent iPSCs. Christen et al. (2010) speculate that amphibian limb regeneration blastemas contain partially reprogrammed cells and that there may be a common mechanism between dedifferentiation and reprogramming.


Sall gene family members (Sall1, 2, 3, and 4) show similar and unique expression patterns in different systems such as organogenesis, limb development, and neurogenesis (reviewed de Celis and Barrio,2009). For example, both Sall1 and Sall4 are expressed in ESC (Sakaki-Yumoto et al.,2006) and in early developing limb bud mesenchyme (Hollemann et al.,1996; Neff et al.,2005). SALL4 can heterodimerize with other family members such as SALL1 (Sakaki-Yumoto et al.,2006) and Sall1, Sall2, and Sall3 have essential redundant functions during murine neurulation (Bohm et al.,2008). Sall1 and Sall3 are involved in Shh linked A/P patterning of developing mouse limbs (Kawakami et al.,2009) and human Sall1 mutations produce Townes-Brocks Syndrome, most likely caused by the dominant-negative effect of truncated SALL1, which includes among its defects thumb abnormalities (Kohlhase et al.,1998; de Celis and Barrio,2009). Functional gene knockout studies of mouse autopod development showed that Sall1 and Sall3 are partially redundant. While Sall1 can compensate for loss of Sall3 and vice versa, double knockouts have no digit 1, similar to Sall4 loss-of-function mutants (Kawakami et al.,2009). Both Sall1 and Sall3 are negatively regulated by Sall4 in ESCs (Yang et al.,2010). Cole et al. (2008) found that while SALL4 can bind the Tcf3, Oct4, and Nanog promoters, SALL1 only binds the Nanog promoter, SALL3 binds only the Oct4 promoter, and SALL2 does not bind to any of the mentioned t-factors (Cole et al.,2008). Sall1 is part of the core pluripotency network that suppresses ectodermal and mesodermal differentiation (Karantzali et al.,2010). These complex interactions between Sall gene family members prompted us to look at the expression profile of the Sall gene family during Xenopus hindlimb regeneration at the regeneration-complete and-incomplete stages utilizing RT-PCR and qPCR assays.

RT-PCR (Fig.1A) and qPCR (Fig. 1B) were used to determine the expression of Sall1, Sall3, and Sall4 during Xenopus hindlimb regeneration at stage 53 and stage 57. Sall2 was not assayed because previous experiments demonstrated that Sall2 is not expressed during Xenopus hindlimb regeneration (data not shown). Amputation was through the mid-zeugopodia and blastemas (from stage-53 limbs) or pseudoblastemas (from stage-57 limbs) were collected for analysis at 0dPA, 1dPA, 3dPA, and 5dPA. At 0dPA, about 1 to 2 mm of tissue from the distal end of the amputated limb was used as the internal tissue control. Ornithine decarboxylase (odc) expression was used as the normalizer for gene expression.

Expression of Sall1 and Sall3, like that of Sall4, was higher in stage-53 0dPA than stage-57 0dPA tissue just proximal to the amputation plane. In stage-53 1dPA blastemas, Sall4 showed a 5.2-fold increase while the levels of Sall1 and Sall3 were about the same as the 0dPA controls. Sall4 expression is up-regulated very early after amputation, being detected by WMISH in the dedifferentiation zone proximal to the amputation plane at the earliest time point assayed at 4 hr PA (Neff et al.,2005) and in a separate experiment at 6 hr PA by qPCR (unpublished data). The 3dPA blastema data was similar, with only Sall4 expressed at higher levels. At 5dPA, expression of all three Sall genes was higher than the 0dPA controls (Sall1, 9-fold; Sall3, 6.5-fold; Sall4, 5.4-fold). These data are consistent with the following interpretation: Sall4 alone is up-regulated at 1dPA because it, but not Sall1 nor Sall3, is involved in dedifferentiation/reprogramming. In addition, Sall4 is the only Sall gene up-regulated in blastemas before Shh is up-regulated (unpublished data), which suggests that Sall4 expression correlates with dedifferentiation while Sall1 and Sall3 expression correlates with patterning. The early Sall4 expression after amputation is probably not a wound response because wounding of a Xenopus limb without amputation does not result in local Sall4 expression (Neff et al.,2005, unpublished data). We assume that reprogramming, growth of the blastema, and patterning are probably not all-or-none phenomena, considering the complexity of the events occurring during limb regeneration, but rather overlapping events such as reprogramming at 1 to 3dPA, maintenance of the undifferentiated cells in the growing blastema at 5dPA, and patterning at 7dPA. We speculate that by 5dPA when a definitive morphological blastema is established, Sall4 along with Sall1 and possibly Sall3 are involved in maintaining the blastema stem cells in the undifferentiated state. Both Sall1 and Sall3 may also be involved in AP patterning at this stage in limb regeneration.

The expression of Sall1 and Sall3, like Sall4, is higher in stage-53 0dPA controls than stage-57 0dPA controls. In stage-57 1dPA pseudoblastemas, Sall4 was expressed 9.9-fold higher than the 0dPA controls while Sall1 and Sall3 were not up-regulated. This level of Sall4 expression is below the level detectable by WMISH (Neff et al.,2005). In stage-57 5dPA pseudoblastemas, Sall4 expression was 27-fold higher than the 0dPA control, while Sall1 and Sall3 expression remained unchanged. The increased expression of Sall4 in stage-57 pseudoblastemas after amputation is supported by data from Christen et al. (2010) who found significant but low levels of Sall4 expression 1dPA and 3dPA in stage-57 pseudoblastemas compared to 0dPA control levels. These data are consistent with the interpretation that in response to amputation at stage 57, dedifferentiation and reprogramming of local cells begin, but the response is diminished compared with that at stage 53 because a smaller population of cells is involved or a reduced response in cells capable of reprogramming. In addition, Sall1 and Sall3 are not expressed at a higher level than 0dPA after amputation of regeneration-incomplete limbs because there is no Shh expression and no AP patterning. We speculate that at 5dPA in stage-57 limbs, Sall4 expression in the pseudoblastemas is restricted to a population of undifferentiated mesenchymal cells that will eventually give rise to the cartilage cells of the hypomorphic spikes.

Dedifferentiation and Sall4 Regulation of Chromatin Structure

Embryonic and other stem cells maintain an epigenetic chromatin structure that determines their level of pluripotency, self-renewal, and differentiation. Genome-wide maps of global chromatin structure show very few differences between ESC and iPSC (Guenther et al.,2010). SALL4 protein is primarily localized to heterochromatin (Bohm et al.,2008) and has been linked to chromatin structure remodeling in ESCs (Yang et al.,2008; Bard et al.,2009; Lu et al.,2009; van den Berg et al.,2010). The mammalian ESC SWI/SNF ATP-dependent chromatin remodeling complex, esBAF, is essential for the core ESC transcriptional network that binds SALL4 (Ho et al.,2009a,b; Lessard and Crabtree,2010). Reprogramming to form blastema stem cells and maintenance of the blastema cells within the limb regeneration blastema also most likely involves chromatin remodeling. Reprogramming is facilitated by chromatin remodeling components of the BAF complex (Singhal et al.,2010). Nuclear proteins of stem cells that can reprogram mouse embryonic fibroblasts include SALL4 and pluripotent cell-specific components of the SWI/SNF complex (Singhal et al.,2010). Overexpression of Smarca4 and Smarcc1 was sufficient to enhance reprogramming. Our microarray analyses showed that Smarca4 RNA was expressed at higher levels in stage-53 5dPA limb blastemas than stage-57 5dPA limb pseudoblastemas, while Smarcc1 RNA was expressed at higher levels in stage-53 3dPA blastemas compared to stage-53 1dPA blastemas (Grow et al.,2006 and unpublished data). The SMARCA4 protein is also expressed at much higher levels in stage-53 3dPA limb blastemas than in stage-57 3dPA limb blastemas (King et al.,2009). Other components of the pluripotency-associated SWI/SNF complex (Singhal et al.,2010) were also differentially expressed in our limb regeneration microarray screens, including Smarce1, Aridia, Smarcc1, SmarchA5, SmarcA6, Smarcd1, and Actl6a (Grow et al,2006; unpublished data) and SMARCD1 protein (King et al.,2009). In contrast to SMARCA4, the SMARCC2 protein is a differentiation- (not pluripotency-) associated SWI/SNF complex component (Singhal et al.,2010) and was expressed abundantly in stage-57 3dPA limb pseudoblastemas. These data are consistent with the view that stage-53 3dPA blastemas possess a pluripotency-associated SWI/SNF complex that includes Sall4, while stage-57 limb pseudoblastemas contain a differentiated state SWI/SNF complex, which may or may not include Sall4.

Epigenetic covalent histone tail modifications that activate, repress, or pose repressed genes for activation (bivalent histone modification is a hallmark of pluripotent cells) are involved in regulating stemness, development, differentiation, and reprogramming (Meissner,2010). Dynamic changes in histone modifications occur during dedifferentiation of the dorsal and ventral irises during newt lens regeneration (Maki et al.,2009). A crucial early event in the initiation of zebrafish fin regeneration involves histone modifications (Stewart et al.,2009). SALL4 is one of the pluripotency factors that regulate epigenetic modifiers in human ESC (Yang et al.,2008; Hemberger et al.,2009). Lineage-specific SALL4 promoter occupancy correlates with specific histone modifications in two distinct stem cell lineages (Lim et al.,2008). SALL4 has been linked to the epigenetic histone modification mechanism for regulation of Bmi-1 expression in hematopoietic and leukemic stem cells (Yang et al.,2007). SALL4 in association with Whsc1 is associated with histone modification that regulates the expression of their target genes in ESC (Nimura et al.,2009).

Signaling Pathways Regulating Sall4 Expression During Limb Regeneration

Limb regeneration in Xenopus involves numerous signaling pathways (Beck et al.,2009), including BMP (Beck et al.,2006), Fgf (Yokoyama et al.,2000), and Wnt (Kawakami et al.,2006). Regulation of Sall4 expression during reprogramming and maintenance of pluripotency in ESC and somatic stem cells and probably during patterning also involves numerous context-dependent signaling pathways (Li et al.,2010). For example, in Drosophila melanogaster spalt/Sal expression is regulated during embryogenesis by Shh, dpp (BMP), and EGF signaling pathways, depending on the context (de Celis and Barrio,2009). Xenopus Sall4 was originally identified in a differential display screen looking at genes expressed in animal caps exposed to activin (a TGF-beta family member) and retinoic acid (Onuma et al.,1999). Yang et al. (2008) used ChIP-chip assays to determine that SALL4 binding genes (genes with SALL4 binding sites) encode components in over 30 different signaling pathways, the most prominent being the Wnt, PTEN, NF-κB, and p53 signaling pathways. Sall4 expression is directly activated by the canonical Wnt signaling pathway (Bohm et al.,2008; Al-Qattan,2010). Sall4 activates Fgf10 synergistically with Tbx5 and Tbx4 (Koshiba-Takeuchi et al.,2006). Unpublished data in our lab indicate that in cultured stage-55 Xenopus hindlimb autopods, Sall4 is up-regulated in a dose-dependent manner by Fgf10. Signal transducer and activator of transcription 3 (STAT3) regulates the expression of Sall4 (Bard et al.,2009). Among the growth factors that signal via STAT3 are LIF, which is necessary to maintain mouse ESC, and IL-6, which is required for compensatory liver regeneration (Tiberio et al.,2008). LIF/Stat3 signaling contributes to the maintenance of the undifferentiated state by preventing lineage-specific differentiation programs, including Sall4 expression (Bourillot et al.,2009). ESCs that are induced to differentiate by eliminating STAT3 signaling express Sall4 in the early phases of their differentiation (Bourillot et al.,2009). In Xenopus limb regeneration, Stat3 is up-regulated in both 3dPA stage-53 blastemas and 3dPA stage-57 pseudoblastemas relative to 1dPA limbs (Grow et al,2006 and unpublished data), observations consistent with the finding that Sall4 is expressed in dedifferentiating multipotent cells. Proteomic analysis during the early phase of Xenopus limb regeneration revealed that the overall protein expression profile of 3dPA blastemas at stage 53 more closely resembles that of multipotent adult stem cells than that of totipotent ESCs (King et al.,2009). In summary, Sall4 responds and regulates numerous signaling pathways that are involved in both stemness and limb patterning.


In 2005, we postulated that Sall4 may be playing a role in dedifferentiation (reprogramming) and in maintaining limb blastema cells in an undifferentiated state. This hypothesis was based on the observed differential pattern of SALL4 expression during Xenopus hindlimb regeneration at the regeneration-complete stage and -incomplete stage. Since then it has been shown that Sall4 is involved in maintenance of ESC pluripotency, is a major repressor of differentiation, and plays a major role in reprogramming differentiated cells into iPSC. Sall4 is a component of the stemness regulatory circuit involving Nanog, Oct4, Sox2 and others in pluripotent ESC and iPSC. However, unlike many such stemness factors, Sall4 is also expressed in multipotent embryonic, fetal, and adult somatic stem cells. In addition, Sall4 plays a role in epigenetic modulation of chromatin structure. These functional qualities make Sall4 a major player in regulating the dedifferentiation/reprogramming events that produce and maintain the kinds of multipotent stem cells that comprise epimorphic regeneration blastemas. Functional analysis of genes activated and repressed by Sall4, of the signaling pathways that regulate Sall4 expression, and of Sall4 function in modulating chromatin structure during epimorphic regeneration has the potential to yield large dividends in the field of regenerative biology and medicine.


Research by the authors on dedifferentiation and Sall4 was supported by the National Science Foundation (grant IOS-0814399) and benefitted from the expert technical assistance of Elizabeth Osborne.