Stem cells and pulmonary metamorphosis: New concepts in repair and regeneration



Adult stem cells are likely to have much more versatile differentiation capabilities than once believed. Numerous studies have appeared over the past decade demonstrating the ability of adult stem cells to differentiate into a variety of cells from non-hematopoietic organs, including the lung. The goal of this review is to provide an overview of the growth factors which are thought to be involved in lung development and disease, describe the cells within the lung that are believed to replace cells that have been injured, review the studies that have demonstrated the transformation of bone marrow-derived stem cells into lung cells, and describe potential clinical applications with respect to human pulmonary disease. © 2005 Wiley-Liss, Inc.

Traditional models of adult stem cell (SC) kinetics have recently been questioned. There is a growing body of evidence to suggest that these cells may not differentiate in a strict, hierarchical fashion, but that they are likely to have much more versatile differentiation capabilities. Numerous studies have appeared over the past decade demonstrating the ability of adult SCs to differentiate into a variety of cells from non-hematopoietic organs, including the lung. In murine models, structurally and functionally normal alveolar epithelial cells derived from donor SCs have been shown to replace injured cells. In some studies, these cells have demonstrated the ability to attenuate or reverse disease manifestations. An exciting field of research has emerged, with the ultimate goal of providing novel treatment options for pulmonary diseases. The goal of this study is to provide a review of the growth factors which are thought to be involved in both lung development and pulmonary disease, describe the cells within the lung that are believed to replace cells that have been injured, review the studies that have demonstrated transformation of exogenously administered SCs into cells of the lung in injured hosts, and describe potential clinical applications with respect to human pulmonary disease.


When the human embryo is approximately 4 weeks of age, the respiratorydiverticulum, or lungbud, appears as an outgrowth from the ventral wall of the foregut. As a result, the internal epithelial lining of the lung and associated airways are of endodermal origin. During its separation from the foregut, the lungbud forms the trachea and bronchialbuds which, by week 5, enlarge to form the right and left mainstem bronchi. As the lung buds continue to grow and penetrate into the coelomic cavity, they are covered by mesoderm, which gives rise to the visceral and parietal pleura (weeks 6–8). The bronchial tree continues to divide into smaller airways through week 26. At about this time, some of the cuboidal cells of the respiratory bronchioles become thin and flat, forming type 1 pneumocytes. They are intimately associated with a growing network of capillaries and are now referred to as terminalsacs or primitivealveoli. During the seventh month, sufficient vascularization is present to provide adequate gas exchange so that a premature infant could survive. Type 2 pneumocytes are developing as well and the amount of surfactant within the alveolar fluid steadily increases, particularly during the last 2 weeks before birth (Sadler, 1995).


Epithelial–mesenchymal interactions

A complex and well-integrated sequence of signaling and regulatory pathways are responsible for the development of the human lung, from lung bud to alveolus. Airway branching is directed by interactions between the epithelium and mesenchyme. Demayo et al. (2002) reported that when distal lung mesenchyme is grafted onto tracheal epithelium, airway branching occurs. Further, tracheal epithelium exhibited all of the ultrastructural characteristics of type 2 pneumocytes and expression of surfactant protein C (SP-C) was noted 24 h post-grafting. Fibroblast growth factors (FGF)-1, -2, and -7 appear to play an integral role in transformation of the tracheal epithelium (Cardoso and Williams, 2001). In contrast, when distal lung epithelium was engrafted with tracheal mesenchyme, the epithelial cells assumed a tracheal phenotype and all distal lung epithelial markers were suppressed. This phenomenon may be organ-specific as others have demonstrated that tracheal epithelium of the developing rodent cannot be induced to form lung buds if grafted with gut mesenchyme (Cardoso and Williams, 2001). Platelet-derived growth factor (PDGF)-A is expressed on the epithelial cells of embryonic lungs but its receptors are found within the mesenchyme surrounding the branching epithelium. Mice deficient in PDGF-A lack alveolar septation, resulting in early neonatal death (Bostrom et al., 1996). This observation provides further evidence that epithelial–mesenchymal communication is vital for proper lung development.

Smooth muscle differentiation in the lung appears to be induced by a change in shape of the peribronchial cell from round to elongated. Embryonic mesenchymal cells from any organ can remain in an undifferentiated state in culture if they are forced to assume a round shape, but can become smooth muscle cells (SMCs) if allowed to elongate. This occurs even if the cells are from an organ without significant smooth muscle, such as the kidney (Yang et al., 1999). Laminin-α1 and eventually Laminin-1, the main component of the basement membrane, forms at the epithelial–mesenchymal interface, inducing mesenchymal cells to elongate and differentiate into smooth muscle. Elongation induces laminin-2 synthesis, which further stimulates myogenesis (Relan et al., 1999).

Lung growth factors and knockout models

The known factors that drive lung development include various transcription factors, signaling molecules, extracellular matrix proteins and their receptors (Demayo et al., 2002). A summary of selected growth factors is found in Figure 1. Proper development cannot occur unless these interactions are timed precisely; disruption of this process can result in altered lung morphogenesis. For example, control of cell proliferation and branching of the airways is dependent on expression of bone morphogenic protein (BMP)-4, which in turn is controlled by FGF-10 (Weaver et al., 2000). Mice lacking FGF-10 do not form structures beyond the mainstem bronchi (Sekine et al., 1999).

Figure 1.

A summary of selected growth factors and their role during embryonic lung development.

In a murine knockout model of the transcription factor Nkx2.1, lung structures beyond the secondary or tertiary bronchi do not form (Minoo et al., 1999). The sonichedgehog (−/−) phenotype is similar to Nkx2.1 (−/−) as these mice also display limited distal cellular differentiation (Pepicelli et al., 1998).

While FGF plays a key role in early lung development, it also participates in alveogenesis; FGF receptor (FGFR)-3 and -4 knockout mice lack alveolar formation (Weinstein et al., 1998).

Retinoids in lung development

Retinoids play a fundamental role in lung development and, much like FGF, produce different outcomes at different stages of development. Retinoid signaling is controlled by a number of factors including retinaldehyde dehydrogenase-2 (Raldh2) which produces active retinoic acid (RA), retinoid receptors (RARs and RXRs), and enzymes that degrade RA. These factors are expressed at different times and locations in the developing lung and tightly control the influence of RA on gene expression and overall development (Cardoso and Williams, 2001). Disruption of RA signaling in mice lacking Raldh2 or in rats that are deficient in vitamin A, results in early embryonic lethality or in lung agenesis (Dickman et al., 1997; Niederreither et al., 2000). Although RA is thought to be critical in early lung development, RA signaling in the epithelium becomes markedly down-regulated when epithelial tubules are branching and differentiating. Wongtrakool et al. (2003) demonstrated that when RARα or β receptors were expressed constitutively in the distal lung, RA signaling persisted, resulting in lung immaturity.


Bronchopulmonary dysplasia

The appearance of certain growth factors in altered quantities or at inappropriate times or locations has been reported in some diseases of the lung, however it is not clear whether this is the cause or effect of the disease itself. It has been postulated that bronchopulmonarydysplasia (BPD) is the consequence of an inflammatory state kindled by various insults including antenatal infection, oxygen toxicity, or volutrauma (Demayo et al., 2002). TNF-α expression is increased in the lungs of preterm neonates at risk for BPD and has been shown to decrease Nkx2.1 gene expression (Wongtrakool et al., 2003). Nkx2.1, previously described as being essential for distal lung formation, is decreased in the lungs of neonates who died of BPD (Stahlman et al., 1996).

Many of the biological effects of inflammation are mediated through nuclear factor (NF)-κB. Muraoka et al. (2000) observed that NF-κB expression by the mesenchyme reduced FGF-10 production and resulted in diminished epithelial branching in the embryonic chick lung. Hence, one might speculate that NF-κB may mediate the abnormal lung development of BPD.

Demayo et al. (2002) propose that BPD may be the end-result of the production of the wrong type of fibroblast. Normally, mild fluid distension of the embryonic lung upregulates parathyroid hormone-related protein (PTHrP) and prostaglandin E2 (PGE2) promoting the production of lipofibroblasts. Lipofibroblasts produce leptin and play a role in type 2 pneumocyte growth and differentiation. Leptin acts to stimulate the production of surfactant. Overdistension, hyperoxia or inflammatory cytokines could cause a decrease in lipofibroblast production with a concomitant increase in myofibroblast production, potentially contributing to the pathogenesis of BPD.


While our knowledge of the immunologic and inflammatory components of asthma is increasing, relatively little is known about the cellular and molecular mechanisms of airways remodeling in the asthmatic lung. Bronchial smooth muscle hypertrophy, transformation of fibroblasts to myofibroblasts and subepithelial collagen deposition are all characteristic histologic changes seen in airways remodeling. Epithelial growth factor receptor (EGFR) expression is increased in the damaged airways of normal subjects and in both normal and damaged airways of asthmatics (Davies et al., 2003). However, EGF levels in the bronchial mucosa of asthmatics are normal and there is little evidence of epithelial hyperproliferation (Demoly et al., 1994). EGFR itself may contribute to an inflammatory state by eliciting IL-8 expression from bronchial epithelial cells (Hamilton et al., 2003). TGFβ-1 in BAL samples of asthmatics is increased. TGFβ-1 is known to promote differentiation of fibroblasts to myofibroblasts, which secrete interstitial collagens and growth factors including endothelin 1 and vascular endothelial growth factor (VEGF), mitogens for smooth muscle and endothelial cells respectively (Richter et al., 2001). TGFβ-1 may also indirectly inhibit epithelial repair (Davies et al., 2003). In the developing lung, EGF and TGFβ-1 act in an opposing fashion. When EGF dominates, airways elongate due to EGF's effect on growth and matrix deposition. When TGFβ-1 dominates, growth and matrix deposition are inhibited and airways begin to branch. In the clinical setting of asthma, where EGF levels are low and TGFβ-1 levels are high, this imbalance may contribute to defective wound repair, collagen deposition and airway remodeling.

The role of VEGF in emphysema and other pulmonary diseases

VEGF is an endothelial cell-specific mitogen which plays a critical role in neovascularization during normal development of the lung and in various pathologic states as well. Knockout of just one allele leads to death in utero in humans due to a lack of cardiovascular development. It is constitutively expressed in alveolar and bronchial epithelium, glandular cells of the bronchi and activated alveolar macrophages (Tsokos et al., 2003), and can be further induced by stimuli including hypoxia and TGF-β. However, it has been reported that BAL fluid VEGF concentrations are decreased in patients with ARDS compared with normal volunteers or patients with non-ARDS lung disease (Karner and Crystal, 2001; Maitre et al., 2001). Ohwada et al. (2003) cultured A549 alveolar epithelial-like cells in hydrochloric acid and observed that proliferation of these cells and VEGF expression were suppressed. Addition of exogenous VEGF restored cell proliferation. When anti-VEGF antibodies or neutralizing antibodies to VEGF receptor subtypes (VEGFR) 1 and 2 were added to acid-treated cells, the proliferative effect of exogenous VEGF was blocked. Whether a low VEGF level contributes to the pathogenesis of ARDS remains to be determined.

Low levels of VEGF may also result in inappropriate cell proliferation in response to chronic injury. For example, investigators have demonstrated that VEGF levels are decreased in emphysema. VEGF-depleted adult mice develop histologic evidence of emphysema (Tang et al., 2002) and when VEGFR-2 is blocked in rats, alveolar spaces become enlarged (Kasahara et al., 2000). The maintenance of pulmonary endothelial cells is also thought to depend on the local balance of VEGF and endostatin in the lung as endostatin directly antagonizes the effects of VEGF. Kanazawa et al. (2003) observed low VEGF levels in induced sputums of patients with emphysema compared with normal controls yet there was no difference with respect to endostatin levels. A lower VEGF to endostatin ratio correlated with a lower FEV1, FEV1/FVC ratio and carbon monoxide diffusing capacity in patients with emphysema. On the other hand, high VEGF levels appeared to correlate with a lower FEV1 in asthma, so the role of VEGF in obstructive lung disease remains unclear.

Pulmonary vascular remodeling is a characteristic feature of emphysema. Santos et al. (2003) measured VEGF levels in SMCs from pulmonary arteries of non-smokers, smokers with normal lung function, and subjects with moderate and severe emphysema. VEGF levels were increased in smokers and patients with moderate emphysema compared with non-smokers. Also, VEGF levels directly correlated with thickness of the vessel wall. In patients with severe emphysema, smooth muscle expression of VEGF was actually reduced compared with the other groups in spite of marked remodeling of the vessels. The authors speculate that a less differentiated SMC expressing higher amounts of VEGF in milder forms of emphysema compared with a well differentiated SMC that expresses lower amounts of VEGF might account for the difference that was noted. Nonetheless, it is not clear whether altered levels of VEGF are pathogenic in emphysema or merely a consequence of the disease process.


Several studies have suggested that subsets of cells located in specific areas of the mammalian airway behave in a manner consistent with tissue-specific SCs. These local or resident SCs are permanently established in a particular organ and are capable of differentiation in the setting of injury. SCs have unlimited differentiation potential and cycle slowly only in the setting of tissue turnover, possibly as a means of avoiding genetic mutation (Engelhardt, 2001). SC give rise to transient amplifying (TA) cells which retain the capacity to replicate yet have acquired differentiated functions. Eventually, TA cells become incapable of proliferation and become terminally differentiated (TD) cells. All of these cells can be labeled with [3H] thymidine or BrdU, but after many cell cycles, TA and TD cells lose the label. Only the cells with a slow turnover will retain label and hence SCs are label-retaining cells (LRC) (Borthwick et al., 2001).

Stem cells of the trachea

The distal and proximal portions of the trachea may contain local SCs in different niches. High cytokeratin expression is viewed as a biochemically primitive phenotype and suggests the presence of an entity with SC capability. Borthwick et al. (2001) demonstrated that the ducts of submucosal glands and gland myoepithelial cells of the proximal murine trachea may contain high cytokeratin-expressing cells. BrdU labeling of airway epithelial cells injured by intratracheal instillation polidolcanol or SO2 revealed labeled cells throughout the entire trachea after 3 days. By day 95, the label within most of these cells became quite faint, whereas cells in the submucosal glands of the proximal trachea remained bright, suggesting the presence of LRCs or local SCs. Investigators removed the surface epithelium from donor tracheas and transplanted the tracheas into recipient mice. Days later, the donor glands appeared larger, and cuboidal epithelial cells appeared within the graft suggesting that gland or gland duct cells could reconstitute the tracheal epithelium. It seems logical that cells responsible for repopulating the airway would exist in a protected niche such as a submucosal gland where they are provided some shielding from toxic airway insults.

It has been speculated that pulmonary neuroendocrine cells (PNEC) may be local SCs. LRCs are located in clusters within the cartilage–intercartilage junctions of the distal trachea. While PNECs are also found here, they are greatly outnumbered by the LRCs, leading investigators to believe that PNECs are associated with but not actually LRCs (Borthwick et al., 2001).

Others have suggested that tracheal lining epithelial cells rather than submucosal glands contain SCs for this compartment. Boers et al. (1998) studied normal human lungs with the proliferation marker MIB-1. Basal cells, located in the epithelium abutting the basement membrane and parabasal cells, just above the basal cells, both stained positively for cytokeratins 5 and 14. Within the larger airways (>0.5 mm), parabasal cells contributed more to the proliferative cell population than basal cells, suggesting that basal cells contain a population of slowly cycling SCs that give rise to parabasal cells, which in turn represent a more rapidly-cycling, transient amplifying (TA) cell population.

Stem cells of the distal airways

Nearly 70 years ago, Max Clara described a non-ciliated and non-mucous producing cell in the distal airway epithelium of humans. These clara cells have characteristic cytoplasmic granules and are able to synthesize and secrete proteins, including clara cell secretory protein (CCSP), via cytochrome P-450-mediated metabolism. Clara cell morphology and location are heterogeneous in studied species. They are found throughout the mouse respiratory tract but are isolated to the distal airways in humans. There is an abundance of data that suggests that clara cells are local SCs of the tracheobronchial epithelium. Evans and colleagues (1973, 1975) were among the first to demonstrate this by using NO2 and O3 to injure the airways of rats. Proliferating cells were then labeled with [3H] thymidine. One hour post labeling, all of the terminal bronchiolar epithelial cells had the morphologic features of clara cells by light and electron microscopy (EM). Three hours later, 68% of the airway cells were ciliated, TD epithelial cells. The investigators concluded that clara cells serve as progenitors for airway epithelial cells.

West et al. (2001) showed that inhalational and intraperitoneal naphthalene produced a dose-dependent cytotoxicity of clara cells. Naphthlene is metabolized by cytochrome P-450, in particular, the 2F2 subtype, resulting in the production of a cytotoxic epoxide. Some clara cells are associated with neuroepithelial bodies (NEB) which also contain PNECs. NEBs are thought to play an important role in regulating embryonic lung growth and maturation through the action of various neuropeptides. NEBs respond to chronic airway injury via hyperplasia of PNECs and by an increase in the number of NEBs. Clara cells associated with NEBs are variant clara cells as they have distinct ultrastructural characteristics, express lower levels of CCSP and are resistant to naphthalene-induced cytotoxicity, likely due to a deficiency in cytochrome P450-2F2.

Giangreco et al. (2004) demonstrated that these NEB-associated variant clara cells have characteristics similar to those of SCs, including the expression of the stem cell antigen (Sca)-1 and the ability to efflux Hoechst dye. In this study, among the murine airway epithelial cells that expressed Sca-1 were cells that also expressed the clara cell marker CCSP. Cytochrome P450-2F2 mRNA was undetectable in these cells suggesting that they are not typical clara cells but rather variant clara cells.

Hong et al. (2001) used inhalational naphthalene to induce airway injury in mice and destroy clara cells. By labeling proliferating cells with [H3] thymidine, they demonstrated that variant clara cells and PNECs were among the LRCs. Variant clara cells were then selectively ablated. The authors found an increase in the number of PNECs and NEBs, but no repopulation of the airway epithelium. Hong suggests that airway epithelium is maintained through the proliferation of clara cells which are capable of self-renewal and differentiation into TD ciliated epithelial cells of the airway. NEB-associated PNECs alone are not capable of renewing clara cell depleted airway epithelium but appear to have important interactions with other cells of the NEB. Variant clara cells are likely required for this task.

Stem cells of the alveoli

Type II pneumocytes (AE II) produce and secrete surfactant. Ultrastructural criteria that help to distinguish these cells from type 1 pneumocytes (AE I) are still considered to be the gold standard and include the presence of lamellar bodies, apical microvilli, cell–cell junctions, and a cuboidal shape. A number of alternative methods have been validated, including cell-type-specific lectins and immunohistochemical markers, however the expression of such markers can be effected by developmental stage and certain pathologic conditions (Fehrenbach, 2001). These versatile cells have long been considered to be the local SCs for the epithelial surfaces within the alveolus, serving as the progenitor for AE I and AE II.

During early embryonic development of the mouse lung, calcitonin gene-related peptide (CGRP), a neuroendocrine cell product, clara cell 10-kD protein (CC10) and SP-A are expressed in a nearly identical pattern in all epithelial cells of the distal airways (Wuenschell et al., 1996). Later on in development, cells assume expression profiles that are consistent with those of differentiated cell types suggesting that AE II, clara cells and neuroendocrine cells are all derived from a common progenitor cell.

In 1954, Macklin (1954) reported that AE II cells proliferate in the setting of acute lung injury. By 1970, there were many reports of increased numbers of AE II in the alveolus after lung injury, suggesting that they may be involved in the reparative process (Uhal, 1997). Kapanci et al. (1969) was the first to describe the AE II as a local SC. Using the hyperoxic monkey lung as an investigational model, he showed that AE I were destroyed and replaced by AE II after 4 days of exposure. Restoration of normal lung architecture occurred upon reexposure to room air and it was noted that some of the AE I had characteristics of AE II, suggesting a transformation from AE II to AE I.

Evans and colleagues (1973, 1975) also demonstrated a transformation of AE II to AE I in the injured lung. Adult rat airways were exposed to NO2 and proliferating cells labeled with [H3] thymidine. After 1 h, nearly all of the labeled cells in the alveoli were AE II by EM. By 24 h, AE I began to appear in increasing numbers and by 48 h, this population increased by nearly 10-fold compared with the 1-h value, accompanied by a sharp decline in the number of AE II. In addition, there was a population of cells that had morphologic characteristics of both cell types. The majority of these transitional cells were labeled with thymidine, suggesting that AE II were responsible for the production of AE I. By 3 days, there was a decline in the number of cells with an intermediate morphology accompanied by a reciprocal increase in AE I. This data and similar work by Adamson and Bowden (1974) suggest that there is a direct transition of AE II to AE I via a transitional cell.

The notion that AE II differentiate to form AE I both in vivo and in vitro is widely accepted. However, it is not clear if cell division mediates this process. In previous experiments, thymidine labeling in AE I was consistently 50% of that in AE II, suggesting AE I were the product of dividing AE II. In these experiments, only cells labeling with thymidine (mitotic cells) were identified; thus, cell differentiation independent of cell division was never excluded as a potential mechanism. Shannon et al. (1992) produced more recent evidence to suggest that alveolar epithelial cells are more plastic than once believed, capable of reverting back and forth independent of cell division. AE II were cultured with fetal rat lung fibroblasts and under certain conditions, these AE II lost their lamellar inclusion bodies, produced less mRNA for surfactant and exhibited a flat AE I-like morphology. These cells, under other culture conditions, could regain lamellar inclusion bodies, produce more mRNA for surfactant and again assume an AE II-like morphology suggesting that pneumocytes may alter their phenotype though a process independent of cell division. This interpretation relies on the assumption that cell division did not occur and this was never specifically addressed in this study.

One model of alveolar epithelial kinetics that incorporates most of the available data was proposed by Uhal (1997). AE II undergo cell division yielding two relatively undifferentiated daughter cells, both capable of reverting to the AE II phenotype or differentiating into AE I. Either phenotype may transform to the other without cell division but reentry into cell cycle requires reversion to the AE II phenotype.

A summary of the proposed local SCs for the different anatomical compartments of the lung is found in Figure 2.

Figure 2.

Proposed local stem cells of different anatomic compartments of the lung and the terminally differentiated (TD) cells that they produce.

Side population cells

Side population (SP) cells are a rare subset of cells that have been shown to reconstitute locally damaged tissue. They are local SCs but are defined as being SP cells based on certain phenotypic and physical characteristics. Irrespective of their tissue of origin, all SP cells carry stem cell antigen (Sca)-1 but lack lineage markers (lin−), which would otherwise define them as differentiated cells. SP cells can be isolated by flow cytometry based on their distinctive ability to efflux Hoechst dye, an ATP-requiring process performed by the cellular membrane transporter breast cancer resistance protein (Bcrp) 1. They have been isolated from various tissues including bone marrow, liver, muscle and lung (Summer et al., 2003). Asakura et al. (2002) replaced skeletal muscle injured by cardiotoxin with locally injected muscle-derived SP cells. Bone marrow SP cells have replaced ischemic cardiac myocytes demonstrating that SP cells can also assume a phenotype different from their site of origin (Jackson et al., 2001).

Summer et al. (2003) reported that SP cells in adult mice comprised roughly 0.03%–0.07% of all lung cells. These cells expressed variable positivity for CD45, a hematopoietic marker. In skeletal muscle, Sca-1+/CD45+ SP cells were shown to have hematopoietic potential whereas Sca-1+/CD45− SP cells had myogenic potential (Kawada and Ogawa, 2001).

CD45+ lung SP cells also expressed CD31 (PECAM-platelet/endothelial cell adhesion molecule), an endothelial marker, and it is speculated that lung SP cell expression of CD31 may indicate that these cells play a role in the production and repair of the pulmonary vasculature (Summer et al., 2003). Others have suggested that endothelial cells and hematopoietic SCs are derived from a common origin and co-expression of both markers on lung SP cells lends weight to this argument. Marrow-derived SP cells that express PECAM-1 were able to reconstitute endothelium in the ischemic heart (Jackson et al., 2001). Kotton et al. (2003) found co-expression of Sca-1 and PECAM-1 in pulmonary arteries, veins, and alveolar capillaries. Whether these cells, or SP cells of the lung for that matter, can behave as progenitors for other cell types remains unanswered by current studies.

A word of caution on interpretation of SP cell studies is indicated. Bone marrow from Akt-1 knockout mice have a reduced number of bone marrow SP cells but fully engraft recipient bone marrow (Mogi et al., 2003). Thus, the SP phenotype may not always correlate with the SC phenotype.


Many studies have shown that bone marrow-derived cells have the capacity to produce a variety of non-hematopoietic cells both in vitro and in vivo. Systemically injected murine bone marrow cells can differentiate into skin (Abedi et al., 2002), muscle (Ferrari et al., 1998; Gussoni et al., 1999), bone (Hou et al., 1999), heart (Orlic et al., 2001a), brain (Eglitis and Mezey, 1997), liver (Petersen et al., 1999; Alison et al., 2000), and lung (Kotton et al., 2001; Krause et al., 2001; Grove et al., 2002; Theise et al., 2002; Ortiz et al., 2003). Appropriate concern has been raised about the meaning of these observations (Lemishka, 2002). Do converted cells function? Is the phenomenon quantitatively significant? Is the donor cell identification and lineage determination correct? These are all critical questions and have been evoked by deficiencies in some of the reports. However, some studies have been able to establish that in defined models, marrow conversion to non-marrow tissue can produce quantitatively significant, functioning cells that can actually reverse disease manifestations. Our group has shown that an infusion of a large number of marrow cells into non-ablated host mice could produce bone cells that persisted and presumably functioned out to 6 months (Nilsson et al., 1999). Perhaps more definitively persuasive are the studies by Lagasse et al. (2000), who used a genetically deficient mouse with tyrosinemia type 1 and liver failure, in which they demonstrated that purified SCs could produce a large number of hepatocytes and cure some mice of their liver disease. Orlic and colleagues (2001a,b,c) have shown that either endogenously mobilized or directly injected purified marrow SCs could restore cardiac function to mice subjected to cardiac ischemia. Donor-derived marrow cells have also been shown to affect stroke manifestations in rats (Zhao et al., 2002).

Reports have raised the possibility that a fusion event between the donor and host cell, rather than cell replacement, could explain some of these results (Ying et al., 2002; Terada et al., 2002). This potential mechanism, however, does not negate the significance of the studies previously mentioned. The issue of which cell among the heterogeneous population of donor cells is mediating conversion is of intense interest. Overall, the phenomenon of marrow conversion into other tissue cells appears well established and the quantitative and functional significance of this is now the subject of a large number of studies.


A number of recent publications have indicated that bone marrow cells have the capacity to convert to a variety of different lung cells (Table 1). The mechanism of this phenomenon remains unknown, but is like the result of many factors working together (Table 2). Hematopoietic SCs may convert into local tissue SCs which in turn convert into TD lung cells as needed. Alternatively, hematopoietic SCs may avoid an intermediate step and convert directly into TD lung cells (Fig. 3). It is also not known if within a heterogeneous population of SCs, each cell has the potential to convert into any lung cell or if only certain SC types convert into specific lung cells (Fig. 4).

Table 1. Conversion of bone marrow-derived cells to lung cells in mice
AuthorInjuryDonor cellsTrackingAnalysis after BMTAnalysisConversion result
  1. TBI, total body irradiation; cGy, centigrey; NOD/SCID, non-obese diabetic/severe combined immunodeficient; Lin−, lineage depleted; MSC, mesenchymal stem cells; WBM, whole bone marrow; β-Gal, β-galactosidase; GFP, green fluorescent protein; FISH, fluorescent in situ hybridization; FACS, fluorescent activated cell sorting; IHC, immunohistochemistry; IF, immunofluorescence; RT-PCR, real time polymerase chain reaction; FC, flow cytometry; AE I, type I pneumocytes; AE II, type II pneumocytes; ATRA, all-trans retinoic acid; GCSF, granulocyte colony stimulating factor.

Krause et al. (2001)TBI: 1050–1100 cGy1 Fr25lin-male, +2 × 104 female WBM cellsY chromosome, chromosome11 mFISH, IHCAlveolar epithelium, 20.3%, AE II (SP-B mRNA+), bronchial epithelium, 3.74%
Kotton et al. (2001)Intratracheal bleomycin1–2 × 106 MSCβ-Gal1–30 dIHCAE I (not quantified)
Theise et al. (2002)TBI: 1200 cGy2 × 105 male WBM or 200 male, CD34+lin− cellsY chromosome1 d–6 m, 8 m for CD34+lin−FISH, IHCAE II, up to 14%, cytokeratin+/Y+ cells lining entire alveoli
Jiang et al. (2002)NOD/SCID, +/− 250 cGy TBIMSC, ? amountβ-Gal4–24 wIFAlveolar epithelium, ∼4% (+/− TBI)
Grove et al. (2002)TBI: 1100–1200 cGy3.5–5 × 106 male, WBM or 2 × 105male GFP WBMGFP, Y chromosome2–11 mFISH, IFAlveolar epithelium, 1%–7%, AE II (SP-B mRNA+), AE I? (cell specific marker not used)
Abe et al. (2003)TBI: 700, 950 cGy1–3 × 106 WBM, 2000 BM SPGFP30–140 dFACS, IHCAE I (cell-specific marker not used); bronchial epithelium (Not quantified)
Voswinckel et al. (2003)TBI: 1100 cGy, +pneumonectomy2–5 × 106 WBMGFP, β-GalTBI then pneumo at 6-8 w, sacrificed 3 w laterIHC, IFNo conversion in the lung
Ortiz et al. (2003)Intratracheal bleomycin from bleo resistant mice5 × 105 male MSCY chromosome7–14 dFISH, RT PCR, IHCAE II (not quantified) decreased inflammation, collagen deposition
Ishizawa et al. (2004)TBI: 1200 cGy, +intratracheal elastase2 × 106 fetal liver cellsGFPTBI then elastase at 3 w, sacrificed 12 d laterIHC, IF% GFP+/CD45− cells/alveoli: ∼5% elastase only, ∼17% elastase/ATRA, ∼19% elastase/GCSF, ∼26% elastase/ATRA/GCSF
Hashimoto et al. (2004)TBI: 1000 cGy, +intratracheal bleomycin4 × 106 WBMGFP28 dIF, RT-PCR, FCIncrease in fibroblasts TBI/bleo+ vs. TBI/bleo−; 27.5% lung cells GFP+/Col 1+
Aliotta et al. (2004)TBI: 0, 500, 900 cGy or 900 + cardiotoxin5(× 106 WBMGFP1,3 mIF, IHCAE I, AE II (Not quantified), Small blood vessels partially replaced?
Dooner et al. (2004)TBI: 900 cGy, +cardiotoxin5 × 106 WBM cultured in cytokines mobilized with G-CSFGFP2 mIFGFP+/CD45− cells up to 35% of all lung cells
Table 2. Potential factors contributing to lung conversion
 Cardiotoxin (cobra venom)
Recipient characteristics
 Immune status
  NOD/SCID phenotype
 Sensitivity to injury
Donor cell characteristics
 Cell type
  Whole bone marrow
  Mesenchymal cells
  Fetal cells
  Various purified stem cell populations
 Number of donor cells infused
 Cell cycle
  Use of cytokines to induce donor cells to enter cell cycle
  Cell cycle related to circadian rhythm
 Timing of infusion
  Before or after induction of chimerism
Time to tissue analysis
Exogenous factors
 Mobilization of bone marrow with G-CSF
 Use of growth factors (all-trans retinoic acid)
Figure 3.

It is not known whether hematopoietic stem cells first convert into pulmonary stem cells which then convert into TD cells of the lung (A) or if they convert directly into TD cells (B).

Figure 4.

It is not known if within a heterogeneous population of stem cells (X), each cell has the potential to convert into any lung cell (A) or if only certain stem cell types (X1, X2, X3) convert into specific lung cells (B).

Injury to the lung appears to be essential for conversion to occur. Davies et al. (2002) took brushings of uninjured nasal passages from female human subjects that had a previous gender-mismatched bone marrow transplant (BMT). Fluorescence in situ hybridization (FISH) for the Y-chromosome in combination with anti-cytokeratin staining revealed that although a median of 2.5% of the nuclei visualized were of donor origin, none were epithelial cells (cytokeratin+). Contrary to these findings, Suratt et al. (2003) looked at lung samples obtained by diagnostic lung biopsy or at autopsy from two female human subjects that had a previous gender-mismatched BMT. They found significant amounts of donor-derived epithelium (2.5%–8.9%) and endothelium (37.5%–42.3%) located predominantly within alveoli.

Of particular interest, donor-derived epithelial and endothelial cells were essentially absent from areas of graft versus host related obliterative bronchiolitis and none of the identified fibroblastic tissue present in the samples appeared to be donor-derived. A third subject had no donor-derived endothelial or epithelial cells that could be found. It is possible that differences in conditioning regimens (which lacked total body irradiation in the third subject) or the brief engraftment period prior to the biopsy in the third subject (50 days vs. 200 and 462 days) could have accounted for this discrepancy.

The findings of Kleeberger et al. (2003) echo these results. Seven human lung allograft recipients were examined for the presence of recipient-derived cells within the graft, all of which were removed for various reasons. Investigators noted recipient-derived AE II (9.1%–20%), bronchial epithelial (5.7%–25.5%), and seromucosal glandular (9.1%–24.2%) cells. Epithelial structures displaying signs of chronic injury, such as squamous metaplasia, showed a markedly higher number of recipient-derived cells, up to 24% compared with only 9.5% in more histologically normal regions.

Radiation injury models

The majority of the investigators who have reported significant conversion of bone marrow-derived cells to lung cells have used total body irradiation (TBI) as a mode of injury to the recipient (Fig. 5). TBI facilitates marrow chimerism, the replacement of recipient bone marrow cells by donor derived bone marrow cells. Our preliminary data suggest a dose-response relationship between chimerism and TBI. There is more histologic evidence of radiation injury to the lung and a higher conversion rate of donor bone marrow cells to lung cells. It is not known whether donor cells engraft in the marrow before homing to injured organs or whether injected donor cells go directly to injured organs after infusion. Jiang et al. (2002), however, demonstrated that bone marrow mesenchymal cells infused into immunodeficient mice were able to differentiate into liver, gut, and lung cells, with roughly 4% of the alveolar epithelium of donor origin, despite very low levels of bone marrow chimerism.

Figure 5.

Injury and conversion. Whole bone marrow carrying the green fluorescent protein (GFP) gene was transplanted into mice that received no irradiation (A) light, H&E, 20×; (B) same sample, fluorescence microscopy, FITC/Rhodamine filters or 1200 cGy of total body irradiation; (C) light, H&E, 20×; (D) same sample, fluorescence microscopy, FITC/Rhodamine filters. There are no donor-derived, GFP+ cells in the uninjured lung (B). Donor-derived, GFP+ cells are present in the injured lung (D, green dots). After labeling with cell-specific antibodies, many of these donor-derived cells were found to be lung cells.

As with other models of acute lung injury, alveolar epithelium damaged by radiation is repopulated as a result of AE II proliferation and conversion into AE I. Theise et al. (2002) transplanted donor bone marrow cells from male mice into female mice exposed to 1200 cGy of TBI. Within 3 days of exposure to radiation, lungs of recipient mice appeared hypocellular, with breakdown of capillaries within alveolar septa and extravasation of erythrocytes into alveolar spaces. Capillary breakdown and extravasation appeared to peak 5 days after the injury. Functioning donor-derived AE II, determined by co-localization of surfactant B mRNA and the Y chromosome were seen as early as 5 days after the injury and BMT, increasing to as high as 14% at 6 months. At later time points, cytokeratin+ donor-derived cells were seen lining entire alveoli and had the morphologic appearance of AE I, suggesting that they were the product of donor-derived AE II.

Krause et al. (2001) transplanted irradiated female mice with a single highly purified SC (Fr25lin-) from a male donor, along with whole bone marrow from female donors. At 11 months, roughly 20% of all of the alveolar epithelium and nearly 4% of the bronchial epithelial cells were derived from the one purified marrow cell. Some alveoli were lined entirely by donor-derived epithelial cells. A portion of the donor-derived alveolar epithelial cells were AE II. Similarly, Grove et al. (2002) transduced whole bone marrow cells with a retrovirus encoding the green fluorescent protein (GFP) and infused them into irradiated hosts, yielding 1%–7% donor-derived, alveolar cells. The viral constructs appeared to be stable as GFP expression was detected in mice as far as 11 months from the time of infusion. Figure 6 demonstrates a lung sample from a GFP+ mouse.

Figure 6.

Lung from a green fluorescent protein+ (GFP) mouse. (100×, FITC filter).

Bleomycin injury models

The exact mechanism of bleomycin toxicity to the lung is not known. AE I are particularly sensitive to bleomycin and are the first alveolar cell to be injured, while AE II have a more variable sensitivity, in part related to cell cycle as those in G0 seem to be resistant to bleomycin, while those that are proliferating and differentiating undergo atypical metaplasia in the presence of the drug. When a net loss of AE I occurs, a fibrotic reparative process is initiated. Bleomycin is also believed to stimulate fibroblasts directly, increasing this fibrotic response (Murray and Nadel, 2000). Intratracheal administration of bleomycin to the lung induces pulmonary fibrosis in mice and has served as a valuable model for research. Ortiz et al. (2003) transplanted bone marrow mesenchymal cells from male bleomycin-resistant mice into female bleomycin sensitive mice after the administration of intratracheal bleomycin. Lungs were then digested and AE II isolated to measure the amount of male DNA by real-time PCR. Male DNA accounted for 1.37 × 10−3% of the total DNA content of AE II in bleomycin-treated, transplanted females. Although small, these findings do represent lung conversion in the bleomycin injury model. A more striking finding was that mice transplanted with bleomycin resistant mesenchymal cells immediately after the bleomycin injury demonstrated a marked decrease in lung inflammation and collagen content compared with mice transplanted 7 days after the injury or mice injured but not transplanted. The mechanism of this phenomenon, whether it is due to replacement of a small number of AE II with the bleomycin resistant phenotype or mesenchymal cell alteration of the microenvironment at the site of injury, is unknown. Kotton et al. (2001) reported a different conversion phenomenon. Bleomycin-exposed mice did not produce donor-derived AE II, but did demonstrate peripheral clusters containing numerous donor-derived AE I. Different results may in part be related to a difference in the number of mesechymal cells infused, mesenchymal cell culture conditions or time to cell infusion after bleomycin exposure.

Combined modes of injury and lung conversion

There is some evidence to suggest that different modes of injury introduced in series may provide an additive effect on conversion. Our group has found that TBI at 500 cGy vs. 900 cGy in mice transplanted with GFP+ whole bone marrow yielded a clear dose response; a higher percentage of GFP+/CD45− cells (presumed to be donor-derived non-hematopoietic) were present with higher doses or irradiation (Aliotta et al., 2004a). A certain percentage of these cells appear to be AE I (Fig. 7), AE II (Fig. 8), vascular endothelial cells, and SMCs by both standard morphology and cell-specific antibody labeling techniques. There is nearly a doubling of the percentage of GFP+/CD45− cells in the lung when mice are subjected to intramuscular cobra venom (cardiotoxin) 7 days after TBI (Aliotta et al., 2004b). Cardiotoxin causes local tissue necrosis and is believed to induce widespread microvascular leakage and pulmonary edema, which may somehow enhance the conversion phenomenon.

Figure 7.

Donor-derived type I pneumocyte. Whole bone marrow carrying the green fluorescent protein (GFP) gene was transplanted into lethally irradiated mice. The featured cell (arrow) is GFP+ (green) and labels with Lycopersiconesculentum antibody (red), indicting that it is a donor-derived type I pneumocyte. This cell is non-hematopoietic as it is CD45− (CD45+ cells are blue). A: GFP and Lycopersiconesculentum are present in the same cell (FITC/Rhodamine/DAPI filters). B: GFP appears green (FITC filter). C: Lycopersiconesculentum appears red (Rhodamine filter). D: The cell does not stain with CD45 antibody (DAPI filter).

Figure 8.

Donor-derived type II pneumocyte. Whole bone marrow carrying the green fluorescent protein (GFP) gene was transplanted into lethally irradiated mice. The featured cell (arrow) is GFP+ (green) and labels with surfactant protein B (SP-B) antibody (red), indicting that it is a donor-derived type II pneumocyte. The cell is also non-hematopoietic, as it is CD45− (CD45+ cells are blue). A: GFP and SP-B are present in the same cell (FITC/Rhodamine/DAPI filters). B: GFP appears green (FITC filter). C: SP-B appears red (Rhodamine filter). D: The cell does not stain with CD45 antibody (DAPI filter).

Elastase lung injury is another useful research model in rodents as it is a potent inducer of emphesematous changes when administered intratracheally. Changes include increased lung volume (reflecting a loss of elastic recoil), larger but fewer alveoli and diminished volume-corrected alveolar surface area due to destruction of alveolar walls. Massero and Massero (Massaro and Massaro, 1997) showed that treatment with all-trans retinoic acid (ATRA) in rats with elastase-induced lung injury actually reversed many of these changes. Ishizawa et al. (2004) used this model in mice whose bone marrow was made almost completely chimeric with TBI and GFP+ fetal liver cells. Mice were then given ATRA and/or mobilized with G-CSF. With elastase and radiation injury alone, 5% of the cells in sampled alveoli were GFP+/CD45−. This number increased to nearly 20% in mice treated with either ATRA or G-CSF, and over 25% in mice treated with both. Investigators provided histologic evidence to suggest reversal of the emphesematous changes in mice treated with ATRA, G-CSF, or both.

One goal of SC transplantation is to produce functioning target organ cells in quantitatively significant numbers that may prevent, halt, or even reverse disease manifestations, as was the case in the work by Ortiz. What is not known is whether donor-derived cells contribute to or possibly worsen disease. Hashimoto et al. (2004) induced chimerism in mice after TBI and infusion of whole bone marrow from GFP+ donors. There was a significant increase in the number of fibroblasts in those treated with bleomycin and TBI compared to those that were treated with TBI alone. Of all lung cells, 27.5% were GFP+ and capable of producing type I collagen, suggesting that donor-derived fibroblasts may have had a role in the fibrosis.

Voswinckel et al. (2003) demonstrated very rare engraftment events in irradiated mice that then had a left-sided pneumonectomy. They used an equivalent number of whole bone marrow cells, a slightly higher amount of TBI (1100 cGy) and their bone marrow was fully chimeric at the time of pneumonectomy. The pneumonectomy lead to compensatory lung growth with restoration of lung volume, alveolar number, alveolar surface area but almost none of the structural components of the lung were donor-derived.

Donor cell characteristics

Cells with SC activity comprise only a very small percent of the total number of cells present in unpurified whole bone marrow. SCs can be separated from this population based on certain phenotypic characteristics, including the presence of stem cell antigen (Sca)-1, C-kit (another stem cell antigen), or the absence of any cell markers (Lin−) that would define the cell as being differentiated. SCs can also be separated based on certain physical characteristics, including their ability to efflux Hoechst or Rhodamine dyes. Based on current available evidence, it is not clear if purified SCs offer an engraftment advantage in the lung compared to whole bone marrow. In unpublished data from our lab, we compared the use of whole bone marrow to various purified bone marrow cell populations, including cells that are Sca-1−, Sca-1+, C-kit−, C-kit+, Lin−, and Lin+, to determine if any one population offers a higher conversion rate. Mice were exposed to 500 cGy of TBI, infused with different populations of GFP+ marrow cells, and later given IM cardiotoxin as an additional injury. Animals were transplanted either with whole bone marrow or marrow that had been run through a flow cytometer to select out cells that were positive for 1, 2, or all 3 SC markers (Lin−, C-kit+, Sca+). Both whole bone marrow and these subsets engrafted recipient marrow at levels of 55%–65% and up to 2% of the lung cells were of donor origin. The degree of lung chimerism was directly proportional to bone marrow chimerism. Lin+, C-kit− and Sca− cells, engrafted recipient marrow at levels of 10%, 1.9%, and 0.62%, respectively, with no GFP+/CD45− cells in the lung. Based on these data, certain populations of whole bone marrow would appear to offer an engraftment advantage over other populations.

Donor cells forced into cell cycle in vitro by a variety of cytokines appear to have an engraftment advantage depending on their point of cell cycle at the time of transplantation. In unpublished data from our lab, Lin− Sca+ cells from a GFP+ mouse were cultured in vitro with IL-3, -6, and -11, and steel factor (a factor which promotes SC proliferation) for 24 or 48 h. They were then transplanted into irradiated recipients that were later subjected to intramuscular cardiotoxin injury and two rounds of bone marrow mobilization with G-CSF. The number of GFP+/CD45− cells within the lung was nearly double (up to 35%) when donor cells were cultured for 24 h compared with uncultured donor cells or those cultured for 48 h. SCs undergo phenotypic changes when they are forced into cell cycle. While cycling, SCs express a group of surface antigens and receptors that are constantly changing. There may be a point in the cell cycle where the SCs surface antigen/receptor profile results in a greater tendency for that cell to home to the injured lung, where a complimentary antigen/receptor profile may exist.


Recent studies have demonstrated that adult marrow SCs have the capacity to cross lineage boundaries in vivo and produce non-hematopoieitc cells in a variety of tissues (Pereira et al., 1995; Bittner et al., 1999; Gussoni et al., 1999; Nilsson et al., 1999; Lagasse et al., 2000; Jackson et al., 2001; Kocher et al., 2001; Orlic et al., 2001a,b,c; Grove et al., 2002; LaBarge and Blau, 2002; Sapienza, 2002; Abe et al., 2003; Badiavas et al., 2003; Direkze et al., 2003; Erdo et al., 2003; Hess et al., 2003; Hoi et al., 2003; Ianus et al., 2003; Kataoka et al., 2003; McBride et al., 2003; Ortiz et al., 2003; Wang et al., 2003; Abedi, 2004; Abedi et al., 2004; Airey et al., 2004; Grinnemo et al., 2004; Koshizuka et al., 2004; Mathews et al., 2004; Murry et al., 2004; Nygren et al., 2004; Wang et al., 2004). These studies have dramatically altered standard models of marrow cell–cell regulation and suggested that adult marrow SCs might have exciting potential for restoration of damaged tissue. Pereira et al. (1995) initially demonstrated that adherent marrow mesenchymal cells have the capacity to produce bone, cartilage, and lung and Nilsson et al. (1999) showed in vivo bone cell production by infused marrow cells. Perhaps most striking were studies by Lagasse et al. (2000), who demonstrated that highly purified bone marrow SCs could produce hepatocytes and restore liver function in the fumaryl acetoacetate deficient mouse (FAH−), a rodent with a fatal genetic tyrosinemia. Orlic and colleagues (2001a,b,c) published studies indicating that purified hematopoietic SCs directly injected into ischemic hearts could result in significant cardiac replacement with salutary functional effects. Marrow conversion to skeletal muscle, pulmonary cells, renal mesanagium, endothelial cells, hepatocytes, and epithelial cells in the skin and gastrointestinal tract have also been described with varying levels of donor cell representation.

Systems to track donor cells have included transplants with GFP positive or β-galactosidase positive transgenic cells. Alternatively, male marrow cells have been tracked in female hosts. There have been impressive results with these techniques showing colocalization of donor markers with lineage specific markers such as desmin (skeletal muscle), tomato lectin (type 1 pneumocytes), albumin (hepatocytes), or cytokeratins (epithelial cells). In this work whole marrow cells, cultured bone marrow mesenchymal SCs, or purified hematopoietic SCs have been studied.

Several rather divisive controversies have enveloped this field, fueled by a series of irrelevant conclusions or red herrings (Anderson et al., 2001; Lemischka, 2002; Wagers and Weissman, 2004). A major problem has been the demand that transdifferentiation be established. While this term has been used by many, at the present time virtually no one believes that transdifferentiation is the operative mechanism for marrow to non-hematopoietic cell conversions and, in fact, it is not clear that transdifferentiation has ever been demonstrated. Unfortunately, there continues a persistent demand that this be demonstrated. This in turn has led to a demand that clonal studies be carried out. This, of course, makes no sense in an intrinsically heterogeneous cell population and the most purified hematopoietic SCs are highly heterogeneous. In fact this is a basic characteristic of the marrow hematopoietic SC.

There have also been criticisms that the number of transformations or conversions is too low to be important, but this is simply not the case, given the above cited results with the FAH-tyrosinemic mouse (Lagasse et al., 2000) and reported conversion rates of 12.5% in cardiotoxin injured skeletal muscle (Abedi, 2004) and up to 35% in cardiotoxin injured lung (Dooner et al., 2004). In addition, total replacement of hair follicles and sebaceous glands in the skin has been observed (Badiavas et al., 2003).

A great deal of negative attention has been directed to the phenomena of cell fusion since the initial reports of embryonic SCs fusing with marrow cells or with neurosphere cells (Ying et al., 2002; Terada et al., 2003). Fusion is of course a normal mechanism in hepatic or skeletal muscle development and with repair after injury. In addition it is well known that macrophages fuse to produce inflammatory cells under inflammatory conditions. Therefore, it is not surprising that fusion has been seen in some of the injury models in which marrow cells have been shown to convert to non-hematopoietic tissue cells. In fact, in the FAH− mouse model fusion is the mechanism of conversion and this results in the cure of some of these mice (Lagasse et al., 2000). However, studies in different models have shown that there can be substantial conversion of marrow cells to hepatic cells (up to 20%) without cell fusion being involved (Ying et al., 2002a,b). There are now numerous studies showing that with particular transplant models substantial conversion rates can be obtained without fusion. In a series of elegant studies employing a Cre/lox recombinase transplant model, Harris et al. (2004) have clearly demonstrated marrow conversion to epithelial cells in the lung, liver, and skin without cell fusion. Thus it appears that fusion may or may not explain the conversion phenomena and that the presence of fusion is very model dependent. In addition, fusion does not preclude therapeutically significant conversions of marrow cells to non-hematopoeitic tissue cells. It should be regarded as an interesting means to an end. The amount of negative commentary on fusion is clearly unwarranted.

An important aspect of plasticity studies is the number of variables which can affect the conversion outcomes, such as the timing of the transplant, the cell dose and number of cell infusions, the model of cell delivery, the functional state of the marrow cell delivered, and the nature of the marrow cell population or subpopulation. However, injury and marrow chimerism seem to be the major factors in obtaining a significant rate of marrow conversions. Injury is probably most important, and the optimum type and extent of injury will vary for each tissue studied. Optimums for injury in the conversion setting have yet to be definitively determined for any tissue. Irradiation is routinely used to induce marrow or blood chimerism, which is essential in most models for conversion. Whether irradiation promotes conversion by depleting endogenous tissue SCs remains an open question.

Unfortunately, a failure to address many or all of these variables appears to underlie prominent “negative” studies on marrow SC plasticity (Castro et al., 2002; Wagers et al., 2002). Reports that failed to demonstrate SC plasticity employed either no injury or the wrong injury. An important point here is that in all the studies on skeletal muscle, lung, and liver that showed conversion events, the level of conversion and whether or not fusion was involved related to the specifics of the model. For instance, without injection of an agent injuring muscle, such as cardiotoxin, conversions are either not seen or are extraordinarily rare, but with local injection of cardiotoxin conversions are always seen and have been observed to exceed 12% of the repaired muscle (Abedi, 2004). It would appear that each tissue probably has its own optimum injury and set of circumstances for marrow conversion. At present, in no tissue has an optimum regimen for conversions been worked out.

In summary, a number of groups have demonstrated high levels of conversion of marrow cells to non-hematopoietic tissue cells, especially in the liver, lung, and skeletal muscle. In some cases there have been clinically important functional effects.


A number of clinical trials have been initiated, mostly for cardiac disease. The results of controlled trials suggest that cell infusion is relatively safe. The only randomized trial in which marrow cells were infused after angioplasty in patients with severe ischemia showed an improvement in ejection fraction at 6 months in the marrow treated patients (Hertenstein et al., 2003).

Patients with chronic refractory wounds have been effectively treated with local installation of fresh and cultured marrow cells (Badiavas and Falanga, 2003). Thus, the capacity of marrow cells to effectively replace cells and increase function in some tissues has clearly been demonstrated and the results suggest that these approaches can be clinically effective.

Based on the current experimental animal data describing marrow SC conversion to normal lung tissue, one can envision application of SC therapy to a number of pulmonary diseases for which there are few effective treatments.

Cystic fibrosis

Grove et al. (2002) have demonstrated transport of a retrovirus via a SC carrier to lung. This study demonstrated that bone-marrow derived SCs can maintain their ability to differentiate into lung epithelium while maintaining long-term transgene expression. Thus, SCs might serve as a vector in diseases where replacement of a protein or DNA would be useful. Cystic fibrosis is such a disease. In vitro studies suggest that a clinical improvement would occur in cystic fibrosis patients if 10%–15% of their airway cells had the cystic fibrosis transmembrane regulator. Delivery of gene therapy for these patients via SC vectors may thus offer new hope for treatment.


In asthma, there are currently no effective preventions or treatments of “airways remodeling.” Airways remodeling consists of epithelial shedding, thickening of the basement membrane, increase in blood vessel cross-sectional area, airway smooth muscles hyperplasia and hypertrophy, mucous gland and goblet cell hyperplasia and increased collagen deposition (Dunnill, 1971; Carroll et al., 1993; Montefort et al., 1993a,b; Jeffrey, 1998; Busse et al., 2000). There is accumulating evidence that the epithelium of asthmatics is structurally and functionally abnormal (Bucchieri et al., 2002; Rosendahl et al., 2002). Holgate and colleagues argue that the communication between epithelium and mesenchyme in asthma is similar to remodeling of airways during embryonic development (Zhang et al., 1999; Davies et al., 2003). Epithelial damage and TH2 cytokines act in concert to disturb the “epithelial–mesenchymal trophic unit,” leading to myofibroblast activation and inflammatory remodeling processes. SC therapy offers the promise of reduction of inflammation, correction of the epithelial and mesenchymal abnormalities, and perhaps correction of the remodeled airways.

Idiopathic pulmonary fibrosis

Usual interstitial pneumonitis affects almost 40,000 patients annually in the United States alone, and there are no effective treatments. Recent data suggests that collagen overproduction may originate from cells derived from bone marrow precursors rather than parenchymal lung fibroblasts (Hashimoto et al., 2004). Whether the inflammation in usual interstitial pneumonitis will supply homing signals for corrective marrow SCs or whether conditioning will be required is unknown. The hope of SC therapy would be replacement of fibroblast foci with normal tissue, reduction of collagen deposition, and restoration of normal lung architecture.


Emphysema is a complex problem because of the need for structural regeneration of basement membrane and replacement of alveolar and endothelial cells. Recent studies have suggested a role for RA in alveolar regeneration (Hind and Maden, 2004). Ishizawa et al. (2004) treated mice with elastase-induced emphysema with ATRA, granulocyte colony-stimulating factor (G-CSF), or a combination of both. Both ATRA and G-CSF promoted lung regeneration and increased bone marrow derived cells in alveoli, but combined treatment had an additive effect, suggesting a role for bone marrow cell mobilization in lung regeneration.

A pilot study of ATRA treatment for human emphysema (Mao et al., 2002) has been completed. In this study, investigators randomized 20 patients with severe emphysema to receive ATRA or placebo for 12 weeks. ATRA did not demonstrate improvement in serial pulmonary function testing, quality of life scores or in the radiographic appearance of the emphysema by high resolution CT scan. The findings of a larger trial have yet to be published.

Acute respiratory distress syndrome

Acute respiratory distress syndrome can result from myriad of local or systemic insults that lead to injury of AE I and pulmonary capillary endothelial cells. Over the ensuing 5–10 days, this injury may lead to a fibroproliferative phase of the syndrome, resulting in pulmonary fibrosis and gross physiologic derangements that necessitate prolonged mechanical ventilation with an attendant increase in morbidity, mortality, and cost.

SC transplantation has been shown to promote wound healing (Badiavas et al., 2003) and to reverse bone marrow fibrosis in myelofibrosis (McCarty, 2004). As previously described, Ortiz et al. (2003) demonstrated a reduction of inflammation and fibrosis with MSC transplantation in the bleomycin-injured lung. These effects may have been the result of partial repopulation of alveolar surfaces, production of inflammatory cytokine antagonists by the MSCs, or scavenging of molecules involved in the fibroproliferative process by the transplanted cells.

Given what is know about ARDS and SC transplantation, it is reasonable to speculate that MSC could partially restore alveolar epithelial surfaces or modulate the inflammatory and fibrotic cascades involved in the pathophysiology of ARDS, and hence improve outcome.

Primary pulmonary hypertension

Primary pulmonary hypertension (PPH) is characterized by multiple abnormalities of cellular function, the most prominent of which include uncontrolled proliferation of pulmonary vascular endothelial cells and proliferation and hypertrophy of pulmonary vascular SMCs. These abnormal growth responses result in a variety of vascular lesions and may be the result of a heightened response to normal levels of growth factors, loss of normal apoptosis, increased expression of vascular growth factors, or a change in the balance of mitogenic versus anti-mitogenic factors. Pulmonary vascular endothelial cells also show significant functional abnormalities. Studies have demonstrated decreased expression of endothelial nitric oxide synthase (eNOS), increased synthesis of the potent vasoconstrictor endothelin-1 (ET-1) (Giaid et al., 1993; Giaid and Saleh, 1995), deficiencies in prostacyclin synthase (Tuder et al., 1999) and functional abnormalities in the voltage gated potassium channels, serotonin transporter, and BMP type II receptor (Yuan et al., 1998; Thomson et al., 2000; Eddahibi et al., 2001). It is unclear whether these impairments are responsible for the development of pulmonary vascular disease or are the result of a damaged pulmonary vasculature, but recent studies demonstrating increased susceptibility to pulmonary hypertension in patients with mutations in the BMPR2 and 5HTT genes suggest a causal relationship (Thomson et al., 2000; Eddahibi et al., 2001).

There is no evidence that current therapies cure the underlying disease and overall mortality remains close to 20% after 3 years of therapy. Lung transplantation remains an option, but the number of potential recipients far exceeds the available supply and 5-year mortality from single lung transplant exceeds the mortality of medical therapy for this disease. Genetic manipulation strategies have been hampered by difficulties in targeting the microvascular circulation; adenoviral vectors appear to be more effective at introducing plasmids to alveolar epithelial cells than to pulmonary endothelium or vascular smooth muscle.

Repopulating the pulmonary circulation with normally functioning endothelial cells is theoretically attractive and the development of methods that favor the localization of transplanted SCs to the pulmonary circulation would result in a powerful tool for targeting the lung for gene therapy. The feasibility of this approach was recently demonstrated by Nagaya et al. (2003). These investigators were able to significantly reduce pulmonary hypertension in monocrotaline-treated rats by infusing human umbilical vein endothelial cells (HUVECs) that had been genetically altered to overexpress adrenomedullin. Interestingly, infusion of untreated HUVECs also reduced pulmonary hypertension, although to a lesser degree.

General treatment approaches

There are a number of pulmonary diseases which may be candidates for SC therapy (Table 3). There are also many different strategies that may be used to achieve a therapeutic effect. Exogenous SCs could be administered intravenously into recipients or instilled directly into diseased airways via bronchoscopy. Endogenous SCs could be repeatedly mobilized from the bone marrow to the lung with G-CSF. This strategy may require inflicting additional injury on the lung to encourage homing of mobilized SCs to the lung; however, the disease itself may provide an adequate environment for sufficient homing to occur. Alternatively, SC homing may be facilitated by arming SCs with antibodies directed to antigens expressed by the injured tissues (Lum et al., 2004). In addition, exogenous SCs could be treated ex vivo with a variety of pulmonary growth factors to enhance conversion to lung cells.

Table 3. Pulmonary diseases that may be candidates for stem cell therapy
DiseaseTherapeutic target
Cystic fibrosisStem cells used as a vector to deliver the correct CFTR gene
AsthmaReduce inflammation and reverse airways remodeling
IPFReplace fibroblast foci with normal lung tissue and reduce collagen production
EmphysemaRestore alveolar epithelial surfaces and vasculature
ARDSModulate inflammation and reduce collagen deposition
Pulmonary hypertensionReplace the pulmonary vasculature with normally functioning endothelial cells


Recent studies have demonstrated that adult marrow SCs are more plastic than once believed, capable of converting into cells of non-hematopoietic organs. The factors which influence this phenomenon are largely unknown but it would appear that injury, through a variety of modalities plays a key role in the homing of these cells and their conversion to cells of the target organ. Investigators have reported the appearance of functionally and structurally normal alveolar epithelial cells derived from donor SCs and, in some studies, the attenuation of inflammation and fibrosis. It is unclear if the quantity of conversion is sufficient to make an appreciable impact on human pulmonary disease. The timing of donor cell infusion with respect to cell cycle, population of cells used (whole bone marrow versus a purified SC population), mobilization with G-CSF, or use of growth factors may augment conversion and yield clinically significant outcomes. It is possible that SC infusion during a proinflammatory or profibrotic process could augment these processes, so human trials must proceed with caution. We are excited by the results of the formative studies reviewed herein and anticipate that SC transplantation may someday provide clinicians with novel treatment modalities for pulmonary diseases that currently have few options.