Bone marrow-derived cells as progenitors of lung alveolar epithelium

Accumulating data demonstrate that bone marrow-derived cells can serve as precursors for differentiated cells of multiple organs. In these studies, adult bone marrow cells give rise to tissue structures previously held to be derived only from one of the original embryonic cell layers (Eglitis and Mezey, 1997; Ferrari et al., 1998; Gussoni et al., 1999; Kopen et al., 1999; Lagasse et al., 2000; Orlic et al., 2001; Pereira et al., 1995; Pereira et al., 1998; Petersen et al., 1999; Prockop, 1997; Woodbury et al., 2000). This has challenged long held views regarding the potential origin of somatic stem cells and the fixed nature of stem cell repertoire. Based on ex vivo experimental assays of unlimited self-renewal and clonogenecity, subtypes of bone marrow-derived cells meet the classical criteria for stem cells (Bianco and Robey, 2000; Lemischka, 1999; Prockop, 1997). In addition, but no less important, progenitor cells with more limited proliferative and differentiation repertoire also exist in the bone marrow (Bianco and Robey, 2000; Lemischka, 1999; Prockop, 1997).

To date, systemically injected mouse bone marrow-derived cells have been demonstrated to differentiate into parenchymal cells of various non-hematopoietic tissues including muscle, cartilage, bone, liver, heart, brain, intestine and lung (Alison et al., 2000; Eglitis and Mezey, 1997; Ferrari et al., 1998; Gussoni et al., 1999; Hou et al., 1999; Kopen et al., 1999; Krause et al., 2001; Lagasse et al., 2000; Nilsson et al., 1999; Orlic et al., 2001; Petersen et al., 1999; Prockop et al., 2000; Woodbury et al., 2000). The engrafted marrow cells are able to adopt the morphologic and molecular phenotype of their new resident organ. In the lung, bone marrow-derived cells have been detected in recipient lung tissue as fibroblast-type cells (Ding et al., 1999; Ono et al., 1999; Pereira et al., 1995; Pereira et al., 1998) and recently as differentiated bronchial epithelium and alveolar type II pneumocytes, raising the possibility of utilizing marrow to generate lung epithelium (Krause et al., 2001). The easy accessibility of bone marrow cells, along with their ability to adopt new phenotypes, has broad implications for disease therapy of many organ systems.

Fresh bone marrow cultured in plastic dishes can be separated into two general populations (Bianco and Robey, 2000; Phinney et al., 1999; Prockop, 1997). Cells that remain floating in the culture dish supernatant media are classical hematopoietic stem cells. Cells that adhere to the plastic culture dish grow into heterogeneous colonies of three discernible morphologic subtypes: small rounded cells positive for factor VIII antigen and CD-31 that may be of endothelial stem cell lineage, small stellate cells positive for mac-1/CD11b and CD-45 that are likely of myeloid lineage, and large polygonal fibroblast-like cells that synthesize matrix proteins such as collagen I, collagen IV, laminin and fibronectin (Conget and Minguell, 1999; Pittenger et al., 1999; Prockop, 1997). This latter cell type can serve as precursor for bone, muscle, fat and cartilage tissue (Bianco and Robey, 2000; Hou et al., 1999; Mackay et al., 1998; Nilsson et al., 1999; Phinney et al., 1999; Pittenger et al., 1999; Prockop, 1997), and can, under certain experimental conditions, transdifferentiate into non-mesodermal cell types (Kopen et al., 1999; Liechty et al., 2000; Prockop et al., 2000; Woodbury et al., 2000). A notable example of transdifferentiation is the adoption of a neural cell morphology and molecular phenotype during culturing (Woodbury et al., 2000) and following direct injection of these cells into the brain (Kopen et al., 1999).

We have explored the fate and phenotype of cultured plastic-adherent bone marrow cells that engraft in lung tissue. Our results demonstrate that engrafted cells can acquire the morphological and molecular features of type I cells of the alveolar epithelium. While limited engraftment occurs in the normal lung, we found that engraftment is facilitated after bleomycin-induced lung injury. These novel findings extend the list of differentiated cell types that bone marrow can adopt in vivo. In addition, they provide a new experimental model for characterizing the molecular signals that control lung epithelial cell differentiation.

Donor bone marrow was harvested from 1 month old B6,129S GTRosa26 or C57BL/6 GTRosa26 female mice. These mice constitutively express the lacZ/neomycin resistance transgene (lacZ/neo) in all tissues. lacZ expression is detectable after X-gal staining, which imparts a bright blue cytoplasmic stain (Friedrich and Soriano, 1991). Recipient mice were strain-matched 1 month old female B6,129SF1 or C57/BL6 wild-type mice. All mice were purchased from Jackson Labs (Bar Harbor, ME). Animal studies were conducted according to protocols approved by the Boston University animal use committee and adhered strictly to NIH guidelines for the use and care of experimental animals.

Whole marrow was harvested by flushing femurs, tibiae, and humeri with ice cold Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Grand Island, NY). Marrow cells were plated in plastic dishes at 10⁶ cells/cm² and re-fed every 2-3 days (DMEM with 10% FBS and 1% penicillin-streptomycin). Subconfluent plastic-adherent cultured cells were harvested at 1 week by trypsinization (0.25% trypsin/1 mM EDTA) for 5 minutes at 37°C followed by gentle scraping. Cells were collected by centrifugation (800 rpm) and washed with PBS.

Recipient mice were anesthetized and treated with intratracheal bleomycin (0.075 Units bleomycin in 0.1 ml PBS) or PBS alone. Plastic-adherent cultured bone marrow cells (1-2×10⁶ in 0.2 ml PBS) were injected into the tail vein 5 days later. At 1, 2.5, 5, 14 or 30 days after marrow cell injection, mice were sacrificed by cervical dislocation followed by transection of the abdominal aorta.

Lungs were fixed for 15 minutes by instillating 0.25% glutaraldehyde in PBS (pH 7.4, 22°C) through a tracheal catheter. The fixative was removed before PBS lavage. X-gal solution (0.1% X-gal, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 2 mM MgCl₂ in PBS, pH 7.4) was instilled. After tracheal ligation, harvested lungs were incubated in X-gal solution overnight, washed with PBS, and fixed in 4% paraformaldehyde (Polysciences, Warrington, PA.) Intact lungs were screened under a dissecting microscope to localize engrafted blue-stained donor cells. Blue “dots” representing clusters of stained cells were microdissected, and embedded in paraffin or plastic (methyl methacrylate) and sectioned by standard methods. In addition, 20 random lung paraffin sections from the right lung of each recipient were processed for analysis.

After sections containing lacZ-expressing cells were photographed, the coverslips were removed and the sections were processed for immunostaining or lectin binding. Before staining, sections were treated with hydrogen peroxide in methanol (0.3%, 15 minutes, 22°C) to quench endogenous peroxidases. For lectin binding, sections were incubated with 1% rabbit serum in PBS for 15 minutes, washed with PBS, and exposed to biotinylated Lycopersicon esculentum lectin (2 μg/ml; Vector Laboratories, Burlingame, CA) for 30 minutes. HRP-labeled avidin (5 μg/ml; Vector Laboratories) was applied for 30 minutes followed by the chromogen diaminobenzadine. The slides were counter-stained with Nuclear Fast Red. For T1α immunohistochemistry, antigen retrieval was performed by heating sections to 90°C in a citric acid buffer (antigen retrieval solution; Vector Laboratories) for 20 minutes, slowly cooling to room temperature, prior to quenching. Sections were blocked with 1% goat serum in PBS (60 minutes), incubated overnight (4°C) with a monoclonal hamster anti-mouse T1α antibody (Developmental Studies Hybridoma Bank, University of Iowa, Hybridoma #8.1.1, www.uiowa.edu/~dshbwww/, courtesy of Dr Andrew Farr) (Farr et al., 1992), and treated with HRP-labeled goat anti-hamster IgG (ICN, Costa Mesa, CA) for 30 minutes at room temperature. Signals were amplified using tyramide (TSA-Biotin System, NEN Life Science Products, Boston, MA) according to the manufacturer’s protocol before exposure to diaminobenzadine (6.5 minutes). Cell proliferation was assessed immunohistochemically by staining for proliferating cell nuclear antigen (PCNA), using an anti-PCNA antibody according to manufacturer instructions (PCNA Staining Kit, Zymed Laboratories, San Francisco, CA).

PCR analysis for the lacZ/neomycin resistance gene (lacZ/neo) (Friedrich and Soriano, 1991) was performed on DNA extracted from recipient tissue using primers that span the fusion gene (forward primer: CCGCATCCCCTGCTGTCCCGTGCA, reverse primer: CTCCCCCAACCCCCTGTCTGCTGT; 94°C 30 seconds, 53°C 60 seconds, 72°C 60 seconds, 35 cycles). The identity of the 140 base pair PCR product was confirmed by sequencing. To assess fidelity of PCR, the mouse surfactant protein C gene (SP-C) was also amplified in each sample. Lung tissue from the donor B6,129GtRosa 26 mouse strain served as positive control and lung tissue from a wild-type recipient B6,129SF1 served as negative control.

Bone marrow obtained as above was plated on plastic chambered slides (3.5×10⁵ cells/well) (Lab-Tek 8 well chambered Permanox Slides, Naperville, IL,) and cultured as above. After 1 week, subconfluent adherent cultured cells were fixed with ice-cold 5% acetic acid in ethanol. Using the above staining methods, cells were immunostained for T1α protein expression. Whole murine femur was fixed in 4% paraformaldehyde, decalcified by immersion in Immunocal (Decal Chemical, Congers, NY) for 2 weeks before washing with deionized water, embedding in paraffin, sectioning and immunostaining for T1α.

10 μg of total RNA, isolated from fresh marrow aspirates and marrow cultured on plastic for 1 week was reverse transcribed (Promega Reverse Transcription System, Madison WI) before PCR amplification of T1α , aquaporin-5 (AQP-5), and surfactant protein-C (SP-C) cDNAs (forward primers: for T1α TGTGGACCGTGCCAGTGTTGTT, for AQP-5 CCTTATCCATTGGCTTGTCGG, for SP-C ATTACTCGGCAGGTCCCAGGAGCCA; reverse primers: for T1α TCATCTTCCTCCACAGGAAGAGGA, for AQP-5 TCTGAGCTGTGGCAGTCGTTC, for SP-C AGATATAGTAGAGTGGTAGCTCTCC; 94°C 30 seconds, 55°C 60 seconds, 72°C 60 seconds, 30 cycles). Using identical amplification conditions, nested PCR was subsequently performed with a second set of primers (forward primers for T1α for AAGGCACCTCTGGTACCAACG, for AQP-5 TCTTCTGGGTAGGACCGATCG, for SP-C TTGTGGTGGTGGTCCTCGTT; reverse primers for T1α ACAATGAAGATCCCTCCGACG, for AQP-5 GAGAGGTGCTCCAAACTCTTCGT, for SP-C AGGTCTCTCCCGGAAGAATC). The identity of bands was confirmed by size estimation and restriction mapping.

A recipient animal was defined as showing lung engraftment of donor-derived cells if at least one lacZ-expressing cell was found by X-gal staining. An engrafted cell cluster was arbitrarily defined as more than 3 engrafted cells seen in one high power field by X-gal staining. The proportion of animals in each group (bleomycin exposed versus PBS exposed) showing lung engraftment or showing cluster pattern engraftment was assessed by performing a Fisher’s exact test. P<0.05 was taken as a statistically significant difference.

X-gal-stained cells were readily apparent in bleomycin-treated lungs at 30 days after injection of adherent lacZ-positive cells. The engrafted cell clusters appeared as blue dots of variable size in the parenchyma of recipient lungs (Fig. 1A). PCR analysis confirmed the presence of the lacZ/neo fusion gene (Friedrich and Soriano, 1991) in the microdissected blue clusters (Fig. 1B). The identity of the expected 140 bp lacZ/neo PCR product was validated by DNA sequencing.

To assess whether donor cell seeding of recipient marrow was associated with lung engraftment, we analyzed recipient bone marrow by PCR and by X-gal histochemistry. We found no evidence of bone marrow engraftment in any of the mice with lung engraftment 2 or 4 weeks after injection. In some animals, donor cells were observed in recipient spleens suggesting that the injected marrow cells did circulate beyond the initially encountered lung capillary bed. In contrast, we did not observe engraftment in recipient livers or hearts.

To examine the cell type of the lung-engrafted lacZ-positive cells, we performed a microscopic analysis of paraffin-embedded recipient lungs that contained blue X-gal-stained clusters (Fig. 2). These studies revealed clusters of flat blue lines along the alveolar surface. This morphology matches the normal location, size, and orientation of type I pneumoctyes of the alveolar epithelium (Fig. 2A,B). For better cellular definition, blue dots from injured recipient lungs were also embedded in plastic. Analysis of plastic sections (1-2 μm) confirmed that the lacZ-labeled donor cells had become flattened cells that lined alveoli, contained ovoid nuclei that bulged slightly into the alveolar lumen, and were located adjacent to type II cells. These are features that uniquely define type I pneumocytes of the alveolar epithelium (Dobbs et al., 1988) (Fig. 2D). In areas where engrafted type I cells abutted type II cells, the blue X-gal staining of the type I cell cytoplasm abruptly ceased at the type I-type II cell interface.

A summary of our findings for animals sacrificed 30 days after cell injection is shown in Fig. 3. Bleomycin-induced lung injury affected both the frequency and the pattern of lung engraftment in recipient mice. There was a statistically significant difference in the proportion of mice showing lung engraftment in the bleomycin-exposed group (4 of 4 mice) compared to the PBS-exposed group (2 of 9 mice; P=0.02). We also found a difference in engraftment patterns between the groups. PBS-exposed mice showed single engrafted cells located either at or near the pleural surface, whereas bleomycin-exposed mice showed engraftment as peripheral clusters of numerous type I pneumocytes in 3 of 4 engrafted mice (P=0.014). Bleomycin-exposed recipients showed 2-6 clusters/lung. All animals that received wild-type, non-lacZ-expressing bone marrow injections did not stain for X-gal.

We performed lectin histochemistry and T1α immunohistochemistry to assess expression of type I cell-specific markers on engrafted lacZ-expressing cells. We first examined lung engrafted cells for binding to the Lycopersicon esculentum lectin. In lung tissue, this lectin binds to the surface of type I pneumocytes, but does not bind to type II cells or lung endothelium (Bankston et al., 1991). A representative lung section containing engrafted cells subjected to lectin staining, is shown in Fig. 4. We found that all lung engrafted X-gal-stained cells bound Lycopersicon esculentum lectin.

We subsequently assessed the expression of T1α in the engrafting cell population in recipient lungs. In normal rodent lung, this protein is expressed only in type I cells and is targeted to the apical surface (Cao et al., 2000; Danto et al., 1992; Ramirez et al., 1997; Rishi et al., 1995; Williams et al., 1996). The hamster anti-mouse T1α monoclonal antibody employed in this analysis has staining characteristics identical to T1α antibodies used on rat tissue, and as expected, specifically identifies a single protein of ∼38 kDa in homogenates of mouse lung (Y. X. Cao, M. I. Ramirez and M. C. W., unpublished). Immunostaining of lacZ-positive donor lungs (Fig. 5A-C) demonstrates that both lacZ and T1α markers can be identified on a single section and illustrates the typical linear staining pattern for a type I pneumocyte marker (Dobbs et al., 1988; Williams et al., 1996). Panels A-C establish the antigen specificity of this antibody after X-gal staining and re-confirm that the T1α protein is not expressed in type II pneumocytes (Fig. 5A), endothelium (Fig. 5B), and alveolar macrophages (Fig. 5C). We immunostained recipient lungs (Fig. 5D,E) and found that the blue engrafted donor cells express T1α , further evidence that these marrow-derived cells adopt molecular phenotypic features of type I pneumocytes.

At 5 and 14 days after injection into bleomycin-treated mice, engrafted cell clusters could be observed; all engrafted cells had a type I cell morphology (Fig. 6). No lacZ-expressing type II pneumocytes were observed at any time point.

To further evaluate the possibility that donor cells engraft first as type II cells and later acquire the type I cell phenotype, bleomycin-injured mice were also sacrificed at 1 and 2.5 days after injection. At these time points, donor-derived cells could be detected both in the pulmonary vasculature (Fig. 7A), and in the injured recipient lung parenchyma. Within the parenchyma, clumps of 1-3 donor cells were found along the alveolar epithelial surface (Fig. 7B-D). These cells appeared elongated with attenuated cytoplasmic extensions, and none possessed the location, shape, or lamellar bodies characteristic of type II pneumocytes when examined by light and phase-contrast microscopy (Fig. 7E). For comparison, no positive X-gal-stained cells were observed at either time point in bleomycin-injured wild-type recipients injected with wild-type (non-lacZ expressing) cultured marrow.

We used PCNA immunohistochemistry (Waseem and Lane, 1990) to assess for evidence of cell proliferation in lung engrafted cells from bleomycin-exposed recipients 1, 2.5 and 5 days after cultured marrow injection (Fig. 7F). We found no positive PCNA staining in engrafted cells. In addition, no mitotic figures were seen in any lung engrafted cell at any time.

Plastic-adherent cultures of donor bone marrow cells display three discernible morphologic subtypes (Fig. 8A). These include small rounded cells, small stellate cells and large polygonal fibroblast-like cells (Phinney et al., 1999; Wang and Wolf, 1990). To examine whether any of these cell types express type I pneumocyte markers during culturing, we performed T1α immunohistochemistry. Approximately 10% of cells are T1α positive after 1 week of culturing on plastic (Fig. 8B). T1α expression appears to be limited to cells with the morphology of the fibroblast-like cell sub-population, as shown by their polygonal shape (Fig. 8C). Within an aggregate of fibroblast-like cells, T1α expression is variable, as not all cells express this marker at detectable levels (Fig. 8D). For comparison, we immunostained cultures for a pan-endothelial marker (Leppink et al., 1989; Phinney et al., 1999). Consistent with reported work, this antigen is expressed in the small rounded cells that likely represent endothelial precursors; this sub-population was T1α negative (data not shown).

Nested RT-PCR analyses on mRNA derived from plastic-adherent bone marrow cell cultures show that T1α mRNA is expressed (Fig. 9). In addition, T1α mRNA is expressed in fresh bone marrow aspirates (Fig. 9). Aquaporin-5, another type I cell-specific marker (Funaki et al., 1998), is also expressed in both marrow cell preparations (Fig. 9). No signal was observed in either cell sample for surfactant protein C (SP-C) mRNA, a marker specific for alveolar type II cells (Fisher et al., 1989).

To further examine expression of T1α in bone marrow cells, we immunostained a decalcified murine femur embedded in paraffin. We found that a population of ovoid cells located in the bone immediately subjacent to the marrow cavity was T1α positive (Fig. 10). There were no cells within the marrow cavity that stained positive for T1α. These findings are in agreement with previous work showing that T1α is expressed in a subset of osteoblasts at the time of their encapsulation in the bony matrix (Wetterwald et al., 1996).

The present study is the second description of a bone-marrow derived cell capable of forming lung alveolar epithelium (Krause et al., 2001), and demonstrates for the first time that cultured bone marrow cells can serve as type I pneumocyte precursors. By injecting lacZ-positive plastic-adherent cultured bone marrow cells into recipient mice, we demonstrated engraftment of marrow-derived cells in recipient lungs as cells possessing both the morphologic and molecular phenotype of type I pneumocytes. While these are unique findings, they are consistent with recent work showing that bone marrow-derived cells can serve as progenitors for differentiated parenchymal cells of organs such as the brain (Kopen et al., 1999; Prockop et al., 2000; Woodbury et al., 2000), liver (Lagasse et al., 2000; Petersen et al., 1999) and epithelia of the gastrointestinal tract and lung (Krause et al., 2001).

The distinct cellular localization, morphology and molecular features of the engrafted cells indicate that the engrafted cells express the type I pneumocyte phenotype, and the PCR data confirm that positive blue X-gal staining arises from locally expressed E. coli-derived lacZ. We considered the possibility that the observed staining patterns resulted from the diffusion of the X-gal reaction product from another cellular source. We did not detect, however, positive staining in any other cell type within the alveolus. This is consistent with previous work showing that cell-specific expression of lacZ in the alveolar epithelium remains localized and does not leak into neighboring cells. (Strayer et al., 1998). Since no staining was observed in lungs from animals that received wild-type bone marrow cells that do not carry the lacZ transgene, positive blue staining cannot be attributed to endogenous murine galactosidase expression. In keeping with these results, previous work showed that endogenous galactosidase activity is only observed at acidic pH (Dimri et al., 1995; Severino et al., 2000; Weiss et al., 1997; Weiss et al., 1999); notably all our experiments were performed at pH 7.4.

We found that animals exposed to bleomycin are more likely to show lung engraftment of bone marrow-derived cells than PBS-treated animals. The oxidative damage produced by bleomycin leads to both alveolar endothelial and epithelial injury, inflammatory exudation, and fibroblast recruitment which in turn lead to enhanced alveolar cell turnover, increased permeability, and localized release of effector substances (Adamson and Bowden, 1979). These events could all conceivably contribute to the engraftment process. In our studies, marrow cell injection coincided with the point at which bleomycin-induced edema and necrosis of alveolar endothelial cells has already occurred, and type I pneumocyte injury is just beginning.

We believe our findings differ from previous work on lung engraftement by Krause et al., because of significant differences in experimental protocols. Krause et al., demonstrated that an uncultured bone marrow-derived stem cell was able to achieve multi-lineage engraftment in irradiated recipient mice, including lung engraftment as bronchial epithelium and type II pneumocytes, 11 months after bone marrow transplantation (Krause et al., 2001). In contrast, we observed lung engraftment of marrow-derived cells only as type I pneumocytes and did not find detectable marrow engraftment. Our use of recipients with intact bone marrow, induction of lung injury with bleomycin, and selective expansion of plastic-adherent cells could all conceivably account for our unique results. These cells, expanded on plastic, constitute only a small percentage of normal marrow. In addition, their functions and phenotypes may be altered by culturing. Finally, whether the type of marrow-derived cell used by Krause et al., was present in our cultures is not known. Our results, therefore, are consistent with the possibility that marrow-derived cells can engraft as other lung phenotypes under different experimental conditions.

No donor-derived type II pneumocytes were detected at any time, but we cannot completely exclude the possibility that engrafted type II cells were present prior to the first time point we studied (24 hours). If donor-derived type II cells were present, they could serve as a transient amplifying population that subsequently gives rise to type I cells, as is known to be the case in lung injuries (Adamson and Bowden, 1974). This possibility seems unlikely, however, since type II cells are a self-renewing local stem cell population. Residual type II cells expressing the genomic lacZ marker should be present if type II cell engraftment had initially occurred. Indeed, previous work studying tritiated thymidine-labeled type II pneumocytes in vivo showed that type II daughter cells remain labeled even after one of the progeny has differentiated into a type I cell (Adamson and Bowden, 1974), and Krause et al.’s report achieving marrow engraftment as type II cells showed these cells were detectable as early as 5 days and as late as 11 months after injection (Krause et al., 2001). It is also important to note that the type II cells in our X-gal-stained donor lungs constitutively express lacZ (Fig. 5A), indicating that any engrafted type II cell should similarly be detectable by X-gal staining.

We identified a clustered pattern of cell engraftment in injured lungs beginning 5 days after injection. These clusters could theoretically arise in several ways: a single marrow-derived cell might engraft in lung tissue and then clonally expand into a cell cluster, multiple individual cells might engraft separately in the same location, or a clump of injected cells could lodge in a lung vessel and engraft en masse. Our data do not yet distinguish between these possibilities. However, in preliminary work, we found no cell clusters present before 5 days after injection, and no cell clumps larger than 3 cells lodged in any pulmonary vessel. Furthermore, we found no evidence of proliferating engrafted cells studied by PCNA staining at 1, 2.5, and 5 days after injection, suggesting that clonal expansion may not be responsible for cluster formation. This clustered pattern of engraftment has also been observed in liver when marrow-derived cells engraft as differentiated hepatocytes during metabolically induced injury (Lagasse et al., 2000). Engraftment as cell clusters is therefore not unique to the lung.

These studies demonstrate that bone marrow cells are capable of expressing markers specific for type I cells, including T1α and aquaporin-5, and the RT-PCR data indicate the presence of T1α -expressing cells within aspirates prior to culturing. In cultured cells, we found T1α -expressing cells within colonies of cells with a polygonal morphology. This cell sub-type is believed to serve as pluripotent stem cells for muscle, bone, and fat (Phinney et al., 1999; Pittenger et al., 1999; Prockop, 1997). Colonies of these cells are believed to evolve in culture as a result of clonal expansion from a single adherent marrow cell (Bianco and Robey, 2000; Wang and Wolf, 1990). By immunostaining whole de-calcified bone, we found, as has been reported (Wetterwald et al., 1996), a T1α -positive long thin cell on the interior aspect of the bone facing the marrow cavity. These cells are believed to be osteoblasts or pre-osteocytes (Wetterwald et al., 1996). We do not yet know whether this cell was present in our original marrow aspirates and subsequently served as a precursor to the population of T1α -positive cultured cells.

The functional relationship between expression of type I pneumocyte markers by adherent bone marrow cells and engraftment potential in the alveolar epithelium is uncertain at this time. Our observations raise intriguing questions regarding the relative role of lung derived cues in establishing a lung epithelial phenotype. It is possible that cultured bone marrow-derived cells display certain features of lung differentiation prior to engraftment. It is also possible that multipotent marrow cells lose their engraftment potential as they express differentiation markers, such as T1α during culturing.

In conclusion, we believe that these studies provide a rationale for the eventual use of bone marrow progenitors in the treatment of conditions in which failure of the intrinsic regenerative capacity of damaged organs is a central pathological feature. In the lung, such an approach may be relevant for conditions associated with acute and extensive injury to the gas exchange surface.

We thank Dr Jerome Brody for thoughtful guidance and Anne Hinds for technical assistance with histology specimens. This work was supported by NIH grants P01HL47049 and U19AI41994.