Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Cystic fibrosis (CF) afflicts 1 in 3,000 live births in the Caucasian population each year and is the most common genetically inherited disorder in this population (1). The primary defect in (CF) is the absence of a chloride channel function facilitated by the cystic fibrosis transmembrane conductance regulator (CFTR) (2). CF lung disease is characterized by abnormally thick mucus secretion and chronic bacterial infections as a result of abnormal regulation of fluid and electrolyte balance in the airways (5). A major complication of this disease results from hypersecretory responses of the surface airway epithelium and submucosal glands to bacterial infection. Such hypersecretory responses in CF may involve both submucosal gland hypertrophy (expansion of existing gland mass) and hyperplasia (the formation of new glands from the surface airway epithelium) (8). The mechanisms of these pathologic hypersecretory responses are largely unknown but are presumed to be due to altered pathways of epithelial differentiation. Studies which attempt to understand the processes that regulate gland development in the airway will also aid in elucidating the pathophysiology of gland hyperplasia and hypertrophy found in this CF hypersecretory phenotype.

In addition to the involvement of submucosal glands in CF hypersecretory responses, submucosal glands have been hypothesized to have functional properties which directly affect the bacterial colonization process in the CF airway. This hypothesis is based on the finding that CFTR mRNA and protein expression in the proximal airway is found at the highest levels in submucosal glands (9). Specifically, defects in CFTR function in serous cells have been suggested to affect the secretion of antibacterial agents such as lysozyme and lactoferrin into the airway through alterations in the chloride and fluid secretory pathways of submucosal glands (9, 10). To this end, submucosal glands have been proposed to be potentially important targets for gene therapy of CF. These regions, which begin to develop in human airways at approximately 15 wk in utero, have been excluded as targets from initial gene therapy trials in adults because they are not readily accessible from the airway lumen. Because of the anatomic inaccessibility of submucosal glands in adults, we propose to test the feasibility of in utero gene therapy approaches to target submucosal gland progenitor cells prior to submucosal gland development.

To rationally develop in utero gene therapy strategies for CF, a concrete understanding of the progenitor cell targets with pluripotent capacity for airway differentiation is imperative. In this study, we have attempted to define an airway progenitor cell population capable of submucosal gland development by identifying transcriptional regulatory genes involved in gland development and morphogenesis. Based on findings in lymphoid enhancing factor 1 (Lef1)-deficient mice which demonstrate that Lef1 expression is required for epithelial-mesenchyme interactions needed to promote hair follicle and mammary gland development (11), we hypothesized that this high mobility group (HMG) transcription factor may also define progenitor cell commitment in the formation of airway submucosal glands. Lef1-deficient mice have normal lung morphology. This finding is not surprising given the fact that no evidence of Lef1 expression has been detected in developing mouse lungs. However, airway cell biology in the mouse differs greatly from that of humans. For example, with the exception of glands in the most proximal regions of the trachea, mouse cartilaginous airways do not contain submucosal glands and their predominant secretory cell type is the Clara cell. In contrast, human cartilaginous airways have abundant submucosal glands and their secretory cell type includes goblet cells. Hence, we have chosen a ferret animal model as an alternative to evaluate expression and potential functional involvement of the Lef1 in airway submucosal gland development and morphogenesis. The ferret is an attractive animal model for studying gland development due to its pattern of gland development, which begins immediately after birth and closely resembles that of human 15-wk in utero submucosal gland development (12). Furthermore, the anatomy and cell types of differentiated ferret submucosal glands are identical (by morphologic criteria) to those found in adult human airways (13, 14).

Lef1 is an HMG transcription factor which has been extensively studied for its regulatory effects on T-cell receptor- gene expression (TCR-) (15, 16). In this setting Lef1 mediates the activation of TCR- gene expression by facilitating the formation of tertiary transcription complex through DNA binding and by directly interacting with two other flanking transcription factors, ETS-1 and CREB. (17). The downstream genes activated by Lef1 have been identified to include cytokeratins in hair follicle formation (18) and the cell adhesion molecule E-cadherin (19).

In the present study, we have identified a specific cellular compartment of airway epithelial progenitor cells which have pluripotent capacity for submucosal gland development. This identification was based on the cell type-specific expression pattern of the Lef1 transcription factor within airway progenitor cells at the earliest stages of gland bud formation. Furthermore, we have demonstrated that in vivo targeting to this progenitor cell population using recombinant retroviral vectors provides stable transgene expression throughout gland morphogenesis. Such studies lay the foundation for the rational development of gene therapy strategies targeting submucosal glands in the treatment of CF airways in utero.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

fLef1 Cloning by RT-PCR

The ferret Lef1 (fLef1) cDNA was cloned by reverse transcription-polymerase chain reaction (RT-PCR) of ferret mRNA. mRNA was reverse-transcribed into cDNA using poly(T) primers followed by PCR amplification with nested primers (20). Primers used for the first round amplification included EL22: 5'GGACCCGGAACTCTGCGCCAC-3' (1022-1042 bp of the mouse Lef1 cDNA sequence) and EL23: 5'-ACGACATTCGCTCTCATTTCTTTCAT-3' (1911-1936 bp of the mouse Lef1 cDNA sequence). PCR conditions for amplification included 500 nM of each primer, 4 mM MgCl₂, 1.6 mM dNTPs, and 0.5 units of Taq DNA polymerase per 50 µl reaction for 35 cycles (cycling between 94 and 72^oC for 30 s at each temperature). Fragments generated from this first round of PCR were used as a template for a second round of PCR with nested primers. Nested primers included EL24: 5'-GATCTGAATTCATCTTCGCCGAGATCAGTCATCC-3' (1092-1114 bp of the mouse Lef1 cDNA sequence; EcoRI site incorporated into primer as underlined) and EL25: 5'-GATCTGGATCCCTTCACGTGCATTAGGTCACTGTC-3' (1809-1832 bp of the mouse Lef1 cDNA sequence; BamHI site incorporated into primer as underlined). The 5' region of fLef1 cDNA was obtained by degenerative PCR from ferret lung genomic DNA using primer EL119: 5'-GATCTGAATTCTCCRCAGCGGAGCGGAGATT-3' (-35 to -15 of mouse and human conserved Lef1 cDNA sense sequence; EcoRI site incorporated into primer as underlined) and primer EL120: 5'-GATGTGGATCCAAGGAGGACTTGATGTCGGCTAAGTCGCC-3' (141-173 bp of the ferret cDNA antisense sequence; BamHI site incorporated into primer as underlined). The PCR products were then unidirectionally cloned into the pCRscript (Stratagene, La Jolla, CA) using the EcoRI and BamHI restriction sites incorporated into primers. In total, three independent clones were sequenced to obtain a consensus devoid of Taq errors. The sequence in Figure 1 represents the composite of these independent fLef1 clones.

View larger version (50K): [in this window] [in a new window]   Figure 1.   Cloning of a partial ferret Lef1 cDNA. Thymus poly-A mRNA was used to clone a partial cDNA to the ferret Lef1 gene (fLef1) using RT-PCR. The 5' sequence encoding exon 1 of the fLef1 gene was cloned by degenerative PCR from ferret genomic DNA. The consensus nucleotide sequence encompassing the first 642 bp of coding sequence (methionine start codon designated at base pair 1 is presented on the right). The amino acid sequence presents homologies between ferret, human, and mouse Lef1 with amino acid 1 as the start methionine. Nonconserved amino acids between these species are represented in bold letters. This fLef1 cDNA was used to generate riboprobes (underlined sequence) for in situ detection of Lef1 mRNA in ferret tracheal tissue sections.

In Situ Hybridization

Four ferret tracheal developmental time points, including 0-6 h, 3 d, 1 wk, and 2 wk, were analyzed for fLef1 mRNA expression by in situ hybridization. Frozen sections (6 µm) were mounted and fixed in 4% paraformaldehyde/phosphate-buffered saline (PBS) for 2 h, then dehydrated through 30, 70, 95, and 100% ethanol. Slides were then treated for 30 min at 30^oC in 50 mM Tris (pH 8.0), 50 mM ethylenediaminetetraacetic acid (EDTA), and 10 µg/ml Proteinase K. After rinsing twice for 30 s in 0.2× standard saline citrate (SSC) they were again fixed in 4% paraformaldehyde/PBS for 20 min at room temperature. Slides were then rinsed in 0.1 M triethanolamine (TEA), pH 8.0, twice for 4 min, followed by acetylation in 0.0025% acetic anhydride containing TEA buffer for 10 min. Finally, slides were rinsed in 0.2× SSC twice for 2 min, followed by dehydration through a graded series of ethanols. Slides were stored under vacuum at 20^oC until analyzed.

Sense and antisense cRNA riboprobes were generated from linearized fLef1 cDNA plasmids by in vitro transcription (507 bases in length). A developmental panel of ferret tracheal sections from 0-, 3-, 7-, and 14-d postnatal ferrets were analyzed by in situ hybridization for Lef1 mRNA. In total, three animals were analyzed for each time point. Slides were prehybridized at 54^oC in 50% formamide, 2.5× Denhardt's solution, 0.6 M NaCl, 10 mM Tris (pH 8.0), 1 mM EDTA, 0.1% sodium dodecyl sulfate, 10 mM dithiothreitol (DTT), and 500 µg/ml Escherichia coli tRNA (all RNase free) for 4 h. Hybridization was performed in prehybridization solution containing 10% dextran sulfate and 1 × 10⁷ cpm/ml of sense or antisense riboprobes for 16-18 h. Coverslips were removed and slides were rinsed in 4× SSC four times for 5 min at room temperature. Slides were then incubated in 0.5 M NaCl, 1 mM EDTA, 20 µg/ml RNase, and 10 mM Tris (pH 8.0) for 30 min at 37^oC and washed in 2× SSC/1 mM DTT four times for 5 min at room temperature, followed by three washes in 0.5× SSC/1 mM DTT for 15 min each at 54^oC. Slides were then dehydrated through a graded series of ethanols, dipped in Kodak photographic emulsion NBT2, and exposed at 4^oC for 3 wk prior to developing and counterstaining in hematoxylin and eosin.

Generation of Newborn Ferret Tracheal Xenograft Model of Submucosal Gland Development

Newborn ferret tracheas were excised at 1-12 h after birth, ligated to flexible plastic tubing, and transplanted subcutaneously in the flanks of nu/nu athymic mice. Tracheal transplants were surgically implanted such that the lumen of the tubing could be accessed for in vivo gene transfer following transplantation (21). Our laboratory has reported similar methodologies used to study in vivo recombinant retroviral (22) and adenoviral (23) gene transfer within proximal airway epithelium of human bronchial xenografts. For determination of viral multiplicity of infection (MOI), the number of surface airway epithelial cells in newborn ferret tracheas was determined by morphometric quantification of epithelial nuclei from 15 independent tracheal sections of two animals. To assess whether the proposed manipulations of this ferret model system significantly altered the time course and extent of submucosal gland development, we compared the differentiated state of submucosal glands (SGs) formed at various times after birth between ferret tracheal xenografts and age-matched littermate native tracheas. Tracheas were harvested from both xenografts and age-matched littermate ferrets and either fixed in buffered formalin for paraffin sectioning or fresh-frozen in optimal cutting temperature (OCT) embedding media for cryosectioning. Submucosal gland development was compared between tracheal xenografts and native ferret tracheas at 1, 3, 5, and 8 wk after transplantation (xenografts) or postnatally (age-matched littermates). Because there is only a single duct leading to the airway surface per gland, we quantitated the number of glands in ferret xenografts and age-matched ferret tracheas by evaluating the number of gland ducts. The abundance of gland ducts was evaluated in every tenth section from three independent xenograft or native tracheal airways. On average, approximately 100 noncontiguous tracheal rings were analyzed from each group.

In Vivo Recombinant Retroviral Gene Transfer to Submucosal Gland Progenitor Cells

In vivo gene transfer studies were performed using recombinant lacZ (CB-LacZ, titer 5 × 10⁵ cfu/ml) and alkaline phosphatase (BA-Alkphos, titer 1 × 10⁸ cfu/ml) retroviral vectors (24, 25). CB-LacZ retroviral constructs contained the -galactosidase reporter gene under the direction of the cytomegalovirus (CMV) enhancer and -actin promoter, while BA-Alkphos retrovirus contained the alkaline phosphatase (Alkphos) reporter gene under the direction of the -actin promoter. Retroviral stocks were generated by harvesting 16-h conditioned media from clonal amphotropic psi-Crip cell lines. All recombinant retroviral producer cell lines were free of helper virus, based on a lacZ mobilization assay (26). Newborn ferret tracheas were excised and ligated to flexible plastic tubing and the lumen was filled with recombinant retroviral CB-LacZ or BA-Alkphos supernatants prior to or just after implantation subcutaneously in nu/nu mice as xenografts. Xenografts were infected with retroviral producer supernatants for 24 h in the presence of 2 µg/ml polybrene. Xenografts were harvested at 1, 2, and 5 wk after transplantation and frozen in OCT. Frozen sections were histochemically stained for -galactosidase or Alkphos activity as previously described (22). Tissues were incubated at 65^oC for 30 min prior to staining for Alkphos (transgene Alkphos is heat-stable, whereas endogenous Alkphos is heat-labile). Following staining, these samples were postfixed and stained in hematoxylin and/or eosin for morphologic analysis. The percentages of Alkphos-expressing ciliated and basal cells were quantitated by morphometric analysis of > 3,000 surface airway epithelial cells from nine independent xenograft samples ranging from 5 to 8 wk after transplantation. Similarly, the percentage of submucosal glands expressing Alkphos transgene was quantitated. In these experiments, > 200 glands were scored for the presence of Alkphos expression. Positive glands were denoted by the presence of Alkphos staining in all or a subset of gland tubules.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

fLef1 cDNA Cloning and Analysis of Expression in Submucosal Gland Progenitor Cells

Using primers derived from sequence homology between the mouse and human Lef1 genes, we cloned the ferret homolog cDNA to Lef1 (Figure 1) and evaluated its expression pattern during submucosal gland development and morphogenesis, using the newborn-ferret model of tracheal gland development. Ferret submucosal gland development initiates during the first week of life following the consolidation of surface airway progenitor cells in the formation of gland buds. fLef1 mRNA expression was evaluated using in situ hybridization with 507-bp cRNA sense and antisense fLef1 probes. In situ mRNA localization studies against a developmental panel of native ferret tracheas demonstrated high levels of fLef1 mRNA expression in newborn-ferret tracheal surface airway epithelial cells, which was confined to submucosal gland-forming buds (Figure 2). This pattern of expression in the surface airway epithelium produced punctate regions of hybridization to antisense fLef1 probes (Figures 2A-2F, 2H, and 2J-2O) which were not detected in sections hybridized to sense (Figures 2G, 2I, and 2P) or RNase-pretreated sections hybridized with antisense probes (data not shown). The earliest stages of gland development in which surface airway epithelial cells condense prior to gland-bud formation expressed Lef1 mRNA as seen in Figures 2A, 2D, and 2J-2O. Although gland formation initiates within the first week postnatally, newly forming gland buds continue to appear within the first 3 wk of life. In our studies, gland buds were most prominent at 3 wk postnatally. The expression of Lef1 mRNA in these newly forming primordial gland buds suggests that transcriptional influences regulating activation of Lef1 expression may define the phenotype of early submucosal gland progenitor cells. As gland buds invaginated into the interstitium, Lef1 expression was found predominantly at the invading tips of these primordial tubules (Figures 2B, 2C, 2H). During later stages of submucosal gland morphogenesis, Lef1 expression remained confined to the most distal ends of elongating tubules between 2 and 5 wk of airway development (Figure 2E). Additionally, Lef1 expression was seen within the elongating tips of tracheal cartilage (Figures 2A, 2B, 2F, 2L-2N).

View larger version (150K): [in this window] [in a new window]   View larger version (116K): [in this window] [in a new window]   Figure 2.   In situ localization of fLef1 mRNA expression during newborn-ferret tracheal development. In situ sense and antisense cRNA probes were generated from the fLef1 cDNA clone outlined in Figure 1. A panel of newborn tracheal developmental tissue sections (0, 3, 7, and 14 d postnatal) were hybridized to antisense (A-F, H, J-0) and sense (G, I, P) fLef1 cRNA probes. Panels represent 3-d (A, D), 1-wk (B, C, F, H), and 2-wk (E, G, I-P) sections. Open arrows mark Lef1-expressing, gland-forming buds and tubules, while solid arrowheads mark non-Lef1 expression regions of developing glandular tubules. Lef1 expression in cartilage (c) is also marked. Panels C, D, H, K, and O represent enlargement of Lef1-expressing regions in panels A, B, F, J, and N, respectively. All slides were exposed for photoemulsion for 3 wk prior to developing. Bars = 30 µm.

Recombinant Retroviral Gene Transfer to Submucosal Gland Progenitor Cells in the Developing Ferret Airway

In the present study we utilized ferret xenografts of native ferret airways transplanted subcutaneously in the flanks of nu/nu athymic mice as a model for evaluating in vivo targeting with retroviral vectors to Lef1-expressing submucosal gland progenitor cells. Because these studies involved the transplantation of native newborn-ferret tracheas into an altered growth environment, we felt it was necessary to confirm that the developmental time course of xenograft tracheal submucosal glands did not differ from that of native newborn-ferret airways. To this end, we compared gland development between 1 and 8 wk of newborn-ferret xenografted tracheas to that of age-matched tracheas from littermate control ferrets. No significant differences were seen between native and xenografted tracheas during the developmental stages of gland formation (compare Figures 3A and 3B with 3E and 3F). In native ferret tracheas, gland buds (or similar gland duct openings) were present at 2 wk postnatally in infrequent numbers (0.07 ± 0.05 glands/tracheal ring) and increased 10-fold by 3 wk (0.77 ± 0.09 glands/tracheal ring). By 5 wk postnatally, the number of glands per tracheal ring decreased to 0.38 ± 0.08, likely indicating a decrease in newly developing glands in the setting of continued tracheal growth (i.e., luminal diameter and length). Age-matched ferret tracheal xenografts at 5 wk demonstrated levels of gland development (0.32 ± 0.10 glands/tracheal ring) equivalent to those seen in native 5-wk ferret tracheas (Figure 3J). Furthermore, no differences could be detected between the extent of gland differentiation (i.e., serous and mucous tubules) between ferret tracheal xenografts and native age-matched ferret tracheas over the course of 8 wk of development (Figures 2C, 2D, 2G, 2H). Additionally, the growth of mesenchymal tissues such as cartilage and interstitial fibroblasts was also unaltered in xenograft transplants. The extent of cartilage growth is most evident by comparing the increase in tracheal luminal diameter of 1-wk with 8-wk xenografts (Figure 3I). These results suggest that the ferret tracheal xenograft model is capable of reproducing levels of surface airway and submucosal gland epithelial development similar to those seen in the native ferret airway. Furthermore, these studies also suggest that the epithelial-mesenchymal interactions necessary for morphogenesis and differentiation of submucosal glands were not significantly altered by changes in systemic factors, such as hormones contributed by the host. The development of this xenograft model of proximal airway submucosal gland development set the stage for evaluating gene targeting to submucosal gland progenitor cells using recombinant retroviral vectors.

View larger version (17K): [in this window] [in a new window]   View larger version (50K): [in this window] [in a new window]   Figure 3.   Submucosal gland development in newborn native and xenograft tracheas. Maturation of submucosal glands within the native ferret trachea at 1 wk (A, B), 8 wk (C), and adult (D). This is compared with the extent of submucosal gland formation of age-matched newborn ferret littermates in which tracheas were implanted as xenografts at birth and harvested at 1 wk (E, F ) and 8 wk (G, H) after transplantation. The extent of organ growth in xenografts harvested at 1 wk (top) and 8 wk (bottom) after transplantation is shown in panel I. The mean (± SEM) number of gland ducts is compared between native ferret tracheas (2, 3, and 5 wk) and ferret tracheal xenografts (5 wk) in panel J. Unmarked arrows point to primordial gland-forming buds and tubules from the surface airway epithelium. M, mucus tubules; S, serous tubules; C, collecting duct. Bars = 30 µm for panels A-H and 2 mm for panel I.

Using recombinant retroviral stocks of BA-Alkphos (1 × 10⁸ cfu/ml) and CB-LacZ (5 × 10⁵ cfu/ml), the lumen of newborn-ferret tracheal xenografts was infused with approximately 50 µl of retroviral supernatant prior to transplantation subcutaneously in nu/nu mice. Although the entire lumen of the xenograft and connected tubing is approximately 50 µl, the volume of a newborn ferret airway lumen is approximately 10 µl. However, since this fluid is reabsorbed by the tracheal implant, it is assumed that the total 50 µl viral dose (2 × 10⁶ cfu for BA-Alkphos and 2.5 × 10⁴ cfu for CB-LacZ) is exposed to the tracheal epithelium. Newborn-ferret tracheal epithelium contained approximately 5 × 10⁴ undifferentiated cuboidal epithelial cells (as determined by morphometric quantification of cell numbers in tissue sections); this level of infection represents an approximate MOI of 40 cfu/cell for BA-Alkphos and 0.5 cfu/ cell for CB-LacZ. Retroviral infection of newborn-ferret tracheas with BA-Alkphos demonstrated transduction efficiencies ranging from 5 to 40% of the surface airway epithelium at 5 wk after transplantation (Figures 4B and 4F). Morphometric quantification of nine independent BA-Alkphos-infected samples demonstrated transgene expression in 8.2 ± 4.5% of basal and 7.1 ± 3.8% of ciliated cells. CB-LacZ-infected xenografts served as negative controls for the specificity of Alkphos staining in these experiments and demonstrated a lack of detectable histochemical precipitate. By 5 wk after transplantation, clones of Alkphos-expressing epithelial cells contained goblet, intermediate, ciliated, and basal cells (Figure 4F), suggesting that pluripotent progenitor cells of the surface airway epithelium had been targeted. One complication encountered with the Alkphos reporter was that the Alkphos protein was localized to the apical portion of ciliated cells. Hence, when viewing the Alkphos transgene expression in surface airway epithelial cells (Figures 4B and 4F), apical expression should not be confused with a lack of transduction. This peculiar localization pattern of Alkphos did not pose a problem in viewing transgene expression in submucosal gland cells. Gene targeting of surface airway submucosal gland progenitor cells was inferred by visible Alkphos transgene expression in gland-forming buds at 1 wk after transplantation (Figures 4C and 4D) and in elongating tubules by 2 wk after transplantation (Figure 4E). Transgene expression was observed in all cells (Figures 4D and 4E) or a subset of cells (Figures 4B and 4C) of a given gland. This variability was likely dependent on the time of infection with respect to initiation of gland-bud formation. Transgene expression in primordial gland buds persisted throughout gland development and was seen in 36 ± 9% of xenografts glands at 5 wk after transplantation (Figure 4B). Due to the 200-fold lower titer of CB-LacZ retroviral supernatants, only few -galactosidase-expressing clones were detected following X-gal staining of CB-LacZ-infected xenografts (data not shown).

View larger version (165K): [in this window] [in a new window]   Figure 4.   Retroviral-mediated gene transfer to submucosal gland progenitor cells in newborn-ferret xenografts. Xenografts were infected with CB-LacZ (panel A) and BA-Alkphos (panels B-F) retroviruses for 24 h; harvested at 5 wk (panels A, B, F), 1 wk (panels C, D), and 2 wk (panel E) after transplantation; and histochemically stained for alkaline phosphatase activity. Closed arrows mark Alkphos-positive gland regions, while open arrows mark negative regions of staining within the same gland. Arrowheads mark Alkphos-positive clones in the surface airway epithelium (Alkphos localizes to the apical borders of ciliated cells but can be seen throughout the cytoplasm of basal and nonciliated columnar cells). Xenografts stained for LacZ demonstrated negligible gene transfer because of the 200-fold lower titer of these stocks in comparison with Alkphos vectors (data not shown). Hence, CB-LacZ-infected xenograft airways serve as negative controls for the specificity of Alkphos staining in these studies. Bars = 30 µm.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

CF has emerged as a model disease for testing gene therapies. However, due to the diversity of CFTR-expressing cellular targets in the lung, innovative approaches must be developed to reconstitute CFTR function in the various cellular compartments which may be needed to correct airways disease. One such cellular compartment which poses difficulties for gene targeting in CF is submucosal glands. Due to the anatomic inaccessibility of submucosal glands in the airway, we proposed to test the feasibility of in utero gene targeting to submucosal gland progenitor cells prior to gland formation. Such approaches require both a knowledge of progenitor-cell targets for submucosal glands and the use of integrating vectors capable of transducing expanding progenitor-cell populations. The present study has attempted to define the progenitor-cell targets for gene therapy of submucosal glands in the airway. We hypothesized that the HMG transcription factor Lef1 may be a useful marker for this progenitor-cell phenotype. This hypothesis was based on the recent finding that Lef1 knockout mice have defective development of organs which undergo epithelial-mesenchymal interactions similar to those seen in submucosal gland development.

To test the hypothesis that Lef1 gene expression may define a subpopulation of surface airway epithelial progenitor cells involved in submucosal gland development, we cloned a portion of the fLef1 cDNA for studies evaluating mRNA expression in developing newborn-ferret tracheas. The ferret provides an attractive model for the evaluation of human in utero submucosal gland development because the airways of this species develop submucosal glands postnatally (12). In situ hybridization studies confirmed our initial hypothesis that Lef1 gene expression is upregulated specifically in surface airway progenitor cells which give rise to gland-forming buds. Based on findings in Lef1 knockout mice, one would predict that Lef1 expression is required for the inductive process of gland development. However, since mice do not contain significant submucosal glands throughout their cartilaginous airways, this hypothesis is not easily testable in the Lef1 knockout model. Nonetheless, preliminary evidence evaluating nasal submucosal glands demonstrates that these glandular structures are absent in Lef1-deficient mice (data not shown). Such findings support the notion that Lef1 expression is required for submucosal gland development.

In summary, these studies have begun to define the cellular phenotype of progenitor-cell targets for gene therapy of submucosal glands in the developing proximal airways. Furthermore, these findings support earlier retroviral lineage work evaluating progenitor-progeny relationships of airway submucosal gland progenitor cells in human bronchial xenografts (27). This previous work has suggested that a subset of surface airway basal cells may represent a progenitor cell type with a capacity for submucosal gland development in the human airway. When considering findings in the present study, one would anticipate that Lef1 expression may define molecular characteristics of this particular gland progenitor cell type. Furthermore, the abundance of this progenitor-cell phenotype in the human bronchial xenograft model was consistent with the infrequent abundance of gland progenitor cells in the airway.

To assess our ability to target this submucosal gland progenitor-cell phenotype, we developed a xenograft model of the newborn-ferret trachea which allowed for the evaluation of in vivo gene targeting with recombinant retroviruses. Studies comparing the extent of submucosal gland development between ferret tracheal xenografts and native tracheas from age-matched control ferrets have demonstrated that the gland developmental processes are intact in this xenograft model. Studies using recombinant retroviruses harboring the recombinant Alkphos transgene have demonstrated successful gene transfer to primordial gland-forming buds. Furthermore, this transgene expression persisted throughout all stages of gland development. In summary, these studies have led to a better understanding of the progenitor-cell phenotypes which must be targeted to achieve persistent gene transfer to submucosal glands in developing proximal airways. Such studies have set the stage for a rational approach to in utero gene therapy strategies which target submucosal glands in CF.