Pulmonary Center, Boston University School of Medicine, Boston, Massachusetts
ABSTRACT
Hoechst-effluxing cells (side population cells) are a rare subset of cells found in adult tissues that are highly enriched for stem and progenitor cell activity. To identify potential stem and progenitor cells during lung development, we generated gene expression profiles for CD45 and CD45+ side population cells in the embryonic day 17.5 lung. We found that side population cells comprise 1% of total embryonic day 17.5 lung cells (55% CD45+, 45% CD45). Gene profiling data demonstrated an overrepresentation of endothelial genes within the CD45 side population. We used expression of several distinct genes to identify two types of CD45 side population cells: 1) von Willebrand factor+/smooth muscle actin+ cells that reside in the muscular layer of select large vessels and 2) von Willebrand factor+/intercellular adhesion molecule+ cells that reside within the endothelial layer of select small vessels. Gene profiling of the CD45+ side population indicated an overrepresentation of genes associated with myeloid cell differentiation. Consistent with this, culturing CD45+ side population cells was associated with induction of mature dendritic markers (CD86). The microarray results suggested that expression of myeloperoxidase and proteinase-3 might be used to identify CD45+ side population cells. By immunohistochemistry, we found that myeloperoxidase+/proteinase-3+ cells represent a small subset of total CD45+ cells in the embryonic day 17.5 lung and that they reside in the mesenchyme and perivascular regions. This is the first detailed information regarding the phenotype and localization of side population cells in a developing organ.
stem cells; myeloid; vascular; microarray; side population; lung development; Vwf; Mpo; Prtn3; CD45
SIDE-POPULATION (SP) cells are defined by their ability to actively efflux Hoechst-33342 dye. This unique property is mediated through the action of cell surface-localized p-glycoprotein multidrug/ATP-binding cassette transporter protein (15, 34). As a result of this capacity, SP cells can be selectively isolated during flow cytometry of Hoechst-stained cells. SP cells, although rare, have been identified in various adult mammalian tissues and organs, including bone marrow, liver, skeletal muscle, brain, spleen, small intestine, peripheral blood, heart, kidney, and lung (4, 15, 21, 27). SP cells have also been identified in the embryonic heart (21).
Notably, SP cell populations are highly enriched for stem or progenitor cell activity. For example, a single adult transplanted bone marrow SP cell can completely reconstitute all blood lineages in radio-ablated mice (22). To date, the specific microanatomical sites in which tissue or marrow SP cells reside have not been identified. This relates, in part, to the rarity of these cells, and the fact that expression of the ABC transporter protein does not selectively mark SP cells. In this regard, we found that the ABC transporter protein is also expressed in distinct cell types that do not efflux Hoechst; in these cells, expression of this molecule is localized to an intracellular site (27).
Bone marrow SP cells uniformly express the pan-hematopoietic marker, CD45, whereas tissue SP cells can be subdivided into CD45-positive (+) and CD45-negative () subsets. The adult lung contains 0.050.07% SP cells, comprised of 6070% CD45+ and 3040% CD45 cells (27). CD45+ tissue SP cells possess hematopoietic progenitor and stem cell activity and are likely bone marrow derived. The activities and fate of tissue CD45 SP cells are less clear. Recent data indicate, however, that progenitors for distinct parenchymal cell types are contained within organ-specific CD45 SP cells (18). For example, testis-derived CD45 SP cells can serve as precursors for differentiated spermatocytes (18).
As an initial strategy to specifically identify stem and progenitor cells involved in lung organogenesis, we examined the developing mouse lung at several gestational ages for SP cells. We found SP cells at all developmental lung stages, with the highest frequency of cells present at embryonic day 17.5 (E17.5), representing 1% of all lung cells. (28). Most notably, the CD45+ embryonic lung SP cells contain functional hematopoietic stem cells, whereas the CD45 population is comprised of cells that can differentiate into complex mesenchymal and vascular structures (28).
To further clarify the role of these cells in lung development, we used oligonucleotide microarrays to evaluate their gene expression profile. In these studies, we identified 31 and 44 genes that were selectively or more highly expressed in E17.5 CD45 and CD45+ SP cells, respectively. Using these results as a guide, we attempted to immunolocalize cells in E17.5 lung tissue sections that displayed expression phenotypes consistent with SP cells. Together, these data indicate the presence of distinct subsets of E17.5 lung CD45 SP cells that reside at specific sites within developing blood vessels. Our expression data and in vitro culture data indicate that CD45+ SP cells represent a small subset of total CD45+ cells within the developing lung, likely functioning as precursors to local cells of myeloid lineage.
MATERIALS AND METHODS
Cell isolation.
Pregnant CD-1 mice (Charles River) were euthanized by isoflurane (Baxter, Deerfield, IL) anesthesia followed by cervical dislocation at E17.5. Embryos were dissected, and lungs were removed before mincing with a sharp razor blade and incubation with collagenase (0.1%; Roche Diagnostics, Indianapolis, IN), dispase (2.4 U/ml; Roche), and CaCl2 (2.5 mM) at 37°C for 1 h. Cells were filtered, collected, and suspended in PBS using methods as previously described (27). These protocols were approved by the Boston University Animal Use Committee and strictly followed National Institutes of Healths Policy on Animal Care and Use.
Fluorescence-activated cell sorting.
Hoechst-33342 (5 µg/ml; Sigma-Aldrich, St. Louis, MO) staining of proteolytically dispersed lung cells was carried out at 37°C for 90 min in DMEM (Invitrogen, Carlsbad, CA) containing 2% FCS and 10 mM HEPES (27). At the completion of staining, cells were placed immediately on ice. Antibody staining was subsequently performed in the dark at 4°C for 30 min with directly conjugated (phycoerythrin; PE) monoclonal rat anti-mouse-CD45 (BD Pharmingen, Lexington, KY). After the staining, cells were washed twice and resuspended in HBSS supplemented with 2% FCS. Dead cells were excluded from flow cytometry analysis on the basis of propidium iodide (PI) staining (2 µg/ml). In all studies, dead cells comprised <515% of total dispersed lung cells. Cells stained with an isotype control antibody were employed as negative controls and to establish gating parameters. All cells were simultaneously analyzed by high-speed flow cytometry for the following six parameters: Hoechst-33342 red, Hoechst-33342 blue, PI positivity, forward scatter, side scatter, and PE-CD45 reactivity.
For the first sort, cells were separated into SPs and main populations (MPs), collected, and subsequently resuspended in HBSS containing 2% FBS on ice. After the first isolation, the sorting procedure was repeated to increase the purity of SP and MP cells. At this time, cells were further separated and collected as four distinct cell populations: SP CD45, SP CD45+, MP CD45, and MP CD45+. Other antibodies used during flow cytometry include allophycocyanin (APC)-conjugated and FITC-conjugated monoclonal rat anti-mouse CD45, FITC-conjugated monoclonal rat anti-mouse intracellular adhesion molecule 2 (Icam2), and FITC-conjugated monoclonal rat anti-mouse CD86 (BD Pharmingen, San Diego, CA).
Sorting was performed on a triple laser instrument (MoFlo; Cytomation, Fort Collins, CO). An argon multiline UV (333363 nm) laser was used to excite Hoechst dye. Fluorescence emission was collected with a 405/30 band-pass filter (Hoechst blue) and a 660 ALP (Hoechst red). A second 488-nm or 640-nm argon laser was used to excite PE, FITC, PI, or APC, respectively. Data were analyzed using Summit V3.1 software.
Amplified microarray.
Oligonucleotide array hybridizations were carried out according to Affymetrix protocols (Gene-chip eukaryotic small sample target-labeling assay version II). Briefly, total RNA was extracted from 10,000 freshly isolated double-sorted E17.5 SP and MP cells using TRIzol (Invitrogen, Carlsbad, CA) supplemented with 10 µg/ml glycogen (Ambion, Austin, TX). Two cycles of RNA amplification were performed for each sample. In the first cycle, double-stranded cDNA was synthesized from total RNA using an oligo(dT)-T7 primer (Affymetrix, Santa Clara, CA) followed by an in vitro transcription reaction to generate complementary RNA (cRNA). cRNA was then used for a second cycle of amplification. During the second-round amplification, the double-stranded cDNA that was generated was converted to cRNA with a biotin-labeling kit [BoiArray Highyield RNA transcript-labeling kit (T7); Enzo Biochem, New York, NY]. Twenty-five to forty micrograms of biotinylated cRNA were obtained from two rounds of RNA amplification from the initial 10,000 cells (50 ng total RNA). The purified biotin-labeled cRNA was fragmented using a fragmentation buffer for 35 m at 95°C. Labeled fragmented cRNA (15 µg) was then hybridized to high-density oligonucleotide mouse arrays (MOE430A) representing 22,690 known genes (Affymetrix). After hybridization, the chip was washed, stained, scanned, and normalized per the manufacturers instructions (1, 19). Affymetrix Microarray Suite 5.0 software (Affymetrix) was employed for data scaling. This program uses an average single density of 500 to normalize results for interarray comparisons. The 3'/5' signal ratios ranged from 0.86 to 1.75. Arrays were performed in triplicate for each cell subset using RNA derived from three independent isolations. Each arrays analysis was performed with Affymetrix Microarray Suite 5.0 software. A detection P value less than 0.05 (P < 0.05) or greater than 0.05 (P > 0.05) was called present or absent, respectively. The percentage of genes with present call ranged from 37.6 to 47.9% for all samples. Intensity signals <100 were defined as background. All genes on the oligonucleotide arrays were input into the query gene files and downloaded as Excel files. All data from our microarray experiments have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus. The accession numbers for these samples are GSM37461, GSM37462, GSM37463, GSM37464, GSM37465, GSM37466, GSM37467, GSM37468, GSM37469, GSM37470, GSM37471, GSM37472, GSM37473, and GSM37474. The list of genes on these arrays is available at http://www.affymetrix.com/analysis/download_center.affx.
Semiquantitative real-time RT-PCR.
To verify results from Affymetrix chip hybridization, total RNA was extracted from freshly isolated double-sorted E17.5 lung SP and MP cells that were subdivided into CD45+ and CD45 groups; semiquantitative real-time RT-PCR (qRT-PCR) was performed on representative genes. RT reactions were carried out using random hexamers and RETROscript kit following the manufacturers manual (Ambion). qRT-PCR was performed with an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) using TaqMan Universal PCR Master Mix (ABI). The TaqMan probes and primers for von Willebrand factor (Vwf) (assay ID, Mm00550376_m1), protein C receptor (Procr) (assay ID, Mm00440992_m1), proteinase-3 (Prtn3) (assay ID, Mm00478323_m1), and myeloperoxidase (Mpo) (assay ID, Mm00447886_m1) were purchased as Assay-on-Demand gene expression products (Applied Biosystems). Expression levels were normalized to 18S ribosomal RNA (18S rRNA). The thermal cycler conditions were as follows: hold for 10 min at 95°C, followed by two-step PCR for 40 cycles of 95°C for 15 s followed by 60°C for 1 min. All samples were performed in triplicate. Amplification data were analyzed using ABI Prism 7000 SDS software (Applied Biosystems). Relative standard curves were used to determine target mRNA levels in different samples according to the manufacturers instructions (Applied Biosystems). For each sample, the relative amount of 18S rRNA was determined by analyzing a separate plate with 18S rRNA standards and primers from the 18S rRNA endogenous control kit (Applied Biosystems).
Immunostaining of E17.5 lung tissue sections and isolated cells.
E17.5 mouse lung tissue was obtained from normal pregnant mice as described above. Lungs were formalin fixed, embedded in paraffin, and sectioned (5 µm) using standard methods. Paraffin was removed from sections by immersing in solvent (Citrisolv; Fisher Scientific, Hanover Park, IL), before rehydration via exposure to graded alcohols and distilled water. Slides were then stored at 4°C before staining. For analysis of sorted cells, 20,000 SP CD45, SP CD45+, MP CD45, or MP CD45+ cells were collected onto positively charged slides by cytocentrifuging at 700 rpm for 5 min before air drying. The slides were then fixed with 4% paraformaldehyde (PFM)/PBS for 10 min, washed with 1x PBS, and stored at 4°C before staining.
For immunostaining, primary antibodies used included rabbit anti-mouse-Vwf (Chemicon, Temeula, CA), rabbit anti-mouse Mpo (Lab Vision, Fremont, CA), mouse monoclonal smooth muscle actin- (Sma)-Cy3-conjugate (Sigma-Aldrich), goat anti-mouse-PR3 (Prtn3) (Santa Cruz Biotechnology, Santa Cruz, CA), goat anti-mouse-Icam2 (R&D Systems, Minneapolis, MN), and monoclonal rat anti-mouse-CD45 (BD Pharmingen). For staining procedures that used monoclonal antibodies, isotype antibodies served as controls.
Secondary antibody detection reagents and kits used for these procedures include biotinylated secondary antibodies with the ABC reagent (Vector Laboratories, Burlingame, CA) or streptavidin-FITC (BD Pharmingen), donkey anti-rabbit-red 594 (Molecular Probes, Eugene, OR), and donkey anti-goat-red 594 (Molecular Probes). For detection of CD45 in tissue sections, the TSA-Biotin System (PerkinElmer, Boston, MA) was employed.
Cell culture.
SP CD45+ cells (10,000) were isolated and collected from E17.5 mouse lung and plated into a chamber slide with a defined hematopoietic differentiation culture medium. This medium contains StemSpan SFEM (Stemcell Technologies, Vancouver, BC, Canada), 1% penicillin-streptomycin (Invitrogen, Carlsbad, CA), mouse-GSF (100 ng/ml; Sigma-Aldrich), IL-4 (20 ng/ml; Stemcell Technologies), and TNF- (20 ng/ml; Stemcell Technologies). Cultures were maintained in humidified air with 5% CO2 at 37°C. Culture medium was changed every 48 h. After 15 days, cells were fixed with 4% PFM/PBS for 15 min, rinsed with PBS, and then stored at 4°C before immunostaining.
Statistics analysis.
Data are presented as means ± SD. Experimental data were compared using Students t-test. Results were considered statistically significant when P < 0.05.
RESULTS
Purification of SP and MP cells.
As a result of their ability to efflux Hoechst dye, SP cells can be identified on density dot plots as a distinct cell population during flow cytometry (15). Using this method, we sought to isolate pure populations of SP cells from the E17.5 mouse lung for gene profiling analysis. At this developmental time point, the percent of SP cells ranged from 0.8 to 1% (Supplemental Fig. S1A; available at the Physiological Genomics web site).1
To derive RNA for microarray studies, cells were sorted twice by high-speed cell sorting; after the first sort, the relative purity of SP cells was 90% (data not shown). The sorting procedure was then repeated, resulting in a relative purity of 98% (Supplemental Fig. S1B). During the second sort, SP and MP cells were further separated into CD45 and CD45+ subsets (Supplemental Fig. S1C). RNA from these four distinct cell populations was extracted, amplified, and then subjected to gene profiling using Affymetrix MOE430A oligonucleotide arrays (representing 22,690 genes). As expected, expression of CD45 mRNA was enriched in both CD45+ subsets, whereas expression of BCRP1/ABCG2 mRNA was enriched in both SP populations.
Molecular signatures of SP cell population in the E17.5 mouse lung.
The analysis of data from three arrays performed on lung cells derived from three distinct isolations demonstrated that the pairwise relationships between SP CD45 and MP CD45cells and between SP CD45+ and MP CD45+ cells, as represented by the squared Pearsons correlation coefficient (R2), were 0.906 and 0.893, respectively, indicating a significant positive linear correlation of gene expression intensity between these two populations. In contrast, the squared Pearsons correlation coefficients were <0.6 between CD45 and CD45+ cell populations.
Our gene expression data displayed a normal distribution for homogenous groups. Thus we employed a Students t-test for statistical analysis of gene expression data. For the purposes of identifying genes that characterize the SP CD45 and CD45+ cell subsets, expression was considered significantly increased if the fold change (FC) was >2.0 (P < 0.05; Students t-test) relative to the corresponding MP cell subset. Notably, a "twofold change" has been commonly used to identify genes that were considered to be significantly changed relative to a defined control (3, 26). Using this criterion, these studies identified 31 genes that were highly expressed in SP CD45 cells from the 22,690 known genes on the array (Table 1). All of these genes were expressed at significantly lower levels in the CD45+ SP and MP cells (Table 1). Genes associated with an endothelial phenotype were highly represented within the SP CD45 group (Vwf, Procr, Vegfc, Icam2, Eng, and Tie1; Table 1). There was also high expression of Procr and Eng, two genes that are notably highly expressed in adult hematopoietic stem cells (16, 8). Interestingly, nine genes highly expressed in adult cardiac SP cells (which are all CD45) were similarly highly expressed in SP CD45 cells in the E17.5 mouse lung (Vwf, Vegfc, Icam2, End1, Cldn5, Kdr, Eng, Tie1, and Emcn-pending; Table 1) (21). There were two transcriptional factors (Sox17 and Cebpd) that were highly expressed in lung SP CD45 cells.
|
|
|
|
|
|
Localization of lung SP CD45+ cells.
To resolve the location of SP CD45+ cells in tissue sections, we first identified the distribution of Mpo-expressing cells in the E17.5 lung. We found scattered, positively Mpo-stained cells distributed throughout the lung mesenchyme (Fig. 5A). Because Prtn3 mRNA is overexpressed in isolated SP CD45+ cells, we determined whether Mpo+ cells also express Prtn3. Immunofluorescence analysis indicated that Prtn3 and Mpo expression localized to the same cell population (Fig. 5, CE). This limited distribution of a CD45+ cell subset is in striking contrast to the distribution of all lung cells that express CD45 (Fig. 5B). To further clarify these findings, SP CD45+ cells were collected onto glass slides and stained for Mpo and Prtn3. Most, if not all, cells stained positively for both of these proteins (Fig. 5, FH). There were no Mpo+ cells found in MP CD45+ cells collected onto slides (data not shown). These findings indicate that SP CD45+ cells can be identified in E17.5 lung tissue by expression of Mpo or Prtn3.
|
DISCUSSION
This is the first detailed information regarding the phenotype and localization of SP cells in a developing organ. In this work, the Pearson correlation coefficients indicated that CD45 status was the best predictor of relatedness between the various cell populations examined, rather than the ability to efflux Hoechst dye. In view of the fact that SP populations are enriched for progenitor and stem cell activity (15, 21), these data are consistent with a progenitor relationship between CD45+ and CD 45 SP cell subsets and their corresponding MP cell subsets. To this end, our gene profiling and localization data suggest that subsets of CD45 SP cells may be involved in vascular development. Of note, E17.5 is a time of rapid expansion of the vascular elements in the developing lung (30). Indeed, we have recently shown that embryonic CD45 SP cells can form mature smooth muscle and primitive vascular networks in vitro (28). Our data further indicate that a CD45+ SP cell population may serve as progenitors for differentiated resident lung cells of myeloid lineage.
We acknowledge that one limitation of the gene profiling strategy used herein is that extremely rare Hoechst-effluxing cell types may have been missed and, as a result, not identified. Moreover, this method will not by itself resolve the relative cellular heterogeneity within each SP cell subset studied. Despite these limitations, we identified 31 genes that were highly expressed in the SP CD45 cell subset. Many genes associated with an endothelial phenotype were highly represented within these 31 genes (i.e., Vwf, Procr, Vegfc, Icam2, Eng, Tie1, Esam-pending, endomucin, C1qr1, Xlkd1, and Agtrl1). Interestingly, nine genes highly expressed in adult cardiac SP cells were similarly highly expressed in SP CD45 cells in the E17.5 mouse lung (Vwf, Vegfc, Icam2, End1, Cldn5, Kdr, Eng, Tie1, and Emcn-pending) (21).
Using the gene profiling data as a guide, we examined expression of several distinct genes that identified two distinct populations of CD45 SP cells residing within the E17.5 lung. These include 1) Vwf+/Sma+ cells that reside in the muscular layer of select large vessels and 2) Vwf+/Icam2+ cells that reside within the endothelial layer of select small vessels. These data do not rule out the possibility that non-dye-effluxing cells are contained within these distinct cell populations. Nevertheless, these data indicate that these populations are markedly enriched for SP cells.
The localization of both Vwf-expressing SP cell subsets, along with the gene profiling data, is consistent with a role for these cells in vessel development. Supporting this is recent work by Summer et al. (28) showing that SP CD45 cells form complex tube-like structures in vitro. As indicated, the Vwf+/Icam2+ SP cells were localized to the endothelial layer of small vessels. The concomitant expression of Icam2 in lung mesenchyme is consistent with previous work suggesting a relationship between endothelial progenitors and lung mesenchyme (10, 12). On the other hand, coexpression of Vwf and Sma within CD45 SP cells residing in the muscular layer of large vessels suggests an ancestral relationship between endothelial and vascular smooth muscle cells. This possibility is consistent with studies indicating the existence of primitive embryonic angioblasts that serve as precursors for various vascular cell types involved in blood vessel development (29). Furthermore, a recent study showed that vascular smooth muscle cells are derived from an FLK1+/TAL1 endothelial-like cell during vasculogenesis (12). Both of these SP CD 45 subsets displayed prominent and restricted expression of Vwf. In adult animals, Vwf, also known as factor VIII-related protein, is a large glycoprotein expressed by endothelial cells throughout the systemic and pulmonary vasculature that is directly involved in blood coagulation (13, 17, 24). The fact that the phenotype of Vwf-deficient mice is solely restricted to abnormalities in hemostasis suggests that expression of this molecule is not required for normal progenitor and stem cell activity in the lung (11). Icam2 is a cell surface adhesion molecule that is also expressed in adult endothelial cells (32). Icam2-deficient mice appear to have normal lung structure at birth but display aberrant regulation of eosinophil accumulation during allergen-induced inflammation (14).
We also identified 44 genes that were selectively and highly expressed in the SP CD45+ cell subset. Many of these 44 genes characterize cells that are differentiating along a myeloid pathway (Mpo, Prtn3, Il8rb, Clecsf9, CD14, Ccr2, Lgals3, and Samhd1). One possibility, therefore, is that SP CD45+ cells represent a heterogeneous group of blood-derived cells that are at various stages of differentiation but present in lung tissue. Accumulated work does indicate that the lung can be a source of signals (i.e., GM-CSF) that induce the differentiation of myeloid precursors (6). Along these lines, mice deficient in GM-CSF have a specific pulmonary phenotype resulting from incomplete alveolar macrophage differentiation. In our studies, we found that isolated SP CD45+ cells were capable of differentiating into cells that express dendritic cell marker (CD86). Taken together, we speculate that E17.5 SP CD45+ cells undergo local differentiation into distinct cells involved in the innate immunity of the postnatal lung.
Our gene profiling, real-time RT-PCR, and immunohistochemical studies indicate that all or most of the cells contained within the SP CD45+ subset express both Mpo and Prtn3. Despite these findings, it is still possible that this subset is composed of a heterogeneous population of cells. In adult cells, Mpo and Prtn3 expression characterizes myeloid progenitors (2, 9). Using expression of these two genes as a guide, we found that SP CD45+ cells represent a small subset of total CD45+ cells within the E17.5 lung and that they reside predominantly in lung mesenchyme and in perivascular sites.
The functional role of Mpo and Prtn3 in myeloid progenitor cells is not evident at this time. Mpo is a major primary granule protein that catalyzes the reaction of hydrogen peroxide with chloride, resulting in production of hypochlorous acid (31). Mpo-deficient mice develop normally but in adult life are susceptible to infection (7, 23). Prtn3 is a neutral serine protease; mice deficient in this molecule have not yet been generated. Both Mpo and Prtn3 are also expressed in mature neutrophils but not in other mature myeloid cells (20, 25).
In conclusion, in this paper we describe the use of gene profiling to characterize and localize subsets of cells derived from the E17.5 lung that possess a capacity to efflux Hoechst dye. A salient feature of isolated cell populations that efflux Hoechst is the fact that they are highly enriched for stem and progenitor cell activity (15, 18, 28, 5). On the basis of this, we suggest that we have identified gene expression profiles that can be used to identify cell types that may serve as progenitors for specific vascular and immune cells in the developing lung. The ability to isolate Hoechst-negative cell types provides a means for further study of these cells in lung development. Finally, we speculate that the genes identified as overrepresented within SP subsets could be used as tools to elucidate pathways and cells controlling lung development.
GRANTS
This work was supported by National Heart, Lung, and Blood Institute Grants PO1-HL-47049-11, RO1-HL-69148-02, and R21-HL-72205.
ACKNOWLEDGMENTS
We thank Dr. Alan Ho for technical assistance during flow cytometry and the staff of the Boston University Medical School microarray core facility.
FOOTNOTES
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: A. Fine, Pulmonary Center, Boston Univ. School of Medicine, 80 East Concord St., R-304, Boston, MA 02118 (e-mail: afine{at}lung.bumc.bu.edu).
10.1152/physiolgenomics.00059.2005
1 The Supplemental Material for this article (Supplemental Figs. S1 and S2) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00059.2005/DC1.
REFERENCES