Introduction

Top Abstract Introduction Materials and methods Results Discussion References

Fibroblasts are the major cell type responsible for the synthesis and turnover of extracellular matrix components in most connective tissues, and both the number and activity of fibroblasts increase in pulmonary fibrosis. There is significant evidence that fibroblasts are heterogeneous with respect to functional properties, and that certain subpopulations of these cells may be clonally selected and expanded in diseased tissues (1). Subpopulations have been separated from cultures of lung and gingival fibroblasts by utilizing differences in binding to Thy-1 antigen, types I and III collagens, and complement component C1q, and these subpopulations have been shown to differ in morphology, proliferation rate, protein synthesis, and response to inflammatory mediators (5). These differences are maintained in subculture, independent of culture conditions and passage number. However, it has not been possible to utilize these differences to identify and distinguish fibroblast subpopulations in vivo.

We have previously separated two human lung fibroblast subpopulations on the basis of differences in their binding to C1q, using cell sorting with an anti-C1q antibody (10). We demonstrated that these subpopulations differ in growth rate, collagen synthesis, and degree of response to transforming growth factor- (TGF-) and interferon- (IFN-) (10). One of these subpopulations was obtained by growing tissue explants in fresh human serum, and expresses the receptor for the collagen domain of the C1q molecule (cC1q-R). The other subpopulation, obtained by growth in heat-inactivated plasma-derived serum, expresses a 51-kD protein that binds to the C1q globular domain (11). The former fibroblast type could be distinguished from other fibroblasts by immunostaining with an anti-cC1q-R antibody, because only these cells express cC1q-R (11). However, in human lung tissue specimens, C1q-R are expressed poorly by fibroblasts, and other cell types express them more strongly, thus preventing the use of these receptors for distinguishing fibroblast types in tissues (Raghu and Narayanan, unpublished data). Here we report the isolation of a differentially displayed gene that is expressed strongly by one of two human lung fibroblast subpopulations. Expression of this gene was not detected in smooth-muscle cells, endothelial cells, or epithelial cells, and it therefore offers promise as a marker for fibroblasts and fibroblast subtypes in cell culture and tissues.

	Materials and Methods

Top Abstract Introduction Materials and methods Results Discussion References

Cells

Human lung fibroblasts used for our studies were isolated from explants of pulmonary lung parenchyma obtained from fetal lung specimens, with approval of the University of Washington Human Subjects Review Committee. The tissue was cultured in medium containing either fresh human serum or complement-inactivated, plasma-derived serum and subjected to fluorescence-activated cell sorting (FACS), as described previously (10, 12). The resulting cells were designated as HH and NL cells (for human serum high and null serum low), respectively. Cultures of the cells were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% bovine serum.

RNA Extraction and Northern Blot Analysis

Total RNA from cells and tissues was extracted with TRIzol reagent (GIBCO, Gaithersburg, MD). For Northern blot analysis, 10 µg RNA was electrophoresed through formaldehyde-agarose gels, transferred to Hybond nylon membranes (Amersham, Arlington Heights, IL), and crosslinked by ultraviolet irradiation. Hybridization was done overnight at 42°C in 50% formamide, according to standard protocols, with ³²P-labeled probes (10⁶ cpm/ml) (13). The blots were stripped and reprobed with a ³²P-labeled oligonucleotide probe for 18S RNA. The levels of messenger RNA (mRNA) were compared densitometrically by normalizing for the amounts of 18S RNA.

Differential Display

Deoxyribonuclease (DNase)-treated RNA was prepared, and complementary DNA (cDNA) and ³³P-labeled polymerase chain reaction (PCR) products were generated with a Delta RNA fingerprinting kit (Clontech Laboratories, Palo Alto, CA). Differential display was performed as described by the manufacturer, using 22 different primer pairs (25- or 26-mers). PCR fragments were generated and separated in 6% denaturing polyacrylamide sequencing gels (14, 15). In each experiment, PCR reactions were conducted in triplicate and bands of interest were excised, reamplified, and subcloned into the plasmid vector PCR-TRAP (GenHunter, Nashville, TN), using standard molecular cloning techniques.

DNA Sequencing

DNA sequencing was done on an ABI automated system (Applied Biosystems, Foster City, CA), using the ABI Prism DNA sequencing kit with two different sets of primers. Multiple sequences were obtained from each clone in both orientations.

	Results

Top Abstract Introduction Materials and methods Results Discussion References

Analysis of gene expression in the HH and NL fibroblast subpopulations, using the 22 different primer sets described previously, yielded seven PCR products that were differentially displayed; six of these seven products were identified as coming from HH fibroblasts and one from NL fibroblasts (data not shown). The most prominent of the six HH fibroblast display products, obtained with the primers 5'-ATTAACCCTCACTAAATGCTGGGGA-3' and 5'-ATTAACCCTCACTAAATCGGGTCATAT-3', was cloned and sequenced. The nucleotide sequence of this product, referred to as LR8, is shown in Figure 1.

View larger version (48K): [in this window] [in a new window]   Figure 1.   Nucleotide sequence of LR8 cDNA. The sequences of the primers are shown in lower case and are underlined.

To confirm the differential expression of LR8 in HH lung fibroblasts, we performed Northern blot analysis. Using the LR8 clone as a probe, we detected a 1.6-kb mRNA whose expression was 7.7 times greater in HH cells than in NL cells (Figure 2). We then studied the expression of LR8 mRNA in vivo and examined the total RNA obtained from normal and fibrotic human lung specimens. Both tissues expressed the 1.6-kb mRNA, but its level in fibrotic lung specimens was 2.7 times greater than that in normal lung (Figure 3).

View larger version (61K): [in this window] [in a new window]   Figure 2.   Northern blot analysis (in duplicate) of total RNA obtained from HH (a) and NL (b) human lung fibroblasts. Top panels: LR8 probe; bottom panels: probe for 18S RNA. By densitometry, HH fibroblasts contained 7.7-fold more mRNA than did NL cells.

View larger version (59K): [in this window] [in a new window]   Figure 3.   Northern blot analysis of total RNA obtained from normal (a) and fibrotic (b) human lungs. Top panels: LR8 probe; bottom panels: probe for 18S RNA.

Lung tissue contains various cell types, including fibroblasts, endothelial cells, epithelial cells, smooth-muscle cells, and alveolar macrophages (AM). To determine whether the LR8 mRNA that we detected in lung tissues was derived from cell types other than fibroblasts, we performed Northern blot analysis on total RNA obtained from cultured human aortic smooth-muscle cells, human umbilical vein endothelial cells, and neonatal foreskin epithelial cells. We detected no significant hybridization signals in these preparations (Figures 4a to 4c). However, Northern blot analysis of RNA from dermal fibroblasts demonstrated the presence of LR8 mRNA (Figure 4d).

View larger version (48K): [in this window] [in a new window]   Figure 4.   Northern blot analysis in duplicate of total RNA obtained from human umbilical vein endothelial cells (a), human aortic smooth-muscle cells (b), dermal epithelial cells (c), and dermal fibroblasts (d). Top panels: LR8 probe; bottom panels: probe for 18S RNA.

A computer search of gene and protein sequences available in the GenBank and Swiss protein database (SWISSPROT) databases, done with the nucleotide sequence of LR8 as a probe, did not identify any sequence with a significant similarity to LR8. However, a second search, of the nonredundant database of expressed sequence tag (EST) sequences maintained by the National Center for Biotechnology (NCBI), revealed five human EST sequences (GenBank accession numbers AA573785, AA309046, AA372940, H711284, and H39231) that were identical or nearly identical to the LR8 sequence. In addition, six mouse EST sequences (GenBank accession numbers AA259324, AA403341, W54253, AA509380, AA116218, and AA4454394) and one pig EST sequence (GenBank accession number F22962) that were closely similar to LR8 were also identified. The five human EST sequences and the LR8 sequence were assembled into a continuous contig of 1,208 bases, using the Sequencer (LifeCodes, Ann Arbor, MI) sequence assembly program, with plurality used to decide base calls. The six mouse EST sequences were similarly assembled into a 958-base continuous contig. Both contigs contained a large open reading frame. The DNA and encoded amino acid sequences of the human and mouse cDNA contigs are shown in Figure 5. The open reading frame and the human sequence predicted an amino acid sequence of 270 residues, whereas the mouse sequence was predicted to be 263 residues. Three gaps of 18 bp, 3 bp, and 6 bp were introduced into the coding regions to maintain optimal nucleotide and amino acid sequence alignment. The nucleotide sequences within the predicted coding regions of the human and mouse cDNAs were 68% identical (534 of 783 nucleotides, with three gaps). The encoded amino acid sequences were identical at 54% of the positions (141 of 261 amino acids, with three gaps). With allowance of conservative substitutions, most consisting of hydrophobic amino acid substitutions, the similarity of the two sequences increased to 62% (161 of 261 amino acids, with three gaps). The single pig EST encoded the C-terminal sequence and had 61% identity (37 of 61 amino acids) and 46% similarity (27 of 59 amino acids) to the C-terminal region of the human and mouse sequences, respectively (data not shown). The human and mouse sequences were predicted to encode proteins of 29,138 and 28,342 D, with eight and 10 cysteine residues, respectively. Neither sequence contained a potential N-linked glycosylation site (NXS or NXT). The results of hydropathy analyses were very similar for both the mouse and human sequences, and predicted four strong hydrophobic domains in each, with an additional hydrophobic domain at the N-terminus that was reminiscent of a signal sequence (Figure 6). Analysis of the predicted amino acid compositions showed that 30% of the amino acids were hydrophobic. These results strongly suggested that both the putative human and mouse proteins are anchored in a membrane, and may traverse the membrane in multiple regions.

View larger version (53K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 5.   Comparison of the nucleotide and encoded amino acid sequences of the composite LR8 human cDNA and its murine homologue.

View larger version (18K): [in this window] [in a new window]   Figure 6.   Hydropathy analysis of LR8 putative protein, done with the Kyte and Doolittle algorithm as implemented in Gene Pro software (Riverside Scientific, Bainbridge Island, WA).

To determine whether the assembled sequence shown in Figure 5 represented a true cDNA, we performed a PCR reaction, using a sense primer upstream of LR8 that represented the predicted N-terminal end, and an antisense primer within LR8 at the stop codon (5'-TCCTAGGATCCAGGCATGCCCAGCCCACC-3', bp 209-229; and 5'-TTACCAAGCTTCACAGGACAATGGCAGTGGAG-3', bp 955-976, respectively; the underlined sequences are additional bases for restriction sites). Human lung fibroblast cDNA was used as the template. As expected, a 789-bp-long product, which encoded the protein minus the signal sequence, was obtained (Figure 7), and its sequence matched the predicted nucleotide sequence of LR8 (data not shown).

View larger version (70K): [in this window] [in a new window]   Figure 7.   Agarose gel (1%) electrophoresis of PCR product obtained with sense primer designed from the assembled cDNA sequence (bp 209-229; see Figure 5) and antisense primer within LR8 and ending with the stop codon (bp 955 to 976; see Figure 5). Cycles were 1 min at 94°C, 33 cycles of 30 s each at 94°C, 3 min at 68°C, and 3 min at 94°C. Lane a: size marker, 123-bp ladder; lane b: PCR amplified product. The arrowhead in lane b indicates where LR8 will migrate.

	Discussion

Top Abstract Introduction Materials and methods Results Discussion References

So far, fibroblast subpopulations have been differentiated in vitro, through such functional properties as proliferation rate and collagen synthesis (1). Differences also exist among cultured fibroblasts in the expression of cell surface proteins such as Thy-1, C1q receptors, and collagens (5- 10). However, whether these differences are maintained in vivo, and how these proteins are regulated is not clear. For example, the receptor for the collagen domain of C1q is expressed differentially by one subpopulation of lung fibroblasts in culture (11); in tissues, however, this receptor is expressed poorly by fibroblasts, whereas other cell types, including endothelial cells, smooth-muscle cells, type II epithelial cells, and AM express this protein more strongly (Raghu and associates, unpublished data). The LR8 gene that we identified by differential display in the present study appears to be a more specific marker for fibroblasts. In addition, its mRNA is differentially expressed in at least two different human lung fibroblast subpopulations. Our results show that LR8 mRNA is also expressed in normal and fibrotic human lung tissues in vivo. It appears that LR8 is expressed at higher levels in fibrotic lungs than in normal tissue (Figure 3). However, these results are preliminary and need additional supporting evidence. Nevertheless, the lack of expression of LR8 in cultured endothelial, epithelial, or smooth-muscle cells suggests that the mRNA detected in lung tissue is most likely derived from fibroblasts. In situ hybridization studies are currently being done to verify this hypothesis. Its expression by dermal fibroblasts indicates that LR8 is also expressed by fibroblasts of other tissues.

The sequence of LR8 is not similar to that of any known protein or protein motif, and therefore no function for this protein has yet been identified. That the DNA for LR8 represents a true cDNA was confirmed by the ability to obtain a product with expected length and sequence through PCR done with primers upstream to and within LR8 (Figure 7). However, analysis of the putative coding sequence suggests that its DNA may encode an integral membrane protein with at least eight cysteines. We are currently establishing systems for the expression of recombinant LR8 protein to develop serologic reagents for further characterization of the LR8 gene product, and for use in functional studies of cells transfected with the LR8 gene.

Recently, a murine fibroblast-specific protein (Fsp1) that belongs to the calmodulin-S100 troponin C superfamily of intracellular calcium-binding proteins was cloned, although its sequence has not been reported (16). This protein is involved in epithelial mesenchymal transformations (17), and its expression is increased in fibrosing kidneys (16). Although we could not compare the DNA sequences of Fsp1 and LR8, the difference in size of their mRNAs (1.6 kb for LR8 versus 0.65 kb for Hsp1) suggests that the genes encoding them are not the same.

We have also identified the murine homologue of LR8, which will allow the study of LR8 in animal models of pulmonary fibrosis. These studies, however, will require additional information about the expression of LR8 in AM and endothelial, epithelial, and smooth-muscle cells derived from the lung, and about the way in which LR8 expression is regulated by molecules present in the local environment of normal and pathologic lungs. The results of the present study suggest that LR8 could be a useful marker for identifying populations of fibroblasts and fibroblast subtypes in vitro and in vivo, in lungs as well as in other tissues.