A number of the nasal epithelial ESTs show very high homology with sequences from the human mitochondrial genome (Table 1), implying that these sequences are derived from mitochondrial DNA. It is surprising, however, that most of these ESTs are not only homologous with the mitochondrial genome but also with chromosomal DNA, and that the same region and extent of homology with mitochondrial and genomic DNA is observed in many cases. Moreover, an individual EST can have very high homology with a HTGS genomic clone from more than one chromosome (see 34-18, 36-62, 5R22 and 36-100 in Table 1). ESTs 34-18 and 9694, for example, have homology with the mitochondrial urf4 gene and with the same genomic clone in chromosome 5. Multiple alignment of ESTs 34-18 and 9694 suggests that these are different sequences and from different regions of the genomic clone.

It is possible that families of very similar genes (or of genes with similar domains) are expressed from mitochondrial and nuclear genomic DNA or that a massive amount of domain fusion has occurred between mitochondrial and nuclear genes. Only one chimeric urf4 cDNA (fused mitochondrial and nuclear DNA) has previously been recovered experimentally from a viral integration, over a decade ago [5]. On the basis of the high homology of each EST with both mitochondrial and nuclear DNA, we suggest that nuclear and mitochondrial domain fusion is not an isolated phenomenon, but is rather common. This remains to be rigorously investigated.

A second group of ESTs are completely homologous with known genes in the human genome (Table 2). We assume that these are either the same gene as their genomic counterpart or a gene containing the same domain. From this evidence it is interesting to note that a number of important genes whose functions are known are also expressed in ciliated epithelial cells, although the significance of this expression is unknown. Although further rigorous experiments are needed to characterize these genes in ciliated epithelium, the probable functions of some important genes are discussed below.

Cytohesin (EST 14-49) is involved in signal transduction pathways and regulates cell adhesion [6]. Expression of this gene may play an important role in the adhesion of epithelial cells during the expansion of the cell layer.

Cyclophilin C (EST 24-51). In response to endotoxin, mice deficient in cyclophilin-associated protein overproduce interleukin-12 and interferon-gamma systemically and tumor necrosis factor-alpha locally. These are proinflammatory molecules that also promote helper T-cell responses [7]. The role of this gene in ciliated epithelial cells remains elusive, however, and an important concern for further investigation.

Epidermal growth factor receptor (EGFR) kinase (EST 6034). EGFR is an important ligand-binding protein and its level is elevated in many tumors [8]. The EGFR gene is a potential oncogene and expression of EGFR kinase may be required during formation of epidermis by epithelial cells.

The Drosophila Staufen genes (EST 6092) are RNA-binding proteins important for RNA transport and localization in the oocyte and neurons in Drosophila. The motor protein dynein (ddlc1) in conjunction with Staufen and Swallow acts as an adaptor for transporting bicoid RNA along microtubules to their minus ends at the anterior pole of the oocyte [9,10]. As dynein genes are highly expressed in ciliated epithelial cells [11] a potential interaction of dynein with the human Staufen homolog could be deduced.

Decay-accelerating factor (DAF, CD55) (EST 9661) protects host cells from the activation of autologous complement on their surfaces. It functions to disable the C3 convertases, the central amplification enzymes of the complement cascade [12]. Expression of this gene in nasal epithelial cells could be explained by the need for protection against antigen-induced complement activation.

A number of ESTs (Table 3) are partially homologous (that is in part of the EST sequence) to domains of known genes and could be of interest. However, further investigation of complete cDNAs and their functions may reveal the true identity of these genes. A few ESTs which are partly homologous with the domains of important genes are discussed below.

EST 24-17 (H3 pseudogene). A stretch of 21 base pairs (bp) of this EST has homology with the pseudogene of histone H3 but not with the normal histone H3 gene. Histone H3 is an important housekeeping protein involved in chromatin packing [13]. It is possible that another H3 RNA is transcribed, which may be different from both the normal H3 gene and the pseudogene.

EST 36-5 (retinoic acid responder). As retinoic acid plays an important role in in vitro ciliogenesis, expression of the retinoic acid responder domain is not unexpected in ciliated epithelial cells, where it may modulate a number of dynein heavy-chain genes during ciliogenesis [14].

EST 9010 (Attractin precursor). The protein attractin is secreted by activated T cells and has also been detected in the central nervous system [15]. It is suggested to be involved in immunity, obesity and pigmentation [16]. The mouse mahogany mutation is caused by a mutation in the attractin gene [17]. A portion (192 bp out of 194 bp) of EST 9010 has a very high similarity or identity with the attractin gene, suggesting that this EST is derived from a gene containing an attractin precursor domain. Expression of such a gene in ciliated epithelial cells is of unknown significance.

EST 906 (MDM2). MDM2 is an oncogene and the MDM2 oncoprotein binds to the p53 protein, inhibiting p53's function as a transcription factor and inducing its degradation. An MDM2-p53 autoregulatory feedback loop regulates the function of the p53 tumor suppressor gene [18]. The significance of the expression of MDM2 or of a gene carrying an MDM2 domain in ciliated epithelial cells remains to be investigated.

EST 907 (Mkp5 dual-specificity phosphatase). Mkp5 is a member of the mitogen-activated kinase (MAP kinase) family (10 genes in total) and has an important role in apoptosis, tumor progression and immune responses [19]. MAP kinases contain a docking motif that increases the efficiency of the reaction [20]. Part of EST 907 (52 bp out of 391 bp) has high similarity with the docking motif, possibly implying that another member of the MAP family could exist.

EST 9667 (transcription regulator protein; HCNGP). This transcriptional regulator has been isolated from adrenal gland (C Jiang, J Shi, C Huang, S Ren, Y Li, J Zhou, Y Yu, S Xu, Y Wang, G Fu, et al., unpublished data; GenBank accession number AF119664). A part of EST 9667 (86 bp out of 87 bp) is highly similar to the HCNGP; expression of this gene in ciliated epithelial cells is of unknown significance.

EST C2s-3 (translocase). Protein translocation across the cytoplasmic membrane has not been studied extensively in mammalian cells. It is also not known how many genes are involved in this process [21]. It is interesting that this EST shows homology (34 bp out of 38 bp) with a translocase gene and may carry out a translocase-like function.

EST 9640 (Sid3177). Sid3177 is a unique gene isolated from the mouse and assumed to be a part of the inactive progesterone receptor complex (N Seki, A Hattori, A Hayashi, S Kozuma, M Muramatsu, T Saito, unpublished data; GenBank accession number AB024935). The high homology of EST 9640 with a domain of Sid3177 suggests that a similar gene(s) is present in humans and is expressed in ciliated epithelial cells.

ESTs 3R1-32, 5r29, 5R5 (genomic scaffold-binding protein). Parts of these ESTs have high homology with the Drosophila scaffold-binding protein [22]. Scaffold-binding proteins are important in the replication and segregation of chromosomes. However, human counterparts of the complete genes have yet to be isolated.

EST 9014 (asthmatic clone). Part of this EST has homology with a cDNA isolated from asthma patients and that appears to be expressed differentially in asthmatics (IC Kilty, PJ Vickers, unpublished data; GenBank accession number AF095856). Expression of such a gene in the upper airway epithelium is of important in the context of identifying genes responsible for asthma.

Twenty ESTs (33%) have no significant homology (Table 4) to sequences in normal databases. The identities of these genes are not known even after 2 million sequences have been accumulated in the dbEST database. BLAST searches against the HTGS database, however, reveals that most of them are highly homologous with known genomic clones. These ESTs have been mapped electronically and their chromosomal locations derived. In recent years, there has been an exponential rise in the number of sequences available in the public databases. Despite this, a high percentage of partial sequences of cDNAs (ESTs) submitted to the databases remain unrecognized (anonymous ESTs). This lack of similarity could be explained [23] in several ways. One explanation is that a different part of the transcript is present in GenBank; second, the transcript represents a novel gene not yet isolated; third, there is alternative splicing of the same gene in different species; fourth, inaccurate sequence data; and/or fifth, the sequence of the transcript has diverged to an extent that it is not recognized as an ortholog.

Human nasal epithelial cells were enzymatically dissociated from biopsies using a protease type XIV (Pronase) digestion overnight at 4°C. Pronase was inactivated by adding fetal calf serum (FCS) or NU-Serum (10% final concentration) and the cell suspension was washed three times in culture medium (Ham's F12- DME 1/1 with cholera toxin (10 ng/ml), streptomycin (50 μg/ml), penicillin (50 IU/ml) and 2% Ultroser G). The cell suspension was preplated on plastic for 1 h at 37°C to remove most of the contaminating fibroblasts [24,25]. Cells were plated in T75 tissue culture flasks on 0.2% collagen gel for monolayer culture and kept at 37°C at 5% CO2 atmosphere in a biological oxygen demand (BOD) incubator. Culture medium was changed three times a week. After three weeks of exponential growth the cultures reach confluence.

Cells were then released from the collagen gel using 200 IU/ml collagenase type IV. Cell clusters, aggregates and cell sheets were washed three times in culture medium to eliminate collagenase and then placed in culture medium at 37°C on a gyratory shaker at 80 rpm to avoid attachment of the cells to the culture flask. During the first week, the medium (the same as used earlier) was changed every day. On the second day, the 2% Ultroser G was replaced by 10% NU-serum [24,25]. After 1 week stable aggregates, spheroids and vesicles were formed and showed no tendency to adhere to the culture flask; the culture was kept stationary for another few weeks. Generally, cilia appear in 2 weeks and cells were used to isolate total cellular RNA after the third or fourth week.

We used degenerate primers for the reverse transcription and PCR amplification (RT-PCR). Primers are initially designed [11] to clone the dynein heavy-chain genes from ciliated epithelial cells and are taken from the P-loop region (ATP hydrolysis region) of dynein heavy-chain genes. Reverse transcription was done with primers 4, 6 and 3R. PCR amplifications were done in round-robin fashion with primers 1 and 4, 2 and 4, 3 and 4, 1 and 6, 2 and 6, 3 and 6, 1 and 3R, 2 and 3R, 3 and 3R, and so on. Primers [11] used in these studies were 1, 5'-TAY GGN TTY GAR TAY YTN GG-3'; 2, 5'-GTN CRR ACN CCN YTN ACN GA-3'; 3, 5'-ACN GGN AAR ACN GAR ACN AC-3'; 4, 5'-CCN GGR TTC ATN GTD ATR AA-3'; 6, 5'-CKN ARN CCR AAR TCR TAR TG-3', 3R, 5'-GTN GTY TCN GTY TTN CCN GT-3'.

Total RNA was isolated from the cultured cells by the method described in [26,27]. RT-PCR was carried out with GeneAmp RT-PCR Kit (Perkin Elmer). Each sample of RNA was routinely treated with DNase I for 6 h at 37°C to remove any genomic DNA contamination. Two micrograms of RNA were reverse transcribed by the downstream primer in 20 μl at 42°C and PCR amplified withthe addition of upstream primer in a 100 μl volume. In all cases, PCR conditions were: for denaturation, 94°C, 4 min; for amplification, 94°C for 1 min; 50°C for 1 min; 72°C for 1 min for 40 cycles; and for elongation, 72°C for 10 min. RNA without reverse transcriptase and water without RNA (plus reverse transcriptase) did not yield any product in any of the PCR reactions.

PCR products were cloned in PCR2.1 vector of the TA-cloning Kit (InVitrogen) and were subjected to blue/white selection. White colonies were checked by PCR for the presence of insert with the vector-specific primers (M13 forward and reverse). Approximately 400 clones were sequenced on an ABI377 Automated Fluorescence Sequencer (Perkin Elmer). Sequences were screened with BLAST for the identity of these clones. Along with the cloning of nine dynein heavy-chain genes [11], a number of non-dynein cDNAs were recovered which were studied in detail. From 400 clones, 63 were selected as unique by the following procedures. The dynein heavy-chain genes (82 clones) were ignored; only one sequence was selected when two or more clones containing similar sequences were obtained; very small sized clones (below 70 bases) were ignored. Only in one case was a 56 bp clone (5R16) selected, as this sequence was not obtained repeatedly.

ESTs were BLASTed against the normal database (in NCBI BLAST page normal database (GenBank) designated as 'nr') and highest similarities with the known genes were taken into account. In cases of homologies with more than one gene, only the gene with the highest homology (number of base pairs, highest similarity and identity) was taken as the homologous gene and functional characterization has been done on the basis of the function of the known gene.

EST sequences were also BLASTed against the HTGS databases and the accession number and chromosome number of the highest-similarity clone were noted. Each clone was searched by accession number in Locus Link [28] in 'GENEBANK MAP' and the clone carrying flanking markers in a particular chromosome was assigned in 'STS MAP'. The distance from P-tel of the chromosome was taken as the map position of the EST.

All ESTs are deposited in databases under the dbEST accession numbers 8451921 to 8451980 and 8452140 to 8452142 and the Genbank accession numbers BG673720 to BG673779 and BG687691 to BG687693.