Although expression of the α5 integrin subunit has been reported in the human corneal epithelium,
16 51 52 53 such primary cultured cells are inappropriate as a model for conducting detailed gene promoter studies, primarily because of the difficulties encountered in both culturing and transfecting these cells in vitro.
52 54 In contrast, RCECs are easy to maintain in culture without the need for feeder cells. Most of all, they can be transfected with high efficiency by a polycationic detergent used as a transfection reagent (Lipofectamine; Invitrogen-Gibco)-mediated gene transfer
31 55 56 which makes them an ideal cellular model to conduct gene promoter studies. As the proliferative state of a particular cell type changes with cell density, probably as a result of altered gene expression, we exploited semiquantitative RT-PCR analyses to investigate whether transcription of the α5 integrin subunit gene is under the influence of the cell density reached by primary cultured RCECs
(Fig. 1) . PCR amplifications were performed on reverse-transcribed total RNA obtained from both subconfluent and postconfluent RCECs. Coamplification of the 18S ribosomal RNA was also performed for normalization purposes. The specific α5 PCR product, which appeared as a single DNA fragment of the expected size (171 bp) under all culture conditions (subconfluent and postconfluent RCECs) was detectable on the gel after 30 cycles of amplification and remained linear up to 34 cycles
(Fig. 1A) . However, a significant reduction was observed for the corresponding PCR amplification cycles in postconfluent cells. Normalization of the α5 signal to that of the 18S rRNA, which appear as a single band of the appropriate size (489 bp) in subconfluent RCECs but as a doublet in postconfluent cells (probably as a result of alternative splicing events occurring at high cell density), provided evidence that the amount of α5 transcript is on average 4.6 ± 0.3 times lower in postconfluent RCECs than in subconfluent cells (as revealed through densitometric analyses of the amplified PCR products). As a control, we also examined whether the expression of a class II gene, the transcription of which is not regulated by Sp1, remains unaltered by the state of cell density. For this purpose, we selected the gene encoding the human α4 integrin subunit, the expression of which has been found only in a restricted number of tissues, including the cornea,
16 both lymphoid and myeloid cells,
57 and differentiating skeletal muscle.
58 We have reported the binding of transcription factors to the α4 basal promoter,
56 of which none belong to the Sp1 family. Besides, no GC-rich Sp1 target sites are found in the sequence of the α4 promoter.
49 The selection of this gene also relies on the fact that cotransfection of the α4 promoter along with either the Sp1 or Sp3 expression plasmids (pPACSp1 and pPACSp3) in Sp1-deficient
Drosophila Schneider cells had no influence at all on the transcriptional activity directed by the α4 promoter (data not shown). As revealed on
Figure 1B (left), PCR amplification yielded the expected 772-bp α4 fragment. However, no alteration was observed in the amount of α4 amplification product between reverse-transcribed mRNAs prepared from postconfluent and subconfluent cells on normalization to the 28S ribosomal RNA. We next isolated total RNA from either postconfluent or subconfluent RCECs and examined the α4 transcript by Northern blot analysis. After only a few hours of exposure, a single mRNA species of approximately 6.5 kb corresponding to the previously reported major α4 transcript
49 was observed just above the 28S rRNA (
Fig. 1B , right). As expected, expression of the α4 transcript was not altered when RNA was isolated from postconfluent RCECs. We therefore conclude that the relative concentration of the α5 mRNA, but not that of the α4 transcript, is considerably reduced when RCECs reach confluence.