In this work, we studied the 5′-UTR sequence of human p27 mRNA and further tested the putative IRES activity using promoterless dicistronic and monocistronic vectors and direct assay of dicistronic mRNAs. Our results indicated that the 5′-UTR sequence of p27 contains promoters that can drive production of abundant monocistronic transcripts with shorter 5′-UTRs. These findings raise questions on the IRES activity of the 5′-UTR sequence of p27 reported previously using the traditional dicistronic DNA assay.
Use of IRES for translating the untranslatable cellular mRNAs with long 5′-UTRs is an interesting concept (29
). The use of such a mechanism for translating polycistronic viral RNAs has been convincingly demonstrated previously (30
). However, whether cellular mRNAs also use this mechanism is currently under debate (15
). The central problem associated with the majority of studies claiming cellular IRES is the use of traditional dicistronic DNA test which does not directly test the translation. Potential promoter activities and alternative splicing could both contribute to the observed ‘IRES’ activity. These potential problems have recently been demonstrated on several previously believed-to-be cellular IRES elements. These IRES activities were shown to be due to cryptic promoters present in the 5′-UTRs (19
) or due to differential splicing (35
) which would create smaller monocistronic transcripts from the dicistronic DNAs for potential cap-dependent translation initiation of the second cistron. In one of these studies, the prevailing cellular IRES sequence of eIF4G (36
) was shown to be an C/E BP beta transcription factor binding site (19
). In another study, the strong IRES activity in the 5′-UTR of XIAP mRNAs was found to be mostly due to alternative splicing (35
). Furthermore, a promoter activity was also found in the 5′-UTR sequence of hepatitis C virus, which has previously been shown to have IRES activities (38
). While these studies do not necessarily exclude in general the existence of IRES in cellular mRNAs per se
, they necessitate more stringent studies such as tests using promoterless dicistronic DNA transfection, dicistronic RNA transfection and RT–PCR for safeguarding any claims of cellular IRES.
Previously, the 5′-UTR sequences of both human and murine p27 were reported to contain IRES activities using traditional dicistronic DNA assays (8
). While the IRES in the mouse p27 was thought to be located in the 217 bases proximal to the AUG start codon, the full IRES activity in human p27 was thought to be located within the 356 bases upstream of the start codon. However, in the present study, we showed that these previously reported IRES activities are likely due to the promoters present in the 5′-UTR sequence of p27 (see below). In the case of human p27, the deletion analysis showed that the 184 bases upstream of the AUG start codon contains about half of the putative IRES activity with a 15-fold stimulation of Fluc/Rluc ratio by the 5′-UTR of p27 compared with the vector (9
). This extent of stimulation is similar to the promoter activity we found using the dicistronic construct containing the 150 bases of human p27 5′-UTR (see below). Furthermore, no stimulation of the second cistron was observed with the 5′-UTR of p27 in the dicistronic RNA assay (). In fact, the translation of the second cistron was inhibited in the dicistronic RNA assay by inserting the 5′-UTR sequence of p27 into the intergenic region. These observations further suggest that the 5′-UTR of human p27 may not contain any significant IRES element as previously thought.
Using both the promoterless dicistronic vector and the standard promoter-testing monocistronic vector pGL3, we demonstrated that the 5′-UTR of p27 contains promoters to stimulate the transcription of its following sequences. Because deletion of the first 114 bases reduced by about half the activity to stimulate the expression of firefly luciferase gene, it is likely that the 575 bases of the 5′-UTR sequence of p27 contain a promoter at the 5′ end, which is absent in the construct harboring the 3′ 461 bases of the 5′-UTR sequence. Indeed, we found by RNase protection assay that a transcription start site exists upstream of our probe (marked as X in ). Furthermore, Kullmann et al
) also found that the 575 bases of the human p27 5′-UTR contains a promoter that is absent in the 461 bases of p27 5′-UTR, consistent with our findings. Further deletion at the 5′ end from −461 to −150 drastically reduced the activity in stimulating the luciferase expression. This observation suggests that the cryptic promoters likely exist in the region between −461 and −150. This conclusion is confirmed by RNase protection assay, which showed that two transcription start sites at positions −125 and −139 are present in both the 575 and 461 constructs but they are missing in the −150 construct (). It is, however, noteworthy that the remaining 150 bases in the 5′-UTR of human p27 still stimulate the expression of the luciferase reporter albeit at a much reduced level compared with the longer sequences, suggesting that another promoter likely exists in the 150 bases upstream of the AUG start codon. RNase protection assay confirmed that there is indeed a transcription start site at about −51 in all three constructs. However, this promoter activity appears to be very weak in the absence of upstream promoters, such as the vector SV40 promoter or promoters present in the upstream sequence of p27 5′-UTR. It, thus, remains to be determined whether these putative promoters have any physiological roles in regulating p27 expression. The use of these promoters may generate transcripts with shorter 5′-UTRs, which can be efficiently translated using the cap-dependent scanning model for translation initiation. Production of these shorter transcripts may be important for the increased expression of p27 during G1 phase. We are currently testing these possibilities and working toward this direction.
Protein factors have been suggested to be involved in the putative ‘IRES’ activity of p27. Several mRNA-binding proteins, HuR, HuD and hnRNP C1/C2 (7
), and more recently PTB (25
), were found to interact with the 5′-UTR of p27 and shown to either repress or enhance p27 expression. However, using in vitro
translation system we found no evidence that PTB stimulates the expression of the second cistron firefly luciferase of the dicistronic RNA containing the 5′-UTR of p27. Furthermore, the finding of cryptic promoters in the 5′-UTR sequence of p27 urges the need to re-evaluate the mechanism how these factors regulate p27 expression. Because all the previous claims are based on transfections using DNA reporter constructs, the effect of these factors on transcription using the promoters in the 5′-UTR of p27 could not be ruled out. Thus, the effect of these factors on the expression of endogenous p27 might in fact be mediated through mechanisms other than the putative ‘IRES’. HuR and HuD are the AU-rich element-binding protein (39
), which may affect the p27 expression through binding to the 3′-UTR of p27 mRNA. However, PTB and hnRNP-C (40
) are the known factors associated with RNA splicing and they may affect the expression of p27 by affecting the splicing of p27 mRNAs. Sequence examination showed that the 5′-UTR of p27 contains two potential splicing acceptor sites close to the AUG start codon. One is located at −114 to −129, while the other is at −37 to −52. Alternatively, these factors may affect the expression of other unknown factors, which indirectly contributes to p27 expression regulation. We propose that the effect of these factors on p27 expression need to be re-evaluated in the future by performing more stringent experiments, such as using transfection of RNA transcripts instead of DNA constructs and by testing the possibility of their effects on transcription and splicing of p27 mRNAs.