The TNF mRNA was the first-identified and remains the best-characterized target for TTP-induced destabilization (13
). That it is a physiologically relevant target was demonstrated by the systemic inflammatory phenotype seen in mice deficient in TTP, which could be prevented almost entirely by repeated injections of anti-TNF antibodies or by interbreeding the TTP-KO mice with mice deficient in both TNF receptors (10
). Nonetheless, the subsequent identification of GM-CSF mRNA as a physiological target for TTP demonstrated that there was more than one target for TTP-induced mRNA destabilization and that this interaction could occur in bone marrow-derived stromal cells instead of in the macrophages thought to be responsible for most TNF secretion (12
). This indicated that TTP might be a more general regulator of ARE-containing mRNA stability, with some targets presumably cell type and stimulus specific.
In an attempt to identify other mRNA members that might be regulated by TTP, we returned to the cell type and stimuli that were used in the initial identification of TTP: cultured mouse fibroblasts, serum deprived and then stimulated with insulin, serum, or phorbol esters (24
). In these cells, TTP mRNA levels are essentially undetectable in the serum-deprived cells, but increase dramatically after serum stimulation to reach maxima at approximately 45 min. We found that the corresponding protein levels were also undetectable at time zero, but reached a more stable plateau after approximately 90 min of serum stimulation. Therefore, we chose 90 min after serum treatment as a time when maximal concentrations of TTP protein would be available for target mRNA binding, while mRNAs encoded by genes rapidly but transiently induced by serum would still be available as potential TTP targets. We anticipated that the single known genetic difference between the two stable fibroblast cell lines used (i.e., the presence or absence of TTP) might permit the identification of novel TTP targets whose mRNAs were abnormally stabilized in the specific absence of TTP. Nonetheless, one limitation of our study is that only a single cell type, a single stimulus, and a single time after stimulation were used; we anticipate that other physiologically relevant TTP targets will be uncovered by using similar approaches with other cell types, stimuli, and timing paradigms.
Since we were simultaneously measuring levels of some 40,000 transcripts in these studies, one might anticipate that the mRNA decay rates for some transcripts might differ based on statistical expectations alone. We attempted to minimize this effect by purifying cellular RNA from several time points after actinomycin D addition, with multiple biological replicates at each time point. Nonetheless, it was critical to confirm any differences in decay rates from the microarray experiments by a second measurement technique: in this case, Northern blotting. In addition, since the conversion from primary embryonic fibroblast cultures to stable cell lines is undoubtedly associated with one or more genetic changes, it was important to revalidate the results with a different pair of cell lines derived from other littermate mice.
It should be noted that our intention was to use microarray and bioinformatics approaches to focus on transcripts that, in our view, had the greatest likelihood of being physiological TTP targets, rather than on discovering systematically the entire population of TTP targets. This approach meant that many interesting candidates had to be set aside while the highest-probability targets were interrogated further. For example, we focused on transcripts containing two or more heptamer TTP binding sites that were conserved between mice and humans. However, we have shown previously that the internal UUU sequence in the heptamer can be expanded to four or contracted to two U's with only twofold decreases in binding affinity (8
), making it possible that sequences with those types of loosened consensus sequence could still be true binding partners. Similarly, in the 250 transcripts significantly stabilized in the TTP-KO cells, there were approximately 42 that contained only a single TTP binding heptamer; we have demonstrated previously that even a single binding site can confer TTP-dependent destabilization on a transcript (21a
). In addition, there were many transcripts that had two or more binding sites that were not perfectly conserved in humans. These could well be important targets in the mouse, or the apparent failure of conservation could have been due to imperfect consensus, inadequate database information, etc. In all of these cases, likely targets that were bypassed because of our focus on high-probability targets will need to be evaluated in future follow-up studies.
The microarray studies also suggested that there were 56 transcripts that were significantly destabilized in the absence of TTP. Some of these may have been a statistical by-product of the large number of comparisons being made. However, there are precedents in the literature for ARE binding proteins that stabilize transcripts upon binding, and we have previously demonstrated that TTP, at high concentrations, can actually protect ARE-containing transcripts from accelerated decay (20
). Thus, it remains theoretically possible that among these 56 transcripts will be some whose physiological decay is actually prevented by TTP binding, a possibility that will also require follow-up experiments.
The top candidate transcript in this study was that encoding Ier3, also known as IEX-1 or gly96 (see reference 30
and references therein). Its 3′-UTR contains several highly conserved potential TTP binding sites, and we showed by direct binding and cotransfection studies that these sites were important in TTP binding as well as in mediating TTP-dependent transcript decay. The Ier3 transcript has in common with the TNF, GM-CSF, and IL-2 mRNAs that it can be rapidly induced in the specific cell type of interest, in this case by serum, and it exhibited a short half-life in this situation. Most of the other validated TTP targets exhibited similar rapid induction and subsequent rapid decay kinetics, suggesting that this type of “immediate-early response” might be characteristic of most if not all physiological TTP targets. Nonetheless, the lack of effect of TTP deficiency on the prototypical immediate-early response transcript, that of c-fos
, as well as its apparent lack of effect on its own fusion transcript, argues that TTP is not involved in the regulation of many other rapidly inducible, rapidly turning over transcripts. It is possible that the other TTP family members, which are more constitutively expressed in this cell type, might be responsible for the decay of mRNAs that might also be more constitutively expressed.
We studied the TTP regulation of the Ier3 transcript in some detail as an example of the studies that will be necessary to establish each of the top nine hits as physiological TTP binding partners and “targets.” This point deserves further emphasis: The other eight “top hits” were evaluated by only one or two Northern blotting experiments, and each will have to be subjected to the same battery of tests as the Ier3 transcript in order to be considered a true, physiological TTP target transcript. The Ier3 transcript fulfilled the same types of experimental criteria that were applied to previous studies of the TNF mRNA: i.e., the mRNA was markedly stabilized in KO cell lines derived from three different pairs of mice; TTP could bind to several ARE binding sites in the transcript; and TTP could be used to promote decay of the Ier3 transcript in cell transfection experiments, in a manner dependent on the presence of the characteristic TTP binding sequences in the transcript and key zinc finger residues in the TTP protein. Interestingly, the Ier3 transcript was not stabilized in analogous experiments using fibroblasts deficient in Zfp36l1 (32
), a TTP family member that is expressed more constitutively in these cells. Considerable further research will be required to determine the effect of the TTP-induced destabilization of this transcript in normal physiology; it remains possible that such an effect will only be observed during some abnormal perturbation of normal physiology.
The Ier3 transcript and protein have been the subject of considerable study since their original cloning as a serum-induced gene (14
). Mice in which the Ier3 gene was disrupted were recently shown to exhibit elevated blood pressure and cardiac hypertrophy (30
), while earlier studies of overexpression in lymphocytes in intact mice demonstrated impaired T-cell apoptosis and increased susceptibility to a lupus-like syndrome (40
). It will be of interest to determine whether the increased Ier3 mRNA stability demonstrated in the TTP-KO cells will be reflected in a contribution to the TTP deficiency phenotype in mice, which, among other characteristics, exhibit a systemic autoimmune syndrome with some lupus-like features (33
). To our knowledge, there have been no reports of systemic overexpression of Ier3 in mice that might help to identify potentially relevant aspects of the TTP deficiency phenotype. One approach might be to remove the ARE, resulting in a stabilized mRNA and enhanced expression; this approach was used successfully in the case of the TNF mRNA, when removal of the ARE resulted in a profound inflammatory state that resembled a more severe form of the TTP deficiency syndrome (18
). We anticipate that a similar approach, applied to the AREs of Ier3 and, possibly, the other transcripts described here, should reveal the physiological significance of these sequence elements in the intact animal.