In this study we established that NIH3T3 mouse fibroblasts showed an increase in HO-1 mRNA steady-state levels in response to acute exposure to NO released from SPER/NO. This increase also correlated with a significant increase in the HO-1 mRNA half-life, which was also observed in response to a prolonged NO exposure using a slow-releasing diazeniumdiolate, DETA/NO or S-nitrosoglutathione. These observations are in agreement with published reports showing strong HO-1 mRNA induction in response to NO in other cell types [8–10, 16, 17] and with its induction and stabilization in human IMR90 fibroblasts [18, 19] and rat aortic smooth muscle cells in response to various sources of NO . On the other hand, we have not been able to demonstrate NO-mediated stabilization of the HO-1 message in HeLa or HepG2 cells (unpublished data), which suggests that the mechanism may be limited to certain cell types.
Treatment with NO, at concentrations that increased HO-1 mRNA stability, prevented cell proliferation during the 24 h after exposure. This observation is in agreement with the ability of NO to cause cell cycle arrest [51, 52]. It also parallels reports where exposure to DETA/NO prevented NIH3T3 cell proliferation in a p53-dependent manner, causing the induction of several growth-arrest genes: p21, GADD45 and Bax . The induction and stabilization of HO-1 mRNA under stress conditions that compromise cell growth is consistent with its proposed role in mediating cell survival during nitrosative and oxidative stress [14, 16, 17].
Also in agreement with this role is the induction and stabilization of HO-1 in response to CdCl2, NaAsO2 and H2O2. These agents, although chemically unrelated to NO, have overlapping effects on cellular physiology through the depletion of cellular glutathione, which shifts the redox equilibrium in the cell [42, 43], and can also affect the cell cycle and viability [54–58]. It is important to note that, in our study, we used exposures to CdCl2, NaAsO2 and H2O2 similar to those previously reported to induce HO-1expression [11–13]. These concentrations were lower than those reported to affect the cell cycle and, in agreement with this, failed to block cell proliferation during 24 h after the exposure. It is then possible that higher concentrations of these agents are required to achieve the degree of HO-1 mRNA stabilization seen with NO. In the case of CdCl2, concentrations of 75 μM and higher were needed to stabilize HO-1 mRNA, but under these conditions, stability was accompanied by massive ribosomal and messenger RNA degradation (data not shown). The increase in the HO-1 mRNA half-life observed in response to the non-NO agents, at concentrations that resulted in a similar HO-1 mRNA induction to that elicited by NO, shows that other sources of cellular stress may regulate HO-1 gene expression post-transcriptionally, and that HO-1 mRNA stabilization is not exclusive to NO. Overall, however, the most dramatic increase in HO-1 mRNA half-life was consistently observed with NO. Therefore it is possible that, in addition to thiol-dependent reactions and oxidative stress, other NO-specific reactions might mediate the post-transcriptional regulation of HO-1.
Because NO treatments that resulted in HO-1 mRNA stabilization were also associated with a decrease in cell proliferation, we assessed HO-1 mRNA stability in response to MMS, a cytotoxic alkylating agent. In agreement with published reports [59, 60], increasing concentrations of MMS caused a decrease in the number of viable cells over time. Interestingly, although chemically unrelated to NO, MMS exerted significant stabilization of the HO-1 mRNA, expanding the list of known HO-1 inducers. In this respect it is noteworthy that MMS can activate mitogen-activated protein kinase (MAPK) pathways in a DNA damage-independent manner, modulated by the intracellular level of glutathione , perhaps through the ability of MMS to alkylate sulfhydryl groups. Therefore, as HO-1 mRNA induction has been reported to depend on the activation of MAPK pathways in some cell types [11, 12, 57, 62], and HO-1 up-regulation is modulated by the intracellular glutathione status (reviewed in ), it is possible that MMS and NO regulate HO-1mRNA induction and stabilization through overlapping mechanisms.
To begin to address the mechanism of post-transcriptional regulation of HO-1 in NIH3T3 cells, and because no evidence for alternative polyadenylation sites within the HO-1 gene was found, we tested whether HO-1 induction by NO is accompanied by the use of an alternative transcriptional start site, which could in turn result in differences in the 5' untranslated region of HO-1 mRNA and affect its stability. Such an effect has been reported for unstable c-myc mRNA and IGFII mRNA, which can be transcribed from different promoters, and whose stability is regulated through the use of alternative transcriptional start sites [46, 47]. Unlike these cases, the HO-1 transcriptional start site was not affected by any of the stabilizing treatments (NO exposure or MMS), consistent with other data showing that HO-1 mRNA post-transcriptional regulation by NO can occur independently of transcription .
We also monitored the effect of NO and other stress inducers on HO-1 mRNA deadenylation, since deadenylation has been described as the rate-limiting step in mRNA decay [23–27]. In stabilizing HO-1 mRNA, NO had only a small effect on the initial rate of poly(A) tail shortening, acting instead to prevent the complete deadenylation of HO-1 mRNA past a 30–50-nt critical poly(A) tail length. A minimum poly(A) tail past which degradation occurs has been previously described for globin and histone mRNAs injected in Xenopus laevis extracts (reviewed in ), although such a critical length is not applicable to all mRNAs, because in this system the half-life of interferon mRNAs was independent of the length of their poly(A) tail . In this regard, it is interesting to note that, a significant amount of the HO-1 mRNA from NO-treated samples remained sufficiently polyadenylated to be retained on oligo(dT) Sepharose 8 h after treatment (Additional File 3: Recovery of polyadenylated (A>30) HO-1 mRNA from NO-treated cells). An alternative possibility is that NO acts to block decapping, or the activity of endonucleases that might be activated when the poly(A) tail is significantly shortened. This mechanism of stabilization of HO-1 mRNA is in contrast to the mRNA post-transcriptional regulation described for NMD and ARE-mediated decay, where the half-life of a transcript containing a premature termination codon or an ARE is regulated through changes in its deadenylation rate [25–27]. In our experiments, we estimated that the >5-fold increase in the HO-1 half-life that occurs in response to NO is unlikely to be the result of differences in the HO-1 deadenylation rate, because the poly(A) tail of HO-1 mRNA in NO-treated cells was shortened with a rate that was only 1.5-fold slower than that of control cells.
Our experiments with MMS show that a different form of cellular stress can also stabilize HO-1, but by a different mechanism that involves a dramatic change in the rate of shortening the poly(A) tail from ~200-nt down. We have also demonstrated that, in spite of strong transcriptional induction of HO-1 in response to CdCl2, this agent did not strongly affect the rate of poly(A) tail shortening and had only a small effect on the HO-1 mRNA half-life. As CdCl2-induced transcription did not change the start site, this experiment further supports the conclusion that stabilization occurs independently of HO-1 mRNA transcription, and excludes the possibility that detection of oligoadenylated HO-1 mRNA through the time course is merely a by-product of elevated HO-1 mRNA levels prior to the addition of AD in the treated samples.
Here we have characterized a mechanism for the post-transcriptional regulation of HO-1 mRNA where stabilization can occur essentially independently of the initial rate of poly(A) tail shortening, by preventing the final steps of deadenylation or a subsequent step in decay. Such regulation has been reported for the stabilization of transcripts in an in vitro decay assay, where purified ELAV proteins did not affect deadenylation of ARE-containing mRNA, but prevented the decay of deadenylated transcripts . Moreover, the differential regulation for the initial rate of poly(A) tail shortening and the terminal deadenylation rate has been described for the GAP-43 mRNA by HuD, a neuronal ARE-binding protein that can prevent poly(A) tail shortening from transcripts containing a long poly(A) tail (A150) [34, 36, 67]. HuD appears to have no effect on the deadenylation and decay rates of GAP-43 transcripts with short tails (A30), perhaps because of lower binding affinity for these transcripts  than for transcripts with longer poly(A) tails . In a similar way, an NO-induced stabilizing factor that would preferentially bind certain mRNAs with short poly(A) tails (A30–50) would affect only the terminal steps of HO-1 mRNA deadenylation. Indeed, some recent experiments demonstrate NO-inducible binding to the 3' UTR of HO-1 mRNA by cellular proteins (A. Rabinovic et al., manuscript in preparation).
Differential regulation of the rate of poly(A) tail shortening and terminal deadenylation could also be achieved though the differential regulation of cellular deadenylases. To date, the activity of PARN, a major cellular deadenylase, has been well characterized in HeLa S100 extracts . In addition to PARN, two yeast deadenylase activities conserved in higher eukaryotes have been characterized: Ccr4/Caf1 and Pan2/Pan3 . While both activities are required for normal deadenylation in yeast, Pan2/Pan3 stops at the last 20–26-nt, perhaps due to its requirement for poly(A) binding protein (Pab1p) as a cofactor, and only the Ccr4/Caf1 complex can process the last phase of deadenylation [23, 70]. The Pan2/Pan3 complex has been identified in HeLa cells, where its activity is also stimulated by polyadenylate-binding protein (PABP) . Thus, regulating the activity of one or the other deadenylase would have distinct effects on the rate of poly(A) shortening and terminal deadenylation, and such a mechanism could account for the differential effects of MMS and NO on the rate of HO-1 mRNA poly(A) tail shortening and terminal deadenylation.