nitric oxide (NO) modulates vascular control but also displays properties relevant to the pathogenesis of autoimmune (16), chronic inflammatory (8), and neurodegenerative disorders (20); transplantation rejection (10); carcinogenesis (1); and sepsis (18). NO is synthesized from anl-arginine substrate by NO synthase (NOS), of which there are three isoforms. NO is produced constitutively by endothelial and neuronal NOS. In contrast, under inflammatory conditions, activation of an inducible NOS (iNOS) enzyme occurs, resulting in NO production over longer time periods and in larger quantities, which may have both cytotoxic and cytoprotective effects (17). The iNOS gene, which is expressed in many tissues in humans and other species, is under the transcriptional control of a number of inflammatory mediators, including cytokines and lipopolysaccharide (17). The human iNOS gene is 37 kb in length and is localized to chromosome 17 (3, 13). Although iNOS isoforms are highly conserved between species, suggesting that all are products of the same gene, there is less homology of the 5′-flanking region (3, 5, 14, 21). Moreover, although there appear to be only small variations in the deduced amino acid sequences between various human cell lines, iNOS gene transcription seems to be regulated differently (2, 5, 11) as it is in other species (14). Furthermore, cytokine combinations leading to iNOS induction may vary between species (14) and between cell types in the same species (2, 5, 12).
Identifying precisely the nature of such cell- and stimulus-specific control of NO production may be important in determining its physiological and pathophysiological roles in different tissues in humans. What is already known? First, a large span of the 5′ region (lying between −3.8 and −16 kb) is required for cytokine-mediated iNOS induction in humans (4, 11), in marked contrast to the murine iNOS promoter region where only 1 kb is necessary to confer inducibility to lipopolysaccharide and interferon (IFN)-γ (22). Second, depending on the cell type studied, different regions of the 5′-flanking sequence support cytokine-mediated iNOS induction (4, 11, 19). Third, the extent of iNOS gene expression differs according to cell or organ type. Fourth, posttranscriptional regulation may be important in determining the steady-state level of iNOS mRNA, at least in some human cell types, a phenomenon that may not be important in other species (11,15). Finally, the human iNOS promoter exhibits differential responsiveness to mixtures of the same two cytokines (i.e., IFN-γ and interleukin-1β) in different cell lines (4).
Despite these recent findings, information concerning the actual control elements of human iNOS gene regulation and tissue specificity is lacking. In this sense, the study by Mellott et al. (13a) published in this issue of the American Journal of Physiology-Lung Cellular and Molecular Physiology provides a significant boost to our understanding. What were their principal observations? First, two epithelial cell lines from two different tissues were shown to have different cytokine requirements for minimal iNOS induction. Thus, in A549 cells derived from pulmonary epithelial cells, there was a minimal requirement for both interleukin-1β and IFN-γ, whereas in the AKN-1 hepatic biliary epithelial cell line, only IFN-γ was required. Second, the authors explored the hypothesis that this difference might reflect tissue-specific molecular control mechanisms for iNOS induction by analyzing the large 5′-flanking region previously shown to be important in this process in humans (4, 11, 19). Large regions of chromatin were screened for regulatory regions with DNase I-hypersensitive (HS) site analysis (6). DNase I-HS sites are a subset of the more generally termed nuclease-HS sites, thought to represent the “open windows” that allow enhanced access of important cis-acting DNA totrans-acting factors. These sites are generally two orders more sensitive to nuclease digestion than the rest of the bulk chromatin. Within a given cell type, nuclease-HS sites fall into two main categories: constitutive and inducible. Several classes of nuclear proteins have been associated with a subset of HS sites including topoisomerases I and II, RNA polymerase II, and transcription factors. With these HS sites as a guide, in vivo footprint analysis was performed to identify potential DNA binding sites of regulatory proteins at single-nucleotide resolution. In essence, this technique involved chemical modification of the N7 of guanine residues with dimethyl sulfate (DMS). These modifications cause rare nicks in the phosphodiesterase backbone of DNA, leading to 5′-ends that could be ligated, allowing PCR to be performed [ligation-mediated PCR (9)]. When a region of the iNOS promoter between −5.8 and −13 kb was examined, large differences in chromatin structure were identified between the pulmonary (A459) and hepatic biliary (AKN-1) epithelial cell lines. In particular, the induction of iNOS with cytokines resulted in the appearance of several inducible HS sites. Although most were common to both cell types, each had unique, tissue-specific sites. Furthermore, taking the chromatin structure data as a whole, HS sites were more likely to be constitutive in the AKN-1 cells and inducible in the A549 cells. As the authors (13a) suggested, these differences may reflect a varying cytokine requirement to minimally induce iNOS mRNA levels. DMS in vivo footprint analysis was performed on a region of chromatin (approximately −5.45 to −4.95 kb) that contained both constitutive, inducible, and tissue-specific sites and because genomic sequence data was available. The authors reported a good correlation between the nature of the HS sites and that of the in vivo footprints. Interestingly, in A549 cells, there was cytokine-induced, increased DMS reactivity at approximately −5.5 kb, an area that maps to inducible HS sites in A549 cells but is constitutively accessible in AKN-1 cells. As the authors hypothesize, this difference may relate to tissue-specific regulation in terms of the minimal requirement of cytokines for iNOS induction. Furthermore, computer analysis identified activator protein-1 and Ets-1 consensus sequences within identified footprints, suggesting that these molecules may be involved in regulation of transcription of the human iNOS gene.
How robust are these data and what do they infer? The advantages of this kind of study are that DNA-protein interactions are preserved and the experiments therefore supply functional information. In this sense, they are complementary to those using transfection methods with deletion analysis of the desired promoter region, in which functional regulation by intact chromatin may be lost. Thus, to a large extent, the results of Mellott et al. (13a) support previous data from transfection studies, but they also highlight differences. Specifically, a region of the promoter shown by others to be cytokine inducible using deletion analysis (4) was inactive in the present study, which employed DNase I-HS site analysis. The authors (13a) argue that this observed difference may be due to the inability of a plasmid-based construct to attain a chromatin structure similar to that of the endogenous gene. How important such in vitro versus in vivo differences are remains unclear. Either way, HS site analysis will narrow down the areas of promoters that are likely to be functional. Could anything be added to enhance the utility of these investigations? Analysis of the functional significance of the cytokine alterations at −5.5 kb in AKN-1 cells might have improved the study, as would the identification of which transcription factors bind to this region.
The clinical significance of these findings is as yet limited. However, it is clear from animal experimentation and early clinical studies that simple blanket inhibition of NOS is unlikely to lead to therapeutic advances in complex inflammatory conditions such as sepsis. Thus NOS inhibition in patients with hypodynamic septic shock has recently been shown to adversely influence outcome (7). Clearly, research into specific regulation of NOS at a tissue level of the kind reported here may at least enable future pharmacological interventions to be directed more specifically and, therefore, effectively. At best, improved understanding of iNOS gene regulation may also provide future opportunities for molecular, tissue-specific manipulation of NO production.
S. J. Wort and J. A. Mitchell were supported by the Wellcome Trust.
- Copyright © 2001 the American Physiological Society