Decreased expression of the NF-κB family member RelB in lung fibroblasts from Smokers with and without COPD potentiates cigarette smoke-induced COX-2 expression

COPD is an obstructive lung disease that is increasing in prevalence worldwide, affecting an estimated 200 million people [56]. While the etiology of COPD is strongly linked to smoke exposure, the underlying pathogenic mechanisms by which smoke causes chronic, aberrant pulmonary inflammation remains poorly defined. Numerous signal transduction pathways, including the classic (p65/p50) NF-κB pathway, contribute to cigarette smoke-induced inflammation. The pro-inflammatory activities of the classic NF-κB pathway are counterbalanced by another REL protein called RelB, initially identified as I-Rel for Inhibitory Rel because of its ability to reduce the transcriptional activities of NF-κB [47]. We have published that over-expression of RelB diminishes the induction of inflammatory mediators, including COX-2 as well as consequent lung neutrophilia caused by cigarette smoke [29,30]. These finding highlight RelB as a potentially important anti-inflammatory REL protein that protects against the deleterious effects of cigarette smoke, raising the possibility that low/absent RelB levels in the lung may predispose some individuals who smoke to aberrant inflammation and the eventual development of COPD. To our knowledge, we are the first to demonstrate that lung fibroblasts from At Risk (smokers with no airflow obstruction) and COPD subjects have significantly less RelB mRNA and protein expression compared to fibroblasts from individuals who are non-smokers (Normal).

One of the more intriguing findings from our study is that while there was a significant decrease in RelB mRNA in lung fibroblasts from At Risk and COPD subjects (Figure 2), there was no significant change in systemic RelB mRNA levels based on smoking status or airflow limitation (Table 2). This latter observation is consistent with our recent publication in which we demonstrated that systemic RelB expression was associated with health outcomes during acute exacerbations in COPD but was not associated with clinical features during stable-state [55]. While this difference between pulmonary and systemic RelB expression may be reflective of the fact that the lung fibroblasts were derived from different subjects compared to those from which systemic RelB was measured, we postulate that this disparity is reflective of fundamental differences in RelB expression, activation and function between structural and immune cells. Whereas RelB is highly expressed and constitutively active in lymphoid cells [57], there is minimal RelB activity in quiescent structural cells such as endothelial cells and fibroblasts [24,58]. In response to certain inflammatory triggers however (e.g. IL-1β, lipopolysaccharide [LPS], CD40L), RelB expression and/or activity is increased which serves to dampen the expression of inflammation-associated proteins such as chemokines and adhesion molecules [24,29,59]. This is not the case with immune cells, where RelB does not control the production of chemokines from LPS-stimulated macrophages [59]. Moreover the anti-inflammatory abilities of RelB in protecting against inflammation are largely attributed to non-hematopoietic cells, particularly fibroblasts. Evidence in support of this includes: (1) the transfer of normal bone marrow into irradiated RelB-deficient mice failed to alleviated the inflammatory syndrome and (2) injection of LPS-stimulated RelB^−/− fibroblasts potently induced inflammation in vivo [59]. Thus, it may be that lung fibroblasts with low RelB expression due to chronic smoke exposure (rather than immune cells) are active participants in the abnormal inflammation associated with cigarette smoke exposure, capable of inciting and perpetuating the pulmonary inflammatory response via the production of inflammatory mediators such as COX-2.

Our observation that lung fibroblasts from individuals who smoke (At Risk) and those with airflow limitation (COPD) have reduced RelB mRNA and protein expression (Figure 2) suggests that cigarette smoke directly reduces RelB levels, a notion further supported by our in vitro experimental data showing CSE exposure decreases RelB protein (Figure 3). Our data also support that the mechanism by which smoke decreases RelB protein is due to degradation by the 26S proteasome (Figure 3D). Consistent with this notion, others have demonstrated that RelB is proteolytically degraded in T cells in a ligand-specific manner [50,60], including the appearance of the lower MW band of ≈ 55 kDa (Figure 2A) [49] and that RelB degradation also occurs in pulmonary cells exposed to hypercapnia [25]. Another unresolved question is why there was a significant decrease in RelB mRNA in smoker-derived (At Risk) lung fibroblasts (Figure 2) whereas exposure to CSE in vitro did not significantly decrease RelB mRNA levels in any of the subject groups (Figure 3). It is possible that chronic long-term exposure to cigarette smoke, characteristic of individuals in the At Risk and COPD categories, is necessary to alter RelB mRNA expression, such that a single exposure to cigarette smoke over a 24 hour time-period, our standard in vitro exposure protocol, is insufficient to alter RelB at the mRNA level.

It is somewhat paradoxical that there was a significant cigarette smoke-induction of Cox-2 mRNA only in COPD cells (and not At Risk), despite the fact that both have similar low RelB expression levels (Figures 2 and 4). At first glance this might indicate that RelB ultimately does not contribute to the regulation of smoke-induced COX-2. However our data utilizing siRNA to knock-down RelB in Normal cells, where there is potentiation of CSE-induced Cox-2 mRNA expression (Figure 5), expands data obtained from our in vitro and in vivo models of smoke exposure [30] to show that RelB contributes to the suppression of cigarette smoke-induced COX-2 in human lung cells. One explanation for these data is that RelB is in fact critical in attenuating COX-2 in naive cells that initially encounter respiratory toxicants such as cigarette smoke, but that RelB alone is insufficient in counter-balancing the deleterious effects associated with chronic, long-term smoke exposure in COPD subjects. It may also be that post-translational modifications of RelB protein occur in COPD, accounting for the differential response between At Risk and COPD-derived cells, both of which have reduced RelB protein levels but divergent transcriptional changes in Cox-2 mRNA. Protein ubiquitination, which is implicated in COPD pathogenesis [61], augments the transactivation potential of RelB to promote NF-κB-dependent transcription [62]. Thus if such modifications account for the difference in response between At Risk and COPD cells, then alteration of RelB levels (via siRNA) in At Risk fibroblasts would be ineffectual in altering smoke-induced Cox-2 expression. There is also no difference in the expression or nuclear localization of p65 between the three groups in response to CSE (Figure 6), making it unlikely that RelB suppression of Cox-2 is via alteration in the canonical NF-κB pathway. It remains possible that co-activators such as p300, necessary for transcriptional induction of Cox-2 by NF-κB [63], are repressed by RelB or altered in COPD, thereby accounting for differences in Cox-2 between At Risk and COPD fibroblasts or the initiation of Cox-2 transcription after RelB knock-down. Moreover, cigarette smoke exposure can cause epigenetic changes in the lung, leading to significant increases in inflammatory proteins associated with COPD pathogenesis [64,65]. Thus it could also be that, in addition to low RelB levels in COPD, there are further epigenetic changes in the lung or additional protein modifications to the RelB protein not identified in this study that render it unable to exert negative control over repeated/chronic exposures. Such epigenetic alterations could also be why there is reduced RelB mRNA in lung fibroblasts from chronic smokers but not after a single 24-hour exposure. Finally, another potential explanation is that a protein partner of the RelB pathway essential for its full inhibitory activities (e.g. p100/p52 or the aryl hydrocarbon receptor [AhR]) [29] may also be inherently absent or defective in COPD subjects, and thus not allow for the full anti-inflammatory abilities of RelB. These and other possibilities are currently being explored.

Recently, a reciprocal relationship between RelB and miR-146a in immunity and inflammation has emerged including the induction of miR-146a by IL-1β [66] and CSE [31]. McMillan and colleagues demonstrated that in adult lung fibroblasts, downregulation of RelB via siRNA decreases the magnitude of IL-1β-induced miR-146a expression [66]. Sato et al. demonstrated that COPD fibroblasts produce less cytokine-stimulated miR-146a compared to fibroblasts from smokers (At Risk) [36]. In our study, both At Risk and COPD fibroblasts also failed to significantly increase miR-146a in response to CSE, and effect that was independent of RelB expression (Figure 8), suggesting that RelB does not contribute to CSE-induction of miR-146a in human lung fibroblasts. This differs from our recently published data utilizing RelB^−/− mouse lung fibroblasts, which has significantly less miR-146a compared to RelB-expressing cells [31]. While this may reflect species-specific differences in basal miR-146a regulation, it is equally likely that even low detectable levels of RelB expression in At Risk and COPD fibroblasts are sufficient to promote basal miR-146a expression. Of the potential systemic markers examined in this study, only miR-146a was associated with health outcomes (Table 3) and increased in COPD (Figures 7 and 9). Our findings that basal miR-146a was higher in both COPD lung tissue as well as systemically in the blood, but not in lung fibroblasts, suggests that cells of hematopoietic origin- and not lung structural cells- contribute to this heightened expression. It is also intriguing that miR-146a was significantly increased in COPD Gold 1 but not with more severe disease (Gold 2+). Whether the increased miR-146a in COPD 1 is a compensatory mechanism or predictive of those individuals who will go on to develop more severe COPD is not known. Further longitudinal assessment of miR-146a utilizing the CanCOLD population may reveal the novelty of miR-146a as a biomarker of COPD progression.

We recognize that there are several limitations with our study, including the cross-sectional nature of the data obtained from the fibroblasts derived from the lung surgical specimens. This ultimately makes us unable to determine if the individuals who are smokers with low RelB will ultimately develop COPD. We also cannot exclude the possibility that medications (e.g. inhaled corticosteroids) taken by the subjects in our study had an impact on the relative expression levels of RelB, as corticosteroids can dampen Cox-2 gene transcription via NF-κB [67]. Finally, another perceived limitation is the reliance on mRNA levels to correlate with clinical parameters, as quantification of blood RelB protein expression remains to be determined. Despite these limitations, our data support that RelB suppresses COX-2 expression upon exposure to cigarette smoke. When considered with our previous work [31], our data implies that RelB and miR-146a may work cooperatively to suppress COX-2 expression in response to environmental toxicants.