The Janus faces of iNOS.

To the Editor: Recently, two groups have reported on the phenotype of transgenic mice with cardiac-specific overexpression of the inducible nitric oxide synthase (iNOS). Although both used the -MHC promoter to target iNOS expression, our model used the tetracyclineregulated system ( MtTA /iNOS ),1 whereas the other used a nonconditional approach.2 Intriguingly, the phenotypes reported varied dramatically, from lethal arrhythmias in our case, to no identifiable abnormality in the second. Additionally, a recent publication from the second group3 includes criticisms of our work that are misleading. The statement that “iNOS activity in [the nonconditional] model is approximately 100to 120-fold higher than. . .in the conditional model” (page 1357)3 is unfounded. A direct comparison between cardiac iNOS activities in the two models is not possible. Heger et al2 presented S-ethylisothiourea-sensitive conversion of arginine to citrulline (control: 1.7 3.1 versus transgenic: 697 238 pmol min 1 mg ), whereas we measured L-NMMA–sensitive Ca -independent activity (1 mmol/L EGTA) (0.056 0.05 versus 4.57 1.36 pmol min 1 mg ).1 While both provide measures of iNOS activity, absolute values varied considerably. Although baseline values in the Heger mice were ostensibly 30-fold higher than ours, it is apparent from these1,2 (and other4) reports that basal cardiac iNOS activity in mice is not different from zero. Thus, it is inappropriate to compare directly the reported iNOS activities in these models. Normalizing iNOS activity using control values is invalid, since fold increases based on a denominator that approaches zero is prone to overestimation. In our view, a more reliable comparison could be made between total NOS activity, for which accurate basal measurements can be made. Unfortunately, total NOS activity has not been reported for the Heger mice. Indeed, based on the published data, we believe that our model exhibited higher (not lower) NOS activity, as evidenced by 10-fold elevation in total L-NMMA–sensitive NOS activity in the heart (ie, in the absence of EGTA) compared with controls.1 Additionally, our model is accompanied by enhanced cardiac ONOO and O2 , suggesting iNOS “uncoupling,”1 as would be expected in the setting of its marked and sustained overexpression. By contrast, the Heger mice exhibited only 2to 3-fold increase in NOx production2, and no evidence of uncoupled activity. Thus, their recent conclusion that “Mungrue et al. . . failed to provide biochemical evidence of iNOS activity in vivo” (page 1357)3 chooses to ignore other biologically important products of iNOS activity, namely ONOO and O2 . In our view, Heger et al show that modest increases in coupled NOS activity with NO as the main product is not harmful, whereas we show that high levels of uncoupled activity associated with free radical production is associated with cardiac pathology. This view is entirely consistent with the literature. Of particular importance, several lines of evidence suggest active selection against the generation of mice with high levels of cardiac iNOS activity. First, our attempts to produce nonconditional -MHC-iNOS mice were not successful after screening 50 founders. Indeed, we would be interested to know how many mice Heger et al screened to generate transgenic founders? Second, in generating the conditional iNOS line, our frequency of genotypepositive founders (1 of 15) was significantly lower than that observed in other lines (1 of 4). Also, in our tTA-dependent iNOS lines, there was no detectable tTA-independent iNOS expression. In comparison, we observe that up to 30% of putatively tTA-dependent transgenes may exhibit some baseline expression (ie, leakiness). The absence of such founders in the iNOS project suggests that leaky expression of iNOS was selected against. Third, in the absence of doxycycline supplementation, there was 50% reduction in the number of MTA /iNOS progeny born.1 Accordingly, in comparing the phenotypes reported for conditional versus nonconditional -MHC– directed iNOS overexpression, the occurrence of adaptive changes in the Heger mice remains another possible explanation for the differences in phenotypes. Instead, Wunderlich et al3 have suggested that nonspecific effects of the tTA or LacZ coexpression could account for the phenotype of our conditional model. This speculation is particularly misleading. First, ageand sex-matched littermate MTA /iNOS controls, which express tTA (but not iNOS) in a cardiac-restricted manner, did not display any evidence of cardiac dysfunction,1 consistent with many other reports.5–9 Second, while doxycycline administration would have suppressed both the expression of iNOS and -gal, the latter is unlikely to have caused our phenotype. -Gal is a widely used reporter gene, and its expression has not previously been reported to generate a cardiac phenotype.6 Moreover, in our study, line 365iC, which had comparable -gal activity but no iNOS expression (possibly due to transgene truncation/deletion), did not display any incidence of premature mortality. Finally, preliminary experiments using an iNOS-specific inhibitor (1400W), completely rescued the lethal phenotype of MTA /iNOS mice (authors’ unpublished data, 2003), providing further confirmation that the cardiac manifestations are a result of increased iNOS activity in this model. In conclusion, while the differences between the two models of enhanced cardiac iNOS remain to be fully explained, it is misleading to dismiss cardiac toxicity as an artifact of the conditional system. Rather, it is more reasonable to suggest that the conditional approach is essential to overcome a lethal phenotype or adaptations that result from high levels of cardiac iNOS overexpression, likely related to uncoupled activity and increased free radical generation. Indeed, the major strength of the tTA-dependent transgenic system is its ability to overcome a well-defined and fundamental weakness of nonconditional transgenic strategies with respect to premature mortality and/or developmental adaptation.