Re: “Regulation of S-Nitrosylation in Aging and Senescence” by Larrick and Mendelsohn (Rejuvenation Res 2019;22:171–174)

Dear Editor:

We read with attention the commentary published last April in Rejuvenation Research by Larrick and Mendelsohn ¹ in which they argue on our recent article ² and compare it with that by Zhang et al. ³ We have appreciated the interest that Larrick and Mendelsohn showed in our findings and the efforts they made to reconcile seemingly conflicting data on the role of S-nitrosoglutathione reductase (GSNOR) and S-nitrosylation in aging. Indeed, Zhang et al. propose that GSNOR increase is a distinctive feature of the prefrontal cortex and hippocampus of old individuals, whereas we report that GSNOR expression gradually decreases in cell senescence, as well as during aging in mice and humans, except for centenarians wherein it is maintained at levels comparable with those of young individuals.

We found the commentary comprehensive; however, we believe that it is unbalanced in certain passages and in the overall conclusions, in which Larrick and Mendelsohn state that GSNOR decreases in senescent cells but, on the contrary, it accumulates throughout mammalian aging. Along this line of reasoning, they support the theory of Zhang et al. and, inevitably, reduce the biological relevance of our results.

This letter is to provide our opinion on some of the controversial aspects debated in their commentary, as well as constructively discuss to explain our point of view and suggest new hypotheses. We hope this reply could make up for this lack, and contribute to offer a more general overview on the role and effects of GSNOR in aging.

There are at least two different issues we would like to point out. The first one, that we could define “semantic,” deals with what we mean by aging. What are the phenotypes that can be considered hallmarks or predictive of aging? This is a question that has still not found any unambiguous answer. Many molecular defects (e.g., impairment of the mitochondrial electron transfer chain or protein quality control) have been discovered to induce cell senescence. ^4,5 A number of gene mutations (e.g., in LMNA locus) have been found associated with a premature aging phenotype in humans. ⁶ Several pathological states have been identified as distinctive of an aged organism (e.g., neurodegenerative diseases). ^7,8 However, (1) how much cell senescence can be a direct cause of human aging; (2) how many single gene mutations can autonomously induce aging; and (3) to what extent are aging-related diseases predictive of, or share genetic abnormalities with aging, still remain heavily debated. Aging is a multifactorial and progressive process resulting from a combination of different events with no necessary univocal cause–effect relations. Therefore, we disagree with Larrick and Mendelsohn when they state that our data on GSNOR and S-nitrosylation are not related to aging but, mostly, to cell senescence. Why shouldn't the molecular mechanisms driving cell senescence contribute large-scale to human aging? Taking this as an assumption, they do not consider the importance of the numerous publications of the past 20 years on: (1) the deleterious effects of S-nitrosylation in cell (mainly in neuron) physiology and (2) Gsnor ^−/− mice characterization, suggesting that GSNOR ablation produces a number of aging-related phenotypes. Actually, Larrick and Mendelsohn cite them in the beginning of their article, but do not refer back to them. For example, they do not mention that Gsnor ^−/− mice show (1) reduced immune response ⁹ ; (2) increased morbidity after endotoxic shock ¹⁰ ; neuromuscular dysfunction resembling atrophic states and neuropathic pain ¹¹ ; and a (3) predisposition to cancer, ¹² which are all distinctive signs of elderliness.

Are these conditions not tightly related with aging? According to Larrick and Mendelsohn, apparently, they are not as tight as those from Zhang et al. supporting the hypothesis that GSNOR acts as a pro (rather than anti)-aging protein. The main reason for this relies on the evidence that transgenic mice, selectively overexpressing GSNOR in the brain (Gsnor ^tg), show defects in long-term potentiation (LTP), a parameter used by neurobiologists to assess processes associated with cognitive dysfunction. To reinforce this evidence, Zhang et al. also demonstrate that old Gsnor ^−/− mice rescue age-related cognitive decline and LTP through a mechanism involving S-nitrosylation of CaMKIIα.

It is worthwhile to mention that besides such mild long-term memory decay—which is the only functional relationship between GSNOR and aging shown by Zhang et al.—no further defects in the synaptic and presynaptic signaling, no deficit in other behavioral tests, and no further signs of aging were reported in Gsnor ^tg mice. On the basis of all this, we are still not unclear why LTP decrease should be a priori considered the most relevant hallmark of aging, but not all the other phenotypes reported in Gsnor ^−/− mice, neither, for instance, the presence of ubiquitin- and α-synuclein-containing protein aggregates that we observed in the brain of young (2 months old) Gsnor ^−/− mice. It is even harder to understand why LTP decrease and aging, at least for Larrick and Mendelsohn, should be necessarily synonymous. This is, indeed, what comes out by reading their article in Rejuvenation Research.

The second issue, that we might define “phenotypic,” is a direct consequence of the first one and deals with the question: “When can we consider an organism aged?” Aging is a process that shows its effects over time and varies among individuals. For this reason, it would be more appropriate and informative to follow the appearance of aging-related phenotypes, over time and individually. Obviously, this approach is extremely demanding if applied to humans, as it would imply to monitor a cohort of people during the entire or a significantly long period of their life. Owing to this, none of the correlations between GSNOR expression and age—both those reported by Zhang et al. (originating from meta-analysis of already existing publicly available data sets) and in our article (performed in total mRNA that we extracted from human samples)—was done in the same individuals over lifespan, but on a mixed population of individuals of different ages. From those analyses, we can only conclude that, in humans, GSNOR has a different age-dependent expression pattern, but not unambiguously state that it is up- or downregulated, or indicate any functional relationships with aging. Correctly, Larrick and Mendelsohn questioned our choice of showing GSNOR expression data in age-grouped segments to highlight the differences of the two subpopulations: young (from 18 to 20 years old) versus old (up to 85 years old). Here, we show the same data as an unbiased series to confirm that, at least in the blood samples that we analyzed, GSNOR mRNA levels significantly decrease over lifespan (Fig. 1).

OPEN IN VIEWER

FIG. 1.

GSNOR mRNA expression in PBMCs obtained from humans of different ages. Values are expressed as fold change with respect to the average of the series. GSNOR expression level is normalized on two different internal standards (actin and L34), and each point represents the mean value of each individual in triplicate. Linear regression and p value are shown. GSNOR, S-nitrosoglutathione reductase; PBMCs, peripheral blood mononuclear cells.

Intriguingly, these differences would disappear if we include long-lived (≥95 years old) individuals who, in fact, express high levels of GSNOR. We intentionally ruled out this cohort from the analyses, as we considered it to be composed of outliers. In contrast, Larrick and Mendelsohn interpret our results as an additional proof of the robustness of Zhang et al. hypothesis (i.e., that GSNOR accumulates over time), and conclude that this might be associated with memory loss distinctive of elderly. We disagree also on this point, mostly on the idea that our cohort of long-lived individuals represents a model of aging-related pathophysiology. Actually, we believe that they are a good model of longevity and healthy aging. This takes on even greater relevance if we consider that the average life expectancy in the Western countries ranges approximately from 75 to 84 years, which is at least 10 years below the age of the selected group of centenarians (95–101 years old). Therefore, any alterations in gene expression that make them differ from middle-aged (60 years old) individuals should be interpreted as beneficial rather than harmful. These changes—like those we observed in GSNOR expression—should be deemed as an exception that concurs to slow down physical decay and extend life span, not as a prototype of detrimental aging.

Similarly, we believe that mice that survive beyond 22 months (i.e., those selected by Zhang et al. for animal experimentation) are much more similar to long-lived rather than old humans. This makes us more confident that—just as we observed in centenarians—the increase in GSNOR expression reported by Zhang et al. in this group of mice could be related to longevity rather than aging. For this reason, we decided to monitor the same groups of animals up to and not beyond 12 months of age. The rationale behind this set of experiments was to have information on the expression of GSNOR over time, but to exclude long-lived individuals from the analyses. Larrick and Mendelsohn reported that Dr. Chen (the corresponding author of Zhang et al.) raised some criticisms on our choice to analyze 12 months old animals, as they are good models of “middle-aged” rather than truly “old” individuals. We believe that Dr. Chen's criticism is correct and, in retrospect, we have to admit that we should have prolonged our analyses. This would have allowed to better define the relationship between GSNOR and aging in mice. However, our data provide a unique and reliable readout on the intraindividual age-dependent modulation of GSNOR expression, as they derive from the same animals followed for 12 months. Based on these results, we are confident that GSNOR expression decreases during aging and, combining these data with those obtained in Gsnor ^−/− mice, we suggest that this negatively impacts health. On the contrary, Larrick and Mendelsohn report that GSNOR expression does not affect longevity, as Dr. Chen did never observe any differences in lifespan between Gsnor ^−/− and Gsnor ^tg mice. Accordingly, the two genetically modified mouse models can get >2 years of age like C57BL/6 wild type strains (the genetic background used to generate Gsnor ^−/− and Gsnor ^tg mice), whose average life span is ∼26 months. This statement is extremely relevant as it substantiates the choice of Zhang et al. to use 22 months old mice as good model of aging. However, they completely neglected the article by Wei et al., ¹² which demonstrates that tumor-free survival of Gsnor ^−/− mice significantly declines just after 20–22 months. This actually indicates that GSNOR ablation does affect the healthy state of mice, especially when other (age-related) morbidity factors emerge. Based on this evidence, the hypothesis that GSNOR deficiency negatively affects general mouse physiology more than it positively sustains LTP in neurons should be considered.

This would not question the validity of the results of Zhang et al., which are straightforward and convincing, but the how they are interpreted. At variance with the commentary from Larrick and Mendelsohn, we suggest that keeping GSNOR expression high might be the result of a selection—centenarians are indeed only few individuals—of a protective effect that gives an advantage against the fatal consequences of excessive S-nitrosylation in some organs, for example, in the brain. ^2,13,14 This hypothesis is further supported by the evidence that GSNOR is the only alcohol dehydrogenase expressed in the nervous system, thus underlining how its activity is extremely important for neuronal physiology. From our point of view, GSNOR increase observed by Zhang et al. during aging might represent the side effect of a protective response; in other terms, the lesser of two evils: “better senile than seriously ill.”