Why were differences in methylating enzyme expression opposite of our hypothesis?
In the present study, we detected a significant effect of native versus non-native population status on the expression of all three genes. As with the trend in Senegal, older (native) populations expressedmore DNMT1 and DNMT3 as well as TET2 than non-native birds. What are reasonable explanations for this pattern, so strong but reverse of expectations? There might be some insight to gain by comparing the most important other drivers of gene expression among the factors we considered (Fig. 1). After tissue, temperature predictability and genetic group were the strongest predictors of expression with native/non-native status significant but a fraction as informative. We cannot explain the effect of genetic group because we had no a priori reason to expect this factor to be so important. This predictor captures the evolutionary history of populations and thus could represent more of a phylogenetic than functional difference.
The very strong effect of temperature predictability, by contrast, is intriguing because it resembles the directionality of the difference we saw for native/non-native status here and in Senegal: birds from sites where plasticity should be more favorable (i.e., the youngest populations) expressed less of the enzymes that regulate methylation. Our expectation was that if DNMT1/3 and TET2 expression were mediators of reversible plasticity (Wu and Zhang 2010, Bogan and Yi 2024), expression of these genes should be highest where conditions are less predictable (McCaw et al. 2020). Even the genetic group effect could be an echo of temperature predictability: despite our ambitious sampling effort, we only collected tissues from 9 countries. More importantly, sparrows from Israel, Senegal and Vietnam are also members of genetic group 2, the most predictable sites we studied (Table 1). Our current country set prevents us from disentangling genetic history and temperature predictability. Nonetheless, we suspect that temperature predictability is the driving force here. Genetic group 1 includes birds from Spain, one of the more predictable climates, but also birds from the Netherlands and Norway, the least predictable climates, and native/non-native status either relates to introduction history alone or is also partly entangled with temperature predictability.
Another potentially useful way of updating our hypotheses given our results involve kinds of plasticity that might be fostered by methylation. Our initial hypothesis was based on environmentally-sensitive, reversible plasticity. Our results may reflect, though, the dispositions of populations to realize canalized, developmental plasticities to predictable cues over development, not plasticity activated and suppressed to match current conditions (Vogt 2017a, Vogt 2017b). In support, there was a consistent effect of latitude on expression of all three genes. Latitudinal effects might reflect a tendency for DNMT/TET expression to track photoperiod-related seasonality (Stevenson 2018), which increases towards the poles. Seasonality as a predictable form of environmental variation is quite distinct from Colwell’s indexes, which capture climaticunpredictability . Perhaps DNMT/TET expression is more important for plastic responses to predictable than unpredictable environmental change (McCaw et al. 2020, Lynch et al. 2016). There could also be an upper limit on the rate at which environmental variation can be transduced into DNMT expression, then methylation, and ultimately reversible plasticity (Snell-Rood et al. 2018). DNA methylation plays a role in the regulation of several seasonal plasticities (Fishman and Tauber 2024) from the neuroendocrine coordination of biorhythms to the recrudescence of reproductive tissues (Sharma et al. 2018). We expected that DNMT expression differences play some role in reversible plasticity, too, but the consistent effects of latitude on expression could suggest that DNMT and TET2 expression levels might instead be set permanently in early development (i.e., epigenetically programmed). By measuring gene expression in only adult birds, we could be capturing the roles these enzymes played in coordinating what would become rhythmic changes in phenotypes in consistently periodic environments (Friston 2010). Measurements in immature animals could have produced very different patterns.
We find this possibility (i.e., the patterns we observed largely being due to the age of the birds we studied) worthy of future study but unsatisfying. First, latitude effects on expression were very small compared to other forces. If seasonality is so important, its effects would probably have been of comparable strength to the other factors. Second, DNMT and TET2 expression (Lynch et al. 2016) and resultant methylation (Sheldon et al. 2020) can in fact be quite dynamic in many vertebrates (see also Schrey et al., this issue). Expression of all three enzymes can change over months, weeks, days, and even hours (Alvarado et al. 2015, Stevenson 2017). To our knowledge, no one has yet evaluated how quickly expression of these enzymes can change in adult house sparrows, but there is no reason to believe that this species would be unlike others. Finally, DNMTs in other species can in fact play roles over the timescales and contexts that underpin our hypotheses (Luo et al. 2012). For instance, in domesticated chickens, strain differences in resistance to Marek’s disease virus were related to DNMT expression; exposure to a novel and lethal environmental stimulus (i.e., the virus) led different individuals to distinctly respond to and cope with the stimulus via plasticity over a fairly short time scale. If broadly applicable, these results suggest that appreciable within-individual variation in the expression of these DNMT/TET is achievable. Indeed, inter-cell type variation in expression of these genes is well-known, and it is partly for this reason we measured expression in three tissues. One thing to mention as well is that we gave all birds in our study a small dose of lipopolysaccharide (LPS) to induce an immune response that is the subject of other projects. We could not study untreated birds for ethical and practical reasons, but it is possible that birds not treated with LPS could have shown different patterns.
A final surprising result that warrants attention and might even be related to the unexpected outcomes we observed among populations are the strong within-individual and among-population correlations in gene expression (Fig. 2). We did not expect such strong relationships among genes given their unique functions, yet these results suggest that the persistence, creation and erasure rates of methyl marks across tissues within individuals are probably similar among birds and across sites. Such strong covariation suggests that these genes might evolve and/or operate as a unit among individuals and populations, which could constrain expression variation consistently, despite the utility of plasticity for unpredictability in some sites.