Acclimation and adaptation to stress are well known types of transgenerational adaptive plasticity 
. Examples include tolerance to several stresses in timberline plants associated with adaptation to UV-B radiation 
; increased tolerance to cold in progeny of Arabidopsis
plants grown at low temperatures 
; and enhanced performance of progeny grown in the light environment of parents in a parental light environment 
. We found that progeny of Arabidopsis
plants exposed to salt, temperature, water and UVC stresses exhibit increased HRF, increased tolerance to stress, and increased DNA methylation.
The heritability of these transgenerational effects in successive generations is still an issue. Although we were able to confirm earlier studies 
reporting that several stresses, including UVC, induced changes in HRF that persisted in the progeny, we could not prove that these effects persisted in successive generations in the absence of stress. In a recent study, Pecinka et al. (2009) reported that transgenerational effects of stress on HFR were stochastic, i.e., highly variable and dependent on the nature of stress 
. Four of 10 stress conditions they tested appeared to be effective: a genotoxic agent bleomycin and chemical zebularine which blocks cytosine methylation induced a persistent increase in HFR. Paraquat which induces oxidative stress increased HRF that did not persist; and mannitol which induces osmotic stress decreased HRF in the progeny; these results are consistent with the effect of drought that we observed. Factors that might account for the discrepancies include transgenic lines used, plant age and growth conditions as well as the exact nature of stress protocols. In conclusion, various stress factors can induce transgenerational changes in HRF; however, these changes do not represent a consistent general response to stress, and moreover, they are not necessarily inherited in the absence of stress. Further, since these measurements depend on the use of transgenes as reporters, the biological significance of observations is still unclear.
Changes in DNA methylation have been proposed to be responsible for adaptation to stress by Arabidopsis thaliana
plants and the pine tree population naturally grown in the vicinity of Chernobyl 
. Common iceplants (Mesembryanthemum crystallinum
) exposed to stress undergo changes in satellite DNA methylation, which results in a switch from C3-type to C4-type carbon dioxide assimilation 
. A direct correlation between the frequency of rearrangements at various disease-resistant gene-like loci and the level of methylation at these loci in response to stress resulting from virus infection was observed before 
. In the present study we found that transgenerational effects of stress on HRF and stress tolerance were associated with changes in DNA methylation. Together with the effects of 5azaC on stress tolerance, this is consistent with the hypothesis that alterations in DNA methylation are required for transgenerational effects that we observed. The exact relationship between DNA methylation and stress is still unclear. While the genome of S1 plants was hypermethylated at the global level, many loci nonetheless exhibited hypomethylation. Moreover, we found no clear correlation between changes in HRF at the transgene locus and methylation of the locus.
Interestingly, several genes known to be involved in HFR or chromatin modifications showed altered methylation in S1_25 and S1_75 plants relative to controls (Table S2
; ). For example, the promoter region of Msh2 involved in mismatch repair and UVH3 involved in UV-damaged DNA repair exhibited a 50% decrease in methylation associated with stress. Similarly, 5 different genes encoding proteins that are involved in histone modification, namely, SUVH2, SUVH5, SUVH6, FLD and UBP26, showed a dramatic increase in methylation associated with stress. SUVH2, SUVH5 and SUVH6 are histone methyltransferases involved in heterochromatic gene silencing 
together with SUVH4 (KRYPTONITE)
control transposon movement, whereas SUVH6
together with SUVH4
control transcribed inverted repeats 
. Flowering locus D (FLD) encodes a protein containing a histone deacetylation domain. Deficiency in FLD results in hyperacetylation of FLC chromatin, up-regulation of FLC expression, and extremely delayed flowering 
encodes an enzyme that removes ubiquitin modifications of histone H2B, facilitates DNA methylation and heterochromatin formation, and is important for endosperm and flowering 
Potential mechanism of transgenerational changes in the progeny of stressed plants.
Our results provide evidence that transgenerational effects of salt stress on HRF and stress tolerance depend on DCL2 and/or DCL3. Small RNA biogenesis in Arabidopsis thaliana depends on several proteins including DCLs. In particular, biogenesis of miRNA requires DCL1, whereas siRNA biogenesis depends on DCL2, DCL3, and DCL4.
Stress is known to induce the differential expression of various small regulatory RNAs 
. Micro-RNAs (miRNAs) seem to be the predominant class of molecules that are induced by abiotic stress such as cold, drought, salt and UV, with many of them being commonly regulated 
. The involvement of siRNAs in abiotic stress response is somewhat less established, the salt-regulated nat-siRNA P5CDH and SRO5 pair being the most well-known example 
. The involvement of siRNA metabolism in the establishment of a new methylation pattern and possibly stress tolerance has been suggested before 
. Recent work by Agorio and Vera (2007) showed the role of AGO4 in the process of resistance of Arabidopsis
to Pseudomonas syringae
; these scientists found that ago4
was sensitive to bacterial infection 
. It is an interesting fact that dcl3
mutants, which are supposedly impaired in the same pathway of siRNA biogenesis, remained tolerant to bacterial infection, thus suggesting the complex process of stress tolerance this pathway is involved in.
Each of the DCL enzymes generates predominantly a particular class of small RNAs. Whereas DCL1 is required for miRNA biogenesis 
, DCL2 is apparently needed for the generation of viral siRNAs 
. DCL3 is involved in processing of endogenous repeats and in the formation of heterochromatic siRNAs 
, whereas DCL4 is required for ta-siRNA biogenesis 
. DCL3-dependent processing of endogenous repeats and the formation of heterochromatic siRNAs can be considered as one of the mechanisms capable of directing RNA-dependent DNA methylation 
The involvement of DCL2 and DCL3 in passing on the memory of stress to progeny may occur at different levels, the main one possibly being the establishment of a differential methylation pattern via the activity of small RNAs. The reason why the picture of the involvement of DCLs in transgenerational response did not become more pronounced can be explained by the substantial functional redundancy of DCLs, suggesting their compensating functions 
In conclusion, we have shown that the progeny of stressed plants exhibit changes in recombination frequency, genome methylation and stress tolerance. Admittedly, we have only documented these changes in HRF within the transgene reporter thus far. No mechanistic link between DNA methylation and the level of homologous recombination at the loci we have studied can be established here. However, we provide the first experimental evidence that the establishment of stress acclimation and stress adaptation correlates with changes in genome methylation and potentially depends on small RNA pathways requiring DCL2 and DCL3. It remains to be determined whether stress induces the expression of smRNAs targeting specific sequences within the plant genome for methylation or repressive histone modifications. Experiments involving various mutants impaired in the establishment/maintenance of methylation patterns, such as drm2, ddm1, met1, cmt3 as well as mutants impaired in biogenesis of miRNAs/siRNAs, will provide further insight into this interesting phenomenon.