Leaf Trichome Distribution Pattern in Arabidopsis Reveals Gene Expression Variation Associated with Environmental Adaptation

Gene expression varies stochastically even in both heterogenous and homogeneous cell populations. This variation is not simply useless noise; rather, it is important for many biological processes. Unicellular organisms or cultured cell lines are useful for analyzing the variation in gene expression between cells; however, owing to technical challenges, the biological relevance of this variation in multicellular organisms such as higher plants remain unclear. Here, we addressed the biological relevance of this variation between cells by examining the genetic basis of trichome distribution patterns in Arabidopsis thaliana. The distribution pattern of a trichome on a leaf is stochastic and can be mathematically represented using Turing’s reaction-diffusion (RD) model. We analyzed simulations based on the RD model and found that the variability in the trichome distribution pattern increased with the increase in stochastic variation in a particular gene expression. Moreover, differences in heat-dependent variability of the trichome distribution pattern between the accessions showed a strong correlation with environmental factors to which each accession was adapted. Taken together, we successfully visualized variations in gene expression by quantifying the variability in the Arabidopsis trichome distribution pattern. Thus, our data provide evidence for the biological importance of variations in gene expression for environmental adaptation.

8 intensity perturbed both trichome and stomata distribution patterns (Fig. S4). However, 1 7 9 seedlings exposed to a modestly high temperature (26°C) showed perturbed trichome 1 8 0 distribution patterns but normal stomata distribution patterns compared with seedlings grown 1 8 1 in plates at 22˚C (Fig. 5), suggesting that gene expression was more varied at 26°C. A further 1 8 2 increase in temperature to 30˚C eventually disturbed stomata distribution patterns. These 1 8 3 results suggest that the trichome distribution pattern is more sensitive to environmental 1 8 4 fluctuations than the stomata distribution pattern. 1 8 5 The results obtained from sampled leaves are likely to reflect the amounts of GL3 1 8 6 gene products. To confirm that the variation of the GL3 protein amounts correspond to the 1 8 7 NND distribution pattern, we evaluated the GL3 protein level in transgenic plants expressing 1 8 8 the GL3 gene fused to yellow fluorescent protein (YFP) gene under the control of the GL3 1 8 9 promoter (GL3-YFP) [14,27]. Since we considered the trichome pattern but not mature 1 9 0 trichomes per se, we measured YFP signals in the cells within the trichome initiation zone 1 9 1 [28]. Variability in the intensity of GL3-YFP fluorescence in Arabidopsis leaves increased at 1 9 2 26°C (Fig. 6). The distribution of GL3-YFP fluorescence showed no significant difference 1 9 3 between plants grown under strong light condition, in which patterns of trichome as well as 1 9 4 pavement cells were altered, and those grown under normal light intensity, as expected. 1 9 5 These results indicate that the trichome pattern reflects the variation in GL3 expression, 1 9 6 unless the stomata pattern change, and is increased at modestly high temperatures. 1 9 7 1 9 8 2.4 Trichome distribution patterns are variable in accessions and climate 1 9 9 Based on the above findings, we speculated that variations in gene expression under 2 0 0 certain conditions could be predicted based on the regularity of the trichome pattern, and this 2 0 1 could have an adaptive significance. We were also curious to determine which natural factors 2 0 2 are responsible for the gene expression variation. We addressed these questions by 2 0 3 comparing the regularity of trichome distribution patterns between of A. thaliana accessions 2 0 4 adapted to the different climatic conditions [29][30][31]. A total of 11 accessions (Don-0, Aitba-1, 2 0 5 Col-0, IP-Tri-0, Yo-0, Ra-0, Van-0, Ler-0, Spro-0, Pi-0, and Kin-0) were grown together in one 2 0 6 9 plate under a normal or modestly high temperature (Fig. 7). Strikingly, the results indicated 2 0 7 that the change in NDD variance varied between accessions. The accessions could be 2 0 8 divided into two groups at 26˚C: the high-variance group, comprising Don-0, Aitba-1, and Col; 2 0 9 and the low-variance group, comprising Ler and other accessions. IP-Tri-0 was the only 2 1 0 accession showed no difference in NND variance distribution between at 22°C and 26°C.

1 1
This led us to another question, i.e., whether gene expression variations would reflect the 2 1 2 specific environment to which an accession is adapted. 2 1 3 To evaluate the relationship between gene expression variation and environmental 2 1 4 factors, we used the BioClim data set, which comprises 19 global land surface datasets 2 1 5 (Worldclim [http://www.worldclim.org/current]) [32]. We calculated the ratio of variance at 2 1 6 26°C to that at 22°C. When gene expression variation increased under mild heat, the ratio of 2 1 7 variances increased beyond 1.0. We analyzed the ratio of variances of 11 accessions and 2 1 8 compared these ratios with BioClim indices (Fig. S5). We found that three indices (BIO1, 2 1 9 BIO10, and BIO11) showed significant positive correlations with the ratio of variances of NND.

0
Interestingly, all of these three indices represent mean temperature under certain conditions. 2 2 1 In particular, BIO1, the index of the mean of annual temperature, was positively correlated 2 2 2 with the variability in the trichome distribution patterns (Fig. 8). These results demonstrate a 2 2 3 strong relationship between gene expression variation and environmental factors, especially 2 2 4 temperature, suggesting that temperature affects gene expression variations. Gene expression variations are mainly classified into two categories based on the 3 1 1 source of causal factors: intrinsic variation, which is observed intrinsically due to 3 1 2 thermodynamics fluctuations, resulting in variable gene expression changes between cells; 3 1 3 and extrinsic variation, which is caused by extrinsic factors, such as environmental stimuli, 3 1 4 and is therefore observed in all cells simultaneously. In our experiments, we could not 3 1 5 determine whether intrinsic or extrinsic variation caused the variability in the trichome 3 1 6 distribution pattern. Variation in the nucleosome structures due to epigenetic modifications is 3 1 1 3 supposedly a major contributor to intrinsic variation [2]. Our results of the chemical 3 1 8 perturbation of histone modification revealed changes in trichome distribution patterns, which 3 1 9 could be an evidence of intrinsic variation in the case of trichome pattern. On the other hand, 3 2 0 altered trichome distribution patterns due to a modest heat treatment may be an example of 3 2 1 extrinsic variation. Therefore, our approach consists of both intrinsic and extrinsic variations. 3 2 2 In conclusion, our finding of a correlation between the trichome distribution pattern 3 2 3 and temperature suggests that natural variation in gene expression is associated with 3 2 4 adaptation to the environment. Further investigation would help to understand the biological 3 2 5 relevance of the trichome distribution pattern, as well as the mechanism underlying gene 3 2 6 expression variation to achieve environmental adaptation. Ohio State University, Columbus, OH, USA), were used in this study. We show data for Col and 3 3 3 Ler, which originated from Col-0 and Ler-0, respectively. Seeds of rol1-1 and rol1-2 mutants 3 3 4 were obtained from ABRC. Seeds of the ap6-1 mutant were kindly provided by P. Wigge (John 3 3 5 Innes Centre, UK) and K. Sugimoto (RIKEN, Japan).

6
Seeds were sterilized for 6 min in a solution of 50% commercial bleach (Kao, Singapore) 3 3 7 containing 6% sodium hypochlorite and then were washed three times with distilled water. The 3 3 8 sterilized seeds were laid out on a sterile filter paper or in 50% Murashige and Skoog (MS) 3 3 9 medium (Wako Pure Chemical Industries, Osaka, Japan) supplemented with 1% sucrose 3 4 0 (Wako Pure Chemical Industries) and 6% gellan gum (pH 5.9; Wako Pure Chemical Industries), 3 4 1 and sterilized 0.1% agarose solution was added to the seeds. Plates containing the seeds were 3 4 2 incubated at 4°C in the dark for 3 days. Subsequently, the plates were transferred to a growth 3 4 3 chamber and incubated for 21 days at 22°C under constant light (1,000 lm/m²).
Initial gene expression difference affects a trichome pattern.
By observing trichome patterns, the initial gene expression variation can be predictable.
Gene expression is varied in each initial cell (Blue).