Overexpressing the HD-Zip class II transcription factor EcHB1 from Eucalyptus camaldulensis increased the leaf photosynthesis and drought tolerance of Eucalyptus

Alteration in the leaf mesophyll anatomy by genetic modification is potentially a promising tool for improving the physiological functions of trees by improving leaf photosynthesis. Homeodomain leucine zipper (HD-Zip) transcription factors are candidates for anatomical alterations of leaves through modification of cell multiplication, differentiation, and expansion. Full-length cDNA encoding a Eucalyptus camaldulensis HD-Zip class II transcription factor (EcHB1) was over-expressed in vivo in the hybrid Eucalyptus GUT5 generated from Eucalyptus grandis and Eucalyptus urophylla. Overexpression of EcHB1 induced significant modification in the mesophyll anatomy of Eucalyptus with enhancements in the number of cells and chloroplasts on a leaf-area basis. The leaf-area-based photosynthesis of Eucalyptus was improved in the EcHB1-overexpression lines, which was due to both enhanced CO2 diffusion into chloroplasts and increased photosynthetic biochemical functions through increased number of chloroplasts per unit leaf area. Additionally, overexpression of EcHB1 suppressed defoliation and thus improved the growth of Eucalyptus trees under drought stress, which was a result of reduced water loss from trees due to the reduction in leaf area with no changes in stomatal morphology. These results gave us new insights into the role of the HD-Zip II gene.

Leaf and photosynthetic traits. Significant alterations were again observed in the leaf morphological and anatomical traits in the transgenic lines EcHB1-2 and EcHB1-10 ( Fig. 2A). These transgenic lines had densely-packed small leaves with increased number of chloroplasts; in these transgenic lines, the leaf area was 0.5-fold, while the mesophyll fractions (ratio of mesophyll tissues in leaf section), number of mesophyll cells, and chloroplasts was 1.3-1.9-fold of the control line GUT5 (Fig. 2B).
As a result of the dense mesophyll tissues and increased number of chloroplasts, the transgenic lines EcHB1-2 and EcHB1-10 had a high area-based photosynthetic capacity compared to the control line GUT5 (Fig. 3A). The maximum photosynthetic rate (A max ), maximum Rubisco carboxylation rate (V cmax ), and photosynthetic electron transport rate (J) in the transgenic lines EcHB1-2 and EcHB1-10 were 1.3-1.7-fold those of the control line GUT5 (Fig. 3B).
We determined the CO 2 diffusional limitations during photosynthesis in the Eucalyptus leaves (Fig. 4). Stomatal conductance (g s ) in the transgenic lines was slightly higher (1.2-fold) or similar to that in the control lines, while mesophyll conductance (g m ) in the transgenic lines EcHB1-2 and EcHB1-10 was 1.8-1.9-fold that of the control line GUT5 (Fig. 4A). Although the stomatal and biochemical limitations were similar between the lines, the mesophyll limitations of leaf photosynthesis were smaller in the EcHB1-10 line compared to the control line GUT5 (Fig. 4B). The higher g m in the transgenic lines EcHB1-2 and EcHB1-10 correlated to a higher chloroplast surface area exposed to intercellular airspaces (S c ), and the higher photosynthetic rate at 1500 µmol m −2 s −1 of PPFD (A 1500 ) in the transgenic lines correlated to a higher g m (Fig. 4C).
The data for photosynthesis and leaf morphological or anatomical traits were pooled for the two transgenic lines that had high EcHB1 expression levels (EcHB1-2 and EcHB1-10), and then compared to those of the control line GUT5 (Table 1). Leaf photosynthetic traits were increased in EcHB1-overexpressed lines except for stomatal conductance (g s ), which corresponded to an increase in the anatomical factors enhancing CO 2 diffusion (S c , 1.6-fold) with no change in the cell wall thickness, which would inhibit CO 2 diffusion. The increase in the number of mesophyll cells (1.4-fold) was accompanied by decreases in the palisade and epidermal cell width (0.8-fold), and an increase in the number of chloroplasts (1.5-fold) was followed by decreases in the width and thickness of the chloroplasts (0.8-0.9-fold). A higher leaf mass per area (LMA, 1.2-fold) while a similar mesophyll thickness was obtained. Stomatal traits including the stomatal size and density were similar between the EcHB1-overexpressed lines and GUT5.
Principle component analyses were performed for the traits that were different between the EcHB1-overexpressed lines and GUT5 (Table 1), among the traits that were related to leaf photosynthetic functions (Fig. 5A), and among the leaf anatomical traits (Fig. 5B). The first main axis (PC1) is considered to be suitable for explaining variabilities in the data for both analyses, because PC1 accounted for 75-87% of the total variability in the data. Gas exchange parameters, such as A max , g m , V cmax , J, and A/E, had negative loadings, and the anatomical traits of S c and LMA also had negative loadings, while the leaf area (LA) had a positive loading over PC1 (Fig. 5A). Overall, the size of the cells or chloroplasts had negative loadings, which corresponded to the negative loading of LA over PC1 (Fig. 5B). In contrast, the number of cells, number of chloroplasts, and fraction of mesophyll cells had positive loadings, with positive loadings of LMA and S c . In these biplots, GUT5 plants had positive scores, while EcHB1-2 and EcHB1-10 had negative scores over PC1. Tree growth traits such as stem diameter, tree height, growth rate in stem diameter and tree height, shoot fresh weight (FW), root FW and leaf number. Blank, gray, and filled bars indicate GUT5, EcHB1-2, and EcHB1-10, respectively. Means with standard errors are shown for 3-12 trees. Differences between lines were analyzed using one-way ANOVA, in which different letters indicate that the difference was statistically significant (P < 0.05).
www.nature.com/scientificreports www.nature.com/scientificreports/ photosynthetic traits on a tree-basis and drought tolerance. The evaporation rate measured at 1500 μmol m −2 s −1 of PPFD (E 1500 ) was smaller in the line EcHB1-2 than GUT on a tree-basis (0.6-fold) and on a leaf-basis (0.5-fold), while the line EcHB1-2 had a similar E 1500 on a leaf-area basis to those in the line GUT5 ( Table 2). On a tree-basis, photosynthesis measured at 1500 μmol m −2 s −1 of PPFD (A 1500 ) and the surface area of the chloroplasts facing intercellular airspaces (S c ) were similar between the line GUT5 and EcHB1-2, although in the transgenic line EcHB1-2, these traits were higher and lower on a leaf-area-basis and a whole-leaf-basis, respectively.
Overall, the drought and following recovery experiment showed enhanced drought tolerance in EcHB1-2 (Fig. 6). The number of leaves was higher in EcHB1-2 than that in GUT5 at all three drought levels during the 4-month drought experiment, except for the 3-month high-drought level (Fig. 6A). The tree height was higher in EcHB1-2 than that in GUT5 at all three drought levels after a 4-month drought (Fig. 6B). Stem diameter was similar in EcHB1-2 and GUT5 at the low-drought level, and differences between EcHB1-2 and GUT5 at the middle-and high-drought levels became less distinct compared to before the drought experiment. After the 3-month recovery experiment, enhanced growth was observed overall in the transgenic line EcHB1-2, in which differences between the lines were more significant under the low-drought level (Fig. 6C). The ratios of tree height, shoot fresh weights and stem fresh weights of EcHB1-2 to those of GUT5 were 1.4-1.5-fold, 1.3-1.4-fold, and 1.2-1.4-fold at the low-, middle-, and high-drought levels, respectively.

Discussion
Significant enhancements in the area-based leaf photosynthetic functions were obtained in Eucalyptus trees overexpressing EcHB1 (Fig. 3). This is the first study showing the effect of an HD-Zip class II transcription factor on the leaf photosynthetic function. Leaf photosynthesis is limited by 1) biochemical activities and 2) CO 2 diffusion through the stomata and mesophyll [6][7][8] . Overexpression of EcHB1 in the leaf induced significant alterations in the fully-expanded mature leaves. Blank, gray, and filled bars indicate GUT5, EcHB1-2, and EcHB1-10, respectively. Differences between lines were analyzed using one-way ANOVA, in which different letters indicate that the difference was statistically significant (P < 0.05).  Table 1). The increase in the number of mesophyll cells is related to an increase in the number of chloroplasts (1.5-fold, Fig. 5B, Table 1), which may induce enhancements in the photosynthetic biochemistry, e.g., the carboxylation enzyme Rubisco that regulates the velocity of carboxylation (V cmax ) and photochemical efficiency and the Calvin cycle that regulates the electron transport rate (J, Fig. 3B). When the relationship between the leaf anatomical traits and biochemical parameters was analyzed, a high V cmax corresponded to a large number of chloroplasts in angiosperms and ferns 8,23 .

Scientific RepoRtS
Overexpression of EcHB1 induced marked enhancement in the mesophyll diffusional conductance, g m (1.8-1.9-fold, Fig. 4A), which is strongly affected by alterations in the mesophyll anatomy (Fig. 5A); the significant increase in the number of chloroplasts (1.5-fold) with a decrease in chloroplast size (Table 1) resulted in a 1.6-fold increase in the chloroplast surface area (S c , Table 1) for CO 2 diffusion 5 . The concurrent increases in S c and g m in the EcHB1-overexpressed lines (Fig. 4C), together with the concurrent increases in g m and A (Fig. 4C), supports the significant role of S c for determining g m , and thus, photosynthesis. On the other hand, overexpression of EcHB1 induced little alteration in the stomatal conductance, g s (Fig. 4A). The stomatal density and size strongly affect stomatal conductance, g s 24 . Low stomatal density as well as large stomatal size reduces g s , because low stomatal density reduces the total stomatal pore area in leaves, while large stomatal size increases the diffusion path length 25 . The small change in stomatal conductance (g s ) by the overexpression of EcHB1 (Fig. 4A) may be due to the lack of alterations in both the stomatal density and size (Table 1). CO 2 diffusion through the stomata and mesophyll imposes almost comparable limitations in photosynthesis in different tree groups including deciduous, evergreens, and conifers 7 . The present study for Eucalyptus trees showed that the mesophyll tended to impose Photosynthetic parameters such as maximum photosynthetic rate (A max ), maximum carboxylation rate (V cmax ), and maximum electron transport rate (J) estimated from light response curves and A-C i curves in the leaves of Eucalyptus. Means with standard errors are shown for 4-12 fully-expanded mature leaves. Blank, gray, and filled bars or symbols indicate GUT5, EcHB1-2, and EcHB1-10, respectively. Differences between lines were analyzed using one-way ANOVA, in which different letters indicate that the difference was statistically significant (P < 0.05).
Overexpression of EcHB1 improves the drought tolerance of Eucalyptus trees (Fig. 6). This improvement in drought tolerance may be related to the alteration in the transpiration of the whole tree. As a result of (1) an unchanged transpiration rate on a leaf-area basis, (2) a significant reduction in leaf area (0.5-fold), and (3) a small increase in the number of leaves (1.3-fold, data not shown), the estimated transpiration rate on a tree basis was reduced (0.6-fold) for the EcHB1-overexpressed lines ( Table 2). This will reduce water loss from trees, and thus, may contribute to prevent the EcHB1-overexpressed lines from leaf-falling. Additionally, we observed alterations in the vessel anatomy in the stems of EcHB1-overexpressed lines, with a decrease in the radial diameter and a high cell wall thickness (data not shown). These changes will enhance stiffness of xylem cells but reduce the hydraulic conductivity in stems 26,27 , and as a result, will also reduce water loss from the trees during drought. Although the lower root biomass in the EcHB1-overexpressed lines (Fig. 1B,C) may cause less water uptake from the soil and thus potentially induce a sensitive response to drought 28 , the effect of reducing water loss ( Table 2) may overcome the effect of less water uptake under the severe drought condition in the present study. The lower rate of leaf-falling during drought (Fig. 6A) may contribute the rapid recovery of the number of leaves after re-watering and thus improves the growth of the trees (Fig. 6C). Some morphological properties of the EcHB1-overexpressed lines on a tree-basis are similar to the Eucalyptus trees with high drought tolerance. Eucalyptus provenances from dry areas in Australia have a higher leaf per stem area ratio 29 , which supports our result of a larger number of leaves with a smaller stem diameter, i.e., a larger number of leaves per stem area, in the EcHB1-overexpressed line EcHB1-2 (Fig. 6A).
The HD-Zip I, HD-Zip II, and HD-Zip III transcription factor networks regulate plant growth responses to environmental conditions through suppression or promotion of cell multiplication, differentiation, and expansion through phytohormone-regulated networks 16,17 . Some Arabidopsis HD-Zip II genes are required www.nature.com/scientificreports www.nature.com/scientificreports/ for shade-induced hypocotyl elongation 30,31 , as well as controlling several auxin-regulated developmental processes including lateral organ polarity 32 . Overexpression of EcHB1, an HD-Zip II transcription factor, induced an increase in tree height, a decrease in tree diameter, and a decrease in root growth in Eucalyptus (Fig. 1B,C), which is in line with the results of a previous study 30 ; in Arabidopsis seedlings, the increasing level of Arabidopsis HD-Zip II gene AtHB2, induces shifts from radial cell expansion to longitudinal cell expansion in the hypocotyl, and involves an inhibition in root growth. The increase in plant height by the overexpressing EcHB1 under the control of CaMV 35 S promoter was also reported in tobacco plants 15 . These results suggest that the functions of EcHB1 for plant growth are similar to those of AtHB2.
Alterations in leaf morphology by the overexpression of EcHB1 (Fig. 2, Table 1) again suggests similar functions of EcHB1 to those of some Arabidopsis HD-Zip II genes including AtHB2. The decreased leaf size in EcHB1-overexpressed Eucalyptus (Fig. 2) supports the results of previous studies which found that overexpression of Arabidopsis HD-Zip II genes, AtHB2, AtHB4, HAT1, HAT2, and HAT3 decreased leaf expansion in Arabidopsis 30,33-36 . The reduction in leaf expansion may be partly due to the reduction in the width of epidermal cells in the leaf-width direction 30 (Table 1). In the present study, we also found increases in the proliferation of leaf mesophyll cells, as well as increases in the number of chloroplasts in the leaf-thickness direction in the Eucalyptus plants overexpressing EcHB1 (Table 1), which suggest additional functions of HD-Zip II genes in leaf development.
In conclusion, overexpression of EcHB1, an homeodomain leucine zipper (HD-Zip) II transcription factor, in the hybrid eucalypt GUT5 (Eucalyptus grandis and Eucalyptus urophylla) induced significant enhancements in leaf-area-based leaf photosynthesis. This improvement is due to both enhanced CO 2 diffusion into chloroplasts and increased photosynthetic biochemical functions, which are affected by alterations in the mesophyll anatomy, e.g., an increased number of chloroplasts per unit leaf area. Overexpression of EcHB1 improved drought

Number of cells and size
Mesophyll cell number (mm −1 section) 336 (14) 457 ( Table 1. Mean values of photosynthetic, anatomical, and morphological traits of the 4-22 fully-expanded mature leaves in the 4-year-old trees. Means with standard errors in parenthesis are shown. Data obtained using the isotope method are presented for mesophyll conductance (g m ). Anatomical data were obtained using light or electron micrographs (n = 25-120). Significant levels were tested by Welch's t-test with differences shown as * P < 0.05, ** P < 0.01 and *** P < 0.001. n.s. means no significant difference (P > 0.09). www.nature.com/scientificreports www.nature.com/scientificreports/ tolerance, which is mainly affected by the reduction in leaf area with no changes in stomatal morphology. Although negative regulations of leaf size by Arabidopsis HD-Zip II genes have been already reported, the positive regulation of cell division and number of chloroplasts by an HD-Zip II gene are new findings, which gave us insights into the role of the HD-Zip II gene.

Methods plant material.
A hybrid Eucalyptus tree GUT5 generated from Eucalyptus grandis and Eucalyptus urophylla was used as the host plant 14 . This GUT5 is a selected clone that has a high ability to regenerate from a transformed callus. GUT5 shows good growth performance in subtropical areas, especially in Brazil where commercial forest plantations of Eucalyptus species are widely performed. Transgenic Eucalyptus GUT5 lines with a CaMV 35S Figure 5. Biplot of the principle component analyses for leaf and photosynthetic traits where the differences between the control and EcHB1-overexpressed lines were statistically significant (see Table 1 and EcHB1-2 were tested by Welch's t-test with * P < 0.05, ** P < 0.01 and *** P < 0.001. n.s. means no significant difference (P > 0.1). www.nature.com/scientificreports www.nature.com/scientificreports/ promoter fused to EcHB1 cDNA (accession No. AB458829) were generated in our previous study 15 . We analyzed control (GUT5) and two transgenic lines (EcHB1-2 and EcHB1-10), in which 3-16 plants were used for the growth and drought experiment and one -three individual were selected for the detailed analysis of leaves. To obtain multiple transgenic Eucalyptus plants with the same genetic background, the stems of Eucalyptus plants were transplanted to 1.5-liter plastic pots filled with soil and then grown in temperature-controlled greenhouses of the Forestry Research Institute of the Oji Holdings Corporation (Kameyama, Japan) and Kyoto Institute of Technology (Kyoto, Japan). The plants were watered daily and fertilized biweekly with liquid fertilizer Hyponex (Hyponex Japan, Osaka, Japan) at a concentration of 1/1000. In the glasshouse of the Forestry Research Institute, the daily minimum and maximum temperatures ranged from 15-25 °C and 25-36 °C, respectively. In the glasshouse of the Kyoto Institute of Technology, the temperature was kept at 25 °C. Prior to the detailed leaf analysis, phenotypes of the plants and the growth rate were measured at the Forestry Research Institute for 12-month-old trees. The growth rates in stem diameter and tree height were calculated as increments of the values per 12 months.  (B) Tree height and stem diameter before and after the 4-month drought treatment. (C) Tree height and shoot and stem fresh weight after 3 months of re-watering. Means with standard errors are shown for 7-25 trees. The differences between GUT5 and EcHB1-2 were tested using the t-test, with asterisks indicating a significant difference with * P < 0.05, ** P < 0.01, *** P < 0.001 and n.s. indicating the difference was not significant. P-values are shown when 0.05 < P < 0.1.

Leaf morphological and stomatal properties.
For measurements of the leaf morphological traits, fully expanded 5 th to 10 th leaves from the top of the branches were obtained during the same period as the photosynthesis measurements were performed (from August to October). Leaf transverse sections were obtained from five leaves (n = 5) at the central part of the lamina, and fixed and embedded in Spurr's resin. Mesophyll anatomy was analyzed in 800-nm-thick transverse sections of leaves, observed using a light microscope (BX51-33, OLYMPUS, Tokyo, Japan), and the cell wall thickness of mesophyll cells was observed in 80-nm-thick transverse sections of leaves with a transmission electron microscope (JEM-1220, JOEL, Tokyo, Japan), as described previously 42 . Leaf anatomical traits were measured from digitized images of leaf sections of the lines GUT5, EcHB1-2, and EcHB1-10, using software 43 (Image J). The surface area of chloroplasts exposed to the intercellular air spaces per unit leaf area (S c ) and mesophyll porosity were estimated as described previously using the images of light micrographs 10 . Briefly, S c was estimated from the ratio of the perimeter length of chloroplasts/mesophyll cells exposed to the intercellular air spaces. Mesophyll fraction was calculated as 1 -mesophyll porosity. The number of mesophyll cells and chloroplasts, mesophyll thickness, mesophyll fraction, and S c were measured from five images for each line (n = 5). Cell size (width and length) of the uppermost palisade mesophyll and the secondary layer of spongy mesophyll, the size of chloroplasts in palisade cells (width and thickness), and upper epidermal cell size (width and thickness) were measured in five cells of each image (n = 25). The thickness of the cell walls covered with chloroplasts was measured at two parts of the palisade cells in 10 images of sections of 80 nm thickness (n = 20) using a transmitting electron microscope (JEM-1220, JEOL Ltd., Tokyo, Japan). The chlorophyll content was estimated using a chlorophyll meter for six leaves (SPAD-502Plus, Konica-Minolta, Tokyo, Japan); the chlorophyll content was shown as SPAD values 44 . Leaf area was measured in four leaves using a scanner (Canoscan 9950 F, Canon, Tokyo, Japan) and image analysis software (Image J), and the leaves were dried at 60 °C for 48 h using an oven (MOV-112, SANYO Electric, Osaka, Japan) and weighed to assess the dry mass to calculate leaf dry mass per leaf area (LMA).
Stomatal density (SD) and stomatal size (length and width) were estimated in the lines GUT5 and EcHB1-10 using light micrographs of the replicas of leaf surface, as described in our previous paper 24 . The fully-expanded 5 th leaves from the top of the branches were obtained in July, and SD was measured in two leaves of the four individuals (n = 8), and stomatal sizes were measured in five closed stomata of two leaves from the three individuals (n = 30). tree-basis photosynthesis and evaporation. The Leaf-area-basis photosynthesis rate and evaporation rate (A 1500 , E 1500 ) measured at 1500 μmol m −2 s −1 of photosynthetic photon flux density (PPFD) and S c were obtained for four leaves of the GUT5 and EcHB1-2 lines (n = 4), as described previously. Then, whole-leaf-basis A 1500 , E 1500 , and S c were calculated using the averaged leaf-area-basis data multiplied by the leaf area of six leaves (n = 6). Tree-basis A 1500 , E 1500 , and S c were calculated using the averaged whole-leaf-basis data multiplied by the number of leaves of the 8-month-old individual trees (n = 15-16).
Drought tolerance experiment. Seven to nine Eucalyptus trees of the lines GUT5 and EcHB1-2 were generated from tree cuttings or calli, and grown in 7.0-liter pots for eight months in the glasshouse of the Forestry Research Institute of Oji Holdings Corporation, and irrigated with sufficient amounts of water daily before the drought experiment. To determine drought tolerance, drought and recovery experiments were successively performed. For the drought experiment, Eucalyptus trees were grown for 4 months under low-, middle-, and high-drought levels. The "low-drought level" was determined as the minimum irrigation level that no leaf