Selection of elite lines of BpGH3.5-transgenic Betula platyphylla using growth adaptability analysis

Chu Wang · Yang Wang · Jianjun Zou ·Rusheng Peng · Qibin Yu · Guifeng Liu ·Jing Jiang

Abstract White birch ( Betula platyphylla) is precious material for pulpwood and widely distributed in 14 provinces of China. Previous study indicated that inhibited expression of a gene encoding an auxin amide synthase,BpGH3.5, in transgenic plants reduced the level of IAAamino acid conjugation, resulting in more free IAA, thereby better growth of birch. Utilizing transgenic- BpGH3.5 lines to increase wood production in a wide range of environments is the goal for breeders. In three f ield trials here, we measured tree height, diameter at breast height, and volume of 16 BpGH3.5-transgenic 7-year-old white birch lines(including 12 antisense strand lines and 4 overexpression lines) and a wild-type white birch line from three sites that varied greatly in their environmental conditions. To select elite BpGH3.5- transgenic lines for each target environment,we used an additive main eff ects and multiplicative interaction model to analyze genotype by environment interaction,growth adaptability and stability. The selection criteria for elite transgenic lines were set as the average volume plus 0.75 times the standard deviation for the tested lines at each test site. Results showed that the eff ect of line and site for height was highly signif icant ( P ＜ 0.01), and the eff ect of line × site was signif icant ( P ＜ 0.05); selected as the elite lines were FG12, FG13 and FG27 at the Maoershan Experimental Forest Farm, FG13 and FG32 at the Shidaohe Forest Farm, and FG3 and FG31 at the Ecological Experiment Forest Farm. These seven high-yield, stable lines can now be tested in production trials or adjacent trial areas with similar environmental conditions, while the high-yield, unstable lines should be tested in production trials in areas deemed suitable for their growth. These results provide guidance on which released transgenic elite lines will grow best in a wide range of conditions.

Keywords Betula platyphylla · BpGH3.5 · Aff orestation test · Growth adaptability

Introduction

Auxin amide synthase (Gretchen Hagen 3,GH3) genes are auxin primary/early response genes, which are closely related to plant growth and development (Guilfoyle et al.1998; Liscum et al. 2002). GH3 mainly regulates the concentration of free indole acetic acid (IAA) by conjugating amino acids to hormones such as IAA in plant tissue cells, thereby regulating the concentration of active IAA(Li et al. 2008; Mellor et al. 2016). Transgenic studies on rice (Oryza sativa) showed that the overexpression ofGH3genes resulted in shorter heights of transgenic lines, and the overexpression ofOsGH3.2,OsGH3.8, andOsGH3.13signif icantly improved the tolerance of the lines to biotic and abiotic stresses (Zhang et al. 2009 ; Du et al. 2012; Ding et al. 2008). The overexpression ofCsGH3.1andCsGH3.6genes in citrus (Citrus sinensis) enhances the resistance toXanthomonas axonopodispv.citriby inhibiting auxin accumulation (Chen 2017).

White birch (Betula platyphylla), widely distributed in 14 provinces in northern, northeastern, northwestern and southwestern China, is a valuable material for pulpwood.Its wood volume is the greatest among broad-leaved trees and is equal to the sum total of all poplar cultivars in China.Genetic engineering and breeding studies on white birch since the 1990s have led to a shortened juvenile period,improved insect resistance and salt tolerance, and variations in leaf type and lignin and chlorophyll content (Zhan et al. 2003; Huang et al. 2013; Li et al. 2013a, b; Zhang et al. 2014; Chen et al. 2018, 2021; Fang et al. 2018). TheBpGH3.5-transgenic white birch that was developed by Yang et al. ( 2015a, b) and the antisense chain transgenic white birch have shorter primary roots and lateral roots than the wild-type white birch (WT), but the root hairs are signif icantly longer and more abundant than those of the WT, but there is no signif icant diff erence in plant growth between the transgenic lines and the WT. In f ield trials of 7-yearoldBpGH3.5-transgenic lines at the White Birch Breeding Station of Northeast Forestry University, tree height, DBH and volume of 22BpGH3.5-transgenic lines (G) and 32BpGH3.5-antisense-transgenic lines (FG) were analyzed (Yu et al. 2019). Results show that the expression ofBpGH3.5gene is inhibited after theBpGH3.5antisense strand is introduced into the birch genome, which weakens the ability ofBpGH3.5to conjugate amino acids to IAA; thus, there is more free IAA, promoting tree growth. Most of the antisense strand lines grew faster than the wild type.

To evaluate the growth adaptability and stability of the aforementionedBpGH3.5-transgenic white birch trees in diff erent environments, we evaluated 7-year-oldBpGH3.5transgenic birch lines that had been started in 2012. The analysis of genotype by environment interaction (GEI) is the main method for screening genotypes for special adaptability to diff erent environments (Li et al. 1998; Li et al. 2001;Nagamitsu et al. 2014; Liu et al. 2015; Shen et al. 2015).Since the GEI complicates the estimation of phenotype and genotype values and leads to bias in the estimate of genetic eff ect, identifying elite transgenic lines and determining their true genetic potential can be a challenge. The additive main eff ects and multiplicative interaction (AMMI) model has been widely used to adjust GEI eff ect for estimation of genetic stability and growth adaptability (Bose et al. 2014).AMMI captures a large portion of the GEI and separate main and interactions and provide meaningful interpretation of the genotype stability. AMMI is a hybrid analysis combining ANOVA for the genotype and environment main eff ect with principal component analysis of GEI. Based on the AMMI, a stability value can then be calculated. The objectives of the study were to (1) identify elite transgenic lines across the diff erent environments, and (2) provide production guidance on the environmental risk for the released transgenic lines.

Materials and methods

Plant materials and growing environments

Four lines were transfected withBpGH3.5overexpress strand (G), 12 lines withBpGH3.5antisense strand(FG), and one line was a WT (Table 4). The transgenic lines were subcultured on a medium containing Woody Plant Medium (WPM) + 6-Benzylamino purine (6-BA)1.0 mg L -1 + 1-Naphthlcetic acid (NAA) 0.02 mg L -1 + Kanamycin (Kan) 50 mg L -1 + Cephalothin (Cef) 250 mg L -1 ;the WT line was subcultured on the same medium but without Kan and Cef. In the early spring of 2011, the aforementioned transgenic birch seedlings were transplanted into rooting medium (WPM + 3-indolebutyric acid (IBA) 0.4 mg L -1 ), then 30 days later to seedling trays (Yang et al. 2011)and placed in the greenhouse for routine management. At the beginning of June, seedlings of the same height were selected and transplanted into 21 cm × 21 cm pots. About 60 plants were transplanted for each line and placed in a plastic shed for conventional water and fertilizer management.

In the early spring of 2012, three field trials were established, each in a randomized arrangement with rows 2 m × 2 m apart. The recipient birch used for the transgenesis was a single plant from the Maoershan elite seed sources in Heilongjiang Province. Therefore, the selection of the plantation trial sites after obtaining theBpGH3.5-transgenic line took into account the similarity of environmental conditions, and the adaptation trials were carried out in three provinces in Northeast China: Heilongjiang, Jilin and Liaoning. They are Maoershan Experimental Forest Farm (Shangzhi, Heilongjiang Province), Shidaohe Forest Farm (Huinan,Jilin Province), and Ecological Experimental Forest Farm(Chaoyang, Liaoning Province) (Fig. 1). Maoershan Experimental Forest Farm is on a small ridge on the western slope of the Zhangguangcai Ridge in the Changbai Mountains. It is in a hilly area of low mountains that stretches from the Songnen Plain to the Zhangguangcai Ridge. The region has a temperate continental monsoon climate. The vegetation belongs to the f lora of the Changbai Mountain in northeastern East Asia, dominated by temperate coniferous and broad-leaved mixed forests. Shidaohe Forest Farm is located in southeastern Jilin Province at the transition zone between the Changbai Mountains and the Songliao Plain. The region is in the northern temperate zone and has a continental monsoon climate in the northern temperate zone. The vegetation is also part of the f lora of Changbai Mountain. The Ecological Experimental Forest Farm, in the northern slope of the Longgang Mountain branch of the Qianshan Mountains, is also within the Changbai Mountains with the same vegetation type. However, it has a temperate continental climate(Table 1).

Fig. 1 Location of the aff orestation trial site in northeastern China

Growth variable measurements

The number of 7-year-old trees varied from 70 to 111 at each site and were measured in the spring of 2018. The test lines and their numbers in the three test sites are shown in Table S1. Height was measured using the Ultrasonic Altimeter and the Users Guide Vertex IV and Transponder T3(Hagl?f, L?ngsele Sweden). DBH was measured with a DBH ruler.

Table 1 Geographical climate factors at the three test sites

Statistical analyses

SPSS version 19.0 (IBM, Armonk, USA) and Excel 2016(Microsoft, Redmond, WA, USA) software for analysis of variance, multiple comparisons and principal component analysis (PCA).

whereY ijkis the height value of repetitionkof lineiat sitej,uis the overall mean,α iis the deviation between the mean and the overall mean for linei(plant main eff ect);Β jis the deviation of environmentjfrom the total mean (environmental main eff ect),λ ris the eigenvalue or singular value of the principal component axis of the interaction eff ectr,representing the part of the interaction square sum that can be explained by this axis,ψ iris the eigenvector value of the line onr,σ jris the environmental eigenvector value on axisr,nis the number of PC axes retained in the model,which means the product term that can contain most of the GEI interaction information number,ρ ijis the residual error

wherenis the number of signif icant IPCs,D g(e)is the score of genotypegor environmenteonnIPCs, andD g(e) is a measures of the relative stability of environmentgor environmente.

Results

Multisite combined variance analysis and main genetic variable analysis

Table 2 presents ANOVA results for growth traits of the 17 tested lines ranged from 3 to 23 trees at the three f ield trials.A signif icant eff ect of line and line × site for height, DBH and volume was found (P＜ 0.05). The eff ect of line × site on height was found to be highly signif icant (P＜ 0.01). Thesite did not signif icantly aff ect the height, DBH and volume for the diff erent transgenic lines. We thus decided to use height for the subsequent stability analysis based on the above results.

Table 2 ANOVA of growth variables of transgenic lines at each test site in northeastern China

From the results of the growth variables for the genotypes(lines), the lines performed best at the Ecological Experimental Forest Farm. Mean DBH was 4.24 cm and volume 0.0037 m 3 higher than at the other sites. The coeffi cient of variation (CV) for height, DBH and volume was the smallest at the Ecological Experimental Forest Farm. This indicated that the transgenic plant not only grew the most in the Ecological Experimental Forest Farm, but also grew uniformly.The height, DBH, and volume growth in Maoershan Forest Farm and Shidaohe Forest Farm had large CVs (Table 3).

Selection of elite transgenic lines at the test sites

Multiple comparisons were made on the growth traits of the 17 tested lines in 3 test sites (Table 4). The selection criteria for elite transgenic lines were set as the mean volume plus 0.75 times the standard deviation for the tested lines at each test site. The volume for lines FG12, FG13, and FG27 at the Maoershan test site was 61.29%, 67.74%, and 77.42%higher than population mean, and 88.24% and 47.06% for lines FG13 and FG32 at the Shidaohe Forest Farm, and 43.24% and 45.95% for lines FG3 and FG31 at the Ecological Experimental Forest Farm, respectively. The aff orestation preservation rate of selected elite lines is above 75%,except for the experimental sites at the Ecological Experimental Forest Farms.

Analysis of growth stability and adaptability

There was a highly signif icant transgenic line and site eff ect(P＜ 0.01) and a signif icant line × site eff ect (P＜ 0.05) on height (Table 5). The results of the PCA showed that the f irst component can explain most of the interaction (73.59%),and was highly signif icant (P＜ 0.01). It also showed that the application of the AMMI model to analyze the interaction eff ects between lines and sites was feasible and accurate.

In a double plot with the average height of each line as the abscissa and the IPC1 value of each line or three test sites as the ordinate (Figs. 2 and 3), we can intuitively understand the interaction between the growth of the tested lines and the environment. The results showed that the degree of dispersion between the test lines in the horizontal axis direction was greater than that between the test sites, which indicated that the high growth performance of the same transgenic line at diff erent test sites diff ered less. In the direction of the vertical axis, when IPC1 = 0 is taken as the dividing line, the G4, G9, WT, FG5, FG10 and the Ecological Experimental Forest Farm were located on the same side of the dividing line. This result indicates that the test site had a positive interaction eff ect on the growth adaptability of the above f ive lines. Correspondingly, Maoershan Experimental Forest Farm and Shidaohe Forest Farm also had positive interactions on FG2, FG3, FG8, FG11, FG12, FG13, FG27, FG29,FG32, FG31, G6, and G7, indicating that these 12 lines have higher growth adaptability at the Maoershan Experimental Forest Farm and Shidaohe Forest Farm. In the horizontal axis direction in Fig. 2, line FG13 had the greatest mean height and G4 the lowest. In the direction of the vertical axis,the absolute value of IPC1 of the G4 is the largest, indicating that the line has the strongest interaction eff ect with the environment, was more sensitive to the growth environment,and had low stability. Lines G9 and FG8 had the smallest absolute IPC1 value, indicating that their growth was less aff ected by environment and they had the highest stability.The mean height of trees at the Shidaohe Forest Farm was the highest, followed by the Maoershan Experimental Forest Farm and the Ecological Experimental Forest Farm (Fig. 3,Table 6). These results are consistent with those of the analysis of the growth variables of the tested lines at each site.

Based on the relative stability parameterDgcalculated according to the linear regression model, the order of height stability of the 17 lines, with lowerDgvalues indicatinghigher stability of the tested line, was G9 ＞ FG8 ＞ FG2 ＞ F G3 ＞ FG12 ＞ FG13 ＞ FG31 ＞ G6 ＞ FG10 ＞ G7 ＞ FG5 ＞ WT＞ FG11 ＞ FG29 ＞ FG32 ＞ FG27 ＞ G4 (Table 6). The growth stability of the tested lines was the highest at the Maoershan Experimental Forest Farm, followed by the Shidaohe Forest Farm and the Ecological Experimental Forest Farm.

Table 3 Statistics for main growth variables for the transgenic lines at the three sites in northeastern China

Table 4 Growth variables and preservation rates of transgenic white birch at each test site in northeastern China

Table 5 AMMI model of test lines at each test site

Fig. 2 Biplot of tree height for each of 17 lines

Fig. 3 Biplot of tree height for each site

In the comprehensive analysis of the 17 tested lines shown in Figs. 2 and 3 and Table 6, lines with “a” in the results of the multiple comparisons of height were considered high yield, the rest were low yield; lines withDgless than 0.5 were stable lines, and the rest were unstable(Table 6). The high-yield, stable lines were FG2, FG3, FG8,FG10, FG12, FG13 and FG31. The high-yield, unstable lines were FG27, FG29 and FG32. The low-yield, stable lines were G6, G7, G9 and FG5. Lines G4, FG11 and WT were low yield and unstable.

Discussion

Growth performance of transgenic lines

Studies have shown that, due to the random insertion of T-DNA, the foreign gene integration sites and target gene copy number in the lines diff ered, resulting in diff erent growth performance of the transgenic lines (Li et al. 2013a,b; Gang et al. 2019a, b). In our present multi-site aff orestation experiments using 16BpGH3.5-transgenic lines and the parent wild-type line of white birch and analysis of growth variables and the interaction between the lines and the environment, we found signif icant line and site eff ect for the growth, and the line × site eff ect was also signif icant. The signif icant eff ects of line, site and their interaction may be mainly due to diff erences in the integration sites of foreign genes and the number of random inserted copies in theAgrobacterium-mediated transformation (Wang et al. 2002). That is to say, the tested transgenic birch lines will have a variety of genetic eff ects due to the diff erences integration sites and insertion copy numbers ofBpGH3.5in their genomes. Thus, each line will express theBpGH3.5to varying degrees, resulting in diff erent growth traits and adaptability and the best-growing lines will diff er among the test sites. In addition, some transgenic lines may experiencetransgene silencing or insertion mutation. For example, G4 is a T-DNA insertion mutant line. The insertion of T-DNA into the promoter region of theBpEIN3 gene cause a premature leaf senescence (Li et al. 2017). The aff orestation test also proved that this line is also the least stable line, ranking last (Table 6).

Table 6 Principal component values and stability parameters of interactions of test lines at each test site

Interaction between genotype and environment

The ultimate goal of the multi-site combined analysis of white birch transgenic lines was to not only to select the lines that grow well in each test site, but also to determine the best transgenic lines for good growth adaptability and stability at each site. This information will serve as a basis for environmental risk assessments ofBpGH3.5transgenic lines for their release for f ield production.

Although regression analysis models have been used to evaluate forest stability, this method is not an independent estimation of environmental effects and genetic effects,and the interaction effects between genotype and environment are not simply linearly superimposed. Therefore,this model is not ideal method for interaction analysis (Li et al. 1998). The AMMI model is an effective model for analyzing regional data of plant species and is widely recognized for analysis of stand stability (Li et al. 2013a, b;Liu et al. 2015). The AMMI model combines the PCA and analysis of variance, which can effectively decompose the interaction effects between genotype and environment,thereby improving the accuracy of the stability analysis of the tested lines (Li et al. 2001, 2013a, b).

Our AMMI model explained 73.59% of the total sum of squares, indicating that the model can accurately evaluate the interaction eff ects between the tested lines and sites. However, the model can only evaluate the stability of each line, not the productivity. Therefore, we combined the AMMI model analysis with a linear regression model analysis to comprehensively evaluate the growth of each line at the three trial sites. The lines were thus divided into four types: high yield and stable, high yield and unstable, low yield and stable, or low yield and unstable. The multipoint combined analysis of transgenic lines can thus determine lines that grow and perform well at each test site to select suitable transgenic lines for productivity trials.For example, for high yield and stable lines FG2, FG3,FG8, FG10, FG12, FG13 and FG31, production trials can be continued at 3 test sites and adjacent areas with similar environmental conditions. For the high-yield and unstable lines, the main focus is on setting up production tests in areas suitable for their optimal growth. For example, the line FG27 can be tested at the Maoershan test site; the FG32 can be tested at the Shidaohe Forest Farm.

Aff orestation preservation rate of transgenic lines in diff erent regions

The aff orestation preservation rate, an indicator of the degree of adaptation of the tested plants to the growth environment and the stability of the transgenic plants, is greatly aff ected by the geographic and climatic conditions. The Ecological Experimental Forest Farm is in a semi-arid area, with an average annual rainfall of only 500 mm and strong winds.It is not a suitable area for white birch, and it has no natural birch forests. As a result, the planting preservation rate for the transgenic birch pilot test at this site was low (55.46%).Even the selected lines FG3 and FG31 that had larger volume growth, had low aff orestation preservation rates (42%and 50%, respectively).

BpGH3.5 antisense chain promotes growth of birch trees

In previous study, we proved that the expression level ofBpGH3.5gene was higher than WT in the overexpression line, and lower than WT in the antisense line (Yu et al.2019). The best lines selected from the above three experimental sites were all trans-antisense strand lines, indicating that the birch lines with low expression of theBpGH3.5gene tend to grow better. The auxin amide synthase gene family(GH3) is a typical auxin primary/early response gene. Most of the family genes can regulate growth and development by adjusting the concentration of free IAA in plants. For example, the hypocotyls and primary roots of theArabidopsis thaliana YDK1/GH3.2overexpressedydk1-Dline were shorter than the control line, and at the same time,its apical dominance was reduced, which was obviously dwarfed (Takase et al. 2004). In addition,GH3gene plays an important role in the process of light signal transmission.TheArabidopsis DFL1/GH3.6-overexpressing linedfl 1-Dhas shorter hypocotyls when grown in the light compared with the wild type, but no signif icant diff erence in hypocotyl length in the dark (Nakazawa et al. 2001). The height of seedlings ofBpGH3.5-transgenic white birch is not signif icantly diff erent from that of WT. However, the primary and lateral roots of the histogenic rooted seedlings transduced with theBpGH3.5positive and antisense chains were signif icantly shorter than those of the wild type, with the average length of their lateral roots being 23% and 36% shorter than that of the wild type, and the content of endogenous IAA is signif icantly lower (Yang et al. 2015a, b). WhenBpGH3.5white birch lines were 7 years old, the tree height, DBH, volume and free IAA content of theBpGH3.5-overexpressing lines (G2, G7, G15, G16, and G20) and antisense lines (FG8,FG9, FG10, FG12, and FG15) was determined again. Results showed that diff erences in the aforementioned traits among the transgenic lines were highly significant (P＜ 0.01).Height, DBH and volume of transgenic lines were higher than the population average, and antisense strands were present in more than 80% of the transgenic lines (Yu et al.2019). Furthermore, the relative expression ofBpGH3.5and the content of free IAA in leaves were measured on f ive overexpression lines and f ive trans-antisense strand lines.By qRT-PCR analysis, the expression ofBpGH3.5in the f ive overexpressing lines was signif icantly higher than in the WT line. On the contrary, endogenousBpGH3.5expression in the f ive trans-antisense strand lines were all signif icantly downregulated. This results indicated that introduction of theBpGH3.5antisense strand in the white birch genome interfered with the expression ofBpGH3.5. The IAA content in theBpGH3.5-overexpressinglines was lower or signif icantly lower than in the WT, whereas the IAA content in the f ive trans-antisense strand lines was signif icantly higher than in the WT (P＜ 0.01); and the mean IAA mean was 52.26%higher than in the WT. This result indicates that the expression ofBpGH3.5is inhibited after theBpGH3.5antisense strand is introduced into the birch genome, which weakens the degree of amino acid conjugation with IAA, and thus the release of more free IAA, thereby promoting birch growth.That is, most of the antisense strand lines grew fast (Yu et al. 2019). Our analysis of the growth in height, DBH,and volume of 7-year-oldBpGH3.5-transgenic birch at the 3 test sites showed that the transgenic lines with good growth stability (high- yield and stability) at all three test sites were allBpGH3.5antisense strand lines, which agrees with our hypothesis. These lines had the best stability, fast growth,and strongest adaptability at the three test sites. They are thus the f irst choices for further production f ield tests at the three sites. Lines G6, G7, G9 and FG5 are low-yield and stability type and varied slightly at the three test sites, but their growth is slow, so they will not be tested in production trials.The results from these valuable newly generated transgenic varieties will guide environmental risk assessments for releasing genetically modif ied lines.

Acknowledgements The authors thank the staff and postgraduate students at State Key Laboratory of Tree Genetics and Breeding(Northeast Forestry University) for their assistance in carrying out the study (with special thanks to Prof. Xiyang Zhao for his guidance).The authors also gratefully acknowledge the support of staff at the Jilin Provincial Academy of Forestry Sciences and Liaoning Poplar Research Institute.