Peach DELLA Protein PpeDGYLA Is Not Degraded in the Presence of Active GA and Causes Dwarfism When Overexpressed in Poplar and Arabidopsis

Controlling the tree size of fruit species such as peach can reduce the amount of labor and input needed for orchard management. The phytohormone gibberellin (GA) positively regulates tree size by inducing degradation of the GA signaling repressor DELLA. The N-terminal DELLA domain in this protein is critical for its GA-dependent interaction with the GA receptor GID1 and the resulting degradation of the DELLA protein, which allows for growth-promoting GA signaling. In this study, a DELLA family member, PpeDGYLA, contains a DELLA domain but has amino acid changes in three conserved motifs (DELLA into DGYLA, LEQLE into LERLE, and TVHYNP into AVLYNP). In the absence or presence of GA3, the PpeDGYLA protein did not interact with PpeGID1c and was stable in 35S-PpeDGYLA peach transgenic callus. The overexpression of PpeDGYLA in both polar and Arabidopsis showed an extremely dwarfed phenotype, and these transgenic plants were insensitive to GA3 treatment. PpeDGYLA could interact with PpeARF6-1 and -2, supposed growth-promoting factors. It is suggested that the changes in the DELLA domain of PpeDGYLA may, to some extent, account for the severe dwarf phenotype of poplar and Arabidopsis transgenic plants. In addition, our study showed that the DELLA family contained three clades (DELLA-like, DELLA, and DGLLA). PpeDGYLA clustered into the DGLLA clade and was expressed in all of the analyzed tissues. These results lay the foundation for the further study of the repression of tree size by PpeDGYLA.


Introduction
Controlling tree size is critical for reducing the amount of labor and input needed for orchard management [1]. Many fruit growers utilize dwarfing rootstocks to control tree size, which also can reduce the length of the juvenile period, achieve a high production efficiency, and allow for mechanized management [2][3][4]. Otherwise, pruning and spraying growth-regulating agents are usually used for orchard management, especially for peach (Prunus persica), which is vigorous in vegetative growth [5]. These horticultural practices used to constrain tree growth are costly in terms of materials, time, and labor. With the rapid development of genetic engineering, the genetic manipulation of tree size through genetic modification may offer a more efficient method than traditional breeding technologies [1]. Therefore, identifying the genes that control tree size will be helpful for peach breeding in the future.

PpeDGYLA Carries an Altered DELLA Motif of DGYLA
In peach, a DELLA family member (Prupe.7G236100) was found to encode a deduced protein of 537 amino acid residues. Multiple sequence alignment showed that this DELLA family member contained a DELLA domain that showed some differences compared with other DELLA proteins ( Figure 1). The DELLA domain is comprised of three conserved motifs (DELLA, LEQLE, and TVHYNP). In this DELLA family member, the three motifs were changed into DGYLA, LERLE, and AVLYNP ( Figure 1). Therefore, this member was named PpeDGYLA. Sequence similarity analysis was conducted using three peach DELLA members (PpeDGYLA and PpeDELLA1/-2) and five Arabidopsis DELLA members (RGA, GAI, and RGL1/-2/-3). PpeDGYLA showed a range of 39.94-44.96 % sequence similarity when aligned with the seven DELLA proteins (PpeDELLA1/-2, AtRGA, AtGAI, and AtRGL1/-2/-3), while the seven DELLA proteins had a range of 50.55-73.59 % when aligned with each other (Table 1). This suggests that PpeDGYLA had a relative low sequence similarity with other DELLA proteins.
DELLA family member contained a DELLA domain that showed some differences compared with other DELLA proteins ( Figure 1). The DELLA domain is comprised of three conserved motifs (DELLA, LEQLE, and TVHYNP). In this DELLA family member, the three motifs were changed into DGYLA, LERLE, and AVLYNP ( Figure 1). Therefore, this member was named PpeDGYLA. Sequence similarity analysis was conducted using three peach DELLA members (PpeDGYLA and PpeDELLA1/-2) and five Arabidopsis DELLA members (RGA, GAI, and RGL1/-2/-3). PpeDGYLA showed a range of 39.94-44.96 % sequence similarity when aligned with the seven DELLA proteins (PpeDELLA1/-2, AtRGA, AtGAI, and AtRGL1/-2/-3), while the seven DELLA proteins had a range of 50.55-73.59 % when aligned with each other (Table 1). This suggests that PpeDGYLA had a relative low sequence similarity with other DELLA proteins.  PpeDGYLA The theoretical isoelectric point (pI) of PpeDGYLA was 5.58 with a calculated molecular mass of 59.16 kDa. PpeDGYLA was predicted to be an unstable hydrophobic protein based on the instability index (II) and the grand average of hydropathicity (GRAVY) (Table S2), and was comprised of an alpha helix, an extended strand, and a random coil with percentages of 46.00 %, 12.48 %, and 41.53 %, respectively (Table S2 and Figure S1).   The theoretical isoelectric point (pI) of PpeDGYLA was 5.58 with a calculated molecular mass of 59.16 kDa. PpeDGYLA was predicted to be an unstable hydrophobic protein based on the instability index (II) and the grand average of hydropathicity (GRAVY) (Table S2), and was comprised of an alpha helix, an extended strand, and a random coil with percentages of 46.00 %, 12.48 %, and 41.53 %, respectively (Table S2 and Figure S1).

PpeDGYLA Was More Stable than PpeDELLA1 after GA 3 Treatment
The DELLA domain mediates the GA-dependent interaction with the GA receptor GID1 and its own GA-induced degradation [16,23]. Multiple sequence alignment showed that several conserved amino acids were changed in the DELLA domain of PpeDGYLA. Therefore, we hypothesized that the GA-dependent interaction with GID1 and the degradation of PpeDGYLA were affected due to the changes in its DELLA domain. Our previous study showed that the GA receptor PpeGID1c could interact with PpeDELLA1 in the presence of active GA, which then induced the degradation of PpeDELLA1 and increased the plant height of peach [5].
In this study, we analyzed the interaction between PpeDGYLA and PpeGID1c and the protein level of PpeDGYLA under GA 3 treatment. In the Y2H assay, yeast containing BD-PpeGID1c and AD-PpeDELLA1 could not grow on SD/-THLA+X-α-gal without GA 3 , but grew normally in the presence of GA 3 . Switching out the DELLA protein, yeast containing BD-PpeGID1c and AD-PpeDGYLA could not grow in the presence or absence of GA 3 (Figure 2A). This suggests that the GA-dependent interaction between PpeGID1c and PpeDGYLA was interrupted.
In this study, we analyzed the interaction between PpeDGYLA and PpeGID1c a the protein level of PpeDGYLA under GA3 treatment. In the Y2H assay, yeast contain BD-PpeGID1c and AD-PpeDELLA1 could not grow on SD/-THLA+X-α-gal without G but grew normally in the presence of GA3. Switching out the DELLA protein, yeast co taining BD-PpeGID1c and AD-PpeDGYLA could not grow in the presence or absence GA3 (Figure 2A). This suggests that the GA-dependent interaction between PpeGID and PpeDGYLA was interrupted.  In peach leaf callus carrying 35S::PpeDELLA1-flag and 35S::PpeDGYLA-flag ( Figure 2B), the protein levels of PpeDELLA1-flag and PpeDGYLA-flag were analyzed using Western blot assay. In the absence of GA 3 , PpeDELLA1-flag and PpeDGYLA-flag were both detected by an anti-flag antibody. In the presence of GA 3 , the PpeDELLA1-flag protein level decreased significantly, while the protein content of PpeDGYLA-flag showed no change ( Figure 2C). These results suggest that PpeDGYLA could not be degraded in the same way as PpeDELLA1 with GA 3 treatment.

PpeDGYLA Overexpression Results in a Severe Dwarf Phenotype in Poplar and Arabidopsis
The high stability of PpeDGYLA in the presence of active GA meant that PpeDGYLA overexpression might have resulted in a more severe dwarf phenotype than when PpeDELLA1 was overexpressed. Because it is very hard to conduct stable transformation of peach, 35S-PpeDGYLA was delivered into poplar, and two transgenic lines highly expressing PpeDGYLA were obtained ( Figure 3C). An empty vector (EV) construct was used as the control. The two transgenic lines showed an extremely dwarfed phenotype compared with the control ( Figure 3A). The leaves of the two transgenic lines were smaller than the control ( Figure 3B). These transgenic plants were transplanted from the WPM medium to soil, and were cultured for another 6 weeks before their height and leaf size were measured ( Figure 3D−F). The plant height of the two transgenic lines were 15 % and 4 % of the control, respectively.
(line 2 and 3) showed a dwarfing phenotype, with line #3 having an extremely dwarf phenotype ( Figure 4). Because of the extremely dwarfed phenotype of line #3, it was d ficult to obtain the seeds and T1 generation plants were used to analyze the phenotyp These results demonstrate that overexpressing PpeDGYLA severely inhibited the grow of transgenic plants.  In Arabidopsis, 35S-PpeDGYLA T1 generation plants were used to analyze the plant height. Similar results were obtained in these Arabidopsis PpeDGYLA transgenic lines. The hypocotyl length and the plant height of the 25-day-old and 35-day-old plants were photographed and measured ( Figure 4). Two Arabidopsis PpeDGYLA transgenic lines (line 2 and 3) showed a dwarfing phenotype, with line #3 having an extremely dwarfed phenotype ( Figure 4). Because of the extremely dwarfed phenotype of line #3, it was difficult to obtain the seeds and T1 generation plants were used to analyze the phenotype. These results demonstrate that overexpressing PpeDGYLA severely inhibited the growth of transgenic plants.

Poplar and Arabidopsis PpeDGYLA-OE Transgenic Plants Were Insensitive to GA 3 Treatment
Our results demonstrated that GA 3 treatment could not induce the degradation of PpeDGYLA. Therefore, we speculated that poplar and Arabidopsis PpeDGYLA transgenic plants might be insensitive to GA treatment. We examined the effect of bioactive GA (GA 3 ) on plant height in PpeDGYLA-OE poplar and Arabidopsis plants. After treatment with 100 µM GA 3 for 7 d, the poplar EV plants were significantly taller than the untreated controls (0 µM GA 3 ), while PpeDGYLA-OE plants showed no obvious change in height ( Figure 5A,B). The same results were observed in PpeDGYLA-OE Arabidopsis after treatment with 1 µM GA 3 for 3 d. After GA 3 treatment, Arabidopsis EV plants had a longer hypocotyl than the control, but PpeDGYLA-OE plants had no obvious change in hypocotyl length ( Figure 6A,B). These results suggest that poplar and Arabidopsis PpeDGYLA transgenic plants were insensitive to GA 3 treatment.

Poplar and Arabidopsis PpeDGYLA-OE Transgenic Plants were Insensitive to GA3 Treatment
Our results demonstrated that GA3 treatment could not induce the degradation of PpeDGYLA. Therefore, we speculated that poplar and Arabidopsis PpeDGYLA transgenic plants might be insensitive to GA treatment. We examined the effect of bioactive GA (GA3) on plant height in PpeDGYLA-OE poplar and Arabidopsis plants. After treatment with 100 μM GA3 for 7 d, the poplar EV plants were significantly taller than the untreated controls (0 μM GA3), while PpeDGYLA-OE plants showed no obvious change in height ( Figure 5A, B). The same results were observed in PpeDGYLA-OE Arabidopsis after treatment with 1 μM GA3 for 3 d. After GA3 treatment, Arabidopsis EV plants had a longer hypocotyl than the control, but PpeDGYLA-OE plants had no obvious change in hypocotyl length ( Figure 6A, B). These results suggest that poplar and Arabidopsis PpeDGYLA transgenic plants were insensitive to GA3 treatment.

PpeDGYLA Interacts with PpeARF6-1/-2
Numerous studies have demonstrated that DELLA proteins repress plant growth mainly by blocking the DNA-binding capacity of growth-promoting factors, such as ARF6, PIFs, and BZR1 [21,28,29]. In this study, we investigated the interactions of PpeDGYLA with PpeARF6, PpePIF8, and PpeBZR1 using Y2H assays. Our previous study showed that PpePIF8 played a positive role in the growth of peach. The homologous genes of ARF6 and BZR1 were identified by constructing phylogenetic trees ( Figure  S2). Two genes related to ARF6 and two related to BZR1 were identified from the peach genome. The full-length PpeDGYLA protein had a transcriptional activation activity, and PpeDGYLA-GRAS (DG-GRAS) was used for Y2H assays (Figure 7). The results showed that PpeDGYLA-GRAS interacted with PpeARF6-1 and -2 in yeast (Figure 7).

PpeDGYLA Interacts with PpeARF6-1/-2
Numerous studies have demonstrated that DELLA proteins repress plant growth mainly by blocking the DNA-binding capacity of growth-promoting factors, such as ARF6, PIFs, and BZR1 [21,28,29]. In this study, we investigated the interactions of PpeDGYLA with PpeARF6, PpePIF8, and PpeBZR1 using Y2H assays. Our previous study showed that PpePIF8 played a positive role in the growth of peach. The homologous genes of ARF6 and BZR1 were identified by constructing phylogenetic trees ( Figure S2). Two genes related to ARF6 and two related to BZR1 were identified from the peach genome. The full-length PpeDGYLA protein had a transcriptional activation activity, and PpeDGYLA-GRAS (DG-GRAS) was used for Y2H assays (Figure 7). The results showed that PpeDGYLA-GRAS interacted with PpeARF6-1 and -2 in yeast (Figure 7).

The Evolutionary Relationship between PpeDGYLA and Other DELLA Members
In this study, our results suggest that PpeDGYLA may be different from the other two peach DELLA family members. Therefore, a phylogenetic tree was constructed using the DELLA members from different land plant taxa, including bryophytes, lycophytes,

The Evolutionary Relationship between PpeDGYLA and Other DELLA Members
In this study, our results suggest that PpeDGYLA may be different from the other two peach DELLA family members. Therefore, a phylogenetic tree was constructed using the DELLA members from different land plant taxa, including bryophytes, lycophytes, gymnosperms, and angiosperms. All DELLA members were divided into three types ( Figure S3). DELLA members from bryophytes and lycophytes clustered into Type 1 and were named DELLA-Like (DL), while DELLA members from gymnosperms and angiosperms were clustered into Type 2 and 3. Type 2 contained DELLA members that mostly contained a DELLA domain starting with the DELLA (DE) amino acid sequence, while Type 3 contained DELLA members that mostly contain DELLA domains starting with DGLLA (DG). PpeDGYLA was clustered into Type 3.
The types of DELLA proteins in each species were analyzed according to the phylogenetic tree ( Table 2). Bryophytes and lycophytes contained just Type 1 DELLA proteins. All of the analyzed gymnosperms and angiosperms contained DELLA members from Type 2. Among the analyzed gymnosperms and angiosperms, there were 12 species that lost the DELLA proteins from Type 3 ( Table 2), including the model species Arabidopsis and poplar.

Transcript Levels of PpeDELLA in Different Tissues
To analyze the tissue-specific expression of PpeDGYLA, eight tissues (different floral organs, leaves, shoot tips, and stems) were collected from peach 'QMH'. qRT-PCR analysis showed that PpeDGYLA was expressed in all eight tissues (Figure 8). Among these tissues, PpeDGYLA transcript levels were highest in the stems and lowest in mature leaves.
The transcript levels of the three peach DELLA family members were compared using the stem, flower, and fruit transcriptome data in the NCBI database (PRJNA723104). The information for these samples was reported in our previous study [30]. The three peach DELLA members were expressed in these samples, and PpeDGYLA showed a higher transcript level in all of the analyzed samples compared with PpeDELLA1 ( Figure S4). This implied, to some extent, that PpeDGYLA plays an indispensable function in the plant growth and development of peach.

Amino Acid Changes in the DELLA Domain of PpeDGYLA May to Some Extent Account for the Severe Dwarf Phenotype of Poplar and Arabidopsis Transgenic Plants
GA promotes plant growth by inducing degradation of the DELLA protein. DELLA is a growth repressor that interacts with growth-promoting factors such as PIF proteins, ARF6 and BZR1 [19][20][21]28]. The DELLA protein is comprised of two domains, the DELLA domain and the GRAS domain. The GRAS domain is responsible for the interaction between DELLA and these growth-promoting factors [31][32][33][34]. PpeDGYLA is highly homologous to other DELLA proteins and contains a conserved GRAS domain and several key alterations with the DELLA domain. Our results demonstrate that overexpressing PpeDGYLA in poplar and Arabidopsis resulted in a dwarf phenotype. A similar result was reported in rice, in which overexpressing SLRL1, a member of the Type 3/DGLLA clade, induced a dwarf phenotype [35]. Moreover, PpeDGYLA could interact with the ARF6 orthologous protein PpeARF6-1/-2 through its GRAS domain. These results suggest that PpeDGYLA is similar to DELLA and plays a negative role in plant growth.
Our previous study showed that overexpressing PpeDELLA1/-2 (two peach DELLA genes) just slightly inhibited plant height (semi-dwarf) in Arabidopsis [27]. In this study, our results show that overexpressing PpeDGYLA resulted in a severe dwarf phenotype in both Arabidopsis and poplar. We hypothesized that the amino acid residue differences within the DELLA domain were responsible for the discrepancy in plant height between the PpeDELLA1/-2 and PpeDGYLA transgenic plants. Our results show that the three conserved motifs (DELLA, LEQLE, and TVHYNP) in PpeDGYLA were changed into DGYLA, LERLE, and AVLYNP. Studies have shown that these three motifs help to mediate the GA-dependent interaction with between the GA receptor GID1 and the GA-induced degradation of DELLA [16,23]. In this study, our results show that the PpeDELLA1 protein, but not PpeDGYLA, could interact with PpeGID1c and be degraded after GA3 treatment. Moreover, poplar and Arabidopsis plants ectopically overexpressing PpeDGYLA were insensitive to GA3 treatment. In Arabidopsis, a 17-amino acid deletion within the DELLA domains of GAI and RGA both caused a GA-unresponsive, severe dwarf phenotype [36]. On the contrary, the dwarf phenotype caused by overexpressing a normal RGA gene could be recovered by GA3 treatment [36]. Interestingly, a single amino acid mutation in the DELLA motif of a grape DELLA protein resulted in a dwarfing phenotype [26]. From apple (Malus hupehensis Redh. var. pingyiensis), the overexpression of Mhgai1 (a mutation of a single amino acid in the DELLA, mutated to   DELLA is a growth repressor that interacts with growth-promoting factors such as PIF proteins, ARF6 and BZR1 [19][20][21]28]. The DELLA protein is comprised of two domains, the DELLA domain and the GRAS domain. The GRAS domain is responsible for the interaction between DELLA and these growth-promoting factors [31][32][33][34]. PpeDGYLA is highly homologous to other DELLA proteins and contains a conserved GRAS domain and several key alterations with the DELLA domain. Our results demonstrate that overexpressing PpeDGYLA in poplar and Arabidopsis resulted in a dwarf phenotype. A similar result was reported in rice, in which overexpressing SLRL1, a member of the Type 3/DGLLA clade, induced a dwarf phenotype [35]. Moreover, PpeDGYLA could interact with the ARF6 orthologous protein PpeARF6-1/-2 through its GRAS domain. These results suggest that PpeDGYLA is similar to DELLA and plays a negative role in plant growth. Our previous study showed that overexpressing PpeDELLA1/-2 (two peach DELLA genes) just slightly inhibited plant height (semi-dwarf) in Arabidopsis [27]. In this study, our results show that overexpressing PpeDGYLA resulted in a severe dwarf phenotype in both Arabidopsis and poplar. We hypothesized that the amino acid residue differences within the DELLA domain were responsible for the discrepancy in plant height between the PpeDELLA1/-2 and PpeDGYLA transgenic plants. Our results show that the three conserved motifs (DELLA, LEQLE, and TVHYNP) in PpeDGYLA were changed into DGYLA, LERLE, and AVLYNP. Studies have shown that these three motifs help to mediate the GA-dependent interaction with between the GA receptor GID1 and the GA-induced degradation of DELLA [16,23]. In this study, our results show that the PpeDELLA1 protein, but not PpeDGYLA, could interact with PpeGID1c and be degraded after GA 3 treatment. Moreover, poplar and Arabidopsis plants ectopically overexpressing PpeDGYLA were insensitive to GA 3 treatment. In Arabidopsis, a 17-amino acid deletion within the DELLA domains of GAI and RGA both caused a GA-unresponsive, severe dwarf phenotype [36]. On the contrary, the dwarf phenotype caused by overexpressing a normal RGA gene could be recovered by GA 3 treatment [36]. Interestingly, a single amino acid mutation in the DELLA motif of a grape DELLA protein resulted in a dwarfing phenotype [26].
From apple (Malus hupehensis Redh. var. pingyiensis), the overexpression of Mhgai1 (a mutation of a single amino acid in the DELLA, mutated to VELLA) or Mhgai2 (DELLA mutated to DELHA) in tomato exhibited GA-insensitive plants with smaller statures and smaller fruits and seeds [37,38]. Together, these results suggest that changes in the DELLA domain of PpeDGYLA stabilized the protein in the presence of active GA and resulted in a GA-insensitive and severe dwarf phenotype.

DGLLA Is an Important Clade in the DELLA Family
Evolutionary analysis showed that ancestral DELLA evolved into three types of DELLA: DELLA-1, -2, and -3 [24]. Our results show that the DGLLA clade was identical to DELLA3. The DGLLA clade family was retained in vascular plants (lycophytes, gymnosperm, monocots, and eudicot), except for the bryophytes [39]. In this study, PpeDGYLA, one member of the DGLLA clade, was expressed in all of the analyzed tissues and had a higher expression level than PpeDELLA1 in the stems, fruits, and flowers. Moreover, poplar and Arabidopsis plant overexpression PpeDGYLA showed a GA-insensitive dwarf phenotype, which is different from the dwarfing phenotype caused by the overexpression of genes in the DELLA1 clade [36]. This implies that DGLLA is an important clade in the DELLA family and may be different from the DELLA1/2 clade in the gene function.
It is worth noting that not all species of eudicots and monocots contained the DGLLA clade. Hernández-García et al. [24] demonstrated that DELLA3/DGLLA was absent in tomato. In this study, our results show that DGLLA members were absent in 12 species among the analyzed eudicots and monocots. This seems to mean that DGLLA was dispensable for these species. However, Li et al. [40] reported two DELLA-like proteins (SlGLD1 and -2) that lacked the DELLA domain and only contained the GRAS domain. Overexpressing SlGLD1 in tomato caused a severe dwarf phenotype. In addition, a common characteristic of DGLLA members from the grass family is a lack of DELLA domain [41]. Together, it was speculated that the function of DGLLA could be carried out by some DELLA-like proteins that lack the DELLA domain.
It is regrettable that the specific roles of DGLLA in plant development and growth are not very clear. Our results show that DGLLA is absent in the well-known model plant Arabidopsis and in tomato and poplar. Therefore, there has been no study until now reporting on the phenotype of plants in which DGLLA is knocked out.

It Is Meaningful to Compare the DELLA Domain between PpeDELLA1/-2 and PpeDGYLA
In an agricultural setting, controlling the tree size at the cultivar level can reduce the amount of labor and input needed for orchard management. Plant growth could be regulated by changing the strength of the interaction between GID1 and SLR1. Eight gid1 mutants with mutations in its coding sequence showed a series of dwarfing phenotypes, with different plant height, in rice [23]. The DELLA domain in the N-terminal portion of the protein was important for the interaction of GID1c and DELLA. The DGLLA member OsSLRL1 had no DELLA domain and resulted in a dwarf phenotype when overexpressed in rice, while overexpressing SLR1-SLRL1 (a fusion protein of the N-terminal portion of SLR1 with SLRL1) resulted in a severe dwarf phenotype of rice [35]. In this study, our results show that several conserved amino acids were changed in the DELLA domain of PpeDGYLA, which abolished the PpeGID1c interacting activity under GA 3 treatment. This implied that the change in these conserved amino acids might have accounted for the interruption of the interaction between PpeDGYLA and PpeGID1c.

Plant Materials and Growth Conditions
The peach cultivar 'QiuMiHong' (QMH) was obtained from the Fruit Tree Germplasm Repository of Henan Agricultural University (Zhengzhou, Henan Province, China). The different tissues collected from 'QMH' were used to analyze the transcript levels of PpeDGYLA. Stigma, calyx, filaments, and petal were from flowers at full bloom from 6-year-old 'QMH' trees. Young leaves, mature leaves, shoot tips, and stems were collected according to Chen et al. [29].
The progeny of Populus deltoides × P. euramericana cv 'Nanlin895 was used for PpeDGYLA transformation and cultured in Woody Plant Medium (WPM) under a 16/8 h light/darkness at 23 ± 1 • C.

Gene Cloning, Sequence Analysis, and Phylogenetic Tree Construction
The total RNA was extracted from leaf samples of 'QMH' trees using an RNA isolation kit (Huayueyang Biotech, Beijing, China) according to the manufacturer's instructions. cDNA was synthesized from RNA using a HiScript III 1 st Strand cDNA Synthesis Kit (+gDNA wiper) (Vazyme, Nanjing, China). The coding regions of the PpeDGYLA, PpeDEL-LLA1, PpeARF6-1/-2, PpePIF8, and PpeBZR1/-2 genes were amplified from the cDNA using specific primers (Table S1). The resulting PCR amplicons were purified and a vector was used for construction.
The amino acid sequences of DELLA-like proteins were downloaded from the Arabidopsis database (TAIR) (https://www.arabidopsis.org/index.jsp) (Accessed on 5 December 2022). All of the protein sequences of different species were downloaded from Phytozome 12 (https://phytozome.jgi.doe.gov/pz/portal.html#) (Accessed on 5 December 2022) or the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/) (Accessed on 5 December 2022) for the construction of the phylogenetic tree. Multiple sequence alignment was performed using DNAMAN. The phylogenetic tree was constructed using the neighbor-joining method of MEGA 6.0 with the default parameters and setting as follows: Bootstrap method, 1000 replicates, and the Poisson model.

Transformation of Poplar and Arabidopsis
The CDS fragment of PpeDGYLA was cloned into the pSAK277 (with 35S promoter) vector to generate a PpeDGYLA-overexpression construct. The recombinant plasmid and empty vector were transformed into wild-type Arabidopsis using the floral dip method [42]. Transgenic Arabidopsis plants were selected and confirmed by qRT-PCR using specific primers of the PpeDGYLA gene. The phenotypic analyses used T1 transgenic plants. The length of hypocotyl and the height of the plants from different PpeDGYLA and SAK277 transgenic lines (EV) were measured and photographed at the age of 7 d, 25 d, and 35 d.
Poplar plants were transformed by infecting leaf discs with Agrobacterium tumefaciens cultures carrying pSAK277-PpeDGYLA [43][44][45]. After co-culturing with Agrobacteria on WPM containing 20 mg/L acetosyringone for 3 d under darkness, the infected leaf discs were transferred onto fresh medium containing 0.5 mg/L kinetin (KT), 0.75 mg/L 2,4-Dichlorophenoxyacetic acid (2,4-D), 400 mg/L cephalosporin, and 50 mg/L kanamycin and 0.7 % (w/v) agar to induce callus. After culturing for about 3 weeks in the dark, the leaf discs with calli were transferred to adventitious-bud-inducing medium containing 0.02 mg/L thidiazuron (TDZ), 400 mg/L cephalosporin, and 50 mg/L kanamycin and 0.7% (w/v) agar. Regenerated shoots were placed in a rooting medium containing 400 mg/L cephalosporin and 50 mg/L kanamycin and 0.7 % (w/v) agar. The rooted plantlets were characterized by qRT-PCR. The positive transgenic plants were acclimatized in pots placed in an artificial climate chamber (NingBo Ledeng Instrument Manufacturing, NingBo, China; 16/8 h light/darkness, 25 ± 1 • C) for 2 weeks and then grown in the culture room (temperature: 25 ± 1 • C; photoperiod: 16/8 h light/darkness). Two transgenic lines (#1, #2) were selected by qRT-PCR for the phenotyping assays. The length of the leaves, the width of the leaves, and the height of the plants from different transgenic lines were measured and photographed at the age of 6 weeks. More than 12 seedlings were used for statistical analysis.

Hormone Treatment
For the poplar experiment, the positive transgenic and empty vector (EV) plants were transplanted into a growth media (peat: vermiculite: perlite = 3:1:1) and grown in the culture room for 2 weeks. Next, the plants were sprayed with 100 µM GA 3 (Solarbio, Beijing, China) or a mock solution three times within 24 h. Seven days after treatment, the phenotype was photographed and the height of the EV and transgenic plants was measured. At least 12 plants were used for every treatment, and the entire experiment was replicated three times.
For the Arabidopsis experiment, PpeDGYLA-overexpressing and EV transgenic Arabidopsis were germinated on MS (Murashige and Skoog, Sigma, Germany) medium for 5 d in an artificial climate chamber (16/8 h light/darkness, 22 ± 1 • C). One group was transferred to new MS medium containing 1 µM GA 3 , another group was transferred to new MS medium containing mock solution. After 3 d of treatment, the phenotypes were photographed, and the hypocotyl lengths of the EV and transgenic seedlings were measured. At least 18 plants were used for every treatment, and the entire experiment was replicated three times. All data were analyzed by ANOVA, Duncan's multiple range tests (at p < 0.05), and Student's t-test using IBM SPSS Statistics 20 (SPSS, Chicago, IL, USA).
The full-length coding regions of PpeDGYLA and PpeDGYLA-GRAS (a deletion of 408 bp in the N-terminal) were cloned into pGBKT7, while that of PpeARF6-1/-2, PpePIF8, and PpeBZR1/-2 were fused into pGADT7. Y2H assays were carried out as above. Transformed yeast cells were grown on SD/-Trp/-Leu (−TL) (Clontech, Mountain View, CA, USA) and SD/-Trp/-His/-Leu/-Ade (−THLA) with X-α-gal media. All of the experiments were replicated three times. All of the primers used for cloning in this study are shown in Table S1.

Analyzing the Stability of PpeDGYLA Protein
The PpeDGYLA sequence was amplified from cDNA of 'QMH' and cloned into the pSAK277-flag vector to produce the PpeDGYLA-flag overexpression construct. Shoots from 2-year-old 'QMH' peach trees were used for the transformation. The shoots were surface sterilized in 75 % ethanol for 1 min and then in 1% sodium hypochlorite for 10 min, followed by washing with sterile distilled water five or six times. The PpeDGYLA-flagoverexpression (OE) peach callus lines were obtained from these shoots according to Xu et al. [46]. The calli were characterized by qRT-PCR.
For GA 3 treatment, the freshly sub-cultured transgenic and control peach calli were cultured on MS agar medium containing 0 µM or 100 µM GA 3 for 48 h. For Western blot, the protein of the treated and control peach calli was extracted using the TCA-acetone method [47]. The protein was detected by immunoblot using anti-Flag antibody (Sigma, Germany) at a 1:4000 dilution. Coomassie brilliant blue staining of the Rubisco small subunit (RbcS) protein was used as the loading control. The Western blot assay was performed according to Wang and Cheng [5,48].

Conclusions
In this study, we identified a DELLA family member, herein named PpeDGYLA, in peach, containing a non-canonical DELLA domain with three conserved motifs (DGYLA, LERLE, and AVLYNP). The PpeDGYLA protein did not interact with PpeGID1c and was stable in 35S-PpeDGYLA peach transgenic callus in the absence or presence of GA 3 . The ectopic expression of PpeDGYLA in polar and Arabidopsis both showed an extremely dwarfed phenotype, and these transgenic plants were insensitive to GA 3 treatment. Among the three clasdes of the DELLA family (DELLA-Like, DELLA, and DGLLA), PpeDGYLA clustered into the DGLLA clade and was expressed in all of the analyzed tissues. Altogether, these results contribute to our understanding of the function of PpeDGYLA and offer potential ways to modulate tree size of peach.