Genome-wide analysis of BURP genes and identi cation of a BURP-V gene RcBURP4 in Rosachienesis

Lufeng Fu Qingdao Agricultural University Zhujun Zhang Qingdao Agricultural University Hai Wang Beijing Forestry University Xiaojuan Zhao Qingdao Agricultural University Lin Su Qingdao Agricultural University Lifang Geng Qingdao Agricultural University Yizeng Lu Shandong Provincial Center of Forest Tree Germplasm Resources Boqiang Tong Shandong Provincial Center of Forest Tree Germplasm Resources Qinghua Liu Qingdao Agricultural University Xinqiang Jiang (  jiangxinqiang8@163.com ) Qingdao Argricultural University https://orcid.org/0000-0003-0727-3354


Introduction
The environments where plants live are ever-changing and adverse conditions can in uence plant growth and development. The unfavorable environmental conditions include internal and external stimuli, such as heat, drought, cold, nutrient de ciency, and excess salt or toxic metals in the soil (Zhu, 2016). Of these, drought stress serves as a main environmental factor, affecting plant natural distribution, limiting plant productivity, and threatening food security (Rampino, 2006). In response, plants have evolved a complicated network involving stress-related pathways that consist of regulatory proteins such as protein phosphatases, protein kinases, and transcription factors, as well as functional proteins including chaperones and signaling components (Zhu, 2002;Broun, 2004;Bohnert et al., 2006;Hirayama & Shinozaki, 2010). Of the two categories of stress-related proteins, functional proteins play a vital role.
The BURP proteins are plant-speci c and known as functional proteins, which play an important role in growth, development, and stress responses of plants (Phillips et al., 2017). The BURP domain, comprised of conserved amino acid sequences, is initially named after BNM2, USP, RD22, and PG1β (Hattori et al., 1998). The BURP proteins contain three domains, including a hydrophobic domain at the N-terminus, a Cterminal BURP domain, and a variable internal region speci c to individual members. The motifs of Cterminal amino acids in BURP proteins are highly conserved, including certain conserved amino acids and four repeat cysteine-histidine (CH) motifs, namely, CHX10, CHX23-37, CHX23-26, and CHX8W (X = any amino acid residue) (Sun et al., 2019). The BURP genes have been investigated among numerous plants, including alfalfa (Li et al., 2016), cotton (Sun et al., 2019), rice (Ding et al., 2009), poplar (Shao et al., 2011), sorghum (Gan et al., 2010), soybean (Xu et al., 2010), and maize (Gan et al., 2010). Phylogenetic analysis further divided BURP proteins into various subfamilies including BNM2-like, USP-like, RD22-like, PG1β-like, and others.
Many BURP proteins have been identi ed and classi ed according to their sequences; however, the knowledge of their functions remains scarce. Previous studies have revealed that BURPs may play a vital role in growth, development, and response to abiotic or biotic stress of plants. A fairly large number of BURPs were proven to participate in plant response to environmental stimuli. AtRD22, a droughtresponsive BURP gene in Arabidopsis, is upregulated by drought and involved in abscisic acid (ABA) signaling and biosynthesis pathways. The AtRD22 expression is proven to be regulated via the interaction of AtMYB2 and AtRD22-BP1 (Abe et al., 2003). The biosynthesis of proteins participating in ABAmediated gene expression is vital in AtRD22 response to drought stress (Yamaguchi-Shinozaki et al., 1993). AtUSP1 is also suggested to be a suppressor in ABA-mediated response to moisture stress (Van Son et al., 2009). In A. thaliana, drought stress response is suppressed by proteins encoded by AtRD22 and AtUSPL1 belonging to the BURP gene family (Harshavardhan et al. 2014). SBIP-355 of tobacco, a homologous gene to AtRD22, may participate in plant defense via the salicylic acid (SA) pathway (Almazroue et al., 2014). BnBDC1 is speci cally expressed in the shoot of Brassica napus; it is induced by ABA, mannitol, and NaCl, while suppressed by SA and UV irradiation (Yu et al. 2004). BgBDC1-4 of Bruguiera gymnorrhiza, with sequences similar to AtRD22, are upregulated by drought, ABA, and high salt treatments (Banzai et al. 2002). The above ndings suggest that BURPs possess critical functions in response to abiotic stimuli and may participate in phytohormone signaling pathways, such as ABA and SA.
Rose (Rosa spp.), as one of the most important horticultural and economic crops, is characterized by high commercial values in pharmaceutical and perfume industries (Raymond et al. 2018). Rose plants are usually in uenced by adverse conditions (Jiang et al. 2009), among which drought is the main factor affecting the growth, development, and survival of rose plants. Drought stress also restricts the fully utilizations of rose plants, especially in arid or semi-arid areas. The rose genome has been sequenced; however, a comprehensive investigation of BURPs in the rose genome has not been performed. In this study, we carried out a genome-wide identi cation of RcBURPs and investigated their structures, chromosomal localizations, cis-regulatory elements (CREs), and phylogeny. In addition, we overexpressed RcBURP4 in Arabidopsis and examined its role under ABA, NaCl, polyethylene glycol (PEG), and drought stress. We also silenced RcBURP4 in rose using the virus-induced gene silencing (VIGS) method. The results obtained from this study reveal the critical role of RcBURPs in rose response to drought stress and unravel the molecular mechanism of RcBURP4, providing a theoretical foundation for future studies on the role of RcBURPs.

Identi cation and sequence analysis of RcBURPs
The hidden Markov model (HMM) pro le of the BURP domain (PF03181) was downloaded from the Pfam database (Finn et al., 2016). Putative RcBURP proteins were searched using HMMER 3.0 with a threshold of 1e − 10 against the R. chinensis (Raymond et al., 2018) and R. multi ora genomes (Nakamura et al., 2017). The online program Expasy (https://web.expasy.org/compute_pi/) was used to determine the molecular weight (Mw) and isoelectric point (pI) of identi ed RcBURP proteins. The subcellular localizations and potential signal peptides of RcBURPs were predicted by using CELLO v2.5 server (http://cello.life.nctu.edu.tw/) (Yu et al., 2006) and SignalP 4.0 server (Wilkins et al., 1999), respectively.

Analyses of gene structure and conserved motifs
The structure (intron/exon) of the RcBURP genes was predicted by TBtools (Chen et al., 2020) based on the comparison between the coding sequence (CDS) and the genomic sequence. The online MEME program (Bailey et al., 2009) was used to analyze the conserved motifs of RcBURP proteins; the parameter settings were: the occurrence rate of a single motif was no greater than one per sequence; the motif width was between 10 and 300 amino acids; the maximum number of identi ed motifs was 10; all other parameters were default.
Chromosomal distribution, gene duplication, and syntenic analyses Chromosomal distributions of RcBURPs were determined by searching against the rose genome database with Circos (Krzywinski et al., 2009). Gene duplication events were examined with MCScanX (Wang et al., 2012) with default parameters. The syntenic relationships among the orthologous BURPs obtained from R. chinensis, Fragaria vesca, and Malus domestica were generated using Dual Systeny Plotter (Chen et al., 2020). KaKs_Calculator v2.0 (Chen et al., 2020) was adopted to calculate the nonsynonymous (Ka) and synonymous substitution rate (Ks) of each duplicated BURP gene, and the results were visualized in TBtools.

Transcriptomic analysis
The expression pro les of RcBURPs have been generated from our previous study (Li et al., 2020). Gene expression was measured by the reads per kilobase per million mapped reads (RPKM) value for each RcBURP. The expression level was normalized by log2 and visualized in TBtools.

Vector construction and Arabidopsis plant transformation
The full-length CDS of RcBURP4 was cloned and recombinated into the pCAMBIA1300 vector controlled by the cauli ower mosaic virus (CaMV) 35S promoter. The binary vector was introduced into the Agrobacterium tumefaciens strain GV3101, which was then transformed into A. thaliana (Col-0) using the oral dip method (Clough and Bent 1998). Seeds of transgenic A. thaliana were collected and sown on Murashige and Skoog (MS) medium containing 50 mg L − 1 hygromycin. Surviving plants were obtained, and the RcBURP4 expression was con rmed with qRT-PCR using speci c primers (Supplementary Table 1). Three independent transgenic lines (OE# 1, 2, and 3) were used for further analyses.
For root growth phenotyping, the sterilized seeds from VC and RcBURP4-OEs were placed on MS solid media. Seeds were vernalized at 4℃ for 3 d, and then transferred to normal growth conditions (light intensity: 100 mol m − 2 s − 1 ; relative humidity: 40-65 %; 23 ± 2 ℃; 16/8 h light/dark) for 7 d. One-week-old seedlings were transferred to MS plates supplemented with NaCl (0, 100, or 200 mM) or ABA (0, 30, or 50 µM). After incubation for 10 d, the primary root length and lateral root number were measured by ImageJ (http://rsbweb.nih.gov/ij/). Six plant individuals of each treatment was used, and the experiments included at least three replicates.
In situ histochemical localization of hydrogen peroxide (H 2 O 2 ) and superoxide anion (O 2 − ) The H2O2 and O2 − levels were determined in one-week-old seedlings of RcBURP4-OEs and VC by diaminobenzidine (DAB) and nitroblue tetrazolium (NBT) staining (Shi et al., 2010 Determination of chlorophyll, malondialdehyde (MDA), and superoxide dismutase (SOD) contents One-week-old seedlings of RcBURP4-OEs and VC were transferred into a tube containing 50% (w/v) PEG6000 or 200 mM NaCl, with H 2 O used as control. The tubes were then vacuum-in ltrated. Chlorophyll content was assayed as described previously by Faragó et al. (2018). MDA content was determined as per Jouve et al. (2007). SOD activity was measured as described by Meng et al. (2014). At least six seedlings from each independent line were included in each experiment.

Drought tolerance assessment
Three-week-old seedlings of RcBURP4-OEs and VC were grown in trapezoidal pots containing a mixture of vermiculite and humus (1:1, v/v) and maintained under long day conditions (16/8 h light/dark). RcBURP4-OEs and VC were grown without watering for 15 d, watered again, and recovered for 7 d; the survival rate was then calculated. Each comparison included more than 20 plants. For water loss assays, leaves of three-week-old seedlings of RcBURP4-OEs and VC were assayed at designated time intervals (light intensity: 100 mol m − 2 s − 1 ; relative humidity: 30-40%; 23-25℃). Water loss rate was calculated as (W0-Wn)/W0, where W0 represents the initial weight of detached leaves and Wn represents the fresh weight of leaves measured at each time point. Three independent biological replicates were conducted for the experiment, with six seedlings from each independent line.

Silencing of RcBURP4 by VIGS
VIGS of RcBURP4 was conducted as per the procedures previously described (Jiang et al. 2014). A speci c 324-bp fragment of RcBURP4 was inserted into the pTRV2 vector (Liu et al., 2002) to generate the TRV-RcBURP4 construct. Discs with a diameter of 1 cm were collected from the center of immature leaves of rose with a hole punch. These leaf discs were immersed into a bacterial suspension containing the tobacco rattle virus (TRV) control (pTRV1 and pTRV2) or TRV-RcBURP4 (pTRV1-RcBURP4 and pTRV2-RcBURP4), and were then vacuum-in ltrated at 0.5 MPa for 30 s. After release of the vacuum, leaf discs were washed and kept in deionized water at 4°C for 3 d, then at 23°C for 1 d. Discs were dehydrated for 12 h and then rehydrated in deionized water for 24 h. The fresh weight and disc area were measured at 12 h dehydration, and 4 h, 8 h, 12 h, and 24 h rehydration. Discs were sampled after 12-h rehydration to obtain the VIGS e ciency by qRT-PCR. Cell counting of samples after 12-h rehydration was performed as per Jiang et al. (2014). The experiments included ve replicates, with at least 16 discs in each replicate.

Statistical analyses
All statistical analyses were performed in SPSS v25.0 (SPSS Inc., USA). Tukey's honestly signi cant difference (HSD) test was conducted to compare the data. The differences at P < 0.05 were considered signi cant.

Gene structure and conserved motifs of RcBURPs
We next examined the gene structure and conserved motifs of the nine RcBURPs and 14 RmBURPs from R. multi ora, which is closely related to R. chinenesis (Fig. 2a). Results of gene structure analyses revealed that the numbers of exon(s) and/or intron(s) were different between members belonging to different subfamilies. As shown in Fig. 2a, RcBURPs contained one to three exons, with most RcBURPs comprising of one (3/9) or two exons (4/9), while two RcBURPs (RcBURP4 and RmBURP9) had three exons. We further investigated conserved motifs of RcBURPs and RmBURPs using the MEME program. Ten conserved motifs were randomly distributed among these BURPs, and most RcBURPs belonging to the same subfamily had similar motifs. Motifs 1-5 were present in all BURPs, except for RmBURP1, RmBURP3, and RmBURP5 (Fig. 2a). Motif 6 and motif 10 were observed in RD22-like, PG1β-like, and BURP-IV subfamilies. In addition, motif 7 was found in both RD22-like and PG1β-like subfamilies. It is worth noting that motif 8 and motif 9 were only present in the PG1β-like subfamily, and members in the PG1β-like subfamily contained all motifs (1-10). Overall, the gene structure and conserved motif composition were similar in members from the same subfamily, which con rmed the reliability of the subfamily classi cation based on phylogenetic analysis.
Three-dimensional modeling was also conducted on the nine RcBURPs (Fig. 2b). All RcBURPs contained multiple α-helix and coil structures (Fig. 2b). In RcBURPs, the most predicted secondary structure was βstrand (9-36% in RcBURPs), whereas the proportion of α-helix was only 8-25%. Results from 3D modeling revealed tertiary structure similarity in these RcBURPs, indicating that they may evolve from the same ancestral sequence and/or be under purifying selection for stabilization during long-term acclimation after the initial divergence.

CRE prediction for RcBURPs
To further investigate the possible role of CREs in the promoter regions of RcBURPs, we used PlantCARE (Rombauts et al., 1998) to identify CREs in the 2000 bp upstream of the TSS in the nine RcBURPs. Based on their functions, the identi ed CREs were classi ed into four groups: abiotic stress-, light response-, phytohormone response-, and plant growth and development-related (Fig. 3a). Phytohormone-responsive CREs included nine types: ABA response elements (ABRE), auxin response elements (AuxRR-core and TGA-element), methyl jasmonate (MeJA) response elements (CGTCA-motif and TGACG-motif), gibberellin response elements (P-box, TATC-box, and GARE-motif), and SA response elements (TCA-element). CREs responding to abiotic stresses contained ve groups, including TC-rich repeats (defense and stress responsiveness), MBS (drought stress), LTR (cold stress), ARE (anaerobic induction), and CCAAT-box (MYB binding site) (Fig. 3a). Additionally, six CREs were identi ed in terms of light response (i.e., GT1motif, G-box, 3-AF3, Sp1, MRE, and ACE). Only a few CREs were identi ed to participate in plant growth and development; no CRE of this kind was found in RcBURP6. Besides, as shown in Fig. 3a, the numbers of G-box (30), ABRE (25), and ARE (24) were greater than those of other CREs. Speci cally, all RcBURPs contained ARE, with the highest number of 5 in RcBURP3 and RcBURP6. Therefore, RcBURPs may participate in rose response to abiotic stresses, light, and phytohormones.

Chromosomal distribution and evolution of RcBURPs
To study the relationship between the expansion and duplication of RcBURPs, all RcBURPs were mapped onto the ve chromosomes of the rose genome (Fig. 3b). Our results showed that the chromosomal distributions of RcBURPs were signi cantly heterogeneous, with most RcBURPs distributed in the distal regions of the chromosomes. Moreover, Chr6 contained four RcBURPs (RcBURP5, RcBURP6, RcBURP7, and RcBURP9), whereas Chr1 contained two (RcBURP1 and RcBURP2), and each of Chr4, Chr5, and Chr7 contained only one RcBURP. To explore the relationship among RcBURPs, gene duplication analysis (e.g., segmental duplication and tandem duplication) was performed by including BURPs from R. chinensis, M.
domestica, and F. vesca (Fig. 3c). In R. chinensis, only one segmental duplication event between RcBURP3 and RcBURP8 was found. Six and seven homologous gene pairs were identi ed among R. chinensis, M. domestica, and F. vesca (Supplementary Table 4), respectively. To further exam the selection pressure in the divergence of BURPs, we calculated the Ka/Ks ratio for the orthologous gene pairs (Supplementary Table 4). The average Ka/Ks ratio was 0.28 for the 15 identi ed orthologous gene pairs, which suggested that BURPs across the species have undergone purifying or stabilizing selection during the evolutionary process.

Expression of RcBURPs and sequence analysis of RcBURP4
To gain insight into the expression pattern of RcBURPs under drought stress, we searched the gene expression data from our previous transcriptomic study (Li et al., 2020). Six treatments included normal conditions (ND) in leaves (NL) and roots (NR), mild drought stress (MD) in leaves (ML) and roots (MR), and severe drought stress (SD) in leaves (SL) and roots (SR). Different expression patterns were observed for the nine RcBURPs during drought stress. All RcBURPs exhibited induced expression levels in rose leaves and roots under the MD condition, while three RcBURPs (RcBURP1, RcBURP3, and RcBURP8) showed repressed expression in leaves under the SD condition. Compared with ND, RcBURP3, RcBURP5, and RcBURP7 showed lower expression levels in rose roots under the SD condition. Interestingly, RcBURP4 showed higher transcript abundance under SR vs. NR, ML vs. NL, MR vs. NR, ML vs. MR, and SL vs. SR, suggesting that this gene may contribute greatly in rose response to drought stimulus. We then selected RcBURP4 for further analysis. Seven BURP-domain containing proteins of six kinds of plants sequence alignments were also performed (Fig. 4b). The results showed that highly conserved amino acids were evenly distributed, including ve prolines, six phenylalanines, an arginine, three threonines, two cystines, three glycines, two glutamic acids, two serines, an aspartic acid, a histidine, two tryptophans, two valines, two lysines, an alanine, a tyrosine, and four CH motifs. The conserved C-terminal sequence of RcBURP4 can be described as CHX 3 YX 3 VX 2 CHX 12 LX 4 GX 6 AXCHX 2 TX 2 WX 3 HX 2 FX 2 LX 3 PX 3 PXCH.

RcBURP4 -OE Arabidopsis was susceptible to ABA during germination and growth
RcBURP4 contained six ABREs and three abiotic stress response CREs (Fig. 3a), indicating that it may contribute greatly in response to drought stimulus through the ABA-dependent pathway. Therefore, we rst overexpressed RcBURP4 in A. thaliana, and three independent RcBURP4-overexpressing lines (OE#1, 2, and 3) with high expression of RcBURP4 were selected for further study. Firstly, we compared the seed germination rates of RcBURP4-OEs and VC on MS media containing different concentrations of ABA (0, 0.5, and 1.0 µM) (Fig. 5a). As shown in Fig. 5b, no signi cant differences were observed in seed germination rates between RcBURP4-OEs and VC without ABA treatment. However, when supplemented with 1.0 µM ABA, the germination rates of OE#1, 2, and 3 were 72.5%, 50.5%, and 39.5%, respectively, signi cantly lower than that of VC (94.5%). Next, to further analyze the root morphology of RcBURP4-OEs under ABA treatment, 7-day-old seedings were placed in MS media containing 0, 30, or 50 µM ABA, and incubated for 10 d (Fig. 5c). The root length increments of OE#1, 2, and 3 were 14, 10, and 9.5 mm, respectively, under 50 µM ABA treatment, which were signi cantly lower than that of VC (3.5 mm; Fig. 5d).
Moreover, no signi cant difference was observed between RcBURP4-OEs and VC in terms of lateral root number (Fig. 5e). In summary, these results suggested that overexpression of RcBURP4 conferred increased ABA susceptibility in germination and post germination stages in transgenic Arabidopsis.

Overexpression of RcBURP4 improved tolerance to salinity in Arabidopsis
To further characterize RcBURP4 function in response to abiotic stress, we investigated RcBURP4transgenic Arabidopsis performance under salt stress. Seeds of RcBURP4-OEs were sown on MS media containing different concentrations of NaCl (0, 100, 150, and 200 mM) (Fig. 6a). The germination rate of RcBURP4-OEs (18.8-41.7%) was signi cantly inhibited after 150 and 200 mM NaCl treatments compared with that of VC (85-95%) (Fig. 6b). We also measured the root length increment and lateral root number of RcBURP4-OEs and VC seedlings under NaCl treatments (Fig. 6c). VC plants maintained higher root length increment under NaCl treatment than did the RcBURP4-OEs lines (Fig. 6d), and no signi cant difference was found between VC and RcBURP4-OEs in terms of lateral root number (Fig. 6e). Moreover, we also tested the accumulated reactive oxygen species (ROS) level of RcBURP4-OEs and VC. Both RcBURP4-OEs and VC plants showed light brown (DAB) and blue (NBT) colors under normal growth conditions. However, when supplemented with 200 mM NaCl, RcBURP4-OEs exhibited darker brown (DAB) and blue (NBT) colors than VC did ( Supplementary Fig. 1). Besides, the H 2 O 2 content (Fig. 6f) and O 2 − production rate (Fig. 6g) of RcBURP4-OEs were 0.16-0.23 µmol g − 1 and 0.48-0.57 nmol g − 1 min − 1 , respectively, signi cantly higher than those of VC. The above results suggest that overexpressing RcBURP4 leaded to decreased ability of plants to scavenge ROS under salt stress.

Performance of RcBURP4-OE lines under osmotic stress
We also investigated seed germination of RcBURP4-OE lines and VC plants under PEG treatment. As shown in Fig. 7a, no signi cant differences were found when plants were grown without PEG6000. However, when supplemented with 12% PEG6000, the germination rates of OE#1, 2, and 3 were 67%, 63%, and 74%, signi cantly higher than that of VC (41%) (Fig. 7b). Additionally, after 50% PEG6000 treatment, brown and blue (Fig. 7c) colors were lighter in RcBURP4-OE lines than in VC plants. Moreover, the H 2 O 2 content and O 2 − production rate of RcBURP4-OEs were 0.11-0.13 µmol g − 1 (Fig. 7d) and 0.33-0.36 nmol g − 1 min − 1 , respectively (Fig. 7e), signi cantly lower than those of VC plants, indicating that RcBURP4-OE lines accumulated less ROS and reduced cell membrane damage under osmotic stress. Besides, RcBURP4-OEs showed higher levels of chlorophyll content (Fig. 7f) and SOD activity (Fig. 7g), but lower levels of MDA content (Fig. 7h) compared to VC plants. The above results demonstrated that RcBURP4 conferred enhanced resistance to osmotic stress in transgenic Arabidopsis.

Overexpression of RcBURP4 increased resistance to drought stress in Arabidopsis
We further examined the performance of RcBURP4-OEs and VC under drought stress. As shown in Fig. 8a, after 15-d drought stress, VC exhibited obvious symptoms of water loss (e.g., withered leaves), while a small amount of green color was observed in RcBURP4-OEs leaves. After 7-d rewatering, a fair number of RcBURP4-OE lines recovered (Fig. 8a), and the survival rates of OE#1, 2, and 3 were 75%, 62.5%, and 87.5%, respectively, signi cantly higher than that of VC (Fig. 8b). Additionally, changes in physiological indices including MDA content were also evaluated. Fifteen days after drought treatment, RcBURP4-OEs exhibited a signi cantly lower MDA content (40.1-52.6 nmol g − 1 ) than VC (87.7 nmol g − 1 ) (Fig. 8c). The water loss rate was lower in RcBURP4-OEs than in VC at the same time points (Fig. 8d). Therefore, these results suggest that the tolerance to drought stress was stronger in RcBURP4-OEs than in VC.
Silencing of RcBURP4 decreased tolerance to dehydration in rose To further reveal the potential role of RcBURP4, we silenced the RcBURP4 expression in rose leaves using the VIGS approach. A. tumefaciens carrying TRV-RcBURP4 and TRV1 (Liu et al., 2002) were co-in ltrated into rose leaf discs to generate RcBURP4-silenced samples. The in ltrated rose leaves experienced 12-h dehydration and 24-h rehydration (Fig. 9a). Firstly, we con rmed the e ciency of VIGS with qRT-PCR. A ~ 65% decrease in the RcBURP4expression was observed in RcBURP4-silenced samples compared to the control (Fig. 9b). Comparing to the control (TRV), rose plants inoculated with TRV-RcBURP4 showed more serious withering (Fig. 9a). Furthermore, silencing of RcBURP4 resulted in signi cantly smaller disc areas during the 12-h dehydration, a reduction of 10.5%, compared to the TRV control (Fig. 9c). In addition, a more signi cant decline was observed in the fresh weight of RcBURP4-silenced discs after the 12-h rehydration compared with the TRV control (Fig. 9d). However, silencing of RcBURP4 led to a greater cell number per microscope visual area of 28% in abaxial, compared to the TRV control (Fig. 9e), indicating that silencing of RcBURP4 reduced tolerance to dehydration in rose.

Discussion
BURPs have been extensively studied in diverse plants (Van Son et al., 2009;Gan et al., 2011;Dinh et al., 2017;Sun et al., 2019), including A. thaliana, Z. mays, Coffea arabica, G. hirsutum, and S. vulgare. Previous studies have revealed that BURPs could be classi ed into eight subfamilies, including RD22-like, PG1β-like, USP-like, BNM2-like, BURP-V, BURP-VI, BURP-VII, and BURP-VIII (Li et al., 2016). The phylogenic tree constructed in this study supported the classi cation of BURPs, with ve distinctive subfamilies being emphasized in our assays (Fig. 1). Our results showed that subfamilies BURP-VI, VII, and VII had close relationships with the RD22-like subfamily, and subfamily USP-like was close to subfamily BURP-V. This is different from results obtained from previous studies (Li et al., 2016), which may be ascribed to the different methods utilized in phylogenetic analyses. Among these subfamilies, the BURP-V subfamily was only present in monocotyledons, whereas the BNM2-like subfamily was only found in dicotyledons, indicating that these BURPs might have originated before the divergence of monocot and dicot plants.
Interestingly, the BURP-V subfamily was only found exclusively in woody plants, such as R. chinenesis, R. multi ora, and G. hirsutum. Taken together, these results suggest that BURPs may share a common ancestor prior to the divergence of higher and lower land plants.
Chromosomal distribution analysis revealed that the nine RcBURPs were not evenly distributed on rose chromosomes; 44.4% of RcBURPs were located on Chr6 and 22.2% were distributed on Chr1 and Chr7, respectively; no RcBURP genes were identi ed on two chromosomes (Chr2 and Chr3). Meanwhile, the BURPs in F. vesca and M. domestica were also unevenly distributed. These results clearly demonstrate a conserved evolution pattern across the Rosaceae plants. Moreover, gene structure and motif analysis of R. multi ora and R. chienesis showed that the BURP proteins contained a highly conserved BURP domain in the C-terminus, especially the four notable CH motifs ( Supplementary Fig. 2). The component and arrangement of motifs identi ed by MEME implied conserved patterns within subfamilies and divergence among subfamilies. Ten motifs were identi ed in subfamily PG1β-like, and more than ve motifs were detected in other subfamilies (Fig. 2a). Meanwhile, members of the BNM2-like subfamily likely missed some motifs in their N-termini. Our results obtained from motif analysis are consistent with those obtained from the gene structure analysis of BURPs.
BURPs have been identi ed in various plants. However, the function of these genes remains largely unknown. Extensive studies have indicated that BURPs may play diverse roles in growth, development, and response to environmental stresses of plants. Members of subfamily USP-like are often related to plant development. In Arabidopsis, AtUSPL1 is expressed in cellular compartments that are crucial for seed and storage protein synthesis in parenchyma cells (Van Son et al., 2009). Members of the RD22-like subfamily are likely to participate in plant response to abiotic or biotic stress. For example, B. napus BnBDC1 is up-regulated by both cold and salt stresses; overexpression of BnBDC1 in Arabidopsis promotes the expression of its downstream genes, conferring enhanced resistance to drought and freezing (Yu et al. 2004). Overexpression of G. max GmRD22 improves tolerance to salt stress in both Arabidopsis and rice. Moreover, GmRD22 is likely to regulate cell wall peroxidases, thereby strengthening cell wall integrity during salt stress (Wang et al. 2012). AtRD22 and AtUSPL1 are induced as part of the ABA-mediated moisture stress response, and their products act to suppress the drought stress response (Harshavardhan et al. 2014). The CaBDP1 expression is highly regulated in C. arabica under cold, drought, ABA, or salt stress. The overexpression of CaBDP1 leads to delayed germination of transgenic Arabidopsis under abiotic stress with the presence of ABA (Dinh et al., 2017), implying CaBDP1 may participate in ABA signaling pathway. The research on the function of members in the BURP-V subfamily is scarce. In our study, overexpression of RcBURP4, a member belonging to subfamily BURP-V conferred higher sensitivity in transgenic Arabidopsis to ABA, salinity, and PEG in germination and post-germination stages. In addition, RcBURP4-silenced rose plants displayed decreased tolerance to dehydration, while Arabidopsis overexpressing RcBURP4 showed improved tolerance to drought at the seedling stage. These results clearly demonstrate that RcBURP4 plays a positive role in plant response to drought stress.

Conclusions
In the present study, BURPs in the rose genome were systematically analyzed. BURPs could be classi ed into eight subfamilies according to results from analyses of phylogeny, gene structure, and conserved motifs. Gene duplication results indicated that segmental duplication may contribute to the expansion of RcBURPs. Arabidopsis overexpressing RcBURP4, a member of the BURP-V subfamily, showed improved sensitivity under high salinity, ABA, and PEG treatments in the germination stage. Moreover, RcBURP4 conferred enhanced tolerance to drought stress in transgenic Arabidopsis. RcBURP4-silenced rose plants displayed decreased tolerance to dehydration. The results obtained from our study expand the understanding of the BURP gene family and provide a candidate gene RcBURP4 for improving rose tolerance to stresses. Figure 1 Phylogenetic analysis of BURPs. Five AtBURPs from Arabidopsis thaliana, 30 GrBURPs from Gossypium raimondii, 17 GaBURPs from G. arboretum, 21 GmBURPs from Glycine max, 14 OsBURPs from Oryza sativa, 18 PtBURPs from Populus trichocarpa, nine RcBURPs from Rosa chinensis, 11 SbBURPs from Sorghum bicolor, and ten ZmBURPs from Zea mays were used for constructing an unrooted tree using MEGA X and visualized by iTOL. Distinctive subfamilies are indicated with different colors. RcBURPs are indicated with red stars.

Figure 2
Phylogenetic tree and predicted three-dimensional (3D) structure of RcBURPs. a Analyses of phylogenetic relationships, gene structures, and conserved motifs of BURPs of Rosa chinensis and R. multi ora. The phylogenic tree was constructed using the neighbor-joining method and BURPs are classify according to their phylogenic relationships. The green rectangles, yellow rectangles, and black lines indicate exons, untranslated regions (UTRs), and introns, respectively. The conserved motifs in BURPs were identi ed by the MEME program. Predicted conserved motifs 1 to 10 are indicated with speci c colors and the consensus sequences for putative motifs are shown in supplementary Fig. 1. b Homology modeling of RcBURPs. Predicted 3D structures of the nine RcBURPs were obtained by Phyre2 homology modeling.   Overexpression of RcBURP4 improved ABA sensitivity in the transgenic Arabidopsis. a Seed germination of RcBURP4-overexpressing lines (RcBURP4-OEs) and vector control plants (VC) under various ABA concentrations. Homozygous T3 seeds of VC and RcBURP4-OEs (OE #1, 2, and 3) were grown on MS plates containing 0, 0.5, or 1.0 μM ABA, incubated at 4 ℃ for 3 d, and maintained at 23 ℃ for germination. Photographs were taken 7 d after planting. b Seed germination rates of RcBURP4-OEs and VC under ABA treatments. c Root growth phenotypes of RcBURP4-OE and VC seedlings exposed to ABA.  Effects of osmotic stress on seed germination and ROS accumulation of  Silencing of RcBURP4 in rose leaf discs. a Leaf disc phenotype of RcBURP4-silenced samples and the TRV control under dehydration and rehydration conditions. Rose leaf discs were in ltrated with Agrobacterium carrying TRV controls (TRV, pTRV1 and pTRV2) or TRVs with a RcBURP4 fragment (TRV-RcBURP4, pTRV1-RcBURP4 and pTRV2-RcBURP4). After the VIGS procedure, leaf discs were dehydrated