Introduction

The pollen tube is one of the fastest growing organs in flowering plants, which responsible for delivering sperm cells through the pistil to the ovules. During its rapid growth, new components of the cell wall and plasma membrane are continuously deposited at the tip of the growing pollen tube. The role of different structural proteins and polysaccharides such as hydroxyproline-rich glycoproteins (HRGPs) are crucial to maintain the integrity of the pollen tube cell wall (Sede et al. 2020).

HRGPs are typical cell-wall proteins that participate in plant growth, development and immunity (Castilleux et al. 2018). The HRGP superfamily can be classified into three subfamilies: arabinogalactan proteins (AGPs), pro-rich proteins (PRPs) and extensins (EXTs) (Wu et al. 2020). Among the three subfamilies, AGPs have the highest degree of glycosylation. Previous studies speculated that two HRGPs (Pbr036330.1 and Pbr010506.1) might be related to gametophytic self-incompatibility (GSI) in pear (Jiao et al. 2018).

AGPs as one of the major family of glycoproteins in plant cell wall may function in gametophytic cell differentiation, male–female interactions and pollen tube growth (Pereira et al. 2016; Su and Higashiyama 2018). Most of AGPs are predicted to be glycosylphosphatidylinositol (GPI) anchored which is a possible mode of release of AGPs to the extracellular medium where AGPs may exert their functions (Pereira et al. 2016). In Arabidopsis, the expression of AGP6 and AGP11 genes is related to the development of pollen grains and the start of the germination process (Coimbra et al. 2009). In tobacco, AGP components of the stylar extracellular matrix (ECM) directly interact with pollen tubes (Cruz-Garcia et al. 2005). It is indicated that AGPs may be involved in pollen grain formation, initiation of pollen grain germination and pollen tube guidance in the transmitting tract of pistil.

Fasciclin-like AGPs (FLAs) are one of the subfamilies of AGP. To date, genome-wide identification and comparative analysis of FLA gene families have been reported in many plant species, including Arabidopsis thaliana (Johnson et al. 2003), rice (Oryza sativa) (Ma and Zhao 2010), wheat (Triticum aestivum) (Faik et al. 2006), cotton (Gossypium hirsutum) (Huang et al. 2008), Chinese cabbage (Brassica rapa) (Jun and Xiaoming 2012), eucalypt (Eucalyptus grandis) (MacMillan et al. 2015), poplar (Populus trichocarpa) (Showalter et al. 2016), textile hemp (Cannabis sativa) (Guerriero et al. 2017). The members of the FLA proteins possess at least one fasciclin domains that consist of 110 to 150 amino acids (aa) that contain two highly conserved regions (H1 and H2) and one [Phe/Tyr]-His motif (Johnson et al. 2011). This domain has been shown to function as adhesion motifs in extracellular matrix (Kim et al. 2000). Most of FLAs also contain a GPI membrane anchor that allows anchoring the FLA molecule to the cell membrane. Evidence from several studies indicates that FLAs are likely to be important during plant growth and development, such as fiber development (Wang et al. 2015), cell wall regeneration (Showalter et al. 2016), cell-to-cell communication (Jun and Xiaoming 2012), glycan stabilization (Xue et al. 2017) and in response to stresses from cold (Takahashi et al. 2016) and hydrogen peroxide (Zhang et al. 2015). Otherwise, AtFLA3 and AtFLA5 may perform function in pollen development and affect pollen intine formation in Arabidopsis (Li et al. 2010). However, the function of FLA in plant sexual reproduction and the development of pollen tube has never been reported in Rosaseae.

Pear, one of the typical GSI plant in Rosaceae fruit trees, is widely cultivated all over the world (Wu et al. 2013). In this study, we have identified 220 FLA family genes from pear and other six species of Rosaceae, and performed comprehensive analysis on their evolutionary history. Moreover, the expression patterns of PbrFLA10/20/21 in reproductive tissues were investigated to speculate their functional roles in pollen tube growth. Suppressing the expression of PbrFLA10/20/21 resulted in the acceleration of pollen tube growth in pear. Overall, the identification and bioinformatic analysis of FLAs in pear will open the way for studying their biological functions in pollen tube growth and development as well as evolutionary history.

Materials and methods

Plant materials

Pear (Pyrus bretschneideri, ‘Dangshansuli’ variety) trees grown in Fruit Experimental Yard of Nanjing Agricultural University, China were used in this study. Root (from seedling), leaf (from bearing tree), fruit (mature fruit), different developmental stages of pollen tubes, the styles (on the day of flowering) and the styles with different pollination were collected and frozen immediately in liquid nitrogen and stored at −80 °C. Anthers of the ‘Dangshansuli’ variety were collected and dried for one day and then stored in silica gel at −20 °C. Pollen grains were cultured in sealed tube with liquid medium at 24 °C for hydration, germination and growth. The medium components were 0.5 mM Ca(NO3)2, 1.5 mM H3BO3, 450 mM sucrose, 15% (w/v) PEG4000 and 25 mM 2-(N-morpholino) ethanesulfonic acid hydrate, with pH 6.0–6.5 (pH was adjusted with KOH). Pollen tubes longer than their diameters were considered germinated. The length of pollen tubes was measured by Image-Pro software, with approximately 150 pollen tubes measured for length. A great number of pollen grains (0.3 g mature dry pollen per sample) were cultured for different time and used for RNA isolating. The different developmental stages of pollen tubes, mature pollen grains (MP) with 0 min post-cultured, hydrated pollen grains (HP) with 40 min post-cultured, growing pollen tubes (PT) with 6 h post-cultured and stopped-growth pollen tubes (SPT) with 15 h post-cultured, were immediately frozen in liquid nitrogen and kept at -80 °C for RNA extractions (Zhou et al. 2016). As for the styles with different pollination, we collected the styles of the ‘Dangshansuli’ variety at 24 h after self-pollination (SP) and cross-pollination with ‘Cuiguan’ variety (CP), respectively. The styles of ‘Dangshansuli’ variety without pollination as control (UP). The pistils were isolated and stored in liquid nitrogen.

Whole-genome identification of FLA genes in seven species of Rosaceae

The genome sequences of seven species of Rosaceae were downloaded from each genome project (Online Resource 4). Then, the 21 protein sequences of identified AtFLAs were obtained (Schultz et al. 2002). Subsequently, the alignment file for the fasciclin domain (PF02469) obtained from the Pfam database (https://pfam.sanger.ac.uk/) (Finn et al. 2010) was used to build a hidden Markov model (HMM) file. Then, a HMM search was conducted against the seven species of Rosaceae protein database by using HMMER3 software (Eddy 2011). Furthermore, NCBI Batch CD-Search (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) (Marchler-Bauer et al. 2015, 2017) was used to retain the candidate protein sequences based on the existence of the fasciclin domain (smart00554, cl02663, pfam02469). Subsequently, the GPI anchor was identified by big-PI Plant Predictor (https://mendel.imp.ac.at/sat/gpi/gpi_server.html) (Eisenhaber et al. 1999). The ExPASy (https://web.expasy.org/compute_pi) was used to predict the pI (isoelectric point) and MW (molecular weight).

Phylogenetic analysis of the FLA genes

Maximum-likelihood (ML) phylogenetic tree was constructed with all 241 full-length FLA protein sequences of seven Rosaceae species and Arabidopsis by IQ-TREE (Trifinopoulos et al. 2016). The best-fit substitution model VT + F + R7 was determined by ModelFinde (Kalyaanamoorthy et al. 2017), and the bootstrap values were 1000. Evolview (version 2) (Zhang et al. 2012a, b), an online visualization tool, was used to beautify phylogenetic tree.

Chromosomal localization and synteny, Ka, Ks and 4DTv analysis of FLA

Five duplication events, whole-genome duplication (WGD)/segmental duplication, singleton duplication (SD), tandem duplication (TD), proximal duplication (PD), and transposed duplication (TRD), of gene pairs among the seven species of Rosaceae were identified by using the DupGen_finder pipeline (Qiao et al. 2019). Kaks_calculator 2.0 (Wang et al. 2010) was used to calculate the Ka, Ks substitution rates with the Yang and Nielsen method (YN method). ParaAT 2.0 (Zhang et al. 2012a, b) was used to calculate corrected 4DTv values of FLA with the Hasegawa, Kishino and Yano method (HKY method) (Hasegawa et al. 1985). The density plot of the distribution of Ks and 4DTv values was plotted using ggplot2 package in R.

Gene structure, conserved motifs and multiple sequence alignment analysis of the pear FLA genes

Gene structural information for PbrFLA genes was obtained from the pear genome database and displayed using TBtools software (Chen et al. 2020). Conserved motifs of the genes were analyzed by the Multiple EM for Motif Elicitation (MEME) program (Liu et al. 2016) with the following parameters: optimum motif width was set to 30–70, and the maximum number of motifs was set to identify 15 motifs. Multiple sequence alignments corresponding to conserved motifs characteristic of the PbrFLA proteins were determined by Jalview with muscle method (Waterhouse et al. 2009).

Expression pattern analysis of PbrFLA genes by transcriptome sequencing

RNA-seq data of pear pollen (PRJNA299117) were acquired from Sequence Read Archive database (SRA, https://www.ncbi.nlm.nih.gov/sra). Clean reads were obtained by fastp software (Chen et al. 2018a, b). Hisat2 (Kim et al. 2015) was used to align clean reads to the reference genome, and featureCounts (Liao et al. 2014) was used to estimate transcript abundance levels. The transcripts per million (TPM) values were used to measure the expression levels of the PbrFLA genes. The RNA-seq expression profile of pear styles with different pollination (https://figshare.com/s/49b94f0fc5ff1d926afe) was obtained from the previous study (Shi et al. 2017). Then, the expression levels were transformed from fragments per kilobase million (FPKM) to TPM using R code. The pheatmap, a function in R, was used to plot heatmaps based the TPM values of each PbrFLA genes with normalization method of Z-score standardization.

Quantitative real-time PCR analysis (qRT-PCR)

The expression patterns of the PbrFLA genes were analyzed by qRT-PCR analysis. Total RNA was extracted using Plant Total RNA Isolation Kit Plus (FOREGENE Co. Ltd.). Then, the total RNA was adjusted to the same concentration, and based on the adjusted RNA, first-strand cDNA was synthesized by using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech Co. Ltd.). The specific primers were designed by using NCBI online software (https://www.ncbi.nlm.nih.gov/tools/primer-blast), which amplified 38 PbrFLA gene sequences (Online Resource 13). qRT-PCR analysis was performed by using the LightCycler 480 SYBR GREEN I Master (Roche). A 20 μl mixed reaction system was performed, each contained with 150 ng of template cDNA, 0.5 μM of each pair primers and 10 μl of LightCycler 480 SYBR GREEN I Master. All reactions were performed in 96-well plates and each cDNA sample had three replicates. qRT-PCR procedure was set as follows: first 5 min at 95 °C for pre-incubation, 55 cycles at 95 °C for 3 s, 60 °C for 10 s, and 72 °C for 30 s, and then 3 min at 72 °C for extension. Finally, the step of fluorescence signal data collection was carried out at 60 °C. Pyrus UBQ was used as the internal control genes (Chen et al. 2015). The data were analyzed by LightCycler 480 Software v1.5.0 (Roche) using the 2−ΔΔCT method (Livak and Schmittgen 2001). The expression data were analyzed and graphed by R.

Antisense oligo deoxynucleotide experiment

Both phosphorothioate antisense oligo deoxynucleotide (as-ODN) and sense control (s-ODN) were designed using RNA fold Web Server (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi). Soligo software was used to calculate efficient candidate as-ODN sequences for suitable target regions (http://sfold.wadsworth.org/soligo.pl). The antisense oligonucleotide experiment was performed as previously described (Moutinho et al. 2001; Chen et al. 2018a, b). The length of the pollen tubes was measured using a Leica DM6B microscope with LASX software. The length of at least 100 pollen tubes was measured on each condition. The sequences of oligonucleotide primers are listed in Online Resource 13.

Construction of gene co-expression network

To investigate the regulatory network between target genes and transcription factors (TFs), random forest regression was estimated for each gene based on the TFs as inputs using the GENIE3 package in Python3 with default parameters (Huynh-Thu et al. 2010). By the GENIE3 algorithm, we calculated the connectivity between target genes and TFs based on the RNA-seq data (different development of pollen tubes and different pollinated styles). For each co-expressed gene pair, the value of connectivity > 0.005 was extracted (Ramirez-Gonzalez et al. 2018). Networks were visualized by using Cytoscape software (Maere et al. 2005).

Statistical analysis

All experimental data were analyzed by t.test() function in R. Two groups of samples were compared using Student's t-test.

Results

Whole-genome characterization and classification of the FLA genes in seven Rosaceae species

Most FLA proteins have low overall similarity among plant (He et al. 2019); therefore, it is difficult to find family members using Basic Local Alignment Search Tool (BLAST) searches. In order to comprehensively identify the candidate FLA genes in pear genome, we searched for proteins with the fasciclin domain across the whole-genome sequence of pear. The seed file of fasciclin (FAS) (PF02469) from Pfam (http://pfam.janelia.org/) was used to obtain the hidden Markov model (HMM) sequence file; then, HMM searches were performed in HMMER3.0 software against the pear protein database (http://peargenome.njau.edu.cn/). A total of 41 candidate PbrFLA proteins were identified. To further verify the reliability of candidate FLA proteins, NCBI Batch CD-Search was used to detect the completeness of fasciclin domain of candidate proteins. As a result, three candidate proteins without FAS domain were removed. Consequently, a total of 38 non-redundant and complete FLA proteins were identified in pear genome for further analysis. To distinguish members of the FLA family, we named each of PbrFLA genes from PbrFLA1 to PbrFLA38, based on their order on the chromosomes (Online Resource 6). Among these 38 PbrFLA genes, 30 genes could be mapped on chromosomes, and eight genes on the unanchored scaffolds.

Using the same strategies, we identified 33, 28, 33, 37, 31 and 20 FLA genes from apple (Malus domestica), peach (Prunus persica), Chinese plum (Prunus mume), sweet cherry (Prunus avium), strawberry (Fragaria vesca) and black raspberry (Rubus occidentalis) genomes, respectively (Online Resource 4). The lengths of the FLAs protein sequences ranged from 100 to 1197 amino acids. The molecular weights and protein isoelectric points (PI) were ranged from 11.02 to 126.24 kD and 4.11 to 11.10, respectively (Online Resource 6).

To clarify the evolutionary relationship of FLA family in seven Rosaceae species and 21 identified protein sequences of Arabidopsis (Schultz et al. 2002), a ML phylogenic tree for the putative FLA proteins was inferred. Based on the classification of subfamilies in Arabidopsis, a total of 241 FLA proteins from eight species were phylogenetically categorized into four groups, namely class I, class II, class III and class IV. Most FLA genes were contained in class IV (96 of 241), followed by class III (70 of 241), class I (60 of 241) and class II (15 of 241) (Fig. 1, Online Resource 5). Presence and number of fasciclin domains and GPI were also used to interpret the classification (Online Resource 6). As previously described, members of class I possess a single fasciclin domain and most (81.48%) have GPI anchored signal. The smallest group is class II and members contain two fasciclin domains (except Ro07_G07830 and MD07G1214900) but have no C-terminal GPI anchor site. Members of subclass III have either one to five fasciclin domains, and most (70.97%) have a C-terminal GPI anchor site. The remaining FLA proteins constitute subclass IV, which contains FLAs that are quite distantly related to the other FLAs and which have no consistent pattern in the number of fasciclin domains or the presence of a GPI signal.

Fig. 1
figure 1

Maximum-likelihood phylogenetic tree and protein features of FLA proteins from Arabidopsis and Rosaseae. The red, yellow, green and blue indicate the four subfamilies (class I, class II, class III and class IV), respectively. The different shapes and colors indicate different spececies

Full size image

Synteny analysis reveals the dates and driving forces for evolution of the FLA gene family

Duplication events have contributed to the evolution of novel functions, such as the production of floral structures, induction of disease resistance, and adaptation to stress (Panchy et al. 2016). To investigate the origin of FLA family genes, we analyzed five modes of gene duplication, whole-genome duplication (WGD), tandem duplication (TD), proximal duplication (PD), transposed duplication (TRD) and dispersed duplication (DSD), in seven Rosaceae species and identified their contributions to the expansion of the FLA gene family. A total of 267 duplicated gene pairs were found in seven Rosaceae species, with the maximum number of gene pairs derived from DSDs (168 gene pairs), followed by TRDs (33 gene pairs), WGDs (31 gene pairs), TDs (27 gene pairs) and PDs (8 gene pairs) (Table 1). Although a high frequency of DSD-genes was detected, we should not ignore the important roles of other types of single-gene duplication such as WGD and TRD in the expansion of the PbrFLA gene family. Notably, the percentage of gene pairs retained following WGD duplication in pear (20.97%) and apple (23.91%) was higher than that in peach (2.94%), strawberry (3.23%), Japanese apricot (5.56%), sweet cherry (5.26%), and black raspberry (5.00%).

Table 1 Numbers of FLA gene pairs from different duplication events in seven Rosaceae intra-genomics
Full size table

To reveal the evolutionary routes of the FLA gene family, we evaluated distribution of the FLA genes on the chromosomes and both inter- and intra-genomic synteny analyses within seven Rosaceae species. The landscape of 454 inter-species collinear orthologous FLA gene pairs among Rosaceae species presented in Fig. 2, suggesting the strong homology of FLA genes in Rosaceae species. Further, we detected six genes (FvH4_2g30720.1, MD08G1134300, MD15G1109400, Pav_sc0003562.1_g160.1.mk, Prupe.1G461700.1.p, Ro02_G04573) that have collinear relationship with PbrFLA10 and seven genes (FvH4_2g30720.1, MD08G1134300, MD08G1131800, MD15G1109400, Pav_sc0003562.1_g160.1.mk, Prupe.1G461700.1.p, Ro02_G04573) that have collinear relationship with PbrFLA20 (Online Resource 9). After that, intra-genomic collinearity of five duplication events was also investigated in each species, and 62 pairs were found in pear, 46 pairs in apple, 34 pairs in peach, 38 pairs in sweet cherry, 36 pairs in Japanese apricot, 31 pairs in strawberry, and 20 pairs in black raspberry (Fig. 3, Table 1). The chromosome localization of collinear orthologous FLA gene pairs was detected (Fig. 3). In pear and apple, the gene pairs were mainly distributed on chromosomes 3 and 15. In pear and apple, the gene pairs were mainly distributed on chromosomes 3 and 15. In Japanese apricot, the gene pairs were mainly distributed on chromosomes 1. In the rest of the four species, the gene pairs were mainly distributed on chromosomes 6. Thus, the FLA of the seven Rosaceae species was distributed unevenly on chromosomes.

Fig. 2
figure 2

Inter-species collinear orthologous FLA gene pairs among seven Rosaceae species. Black lines along the circumference of the circle mark the gene positions. The lines in different colors inside the circle indicate collinearity relationships among the genes from seven Rosaceae species

Full size image
Fig. 3
figure 3

Intra-genomic collinearity and localization of the FLA genes in seven Rosaceae. The circle in different colors indicate different species and the lines in different colors inside the circle indicate different duplication modes. Whole-genome duplication (WGD), tandem duplication (TD), proximal duplication (PD), transposed duplication (TRD), dispersed duplication (DSD). Intra-genomic collinearity FLA genes pairs in seven Rosaceae species were mapped on the different chromosomes

Full size image

The Ks value is used to estimate the evolutionary history of WGD events (Blanc and Wolfe 2004). In this study, there were two peaks with Ks value at approximately 0.15–0.30 and 1.5–1.8 in pear and apple, respectively (Fig. 4b). In detail, nine WGD-pairs of pear and seven WGD-pairs of apple fell in the range from about 0.15 to 0.30. Simultaneously, three WGD-pairs of pear and three WGD-pairs of apple mapped to the region with values from about 1.5 to 1.8 (Online Resource 8). Furthermore, the recent WGD event with 4DTv (four-fold degenerate site transversion) of ∼0.08 and an ancient WGD event with 4DTv of ∼0.5 (Wu et al. 2013). From the distribution of 4DTv (Fig. 4b, Online Resource 8), there were two peaks with a 4DTv value at approximately 0.08 and 0.5 in pear and apple, respectively. This result further supported the conclusions of Ks analysis, which indicated the FLA gene family had undergone two WGD events in pear and apple. By combining Ks and 4DTv analysis, we concluded that six (PbrFLA26-PbrFLA12, PbrFLA12-PbrFLA31, PbrFLA16-PbrFLA19, MD09G1119100-MD17G1110000, MD13G1028200-MD16G1031100, MD03G1179200-MD11G1198600) and three WGD-pairs (PbrFLA20-PbrFLA11, MD03G1114600-MD04G1100000, MD08G1134300-MD15G1109400) were duplicated during the recent and ancient WGD duplication event, respectively. The Ka/Ks value equals to one showed neutral evolution, positive selection when the Ka/Ks value is greater than one, and purifying selection when the Ka/Ks value is lower than one (Yang 2007). Except Pav_sc0000094.1_g360.1.br-Pav_sc0000094.1_g390.1.br, Pav_sc0001106.1_g220.1.mk-Pav_sc0006377.1_g040.1.mk, Pav_sc0001009.1_g520.1.br-Pav_sc0001640.1_g210.1.br, most Ka/Ks ratios of the FLA gene pairs were lower than one demonstrating FLA primarily evolved under purifying selection (Fig. 4a, Online Resource 8).

Fig. 4
figure 4

The evolution analysis of PbrFLA gene family. a Comparison of Ka/Ks values for different modes of gene duplications. b The distribution of mean Ks and 4DTv value of FLA duplicated gene pairs in pear (red line) and apple (blue line)

Full size image

Gene structural and conserved motif analysis of PbrFLA

Gene structure analysis provides valuable information to interpret the phylogenetic relationship of members within gene family (Li et al. 2009). To identify gene structures and evolutionary trajectories of FLA genes in pear, we investigated exon–intron compositions of the 38 PbrFLAs. Analysis of the genomic DNA sequences showed that none of the PbrFLAs contained intron except that PbrFLA12, PbrFLA19 and PbrFLA29 had one intron (Fig. 5a). As previously reported that introns in FLA genes were lost during plant evolution, especially from green algae to land plants (He et al. 2019). Most members (79%) of class III had two untranslated regions (UTR), whereas no UTR regions were detected from other three classes except PbrFLA14 and PbrFLA30 (Fig. 5a), suggesting that PbrFLAs in class III might have unique and significant biological functions.

Fig. 5
figure 5

The phylogenetic relationship, conserved motifs of PbrFLAs and gene structure of PbrFLAs. a The gene structure of PbrFLA genes, including intron, CDS, UTR. The black line indicates intron; the green rectangle indicates CDS; the yellow rectangle UTR. b The conserved motifs in PbrFLA proteins of pear identified by MEME. Fifteen motifs were indicated by different colors

Full size image

To further provide information to understand the sequence and structure of PbrFLA proteins in pear, MEME tool was used to predict the conserved motifs of PbrFLA proteins. A total of 15 conserved motifs were identified in our study, named from motif 1 to motif 15 (Online Resource 7). As shown in Fig. 5b, most of members of PbrFLA protein contained conserved motifs 1, 4, 6 and 8, while the others were unique to one or few classes. For example, all members of class I contained both motifs 1, 4 and 6; motifs 10 were present only in class II; except for PbrFLA11 and PbrFLA12, the N-terminal and C-terminal of all proteins had motif 1 and motif 8 in class III, respectively; class IV was clearly less closely related to the other subclasses, and motifs 2, 5, 7, 9, 14 and 15 were unique to this class. It is noteworthy that all genes except for class IV had motif 4, suggesting that motif 4 was critical for class I, class II and class III members function.

Multiple sequence alignment of the putative fasciclin domains in 38 PbrFLA proteins were conducted using Jalview software. The predicted fasciclin domains of PbrFLAs possess two highly conserved regions (H1 and H2) of approximately 10 amino acids each, and the [Tyr Phe] His ([Y/F]H) motif locating between these two regions (Online Resource 1). The Thr residue in the H1 region is highly conserved and the other conserved residues such as Val/Ile/Leu and Asn/Asp that may play a role in maintaining the structure of the fasciclin domain and/or cell adhesion (Kim et al. 2000). Small hydrophobic amino acids, such as Leu, Val and Ile, are abundant in the H2 region. In the [Y/F]H motif, His and Pro residues are also relatively conserved. In most cases, H1 and H2 also contain some other conserved residues, such as Val, Ile and Leu. There are about 105 amino acids between H1 and H2. On the contrary, the similarity of amino acid sequence between H1 and H2 is low, though some residues, such as Gly, Asn, Thr and Ser, are found to be conserved in most PbrFLAs.

Expression patterns of the PbrFLA family

The expression patterns of PbrFLAs were investigated in six different tissues of pear by qRT-PCR. Except PbrFLA12/15/35 which showed no expression in any of the six different tissues, 35 PbrFLAs exhibited obviously tissue-specific patterns and were mainly expressed in pollen, root and stem (Fig. 6a). Of note, PbrFLAs in class III were preferentially expressed in the pollen compared to other classes (Fig. 6d). Meanwhile, PbrFLA10/20/21 presented a relatively higher expression level in pollen (Online Resource 10). Subsequently, RNA-seq data were utilized to analyze the expression patterns of PbrFLA during different stages of pollen tube and different pollinated styles of pear (Zhou et al. 2016; Shi et al. 2017). As shown in Online Resource 2, the heatmap of the PbrFLA genes expression levels was hierarchically, and 23 (60.53%) PbrFLAs were expressed in pollinated pistil, and 25 (66.79%) were expressed in pollen, indicating the significant role of PbrFLAs in male–female interactions and growth of pollen tube. Notably, the three genes (PbrFLA10/20/21) have high expression in self-pollinated style (SP) and in stop growing pollen tube (STP) (Fig. 7, Online Resource 11, Online Resource 12). To verify the reliability of the RNA-seq data, nine PbrFLA genes were selected to conduct a qRT-PCR experiment to investigate the expression levels in growing pollen tubes and styles after pollination, respectively. The expression levels of most PbrFLA genes were consistent with the RNA-seq data (Fig. 7).

Fig. 6
figure 6

Expression analysis of PbrFLAs by qRT-PCR in root, stem, leaf, fruit, pollen, and style. The color scale represents the calculated Z-score. Orange indicates high expression. Root (from seedling); leaf (from bearing tree); fruit (mature fruit); pollen (mature pollen grain); style (the style on the day of flowering). a Expression of PbrFLAs in all class. b Expression of PbrFLAs in class I. c Expression of PbrFLAs in class II. d Expression of PbrFLAs in class III. e Expression of PbrFLAs in class IV

Full size image
Fig. 7
figure 7

qRT-PCR and RNA-seq data analysis of the PbrFLA genes in different developmental stages of pear pollen tubes and the styles with different pollination. The right y-axis represents the TPM values of RNA-seq; the left y-axis represents the relative expression levels. a The x-axis represents the different developmental stages of pollen tubes. HP, PT, and SPT correspond to three different developmental stages: HP hydrated pollen grains, PT growing pollen tubes, and SPT stop growing pollen tubes. b The x-axis represents the different pollinated styles. UP, SP and CP correspond to three different pollinated style: UP un-pollinated styles, SP self-pollinated styles, CP cross-pollinated styles

Full size image

Functional analysis of candidate PbrFLA genes in pear pollen tube growth

In order to confirm the contribution of PbrFLA10/20/21 to pear pollen tube growth, we applied antisense oligo deoxynucleotide (ODN) technique (Mizuta and Higashiyama 2014; Chen et al. 2018a, b) in pear pollen. As shown in Fig. 8c, comparing the negative control sense ODN (s-ODN), antisense ODN (as-ODN) was able to downregulate the expression levels of PbrFLA10/20/21. The specific knockdown of PbrFLA10, PbrFLA20 and PbrFLA21 led to a significant acceleration in pollen tube length comparing to the control groups (control, cytofectin and s-ODN) (Fig. 8a, Fig. 8b). It is suggested that pear pollen tube growth might be regulated by PbrFLA10, PbrFLA20 and PbrFLA21.

Fig. 8
figure 8

Knockdown of the expression level of PbrFLA10/20/21 caused the acceleration of pear pollen tube growth. a Pollen tube growth when treated with as-ODN and control groups (control, cytofectin and s-ODN). Bars equal 100 μm. b Treatment with PbrFLA10/20/21 antisense oligonucleotides mixture (as-ODN) resulted in the acceleration of pollen tube growth. s-ODN was transfected with sense oligonucleotides in the presence of cytofectin for the negative control. **p < 0.01 and *p < 0.05 by Student's t-test. Data are the mean ± s.e.m., n = 3 biological replicates. c Expression levels between control groups and as-ODN-mediated knockdown of PbrFLA10/20/21 expression

Full size image

Gene co-expression network analysis

To further investigate the expression regulation of the PbrFLA10, PbrFLA20 and PbrFLA21 in pear, a co-expression network of the FLA genes was constructed. First, we obtained all the transcription factors (TFs) in pear from PlantTFDB (http://planttfdb.cbi.pku.edu.cn/). Then, based on the RNA-seq data, we calculated co-expression relationship between the target genes (PbrFLA10/20/21) and TFs by GENIE3 algorithm that exploits variable importance scores derived from random forests to identify the regulators of each target gene (Huynh-Thu et al. 2010). 50 pairs between PbrFLA10/20/21 and TFs have co-expression relationships (Online Resource 14). Gene pairs were used to construct the co-expression network (Online Resource 3). As a result, there were seven, five and four TFs that revealed a strong co-expression relationship with PbrFLA10, PbrFLA20 and PbrFLA21, respectively. According to the connectivity values, Pbr005306.1 (C2H2), Pbr000646.1 (EIL) and Pbr001365.1 (MYB_related) have the strongest co-expression correlation with PbrFLA10, PbrFLA20 and PbrFLA21, respectively.

Discussion

As crucial role of the plant cell development, some members of the FLA gene family have been extensively studied in plants. However, there have been few such efforts to annotate the FLA gene family in Rosaseae. In this study, we have identified 220 FLA genes which encoded a conserved fasciclin domain in pear and other six species of Rosaceae. Except for black raspberry, the number of FLA genes in the Rosaceae was much higher than those reported in Arabidopsis thaliana. According to the classification of FLAs in Arabidopsis (Johnson et al. 2003), phylogenetic analysis classified the Rosaceae FLA proteins into four subgroups (class I, class II, class III and class IV), suggesting that the evolution of different subfamilies was relatively independent. Previous studies have reported that class III genes may play crucial roles in strong growth tissues, such as male gametophyte (He et al. 2019). For example, AtFLA3 was involved in microspore development, and its knock-down plants showed reduced female fertility (Li et al. 2010). MTR1 regulates male reproductive development in rice (Tan et al. 2012). AtFLA1 might play an important role in root development (Johnson et al. 2011).

WGD event, important for plant adaptability in stress, is an extreme mechanism of gene duplication that leads to a sudden increase in both genome size and the entire gene set (Panchy et al. 2016). TRD event is presumed to arise through distantly transposed duplications occurred by DNA-based or RNA-based mechanisms, and is significant for plant to obtain new functions or to partition existing functions of ancestral genes (Wang et al. 2012). DSD events happen through unpredictable and random patterns by mechanisms that remain unclear, generating two gene copies that are neither neighboring nor collinear (Ganko et al. 2007). In this study, although a high frequency of DSD-genes was detected, WGD and TRD events were also the crucial force of expansion for FLAs in the seven Rosaceae species. Especially, WGD duplication played an important role in the expansion of the FLA gene family in pear and apple. Based on previous reports in our lab, the genome of pear and apple has undergone two rounds of duplications: the recent WGD duplication occurred approximately 30–45 MYA (Ks ~ 0.15–0.30) in pear and apple, while the ancient WGD from γ triplication event occurred approximately ~ 140 MYA (Ks ~ 1.5–1.8) (Wu et al. 2013). Our results of Ks and 4DTv values of FLA paralogs showed that the FLA gene family had undergone the two WGD/segmental duplication events, suggesting that FLA gene family was of early origin and differentiated by adaptation to different types of environment in the evolution process in pear and apple. The Ka/Ks ratios of all duplicated FLA pairs were less than one (except three gene pairs in sweet cherry), which implied that the FLA is mainly undergoing purifying selection.

In general, the gene expression patterns can provide important clues to predict the gene function. Except for the function of AtFLA3 and MTR1 in reproductive organs and AtFLA1 in root. AtFLA8 and AtFLA10, two class III FLAs, are specifically expressed during seed development (Costa et al. 2019). Some EgrFLAs were also highly expressed in stems (MacMillan et al. 2015). In our study, qRT-PCR expression profiling was conducted to explore PbrFLA gene expression patterns in six different tissues. This result indicated that most PbrFLA transcripts were detected in transcriptional abundance in root and pollen, suggesting these genes might have important roles in tissues undergoing growth. Comparing other tissues, 10 PbrFLAs that most belonged to class III were differentially high expressed in the pollen, suggesting class III of FLAs might be related to the development of reproductive organs.

Previous study reported that FLAs containing single-FAS domain may have conserved roles in secondary cell wall biology and properties (MacMillan et al. 2015). In Arabidopsis, AtFLA3, which also contains a single-FAS domain and only one region of arabinogalactan addition sites (a class III AtFLA), was required for cellulose deposition in the intine layer of pollen grain (Li et al. 2010). The phylogenic tree and conserved motifs showed that PbrFLA10, PbrFLA20 and PbrFLA21 were grouped into the same subclass with AtFLA3 (Fig. 1), contained highly conserved domains (Fig. 5) and shared similar gene structure and protein lengths (Table 2). Thus, we explored their expression patterns in differently stages of pollen tubes and in styles with different states of pollination. PbrFLA10, PbrFLA20 and PbrFLA21 were abundantly expressed in the self-pollinated styles (SP) and stop growing pollen tubes (STP) (Fig. 7). Repression of PbrFLA10/20/21 resulted in the acceleration of pear pollen tube growth. Based on previous researches that FLA function exists a genetic redundancy (MacMillan et al. 2010). We speculated that PbrFLA10/20/21 might be functional redundancy and play roles in pollen tube growth by affecting cellulose deposition. The high expression of PbrFLA10/20/21 seems to be involved in fundamental self-incompatibility response mechanisms. More detailed analyses are required to confirm what types of metabolism are affected and if full loss of function of the PbrFLA10/20/21 locus will lead to a more severe defect on pollen tube function.

Table 2 Features of FLA genes identified in pear
Full size table

MYB, C2H2 and EIL transcription factors play important roles in reproductive tissues development. In Arabidopsis thaliana, AtMYB26 (Steiner-Lange et al. 2003), AtMYB103 (Higginson et al. 2003), AtMYB32 (Preston et al. 2003) and AtMYB80 (Phan et al. 2011) are reported to be involved in regulation of anther or pollen development. In apple, MdMYB39L is a key role in sorbitol‐modulated stamen development and pollen tube growth by regulating its downstream target genes (Meng et al. 2018). In Chinese cabbage, BcMF20, a C2H2-type zinc-finger transcription factor, may transcriptionally regulate tapetal-specific genes involved in pollen wall development, which then regulate pollen wall material secretion and subsequent wall development and tapetal programmed cell death (Han et al. 2018). In tobacco, TEIL (Tobacco EIN3-Like) gene is a tobacco homologue of Arabidopsis Ethylene Insensitive 3 (EIN3) and participates in the regulation of flower shape development (Hibi et al. 2007). Combined with the connectivity values of the gene co-expression analysis (Online Resource 3, Online Resource 14), we predicted that Pbr005306.1 (C2H2), Pbr000646.1 (EIL) and Pbr001365.1 (MYB_related) were the most likely to regulate PbrFLA10, PbrFLA20 and PbrFLA21, respectively.

Conclusion

In conclusion, this work presented the phylogenetic relationships, evolutionary history, and expression patterns of PbrFLA family genes. Functional validation indicated that PbrFLA10/20/21 could affect the growth of pear pollen tubes, suggesting their important roles in the growth of the pear pollen tube. This work was able to help us to understand the evolution of the FLA gene family and provided a foundation for future molecular biology on FLA genes during pollen tube growth.