The ‘Shatian pummelo’ (Citrus grandis) fruits, leaves, flowers and other tissues were harvested from the Centre of Citrus Plant at Huazhong Agricultural University (Wuhan, China). The pummelo trees were maintained under standard horticultural management and prevention of plant diseases and insect pests. At the popcorn stage, flowers were picked for anther collection, and the anthers were dried in a 28°C oven. The dried anthers and released pollen were collected into 1.5 or 2.0 ml centrifuge tubes sealed with silica gel and stored in -20°C refrigerator for further use. Tobacco plants (Nicotiana benthamiana) were grown in the growth chamber at 23-25°C with 12h light/12 h darkness.

In order to identify the STP genes in C. grandis, the whole-genome data of pummelo (Citrus grandis (L.) Osbeck.cv. ‘Wanbaiyou’ v1.0) was downloaded from the Citrus Pan-genome to Breeding Database website (http://citrus.hzau.edu.cn/index.php), which was used to obtain the gene sequences and gene annotations. The Hidden Markov Model (HMM) of the Sugar_tr domain (PF00083) from HMMER (https://www.ebi.ac.uk/Tools/hmmer/search/hmmscan) was used to search the pummelo protein database at a standard E-value < 1.0 × 10⁻⁵ (Finn et al., 2011; Chen et al., 2020; Mistry et al., 2021). A total of 53 hypothetical CgSTP proteins were identified. Furthermore, the conserved domain composition of 14 AtSTP protein sequences were analyzed by CD-search. Then, all the 53 putative protein sequences filtered to submit to the National Center for Biotechnology Information (NCBI, https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi). Finally, 14 STP family members of C. grandis were screened out. They were named CgSTP1 to CgSTP14 based on their relationship with members of the STP family in A. thaliana.

To explore the evolutionary relationship of CgSTPs between C. grandis and A. thaliana, a phylogenetic tree was constructed by maximum likelihood (ML) method using MEGAX64 software based on the protein sequences of 14 CgSTPs from pummelo and 14 AtSTPs from A. thaliana. The final tree is then beautified through the ITOL website. Among them, the AtSTP protein sequences were downloaded from the The Arabidopsis Information Resource website (https://www.arabidopsis.org/). In order to better understand the conservation of STP genes in evolution, the collinear correlation analysis between species was carried through the MCScanX (Wang et al., 2012) of TBtools software by offering gene annotation and the whole genome sequence of A. thaliana and C. grandis.

The physicochemical properties of the proteins, which included amino acid number (AA), molecular weight (MW), theoretical isoelectric point (PI), and grand average of hydropathy (GRAVY), were obtained in the ExpasyProtParam server (https://web.expasy.org/compute_pi/) (Wilkins et al., 1999). The number of transmembrane domains was predicted by the NovoPro url (https://www.novopro.cn/tools/tmhmm.html). The WoLF PSORT server (https://wolfpsort.hgc.jp/) was applied to predict protein subcellular localization. The gene structure annotation and CDS files of C. grandis were downloaded from the Citrus Pan-genome to Breeding Database website, and the STP genes information was extracted by TBtools using accession number, followed by the Gene Structure Display Server 2.0(http://gsds.gao-lab.org/)to visualize the exon–intron structure of these genes (Hu et al., 2015).

To further explore the gene structure of CgSTPs, the MEME Suite web server (https://meme-suite.org/meme/tools/meme) (Bailey and Elkan, 1994; Bailey et al., 2009) was used to predict their protein sequences with a maximum number of motif groups of 12 for conserved motifs, any number of repeats, and an optimal width of the motif ranging from 15 to 60 amino acids. The promoter sequences (2 kb of genomic DNA sequence upstream of the translation initiation codon) of the CgSTP genes were obtained from C. grandis genome files and submitted to the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) to predict cis-elements in the promoter (Rombauts et al., 1999; Lescot et al., 2002).

Each 0.2g pummelo tissues, including leaves, flowers and juice sac were snap frozen and ground into fine powder in liquid nitrogen with three independent repetitions. RNA from common tissues was extracted using the Ominplant RNA Kit (Cwbio. Jiangsu, China). And for polyphenol polysaccharide containing tissues, RNA was extracted using the RNAprep Pure Plant Plus Kit (TIANGEN, Wuhan, China) according to the instructions. The concentration and quality of the extracted RNA were confirmed via spectrophotometer and agar gel electrophoresis. Then the EasyScript One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China) reverse transcribed RNA into cDNA. Finally, three technical replicates of qRT-PCR were performed using SYBR Green Supermix kit according to the manual via the Applied biosystem^®QuantStudio™ 7 Flex Real-Time PCR System (ABI, Los Angeles, CA, USA), and 2 ^-ΔΔCT method (Udvardi et al., 2008) was used to calculate and analyze the obtained data.

The CDSs of the CgSTP genes were amplified using gene-specific primers without stop codons ( Supplementary Table 1 ). The homologous recombination approach (Rozwadowski et al., 2008) was used to concatenate these target genes with the YFP101 vector, which contained a C-terminal yellow fluorescent protein (YFP) driven by 35S promoter. After the correct comparison between the sequencing results of Tsingke Biological Company (Wuhan, China) and the CDS of the genome through DNAMAN software, the recombinant plasmid was transferred into Agrobacterium tumefaciens GV3101 (Krenek et al., 2015) (Weidi, Wuhan, China). Then, A. tumefaciens containing the recombinant plasmid was injected into tobacco leaves for transient expression (Kato et al., 2002). Finally, the scanning confocal microscope (Leica TCS-SP8, Wetzlar, Germany) was used to image YFP fluorescence with an excitation wavelength of 514 nm and emission at 520-551 nm.

The CDSs of the target gene were cloned with gene-specific primers containing stop codons ( Supplementary Table 1 ). The yeast expression vector pDR196 was ligated with the gene of interest by homologous recombination, followed by transformation of the recombinant plasmids into the hexose transport deficient yeast strain EBY.VW4000 by lithium acetate method (Soni et al., 1993). The empty vector pDR196 was used as negative control, and the recombinant vector pDR196-AtSTP13 was served as a positive control (Riesmeier et al., 1992; Rottmann et al., 2016; Li et al., 2020). The transformed cells in EBY.VW4000 were pre-incubated in liquid Synthesis Defect (SD)-Ura medium supplemented with 2% maltose (w/v) as the sole carbon source until the OD₆₀₀ value reached 0.6-0.8 (Wieczorke et al., 1999). Four serial dilutions (10×) of yeast cells were then plated on solid SD-Ura medium containing 2% maltose or glucose as the sole carbon source. The samples were cultured at 30°C for 3 days and then observed and photographed.

The CDS(s) of CgSTP genes were cloned with gene-specific primers without termination codons ( Supplementary Table 1 ) and inserted into the overexpression vector pK7WG2D with the 35S promoter (Pi et al., 2022). After that, the final recombinant plasmid was transformed into Agrobacterium tumefaciens GV3101, which was further used for injection of tobacco leaves for transient overexpression. The control with empty vector and the recombinant plasmid carrying the target genes (CgSTP(s)-2D) were transformed into the same leaf, as control in the left and CgSTP(s)-2D in the right part of the leaf. Samples were taken on the third day after transformation. The relative gene expression level was determined by RT-PCR assay (the primers were listed in Supplementary Table 1 ). Then, the content of monosaccharides was measured by GC method as described previously (Liu et al., 2022b). Soluble sugars were extracted in 75% methanol with ribitol (0.12 mg per sample) added as an internal standard and then derivatized sequentially with methoxyamine hydrochloride and N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA). After derivatization, the metabolites were analyzed using a GC equipment (Fuli GC-9720 Pluse, Zhejiang, China) with a HP-FFAP column (30.00 m * 0.32 mm * 0.25 μm) and a 5 m Duraguard column (Agilent Technologies, Palo Alto, CA, USA). Sugar content was quantified based on standard curves generated for each sugar (glucose and fructose) and an internal standard. All results were determined at least three biological replicates.

The pollen grains were cultured in liquid germination media (0.02%[w/v]MgSO₄, 0.01%[w/v]KNO₃, 0.03%[w/v]Ca(NO₃)₂, 0.01%[w/v]H₃B0₃, 15%[w/v]PEG-4000, 10%[w/v]Sucrose, and liquid NaOH to pH 6.0) (Liang et al., 2017) for pollen tube length assay. Only 1 μl of transfection agent (Biosharp, China) and 1 μl of transfection agent plus 1 μl of oligonucleotide primers (ODN concentration of 100 μM) were added as the control group, and 1 μl of transfection agent plus 1 μl of antisense oligonucleotide primers were added as the experimental group (Meng et al., 2014). Three independent experiments were performed for each treatment, and the same amount of pollen grains were added for germination culture at 25°C in a constant temperature and dark environment. At the 4 h, 6 h and 8 h of culture, the germinated pollen grains were observed and photographed by a type microscope, and then each biological replicate at least 100 germinated pollen grains were measured via image software. Supplementary Table 1 lists the oligonucleotide primer sequences used in this study. For gene expression assay, individual duplicate pollen tubes were collected into a 10 ml centrifuge tube and centrifuged under 12000g at 4°C for 10 min. The samples could be re-suspended once in pure water and centrifuged again to remove the supernatant under the same conditions, after that the samples were snap frozen in liquid nitrogen for RNA extraction and gene expression assay by qRT-PCR.

Using the Hidden Markov Model (HMM) and the conserved domains of MSF_STP (cd17361) and Sugar_tr (pfam00083) in STPs, a total of 14 STP family members were screened in the C. grandis genome database via the HMMER and Batch CD-Search websites ( Table 1 ; Figure 1 and Supplementary File 1 ). Following the distant genetic relationship with Arabidopsis STP family genes, the 14 pummelo CgSTPs were named as CgSTP1 - CgSTP14 ( Table 1 ). Furthermore, the phylogenetic tree was constructed based on the protein sequences encoded by CgSTP genes in C. grandis and AtSTP genes in A. thaliana via MEGAX64 ( Figure 1B ). The results showed that the STP proteins could be classified into four groups ( Figure 1B ). Among them, tow CgSTPs (CgSTP7 and CgSTP14) together with two AtSTPs from Arabidopsis were attributed to Group I. Group II consisted of four CgSTPs (CgSTP2, CgSTP6, CgSTP8 and CgSTP13), and four AtSTPs. Four CgSTPs (CgSTP3, CgSTP5, CgSTP9 and CgSTP10) clustered with two AtSTPs in Group III. Group IV contained four CgSTPs (CgSTP1, CgSTP4, CgSTP11 and CgSTP12) and six AtSTPs. Furthermore, a total of two sister pairs of CgSTPs were observed in the phylogenetic tree, including CgSTP5 - CgSTP9 and CgSTP1- CgSTP12 ( Figure 1B ). The result indicated that the STP family members are closely related, which is also a good reference for renaming. Moreover, the position and evolutionary relationships of CgSTP members at the chromosomal level were visualized by TBtools based on the gene structure annotation file ( Supplemental Figure 1 ). Interspecific co-linearity analysis of STP genes revealed that eight pairs of genes are orthologous in two species between C. grandis and A. thaliana, indicating that the STP family was widespread in higher plants and strongly conserved during the evolutionary process.

According to the genomic information of C. grandis chromosomes, the distribution of 14 STP genes on the chromosome was analyzed. It was found that all CgSTP members were unevenly distributed on seven chromosomes, with the most abundantly distributed 7 genes on chromosome 9 ( Table 1 and Supplemental Figure 1 ). Protein sequence characteristics, including number of amino acids (AA), molecular weight (Mw), theoretical isoelectric point, grand average of hydropathicity (GRAVY), transmembrane domain (TMD) and subcellular localization prediction, were also analyzed. As shown in Table 1 , the average STP protein length was 504 aa, with the longest 524 aa from CgSTP13 and the shortest 353 aa from CgSTP8. Accordingly, the molecular weight of these proteins ranged from 38.7 kDa (CgSTP8) to 57.9 kDa (CgSTP11). The theoretical isoelectric points ranged from 8.86 (CgSTP11) to 9.56 (CgSTP9), which were all weakly basic. According to the relevant principle of grand average of hydropathicity (GRAVY), CgSTP1, CgSTP7, CgSTP10, CgSTP11, CgSTP12 and CgSTP13 were defined as amphoteric proteins with the value between -0.5 to 0.5, while the other proteins were hydrophobic proteins with more positive value ( Table 1 ). More importantly, the number of CgSTPs transmembrane domains was presented from 8 to 12 based on the prediction by the NovoPro website. The vast majority of CgSTP members had 12 transmembrane domains, which was consistent with the characteristics of the MFS superfamily (Yan, 2013). In addition, the subcellular localization prediction of the protein sequences of CgSTP genes was performed via the WoLF PSORT ( Table 1 ). Most of the CgSTP proteins are predicted to be plasma membrane proteins, but some are vacuolar membrane proteins, including CgSTP3, CgSTP5, CgSTP8 and CgSTP10, which are different from the localization of STP proteins in other species. These results indicate that CgSTP proteins may play different roles in sugar transport relying on their protein characterization and subcellular localization.

To understand the structure of CgSTP genes, exons, introns, and untranslated regions were analyzed via GSDS2.0 and TBtools. It was found that CgSTP1, CgSTP2, CgSTP3, CgSTP4, CgSTP6, CgSTP8, and CgSTP12 didn’t have upstream and downstream UTRs. While CgSTP5 didn’t have upstream UTR, and the other six members all contained upstream and downstream UTRs ( Figure 2B ). Besides, all members contained exons and introns, but differed in specific numbers and locations. The exon number of the CgSTP genes ranged from 2 to 5 ( Figure 2B ). CgSTP13 gene had five exons, whereas CgSTP12 only had two exons. The different of gene structures might contribute to the functional diversity of closely related STP genes. Moreover, 12 putative conserved motifs were predicted through MEME analysis in most CgSTP proteins, and all these motifs were arranged in a stable order in the protein sequence (motif 2, motif 4, motif 10, motif 3, motif 7, motif 5, motif 6, motif 9, motif 12, motif 1, motif 8, and motif 11) ( Figure 2A ). The length of the conserved motifs ranges from 16 to 50 amino acids, and these 12 motifs are contained in all CgSTP proteins, except for CgSTP8 only has 8 motifs ( Figure 2A ; Supplementary Table 2 ). The high uniformity of these conserved motifs fully reflected the relatively conserved function of CgSTP proteins in the evolutionary process.

Cis-elements are important molecular switches involved in the regulation of gene transcription during plant growth and development or abiotic stress response. To detect regulatory factors and predict cis-elements of CgSTPs, each of the 2 kb promoter region (upstream of the start codon of the gene) of 13 CgSTP genes was retrieved from the C. grandis genome sequence. But for the CgSTP2 gene, the identified sequence was shorter than 2 kb (1047bp), since the presence of another gene located less than 2 kb upstream of CgSTP2 start codon site ( Supplementary File 2 ). Finally, 14 cis-elements in the promoter region of CgSTP genes were predicted by PlantCARE database ( Supplementary Figure 2A ). A heat map was further constructed to show the frequency of different cis-elements ( Supplementary Figure 2B ). These predicted cis-elements respond to different phytohormones (Gibberellin, Abscisic acid, Auxin, Methyl jasmonate, Salicylic acid) and various environmental factors, as well as abiotic and biotic stresses. Among them, hormone-related elements were widely presented in the promoter region of CgSTPs, so it was suspected that hormones play a potential role in regulating the expression of CgSTP genes. Besides, the anaerobic induction regulator is presented in the promoters of all genes except CgSTP2, which suggested that these genes might be induced by anaerobic conditions, such as waterlogging stress. In addition, some the other gene-specific cis-elements were also identified in the promoter region of the CgSTP genes, like the MYB binding site involved in drought-inducibility, low-temperature responsiveness, and the cis-acting element involved in defense and stress responsiveness, which provided the possibility to regulate gene expression and respond to various conditions of CgSTPS. Overall, cis-acting element analysis will provide a good reference for further studies on transcriptional regulation of CgSTP members.

The spatiotemporal expression is a critical aspect in determining gene function. Accordingly, three different tissues of leaf, flower and juice sac of ‘Shatian pummelo’ were collected to perform the tissue expression pattern analysis of STP family members via qRT-PCR assay ( Figure 3 ). The results showed that CgSTPs were widely expressed in leaf, flower and juice sac. Among them, CgSTP7, CgSTP9, CgSTP11, CgSTP13 and CgSTP14 had a higher expression level in two or three different tissues ( Figure 3 ). For instance, CgSTP13 was highly expressed in leaf and flower tissues ( Figures 3A, B ), while CgSTP7 was highly expressed in leaf and juice sac tissues ( Figures 3A, C ). CgSTP9, CgSTP11 and CgSTP14 were highly expressed in leaf, flower and juice sac ( Figure 3 ), implying that these members should have a universal functions in different tissues. What’s more, the tissue-limited expression pattern was also presented that CgSTP4 and CgSTP6 were specifically expressed in flower ( Figure 3B ). For some other members were expressed at relatively low levels in all three tissue, such as CgSTP1, CgSTP2, CgSTP8 and CgSTP10, which might be expressed at specific developmental stage or in response to certain stimuli such as abiotic and biotic stresses. Taken together, these expression patterns in different tissues suggested that CgSTPs should have distinct physiology functions throughout the plant.

To investigate the transport function of CgSTPs, the subcellular localization was first identified. According to the expression level, the 6 higher expressed genes (CgSTP4, CgSTP7, CgSTP9, CgSTP11, CgSTP13, CgSTP14) were selected for further analysis. The cell membrane subcellular localization of CgSTPs were observed by co-expressing the YFP fusion proteins and the plasma membrane marker (PM marker) in tobacco leaves ( Figure 4 ), which is consistent with the prediction results of the WoLF PSORT ( Table 1 ). Based on previous reports, STPs had hexose sugar (like glucose and fructose) transport activity (Buttner et al., 2000; McCurdy et al., 2010; Liu et al., 2018; Kong et al., 2022). To characterize the transport properties of the CgSTP proteins, the monosaccharide uptake incompetent yeast mutant strain EBY.VW4000 (only grow on maltose medium) was applied for sugar uptake assay (Wieczorke et al., 1999). For that, the yeast expression vector pDR196 containing CgSTP4, CgSTP7, CgSTP9, CgSTP11, CgSTP13, CgSTP14 were respectively transformed into the EBY.VW4000, and the empty vector and AtSTP13 were used as negative and hexose-uptake positive control in yeast growth assay. Glucose was added as sole carbon source to detect the sugar transport activity of CgSTPs. All yeast cells grew well on the synthesis deficient (SD-Ura) medium containing 2% maltose ( Figure 5 ), indicating the successfully expression of vector in yeast. As expected, negative control yeast cells transformed with empty vector pDR196 did not grow on glucose medium, while positive control yeast cells transformed with CgSTP13-pDR196 grew normally on glucose medium ( Figure 5 ). For yeast cells carrying CgSTP4, CgSTP11, CgSTP13, and CgSTP14 could be grown on SD medium containing glucose, suggesting that these four CgSTP proteins could have glucose transport activity. In addition, the other two CgSTP members (CgSTP7 and CgSTP9) were unable to recovery yeast growth on glucose medium ( Figure 5 ), implying that these two CgSTP members didn’t have or had a weak hexose transport activity. The subcellular localization and the verification of the monosaccharide transport activity of CgSTPs would provide a foundation for further physiological function research.

To verify and analyze the sugar accumulation function of CgSTPs, the genes that have been confirmed to have transport activity and highly expression level in fruit or leaf tissues were selected. The overexpression vectors of pK7WG2D-CgSTP11, CgSTP13, CgSTP14 were constructed under the 35S promoter. After that, the final expression vectors containing CgSTP11/13/14, and the empty vector control were transformed into tobacco leaves for transient overexpression ( Supplementary Figure 3 ). In consideration of the hexose transport activity, the main monosaccharides’ (glucose and fructose) content were determined by GC-FID method (Li et al., 2015b). As shown in Figure 6 , the glucose and fructose contents with the overexpression of target CgSTPs were increased in comparison with the control. For CgSTP11 and CgSTP14-OE samples, both glucose and fructose had a significantly higher level, implying that they did have the function of accumulating monosaccharides in plant. Overall, the transient overexpression assay provide the important clues for further more detail research of CgSTPs in citrus species.

According to the previous report, sugar transport proteins were essential for pollen tube fast growth (Cheng et al., 2015; Rottmann et al., 2018; Li et al., 2020). In this study, it was found that CgSTP4 had the specifically high expression level in flower ( Figure 3 ). Therefore, the expression profile of CgSTP4 in flower and various tissues of flower, including pollen, pollen tube, receptor, style, filament and petal of C. grandis, were further detected by qRT-PCR. The results revealed that CgSTP4 was highly expressed in pollen grains and pollen tubes ( Figure 7A ). Given the importance of sugar absorption for pollen tube growth, we explored the potential role of CgSTP4 in pollen tubes via the antisense oligonucleotide transfection assay (Meng et al., 2014; Meng et al., 2018). Compared with control and sense oligonucleotide transfection, the expression of CgSTP4 can be significantly inhibited by as-ODN treatment during the germination of pollen in ‘Shatian pummelo’ ( Figure 7B ). As expected, the pollen tube length was shorter with the lower expression of CgSTP4 ( Figure 7C ), indicating that CgSTP4 plays an important potential role in the sugar uptake for the growth of pollen tubes in C. grandis.