The apple fruit samples utilized in this study were collected from 10-year-old ‘‘Hanfu’’ apple trees (M. domestica) cultivated at the Shenyang Agricultural University experimental station (123.88°E, 41.88°N). The ‘Orin’ calli in this study were grown on MS medium with 0.8 mg·L⁻¹ 6-BA and 1.5 mg·L⁻¹ 2,4-D, in darkness at 25°C. Nicotiana benthamiana plants used for subcellular localization were cultivated with a 16 h/8 h (light/dark) photoperiod at 25 ± 2°C and a relative humidity of 50%–70%. In this study, tomato (Solanum lycopersicum L. ‘Micro-Tom’) seeds were used for genetic transformation. Both wild-type (WT) plants and transgenic tomato lines were grown in an artificial climate chamber under a 16 h/8 h (day/night) photoperiod with a temperature of 25 ± 2°C/18 ± 2°C (day/night) and a relative humidity of 50%–70%. All collected samples were immediately frozen in liquid nitrogen and stored at -80°C until use.

RNA extraction, cDNA synthesis, and qRT-PCR were performed using our previously described method (Nie et al., 2022). All primers used are presented in Table S1 . Gene expression levels were analyzed using the 2^-ΔΔCT method (Livak and Schmittgen, 2001). The experiment was repeated three times.

The sequence of MdSWEET23 was acquired from the Genome Database for Rosaceae. The cDNA of MdSWEET23 was amplified from ‘‘Hanfu’’ apple fruits. The pRI101-MdSWEET23-GFP recombinant plasmid was constructed for subcellular localization and genetic transformation. The sequence of MdSWEET23 was submitted to the online websites MOTIF search (https://www.genome.jp/tools/motif/), DeepTMHMM (https://dtu.biolib.com/DeepTMHMM), and ProtParam tool (http://web.expasy.org/protaram/) for analysis of conserved motifs, transmembrane domains, molecular weight, theoretical isoelectric point (PI), and grand average of hydropathicity. Separate multiple sequence alignments were performed for MdSWEET23 and 17 Arabidopsis SWEET sequences (AtSWEET1-17) using ClustalW in MEGA X software. Subsequently, a phylogenetic tree was constructed using the Neighbor-joining method (Kumar et al., 2018).

The pRI101-MdSWEET23-GFP fusion protein was transiently expressed in Nicotiana benthamiana and onion epidermal cells as previously described (Zhang et al., 2021). Empty vectors expressing non-targeted GFP were used as controls. Fluorescent signals were observed and photographed using a fluorescent microscope system (NikonNi-E, Nikon, Japan).

An 1863 bp sequence upstream of the start codon (ATG) of the MdSWEET23 gene was amplified from the apple genome DNA. The primers used for this extraction are listed in Table S1 . The sequences of the cloned promoter were submitted to the PlantCARE website for online analysis of cis-acting elements. The PCR product was cloned into the pBI121 vector upstream of the β-glucuronidase (GUS) gene to construct the pBI121-pMdSWEET23 -GUS recombinant plasmid. The cDNA of MdSWEET23 was cloned into the plasmid pBI121-pMdSWEET23-GUS downstream pMdSWEET23 to construct the pBI121-pMdSWEET23 -MdSWEET23-GUS plasmid. The recombinant plasmids were transformed into ‘‘Hanfu’’ apple fruits using the Agrobacterium tumefaciens LBA4404 and the method described by Spolaore et al. (2001). After 3 days of storage in the dark, the injected fruits were sampled for GUS histochemical staining. Subsequently, the tissues expressing GUS were examined and photographed.

Yeast functional complementation assays were conducted following the methods described in a previous study (Loqué et al., 2007). Briefly, the recombinant plasmid pDR196-MdSWEET23 was constructed, with the primers used in the construction process presented in Table S1 . Subsequently, the recombinant plasmid pDR196-MdSWEET23 or the empty pDR196 vector (as a control) was individually introduced into the yeast mutant strains, SUSY7/ura and EBY.VW4000. SUSY7/ura and EBY.VW4000 yeast cells carrying pDR196-MdSWEET23 were inoculated on a solid SD/Ura medium supplemented with different carbon sources. Transformants were incubated at 30°C for 2–4 days and then observed and imaged.

A 210 bp fragment of MdSWEET23 cDNA fragment corresponding to bases 675 bp–885 bp of MdSWEET23 was PCR amplified from ‘‘Hanfu’’ apple fruit cDNA. The PCR products obtained were successfully ligated into the pTRV2 vector. Agrobacterium strain GV3101 was separately transformed with pTRV1, pTRV2, and pTRV2-MdSWEET23. Subsequently, TRV1- and TRV2- MdSWEET23 Agrobacterium strains were co-inoculated into 140 days after bloom (DAB) ‘‘Hanfu’’ apple fruit by infiltrated using a needleless 1 mL syringe at the maximum diameter of apple fruit, with each fruit being injected three times at evenly distributed locations. The fruits injected with pTRV1 and pTRV2 Agrobacterium strains were used as controls. For each treatment, 10 fruits were injected and bagged. After 5 days, RNA was extracted from the samples for qRT-PCR analysis to detect the MdSWEET23 silencing efficiency. Three fruits with high gene silencing efficiency were selected to determine sorbitol, fructose, glucose, and sucrose contents and MdSDH5, MdSDH6, and MdSOT1 expression levels. The experiment was repeated three times.

The pRI101-MdSWEET23-GFP was transformed into ‘‘Orin’’ calli using the Agrobacterium strain LBA4404, and then the transformed calli was cultured on MS medium containing 50 mg·L^-1 kanamycin for transgenic selection. The transgenic lines were confirmed using genomic PCR analysis with specific primers shown in Table S1 . Transcript levels of MdSWEET23 in these lines were examined using qRT-PCR. Samples were collected to determine the sorbitol, fructose, glucose, and sucrose contents. The experiment was repeated three times.

Micro-tomato transformation was conducted using the methods described by Sun et al. (2006). Two transgenic lines of the T3 generation were selected and used for the phenotypic observation. Plant height was determined 60 days after seeding. Net photosynthetic rate (Pn) was determined using the portable photosynthesis system (CIRAS-2, PP systems, USA) at photosynthetically active radiation of 800 μmol·m^-2·s^-1. The leaf chlorophyll concentration was measured with a SPAD-502 meter (Konica Minolta Sensing, Japan) and expressed as SPAD values. Fruits at the mature green (MG), BC, and red ripe (RR) stages ( Figure 1A ) were used for analysis of sugar contents, analysis of sugar-metabolizing enzyme activities, and the expression levels of sugar transporter genes. The experiment was repeated three times.

Seedlings of the WT and two MdSWEET23-overexpressing (OE) transgenic plants were grown under normal conditions for 60 days, followed by cold treatment (3.5°C) for 3.5–4 h. Leaves were collected to measure the levels of relative electrolyte leakage (REL), soluble sugar, and starch. The REL was determined by the method of Hu et al. (2022). The experiment was repeated three times.

The enzymatic activities of cell wall invertase (CWIN), sucrose synthase (SS), cytosolic invertase (CIN), fructokinase (FK), vacuolar invertase (VI), and sucrose phosphate synthase (SPS) were measured following the kit instructions of Suzhou Grace Biotechnology Co. The soluble sugar and starch content were determined by the methods of Zhang et al. (2021). The sorbitol, fructose, glucose, and sucrose contents were determined using high-performance liquid chromatography (HPLC) (1260 series, Agilent Technologies), a method referenced by Li et al. (2020).

All experiments were conducted with three independent biological replicates. Error bars represent the standard deviation (SD) of the three replicates, and values are presented as the mean ± SD. All statistical analyses were performed using IBM SPSS Statistics 23. Additionally, a t-test for independent samples was used for significant differences between the two groups of data, and significant differences were expressed as * P<0.05. One-way analysis of variance was performed on the data using Duncan`s multiple range test, and the results are indicated by letters. Histograms were plotted using Sigma Plot 10.0.

The MdSWEET23 cDNA was cloned from the total RNA of “Hanfu” apple fruits, and it consists of an 897 bp open reading frame that encodes a polypeptide of 298 amino acids ( Table S2 ). The polypeptide has a molecular weight, an isoelectric point, and a grand average hydropathicity of 33.48 kDa, 8.82, and 0.497, respectively ( Table S3 ). MOTIF search and TMHMM analysis showed that MdSWEET23 contained seven α-helical transmembrane domains with two conserved MtN3/saliva motifs ( Figures 2A, B ), which are unique to the typical plant SWEET proteins. Phylogenetic analysis revealed that MdSWEET23 has the highest similarity to AtSWEET13 with a similarity of approximately 50.50% ( Table S4 ), and MdSWEET23 is classified as a member of the Clade III in the SWEET family ( Figure 2C ). Han et al. (2017) reported the 2.8-Å resolution crystal structure of AtSWEET13 in the inward-facing conformation with a substrate analog, 2′-deoxycytidine 5′-monophosphate, bound in the central cavity. Four residues (Val23TM1, Ser54TM2, Val145TM5, and Ser176TM6) inside the binding pocket in AtSWEET13 lead to a larger cavity than monosaccharide transporters, which have Leu, Asn, Met, and Asn in the corresponding sites, and these residues are most likely involved in substrate selectivity. MdSWEET23 has the same residues as AtSWEET13 in the corresponding sites by amino acid sequence alignment ( Figure 2D ). Thus, it is assumed that the transport activity of MdSWEET23 toward sucrose may be higher than that toward glucose.

The promoter regions (1863 bp) of MdSWEET23 were cloned ( Table S5 ), and its cis-acting elements were analyzed using the PlantCARE program. Hormone-, stress-, and light-responsive and organ-specific cis-acting elements were identified in the MdSWEET23 promoter. Thus, it is speculated that the expression of MdSWEET23 may be in response to light, hormones (such as MeJA, auxin, and ABA), and abiotic stresses (such as low-temperature and drought). Phloem tissue-specific cis-acting elements (as-1) were also identified in the MdSWEET23 promoter, which implied that it might be a phloem-specific promoter ( Table 1 ).

To confirm the spatial expression patterns of MdSWEET23, we fused its promoter sequence with the GUS gene to generate two constructs, pMdSWEET23 -GUS and pMdSWEET23 :MdSWEET23-GUS. The GUS activity in transgenic apple fruits was predominantly detected within the vascular bundle region, particularly in the sepal and carpel vascular bundles ( Figure 3 ).

To determine the subcellular localization of MdSWEET23, a MdSWEET23-GFP fusion protein was transiently expressed in both Nicotiana benthamiana and onion epidermal cells. Fluorescence signals from the MdSWEET23-GFP fusion protein were only detected on the plasma membrane in Nicotiana benthamiana and onion epidermal cells ( Figure 4B, D ), indicating that MdSWEET23 is likely localized on the plasma membrane.

The functionality of MdSWEET23 in sugar transport was examined by transforming yeast strains, including EBY.VW4000 and SUSY7/ura3 with the pDR196-MdSWEET23 fusion vector. The yeast mutant strain, EBY.VW4000 displayed the inability to grow on monosaccharides but exhibited growth on maltose (Wieczorke et al., 1999). In contrast, the mutant strain SUSY7/ura3 could not use external sucrose as the sole carbon source (Riesmeier et al., 1992). A spotting assay showed that EBY.VW4000 cells transformed with pDR196-MdSWEET23 and the empty pDR196 vector (as negative control) exhibited comparable growth rates on the culture medium containing 2% (w/v) maltose, galactose, glucose, and fructose ( Figure 5A ). The SUSY7/ura3 cells transformed either with pDR196-MdSWEET23 or the empty vector pDR196 could grow well on 2% (w/v) glucose, whereas pDR196-MdSWEET23 transformants grew better than the negative control when given 2% (w/v) sucrose ( Figure 5B ). These results indicate that MdSWEET23 is involved in sucrose transport.

The expression of MdSWEET23 in VIGS-treated fruits was examined using qRT-PCR. The results showed that VIGS treatment resulted in a significant reduction in the transcript level of the MdSWEET23, which was only 25.28% of the control level ( Figure 6A ). Silencing of MdSWEET23 significantly reduced sucrose and sorbitol contents in fruits; however, it had no significant effect on the accumulation of fructose and glucose ( Figure 6B ). MdSWEET23 silencing resulted in a significant increase in the expression of MdSOT1, which was approximately 4 times higher than that of the control, while the expression of MdSDH6 was significantly decreased ( Figure 6C ). MdSOT1 functions in the unloading of sorbitol in apple fruits (Wei et al., 2014). While, MdSDH6 encodes a sorbitol dehydrogenase (SDH) that converts sorbitol to fructose (Wang et al., 2009). Therefore, the silencing of MdSWEET23 not only reduced the accumulation of sucrose in ‘‘Hanfu’’ apple fruits but also affected the transport and metabolism of sorbitol.

A pRI101-MdSWEET23 recombinant plasmid was constructed and transformed into ‘‘Orin” calli to verify the function of MdSWEET23. After the PCR identification, four overexpression lines were obtained, and OE-1 and OE-5 were selected for further experiments ( Supplementary Figure 1A ). The growth of ‘‘Orin” calli lines of OE-1 and OE-5 is shown in Supplementary Figure 1B . The contents of soluble sugars and glucose in both OE-1 and OE-5 lines were significantly higher than those in the WT plants ( Figures 6D, E ). Since only sucrose was added as a carbon source in the medium, the accumulation of glucose in the calli might be due to sucrose hydrolysis. The overexpression of MdSWEET23 in the “Orin” calli promoted the accumulation of soluble sugars.

To determine the impact of MdSWEET23 expression on sugar accumulation in plants, transgenic tomato lines overexpressing MdSWEET23 driven by the CaMV35S promoter were generated. After PCR identification, eight MdSWEET23-overexpression (MdSWEET23-OE) lines were obtained ( Supplementary Figure 2 ), and OE2-32 and OE3-11 in T3 generation were used for further experiments. Compared with WT plants, the plant height in lines OE2-32 and OE3-11 decreased by 24.42% and 36.88%, respectively ( Figures 7A, B ), and Pn of mature leaves decreased by 45.43% and 53.84%, respectively ( Figure 7E ). The contents of sucrose, glucose, fructose, and starch in mature leaves of lines OE2-32 and OE3-11 were significantly higher than those in WT plants. Particularly, the starch concentration in these lines was 2.51 and 2.19 times higher than that in WT plants, respectively ( Figure 7C ).

We examined the sugar content in MG, BC, and RR fruits ( Figure 1A ) of MdSWEET23-OE lines and WT. The sucrose content in MG fruits of OE2-32 and OE3-11 was significantly higher than that of WT, but no significant differences were observed in glucose and fructose contents between them ( Figure 1B ). At the BC stage, MdSWEET23-OE lines exhibited significantly higher levels of sucrose, glucose, and fructose than the WT ( Figure 1B ). In contrast, the sucrose, glucose, and fructose contents of RR fruits showed no significant difference between MdSWEET23-OE lines and WT ( Figure 1B ).

To investigate the reasons for the sugar level alteration in MdSWEET23-OE lines, we further examined the activity of sugar metabolism-related enzymes and the expression levels of genes related to sugar transport. In MG fruits, no significant difference was found in enzyme activities between MdSWEET23-OE lines and WT ( Figure 1C ). In BC fruits, the CWIN and SS activities of MdSWEET23-OE lines were significantly higher than those of the WT, with 3.84 and 1.56 times increases, respectively. In RR fruits, the CWIN activity of MdSWEET23-OE lines was significantly higher than that of the WT, with a 1.66-fold increase, while the FK activity was significantly lower than that of the WT, at approximately 86% of the WT.

Compared with WT, the HT1 and HT2 expression levels in MdSWEET23-OE lines were notably higher in BC fruits, while no significant changes were observed in their expression in RR and MG fruits. In BC fruits of MdSWEET23-OE lines, the expression of SUT2 was upregulated, while the expression of SUT1 was downregulated; and the transcript levels of SUT4 were notably downregulated in MG, BC, and RR fruits of the MdSWEET23-OE lines. The overexpression of MdSWEET23 also affected the expression of several SWEETs ( Figure 1D ).

Under cold stress (3.5°C for 4 h), no significant change was found in the leaves of OE2-32 and OE3-11 lines; however, the leaves of WT showed wilting symptoms ( Figure 8A ). The release of electrolytes can characterize the damage to plant cells (Hu et al., 2022). Before cold stress treatment, there was no significant difference in REL levels between the two transgenic lines and WT ( Figure 8B ). After cold treatment, the REL levels of OE2-32 and OE3-11 lines were lower than those of WT ( Figure 8B ). We also compared the changes in sugar and starch contents of leaves between MdSWEET23-OE lines and WT before and after cold treatment. The results showed that the soluble sugar content in leaves of MdSWEET23-OE lines was only slightly higher than that of WT under cold treatment ( Figure 8C ). The starch content in leaves of MdSWEET23-OE lines and WT all decreased when exposed to cold treatment; however, the starch content in leaves of OE2-32 and OE3-11 decreased from 58.27 to 28.61 mg·g^-1 FW (by 50.89%) and from 53.81 to 29.16 mg·g^-1 FW (by 45.81%), respectively, whereas that of WT only decreased from 20.27 to 13.23 mg·g^-1 FW (by 34.73%) ( Figure 8D ).

Fruit growth is a high-priority sink in assimilate partitioning. An apoplasmic phloem unloading strategy is employed in the fleshy fruits of apples, which requires sugar transporters to export sugar into the apoplasm. Many SWEETs function in flesh fruits, such as SlSWEET7a and SlSWEET14 in tomato (Zhang et al., 2021), VvSWEET10 in grape (Zhang Z. et al., 2019), and MdSWEET9b in apple (Zhang et al., 2023). CsSWEET7a localized on the plasma membrane in companion cells (CCs) of the phloem has recently been reported to be involved in sugar phloem unloading in cucumber fruit by exporting hexoses from CCs to the apoplasmic space (Li et al., 2021). In our previous study, MdSWEET23 could play an important role during the progress of apple fruits maturation with starch breaks down and soluble sugars accumulate massively (Nie et al., 2022). Here, MdSWEET23, isolated from a cDNA library of apple fruits, belongs to clade III and is phylogenetically most closely related to AtSWEET13 ( Figure 2 ). AtSWEET13, which also belongs to clade III, is plasma membrane-localized and transports both sucrose and GA (Kanno et al., 2016). But a recent study found that sucrose rather than GA transported by AtSWEET13 and AtSWEET14 directly affects anther development and seed yield (Wang et al., 2022). Our heterologous expression assay in yeast showed that MdSWEET23 functioned in sucrose transport rather than glucose and fructose transport and could restore SUSY7/ura3 growth on media supplemented with sucrose ( Figure 5 ). Moreover, the overexpression of MdSWEET23 in ‘‘Orin” calli promoted sugars accumulation in calli, which grown on MS medium with sucrose as the carbon source ( Figure 6 ). These results indicate that MdSWEET23 is a functional sucrose transporter. In previous studies, it has been found that most SWEETs in clade III are capable of transporting sucrose. Such as AtSWEET11, 12, and 13 in Arabidopsis (Chen et al., 2012; Han et al., 2017), SbSWEET13a, 13b, and 13c in sorghum, and StSWEET11 in potato (Abelenda et al., 2019), but their functions are different.

The functions of SWEETs also largely depend on their subcellular and tissue localization. SWEET proteins are mainly localized on the plasma and the tonoplast, while, a few members are also localized on the endoplasmic reticulum and the Golgi membranes (Chen L. Q. et al., 2015; Jeena et al., 2019; Breia et al., 2021). AtSWEET13 (Kanno et al., 2016) in Arabidopsis, PbSWEET4 (Ni et al., 2020) and PuSWEET15 (Li et al., 2020) in pear, BsSWEET15 and BsSWEET16 (Lu et al., 2022) in Bletilla striata, and SlSWEET14 (Zhang et al., 2021) in tomato, are also localized on the plasma membrane. The localization of sugar transporter ClVST1 shifted from tonoplast in wild-type watermelon to plasma membrane in sweet watermelon, which increased sugar sink potency (Ren et al., 2020). MdSWEET23 was localized on the plasma membrane ( Figure 4 ) and mainly expressed in the vascular bundle, especially in the sepal and carpel vascular bundles of apple fruit ( Figure 3 ). The carpel vascular bundle is the major channel for nutrients transporting toward the seeds, and the sepal vascular bundle is the main channel for nutrient and water transport for fruit expansion (Wang et al., 2015). In Arabidopsis, AtSWEET11, 12, and 15 are all highly expressed in the seed coat and play key roles in seed development (Le Hir et al., 2015). Therefore, it is speculated that MdSWEET23 may participate in apple seed development. The expression of MdSWEET23 in the sepal vascular bundle region indicates that it may be involved in the unloading of sucrose to the fleshy tissue of apple fruits. In apple, MdSWEET9b functions on the SE/CC, and its surrounding parenchyma cells is involved in fruit sugar accumulation (Zhang et al., 2023). In cucumber, CsSWEET7a, localized to the plasma membrane in CCs of the phloem, is involved in sugar phloem unloading in cucumber fruit by removing hexoses from CCs to the apoplasmic space (Li et al., 2021).

The silencing of MdSWEET23 in “Hanfu” apple fruits significantly reduced sucrose and sorbitol content, which further confirmed the involvement of MdSWEET23 in sucrose transport and accumulation in apple fruits. Further, whether the decrease in sorbitol content ( Figure 6B ) was due to MdSWEET23 silencing and whether MdSWEET23 was involved in sorbitol transport and accumulation in apple fruits? The expression of MdSOT1 significantly increased, whereas that of MdSDH6 substantially decreased in the MdSWEET23-silenced fruits. These results indicate that MdSWEET23 silencing affects sorbitol metabolism and transport in apple fruits. However, more direct evidence is needed to prove whether MdSWEET23 can transport sorbitol. In apple, the carbohydrates used for long-distance transportation are sucrose and sorbitol. Sucrose and sorbitol arriving in the SE/CC complex are pumped into the apoplasmic space by SUT/SOT (Fan et al., 2009; Chen et al., 2012) after that sucrose is broken down to hexose by CWIN, which was transported to the phloem parenchyma cells by the hexose transporters (Zhang et al., 2004). We speculate that MdSWEET23 may play a role in sucrose or sorbitol transport along the sugar concentration gradient from the apoplasmic space to the phloem parenchyma cells.

Sugar transport can influence the distribution of carbohydrates throughout the plant, which is important for plant growth, and the disorganization of sugar distribution will lead to abnormal plant growth. The overexpression of ZjSWEET2.2 in jujube led to a significant increase in leaf photosynthetic assimilate (Geng et al., 2020), whereas the overexpression of PbSWEET4 in strawberry reduced sugar content in leaves (Ni et al., 2020). Klemens et al. (2013) found that the overexpression of AtSWEET16 in Arabidopsis resulted in significantly increased growth compared to the WT. While, overexpression of OsSWEET14 and OsSWEET5 in rice significantly reduced plant height (Zhou et al., 2014; Kim et al., 2021). Silencing of SlSWEET7a and SlSWEET14 in tomato resulted in a 35%–48% increase in plant height compared to the WT (Zhang et al., 2021). The ectopic overexpression of MdSWEET23 in tomato led to an over-accumulation of carbohydrates in leaves, with a decrease in photosynthetic rate and plant height ( Figure 7 ). In apple, MdSWEET23 is expressed mainly in the sepal and carpel vascular bundles of fruits (sink organs) ( Figure 3 ). While, in transgenic tomato lines, the expression of the MdSWEET23 was driven by the CaMV35S promoter, a constitutive promoter, and thus MdSWEET23 expression in tomato lacks temporal and spatial regulation. In source organs, SWEETs are mainly responsible for transporting sucrose from the synthesis site to the loading site (Chen et al., 2012). The expression of MdSWEET23 in tomato functional leaves (source organ) may have affected the loading process of photosynthetic assimilates, too much sucrose was transported to the loading site and caused the abnormal accumulation of starch and soluble sugars in mature leaves. The increases in soluble sugar and starch content in leaves negatively regulate the photosynthetic rate, causing a decrease in the net photosynthetic rate of the leaves (Sonnewald, 2001). In maize, CST1 belongs to the Clade I of the SWEET protein family, is specifically expressed in subsidiary cells, and positively regulates stomatal opening and source capacity at the grain-filling stage (Wang H. et al., 2019). At the seedling stage, phenotypes of the MdSWEET23-OE plants might partially or wholly be ascribed to the disordered sugar distribution.

Previous genome-wide studies in different plants have found that SWEETs exhibit variable expression patterns under cold treatments, which suggests that SWEETs are involved in cold-induced sugar-signaling responses (Zhang W. et al., 2019; Jiang et al., 2021). Overexpression of AtSWEET16 (Klemens et al., 2013) and AtSWEET17 (Chardon et al., 2013) in Arabidopsis elevated sugar content and increased cold tolerance. Hu et al. (2022) found that cold-tolerant cucumber had a higher expression level of CsSWEET2 than cold-sensitive cucumber, and the overexpression of CsSWEET2 in Arabidopsis enhanced the cold resistance. The MdSWEET23-OE lines showed higher resistance to cold stress. Ectopic overexpression of MdSWEET23 accelerated starch degradation in leaves under cold stress, but leaf soluble sugar content in OE lines showed no significant difference with WT. We speculated that the overexpression of MdSWEET23 in tomato may have resulted in facilitating the process of sugar loading during cold treatment, which in turn allowed more sugar to be transported out of the source organ, and could potentially influence starch hydrolysis through the feedback regulation mechanism. Overexpression of MdSWEET23 in tomato may alter sucrose transport and distribution in OE lines under cold stress, thereby maintaining sugar homeostasis in response to cold stress. The possible mechanisms of increased cold tolerance of MdSWEET23-OE tomato lines need to be further investigated.

Based on the characteristics of MdSWEET23 in substrate transport, subcellular and tissue localization ( Figures 3 – 5 ), we speculated that MdSWEET23 could be involved in the uptake of sucrose through the plasma membrane in sepal and carpel vascular bundles of apple fruits. Ectopic overexpression of MdSWEET23 in tomato significantly increased the sugar content of the BC fruits but had no significant effect on the sugar content of the RR fruits ( Figure 1 ). In pear, overexpression of PuSWEET15 led to a significant increase in sucrose content in fruits (Li et al., 2020). Similarly, overexpression of VvSWEET10 resulted in a significant enhancement of sugar content in grape calli and tomato fruits (Zhang Z. et al., 2019). Silencing of SlSWEET7a and SlSWEET14 in tomato increased the sugar content of the fruits(Zhang et al., 2021). Sugars can act as signaling molecules that control distinct aspects of plant development. Several recent studies have reported that downstream sugar transporters could affect the sugar metabolism, by influencing upstream sugar metabolic enzyme activities and expression of sugar transporters through the feedback regulation mechanisms. For example, in cucumber, the overexpression of CsSWEET7a could shift the enzymatic reactions to produce more hexoses due to its overexpression leading to more hexose being unloaded from CCs, and less hexose retained in the cytosol of CCs (Li et al., 2021). MdERDL6-1, a tonoplast H+/glucose symporter, was highly expressed in apple fruits, and the overexpressing of it up-regulated expression of the H+/sugar antiporter genes, TST1 and TST2, by exporting glucose from vacuole to cytosol, which in turn led to increased glucose, fructose and sucrose in transgenic apple and tomato leaves and fruits (Zhu et al., 2021). In this study, we observed that overexpression of MdSWEET23 in tomato could affect upstream sugar metabolic enzymes and the expression of sugar transporters. During the hexose accumulation phase in tomato, sugar unloading is the apoplastic pathway (Ruan and Patrick, 1995). Sucrose arrived in the SE/CC complex unloaded into the apoplasm by SUTs or catabolized to hexose by CWIN, and then hexose was transported to parenchyma cells by HTs (Zhang et al., 2021). In BC fruits of the MdSWEET23-OE tomato lines, more sucrose may unload into the apoplasm by MdSWEET23, leading to high CWIN activity to hydrolyze sucrose into hexoses( Figure 1C ), which might be the cause of more accumulation of hexose in fruits of transgenic lines. Meanwhile, the expressions of hexoses transporter genes HT1, HT2, and HT3 were also stimulated ( Figure 1D )., although CWIN activity remained significantly higher as compared to WT. Ectopic overexpression of MdSWEET23 may generate an unusual situation affecting the expression of sucrose transporter genes SUT1, SUT2, and SUT4 ( Figure 1D ). Down-regulated SUT1, SUT2, and SUT4 in RR fruits may negatively affect sucrose unloading, thereby influencing the enrichment of sucrose against the concentration gradient in fruits. Ectopic overexpression of MdSWEET23 in tomato caused altered sugar metabolism and distribution but did not increase sugar sink potency in tomato fruits.