A Sugar Transporter Takes Up both Hexose and Sucrose for Sorbitol-Modulated In Vitro Pollen Tube Growth in Apple[OPEN]

A sugar transporter protein, STP13a, mediates the uptake of both hexose and Suc for in vitro pollen tube growth in apple, with its expression regulated by sorbitol via a MYB transcription factor. Rapid pollen tube growth requires uptake of Suc or its hydrolytic products, hexoses, from the apoplast of surrounding tissues in the style. Due to species-specific sugar requirements, reliance of pollen germination and tube growth on cell wall invertase and Suc or hexose transporters varies between species, but it is not known if plants have a sugar transporter that mediates the uptake of both hexose and Suc for pollen tube growth. Here, we show that a sugar transporter protein in apple (Malus domestica), MdSTP13a, takes up both hexose and Suc when expressed in yeast, and is essential for pollen tube growth on Glc and Suc but not on maltose. MdSTP13a-mediated direct uptake of Suc is primarily responsible for apple pollen tube growth on Suc medium. Sorbitol, a major photosynthate and transport carbohydrate in apple, modulates pollen tube growth via the MYB transcription factor MdMYB39L, which binds to the promoter of MdSTP13a to activate its expression. Antisense repression of MdSTP13a blocks sorbitol-modulated pollen tube growth. These findings demonstrate that MdSTP13a takes up both hexose and Suc for sorbitol-modulated pollen tube growth in apple, revealing a situation where acquisition of sugars for pollen tube growth is regulated by a sugar alcohol.


INTRODUCTION
Pollen germination and rapid tube growth is essential for the reproductive success of flowering plants because it delivers the sperm cells to ovules for double fertilization. Once pollen grains land on the stigma, compatible pollen grains germinate after hydration and grow into the transmitting tissue in the style when conditions are favorable. Early processes in pollen germination and initial tube growth may rely on nutrient storage in the pollen grain (Read et al., 1993), but due to symplastic isolation of the pollen tube, subsequent tube growth requires uptake of sugars from the apoplast of the transmitting tissue. Suc unloaded via symplast from the phloem effluxes into the apoplast via Sugars Will Eventually be Exported Transporters (SWEETs), SWEET9 and SWEET10, with Glc efflux possibly mediated by SWEET1, in the transmitting tissue (Chen et al., 2010;Werner et al., 2011;Rottmann et al., 2018c). The released Suc is either directly taken up by Suc transporters (SUTs/SUCs), or converted to Glc and Fru by cell wall invertase first and then taken up by sugar transporter proteins (STPs), into the growing pollen tube (Goetz et al., 2017;Rottmann et al., 2018c).
Conversion of Suc to hexoses by cell wall invertase is required not only for pollen development but also for pollen germination and tube growth in many plants. In tobacco (Nicotiana tabacum; Goetz et al., 2001), Arabidopsis (Arabidopsis thaliana; Hirsche et al., 2009), and oilseed rape (Brassica napus; Engelke et al., 2011), pollen development was arrested by tissue-specific antisense repression of cell wall invertase, leading to male sterility. When a tobacco cell wall invertase inhibitor, NtCIF, was overexpressed under the control of the promoter of a cell wall invertase gene, NtcwINV2 (Nin88), pollen grains were fully developed, but had lower invertase activity and consequently lower germination and tube growth (Weil et al., 1994;Greiner et al., 1998;Hirsche et al., 2009). Uptake of hexoses derived from Suc into pollen tube is mediated by plasma membrane-localized STPs. STP4, STP6, STP8, STP9, STP10, and STP11 are primarily responsible for Glc uptake into pollen tubes of Arabidopsis (Rottmann et al., 2018c). In cucumber (Cucumis sativus), antisense repression of HT1, a pollen-specific hexose transporter with high sequence similarity to AtSTP11, inhibits pollen germination, tube growth, and seed development .
Direct uptake of Suc by SUTs/SUCs into pollen grains has also been demonstrated to be crucial for pollen germination and tube growth. In tomato (Solanum lycopersicum), antisense repression of LeSUT2 led to reduced pollen germination and tube growth, compromising fruit and seed development (Hackel et al., 2006). Mutants of Arabidopsis SUC1 (Sivitz et al., 2008) and rice (Oryza sativa) SUT1 (Hirose et al., 2010) have poor pollen germination and segregation distortion. In cucumber, antisense repression of SUT1 impairs pollen development, leading to male sterility (Sun et al., 2019). Decreased Suc uptake appears to be responsible for reduced pollen germination in transgenic tobacco plants when tissue-specific overexpression of NtCIF caused a moderate reduction in cell wall invertase activity; further reduction in cell wall invertase activity led to decreased uptake of Glc and additional drop in pollen germination (Goetz et al., 2017). In vitro pollen germination experiments show diverse responses of pollen germination to individual sugars as carbon source, ranging from growing equally well on Glc, Fru, and Suc for petunia (Petunia hybrida; Ylstra et al., 1998) to almost complete inhibition on Glc and Fru relative to Suc for Arabidopsis , with many species (e.g. tobacco, tomato, apple) in between. With this highly species-dependent sugar requirement for pollen germination and tube growth Rottmann et al., 2018c), it is expected that the reliance of pollen germination and tube growth on cell wall invertase and types of sugar transporters (SUTs/SUCs versus STPs) varies between species. It is not known, however, if plants have a sugar transporter that takes up both hexose and Suc for pollen germination and tube growth.
Suc and Glc may also serve as signals for pollen germination and tube growth in addition to being carbon sources. Although Glc as sole carbon source is sufficient for pollen germination in tobacco, tube elongation requires Suc as a metabolic signal (Goetz et al., 2017). Consistent with this signaling role of Suc, antisense suppression of SUCROSE NON-FERMENTING-1-RELATED KINASE1 (SnRK1) was demonstrated to lead to abnormal pollen development and male sterility in barley (Hordeum vulgare; Zhang et al., 2001). In Arabidopsis, Glc inhibits pollen tube growth via hexokinase 1-dependent signaling (Rottmann et al., 2016(Rottmann et al., , 2018c. Many plant species synthesize sugar alcohols in addition to Suc, which is estimated to contribute to 30% of the primary carbon production on a global scale (Bieleski, 1982). The presence of sugar alcohols in these species raises an interesting question as to how the signaling of sugar alcohols operates in the regulation of sugar metabolism and utilization for plant growth and development. In vitro pollen germination and tube growth represents an excellent model system for studying sugar/sugar alcohol signaling because of its simplicity and amenability to experimental manipulations  although differences in gene expression profiles were reported to exist between the in vitro and in vivo systems (Qin et al., 2009).
In pome and stone fruits of the Rosaceae family including apple (Malus domestica), pear (Pyrus communis), peach (Prunus persica), apricot (Prunus armeniaca), plum (Prunus domestica), and cherry (Prunus avium), sorbitol is a major photosynthate and phloem transport carbohydrate. Sorbitol is synthesized via a twostep process in the cytosol of source leaves: conversion of Glc 6-phosphate to sorbitol 6-phosphate by aldose-6-phosphate reductase (A6PR) followed by dephosphorylation of sorbitol 6phosphate to sorbitol (Negm and Loescher, 1981;Zhou et al., 2003). Sorbitol enters the phloem along with Suc via passive, symplastic loading for long-distance transport (Reidel et al., 2009;Fu et al., 2011). In sink tissues such as shoot tips or fruit, sorbitol is converted to Fru by sorbitol dehydrogenase (Negm and Loescher, 1979;Yamaguchi et al., 1994). When sorbitol synthesis is decreased by antisense repression of A6PR in apple leaves, more Suc is transported to sink organs such as shoot tips and developing fruits and the corresponding upregulation of Suc metabolism keeps tree growth and fruit development largely homeostatic (Cheng et al., 2005;Zhou et al., 2006;Li et al., 2018). However, in the flowers of the transgenic trees, decreased sorbitol level leads to abnormal stamen development and reduced pollen germination and tube growth via a MYB transcription factor, MYB39L, clearly indicating a signaling role of sorbitol in stamen development and pollen tube growth (Meng et al., 2018a). In this work, we report the characterization of an STP that takes up Suc as well as hexose and is essential for sorbitol-modulated pollen tube growth in apple.

Antisense Repression of MdSTP13a Decreases Pollen Tube Growth on Glc
In earlier work, transgenic 'Greensleeves' apple trees with decreased sorbitol synthesis were found to have abnormal stamen development and reduced pollen tube growth. Four putative sugar transporters (MdHT1.4-1.7) were among the downregulated genes in developing anthers and growing pollen tubes of the transgenic lines (Meng et al., 2018a). Phylogenetic analysis of these transporters along with 14 Arabidopsis STPs indicate that they have highest sequence similarity to AtSTP13 (Supplemental Figure 1), and therefore they are renamed as MdSTP13a, b, c, and d, corresponding to MdHT1.7, 1.6, 1.5, and 1.4 (Wei et al., 2014).
Considering the importance of sugar uptake for pollen tube growth, we explored the potential role of these four sugar transporters via antisense oligonucleotide transfection (Meng et al., 2014b). Transfection of wild-type 'Greensleeves' pollen with antisense oligonucleotide decreased the expression of the corresponding sugar transporter during pollen tube growth without altering the expression of other three sugar transporter genes (Figures 1A to 1D; Supplemental Figure 2), but only antisense transfection with oligonucleotides targeting MdSTP13a led to significant reduction in pollen tube growth on 5% Glc (Figures 1E to 1I). This indicates that MdSTP13a is essential for in vitro apple pollen tube growth on Glc. Unlike AtSTP13, which is expressed in leaves (Schofield et al., 2009;Yamada et al., 2016), MdSTP13a is specifically expressed in stamens and pollen tubes of apple flowers (Supplemental Figure 3; Meng et al., 2018a), consistent with its role in pollen tube growth.
To determine the cellular location of the MdSTP13a protein, a 35S:MdSTP13a-GFP fusion construct was transiently expressed in apple callus protoplasts. The GFP signal from the fusion protein was colocalized to the plasma membrane with the plasma membrane Deep Red marker in contrast with the wide distribution throughout the protoplast in the 35S:GFP control (Figures 2A and 2B). Considering that MdSTP13a is highly expressed in stamens and pollen tubes (Meng et al., 2018a), we also determined the localization of MdSTP13a in pollen tubes by biolistically expressing MdSTP13a-GFP under a pollen-specific promoter Lat52 (Albani et al., 1991;Khurana et al., 2012). A clear GFP signal from the fusion protein was detected in the pollen tube's plasma membrane with the FM 4-64 red marker ( Figure 2C). These results indicate that MdSTP13a is a plasma membrane protein.

MdSTP13a Transports Hexose and Suc Competitively in Yeast
To characterize the transport properties of the encoded protein, we expressed MdSTP13a via the yeast expression vector pDR196 (I) Photos of pollen tube growth at 120 min. Scale bar 5 50 mm. Data are mean 6 SE, n 5 3. Different letters (a, b) indicate significant difference between groups using Tukey's Honest Significant Difference test at P < 0.05 after ANOVA. in a hexose transport-deficient yeast strain, EBY.VW4000, which cannot grow on hexose but can grow on maltose (Wieczorke et al., 1999). As expected, VW4000 with empty vector pDR196 did not grow on 2% (w/v) Glc, but expression of MdSTP13a enabled its growth on Glc ( Figure 3A). 14 C-Glc uptake analysis showed that MdSTP13a-expressing EBY.VW4000 took up 14 C-Glc, whereas the empty vector control did not ( Figure 3B). Over the pH range examined (2.0 to 8.0), maximum 14 C-Glc uptake occurred at 5.5 to 6.0 ( Figure 3C), which is consistent with the pH optima reported for other members of the STP family (Rottmann et al., 2016(Rottmann et al., , 2018a. Based on the response of 14 C-Glc uptake to Glc concentrations, the K M value of MdSTP13a for Glc was estimated to be 157.3 6 12.6 mM, with a maximum uptake rate (V max ) of 26.6 6 1.3 nmol min 21 mg 21 cells at pH 6.0 ( Figure 3D) via the nonlinear regression applied to Michaelis-Menten kinetics analysis.
To determine the substrate specificity of MdSTP13a, the possible uptake of other sugars was assessed by measuring the uptake of 14 C-Glc in the presence of a 10-fold excess of nonradioactive sugars as shown in Figure 3E. Other hexoses, Gal, Fru, and Man, reduced the uptake rate of 14 C-Glc to 56%, 53%, and 50%, respectively. As addition of nonradioactive Glc decreased 14 C-Glc uptake to 30% of the control, the data indicate that these sugars are transported by MdSTP13a, but at a lower affinity than Glc. By contrast, pentoses such as Xyl, Ara, and Rib, disaccharides including turanose, maltose, trehalose, and melibiose, and trisaccharides such as raffinose and the main sugar alcohol in apple, sorbitol, did not alter 14 C-Glu uptake. However, 14 C-Glc uptake was significantly reduced in the presence of nonradioactive Suc (to 29% of the control) and its analog esculin (to 49% of the control), suggesting that MdSTP13a has the potential for transporting Suc. In addition, low concentrations of the proton uncoupler, carbonyl cyanide m-chlorophenyl hydrazone (CCCP), significantly decreased 14 C-Glc uptake (to 18%), but p-chloromercuriphenylsulfonic acid (PCMBS), a sulfhydryl reactive-reagent, did not ( Figure 3E). This suggests that sugar uptake via MdSTP13a occurs in symport with protons, as has been shown for other STPs Rottmann et al., 2016Rottmann et al., , 2018a.
We further analyzed 14 C-Glc uptake in response to a range of Suc concentrations (0 to 750 mM) to examine the inhibition type and kinetics of MdSTP13a by Suc. Increasing Suc concentration did not alter the V max value for 14 C-Glc, but decreased its K M value, with a K I (the inhibition constant) value of 246 6 13 mM ( Figure 3F). These results suggest that Suc is a competitive inhibitor for Glc uptake by MdSTP13a.
To directly assess the uptake of Suc by MdSTP13a, we expressed MdSTP13a in the hexose transport-and SUC2deficient Saccharomyces cerevisiae strain CSY4000, which was generated from EBY.VW4000 for characterizing Suc transporters as well as hexose transporters (Rottmann et al., 2016). CSY4000 with empty vector pDR196 grew only on maltose, but complementation of CSY4000 with MdSTP13a enabled growth on Suc and Glc ( Figure 4A). As predicted, MdSTP13a-expressing CSY4000 took up 14 C-Suc, whereas the empty vector control did not ( Figure 4B). The optimal pH for 14 C-Suc uptake by MdSTP13a is ;6.0 ( Figure 4C), similar to that for 14 C-Glc uptake ( Figure 3C). However, the K M value for Suc (66.9 6 1.8 mM) is lower than that for Glc ( Figure 4D). The V max value for Suc (9.4 6 0.2 nmol min 21 mg 21 cells) is ;1/3 of that for Glc. Inhibition of 14 C-Suc uptake by nonradioactive sugars is similar to that of 14 C-Glc uptake: Hexoses such as Glc, Fru, Gal, and Man, and Suc and its analog esculin inhibited 14 C-Suc uptake whereas other sugars did not ( Figure 4E). Similar to 14 C-Glc uptake, 14 C-Suc uptake was significantly inhibited by CCCP, but not by PCMBS. To further confirm the inhibition of 14 C-Suc uptake by Glc, we assayed 14 C-Suc uptake over a range of Glc concentrations (0 to 750 mM). Increasing Glc concentration did not alter the V max value for 14 C-Suc uptake, but decreased the K M value, with a K I of 296 6 16 mM ( Figure 4F), exhibiting typical competitive inhibition kinetics.
To ensure our 14 C-sugar assays are reliable, two previously characterized sugar transporters-an STP from Arabidopsis, AtSTP13 (Nørholm et al., 2006) and a Suc transporter from potato (Solanum tuberosum), StSUT1 (Krügel et al., 2013)-were expressed in CSY4000 via pDR196 and included as additional controls. Esculin assays showed that CSY4000 expressing MdSTP13a or StSUT1 took up esculin whereas CSY4000 expressing AtSTP13 or pDR196 empty vector did not (Supplemental Figure 6A). Consistent with their previously reported substrate specificity, 14 C-Glc uptake by AtSTP13 was inhibited by Glc, but not by Suc or esculin, whereas 14 C-Suc uptake by StSUT1 was inhibited by Suc and esculin, but not by Glc (Supplemental Figures  6B and 6C).
We also expressed MdSTP13a in the widely used Suc uptakeand SUC2-deficient yeast strain SUSY7/ura3 (Riesmeier et al., 1992) to further verify its ability to take up 14 C-Suc, with StSUT1 expressed in SUSY7/ura3 as a positive control. Similar to the results obtained with CSY4000-MdSTP13a, 14 C-Suc uptake by SUSY7/ura3-MdSTP13a was inhibited by Glc, Suc, and esculin. By contrast, 14 C-Suc uptake by SUSY7/ura3-StSUT1 was inhibited by Suc and esculin, but not by Glc (Supplemental Figures  6D and 6E). Taken together, these results indicate that MdSTP13a is a dual transporter that takes up Suc as well as hexose.

Antisense Repression of MdSTP13a Reduces Pollen Tube Growth on Glc and Suc, But Not on Maltose
Based on the substrate specificity of MdSTP13a obtained in yeast, we hypothesized that MdSTP13a is essential for apple pollen tube growth on Suc as well as Glc but not on maltose. To test this hypothesis, we conducted pollen tube growth experiments on the three sugars by altering the expression of MdSTP13a via oligonucleotide transfection. In liquid germination media containing 5% Glc, Suc, or maltose, antisense oligonucleotide transfection with MdSTP13a (as-MdSTP13a) significantly reduced its transcript level during pollen tube growth compared with the control (transfection agent only) and sense oligonucleotide transfection (s-MdSTP13a; Figure 5A). This led to significantly slower pollen tube growth on both Glc and Suc media, but not on maltose medium ( Figure 5B; Supplemental Figure 7). This is consistent with the finding in yeast with heterologous expression of MdSTP13a that both Glc and Suc are taken up by MdSTP13a, whereas maltose is not ( Figures 3D and 4D).
As extracellular invertase can break down Suc into Glc and Fru in supporting pollen tube growth on Suc medium, we used two approaches to determine if the conversion of Suc to Glc and Fru plays a significant role in providing carbon for apple pollen tube growth on the Suc medium. First, we directly measured sugar composition of the Suc medium over the duration of pollen tube growth. Suc concentration remained essentially unchanged during the 2-h pollen tube growth period (Supplemental Figure 8A). There was trace amount of Glc and Fru in the Suc medium at the beginning of pollen tube growth (presumably from impurity of the Suc used), and their concentrations increased in the first 1.5 h and then leveled off. However, the concentrations of Glc (D) Michaelis-Menten kinetics analysis of 14 C-Glc uptake in relation to Glc concentration. K m 5 153.7.9 6 12.6 mM; V max 5 26.6 6 1.3 pmol min -1 mg -1 cells. The kinetic constants were derived from untransformed data via nonlinear regression using the software R (https://rpubs.com/RomanL/6752). (E) 14 C-Glc uptake of MdSTP13a expressed in EBY.VW4000 as affected by competing sugars and metabolic inhibitors. Competing sugars and metabolic inhibitors (CCCP and PCMBS) were added 30 s before addition of labeled 14 C-Glc at a final concentration of 1 mM and 50 mM, respectively, and 14 C-Glc uptake was assessed for 4 min. Values are relative to that obtained in the absence of any competing sugar or metabolic inhibitor (Control as 100%). (F) Lineweaver-Burk plots of 14 C-Glc uptake of MdSTP13a expressed in the EBY.VW4000 under different concentrations of Suc. The kinetic constants were derived from untransformed data via nonlinear regression using the software R. The estimated K m 5 148.1 6 10.2 mM, V max 5 25.0 6 1.4 pmol min -1 mg -1 cells without Suc; K m 5 184.2 6 4.5 mM, V max 5 25.1 6 1.5 pmol min -1 mg -1 cells with 250 mM of Suc; K m 5 236.3 6 5.9 mM, V max 5 24.7 6 1.5 pmol min -1 mg -1 cells with 500 mM of Suc; and K m 5 355.7 6 9.2 mM, V max 5 25.2 6 1.4 pmol min -1 mg -1 cells with 750 mM of Suc. K I 5 246 6 13 mM. Data are mean 6 SE, n 5 3. Different letters (a, b, c, d) in (E) indicate significant difference between groups using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA. FW, fresh weight. and Fru were still very low even at the end of the 2-h pollen tube growth period (at ;120 mg/mL), suggesting that Suc cleavage by cell wall invertase is very limited. Based on the uptake kinetics of MdSTP13a for Glc and Suc and their competitive inhibition, we predicted that the contribution of Glc/Fru uptake to pollen tube growth on the Suc medium would be negligible. When pollen grains were germinated on a Glc and Fru medium at 120 mg/mL each, with or without mannitol included to adjust the osmotic potential to that of 5% Suc, pollen tube growth was essentially the same as the no-sugar control whereas pollen tube was significantly longer in the Suc medium during the first 2-h period and beyond (Supplemental Figure 8B). This clearly demonstrates that, although the initial pollen tube growth relies primarily on nutrients stored in the pollen grains, their subsequent growth is mainly supported by uptake of Suc from the Suc medium. Second, we selected the most highly expressed cell wall invertase, MdCWI2 (MDP0000268052), based on our RNAsequencing data (Meng et al., 2018a), followed by confirmation with reverse transcription quantitative PCR (RT-qPCR), and transfected pollen of wild-type 'Greensleeves' apple to determine if suppressing its expression leads to a decrease in pollen tube growth. Antisense MdCWI2 oligonucleotide transfection of wildtype pollen significantly decreased its transcript level, but did not alter pollen tube growth (Supplemental Figure 9). Collectively, these results indicate that conversion of Suc to Glc and Fru contributes very little to pollen tube growth on the Suc medium within the first 2 h in apple.
As SUTs/SUCs have been demonstrated to be essential for pollen tube growth in tomato (Hackel et al., 2006), Arabidopsis (Sivitz et al., 2008), rice (Hirose et al., 2010), and cucumber (Sun et al., 2019), we determined SUT's role in pollen tube growth in apple on Suc medium. The most highly expressed SUT in pollen, MdSUT2 (MDP0000850943), was selected for transfection of wild-type 'Greensleeves' apple pollen based on our RNAsequencing data (Meng et al., 2018a), followed by confirmation with RT-qPCR. Antisense MdSUT2 oligonucleotide transfection (D) Michaelis-Menten kinetics analysis of 14 C-Suc uptake in relation to Suc concentration. K m 5 66.9 6 1.8 mM, V max 5 9.4 6 0.2 pmol min -1 mg -1 cells. The kinetic constants were derived from untransformed data via nonlinear regression using the software R (https://rpubs.com/RomanL/6752). (E) 14 C-Suc uptake of MdSTP13a expressed in CSY4000 as affected by competing sugars and metabolic inhibitors. Competing sugars and metabolic inhibitors (CCCP and PCMBS) were added 30 s before addition of labeled 14 C-Suc at a final concentration of 1 mM and 50 mM, respectively, and 14 C-Suc uptake was assessed for 4 min. Values are relative to that obtained in the absence of any competing sugar or metabolic inhibitor (Control as 100%).
(F) Lineweaver-Burk plots of 14 C-Suc uptake of MdSTP13a expressed in CSY4000 under different concentrations of Glc. The kinetic constants were derived from untransformed data via nonlinear regression using the software package R. The estimated K m 5 66.2 6 5.5 mM, V max 5 9.2 6 0.2 pmol min -1 mg -1 cells without Glc; K m 5 113.9 6 3.8 mM, V max 5 9.3 6 0.2 pmol min -1 mg -1 cells with 250 mM of Glc; K m 5 148.1 6 7.8 mM, V max 5 9.4 6 0.4 pmol min -1 mg -1 cells with 500 mM of Glc; and K m 5 193.6 6 13.9 mM, V max 5 9.2 6 0.2 pmol min -1 mg -1 cells with 750 mM of Glc. K I 5 296 6 16 mM. Data are mean 6 SE, n 5 3. Different letters (a, b, c) in (E) indicate significant difference between groups using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA. FW, fresh weight. (D) Glc-induced changes in current density (top) obtained from individual pollen tube protoplasts transfected with s-MdSTP13a (n 5 8), as-MdSTP13a (n 5 8) or Control (n 5 6) and Suc-induced changes in current density (bottom) obtained from individual pollen tube protoplasts transfected with s-MdSTP13a (n 5 8), as-MdSTP13a (n 5 8) or Control (n 5 7). Data are mean 6 SE, n 5 3 in (A) and (B). Different letters (a, b) indicate significant difference between groups using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA. significantly decreased its expression, but did not alter pollen tube growth (Supplemental Figure 10).
To confirm that MdSTP13a takes up both Glc and Suc during pollen tube growth, we used the patch-clamp technique to monitor transporter activity across the plasma membrane of isolated pollen tube protoplasts in response to antisense oligonucleotide transfection of MdSTP13a (as-MdSTP13a). As transmembrane H 1 gradient drives the transport capacity of H 1dependent transporter, we chose to use pH 7.5 for the pipette medium and pH 5.5 for the bath medium. Pollen tube protoplasts were isolated from the control or transfected with fluorescent probe cy5-labeled as-MdSTP13a or s-MdSTP13a oligonucleotide, and were selected under a fluorescence microscope for patch-clamp. When Glc was added to the bath solution to a final concentration of 50 mM, elicited transmembrane currents of 0.46 6 0.04 pA/pF and 0.42 6 0.02 pA/pF were recorded in the control and s-MdSTP13a transfected protoplasts, respectively, whereas a significantly smaller current (0.29 6 0.03 pA/pF) was detected in the as-MdSTP13a-transfected protoplasts (Figures 5C and 5D; Supplemental Figure 11). Similarly, when Suc was added separately to the bath solution to a final concentration of 50 mM, a significantly smaller transmembrane current was recorded in the as-MdSTP13a oligonucleotide-transfected cells (0.31 6 0.05 pA/ pF) than in the control (0.56 6 0.03 pA/pF) and s-MdSTP13a (0.58 6 0.03 pA/pF) oligonucleotide-transfected cells ( Figures 5C and  5D). As the currents were recorded within a few minutes upon adding respective sugars for protoplasts (cell wall-free), the Sucelicited current is unlikely the result of any hexose converted from Suc. So measurements of Glc-and Suc-elicited transmembrane currents indicate that MdSTP13a functions as a sugar/proton symporter in mediating the import of both Glc and Suc into the growing pollen tubes.
Esculin has been used to mimic the uptake of Suc in both yeasts and plants (Gora et al., 2012;Zanon et al., 2015;Nieberl et al., 2017;Rottmann et al., 2018b;Patzke et al., 2019). We conducted esculin assays on pollen transfected with as-MdSTP13a or s-MdSTP13a (as a control) to further verify the uptake of Suc by MdSTP13a for pollen tube growth. Incubation of the developing pollen tubes with 1 mM of esculin for 40 min showed that as-MdSTP13a-transfected pollen tubes had significantly lower fluorescence intensity than s-MdSTP13a-transfected pollen tubes (Figure 6), indicating as-MdSTP13a transfection led to a significant reduction in the uptake capacity for Suc by pollen tubes.

Pollen with Lower Sorbitol Have Reduced Pollen Tube Growth on Glc and Suc, but not on Maltose
In earlier work, transcript levels of MdMYB39L, a MYB transcription factor, and its several potential target genes including MdSTP13a were found to be downregulated in developing stamens and growing pollen tubes of the two transgenic lines (A4 and A10) with antisense suppression of A6PR expression. The decreased sorbitol synthesis led to abnormal stamen development and reduced pollen tube growth in the antisense lines via MdMYB39L and application of exogenous sorbitol partially restored the phenotype (Meng et al., 2018a). Measurements of sugar concentrations in mature pollen grains confirmed that sorbitol concentration was significantly lower in both A4 and A10 than in wild type, with A10 having slightly lower sorbitol than A4, whereas no difference was detected in concentrations of other sugars (Supplemental Figure 12). We reasoned that, if MdSTP13a is essential for sorbitol-modulated pollen tube growth, pollen tube growth of the transgenic lines would be dependent on whether the carbon source used in the medium is a substrate of MdSTP13a or not. We conducted pollen tube growth assays in media with Glc, Suc, or maltose as sole carbon source. Transcript levels of both MdMYB39L and MdSTP13a were significantly lower in A4 and A10 (B) Quantification of esculin uptake in pollen tubes transfected with s-MdSTP13a or as-MdSTP13a. Relative fluorescence was measured via the software ImageJ (https://imagej.nih.gov/ij/). Data are mean 6 SE. ** Represents significant differences in comparison with control using Student's t test at P < 0.01. Data are mean 6 SE, n 5 10. ** Represents significant differences using Student's t test at P < 0.01. pollen tubes than in the wild-type control regardless of the carbon source used (Figures 7A and 7B), with that of MdSUT2 and medium sugar composition unaltered between genotypes on Suc medium (Supplemental Figure 13). However, A4 and A10 exhibited slower pollen tube growth relative to the wild-type control on Glc and Suc media, but not on maltose medium ( Figure 7C; Supplemental Figure 14). We also measured the Glc-and Sucelicited transmembrane currents in pollen tube protoplasts isolated from wild type (control), A4, and A10 lines by patch-clamp. Significantly smaller currents were detected in pollen tubes of A4 and A10 transgenic lines compared with those of the wild-type control upon adding 50 mM of Glc or Suc (Figures 7D and 7E; Supplemental Figure 15). These data are consistent with those when the transcript level of MdSTP13a was suppressed with antisense oligonucleotide transfection ( Figures 5C and 5D). Taken together, these data demonstrate that transgenic pollen with decreased sorbitol level have reduced pollen tube growth on Glc and Suc, and the reduced growth is associated with decreased MdSTP13a-mediated uptake of Glc and Suc into the growing pollen tubes. The lack of difference in pollen tube growth between the two transgenic lines and wild type on maltose is consistent with maltose not being a substrate of MdSTP13a.
Esculin uptake assays during pollen tube growth showed that both A4 and A10 pollen tubes had significantly lower fluorescent intensity than the wild-type control pollen tubes (Supplemental Figure 16), indicating transgenic pollen with decreased sorbitol levels have significantly lower uptake capacity for Suc to support pollen tube growth. The corresponding downregulation of MdSTP13a expression to antisense oligonucleotide transfection of MdMYB39L obtained earlier (Meng et al., 2018a) and parallel changes in transcript levels between MdMYB39L and MdSTP13a during pollen tube growth suggest transcriptional regulation of MdSTP13a by MdMYB39L. We found two typical MYB binding sites, MBS1 (CAACTG) and MBS2 (CGGTCA), in the promoter region (1,017 bp) of MdSTP13a via sequencing analysis ( Figure 8A). Subsequently, yeast onehybrid (Y1H) assays were performed to determine if MdMYB39L binds to the MdSTP13a promoter. MdMYB39L was inserted into the pGADT7 vector, and various lengths of promoter fragments of MdSTP13a were generated and inserted into the pAbAi vector as shown in Figure 8B. Yeasts cotransformed with pGADT7-MdMYB39L and pAbAi-MdSTP13a-Pro, containing either one or both MBS binding sites, grew normally in SD-Leu-Ura/200 ng mL 21 aureobasidin A (AbA) plates, but the negative control or the promoter region without MBS motif did not ( Figure 8C).
Electrophoretic mobility shift assay (EMSA) was performed using prokaryon-expressed and purified MdMYB39L-GST fusion proteins. An oligonucleotide probe from the MdSTP13a promoter containing either MBS site was used for DNA-affinity trapping. When the MdMYB39L protein was incubated with biotin-labeled MdSTP13a promoter oligonucleotides, a DNA-protein complex with slower mobility was observed for promoter fragments containing either MBS site ( Figure 8D, lanes 3 and 4), but not for promoter fragments without MBS ( Figure 8D, lanes 1 and 2). This suggests the specific binding of MdMYB39L to MdSTP13a promoter requires the MYB motif.
To verify in vivo binding of MdMYB39L to the MdSTP13a promoter, chromatin immunoprecipitation (ChIP)-PCR assay was conducted using transgenic apple calli overexpressing MdMYB39L-GFP and GFP control. The MBS promoter regions of MdSTP13a, but not the other cis-acting regulatory elements (CAREs, as negative controls), were enriched by ChIP-PCR in the 35S:MYB39L-GFP transgenic calli compared with the 35S:GFP control ( Figure 8E). These results support the specific binding of MdMYB39L to the MBS motifs in the promoter of MdSTP13a.
To confirm that MdMYB39L transcriptionally regulates MdSTP13a during pollen tube growth, we suppressed the expression of MdMYB39L via antisense oligonucleotide transfection of MdMYB39L (as-MdMYB39L). Compared with the sense oligonucleotide transfection (s-MdMYB39L) and transfection agent control, as-MYB39L significantly decreased the transcript level of both MdMYB39L and MdSTP13a ( Figures 9A and 9B), without affecting MdSUT2 expression (Supplemental Figure 17). This led to a significant reduction in pollen tube growth on Glc and Suc medium, but not on maltose medium ( Figure 9C; Supplemental Figure 18). These results demonstrate that MdMYB39L regulates the expression of MdSTP13a in modulating pollen tube growth by altering MdSTP13a's capacity for the uptake of Glc and Suc.

Sorbitol-Modulated Pollen Tube Growth Is Dependent on both MdMYB39L and MdSTP13a
It was previously demonstrated that addition of sorbitol to in vitro pollen germination medium partially restored pollen tube growth of A6PR antisense lines A4 and A10 on Suc and hexoses (Meng et al., 2018a). We predicted that, if sorbitol-modulated pollen tube growth is dependent on both MdMYB39L and MdSTP13a, antisense suppression of either of them would diminish the partial restoration of pollen tube growth by addition of sorbitol in the germination medium.
When 25 mM of sorbitol was added to the pollen germination medium with 5% Suc in combination with sense or antisense oligonucleotide MdMYB39L transfection, only transfection with as-MdMYB39L significantly decreased the transcript level of MdMYB39L in wild-type pollen and consequently reduced pollen tube growth (Figure 10), which is similar to the result obtained previously without sorbitol addition (Figure 9). For A4 and A10 pollen, addition of sorbitol increased the expression of both MdMYB39L and MdSTP13a and partially restored pollen tube growth, but this partial restoration was blocked by as-MdMYB39L transfection that significantly reduced MdMYB39L transcript level relative to s-MdMYB39L transfection (Figure 10; Supplemental Figure 19). Similar results were obtained when 5% Glc was used as the carbon source (Supplemental Figure 20). These data confirm that sorbitol modulates pollen tube growth via MdMYB39L.
To determine if sorbitol-modulated pollen tube growth via MdMYB39L requires the uptake of sugars mediated by MdSTP13a, we added 25 mM of sorbitol to the pollen germination medium with 5% Suc in combination with sense or antisense oligonucleotide MdSTP13a transfection. In wild-type pollen, only transfection with as-MdSTP13a significantly decreased the transcript level of MdSTP13a, leading to reduced pollen tube growth. Addition of sorbitol increased the expression of both MdMYB39L and MdSTP13a in A4 and A10 pollen tubes and partially restored their growth, but this partial restoration was blocked by as-MdSTP13a transfection that significantly reduced MdSTP13a transcript level relative to s-MdSTP13a transfection (Figure 11; Supplemental Figure 21). Similar results were obtained when 5% Glc was used as the carbon source (Supplemental Figure 22). Combined with the data obtained on transcriptional regulation of MdSTP13a by MdMYB39L (Figures 8, 9, and 10), these results clearly indicate that sorbitol-modulated pollen tube growth requires MdSTP13a-mediated uptake of Suc and Glc, which is regulated by MdMYB39L in response to sorbitol.

DISCUSSION
In earlier work, we found that sorbitol acts as a signal regulating stamen development, pollen germination, and tube growth in apple via MYB39L (Meng et al., 2018a). One of the genes downregulated by decreased sorbitol level in stamens and pollen tubes is MdSTP13a, an STP homologous to STP13 in Arabidopsis.
Here, we demonstrate that MdSTP13a takes up Suc as well as hexose for in vitro pollen tube growth and is transcriptionally regulated by MdMYB39L in response to sorbitol. This reveals a situation where a plasma membrane-localized sugar transporter serves as a dual transporter for hexose and Suc, and is modulated by a sugar alcohol in mediating sugar uptake for a key plant developmental process.

MdSTP13a Is a Dual Sugar Transporter Essential for Apple Pollen Tube Growth
MdSTP13a was first identified as one of the downregulated genes by decreased sorbitol level in stamens and pollen tubes of transgenic apple trees with antisense repression of A6PR (Meng et al., 2018a). Suppression of MdSTP13a expression via antisense oligonucleotide transfection of wild-type pollen indicates that it plays an essential role in pollen tube growth on Glc (Figure 1). MdSTP13a localizes to the plasma membrane of pollen tubes ( Figure 2). 14 C-Glc uptake assays of MdSTP13a expressed in hexose transport-deficient yeast strain EBY.VW4000 confirmed that it transports Glc and other hexoses, which is similar to Arabidopsis STP13 (Nørholm et al., 2006). The affinity of MdSTP13a for Glc (K M 5 153.7 mM) is lower than that of AtSTP1, STP4, STP6, STP8, STP9, STP10, STP11, and STP12 (K M 5 7.6 to 84 mM) and CsHT1 (K M 5 107.3 mM), but higher than that of AtSTP3 (K M 5 2 mM; Truernit et al., 1996;Büttner et al., 2000;Sherson et al., 2000;Schneidereit et al., 2003Schneidereit et al., , 2005Scholz-Starke et al., 2003;Cheng et al., 2015;Rottmann et al., 2016Rottmann et al., , 2018a. The first clue that MdSTP13a also functions as a Suc transporter came from inhibition of its 14 C-Glc uptake by Suc and its analog esculin, and subsequently this inhibition was determined to be competitive in nature by 14 C-Glc uptake assays with increasing Suc concentrations (Figure 3). 14 C-Suc uptake assays on hexose transportand SUC2-deficient yeast strain CSY4000 complemented with MdSTP13a confirmed the uptake of Suc and its competitive inhibition by Glc (Figure 4). Additional verification of the competitive inhibition between Suc and Glc was obtained by 14 C-sugar uptake assays of MdSTP13a expressed in the widely used Suc uptakeand SUC2-deficient yeast strain, SUSY7/ura3 (Supplemental Figure 6D). Interestingly, the affinity of MdSTP13a for Suc (K M 5 66.9 mM) is close to the higher end of high-affinity dicot SUTs/ SUCs (K M 5 66 mM to 2 mM; Stadler et al., 1995;Stadler and Sauer, 1996;Kühn et al., 1997;Barker et al., 2000;Weise et al., 2000;Knop et al., 2004;Sivitz et al., 2007;Kühn and Grof, 2010) and much higher than that of other SUTs/SUCs (K M 5 2 to 20 mM; Rae et al., 2005;Reinders et al., 2006). Collectively, these data demonstrate that MdSTP13a is a dual transporter for hexose and Suc.
MdSTP13a is specifically expressed in stamens and pollen tubes (Meng et al., 2018a;Supplemental Figure 3). Several lines of evidence support its uptake of Suc as well as Glc for pollen tube growth. First, when Suc, Glc, or maltose was used as sole carbon source for wild-type apple pollen transfected with antisense oligonucleotide of MdSTP13a, reduced pollen tube growth was detected on both Suc and Glc but not on maltose. This is consistent with the substrate specificity of MdSTP13a obtained in yeast that both Suc and Glc are taken up by MdSTP13a but maltose is not (Figures 3E and 4E). Second, hexoses derived from Suc cleavage by cell wall invertase contributes very little to apple pollen tube growth on Suc medium. Measurements of sugar composition of the Suc medium over the duration of pollen tube growth found very low concentrations of Glc and Fru, and even the highest concentrations of Glc and Fru detected did not support pollen tube growth beyond the no-sugar control (Supplemental Figure 8). Transfection of pollen with antisense oligonucleotide of the main cell wall invertase gene, MdCWI2, did not alter pollen tube growth either (Supplemental Figure 9). Third, patch-clamp analysis of pollen protoplasts showed significantly lower transmembrane currents elicited by Suc and Glc in the wild-type pollen tubes transfected with antisense oligonucleotide of MdSTP13a compared with the sense MdSTP13a-transfected control (Figures  (E) Glc-induced changes in current density (top) obtained from pollen tube protoplasts of CK (n 5 8), A4 (n 5 6), and A10 (n 5 6) and Suc-induced changes in current density (bottom) obtained from pollen tube protoplasts of CK (n 5 8), A4 (n 5 7), and A10 (n 5 7). Data are mean 6 SE, n 5 3 in (A), (B), (C), and (D). Different letters (a, b, c) indicate significant differences between genotypes using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA. 5C and 5D). As protoplasts do not have cell walls and the measurements of individual protoplasts were made in a few minutes, cleavage of Suc into hexoses is unlikely to contribute significantly to the Suc-elicited current. Finally, esculin assays indicated that antisense MdSTP13a transfection significantly reduced the uptake capacity for Suc by pollen tubes (Figure 6). MdSTP13a-mediated uptake of Suc being primarily responsible for apple pollen tube growth on Suc medium is also corroborated by the fact that antisense suppression of the most highly expressed SUT in apple pollen, MdSUT2, did not affect pollen tube growth (Supplemental Figure 10) and the transcript level of MdSUT2 was not altered by antisense suppression of A6PR or MdMYB39L (Supplemental Figures 13 and 17).
The reliance of apple pollen tube growth almost entirely on direct uptake of Suc rather than cleavage of Suc by cell wall invertase on Suc medium contrasts with that of species such as tobacco (Goetz et al., 2001), Arabidopsis (Hirsche et al., 2009), and oilseed rape (Engelke et al., 2011), where cell wall invertase is found to be essential for pollen development and pollen tube growth. For petunia, almost all the Suc is converted to Glc and Fru within 13 h in the Suc medium (Ylstra et al., 1998). Even in these cell wall invertase-dependent species, however, uptake of both hexose and Suc takes place simultaneously during in vitro pollen tube growth as demonstrated in tobacco (Goetz et al., 2017) before all the Suc is converted to hexoses. In all the species reported so far, uptake of hexose and Suc for pollen tube growth is mediated by hexose-transporting STPs and SUTs/SUCs, respectively. However, uptake of both hexose and Suc is mediated by MdSTP13a in apple. This dual transport function might have reduced the need for reliance on cell wall invertase for conversion of Suc to hexoses before uptake in this species during in vitro pollen tube growth. It should be noted, however, that the stigma exudates of apple flowers contain Glc, Fru, and Suc (Pusey et al., 2008). The hexose is either the result of cell wall invertase activity in the stigma tissue or efflux from the stigma cells after hydrolysis of Suc by neutral invertase inside the cells. The sugar profile in the apoplast of transmitting tissues remains to be characterized for apple, but regardless of the ratio between hexose and Suc MdSTP13a is most likely the main sugar transporter that takes up sugars for pollen tube growth in vivo inferred from our in vitro data.
How MdSTP13a evolved for uptake of both hexose and Suc in apple is not clear. MdSTP13a is homologous to AtSTP13, but AtSTP13 transports only hexose and is implicated in programed cell death (Nørholm et al., 2006). MdSTP13a does not resemble any of the SUTs/SUCs reported, but has a high affinity for Suc than for Glc (Figures 3 and 4). Electrophysiological analyses of Arabidopsis tonoplast monosaccharide transporters (TMT), TMT1 and TMT2 (later renamed as tonoplast sugar transporter [TST1 (E) ChIP-PCR confirmation of the binding of MdMYB39L protein to MBS1 and MBS2 in the MdSTP13a promoter. The MdMYB39L-DNA complex was coimmunoprecipitated from MdMYB39L-GFP transgenic apple calli using a GFP antibody, with empty GFP vector transgenic apple calli as a negative control. Data are mean 6 SE, n 5 3. ** Represents significant differences using Student's t test at P < 0.01. and TST2]) and sugar beet (Beta vulgaris) TST1 indicate that they are sugar/H 1 antiporters for vacuolar uptake of both hexoses and Suc whereas BvTST2.1 is specialized for Suc uptake into the vacuole in sugar beet taproots (Schulz et al., 2011;Jung et al., 2015). It is suggested that the substrate specification toward Suc by BvTST2.1 is most likely the result of breeding for high Suc levels in this crop over the last 200 years from the broader substrate specificity of TST-type transporters for both monosaccharides and disaccharides (Jung et al., 2015). Recently, a plastidic sugar transporter, pSuT1, previously named as "vacuolar Glc transporter," VGT3, in Arabidopsis, was found to use both Glc and Suc as substrates based on transport analyses of yeast cells expressing a truncated, vacuole-targeted version of pSuT1 (Patzke et al., 2019). Nonaqueous fractionation of subcellular sugar distribution supports the role of pSuT1 in mediating Suc export from the chloroplast for inflorescence development and cold stress tolerance. All these three types of transporters-STPs, TSTs, and VGTs-belong to the large monosaccharide sugar transporter-like family, but apple MdSTP13a is the only one that has been found to transport both hexose and Suc in symport with protons across the plasma membrane. It would be of great interest to determine if orthologs of MdSTP13a in other sorbitol-synthesizing species in the Rosaceae family have similar transport properties, and what sequences/structures define the broader substrate specificity of MdSTP13a compared with other STPs that transport only hexoses.

MdSTP13a is Transcriptionally Regulated by MdMYB39L, and MdSTP13a-Mediated Sugar Uptake Is Required for Sorbitol-Modulated Apple Pollen Tube Growth
Decreased sorbitol levels in stamens of A6PR antisense plants were previously demonstrated to result in abnormal stamen development and reduced pollen germination and tube growth via MYB39L (Meng et al., 2018a). Bioinformatics analysis identified MdSTP13a as one of the potential target genes of MYB39L. (C) In vitro pollen tube growth of pollen transfected with s-MdMYB39L, as-MdMYB39L, or Control on 5% Glc, Suc, or maltose medium. Data are mean 6 SE, n 5 3. Different letters (a, b) indicate significant difference between groups using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA.
Subsequently, Y1H assay, EMSA, and ChIP-PCR analysis confirmed the binding of MYB39L to the MYB sites in the promoter of MdSTP13a (Figure 8). Antisense repression of MdMYB39L via oligonucleotide transfection decreased the transcript level of MdSTP13a and pollen tube growth of wild-type apple pollen ( Figure 9) and blocked partial restoration of pollen tube growth of A6PR antisense lines A4 and A10 by addition of sorbitol (Figure 10), providing the mechanism for transcriptional regulation of MdSTP13a in response to sorbitol level.
The requirement of MdSTP13a-mediated sugar uptake for sorbitol-modulated pollen tube growth in apple is supported by multiple lines of evidence. First, A6PR antisense lines have decreased transcript levels of MdMYB39L and MdSTP13a during pollen tube growth regardless of sugar type used in the medium, and have reduced pollen tube growth on Glc and Suc, but not on maltose ( Figures 7A, 7B, and 7C). In addition, patch-clamp analysis of pollen tube protoplasts showed that pollen tubes of the A6PR antisense lines have significantly lower transmembrane currents elicited by Glc and Suc and lower esculin uptake ( Figures  7D and 7E; Supplemental Figure 16). Second, transfection of wildtype pollen with antisense oligonucleotide of MdMYB39L significantly decreased the transcript levels of MdMYB39L and MdSTP13a regardless of sugar type, and reduced pollen tube growth on Glc and Suc, but not on maltose (Figure 9). In this and all other cases (Figures 5 and 7), treatment (antisense repression) effects on pollen tube growth in relation to sugar type (Glc and Suc versus maltose) match the substrate specificity of MdSTP13a obtained in yeast ( Figures 3D and 4D), i.e. both Glc and Suc are taken up by MdSTP13a whereas maltose is not. This has allowed us to use maltose as a control to demonstrate that sorbitolmodulated pollen tube growth via MdMYB39L relies on MdSTP13a only when the carbon source used is a substrate of MdSTP13a such as Suc and Glc. Both Glc and Suc are relevant to the in vivo sugar supply as they are present in the stigma exudates of apple flowers (Pusey et al., 2008) and A6PR antisense lines (A4 and A10) have reduced pollen tube growth in vivo (Meng et al., (C) Responses of pollen tube growth to the addition of sorbitol in combination with s-MdMYB39L or as-MdMYB39L transfection. Data are mean 6 SE. n 5 3. Different letters (a, b) indicate significant differences using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA. 2018a). Maltose fully supports apple pollen germination and tube growth as demonstrated, but uptake of maltose is apparently achieved by a separate transport system, not by MdSTP13a, and as a result maltose-supported pollen tube growth is not dependent on MdSTP13a, MdMYB39L, or sorbitol. Finally, addition of sorbitol partially restored pollen tube growth of A6PR antisense lines on Suc and Glc media, but this restoration was blocked by transfection with antisense oligonucleotide of MdSTP13a (Figure 11).
Both Suc and Glc have been found to act as signals regulating pollen germination and tube growth. Suc is required for tobacco pollen tube elongation as a metabolic signal (Goetz et al., 2017), most likely via the SnRK1-mediated signal transduction pathway as antisense suppression of SnRK1 led to abnormal pollen development and male sterility earlier (Zhang et al., 2001). Glc inhibits pollen tube growth of Arabidopsis in a hexokinase 1-dependent manner (Rottmann et al., 2016(Rottmann et al., , 2018c. However, in apple, a sorbitol synthesizing and transporting species, sorbitol appears to fulfill the signaling role for pollen germination and tube growth. Sorbitol accounts for 60% to 80% of the photosynthates produced in apple source leaves (Bieleski, 1982;Cheng et al., 2005) and stores more energy than Suc and Glc on an equivalent molar carbon basis (Loescher, 1987), making it an ideal candidate for representing the plant energy status. The high energy requirement for pollen development and tube growth might have prompted apple to adopt sorbitol as a signaling molecule for these developmental processes although sorbitol is much less abundant than Suc in mature pollen grains (Supplemental Figure 12). Based on the findings presented here and elsewhere (Meng et al., 2018a), we propose the following model for sorbitol-modulated pollen tube growth (Figure 12): Sorbitol present in the pollen grain or taken up from the apoplast of the surroundings tissues or in vitro medium presumably by polyol/monosaccharide transporters upregulates the expression of MdMYB39L in the nucleus, and subsequently MdMYB39L binds to the promoter of MdSTP13a to activate its expression. The MdSTP13a protein is targeted to the (C) Responses of pollen tube growth to the addition of sorbitol in combination with s-MdSTP13a or as-MdSTP13a transfection. Data are mean 6 SE. n 5 3. Different letters (a, b) indicate significant differences using Tukey's Honest Significant Difference test at P < 0.05 after one-way ANOVA.
plasma membrane and mediates the uptake of both Suc and hexose in a competitive manner to provide energy for pollen tube growth. How sorbitol is sensed and the signal is transduced to alter the expression of MdMYB39L warrants further work. As sorbitol was recently found to modulate resistance of apple leaves to Alternaria alternata by regulating the expression of an NLR gene, NLR16, via a WRKY transcription factor, WRKY79 (Meng et al., 2018b), it appears that sorbitol serves as a signal for biotic stress tolerance as well as for plant development. In light of the presence of both Glc and Suc signaling in apple (Hu et al., 2016;Liu et al., 2017), how sorbitol signaling interacts with the network of hexose/Suc signaling in sugar alcohol synthesizing species deserves further research.

Plant Material and Growth Conditions
Untransformed wild-type 'Greensleeves' apple (Malus domestica; CK) and two transgenic lines with antisense suppression of MdA6PR (A4 and A10) were used in this study. Transcript levels of MdA6PR in the leaves of A4 and A10 are ;10% of that in CK; both A4 and A10 have significantly lower concentrations of sorbitol in leaves and stamens, with sorbitol level being lower in A10 than in A4 in stamens (Cheng et al., 2005;Wu et al., 2015;Meng et al., 2018a). The trees were 6 years old and grown on M.26 rootstock in 20liter containers outside at Cornell Experimental Orchards. They were arranged in a completely randomized design in three replicates per genotype, with two trees per replicate. The trees were maintained under standard horticultural management and disease and insect control. At the popcorn stage, flowers were harvested for anther collection and the anthers were dried under incandescent lights. The dried anthers with the released pollen grains were stored at 220°C.
Calli derived from apple cv Orin were used for ChIP-PCR assays. They were cultured on Murashige and Skoog (MS) medium with 30 g/L of Suc, 1.5 mg/L of 6-BA, and 0.5 mg/L of indole-3-acetic acid at 25°C in the dark.

Phylogenetic Analysis of MdSTP13s and Promoter Analysis of MdSTP13a
Phylogenetic analysis of the protein sequence of MdSTP13a was conducted along with those of the Arabidopsis STP family members, which were downloaded from The Arabidopsis Information Resource (https:// www.arabidopsis.org/). The phylogenetic tree was built by the neighborjoining method using the software MEGA7 (https://www.megasoftware. net/). The alignment used to generate the phylogenetic tree is supplied in the Supplemental File.

In Vitro Pollen Tube Growth Experiments
Pollen grains were hydrated for 15 min and then cultured in liquid germination medium (0.01% [w/v] H 3 BO 3 , 0.015% [w/v] CaCl 2 , and 5% [w/v] sugar, at pH 5.8) as described by Meng et al. (2014b). To test the effect of carbon source on pollen tube growth, 5% (w/v) Glc, Suc, or maltose was used as indicated. Pollen grains were incubated at 25°C in the dark for 120 min to determine pollen tube growth, and pollen tubes were photographed by light microscopy at 60, 90, and 120 min. Each treatment was replicated three times, with at least 100 germinated pollen grains measured per biological replicate (corresponding to those in the field experiment) at each time-point. Germinated pollen grains were also collected for RNA extraction and gene expression analysis through RT-qPCR. All pollen tube growth experiments were repeated three times. Statistical analysis results are provided in the Supplemental Data Set.

Antisense Oligonucleotide Transfection
Antisense oligonucleotide transfection was performed on pollen grains as described by Meng et al. (2014a) to suppress target gene expression. For each gene, a phosphorothioated antisense oligodeoxynucleotide was designed and its sense oligodeoxynucleotide was used as a control. After being hydrated for 15 min, the pollen grains were treated with antisense, sense oligodeoxynucleotide, or transfection agent alone (Lipofectamine 3000; Invitrogen), and then cultured in liquid germination medium for 120 min. The oligodeoxynucleotide sequences used in this study are listed in the Supplemental Table. RNA Extraction and RT-qPCR A fixing solution (4% [v/v] formaldehyde, 10% [w/v] Suc, 50 mM of 2,2'piperazine-1,4-diylbisethanesulfonic acid, 5 mM of MgSO 4 , and 0.5 mM of CaCl 2 at pH 7.0) was added to liquid pollen germination medium, and pollen tubes were collected by centrifugation at 12,000g for RNA isolation. Total RNA was extracted from pollen tubes using the modified CTAB method as described in Gasic et al. (2004). After digestion with DNase I (Thermo Fisher Scientific), 2 mg of total RNA was reverse-transcribed to cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). RT-qPCR analysis was performed with iQ SYBR Green Supermix in an iCycler iQ5 system (Bio-Rad) following the manufacturer's instructions. Each reaction was replicated three times per biological replicate, with ACTIN as an internal reference gene. The relative expression of each gene was calculated using the 2 2DCT method (Udvardi et al., 2008). Sequences of the primers for RT-qPCR used in this study are listed in the Supplemental Table.

Subcellular Localization of MdSTP13a
Subcellular localization of MdSTP13a was conducted in both apple callus protoplasts and pollen tubes. The coding sequence (CDS) of MdSTP13a MdSTP13a protein resides in the plasma membrane of pollen tubes and mediates the uptake of Suc and hexose in a competitive manner to provide energy for pollen tube growth. Sorbitol present in the pollen grain or taken up from the apoplast of the surroundings tissues or in vitro medium presumably by polyol/monosaccharide transporters upregulates the expression of MdMYB39L in the nucleus, and subsequently MdMYB39L binds to the promoter of MdSTP13a to activate its expression. was amplified by PCR using the primers MdSTP13a-GFP-F/R (Supplemental Table), and the sequenced amplicon was inserted into the SacI/Hind III cloning site of a modified CaMV35S-GFP vector (Lin et al., 2009) to generate p35S:MdSTP13a-GFP. After purification with a Nucle-oBond Xtra Midi kit (Macherey-Nagel), both p35S:MdSTP13a-GFP and p35S:GFP were introduced into apple callus protoplasts as described for Arabidopsis (Arabidopsis thaliana; Sheen, 2001). The transformed protoplasts were incubated at 23°C in the dark for 16 h. For localization analysis in pollen tubes, MdSTP13a was cloned into the Lat52:pUC-eGFP vector, which contained the pollen-specific promoter of Lat52 (Albani et al., 1991;Khurana et al., 2012). Then the recombinant plasmid was mixed with spermidine (at a final concentration of 0.1 M), standard concentration of gold, and CaCl 2 (at a final concentration of 2.5 M). The mixture was oscillated for another 10 min, and then was transformed into the 15-min hydrated apple pollen by particle bombardment. The transformed pollen grains were subsequently cultured on germination medium in the dark for 4 h. The GFP signal was detected at 500 to 530 nm using a confocal laserscanning microscope (LSM 700; Carl Zeiss) after excitation at 488 nm. The plasma membrane stains, Deep Red (C10064, Thermo Fisher Scientific; Excitation/Emission: 649/666 nm) and FM 4-64 (T-13320, Thermo Fisher Scientific; Excitation/Emission: 558/734 nm), were used for colocalization of MdSTP13a protein in apple callus protoplasts and pollen tubes, respectively.

Functional Characterization of MdSTP13a by Heterologous Expression in Yeast and 14 C-Sugar Uptake Assays
The CDS of MdSTP13a and AtSTP13 were cloned into the yeast expression vector pDR196 using the primers of MdSTP13a-196-F/R and AtSTP13-196-F/R, respectively (Supplemental Table), and then transformed into the hexose transport-deficient yeast strain EBY.VW4000, hexose transportand SUC2-deficient yeast strain CSY4000, and Suc uptake-and SUC2deficient yeast strain SUSY7/ura3 using the lithium acetate method, with empty vector pDR196 as a control (Riesmeier et al., 1992;Wieczorke et al., 1999;Rottmann et al., 2016). The EBY.VW4000 and CSY4000 positive clones were spotted in serial dilutions and cultured on the SD/-Ura plate, which was supplemented with 2% (w/v) maltose, Glc, or Suc as sole carbon source to test yeast growth. Yeast cultures were incubated 3 to 5 d at 30°C before photography.
Sugar uptake assays were performed as described in Cheng et al. (2015). Briefly, the transformed yeast strains carrying either recombinant plasmid or empty pDR196 were grown in liquid medium to an OD 623 value of 0.8, harvested by centrifugation at 1,240g, and then washed twice with 25 mM of phosphate buffered saline (PBS at pH 5.5). The collected yeast cells were resuspended in the same PBS buffer to an OD 623 value of 20 (;0.1 mg yeast/100 mL) for sugar uptake assay. Sugar uptake was detected by adding 14 C-Glc or 14 C-Suc (0.02 mCi) into the yeast cells to a final specified concentration (100 mM) and incubated at 30°C in a water bath of a shaker. After a timed incubation, yeast cells were collected via vacuum filtration onto 0.8-mm glass fiber filters (Whatman cat. no. 1827-024; GE Healthcare) and washed three times with 5 mL of ice-cold distilled water. The filters were placed into the scintillation solution (Ecoscint H) for liquid scintillation counting on a Multi-Purpose Scintillation Counter (model no. LS 6500; Beckman Coulter).

Isolation of Pollen Tubes Protoplasts and Analysis of Sugar Transport Activity via Patch-Clamp
Pollen grains were germinated and grown in 2 mL of liquid germination medium (0.01% [w/v] H 3 BO 3 , 0.015% [w/v] CaCl 2 , and 5% [w/v] sugar at pH 5.8) at 25°C for 1 h. The liquid germination medium was removed and replaced with the same volume of enzyme solution, which contained 0.1% (w/v) macerozyme R-10 (Onozuka), 0.2% (w/v) Cellulase RS-10 (Onozuka), 0.07% (w/v) Pectolyase Y-23 (Seishin), and 1% (w/v) BSA (Sigma) in basic solution (200 mM of KCl, 1 mM of CaCl 2 , and 2 mM MgCl 2 , at 1,200 mOsm). After incubation for ;2 to 5 min at 25°C, the enzyme solution was diluted with bath solution to stop the digestion, and filtered through 100-mm nylon mesh. The released pollen tube protoplasts were centrifuged at 500g for 5 min and washed twice with basic solution.
The whole-cell mode was used for patch-clamp electrophysiology as described in Wingenter et al. (2010) and Schulz et al. (2011). Both bath and pipette solutions contained 200 mM of KCl, 1 mM of CaCl 2 , and 2 mM MgCl 2 . The pH values were either adjusted with 10 mM of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid-Tris to pH 7.5 for pipette solution or with 10 mM of MES-Tris to pH 5.5 for bath solution. Kimax-51 glass capillaries (Kimble Products) were used to make patch pipettes, which were characterized by a resistance in the range of 3 to 5 mV in the bath solution. The whole-cell currents were recorded for 2.5 to 5 min after the whole cell configuration was achieved using an Axopatch-200B amplifier (Axon Instruments) connected to a computer via an interface (TL-1 DMA Interface; Axon Instruments). For the recordings, the holding potential was clamped to 0 mV and low-pass-filtered at 200 Hz. Current amplitudes of individual protoplasts were normalized to the whole-cell membrane capacitance C m to allow comparison of current magnitudes among different sizes of protoplast. To measure Glc-or Suc-induced currents, Glc or Suc was added to the bath solution to a final concentration of 50 mM after the whole-cell configuration was achieved, and the whole-cell currents were recorded. The software pCLAMP (v10.2; Axon Instruments) was used to acquire and analyze the whole-cell currents. The software SigmaPlot v11.0 (Systat Software) was used to draw current density voltage plots and for data analysis.

Esculin Uptake Assay of Yeast and Pollen Tubes
The yeast strain CSY4000 expressing MdSTP13a, AtSTP13, StSUT1, or empty vector pDR196 was used for esculin uptake assay following Gora et al. (2012). Briefly, yeast cells cultured overnight were collected by centrifugation at 1,240g, and then washed three times with 25 mM of PBS (at pH 5.5). They were recultured in PBS buffer containing 1 mM of esculin in a shaker at 200 rpm for 1 h at 30°C. Subsequently, the yeast cells were harvested by centrifugation and washed three times with PBS buffer to remove the leftover esculin. The yeast pellets were resuspended in PBS buffer and the esculin signal was detected at 450 nm using a confocal laserscanning microscope (LSM 700; Carl Zeiss) after excitation at 426 nm.
Pollen grains were germinated on 5% (w/v) maltose medium for 60 min. The maltose medium was chosen over 5% (w/v) Suc because the transcript level of MdSTP13a was decreased by as-MdSTP13a transfection or antisense repression of A6PR to a similar degree on Suc and maltose ( Figures  5A and 7B). This made it possible for esculin uptake to take place without any competition from Suc or any disturbance that rinsing off high concentrations of Suc from 5% (w/v) Suc medium would have on pollen tube function. Esculin was then added to the germination medium to a final concentration of 1 mM for a 40-min incubation. The pollen tubes were subsequently rinsed and mounted onto glass slides in the liquid medium without esculin. Fluorescence was detected with a model no. BX61 fluorescence microscope (Olympus) at an emission wavelength of 454 nm and 670 nm after excitation at 426 nm and 635 nm for esculin and Cy5labeled oligonucleotides, respectively. The relative fluorescence of esculin was analyzed by the software ImageJ (https://imagej.nih.gov/ij/).

Y1H Assay
The Yeast One-Hybrid System-Matchmaker Gold Kit (catalog No. 630491; Clontech) was used for Y1H assay. Fragments of MdSTP13a promoter with various lengths generated via PCR were cloned into the pABAi vector using the primers listed in the Supplemental Table. The vectors were transformed into yeast strain Y1H Gold after linearization. Positive clones were selected through PCR and used to determine the minimal inhibitory concentration of AbA. The CDS of MdMYB39L was cloned into the pGADT7 vector to generate pGADT7-MdMYB39L (Primers: MdMYB39L-Y1H-F/R; Supplemental Table), which was then transformed into yeast strain Y1H Gold competent cells carrying pAbAi-MdSTP13a promoter. Cotransformed positive clones were spotted in a series of dilutions (1:1, 1:10, 1:100, and 1:1,000) and cultured on the SD/-Leu-Ura plate with or without 200 ng/mL of AbA at 30°C for 3 to 4 d. A pair of pGADT7-p531pAbAi-p53 was included as a positive control whereas pairs of pGADT7-MdMYB39L1pAbAi and pGADT71pAbAi-MdSTP13a Promoter were used as negative controls.

EMSA
The coding region of MdMYB39L was amplified via PCR (Primer: MdMYB39L-GST-F/R; Supplemental Table) and cloned into EcoRI/XhoI site of the pGEX-4T-1 vector. The recombinant plasmid (MdMYB39L-GST) was transformed into Escherichia coli strain BL21 (DE3). The expression of the MdMYB39L-GST fusion was induced by addition of 0.1 mM of isopropyl-D-1-thiogalactopyranoside in a 5-h incubation at 30°C. Production and purification of MdMYB39L-GST fusion protein was performed using Glutathione Sepharose 4B (cat. no. 17-0756-01; GE Healthcare). EMSAs were performed using the LightShift Chemiluminescent EMSA Kit (cat. no. 20148;Thermo Fisher Scientific) according to the manufacturer's instructions. Complementary pairs of 59-end biotin-labeled oligonucleotides of MdSTP13a promoter sequence (CARE1, CARE2, MBS1, and MBS2) were annealed and used as probes for the EMSA. The protein-DNA samples were then separated on 6% (w/v) polyacrylamide gels and signals were captured with a ChemiDoc XRS (Bio-Rad).

Apple Callus Transformation and ChIP-PCR Assay
Apple callus transformation was performed as described in An et al. (2012) with minor modifications. Full-length MdMYB39L cDNA was amplified using primers MdMYB39L-CHIP-F/R (Supplemental Table), and cloned into pGWB451 vector with a C-terminal GFP tag through Gateway BP and LR reactions (BP Clonase cat. no. 11789013 and LR Clonase cat. no. 11,791,019; Invitrogen). pGWB451-MdMYB39L and the empty vector were transformed into Agrobacterium GV3101, and positive clones were screened via colony PCR for callus transformation.
Three-week-old "apple cv Orin" calli were collected and suspended in the liquid medium of Agrobacterium GV3101 harboring pGWB451-MdMYB39L or the empty vector for 15 min with shaking (140 rpm) at 25°C. The calli were cocultured on solid MS medium (4.34 g/L of MS, 1.0 mg/ L of 6-BA, 1.0 mg/L of 2, 4-D, 30 g/L of Suc, and 7 g/L of agar at pH 5.8) at 25°C in the dark. After 2 to 3 d, the transgenic calli were washed three times with sterile water and then cultured on solid MS medium that contained 250 mg/L of carbenicillin and 30 mg/L of kanamycin for selection. ChIP-PCR was performed using a Pierce Agarose ChIP Kit (cat. no. 26,156;Thermo Fisher Scientific). Approximately 1 g of MdMYB39L-GFP or GFP control transgenic calli was cross linked in 1% (v/v) formaldehyde. Subsequently, the immunoprecipitate was used to isolate the protein-DNA complex with a GFP antibody (product no. PA-980A, lot No. RH236759; Invitrogen). The DNA was purified from the protein-DNA complex using the ChIP kit, and the abundance of each DNA fragment was quantified through RT-qPCR using the primers listed in the Supplemental Table  (

Analysis of Sugars via Gas Chromatography-Mass Spectrometry
Sugar composition and concentrations of pollen grains and Suc germination medium during pollen tube growth were analyzed by gas chromatography-mass spectrometry (GC-MS). For the pollen grains, sugars were extracted from 20 mg of pollen in 75% (v/v) methanol, with ribitol added as internal standard (Wang et al., 2010). After fractionation of the nonpolar metabolites into chloroform, 5 mL and 20 mL of the aqueous phase was taken for analysis of Suc and sorbitol/hexoses, respectively. For the 5% (w/v) Suc medium, 10 mL was taken at 0, 60, 90, and 120 min, diluted 40 times, with ribitol added as internal standard, and then 2 mL and 20 mL were used for analysis of Suc and hexoses, respectively. The samples were dried under vacuum without heat, and derivatized with methoxyamine hydrochloride and N-methyl-Ntrimethylsilyl-trifluoroacetamide sequentially for separation and analysis with a model no. 7890A GC/5975C MS (Agilent Technology) on a DB-5MS capillary column (20 m 3 0.18 mm 3 0.18 mm) with a 5-m Duraguard column (Agilent Technology) as previously described by Wang et al. (2010).

Supplemental
Supplemental Figure 8. Sugar composition of 5% Suc medium and pollen tube growth.
Supplemental Figure 9. Pollen tube growth in response to MdCWI2 antisense oligonucleotide transfection on 5% Suc medium.
Supplemental Figure 10. Pollen tube growth in response to MdSUT2 antisense oligonucleotide transfection on 5% Suc medium.
Supplemental Figure 11. Recordings of currents across pollen tube protoplasts transfected with sense or antisense oligonucleotide of MdSTP13a (s-MdSTP13a, as-MdSTP13a), or transfection agent alone (Control) before addition of Glc or Suc.
Supplemental Figure 12. Concentrations of sugars in mature pollen grains of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple.
Supplemental Figure 13. MdSUT2 expression during pollen tube growth of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple on 5% Suc medium and sugar composition of the medium.
Supplemental Figure 14. Photos of pollen tubes of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple after 2 h of growth on media with different sugars.
Supplemental Figure 15. Recordings of currents across pollen tube protoplasts of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple before addition of Glc or Suc.
Supplemental Figure 17. Expression levels of MdSUT2 in pollen transfected with sense or antisense oligonucleotide of MdMYB39L (s-MdMYB39L, as-MdMYB39L), or transfection agent alone (Control) during pollen tube growth on 5% Suc medium.
Supplemental Figure 18. Photos of pollen tubes transfected with sense or antisense oligonucleotide of MdMYB39L (s-MdMYB39L, as-MdMYB39L) or transfection agent only (Control) after 2 h of growth on media with different sugars.
Supplemental Figure 19. Photos of pollen tubes of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple in response to addition of sorbitol (25 mM) in combination with MdMYB39L sense or antisense oligonucleotide transfection (s-MdMY-B39La, as-MdMYB39L) on 5% Suc medium, with the 5% Suc medium only as control.
Supplemental Figure 20. Transcript levels of MdMYB39L and MdSTP13a and pollen tube growth in response to addition of sorbitol in combination with MdMYB39L sense or antisense oligonucleotide transfection on Glc medium. Figure 21. Photos of pollen tubes of wild-type control (CK) and A6PR antisense lines (A4 and A10) of 'Greensleeves' apple in response to addition of sorbitol (25 mM) in combination with MdSTP13a sense or antisense oligonucleotide transfection (s-MdSTP13a, as-MdSTP13a) on 5% Suc medium, with the 5% Suc medium only as control.

Supplemental
Supplemental Figure 22. Transcript levels of MdMYB39L and MdSTP13a and pollen tube growth in response to addition of sorbitol in combination with MdSTP13a sense or antisense oligonucleotide transfection on Glc medium.
Supplemental Table. List of primers and oligodeoxynucleotide sequences.
Supplemental File. Alignment used to generate the phylogenetic tree in Supplemental Figure 1.
Supplemental Data Set. Statistical analysis results.

ACKNOWLEDGMENTS
The Agilent GC-MS system used in this work was generously donated by David Zimerman (1954 Cornell University PhD in Pomology). We thank John Ward at the University of Minnesota for pDR196-StSUT1 construct, Ruth Stadler at Friedrich Alexander University of Erlangen-Nürnberg for yeast strain CSY4000, Christina Kühn at Humboldt University of Berlin for yeast strain SUSY7/ura3, Zhengxian Zhang at China Agricultural University for yeast strain EBY.VW4000, Takaya Moriguchi of the National Institute of Fruit Tree Science in Japan for apple cv Orin calli, Richard Gaisser for maintaining trees, Xiaochun Wu at Cornell University for technical assistance, and Robert Turgeon and Dagang Hu at Cornell University for critical reading of the article. This work was supported by the Cornell University Agricultural Experiment Station.