A spatiotemporal single-cell transcriptomic atlas reveals fertilization-induced auxin signaling driving strawberry fruit set and development

Abstract

Fertilization-triggered fruit set and morphogenesis constitute a complex developmental program involving diverse cell populations within fruit tissues. However, the cellular heterogeneity of this process has long obscured the precise contributions of distinct cell types. Here, by using woodland strawberry (Fragaria vesca) as a model, we integrated spatial transcriptomics, single-nucleus RNA sequencing (snRNA-seq), and spatial metabolomics to construct a comprehensive and high-resolution single-cell atlas of strawberry fruit, spanning post-fertilization, fruit set, and fruit enlargement stages. Pseudotime trajectories of the three main cell types revealed substantial developmental state transitions, with specific expression patterns observed for auxin signaling components. Integration of spatial metabolomics and the ProDR5::RUBY stable reporter line reveals an auxin distribution asymmetry that shapes fruit morphogenesis. This distribution asymmetry is jointly correlated with auxin transport components FvePIN1 and FvePIN5, with distinct temporal specificity in the receptacle. In response to auxin, FveARF6 acts as a key positive regulator of fruit development, as demonstrated by stable transgenic lines: FveARF6-RNAi, FveARF6-CRISPR knockout, and FveARF6-overexpression. Together, this work provides a spatiotemporal framework for understanding the cellular and molecular mechanisms underlying fertilization-induced fruit development and morphogenesis.

Fertilization-triggered fruit set and morphogenesis is a complex process involving multiple cell types. Here the authors use spatial transcriptomics and single nucleus RNA-seq to map fertilization induced changes in woodland strawberry.

Introduction

Fleshy fruits, characterized by the differentiation of the pericarp and accessory tissues into succulent structures, constitute a cornerstone of global nutrition and agricultural economy. Despite their agronomic and ecological prominence, the molecular choreography governing fruit set, the commitment to post-fertilization development, and fruit morphogenesis, which sculpts tissue architecture and organ identity, remains among the most enigmatic frontiers in plant developmental biology. This knowledge deficit stems from three convergent challenges: the temporal precision of initiation cues, the structural complexity of fruit tissues, and the daunting genomic heterogeneity of major crops, including allo-octoploid strawberry¹.

Critically, fleshy fruit development shares deeply conserved mechanisms, such as fertilization-triggered phytohormone biosynthesis and signaling, yet exhibits remarkable diversity in developmental trajectories. In botanical fruits, such as tomato (Solanaceae) and peach (Rosaceae), the fleshy part is derived from the ovary wall, whereas in accessory fruits like strawberry (genus Fragaria), the fleshy fruit develops from the receptacle. It is believed that this process is orchestrated by a complex interplay of delicate phytohormone coordination, cryptic transcriptional cascades, and epigenetic regulation². Up to now, how fertilization-induced signaling activates the fruit-specific developmental program in a particular floral tissue is still largely unknown.

The functional dissection of these mechanisms has been impeded by limitations in the classic model, Arabidopsis, which, while instrumental in illuminating dry fruit dehiscence mechanisms, lacks relevance to fleshy physiology. Up to now, most studies on fruit set and subsequent development were conducted in tomato³, a typical botanical fruit. The molecular mechanism of fruit set and development in accessory fruits has not been fully elucidated. This void positions the diploid woodland strawberry (Fragaria vesca, F. vesca) as an emerging Rosetta Stone-its miniature fruit, diploidy, rapid lifecycle, and sequenced genome enable CRISPR-based functional genomics⁴ and single-cell transcriptomics⁵. The distinctive accessory fruit form also offers an excellent system to study how seed-derived signals trigger the receptacle to develop into a fleshy fruit. Accessory fruit development in the wild strawberry F. vesca closely mirrors the receptacle-to-flesh transformation in cultivated octoploid strawberries, highlighting F. vesca as a convenient model to study cultivated strawberry as well as other fruit and vegetable crops.

The development of strawberry fruit requires fertilization-induced auxin signaling, as the removal of achenes completely abolished receptacle enlargement, whereas exogenous auxin application restored the receptacle's development into a fleshy fruit⁶. Successful fertilization is essential for strawberry fruit quality and yield, as unfertilized achenes prevent local phytohormone production, causing uneven growth and ultimately deformed fruit flesh. However, the pivotal role of fertilization-induced auxin signaling in strawberry fruit set and enlargement remained mechanistically unresolved for over six decades. Relatively recent transcriptomic profiling posited that auxin and gibberellin (GA) biosynthesis genes (such as YUC5, TAR1, GA20ox3) were upregulated in ghost (seed coat and endosperm) upon fertilization, and the synthesized auxin and GA were then transported to the receptacle⁷. More recently, AGL62 and AGL80, both MADS-box transcription factors, were identified to activate auxin synthesis in the endosperm, revealing the earliest molecular mechanism of fertilization-induced auxin biosynthesis in the fertilized endosperm⁸. In the receptacle, we also identified a compensatory feedback loop, in which auxin response factor ARF8 represses GA biosynthesis through the “RGA1-ARF8-GiD1c” module to prevent fruit over-expansion⁹. Nevertheless, key questions remain unanswered, particularly on how endosperm/seed-derived auxin is transported to the receptacle, how transporters establish a heterogeneous auxin distribution in the receptacle to pattern specific tissue types, and how the receptacle cells respond to such asymmetric auxin distribution in shaping the final fruit shape and texture.

Single-nucleus RNA sequencing (snRNA-seq), which bypasses protoplast isolation by profiling nuclei, avoids stress-induced transcriptional artifacts and has been applied in the integration of multiple tissues to generate comprehensive cellular and molecular landscapes^10,11. However, snRNA-seq alone lacks spatial context, constraining the precise assignment of cell identities within complex tissues. These limitations are especially pronounced in fleshy fruits, where intricate tissue architecture, high metabolite content, and extreme cellular hydration pose formidable technical challenges for studies at single-cell resolution. Spatial transcriptomics RNA sequencing (spRNA-seq) addresses these constraints by combining spatially resolved gene expression with high-resolution tissue imaging, thereby illuminating developmental processes in diverse plant systems, including Arabidopsis morphogenesis¹², maize ear patterning¹³, rice embryogenesis¹⁴, soybean root nodule maturation¹⁵, pitaya fruit senescence¹⁶, and cucumber floral bud determination¹⁷. Thus, integrating snRNA-seq with spatial transcriptomics offers a powerful strategy to interrogate large, heterogeneous, and metabolite-rich tissues such as strawberry fruit, owing to the approach’s single-cell resolution, robustness, mature tissue-sectioning workflows, and efficient mRNA capture. Notably, a spatial transcriptomic atlas at the single-cell level from fruit set to subsequent growth has yet to be established, leaving the cellular logic underlying these processes unresolved.

In this study, we constructed a comprehensive cell atlas of strawberry fruit encompassing major developmental stages to explore the molecular basis of fruit development and morphogenesis. Trajectory analysis and gene expression profiles revealed aging states and expression trends of auxin-signaling genes along fruit development. We identified two auxin efflux carriers, FvePIN1 and FvePIN5, that may collaboratively contribute to auxin distribution asymmetry in cortex and pith. By leveraging stable transgenic analyses, we characterized FveARF6 as a key transcriptional regulator of fertilization-induced auxin responses during fruit development. FveARF6 interacts with FveIAA26, which is degraded in response to exogenous NAA treatment. This study provides a spatiotemporal map of the signaling network underlying fleshy fruit development and morphogenesis in the receptacle. Our findings are broadly relevant to fruit formation in flowering plants and provides a scientific framework for enhancing fruit size and quality through genome editing-assisted breeding strategies.

Results

Staging fruit development in F. vesca

The strawberry fleshy fruit is derived from a modified stem tip (receptacle), with seed-bearing ovaries (achenes) dotting the surface of a supportive receptacle¹⁸. Together, the receptacle and the achenes constitute the strawberry fruit. The developmental progression and staging of F. vesca accession ‘Yellow Wonder 5AF7’ flower and early fruit was described previously^7,18. On the basis of morphological and histological features, we refined the stages further, classifying floral and fruit development into nine distinct stages (Fig. 1a): At stage 0, floral buds hardly expose opaque white petals and yellow-orange anthers inside, which is equivalent to floral stage 11 described previously¹⁸. Stage 1 corresponds to partially opened flowers at the onset of anthesis, equivalent to floral stage 12¹⁸. Stage 2 (2–4 days post-anthesis, DPA) possesses fully opened flowers with stigmas covered with yellow pollen grains, whereas stage 3 (6–7 DPA) is marked by petal abscission and stigma browning. At stage 4 (8–10 DPA), both achenes and receptacle enlarge, representing a key developmental transition. Stage 5 (10–15 DPA) is characterized by continued fruit enlargement and the onset of hollowness. At stage 6 (16–20 DPA), the receptacle attains its maximum size during the green fruit phase. By stage 7 (20–-25 DPA), receptacle softening begins, and achenes transition from green to yellow in the yellow-fruited accession (YW5AF7) and turn pink in the red-fruited accession (Rugen); this stage is similar to the white stage (also called turning stage) defined previously¹⁹. At stage 8 (25–30 DPA), the receptacle reaches full maturity, exhibiting increased softness and characteristic aroma. Among these stages, stage 2 (immediately post-fertilization) and stage 3 (fruit set) represent critical developmental turning points. Semi-thin sections and histological analysis with Safranin O-Fast Green staining reveal incipient lignin deposition in the outer achene layer (Fig. 1b and Supplementary Fig. 1), alongside marked increases in receptacle and achene diameters from stage 3 onward (Fig. 1c–f). Continued fruit enlargement during stages 4 and 5 highlights this period as a second pivotal phase in morphology. To dissect the mechanisms underlying fruit development and morphogenesis, we focused on stages 2, 3, and 5, representing post-fertilization, fruit set, and fruit enlargement, respectively, and profiled their molecular programs at single-cell resolution.

Fig. 1. Systematic staging of F. vesca fruit development.

a Side (top row), top (second row), and inner (third row) views of strawberry fruit development from stage 0 to stage 8. Stage 0 represents pre-fertilization. Floral buds hardly expose opaque white petals and yellow-orange anthers inside. At stage 1, flowers partially open at the onset of anthesis. At stage 2, fertilization has finished. Stigmas are covered with yellow pollen grains. Stage 3 marks the petal abscission and stigma browning. At stage 4, both the receptacle and achenes continue to expand. By stage 5, fruit enlargement continues. The fourth row, achenes isolated from fruits across different developmental stages. Scale bars: 1 mm and 0.2 mm (the fourth row). b Semi-thin sections and micrographs of early-stage fruits. Scale bar: 1 mm. The horizontal and vertical lines indicate the horizontal diameter and vertical diameter. c,d Quantification of fruit vertical (length) and horizontal (width) diameters across developmental stages. Data are presented as mean ± SD. Different lowercase letters above the bars indicate statistically significant differences among groups (P < 0.01, **), as determined by one-way ANOVA followed by Fisher’s LSD post-hoc test. e,f Quantification of achene vertical (length) and horizontal (width) diameters across developmental stages. Data are presented as mean ± SD (n represents biological replicates). Different lowercase letters above the bars indicate statistically significant differences among groups, as determined by one‑way ANOVA followed by Fisher’s LSD post‑hoc test (P < 0.01, **). Groups sharing the same letter are not significantly different from one another.

Cell atlas of F. vesca fruit development

To systematically elucidate the gene expression patterns within the fruit of F. vesca across the development process, we performed snRNA-seq on meticulously dissected whole fruit tissues, including the receptacle and achenes, across post-fertilization (stage 2), fruit set (stage 3), and fruit enlargement (stage 5) (Fig. 2a). Two biological replicates at each developmental stage exhibited a concordance with Pearson’s coefficients > 0.91, confirming the reproducibility of the snRNA-seq data (Supplementary Fig. 2). After quality control and filtering, we finally obtained 32,017 cells expressing 57,445 genes. A mean of 1442 genes and a median of 1608 unique molecular identifier (UMI) counts were detected (Supplementary Data 1). All samples were merged for cell clustering and embedding, yielding a total of 21 distinct clusters (Fig. 2b). Differential expression analysis was performed to identify marker genes for each cell cluster (Fig. 2c and Supplementary Data 2–5), which were then used to assist the cell cluster annotation. Among them, VARIANT IN METHYLATION 1 (VIM1, FvH4_5g24020) from cluster 2 was expressed in highly dividing cells (Liu, S., Yu, Y., Ruan, Y., Meyer, D., Wolff, M., Xu, L., Wang, N., Steinmetz, A., and Shen, W.-H. (2007). Plant SET- and RING-associated domain proteins in heterochromatinization. The Plant Journal 52, 914-926). DELAY OF GERMINATION 1 (DOG1, FvH4_3g07640), SUCROSE SYNTHASE 2 (SUS2, FvH4_4g18710), and PER1 (1-cysteine peroxiredoxin, FvH4_4g21120) were identified in cluster 4 (Supplementary Figs. 3, 4), and PER1 was reported to function in seed dormancy, expressed in the aleurone layer^20,21. RUBISCO BISPHOSPHATE CARBOXYLASE (RBCS2B, FvH4_6g26030) from cluster 5 was involved in the photosynthetic apparatus (Fig. 2c and Supplementary Fig. 4). Examples of marker genes for cluster 8 include PHLOEM INTERCALATED WITH XYLEM (PXY, FvH4_7g27270), SIEVE ELEMENT OCCLUSION-RELATED 1 (SEOR1, FvH4_5g06680), and TARGET OF MONOPTEROS 5 (TMO5, FvH4_4g07050) in vascular cells (Supplementary Fig. 3c). ALPHA-VPE (FvH4_6g36630) and STK (FvH4_5g32540), which are specifically expressed in integument (Coen, O., Fiume, E., Xu, W., De Vos, D., Lu, J., Pechoux, C., Lepiniec, L., and Magnani, E. (2017). Developmental patterning of the sub-epidermal integument cell layer in Arabidopsis seeds. Development 144, 1490-1497.)(Lee, T.A., Illouz-Eliaz, N., Nobori, T. et al. A single-cell, spatial transcriptomic atlas of the Arabidopsis life cycle. Nat. Plants 11, 1960–1975 (2025). https://doi.org/10.1038/s41477-025-02072-z, were specifically expressed in cluster 19 (Fig. 2c and Supplementary Fig. 5).

Fig. 2. snRNA-seq analysis of strawberry fruits across developmental stages.

a A schematic workflow of snRNA-seq and spatial transcriptomics profiling performed on strawberry fruits across developmental stages 2, 3, and 5. b UMAP visualization of 21 cell clusters identified by snRNA-seq at three different stages. Each dot represents a single cell, and distinct colors denote different clusters. c Marker gene expression across cell types identified from integrated snRNA-seq data. The dot size corresponds to the proportion of cells expressing the gene, and the color intensity indicates the average expression level.

Spatial transcriptomics facilitated cell type annotation and spatiotemporal analysis in developing strawberry fruit

To obtain a comprehensive annotation of the transcriptomics atlas, we further carried out 10× Visium spatial transcriptome sequencing (spRNA-seq) spanning the corresponding developmental stages in snRNA-seq, at post-fertilization (stage 2), fruit set (stage 3), and fruit enlargement (stage 5). Cryo-sectioned fruit tissues were placed onto spatially barcoded microarray slides covering ~5000 capture spots with 10× Visium platform (Supplementary Fig. 6a–d), enabling mRNA molecules to be spatially indexed, sequenced, and computationally mapped to their precise tissue coordinates (Fig. 3a). Relatively high reproducibility was also obtained between two biological replicates in the spRNA-seq data (Supplementary Fig. 6e). From the spatial transcriptomic profiling, we obtained 7654 spatial bins expressing 47,303 genes, with a median of 4262 counts and a median of 2250 features (Supplementary Fig. 6b–d and Supplementary Data 6). A total of 14 distinct cell types were classified based on the expression and spatial information from the chips, assisted by manual tuning (Fig. 3a and Supplementary Fig. 7). UMAP visualization revealed relatively independent territories of the 14 spatial cell types, reflecting different expression patterns (Supplementary Figs. 7, 8). From the spatial atlas, achenes (containing seed and ovary wall) could be detected at stage 2 and stage 3. Receptacle tissues were separated into three cell layers. First, Fruit-Epidermis occupies the outermost circumference of the fruit. Second, Parenchyma cells were clustered into three subtypes, including Cortex-1, Cortex-2, and Bundle-Sheath. Third, Fruit-Bundle cells were distributed in the cortex (Fig. 3a and Supplementary Figs. 8, 9). Marker genes for each cell type displayed specific expression patterns (Supplementary Fig. 10 and Supplementary Data 7–10). Most significantly, two cell types (Diamond, Fruit-Bundle-Adjacent) were identified in our spatiotemporal atlas. The Diamond cell type, representing the center and basal position of the receptacle fruit, consists of a cluster of cells forming a diamond shape that are positioned at the critical nexus between the stem and the receptacle; the Fruit-Bundle-Adjacent represents distinct layers of cells situated between the Fruit-Bundle and the cortex of the fruit. Other distinctly resolved cell populations were annotated as Sepal (SE), Pedicel (PE), Pedicel-Bundle, and Carpel (Ca).

Fig. 3. Extensive validation of correspondence between snRNA-seq and spRNA-seq of strawberry fruits across developmental stages.

a Cell type annotations overlaid on corresponding H&E-stained fruit section images at stage 2, stage 3, and stage 5. Each spot is colored according to its annotated cell type. b Heatmaps showing expression of the top 200 marker genes identified from snRNA-seq (top) and spRNA-seq (bottom). Color scale indicates scaled average expression per cell type. c UMAP projection of snRNA-seq (left), spatial expression map (middle), and RNA in situ hybridization (right) of the Fruit-Epidermis marker gene FvH4_3g24570 at stage 5. Scale bars: 1 mm (overview) and 0.05 mm (magnified inset). Representative images from three independent experiments with similar results are shown. d UMAP projection (left), spatial expression (middle), and RNA in situ hybridization (right) of the Seed marker gene FvH4_4g05050 at stage 2. Scale bars: 1 mm (overview) and 0.1 mm (magnified inset). Representative images from three independent experiments with similar results are shown. e UMAP projection (left), spatial expression (middle), and RNA in situ hybridization (right) of the Vascular-Bundle marker gene FvH4_3g36020. Scale bars: 1 mm (overview) and 0.05 mm (magnified inset). Representative images from three independent experiments with similar results are shown. f UMAP projection (left), spatial expression (middle), and fluorescence in situ hybridization (FISH) of the Cortex marker gene FvH4_3g01740 at stage 3. Scale bars: 0.05 mm. White arrows indicate the FISH signal; magnified views are shown on the right. FISH was performed on 4-μm cryosections of stage 3 fruits using Cy3-labeled probes and tyramide signal amplification (TSA). Nuclei were counterstained with DAPI (blue). Representative images from three independent experiments with similar results are shown. g UMAP of all single cells after integration, showing 21 clusters categorized into 7 major cell types (colored) that correspond to spRNA-seq defined cell types. h Sankey diagram illustrating the proportion of each cell type derived from spRNA-seq (left) and snRNA-seq (middle) across stages 2, 3, and 5. Line width is proportional to the relative cell abundance.

To verify the cell type annotation of the snRNA-seq atlas, we employed the top 200 specific marker genes from the single-cell atlas and assessed their corresponding expression patterns in spRNA-seq data. Marker genes for Seed, Vascular-Bundle, Sepal & Pedicel, and Carpel cells showed specific correspondence between the two datasets, facilitating the annotation (Fig. 3b and Supplementary Data 11). Additionally, we observed that only the Parenchyma cell type was identified by snRNA-seq, while Cortex/Diamond/Pith cell types were classified from spRNA-seq data (Fig. 3b), indicating the necessity of spRNA-seq for comprehensive cell type characterization. In situ hybridization, fluorescence in situ hybridization (FISH), and spatial projection of marker genes identified from snRNA-seq data validated the mapping of the two platforms (Supplementary Data 12-13), both exhibiting high accuracy in cell type annotation (Fig. 3c–f and Supplementary Figs. 5, 11, 12). Based on the marker gene expression results, the 21 cell clusters identified from the snRNA-seq data were categorized into seven distinct cell types, including Seed cells (sc-clusters 4, 10, and 19), Epidermal cells (sc-clusters 15, 16), Vascular-Bundle cells (sc-cluster 8), Parenchyma cells (sc-clusters 13, 14, and 20), Sepal & Pedicel cells (sc-cluster 5, 7, and 17) and Carpel cells (sc-clusters 2, 3, 12, 18). sc-cluster 6, 9, 21 were not directly part of the fleshy fruit tissues and were collectively annotated as others (Fig. 3g and Supplementary Fig. 13). Comparison of spRNA-seq and snRNA-seq revealed cell type correspondence and proportional changes: the proportion of Parenchyma and Epidermis cells at stage 5, and the proportion of Vascular-Bundle cells at stage 3 increased (Fig. 3h and Supplementary Fig. 14). In this study, we focus on cell types derived from the strawberry fruit that represent Epidermis, Parenchyma, and Vascular-Bundle cells to be studied in the next sections.

Pseudotime analysis revealed the progression of developmental trajectories across different stages of fruit in F. vesca

Using a single-cell atlas of the seed and receptacle, we investigated the key factors and developmental trajectories during fruit development. Trajectory analysis of VascularBundle, Parenchyma, and Epidermis cells across three stages revealed that the sampling stage, rather than cell type, was the predominant factor driving cell expression changes (Fig. 4a–f and Supplementary Fig. 15). Weighted gene co-expression network analysis (WGCNA) identified that the three cell types tended to cluster more densely as the staging progressed (Fig. 4g), confirming the principle of staging as a factor in fruit development. To validate the stage-specific analysis, we integrated the gene expression profiles with previously published hand-dissected mRNA-seq data⁷. The expression pattern of genes specifically enriched in Parenchyma cells overlapped with previously published differentially expressed genes in cortex tissue (Fig. 4h and Supplementary Fig. 16a). Collectively, these results strongly suggested tight coordination of developmental stages among the three cell types. For instance, the GA signaling gene FveRGA1(FvH4_4g341100), a key regulator in fruit set⁹, was highly expressed at an early stage in module 5 (Supplementary Fig. 16b and Supplementary Data 14).

Fig. 4. Developmental trajectories uncover stage-specific divergence of Parenchyma and Vascular-Bundle lineages in F. vesca fruit.

a Visualization of the developmental trajectory of Vascular-Bundle cells colored by pseudotime. b Developmental trajectory of Vascular-Bundle cells colored by cell state. c Developmental trajectory of Vascular-Bundle cells, split by developmental stage. d Visualization of the developmental trajectory of Parenchyma cells colored by pseudotime. e Developmental trajectory of Parenchyma cells, with colors indicating distinct cell populations. f Developmental trajectory of Parenchyma cells, split by developmental stage. g Weighted gene co-expression network analysis (WGCNA) of Epidermis, Parenchyma, and Vascular-Bundle cell clusters revealed 35 distinct co-expression module eigengene (ME) across the three developmental stages. The color scale indicates the strength of association. h Left: heatmap showing gene expression in cortex and pith from bulk RNA-seq ⁷ across developmental stages. Right: heatmap of gene expression from snRNA-seq in Parenchyma, Epidermis, and Vascular-Bundle cells at stage 2, stage 3, and stage 5. i Gene Ontology (GO) enrichment of genes differentially expressed in module 10, which is enriched in Parenchyma cells at stage 2. Enriched GO terms were identified using the hypergeometric test. j Expression of auxin signaling genes along pseudotime in Vascular-Bundle cell populations.

To further explore important genes involved in this coordination, we incorporated stage information as a parameter into WGCNA and identified cell-type-enriched and stage-specific modules (Supplementary Fig. 16b). Gene Ontology (GO) enrichment analysis of these stage-related modules revealed pathways associated with DNA integration and cell wall organization or biogenesis, as well as terms related to nucleotide metabolism and hormone binding (Epidermis cells of ME 22 in Supplementary Fig. 16c and Supplementary Data 15–20). Gene expression and GO enrichment analysis also revealed the auxin signaling and response pathways associated with cell-type-specific genes (Fig. 4i and Supplementary Data 21–22, Supplementary Fig. 16d-e). We focused on vascular bundle tissue, which also serves as the structural framework for auxin signaling transduction, and revealed stage-dependent changes in auxin signaling components in pseudotime analysis (Fig. 4j, Supplementary Figs. 16f, 17, 18, and Supplementary Data 23). Collectively, these results suggest that auxin metabolism and signaling play an important role in strawberry fruit development.

Auxin efflux carriers PINs are associated with auxin distribution asymmetry in the receptacle to drive fruit morphogenesis

Auxin transport from seeds enclosed inside the achenes to the receptacle is mediated by dedicated auxin transport machinery. To illustrate how the auxin distribution asymmetry in the receptacle was established and which specific auxin transporters may be involved, we examined the spatial expression pattern of auxin transporter genes (including PIN-FORMED PINs, auxin transporter-like protein AUX1, and ABC transporter protein Brefeldin A-inhibited guanine nucleotide-exchange protein BIG) (Supplementary Fig. 19 and Supplementary Data 24). Consistent with auxin localization, genes were enriched in Seed (PIN2, PIN4, and AUX-LIKE3) and Bundle-Sheath (PIN8, AUX1) at stage 2, in Pedicel-Bundle (PIN1, PIN5, AUX-LIKE1, AUX-LIKE2), Pedicel (BIG1, PIN3, and AUX-LIKE2), and Bundle-Sheath (AUX1, BIG2, AUX-LIKE3) at stage 3, and constrained to Fruit-Bundle (PIN1) and Fruit-Bundle-Adjacent (PIN1, PIN3, PIN5, PIN8, AUX-LIKE1, AUX-LIKE3, and BIG2) at stage 5 (Supplementary Fig. 19). Spatiotemporal expression patterns of all auxin transporters could be classified into 11 distinct modules (Types 1–11; Supplementary Data 25), including Bundle-Sheath-specific expression (T1) at early stages (stages 2 and 3) and Fruit-Bundle-specific expression (T3) at later stages (stage 5) (Fig. 5a). Notably, two putative auxin efflux carriers, FvePIN1 and FvePIN5, exhibited a shift in expression pattern: initially localized in the Seed and Pedicel-Bundle cells at early developmental stages, they became predominantly enriched in the Fruit-Bundle and Fruit-Bundle-Adjacent cells by stage 5 (Fig. 5b and Supplementary Fig. 20). This spatial redistribution suggests a role for FvePIN1 and FvePIN5 in restricting auxin to specific receptacle domains during fruit development.

Fig. 5. Auxin efflux carriers FvePIN1 and FvePIN5 coordinate auxin transportation and auxin distribution to modulate fruit development and morphogenesis.

a Spatial expression patterns of auxin transport-related genes across developmental stages. T1 and T3 indicate distinct temporal expression types. b Heatmap showing expression patterns of auxin transport-related genes during fruit development. c Spatial expression maps of FvePIN1 (top) and FvePIN5 (bottom) across developmental stages. Color scale indicates normalized expression level. d UMAP projection of FvePIN1 and FvePIN5 from snRNA-seq data (left) and spatial transcriptomic data (right) at stage 5. VS represents Vascular-Bundle in snRNA-seq. FB represents Fruit-Bundle in spRNA-seq. FBA represents Fruit-Bundle-Adjacent in spRNA-seq. e In situ hybridization of FvePIN1. Enlarged views of the boxed region are shown to the right; sense probe served as negative control. Scale bars: 1 mm (left), 0.1 mm (middle) and 0.5 mm (right). VS represents the Vascular-Bundle. f In situ hybridization of FvePIN5. Enlarged views of the boxed region are shown to the right; sense probe served as a negative control. Scale bars: 1 mm (left), 0.05 mm (middle), 0.5 mm (right). VS represents the Vascular-Bundle. Three independent replicates were performed for in situ hybridization. g GUS staining driven by FvePIN1 (left) and FvePIN5 (right) promoters in strawberry fruits (inner and outer tissues). Scale bars: 2 mm (FvePIN1 and FvePIN5 inner), 1 mm (FvePIN5 achene). Three independent replicates were performed. h Fruit phenotypes of strawberry after NAA or NPA treatment. Scale bars, 2 mm. i Quantification of hollow area in fruits treated with NAA or NPA, measured using an online tool (www.smartagriculture.top). Data are presented as box plots with individual data points overlaid as dots. The center line of each box represents the median, and the bounds of the box indicate the 25th and 75th percentiles. The whiskers extend to the minimum and maximum values. Statistical significance was assessed by one-way ANOVA followed by Tukey’s HSD post-hoc test for multiple comparisons. * represents P < 0.05; ** represents P < 0.01; *** represents P < 0.001; **** represents P < 0.0001. j Receptacle length (left) and receptacle width (right) after NAA or NPA treatment. Data are presented as box plots with individual data points overlaid as dots. The center line of each box represents the median, and the bounds of the box indicate the 25th and 75th percentiles. The whiskers extend to the minimum and maximum values. Statistical significance was assessed by one-way ANOVA followed by Tukey’s HSD post-hoc test for multiple comparisons. * represents P < 0.05; ** represents P < 0.01; *** represents P < 0.001; **** represents P < 0.0001.

To resolve the spatiotemporal activity of FvePIN1 and FvePIN5, we profiled their spatial expression across developmental stages. FvePIN1 and FvePIN5 were specifically enriched in Pedicel-Bundle cells at stage 2 and stage 3, as shown by the heatmap (Fig. 5b) and spatial projection (Fig. 5c). Integration of snRNA-seq, spatial transcriptomics (Fig. 5d), and in situ hybridization (Fig. 5e, f) confirmed Fruit-Bundle-restrictive expression of FvePIN1 and Fruit-Bundle-Adjacent expression of FvePIN5 at stage 5. Promoter-reporter assays in both stable transgenic lines (ProFvePIN1::GUS and ProFvePIN5::GUS) and in transiently expressing strawberry fruits (ProFvePIN1::PIN1-GFP and ProFvePIN5::PIN5-GFP) further revealed strong activity in Fruit-Bundle and Seed tissues (Fig. 5g and Supplementary Figs. 21, 22). GO enrichment analysis of genes specifically expressed in Fruit-Bundle at stage 5 also revealed an overrepresentation of membrane-associated pathways (Supplementary Fig. 23 and Supplementary Data 26–27). Consistent with their Fruit-Bundle localization, FvePIN1 was transcriptionally upregulated at stage 3, whereas FvePIN5 was predominantly activated in Fruit-Bundle-Adjacent at stage 5 (Fig. 5b and Supplementary Fig. 24a, b), suggesting that they may function in different temporal phases and spatial cell types to mediate distinct auxin distribution patterns during fruit set and morphogenesis.

Auxin-mediated cell division and proliferation constitute a core mechanism underpinning fruit development and morphogenesis²². We therefore hypothesized that altered spatial deployment of auxin transporters and signaling proteins could underlie developmental defects, including misshapen and hollow fruit. Accordingly, we focused on the causal basis of hollow fruit, a common growth defect observed in many strawberry varieties. Histological analysis of F. vesca fruits indicated that the fruit hollowness primarily results from reduced cell division in the pith tissue, which apparently exhibited a looser cellular arrangement (Fig. 1a and Supplementary Fig. 24e). To further test this, we constructed a co-expression network of FvePIN5, which revealed strong associations with hormone- and cell-division-related genes (Supplementary Fig. 24c and Supplementary Data 28). The exogenous auxin (1-naphthaleneacetic acid, NAA) and the auxin transport inhibitor (N-1-naphthylphthalamic acid, NPA) were applied to F. vesca fruits at stage 3, and the phenotype was evaluated at stage 5. NAA treatment markedly suppressed fruit hollowness and increased fruit width, whereas NPA treatment promoted hollowness in strawberry fruits (Fig. 5h–j), underscoring the central role of auxin transport in specialized strawberry fleshy fruit development.

To directly assess the roles of FvePIN1 and FvePIN5 in fruit development and morphogenesis, we transiently co-expressed these genes in cultivated strawberry fruits. Phenotypic assessment revealed that co-infiltration of Agrobacterium carrying FvePIN1-OE and FvePIN5-OE consistently suppressed the formation of hollow fruits (Supplementary Fig. 24d, e). These findings suggest a model in which FvePIN1- and FvePIN5-mediated auxin transport from seeds into the receptacle contributes to an asymmetric auxin distribution between the cortex and pith. This spatial asymmetry leads to weaker cell division in the pith relative to the cortex at stages 3 and 5, thereby likely giving rise to hollow fruit formation. Collectively, our results suggest that Fruit-Bundle-specific expression of FvePIN1 and FvePIN5 coordinates auxin distribution between cortex and pith to modulate fruit development and morphogenesis.

Comparative spatiotemporal transcriptomics and spatial metabolomics reveal a correlation between ARF6 and an asymmetric auxin distribution pattern during F. vesca fruit set

In strawberry, fertilization induces auxin biosynthesis in the endosperm within a seed that is enclosed inside the achene. This endosperm-derived auxin is subsequently transported to the receptacle to drive its initiation and expansion into fruit flesh^8,9. Auxin has long been recognized as a central regulator of cell division, yet most studies have examined its role at the whole-organ or tissue level during fleshy fruit development. To investigate the spatiotemporal regulation of auxin signaling in mediating receptacle fruit development and morphogenesis, we first cataloged auxin metabolism, signaling, and response genes in the strawberry genome and profiled their expression patterns from fertilization to enlargement (Supplementary Figs. 25, 26 and Supplementary Data 24). Consistent with previous reports^23,24, the auxin biosynthetic enzymes YUC4 and YUC10 were specifically expressed in the Seed cell type at stage 2 (Supplementary Fig. 25), confirming the reliability of our analysis. Notably, spatiotemporal expression patterns of auxin-responsive genes revealed a transition in responsive cell types: in stage 2, enrichment was observed in the Fruit-Epidermis, Cortex, and Bundle-Sheath; in stage 3, expression shifted to the Bundle-Sheath, Diamond, Pedicel, and Pedicel-Bundle; and by stage 5, auxin-responsive genes became more restricted to the Fruit-Bundle and Fruit-Bundle-Adjacent cell types (Supplementary Figs. 26, 30). These findings delineate a progressive restriction of auxin signaling regions from a specific cell type to another as fruit development proceeds.

To elucidate the spatial distribution of auxin in F. vesca fruit, we generated stable transgenic lines expressing the ProDR5::RUBY reporter. The RUBY system was designed based on a system previously described in Arabidopsis thaliana (He et al., 2020) and introduced into F. vesca (Supplementary Table 1). We first evaluated the reliability of the RUBY reporter in monitoring auxin distribution in F. vesca plants. RUBY-derived red pigmentation was consistently observed in the stele of the meristematic zone (Supplementary Fig. 27a), the tip of the lateral root primordium (Supplementary Fig. 27b), and the base of emerged young lateral roots (Supplementary Fig. 27c), which recapitulates the reported spatial distribution of auxin maxima in F. vesca roots (Feng, J., Dai, C., Luo, H., Han, Y., Liu, Z., and Kang, C. (2019). Reporter gene expression reveals precise auxin synthesis sites during fruit and root development in wild strawberry. Journal of Experimental Botany 70, 563-574). The RUBY signal revealed that auxin was broadly distributed across the receptacle (cortex and pith) and floral organs (stamens and pistils) at early stages (stage 2 and 3), but became more abundant in the newly emerged Fruit-Bundle, Fruit-Bundle-Adjacent, and Pedicel-Bundle cell types at the later stage (stage 5) (Fig. 6a and Supplementary Fig. 28a, b). Quantification of RUBY-derived red pigmentation revealed a continuous increase from stage 4 to stage 7 in both seeds and fruit vascular bundles (Supplementary Fig. 28c, d), indicating dynamic changes in auxin distribution during fruit development. This distribution was independently validated in ProDR5::GUS reporter lines, which showed Fruit-Bundle-specific auxin enrichment at late stages (Supplementary Fig. 28e). To further validate spatial auxin distribution in cortex, pith, and seed tissues, we performed spatial metabolomics via the AFADESI-MSI platform coupled with LC-MS analysis of indole-3-acetic acid (IAA) at stages 0, 2, 3, and 5. Consistent with transgenic RUBY and GUS reporter assays, IAA was broadly distributed in the Cortex and Diamond at stage 3 but became largely confined to the Fruit-Bundle and Fruit-Bundle-Adjacent regions at stage 5 (Fig. 6b and Supplementary Fig. 28f).

Fig. 6. Comparative spatiotemporal transcriptomics reveal that FveARF6 expression alignments with auxin distribution asymmetry and mediates receptacle development and morphogenesis in F. vesca.

a Phenotypic characterization of ProDR5::RUBY transgenic F. vesca fruits across developmental stages 2 to 7. Red triangles indicate RUBY signals in longitudinal sections of seeds. Scale bars, 2 mm (top two rows) and 0.2 mm (bottom row). b Spatiotemporally resolved profiling of indole-3-acetic acid (IAA) via spatial metabolomics using the AFADESI-MSI platform (m/z = 272.1744). Scale bars, 2 mm. c Spatial visualization of genes involved in auxin signaling across stages 2 to 5. d UMAP projection of FveARF6 and FveIAA26 expression in spatial transcriptomics data across stage 2, stage 3, and stage 5. e Heatmap showing expression patterns of FveARF6 and FveIAA26 across three developmental stages. Abbreviations: BS Bundle-Sheath, C1 Cortex-1, D Diamond, Epi Fruit-Epidermis, Pe Pedicel, PB Pedicel-Bundle, Pi Pith, Se Sepal, SD Seed, Ca Carpel, S-T Sepal-Tip, C2 Cortex-2, FB Fruit-Bundle, FBA Fruit-Bundle-Adjacent. f Spatial visualization of FveARF6 expression from stage 2 to stage 5. The color scale denotes normalized expression levels. g Schematic diagram of auxin distribution (upper arrow) and FveARF6 expression patterns (lower arrow).

The correlation between IAA distribution and the expression pattern of cell cycle genes was further verified in the receptacle fruit. We identified cell cycle genes in F. vesca and categorized them into four groups based on the expression pattern in G0/G1-phase, S-phase, G2-phase, M-phase²⁵. Genes highly expressed in the S-phase are abundant in Seed and Fruit-Epidermis cells at stage 2, and enriched in Pedicel-Bundle and Fruit-Epidermis at stage 3, whereas they shifted to Cortex-1, Fruit-Bundle, and Fruit-Bundle-Adjacent at stage 5 (Supplementary Fig. 26d and Supplementary Data 25, 29, 30). G2- and M-phase exhibited a similar expression shift from Seed and Fruit-Epidermis to Fruit-Bundle and Fruit-Bundle-Adjacent cells (Supplementary Fig. 26e–g; Supplementary Data 30). Cells expressing G0/G1-phase genes were enriched in the Pith at stage 3 and in the Bundle-Sheath at stage 5, indicating that these cells primarily undergoing cell expansion rather than proliferation at these stages (Supplementary Fig. 26f and Supplementary Data 30). The specific expression patterns of the cell cycle genes indicate that active cell division shifts from the outer cell layers (Seed, Fruit-Epidermis) to the Fruit-Bundle and Fruit-Bundle-Adjacent layers at stage 5.

To identify specific auxin-responsive genes mediated by asymmetric distribution of auxin across distinct cell types and developmental stages, we classified auxin-responsive genes into 11, 7, and 5 expression types at stage 2, stage 3, and stage 5, respectively (Supplementary Data 25). We analyzed the expression pattern of all candidate auxin-responsive genes in auxin-enriched cell types, focusing on Bundle-Sheath (T9) and Fruit-Epidermis (T10) at stage 2; Bundle-Sheath (T3, T4) at stage 3; Fruit-Bundle-Adjacent (T4) and Fruit-Bundle (T5) at stage 5 (Fig. 6c and Supplementary Figs. 29, 30). Specifically, we identified two key candidate genes, FveARF6 and FveIAA26, that exhibited a similar expression pattern (Fig. 6d–f and Supplementary Fig. 31a–b). This pattern coincided with the spatiotemporal distribution of auxin during the early (stages 2 and 3) and late (stage 5) developmental stages (Fig. 6g). These results suggested that FveARF6 may serve as a key regulator determining receptacle development and morphogenesis in response to auxin distribution.

To functionally validate the role of FveARF6 in receptacle development, we generated stable transgenic FveARF6 knock-out (KO) lines using a CRISPR/Cas9 system⁴, as well as FveARF6 overexpression (OE) lines and FveARF6-RNAi silencing lines, all generated via Agrobacterium-mediated stable transformation. Genotyping and quantitative phenotypic analysis confirmed the successful gene knockout, overexpression, and silencing of FveARF6 (Supplementary Table 2 and Supplementary Fig. 32a–c). Both CRISPR/Cas9 knockout and RNAi silencing lines exhibited significantly reduced receptacle width and length from stage 2 to stage 5, while overexpression of FveARF6 significantly increased both receptacle length and width (Fig. 7a, b and Supplementary Fig. 32d–e). Real-time PCR showed that FveARF6 was induced by in vitro auxin treatments after 72 hours in developing F. vesca fruits (receptacle and achenes) (Supplementary Fig. 31c–d). Together, these results indicate that FveARF6 acts as a positive regulator of receptacle growth in response to auxin signaling.

Fig. 7. FveARF6 acts as a positive regulator of receptacle growth and physically interacts with FveIAA26.

a Phenotypic comparison of FveARF6-CRISPR, FveARF6-overexpression, and wild type fruits at stages 0, 2, 3, and 5. Scale bars, 2 mm. b Quantification of receptacle size from FveARF6-CRISPR, FveARF6-overexpression, and wild type fruits across stage 0, stage 2 (normalized to petal size), stage 3, and stage 5. In the box plots, all data were presented as mean ± SD. Statistical significance was assessed by two-way ANOVA followed by Tukey’s multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. c Luciferase complementation image (LCI) assay in N. benthamiana leaves. d Yeast two hybrid assay testing the interaction of FveARF6 and FveIAA26. Positive interaction is shown by growth on selection medium -tryptophan (Trp), -leucine (Leu), -histidine (His), and -adenine (Ade). pGADT7-T and pGBKT7-53 were used as a positive control. EV stands for empty vector. e Pull-down assay to detect the interaction between FveARF6 and FveIAA26. Representative images from three independent experiments with similar results are shown. f Cell-free protein degradation assay. Recombinant IAA26-GST was expressed in E. coli and purified. Leaf mesophyll protoplasts of F. vesca were treated with auxin (NAA) and lysed. Protein degradation was monitored by immunoblotting with an anti-GST antibody. Representative images from three independent experiments with similar results are shown.

Gene expression profiling across different cell types revealed co-expression of FveARF6 and FveIAA26, a putative Aux/IAA family member (Fig. 6d–e). To test whether FveARF6 physically interacts with FveIAA26, we performed yeast two-hybrid assays, co-luciferase (co-LUC) assays in Nicotiana benthamiana, and pull-down assays. All three assays showed a specific interaction between FveARF6 and FveIAA26 (Fig. 7c–e). We further examined the auxin-dependent instability of FveIAA26 using cell-free degradation assays. Recombinant FveIAA26-GST protein was rapidly degraded within 2 hours of NAA (0.1 mM) treatment (Fig. 7f). This suggests that the FveARF6-FveIAA26 module also follows a canonical auxin response pathway in plants.

Finally, to explore the molecular mechanism underlying FveARF6-mediated fruit development, we performed bulk RNA sequencing on fruits of FveARF6-CRISPR knockout lines. GO analysis revealed differential expression of genes enriched in DNA-binding transcription factor activity and cell wall modification pathways (Supplementary Fig. 32f). Together with the WGCNA results (Supplementary Fig. 31e; Supplementary Data 31), these findings suggest that FveARF6 triggers receptacle development through transcriptional modulation of cell wall-associated genes.

Discussion

Fruits represent not only a pivotal evolutionary innovation for seed protection and dispersal, but also hold immense agricultural and economic value for humans². However, the molecular mechanisms of fruit set and subsequent morphogenesis at the molecular and cellular level remain poorly understood, largely due to the intricate fruit tissue architecture, high metabolic complexity, and the absence of robust cell-type markers.

In this study, by integrating snRNA-seq, spatial transcriptomics, and spatial metabolomics, we constructed a spatiotemporal single-cell resolution atlas for fertilization-induced fleshy fruit development and morphogenesis, spanning post-fertilization (stage 2), fruit set (stage 3), and fruit enlargement (stage 5). We refined the classification of the cortex into three subtypes (Cortex-1, Cortex-2, and Bundle-Sheath) and the vascular bundle tissue into two subtypes (Fruit-Bundle and Fruit-Bundle-Adjacent). In addition, we identified two new cell types, termed Diamond and Fruit-Bundle-Adjacent (Fig. 3). Together, these findings underscore the remarkable cellular diversity of the F. vesca receptacle and provide a single-cell-based conceptual framework for dissecting the mechanisms underpinning fruit development and fruit type diversification within the Rosaceae family².

Based on ProDR5::RUBY reporter lines and a spatial metabolomics platform, we showed that at stage 2, seed-derived auxin is transported to the receptacle, where it accumulates in the Cortex and Diamond tissues. At stage 3, the Fruit-Bundle and Bundle-Sheath are enriched in transcripts of the major auxin efflux carriers FvePIN1 and FvePIN5, which presumably mediate directional auxin transport toward the vascular bundle and the whole fruit. At stage 5, auxin is redistributed into a new cell layer-the Fruit-Bundle-Adjacent layer on the cortex side, establishing an asymmetric auxin distribution between the cortex and pith (Fig. 5a, b). Additionally, inhibition of FvePIN1 and FvePIN5 by NPA disrupted auxin transport and caused canonical fruit developmental defects, such as deformed and hollow fruit. Conversely, co-expression of FvePIN1 and FvePIN5 suppressed fruit hollowness (Fig. 5h), suggesting that spatial auxin transport and distribution determine fleshy fruit development and specificity.

In our study, FveARF6 was identified as a key positive regulator of receptacle development that responds to this auxin asymmetry, based on the following evidence: 1) Spatiotemporal expression: FveARF6 exhibits an expression pattern closely mirroring auxin distribution across fruit developmental stages. Specifically, at stages 2 and 3, FveARF6 is highly expressed in Cortex and Diamond; at stage 5, its expression is predominantly restricted to the Fruit-Bundle and Fruit-Bundle-Adjacent (Fig. 6d–f); 2) Genetic evidence: CRISPR knockout, overexpression (OE), RNAi knockdown, and stable transgenic lines of FveARF6 all indicate that FveARF6 functions to promote fruit development (Fig. 7a, b); 3) Auxin response: Exogenous application of the synthetic auxin NAA substantially degrades its inhibitor FveIAA26 protein and activates FveARF6 (Fig. 7c–f). Our findings reveal that, beyond canonical auxin response factors involved in general growth, FveARF6 acts as a tissue-specific interpreter of auxin gradients, linking asymmetric auxin distribution to differential receptacle tissue differentiation (e.g., cortex, diamond, fruit-bundle tissues). This provides a refined model for understanding how spatial auxin cues are decoded into organ morphogenesis in strawberry fleshy fruits. Our study employs strawberry, an accessory fruit, as a mechanistic case study to elucidate the spatiotemporal regulatory mechanisms of fleshy fruit development, offering a framework for dissecting the developmental mechanisms of other fruit types.

Single-cell RNA sequencing has been successfully deployed across various plant species, yielding insights into vegetative structures such as leaves^26,27, roots²⁸, hypocotyls²⁹, and shoot apices³⁰, as well as select reproductive tissues including inflorescences^31,32, anthers³³, fruit¹⁶, and zygotic embryos³⁴. A recent study identified an auxin-triggered mechanism underlying post-fertilization morphological remodeling of fruits in the Brassicaceae²². However, the cellular trajectories underlying fleshy fruit development and morphogenesis remain uncharacterized. To date, investigations of fertilization-induced fruit development and enlargement have largely been restricted to tissue-level resolution^8,9, lacking a single-cell perspective.

Constructing single-cell atlases for large-size, heterogeneous, and metabolite-rich tissues such as fruits remains challenging. The recent development of BMKMANU DG1000 snRNA-seq, combined with the 10x Visium platform for spatial mapping, offers a powerful solution for studying fruit development. This integrated approach benefits from high tissue compatibility, established sectioning workflows, and efficient mRNA capture. Here, we present the spatiotemporal single-cell atlas of strawberry fruit set and morphogenesis. Our analysis reveals remarkable cellular heterogeneity, comprising 14 distinct cell types, including two previously unknown populations, Diamond and Fruit-Bundle-Adjacent, which serve as key regulatory hubs for receptacle initiation in strawberry. This spatial single-cell transcriptomic atlas for fleshy fruit initiation and development in fruit crops will help propel fruit crop research into the single-cell era.

In general, fruit development progresses through three major phases: fruit set, fruit enlargement, and fruit ripening, with fruit set being a critical determinant of subsequent development^35,36. Fruit set is set up after successful fertilization of an open flower, triggering hormonal signal synthesis in the seed and transport to the fruit flesh tissues to orchestrate downstream developmental processes³⁵. In the 1950s, Nitsch found that the fertilization-induced auxin played a key role in initiating strawberry fruit development⁶. Genome-wide transcript profiling in diploid strawberry, F. vesca, revealed that auxin synthesis genes were highly expressed in the ghost (mainly endosperm)⁷. Importantly, the endosperm has been pinpointed as the fertilization-sensing and auxin-producing tissue in Arabidopsis^37,38 as well as in F. vesca⁸. It is still unclear how seed endosperm-derived auxin is transported to the receptacle and how this auxin subsequently activates the receptacle to become enlarged and fleshy. Prior studies revealed a feedback regulatory loop involved in auxin and GA crosstalk in the receptacle that downregulates fruit enlargement⁹. In this study, using a spatiotemporal transcriptome atlas and stable transgenic auxin reporters (ProDR5::RUBY), we characterized the heterogeneous distribution of auxin and auxin-responsive genes within the receptacle, addressing outstanding questions regarding auxin distribution asymmetry, multi-cell-type differentiation, and the determination of final fruit architecture.

Our analysis identified FveARF6 as a critical auxin signaling component in activating cell division and cell expansion across all cell types at early stages (stages 2 and 3), but specifically in the new cell type Fruit-Bundle-Adjacent at late stages (stage 5) (Fig. 6d–f), coinciding with the auxin distribution asymmetry in the receptacle. Although ARF6 homologs are known to regulate floral organ maturation and fruit initiation in Arabidopsis^39–42, ARF6 and ARF8 have also been shown to function redundantly in fruit valve development through association with the CrSTM promoter in Capsella ²². Our work uncovers a neo-functionalized or expanded role of ARF6 in receptacle (a major fleshy tissue) development in strawberry or related Rosaceae species. This advances comparative developmental biology by illustrating how an orthologous transcription factor family has been co-opted to control a lineage-specific fruit structure in both dry and fleshy fruits.

Following fertilization, a gradual increase in IAA content was observed in the seeds, consistent with a previous report⁴³. At stage 5, the emergence of Fruit-Bundles restricts PIN-mediated auxin transport routes toward the receptacle and pedicel. In this process, FvePIN1 is predominantly expressed in  the Fruit-Bundle at stage 3, whereas FvePIN5 is highly and specifically expressed in the Fruit-Bundle-Adjacent (toward the cortex side) at stage 5, suggesting that both PINs coordinately modulate auxin transport pattern (Fig. 5), leading to the asymmetric auxin distribution in strawberry receptacle fruit. Overall, our spatiotemporal transcriptomic analysis of strawberry fruit development has uncovered a fertilization-induced, phytohormone-mediated signaling pathway that regulates fleshy fruit development. Furthermore, through reverse functional validation of pathway disruption, specifically its link to fruit hollowness, we demonstrate the utility of the single-cell resolution spatial atlas generated in this study. These findings not only illustrate the power of spatiotemporal transcriptomics in uncovering developmental mechanisms but also provide a new framework for fundamental research in fruit crops in the future.

Our spatiotemporal single-cell atlas revealed a previously unrecognized “two-phase” developmental trajectory for strawberry receptacle fruit development, in which the cortex and pith undergo distinct developmental pathways, and collaboratively determine fleshy development and fruit morphology. Specifically, after fertilization, auxin is synthesized in the endosperm within the seeds and transported to the cortex, pith, and Diamond cells, distributing concurrently across these tissues. This auxin accumulation activates downstream signaling gene (FveARF6), triggering first-phase cell division and expansion in all these tissues during early stages (stage 2 and 3), thereby driving the initial enlargement of both the cortex and pith. However, by late stage 5, the fully formed Fruit-Bundle constrains auxin transport towards the Pedicel-Bundle (likely via PIN1) and favors auxin efflux to the cortex through the Fruit-Bundle-Adjacent region via PIN5. This is consistent with auxin flow and triggers second-phase of cortex development at late stage 5. In contrast, auxin distribution and ARF6 expression remain low in the Diamond and Pith tissues, limiting their second-phase growth. This unparalleled two-phase developmental pattern underlies fruit hollowness formation in F. vesca and many cultivated strawberry cultivars, primarily due to restricted PIN expression and auxin distribution. Moreover, late-stage cortex development proceeds horizontally (from the Fruit-Bundle-Adjacent region towards the Fruit-Epidermis), whereas pith development follows a vertical direction (from Diamond cells towards the pith). These tissues respond to auxin accumulation along separate trajectories, ultimately shaping the overall morphology of the receptacle fruit (Fig. 8).

Fig. 8. A model of fertilization-induced auxin signaling orchestrating fruit development and morphogenesis in strawberry.

Fruit development is spatiotemporally regulated by auxin distribution and specific genetic regulators. Initially, seed-derived auxin is transported into the receptacle at stage 2, where it accumulates in the cortex and Diamond. After stage 3, the Fruit-Bundle differentiates from the Bundle-Sheath layer and becomes enriched in transcripts of the auxin efflux carriers FvePIN1 and FvePIN5, driving directional auxin transport toward the Pedicel-Bundle and establishing an asymmetric auxin distribution between the cortex and pith. FveARF6 was identified as a key gene and displays a spatiotemporal expression pattern that closely mirrors auxin distribution across the fruit developmental stages: at stage 2, it is highly expressed in the Cortex and Diamond, promoting cortex growth and Diamond-mediated coordination of pith development; by stages 3 and 5, its expression shifts to the Fruit-Bundle and Fruit-Bundle-Adjacent cells, sustaining cortex development. Reduced auxin accumulation in the Diamond and pith at these stages leads to differential cell division, ultimately causing hollow fruit formation.

PINs coordinates cellular auxin levels through their different polar membrane localization⁴⁴. In the shoot apex of Arabidopsis, the subcellular localization of PIN1 was oriented by mechanical signals, revealing a tight coupling between morphogenesis and auxin transport⁴⁵. The vascular tissues in plants determin the programmed growth and morphogenesis^46,47. Our findings highlight the Fruit-Bundle as a potential auxin-enriched nexus that constrains the expression of PINs, and this enrichment is spatially correlated with post-fertilization fleshy fruit shape remodeling. Yet, we still do not know how the fate of these Fruit-Bundle cells is determined in strawberry fruit, as well as the specific downstream signaling gene of the transcription factor FveARF6 that orchestrates receptacle development and morphology.

The elucidation of auxin transport, distribution, and response mechanisms in the receptacle not only advances fundamental knowledge of fruit set and morphogenesis, but also holds direct implications for strawberry breeding and cultivation. By dissecting the genetic factors and signaling pathways that govern fruit initiation, particularly how seed-derived auxin is directionally transported via FvePIN1/FvePIN5 to establish an asymmetric distribution among cortex, diamond, and pith, our study provides a molecular framework for understanding the earliest determinants of fruit development. This knowledge is instrumental in tackling long-standing industry challenges, such as malformed and hollow fruits, widespread problems that occur in many fruit and vegetable crops, including watermelon, lettuce, potato, tomato, cucumber, sugar beet, and other crops. These defects often arise from disrupted auxin flux, as demonstrated here by NPA mediated inhibition of PIN carriers, and our findings pinpoint the specific cell types (e.g., Fruit-Bundle-Adjacent) and genes (FvePIN1 and FvePIN5) whose misregulation correlates with fruit hollowness. Furthermore, the genetic basis of strawberry fruit shape and size, traits of paramount commercial value-can now be traced to the spatiotemporal activity of auxin-responsive regulators. Notably, we reveal a previously undocumented a secondary rapid fruit expansion occurring soon after stage 3 (Fig. 1), a critical phase that accounts for a substantial portion of final fruit volume. Our integrated single‑nucleus and spatial transcriptomic atlas shows that this late‑stage expansion coincides with the sustained expression of FveARF6 and its downstream targets within the Fruit‑Bundle lineage, suggesting an auxin‑driven reactivation of cell expansion distinct from early enlargement. By linking these molecular events to whole‑fruit architecture, our results offer actionable targets for marker‑assisted breeding of uniformly shaped, non‑hollow, and optimally sized strawberries and provide a roadmap for manipulating hormonal distribution to enhance fruit development and quality.

Methods

Plant materials and culture conditions

The wild strawberry F. vesca Yellow Wonder 5AF7 (YW5AF7) was used for the study. The strawberry cultivar ‘Monterey’ was used for functional identification. Plants were grown in a growth chamber at a temperature of 25 °C/22 °C and light/dark of 16 h/8 h for day/night. The relative humidity was about 50%.

Tissue sectioning of strawberry fruit

Tissue sectioning was performed following a previously described protocol⁴⁸. Early-stage strawberry fruits were fixed in FAA containing 50% (v/v) ethanol, 5% (v/v) acetic acid, and 3.7% (v/v) formaldehyde. The tissue and fixative were placed in a vacuum for 15 min for better penetration. After releasing the vacuum, the tissue was placed at 4 °C overnight. The tissue was dehydrated with 50% ethanol, 60% ethanol, 70% ethanol, 85% ethanol, 95% ethanol, and 100% ethanol (twice). Next, the tissue was permeated with 25% xylene/75% ethanol, 50% xylene/50% ethanol, 75% xylene/25% ethanol, and 100% xylene (three times). After infiltration with paraplast, the tissues were placed at room temperature and then at 42 °C. Next, the tissues were placed in 100% paraffin wax in an oven at 60 °C. Fresh paraffin solution was changed twice a day for 3 days, and then a wax block was poured. The tissues were sectioned at 10 µm thickness and stained by Safranin-O/Fast Green. All tissue samples were sectioned using a Thermo Scientific HM 355S automatic rotary microtome. The images were captured with a ZEISS Axiolab 5 microscope.

10× Visium spatial RNA-seq

10× Visium Spatial RNA-seq was conducted with the help of BioMarker Company (Qingdao). Two replicates of stage 3 fruits and one replicate each of stage 2 and stage 5 fruits were embedded and sectioned. Tissue sections were adhered to the 6.5 × 6.5 mm chip area with 5000 clusters (spots) containing barcoded primers (10× Genomics). The embedded strawberry fruits were sectioned at 10 μm for stage 3 and 12 μm for stage 5, and systematically placed on chilled Visium slides (3000394, 10× Genomics) and Visium Spatial Gene Expression slides (2000233, 10× Genomics). The spatial cDNA library was prepared with an automated MBS 8000 system⁴⁹. Sequencing was performed on the Illumina NovaSeq 6000 (Biomarker Technologies Corporation, Beijing).

Single-nucleus RNA-seq

F. vesca fruits (including achenes) at stage 3 and stage 5 were washed 2–3 times in 4 °C pre-cooled 1×PBS on ice, transferred to a sterile RNase-free culture dish, and cut into small pieces (about 0.5 mm²) with 10 cm surgical scissors. After incubation at 37 °C with a hydrolysate solution, the cells were filtered through a 40 μm cell sieve. The cell suspension was centrifuged at 300 g at 4 °C for 5 min. Discard the supernatant and add 1 mL of resuspension to the centrifuge tube to resuspend the cells. Then,  3 mL of pre-cooled erythrocyte lysis buffer (Solarbio), blowing and suction evenly. Incubated at 4 °C for 5–10 min, centrifuged at 300 g at 4 °C for 5 min immediately after completion and remove the supernatant and add 1 mL of resuspension buffer (pre-cooled 1X PBS) to fully resuspend the cells. Adjust the cell concentration to 700–1200 cell/μL and cell activity to over 85%, and place cells on ice.

SnRNA-seq libraries were prepared with BMKMANU DG1000 GEM Kits on the Droplet Generation Equipment (BMKMANU). The cell suspension was loaded onto the BMKMANU C1000 Chip within 30 min, and the Droplet Generation Equipment was used to generate single-cell gel beads in emulsion (GEMs) according to the manufacturer’s recommendations. Barcoded, full-length cDNA was generated, and libraries were constructed according to the performer’s protocol. Sequencing was performed on the Illumina NovaSeq^TM X Plus with a sequencing depth of at least 50,000 reads per cell and 150 bp (PE150) paired-end reads (performed by Biomarker Technologies Corporation, Beijing, China).

RNA in situ hybridization and fluorescence in situ hybridization (FISH) assay

In situ probes were synthesized by PCR amplification of cDNA using gene-specific primers (Supplementary Data 32) containing the T7 and SP6 RNA polymerase binding sites. Digoxigenin-labeled UTP (11175025910) was used for in vitro transcription. In situ hybridization was performed following the protocol adapted from a previous report⁵⁰: Floral buds were placed in 3.7% FAA on ice and were fixed by vacuum infiltration. The fixed samples were dehydrated through a graded ethanol series  (50, 70, 85, 95%, and 100% EtOH) at 4 °C for 90 min at each EtOH grade. They were renewed in 100% EtOH and left overnight at 4 °C. After being renewed with 100% EtOH, and replaced with 50% EtOH/50% Histoclear, 100% Histoclear was repeated three times, and fresh Paraplast was added and left overnight without shaking. The paraplast chips were melted at 42 °C, and fresh molten paraplast was replaced at 60 °C overnight. Fresh paraplast was renewed 2 times a day for 3 days and placed in tissue molds. The embedded tissues were sectioned and blocked in glass slides. After air-drying, 4 volumes of hybridization solution were added to the probe. Slides were then placed in boxes on tissues soaked in 2 × SSC/50% formamide. Boxes were sealed with tape and placed at 50 °C overnight. The slides were washed and blocked before antibody application  and incubate for 2 h at room temperature. Slides were washed and treated with a substrate solution containing BCIP/NBT for signal detection. Images were captured  with a microscope slide scanner (PANNORAMIC MIDI Ⅱ, 3DHISTECH Ltd) and ZEISS Axiolab 5 microscope.

The RNA fluorescence in situ hybridization (FISH) assay was performed with the SweAMI signal amplification system at ServiceBio Inc (Wuhan, China). Tissue samples were fixed at 4 °C for at least 12 h, then dehydrated through a graded ethanol series, cleared in xylene, and embedded in paraffin wax. The embedded tissues were sectioned at 4 μm thickness, then dewaxed, rehydrated, and rinsed in DEPC-treated water. Antigen retrieval was performed by microwave heating in citrate buffer (pH 6.0) for 7 min at medium power, followed by a 6-min pause and then 10 min at low power. After digestion with proteinase K (20 μg/mL) at 40 °C for 5 min, the sections were washed three times with PBS (5 min each). Pre-hybridization was carried out with pre-hybridization solution at 40 °C for 1 h. The sections were then hybridized overnight at 40 °C containing the specific SweAMI probes (labeled with Cy3; 500 nM). After hybridization, slides were washed sequentially with SSC at room temperature. Subsequently, the sections were incubated with pre-warmed branched-DNA amplifier probes at 40 °C for 45 min, followed by washing with a graded SSC series at 40 °C for 5 min each. Signal probes (Cy3-labeled, diluted 1:200) were then applied and incubated at 40 °C for 3 h. Nuclei were counterstained with DAPI for 8 min in the dark. The slides were imaged using a 3DHISTECH Pannoramic MIDI. DAPI was excited at 330–380 nm (emission 420 nm, blue), and Cy3 at 510–560 nm (emission 590 nm, red).

Spatial metabolomics

Spatial metabolomics analysis was conducted by MetWare Biotechnology Co., Ltd. (Wuhan, China). Specifically, frozen tissue samples were fixed with three drops of distilled water prior to sectioning. Sections of 25 μm thickness were obtained using a Leica CM1950 cryostat (Leica Microsystems) precooled to −20 °C. Tissue sections were mounted on adhesive glass slides and dried under vacuum for 30 min. These samples were sent for Mass spectrometry imaging: Ambient field desorption electrospray ionization mass spectrometry imaging (AFADESI-MSI) was performed on an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific) with the following parameters: solvent flow rate, 6 μL min⁻¹; nitrogen gas flow, 45 L min⁻¹; spray voltage, 7 kV; sample-to-sprayer and sprayer-to-ion-transfer-tube distances, 3 mm each. The mass spectrometer was operated at a resolution of 60,000 (FWHM at *m/z* 200) and a scan range of *m/z* 70–1000. Automatic gain control (AGC) was set to 1 × 10⁶ charges, maximum injection time to 200 ms, S-lens voltage to 55 V, and capillary temperature to 350 °C. Tissue sections were raster-scanned continuously along the *x*-axis at 0.35 mm s⁻¹ with 0.5 mm step intervals in the *y*-axis.

Cell atlas construction

The raw snRNA-seq data and spRNA-seq data were first aligned using Cell Ranger (10× Genomics) and Space Ranger (v1.2.2;10× Genomics)⁵¹ for calculation of cell expression matrices. The genome and GTF annotation files of strawberry were downloaded from F. vesca Whole Genome v4.0.a2 (http://www.rosaceae.org/species/fragaria_vesca/genome_v4.0.a2)⁵². The gene-cell matrices were subsequently loaded into the Seurat⁵³ package (v.3.2.0), which was implemented in R (v.4.0.2). Low-quality cells or bins were filtered manually. The clean and normalized data obtained above were combined using the merge command in the Seurat package respectively for snRNA-seq and spRNA-seq data. Then the merged datasets were split by sample and processed using SCTransform for further anchor gene identification. Finally, cell expression objects of different samples were merged by IntegrateData for data integration and batch effect correction. The corrected objects of snRNA-seq and spRNA-seq data were subjected to principal component analysis (PCA) dimension reduction and cell embedding with UMAP and t-distributed stochastic neighbor embedding (TSNE) in the same two-dimensional space. The resulting cell clusters were classified using FindNeighbors and FindClusters.

Trajectory analysis

Pseudotime trajectories were constructed by Monocle2 (v.2.30.1). Differentially expressed genes were identified according to the FindAllMarkers in the R package Seurat (min.pct = 0.25; avg_log2FC.threshold = 0.6) and filtered by p_val_adj < 0.01. Dimensionality reduction, clustering and pseudotime trajectory inference were performed using the newCellDataSet and orderCells, respectively, with the default parameters.

Mass spectrometry experiments

Sample preparation was performed as previously described⁵⁴. Ground samples (0.5 g) were extracted with 2.5 mL of pre-cooled 50% (v/v) aqueous acetonitrile containing 40 µL of isotopically labeled IAA as the internal standard. The mixture was vortexed for 1 min, ultrasonicated at low temperature (4 °C) for 15 min, and centrifuged at 14,800 g for 5 min at 4 °C. The pellet was re-extracted using the same procedure. The combined supernatants (approximately 5 mL) were diluted with water to a final volume of 50 mL (5% acetonitrile) and subjected to solid-phase extraction (SPE). SPE was performed using an automated high-throughput system with Oasis HLB cartridges (Waters, WAT094226), which were preconditioned with 3 mL of methanol and 3 mL of water. After sample loading at a flow rate of 3 mL min⁻¹, the cartridges were washed with 5 mL of water and eluted with 3 mL of dichloromethane containing 0.5% (v/v) formic acid and 2.5% (v/v) methanol (prepared as 3 mL CH₂Cl₂, 15 µL formic acid, and 75 µL methanol). The eluate was combined with 1 mL of methanol and evaporated to dryness under a gentle stream of nitrogen. The dried residue was reconstituted in 0.2 mL of methanol, vortexed for 30 s, filtered through a 0.22 µm PTFE membrane, and analyzed by ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS).

Chromatographic separation was performed on an Acquity UPLC HSS T3 column (2.1 × 150 mm, 1.8 µm). The mobile phases consisted of water containing 0.1% (v/v) formic acid (A) and 0.1% (v/v) formic acid-acetonitrile (B). The gradient elution program was as follows: 0–1 min, 10% B; 1–13 min, 10–60% B; 13–14 min, 60–95% B; 14–17 min, 95% B; 17-18.1 min, 10% B; and 18.1–20 min, 10% B. The flow rate was 0.2 mL min⁻¹, and the injection volume was 2 µL at 35 °C.

Mass spectrometry was performed on a triple-quadrupole mass spectrometer equipped with a heated electrospray ionization (H-ESI) source, operating in both positive- and negative-ion switching modes. The spray voltages were set at 4,000 V (positive ion) and 3,000 V (negative ion). The sheath gas and auxiliary gas pressures were maintained at 35 and 10 arbitrary units, respectively. The ion transfer tube temperature and vaporizer temperature were both set at 350 °C. Specific ion transitions and compound-dependent parameters for each analyte were provided in the Source Data.

NAA treatment

To examine the auxin response of FveARF6, the wild strawberry (F. vesca) cultivar ‘Yellow Wonder’ with consistent growth status was selected, and its flowers at stage 2 were evenly sprayed with 0.1 mM NAA solution onto the flower surfaces. Samples were collected at fixed time points (0, 3, 6, 9, 12, 24, 48, 72, 120, and 168 h). During sampling, sepals, petals, and stamens were removed, and only the receptacles were retained. Three biological replicates were collected at each time point.

For the auxin-mediated fruit development and morphogenesis experiment, a 50 mM stock solution of NAA (PhytoTech, N600) and NPA (Sigma-Aldrich, 33371) were prepared in ethanol and diluted with two drops of Tween 20 in water before use. The final concentrations were 500 µM for NAA and 100 µM for NPA. Control treatments were performed with water. Starting from the S3 stage of fruit development, the fruits were sprayed with the respective hormone solutions every 2 days until the day 12 of sampling. The treated fruits were photographed using a stereomicroscope to document any morphological changes.

Vector construction and stable transformation of FveARF6-RNAi, FveARF6-OE, and FveARF6-CRISPR-edited plants

Vector construction was performed based on the previously described JH4-JH12 system⁴. FveARF6-RNAi sequence is listed in Supplementary Table 2. The target sequence was amplified and cloned into JH4 (containing an sgRNA cassette driven by the U6 promoter) as an entry vector and then into JH12 (Cas9 driven by the AtUBQ10 promoter) through Gateway cloning. The full-length coding sequence of FveARF6 (FvH4_3g05010) was cloned and recombined into the JH23 vector⁹ at the KpnI and SalI sites. The final vector containing the FveARF6 target sequence was transformed into the GV3101 strain and infiltrated into the callus of F. vesca. After screening with hygromycin B and verification by CAS9 expression and Sanger sequencing, the FveARF6-edited plants were ready for phenotype observation and RNA-sequencing analysis. Two FveARF6-CRISPR lines (+ 1, +1) and (-16, -16) were subjected for RNA-seq by using the Illumina platform.

Subcellular localization

For experiments in Nicotiana benthamiana, each construct was transformed into Agrobacterium tumefaciens strain GV3101, and resuspended in infiltration buffer (10 mM MES, 10 mM MgCl₂, and 200 μM acetosyringone) to an OD₆₀₀ of 0.6. FvePIP2a-mCherry was used as a plasma membrane marker⁵⁵, and CD3-959 (ER-rk) was applied as the ER marker⁵⁶. The bacterial suspension containing the target construct was mixed 1∶1 (v/v) with a suspension containing the organelle marker, and the mixture was infiltrated into the leaves of 4-week-old Nicotiana benthamiana.

For localization in F. vesca protoplasts, protoplasts were isolated from callus in an optimized enzyme mixture (containing 1.5% Cellulase R-10, 0.75% Macerozyme R-10, 0.5% Snailase, and 0.4 mol L⁻¹ D-mannitol) for 10 h at 28 °C. After filtration through nylon mesh and mixing with CPW21 solution, the protoplasts were subjected to density gradient centrifugation in CPW7 solution (0.2 mmol·L⁻¹ KH₂PO₄, 1 mmol·L⁻¹ KNO₃, 2.08 mmol·L⁻¹ MgSO₄, 0.001205 mmol·L⁻¹ KI, 0.000012 mmol·L⁻¹ CuSO₄·5H₂O, 1.35 mmol·L⁻¹ CaCl₂ and 400 mmol·L⁻¹ mannitol). Purified protoplasts were collected and transfected with 40% polyethylene glycol-4000 (PEG4000) in a solution containing 0.2 mol·L⁻¹ mannitol and 0.2 mol·L⁻¹ MgCl₂ for 20 min⁵⁷. After transfection, the protoplasts were washed with W5 solution (2 mmol·L⁻¹ MES, 125 mmol·L⁻¹ CaCl₂, 5 mmol·L⁻¹ KCl, 154 mmol·L⁻¹ NaCl) and incubated in the dark for recovery. Fluorescence signals were observed, and images were acquired by a Nikon A1R HD25 confocal laser scanning microscope.

Functional testing of FvePIN1 and FvePIN5 by transient overexpression and stable transformation

Full-length cDNAs of FvePIN1, FvePIN5, and FveARF6 were cloned into the overexpression vector JH23⁹ to generate constructs JH23-PIN1, JH23-PIN5, and JH23-ARF6. UBQ::PIN1-GFP/UBQ::PIN5-GFP were cloned into JH23-GFP with KpnI and AscI, and FveARF6 was cloned into JH23-GFP with KpnI and SalI. ProDR5::RUBY vector was generally provided by Dr. Yubing He⁵⁸. To generate the ProPIN1::PIN1-GFP and ProPIN5::PIN5-GFP fusion constructs, the promoter sequences of PIN1 and PIN5 were amplified from F. vesca genomic DNA, and the corresponding coding region (from ATG to the last codon before the stop codon) was amplified from F. vesca cDNA. For each gene, the promoter fragment and the coding sequence were fused by overlap-extension PCR to yield a contiguous ProPIN-PIN amplicon. These amplicons were then individually cloned into the PM999-GFP vector using HindIII and XbaI restriction sites, in-frame with the N-terminus of GFP to avoid any frameshift. The C-terminal GFP fusion was used for localization analysis, whereas functional assays were performed with the untagged protein. All constructs were verified by Sanger sequencing, and the primer sequences are provided in Supplementary Data 32.

Constructs were transformed into Agrobacterium tumefaciens strain GV3101, which was cultured overnight in LB medium supplemented with rifampicin and kanamycin at an OD₆₀₀ of 0.8–1.0. After centrifugation at 5000 g for 5 min, the bacterial pellet was resuspended in infection buffer (10 mM MgCl₂, 10 mM MES, pH 5.4, 100 μM acetosyringone) to an OD₆₀₀ of 0.6. The bacterial suspension was incubated in the dark at room temperature for 2 h before being injected into diploid strawberry fruits and octoploid strawberry fruits. For diploid strawberries, the injections were performed at the small green fruit stage (Stage 4), with tissue sampling collected approximately 7 days post-injection at the mature fruit stage (Stage 6). For octoploid strawberries, injections were performed at the large green fruit stage, with sampling conducted 7 days later at the mature stage. The JH23-eGFP empty vector was used as a control. Stable transformation of ProDR5::RUBY into F. vesca was performed following the protocol as reported previously⁹. The final construct containing the target sequence was introduced into Agrobacterium tumefaciens strain GV3101, and subsequently used to infiltrate F. vesca callus. After co-cultivation, the transformed tissues were screened on medium containing 2 mg/L hygromycin.

Co-immunoprecipitation

Recombinant constructs encoding FveARF6-HA and FveIAA26-mCherry in the pCAMBIA1300-BS vector backbone were generated and individually transformed into Agrobacterium tumefaciens strain GV3101. Agrobacterium harboring the respective constructs were mixed in the desired combinations, resuspended in infiltration buffer, and infiltrated into the leaves of Nicotiana benthamiana plants. At 2–3 days post-infiltration, the infiltrated leaf tissues were harvested and lysed using protein extraction buffer (P0013; Beyotime Biotechnology, Shanghai, China). The cleared lysates were incubated with anti-HA Nanobody Magarose Beads (KTSM1335; AlpalifeBio, Shenzhen, China) at 4 °C for 3–4 h with gentle rotation to capture HA-tagged protein complexes. After incubation, the beads were washed three times with 1× PBS to remove non-specifically bound proteins. Bound proteins were eluted by boiling the beads in SDS loading buffer at 95 °C for 5–10 min. The eluted samples were subjected to immunoblot analysis using anti-HA (Cat. No. EPR22819-101, AB236632, Lot No. 1006247-32, Abcam, diluted as 1:5000) and anti-mCherry (Cat. No. 26765-1-AP, Lot No. 00176460. Proteintech, diluted as 1:5000) antibodies, respectively.

Cell-free degradation assay

The recombinant construct pGEX-4T-1-BS-FveIAA26 was transformed into Escherichia coli Rosetta-gami (DE3) pLysS. Positive transformants were cultured and induced with 0.5 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) at 16 °C for 14–16 h. The recombinant GST-FveIAA26 protein was then purified. To obtain cellular lysates, mesophyll protoplasts were isolated from the cultivated strawberry (Fragaria×ananassa cv. ‘Benihoppe’) and treated with 0.1 mM NAA for 24 h. The treated protoplasts were collected, resuspended in lysis buffer (P0013, Beyotime, Shanghai, China), and mixed with the purified GST-FveIAA26 protein. Aliquots were withdrawn at various time points (0, 0.5, 1, 2, and 3 h), and the reactions were terminated by the addition of SDS loading buffer followed by boiling at 95 °C for 5–10 min. Immunoblotting was performed using an anti-GST antibody (Cat. No. EPR4236, AB111947, Lot No. 1043950-27, Abcam) at a 1:10,000 dilution.

Measurement of the hollow area ratio

The proportion of hollow area in F. vesca was measured using  an online web tool (https://www.smartagriculture.top/) developed by the Smart Agriculture and Integrated Weed Management group of Peking University Institute of Advanced Agricultural Sciences. The hollow area ratio in cultivated strawberry fruits was quantified by manual annotation of strawberry regions and hollow regions using image processing methods, and calculating the proportion of hollow-region pixels relative to total strawberry tissue pixels.

Accession numbers

The sequence data for the F. vesca genes can be found in Genome Database for Rosaceae (GDR; www.Rosaceae.org). The genes name and IDs (genome version 4.0.a2) used in this study: FvePIN5 (FvH4_6g00660),FvePIN1 (FvH4_5g17310), FveGH3.12 (FvH4_2g24250), FveARF5 (FvH4_2g38760), FveGH3.17 (FvH4_4g22430), FveTAR2 (FvH4_5g05880), FveIAA14b (FvH4_2g22520), FveYUC10 (FvH4_2g24750), FveGH3.5 (FvH4_3g21460), FveARF16c (FvH4_7g19350), FveIAA31 (FvH4_4g21780), FveIAA12 (FvH4_4g04700), FveGH3.11 (FvH4_7g16030), FveGH3.1 (FvH4_2g04750), FvePIN4 (FvH4_4g32450), FveIAA19 (FvH4_6g30860), FveLAX3 (FvH4_2g01570), FveIAA26 (FvH4_6g32450), FveGH3.6 (FvH4_3g04010), FvePIN8 (FvH4_6g49230), FveARF1b (FvH4_2g24730), FveIAA11 (FvH4_2g20500), FveAFB2 (FvH4_6g54140), FveIAA16 (FvH4_6g02860).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

J-H.Z. and X.H. conceived and designed the experiments; Z-M.M. and S-L.G. performed the spatial-transcriptomic and snRNA-seq; Z-M.M. and M.Z. performed the bioinformatic analysis; Z-M.M., G-Z.T., W-L.L., Y-Q.Z., X-D.Z. and C-C.Z. performed the functional test and characterized the phenotype of ProDR5::RUBY transgenic strawberry plants. X.L. drafted the model; Y-T.G in charge of strawberry materials management; W-X. and L.L. provided assistance on GUS staining; C-Y.K. provided the ProDR5::GUS, ProFvePIN1::GUS, and ProFvePIN5::GUS transgenic F.vesca materials; J-L.Y. and T.L. helped with software; Z-M.M. and J-H.Z. drafted the original manuscript; Z-C.L., H-L.L. and P-B.X. provided advice for the experimental design and revised the manuscript; J-H.Z. and X. H. approved the final version of the manuscript and Supplementary Information; J-H.Z. provided the funding support. All authors have read and approved the final manuscript.

Peer review

Peer review information

Nature Communications thanks Yang Dong, and Haiyan Hu for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Science Foundation of China (32272692, 32572998); Key R&D Program of Shandong Province (2025LZGC026); Project ZR2022QC054 and ZR2024QC094 of the Shandong Provincial Natural Science Foundation; Taishan Scholars Program (J.Z.); China Postdoctoral Science Foundation (2025M773737), and funding from Peking University Institute of Advanced Agricultural Sciences/Shandong Laboratory of Advanced Agricultural Sciences at Weifang.

Data availability

The scRNA-seq and spRNA-seq data generated in this study have been deposited in the NCBI SRA database under accession code BioProject number PRJNA1311706. The Metabolomics data are available at MetaboLights under accession code MTBLS15677. Source data are provided with this paper.

Competing interests

The authors declare no conflict of interest.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77874-6.