All F. × ananassa plants used in this study were maintained and grown in shaded greenhouses (standard or for GMOs in the case of transgenic plants) under natural sunlight and temperature conditions at IFAPA, Málaga, Spain. Temperature conditions were recorded in an outdoor meteorological station 1500 m away from the shaded greenhouses. For the different experiments, plants were clonally propagated from runners during June-September and potted in October/November in 22 cm pots with a mixture of universal substrate and river sand (3:1 v/v).

For analysis of gene expression in the different plant tissues, leaf, root, crown, flower and green, white and ripe fruit were sampled as shown in Supplementary Figure 1. They were sampled under SD conditions (9 h38 min/14 h21 min day/night; average maximum and minimum temperature from previous 4 weeks: 17.7/9.7°C) from cultivar Chandler. Three biological replicates were collected, each consisting of 4–6 plants.

For comparison of short day and long day (SD/LD) conditions, leaf and crown samples from seasonal flowering (SF) cultivar Chandler and perpetual flowering (PF) cultivar Selva were harvested in December 21st 2012 (9 h38 min/14 h21 min day/night; average maximum and minimum temperature from previous 4 weeks: 17.7/9.7°C) and June 21st 2013 (14 h40 min/9 h19 min day/night; average maximum and minimum temperature from previous 4 weeks: 26.7/16.5°C) for SD and LD photoperiods, respectively. Three biological replicates were collected at Zeitgeber time (ZT) 6, each consisting of 4–6 plants.

For analysis of circadian rhythmic expression, leaf samples from ‘Chandler’ and ‘Selva’ were collected in two separate experiments, at SD and LD conditions. SD sampling took place on December 22nd 2013 (9 h38 min/14 h21 min day/night; average maximum and minimum temperature from previous 4 weeks: 18.2/8.8°C) and LD on June 23rd 2014 (14 h40 min/9 h19 min day/night; average maximum and minimum temperature from previous 4 weeks: 29.6/19.4°C). The first time point was sampled at dawn and then at 4, 8, 12, 16, 20, and 24 h. Three biological replicates consisting of leaves from at least three plants were collected at each time point.

For evaluation of transgenic phenotypes, 9 clones of each transgenic line were propagated from runners as described above. Nine clones from two pGUS lines (pGUS2 and pGUS3) and from Camarosa wild type (WT) were grown and used as controls. For gene expression studies in 35S:FaSOC1 lines, young leaf (first leaf with fully expanded leaflets) and crown tissues from 3 biological replicates were collected under natural SD conditions (November 29th 2017, 9 h52 min/14 h17 min day/night; average maximum and minimum temperature from previous 4 weeks: 21.1/10.6°C) at ZT 3, each replicate being a pool from 3 plants. As initial evaluation of T₀ transgenic plants revealed similar phenotypes in lines pGUS2, pGUS3, and WT non-transformed plants, only the pGUS3 control was kept as control for gene expression studies.

Phenotypic data were analyzed using one-way ANOVA followed by Tukey’s test. A value of P < 0.05 was considered as statistically significant. Graphs and statistical analyses were performed using the GraphPad Prism 8 software.

Total RNA was extracted from three biological replicates following a CTAB protocol (Gambino et al., 2008) from 200 mg of frozen powdered samples from vegetative tissues or 300 mg from fruit samples. Residual DNA from RNA samples was removed using Invitrogen™ TURBO DNA-free™ Kit and cDNA was synthesized from 1 μg of RNA using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Gene expression levels were analyzed in triplicate by quantitative real-time PCR (qPCR) with the SsoFast EvaGreen supermix (Bio-Rad) and a CFX96 real-time PCR system (Bio-Rad) using a standard two-step program of 40 cycles, annealing at 60°C. The relative expression was calculated using the geometric mean of the housekeeping genes DBP and GAPDH for normalization and the 2^–ΔΔCt method (Pfaffl, 2001) unless stated otherwise in the figure legend. Primer sequence information for all analyzed genes is listed in Supplementary Table 1.

GraphPad Prism was used for statistical analysis. If data points passed the D’Agostino-Pearson omnibus K2 normality test (P < 0.001) an ordinary one-way ANOVA was employed followed by either a Fisher’s LSD test to compare preselected pairs of columns (SD/LD genotypes and GA treatment experiments), or a Tuckey’s test to compare each column mean with every other mean (tissue expression) or, when comparing each group to a control group, a Dunnett’s test (expression in transgenic lines). When data points didn’t pass the normality test, p-values were determined by a Kruskal Wallis test, followed by Dunn’s post-test for multiple group comparisons. Mean and SEM of the 3 biological replicates were plotted. *P < 0.05; ^**P < 0.01; ^***P < 0.001; ^****P < 0.0001.

For overexpression of FaCO and FaSOC1, ORFs were cloned from cv. Selva leaf and crown tissues, respectively. PCR amplification was performed using Pfu polymerase 5-prime and DNA fragments cloned in pGEMT-easy (Promega) and sequenced. To generate the 35S:CO and 35S:FaSOC1 overexpression constructs, each gene was cloned into the SalI and XbaI restriction sites of the pBINPLUS vector (van Engelen et al., 1995). A 35S:GUS construct in the same pBINPLUS vector was used as control.

Leaf explants from in vitro ‘Camarosa’ plants were transformed with Agrobacterium tumefaciens LBA4404 harboring the 35S:FaCO, 35S:FaSOC1 or 35S:GUS constructs as previously described (El Mansouri et al., 1996). Transgenic shoots were selected on medium containing 50 mg/L kanamycin and 500 mg/L carbenicillin and grown in vitro in a culture chamber under cool-white light (at 15 mE) and a long-day photoperiod (16-h light/8-h dark) at 22°C. Resistant plantlets were acclimated to soil conditions, then transferred into 22 cm pots and grown in confined greenhouse under natural environmental conditions. Preliminary evaluations on the first year were performed with only one replicate of each transgenic line. For successive years, plants were propagated during the summer and 9 plants of each line and controls were transferred to the greenhouse. Phenotype evaluation was performed during the growing season, from December to September.

Gibberellin treatment was performed on 6 weeks old clonally propagated F. × ananassa ‘Camarosa’ plants grown under natural greenhouse conditions. Six plants were used for each of the 3 biological replicates. GA₃ (SIGMA S7645) was first dissolved in ethanol at 50 mg/mL and a diluted 1:1000 working solution was made in 0.1% Tween-20 (GA₃ final concentration 50 mg/L). Controls were treated with a 0.1% Tween-20 0.1% ethanol solution. Plants were sprayed to runoff twice, with the first application conducted at day 0 and a second one at day 2. Crown and young leaf samples were collected at day 3 (October 7th, 11 h41 min/13 h18 min day/night, average maximum and minimum temperature from previous 4 weeks: 26.4/16.4°C) at ZT 3.

As a starting point for our study on the genetic pathways controlling the transition to flowering in response to day length in cultivated strawberry, we compared the expression of selected flowering time genes in the SF and PF cultivars Chandler and Selva, respectively. Light and photoperiod sensing occurs mainly in leaves and this signal is transmitted to plant meristems to determine their fate (Song et al., 2015). The strawberry plant shoot, also called crown, contains the SAM at terminal position and AXMs at the basis of each leaf. Expression analyses were performed in these two tissues, leaves and crowns (Supplementary Figure 1), collected under natural SD or LD photoperiods. The phenological phase of the meristems was determined by evaluating the expression of the major floral meristem identity genes FaLFYa, FaAP1, and FaFUL, as well as the repressor of floral transition FaTFL1 (Figure 1A). Expression levels of FaLFYa and FaAP1 were higher in crowns from plants collected under the inductive SD photoperiods in both cultivars compared to LD, whereas FaFUL induction was only detected in crowns from ‘Selva.’ On the other hand, the floral inhibitor FaTFL1 was downregulated in SD crowns, coinciding with FaAP1 and FaLFY induction. However, FaTFL1 upregulation under LD was only observed in crowns from ‘Selva.’ Rather than cultivar differences in the signaling pathways, these results are more likely to reveal differences in the timing of phase transitions between them. Altogether, these results suggest that only plants grown in SD presented floral meristems in their crowns, and that the transition to flowering, and later in LD to the vegetative phase, lagged behind in ‘Chandler’ compared to ‘Selva.’ Additionally, the floral meristem identity genes were not upregulated in the LD sampled meristems from the PF cultivar ‘Selva,’ indicating that the transition to the second reproductive phase had not taken place when samples were collected in late June.

In leaf samples harvested under LD, coinciding with the beginning of summer, we observed a clear induction of FaCO in both cultivars that correlates with higher FaFT1 expression in leaves (Figure 1B). In ‘Chandler,’ FaCO upregulation in leaves parallels a drop in FaTEM expression in the same tissue (Figure 1B). Although not statistically significant, a similar trend in FaTEM expression was observed in ‘Selva.’ As proposed in F. vesca (Koskela et al., 2012; Rantanen et al., 2014), in cultivated strawberry FaFT1 induced in leaves might act as a long-distance signal and move to the apical meristems. However, in the presence of functional FaTFL1, the overall effect is floral repression and promotion of the vegetative state, as reflected by the low transcript level of the floral markers FaLFYa and FaAP1 (Figure 1A). Noteworthy, the most prominent change in gene expression under LD in the two cultivars was the strong FaSOC1 induction detected both in leaves and crowns (Figure 1B).

In the transition to winter, under SD, FaFT2 and FaFT3 expression showed a sharp peak in ‘Chandler’ leaves. In ‘Selva,’ a moderate similar trend was detected, suggesting the rise in expression of these genes might be transient and could have occurred earlier in ‘Selva’ (Figure 1B). Additionally, under flowering inductive SD, all three FaFT genes were induced in crowns of both cultivars, although FaFT2 induction was only clearly detected in ‘Chandler’ (Figure 1B). FaFTs upregulation in the meristems is compatible with them acting upstream of the floral identity genes FaLFYa and FaAP1 and thus promoting floral induction and inflorescence development under SD in cultivated strawberry (Figure 1A).

In the two F. × ananassa cultivars analyzed in this study, FaCO and FaSOC1 expression is photoperiodically regulated. LD induced the expression of both genes, suggesting FaCO and FaSOC1 might be involved in the regulation of the seasonal alternation between flowering and runnering in response to day-length. Therefore, we decided to focus our work on these two transcription factors.

First, we analyzed the expression of FaCO and FaSOC1 in different tissues collected under SD from cultivar Chandler. As shown in Figure 2A, FaCO expression was specific to the aerial tissues of the plant, with leaves showing the highest FaCO transcript levels. This high expression is compatible with FaCO forming part of the photoperiod measurement system in leaves, as described in other species. Elevated FaCO expression was also detected in flowers and decreased during fruit development and ripening. In the case of FaSOC1, transcripts were detected in all tissues analyzed, although expression was highest in leaf and root, followed by crown, and it was markedly lower in reproductive organs (Figure 2B).

Then, the circadian expression of FaCO and FaSOC1 was analyzed in leaves collected from the SF and PF accessions ‘Chandler’ and ‘Selva’ grown under SD and LD photoperiods (Figures 2C–F). FaCO daytime expression rhythm was similar to the one described for its ortholog FvCO (Rantanen et al., 2014), showing a peak of expression at the end of the dark period then quickly declining at dawn. The same pattern was observed under LD and SD photoperiods and in the two genotypes analyzed, and thus, the different flowering habits cannot be explained by differences in FaCO diurnal pattern of expression.

On the contrary, diurnal FaSOC1 expression did not show any clear rhythm in none of the photoperiods tested nor cultivar-dependent differences. These results stand in contrast to previous observations by Koskela et al. (2016) in the SD cultivar Honeoye. They did not observe any diurnal oscillation under SD either, but in LD FaSOC1 was slightly up-regulated in the morning. As occurred in the case of FaCO, the differences in flowering behavior between genotypes can’t be explained by different circadian expression patterns of FaSOC1.

To further investigate the molecular mechanisms involved in the tradeoff between runnering and flowering and the role of FaCO in these processes in cultivated strawberry, we generated transgenic plants of the SF cultivar Camarosa ectopically expressing FaCO under the control of the CaMV 35S promoter. FaCO cDNA was amplified from leaf samples and sequence analysis revealed it corresponds to homoeolog FxaC_23g53510 from ‘Camarosa’ chromosome Fvb6-2 (Edger et al., 2019). Several putative transgenic shoots from independent transformation events were obtained, from which nine PCR positive lines were acclimated and transferred to the greenhouse for initial evaluation and selection. FaCO overexpression was confirmed by RT-qPCR, with transgenic lines showing expression levels 7–80 times higher than the control (Supplementary Figure 2A). Flowering initiation was evaluated in these plantlets coming directly from tissue culture and therefore only one clone per transgenic line was available. A wide variation in the number of days before flowering was found among the nine transgenic lines (Supplementary Figure 2B). With the exception of line CO OverExpressor (COE)-14, all transgenic lines were characterized by less vigor, showing smaller plant size and slightly smaller leaves. Runner capacity was also affected at this stage. Three of the nine transgenic lines evaluated (lines COE-22, -40, and -42) were not able to produce runners, three had an intermediate phenotype (COE-12, -15, and -21) and only three (COE-11, -13, and -14) generated a similar number of runners than the controls, although the severity of this phenotype did not correlate with transgene expression level (Supplementary Figure 2A).

Lines unable to runner cannot be propagated and therefore are lost, hampering further evaluation of the effect of FaCO overexpression on runnering capacity beyond the first generation. Among the lines able to runner, COE-11, -12, and -13 were chosen as representative of the phenotypic variability observed in T0 lines and vegetatively propagated for further analyses. A total of nine plantlets per transgenic or control lines (WT and 35S:GUS) were grown in a greenhouse under natural light and temperature conditions. The three selected lines maintained a slightly smaller plant size throughout next generations (Figure 3). Winter flower initiation was slightly delayed in COE lines compared to control plants (Figure 3A), although statistically significant differences were only observed between pGUS control and lines COE-12 and 13, which flowered an average of 16 or 14 days later, respectively. Despite this subtle delay in the emergence of the first flowers, no differences in the cumulative number of flowers were observed (Figure 3B). Following the flowering season, runner initiation occurred slightly sooner in lines COE-11 and 13 (Figure 3C), but both of them ended up producing a similar number of runners. On the contrary, line COE-12, that initiated runners at the same time than controls, produced significantly less number (Figure 3D), as already observed during the first season (Supplementary Figure 2A). Altogether, these results point to a marginal and contrasting role of FaCO in the control of meristem fate, at least under the growing conditions used in this work.

FvSOC1, the FaSOC1 ortholog from the diploid relative F. vesca, is known to alter the balance between runnering and flowering toward runnering, acting as a central hub where the fate of the meristem is established (Mouhu et al., 2013). To determine whether FaSOC1 has a similar role in the cultivated strawberry, it was overexpressed in F. × ananassa cv. Camarosa. FaSOC1 cDNA used for transformation was amplified from crowns and corresponded to the homoeolog FxaC_25g18220 from ‘Camarosa’ chromosome Fvb7-2 (Edger et al., 2019). Eight PCR positive lines were selected and transferred to soil, all of them showing markedly higher FaSOC1 expression in leaves than the control (Supplementary Figure 3A).

In a first-year evaluation, all 35S:FaSOC1 lines exhibited increased vegetative development except line SOC1 OverExpressor (SOE)-1, the one with lower FaSOC1 expression (Supplementary Figure 3C). A correlation between transgene expression level and phenotypic severity was observed and thus, lines SOE-20 and -21, the higher overexpressors, also showed the most severe phenotype. As occurred in 35S:FaCO T₀ plants, a wide variability in flowering time was found among the eight 35S:FaSOC1 lines. Four of them were late flowering, with SOE-7, -9, and -20 being particularly late, as they flowered at least 60 days later than the controls (Supplementary Figure 3B).

Lines SOE-7, -9, -20, and -21 were clonally propagated from runner cuttings. Nine plants per line, except SOE-7, were subjected to further molecular and phenotypic evaluation during the growing season, from December to September, in successive years. The number of days until the emergence of the first flower and the total number of flowers per plant were scored, showing FaSOC1 overexpression had an evident impact on flowering induction (Figure 4). All 35S:FaSOC1 lines flowered later than the controls (Figure 4A). Lines SOE-9, -20, and -21 flowered an average of 20, 25 and 9 days later than the WT, which in turn flowered 4 days later than the pGUS control, although statistically significant differences were only observed between both controls and SOE-9 and -20. Additionally, SOE-20 presented a significantly reduced number of flowers compared to control lines.

A very distinctive phenotype of the four SOE-7, -9, -20, and -21 lines was the production of runners in winter, immediately after separation from mother plants and before floral induction (Figure 4H). During summer LD, the three evaluated lines also produced runners significantly earlier than the controls (Figure 4C). Control lines started runners 223 or 227 days after planting while transgenic lines produced runners after 188–203 days. Although transgenic lines generated a slightly higher number of runners per plant, no significant differences were observed. The more vigorous appearance of 35S:FaSOC1 lines was not a consequence of higher amount or larger leaves but rather more erect and taller plants. Although all of them were higher than the controls (Figure 4H), statistically significant differences were only observed in SOE-9 and -20 (Figure 4E). This phenotype was the reflection of considerably longer leaf petioles displayed by all SOE lines (Figure 4F), which also presented longer inflorescence peduncles (Figure 4G).

Ectopic and constitutive overexpression of FaSOC1 negatively affected floral and fruit development (Supplementary Figure 4), increasing the incidence of malformed fruits and aborted flowers. Flowers within the inflorescences were more compact and presented defects in the sepal and petal whorls. Sepals were bigger whereas petals, although present, were reduced in size and senesced prematurely compared to the WT. Fruits were in general smaller and in the majority of cases did not develop into fleshy fruits. Line SOE-20, the one with higher transgene overexpression and stronger vegetative phenotype, was also the most affected in fruit development. The majority of SOE-20 fruits got completely dry in the green stage and only a small percentage developed further into dry pinkish fruits that remained attached to the plant. Line SOE-9 presented an intermediate phenotype and developed few wild type ripe fruits, while about half the fruits in line SOE-21 ripened into red fleshy fruits. Noticeably, achenes remained green in all three transgenic lines (Supplementary Figure 4).

Taken together, the photoperiodic pattern of FaSOC1 expression and the phenotype of 35S:FaSOC1 lines, point to FaSOC1 acting as a negative regulator of flower induction and promoter of vegetative growth in cultivated strawberry.

Since the transgenic plants overexpressing FaSOC1 have a late flowering phenotype, the expression of flowering time genes, including FaCO, FaTEM, FaFT1-3, FaTFL1, FaLFYa, FaAP1, and FaFUL, was tested in all of them. Expression analysis were performed in young leaf and crown tissue collected under natural SD conditions. Overexpression of FaSOC1 in leaf and crown of transgenic plants used in these experiments was confirmed by RT-qPCR (Figure 5A). As observed in the first season (Supplementary Figure 3A), SOE-20 was the line with highest transgene expression, followed by SOE-21 and then SOE-9.

FaCO expression levels were similar in 35S:FaSOC1 and control plants in the two tissues analyzed (Figures 5B,C), supporting the model proposed in F. vesca that places FvCO upstream of FvSOC1 (Kurokura et al., 2017). However, FaTEM was downregulated in leaves of all three SOE lines, although differences were only statistically significant in SOE-21. FaTEM expression in crowns, whereas not as evident as in leaves, also tended to be downregulated in the three transgenic lines (Figures 5B,C). The only FaFT family member detected in leaves of these SD grown plants was FaFT1, but its expression level was low and not affected by FaSOC1 overexpression (Figure 5B).

The expression of the floral meristem genes FaLFYa, FaAP1, and FaFUL was significantly downregulated in 35S:FaSOC1 crowns compared to the control (Figure 5C), in agreement with their respective phenological stage. Whereas controls were flowering, SOE lines showed active vegetative development at the time of sampling. Additionally, FaTFL1 expression was induced in lines SOE-20 and -21 (Figure 5C), supporting FaTFL1 activation downstream of FaSOC1 also happens in F. × ananassa as previously described in F. vesca (Mouhu et al., 2013).

Along with expression changes in floral meristem identity genes, the most dramatic changes in gene expression observed in 35S:FaSOC1 crowns affected FaFT2 and FaFT3, with transcript levels of both genes greatly reduced in all SOE lines (Figure 5C). In contrast, FaFT1 expression was similar in SOE and control lines (Figures 5B,C).

‘Camarosa’ plants overexpressing FaSOC1 showed a characteristic elongated phenotype and initiated runners during winter short days, as previously observed in F. vesca overexpressing FvSOC1 (Mouhu et al., 2013), resembling the effect of GA-treated plants (Thompson and Guttridge, 1959; Hytönen et al., 2008). Indeed, in F. vesca, FvSOC1 positively regulates the expression of the GA biosynthetic genes FvGA20ox and FvGA3ox required for runner differentiation (Mouhu et al., 2013). In order to evaluate if FaSOC1 modulates GA metabolism, the expression of FaGA20ox2, FaGA20ox4, FaGA3ox1, FaGA3ox2, and FaGA2ox1 was analyzed in leaves and crowns of SOE lines grown under SD.

FaGA20ox2 and FaGA3ox2 were not detected in any of the tissues analyzed, neither in control nor in transgenic lines. FaGA20ox4 was not detected in leaves but, although close to detection limits, it could be amplified from crowns. In lines SOE-9 and SOE-21 FaGA20ox4 transcript level was similar to the control (Figures 6A,B). Only in the most severe SOE-20, FaGA20ox4 tended to be upregulated but transcript levels were highly variable among the three biological replicates and thus not statistically significant (Figure 6B). FaGA3ox1 expression in leaves and crowns was similar in SOE and control lines (Figures 6A,B).

As active GA levels are the result of their biosynthesis as well as their deactivation rates, FaGA2ox1 expression, an oxidase involved in GA deactivation, was also analyzed. FaGA2ox1 transcripts were detected in leaves and crowns but the expression level, as happened with FaGA20ox4 and FaGA3ox1, was not affected by FaSOC1 overexpression (Figures 6A,B). Hence, our results do not support a role for FaSOC1 as an inducer of active GA accumulation in F. × ananassa, as was shown for its ortholog FvSOC1 in F. vesca (Mouhu et al., 2013).

Alternatively, as a similar constitutive GA response phenotype was described in the suppressor of runnerless (srl) F. vesca mutant (Caruana et al., 2018), the vegetative phenotype of SOE lines could also be explained by lower expression of the GA signaling repressor FaRGA1 instead of activation of GA biosynthesis. However, no significant differences in FaRGA1 expression were detected between the control and transgenic lines (Figures 6A,B), and therefore, the vegetative phenotype of SOE lines cannot be explained by neither increased endogenous active GAs levels nor reduced FaRGA1 expression.

Our expression studies indicate that the elongated phenotype and increased runner production observed when FaSOC1 was overexpressed was not due to a rise on GA biosynthesis as previously reported in F. vesca (Mouhu et al., 2013). Alternatively, we hypothesized that FaSOC1 could be a target gene of GA signaling in F. × ananassa as it has been shown in Arabidopsis (Moon et al., 2003). To test this possibility, F. × ananassa plants were sprayed with 50 ppm GA₃ and gene expression analysis was conducted on GA₃- and mock-treated plants. Since there is strong evidence for negative feedback control of the expression of GA20ox and GA3ox genes by GA (Hedden and Phillips, 2000), the expression of FaGA20ox4 and FaGA3ox1 was analyzed. In our conditions, FaGA20ox4 expression did not respond to GA application but FaGA3ox1 was notably down-regulated in GA₃-treated samples, both in leaves and crowns, indicating that FaGA3ox1 is involved in the negative feedback regulation of the GA biosynthetic pathway in F. × ananassa (Supplementary Figure 5). Despite plants were responding to GA treatment, identical FaSOC1 transcript levels were found in control and treated plants (Figure 7A).

To further investigate the molecular events downstream of GA perception, the expression of other flowering time genes was examined. Transcript accumulation of FaLFYa, FaTFL1, FaFT1, and FaFT2 did not significantly differ between treated and untreated samples (Figure 7A). In contrast, we observed significant differences in the accumulation of FaCO, FaTEM, and FaFT3 (Figure 7A). In leaves, FaCO and FaTEM expression followed opposite regulation, with GA₃ mimicking the effect of LD conditions on their expression levels (Figures 1B,7A). FaCO levels raised after GA₃ treatment whereas FaTEM dropped in the same samples. In crowns, a two-fold induction of FaFT3 was observed in response to GA₃, suggesting that FaFT3 upregulation in the AXMs might be required for the transition to the vegetative phase in F. × ananassa.

In strawberry, active GA levels, together with photoperiod and temperature, regulate the transition to the vegetative phase. In F. vesca, the signaling cascade mediating these effects has been partially elucidated, with LD upregulating FvSOC1 which, in turn, induces FvGA3ox and FvGA20ox gene expression, resulting in higher GA accumulation (Mouhu et al., 2013). In F. × ananassa, GA biosynthesis is required to induce runner production under LD (Hytönen et al., 2009) but, according to our results, its activation is not mediated solely by FaSOC1. Still, GA metabolism genes might be photoperiodically regulated through a FaSOC1 independent pathway. In order to explore how photoperiod regulates GA metabolism in F. × ananassa and putative differences between cultivars with contrasting flowering habits, FaGA3ox1, FaGA20ox4, and FaGA2ox1 expression was analyzed in ‘Selva’ and ‘Chandler’ under SD and LD natural conditions.

FaGA3ox1 expression in leaf and crown was promoted by SD (Figure 7B). By contrast, FaGA20ox4, which in F. vesca is essential for runner production (Tenreira et al., 2017), showed opposite regulation. In leaves, FaGA20ox4 transcripts were barely detected under any of the photoperiods tested. In SD crowns, FaGA20ox4 was only detected in one biological replicate from the SF cultivar Chandler (Figure 7B). However, LD conditions considerably upregulated FaGA20ox4 accumulation in crowns from the two cultivars. Interestingly, in parallel to FaGA20ox4 induction in LD crowns, a sharp increase in FaGA2ox1 expression was detected exclusively in the PF cultivar Selva (Figure 7B). This FaGA2ox1 upregulation merely in the PF cultivar suggests a putative mechanism to regulate continuous flowering in F. × ananassa, as FaGA20ox4 induction in LD-grown meristems would be likely counteracted by the high level of FaGA2ox1 expression, and therefore GA deactivation would neutralize its biosynthesis.

FaSOC1 expression is detected in a range of vegetative and reproductive tissues, although expression levels are notably lower in flowers and fruits indicative of a major role during vegetative development. No significant diurnal rhythm was observed in its expression neither in SD nor in LD. In contrast, FaSOC1 expression showed a marked response to day length in both leaf and crown and in ‘Chandler’ and ‘Selva,’ with an average six-fold induction under long photoperiods, again supporting its role during the vegetative phase as described for F. vesca FvSOC1 (Mouhu et al., 2013). Overexpression of FaSOC1 in the SD cultivar Camarosa delayed flowering, confirming the role of FaSOC1 as a strong repressor of flowering. At the transcriptional level, FaSOC1 overexpression downregulates the expression of FaFT2 and FaFT3 in crowns, with a concomitant decline in the accumulation of floral identity genes such as FaAP1, FaLFYa, and FaFUL (Figure 5B). Mouhu et al. (2013) showed FvSOC1 activates the transcription of FvTFL1 and that FvTFL1 function is required for the FvSOC1-induced flower repression in F. vesca. In our SOE lines, FaTFL1 upregulation was not consistent in all transgenic lines (Figure 5B), being only detected in SOE-20 and -21, the ones with higher transgene expression level. A yet uncharacterized SOC1 independent mechanism of TFL1 regulation has been proposed in F. vesca acting at low (<11°C) and high temperatures (>23°C) (Rantanen et al., 2015) or in F. nilgerrensis at 11–18°C (Fan et al., 2022), suggesting this other temperature-responsive pathway could be acting in ‘Camarosa’ in our growing conditions. Since the late flowering phenotype was similar in the three transgenic lines (Figure 4A) and it did not correlate with FaTFL1 expression level, we concluded that in ‘Camarosa,’ FaSOC1-dependent FaFT2 and FaFT3 repression in crowns might have a major role mediating flowering repression. Additionally, FaTFL1 activation is only achieved at higher FaSOC1 transgene levels and do not correlate with the strength of the flowering time phenotype. These results highlight once more the importance of the TFL1/FT balance in the control of meristem fate in strawberry (Gaston et al., 2021) and suggest FaSOC1 modifies this balance to promote the vegetative state in F. × ananassa mainly through FaFT2 and FaFT3 downregulation.

In addition to delaying flowering, overexpression of FaSOC1 has various effects on growth and development during the vegetative phase. In SF octoploid cultivars, under natural growing conditions, a sharp upregulation of FaSOC1 expression in leaves and crowns correlates with the induction of runner formation and petiole elongation, which are normal photoperiodic responses to increasing day lengths (Guttridge and Thompson, 1964). We show here that FaSOC1 is involved in the regulation of both processes as petioles are significantly longer and the emergence of runners takes place 3–6 weeks earlier in 35S:FaSOC1 than in control lines, although the final count of runners is only slightly increased at the end of the season (Figure 4). Similar morphological and physiological effects are observed after exogenous GA treatment (Thompson and Guttridge, 1959; Guttridge and Thompson, 1964; Hytönen et al., 2009). In fact, the phenotype of F. vesca plants overexpressing FvSOC1, comparable to that in our 35S:FaSOC1 lines, is due to an upregulation of GA3-ox and GA20-ox genes, leading to increased GA accumulation (Mouhu et al., 2013). However, we could not confirm a similar mechanism in ‘Camarosa’ SOE lines, as none of the tested genes in the GA metabolic pathway presented differential expression compared to the control (Figure 6). The activity of the GA biosynthetic enzyme GA20ox is of particular importance in determining GA concentration in many plant species. In F. vesca, FvGA20ox4 activation in AXM under LD has been shown to be required for stolon development. Whereas at 18°C this activation occurs via an FvSOC1-dependent photoperiodic pathway (Mouhu et al., 2013; Tenreira et al., 2017; Andrés et al., 2021), FvGA20ox4 upregulation also occurs independently of FvSOC1 (Andrés et al., 2021; Fan et al., 2022). The latter situation is therefore in line with our observations in cultivated strawberry. Although FaSOC1 overexpression did not affect that of FaGA20ox4, this gene showed a marked photoperiodic response in crowns in response to LD (Figure 7B), suggesting it plays a critical role in the accumulation of active GA required for inducing the vegetative phase, in agreement with its role in the diploid species. Furthermore, an interesting difference between the PF cultivar Selva and the SF cultivar Chandler was observed in FaGA2ox1 expression, coding for a GA inactivation enzyme (Figure 7B). FaGA2ox1 transcript accumulation in ‘Chandler’ was low under SD and LD photoperiods, and equivalent to that in ‘Selva’ under SD. However, a marked upregulation of FaGA2ox1 was detected in ‘Selva’ under LD. Increased FaGA2ox1 activity is expected to decrease the levels of bioactive GAs, negatively affecting runner production and promoting branch crown development, aiding continuous flowering in ‘Selva.’

In an attempt to explain the constitutive GA response in SOE lines, we considered the possibility of FaSOC1 being a mediator of GA signaling, activated downstream of GA perception, similar to Arabidopsis SOC1, which is induced by GAs (Lee and Lee, 2010). We could not confirm this hypothesis as, in our studies, FaSOC1 expression was equivalent in GA-treated and mock plants (Figure 7A). However, although plants were sensing and responding to GA, a weakened response to exogenous GA cannot be completely ruled out under our photoperiodic conditions, as daylight was slightly below 12h at the time of the first application. Previous studies showed F. × ananassa plants are able to respond to exogenous GA treatment under SD (Guttridge and Thompson, 1959), but also a reduced sensibility has been observed in short compared to long photoperiods (Hytönen et al., 2009).

Considering FaSOC1 expression does not respond to GA treatment, we therefore hypothesized that GA signaling and FaSOC1 might have common target genes that would account for the vegetative phenotype observed in SOE lines. Among the flowering genes evaluated in this work, FaTEM1 expression is downregulated in leaves of GA treated and 35S:FaSOC1 plants. Notably, a similar trend was observed in ‘Selva’ and ‘Chandler’ grown under LD (Figure 1B). The fact that FaTEM was downregulated in different scenarios, all of them promoting vegetative development, pointed to a putative role of this factor in the repression of flowering in F. × ananassa. TEM proteins have been proposed to act as regulators of the juvenile vegetative phase in species as diverse as antirrhinum, Arabidopsis or olive (Sgamma et al., 2014), including another perennial Rosaceae such as loquat (Peng et al., 2021). In Arabidopsis, TEM1 represses photoperiodic flowering at early stages of vegetative growth, then its expression levels progressively decay throughout development to reach a minimum at the time of floral transition, thus allowing FT activation by CO (Castillejo and Pelaz, 2008). Recently, it has been shown that heterologous overexpression of apple MdTEM in F. vesca H4 leads to FvFT1 induction, whereas the opposite effect is observed in RNAi-MdTEM1 lines (Dejahang et al., 2022). In 35S:MdTEM1 and 35S:MdTEM2 lines, generated in the fvtfl1 mutant background of H4, elevated FvFT1 expression levels in leaves correlated with an extremely early flowering (Dejahang et al., 2022), phenocopying FvFT1 overexpression in H4 (Rantanen et al., 2014). However, in a TFL1 WT genotype, FT1 upregulation in leaves would be expected to induce TFL1 in the SAM and repress flowering.

In F. vesca H4, FvCO is required for FvFT1 induction in leaves in response to long photoperiods, functioning as a strong promoter of flowering under LD as there is no functional FvTFL1 (Kurokura et al., 2017). However, in SF genotypes carrying a functional TFL1, the role of CO in the photoperiodic control of the alternance between vegetative and reproductive development needs further investigation. According to the model proposed in F. vesca, and considering ‘Camarosa’ harbors a functional allele of the flowering repressor FaTFL1, an expected outcome of FaCO overexpression in ‘Camarosa’ was the activation of the FaFT1-FaSOC1-FaTFL1 module and consequently, flowering repression and promotion of vegetative development. As shown in Figure 1, in F. × ananassa, we observed a correlation in the upregulation of FaCO and FaFT1 in leaf, and FaSOC1 and FaTFL1 in crowns by LDs. However, overexpression of FaCO in SF ‘Camarosa’ only causes a subtle delay on flowering time and a similar discrete advance of runnering (Figure 3A), resulting in a slightly extended vegetative phase. These results might indicate the existence of an additional factor that prevents vegetative development under SD even in the presence of elevated FaCO. A putative candidate could be FaTEM, as it has been shown to repress FaFT1 expression (Dejahang et al., 2022) and it is upregulated in response to SD (Figure 1A). As a result, elevated FaTEM expression levels would not allow FaFT1 activation by FaCO. Additionally, in contrast to the moderate effect of FaCO in prolonging the vegetative phase, a negative effect of FaCO overexpression on runner capacity was detected in most of the T₀ transgenic lines and further observed in COE-12 (Figure 3D) in the following season. Although T₀ phenotypes might be affected by the fact that those plants come directly from in vitro culture and maybe not related to transgene expression, the possibility of FaCO negatively affecting runner initiation should not be discarded and, in that case, the phenotypes of the three selected runnering lines might not fully reflect FaCO function. In depth characterization of 35S:FaCO transgenic lines awaits in order to elucidate the molecular mechanisms regulating FaCO effects on flowering and runnering.

In this work we propose a role for FaCO and FaSOC1 as weak and strong repressors, respectively, of the photoperiodic flowering pathway in F. × ananassa. Additionally, our expression analysis performed in the SD and PF cultivars, ‘Chandler’ and ‘Selva’ respectively, shed some light into the molecular mechanisms leading to flower induction in both of them. In many different species, FT genes have been shown to be major components of the florigen, a graft-transmissible signal produced in the leaves that induces flowering at the shoot apex in response to inductive photoperiods (Andrés and Coupland, 2012). In addition to regulating flowering, FT genes have been implicated in a range of physiological processes, such as promotion of vegetative development in poplar (Böhlenius et al., 2006), tuberization or bulbing in potato and onion, respectively (Navarro et al., 2011; Lee et al., 2013), or more pleiotropic roles as general regulators of growth, like in tomato or maize (Lifschitz et al., 2006; Danilevskaya et al., 2011). In the strawberry genome, three FT homologs have been identified (Nakano et al., 2015), however there is still scarce knowledge about the physiological process controlled by each FaFT member, or if variations in their expression pattern are able to explain differences in the vegetative and flowering responses of SD or PF cultivars. Most recent studies overexpressing F. vesca FvFT2 in F. vesca and F. × ananassa suggest that FaFT2 might act as florigen under SD in cultivated strawberry (Gaston et al., 2021).

In SF cultivars like ‘Chandler,’ floral induction usually takes place at late fall, when diurnal light periods and temperatures become lower, whereas in PF cultivars as ‘Selva,’ flowers are initiated continuously throughout the growing season from spring until late autumn. The length of the flowering period is also variable among cultivars within the SF or PF habits and dependent on genetic and environmental conditions (Stewart and Folta, 2010; Heide et al., 2013; Labadie et al., 2019). In the present study, we have conducted an expression analysis of the three FaFT homologs in crowns and leaves of F. × ananassa ‘Selva’ and ‘Chandler’ grown under natural conditions and collected in SD (December 21st) or LD (June 21st). Under floral promoting SD, higher expression levels of the three FaFT homologs, with the exception of FaFT2 in ‘Selva,’ are observed in crowns, and correlate with the induction of floral identity genes such as FaAP1 and FaLFY and downregulation of the floral repressor FaTFL1 (Figures 1A,B). FaFT1 and FaFT3 upregulation in shoot tips under floral inductive conditions have been previously described in other octoploid cultivars (Nakano et al., 2015; Koembuoy et al., 2020). SD specific FaFT2 induction in shoot tips was also documented by Nakano et al. (2015) after 7 days of photoperiod treatments, although no differences were observed after 21 days. Interestingly, FaFT2 and FaFT3 induction in crowns is abolished when FaSOC1 is overexpressed (Figure 5B), suggesting SD dependent FaSOC1 downregulation (Figure 1) is required to allow activation of the floral promoter FaFT genes. Alternatively, FaSOC1 might directly repress FaFT2 and FaFT3 in meristems.

Gene expression changes in crowns are expected to occur in response to a long-distance signal coming from leaves (Colasanti and Sundaresan, 2000; Andrés and Coupland, 2012). Accordingly, we observed SD photoperiods induce the accumulation of FaFT2 and FaFT3 transcripts in leaves, being more evident in SF ‘Chandler’ than in PF ‘Selva’ at the time sampling was performed (Figure 1B). Upregulation of FvFT3/FaFT3 by SD was not detected in leaves in previous studies by Nakano et al. (2015) or Gaston et al. (2021). In this latter study, FaFT2 expression was equivalent in SD and LD leaf samples from the F. × ananassa Japanese cultivar Nyoho. In contrast, Gaston et al. (2021) detected a transient upregulation of FvFT2 under SD conditions in a PF, but not in a SF, F. vesca genotype. In their monthly time-course performed from June to November under natural environmental conditions, FvFT2 peaked in October. The observed discrepancies might be due to natural variation among genotypes. Alternatively, peaks might be easily missed due to their transient nature. FaFT2 induction in PF ‘Selva’ could have already happened when we sampled in December, whereas FvFT2 rise in the SF cultivar used in Gaston et al. (2021) could have taken place after their last sample in November. According to our qPCR data, upregulation of FaFT2 and FaFT3 in leaf is compatible with both of them having partially overlapping roles as florigens. Indeed, FvFT2 has been already shown to function as a strong floral promoter when overexpressed in F. vesca and F. × ananassa (Gaston et al., 2021; Sabbadini et al., 2021). In the same study, FvFT3 overexpression only caused a modest advance of flowering in F. vesca. In contrast, when FvFT3 is overexpressed in F. × ananassa, it promotes the vegetative state of AXM and thus runner development (Gaston et al., 2021; Sabbadini et al., 2021). This apparently contradictory result, with FaFT3 expression in leaves associated with floral inductive conditions, and the promotion of runner production when overexpressed, raised the possibility of FaFT3 fulfilling different roles, as floral inducer in leaves and SAM and as runner promoter in AXM. In F. × ananassa, GA is a key signal determining bud fate and needed for runner initiation (Hytönen et al., 2009). Therefore, we quantified FaFT1-FaFT3 expression in GA-treated samples. Interestingly, FaFT1 and FaFT2 expression do not respond to GA, but a two-fold FaFT3 induction is observed specifically in crowns after exogenous GA application (Figure 7A), supporting a key role for FaFT3 in the regulation of the vegetative developmental fate of AXM downstream GA perception.

As for FaFT1, its expression pattern follows opposite photoperiodic regulation in crowns compared to leaves, suggesting it might also fulfill tissue specific roles. Similarly to F. vesca FvFT1 (Koskela et al., 2012), FaFT1 is induced in leaves under LD, promoting the vegetative state of plant meristems. However, under SD, when FaTFL1 expression drops in the meristems, FaFT1 can act as floral inducer in crowns as previously shown in F. vesca H4 in LD or when FvFT1 is overexpressed (Rantanen et al., 2014).

Altogether, our results suggest the three F. × ananassa FT proteins have overlapping and particular roles in the regulation of the alternance between flowering and vegetative development. FaFT2 is proposed as the florigenic signal transiently produced in leaves under SD, which likely exerts its function redundantly with FaFT3. At the SAM, FaFT1-3 induction under SD indicates overlapping roles in the promotion of flowering transition. Additionally, the transition to the vegetative phase would require upregulation of FaFT1 and FaFT3 in leaf and AXMs, respectively.

Overall, this work supports that F. × ananassa FaSOC1 has retained FvSOC1 role as repressor of flowering and promoter of vegetative development. However, signaling events downstream FaSOC1 show particularities compared to F. vesca that hinders direct transfer of knowledge acquired in the diploid species to cultivated ones. On the contrary, FvCO has been shown to promote flowering while repressing runner development in fvtfl1 F. vesca accessions, whereas our study suggests that FaCO responds to LD photoperiods as in F. vesca with a different outcome, as flowering was marginally delayed in COE lines. Comparison of FvCO and FaCO protein sequences did not reveal substantial differences able to explain the contrasting phenotypes in the two Fragaria species. A total of 13 aminoacid substitutions were found in the coding sequence, 6 non-synonymous, but none of them affected conserved residues (Supplemental Figure 6). The presence of a functional FaTFL1 allele in the SF cultivar Camarosa and COE lines is an obvious difference with reported studies in F. vesca and if its activation occurs downstream of FaCO, as it has been suggested for FvTFL1, a more dramatic repression of flowering was expected. The subtle negative effect on flowering of FaCO overexpression and the contrasting effects on runner initiation suggest the existence of putative additional factors that counterbalance its activity, particularly under SD conditions.

Moreover, our results provide a groundwork for detailed characterization of factors that may play a critical role in the tradeoff between flowering and runnering in F. × ananassa. One of such factors is FaTEM, which expression pattern is compatible with a role as repressor of vegetative development under SD, and thus promoter of flowering, likely through preventing FaCO-dependent FaFT1 activation in leaves. Additionally, LD induction of FaGA2ox1 in crowns specifically in PF ‘Selva’ suggests a putative mechanism to lower active GA levels in PF genotypes in order to enable successive rounds of flowering. Lastly, detailed expression of FaFT genes suggest the three homologs from F. × ananassa might have overlapping roles inducing flowering in the apical meristem under SD, but also functional divergence in the case of FaFT1 and FaFT3 is proposed. Although further studies are needed to clarify these mechanisms, our work broadens our knowledge on the regulatory pathways controlling flowering and vegetative growth in cultivated strawberry in response to photoperiod, and how these compare with flowering regulation in annual plants or the diploid relative F. vesca.