Fruit growth
Fruit on the day of anthesis is shown in Figure 1a. Fruit at 3 days after anthesis is shown in Figure 1b; at this time point they had not begun exponential growth, which is a typical symptom of growth suppression of parthenocarpic fruit. The growth of fruit based on length is shown in Figure 2a. The onset of exponential growth was set at the time at which RGR became larger than 2% h–1. Some fruits started to grow exponentially at various DAA, whereas other fruits did not grow exponentially. The distribution of the DAA of the onset of the exponential growth is shown in Figure 2b. About 60% of the fruits did not exhibit exponential growth, and most of them eventually became soft: they were regarded as having aborted. The distribution of the DAA of the fruit abortion is shown in Figure 2c.
Gene expression analysis
At 8 DAA, the levels of the transcripts of all genes were high only in one fruit, and were close to zero in the remaining fruits. This was because most slow fruits aborted, and some slow fruits were larger than the marketable size, in which the RGR decreases normally. Therefore, gene expression was determined from 0 to 6 DAA.
The variation in transcript levels was very large. One possible explanation for this observation is that the extent of the sugar starvation fluctuated greatly among fruits. In this case, the correlation between the transcript level of the genes will be high. The visualized correlation matrix of the genes and RGR at 2–6 DAA is shown in Figure 3. The correlation coefficients (r) of the transcript level of the genes and RGR were moderately negative, with the exception of that obtained for HB (its absolute r value was very low). This negative r was consistent with the relationship between growth cessation and gene expression. The r values among the transcript levels of the AS, CsSEF1, CsFDI1, PMS, and uk4 genes were relatively high, suggesting that these genes are regulated similarly. Remarkably, the r value between CsSEF1 and uk4 and that between CsFDI1 and uk4 were very high.
The variation of the data was much larger than that reported previously (Tazuke and Asayama 2013; Tazuke et al. 2015; Tazuke et al. 2017). The study of the effect of salinity on fruit growth revealed that the RGR of a fruit in which growth was suppressed by salinity tended to fall to about 1.5% h–1 (Tazuke 1997). Thus, for each DAA, the fruits were classified into two groups: those with an RGR >1.5% h–1 (referred to as “fast fruit” hereafter), and those with an RGR <1.5% h–1 (referred to as “slow fruit” hereafter). The average and standard error of each group were calculated. The expression of genes obtained using the subtraction method is shown in Figure 4. The expression of genes obtained by RNA-seq is shown in Figure 5.
For the AS gene, the transcript levels in both fast and slow fruit were high from 2 DAA compared with that observed on the day of anthesis. In slow fruit, the transcript level tended to increase at 6 DAA, whereas in fast fruit, the transcript level decreased markedly. (Fig. 4a).
For the CsSEF1 gene, the transcript level of fast fruit remained low from 0 to 6 DAA. In slow fruit, the transcript level began to increase at 4 DAA and became as high as 50 times the transcript level detected at 0 DAA, although the variance was very large (Fig. 4b).
For the CsFDI1 gene, the pattern of change in transcript levels was similar to that of CsSEF1, although the transcript level of slow fruit at 6 DAA was only about five times the transcript level detected at 0 DAA; however, its variance was very large (Fig. 4c).
For the HB gene, the transcript level increased markedly at 6 DAA. It was higher in fast fruit and about 1000 times higher than the transcript level detected at 0 DAA (Fig. 5a).
For the PMS gene, the transcript level of slow fruit was higher than that of fast fruit and increased with the DAA. The variation in the slow fruit category was very large. The transcript level of fast fruit was relatively high (Fig. 5b).
For the NAC gene, the variability of the slow fruit was very high at 2 DAA and the transcript level of slow fruit was very low at 4 DAA. The transcript level of the slow fruit at 6 DAA was about 200 times the transcript level detected at 0 DAA. The transcript level of fast fruit remained low (Fig. 5c).
For the NA3 gene, the pattern of the change in the transcript level was similar to that of CsSEF1. The transcript level of slow fruit at 6 DAA was about 400 times the transcript level detected at 0 DAA (Fig. 5d).
Finally, for the uk4 gene, the pattern of the change in the transcript level was similar to that of CsSEF1. The transcript level of slow fruit at 6 DAA was about 100 times the transcript level detected at 0 DAA (Fig. 5e).
We identified three patterns of gene expression—type I (AS and PMS), type II (CsSEF1, CsFDI1, NA3, and uk4), and type III (HB and NAC)—based on the time course of gene expression and its correlations.
Sugar levels
In both fast and slow fruit, the sucrose concentration was very low (data not shown). The sum of the concentrations of glucose and fructose was regarded as the hexose concentration and is presented in Figure 6. In fast fruit, hexose concentration was low at 0 DAA and increased steadily with DAA. In slow fruit, hexose concentration became even lower with increasing DAA.