Visualization and morphological analysis of the void distribution in pea sprouts before and after postharvest regrowth
2026 Volume 19 Issue 1 Pages 15-23
Details
2026 Volume 19 Issue 1 Pages 15-23
The effects of regrowth on the internal structure of green pea sprouts. Using X-ray micro-CT, void distribution was successfully visualized and characterized in sprouts before and after regrowth. In the CT images, a cavity appeared at the center of the stem after regrowth. Individual void analysis using Avizo software for three-dimensional void structures showed a shift in the void volume distribution after regrowth with the emergence of irregularly shaped voids. However, textural analysis indicated a significant decrease in crispness after regrowth. This study suggested that the decrease in crispness correlated with void distribution modifications. In addition, metabolome analysis revealed the downregulation of mannitol and other carbohydrates, indicating reduced photosynthetic activity and sustained aerobic respiration in regrown sprouts.
In recent years, micro-scale vegetable production has evolved, giving rise to early growth forms called sprouts and microgreens (Cano-Lamadrid et al., 2023; Du et al., 2022; Galieni et al., 2020). Green pea sprouts (Pisum sativum L.) are widely consumed in Asian countries, such as Thailand, China, and Japan, because of their rich nutrient and bioactive compound contents (Suathong et al., 2024). These are typically cultivated in plant factories that are less affected by weather conditions. Hence, they are harvested throughout the year, and their production quantity and price are highly stable. In Japan, green pea sprouts are packaged with attached roots to extend their shelf life. In households, roots, which are inedible parts of the sprouts, are often used for “Regrowing”. Regrowing green pea sprouts involves cultivating leaves and stems anew using the roots and side shoots remaining after initial use. This process is easily performed by soaking the samples in water (Nishikawa et al., 2023). The advantages of regrowth include cost savings on food by utilizing the regrown parts for cooking as well as the joy and satisfaction derived from growing plants. Consequently, regrowing green pea sprouts is widely embraced in households in practices such as kitchen gardening in Japan. However, regrown green pea sprouts are said to have a different texture, including reduced crispness. Elucidating the underlying causes may provide a way to develop methods to enhance the palatability of green pea sprouts, potentially expanding their use as an effective technique for vegetable scrap. Unfortunately, the current situation is characterized by a scarcity of academic research on vegetables, including green pea sprouts, which are grown through regrowth.
Fillion et al. (2002) used the term “crispy” to describe the light and thin texture of vegetables, with celery as a fitting example of a fiber-rich stem vegetable embodying this characteristic. Suathong et al. (2024) also included crispness as a factor indicating a good quality of green pea sprouts noted during a sensory evaluation. Typically, the factors influencing the crispness of food products include material properties (physical properties and composition of constituents) and structure (Primo-Martin et al., 2008). In particular, internal structural elements, such as pore thickness and porosity, have been linked to crispness (Luyten et al, 2006; Primo-Martin et al., 2008; Primo-Martín et al., 2010). Hosoi et al. (2018) reported that the distribution of pores in food influences texture formation. In the case of fruits and vegetables, it is well documented that the physical properties and composition of the cell wall as well as the turgor pressure of the cell membrane affect the texture (Harker et al., 1993; Wang et al., 2018; Zdunek et al., 2014). However, limited research suggests that void formation in tissues affects vegetable texture.
Given that fruits and vegetables, such as pea sprouts, form various void networks in their tissues for gas and mass transfer (Yamauchi et al., 2015), it is highly probable that changes in these void networks contribute to texture deterioration after regrowth. With the widespread adoption of X-ray micro-CT in recent years, the analysis of the internal structure of food products has advanced significantly. Nugraha et al. (2019) created 3-D porosity maps of horticultural products, such as eggplants, turnips, apples, and pears, using X-ray CT. We assumed that this technology could be employed to analyze the numerous voids distributed within the tissue of pea sprouts more precisely. In this study, X-ray CT images were used to characterize the voids in the tissues of regrown pea sprouts to elucidate the influence of intra-tissue void distribution on texture formation in pea sprouts.
Metabolome analysis is a practical approach for investigating the causes of changes in the quality of fruits and vegetables and has been used frequently in recent years. For example, Kuroda et al. (2024) clarified the metabolic changes that occur when strawberries are grown in a potassium-deficient environment through analysis based on primary metabolites. Since regenerative cultivation is usually carried out in environments without enough nutrition, it was assumed that the metabolism important for tissue formation could be inhibited. In this study, we attempted to elucidate the causes of changes in the quality of fruits and vegetables by presenting the results of metabolome analysis.
Green pea sprouts (Pisum sativum L.) purchased from a local market in Gifu city, Japan were utilized for the experiment. The sprouts were trimmed at the top of the leaves near the roots and separated into stems and roots. The roots were placed in a plastic container and supplied with distilled water, sufficient to submerge them. Regeneration was performed in an incubator at 25 °C with a 12-h light/dark cycle over a period of 5 days. The distilled water was changed daily during cultivation. The stems were cut 3 cm from the surface and used as pea sprout samples before and after regrowth.
2.2. Texture analysisTexture analysis was conducted based on the method described in a previous study (Yoshioka et al., 2009) with slight modifications. A texture analyzer (TA-XT plus, Stable Micro Systems Ltd., UK) equipped with a wedge-shaped plunger was used for the measurements. The stem of the sample was placed on a stage perpendicular to the long axis of the wedge. The plunger was lowered at a rate of 1 mm/s until 70 % of the initial sample thickness was reached. During compression, load changes were recorded at intervals of 0.005 s. To determine the load fluctuation and crispness of the sprout, the fluctuation index (Dt) (N) and crispness index (CI) (N) were defined using Eqs. (1) and (2).
| (1) |
| (2) |
Where Ft, Ft−1, and Ft+1 represent the load (N) at times t s, t − 0.005 s, and t + 0.005 s, respectively. The Dt value was plotted against time to capture the frequency of variation in the measured mechanical properties of the object. The variation in Dt over time is shown in Fig. 1. In this study, peak detection was performed using the Signal Processing Toolbox in Matlab R2023a (MathWorks Inc., USA), and the number of peaks was counted.
To capture changes in tissue structure, such as porosity, the samples were observed using X-ray micro-CT (Skyscan 1172, Bruker Corp., USA). To prevent drying during imaging, the samples were inserted into polyethylene containers with a length of 15 mm and a diameter of 4 mm. The polyethylene containers were secured at the stage using synthetic rubber. The imaging conditions were set as follows: a tube voltage of 49 kV, tube current of 148 μA, and voxel size of 3.75 μm. The obtained images were reconstructed using NRecon software (Bruker Corp.) to obtain cross-sectional images.
The obtained cross-sectional images were analyzed using Avizo 9.4 (Thermo Fisher Scientific Inc., USA). Initially, the correction was performed using the Median Filter, and then high-intensity regions (GS > 96) representing tissue cells and low-intensity regions (GS ≦ 95) representing air were extracted via thresholding. The high-intensity regions revealed a cylindrical object with pores. The porosity, average object thickness, and average space thickness were calculated using the Avizo Analysis Module. Furthermore, for low-intensity regions, the borderkill module was applied to extract only the pores inside the pea sprout tissue. These objects were labeled, and the distribution of the pores was visualized. The shape parameters for individual pores were calculated using the Label Analysis module, with a focus on obtaining the volume (V) and Shape_VA3D. Here, Shape_VA3D is defined as Eq. (3).
| (3) |
Where A indicates the boundary area of each void (μm2). Shape_VA3D represents a perfectly spherical shape when it is equal to 1 and indicates larger values based on the complexity of the shape.
2.4. Metabolome analysisThe pea bean sprouts were placed in a 50 mL centrifuge tube and frozen with liquid nitrogen. The frozen samples were then freeze-dried for three days using a vacuum freeze-dryer (FDU-1200, EYELA, Japan). Subsequently, the pea bean sprout samples were powdered using a bead crusher. The powder (10 μg) was obtained in a 1.5 mL tube, and 10 μL of methyl methanesulfonate (internal standard at 1 mg/mL) was added. Next, 1 mL of a mixture of water:methanol:chloroform in a ratio of 1:2.5:1 was added to the tubes to denature the proteins. The mixture was thoroughly mixed using a bead crusher (Shake Master Neo, Bio Medical Science Inc., Japan) and then centrifuged at 4 °C, 16,000 G for 3 min (Cooled Centrifuge 1720, KUBOTA Corp., Japan). The supernatant (900 μL) was collected in a 1.5 mL tube. Milli-Q water (Merck KGaA, Germany) was added to the tube containing the supernatant to denature any remaining protein. This solution was mixed well using a vortex mixer and then centrifuged again at 4 °C, 16,000 G for 3 min. After centrifugation, 350 μL of the supernatant was collected in a tube. This was placed in a centrifugal evaporator (CVE-2100, EYELA) to evaporate the methanol in the solution. The remaining solution was frozen and dried using a freeze-dryer. To the freeze-dried sample, 100 μL of a 20 mg/mL methoxyamine-pyridine solution was added, and the cap was wrapped with Parafilm. The mixture was processed for 20 min using an ultrasonic agitator until the residue was dispersed. The heating shaker was set to a temperature of 30 °C, and the mixture was processed for 90 min at 1,200 rpm. After processing, 50 μL of MSTFA was added to the tube, the shaker temperature was set to 37 °C, and the tube was shaken at 1,200 rpm for 30 min. Subsequently, the mixture was centrifuged at 16,000 G for 3 min, and 100 μL of the supernatant was collected in a GC-MS vial as the analytical sample.
Metabolites in the samples were detected using a GCMS-TQ8040NX (Shimadzu Corp., Japan) and analyzed according to the Smart Metabolites Database (Shimadzu Corp.). The injector temperature was set to 280 °C. Subsequently, the GC oven was maintained at 100 °C for 20 min and ramped up to 320 °C at 10 °C/min. This temperature was again maintained for 11 min. Helium was used as the carrier gas, with a GC injection volume of 1 μL, a DB-5 column (Agilent Technologies Japan, Ltd., Japan) (30 m of length, 0.25 mm of inner diameter, 1 μm of film thickness), and a column flow rate of 1.10 mL/min in split mode (split ratio 1:25). The interface temperature was set at 280 °C, and the ion source temperature was set at 200 °C. All experiments were conducted in triplicates.
2.5. Statistical analysisDifferences in each parameter before and after regenerative cultivation were evaluated using t-tests. In addition, Spearman's rank correlation coefficient was obtained to evaluate the relationship between texture parameters and structural properties. Analyses were performed using R i386 3.5.1 (R Development Core Team, 2018).
The data obtained from the GC-MS/MS analysis were normalized using sample weights and internal standards. Principal component analysis (PCA) was performed using MetaboAnalyst 5.0 (Xia Lab, 2021) to identify the metabolites contributing to the separation before and after regenerative cultivation. Principal component loadings were calculated for each metabolite, forming the basis for the selection of important metabolites.
Figure 2 (a) shows the CI values of the pea sprouts before and after regrowth. The CI values of pea sprouts after regrowth decreased significantly. Figure 2 (b) shows the peak frequencies of the samples before and after regrowth. The number of peaks per unit distance increased in the regrown samples. The CI value can express crispiness and crunchiness, which are considered difficult to determine using the maximum load (Masumoto et al., 2018; Yoshioka et al., 2009). The CI value was used to examine the texture of cucumbers (Horie et al., 2004) and persimmons (Masumoto et al., 2018). These parameters were calculated based on values corresponding to the second derivative of the time variation of the load during the breaking test, as expressed in Eq. (1). In homogeneous materials, physical property would not be fluctuated, and the load change during the breaking test increases monotonically, whereas, in heterogeneous structures, the load change is expected to fluctuate according to the physical property distribution. In this study, fluctuations in the physical properties owing to the cellular structure and voids were assumed. Interestingly, in the regrown sprouts, fluctuations were frequent, yet the CI values were not substantial. One factor was the narrow range of variation, which suggests reduced tissue robustness after regrowth cultivation.
Figures 3 (a) and (b) show the reconstructed X-ray micro-CT images of the samples before and after regrowth. Before regrowth, the entire stem of the sprout was densely packed with cells. However, after regrowth, coarse cavities appeared at the center. Figures 4 (a) and (c) show that the porosity and average space thickness increased significantly after regrowth. However, the average object thickness was not significantly different from that shown in Fig. 4 (b). The regrown sprouts had a greater proportion of voids in the overall structure, suggesting that the voids may have been larger than before regrowth. Changes in these parameters and the occurrence of a large central cavity were attributed to the lack of the necessary nutrients. Because only distilled water was supplied to the peas during regrowth, the availability of the necessary nutrients was limited to what was present in the peas. Therefore, we assumed that if pea sprouts were grown repeatedly from the same pea, the nutrients obtained from the peas would be reduced, inducing changes in the tissue structure. Insufficient photosynthesis is also considered as a possible cause. Photosynthesis plays an important role in plant growth and survival. Plant stress regulates photosynthesis and reduces yield (Chauhan et al., 2023). The stress to which plants are subjected is regarded as a threat to agriculture worldwide, and research on the stress response of plants has been actively conducted (Panahirad et al., 2023; Sharma et al., 2019). During regrowth, pea sprouts are grown by feeding distilled water to roots cut above the leaves growing near the roots of the pea sprouts. Therefore, seedlings grow with few leaves, which are mainly used for photosynthesis. Hence, photosynthesis is insufficient and, consequently, the synthesis of carbon sources is insufficient. Therefore, we hypothesized that the lack of nitrogen, phosphate, potassium, or carbon, which are essential for plant growth, may have led to a decrease in volume and a change in tissue structure with large voids.
To obtain a more detailed understanding of the changes in void characteristics owing to regrowth, individual void parameters were determined in addition to the overall parameters, as described above. Figures 3 (c) and (d) show the three-dimensional distribution of voids in the samples visualized using Avizo 9.4. (Thermo Fisher Scientific Inc.). We successfully determined the morphological parameters of the individual voids using label analysis. The number of voids extracted was 23,472 ± 628 before regrowth and 14,308 ± 2,566 after regrowth in the scanned region of the stem. Figure 5 shows the histograms indicating the volume distribution of the individual voids. In Fig. 5 (a), a histogram based on the number of voids reveals the presence of numerous voids in the range of 102 to 104 μm3. The total number of voids was higher before regrowth cultivation, and the number of voids in each size fraction followed a similar trend. Figure 5 (b) shows the proportion of each void volume fraction to the total void volume. In this figure, voids in the range of 105 to 107 μm3 constituted the majority. Notably, after regrowth cultivation, voids in the range of 106 to 107 μm3 represented the largest proportion, revealing a distinct trend compared to that observed before regrowth. This tendency probably contributed to the increase in the average space thickness. These findings suggest significant alterations in void structure, with the post-regrowth scenario displaying a notable shift in the distribution of void volumes. The observed changes in the number and proportion of voids provided valuable insights into the impact of regrowth cultivation on the internal structure of the studied material.
Values are expressed as mean ± S.E. (n = 9).
Figure 6 shows the relationship between the volume of voids and Shape_VA3D before and after regrowth. Shape_VA3D was used for the shape analysis of the mortar pores (Niu et al., 2022) and sand grains (Markwitz et al., 2020). Both before and after regrowth, the majority of voids developed complexity with increasing volume. Ando et al. (2021) visualized the voids in mung bean sprouts using CT and analyzed their shape characteristics, which indicated that the complexity increased with increasing volume, as in the present study. Therefore, it was confirmed that voids in the stems of sprouted vegetables developed according to a certain tendency. However, in the region from 1 to 10 of Shape_VA3D, a void group demonstrating a different shape behavior from that of the main group emerged. This effect became more pronounced after regrowth, reaching a volume of 107 μm3. Therefore, regrowth promotes the formation of irregular voids with relatively simple shapes. This is attributed to the large voids at the center of the sprouts, as shown in Fig. 3.
Table 1 shows the correlation coefficients between the void characteristics and crispness factors. Porosity is the most common parameter that indicates the void characteristics of food. Therefore, many researchers have used porosity to evaluate the quality of food, including vegetables (Imaizumi et al., 2017; Li et al., 2024; Nugraha et al., 2019). In addition, Li et al. (2024) showed a correlation between the crispness and porosity of the dried vegetables. However, in the present study, crispness factors were not correlated with porosity. As for the average object thickness, small changes attributed to regrowth showed that this factor did not seem to contribute to texture degradation. A negative correlation (R = −0.837) was found between pore thickness and the CI value, and a positive correlation (R = 0.756) was found between pore thickness and peak frequency. From these results, it can be concluded that crispness was enhanced not simply by the amount of air present but also by the dense pore dispersion, which increased the fluctuation of the physical properties. In terms of the correlation between the void size fractions, V1–V4 showed relatively strong correlations with these parameters. Therefore, void size from 102 to 105 μm3 was considered important for the development of crispiness. In contrast, V6 exhibited an extremely low correlation. This region corresponded to the abnormally formed voids mentioned above, and these coarse voids did not contribute to crispiness.
| CI value | Peak frequency | ||||
|---|---|---|---|---|---|
| Porosity | −0.329 | 0.404 | |||
| Average object thickness | 0.088 | −0.092 | |||
| Average space thickness | −0.837 | *** | 0.756 | *** | |
| Number of voids at volume fractions | V1 | 0.767 | *** | −0.746 | *** |
| V2 | 0.800 | *** | −0.692 | ** | |
| V3 | 0.748 | *** | −0.672 | ** | |
| V4 | 0.752 | *** | −0.639 | ** | |
| V5 | 0.690 | ** | −0.529 | * | |
| V6 | −0.064 | 0.201 | |||
V1 ≤ 102, 102 < V2 ≤ 103, 103 < V3 ≤ 104, 104 < V4 ≤ 105, 105 < V5 ≤ 106, 106 < V6 ≤ 107 (μm3)
To explore the causes of the changes in the void characteristics induced by regrowth, primary metabolites were analyzed by GC-MS/MS, and 122 components were successfully identified (Table S1). Figure 7 (a) shows an overview of these substances as a volcano plot, with metabolites for which p < 0.05 and fold change FC > 1.5 are colored. Among these, 45 and 28 metabolites showed downregulated and upregulated expression, respectively. The most notable substance here was mannitol, which had a fold change of −153.93. The presence of mannitol has been confirmed in Fabaceae (bean, pea), and it is assumed to act as a photoassimilate and phloem-translocated carbohydrate (Bourne, 1958; Patel et al., 2016; Stoop et al., 1996). It is not difficult to imagine that the sprouts that lost their leaves after the initial use had severely reduced photosynthetic activity, whereas mannitol was metabolized in the sink. KEGG pathway analysis showed that these metabolites were mainly associated with carbohydrate metabolic pathways (Fig. 7 (b)).
The GC-MS/MS results were subjected to PCA and visualized as score plots in Fig. 7 (c). The first two principal components (64.5 % and 10.2 % of PC1 and PC2, respectively) accounted for accumulative variance contributions of 74.7 %, representing a substantial portion of the information. PCA separated the regrown samples from the original samples, confirming the veracity of the results of metabolome analysis. Differences before and after regrowth were identified, particularly for PC1. Substances that exhibited positive PC loadings were predominantly nucleotide metabolites such as pyrimidines and purines. In plants, these substance fluctuations may appear as a stress response (Chen et al., 2023; Stewart et al., 1971; Yang et al., 2024). In contrast, the negative PC loadings for PC1 were primarily associated with carbohydrate and amino acid metabolism. For instance, mannitol, mannose, and sucrose are involved in the photosynthesis and transportation of carbohydrates (Patel et al., 2016). The decrease in these substances in the regrown sprouts probably occurred because the leaves were cut off, and sufficient photosynthesis was inhibited, as described above. In addition, advanced consumption of glycolytic and TCA circuits due to aerobic respiration was suggested by changes in glucose-6-phosphate, amino acids, and other substances. Consequently, the carbon source required for the construction of the cell wall tissue was considered insufficient.
Pea sprouts are widely consumed in Asian countries, particularly Japan, where the common practice of regrowing them for reusing inedible parts is prevalent. However, the potential loss of the characteristic crisp texture when subjected to regrowth is concerning. This study aimed to explore the causes of this texture deterioration by employing X-ray μCT for an in-depth analysis of internal structures, focusing specifically on the void characteristics. The successful visualization of voids and the acquisition of characteristic parameters shed light on the previously unexplored realm of void structures. Correlation analyses were conducted to reveal the associations between crisp texture and void characteristics. The importance of void distribution in vegetable texture formation is evident, suggesting its potential as a quality indicator for vegetables. Additionally, metabolome analysis provided insights into the metabolic pathways that could influence void analysis. The significance of voids and the applied analytical techniques, along with the findings of this study, will contribute to the advancement of high-level vegetable utilization.
This work was supported by JSPS KAKENHI Grant Number 23K14045 and 21H04748.
The authors declare no conflicts of interest.
(URLs on references were accessed on 26 January 2026.)
