In this study, wild-type AB strain zebrafish were bred in natural pairs. Zebrafish aged four days post-fertilization (dpf) were chosen for antianxiety activities and memory-improving [responsiveness and acetylcholinesterase (AChE) production] evaluation. Zebrafish aged one dpf were used to evaluate the samples’ protective abilities to the central nervous system (CNS). Translucence melanin allele of Albino-type zebrafish bred in natural pairs aged three dpf were used to evaluate the antioxidant efficacy.

All zebrafish were raised in 28°C fish farming water (water quality: 200 mg of instant sea salt per 1 L reverse osmosis water, conductivity: 450–550 μS/cm; pH: 6.5–8.5; hardness: 50–100 mg/L CaCO₃) as culture, license number of experimental animals used: SYXK (Zhejiang) 2022-0004. The feeding practice was carried out following the requirements of the international AAALAC certification (certification number: 001458).

The anxiety zebrafish model was established by adding 1-(3-chlorophenyl) piperazine hydrochloride (mCPP, 20.0 μg/mL, 1 h incubation at 28°C and washed off) (10, 11). After that, in the positive drug group, Selegiline was added (20.0 μg/ml, 24 h incubation at 28°C). The OS model was established by adding menadione (2.25 μM, 22 h incubation at 28°C) to the culture. In the positive drug group, N-Acetyl-L-cysteine (NAC) was also added (62.5 μg/ml, 22 h incubation at 28°C). The zebrafish NSI model was established by adding Mycophenolate mofetil (0.4 μM, 48 h incubation at 28°C) to the culture. In the positive drug group, Glutathione was also added (615 μg/ml, 48 h incubation at 28°C). The zebrafish AD model was established by adding aluminum chloride (140 μM, 24 h incubation at 28°C) to the culture. Donepezil hydrochloride was also added to the positive drug group (3.33 μg/ml, 24 h incubation at 28°C). The nutrients in each test were incubated under the same condition as their respective positive drug group. The control (blank control, culture-only) was set up simultaneously with the other groups.

Four positive drugs were used in this experiment: Selegiline, NAC, Glutathione, and Donepezil hydrochloride are commercially available synthetic chemicals. The nutrients, including Stachyose, Lactic acid bacteria-Lactobacillus Plantarum (PS128), Walnut Peptide (WP), Desert Cistanche (DC), and Theanine (Th) were all plant extracts. Casein Peptides (CP) were originated from milk proteins. All the nutrients are commercially available. Their combinations were nutrient samples with determined concentrations. The names and concentrations of the positive drugs, individual nutrients and combinations tested were summarized in Table 1. Their doses in zebrafish and the estimated doses in humans were also listed.

In this test, four dpf wild-type AB zebrafish were randomly selected and raised in the six-well plates with 30 fish per well (for all groups). After treatments, all zebrafish were transferred into a 96-well plate (one tail/well) and 200 μl of culture solution was added to each well. A behavioral analyzer was then used to detect the total moving distance of zebrafish for further analysis.

In this test, the wild-type AB strain zebrafish of four dpf were randomly selected and raised in the six-well plate, with 30 zebrafish per well (for all groups). After treatments, the cortisol concentration of zebrafish (three replicates per treatment) was measured using the enzyme-linked immunosorbent assay (ELISA kit, Supplementary Table 1: Instruments and consumables). The data were then collected and analyzed.

The translucent albino zebrafish with three dpf melanin allele mutations were randomly selected and raised in six-well plates. Thirty zebrafish were kept in each well (for all groups). After treatments, zebrafish were transferred into 24-well plates and stained with specific reactive oxygen species (ROS) fluorescent dye. After staining, 10 zebrafish were randomly selected from each test group and placed under a fluorescence microscope for imaging. Nis-elements D 3.20 advanced image processing software was used for analysis. The fluorescence intensity at the zebrafish yolk sac was measured for further analysis.

Zebrafish of the wild-type AB strain of one dpf were randomly selected and raised in six-well plates with 30 zebrafish per well (for all groups). After treatments, zebrafish were stained with acridine orange (AO, Supplementary Table 1: Instruments and consumables). After staining, 10 zebrafish were randomly selected from each group and placed under a fluorescence microscope for imaging. Images were processed by Nis-elements D 3.20 advanced image processing software. After that, the fluorescence intensity in the zebrafish CNS was measured.

Statistical analysis was conducted using SPSS v.20 software (IBM Ltd., UK). Data following normal distributions were represented as means ± standard errors of the mean (SEM). Differences among groups were analyzed by one-way ANOVA, followed by the post hoc Tukey Dunnett’s multiple comparison test. In the non-parametric test, the Kruskal–Wallis single-factor ANOVA test (K samples) was used. Comparative analysis of the two groups: independent t-tests and non-parametric tests were used for normal distributions. Mann–Whitney tests were used for non-normal distributions. The correlations between each effect were analyzed statistically by Pearson’s correlation. The area under curve (AUC) of velocity in 20 min was used for correlations between reactivity and other effects. When putting all the indexes together in one graph, we normalized every index to control to compare the full impact of each combination. P values less than 0.05 were considered significant in all analyses. Data were collected from at least three biological replicates.

We used mCPP to increase zebrafish’s anxiety-like behaviors and stimulate the production of cortisol (10, 12). The zebrafish’s movement (representative trace maps in Figure 1A and Supplementary Figure 1A) reflected the zebrafish’s anxiety level (13). From the results in Figure 1B and Supplementary 1B, the distance in the model group was significantly longer than the control (P < 0.001), indicating that the zebrafish anxiety model was effectively established. As expected, zebrafish exhibited a shorter distance in the Selegiline (an irreversible monoamine oxidase (MAO)-B inhibitor that increases 5-HT levels) (14) group than in the model or the control. Based on the preliminary screening (Supplementary Figure 1B), we selected several nutrients and their combinations for further tests.

Figure 1B revealed that all the individual nutrients and combinations exhibited shorter distances than the model group, indicating their antianxiety effects. Among them, Stachyose at 500 μg/mL was the most effective nutrient, with comparable activity to the positive drug. For the combinations, firstly, we noticed that when the ratio of WP + DC was adjusted based on weight (1:1, 1:3, and 3:1), their effects varied and, at 93.7 + 31.2 μg/ml (this concentration was selected for downstream tests) showed the highest capability which is comparable to the positive drug. Secondly, the effects of the combinations were dose-specific but not dose-dependent in all groups. For instance, in Stachyose + PS128 group and WP + Th group, the increase of both components’ concentrations did not promote their antianxiety capabilities. Thus, in Stachyose + PS128 and WP + Th groups we selected the lowest concentrations (83.3 + 500 and 14.2 + 33.3 μg/ml, Stachyose + PS128 is more effective, P = 0.003, Figure 1B) for downstream tests. Nevertheless, the concentration increases in the combination of WP + CP promoted the antianxiety effect, showing a preliminary dose dependence.

The production of cortisol in each group was also measured to show the zebrafish’s anxiety level (15). As expected, most tested nutrients and combinations showed lower cortisol levels than the model group, with Stachyose + PS128 (83.3 + 500 μg/ml) being the most effective combination (Figure 1C). Surprisingly, specific concentrations of WP + DC and WP + CP combinations did not inhibit the production of cortisol, which was not consistent with the zebrafish’s moving distance. Despite this, a positive correlation (r = 0.719, P < 0.001) was still observed between the distance and the cortisol level (Figure 1D), showing that the antianxiety effects assessments were reliable.

It was reported that biological processes such as oxidative stress are partly affected by mood states (16). Therefore, we continued to investigate the antioxidant effects of the nutrients. Menadione was used to induce oxidative stress, which triggered green fluorescence expression in the yolk sac of zebrafish (Figure 2A). The increase in fluorescence intensity represents a decrease in antioxidant function (17). The intensity (Figure 2B) was observed to be much higher in the model than in the control, indicating that the model establishment was successful. All the combinations showed antioxidant effects. Although none of them could recover the model to the level of the control, Stachyose + PS128 and WP + Th combinations showed similar effects as the positive control drug NAC, a glutathione precursor, which modulates glutamatergic transmission and targets oxidative pathways (18). Their antioxidant effects were also affected by concentrations: in WP + CP, the higher concentration at 56.7 + 100 μg/ml was more effective than the lower concentration, indicating a preliminary dose-dependence (Figure 2B).

It was also reported that there are links between anxiety, oxidative stress and neuron damage (19, 20). Thus, we continued to investigate the central neuroprotective effect of nutrients. Mycophenolate mofetil was used to induce NSI in zebrafish, which triggered green fluorescence (Figure 3A) in apoptotic cells at CNS (21). Firstly, in the model group, the fluorescence intensity (Figure 3B) was almost three times higher than the control, showing the model establishment was successful. Secondly, the combination of WP + Th showed the strongest activity in reducing central neural apoptosis, close to the positive drug Glutathione, whose depletion is a common feature of apoptotic cell death (22). Unlike the antianxiety tests, the other combinations (WP + DC, WP + CP, and Stachyose + PS128) exhibited a similar but much lower protective effect than WP + Th. A preliminary dose-dependent effect was observed in WP + CP and further tests with more concentrations are needed. Noteworthily, when we plotted the antioxidant effect of the nutrients against their central neural apoptosis level, a good positive correlation was established (r = 0.897, P = 0.003, Figure 3C), showing that both effects from the nutrients were highly related.

Apart from the above tests, we further investigated the memory-improving abilities of the nutrients, as anxiety was reported to cause memory damage (23). During model establishment, aluminum chloride was used to build zebrafish’s AD model: Excessive intake of aluminum chloride causes toxic effects on the nervous system, resulting in abnormal zebrafish behavior (24). It also destroys the cholinergic nerve function, increases the production of amyloid, reduces learning ability and causes memory loss (24, 25). Donepezil hydrochloride, an AChE inhibitor, was used as the positive drug in this test. The nutrients’ memory-improving abilities were evaluated by measuring the velocity of zebrafish using the response capability test (26). From the graph shown in Figure 4A, the mean velocity of the model group was the lowest of all, demonstrating the effectiveness of the modeling. The results showed that the nutrients improved the mean velocity, which was slower and less stable under light (0–5 and 10–15 min) than in the dark (5–10 and 15–20 min). WP + CP significantly improved zebrafish’s response capability, with a preliminary concentration-dependent effect. It is noteworthy that WP + Th showed the most substantial capability in improving velocity, even recovered to a higher level than the control.

Furthermore, zebrafish with nervous system damage were reported to have slower swimming speed; But after treatment, their swimming speed was recovered with the stimulation of light-dark alternation (27). Consistent with this report, the difference in velocity between light and dark was much higher in nutrient combinations (e.g., WP + Th) than in the model (Figure 4B). The mean velocity was further broken down into cycles of 5 min in Figure 4C. The patterns of the results in each period were similar, further proved that the effects of nutrients in the shorter period are consistent. Another crucial memory-improving index was the production of AChE (28), the intensity of which (Figure 4D) confirmed the effectiveness of all the combinations, with WP + CP at a higher concentration being the most effective. When we plotted the AChE intensity against the AUC of the mean velocity of all groups, a good negative correlation was observed (r = −0.836, P = 0.009, Figure 4E), indicating the memory-improving assessments were reliable.

With all these data in hand, we further analyzed the relationships between different effects of the nutrient combinations. The results of all effects were consolidated in Figure 5A, from which we observed that the pattern of the effects from test groups are similar and most of the nutrients improved their respective models. Among them, WP + Th was almost the most promising in all tests. When the value of individual effect was plotted against each other, good correlations (17 out of 21 have absolute r value larger than 0.5, representative plots were shown in Supplementary Figure 2) were observed (Figure 5B), suggesting the effects of the nutrient combinations were universal and highly related.