Optimization of extraction conditions and determination of purine content in marine fish during boiling

PCA method

The results showed that acid concentration, hydrolysis temperature, hydrolysis time, and the liquid-solid ratio had influences on the extraction percent of the purines. As shown in Fig. 2, the extraction percent of the purine decreased gradually with increasing PCA concentration, eventually beginning to increase once the PCA concentration reached 60%. When the concentration of PCA was 80%, the extraction percent was higher. The extraction percent also reached a maximum at a hydrolysis temperature of 80 °C, beyond this point it decreased slightly. This may be due to the destruction of some purine bases at high temperatures (Jamil, Halim & Sarbon, 2016). In addition, the extraction with the PCA method was the best with a hydrolysis time of 55 min and a liquid-solid ratio of 60:1. Finally, level ranges of acid concentrations (70–90%), temperatures (70–90 °C), hydrolysis time (50–60 min) and liquid-solid ratio (50:1–70:1 mL/g) were selected for BBD.

Mixed-acid method

The effects of hydrolysis temperature, the liquid-solid ratio, FA concentration, TFA concentration, and hydrolysis time on the extraction percent of the four purine bases are shown in Fig. 3. An increase in hydrolysis temperature improved the extraction efficiency gradually, until a maximum extraction percent was reached at about 90 °C. The extraction percent of the purine increased firstly and then decreased with the increase of TFA concentration, with the extraction percent reached the maximum at 85%. The extraction percent of the purines fluctuated with increasing liquid-solid ratio, and the highest extraction percent was obtained when the sample was hydrolyzed for 10 min. Moreover, the changes of FA concentration had little effect on the extraction percent of purine. Finally, according to the results of single factors design, the selected low and high values for hydrolysis temperature, liquid-solid ratio, FA concentration, TFA concentration, and hydrolysis time were (80, 100 °C), (20:1, 40:1 mL/g), (75%, 85%), (80%, 90%), and (5, 15 min) in the PBD, respectively.

Plackett–Burman screening studies

The workload of a five factors response surface design was large, so its variables need to be screened. PBD was an effective method for screening the best candidate factors (Vasiee et al., 2016). Based on the result of single-factor design, a two-level Plackett–Burman design of 12 runs were inducted in this paper to confirm the important factors that significantly affect the extraction percent of mixed-acid. The standard ANOVA was calculated from experimental runs (Table 3). The analysis indicated A-hydrolysis temperature, D-TFA concentration and E-hydrolysis time had significant influence on extraction percent of purine. F-value of the model was 18.76 which indicated the model was significant. The value of p < 0.05 implied that the model terms were significant. Moreover, the predicted R² (0.7596) was in reasonable agreement with the adjusted R² (0.8898) as the different was <0.2. “Adeq precision” was used as a tool to measures the signal to noise ratio, that greater than four was desirable. In this study, a ratio of 13.549 implied an adequate signal and this model could be used to navigate the design space. As shown in Table 3, the contribution of hydrolysis time (38.76%) was the highest, which was followed by hydrolysis temperature (36.42%), TFA concentration (15.61%), FA concentration (3.04%), and liquid to solid ratio (0.16%). Moreover, as showed in Fig. 4, the variables of E, A, and D presented a positive effect on the total purine extract. By analyzing the data obtained from PBD, the insignificant variable was ignored and the variables of extraction temperature, time and the concentration of TFA were selected for further study by BBD to attain the optimal extraction process.

PCA method

There were a total of 29 runs for optimizing the four independent parameters in the current BBD. Table S1 shows the experimental design used for the study. The data evaluated by ANOVA were presented in Table 4. These data were fitted to nonlinear quadratic models for percentage yield of total purine content. The fitted model equation is: $$\begin{array}{l}\text{Total\hspace{0.17em}purine\hspace{0.17em}yield}\left(\%\right)=85.85+0.47\text{A}+0.063\text{B}+0.27\text{C}-0.34\text{D}-0.085\text{AB}-0.52\text{AC}+\\ 0.36\text{AD}+0.015\text{BC}-0.09\text{BD}+0.42\text{CD}-2.71{\text{A}}^2-0.67{\text{B}}^2-1.01{\text{C}}^2+\\ 0.001583{\text{D}}^2\end{array}$$

The Model F-value of 18.89 showed that this model was significant at less than 0.0001. The lack of fit F-value of 1.98 and the p-value of 0.2661 implied it was unimportant due to relative pure error. The not significant of value of lack of fit indicated the model was fitted with good prediction (Singh, Bajar & Bishnoi, 2017). In addition to, the value of the determination coefficient (R²), adjusted coefficient (Adj R²), and coefficient of variation were 0.9497%, 0.8994%, and 0.56%, respectively. The R² value of the quadratic regression model showed that only 0.0503 of the total variations were not explained by the model. Adjust R² of 0.8994, which also confirmed that the model was highly significant. Besides, the low value of the coefficient of variation clearly indicated a better precision and reliability in conducted experiments. The ANOVA indicated that the independent variables studied hydrolysis time, liquid-solid ratio, PCA concentration were significant factors (p-value <0.05) (Karacabey & Mazza, 2010), and the quadratic terms (AC-PCA concentration and hydrolysis time) significantly affected the total purine yield.

Contour plots were constructed to study interaction effects of the factors on the responses (Yin & Dang, 2008). Contour plots figures were drawn for the pair-wise combination of the four variables while keeping other factors at its medium value. Different shapes of contour plots indicated different type of interactions between the factors. The elliptical shape showed that interactions between corresponding factors were significant, although the reciprocal interactions between the corresponding factors are insignificant as represented by a circular contour plot (Kamble et al., 2018). The quadratic equation and ANOVA showed that AC (PCA concentration and hydrolysis time) have strength interaction. The interaction between PCA concentration and hydrolysis temperature was represented by contour plot shown in Fig. 5. As excepted, the contour plot was elliptical, which indicated the interaction among them was significant. Fig. 5 showed that PCA concentration exhibited a significant effect whereas hydrolysis time. An increase in total purine yield resulted when the concentration of PCA (A) was added in the range from 70 to 82 °C. Likewise, as the time of hydrolysis (C) increased in the range from 50 to 56, total purine extraction percent increased. Finally, the optimum conditional for PCA purine extraction method given by the software of Design-Expert 8.0.6 was acid concentration 80.97%; temperature 79.78 °C; hydrolysis time 54.20 min; liquid-solid ratio 50.78:1 and the extraction percent of total purine was 86.15. Then, according to the actual operating conditions, the optimum predicted process parameters were adjusted to acid concentration 80%; temperature 80 °C; hydrolysis time 55 min; liquid-solid ratio of 50:1 and the next verification test was carried out.

Mix-acid method

As shown in Table 3, three most important variables affecting extraction efficiency of total purine were hydrolysis time, the temperature of hydrolysis and TFA concentration, according to the screening by Plackett–Burman design. These three factors were further optimized by BBD. The equations below show the relationship of the three factors and extraction yield.

The experimental design and ANOVA result of the model was shown in Table S2 and Table 5. The value of p less than 0.001 and non-significant value of lack of fit (F = 0.59, p-value = 0.6523) indicated the model is fitted with good prediction. The model F-value of 40.10 indicated the model was significant and only a 0.01% chance that “Model F-value” this large could occur due to noise. Moreover, the model has a high value of R² (0.9810), Adj R² (0.9565), and Pre R² (0.8858), indicating a good agreement between the experimental results and predicted values of the response. Furthermore, in this paper “Adeq precision” of 17.524 > 4, which implied this model can be used to navigate the design space. At the same time, the p-values of A, B, C, AB, AC, BC, A², B², and C² were all lower than 0.05 demonstrating that they were significant variables.

The mutual effect of TFA concentration and hydrolysis temperature was represented in Fig. 6A. The counter plots show elliptical shaped indicating the significant interaction. Moreover, in a certain range, the extraction percent of total purine increased with TFA concentration and hydrolysis temperature. Increased TFA concentration and hydrolysis temperature led to the increase in total purine yield, but slightly decreased when exceeded the threshold levels of 89.00% and 92.00 °C. The interactions between hydrolysis time and temperature was significant because of elliptical shape of contour plot (Fig. 6B). An increase in the yield of total purine could be achieved with the controlled hydrolysis time was 8–13 min. Similar interactive effects were observed when TFA concentration and hydrolysis time were considered (Fig. 6C). It was obvious that the extraction percent of total purine was the highest when the hydrolysis time and TFA concentration were 10.20 min and 87.00%, respectively. Finally, BBD experiment predicted that the highest extraction percent was 104.51%, and the extraction conditions were as follows: mixture acid 88.00% TFA/80.00% FA (v/v, 1:1); temperature 91.83 °C; hydrolysis time 9.43 min; liquid-to-solid ratio 30:1. With some modifications, TFA concentration of 90.00%, temperature of 90.00 °C, hydrolysis time 10.00 min were used to verification test.

Determination of the optimal method

In order to determine the best extraction method, the extraction yields of the two methods were compared; the results were given in Table 6. The actual extraction yield of total purine were 83.75% ± 2.65% (PCA method) and 103.94% ± 3.85% (mix-acid method), respectively. The relative error with the predicted value were less than 5%. The experimental results were well agreement with the predicted results. The result indicated that the model developed was considered to be accurate and reliable. In addition, the extractions of the total purine by the mixed-acid method were higher than those by the PCA method. These differences were likely due to protective effects of FA on purine bases. Moreover, TFA could promote complete dissolution of purine bases. In addition, the mixed-acid treatment was time saving and could effectively avoid the production of toxic chlorine. Therefore, the mixed-acid method was chosen for the extraction of purine bases from different marine fish.

Method validation

The linearity, range, and LOD of the method were subsequently calculated; results are shown in Table 7. Excellent linearity was observed for the quantification of the four purine bases. The LODs ranged from 0.0118 to 0.0774 mg/L, which were considered excellent. Furthermore, the R² values were all above 0.9999, indicating a good linearity.

Standards were used to evaluate the precision of the method. For the repeatability and recovery tests, Scophthalmus maximus (dorsal muscles) samples were prepared by the method described previously. As shown in Table 8, the deviation in results were less than 0.69%, which indicated that the HPLC method was highly reproducible. The mean repeatability of the method for quantification of adenine, guanine, hypoxanthine, and xanthine were 1.32%, 0.87%, 0.83%, and 1.77%, respectively. By this method, average recoveries ranged from 94.90% to 104.51%. These results suggest that the mixed-acid method avoided damaging the purine bases, and confirm the validity of the method.

Purine contents in edible parts of marine fish

The contents of the four purines in different edible parts of marine fish were determined by the method described above; results are shown in Table 9. Among the marine fish evaluated, the total purine content (dorsal muscles + abdominal muscles + skin + entrails + eyes) was highest in Sphyraena. In contrast, Rajiformes exhibited the lowest total purine content. Considering only the edible parts of the fish (muscles, skin, and eyes) Scophthalmus maximus, Sphyraena, and Sardinella had the highest purine contents. Rajiformes and Anguillidae had the lowest purine contents and would thus be more suitable for a low-purine diet. To control the concentration of serum uric acid and avoid hyperuricemia and gout, it is best to reduce the consumption of high-purine fish such as Scophthalmus maximus and Sphyraena.

The purine contents in fish viscera were significantly higher than those in muscles. The total purine content in viscera ranged from 1,085.61 to 2,719.72 mg/kg, but the muscle samples contained only 551.33–1188.12 mg/kg (dorsal muscles) and 376.94–1302.82 mg/kg (abdominal muscles) total purines, with Sphyraena and Scophthalmus maximus exhibiting higher total purine contents than the muscles samples of other fish. This may be due to the higher visceral metabolic rate in the intestine of these species, which could promote the formation of purine bases. Furthermore, all muscle samples (dorsal muscles and abdominal muscles) were found to contain higher amounts of hypoxanthine than the other three purine bases. This observation is consistent with the report by Qu et al., (2017), who found that hypoxanthine content was the dominant purine in the muscles of Lateolabrax japonicus, followed by adenine, guanine, and xanthine.

The skins of the marine fish contained large amounts of guanine and hypoxanthine, which accounting for 74.22–95.48% of the total purine content. The total purine content of skin varied significantly between the different species of marine fish. Anguillidae skin exhibited total purine content of 486.56 mg/kg, whereas the purine content of Sphyraena skin was 10 times higher. This might be related to the different habitats and growth patterns of the marine fish. Theoretically, the four purine bases can be transformed to uric acid in an equal manner, but hypoxanthine and adenine exhibit the greatest hyperuricemic effect (Clifford et al., 1976). The main difference between the purine contents in viscera compared with that of the eyes was the dominant purine. As seen in Table 9, viscera were rich in adenine, guanine, and hypoxanthine, whereas eyes contained the highest level of guanine. Furthermore, it is worth noting the purine content in fish eyes, which ranged from 1,010.64 to 3,990.19 mg/kg. In some regions, especially in China, consumers commonly eat fish eyes. Considering the high purine content of fish eyes and the pain and potential joint damage caused by hyperuricemia and gout, the consumption of fish eyes should be avoided, even though they are rich in nutrients such as collagen and unsaturated fatty acids.

Changes in the purine content of muscle during boiling

Previous studies have demonstrated that processes such as boiling, steaming, and microwaving can significantly reduce the purine content in foods. Among these processes, boiling is one of the most effective approaches to decreasing purine content. However, most studies on the effect of boiling on purine levels have mainly focused on meats such as chicken and beef, while less attention had been paid to the effect in aquatic organisms (Brulé, Sarwar & Savoie, 1989; Young, 1983). The changes in purine content during the boiling of marine fish measured during this study are shown in Fig. 7.

Hypoxanthine was the major purine base measured in the muscle samples (Fig. 7). In the dorsal and abdominal muscles of Scophthalmus maximus, the amounts of adenine, guanine, and hypoxanthine decreased after boiling in water, but no changes were observed in the amount of xanthine present (Fig. 7). Among the purines measured, hypoxanthine was found to undergo the greatest reduction, from 867.90 to 258.24 mg/kg (70.24%) in dorsal muscles and from 933.94 to 442.98 mg/kg (52.57%) in abdominal muscles. Several studies have determined that hypoxanthine and free hypoxanthine-related compounds have good solubility and can be released easily from foods during cooking (Brulé, Sarwar & Savoie, 1989; Young, 1982). Our findings are also comparable to those of Lou, Lin & Benkmann (2001), who found that free purine bases in grass shrimp could be easily released and transferred to the cooking liquid during the cooking process. Furthermore, the contents of adenine and guanine decreased slightly after boiling; the amounts of adenine and guanine released into the cooking water were 35.89% and 15.92%, respectively, in dorsal muscles, and 32.21% and 29.18%, respectively, in abdominal muscles. The xanthine contents in muscles were relatively low to begin with, and exhibited nearly no change after boiling.

As for the effect of boiling time on changes in purine content, the contents of the four purine bases decreased more extensively with prolonged treatment time, and most purines in the dorsal muscles were reduced within 12–15 min. In comparison, the total purine and hypoxanthine contents in abdomen muscles decreased significantly within 0–3 min, declining by 64.63% and 49.13%, respectively (Fig. 7).

Changes in purine content of skin during boiling

Changes in the purine content of the skins were similar to those of muscles. Boiling reduced the amounts of purine base presents. As shown in Fig. 8, boiling led to a slight decrease in total purine content, from 1,596.68 to 1,117.41 mg/kg in Scophthalmus maximus skin. Hypoxanthine decreased rapidly (by 41.47%) in the first 3 min, after which time the rate of removal slowed down. Guanine was the major purine base in the skin samples; however, adenine and guanine decreased only slightly after boiling, declining by 35.71% and 23.20%, respectively. Boiling had the greatest effect on the removal of hypoxanthine among the four purines, with the content decreasing from 324.04 to 82.51 mg/kg.

Change in purine content of the cooking liquid during boiling

In order to explore the transfer of purine between the fish parts and the cooking water during boiling, the changes in purine content of the cooking water were also determined. As shown in Fig. 9, with increased boiling time, the contents of the four purines in the cooking liquid increased gradually over 15 min. The total purine content of the water increased remarkably within the first 3 min. The cooking liquid contained abundant levels of hypoxanthine, which was measured at concentrations approximately three times higher than those of the other three purines combined. The hypoxanthine content significantly increased within the first 9 min and remained stable thereafter. There was no significant change in xanthine content over 15 min of boiling. The adenine and guanine contents in the cooking liquid increased sharply from 3 to 15 min, by 161.99 and 180.10 mg/kg, to final contents of 200.26 and 230.94 mg/kg, respectively. These results confirm that the purines contained in the edible parts of the marine fish were transferred to the cooking liquid. This result agrees with the findings of Young (1982), who reported that the level of hypoxanthine in tissues decreased during cooking as it was removed by the cooking liquids.