3.1. Effect of the concentration gradient of oleic acid and palmitic acid on propionic acid production
The VFAs yield is an important index to measure the performance of an acid production system. As shown in Fig. 1 (a), the VFAs yield of group Q2 (2382.22 mg/L) and group Q6 (2167.88 mg/L) were higher than the blank control group Q1 (2083.74 mg/L), and they were 14.3% and 4% higher than that of the blank control group, respectively. This indicated that LCFAs could promote acid production when the concentration of LCFAs was at an appropriate value, and the promotion effect of oleic acid was stronger than that of palmitic acid. The promotion was led by the β-oxidation of the oleic acid or palmitic acid, which contained VFAs as oxidation products that raising the total yield of VFAs [20]. The acid production of the other experimental groups was lower than that of the blank control group, indicating that LCFAs inhibited the acid production capacity of the acid production system at the corresponding concentration of LCFAs.
In addition, it can be observed from Fig. 1 (a) that the time of VFAs production peak in each experimental group had a certain trend. The peak value of VFAs production in the blank control group was on the 6th day, that in oleic acid group was on the 6th, 11th and 5th day, and that in palmitic acid group was on the 9th, 10th and 11th day. Except for group Q2 and Q4 in the oleic acid group, acid production's peak value in other experimental groups appeared lag compared with that in the blank control group. The lag period of the palmitic acid test group was longer than that of oleic acid. Besides, the lag period increased with the increase of palmitic acid concentration, which showed its delaying effect on the acid production system. The reason for the delayed effect is that when oleic acid/palmitic acid was attached to the surface of acid-producing bacteria, undegraded LCFAs affect the interaction of internal and external substance exchange and metabolism of acid-producing bacteria. The reason why the lag time of the palmitic acid test group was longer is that the degradation rate of palmitic acid as saturated acid is much slower than that of oleic acid (unsaturated acid) [20–22]. Therefore, it can be concluded that the palmitic acid would accumulate on the surface of acidogenic bacteria for a longer time. When the concentration of attached LCFAs was so much that bacteria could not degrade it and beyond the tolerance range of bacteria, it showed the inhibitory effect of LCFAs on the acid production system.
As shown in Fig. 1 (b), palmitic acid negatively affected propionic acid production in this study. However, the inhibition of palmitic acid on propionic acid production capacity in the acid production system slowed down with the increase of palmitic acid dosage, and the propionic acid production curve in the palmitic acid test group showed a similar trend. The yield curve of propionic acid in the oleic acid group was divided into two movements. When the dosage of oleic acid was 4g/L in group Q2, the peak concentration of propionic acid was 2027.57mg/L, and that in the blank control group Q1 was 1533.76mg/L. Compared with group Q1, the production of propionic acid in group Q2 was 1.3 times higher than that in group Q1, showing that propionic acid production was significantly improved. When the dosage of oleic acid was 8g/L and 12g/L, it offered an inhibitory effect on propionic acid production, and the propionic acid production curve was similar to that of the palmitic acid group. According to Sousa et al., when oleic acid was the only carbon source in the anaerobic system, no other VFAs were found as degradation products except acetic acid[23]; therefore, the possibility of conversion of oleic acid to propionic acid was ruled out. The reasons for the above results can be explained from the following two aspects: a) In the process of oleic acid decomposition, some factors can promote the production of propionic acid existed [17]; b) According to some studies, oleic acid and palmitic acid had an inhibitory effect on propionic acid consuming bacteria, but oleic acid had a stronger inhibitory effect on propionic acid consumption[24, 25]. Hence, it can be considered that within the concentration range that acid-producing bacteria can tolerate, oleic acid could promote the production of propionic acid. And on the other hand, oleic acid reduced the consumption of propionic acid, thus forming the phenomenon of propionic acid accumulation.
3.2. Composition of VFAs in each experimental group
In this short-term batch test, when the production of VFAs in each acid-producing system reached its peak, it was indicated that at this moment, the key microbial community in the system came to a state that exerting the strongest acid-producing capacity, and the dominant acid-producing bacteria showed strong vitality. Table.2 showed the composition of VFAs in the system when each test group reached the peak production of VFAs.
Table.2 The VFAs composition under the maximum concentration in each group
Items
|
No.
|
Content
(g/L)
|
a VFAs
(mg/L)
|
b HAC
(%)
|
b HPr
(%)
|
b HIsobu
(%)
|
b HBu
(%)
|
b HIsova
(%)
|
b HN-va
(%)
|
None
|
Q1
|
0
|
2198.21
|
20.98 ± 0.3
|
64.50 ± 0.9
|
0.75 ± 0.2
|
6.73 ± 0.8
|
1.77 ± 0.4
|
5.28 ± 1.1
|
oleic acid
|
Q2
|
4
|
2382.22
|
6.42 ± 0.2
|
76.50 ± 0.5
|
0.56 ± 0.2
|
5.99 ± 0.3
|
1.92 ± 0.2
|
8.61 ± 0.6
|
|
Q3
|
8
|
1303.08
|
27.12 ± 1.1
|
11.31 ± 0.9
|
0.45 ± 0.1
|
41.90 ± 0.4
|
0.90 ± 0.2
|
18.31 ± 0.7
|
|
Q4
|
12
|
1510.66
|
10.60 ± 0.8
|
15.80 ± 0.4
|
1.04 ± 0.6
|
48.54 ± 0.7
|
1.31 ± 0.5
|
22.71 ± 0.3
|
palmitic acid
|
Q5
|
5
|
1806.22
|
21.06 ± 1.2
|
31.88 ± 0.6
|
0.30 ± 0.1
|
35.73 ± 0.6
|
1.54 ± 0.3
|
9.50 ± 0.6
|
|
Q6
|
10
|
2167.87
|
20.19 ± 1.1
|
22.27 ± 0.9
|
1.24 ± 0.2
|
39.94 ± 0.8
|
1.38 ± 0.3
|
14.98 ± 1.1
|
|
Q7
|
15
|
1986.67
|
19.49 ± 0.5
|
26.72 ± 0.8
|
0.62 ± 0.2
|
41.66 ± 0.3
|
2.59 ± 0.6
|
8.92 ± 0.3
|
a VFAs represents the maximum concentration in each experimental group.
bAc, acetic acid; Pr, propionic acid; Isobu, Isobutyric acid; Bu, Butyric acid; Isova, Isovaleric acid; N-va, N-valeric acid.
|
The composition of VFAs was closely related to the type and concentration of LCFAs during fermentation. It can be observed from Table.2 that acetic acid, propionic acid, n-butyric acid, and n-valeric acid are the main components of VFAs in the acid production system. Since the REDOX degrees of glycerol and propionic acid are numerically equal(γ = 4.67 for both)[26], in the blank test group Q1, glycerol was the main carbon source, propionic acid was the main VFAS component. Therefore, group Q1 showed propionic acid-type fermentation. While with the dosage of LCFAs, the other groups showed a great difference. In the oleic acid experimental group, when the dosage of oleic acid was 4g/L, propionic acid production accounted for the absolute advantage. But it dramatically reduced as the concentration of oleic acid increased; instead, n-butyric acid and n-valeric acid content increased, and the acid production system changed from propionic acid-type fermentation to butyrate-type fermentation. In the palmitic acid test group, butyric acid was the main VFAs product no matter the concentration of palmitic acid. And all the palmitic acid experimental groups showed butyrate-type fermentation. This phenomenon was attributed to the fact that acid-producing bacteria were the main undertakers of VFAs production. The addition of LCFAs changed the original living environment of the microbial community and brought a change shock. The corresponding microbial community was formed dominant flora through transformation and reproduction to adapt to the lash of the living environment. Thus, different acid-production fermentation types emerged.
3.3. Effects of oleic acid and palmitic acid on pH and ORP in the acid production system
PH and ORP are the key parameters reflecting the acid production performance of the acid production system because they affect the microbial community and metabolic pathway and determine the production efficiency as well as the composition of VFAs in the process of anaerobic fermentation[27, 28]. Figure 2 shows the changes of pH and ORP of each experimental group during fermentation.
It can be seen from Fig. 2(a) that in the early stage of fermentation, the substrate was hydrolyzed and acidified rapidly under the action of microorganisms to produce a large number of organic acids, and the pH of all experimental groups showed a downward trend. With the increase of fermentation time, the pH of each experimental group showed a different state. In group Q1, the pH trend downward at first but increased on the 8th day and then gradually stabilized. During the fermentation period, the pH range of group Q1 remained at 4.0-4.7. As for the oleic acid group, the pH fluctuated greatly during the fermentation period, ranging from 3.5–4.9. In group Q2, the pH showed a continuous downward trend from day 3 to day 8, and then reached the minimum pH value of 3.53 on the 8th day, finally rose in a broken line, showing that the degree of acidification was higher when the concentration of oleic acid was 4 g/L. And for the palmitic acid group, the pH ranging from 4.3–4.8 tended to be stable during the fermentation period. The phenomenon mentioned above indicated that oleic acid had more effect on pH than palmitic acid. In this way, the suitable pH values in the oleic acid favored the natural selection of propionic acid-producing bacteria over other microorganisms [29].
As for ORP, it can be found from Fig. 2 (b) that there was no significant difference in ORP range of the same type of LCFA: The ORP range of the blank control group was − 7-131mV, 108–297 mV in the oleic acid group, and − 166 − 139 mV in the palmitic acid group. Thus, it can be seen that the increase of ORP in the acid production system was mainly affected by LCFAs components but had little relationship with its concentration. Besides, the results showed that the ORP of the oleic acid test group was higher than that of the palmitic acid test group, indicating that the oxidation of the oleic acid test group was stronger than that of the palmitic acid test group.
To sum up the above phenomena, the level of ORP was not related to the concentration of LCFAs, but more related to the types of LCFAs. And compared with palmitic acid, oleic acid had a greater impact on the pH of the acid production system, and the ORP level was higher, showing stronger oxidation. Due to the lack of an electron transport system in the acid-producing system of anaerobic fermentation, a large amount of NAD+ is usually produced during the oxidative dehydrogenation of the substrate. Therefore, NAD+/NADH is a standard REDOX pair in the intracellular metabolism of acid-producing bacteria[30]. ORP is the main index of electron transfer and REDOX balance of acid-producing bacteria in the reaction process [31]. Hence, according to the experimental phenomenon of higher ORP levels in the oleic acid test group, it can be concluded that the ratio of NAD+/NADH in the oleic acid test group was higher than that in the palmitic acid test group. This phenomenon was related to the β-oxidative degradation of LCFAs. As an unsaturated fatty acid, Oleic acid degrades faster than palmitic acid and releases NAD+ during the β-oxidative degradation[22]. The main substrate of each test group was glycerol in this experiment. With microbial-related enzymes, the glycerol was successively decomposed into dihydroxyacetone phosphate(DHAP), phosphoenolpyruvate(PEP), pyruvate and a large amount of NADH(Fig. 3). Thus, with the appropriate dosage of oleic acid (in this test, the dosage of oleic acid is 4g/L), the β-degradation of oleic acid facilitated the conversion of glycerol to the production of large amounts of NADH. To maintain the balance of NAD+/NADH ratio in the acid-producing system, acid-producing bacteria would relieve the accumulated pressure of NADH through propionic acid production. This is consistent with Ren et al.[30].
3.4. Microbial community structure in oleic acid and palmitic acid systems
The final samples selected for microbial detection in this experiment were: the microbiological samples of blank control group Q1, group Q2, and group Q6 when they reached acid production peak. The analysis data were shown as A1, A2 and A3, respectively.
3.4.1. Microbial diversity and richness
Table.3 reflects the microbial community structure of the selected experimental group when the acid production was stable(i.e., when the VFAs production was the highest). Among them, OTUs, Chao, and Shannon indices are commonly used to measure microbial richness. The higher the index values, the higher the species richness of the samples. The high biodiversity of the acidogenic system is conducive to the stability of the acidogenic system and has a positive effect on the degradation and transformation of substrates in the acidification stage[33]. It can be seen from Table.3 that the richness of microbial community in the Q2 and Q6 was not far different from that in Q1. That means the microbial community in the acid-producing system was not significantly affected at the corresponding LCFAs concentration, which is the main reason for the similar total VFAs yields of the two experimental groups and the blank control experimental group. Besides, the Chao value showed that LCFAs promoted the diversity of microbial communities in the system.
Table 3
Diversity indices of bacterial gene sequences from samples taken
Sample
|
OTUs
|
Shannon
|
Chao
|
Ace
|
Simpson
|
Coverage
|
A1
|
1396
|
4.07
|
1613.32
|
1609.89
|
0.06
|
0.99
|
A2
|
1495
|
4.00
|
1695.31
|
1678.44
|
0.09
|
1.00
|
A3
|
1522
|
4.36
|
1685.78
|
1681.00
|
0.07
|
1.00
|
3.4.2. Microbial community structure
Sequence distribution of bacterial communities with relatively high abundance in the three groups of samples at the Genus level is shown in Fig. 4. Among them, the species whose abundance ratio was less than 1% in the sample were classified as Other. The results showed that the three groups of samples had high similarities in bacterial composition. However, the dominant microbial community of each group had a great difference. When the acid-producing system was stable, the main bacterial community in group Q1 were Clostridium_sensu_stricto, Acetobacter, Anaerosinus, and Propionispira. Their relative abundance was 31.81%, 10.84%, 10.33%, and 7.49%, respectively. The main bacterial populations in the oleic acid test group Q2 were Propionispira, Anaerosinus, Lactobacillus, Clostridium_Sensu_Stricto, and their relative abundance was 25.43%, 11.94%, 9.18%, and 6.84%, respectively. Although relatively abundant, some bacterial genera, such as Megasphaera, still be involved in the crucial reaction of oxidizing glycerol to lactic acid and then converting it to propionic acid[34]. The main bacterial populations in the palmitic acid test group Q6 were Clostridium_sensu_stricto and Anaerosinus, whose relative abundance was 38.43% and 11.76%, respectively.
The results showed that the types of LCFAs had important effects on the key propionic acid bacteria. For example, Propionispira is a typical propionic acid producer [12], usually found in anaerobic fermentation reactors that produce propionic acid using organic substances. It can be seen from Fig. 4 that the Propionispira was dominant in group Q2, but the relative abundance of Propionispira in group Q6 was only 1.47%. Besides, Lactobacillus is also a propionic acid producer [35]; it had a high abundance in group Q2, a small amount in group Q1, but a very low quantity in group Q6. Combined with the acid production results, the rank of propionic acid production from high to low in the three experimental groups was Q2 > Q1 > Q6. This phenomenon indicated that propionic acid-producing bacteria had good adaptability in the oleic acid environment and could give full play to its propionic acid production capacity; however, it was seriously inhibited by palmitic acid.
Another interesting phenomenon was that the dominant bacteria in group Q1 and group Q6 were both Clostridium, a kind of butyric acid-producing bacteria[36]. Theoretically, butyric acid should be the main component of VFAs in group Q1 and Q6, because butyric acid bacteria dominate in abundance. However, in contrast to group Q6, the main VFAs component in group Q1 was propionic acid and acetic acid, indicating that there existed some kinds of bacterial communities competing or antagonizing with Clostridium in group Q1. From the comparison of relative abundance, it can be inferred that Acetobacte is the competitive flora of Clostridium. The above analysis shows that the same seed sludge evolves different dominant community structures under different LCFAs species, resulting from microbial communities adapting to the environment, and there was the interaction between microbial communities.
3.5. Analysis and prediction of optimal concentration of oleic acid in propionic acid production by Response Surface Method
From the above analysis, it can be determined that the appropriate concentration of oleic acid was conducive to propionic acid production. Design Expert 10 was used to perform response surface analysis to define further the oleic acid concentration that could promote propionic acid production. Set up propionic acid yield(Y) as the dependent variable, oleic acid concentration(A) and fermentation time(B) as independent variables. The analysis results were shown in Table. 4 and Fig. 5.
Table 4
Fitting and ANOVA result for the oleic acid test group model.
Source
|
Sum of Squares
|
df
|
Mean Square
|
F Value
|
p-value Prob > F
|
Model
|
1.63E + 07
|
9
|
1.81E + 06
|
44.84
|
༜0.0001
|
A-LCFA concentration
|
1.02E + 07
|
1
|
1.02E + 07
|
252.21
|
༜0.0001
|
B-fermentation time
|
6.43E + 04
|
1
|
6.43E + 04
|
1.59
|
0.2146
|
AB
|
5.24E + 05
|
1
|
5.24E + 05
|
12.97
|
0.0009
|
A²
|
3.39E + 06
|
1
|
3.39E + 06
|
84.04
|
༜0.0001
|
B²
|
3.32E + 05
|
1
|
3.32E + 05
|
8.21
|
0.0067
|
A²B
|
2.19E + 05
|
1
|
2.19E + 05
|
5.43
|
0.0252
|
AB²
|
3.42E + 05
|
1
|
3.42E + 05
|
8.46
|
0.0060
|
A³
|
5.68E + 06
|
1
|
5.68E + 06
|
140.61
|
༜0.0001
|
B³
|
1.14E + 05
|
1
|
1.14E + 05
|
2.83
|
0.1009
|
Residual
|
1.53E + 06
|
38
|
40373.04
|
|
|
Cor Total
|
1.78E + 07
|
47
|
|
|
|
Note: R-Squared = 0.9139, Adj R-Squared = 0.8936, coefficient of variation(C.V.) = 22.63%,
Adeq Precisior = 21.319
|
Evaluate the multiple regression fitting model according to the analysis of variance(Table. 4), the F value of the model was 44.84, and the value of Pmodel >F was less than 0.0001. The fitting results of the software showed that the response of the model to propionic acid concentration was highly significant. In addition, the coefficient of variation(C.V.) of the model was 22.63%, and the Adeq precision was 21.319, which was more than 4, indicating that the model had high reliability and accuracy. According to the corresponding F value obtained by ANOVA, the concentration of oleic acid(F = 252.21) was higher than that of fermentation time(F = 1.59), indicating that the concentration of oleic acid was a more important factor affecting the production of propionic acid. The test data were fitted by multiple regression, and the mathematical regression formula generated by the coding factor(Table. 4) was shown in Eq. (1).
Y = 524.51-2482.10A + 175.99B-409.65AB + 1830.03A2-332.302-413.10A2B + 482.90AB2 + 3
379.31 A3 + 338.98B3 (1)
The "optimization" tool in software Design Expert 10 was used to analyze and predict the optimal conditions for propionic acid production (Fig. 5): When the concentration of oleic acid was 2.3g/L and the fermentation time was on the 7.4th day, the expected yield of propionic acid was 2028.16mg/L.