The P. pastoris GS115 strain containing the pPICHKAPel vector was used for the heterologous expression of pectate lyase (Pel from Bacillus sp. RN1) under control of the AOX1 promoter G/Pel (Zheng et al., 2020). The P. pastoris GS115 strain containing the pPICZαAphy vector was used for heterologous expression of phytase (Phy from C. amalonaticus CGMCC 1696) under the control of the AOX1 promoter G/Phy (Weidner et al., 2010). The pre-inoculum for bioreactor cultivation was yeast extract peptone dextrose.

The SCC experiment was conducted in a 3-l Sartorius Biostat A Plus device (Sartorius Biotech, Gottingen, Germany). Initial and replacement media consisted of basal salt medium (BSM) added at 4.35 ml trace element solution PTM₁ per liter as described elsewhere (Zhao et al., 2008), and the inoculation concentration was 8% (v/v). BSM contained 27 ml·l⁻¹ 85% H₃PO₄, 40 g·l⁻¹ glycerol, 18 g·l⁻¹ K₂SO₄, 14.9 g·l⁻¹ MgSO₄·7 H₂O, 4.13 g·l⁻¹ KOH, and 0.93 g·l⁻¹ CaSO₄, supplemented with 4.35 ml·l⁻¹ PTM1 trace salts containing 65 g·l⁻¹ FeSO₄·7H₂O, 20 g·l⁻¹ ZnCl₂, 6 g·l⁻¹ CuSO₄·5H₂O, 3 g·l⁻¹ MnSO₄·H₂O, 0.5 g·l⁻¹ CoCl₂, 0.2 g·l⁻¹ MoNa₂O₄·2H₂O, 0.2 g·l⁻¹ biotin, 0.09 g·l⁻¹ KI, 0.02 g·l⁻¹ H₃BO₃, and 5.0 ml·l⁻¹ 98% H₂SO₄.

The fermentation process was divided into four phases, as shown in Figure 1. Phases I and II were the cell growth stage in which DO was maintained at >20% by controlling the agitation rate (200–1,000 rpm, phase I), with a feed rate of 50% (w/w) glycerol containing 12 ml·l⁻¹ PTM1 solution (phase II); temperature was maintained at 30°C. pH was maintained at 5.5 by feeding with 25% ammonium hydroxide. Phases III and IV were the induction stages, in which methanol containing 12 ml·l⁻¹ PTM1 solution was fed to produce target proteins. The temperature was controlled at 25°C, pH was adjusted to 6.0, and DO is maintained at >20%. All SCCs followed the same procedure in the first cycle as general batch-fed cultivation. However, at the end of the first cycle, part of broth was removed so that 0.75 l remained in the bioreactor, and then BSM was added until reaching a total volume of 1.5 l. After the second cycle, the above operation was repeated. This process removed phases I, II, and III, leaving only phase IV. The SCC of reused cells has an additional sedimentation stage (phase V) for distinguishing and collecting cells. At the end of phase V, the supernatant was removed, leaving only 0.35 l fermentation broth (15% of the total fermentation broth). Then, low-salt basal salt medium (LSM) containing 4.35 ml PTM1 per liter was added to a total volume of 1.5 l. The LSM contained 27 ml·l⁻¹ 85% H₃PO₄, 9 g·l⁻¹ K₂SO₄, 7.45 g·l⁻¹ MgSO₄·7 H₂O, 2.065 g·l⁻¹ KOH, and 0.465 g·l⁻¹ CaSO₄.

Sample (10 ml) were collected every 4 h (phases I and II) or 6 h (stage III) and characterized by absorbance at 600 nm in 2 ml cuvettes.

Pectate lyase activity was determined according to the method described by Zheng et al. (2020), with one modification: measuring the absorbance change at 235 nm with 2 mg PGA ml-1 as the substrate in 50 mM glycine–NaOH (pH 10.0) buffer containing 1 mM CaCl₂ for 40°C, 30 min. One unit (U) of pectin lyase activity was defined as the amount of enzyme that is required to produce unsaturated oligogalacturonide equivalent to 1 μmol of unsaturated diglucuronide min–1 using a molecular extinction coefficient of 4,600 M⁻¹ cm⁻¹ at 235 nm (Zheng et al., 2020).

A 1-ml cell resuspension (OD₆₀₀ = 40) was centrifuged at 6,000 rpm for 1 min. The cells were collected and washed three times in 10 mM phosphate-buffered saline (PBS, pH 7.4), and OD₆₀₀ was adjusted to 5. Then, a 10-μl aliquot of 5 mmol propidium iodide (PI) was added to 200 µl cell suspension and incubated on a shaker for 20 min at 37°C. The cells were washed three times and resuspended in 1.5 ml PBS (pH 7.4). The cell suspension was analyzed by flow cytometry (Beckman Coulter, Fullerton, CA, USA). One hundred thousand cells per sample were counted and analyzed using Exp032 software (Beckman Coulter). The mean value of fluorescence intensity (X-mean) of the sample could be calculated by software FlowJo v10.

We developed an improved, greener, and more energy-efficient cultivation process for recombinant protein production from Pichia pastoris. In regular semi-continuous cultivation (SSC), the period of fermentation was shortened by omitting the growth phase (Figure 2) after the second cycle. This time would be further prolonged if higher cell concentrations are used for initial induction, combined with the methanol adaptive phase (phase III).

The strain G/Pel was used to produce pectate lyase in a 3-l bioreactor after the growth stage, which was evaluated by running eight cycles. The levels of pectate lyase gradually decreased within the eight cycles. As the cycle changed, the production capacity gradually declined. At the end of the fifth cycle, pectinase activity was only 79% of the first cycle. The patterns of cell growth were the opposite of those of enzyme activity. From the second cycle, cell growth gradually accelerated. Although the number of cells gradually increased, the number of vitality cells gradually declined (Figure 3), accounting for the gradual decline in production capacity. Similar results were observed for phytase production (Supplementary Figure S1), and the production capacity decreased even more. which was due to reduced productivity as cells age (Verbelen et al., 2009). At the end of the fourth cycle, the enzyme activity was only 23.8% of that in the first cycle.

However, the mass of cells was drained from the broth each cycle, and the vitality cells were difficult to separate from the remaining broth, leading to the accumulation of aging cells. Dead cells would inevitably affect productivity.

To avoid the significant decrease in production capacity due to cell aging in SCC, we conducted semi-continuous cultivation with sedimentation to separate vitality cells from the broth. Sedimentation time is a principal element. As shown in Figure 4, the longer the sedimentation time, the better the separation of cells. Remaining in the lowest layer by 15% of fermentation broth to collect cells, the cell recovery reached 28.9 ± 0.45% after 24 h of sedimentation. Over time, the cell recovery rate slowly increases. These results indicate that the G/Pel cells were better for precipitation of solid/liquid separation when the fermentation broth stood for 24 h.

Moreover, 15% fermentation broth remained in the bottom to collect cells, which were then added to fresh LSM for the subsequent fermentation cycle. The average cell recovery reached 39.5 ± 4.4%. The mass of G/Pel cells grew to reach about 300 OD₆₀₀, and the maximum production of pectate lyase was about 60 U/ml until the eighth cycle (Figure 4). The growth of cells and pectate lyase production were stable in each cycle because aging cells were discarded (Figure 5). Moreover, most vitality cells remained for the next cycle. There was almost no loss of pectate lyase activity after standing for 24 h, indicating that the process can be applied to culture G/Pel (Figure 4). In addition, the same process was used for the production of phytase, with almost identical results (Supplementary Figure S2). Compared to fed-batch fermentation, we saved 140 h and substantial water and electricity within the eight cycles. However, fermentative broths need to stand for 24 h between cycles, which is time consuming and increases the risk of contamination. The results were similar to Pfeffer et al. (2006); the semi-continuous process would collect more products and reduce.

Yeast cells have a negative surface charge (Figure 6A) (White and Walker, 2011; Kregiel et al., 2012), which affect the sedimentation efficiency of yeast due to the repulsion of the same charge (Benabdesselam, 2010; Tibayrenc et al., 2011). The positive charges of metal ions can interact with the cell surface to form a salt bridge (Mill, 1964; Jayatissa and Rose, 1976; O'Mahony et al., 2006; Goossens et al., 2015). We applied this concept, using Ca²⁺ and Fe³⁺ ions.

To deliver these ions, we added CaCl₂ and FeCl₃. First, we determined their optimal concentrations. As shown in Figure 6, the effects of flocculation precipitation of P. pastoris cells increased over 12 h with the concentration of Ca²⁺ ions. When the concentration of CaCl₂ was 200 mM, the supernatant was clarified and sedimentation was the best. Cell recovery reached 29.0 ± 0.45%, almost as good as natural sedimentation for 24 h. Using the same process, we found that 50 mM FeCl₃ was optimal.

We applied these to accelerate G/Pel sedimentation for SCC. As shown in Figure 7, the sedimentation rate of G/Pel was significantly accelerated. After sedimentation for 12 h, the layer of vitality cells was collected for the next fermentative cycle, in which the cell density attained about 150 OD₆₀₀ at a total volume of 1.5 l. The average cell recovery was 40.7 ± 6.5% and 38.2 ± 2.5% using Ca²⁺ and Fe³⁺, respectively, almost the same as natural settling for 24 h. However, the effects of Ca²⁺ ion precipitation did not meet expectations. After three cycles, the growth of cells and the production of pectin lyase decreased because Ca²⁺ ions accumulated as the culture was recycled. SCC obtained the best results with Fe³⁺ ion sedimentation. When a fermentative cycle was finished in 96 h, 2 M Fe³⁺ ions were added to the broth, maintained at a concentration of 50 mM ions (2.25 l in the ending fermentation), and then used to accelerate G/Pel cell sedimentation for 12 h. The vitality cells (Figure 8B) were collected for recycle fermentation, and the supernatant was used for pectate lyase preparations. Running eight cycles, the process showed optimal semi-continuous fermentation (Figure 7). It also cut down on processing time and the use of resources, water, and electricity. When SSC with Fe³⁺ ions was applied to the G/Phy strain, similar results were obtained (Supplementary Figure S3).

To demonstrate the benefits of cell recycling, the economic and environmental efficiency of all processes was evaluated from the viewpoint of resource consumption, time cost, production efficiency, and waste cells produced. The product recovery, cell recovery, cell concentration of broth, yield of methanol in cells γx/m, yield of methanol in phytase γ_P/m, and time cost of each cycle are given in Supplementary Table S1.

The average product recovery of SCC was about 71.5% lower than 88.1%, 87.6%, and 87.2% of NS24, CCa, and CFe. That means that the improved process can collect more protein products in one cycle. The average cell recovery of SCC was about 37.4%, versus 38.8%, 41.2%, and 38.2% for NS24, CCa, and CFe, respectively. In other words, the number of cells collected by sedimentation could meet the production requirements of the next cycle as in standard SCC. The cell concentration of the supernatant was used to evaluate the effects of the process on downstream solid–liquid separation; a lower cell concentration of the supernatant would help reduce the cost of later separation. The results showed that the order of average cell concentration (OD₆₀₀) of the supernatant was CCa (156.1 ± 22.0) < NS24 (161.3 ± 11.8) < CFe < SCC (168.0 ± 13.8) < batch-fed cultivation (311.5 ± 47.1). The concentrations of vitality cells were much lower than those of batch-fed cultivation and the SCC process, indicating that it could effectively reduce the cost of downstream solid–liquid separation.

The biomass on methanol yield coefficient γx/methanol and product on methanol yield coefficient γ_P/methanol were used to evaluate the utilization efficiency of methanol. The average γ_x/methanol value of SCC, NS24, CCa, and CFe was about 0.178 ± 0.08 g·DCW·g⁻¹ methanol⁻¹ in the first cycles. In other words, the γ_x/methanol of G/Pel in batch-fed cultivation was 0.178 ± 0.08 g·DCW·g⁻¹ methanol⁻¹. Except for CCa, the γ_x/methanol of other processes was larger than the first cycle in the subsequent cycle. Almost all of the values were above 0.22. The average γ_P/methanol of SCC, NS24, CCa, and CFe was about 273.5 ± 7.5 u·g⁻¹ methanol⁻¹ in the first cycles. The γ_P/methanol of SCC started to go down after the second cycle, and the processes maintained high γP/methanol values, all greater than or equal to the first cycle, and an average γx/methanol of 281.2 ± 23.4, 284.5 ± 17.4, and 285.5 ± 13.8 u·g⁻¹ methanol⁻¹, respectively. These results indicate that all processes increased product recovery and the utilization efficiency of methanol.

Next, SCC, NS24, CCa, and CFe were used to conduct a comprehensive evaluation of the entire process of G/Pel culture, as shown in Table 1. The first cycle was batch-fed cultivation, set as a standard to evaluate the different processes. The consumption of glycerol, methanol, inorganic salts, the time cost, and the amount of waste yeast were compared under the condition of producing the same amount of phytase. The consumption rates of methanol glycerol and inorganic salt were 19.7 ± 1.4% of batch-fed cultivation, respectively, and the amount of waste yeast produced and time cost were almost the same as batch-fed cultivation. When producing the same pectate lyase, methanol consumption even reached 130% of batch-fed cultivation. SCC has no apparent efficiency advantages. However, NS24 had a significant advantage over SCC and batch-fed cultivation in terms of vitality cell collection. When producing the same amount of pectate lyase, NS24 only needed 15.2 ± 0.2% glycerol, 56.3 ± 0.1%, 86.3 ± 0.1% time cost, and produced 78.8 ± 0.6% waste yeast. Although methanol consumption was not lower, NS24 was more economical and greener. Adding Ca²⁺ to accelerate G/Pel sedimentation was not suitable for recycling. Although the process reduces glycerol consumption, inorganic salt consumption, time cost, and amount of waste yeast, the decrease in the growth rate of G/Pel meant that the process could not be run for long. CFe further reduced the time cost based versus NS24, which only cost 81.5 ± 0.0% time of batch-fed cultivation while producing the same pectate lyase, less 86.3 ± 0.1% of NS24.

Compared to batch-fed cultivation, NS24 and CFe are significantly green and efficient; CFe, in particular, was at least 20% cheaper than batch-fed cultivation and SCC. When we changed the strains from G/Pel to G/Phy, the SCC, NS24, and CFe culture of G/Pel and G/Phy yielded similar results. This comprehensive evaluation indicates that NS24, CCa, and CFe consume fewer resources, decrease time cost, and produce less waste yeast than batch-fed cultivation and general SCC (Supplementary Tables S2, S3).

In sum, regardless of the fermentation process of G/Phy or G/Pel, when cells can be recycled, the same results are always obtained: less time cost, less resource consumption, and less waste yeast generation. In other words, the process of cell recycling is greener and more economical.