3.4 Comparison of production processes
When all processes were compared, we observed that the exponential growth phase occurred during the first 3 h in each case. Until this point, all processes were in the batch phase, and for this reason, they presented the same µmax , 1.17 ± 0.06 h-1, which was similar to other pneumococcus strains (Gogola-Kolling et al., 2014; Liberman et al., 2008).
As expected, glucose consumption was almost 4 times higher in the perfusion-batch, since the cultivation time is longer than in batch (Figure 3). Moreover, the production of lactate was almost 5 times higher and biomass was 3 times greater, measured by OD and dry cell weight. Furthermore, the production of lactate was almost 1:1 of glucose consumption. Dry cell weight was greater in batch process until 4 h, then, it became higher in the perfusion-batch due to growth arrest in the batch process. Table 2 compares our results from the batch, fed-batch and perfusion-batch.
All processes presented an equivalent average of biomass yield on glucose, with statistically similar Y X/S . The mean Y X/S was 0.15 g dry cell weight / g glucose, similar to other LAB as Lactobacillus delbrueckii in continuous process with cell recycling (Ohleyer et al., 1985),Lactobacillus casei in batch, fed-batch and continuous process without recycling (Aguirre-Ezkauriatza et al., 2010), and similar to other pneumococcus (Liberman et al., 2008).
Total acetate production was higher in the perfusion-batch process, butY acet/S was statistically higher for the batch process (0.41 g acetate/g glucose), which indicates the nutritional limitation of the batch culture, since acetate production occurs mainly in low glucose concentration (Carvalho et al., 2013).
Y lac/S and Y lac/X were significantly higher in the perfusion-batch process. Ylac/S obtained here in the perfusion-batch (0.97 g lactate/g glucose) was as high as reported for LAB used in lactate manufacture, such as L. delbrueckii subsp. delbrueckii ,L. paracasei and L. lactis subsp. lactis , which presented Y lac/S = 0.91 g/g (John et al., 2007), or Lactobacilllus sp. strain RKY2, which presented Ylac/S = 0.93-0.97 g/g in continuous process with cell recycling with D = 0.04-0.36 h-1 (Wee and Ryu, 2009). Although the conversion of glucose into lactate was high, Ylac/X was lower than other LAB as Lactococcus lactis (Parente et al., 1994), indicating the conditions employed here favored cell growth rather than lactate production, which is in accordance with our goal.
The perfusion-batch exhibited productivity 2 fold lower for biomass (PX ) due to the higher culture medium volume used in this process, and 3.5 fold higher for lactate (Plac ) in comparison to batch and fed-batch protocols. The PX of perfusion-batch was also lower when compared to other pneumococcus strain (Gogola-Kolling et al., 2014), and other Streptococcus (Taniguchi et al., 1987). WhereasPlac observed in all processes here was similar to other studies using LAB, such as in the batch process ofEnterococcus faecalis (Wee et al., 2004) or continuous process of serotype 14 pneumococcus strain (Gogola-Kolling et al., 2014), it was lower than other LAB cultivated using continuous process with cell-recycling, as Lactobacillus paracasei (Xu et al., 2006),Lactobacillus helveticus (John et al., 2007) orLactobacillus rhamnosus (Kwon et al., 2001), which is in accordance with our goal of optimizing cell growth, but not the lactate production.
Here, we also estimated the total protein production and number of vaccine doses (Table 2). The perfusion-batch integrated to the cell separation had almost threefold more protein in the concentrated product, as observed for biomass production, which could generate 3 times more human vaccine doses per lot than the simple batch process harvested at OD 10. Despite the fact that batch processes consume less glucose and other reagents, and spend less time to produce one lot (considering cultivation and downstream process), as observed in Table 3, it would be necessary to perform 3 batch processes to obtain the same biomass as in 1 perfusion-batch (Figure 4). As consequence, the number of doses produced by lot in the perfusion-batch will be also about 3 times higher. Therefore, the batch protocol would be more expensive due to the cost with medium, reagents, cleaning process, employees, infrastructure, etc. Consequently, the perfusion-batch would minimize downtime, which is also costly. In addition, the perfusion-batch would produce 4 times more doses than the previously developed simple batch process, in which the cells were harvested at lower concentration, when OD reached 6.0 (Gonçalves et al., 2014).