Parameter variables mainly number of passes, biomass concentration and pulse pressure are primary concerns that would affect the outcome of P. pastoris cell disruption and the release of HBsAg specific protein. To that extent, optimizations of these parameters are essential for efficient extraction of the protein. With the advantage of statistical design (RSM), it provides an alternative methodology to optimize a particular process by considering the mutual interactions among the factors and to give an estimate of the combined effects of these factors on the final results, which could lead to simplification of process optimization and cheaper production costs. In addition, significant factors that influence the responses to the greatest extent could also be determined.
Among the three significant factors tested, pulse pressure gave the most significant effect compared to that of number of passes and biomass concentration in both cell disruption capability and in release of HBsAg. This indicates that pulse pressure was very important for maintaining the optimal level in the recovery of HBsAg while achieving the best possible cell disruption capability. Pulse pressure has been widely investigated previously in the improvement of cell disruption involving high pressure homogenizer on various types of microorganism. Cell disruption is a two-step process that involves primary rupture of the cell envelope involved a point break, followed by further breakage of the cell wall and degradation of cellular debris . When pressure is applied, disruption increases with increasing level and in certain cases, complete disruption of the initial cell load is achieved. For instance, Kleinig and Middleberg  demonstrated for yeast cells, disruption only occurs beyond the pressure of 560 bar. Donsě et al.  pointed out that the linear trend of cell disruption kinetics is dependent on pulse pressure to a certain level. For the homogenization of baker’s yeast, a 4-fold increase in degree of cell disruption was observed at 2,500 bar as compared to 500 bar. This pressure dependence were due to the difference in cell membrane and cell wall properties especially for yeast cells, which are considerably harder to disrupt due to its overall very thick and resistant cell wall structure [35, 36]. In this matter, Harrison et al.  reviewed that yeasts are harder to disrupt as compared to Gram positive and negative bacteria. From Figure 2 B and C, it is evident that experimental data reported in response surface plot as a function of the pressure effect seem to fit well by a straight line suggesting a first order of kinetics for cell disruption capability. In this case, CDR depends linearly on the applied pulse pressure of homogenization regardless of the influence from number of passes and biomass concentration. It was noticed for multi pass homogenization, cell disruption efficiency reduces after each pass tending towards a plateau of an asymptotic value of disruption . This can be attributed to the natural distribution of individual cell resistance to pressure, since the most resistant cells of the initial microbial population are likely to survive the pulses of pressure applied. Therefore, when all the weakest cells are destroyed, no further disruption would occur, even if the number of passes increases . In this case, we supposed that the lack of significance changes in CDR under the determined range was due to the achieved plateau level even at the lowest number of pass for the mentioned pulse pressure range. A threshold value for multiple number of passes effect exists, above which effectiveness of homogenizer process would not be seen. In spite of that, our experimental data appear to be in conflict with the results reported in the literature. Several authors indeed supported the hypothesis that in number of passes treatment, each pass is additive. Each pass would cause the same rate of cell disruption [14, 39]. It was observed that successive rounds of high-pressure homogenisation have an additive effect on viability reduction of Yersinia enterocolitica and Staphylococcus aureus.
Also in this case, our results are in agreement with the findings from several authors in that the cell has no discernible influence on cell disruption efficiency over a wide range of cell concentrations [19, 40]. For yeast disruption process, it was determined that the process was independent from effects of biomass concentration at a range below 750 g/L . In addition, Van Hee et al.  further concluded that biomass concentration does not have a significant effect on cell rupture in the concentration range of 0.06-115 g/L for high pressure homogenizer. It is not unexpected as the main factor in determining the kinetics of cell disruption is the natural distribution of individual cell resistance to high pressure homogenization rather than cell concentration . It is evident that the biomass concentration used in the experiments was in the mentioned ranges, thus the observation appears to be valid. Instead, for the combinative effects on number of passes and biomass concentration, it leads to a clearer trend of dependency. At a fixed pulse pressure, the lower the biomass concentration and number of passes, the higher the cell disruption capability was achieved (Figure 2A). This behaviour could more accurately ascribe the conditions observed in the previous effect of number of passes, at which weaker or more sensitive cells are selectively disrupted in initial few passes below threshold value. Subsequent passes would treat more resistant cells, resulting in lower cell disruption capability . In another approach, Peleg and Cole  suggests the hypothesis that cell disruption capability is derived from the cumulative form of resistances or sensitivities distribution within cell population. This approach takes into account of the coexistence of subpopulation which is more resistant to stress or protected by several factors, such as dead cells or secondary product of cellular metabolism. It is also worth noting that the combination effects of number of passes and biomass concentration enhanced the cell disruption capability achieving a higher CDR rate approximately with 18 times number of passes and 6.72 g/L biomass concentration. Nonetheless, the influence was minor in the presence of the pulse pressure effect.
A different behaviour was observed in the release of specific protein from cell disruption process, with the significant factors involving not only pulse pressure but also number of passes and biomass concentration. From Figure 3, all three graphs produce a ‘mountain shaped’ response surface plots which implicates a co-dependency among the variables. Pulse pressure under the influences of number of passes and biomass concentration was observed to enhance specific protein release up to a certain level, above which the release of specific protein curve level off to a decreasing value (Figure 3 B and C). In the literature numerous examples of protein release curves under the influence of pulse pressure which are not governed by the first order of kinetics and gave rise to a non-linear behaviour were reported [19, 37]. For example, Keshavarz et al.  observed that the release of specific protein alcohol dehydrogenase (ADH) increased to a curvature level accordance to pulse pressure from cell disruption of Rhizopus nigricans. Results in support of decrease in the release of specific protein after threshold saturation in the increase of pulse pressure were also reported. Duerre and Ribi  concluded that a pressure range of 1,034 bar was sufficient for maximum cell disruption and speculated that degradation of protein can occur when cell disruption was performed at very high pressure range (≥1,734 bar). Ho et al.  speculated that the decrease in specific protein release might due to breakage or degradation of the specific protein resulting in ruptured, degraded or unformed particles that has a loss of antigenicity from the samples obtained. In another instance, study performed by Lovitt et al.  has shown that maximum specific activity was achieved at pulse pressure of 1,000 bar disruption from baker’s yeast. At higher pressures, up to 2,400 bar, enzyme activity released fell progressively. The observations are consistent with greater quantities of protein becoming solubilized at highest operating pressures but that the specific protein levels in solution remained stagnant or even decline at very high pressures. At the same time the protein released rose rapidly as the operating pressure increased to 1,380 bar, until 2,750 bar where over 85% of the total protein present had been released into a soluble form, thus diluting the specific protein . On the basis of this assumption, it would explain the results obtained for experimental run 7–8 and 14. Under the influence of high pressure range (≥ 1,000 bar), high total soluble protein was released while low specific protein was detected, resulting in low selective product recovery (Figure 4). On the other hand, the loss or decrease in specific protein release could be caused by foam formation during homogenization where large fraction of the specific protein were contained in the foam that was not included in the samples leading to inaccurate sampling . However, this is unlikely as the foam formation in our experiments is only of a minute amount compared to the volume used in the experiment runs. Nevertheless, strong reactivity signal was detected in samples derived from pulse pressure of 1,000 bar and at a biomass of 7.70 g/L. Meanwhile, several studies have reported that increase in pressure above a certain range would result in micronisation of cell debris [40, 44, 45] and that higher protein release was observed due to the release of insoluble protein complex, peptides, glycopeptides and amino acids . This is coherent with our experimental data, where higher ratio of selective product recovery was achieved in lower pressure range (≤600 bar) in comparison to higher pressure range (≥1,000 bar) due to higher total soluble protein released.
Based on the kinetic-rate law model reported by Hetherington et al.  and observations made by Donsě et al. , it is possible to predict that the release of specific protein is dependent on the coherent influences of number of passes and pulse pressure. On the basis of such assumption, increasing either one factor would result in the same outcome of protein released and regulation in the increase of both factors would dramatically enhance the release of specific protein to a certain level. From Figure 3B, it is clear that our result is in agreement with the findings in the literature that indeed, additional passes are required for better release of specific protein corresponding to pulse pressure used [17, 19, 22, 46]. In general, a single pass system is thought to be best for the product; however, at a given pressure, multiple passes may be beneficial for further downstream processing depending on the location of the specific protein in the cell. The rate of release can be faster than, equal to, or slower than the overall protein release [17, 21, 29]. It has to be pointed out that the disruption of cells to a state where membrane associated specific proteins are released are notably to be more difficult. For example, a membrane associated enzyme cytochrome oxidase from Pseudomonas aeruginosa would require 3 times or more passes compared to unbound intracellular enzymes that could normally be released in a single pass . In our case, HBsAg produced from P.pastoris are membrane associated protein  which suggests for the reason of relatively higher number of passes. In addition, difficulty of disruption was also encountered for the difference in growth phase and culture medium. It has been reported that stationary phase cells [16, 24, 48] and cell grown in complex medium containing yeast extract are more difficult to disrupt . Considerably, similar number of passes (20 times) was also observed to be used in producing efficient breakdown of nano-particles using the Avestin homogenizer  while noticeably much more passes (3,000 times) was used in cell disruption of yeast cells prior to purification . Further increase in the number of passes showed a different behaviour for the release of specific protein, with a reduction in the release of HBsAg was observed. Data reported in the literature from cell disruption using high pressure homogenizers in extraction of intracellular material postulated that intracellular material released is not susceptible to shear damage in a significant degree [22, 27, 45]. Losses are normally due to susceptibility to degradation or inactivation by other factors such as proteases . However, considering that these data were based on low number of passes (1–5 times) this explanation would not properly represent the observations made in this study. Conversely, membrane associated proteins and multi-enzyme complexes were reported to be shear sensitive. For example, in the isolation of a multi-enzyme membrane-associated complex, alkane hydroxylase by homogenization of Pseudomonas putida was reported to have badly damaged the enzyme . In another instance, Augenstein et al.  concluded that in the release of shear sensitive specific protein from Bacillus brevis, degradation of the specific protein occurs beyond a certain number of passes depending on pulse pressure.
Biomass concentration in the combination with pulse pressure and number of passes was shown to influence the outcome of specific protein (Figure 3 A and C). The extent of initial load contributes to the increase and decrease in specific protein release in both cases. Under the influence of number of passes, Save et al.  demonstrated that protein release are only 7 times even though cell concentration was increased to 100 times. Thus, it was speculated that the number of passes in protein release are dependent on biomass concentration and dilution rate of the sample. Mosqueira et al.  explained that the viscosity of the whole cell suspension is a function of biomass concentration which is an important factor in determining the increase and decrease of specific protein release [40, 55]. At a fixed pressure, repeated passes through the homogenizer gradually increases the viscosity of the homogenate, with the extent of the increase depending on the initial biomass concentration. This phenomenon is thought to occur due to cell debris fragmentation and the release of intracellular soluble components, including protein, nucleic acids and polysaccharides which will rise concomitant with increased number of passes . In another instance, Shamlou et al.  highlighted that the direct collision between the cells and the walls of the valve involves a maximum impaction force that would only occur in low viscosity, which is an important factor in the disruption of yeast cells .
Some studies have shown degradation of high molecular weight compounds subjected to strong shearing forces in a viscous flow [36, 57]. Floury et al.  found that frictional forces encountered by the high fluid viscosity during high-pressure homogenization induced irreversible degradation of long molecules. Undeniably, there are several authors that supported the hypothesis of which cell disruption under high pressures is independent of biomass concentration [19, 37]. Additionally, a non-linear correlation between cell disruption capability and the specific protein release was observed for cell disruption process. This might be attributed to the influences of the effects from the variables studied. A similar condition was Foster  in the study on the effects of biomass concentration where increasing cell concentration was found to reduce cell disruption efficiency measured in optical density. However, on assaying the product release, it was found that product concentration actually increased up to 10-fold. Thus, it was concluded that measurement on cell disruption capability alone would not sufficiency evaluate disruption efficiency. Further studies and a careful analysis must be performed in order to clarify the relationship that the cell disruption capability exerts on the release of specific protein.
Also in our case, it is important to address the temperature rise in high pressure homogenization; as a significant rise would cause detrimental effects on process samples. Notably, energy generated from high pressure homogenizers are typically dissipated as heating of the process fluid . The homogenization experiments were performed at an initial temperature suspension ranging from 4-20°C with the aim to exclude the possible thermal effects. During the cell disruption process with increased volume of 2 L, temperature was seen to rise from an initial 4°C start-up. However, it was able to be maintained at ± 20°C throughout the 20 passes with the integrated heat exchanger attached to water bath for the Avestin homogenizer. The temperature maintained was well below temperature zone where heat inactivation or protein degradation can become involved. Wuytack et al.  demonstrated that temperature was able to be maintained at ± 18°C during cell disruption process at 3,000 bar using an Avestin homogenizer and is consistent with other high pressure homogenizers where prolong duration of cell disruption process at 1,750 bar for 1 h may still be sustain a temperature of ± 14°C . Interestingly, it was observed that further up-scaling have considerable little effect on the results obtained from the process samples. HBsAg released was demonstrated to have retained its antigenic properties (Figure 6). This was however not unexpected, as the same parameters were used in the up-scaling process with the only exception of volume difference. This is in agreement with the findings of several authors that reported for a fixed pressure and number of passes through the homogenizer, percentage of protein release is unaffected by the type and the scale of operation [28, 35]. Under the conditions explored here, the recovery of HBsAg could be performed with different conditions depending on the requirements of the objectives. For instance, high selective product recovery would be generally preferred for further downstream processing as it reduces the impurities in chromatography system and also easier for centrifugation and dead end filtration as the cells are still intact but at a cost of lower specific protein release.