A reduction in temperature was previously shown to improve phage titres in bioprocessing (Greico et al, 2012, Ali et al, 2019), with temperature reduction hypothesised to aid efficiency in viral protein synthesis (Bleckwenn et al, 2005). However, the effect of the method of temperature reduction at the point of infection remains unclear. By examining three methods of temperature reduction (methods 1-3) this study aimed to provide insight into culture maintenance at the point of infection for improved infection kinetics and phage titres in bioprocessing. Within the study, host cell density (~1x107-1x109 cfu/ml) and a range of reduced temperatures (22.3-28ºC) at the point of infection were also investigated for effects on infection kinetics and harvest titres. Additionally, experiments in stirred tank bioreactors provided an insight towards scale-up of phage bioprocesses.
Small scale evaluation of temperature reduction method
In bench scale phage infection experiments, the simplest method of applying temperature reduction is to allow the shaking incubator to reduce to the desired temperature once culture infection has taken place. However, this will not be an applicable method for a scalable bioprocess for large scale production of phage.
To provide information on the best method for temperature reduction at bench scale (20ml shake flask cultures) with a view towards informing development of scalable culture processes, three methods of culture maintenance during early infection and temperature reduction were investigated. The methods of temperature reduction were Method 1, culture maintained in the shaking incubator during temperature reduction (from 37 – 28ºC); Method 2, culture maintained at room temperature (22ºC) during shaking incubator temperature reduction (to 28ºC); and Method 3, culture maintained in a static 28ºC incubator during shaking incubator temperature reduction (to 28ºC). These methods were applied to two different host-phage culture processes; E. coli – T4 phage and S. aureus – Phage K. The temperatures of both the cultures and the environment were monitored during early infection and the results are shown in Figure 1.
Under Method 1, the shaking incubator was shown to take 8 minutes to reduce from 37ºC to 28ºC, whereas the culture temperatures of both organisms reached the desired 28ºC temperature 14 minutes after infection and stabilised by 18 minutes post-infection (Fig 1, A-B). The observed slower rate of culture temperature reduction compared to environment reduction was expected due to the heat transfer from incubator to shake flask culture. Under Method 2, as expected culture temperature reduced more rapidly when maintained at room temperature during early infection. These cultures achieved desired culture temperature at 5 minutes post-infection (Fig. 1 C-D) and were then transferred back to the shaking incubator to complete the infection cycle. Under Method 3, when cultures were maintained in a pre-set 28ºC static incubator for a slower and more controlled temperature reduction profile compared to method 2, the desirable culture temperature was achieved at 11 minutes post-infection. Temperature shock to the culture can have a negative impact on the host cells or phage and therefore it may be more desirable to have a more controlled initial reduction in temperature (Jończyk et al, 2011).
Previous work has shown that by adapting the conditions at which the phage infection takes place; MOI, agitation, time and the temperature of infection, a consistent phage titre of ~5x1013 and 1x1012 pfu/ml can be achieved for T4 phage and phage K respectively (Ali et al, 2019). The data presented in figure 2 shows increases in harvest titre can be achieved by altering the reduction temperature profile during early infection i.e. method 3 significantly improved harvest phage titres for both host-phage systems. Method 3 consistently achieved titres of 7.89 x 1013 ± 1.9 x 1013 pfu/ml and 1.03 x 1013 ± 1.77 x 1013 pfu/ml for T4 and phage K respectively. There was a statistically significant increase in harvest titres between each method (1-3) for T4 phage and method 2-3 only for phage K (p<0.05). It is unclear as to why reducing the temperature at the point of infection improves phage titre across two different host-phage systems. However, to elucidate some of the underpinning mechanisms at the point of infection that may be affecting these differences in final titre and process yield, the phage infection kinetics in terms of adsorption and burst size were assessed.
Within bacteriophage fermentation literature, titres of 1 x 1012 pfu/ml are currently some of the greatest achievable levels (Bonilla et al, 2016). Previous studies have shown that optimising conditions at which phage fermentation takes place can improve the titre achievable (Greico et al, 2009, Greico et al, 2012). However, by examining the conditions at the point of infection this study has shown a further increase by 1 order of magnitude in harvest phage titres depending on the method of initial temperature reduction during early phage infection (method 3). These methods and titre gains are applicable to small scale fermentations and will help to inform scalable fermentation. Whilst no previous work has examined a similar mechanism, a hypothesis for the improvement in titre may be if the optimal temperature for T4 and phage K infection is 28oC, maintaining the culture at the most desirable temperature could improve titre by slowing intracellular processes involved in viral propagation, however this is yet to be determined experimentally. Additionally, by preventing a shock to the culture i.e. significant temperature drops which could affect host cell metabolism, efficiency of replication and the ability of phage to propagate. This hypothesis is somewhat supported by the data shown here where reducing the potential for cellular shock using a slower and more controlled temperature reduction profile improved phage titre, but requires further investigation to determine the biological mechanism of action.
Adsorption and burst analysis
An important study in the kinetics of phage infection is the adsorption of the phage to the host cell and the burst size. It is known that phage adsorption is dependent on the culture conditions (Zaburlin et al, 2017) and differences in phage adsorption kinetics may offer some mechanistic explanation for the differences in titre achieved with the various methods of temperature reduction at the point of infection. Figure 3 shows the adsorption of phage to their host cell for each of the infection methods (1-3). Figure 3A shows that each of the methods had a negligible difference in the adsorption of T4 phage for the first 6 minutes. Thereafter, a statistically significant difference was observed in the rate of adsorption between methods 2 and 3 compared to method 1 where the culture was agitated throughout early infection. There was very little difference between the cultures without agitation (methods 2 and 3) that were held at room temperature or in the static 28ºC incubator respectively. Figure 3B shows a greater degree of variation was observed in the phage K adsorption compared to the T4 phage adsorption. A significantly greater rate of adsorption was observed in incubated cultures (methods 1 and 3) compared to cultures held at room temperature (method 2) from 3 minutes post-infection onwards (p<0.05). Agitation appears play less of a role in influencing phage K adsorption compared to T4 phage adsorption, with no significant difference observed between 7-10 minutes post-infection between methods 1 (agitated) and 3 (static) for phage K whereas only the agitated method influenced T4 phage adsorption.
A further important study in bacteriophage kinetics is the burst size, the increase in phage after the bacterium has been infected with phage (Wang, 2006). Table 1 below shows the average burst size achieved from each phage infection for each of the three temperature reduction methods. Burst size for method 3 was statistically significantly increased compared to methods 1 and 2 (p<0.05) for both phage-host systems. A lower average burst was observed for method 2 compared to method 1 for both phage-host systems although only statistically significant for phage K. This lower burst may be related to reduce control in the temperature reduction profile and possible temperature shock to the culture with method 2 where cultures were removed from 37oC incubation to room temperature (22oC). The temperature reduction during early infection is more gradual for method 1, achieving the desired 28oC culture temperature over 14 minutes compared to 5 minutes for method 2.
Fister et al (2016), previously examined the effect of environmental factors on the adsorption and yield of phage, their data showed how reducing the pH or the addition of salts can negatively influence the adsorption rate of Listeria phage P-100 although it wasn’t significant. The data presented in this study shows that culture temperature during early infection and the method of exposure to a change in temperature can influence adsorption rates in multiple host-phage systems. Although there are differences between the level of influence of temperature on the different host-phage systems, for example different burst sizes for T4 vs. phage K with method 3 temperature reduction and differences in response to agitation for T4 phage. The data shows evidence that a gradual reduction in temperature positively influences adsorption rate and burst size during early infection. Other environmental factors, such as MOI, pH and agitation may also have a significant effect on the phage-host adsorption (Silva & Sauvageau, 2014).
An additional aspect when assessing the point of infection is the growth of the host and the point in the host’s growth cycle at which phage infection takes place. It is established in literature that phage infection should be introduced within early logarithmic growth phase and OD600nm 0.2-0.4 is commonly used (Bryan et al, 2016, González-Menéndez, et al 2018).
Higher host cell densities result in lower harvest phage titres being achieved, thought to be related to fewer phage interactions with host cells and reduced levels of infection, whilst lower host cell densities are too dilute and do not contain sufficient cell numbers to effectively propagate phage (Kasman et al, 2002, Kick et al, 2016). However, it is unclear whether there is an optimal point within this established cell density range nor whether this potential optimal point differs between host – phage systems. Establishing standardised methodology with control points such as at the point of infection will help to inform standardised bacteriophage manufacturing processes with minimal variation in desirable outputs. E. coli and S. aureus cultures were grown to a range of optical densities within and outside the pre-determined range (covering the growth cycle) before phage infection with T4 and phage K respectively, to assess the effect of infection point and the effect on final phage titre and burst size in moving towards a more standardised infection point for these systems.
Figure 4 shows the greatest titre for each phage was observed at an infection point of 0.25OD600nm and the titre achieved was statistically significant higher than the titres achieved from adjacent infection points, 0.3OD600nm and 0.2OD600nm, for both T4 (p<0.001; p<0.003) and phage K (p<0.004; p<0.004) and for all other infection points. At 0.25OD600nm, a titre of 2.2x1013 and 2.5x1011 was achieved for T4 and phage K respectively. The phage burst sizes were calculated for each infection and show an expected clear trend in alignment to the titre results with an increase in burst related to increased output titre. Distinct peaks in burst size aligned to the greatest phage titres achieved at 0.25OD600nm for each phage. Statistically significantly lower burst sizes were achieved between 0.25 and 0.2/0.3OD600nm and 0.25 - 0.05/0.5/1OD600nm for both phage, p<0.05. The results show the importance of host cell density at the point of infection upon maximising output titres and that parameters such as host cell density may be comparable in their optimum for multiple host-phage systems. Although individual phage may respond differently to the point of infection, it is plausible to consider that 0.25OD600nm could be used as a standardised host density at the point of infection for E. coli or S. aureus host production systems. Whether a standardised host density could be applied more widely to further species or host-phage systems would require experimental determination.
An additional experiment was carried out to determine the effect of phage infection once the host culture had stabilised at 28oC (under standard conditions phage inoculation occurs immediately prior to temperature reduction). Previously, figure 1 showed that under Method 1, where cultures were kept in a shaking incubator during temperature reduction, it took 14 minutes for cultures to stabilise at 28°C. During this experiment, 0.25OD600nm cultures were inoculated with phage once they were stable at 28°C following Method 1 temperature reduction. The experiment was carried out in triplicate and with duplicate phage plaque assay titre measurements (n=6) and single burst size measurements (n=3). The results showed there was no significant improvement on the T4 or phage K titre, 8.17x1012±4.05x1012 and 2.67x1012±8.73x1011 pfu/ml. T4 phage gave a more variable and reduced burst size of 107.7±19.9 whilst phage K gave a reduced burst size of 72.1±6.2 when compared to the bursts achieved at 0.25OD600nm above (Fig. 4). The results suggest there is no advantage in postponing phage inoculation for culture temperature reduction to stabilise and taken together with the previous phage adsorption results (Fig. 3) that inoculation followed by temperature reduction may be more beneficial to the very early phage infection process.
Scale up model evaluation of the temperature kinetics
The small-scale shake flask experiments allow for high throughput approach and fewer resources when compared to larger stirred-tank bioreactor systems. However, small-scale shake flasks are not fully representative of the industrial-scale systems that will be employed for the production of bacteriophage, given the differences in fermentation dynamics relating to increased volume and impellor-based agitation. Fermentation of bacteriophage is moving towards larger scale STR cultures rather than shake flask culture (Agboluaje & Sauvageau, 2018) as well as other scalable methods and as such it is important to examine small scale outcomes informing bioprocess development in relevant scalable culture systems. This type of study is also important to examine the differences between the culture systems in terms of bioprocess outcomes to generate knowledge on the advantages and limitations of the translation.
Although the greatest titre was achieved using Method 3 in the small-scale experiments, this method of static incubation is not practicably replicated in a 5L bioreactor. However, further experimentation beyond the scope of this study is required to investigate the effects of stirring and agitation dynamics on early phage infection. The Method 3 results suggest that suspending agitation during early infection / temperature reduction may be beneficial for phage adsorption, burst and related harvest titres in scalable bioreactor cultures.
Initially, the process from the small-scale Method 1 experiments was replicated in a scaled up stirred tank bioreactor model. Bioreactor cultures were maintained at 3L working volume with 100% dO2, pH7.0 and 225rpm / 150 rpm impellor stirring for E. coli / S. aureus respectively. Cultures were infected with phage at 0.25OD600nm, MOI of 2.5 / 0.1 respectively and immediately cooled from 37ºC to 28oC. Figure 5 shows the difference in phage adsorption during early infection between the shake flask and bioreactor culture systems. The shake flask cultures generally gave a greater rate of adsorption for phage K (Fig. 5B), whereas T4 phage showed initially there was a higher level of free phage in the STR cultures but after 7 minutes there was less free phage in the shake flask model (Fig. 5A). The experiments were completed in duplicate and each sample enumerated with duplicate plaque assays. Figure 5 C and D show the temperature reduction in the stirred tank reactor culture. They show that the culture took around 14 minutes to reduce to the desired 28oC temperature. The action of the water jacket and agitation will allow the heat transfer throughout the culture and consequently the reduction in temperature. Given that ultimately, manufacturers may look towards continuous culture, future experiments may benefit from replicating this process in a scale up continuous system, i,e feeding cells into a 28oC reactor.
It was important to note the final titres achieved in the experiments in the 5L bioreactor. The experiment was completed in duplicate with triplicate plaque assay measurements and duplicate burst size measurements. There was a statistically significant increase in titre between method 1 in shake flasks and the 5L scale up model for T4 phage, 5.11 x 1012 ± 1.5 x 1012 pfu/ml and 4.33 x 1014 ± 9.4 x 1013 pfu/ml p<0.01 using an ANOVA whilst phage K gave a statistically significant increase between method 1 in shake flasks and the 5L scale up model for phage K 1.1x1012±1.02x1012 and 8.5x1012±8.99x1012 pfu/ml respectfully, p=0.001 using an ANOVA. This increase in titre between the scale up model and the shake flasks can be accounted for by the change in system used. The stirred tank system offers the advantage of a more controlled environment that the infection can take place in which will consequently an increase in titre is seen. Previous work has shown the improvement in titre between the two culture systems using the same conditions, given the more controlled environment (Ali et al, 2018). Figure 1 showed that the culture took around 14 minutes to reduce from 37oC to 28oC and stabilise, similar to the scale up model. The graphs show a statistically significantly higher rate of adsorption in the first 6 minutes for T4 phage in the shake flask compared to the 5L bioreactor, figure 5A and B. It is interesting to note this as although the conditions were kept the same, the agitation between the culture systems will differ despite the same agitation rate being used, 225rpm E. coli, 150rpm, S. aureus. Additionally, phage K showed a statistically significantly higher rate of adsorption in the shake flask after 2 minutes of infection between each time point p<0.05 using a paired t test. The sheer damage to the host cells and difference in mixing patterns seen in the STR due to the impellor may cause differences in the adsorption between systems (Borys et al, 2018). The graphs show that after 10 minutes, around 99% of phage had adsorbed to their host cell. The difference in the mixing between shake flasks and STR, may affect the adsorption of phage to their host organism and it was interesting to note the differences seen between the culture systems. The harsh agitation caused by impellors may account for the reduced higher free phage in the stirred tank culture despite the same agitation rate used across the systems (150rpm – S. aureus/phage K, 225rpm E. coli/T4 phage).
The duplicate experiment in the STR gave an average titre of 4.33x1014 ± 9.4x1013 pfu/ml and 4x1012 ± 1.41x1012 pfu/ml for T4 phage and phage K respectively. The advantage of using a scale up model is two-fold. Firstly, the increase in yield between the shake flask and STR model within their respected bioprocesses and secondly the improvement in maintenance of conditions which improves the overall bioprocess. This can be seen from the titre achieved as in the shake flask model, figure 2, the greatest titres achievable were 7.89x1012 ± 1.96x1012 and 1.03x1013 ± 1.77 x1012 pfu/ml for T4 phage and phage K respectively. It is interesting to note the burst size achieved in table 1. In the 5L bioreactor, T4 gave a duplicate average burst size of 97.45 ± 7.3 whilst phage K gave an average burst size of 106.1 ± 8.4. The statistically significantly increased burst in the shake flask whilst kept in a static incubator, compared to the 5L for T4 is interesting to note and could be accounted for in the difference in mixing or pH control. Although no significant difference was seen in the phage K burst, a higher titre was achieved in the 5L culture compared to the shake flask culture. The maintenance of the conditions in the STR would allow an overall higher titre to be achieved when the experiment was run to completion. One study has previously shown that reducing the temperature of infection gave a higher titre at 25oC compared to 37oC (Shan et al, 2014). Additionally, Vasina & Baneyx (1997) showed that reducing the temperature from 37oC to 30oC after E. coli was grown to mid log phase allowed greater protein production to be found. Hypotheses were given stating that reducing the temperature allowed metabolic processes to be extended and therefore the host organisms could replicate for longer periods of time. Monoclonal antibody titres have also been shown to improve when grown at 31oC compared to 37oC (Boigard et al, 2018). Previous work has shown significantly improved yields using an STR over shake flask due to pH and oxygen control. Changes in pH will affect the ability of the host organisms to propagate themselves and the phage (Nobrega et al, 2016). Moreover, there will be differences in the mass transfer within the culture, between the shake flask and 5L STR, which may also cause gradients again limiting the propagation ability which can be mitigated in bioreactors as they have less mass transfer limitations (Klockner & Buchs, 2012). Given the length of the culture and the improved control in the process throughout, compared to the shake flask model, it may help to explain why greater titres and burst sizes were seen in the STR.