Growth media is essential to any form of microbial culture, providing the nutrition required for optimal growth. There are a number of different options available, with Lysogeny Broth (LB) (Bertani 2004), Super Optimal Broth (SOB) (Hanahan 1983) and Terrific Broth (TBr) (Tartof 1987) being the most commonly used. However most are rich and undefined media, containing extracts such as yeast or beef that have an unquantifiable and highly variable composition. These extracts are also generally more expensive, complicate recovery and, due to their variable composition, result in significant batch-to-batch variation (Lee 1996; Moser et al. 2012). In the literature researched, TBr contributed to the highest amount of culture growth with Escherichia coli, with Losen et al. (2004) stating that TBr lead to an increase of 5x biomass when compared to LB. Islam (2007) produced similar results, with a significantly higher soluble protein yield in TBr than LB. This was put down to having glycerol as a defined carbon source. Furthermore, it is suggested that glucose is a poor choice due to E. coli excreting acetic acid as a by-product of glucose consumption, lowering pH and reducing growth (Islam et al. 2007; Losen et al. 2004; Marini et al. 2014). Glucose however is not the only issue. Singh et al. (2017) suggests that the carbon and nitrogen source are the most important components of the media as they can affect the type and amount of product produced. Other studies have concluded that E. coli develops a media history, adapting to different medias over time, showing variations in ribosome and RNA polymerase efficacy due to the medias amino acid makeup (Ehrenberg et al. 2013; Paliy and Gunasekera 2007).

Inorganic ions can also play an important role in the growth of cultures. Studier (2005) carried out an exhaustive study on inducer effects in media, investigating a number of variables as well as the presence of inorganic ions. The data collected showed that phosphate promoted kanamycin resistance, while sulphate supported optimum growth. However, on the contrary limiting magnesium concentrations allowed the cell culture to grow to a higher OD600.

It is known that with an increased amount of cell growth, there is generally a higher yield of recombinant protein (Khan et al. 2009). If E. coli are made to produce protein at too high a rate however, inclusion bodies will form which is deemed inefficient due to the complex process of refolding them into functional proteins (Marini et al. 2014). Marini and colleagues (2014) showed that even with higher cell density, functional protein expression had no change. Therefore, there must be a point of optimal cell growth that provides the highest amount of functional protein whilst causing the least amount of inclusion body formation or incorrect protein folding. Antibiotic selection can also impact on the protein yield. Using kanamycin in higher concentrations has been shown to increase plasmid stability and allow maintenance of higher plasmid copy numbers due to selection pressures (Kelly et al. 2009).

Control of pH is essential for growth mediums. All forms of bacterium have optimum pH’s, even at the extremes (acidophiles and alkaliphiles). However, in current synthetic biology many of the used chassis are generally classified as neutrophiles, so maintaining a pH of between 6-8 is essential. Presser et al. (1997) carried out an in-depth study of E. coli growth rates modelling the growth as a function of pH and lactic acid concentration. From this, E. coli was determined to have a pH boundary of 4.0, beyond which resulted in no growth. Lactic acid was found to be inhibitory in high concentrations, something required to consider during scale-up. This links back to aforementioned studies that showed the importance of carbon source, with glucose instigating a drop of pH and inhibition of E. coli growth (Islam et al. 2007; Losen et al. 2004; Marini et al. 2014). However, for very niche experimentation, suboptimal pH may play an important role in experimental design. Maurer et al. (2005) discussed how pH regulates a number of genes, including flagellar motility, catabolism and oxidative stress in E. coli. High pH 8.7 was found to repress membrane proteins, chemotaxis and flagellar motility. Low, acidic conditions of pH 5.0 were found to increase metabolic rates. In conclusion, experimentation entailing genetic circuits and protein production can be said to be drastically affected by pH. pH therefore must be defined in the experimental method for reproducible results.

If synthetic biology is going to follow a BDA approach, robotics will need to be implemented into microbial growth workflows. For high-throughput, this requires the use of multi-well plates and much smaller volumes than standard batch microbial growth techniques. This opens up an entirely new area of irreproducibility and standardising these experiments is therefore crucial to further understanding. Additionally, in a study of 49 papers in the field of synthetic biology it was estimated that upwards of 80% of papers did not provide complete information to fully reproduce their data (Chavez et al. 2017).

In microtiter plates, there are two main variables that need to be optimised; the oxygen transfer rate (OTR) and in turn, the overall mass transfer (KLa). Both of these variables have a significant inhibiting effect on microbial growth if not optimised. Fortunately in microtiter plates, both variables can be optimised in tandem by employing the same methods. OTR has been found to increase with increasing well size and a decrease in fill volume, as would be expected due to the reliance of surface aeration (Running and Bansal 2016; Hermann et al. 2003; Schiefelbein et al. 2013). Hermann and colleagues (2003) presented a clear correlation that a decreasing fill volume results in an increased OTRmax, that is only accentuated with increasing shake speeds. However, increasing the media viscosity can also decrease the OTR (Giese et al. 2014; Klöckner et al. 2013; Running and Bansal 2016), so shaking and baffling are essential to the optimisation of growth. If multiple wells have different viscosities a compromise must be made.

Shaking in microtitre plates needs to surpass the critical shaking frequency, whereby the centrifugal force exceeds that of the interfacial surface tension (Hermann et al. 2003; Kensy et al. 2005). Funke and colleagues (2009) states that below 500 rpm, this is not reached and no significant increase in OTR is seen. Unfortunately the data is not shown, but due to their significant OTR increases from 500-1000 rpm, this suggests it is reliable. The shaking diameter for their results only covers 3 mm, but previous work has conflicting results, using a larger shaking diameter and lower rpm OTR. In two papers, a 300 rpm and shaking diameter of 50 mm was shown to increase OTR significantly, with a shaking diameter of 25 mm showing a 3x decrease when compared with 50 mm (Duetz et al. 2000). However there was some splashing in larger wells (Duetz and Witholt 2004). On the contrary, Hermann et al. (2003) found that a shaking at 300 rpm at 25 mm produced no significant difference in OTR than if not shaken but also confirmed that any higher than 400 rpm at 25 mm would cause liquid spillage.

Baffling changes the flow characteristics of wells, increasing the turbulence and mixing. In microtitre plates, baffling is not standardised as in shake flasks, so the amount of laboratory’s with access to intentionally baffled microtiter plates is limited. Baffling in microtitre plates can also increase the chance of ‘out-of-phase phenomena’, where the flow of liquid creates an unmixed space at the bottom of the well (Büchs et al. 2001). Funke et al. (2009) designed 30 different well shapes, with a gradually increasing number of edges/baffles. From this, standard spherical wells were found to have the worst OTRmax and KLa, whilst the novel 6 edged petal shape allowed maximal OTR. Realistically, the 6 edged petal shape used is not commercially available so it would be unacceptable to suggest this use is common place. Despite this, other research groups have found that square wells have a baffle-like effect (Duetz 2007; Duetz and Witholt 2004; Hermann et al. 2003).

Wells are usually covered to prevent evaporation throughout experimentation. The most common are oils, lids, stickers and seals. The type of cover used can significantly affect the growth rates and overall experimental data (Chavez et al. 2017). Oil has been shown to prevent evaporation entirely, however it significantly lowers the OTR and reduces protein expression, making it sub-optimal for most synthetic biology uses (Chavez et al. 2017). Chavez and colleagues (2017) found that lid and sticker covers allowed for the greatest OTR and protein expression and that lid covering also caused the highest evaporation rate, specifically in the four corner wells. Sealing the plate is another option, however this method has only been to shown to reduce the OTR with minimal evaporation prevention (Zimmermann et al. 2003; Sieben et al. 2016).