Background

Despite heterologous protein expression being widely used in biotechnological research, optimization of protein expression protocols can be a very lengthy and costly process (Bill, 2014; Rosano and Ceccarelli, 2014). Once determined, optimal expression conditions are not readily translated to other fermentation scales or proteins of interest, thereby limiting method transfer to industrial scales, as well as increasing process development timelines (Mühlmann et al. 2017). After protein expression, many applications often require protein purification that calls for using expensive specialized equipment. Specifically, cell lysis for protein extraction is still mostly performed using mechanical methods, such as sonication or homogenization, which may not yield consistent results and limit high-throughput applications (Foster, 1992; Cai et al., 2008). Alternative methods, such as chemical methods orengineering “autolysis” microbial strains that lyse after cell harvesting, exist but are poorly controlled and do not account for oligonucleotides removal to reduce sample viscosity, which can complicate sample handling (Chien and Lee, 2006; Cai et al., 2008).

To address these, we recently described a two-stage protein expression process (Menacho-Melgar et al., 2020b). In this method, we use engineered strains that do not produce organic acid byproducts (e.g., acetate), known to negatively affect cell growth and protein expression (Jian et al., 2010), in combination with plasmids, where the expression of the protein of interest is under the tight control of a low phosphate inducible promoter (refer to Figure 1 for an example annotated plasmid sequence) (Moreb et al., 2020). In two-stage fermentations, protein expression is triggered upon entry into the stationary phase, when phosphate concentration in the media becomes depleted, which (i) stops cell growth (stage 1) and (ii) induces protein expression (stage 2) (Figure 2A). Two-stage fermentations decouple cell growth from protein expression, which results in high protein titers and enables the use of already optimized standard protocols across different scales and proteins of interest (Burg et al., 2016; Decker et al., 2020). In other words, two-stage protein expression does not require optimizing induction conditions, as opposed to methods that induce protein expression during growth. Additionally, this approach has enabled higher protein titers than growth-associated protein expression.

Figure 1. Example two-stage expression plasmid. A. Expression of the protein of interest (GFPuv, green) is driven by a low-phosphate inducible promoter (yibDp, blue). The plasmid has the high copy pUC origin (pBR322 derivative, purple) and a kanamycin resistance marker (yellow). B. Partial yibD promoter sequence showing the -10 and -35 boxes (highlighted in boxes) and the start of transcription (capitalized). Start and end bases for each feature are annotated. For a complete plasmid sequence, see https://www.addgene.org/browse/sequence/271197/ under the ‘Sequence’ tab.

Figure 2. Autolysis and Autohydrolysis in two-stage protein expression. A. In stage 1 “growth”, cells are grown in AB-2 media. In stage 2, “production” gets triggered upon phosphate depletion in the media, when cells enter the stationary phase, activating expression of the protein of interest. B. Expression of the autolysis/autohydrolysis machinery, that is chromosomally integrated and consists of benzonase and lysozyme, also occurs at this stage. C. After induction, lysozyme remains inside the cytoplasm and benzonase is located in the periplasmic space. D. Only upon addition of detergent (Triton-X) and a freeze-thaw cycle, the cells lyse and the endonuclease is able to digest the host’s DNA. Reprinted with permission from Menacho-Melgar et al. (2020a). Copyright 2020 John Wiley and Sons.

Additionally, we built upon two-stage protein expression and engineered an autolysis/autohydrolysis strain, which expresses a lysozyme (from the lambda R gene) in the cytoplasm and a nuclease (from S. marcescens nucA gene, commercially known as Benzonase^TM) in the periplasm, during the protein production stage (Figure 2A-2C) (Menacho-Melgar et al., 2020a). These chromosomally integrated enzymes do not negatively affect the titer of the protein of interest (Menacho-Melgar et al., 2020a). Since this system is tightly controlled, no cell lysis will occur during the growth phase or after harvesting, as opposed to other BL21-based autolysis strains (Menacho-Melgar et al., 2020a). Only upon resuspension in a lysis buffer containing 0.1% detergent and a freeze-thaw cycle, over 90% of the protein is released (compared to a 20-minute sonication protocol, where further sonication does not lead to more protein release) and all of the oligonucleotides in the cell lysate are removed (Figure 2D) (Menacho-Melgar et al., 2020a). Importantly, this method enables the simultaneous lysis of several samples, which can be essential in high-throughput applications (Menacho-Melgar et al., 2020a). Although this method has been validated across different fermentation scales (Menacho-Melgar et al., 2020a), fermentations in microtiter plates and bioreactors require specialized equipment and accessories not readily accessible, whereas fermentations in shake-flasks are more widely and routinely used. Thus, in this protocol, we have described a detailed protocol for performing two-stage fermentations only in shake flasks, as well as protein extraction using our autolysis/autohydrolysis strain and a plasmid expressing GFPuv (Addgene, #127078), as an example (Menacho-Melgar et al., 2020a).