High-throughput preparation methods of crude extract for robust cell-free protein synthesis

Crude extract based cell-free protein synthesis (CFPS) has emerged as a powerful technology platform for high-throughput protein production and genetic part characterization. Unfortunately, robust preparation of highly active extracts generally requires specialized and costly equipment and can be labor and time intensive. Moreover, cell lysis procedures can be hard to standardize, leading to different extract performance across laboratories. These challenges limit new entrants to the field and new applications, such as comprehensive genome engineering programs to improve extract performance. To address these challenges, we developed a generalizable and easily accessible high-throughput crude extract preparation method for CFPS based on sonication. To validate our approach, we investigated two Escherichia coli strains: BL21 Star™ (DE3) and a K12 MG1655 variant, achieving similar productivity (defined as CFPS yield in g/L) by varying only a few parameters. In addition, we observed identical productivity of cell extracts generated from culture volumes spanning three orders of magnitude (10 mL culture tubes to 10 L fermentation). We anticipate that our rapid and robust extract preparation method will speed-up screening of genomically engineered strains for CFPS applications, make possible highly active extracts from non-model organisms, and promote a more general use of CFPS in synthetic biology and biotechnology.

To produce proteins, E. coli S30 extract-based CFPS systems harness an ensemble of catalytic components from crude lysates of cells. Cellular lysates contain active enzymes, as well as helper factors for transcription, translation and protein folding. Activated catalysts within the cell lysate act, in coordination, as a working chemical factory to synthesize and fold desired protein products upon incubation with essential substrates (e.g., amino acids, nucleotides, DNA or mRNA, energy substrates, cofactors, and salts). A variety of cell extract preparation methods have been developed for CFPS. In general, cells are harvested during mid-exponential growth of a low-density fermentation and then processed by high-pressure lysis (,20,000 psig), two 30,000 3 g centrifugations (to remove cell wall fragments and genomic DNA), a run-off reaction (incubation of the clarified extract but no exogenous mRNA or DNA), and dialysis (to provide a suitable storage buffer) 42,43 (Figure 1). Although first described in 1963 44 , many minor changes have been made to the original extract preparation procedures to streamline the process, improve productivity, and reduce background expression (Figure 1) 45,46 .
However, the cell extract preparation procedure generally remains a time and labor-intensive process. In addition, extract preparation procedures have been difficult to standardize, leading to different extract performance within and across laboratories. Furthermore, since most extract preparation procedures involve a large volume of cell culture (.1 L) and costly high pressure disruption equipment to lyse cells (e.g., French Press or impinge homogenizer), obtaining and utilizing a large number of cell extracts from different source strains is not yet routine. This may be important for studying the systems effect of numerous genomic modifications to identify negative and positive effectors of protein synthesis for basic and applied research.
To address some of the aforementioned issues, Shrestha et al. recently streamlined a CFPS extract preparation method using equipment common to most biotechnology laboratories ( Figure 1). Specifically, they used shake flask growth and sonication to reduce cost and variability, eliminating the need for specialized and expensive growth and lysis equipment 47 . This was a significant advance. However, their study was insufficient to provide general guidelines for optimal conditions needed to (i) lyse different cell suspension volumes, (ii) use different E. coli strains, and (iii) prepare lysates in a high-throughput manner from disposable mini-culture tubes.
Here, we sought to build off the Shrestha et al. method to develop a rapid, robust, and high-throughput extract preparation procedure using sonication that could reduce the time to make numerous extracts to ,1 day per 100 extracts, and do so consistently. The goal was to be able to generate ,100 mL of extract (which is sufficient for ,25 CFPS prototyping reactions at the 15 mL scale). We expected this could be achieved at the 10 mL culture volume scale. Our technology development study involved three steps. First, we explored the effect of a numerous sonication variables (energy input, sonication burst time, cooling time, etc.) on extract performance in CFPS. Second, after fixing the aforementioned sonication parameters, we mapped CFPS yields as a function of sonication energy input versus cell suspension volume. Specifically, we generated 144 cellular extracts of a commercially available E. coli strain BL21 Star TM (DE3) and a genomically recoded strain from strain K12 MG1655. Looking at multiple strains enabled us to assess strain differences on lysis and extract performance and ascertain the potential for our protocol to be general. Third, concerning the robustness of our method to make active crude cell extracts, we characterized our approach across a wide range of culture volumes (from 10 mL culture tubes to lab scale fermentation (10 L)). We additionally showed the ability to prepare lysates across multiple extract volumes (100 mL to 30 mL) with identical CFPS performance. In sum, our work yielded an efficient and accurate sonication method that has the potential to be applied to any strain of E. coli. Furthermore, it cataloged the impact of optimizing multiple extract preparation procedures together, which is expected to provide utility for developing CFPS systems from many prokaryotic species (e.g., Streptomyces, Pseudomonas, etc.).

Results and Discussion
Preparation of crude cell extract by sonication. We began our investigation by assessing the impact of total sonication energy  input on cell lysis and crude extract performance, which we defined by the total amount of protein produced. To do this, we first cultured 1 L of BL21 Star TM (DE3) cells in 2.5 L baffled tunair shake flasks, which showed similar growth rate as compared to 10 L fermentation (Supplementary Table S1). Then, thawed cell suspensions of different volumes were lysed using multiple sonication input energies (expressed in Joules), with a constant sonication burstcooling (at 4uC) cycle of 45 s-60 s (s: seconds). Cell resuspensions were not stirred during sonication. Next, crude S30 extracts were prepared from these cell lysates. Subsequently, 15 mL batch-mode transcription and translation (TX-TL) reactions were performed for 4 hours at 37uC to determine the amount of superfolder green fluorescent protein (sfGFP) synthesized. These experiments showed that extract performance depended greatly on the sonication energy input (Figure 2A). Our experiments suggest that if too little sonication energy is used, not enough cells are burst, which results in crude extracts with lower total E. coli protein concentration (Supplementary Figure S1). On the other hand, higher sonication energy levels sufficiently lyse the cellular suspension, but can also deactivate the catalysts present in the extract. Most likely, this is due to the heat shock introduced after many sonication-cooling cycles. Consistent with these design rules, we observed different optimal total sonication energy inputs in relation to different cell suspension volumes (556 J for 1.5 mL and 309 J for 0.5 mL).
We hypothesized that a given cell suspension volume with the same total sonication energy input would exhibit the same protein biosynthesis performance if heat shock was avoided. In other words, a consistent amount of energy is needed to lyse a defined amount of cells. We therefore investigated sonication-cooling intervals for preparing small volumes of cell extract, being careful to avoid any detrimental effect on protein synthesis due to an excessive flow of heat during cell disruption. In particular, to achieve 556 J of energy input for 1.5 mL cell suspension, we monitored four different cycles of the following sonication-cooling conditions; 40 s-60 s, 30 s-40 s, 20 s-25 s, and 10 s-10 s. Consistent with our hypothesis, as long as the total energy input per cell suspension volume was constant, we observed no significant difference in extract activity under different sonication burst periods and 4uC cooling intervals ( Figure 2B). These data suggest that highly active extracts can be produced with min-imum sample heating by identifying an optimum energy input per cell suspension volume and then use of a short sonication-cooling cycle (10 s-10 s).
Optimization of cell extract preparation procedure with different bacteria strains. After we defined a reproducible sonication cell lysis strategy to generate highly active extracts, we decided to perform a series of optimization experiments to identify conditions for robust and consistent extract preparation. We chose to explore the impact of several pre-and post-lysis steps, such as: culture harvest time, the ratio between wet cell pellet weight and buffer A, centrifugation speed, and run-off reaction time. In addition, we chose to study two strains (BL21 Star TM (DE3), as above, and also a genomically recoded derivative of K12 MG1655 (strain C495 48 )). This allowed us to see if our protocol was robust to both B-and K-strains of E. coli.
The composition of the cellular machinery at the time of harvest directly affects the CFPS potential of the crude extract. In E. coli CFPS systems, it has been shown that exponential growth phase is the best time to harvest the cells during culture because the translation machinery is most active 49,50 . However, cells harvested at a higher optical density result in a larger cell mass. In turn, leading to an increased quantity of total crude extract prepared per fermentation. Accordingly, we grew E. coli cells to 2.5, 3.1, 3.5, 4.0, and 5.5 OD 600 , which spanned a range of mid to late exponential growth in our culture media, and prepared individual batches of crude extract originating from each of these cultivations. Specifically, we prepared a set of 1.5 mL crude cell extracts with 556 J of energy input using sonication-cooling cycles of 10 s-10 s. Overall CFPS activity from BL21 Star TM (DE3) strain lysates was fairly consistent at any OD 600 points during exponential growth phase (OD 600 2.5 , 5.5) ( Figure 3A). In contrast, strain C495 showed that the most active extracts were obtained from culture harvested at mid-exponential phase (468 6 14 mg mL 21 for OD 600 2.5-3.5 compared to 292 6 2 mg mL 21 for OD 600 of 4.0-5.5) ( Figure 3A). While the reason for the significant impact of harvest OD on strain C495 remains unknown, the 40% reduction in activity might be related to the slower growth of these cells as compared to BL21 Star TM (DE3). C495 has a 30% increase in doubling time as compared to BL21 Star TM (DE3) (Supplementary Table S1). We went on to show that CFPS activities from both strain lysates were reduced if the cells were from crude extracts prepared with different sonication energy input (in Joules) in 1.5 mL and 0.5 mL cell suspensions. Numbers on the x-axis are total input energy. Fluorescence of sfGFP was measured following a 4 hour CFPS reaction. (B) Protein synthesis activities from crude extracts prepared with different sonication-cooling intervals. For each extract, the same sonication energy input value (556 J) and 1.5 mL cell suspension was used. Energy accumulation was monitored during lysis and then sonication was stopped when the total energy input is reached to 556 J. ''C,'' positive control, cell extract prepared by 10 L fermentation for culture and homogenization to lyse cells. sfGFP reported was based on the amount of active (fluorescent) sfGFP following a 4 hour CFPS reaction. 95% of the total sfGFP synthesized was active. Values represent averages and error bars represent standard deviation for at least 3 independent experiments. www.nature.com/scientificreports harvested in stationary phase (Supplementary Figure S2). Taken together, our data highlight the importance of cell harvest as a key parameter for optimization and batch-to-batch variability. Next, we compared the ratio of cell pellet to buffer suspension for preparing highly active cell extracts by sonication cell lysis. Ratios of 151 and 151.27 (wet cell pellet weight in g to volume of buffer in mL) showed the highest extract activity after lysis ( Figure 3B).
Beyond assessing the impact of pre-lysis steps, we also investigated post-lysis extract preparation steps. Specifically, we evaluated the impact of centrifugation speed on the activity of prepared extracts, as this represents a critical step in streamlining crude extract preparation. The centrifugation speed was not critically influential on the activity of cell extract for CFPS. We observed that speeds over 10,000 RCF (4uC for 10 min) for the first centrifugation provided no significant difference in activity among extract samples ( Figure 3C), noting that these speeds were sufficient to remove all un-lysed cells (see below).
We next evaluated the effect of the run-off reaction, which consists of supernatant incubation at 37uC with 250 rpm agitation for a specified time after the first centrifugation. Strikingly, lysate performance in CFPS by the two different E. coli strains showed very different activity. While extracts from strain BL21 Star TM (DE3) exhibited a gradual decrease of CFPS activity with extended run-off reaction times, those from strain C495 showed the run-off reaction time as an essential parameter to enhance CFPS performance ( Figure 3D). Notably, the run-off reaction step was followed by a second centrifugation (10,000 RCF at 4uC for 10 min), which we found to be important for high CFPS activity when using the run-off reaction. In addition, particular attention was taken to not transfer any cellular debris into the final lysate following the run-off reaction in order to maintain active cellular lysates. By avoiding cellular debris, we were also able to produce lysates without remaining E. coli cells. This requires care in separating the supernatant from the cell pellet at the expense of not collecting all of the lysate.
We next tested the possibility of overexpressing T7 RNA polymerase within the source strain for CFPS as previously reported 45 . Specifically, we compared CFPS activities between T7 RNA polymerase overexpressed BL21 Star TM (DE3) cell extract and purified T7 RNA polymerase that was exogenously supplied to extracts from an uninduced BL21 Star TM (DE3) strain (Supplementary Figure S3). Although T7 RNA polymerase overexpressed extracts showed slightly higher CFPS activity, purified T7 RNA polymerase was used for the experiments below so that we could compare results from both E. coli strains: BL21 Star TM (DE3) and C495. This is because the C495 strain lacks genomically integrated T7 RNA polymerase.
Fine mapping of sonication energy input versus cell suspension volume. After we evaluated all the steps in our sonication based cell extract preparation, we subsequently determined the ideal balance between sonication energy input and cell suspension volume. This represents a step forward compared to previous studies where cell suspension volume was kept constant 47 . Our goal was to reduce the overall volume for each step of the preparation procedure, from cell growth to crude cell extract preparation, such that one might imagine using culture tubes to prepare source cells for tens to hundreds of mutants and their cell lysates in high-throughput. The key point was to identify a correlation between sonication energy input and volume, such that we can predict an optimal parameter configuration to generate highly active extract from different cell strains and suspension volumes. To do this, we mapped the sonication energy input versus cell suspension volume by generating 144 extracts for the BL21 Star TM (DE3) and C495 strains. For each strain, we observed maximal energy inputs for each cell suspension volume when preparing highly active crude cell extract (Figure 4). For illustration purposes, 5 and 6 short sonication-cooling cycles (10 s-10 s) were needed at the optimum energy input for 0.5 mL cell suspensions from the BL21 Star TM (DE3) and C495 strains, respectively. While small volumes of cell suspension were very sensitive to energy input (i.e., the maximum values were tightly banded in Figure 4), large volumes of cell extracts showed a relatively high tolerance against high-energy input (Supplementary Figure S4). Thus, to obtain highly active crude cell extracts from the small volume, a precise titration of energy input must be imposed.
Notably, the two strains also showed different maximization profiles. The C495 strain has a broader tolerance of sonication energy inputs per cell suspension volume when compared to BL21 Star TM (DE3). Even though both E. coli strains showed different patterns, the correlation of energy and volume exhibited a linear trend in both strains ( Figure 4). In accordance with the linear trend, we were able to predict the optimal amount of energy input from user-designated volume for highly active extract. For instance, in follow-up experiments, we built a calibration line using only a few points to choose parameters for efficient extract preparation. As shown in Table 1, except for the smallest volumes (100 mL), predicted energy input for the highest activity of cell extract is well fitted to the experimental data (globally found to be within 5.2 6 4.4%). Next, we showed the ability to prepare lysates across multiple extract volumes (100 mL to 30 mL) with identical CFPS performance ( Figure 5A). Overall, these data highlight that our approach is robust and predictive. Moreover, it outlines a detailed, generalized method to characterize and prepare extracts from multiple strains using sonication-based lysis.
High-throughput cell extract preparation. Upon demonstration of a generalizable approach for maximizing extract quality by sonication cell lysis, we then examined the ability to make active crude cell extracts from small to large culture volumes (10 mL culture tubes to 10 L lab scale fermentation). This is important because our standardized protocol for small culture volumes makes CFPS an accessible tool to many researchers with inexpensive equipment. While our method does require a sonicator with the ability to measure energy input in Joules, we have identified many manufacturers are available to meet this requirement. In addition, our method is simple and can be executed in high-throughput fashion for high-throughput protein expression, testing bacterial strains harboring different genomic mutations, and perhaps for rapid design-build-test cycles in the laboratory to assess genetic constructs or circuits (i.e., biomolecular breadboarding) 51 .
To test small to large culture sizes, we cultured BL21 Star TM (DE3) cells in multiple culture volumes (10 L, 1 L, 500 mL, 100 mL, 50 mL, and 10 mL) and prepared cell extracts from this strain exploiting the linear correlation found in the heat map ( Figure 4). We used a steamin-place fermentor for 10 L culture, 2.5 L baffled tunair shake flasks for large culture (1 L and 500 mL) and 300 mL baffled tunair shake flasks for small cultures (100 mL and 50 mL). For the 10 mL culture, we grew the cells in standard glass culture tubes. In order to ensure Figure 4 | Mapping CFPS yields across the landscape of total sonication energy input and cell suspension volume demonstrates the ability to generate highly active extracts. CFPS activity was mapped for two E. coli strains, showing a linear trend for the highest CFPS activity. The color code represents relative CFPS activities (100% (red) to 0% (purple)) for each designated energy and volume. Linear trends for total energy inputs covering the highest CFPS activity are computed as follows: consistency across different cultivation conditions, we carefully monitored cell growth rates and harvested all cells at an OD 600 of 3. Cells cultured in the 10 L fermentor or shake flasks showed similar growth rates within 13% variation. Whereas, the growth rate of the 10 mL culture in the culture tube was 44% slower (Supplementary  Table S1). We hypothesized that the slower growth rate in the culture tube was due to inefficient mixing and reduced oxygen transfer. Given the slower growth rate, we might have expected that the cell extract from the 10 mL culture tube would have had lower activity. Instead, all cell extracts from different culture volumes have the same CFPS activity ( Figure 5B). These data suggest that there may be a common mechanism limiting CFPS. One possible explanation is that dilution of translation factors is a limitation. Underwood, Swartz, and Puglisi have showed this limitation previously 52 . At longer batch reaction times, small energy molecules and phosphate accumulation in the PEP based energy regeneration system used here are expected to limit CFPS yields, as reported previously 4,53-55 .
The results described above are important due to the practical implications that an extract preparation from 10 mL culture will provide, in particular, by reducing the time and effort for making numerous different extracts. The wet weight of cells from a 10 mL culture tube was about 0.07 g at harvest OD 600 3.0, which was a sufficient quantity to prepare a 100 mL cell suspension for sonication. Only a few seconds (,5 s) and a sonication energy input of about 25 J were necessary to efficiently lyse 100 mL and generate a highly efficient cell extract for CFPS. While the BL21 Star TM (DE3) cells demonstrated similar protein synthesis yields from 10 mL to 10 L cultures, extracts from strain C495 did not perform well at the 10 mL scale for unknown reasons (data not shown).

Conclusions
In summary, we developed a rapid, robust, and high-throughput sonication based cell extract preparation method. We identified that sonication energy input is a critical parameter that needs to be modulated to achieve efficient and highly active CFPS. We also defined a systematic approach to develop sonication based cell lysis procedures adaptable for multiple bacterial strains. By detailed mapping of sonication energy input versus cell volume suspension, we discovered a linear correlation in the fitness landscape that we found useful to extend our approach to a variety of cell suspensions (from the 100 mL Energy differences between predicted and observed 100% productivity.  to the 30 mL scale). In addition, we showed a scaling factor of ,10 3 for different cell culture volumes (i.e., CFPS yields were similar from source strains grown from the 10 mL to the 10 L scale), reducing the cell growth volume to as low as 10 mL for the BL21 Star TM (DE3) strain. We successfully minimized culture volume without compromising CFPS performance. Therefore, our method can significantly reduce the time for generating highly active extracts, which have previously been based on expensive and laborious equipment for culture and cell lysis. The sonication based extract generation reported here can accelerate the time for generating 100 extracts to potentially ,1 day (including culture time) as compared to about 250 days with our previous approach that required a 10 L fermentor and high pressure impinge homogenizer (Supplementary Figure S5). Finally, we note that lysates prepared by sonication can produce high protein titers in CFPS. While CFPS reactions reported throughout the paper were carried out for only 4 hours for rapid prototyping and consistency, Supplementary Figure S6 shows that protein yields are ,1 mg/mL from sonicated lysates when the reactions are allowed to go to completion (20 h). The cell extract preparation method demonstrated in this study not only provides a readily available, consistent approach for preparing E. coli based crude cell extract, but also an opportunity for fast characterization of genomically engineered E. coli extracts in CFPS at high yields. We believe that this work will help to overcome current limitations in preparing cell extracts for high-throughput processes. We also anticipate that this method will greatly increase the 'design-build-test' cycles for a wide variety of synthetic biology studies. In order to lyse cells by sonication, thawed and suspended cells were transferred into 1.5 mL microtube and placed in an ice-water bath to minimize heat damage during sonication. The cells were lysed using a Q125 Sonicator (Qsonica, Newtown, CT) with 3.175 mm diameter probe at frequency of 20 kHz and 50% of amplitude. The input energy (Joules) was monitored and recorded during sonication. The lysate was then centrifuged once at 12,000 RCF at 4uC for 10 min. The run-off reaction (37uC at 250 rpm) and second centrifugation (10,000 RCF at 4uC for 10 min) were performed for strain C495. However, in the case of strain BL21 Star TM (DE3), the run-off reaction and second centrifugation step were not needed. The total amount of protein in cell extract was quantified by Bradford assay. All extracts contained 40.7 6 2.6 mg/mL of total E. coli proteins. All of prepared cell extract was flash frozen in liquid nitrogen and stored at 280uC until use.

Materials
Cell-free protein synthesis. The CFPS reactions were carried out in a 1.5 mL microtube in the incubator. The standard reaction mixture for CFPS consists of the following components in a final volume of 15 mL: Quantitative analysis of synthesized protein.
The amounts of the cell-free synthesized protein were quantified from the TCA-insoluble radioactivity calculation using a liquid scintillation counter (MicroBeta2, PerkinElmer, Waltham, MA) as described elsewhere 55,57 . Fluorescence of active sfGFP was measured by multi well plate fluorometer (Synergy2, BioTek, Winwooski, VT). 2 mL of cell-free synthesized sfGFP was mixed with 48 mL of purified water and placed in a flat bottom of 96-well half area black plate (Corning Incorporated, Corning, NY). Excitation and emission wavelength for fluorescence of sfGFP were 485 and 528 nm, respectively. Total and cell-free synthesized protein was determined by Coomassie-blue staining analysis on a 4-12% SDS-PAGE gel (Life Technology, Grand Island, NY).