Heat-stable preservation of protein expression systems for portable therapeutics production

Many biotechnology capabilities are limited by stringent storage needs of reagents, largely prohibiting use outside of specialized laboratories. Focusing on a large class of protein-based biotechnology applications, we address this issue by developing a method for preserving cell-free protein expression systems under months of heat stress. Our approach realizes an unprecedented degree of long term heat stability by leveraging the sugar alcohol trehalose, a simple, low-cost, open-air drying step, and strategic separation of sets of reaction components during drying. The resulting preservation capacity opens the door for efficient production of a wide range of on-demand proteins under adverse conditions, for instance during emergency outbreaks or in remote or otherwise inaccessible locations. As such, our preservation method stands to advance a great number of different cell-free technologies, including remediation efforts, point of care therapeutics, and large-scale biosensing. To demonstrate this application potential, we use cell-free reagents subjected to months of heat stress and atmospheric conditions to produce sufficient concentrations of a pyocin protein to kill Pseudomonas aeruginosa, one of the most troublesome pathogens for traumatic and burn wound injuries. Our work makes possible new biotechnology applications that demand both ruggedness and scalability.


19!
The vast majority of biotechnology and synthetic biology capabilities remain limited to the 20! laboratory due to issues with reagent stability, robustness, and safety concerns. This is 21! particularly true of protein-based applications, which often capitalize on the inherent properties 22! 1 d (Fig. 2a), individually stored extract and reaction buffer still exhibited significant protein 115! expression capacity even after 3 d (Fig. 2b-c). This suggested a potential benefit to separately 116! preserving extract and reaction buffer. Also, although extract was less stable than reaction buffer 117! in liquid form ( Fig. 1c-d), dried extract was more stable than dried reaction buffer (Fig. 2b-c).

118!
Specifically, even after 40 d, dried extract exhibited less than a 10-fold decrease in expression 119! yield, whereas dried reaction buffer completely lost viability after 6 d. 120!

121!
While simply drying cell extract improved stability, the significant decreases in the efficacy of 122! extract and reaction buffer over a month suggested the need for identifying preserving agents.

123!
For this, we selected the non-reducing sugar alcohol, trehalose. We chose trehalose because it is 124! implicated in anhydrobiosis in natural organisms such as the tardigrade (water bear or moss 125! piglet) 22 , has been widely used in preservation applications 23 , and can be produced in a cost 126! effective manner 24 . As expected, trehalose dramatically extended the shelf life of preserved cell-127! free systems. Whereas reaction mixture dried without trehalose lacked viability after 1 d, 128! reaction mixture dried with trehalose did not undergo a full 10-fold reduction in yield until after 129! 40 d at 37°C (Fig. 2a). Cell extract efficacy was similarly stabilized over longer time periods 130! (over three months) when dried with trehalose ( Fig. 2b), as was dried reaction buffer which, with 131! the addition of trehalose, maintained full stability for 20 d before gradually declining (Fig. 2c).  Fig. 2), only the phosphate donor (here, creatine phosphate) was identified as a reagent that 138! should be preserved separately . Separate preservation of the 139! phosphate donor yielded a major improvement in long-term reaction buffer efficacy, particularly 140! after 20 d (Fig. 2d vs. Fig 2c).

142!
Having demonstrated successful expression from preserved cell extract and fresh reaction buffer 143! and from preserved reaction buffer and fresh cell extract, we next tested system performance 144! when preserved cell extract is mixed with preserved reaction buffer. Figure 2e shows expression 145! yield vs. storage time at 37°C for cases in which all expression components (extract, reaction 146! buffer, creatine phosphate, and magnesium acetate) are preserved, heat stressed, and 147! reconstituted (see also Supplementary Fig. 5). In all cases, expression yield is initially lower 148! when preserved extract is used (Fig. 2e) as opposed to fresh extract (Fig. 2d). This is at least 149! partially due to the fact that, when both extract and reaction buffer components are preserved, 150! reconstituted, and combined, the final concentration of trehalose is higher -a consequence of 151! including trehalose in both the extract and the buffer preservation procedures. High 152! concentrations of trehalose can lower protein expression rates ( Supplementary Fig. 6).

153!
Nonetheless, significant expression is realized for over three months of storage. In a similar 154! experiment for a longer time period, we found that, although expression begins to decrease after 155! approximately two months of exposure at 37°C, significant fluorescence resulting from EGFP 156! expression is realized over eight months of exposure at 37°C (Supplementary Fig. 7 Figure 2f summarizes our overall cell-free preservation approach: 1) cell extract with added 162! trehalose is sterilized and dried, 2) reaction buffer prepared without creatine phosphate, without 163! magnesium acetate, with trehalose, and optionally with PEG is sterilized and dried, 3) creatine 164! phosphate and magnesium acetate are then stored separately, 4) all reagents are reconstituted 165! with water, mixed in the appropriate ratios, and then combined with a chosen DNA construct to 166! express the protein of interest. As an alternative to this preservation protocol, Figure 2a shows 167! that the simple drying of reaction mixture combined with trehalose along with the separate 168! storage of magnesium acetate also constitutes a viable approach to preservation. Nonetheless, 169! the approach in Figure 2f offers flexibility advantages. Namely, different buffer variants could 170! be selected at the time of reconstitution to optimize expression for particular protein(s). 171!

172!
Having established a cell-free reagent preservation approach, we sought to demonstrate 173! therapeutic application potential. For this, we used our preserved protein expression system to 174! produce pyocin S5, a 56.1 kDa protein, which, upon proper folding, forms lethal pores in 175! susceptible strains of the pathogen P. aeruginosa 25,26 . To test the long-term therapeutic efficacy 176! of our system, we used both plate clearing and broth dilution assays. The plate clearing 177! experiments in Figure 3a,c show that complete clearing was achieved at all time points 178! throughout the 136 d of heat stress. Likewise, the broth dilution experiments (Figure 3b,d) show 179! that the survival fraction of P. aeruginosa was reduced by 4 logs or more (depending on 180! trehalose concentration) using a 20-fold dilution of pyocin produced from 136-day-old, heat-181! stressed (37°C) cell-free reagents (extract, reaction buffer, magnesium acetate, and creatine 182! phosphate). Impressively, throughout the 136 d experiment a 100-fold dilution of pyocin 183! ! 9! produced from reconstituted cell-free reagents was sufficient to achieve a 100-fold reduction in 184! the P. aeruginosa survival fraction. To further quantify these results, we calculated MIC values 185! and found that they ranged within an order of magnitude for the duration of the 136-day 186! experiment (Supplemental Note 2, Supplementary Figure 8 and strategic separation of components to drying approaches like lyophilization will be important 202! for large-scale operations. An additional potential improvement of our approach includes the use 203! of antibiotics for reagent sterilization. Here, we chose membrane filtration simply to avoid 204! potential interference with our application demonstration of killing bacteria. However, sterile 205! ! 10! filtration has been previously associated with long term decreases in cell-free expression yield 27 . 206! Therefore, alternative sterilization approaches may not only lend themselves towards better 207! scalability 28 , but also further improve preservation 28 . 208!

209!
Our demonstrated, long-term heat stability of protein expression opens the door to a host of 210! applications in the realms of therapeutics production and delivery, bioremediation, and 211! biosensing. Of particular relevance, the potential end of the antibiotic era has brought increased 212! attention to the production of alternative therapeutics 16 , including pyocin-and colicin-like 213! proteins and more complicated targets like whole phages 29 , antibiotic peptides 30 , and vaccine 214! components 31 . Previous studies have demonstrated efficient cell-free production of these types 215! of therapeutics, and recently, novel fluidics platforms have been developed to facilitate the 216! expression and purification of proteins using cell-free systems 32, 33 . Our work fills a key need to 217! make practical use of these cell-free capabilities and platforms, as it addresses the limiting 218! factors of stability, storage, and distribution associated with protein therapeutics, and enables 219! production of these novel antimicrobials on site. To establish an initial preservation performance baseline, we conducted experiments to 305! determine the inherent shelf life of cell-free reaction mixture (Fig. 1). To characterize reaction 306! mixture stability, we first sterilized reaction mixture through 0.22 µm filtration to prevent initial 307! fouling that may obscure inherent biochemical stability. Aliquots of filtered reaction mixture 308! were then stored at room temperature and at 37°C. At different time points, the plasmid pUC-309! T7tet-T7term was added to express enhanced green fluorescent protein (EGFP) from a T7 310! promoter variant. Using a microplate reader, the fluorescence intensity after 5 h of incubation 311! was measured to quantify expression, as described in the GFP expression assay section above. 312!

313!
To gain a deeper understanding of reaction mixture degradation, we also characterized cell 314! extract and reaction buffer stability separately. To quantify cell extract stability, we stored 315! aliquots at room temperature and at 37°C. At different time points, we added fresh reaction 316! ! 15! buffer and the pUC-T7tet-T7term expression construct and assayed fluorescence as was done to 317! test the full reaction mixture (Fig. 1c). After 4 d at room temperature, cell extract efficacy had 318! decreased by a factor of 31.1. After only one day under heat stress, the extract completely lost 319! viability and generated negligible fluorescence in our expression assay. To test reaction buffer 320! stability, we similarly added fresh extract and the expression construct and quantified 321! fluorescence at different time points (Fig. 1d). At room temperature, buffer efficacy remained 322! high for 43 d and then rapidly decreased. However, after only 9 d at 37°C, the buffer efficacy 323! was reduced by a 41.5-fold. 324!

Preservation and reconstitution 326!
We utilized a simple open-air drying approach. This approach was chosen due to its simplicity 327! and low cost. Also, the fact that the final product is not trapped in a porous material such as 328! paper facilitates transport and large-scale applications. All of the following approaches could be 329! adapted to other drying methods, such as freeze-drying or spray-drying, if desired. Prior to 330! drying, all reagents were filtered using 0.22 µm spin filters (Millipore). All drying was 331! performed by pipetting small aliquots (14-28 µL) on a silicon sheet (Silpat). The sheet was then 332! placed in a 37°C incubator. Dried samples were recovered using a razor and transferred to 333! DNAse/RNAse free 1.5 mL microcentrifuge tubes. Temperature and humidity were quantified 334! using an Inkbird THC-4 digital sensor. For instance, for the experiments in Figure 3, the average 335! temperature was 37.1°C (σ=0.8°C), and the average relative humidity was 20.5% (σ=4.3%). components to perform protein expression reactions. Pyocin S5 was expressed using the T7 359! expression plasmid pT7-pyoS5, and EGFP was expressed using pUCT7tet-T7term. These 360! reactions were incubated in microcentrifuge tubes in a rotator at 30°C. After 6 h, different 361! dilutions of the pyocin reaction were prepared, whereby the EGFP reaction product was used to 362! dilute the pyocin reaction product. 363!

364!
For each time point in Figure 3, 14 samples were set up for analysis by plate clearing and broth 365! dilution assays. Of these 14 samples, 13 consisted of two-fold serial dilutions of the pyocin cell-366! free reaction product, using the cell-free EGFP reaction product to dilute the pyocin product.

367!
The final sample consisted of the cell-free EGFP reaction product, which served as a negative 368! control. Broth microdilution methods were adapted from Wiegand et al 34 , and plate clearing 369! assays were based on a combination of the Kirby Bauer method and previously used assays for 370! characterizing pyocin S5 25 . For both plate clearing and broth dilution, three colonies of P. 371! aeruginosa BE171 35 were cultured in 2 mL casamino acids (CAA) media (5g/L Difco iron-poor 372! casamino acids, 0.25 g/L MgSO 4 *7H 2 O, and 1.18 g/L K 2 HPO 4 ) and grown to log phase in a 373! 37°C shaking incubator. Cultures were diluted in CAA to prepare a 2 mL stock with an OD600 374! of 0.1. For the plate clearing assays, petri dishes (100 mm diameter) were prepared with 25 mL 375! CAA and 1.5% agar. These CAA plates were inoculated by swabbing approximately 200 µL of 376! culture diluted to an OD 600 of 0.1 with a sterile cotton tipped applicator. Once the swabbed 377! solution dried, 10 µL drops of each of the 14 cell-free reaction samples were dispensed (7 378! samples per plate). Plates were then incubated at 37°C for approximately 15 h and subsequently 379! imaged using a G:Box Chemi XX9 imager. For broth dilution assays, a bacterial suspension of 380! 5*10 5 CFU/mL was prepared, and 95 µL of this suspension was aliquoted to wells of a 96 well 381! microplate. Then, 5 µL of each cell-free reaction sample was added to the cell suspension in