Microbial mutagenesis by atmospheric and room-temperature plasma (ARTP): the latest development

Although rational genetic engineering is nowadays the favored method for microbial strain improvement, random mutagenesis is still in many cases the only option. Atmospheric and room-temperature plasma (ARTP) is a newly developed whole-cell mutagenesis tool based on radio-frequency atmospheric-pressure glow discharge plasma which features higher mutation rates than UV radiation or chemical mutagens while maintaining low treatment temperatures. It has been successfully applied on at least 24 bacterial and 14 fungal species, but also on plants, dinoflagellates, and other microbial communities for the improvement of tolerance to medium components, to increase cellular growth and production of cellular biomass, to enhance enzyme activity, and to increase the production of various chemicals. Achievements like 385.7% of acetic acid production enhancement in Acetobacter pasteurianus give this new mutagenesis tool a promising future. However, certain questions remain regarding optimal operational conditions, the effects at subcellular levels, and standard operation procedures, which need to be addressed to facilitate applications of ARTP in microbial breeding and other fields such as evolution of enzymes.


Introduction
Genetic enhancement of whole-cell systems is of great interest in microbial biotechnology. Rational design in genetic engineering has become during the last decades the preferred method of choice with the CRISPR/Cas9 system being the most recent advance (Barrangou and Horvath 2017). However, random mutagenesis is still in many cases the only feasible approach, for instance, if the corresponding genomic locus to the desired phenotype is unknown or if a complex genetic regulation is underlying. Random mutagenesis is also superior from a commercial standpoint as altered strains are classified as genetically unmodified which requires no attention to legal regulations. Besides, products labeled as "GMO free" are currently much better accepted by the public (Twardowski and Małyska 2015).
Although random mutagenesis can be done biologically using methods like proofreading-deficient polymerase (Kim et al. 2013), it is typically achieved by application of chemical mutagens like alkylating agents (Lackmann and Bandow 2014) or azides (Yolmeh and Khomeiri 2017) and physical mutagens like conventional radiations (Yolmeh and Khomeiri 2017) or heavy particle beams (Lackmann and Bandow 2014). The induced DNA damages are subsequently repaired by the SOS system leading to DNA alterations (Bridges 2001). However, almost all tools for random mutagenesis are to a certain extent risky to handle. Whereas chemical mutagens expose the user to high (e.g., ethyl methanesulfonate and methylnitronitrosoguanidine) or very high toxicity (e.g., ethidium bromide and sodium azide), radiation (e.g., X-ray and γ-ray) exposes the user to a permanent danger which requires additional control measures such as radiation badges and radiation barriers. In addition, the generation of radiation requires often bulky and expensive equipment, specialized knowledge for operation and additional security measures to restrict access. Physical plasmas are fully or partially ionized gases which are consisting of neutral and charged particles. They are generally classified into high-and low-temperature plasmas with variations in working gas (e.g., helium, argon, nitrogen, and oxygen), pressure (0.1 Pa up to atmospheric pressure), electromagnetic field, discharge configuration, and temporal behaviors (Bogaerts et al. 2002). Radio-frequency atmospheric-pressure glow discharge plasma (RF APGD) is one of the cold atmospheric plasmas of increased interest which can be produced in between two electrodes driven by a radio-frequency power supply at atmospheric pressure (Tendero et al. 2006). Room-temperature plasma received first attention in applied research for its possible use as a sterilization method for heat-sensitive surfaces (Kramer et al. 2015). However, the sub-lethal treatment suggested also a possible application in the field of mutagenesis. In a quantitative approach using flow cytometry analysis based on the umu test, the ARTP system showed greater DNA damage with higher mutation rates than UV radiation or the application of two selected chemical mutagens (Zhang et al. 2015c). After initial studies proved the successful application of RF APGD plasma jets in microbial breeding , an atmospheric and roomtemperature plasma mutation system (ARTP) was constructed and made commercially available. The ARTP system requires helium (> 99.99%) as a working gas and a typical electric socket as a power source. With a weight of 100 kg and a size of 33 cm × 33 cm × 33 cm, it is of manageable dimensions, requires minimal training, and produces no hazardous radiation, toxic chemicals, or any environmentally harmful waste.
Here, we summarize the progress of ARTP-induced mutagenesis with focus on the most recent reports since the last review ) and the progress in understanding the underlying mechanisms. In the first parts, we elucidate the technical and physico-chemical background in a comprehensive manner. Following this, we focus in the second part on the description and extensive review of the technical parameters and the combination of the ARTP with other mutagenesis methods to enable ARTP operators to understand and optimize the process. Subsequently, a summary of the biotechnological applications is given in the third part. Finally, we close with the concluding remarks to reveal knowledge gaps and future prospects.

Technical and physico-chemical background of ARTP
The ARTP mutagenesis system was initially invented by the Department of Chemical Engineering of Tsinghua University and commercialized by Si Qing Yuan Biotechnology (now Tmax Tree Co., Ltd.) with a vertically mounted RF APGD plasma jet generator as the core component ( Fig. 1a)  . It is currently the only commercially available system and its distribution has been mainly restricted to its origin China. Inside the ARTP apparatus, helium is ionized by a radio-frequency electric field and blown through a nozzle onto a microbial sample on a metal plate fixed to an adjustable platform. The use of atmospheric-pressure plasma in this setup removes the necessity of an expensive vacuum system. Due to the low breakdown voltage, the plasma maintains discharge uniformity, generates little amounts of UV radiation, and assures, in combination with the cooling of the cathode, a biologically compatible gas temperature. A continuous gas flow ensures little mixing with the surrounding air which minimizes the generation of germicidal ozone.
The exact cause of mutagenesis by physical plasma has not yet been fully clarified, but various experiments were carried out on isolated DNA and on DNA in vivo. Besides UV radiation, the generation of reactive chemical species was identified as the most probable cause (Li et al. 2008). Experiments on isolated plasmid DNA showed a fast degradation after RF APGD helium plasma treatment with damage severity depending on the plasma-generating conditions (Li et al. 2008). However, DNA in vivo is also protected by other biomolecules shielding the DNA from damages caused by oxidative stress. Systems like the SOS repair system in prokaryotes are aiding the damaged DNA to maintain its integrity while possibly introducing mutations. In this manner, genes involved in the SOS DNA repair system and genes that are known to respond to oxidative stress were found to be activated in response to physical plasma (Winter et al. 2013). Besides those direct interactions between reactive chemical species and DNA, also indirect reactions with other biomolecules and reactive chemical species leading to mutagenic products were suggested as possible causes for DNA base alterations and the induction of single-or doublestrand breaks (Arjunan et al. 2015). While the impact of plasma on DNA is straightforward, the influence of plasma composition and the resulting critical treatment severity is more complex. The amount of UV radiation and the composition of the generated reactive chemical species depends mainly on the employed gases. While noble gases like argon or helium are generating less UV radiation, the addition of nitrogen or oxygen to noble gases increases the amount of reactive chemical species. Hydroxyl radicals (·OH), atomic oxygen (O), ozone (O 3 ), and superoxide anions (O 2 ·− ) are the most frequently produced reactive chemical species. Furthermore, especially in contact with aqueous solutions, nitrous acid (HNO 2 ), nitric acid (HNO 3 ), hydroxide anions (OH − ), and hydrogen peroxide (H 2 O 2 ) are produced (Lackmann and Bandow 2014). When using helium as a working gas, helium metastables (He * , He 2 * ), helium ions (He + , He 2 + ), and nitrogen molecular ions (N 2 + ) were additionally identified ). Maintaining microbial cell functionality and integrity is required to pass on random mutations to the next generation. However, a high plasma treatment severity can cause irreparable damages to DNA, proteins, and the whole cell, therefore, requiring careful adjustment of the plasma-generating parameters. Enzyme inactivation was suggested to be mainly caused by a degradation of chemical groups, oxidation of catalytic centers, and the overall destruction of the secondary structure (Misra et al. 2016). Direct exposure to room-temperature plasma resulted also in the removal of extracellular matrix followed by etching of whole cells as shown by scanning electron microscopy (Lackmann et al. 2013). Cell etching was monitored in another study by atomic force microscopy. Extended plasma treatment eventually led to a rupture of the cell wall. In both studies, UV treatment alone did not lead to an extended damage of the cell wall (Pompl et al. 2009) which only left reactive chemical species as the primary cause.

Experimental workflow
The ARTP system was designed to allow easy adaptation to different microorganisms. Therefore, a large range of parameter combinations are possible. However, not all possible parameter combinations lead to biologically compatible plasma treatment conditions. The typical workflow (Fig. 1b) starts with the preparation of a defined suspension of microorganisms followed by the ARTP treatment. Subsequently, the treated microorganisms are re-suspended, adequately diluted, and streaked out on agar plates to yield single colonies. The colonies are then screened according to the targeted phenotypes (e.g., photometric test). However, due to the multitude of parameter combinations, it can be challenging to identify optimal plasma mutation conditions leading to single, isolated colonies under maximal usage of the agar surface. Overall, the following three categories are of importance. First, there are the intrinsic but variable parameters of the ARTP system such as the helium gas flow rate (standard liter per minute, SLPM), the gap space between nozzle and sample plate (millimeter, mm), the energy input (Watt, W), and the treatment time (seconds, s). Second, the sample itself causes variations Typical sequence of an ARTP mutagenesis workflow through the kind of microorganism to be manipulated, the cell/spore concentration, the used suspension liquid, and its volume. Third, the subsequent post-treatment of the sample after plasma treatment including the resuspension liquid and time, the subsequent dilution, and the volume used for spreading on agar.

ARTP system variables
For the mutagenesis of biological samples, the ARTP manufacturer suggests standard values for each adjustable parameter to achieve biologically compatible conditions and the generation of reactive chemical species. In compliance with the suggested conditions, a high density of electrons and metastable helium (He*, He 2 *), He + , He 2 + and N 2 + are generated . A helium gas flow rate of 10 SLPM or higher is suggested to avoid interference between the plasma jet and surrounding air as this leads to the formation of germicidal ozone. However, in a prior study, a positive correlation between gas flow rate and the reactive chemical species concentration was shown by the increased damage after treatment of isolated plasmid DNA with plasma generated at flow rates between 5 and 30 SLPM. A 2 mm distance between sample and nozzle is recommended. Larger distances were shown to decrease the concentration of reactive chemical species as shown by the decreased damage severity with distances between 2 and 10 mm. An energy input of around 120 W is suggested to be used. The energy input was found to be positively correlated with the generation of reactive chemical species as proven earlier by damage severity of plasmid DNA treated with plasma between 10 and 120 W (Li et al. 2008). However, increased energy input leads also to a rise in temperature which is especially of significance for the survival of the whole cells. Within an energy input of 40-200 W, temperature was found to be within a biologically compatible range between 36 and 57 °C ). The last intrinsic parameter of the ARTP system and the most important parameter for each experimental setup is the treatment time. It is primarily used to determine the dose of the reactive chemical species to the microorganism under selected conditions. Direct treatment of isolated DNA under identical plasma-generating conditions with time variations between 0.5 to 10 min was shown to increase DNA damages (Li et al. 2008). However, it can be assumed that due to the high complexity and the presence of more sensitive molecules, the effective treatment time for whole cells is situated at a shorter exposure time. Summarizing the applied values of the operational parameters in the literature, a high variety from the suggested standard values is present. However, as some studies do not report their parameter values, a representative distribution cannot be derived. A wide variation of the helium gas flow rate can be found. While many studies selected the suggested helium gas flow rate of 10 SLPM, an equal number of studies reported a helium gas flow rate of 15 SLPM. Single studies went as low as 2 SLPM (Zong et al. 2012;Ren et al. 2017). The suggested distance between sample and nozzle of 2 mm was used in almost all reports. Single cases reported a further distance of 1 cm (Zong et al. 2012;). The energy input is a critical parameter due to its ability of heating the sample to undesirable temperatures. In the reviewed studies, the energy input was, as suggested, typically at 100 or 120 W. One single study chose a low-energy input of 40 W which might explain the extended required treatment time . Three studies applied 180 W either for spores or filamentous fungi which are expectably less affected by higher temperatures due to their higher resistance (Xu et al. 2011(Xu et al. , 2012Wang et al. 2016a). Treatment time is used as the main parameter to adjust the exposure of the sample to plasma while keeping all other parameters constant. Therefore, the treatment time in the reviewed studies is mainly dependent on the other parameters preset. When other parameters were chosen to generate less reactive chemical species, this resulted in the requirement to select a longer treatment time. However, the exact impact of a single parameter cannot be estimated as in most studies multiple parameters were significantly changed and no systematic approaches were applied.

Sample-related variables
The kind of microorganism (e.g., species, Gram-positive/ Gram-negative) and its current developmental stage (e.g., viable cell or spore) were shown to be of significance in case of plasma sterilization. For the ARTP, the treatment time under otherwise constant operation parameters should be evaluated for each strain by establishing a death rate curve. The suggested range for the death rate curve for bacteria is in general estimated to be effective in a range between 15 and 120 s, for Actinomycetes between 30 and 180 s, for fungi between 60 and 360 s, for yeasts between 30 and 240 s, and for microalgae between 5 and 150 s. However, as no research involving the systematic sub-lethal, helium-based plasma treatment of microorganisms via ARTP has been conducted yet, only the current findings regarding plasma sterilization combined with the time suggestions mentioned above can give a general indication for ARTP mutagenesis. Sterilization studies for different species involving Gram-positive and Gram-negative microorganisms either concluded a greater resistance of Gram-positive microorganisms or no difference at all. A study investigated the survival after plasma treatment of the Gram-negative Escherichia coli, the Gram-positive Staphylococcus aureus, the yeast Saccharomyces cerevisiae, and Bacillus subtilis spores using helium/oxygen as a working gas. While D values for E. coli and S. aureus were similar to 18 and 19 s, respectively, S. cerevisiae showed with 115 s and B. subtilis spores with 840 s, a much higher resistance to the plasma treatment (Lee et al. 2006). In another sterilization study with argon plasma involving 11 different bacterial strains, susceptibility of Gram-negative species was demonstrated to be generally greater than that of Gram-positive species. However, the susceptibility of Gram-positive species was determined to be strain specific (Ermolaeva et al. 2011). On contrary, two E. coli strains and one Listeria monocytogenes strain were exposed to dielectric barrier discharge atmospheric cold plasma with a variety of working gas mixtures containing O 2 , N 2, and CO 2 . The Gram-positive L. monocytogenes was inactivated more rapidly than the Gram-negative E. coli strains and a difference in inactivation time between the E. coli strains was found (Lu et al. 2014). These inconclusive findings can be reasoned by variations in plasma generation, but might also elaborate a strain specificity rather than the Gram classification. However, while fungi exhibited a greater resistance than bacteria, spores have clearly the greatest resistance against plasma treatment. For the preparation of appropriate cell concentrations for the ARTP treatment, it is suggested to harvest cells or spores from a culture during the logarithmic phase and wash 2-3 times with sterile solution. Subsequently, the resulting OD 600 should be adjusted to 0.6-0.8 or to 10 6 -10 8 cells × ml −1 . It must be emphasized that the optical density can only give a rough indication of the cell concentration and is heavily dependent on strain, cell type, and growth media. A cell count would be only possible in case of non-motile cells. Therefore, it would be instead advisable to use a constant, appropriate, and high optical density under consideration of the subsequent dilution steps with the aim to obtain single colonies on the final agar plate. The suspension liquid itself is only recommended to be sterile while the volume transferred to the sample plate should be in the range of 10-20 µl. In case of extended treatment times or small volumes, an addition of 5% glycerol to the suspension liquid is suggested to minimize sample evaporation. One study investigated the impact of suspension composition on the severity of the plasma treatment. DNA samples were suspended after plasma treatment in water, phosphate-buffered saline (PBS), and media which consisted of carbonate buffer, salts, amino acids, phenol indicator, and vitamins. While water did not protect the sample, a minimal protection through PBS was observed. In contrast, the media protected the DNA sample to a large extent (Leduc et al. 2009). A later study detected a similar degree of DNA damage in PBS and aqueous solution; however, the formation of strand breaks in supercoiled DNA was slower in PBS than in water. In contrast, a tris-EDTA buffer prevented DNA damages greatly (O'Connell et al. 2011). These findings can be attributed to the buffer itself and to additional compounds within the buffer acting as radical scavengers. Furthermore, the effects of certain parameters interacting with each other should not be neglected. Besides the already mentioned energy input which leads to a temperature rise and subsequently to reduced cell viability, especially the energy input and the helium gas flow rate but also the chosen suspension liquid leads to an increased evaporation. This results in a change of the initial conditions like buffer and cell concentration and changes the overall experimental setup. For sample-related variables, only limited conclusions from the available studies can be drawn as most factors are estimated to exhibit weaker effects than the earlier mentioned ARTP system variables. The microorganism and its current state of development were found to have a strong impact. Fungal species tend to be more resilient than bacterial species and dormant bodies such as spores exhibited the highest resistance to ARTP plasma treatment. However, no conclusions about differences between Gram-positive and Gram-negative species or even amongst strains of the same species can be drawn. For most studies, cell concentration was adjusted to the suggested OD 600 of 1 and 1.5  or even 2 (Lu et al. 2011). As suspension liquid, a physiological saline was mainly employed followed by 50 mM PBS buffer at pH 6 or 7. In a single study 10%, glycerol was used (Wang et al. 2016c). Mostly, the suggestion of applying a volume of 10 or 20 µl suspension liquid was implemented, while very few employed volumes as high as 50 µl (Hua et al. 2010;Lu et al. 2011) and even up to 100 µl . High volumes seemed to have been used especially to counteract evaporation processed as they were used mostly together with long treatment times above 1 min and at high gas flow rates. In contrast, some cases reported the pre-drying of the sample on the sample plate prior to plasma treatment (Liu et al. 2013;Wang et al. 2014a, b).

Post-treatment variables
After the ARTP treatment, the sample plates are submerged in resuspension liquid. A resuspension volume of 1 ml and a vortexing time of at least 1 min are recommended to ensure complete resuspension. The resuspension liquid acts as a radical scavenger equally to the suspension liquid during the ARTP treatment. Furthermore, the ratio between the treated volume and the resuspending media was found to be of importance. Cell samples treated with a dielectric barrier discharge plasma were affected more at lower than at higher dilutions indicating that the concentration of reactive chemical species and the length of the exposure are relevant (Kalghatgi et al. 2011). However, there are no reports on the influence of the vortexing time or intensity on the survival of plasma-treated cells. The subsequent dilution range before spreading the sample on agar is dependent on the plasma treatment severity and the initial cell concentration. In general, a consistent post-treatment procedure and especially homogeneous cell dilutions needs to be ensured to obtain isolated colonies in high reliability. Data describing the post-treatment are limited. Overall, the used resuspension liquids were the same as the suspension liquid used for the plasma treatment. The volume was set between 0.8 and 2 ml, while in one case, even 5 ml was used ). This determines the ratio, as discussed earlier, which is possibly important to stop further exposure of the sample to reactive chemical species. More information about post-treatment details such as vortex intensity, dilution range, or spread volume were rarely given.

Combined usage of the ARTP system with other methods
Most studies reported a single ARTP treatment to yield the desired target. However, some studies reported an iterative ARTP treatment. While an iterative treatment for eight times (Jiang et al. 2017) led to a steady increase of the desired phenotype, 30 repetitions were only used to produce a larger pool of possible mutants (Luo et al. 2017b). ARTP treatment was also combined together with other treatments such as chemical mutagenesis with diethyl sulfate , nitrosoguanidine  or radiation mutagenesis with UV and UV-NaNO 2 ). However, due to the lack of comparative studies, it cannot be concluded if an iterative ARTP treatment or a combination of ARTP with traditional mutagenesis tools is superior over a single ARTP treatment. Apart from applying ARTP in straightforward mutagenesis approaches, ARTP was also successfully used as a tool for the construction of libraries for genome shuffling (Xu et al. 2012;Zhang et al. 2015b;Gu et al. 2017).

Biotechnological applications
The core component of ARTP, the RF APGD plasma jet generator, was first used with helium as working gas in a proof of concept study for the generation of Streptomyces avermitilis mutants in 2010 . Since then, a multitude of case studies were published in Chinese and international journals with the majority of more than 80% published after 2014 (Fig. 2).

Microorganisms
Between 2010 and 2018, a total of 42 different microbial species in 73 case studies can be found in international literature (Table 1) which include Gram-negative bacteria (9 species/17 studies), Gram-positive bacteria (15 species/26 studies), filamentous fungi (6 species/8 studies), yeast (8 species/15 studies), species not belonging to the classes mentioned above (3 species/6 studies) and even a whole microbial community. Together with studies published in Chinese language, more than 300 ARTP case studies are estimated to be currently available. Almost all species are classified as Biosafety Level 1 strains, and most are used in biotechnological research and industry like E. coli and B. subtilis.

Mutagenesis targets
Overall, mutagenesis targets can be classified into six groups, namely, (i) increase of microbial tolerance to certain growth conditions and media compounds, (ii) growth increase and optimization of biomass-related parameters, (iii) enzyme activity increase, (iv) butanol production increase, (v) production increase of organic acids and their derivates, and (vi) production increase of specialty chemicals mainly used in pharmaceutical products. As there are numerous positive achievements reported, only selected outstanding examples can be outlined in detail. In a study comprising three publications involving the yeast Rhodosporidium toruloides, three mutants were isolated which were resistant to the growth inhibitors vanillin, furfural, and acetic acid and, therefore, could grow in sugar cane bagasse hydrolysate ). Subsequently, the mutant with the highest tolerance towards inhibitors was identified (Qi et al. 2014) and further investigated using a mixed transcriptome/ proteome approach. Thirty-nine genes were identified with a larger than fivefold increased transcription level belonging to the cluster of nucleotide excision repair, glycolysis, spliceosome assembly, and MAPK signalling (Qi et al. 2017). In another study, a B. subtilis mutant showed a 37.9% increase in extracellular protein concentration which resulted in an overall higher recombinant protein    secretion (Ma et al. 2015). The increase in butanol production by ARTP was shown several times such as in a mutant which exhibited a 33% higher ABE solvents production and a 25% higher butanol production (12.53 g l −1 total ABE solvents, 4.9 g l −1 acetone, 13.71 g l −1 butanol, and 0.19 g l −1 ethanol) when compared to its wild type (Kong et al. 2016). The enzymes of the cellulase complex in Trichoderma viridae were investigated individually after ARTP treatment. A 2.38-fold increase of filter paper activity, 2.61-fold increase of carboxymethyl cellulase activity, 2.18-fold increase in β-glucosidase activity (EC 3.1.2.21), and a 2.27-fold increase in cellobiohydrolase (EC 3.2.1.91) activity was reported (Xu et al. 2011). The organic acid l-lactic acid produced by an ARTP mutant of Bacillus coagulans reached a titer of 90.2 g l −1 from 100 g l −1 xylose, while 23.49 g l −1 from corn stover with a yield of 83.09% was achieved (Zheng et al. 2014). The carotenoid lycopene, a specialty chemical used as a pigment and antioxidant, showed a 55% concentration increase to 19.3 g l −1 in Blakeslea trispora mutants (Qiang et al. 2014).

Concluding remarks
ARTP has been proven to be a reliable and effective microbial breeding system leading to high frequency of random mutations induced by reactive chemical species which are produced by the helium-based atmospheric and roomtemperature plasma. Its application has been reported to be successful in various studies involving diverse bacterial and fungal species for improving strain properties such as tolerance to medium components and growth conditions and increasing production of valuable products like cellular biomass, enzymes, bulk, and specialty chemicals. However, there are still theoretical and technical questions open for clarification, whose absence might delay the widespread use of the ARTP mutagenesis system including: 1. The factors involved in the operation of the system, their interactions, and the severity of their impact on the resulting mutagenic effects. Developing an automatic and quantifiable parameter setup for effective plasma treatment would be an essential prerequisite for wider applications of the ARTP mutagenesis breeding system. 2. The systematic clarification of the susceptibility of different microorganisms and their growth stages to the ARTP treatment are required for the development of Standard Operation Protocols. Unfortunately, many case studies did not indicate precisely the status of all operational parameters making it difficult for a susceptibility estimation of different microorganisms to the sub-lethal plasma treatment. 3. More omics studies are needed to gain a deeper scientific insight of the mutagenic impact of ARTP to widen the applications of ARTP-induced gene mutations not only in cellular breeding but also in other fields such as directed evolution of enzymes. The combined use of ARTP with other mutation methods such as γ-ray or with rational mutations such as CRISPR/Cas9 might give additional advantages in improving microbes, which is worth investigating.