Evaluating the Toxicity of Ionic Liquids on Shewanella sp. for Designing Sustainable Bioprocesses

Ionic liquids (ILs) are widely used as reaction and separation media in many technologies due to their unique and advantageous physicochemical properties. Thus, further studies approaching the study of the toxicity of these materials are required. Moreover, they are utilized in devices in which microorganisms such as Shewanella sp. act as biocatalysts. Thus, in this work, the toxicity of 69 ILs on the marine bacterium Shewanella sp. was tested. Specifically, the ILs analyzed were based on the cations imidazolium, pyridinium, pyrrolidinium, piperidinium, morpholinium, oxazolinium, phosphonium, ammonium, and sulfonium, in combination with different anions. The toxicities of this wide group of ILs on Shewanella sp. were determined using two methods: 1) the agar disk-diffusion test and 2) the growth inhibition test in liquid media. The relationship between toxicity and IL chemical structure was elucidated. A decrease in the hydrophobicity of the alkyl chain length was found to be a key factor to reduce IL toxicity. On the other hand, phosphonium-based ILs containing long alkyl chains were shown to be largely incompatible with Shewanella sp.


INTRODUCTION
The genus Shewanella was first described in 1931. This microorganism responsible for butter putrefaction was initially named as Achromobacter putrefaciens but after further studies and biochemical characterizations, the species isolated in 1931 was then reclassified into a new genus, Shewanella (Derby and Hammer, 1931;MacDonell and Colwell, 1985). So far, all members belonging to this genus are facultatively anaerobic, which allows them to survive in a wide variety of environments, although they are usually found in aquatic or marine habitats (Hau and Gralnick, 2007). One of the main characteristics of the Shewanella genus is its capability to grow at low temperatures. Although most of the strains are psychrotolerant with optimum growth temperature above 16°C, many strains are psychrophilic and can easily grow below this temperature (Hau and Gralnick, 2007). The capability of Shewanella to use several external electron acceptors to respire, which enables this genus to live in extreme conditions, as well as the possibility of being genetically modified, has boosted its application in a wide variety of technological fields (Hau and Gralnick, 2007;Marsili et al., 2008;Zou et al., 2019), from bioremediation of polluted environments (Varia et al., 2014;Zhang et al., 2018) and biosensing (Webster et al., 2014) to bioenergy production in microbial fuel cell technology (Jorge and Hazael, 2016;Kumar et al., 2016).
Ionic liquids (ILs) are combinations of cations (i.e., imidazolium, pyridinium, and ammonium) and anions (i.e., hexafluorophosphate, tetrafluoroborate, and bis [(trifluoromethyl)sulfonyl]imide) that remain liquid at temperatures below 100°C. In recent years, these compounds have drawn the attention of the scientific community due to their unique properties such as negligible vapor pressure, chemical and thermal stability, nonflammability, high ionic conductivity, wide electrochemical potential window, and solvation ability, among other features (Ventura et al., 2012b). ILs are also considered clean and green alternatives to organic solvents mainly due to their low vapor pressure, which prevents their release into the atmosphere. Besides, their physical and chemical properties can be adapted to specific applications by the fine-tuning of the cation structure and by selecting proper cation-anion combinations. Since a wide variety of ion combinations are possible, it is of the highest importance to outline rational guidelines to develop technologically suitable but also environmentally harmless ILs (Latała et al., 2009).
In this context, this work investigates the biocompatibility of ILs containing different cation and anion structures toward Shewanella sp. Thus, the toxicity of a total number of 69 ILs based on imidazolium, pyridinium, pyrrolidinium, piperidinium, morpholinium, oxazolinium, phosphonium, ammonium, and sulfonium combined with different anions has been analyzed in solid and liquid media. The results obtained are in-depth discussed and related to the structural characteristic of the ILs, which enables identifying key factors for the design of environmentally friendly ILs with lower ecotoxicological risks than conventional organic solvents.

Ionic Liquids
The ILs investigated in the current work are grouped according to the type of cation present in their structure. Tables 1-8 include the name of each IL analyzed, together with its structure and water solubility. These tables also include the nomenclature used for each IL (abbreviation name). All ILs were supplied by IoLiTec (Ionic Liquids Technologies).
Culture of Shewanella sp.
The strain of Shewanella sp. was provided by the Aquatic Ecology Laboratory (University of Murcia, Spain) and kept on Luria-Bertani Agar (LB). A single colony was used to grow the inoculum on LB broth at 25°C with shaking at 200 rpm. All chemicals and media reagents were purchased from Sigma-Aldrich.

Toxicity Analysis
The toxicity of 69 ILs toward Shewanella sp. was analyzed by using two different methods: 1) agar disk-diffusion test and 2) toxicity in liquid media. Both methods are described below.

Agar Disk-Diffusion Test
The agar disk-diffusion method was used for the first time in 1940 by Heatley (Heatley, 1944) being now the most common technique for determining the antimicrobial properties of many compounds (Balouiri et al., 2016). So far, a wide variety of bacteria has been successfully tested by this method by varying the culture media and the incubation conditions as well as the evaluation criteria for the inhibition zones (Clinical and Laboratory Standards Institute, 2015). This method consists of   inoculating agar plates with a standardized inoculum of a specific microorganism. Then, the test compound embedded in small pieces of filter paper was placed on the agar surface and the Petri dishes were incubated under suitable conditions for the growth of the microorganisms. Finally, the antimicrobial activity of the test compound was evaluated according to the diameter of the inhibition growth zone. Although this method is not entirely satisfactory to determine the minimum inhibitory concentration (MIC), it brings many advantages over other techniques such as its simplicity and suitability to assess the compatibility between a wide variety of compounds and microorganisms as well as its low cost (Balouiri et al., 2016). Due to the great interest of the scientific community in the search for new and safe applications of ILs, the agar disk-diffusion method has also been used to determine the biocompatibility of these green solvents (Rebros et al., 2009;Ventura et al., 2012a). Thus, this procedure was selected to evaluate the inhibitory capacity of a wide variety of ILs on the growth of Shewanella sp. To this end, 100 µl of Shewanella sp. culture incubated overnight was spread on an LB agar plate. Each IL was added to sterile, preweighted disks of filter paper (around 6 mm of diameter), and then placed on the surface of the Petri agar dish. ILs in solid-state were weighed and then sterile-distilled water was added dropwise until the salt was completely dissolved. An aliquot of 5 µl of the salt solution was added to the disk of filter paper and the amount embedded ranged between 0.004 and 0.0125 g. All Petri dishes were incubated overnight in a static incubator at 25°C, and the radius of the inhibition zone around the filter paper was measured according to Rebros et al. (2009). Each IL was tested in triplicate and the inhibition zones reported in the present work are the average of the three replicates unless otherwise stated.

Toxicity in Liquid Media
To delve into the biocompatibility of the ILs with Shewanella sp., an alternative method based on measuring the growth rates in LB broth medium containing yeast extract, peptone, glucose, and NaCl in the presence and absence of IL was used. In order to decrease the viscosity, water-miscible ILs were weighed and mixed with sterile-distilled water to produce a stock solution (95% w/v). Viscous ILs along with those with unusual phase behavior were also prepared in the same way. The stock solutions were vortexed before adding an aliquot of 40 µl to the microcultures up to a final concentration of 2% (v/v). Nonviscous water-immiscible ILs were added directly to the cultures by volumetric measurement to produce a biphasic system 2% v/v. The cultures were inoculated with 100 µl from an overnight culture of Shewanella sp. grown in the same medium. The flasks were incubated for 24 h at 25°C with continuous and intensive shaking.
The optical density (OD) of each culture was measured every 2 h and the growth curve was determined for each case. The specific growth rate (µ, h −1 ) was calculated by selecting two points at times t 1 and t 2 in the exponential growth phase and applying the following equation:      This parameter is shown as a percentage of the growth rate in control cultures without ILs. The specific growth rates in % (µ) are reported as a percentage of the growth rate in control cultures grown without ILs (µ 0 0.13 h −1 ) applying the following equation: The specific growth rates reported are the mean of triplicate, unless otherwise stated.

RESULTS AND DISCUSSION
As preliminary information, three types of inhibition were observed in the presence of ILs. The first type corresponds to symmetric inhibition when the IL diffused correctly in the agaragar ( Figure 1A). In other cases, the asymmetric inhibition zones offer a variable zone of clearance, rather than a uniform circle around the filter paper, which can be attributed to nonuniform diffusion ( Figure 1B) (Rebros et al., 2009). Finally, there are several ILs that do not form an inhibition zone ( Figure 1C). The results obtained with the agar diffusion test were compared with the toxicity test in liquid media. As an example, Figure 2 shows the effect of different ILs on the growth of Shewanella sp. The strain was grown in LB broth with and without the respective The effect of the ILs on pH change was checked in cultures and inoculated controls using a universal indicator (20 ml) added before and after growth. In general, the influence on pH was small (changes of < 0.5 pH units). On the other hand, as can be expected, the toxicity effect is variable and greatly depends on the chemical structure of each IL. Tables 1-8 include the ILs studied and are organized by type of cation. The structure of the IL and its water solubility, the inhibition zone, and the specific growth rate (µ, h −1 ) are also presented. The influence of the alkyl substituent of the cation and the type of the cation and anion are systematically analyzed below with the aim to establish toxicitystructure relationships.

Influence of the Alkyl Substituent of the Cation of Ionic Liquid on Toxicity Toward Shewanella sp.
Although there are no data available in the literature related to the toxicity of ILs toward Shewanella sp., several works have shown that, for other microorganisms, the toxicity is directly correlated with the chain length of the alkyl substituent in the cation (Couling et al., 2006;Luis et al., 2007;Romero et al., 2008;Pretti et al., 2009;Pérez De Los Ríos et al., 2017). The influence of an increasing chain length of R1 or R2 in the imidazolium cation moiety on the cytotoxicity in marine bacteria and two types of mammalian cell cultures was also evident in HeLa cells (Stepnowski et al., 2004;Ranke et al., 2007). This effect is currently known as "side-chain effect" (Matzke et al., 2010). For highly lipophilic cations, cytotoxicity does not significantly increase with lipophilicity anymore. It is well known that lipophilicity relationships with biological activity are only linear over a restricted range (Pernak et al., 2003). For Shewanella sp., we also found that behavior. In the case of imidazolium cation combined with dicyanamide, a decrease of the specific growth rate and/or an increase in the inhibition radius was found for the series − ]. According to these results, a higher hydrophobic character in ILs would increase the possibility of their interaction with cell membrane phospholipid bilayers and the hydrophobic domains of the membrane proteins, leading to the disruption of the membrane physiological functions and, consequently, to cell death (Stepnowski et al., 2004;Latała et al., 2005;Ranke et al., 2007;Hernández-Fernández et al., 2015).
In − ] (considering both the inhibition radius and the specific growth rate). In this case, an increase in the alkyl chain length also involves a decrease in water media solubility, reducing the effective concentration of the IL in the medium. Besides, the toxic effect of the ILs could depend on the synergy effect of the combined anion and cation and not exclusively on the cation.

Effect of the Ionic Liquid Cation on Toxicity Toward Shewanella sp.
In order to study the effect of the cation, ILs with the same anion and different cation will be compared. This is the best way to systematically study the toxicity of ILs; however, it is important to point out that toxicity synergy effects could occur between anions and cations, and consequently the sequence of toxicity could be modified for different anion families. In this case, it is difficult to separate individual anion and cation contributions to toxicity. In general and in terms of cation toxicity for other microorganisms, ILs containing aromatic cations, such as imidazolium and pyridinium cations, have shown higher levels of toxicity in comparison to those including nonaromatic cations like pyrrolidinium. A higher hydrophobic character of aromatic cations could increase the possibility of interaction with the cell membrane (Stepnowski et al., 2004;Latała et al., 2005;Ranke et al., 2007). Furthermore, the planarity of the cation ring in imidazolium and pyridinium appeared to be also a relevant parameter for increasing IL toxicity, as reported in (Viboud et al., 2012). This fact could be due to the lower steric hindrance of the aromatic cation, which may favor the interactions with the lipid membrane to a greater extent. − ]. The inclusion of the methyl substituent in the R2 position of the imidazolium ring could reduce the acid proton in 2, so lowering the toxicity of the IL.
The effect of pyridinium cations was studied by comparison of imidazolium-based ILs with equal alkyl chain length and anions. In all cases, a higher toxicity was found for the imidazolium cation with respect to the pyridinium cation by comparing − ]; in this case, the toxicity of the pyridinium cation was lower than that of the [Moxa + ] cation, considering both the inhibition radius (0 vs. 1.1 cm, respectively) and the growth rate (170.7 vs. 25.4%, respectively).
For ILs based on the ammonium cation, different toxicity values were found depending on the alkyl chain length of the cation and the type of the anion. The toxicity values ranged from the highest toxicity for C1EG, C1EG1, and [N 8881 + ][Cl − ] (inhibition radius of 3 cm and 0% growth rate) to the lowest toxicity for [N 8881 + ][CF 3 SO 3 − ] (1.4 cm of inhibition radius and 179% of growth rate). Wood and Stephens (2010) found that quaternary ammonium salts were generally toxic to E. coli. However, it was evidenced that decreasing alkyl chain lengths reduced the toxicity. Our experiment reveals that quaternary ammonium salts are generally toxic, maybe due to their high lipophilic character. However, [Chol + ][H 2 PO 4 − ] and [221PG] (TEGO®-IL-P9) displayed low toxicity toward the strain tested, not inhibiting the bacterial growth in the agar diffusion test. In these cases, the cation contains heteroatoms and short alkyl chains.
Phosphonium-based ILs have been less frequently studied in comparison to other groups despite their high industrial interest (for example, in biotransformation processes such as xenobiotics degradation) (Cieniecka-Rosłonkiewicz et al., 2005). The toxicity results obtained for the phosphonium family suggest that long alkyl chains promote higher toxic effects toward the bacterium (Matzke et al., 2010). The same result was found by Stolte et al. (2008) for phosphonium cations with longer chains (C12, C16, and C18), with a strong inhibitory potential to the used inoculum. The high toxicity of phosphonium halides against Vibrio fischeri was also proved (Couling et al., 2006). However, there are possibilities to decrease the antimicrobial activity using long alkyl chains, e.g., phosphonium based on alkyl chains of 8 and 14 carbons conjugated with the chloride anion. The exchange of the halide by other anions resulted in a loss of antimicrobial activity (Cieniecka-Rosłonkiewicz et al., 2005). In this study, we found that 9 out of 12 phosphonium ILs were very toxic toward Shewanella sp. (large radii of inhibition zone and µ of 0% for the bacterial growth in liquid phase). In the case of phosphonium-based IL, the toxicity values ranged from 5.5 cm of inhibition radius and 0% growth rate ([P 14,6,6,6 + ][Br − ]) to 1.8 cm of inhibition rate and 108.3% of growth rate ([P 14,6,6,6 + ] [BF 4 − ]). For sulfonium-based ILs, only one compound was assayed, offering high toxicity since no growth of Shewanella was observed and the inhibition radius was 1.5 cm.
Effect of the Ionic Liquid Anion on Toxicity toward Shewanella sp.
The effect of the anion composition on the toxicity was analyzed by comparing ILs with different anions and the same cation. As commented above, synergy effects between anion and cation can occur, which makes it difficult to completely isolate individual anion and cation contributions. Several authors have reported that the toxicity is directly correlated to the cation nature, while the anion seems to modulate the toxicity to some extent and in specific cases (Ranke et al., 2004;Couling et al., 2006;Luis et al., 2007;Romero et al., 2008;Pretti et al., 2009 [P 6,6,6,14 + ] yielded growth rates from 0 to 155%. This fact could be attributed to the dependency of the toxicity values on the IL cation and the possible synergy toxicity effects between the cation and anion. Choosing the Less Toxic Ionic Liquids for Shewanella sp. Applications Two general possibilities for the joint application of ILs and Shewanella sp. in chemical processes are 1) the use of ILs as reaction media in biotechnological processes or 2) the use of ILs as extractants in biphasic systems with Shewanella sp. in an aqueous phase. For the first scenario, the key aspect is to find an IL compatible with this bacterium. The results obtained in this study in terms of biocompatibility have been obtained from a wide range of compounds, but they can also help us to qualitatively predict the toxicity of other nonanalyzed ILs. From this work, we can know that several ILs could serve as a medium for Shewanella sp. or could be used in different applications such as microbial fuel cells, in which Shewanella sp. act as biocatalysts and ILs are used for membrane preparation. These selected ILs have a growth rate higher than 150% and an inhibition radius equal to zero. It is important to remark that in all cases these ILs were water-soluble, which relates the idea of water solubility to biocompatibility. This is highlighted by the fact that 19 out of 29 insoluble ILs allowed no growth, while only 17 out of 40 watersoluble ILs showed no growth for Shewanella sp. Furthermore, only two water-insoluble ILs provided growth rates over 100%, [N 8,8,8 It is important to note that it is hard to find conventional organic solvents that extract polar compounds from the water with high efficiency since they would need to be water-insoluble and polar, two properties barely possible in conventional organic solvents. These limitations can be overcome in the case of using ILs as extractant medium from water solutions. Furthermore, the ILs would have to be biocompatible with Shewanella sp. In this case, the two ILs, [N 8,8

Mechanisms of Ionic Liquid Toxicity
Relatively, few mechanisms have been suggested to explain the toxicity of ILs toward marine bacteria, but membrane disruption is considered the most common. ILs at concentrations several orders of magnitude higher than those used in our study have been shown to disrupt synthetic membranes (Sena et al., 2010). In aquatic systems, the side-chain effect could be observed even with the shorter side chains, e.g., IM12 (1-ethyl-3-methylimidazolium) and IM14 (1-butyl-3-methyl-imidazolium), which means that the EC50 values for IM12 are higher than for IM14 (Stolte et al., 2007a;Stolte et al., 2007b). Our results support the membrane disruption mechanism by toxic ILs since, in general, increasing toxicity is observed with increasing IL hydrophobicity and alkyl chain length in the cation (which also contributes to IL hydrophobicity). Furthermore, the inclusion of a heteroatom such as oxygen leads to reducing hydrophobicity and thus toxicity. Nevertheless, other molecular mechanisms cannot be excluded when explaining the toxicity of ILs.

CONCLUSIONS
This work assesses the toxicity of a high number of ILs toward Shewanella sp. in order to analyze their biocompatibility and possible joint application for bioprocess implementation. We observed that the adequate combination of cations and anions can make the whole molecule safe for alive bacterial cells in bioprocess devices. Different and wide ion combinations are possible, by selecting cation-anion combinations suitable to specific applications, which highlights the necessity of screening IL toxicity toward the design of biocompatible compounds.
The results obtained allowed establishing several toxicitystructure relationships. According to the results obtained, the immiscibility of ILs in water has an important effect on the toxicity behavior of the ILs since 66% of the water-immiscible ILs tested were highly toxic, whereas only 42% of immiscible ILs were highly toxic. Furthermore, it was observed that the length of the alkyl chain of the cation presented an important impact on IL toxicity. As a general trend, an increase in the alkyl chain length, which involves an increase in the hydrophobicity, contributes to rising IL toxicity.
It was also determined that several ILs could be used as reaction media in the presence of Shewanella sp. based on their biocompatibility and water miscibility, specifically for Shewanella sp. This work demonstrates the potential combination of IL technology and Shewanella sp. The results of this study are encouraging and also suggest that new fields could be explored with Shewanella sp. and IL materials.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material; further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
HK: Data curation, formal analysis, investigation, methodology, writing-original draft. AR: conceptualization, resources, funding acquisition, supervision, writing-review and editing. MS and VO: data curation, formal analysis, writing-original draft, writing-review and editing. MK: funding acquisition, supervision. JH: writing-review and editing. FH: conceptualization, funding acquisition, project administration, resources, supervision, writing-original draft, writing-review and editing.