Arsenic (As) is a highly toxic, carcinogenic, clastogenic and teratogenic metalloid (Slyemi and Bonnefoy, 2012). Arsenic occurs primarily as inorganic forms of pentavalent arsenate As(V) and trivalent arsenite As(III), with the latter being regarded as the most mobile and toxic form (Yang et al., 2012). Arsenicals generated from natural and anthropogenic sources are the widely distributed contaminants of freshwater, groundwater and seawater (Stolz et al., 2010; Slyemi and Bonnefoy, 2012; Rodríguez-Lado et al., 2013). Combustion of fossil fuels, mining, and applications of arsenic-containing pesticides/herbicides account for most of the arsenic pollution sources (Stolz and Oremland, 1999). Microorganisms are the principal drivers of arsenic chemical speciation by redox transformations and influence arsenic mobility and toxicity. Most of bacteria and archaea virtually carry arsenic-resistance (ars) genes that potentially confer resistance to As(V) and/or As(III) (Jackson and Dugas, 2003). The phenomenon (widespread occurrence of ars genes) indicates the ubiquitous distribution of this toxic metalloids in nature.

Some of microorganisms have a significant impact on the biogeochemical transformations of arsenic and are considered as bioremediation reagents for arsenic contamination (Zhang et al., 2015). Increasing interests have been drawn to investigate microbial detoxification of arsenic compounds, particularly to study the arsenic metabolic mechanisms. Furthermore, some microorganisms such as chemoautotrophic arsenite oxidizers oxidize As(III) to As(V) to gain energy for cell growth when fixing inorganic carbon (CO₂). Instead, microbes that use As(V) as an electron acceptor in anaerobic respiration lead to the production of As(III), indicating that arsenic plays various roles in microbial metabolism (Heinrich-Salmeron et al., 2011; Kruger et al., 2013). Arsenic metabolism and its genetic determinants have been investigated in many Gram-positive and Gram-negative bacteria (Kruger et al., 2013). Bacteria acquired several different arsenic-metabolic strategies including arsenite oxidation (aio system) in arsenite oxidizers which oxidize As(III) to As(V) (Oremland and Stolz, 2003; Heinrich-Salmeron et al., 2011), anaerobic arsenate respiration (arr system) in arsenate-reducing microbes which respire and reduce As(V) to As(III) (Oremland and Stolz, 2003; Páez-Espino et al., 2009), arsenate system (ars system) in arsenate-resistant microbes which reduce cytoplasmic toxic As(V) to As(III) (Oremland and Stolz, 2003; Kruger et al., 2013), and arsenic methylation (arsM system) in arsenic methylating bacteria which convert inorganic arsenic into methylated arsenic (Qin et al., 2006; Kruger et al., 2013; Zhang et al., 2015). Arsenic-metabolic genes are assembled with multiple arsenic operons in some bacteria. For example, both arr and ars operons are found in Shewanella sp. (Saltikov and Newman, 2003); Thiomonas sp. possess two operons (aio and ars system) (Arsène-Ploetze et al., 2010); Herminiimonas arsenicoxydans has aio and ars (Muller et al., 2007); Cyanidioschyzon sp. 5508 has aio, arr, and ars operons (including arsM) (Qin et al., 2009). Arsenic-resistant microorganisms possibly benefited from multiple arsenic-resistance operons, e.g., they can utilize different detoxification strategies under the complex environments.

With extraordinary metabolic versatility, R. palustris has been widely studied for wastewater treatment, bioremediation, hydrogen production and electricity generation (Zhao et al., 2011; Fu et al., 2014; Liu et al., 2015). R. palustris is possibly exposed to a high concentration of arsenic under the above field application conditions. Despite considerable research on the arsenic-metabolic mechanisms in chemotrophic bacteria, only few studies have been conducted in anoxygenic phototrophic bacteria. R. palustris CGA009 has three arsenic-resistance operons (ars1, ars2, and ars3) in the chromosome (Qin et al., 2006). Deciphering the arsenic resistance mechanism(s) in anoxygenic phototrophic bacteria may facilitate engineering stronger arsenic resistant strains with desirable features in industrial applications. The objectives of the present study were as follows: (i) to examine the arsenic resistance capacity; (ii) to exam the differential regulation of the three arsenic operons; (iii) to analyze the arsenic speciation in anoxygenic phototrophic bacteria by using R. palustris CGA009 as a model organism; (iv) to propose a working model for arsenic resistance in R. palustris CGA009.

Effects of As(V) (0.5~6.0 mM) and As(III) (0.5~2.5 mM) on the bacterial growth of R. palustris were tested in modified Ormerod medium and under anaerobic–light conditions (Figure 1). Negligible difference was observed between the culture supplemented with 0.5 mM of As(V) and the control (no AS(V)), indicating that a low concentration of As(V) was not toxic to R. palustris. The cells could retain at least 60, 70, and 80% of the control growth when 3.0 mM, 2.0 mM and 1.0 mM of As(V) were added to the culture respectively, indicating a good resistance to As(V). However, higher concentrations of As(V) (i.e., 4.0 mM to 6.0 mM) severely inhibited the bacterial growth with the most significant inhibition (nearly 100%) at the concentration of 6.0 mM. R. palustris was more sensitive to environmental As(III) than that in As(V) as shown in Figure 1B. For example, 2.0 mM and 2.5 mM of As(III) inhibited up to 90 and 100% of the bacterial growth, respectively. However, cells retained at least 70 and 60% of the control growth when 1.0 and 1.5 mM of the As(III) were added, respectively. Furthermore, we studied the relationship between the growth rate and the arsenic concentrations (Figure 2). Median effective concentration (EC₅₀) values for As(V) and As(III) were estimated to be 2.44 and 1.55 mM, respectively.

LV et al. investigated the arsenic resistance in three purple non-sulfur bacteria and reported the EC₅₀ for As(V) and As(III) Rhodopseudomona palustris CQV97 (2.4 and 0.9 mM, respectively), Rhodobacter azotoformans 134K20 (1.6 and 0.2 mM, respectively) and Rhodobacter capsulatus XJ-1 (1.9 and 0.3 mM, respectively) (Lv et al., 2012a,b). The EC₅₀-value for arsenic in strain CQV97 was higher or comparable to E. coli, P. aeruginosa, Rhodococcus equi, and Methylosinus trichosporium. However, compared to those bacteria with a “high” arsenic resistance capability (>100 mM), there is still considerable room for further improvement.

The genome-mining analysis showed that arsRBC operon existed in seven R. palustris strains (CGA009, HaA2, TIE-1, DX-1, BisB5, BisA53, and BisB18) (Figure 3), while additional arsRM operon was only found in strains CGA009, HaA2, TIE-1, DX-1, and BisB5 (Figure 3) (Lv et al., 2012a). The first operon resembles ars operon (named ars1 operon, arsRCBH type), consisting of arsenite-responsive transcriptional regulator gene (arsR1), arsenate reductase gene (arsC1), arsenite efflux pump gene (arsB) and NADPH-dependent flavin mononucleotide reductase gene (arsH). The second operon [named ars2 operon, arsRRCC (acr3) type] contains two arsR genes (arsR2 and arsR3), two As(V) reductases genes (arsC2 and arsC3), one arsenite permease gene (acr3), and two genes encoding the hypothetical proteins. Both ars1 and ars2 operons encode these proteins that perform As(V) reduction and As(III) extrusion mechanisms. The third operon (named ars3 operon, arsRM type) consists of arsenite-responsive transcriptional regulator gene (arsR4) and As(III)-methyltransferase gene (arsM). ars3 operon encode methylation protein that methylate As(III) to a number of methylated intermediates such as onomethylarsenite [MMA(III)] and dimethylarsenite [DMA(III)]. It should be note that arsenic resistance in microorganisms did not show a direct correlation with arsenic operon number(s) (Kruger et al., 2013). For example, at least five arsenic operons were found in Herminiimonas arsenicoxydans while its tolerance to arsenic was much lower than that in Corynebacterium glutamicum which has only two arsenic operons. Despite of their widespread distribution in various bacteria, the arsenic multi-operons were understudied. Due to arsenic multi-operons co-existing in one microorganism, it is difficult to define the arsenic resistance mechanisms. Therefore, it is necessary to further investigate how these arsenic multi-operons were differentially regulated and their contributions to arsenic speciation.

In this study, we preliminary investigated their respective gene expression in the three ars operons by using quantitative real-time PCR in order to understand which one was actively involved in arsenic resistance in R. palustris CGA009. It should be noted that studying ars operon regulation by As(V) is difficult because it can be quickly reduced to As(III) by As(V) reductase in vivo. Therefore, we first examined the effect of As(III) (rather than As(V)) on the expression of ars1, ars2 and ars3 operons. To do that, we selected the functional genes arsB, arsC2, and arsM as marker genes for ars1, ars2, and ars3 operons, respectively (Figure 4). Gene expression of arsB, arsC2, and arsM were undetectable when As(III) was absent, indicating the three operons were not expressed without arsenic induction. arsB was not transcribed when As(III) was added to the culture at concentrations ranging from 0 to 1.0 mM. Remarkably, the expression of ars2 operon (arsC2) was readably detected when cells were even exposed to a low level of As(III) (0.01 μM). Compared to the control, its transcript level was increased 9.5, 23.8, and 126.8–fold when 0.01, 0.1, and 1.0 mM of As(III) were added respectively, demonstrating that ars2 expression was up-regulated by As(III). However, arsM was not transcribed when the cells were exposed to 1.0 μM of As(III), indicating that gene product of arsM was not critical for detoxifying As(III) under low environmental As(III) conditions (<1.0 μM). However, its expression level was 3.86-fold higher than the control when 1.0 μM concentration of As(III) was present in the medium (Figure 4). Furthermore, arsM transcript level increased with the increasing environmental As(III) concentration (Figure 4). It should be noted that the expression levels of arsM and arsC2 were almost equal when As(III) concentration reached a 1.0 mM, showing that R. palustris CGA009 probably required expression of ars2 and ars3 operons to detoxify the high concentrations of As(III).

The expression dynamics of arsB, arsC2 and arsM (ars1, ars2, and ars3 operons) were investigated in R. palustris CGA009 at different phases of growth (Figure 5). arsC2 expression was highly induced by arsenate at the log-phase (between 43 and 60 h); the expression level decreased when cells entered the stationary phase (67–80 h). A similar expression pattern was found in arsM; however, a relatively low level of expression of arsM was recorded during the whole growth phase, compared to that in arsC2. ars1 transcript level was much lower than those in ars2 and ars3 at different phases of growth.

Unfortunately, only few studies have been conducted to exam the differential regulation of arsenic multi-operons by arsenic. In bacteria and archaea, ars operons are often controlled by ArsR (Figure 3). When As(III) is absent, ArsR binds to the operator/promoter region of the operon and thus repressed the downstream genes' transcription. When As(III) is available, ArsR binds to it and goes through conformational changes, resulting in dissociation from the operator/promoter region. In our transcript analysis experiments, we could not detect a significant ars1 expression (Figure 4). However, this observation was not new. For example, in C. glutamicum, only arsC1' was constitutively transcribed though there were two arsenic resistance operons (ars1 and ars2) (Villadangos et al., 2011). Furthermore, the expression of ars3 operon in R. palustris CGA009 was induced with the presence of 1.0 μM of As(III), indicating it may only contribute to the arsenic detoxification under the higher concentrations of As(III) (see next). However, the expression pattern was not similar to those previously observed in Synechocystis sp. PCC 6803 and P. alcaligenes NBRC14159 (López-Maury et al., 2003; Zhang et al., 2015). Synechocystis sp. PCC 6803 has at least two operons (ars and arsM) on its chromosome. ars operon was regulated by ArsR though arsR was located far away from ars. DNase I footprinting experiments indicated that ArsR binds to two 17-bp direct repeats (ATCAA(N)6TTGAT) in the promoter-operator region. However, the upstream sequence of arsM does not have the 17-bp repeats (López-Maury et al., 2003). Authors proposed that ArsM is constitutively expressed whilst how ArsM expression is regulated has yet to be investigated. In P. alcaligenes NBRC14159, PaarsM was expressed in the absence of As(III) and the expression was further enhanced by As(III) exposure (Zhang et al., 2015). Our results revealed that the expression of the different arsenic operons were affected by growth cycle: the maximal expressions of both ars2 and ars3 operons in CGA009 appeared at early log-phase (43 h) and maintained at high level during the growth of middle log-phase (55 h). It is different from expression patterns reported in the earlier studies in Shewanella sp. ANA-3 (Saltikov et al., 2005). The highest arr expression appeared at the log-phase while peak expression level of ars was observed in the stationary phase, respectively. That transcription of the ars operon in Shewanella sp. ANA-3 could involve other factors, such as those related to quorum sensing and energy production. In fact, a quorum-sensing-based response was shown to be a second regulatory circuit for aio transcription in A. tumefaciens (Kashyap et al., 2006).

HPLC-ICP-MS analysis in middle log-phase demonstrated that As(V) was reduced to As(III), indicating that strain CGA009 detoxified arsenate by reducing As(V) to As(III) by ArsC (Figure 6). However, we failed to detect dimethylarsine (DMA), monomethylarsine (MMA) even in the growth stage where arsM transcript approached to the highest level (Figure 6). DMA and MMA, the intermediates produced in As(III) methylation process did not accumulate in the cells and were immediately converted into volatile trimethylarsine (TMA). Thereafter, TMA were rapidly expelled to extracellular space (Qin et al., 2006). In the same study, Qin et al. heterologously expressed the arsM gene from R. palustris CGA009 in an arsenic-sensitive strain of E. coli. Their results showed that ArsM catalyzed the formation of a number of methylated intermediates from As(III), with TMA as the end product and increased the arsenic resistance, indicating that it was very possible for arsM to be functional in vivo (Qin et al., 2006).

A working model for the arsenic detoxification by the arsenic multi-operons was proposed in R. palustris (Figure 7). In this model, the As(V) enters into cells through inorganic phosphate (Pit) or phosphate specific transport (Pst) systems (Kruger et al., 2013); once As(V) arrives inside the cells, it is reduced to As(III) by ArsC produced from ars1 and/or ars2 operon (possible at a low expression) (Figure 3). As(III) formed in the reaction then inactivates ArsR, initiating the transcription of the arsRRCC (acr3) operon (ars2). Due to the two copies of arsC in ars2, it allows to reduce arsenate more promptly, leading to As(III) accumulation in the cells. If the accumulated As(III) could not be expelled out of the cells by arsenite permease (Acr3), the increasing As(III) triggers the ars3 transcription by releasing ArsR which originally binds arsM promoter/operator. Arsenic resistance genetic units ars2 and ars3 in R. palustris contribute to detoxifying arsenic at a high dose because their transcription level is comparable when cells are treated with 1.0 mM arsenite. With cooperation of ars2 and ars3, R. palustris detoxifying As(III) by extruding it out of the cell by As(III) transporter (Acr3) or transforming As(III) to volatile methylated As(III) (TMA) by ArsM. It is reasonable to assume that ars2 is more important when cells are exposed to lower levels of arsenic due to its activity expression. However, a relatively complex arsenate detoxification system described here suggests that R. palustris growing in the natural environment must be equipped to deal with rapidly changing and perhaps relatively high levels of arsenic.

This study provided a novel insight into arsenic resistance mechanisms in R. palustris CGA009, a member of anoxygenic phototrophic bacteria. R. palustris possessed good arsenic resistance which possibly linked to the arsenic multi-operons in the chromosome. Our results showed ars2 and ars3 operons were upregulated by increasing As(III) concentrations while ars3 operon was only expressed when exposed to the arsenic concentrations more than 1.0 μM. However, the expression of ars1 operon was very low, indicating that ars1 may not be actively used. Collectively, our preliminary data showed that arsenic was possibly transformed by the combination mechanisms of the cytoplasmic As(V) reduction, As(III) extrusion and arsenic methylation when exposed to arsenic at a high concentration.