Toxicity of oxyanions of selenium and of a proposed bioremediation intermediate, dimethyl selenone

INTRODUCTION

The reduction and biomethylation of selenate (SeO) and selenite (SeO) by microorganisms is not clearly understood. Proposed mechanisms suggest that dimethyl selenone [(CH₃)₂SeO₂] is an intermediate of this reduction pathway [1-3]. Selenium-resistant microbes reducing selenium oxyanions to form elemental selenium and/or methylated, volatile selenide products can decrease the concentration of selenium compounds in solution and thus decrease the toxicity of these compounds in their immediate environment [4, 5].

The toxicity of selenium oxyanions and some of the volatile products of bioconversion has been examined, and though the results are somewhat variable depending on the organism tested, the general consensus has been that selenate is less toxic than selenite and that dimethyl selenide, (CH₃)₂Se, is less toxic than both of these oxyanions [4-9]. Recently the comparison of organoselenium production by selenium-resistant bacterial cultures of Pseudomonas fluorescens K27, amended with either selenate, selenite, or dimethyl selenone, showed that the cultures produced significantly more dimethyl selenide and dimethyl diselenide when amended with dimethyl selenone compared to those with added selenate or selenite [3]. This suggested to us that dimethyl selenone might indeed be viable as an intermediate in this reduction process from Se⁶⁺ or Se⁴⁺ to elemental Se or selenide, and if it were produced in a bio-remediation mechanism, would therefore aid in the detoxification of the immediate environment.

The discovery of biological reduction of dimethyl selenone by anaerobic bacteria sparked our interest in the relative toxicity of this proposed intermediate compared to the oxyanions of selenium most commonly found in the environment, selenate and selenite. To examine this toxicity, tests with these three chemical species were performed (1) with a commercial bioassay (Microtox®) that uses the bioluminescent bacteria Vibrio fisheri [10-13] and (2) by comparing growth inhibition and growth rate inhibition using P. fluorescens K27, a selenium-resistant bacterium isolated from Kesterson Reservoir in the San Joaquin Valley of California, USA [14]. A selenium-resistant bacterium was used for the growth inhibition and change in growth rate experiments because previous experience with this microbe [3, 15] has shown that it has some ability to grow in the presence of all three reagents, SeO, SeO, and (CH₃)₂SeO₂, and therefore could be examined in this way. Also, we wanted to compare the toxicity of (two of) these chemical species on a selenium-resistant microbe to determine whether the relative toxicity of these compounds was significantly different from what has been reported in the literature from studies that used toxicity assays with non-selenium-resistant organisms. As far as we know this is the first report of toxicity experiments using dimethyl selenone.

MATERIALS AND METHODS

Growth inhibition

The method used to determine growth inhibition is based on a procedure described by Bringmann and Kühn [19] and Strotmann et al. [20]. Generally, growth inhibition (GI) is expressed as

where OD₅₂₆ of all poisoned samples is determined when the control reaches the late exponential phase of growth [20, 21]. Because the bacteria were poisoned in the early exponential phase of growth and the experiment was terminated just before the control reached its stationary phase, the following formula was used to calculate the percentage of growth inhibition in our experiments:

The initial OD (OD_initial) was measured at the point that the bacteria were amended with the toxicant, and OD_final was measured just before the control reached the stationary phase. For this method, the EC50 values were determined as the toxicant concentrations at which GI = 50% calculated in the linear range of growth inhibition plotted versus log(concentration).

Specific growth rates (SGRs) were determined in the exponential phase of growth of control and test samples and were calculated using the following formula [22]:

Specific growth rates are expressed per hour (h⁻¹). The EC50 values were calculated as the concentration of toxicant that causes a 50% decrease in SGR as compared to growth of the unpoisoned control calculated in the linear range of growth rate plotted versus concentration.

Microtox

The Microtox toxicity experiments were carried out following the manufacturer's procedures in a sodium chloride solution with pH of approx. 7 [23]. All tests were performed at 15°C using the Microtox dedicated spectrophotometer (λ = 490 nm), which provides subambient cooling for 15 sample and control cuvettes during the testing period (Model 500 Analyzer). Selenate and DMSeO₂ experiments were performed at 5-, 10- and 15-min exposure times. Selenite was so toxic to this bacterium at the concentrations tested that only 5-min tests were carried out [24]. For longer tests the final amount of light measured at the test's end was so small that it introduced unacceptable amounts of error into the subsequent calculation of EC50s. Lower selenite concentrations were not examined in the longer time tests but would probably have solved this problem. Following the manufacturer's protocol, each test used four different toxicant concentrations and a reagent blank. These five bioluminescence readings were used to determine the concentration at which the light output dropped to 50% of that of the control.

The result of this acute toxicity test is also an EC50, defined as the concentration of each compound (calculated as ppm Se) necessary to cause a decrease in bioluminescent light output of 50% at the final time of each test. All EC50 values were calculated as means at a 95% confidence level using the Student's t test [11].

Table Table 1.. Acute toxicities^a of seleniferous compounds to Vibrio fisheri

Compound	EC50 (ppm Se)	95% CI	Time (min)	n
SeO	110	±13.6	5	7
(CH3)2SeO2	914	±77.6	5	8
	490	±29.5	10	8
	375	±29.7	15	8
SeO	7,725	±1,591	5	5
	5,762	±1,620	10	5
	5,669	±2,811	15	5

• ^a EC50 is the concentration of each compound necessary to cause a decrease in bioluminescent light output of 50% at the final time of each test.

Microbics' protocol [23] suggests determining the toxicity of a standard toxicant with a well-established EC50 to check the performance of the Microtox system. Phenol (100 ppm C₆H₅OH in H₂O; provided by Microbics and double-checked with our own reagent, from Matheson, Coleman, and Bell, Norwood, OH, USA) was used for that purpose in 5-min tests.

RESULTS

Growth inhibition

Figure 1 shows the overall results of growth inhibition experiments with P. fluorescens K27 amended with selenate, selenite, and DMSeO₂. Linear regressions of growth inhibition versus log(concentration) were used to determine the EC50 values recorded in Table 2; linear range, number of data points (n), and linear regression coefficients (r) are also given in Table 2.

Table Table 2.. Toxicities of seleniferous compounds toward Pseudomonas fluorescens K27 given as the EC50, the toxicant concentration that leads to a 50% growth inhibition or to a 50% decrease in specific growth rate

Method	(CH3)2SeO2	SeO	SeO
Growth inhibitiona
EC50 (ppm Se)	21	61	276
r	0.999	0.988	0.998
range (ppm Se)	log(7.9-39.5)	log(39.5-79)	log(39.5-1,185)
n	6	4	9
Growth rate inhibitionb
EC50 (ppm Se)	13	53	412
r	0.962	0.970	0.967
range (ppm Se)	3.95-39.5	15.8-158	323.7-1,580
n	6	8	12

• ^a The EC50 for growth inhibition (GI) was determined as the toxicant concentrations at which GI = 50% calculated in the linear range of growth inhibition plotted versus log(concentration).
• ^b The EC50 for growth rate inhibition was calculated as the concentration of toxicant that causes a 50% decrease in specific growth rate as compared to growth of the unpoisoned control calculated in the linear range of growth rate plotted versus concentration.

Specific growth rate

Following the growth of P. fluorescens K27 in anaerobic cultures amended with varied concentrations of selenate, selenite and DMSeO₂ allowed the determination of the corresponding SGRs as displayed as an overview in Figure 2. The average SGR of unpoisoned cultures of P. fluorescens K27 grown anaerobically at 28°C in DM-N medium was determined to be 0.126 ± 0.017 h⁻¹ (n = 50) and is represented by the upper labeled horizontal line in Figure 2. The EC50 values (Table 2), where the organism is growing at only half the SGR of the control, were determined by linear regression of SGR versus concentration. The plot in Figure 2 gives an indication of the effects of increasing concentrations of the selenium species on SGR. Observing SGRs also revealed that selenite amendment in the range of 10 to 100 ppm slightly enhanced the growth of this bacterium in this medium. Neither at concentrations of ⩾79 ppm Se (1.0 mM, DMSeO₂) nor at concentrations of ⩾3,950 ppm Se (50 mM, selenite) was any K27 bacterial growth ever observed. In contrast, even at 15,800 ppm Se (200 mM, selenate), P. fluorescens K27 was still able to grow, but with extended lag phases of 4 to 8 weeks.

It must be noted that all reported selenium concentrations in this work are the nominal, that is, actual amended, selenium concentrations. No tests were performed to determine whether the bioavailability of these toxicants in solution had been affected by other components in the media; however, our use of very simple, well-defined media (DM-N medium for work with K27 or a simple salt solution for the V. fisheri assay) was an effort to minimize these effects [26].

Table Table 3.. Comparison of relative toxicities of seleniferous compounds

Subject	Time	Selenite (ppm Se)	Selenate (ppm Se)	Dimethyl selenone (ppm Se)	Reference
Vibrio fisheri	5 min	EC50 110	EC50 7,725	EC50 914	This work
	5 min	EC50 215	—	—	[36]
	15 min	EC50 33.4	—	—	[30]
	20 min	EC50 49.1	EC50 > 1,100	—	[37]
Daphnia magna	48 h	LC50 3.02	LC50 4.07	—	[6]
Chironomus riparius	48 h	LC50 7.95	LC50 16.2	—	[6]
Pseudomonas fluorescens K27				EC50 21
growth inhibition	10 h	EC50 276	EC50 61		This work
Pseudomonas fluorescens K27				EC50 13
growth rate inhibition	up to 60 d	EC50 412	EC50 53		This work
Pseudomonas fluorescens wild				—
type growth rate inhibition	2 h	EC50 52	—		[30]
Green alga	24 h	EC50 143	EC50 61.5	—	[24]
Chironomus decorus	48 h	LC50 48.2	LC50 23.7	—	[32]
Chick embryos	17 d	LD50 0.3	LD50 0.13	—	[29]

DISCUSSION

The advantages of using short-term bacterial motility [27, 28] or bioluminescence bioassays [10, 11] for toxicity are obvious: quick results and high throughput, high sample population (∼10⁶ bacteria per sample), and cost efficiency. Uncertainty with this method may stem from turbidity and sample color complications because the actual measurement is spectrophotometric. Procedural protocols have been developed to allow colored or turbid samples to be evaluated [23]. Because the dilution volumes are on the microscale, small errors in the pipetting reagents involved may have large consequences, unlike with more conventional large-scale bioassays. For the acute tests carried out in this study, no information can be gained on the effects of the toxicant on growth and cell division of the microorganism.

The data in Table 1, determined using V. fisheri, show that selenite had the lowest EC50 among the three compounds tested (110 ppm Se; n = 7; t = 5 min). Selenate is substantially less toxic to this organism, with EC50 of 7,725 ppm Se for the same 5-min test (n = 5). Dimethyl selenone's 5-min average EC50 was 914 ppm (n = 8) and falls between that of the two more oxidized species. Selenium toxicity data from the literature and our results are listed in Table 3. One additional point is that the Microtox assay experiments, which used a manufacturer-supplied unbuffered culture “medium,” may be affected at the highest selenite concentration by a substantial pH change from neutral. For V. fisheri this is considered to be unimportant at a pH as high as 8 [23]; however, a test showed that the highest selenite concentration of the four used in each run had a pH of 8.8. (Selenite anion is the strongest base of the three toxicants examined.) Therefore, although the relative toxicity order of the three selenium species examined is certainly valid, a small deviation from the absolute value of the 5-min selenite EC50 is possible.

The lower LC50 data reported by Ingersoll et al. [6] and the other data in Table 3 are in line with the increasing toxicity seen in Table 1 with increasing assay times (>5 min). The longer the organisms were exposed, the lower the calculated EC50 [24]. Interpreted in this way, our much shorter V. fisheri tests appear reasonable and in accord with 24- and 48-h data and even the 17-d data of one research group [29].

Incubation times were also long for growth and growth inhibition experiments with P. fluorescens K27. As Figure 1 shows, the growth inhibition method led to a typical inhibition graph [20] and allowed the determination of EC50 values with good linear regressions (Table 2). However, unlike many other organisms reported in the literature, selenate was more toxic than selenite for P. fluorescens K27. This strain, isolated from a selenium-contaminated site (Kesterson Reservoir, San Joaquin Valley, CA, USA) [14] tolerates more selenite than the wild type (EC50 = 52 ppm, determined by growth rate inhibition in the first 2 h after the addition of the toxicant) [30]. It is surprising that selenate is more toxic to this microbe than selenite because the main selenium species in the San Joaquin Valley is selenate [31]. The selenium concentrations reported for that site are, however, much lower than the EC50 values for selenate and selenite and P. fluorescens K27 reported here: as selenium these naturally occurring amounts range from less than 0.2 ppm [31] to 1.4 ppm [14].

Unlike for the V. fisheri assays, growth inhibition experiments, and the growth rate inhibition method suggested by Paran et al. [30], no time limit was set to study the effect of varied concentrations of seleniferous compounds on the specific growth rate of P. fluorescens K27 in the research reported here. Following growth until the cultures had reached the stationary phase showed a drastic decrease in growth rate for all three selenium compounds in the same concentration ranges where growth inhibition occurred (Fig. 2 and Table 2) but also revealed that even at 15,000 ppm Se (200 mM, selenate), K27 was still able to grow (Fig. 2). It has to be mentioned, though, that beyond the selenate concentrations causing the steady decrease of the specific growth rate, longer and longer lag phases were observed with those increasing selenate concentrations. At 200 mM selenate, the apparent lag phase was between 4 and 8 weeks. In sterile control experiments no bacterial growth was ever observed, showing that the cultures growing in the presence of 100 and 200 mM selenate were P. fluorescens K27 and not simply contaminations. The long lag phases indicate that at such high concentrations, highly resistant strains of K27 were enriched. This was verified by streaking the 200 mM amended cultures and unpoisoned cultures of P. fluorescens K27 on plates containing 200 mM selenate. Colonies appeared within days when inoculated with cultures previously grown in the presence of 200 mM selenate, whereas on plates inoculated with original, unpoisoned cultures, it took weeks for the first colonies to grow. Neither for selenite nor for DMSeO₂ was the development of such high resistance ever observed.

Comparing growth inhibition and growth rate inhibition of P. fluorescens K27 (Table 2), the EC50 values lie in the same order of magnitude. However, only for selenate were the numbers relatively close. Determined by growth rate inhibition, the EC50 of DMSeO₂ was almost 40% lower than the value obtained by using the growth inhibition method, whereas for selenite it was almost 50% higher. In comparison, the observation of a negative and a positive deviation indicates that the two methods indeed lead to different results and not just to a systematic difference between the two methods involved. Considering the results of our growth rate inhibition experiments, the term growth inhibition was found to be somewhat misleading: 100% growth inhibition was determined at concentrations of 10 mM selenate and 15 mM selenite, even though growth occurred far beyond these concentrations. The limitation in this method is the definition of growth inhibition, which depends on the relative growth of an unamended control and does not, in any way, take an extended lag phase into account.

In addition, comparing the highest concentrations where growth was observed, DMSeO₂ (39.5 ppm Se) is still the most toxic to P. fluorescens K27 among the three selenium compounds investigated; however, selenite (2,370 ppm Se) is more toxic than selenate (15,800 ppm Se), as could generally be expected from the literature. On the other hand, the 50% growth inhibition concentrations determined by Ibrahim and Spacie [24] and the LC50 and LD50 values for these two selenium oxyanions, as reported by Maier and Knight [32] and Palmer et al. [29] respectively, show that selenate was more toxic than selenite in their tests.

In their work with green algae, Ibrahim and Spacie [24] recognized the difficulty in comparing their EC50 values with the general toxicity trend of SeO and SeO in the literature and specifically with data reported for another algae bioassay [33]. In their discussion they mention the importance of investigating the “mechanism(s) whereby these chemical forms cause bioeffects….” [24]. Our examination of the relative toxicity of DMSeO₂, Challenger's [1] proposed intermediate, as compared to SeO and SeO is an effort to do this. Although still combined with oxygen, selenium in DMSeO₂ can be assigned a formal oxidation state of zero, whereas that of selenide is more reduced at -2 [34]. Yet, unlike elemental selenium, DMSeO₂ is freely soluble in water. The intermediate toxicity of (CH₃)₂SeO₂ in an assay using V. fisheri suggests that if these bacteria do indeed supply electrons to reduce and detoxify oxidized forms of selenium to DMSeO₂, then this chemical species as an intermediate would be less toxic than the starting material if that oxyanion were selenite but more toxic than the initial oxyanion if it were selenate. In reference to the following discussion, it must be remembered that the Microtox assay's organism, V. fisheri, is not considered to be selenium resistant because it has been exposed to selenium at levels present in its marine environment (≪1 ppm Se). Furthermore, we have no information on how this microorganism responds metabolically to contact with selenium; therefore, the above discussion is speculation on our part.

The results of the growth inhibition and growth rate inhibition with a selenium-resistant bacterium tell a different story. Toward P. fluorescens K27, DMSeO₂ is more toxic than both selenium oxyanions using these two assays. Based on these results alone one might rule out DMSeO₂ as a possible intermediate of the selenium reduction pathway. However, our recent results with amending bacterial cultures have shown that DMSeO₂ is converted into less toxic [8, 9] and less water-soluble volatile organoselenides at much higher rates than selenite and selenate by both P. fluorescens K27 [3] and several strains of purple nonsulfur bacteria (V. Van Fleet-Stalder, unpublished data). This means that if DMSeO₂ is an intermediate in the bioreduction process for bacteria that have developed a resistance mechanism, the rate-limiting step in the formation of volatile organoselenides from selenite or selenate would probably lie between selenite (or selenate) and DMSeO₂. Thus, relative toxicity aside, high concentrations of DMSeO₂ would not be allowed to build up and do damage in the cell because DMSeO₂ is so readily converted to more reduced, volatile, and less-soluble forms.

Anaerobic, selenite-amended cultures of P. fluorescens K27 turn brick red, most likely from formation of red elemental selenium, whereas selenate-amended cultures do not, and recent work with this organism has shown that the amounts of Se⁰ produced are at least 10 to 20 times larger in selenite-compared to selenate-amended cultures for the same incubation time. Furthermore, this same work has shown that selenite-amended cultures of K27 produce organoselenium compounds in the log phase, whereas in selenate-amended cultures this production occurs in the stationary phase of growth [35]. Together these data, relative toxicities, relative production of organoselenium species, and the growth phase in which they are produced, indicate that selenite and selenate are likely to be metabolized through different biochemical pathways by P. fluorescens K27. Currently, we are examining the relative distribution of the entire suite of selenium-containing species, Se⁰, Se(IV), Se(VI), and organoselenium compounds, in K27 cultures amended with selenium-containing oxyanions. The distribution of each chemical member in selenite- and selenate-amended cultures may shed even more light on the possible mechanisms of selenium toxicity and bioremediation pathways.