Novel and unexpected bacterial diversity in an arsenic-rich ecosystem revealed by culture-dependent approaches

Acid Mine Drainages (AMDs) are extreme environments characterized mostly by heavy metal contaminations and very acidic conditions. It has been already shown that the presence of metals can lead to a decrease of 99.9% of the overall bacterial diversity in soil [1]. The combination of high metal concentration and acidic pH in AMDs further trigger this process, and various studies concluded on the occurrence of a low in situ bacterial diversity [2–4].

The Reigous is a small creek (pH 2.7-3.4) flowing down the ancient mining site of Carnoulès (Gard, France) characterized by high arsenic (up to 350 mg.l^-1) and iron (up to 2700 mg.l^-1) concentration in waters [5]. A follow-up study linked the progressive arsenic removal to bacterial activities, especially their As(III)-oxidizing activities [5]. In order to understand the biological processes occurring in situ, several studies aimed at determining the bacterial diversity in the water body or in the sediment.

In the water, both culture-dependent and culture-independent approaches had already been undertaken. The culture-independent studies pointed out the low procaryotic diversity [3, 6, 7], characterized by a stable composition over time. In parallel, only three bacterial genera were isolated from the water, corresponding to the genera Thiomonas sp., Acidithiobacillus sp. and Burkholderia sp. [8–11]. Thiomonas species was shown to be involved in As(III) oxidation [8, 11–13] whereas the Acidithiobacillus ferrooxidans strain performed iron Fe(II) oxidation [10].

In the soft sediment lying below the running water, up to now only culture-independent approaches were undertaken. A recent study, combining global metagenomic, metaproteomic and RT-PCR, and deciphering both the bacterial diversity present and the community function, highlighted a low bacterial diversity in the sediment allowing the reconstruction of 7 nearly complete genomes (called CARN1 to CARN7), 5 of which representing uncultured bacteria [14]. In-depth analysis of the genomes as well as metaproteomic and RT-PCR resulted in an integrated model of the community function illustrating potential inter-species interactions. This powerful strategy has clearly enhanced the comprehension of the community function. However, in the proposed model, each of three crucial functions was each linked to only one bacterium i.e. the As(III) oxidation only to CARN2, cobalamin biosynthesis only to CARN1/4 (both belonging to Candidatus Fodinabacter communificans) or cellulose degradation only to CARN6 [14].

The aim of the present work was to investigate the bacterial diversity in the sediment by culture-dependent approaches. In a first step, we implemented various strategies to access to the largest cultured bacterial diversity. They integrated modern cultivation methods such as the use of 1) a mineral base that mimicks the AMD water conditions, 2) low organic carbon content to avoid growth inhibition of slow-growing bacteria by fast-growing bacteria, 3) less stringent growth conditions (particularly in terms of pH) to enhance bacterial growth, 4) gellan gum instead of agar as a solidifying agent, and 5) innovating techniques such as the Soil Slurry Membrane System (SSMS) [15]. These culture-dependent approaches allowed the detection of 16 genera that were never detected previously in Carnoulès, thus increasing the overall bacterial diversity by 70%. These results demonstrated that AMD bacterial diversity is larger than previously recognized. They also illustrate that culture-dependent methods remain crucial to determine the bacterial community composition and are really complementary to genomic methods to improve our understanding of natural ecosystems.

In a second step, we investigated the functional redundancy in the cultured bacterial community for important functions i.e. As(III) oxidation, cobalamin biosynthesis, and cellulose degradation. These results showed that by maintaining the integrity of functional processes within the bacterial community, the AMD ecosystem gains more stability and robustness than previously thought.

Our study provided evidences that culture-dependent approaches enable the characterization of a different diversity compared to the one obtained by culture-independent approaches, highlighting the complementarity between the 2 approaches. We also pointed out that the community structure is not as simple as previously established (a 70% increase in overall diversity). Functional experiments showed that important community functions, such as cobalamin biosynthesis, the degradation of cellulose and the oxidation of As(III) are redundant in the ecosystem thus increasing the functional robustness essential for any ecosystem. Additionally, the isolation of neutrophilic or even alkaliphilic strains further highlights the capability carried by the whole community to adapt to in situ conditions changes such as an increase of the pH, thus improving the knowledge of the system resilience. We showed that it remains crucial to associate culture-dependent and culture-independent approaches to gain an integrative view of the community structure and function.

However, measuring the exact in situ role of each non-dominant species remains hard to determine, since they are hardly detectable with standard molecular techniques. Specific FISH-probes experiments can be performed more easily after isolation, but the relative abundances remains unknown, since no (or too few) DNA was recovered from the isolated genera. The determination of the full extent of the microbial diversity is therefore still challenging, and considerable efforts in terms of technologies and work have to be undertaken to approach this aim.

Reviewer's report

Reviewer number: 1

Referee 1 : Prof. Sandor Pongor, International Centre for Genetic Engineering and biotechnology, Trieste, Italy

Report form:

Multispecies microbial communities are a major form of life so it is crucially important to understand the factors underlying their stability. Metagenomic approaches are playing an increasingly important role in studying complex microbial communities and the research community is often led to believe that bacterial communities are what we see through culture-free metagenomic analysis. Delavat and associates have analyzed the microbial community of an arsenic rich Acid Mine Drainage (AMD) site in Carnoulès (France), a site which was believed to be of low biodiversity, based on previous metagenomic studies. Delavat and associates showed, using specific culture media, that the number of species is 70% higher than previously thought and the community includes new phylogenetic groups, such as the subclass Actinobacteridae or the order Rhizobiales which were not detected by culture-free methods. While the work is well presented and adequately documented, and shows very well the difference that exists between cultures and culture free methods in this particular case, there are a number of subjects that the authors may want to expand.

First, what is the main interest of the Carnoulès site as compared to other AMDs, and how do the present results compare to data (especially the comparison of cultured vs. non-cultured approaches) obtained on other AMD sites.

Authors’ response: Carnoulès is an interesting study site, essentially because this site is monitored since more than 15 years (first paper published in 1996 [47], 25 papers since according to Scopus). Chemists, microbiologists, mineralogists work together to elucidate how the whole system functions. Physico-chemical characteristics of sediments and waters are monitored since years and the question of the role of bacteria in the natural attenuation occurring in situ has been the subject of numerous studies. A second point concerns the origin of this AMD, which does not only result of the action of rainwater. The Reigous spring is located below the sterile mining residues, and drains away through them. This dynamic system is maintained throughout the year.

To answer to the second part of the question, we put forward that most of the studies focused on the bacterial species richness in AMDs relied on culture-independent approaches carried out on the AMD water. They have all described a bacterial ecosystem with few species. Moreover, the bacterial species detected in these different studies, carried out over the years and all year round (winter, summer…) were similar, strongly suggesting a temporal stability of the bacterial community composition. However, it is true that we cannot rule out any partial variations over time, especially within the rare biosphere [48].

Thus, our study focused on the AMD sediment, and implemented only culture-dependent approaches giving us the opportunity to explore differently an AMD.

The comment concerning the comparison of our data with those obtained in other AMDs lead to mixed answers: unfortunately there is, to our knowledge, only three studies[4, 49, 50]comparing the bacterial diversity obtained by culture and by molecular approaches in AMDs. Nevertheless, they all used only one isolating medium and did not aim at determining the largest cultured bacterial diversity. Indeed, AMDs are often either 1) screened for the bacterial diversity through molecular approaches [3, 51, 52]or 2) screened by culture-dependent approaches for targeted bacteria, such as iron- or arsenic-oxidizing bacteria [8, 18]. Thus, AMDs were never investigated in order to determine the largest cultured bacterial diversity.

Second, what are spatiotemporal variations of the community composition, can one expect a change in species composition with respect to space and, for instance, the time of the year.

Authors’ response: See the above comment for the answer.

Quality of written English: Needs some language corrections before being published

Authors’ response: The manuscript has been completely revised for corrections.

The authors would like to thank Reviewer 1 for his comments.

Reviewer number: 2

Referee 2 : Dr. Eugene Koonin, NCBI, NLM, NIH, United States of America

Report form:

The findings described here are of interest - in my opinion, mostly from the methodological standpoint, demonstrating the complementarity of culture-independent and targeted culturing approaches in the characterization of a microbiota. This seems to be becoming a common theme in modern microbiological studies. It is unclear to me in what sense the results described here are unexpected as claimed in the title. I think added microbial diversity detected by culturing is exactly what one should expect which does not make the findings unimportant.

Authors’ response: The term “unexpected” was used because, as Reviewer 1 pointed out “the research community is often led to believe that bacterial communities are what we see through culture-free metagenomic analysis”. In this sense, the detection of bacteria previously undetected in numerous culture-independent studies is not expected. It is however true that one could expect that the microbial diversity detected by both approaches would be different. But, studies aiming at comparing culture-independent and –dependent approaches often led to the conclusion that the bacterial diversity detected by molecular approaches is much higher than the one detected by culture [53, 54]. Our results provide evidence of a 70% increase of the overall bacterial diversity thanks to culture-dependent approaches. These reasons led us to use the word “unexpected”.

Also, there is some language in the manuscript that I find irrelevant. For instance (from the Conclusion section of the Abstract): “The isolated bacteria may be part of the biosphere which remained previously undetected due to molecular biases”. Obviously, the newly isolated bacteria have not been detected previously, obviously, they are part of the biota (not biosphere). What is the specific meaning of this sentence?

Authors’ response: This has been corrected in the manuscript, by adding the missing word “rare” before “biosphere”. Indeed, the bacteria detected in this study may be part of the “rare biosphere” as defined by Sogin et al. [27]. Members of this rare biosphere can remained undetected because of molecular biases exposed in the manuscript.

Quality of written English: Needs some language corrections before being published.

Authors’ response: The manuscript has been completely revised for corrections.

The authors would like to thank Reviewer 2 for his comments and for the interest expressed for the study. We fully agree with him to say that coming back to culture is becoming a common theme, and a complement to molecular approaches.

Reviewer number: 3

Referee 3 : Dr. Brett Baker (nominated by Dr. Purificacion Lopez-Garcia), Department of Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI, USA.

Report form:

Delavat et al. employed culturing-based approach in an attempt to extend our knowledge of diversity and metabolisms of an AMD impacted creek with a pH range of 2.7-3.4. Previous studies of this creek and other AMD sites (Iron Mountain being the most extensively studied) have shown that it is low in species richness, not “low biodiversity” as the authors state.

Authors’ response: Indeed, even if the term “biodiversity” is frequently used for bacterial diversity (for review, see [55]), “species richness” is more appropriate and is used in the revised manuscript.

I have several issues with the studies experimental design and the conclusions drawn from the results. If the pH of the creek is 2.7-3.4 why would you use media that ranges from 3.5-9.8, which is not even in the range of the creek? 17 of the 23 isolates listed in Table 1 were obtained on media with pH 5.5 and up.

Authors’ response: The preparation of media covering a large range of pH should allow the isolation of non-acidophilic bacteria or even alkaliphilic ones, that could not grow on acidic media. That is exactly what we obtained, since the majority of them were not isolated at pH 3.5 (close to the in situ pH), but at higher pH. Media such as R2A (pH 7.0) are frequently used for the isolation of bacteria, even in acidic environments. For example, this medium was used to isolate Thiomonas strains [8]which 1) are known to play an important role in the community function of Carnoulès, by oxidizing As(III) and 2) are not extreme acidophilic bacteria [56]. Moreover, we assume that some bacteria isolated here may not currently play important roles in situ. However, they can act as a “seed bank”, becoming important when the environmental conditions change [57].

The authors state that this was an attempt to uncover diversity not previously seen in AMD. However, proving that these are truly members of the creek would require more evidence which could be provided in the form of FISH analyses of creek samples. That being said I don’t believe that most of the isolates are endemic to the site being studied. Bacillus, Arthrobacter, Micromonospora, Pseudomonas, etc. are common contaminants and are surely ubiquitous to laboratories.

Authors’ response: We agree with Reviewer 3. Bacillus, Arthrobacter, Micromonospora and Pseudomonas genera are common contaminants. Nevertheless, they were also previously detected in the Rio Tinto [58]or in arsenic-contaminated soils [59]. Moreover, the genome of Micromonospora sp. X14 contains a fragment of the aioA gene sharing 100% nucleotidic similarity with the aioA gene of Thiomonas sp. CB2 isolated previously from Carnoulès. This indicates a horizontal gene transfer that occurred in situ. Thus, Micromonospora sp. X14 is certainly a Carnoulès indigenous bacterium.

However, FISH experiments could provide other informations, especially toward the proportion of each genus compared to the whole community. These informations would allow to determine if the isolated bacteria indeed belong to the “rare biosphere”.

Another concern stems from the fact that many of the isolates are highly related to those presently known, eg. 97-99% similarity. If these were in fact previously undetected low abundance community members you would expect some novelty from AMD.

Authors’ response: Members belonging to 3 genera, as defined by Tindall [23], were isolated in this study. One cannot expect only novelty when exploring the cultured bacteria. Most of the bacteria isolated here are newly detected in Carnoulès, but this does not mean that all of them are completely new in databases. It only means that they were not detected, maybe because of molecular biases, and/or maybe because they were, in the Carnoulès conditions, not in sufficient abundance to be detected by molecular approaches. But related strains can become abundant in other environments, leading to their detection by PCR.

The authors state in the abstract,“This work highlights that bacterial diversity in AMDs is much higher than previously envisaged”. Sure, assuming these isolates originate from the creek which is difficult to say. It possible that they are transient members of the community and likely not metabolically active.

Authors’ response: We agree with Reviewer 3, we cannot rule out that some of these bacteria may be “transient”, in the same manner to the possible “transient” presence of strains detected by culture-independent approaches. Many studies focused on determining the bacterial diversity in any environment in a single time point. Unless repeated studies, the observed biodiversity is a snapshot of what is present at a given time. This is true for both culture-dependent and –independent approaches. An interesting approach would be to repeat this culture-dependent study over time.

As has been shown all environments have a tail of “rare” species, if these genera are present in the rare component of the community it likely means they are not major contributors to the ecology of the system. Generally the greatest contributions to the field come when isolates are obtained from the dominant members of the community.

Authors’ response: This is true in general. However, some biological functions can be carried by members of the rare biosphere, as presented in the main text (see[31]for a good example). In addition, the bacteria isolated here may serve as a “seed bank”, becoming important in case of environmental changes. They are therefore important for the community stability by maintaining every ecological processes.

Authors’ response: We understand the Reviewers’ point of view, but the results provided evidences that our isolates come from Carnoulès, the best example being Micromonospora sp. X14.

L19-20 – what is a “cultured compartment” of an ecosystem?

Authors’ response: We replaced “compartment” with “bacteria”.

L21 – What makes you think that obtaining isolates from a environment will enable the “access to the largest bacterial diversity”.

Authors’ response: We added the term “cultured” in “access to the largest cultured bacterial diversity”.

L29-30 – how is a genera “unknown in databases”, I think you meant to say they are novel?

Authors’ response: We replaced “unknown” with “novel”

I found the text in Figure 1 to be too small to read.

Authors’ response: Unfortunately, we cannot change the size of the characters in the tree labels without having an enormous tree (due to the number of characters in some labels), making impossible to better define it. However, when enlarged on the screen, it is still possible to read the labels without problem due to pixelisation.

Quality of written English: Needs some language corrections before being published

Authors’ response: The manuscript has been revised for corrections.

The authors would like to thank Reviewer 3 for his comments, which enabled us to precisely explain how we worked. This discussion is even more interesting, since the research community will be able to easily see the gap, but also the complementarity, between molecular and culture approaches and the point of views of the respective researchers.

Reviewers’ response:

This study is of interest in its uniqueness of design. Honestly, upon first reading it was not clear to me why Delavat et al. used media with pH ranging from 3.5 to 9.8 when the creek ranges in pH from 2.7-3.4. They state their motivation for this was to obtain a larger bacterial diversity (and novelty) from this community containing low species richness, as well as determine possible ecological/metabolic roles of low abundance community members. Assuming these Bacteria are not too transient (just passing through), it is clear that the Bacteria they obtained are not dominant members of the community, based on prior studies. Thus, they suggest that these species may be components of the rare biosphere in the creek.

This topic is timely, as the role of the rare biosphere is of considerable interest. Are rare members primarily transient and not particularly active? To what extent do they play key ecological roles, such as keystone species? In this study, they do address possible metabolic roles of these isolates by testing for cellulose degradation and arsenic oxidation for example. However, these experiments were run in media with pH that is not representative of the environment they were recovered from. Thus, it is not likely that these organisms would be too happy do this at lower pH. Therefore, it is possible that these species are part of the “seed bank” and during seasonal disturbances they may be important in maintaining stability of the community. I know for a fact that Actinobacteria are common to other AMD sites. It would be interesting to know if these species are common to rare members of acid mine drainage, that would suggest they are stable members of the communities seed bank and are important in some capacity. They also point out that this study is unique in that they were able to obtain greater diversity by culturing than by culture-independent methods, a 70% increase. This was not that convincing to me, as previous methods (eg. clone libraries) were limited by sensitivity. I bet if they were to high-throughput (tag-sequence) 16S rRNA genes from the same sample they were uncover far more.

Quality of written English: Acceptable

Reviewers’ response: We would like to thank reviewer 3 for his comments, and apologize for the English used in our previous answers. He is fully right when he says “Thus, it is not likely that these organisms would be too happy do this at lower pH”. In fact, the strain Q8 belonging to the genus Paenibacillus (isolated in this study) was studied for its capability to degrade polymers at lower pH. We demonstrated that this strain could degrade starch and xylan under a wide pH range. However, its polymer degrading activity significantly decreased at a pH under 4 [16].

Some of our isolates were detected for the first time in AMDs, whereas others are frequently detected in such environments, suggesting that “they are stable members of the communities seed bank and are important in some capacity” as reviewer 3 suggested. X11 is the most remarkable example, since members of this previously uncultured genus were frequently detected in AMDs and other acidic environments (“Rio Tinto” in Spain, “Lower Red Eyes” in Pennsylvania, “Wheal Jane” in England (NCBI-nr BLAST, unpublished results)). This suggests a possible role of X11 and relatives in the functioning of these ecosystems.

Again, we agree with reviewer 3 when he says: “I bet if they were to high-throughput (tag-sequence) 16S rRNA genes from the same sample they were uncover far more”. We indeed think that high-throughput methods could/will uncover some of the genera detected by culture, and this would be an interesting comparison to be done.