Application of urea–agarose gel electrophoresis to select non-redundant 16S rRNAs for taxonomic studies: palladium(II) removal bacteria

The 16S ribosomal RNA (rRNA) gene has been the most commonly used sequence to characterize bacterial communities. The classical approach to obtain gene sequences to study bacterial diversity implies cloning amplicons, selecting clones, and Sanger sequencing cloned fragments. A more recent approach is direct sequencing of millions of genes using massive parallel technologies, allowing a large-scale biodiversity analysis of many samples simultaneously. However, currently, this technique is still expensive when applied to few samples; therefore, the classical approach is still used. Recently, we found a community able to remove 50 mg/L Pd(II). In this work, aiming to identify the bacteria potentially involved in Pd(II) removal, the separation of urea/heat-denatured DNA fragments by urea–agarose gel electrophoresis was applied for the first time to select 16S rRNA-cloned amplicons for taxonomic studies. The major raise in the percentage of bacteria belonging to genus Clostridium sensu stricto from undetected to 21 and 41 %, respectively, for cultures without, with 5 and 50 mg/L Pd(II) accompanying Pd(II) removal point to this taxa as a potential key agent for the bio-recovery of this metal. Despite sulfate-reducing bacteria were not detected, the hypothesis of Pd(II) removal by activity of these bacteria cannot be ruled out because a slight decrease of sulfate concentration of the medium was verified and the formation of PbS precipitates seems to occur. This work also contributes with knowledge about suitable partial 16S rRNA gene regions for taxonomic studies and shows that unidirectional sequencing is enough when Sanger sequencing cloned 16S rRNA genes for taxonomic studies to genus level.


Dr. Ana Filipa Benedito Assunção
Abstract: The 16S rRNA gene has been the most commonly used sequence to characterize bacterial communities. The classical approach to obtain gene sequences to study bacterial diversity implies: cloning amplicons, selecting clones and Sanger sequencing cloned fragments. A more recent approach is direct sequencing of millions of genes using massive parallel technologies, allowing large-scale biodiversity analysis of many samples simultaneously. However, currently this technique is still expensive when applied to few samples; therefore the classical approach is still used. Recently we found a community able to remove 50mg/L Pd(II). In this work, aiming to identify the bacteria potentially involved in Pd(II) removal, the separation of urea/heat-denatured DNA fragments by urea-agarose gel electrophoresis was applied for the first time to select 16S rRNA cloned amplicons for taxonomic studies. The major raise in the percentage of bacteria belonging to genus Clostridium sensu stricto from undetected to 21% and 41%, respectively for cultures without, with 5mg/L and 50mg/L Pd(II) accompanying Pd(II) removal point to this taxa as a potential key agent for the biorecovery of this metal. Despite sulphate-reducing bacteria were not detected, the hypothesis of Pd(II) removal by activity of these bacteria cannot be ruled out because a slight decrease of sulphate concentration of the medium was verified and the formation of PbS precipitates seems to occur. This work also contributes with knowledge about suitable partial 16S rRNA gene regions for taxonomic studies and shows that unidirectional sequencing is enough when Sanger sequencing cloned 16S rRNA genes for taxonomic studies to genus level.  (Woese andFox, 1977, Fox et al, 1977). Subsequently, rRNA genes have been the most commonly used sequences in phylogenetic, taxonomic and population studies. Thus, the massive work carried out in the last thirty years on DNA sequencing has led to the accumulation of information on these sequences for a large number of organisms.
Although different bioinformatics tools have been developed to analyze the sequences, the principle of the process is unique and can be summarized as follows: highly conserved regions supporting the constancy of the rRNA genes complex secondary structure and function are used to ensure positional homology in sequence alignments, which in their turn are used, taking advantage of the interspersed hypervariable regions, for the attribution of taxonomic classifications and for the construction of phylogenetic trees supporting evolutionary hypothesis.
The use of primers for the conserved domains flanking the hypervariable regions enables robust specific PCR amplifications of target sequences. Thus, the PCR became the preferred approach to obtain rRNA gene sequences to analyze natural or cultured populations and, relying on the objectives of the studies, the different strategies are mainly distinguished by the target genes and by the primers used to amplify them .   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 In general, for bacteria and archaea the 16S rRNA gene encoding the 16S rRNA small subunit has been the most important target sequence for these types of studies (Yarza et al, 2008;. In this case, PCR universal primers for the 16S rRNA gene are generally used when the aim is to characterize all population (e.g. Weisburg et al, 1991;Baker et al., 2003). When the objective is to characterize just a taxonomic group of organisms, specific primers for that group have to be designed. For example, specific primers for Sulphate Reducing Bacteria (SRB) 16S rRNA genes have already been designed and used for phylogenetic, taxonomic and population studies (e.g. Devereux et al, 1989;Castro et al, 2000;Daly et al, 2000;Karr et al, 2005). Another possibility is the use of genes that are only present in the group of organisms to be studied. For example, in the case of SRB, the dsr gene, encoding the enzyme dissimilatory sulphite reductase (DSR) responsible for the central energy conserving step of sulphate respiration (Odom et al, 1984), has proven to be a good alternative (e.g. Wagner et al, 1998;Karr et al, 2005).
The most commonly used strategy in the past (which therefore can be considered a classical approach) to obtain a number of DNA sequences to study bacterial diversity implies the following several steps: (1) PCR amplification of target genes (usually the 16S rRNA gene) or parts of them in a sample; (2) cloning the amplicons by insertion in a vector and transformation into Escherichia coli (E. coli); (3) selecting a number of transformed colonies; (4) multiplying the number of copies of each cloned amplicon by growing E. coli pure cultures and purifying the plasmids or by direct PCR and purifying the amplicons and (5) finally sequencing the cloned fragments through the Sanger method. The negative aspect of this strategy is the time and the cost associated to it when the objective is high-depth sampling to detect rare taxa in complex natural or cultured populations.
Thus, the classical approach is still applied in studies aiming to characterize the main taxa present in only one or in few bacterial communities and when the identification of rare taxa is not important. In these cases usually a DNA fingerprinting analysis of 16S rRNA gene cloned amplicons is carried out to identify different clones, avoiding sequencing similar ones and therefore reducing the costs in DNA sequencing. The fingerprinting methods most applied for the selection of non-redundant cloned amplicons to characterize bacterial communities are: Denaturing-or Temperature-Gradient Gel Electrophoresis (DGGE or TGGE) (Fischer and Lerman, 1979;Rosenbaum and Riesner, 1987), Single-Stranded Conformation Polymorphism (SSCP) (Orita et al, 1989) and Amplified Ribosomal DNA Restriction Analysis (ARDRA) (Dijkshoorn et al, 1998). For example, Karr and colleagues (2005) to explore the biodiversity of SRB in Lake Fryxell located in Antarctica have amplified by PCR the dsr genes on water and sediments from this extreme environment and have separated the polymorphic products to be cloned and sequenced by DGGE.
Nevertheless, all these fingerprinting methods referred above have some drawbacks. For the DGGE analysis: the gels are made of acrylamide, which has carcinogenic effects; making gels with denaturing gradients is difficult and requires wide experience in the preparation of acrylamide gels; the amount of eluted amplicom in each gel excised band is low and thus usually a PCR reamplification is necessary before cloning the fragments to be sequenced, which increases the number of amplification errors. For the TGGE, the drawbacks are similar to those in DGGE and though the preparation of gels is relatively simpler, a special electrophoresis system allowing temperature gradients is necessary. The SSCP method also implies simple acrylamide gels, however the sensitivity for polymorphisms detection is applicable only in products with sizes up to about 300 bp. Regarding the selection of clones by ARDRA, the principal disadvantage is that the detection of polymorphisms does not cover the entire length of the nucleotide sequences, but is limited to the regions recognized by the restriction enzymes used .   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 Palladium, a Platinum Group Metal (PGM), is a metal with high economic value due to the limited global resources and high demand, mainly due to its use in catalytic processes (Deplanche et al., 2014 (Yong et al., 2002;Baxter-Plant et al., 2003) and Vargas and colleagues (2004) Windt et al., 2005). Cells of Bacillus sphaericus also proved to accumulate high amounts of toxic metals, including Pd (Pollmann et al., 2006).
More recently, De Corte and colleagues (2012) discussed the different bio-Pd precipitating microorganisms in which they included all the bacteria mentioned above as well as Citrobacter braakii (Hennebel et al., 2011) and Clostridium pasteurianum (Chidambaram et al., 2010).
Our research group reported for the first time a Pd(II)-resistant mixed bacterial culture enriched from a sludge sample from a municipal wastewater treatment plant (WWTP) able to remove 18mg/L of Pd(II) from an aqueous medium and the phylogenetic analyses showed that this culture was mainly composed by Clostridium species (Martins et al., 2013). Recently we found a new bacteria community, also enriched from a WWTP sludge sample, resistant and able to remove even higher concentrations of Pd(II): up to 50 mg/L.
In this work, aiming to identify bacteria potentially involved in Pd(II) removal and to understand the evolution of the bacterial consortium when the 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 Pd(II) concentration is increased, the new consortia was grown in the absence of Pd(II) and in the presence of 5mg/L and 50mg/L of Pd(II). Then, having only three samples, and because the aim was to identify the main groups of bacteria that constitute the bulk of the communities and not rare taxa, we characterized the bacteria consortia based on 16S rRNA gene sequences selected following the above considered classical approach. To that purpose, a new simple and sensitive method to detect polymorphisms based on urea-agarose gel electrophoresis (Hegedüs et al, 2009) was tested and applied to select representative cloned amplicons for Sanger sequencing in order to taxonomically classify their origin and thus characterize the major bacterial groups in the consortia. Moreover, we compared the taxonomic classifications retrieved with sequences from one half side of the 16S rRNA gene with classifications obtained with sequences from the other half side of the gene obtained by sequencing cloned amplicons with two universal plasmid primers flanking the cloning site, to study the feasibility of using only one of the primers to reduce sequencing costs.

Biological reactors
All assays and the initial bacterial community enrichment were performed in batch reactors using glass bottles (120 mL) containing 100 mL of nutrient medium with pH adjusted to 7.5 ± 0.2 under anaerobic conditions at room temperature (21 ± 1 o C). In order to achieve the anaerobic conditions before inoculation the medium was purged with nitrogen gas and after inoculation about 10mL of liquid paraffin was added. The bottles were sealed with butyl rubber stoppers and aluminium seals and incubated at room temperature.
Growth media and material used in batch experiments were sterilized by autoclaving.

Source and enrichment of the bacterial community
The bacterial consortium used in the present study was enriched from a sludge sample from a wastewater treatment plant, located in Lagos, in southern Portugal.
The medium used for this enrichment was Postgate E (Postgate, 1984) without agar and supplemented with resazurin as redox indicator (0.03 g/L).
Postgate E is a medium developed for SRB, a group of microorganisms known to be able to reduce sulphate to sulfide, thus removing metals from aqueous media as insoluble metal sulphides.

Experimental frame
The medium used in Pd(II) removal experiments was based on Postgate C (Postgate, 1984), which is also a growth medium for SRB, but modified in order to maintain Pd(II) soluble. First, a culture was grown in this medium but without Pd(II) to acclimatize the bacteria. This culture was prepared using 10% (v/v) inoculum harvested from the enrichment by centrifugation at 2500 xg for 10 min and washed with modified Postgate C medium without Palladium.
The modified Postgate C medium contained 0.5 g/L KH 2 PO 4 , 1 g/L NH 4 Cl, 4.5 g/L Na 2 SO 4 , 0.06 g/L CaCl 2 .6H 2 O, 1 g/L yeast extract, 0.0072g/L  In order to evaluate the relation between Pd(II) removal and the bacterial growth, abiotic assays were made in exactly the same conditions as the cultures with Pd(II), but without bacterial inoculum.
The biotic assays (with bacterial inocula) in growth media with Pd(II) were performed with four replicates. The biotic assay without Pd(II) and all abiotic assays were carried out with two replicates.

Analytical methods
The Optical Density (OD 600 ) was determined weekly in order to monitor the bacterial growth. In addition, aiming to monitor an eventual growth of SRB, the oxidation-reduction potential (E h ) and the sulphate concentration were also measured weekly. The pH was monitored due to its importance as a limiting factor and palladium concentration was determined to evaluate its removal from the growth media.
The samples from batch cultures were collected using a sterile syringe and OD 600 was immediately measured in each sample. Then, the samples were centrifuged at 2500 xg for 5 min and the supernatant was used for the remaining analysis. Redox potential (E h ) and pH were determined using a pH/E Meter (GLP 21, Crison). Sulphate concentration was quantified by UV-visible spectrophotometry (Hach-Lange DR2800 spectrometer) using the method of SulfaVer®4 from Hach-Lange. Palladium concentration in the media was determined by flame atomic absorption spectroscopy (Flame-AAS) using an Analyticjena novAA 350 model spectrometer.
The precipitates were obtained collecting the samples by centrifugation at 2500 xg for 10 min, the pellet was washed with ethanol 70% (stirred and centrifuged 2500 xg for 20min, 3 times) and then dried in vacuum (Binder, VDL) at 37ºC ± 1ºC. In order to confirm the particles size, morphology and position in relation to cells, a Transmission Electron Microscopy (TEM) analysis was made using a Hitachi, H8100 model, with a LaB6 filament. This analysis was coupled to an Energy Dispersive X-ray Spectrometer (EDS) for light elements,  The PCR products were analyzed by electrophoresis in 1% (w/v) agarose gels in 1x TAE buffer (AMRESCO Thirty two transformed (white) colonies were randomly selected from cultures without Pd, with 5 mg/L Pd(II) and with 50 mg/L Pd(II) for subsequent taxonomic classifications and consortia characterization. In order to rapidly multiply and isolate the cloned products a PCR with vector specific primers SP6 and T7 was carried out directly from bacteria by touching the colony with a pipette tip and submerging it in the reaction mixture. The PCR was carried out in a thermocycler with the following conditions: denaturation of 94 o C for 3 min, followed by 30 cycles of 94 o C for 1 min, 55 o C for 1 min and 72 o C for 2 min and a final step of 5 min at 72 o C.

Screening of cloned amplicons by urea-agarose gel electrophoresis
Urea-agarose gel electrophoresis The procedure to prepare and run the samples in urea-agarose gel electrophoresis was adapted from Hegedüs and colleagues (2009)  patterns for their: (1) identity similarities calculated with aligned sequences trimmed for quality and cropped to the same size, (2) position on a phylogenetic tree and (3) taxonomic classifications.

Amplicons purification and Sequencing
PCR products were precipitated with absolute ethanol, washed with 70% (v/v) ethanol and resuspended in Mili-Q water. The DNA was quantified using a NanoDrop 1000 Spectrophotometer (Thermo Scientific) and Sanger sequenced with the primers SP6 and/or T7 using a capillary electrophoresis sequencing system (Genetic Analyzer, Applied Biosystems, Model 3130xl). Based on the analysis of the chromatograms the obtained sequences were cropped to eliminate the beginning and the ending regions with doubtful profiles using the program BioEdit Sequence Alignment Editor (Hall, 1999). Gene regions Ten cloned 16S rRNA gene amplicons from the consortium grown with 5 mg/L Pd(II) were sequenced with SP6 and T7 vector universal primers flanking the cloning site and taxonomic classifications obtained independently with both gene parts were compared.

Phylogenetic tree construction
The phylogenetic tree was constructed with the 16S rRNA partial gene sequences obtained for the consortium grown with 5 mg/L Pd(II) (several exhibiting equal gel migration patterns) and a set of 57 16S rRNA reference gene sequences.
The set of 57 reference sequences was previously chosen from a database of 7081 complete 16S rRNA sequences (DataSet S2) identified by Větrovský and Baldrian (2013) in publicly available complete bacterial genomes.
The selection was carried out by local BLAST search with BlastStaion (version 2.0) to characterize a sludge bacterial community enriched from a wastewater treatment plant (WWTP) located in Algarve, Portugal (unpublished data), such as the initial consortium used in the present work.
After trimming low-quality ends, the partial sequences of cloned genes were oriented towards the 16S rRNA Open Reading Frame (ORF) and aligned with the 57 reference sequences using the CLUSTALW Multiple Sequence Alignment tool available online at http://www.genome.jp/tools/clustalw/. The accuracy of these alignments was confirmed by careful observation using the program BioEdit (Hall, 1999).

Bacterial growth
No significant pH variation was observed in both abiotic and biotic assays with values very close to neutral (7.0 to 7.3) and, as shown in the graphs 1, 2 and 3 (Online Resource 1), all the bacteria communities grown in the absence of palladium and in the presence of 5 and 50 mg/L of Pd(II) showed fast growth, reaching OD 600 values above 0.6 after 6 days, and stayed active during the experiment, with even higher OD 600 values.
Another evidence of bacterial growth and activity was that despite during the experiment the redox potential (E h ) decreased in all assays, the magnitude of this decays was clearly different when comparing biotic and abiotic assays. In the biotic assays the E h values decreased drastically from +99, +101 and +199 mV in the beginning of experiments to -189±10.6, -208±4.0 and -317±6.7 mV at the end, respectively for cultures without Pd(II), with 5mg/L Pd(II) and with 50mg/L Pd(II). In the abiotic controls prepared with media containing 5mg/L Pd(II) and 50mg/L Pd(II) the E h varied, respectively from the initial +101 and This reducing agent is used because the chosen culture medium is appropriate for SRB and one of the major prerequisites for cultivating these bacteria is that the E h must be negative. Thus, the much more pronounced declines of E h in the biotic tests indicate biological production of one or more reducing agents. That could be an indication H 2 S production by SRB activity, however despite the use of a growth medium which composition was based on a medium for SRB, the sulphate was not substantially consumed (Online Resource 1 -graphs 1, 2 and 3), suggesting that SRB were not present or, more probably, were in a minority.
The highest E h decrease obtained for the consortium grown with 50mg/L of Pd(II) can probably be due to a higher Pd(II) removal.

Palladium(II) removal
Regarding Pd(II) removal, the community grown with 5mg/L of Pd(II) showed ability to remove 91% of this metal after 21 days of incubation (Online The decrease of Pd(II) concentration in the biotic assays was accompanied by formation of dark-colored precipitates while in the abiotic controls precipitate's formation was not detected.

Precipitates analysis
TEM analysis allowed concluding that the particles composing the precipitates are nanoparticles with sizes between 12 and 32 nm distributed in agglomerates along the bacterial cells and also individualized and presenting a spherical morphology (Online Resource 3 -pictures a b and c). The coupled EDS analysis detected the Pd and S elements in the particles, which is a strong indication that the particles are effectively palladium sulfide (PdS) (Online Resource 3 -picture d).
Peaks corresponding to the carbon and copper elements were also detected in the EDS spectrum (Online Resource 3 -picture d). However, these elements are components of the supporting grid and the respective peaks are detected in background areas.

PCR amplification and cloning 16S rRNA gene amplicons
The agarose gel electrophoresis of PCR products amplified using the

Screening of cloned amplicons by urea-agarose gel electrophoresis
Aiming to avoid sequencing all cloned products selected for each consortium, the simple and sensitive DNA fingerprinting analysis method based on urea-agarose gel electrophoresis described by Hegedüs and colleagues (2009) was tested and applied to identify similar cloned amplicons. Ureia-agarose gel electrophoresis The analysis of the cloned 16S rRNA amplicons by urea-agarose gel electrophoresis DNA fingerprinting allowed to group clones according to their migration patterns. After heat denaturation in the presence of 8M urea, the two strands of the cloned 16S rRNA gene fragments migrated differently in the 1M urea containing agarose gels in the size range of 0.8 to 1,2 Kb (Fig. 1) The high identity similarities calculated between sequences of amplicons with the same gel pattern type was the first sign to confirm this efficiency.
Fourteen sequences from eight cloned amplicons with gel pattern type 5Pd-gpt-2 revealed very high identity similarities (between 99,1 and 100%). Four sequences from three cloned amplicons with gel pattern type 5Pd-gpt-3 also showed very high similarities (98,9 to 100%). Six sequences from four cloned amplicons with gel pattern type 5Pd-gpt-1 still showed high similarities but at a lower level (

16S rRNA based taxonomy
There are two major aspects of bacterial community studies using 16S rRNA gene sequences determining the extent to which these studies are effective: the gene regions that are used and the method used for taxonomic classification.

Gene regions
The sequences obtained with SP6 and T7 primers (from both sides of the vector cloning site) covered different hypervariable regions of the 16S rRNA gene according to its orientations in the cloning vector, with sizes between 477 and 934 bp.
Looking to taxonomic results achieved with the RDP classifier for these 20 sequences it can be seen that sequences either covering the first half part of the gene or covering the second part of the gene allowed identical taxonomic classifications to the last level (genus), except in two cases (clones 5Pd-c-2 and 5Pd-c-12) for which chimerical sequences were detected (Online Resource 4 -Spreadsheet 1). Moreover, the same classification results were obtained when using these same 20 sequences trimmed to equal sizes: on one hand to cover only hypervariable regions V1 to V3 completely and 10% of V4 and on the other hand to cover 10% of hypervariable region V6 and the full length of V7 to V9 (data not shown). Seen this, the 16 rRNA amplicons cloned to characterize the other 2 bacterial consortia (grown without Pd(II) and with 50 mg/L Pd(II)) were sequenced just with the T7 universal primer.
Among all 42 sequences used in this work to characterize the bacterial communities (Online Resource 4 -Spreadsheet 1, 2 and 3) only one didn't cover completely either one or the other of these regions. The exception is the sequence of clone 5Pd-c-6, for which a part spanning region V1 and about 9% of region V2 was cropped in the quality inspection step. Nevertheless, this sequence corresponds to an amplicon from one of the groups of sequenced amplicons with identical gel migration patterns and its classification was like the others from that group (order Clostridiales, family Clostridiaceae), excluding those deciphered to be chimeric.

Communities' characterization
The first thing that stands out when analyzing the composition of bacterial communities on cultures without Pd(II), with 5 mg/L Pd(II) and with 50 mg/L Pd(II) ( and they were not detected in the consortium grown with 50 mg/L Pd(II).
Moreover, interestingly, the genus Arcobacter from family Campylobacteraceae is a group that stands for being present in a large proportion (35.71%) in the culture with 5 mg/L Pd (II) and not being detected in cultures without Pd (II) and with 50 mg/L Pd (II).
Thus, it may be considered that the urea-agarose gel electrophoresis DNA fingerprinting proved to be an efficient method to choose similar 16S rRNA amplicons and avoid redundancy in taxonomic studies.

16S rRNA based taxonomy -Gene regions
Though sequence analysis of the 16S rRNA gene has been widely used to perform taxonomic studies, its hypervariable regions exhibit different degrees of sequence diversity and no single hypervariable region is able to distinguish among all bacteria (Chakravorty et al, 2007). Sequencing the entire 1500bp 16S rRNA gene is necessary when describing a new specie or to distinguish between certain particular taxa or strains. However, generating smaller sequences of about 500bp length is much less expensive and, for example, for most clinical bacterial isolates the initial 500bp sequence provides adequate differentiation for identification (Clarridge, 2004 along with tools to allow researchers to analyze their own rRNA gene sequences (Cole et al, 2014). It includes the tool that was used for taxonomic classifications in this work, the RDP Classifier, which rapidly assigns sequences into taxa with a bootstrap value as an estimate of confidence for each assignment (Wang et al, 2007). The overall accuracy of the RDP Classifier was estimated by Wang et al (2007) to be above around 95% down to the Family level and above around 90% to the genus level, either for 400bp randomly chosen segments or for full length genes (Table 3). In this work, the consistent taxonomic classifications obtained with the RDP Classifier for partial sequences from the beginning and from the end of the 16S rRNA gene, with sizes around 600 and 490 bp, respectively, contribute to confirm this robustness.

Communities characterization and palladium(II) bio-removal
Few studies have been focused on PGM recovery using mixed cultures and to our knowledge just our group, in this work and in another published in 2013 by Martins and colleagues, studied the use of resistant-bacterial communities instead of pure cultures for remediation of palladium (Table 4).
The applicability of pure cultures is limited since in an industrial process it is not easy to maintain the sterile conditions necessary to prevent external microbial contamination. The bacterial performance maintenance for a long time is also a problem due to the susceptibility of pure cultures even to small variations in the conditions. Therefore, the use of mixed cultures in biological metals removal or recovery systems is more realistic for future applications.
SRB are known to be able to remove metals from wastewaters and have been reported to have potential for Pd(II) removal (Yong et al., 2002;Baxter-Plant et al., 2003;Vargas et al., 2004). Thus, the initial community was enriched in a medium that favours the SRB growth (Postgate E). After that the Pd(II) bioremoval essays were performed in a growth medium which components were based on the composition of Postgate C, a medium also specific for SRB, but with modifications to guarantee Pd(II) solubility. However, perhaps due to these modifications in the growth medium and/or due to the possibility of the sludge used in the enrichment not being a good source of SRB, the bacterial communities developed during the assays did not show significant sulphatereducing activity. The reasons for this will not be discussed here as the focus of this paper is not the study of SRB but is the characterization of bacterial communities able to remove Pd(II) from aqueous media.
In the present work a bacterial consortium resistant and with ability to remove 98% Pd(II) from media with 50mg/L of this metal was identified, while only 60% removal from a culture with 18mg/L of Pd(II) was achieved in the first (and so far only) work reporting a mixed bacterial community resistant and able to remove this metal. Looking to works with pure cultures, only two cases of experiments with Pd(II) concentrations higher than 50 mg/L are found: 88% removal from an initial concentration of 213mg/L was achieved with 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 Desulfovibrio desulfuricans (Yong et al, 2002;Baxter-Plant et al, 2003) and 99% removal was attained with Clostridium pasteurianum BC1 from an initial concentration of 100 mg/L (Chidambaram et al, 2010). However, in the former case the work was made with resting cells biomass and in the second case the palladium reduction was achieved just in one minute after mixing the metal solution with the bacterial culture, while in our work the Pd(II) removal occurred during the growth of the bacterial community.
The community grown with 50 mg/L Pd(II) led to 91% removal of this The most representative bacteria (40,91%) in the consortium grown with 50 mg/L Pd(II) and whose representation in the community has increased dramatically with increasing Pd(II) into the culture medium belongs to the Clostridium genus (Table 2), corroborating the idea of its resistance to Pd(II) and emphasizing its potential for palladium bio-recovery, as already reported by other authors. Pure cultures of a specie from this genus (Clostridium pasteurianum BC1) have already been used to reduce Pd(II) ions, being the palladium precipitated on the cell wall and in the cytoplasm (Chidambaram, et al., 2010). Moreover, colleagues of our research group described a mixed bacterial community mostly composed by Clostridium species as resistant to and able to remove Pd(II) (Martins et al, 2013) and other colleagues of our group have recently reported mixed communities able to remove copper, zinc, and iron, in which a Gram-positive population mostly assigned to Clostridium   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 spp. initiated metal bio-removal based on sulfide generation from components of the medium (mainly sulphite) but not from sulphate (Alexandrino et al, 2014).
In our work, the strong indication of PdS precipitates formation despite just a slight decrease of sulphate and the growing of a bacterial community mostly composed by bacteria belonging to Clostridium genus during the Pd (II) removal leaves open the possibility of metal bio-removal by Clostridium spp.
based on sulfide generation from other then sulphate components.
Another explanation can be the production of sulfide from sulphate by SRB that were in such small percentages in the bacterial communities that were Together, the much more pronounced decrease of the E h in biotic tests than in abiotic tests, which suggests that there was biological production of at least one reducing agent, and the strong indication that the precipitates formed in the biotic assays are PdS particles, suggest that the reducing agent H 2 S was produced by growing bacteria. This corroborates both pointed hypotheses of being Clostridium spp. or SRB the responsible agents for the biological removal of Pd (II) from the culture medium. As for the other bacterial genera detected in the consortium grown with 50 mg/L Pd(II), they are known to be adapted to extreme environments, but it is the first time they are referred as palladium resistant. Clostridium XlVa has been reported among the most abundant genus occurring in studies regarding remediation of Acid Mine Drainage (AMD) (Sánchez-Andrea, 2014), which is characterized by being contaminated with high concentrations of metals.
Concerning Hydrogenoanaerobacterium as metals resistant bacteria, no references were found in the literature. Nevertheless, this taxonomic group has been identified as butanol-and isobutanol-tolerant bacteria (Kanno et al, 2013).
Solitalea bacteria present several properties, including anaerobic growth and nitrate reduction. These bacteria have been detected in extreme environments such as in a denitrification reactor and in saline-alkaline lakes (Zhu et al, 2015;Silva 2015). Moreover, solitalea-like bacteria have been described as not being influenced by antibiotic (neomycin and streptomycin) treatments (Kopecky et al, 2014), which is another evidence of their resistance to extreme conditions. The Alkaliphilus species usually survive in certain extreme environments. For example, the Alkaliphilus transvaalensis are strictly anaerobic and extremely alkaliphilic (Kobayashi et al., 2007) and some Alkaliphilus species are alkaliphilic metal-reducing bacteria [Fe(III), Cr(VI), Co(III), U(VI) and Se(VI)], as Alkaliphilus metalliredigens (QYMF), and have been used in metal reduction and biomineralization processes (Roh et al., 2007). Thus, these authors' results together with our results suggest that all these groups of bacteria are resistant to various types of extreme environments, which make them interesting as targets for the study of resistance mechanisms and eventually for biotechnological applications, such as metals bioremediation and biorecovery.
In the bacterial community grown with 5 mg/L of Pd (II), beyond bacteria from the genus Clostridium, the Parabacteroides genus was the predominant, followed by the Arcobacter genus. These two bacteria genus were not identified before as resistant to Pd(II), thus they are also candidates to be further studied as putative efficient Pd(II) removal agents.
Moreover, the detection of a high percentage (35.71%) of bacteria from genus Arcobacter only in the consortium grown with 5mg/L of Pd(II) is very interesting because this suggests that for this taxa the presence of some

Conclusions
In this work we verified that the simple fingerprinting method of running urea/heat-denatured cloned 16S rRNA amplicons on urea-agarose gel electrophoresis allows identifying clones with sequences whose similarities lead to the same taxonomic classifications, at least to the genus level, and therefore proved that it can be used for that purpose instead of the most applied methods up to now: DGGE, TGGE, SSCP and ARDRA, which have some drawbacks.