A stable and sediment-free, benzene mineralizing, sulfate-reducing culture that resisted repeated attempts at isolation was examined using molecular approaches such as traditional cloning and sequencing and a direct PCR fingerprinting method for 16S rRNA genes. Despite the culture's long exposure to benzene as the only carbon and energy source (over 3 years) and repeated dilutions of the original enrichment, this consortium has remained relatively complex. Cloning and sequence analysis identified 12 unique small subunit rRNA genes. The 16S rRNA genes belong to different eubacterial phyla, including Proteobacteria, Cytophagales and Gram-positives. There is one deeply branching clone which is not closely related to any known, sequenced, bacterium. A different clone, however, is closely related to a known sulfidogenic, aromatic hydrocarbon degrader. To assess 16S rRNA gene cloning efficiency, a fingerprinting method based on fluorescent, end-labeling of PCR product (16S rRNA genes) and screening by restriction length polymorphism analysis (RFLP) was employed. The data obtained indicated that we had cloned and characterized nearly all of the eubacterial 16S rRNA genes amplified from the consortia.
Complete mineralization of benzene under sulfidogenic, methanogenic and iron-reducing conditions has recently been clearly demonstrated [1–5]. The isolation of pure cultures which carry out this transformation has been, however, thus far unsuccessful. Pure cultures, of course, are important for gaining a detailed understanding of the anaerobic physiology, genetics and biochemistry of benzene degradation. While the pathway for anoxic toluene mineralization has been studied by a number of groups (as reviewed in ), and the mechanisms involved in ring cleavage of benzoic acid (a central intermediate in the metabolism of many aromatic compounds) have been demonstrated , the route by which benzene itself is degraded is still unclear. Vogel and Grbic-Galic  reported that a methanogenic culture incorporated oxygen from water as a hydroxyl group into benzene, forming phenol. The amounts of phenol detected, however, were very small relative to the concentration of benzene added and it is not certain that phenol represented a true intermediate in the pathway. More convincing evidence for the production of phenol during benzene mineralization in methanogenic consortia has been reported recently . In sulfate-reducing cultures actively degrading benzene however, Lovley et al.  were unable to detect any extracellular intermediates. Investigations of degradation pathways in mixed cultures such as these are always confounded by the difficulty in knowing which organisms are carrying out the reaction being studied.
Isolation of pure cultures from anaerobic consortia can often be difficult. For example, methanogenic cultures degrading substrates other than simple fatty acids generally function as syntropic associations with the fermentative organism responsible for the initial attack on the substrate and the methanogen metabolizing the resulting volatile fatty acids or hydrogen . Hence, isolation of the organism responsible for the initial metabolism may be problematic. Iron-reducing bacteria are also difficult to isolate since they often grow closely bound to the insoluble iron particles that they use as electron acceptors. This makes them difficult to isolate by standard techniques such as plating or serial dilution which ideally require diluting a homogeneous culture to obtain pure cultures. It can also be difficult to provide sufficient quantities of iron in a soluble (non-particulate) form without adding high concentrations of chelators which may be toxic or act as a competing carbon source. Sulfate-reducing bacteria (SRB) are generally isolated by serial dilution (either in liquid media or soft agar) in sealed anaerobic tubes since they require very low reduction potentials . As many SRBs have very slow growth rates (especially those known to degrade aromatic compounds), this technique can result in selection for bacteria growing on impurities in the media rather than the provided substrate. Finally, benzene is toxic. Only minute amounts of benzene can be added at any one time, and sufficient microbial biomass for isolation develops very slowly.
Given all these reasons, alternative methods to examine the microorganisms would be useful and molecular tools can provide information about the organisms that we would not otherwise be able to determine. The approach of using 16S rRNA gene sequences has been applied to characterize the sulfate-reducing communities present at oil fields  and to identify bacteria present in a co-culture capable of reducing sulfate after exposure to oxygen . In the latter study, the information from analyses of 16S rRNA genes was used to devise an isolation scheme which successfully separated the two strains. Kane et al.  used a combination of 16S rRNA cloning and phylogenetic probes to follow the progress of the enrichment and isolation of a sulfate-reducing bacterium.
In the present study we examined a well established sulfate-reducing consortium which completely mineralizes benzene to carbon dioxide , but has so far resisted numerous attempts to isolate the bacterium or bacteria responsible. Both cloning and sequencing of 16S rRNA genes and direct fluorescent fingerprinting of 16S rRNA genes were used to identify the members of the bacterial consortium which carries out benzene degradation.
2Materials and methods
The enrichment of a benzene-degrading, sulfate-reducing consortium, from Guaymas Basin sediment was reported previously . All experiments reported here used stable, sediment-free cultures resulting from repeated dilution of and benzene addition to the original enrichment over the course of 3 years. The cultures had been significantly diluted over time and contain approximately 10−7 of the original sediment. The mineral salts medium used for growth was a modification of Widdel's sulfidogenic medium  with 23 g NaCl and 1.0 g MgCl2-6H2O per liter. Cultures were maintained by the addition of pure benzene (Aldrich Chemical Co., Milwaukee, WI) to a concentration of ∼100 μM with a microliter syringe (Hamilton Co., Reno, NV) whenever the previous supply had been exhausted. Loss of benzene was monitored by GC-FID as previously described for toluene . Strict anaerobic technique was followed at all times.
2.2DNA extraction and amplification
DNA was extracted from a cell pellet resulting from centrifugation of 5 ml of active culture using a procedure modified from Kerkhof and Ward . The cell pellet was first frozen at −20°C, and then resuspended in 225 μl of cold buffer (50 mM glucose/10 mM EDTA/25 mM Tris (pH 8.0)). To this suspension, 100 μl of the same solution containing 10 mg ml−1 lysozyme and 75 μl of 500 mM EDTA (pH 8.2) were added. The mixture was incubated at room temperature for 10 min before lysing the cells with 50 μl of 10% sodium dodecyl sulfate (SDS). The resulting cell lysate was extracted twice with 800 μl of Tris (pH 8.2) saturated phenol:chloroform:isoamyl alcohol (25:24:1) with 0.1% 8-hydroxyquinoline. The mixture was vortexed vigorously to form an emulsion and centrifuged at 14 000×g for 3 min before removing the non-aqueous phase. Sodium acetate (50 μl of 3.0 M) along with 2 μl of 20 mg ml−1 glycogen was added to the aqueous phase and the DNA was precipitated with 1 ml of 100% ethanol. DNA was pelleted at 14 000×g for 15 min at 4°C, dried under vacuum, then resuspended in 100 μl Tris-EDTA buffer.
Amplification of 16S rDNA used the universal eubacterial primers (27 forward (5′ cau cau cau cau AGA GTT TGA TCC TGG CTC AG 3′) and 1525 reverse (5′ cua cua cua cua AAG GAG GTG WTC CAR CC 3′))  containing a uracil rich cloning region on the 5′ end. The following amplification parameters were used: initial denaturation at 95°C for 5 min, then 94°C for 0.5 min, 55°C for 0.5 min and 72°C for 1.5 min (20–30 cycles), and a final extension at 72°C for 10 min in a Perkin-Elmer Gene Amp PCR system 2400 thermal cycler (Perkin-Elmer, Foster City, CA).
2.3Cloning, screening and sequencing
Cloning of 16S rDNA used the Cloneamp System (Gibco BRL, Gaithersberg, MD) as per the manufacturer's instruction. The resulting plasmids were inserted into E. coli (DH5α) by electroporation using a BTX Transporator Plus (BTX Inc., San Diego, CA) following the manufacturer's instructions. Cells were exposed to a 5 ms pulse of 1.3–1.5 kV in a cuvette with a 1 mm gap width. These transformed cells were then cultured on LB plates with ampicillin (100 μg ml−1).
Colony screening used plasmid extraction and digestion with the restriction enzyme HaeIII (Boehringer-Mannheim, Indianapolis, IN) to identify unique banding patterns. Extractions used a modified alkaline lysis miniprep protocol . The restriction digests were separated using Methaphor agarose (FMC Corp., Rockland, ME). Plasmids bearing a unique restriction banding pattern were re-purified using a Flexiprep kit (Pharmacia, Piscataway, NJ) and sequence was determined by automated techniques (Perkin Elmer-ABI, Foster City, CA) with the following primers: 27 forward (5′ AGA GTT TGA TCC TGG CTC AG 3′), 519 reverse (5′ GWA TTA CCG CGG CKG CTG 3′), 530 forward (5′ GTG CCA GCM GCC GC GG 3′), 907 reverse (5′ CCG TCA ATT CMT TTR AGT TT 3′), 926 forward (5′ AAA CTY AAA KGA ATT GAC GG), 1492 reverse (5′ GGT TAC CTT GTT ACG ACT T 3′) .
2.4PCR fingerprinting of 16S rRNA genes
In order to determine the number of distinct 16S rRNA genes available for cloning in DNA extracted from the consortia, we employed a direct DNA fingerprinting method similar to the approach of Avaniss-Aghajani et al. and Liu et al. [20, 21]. The method is based on amplification of 16S genes using a primer fluorescently labeled with 6-FAM (5-carboxy-fluorescein) on the 5′ end (Gibco-Life Technologies, Gaithersberg, MD). In this case, the labeled primer used was 27 forward and the PCR was performed with 1525 reverse as described above. This labeled PCR product was then digested for 6 h with the restriction endonuclease MnlI (New England Biolabs, Beverly, MA) to produce a mixture of variable length, end-labeled 16S rRNA fragments. The labeled fragments were electrophoretically separated on a polyacrylamide gel in an ABI model 373 automated sequencer and the data analyzed using Genescan software (Perkin-Elmer, Foster City, CA).
Ideally, each 16S rRNA gene will produce a unique end-labeled fragment which differs in length from all other 16S rRNA genes. In practice, some resolution is lost since no restriction enzyme can differentiate between all possible 16S rRNA genes. The number of fragments observed, however, can provide a lower estimate of the various 16S rRNA genes present in the DNA extracted from the consortia . For estimation of our cloning efficiency, the cloned 16S rRNA genes were also amplified, digested, separated and compared with the fingerprint generated directly from consortial genomic DNA.
2.5Construction of phylogenetic trees
All unique sequences were aligned using the Ribosomal Database Project (RDP)  alignment function and GDE software ver. 2.2 . Phylogenetic trees were constructed from unambiguously aligned sequences (947 bases of the bacterial alignment, 1124 bases of the delta proteobacterial alignment) using fastDNAml . Bootstrap values for the delta proteobacterial tree were determined from 100 iterations of the maximum likelihood calculation. Similarity values were calculated from the unambiguously aligned sequences by constructing an uncorrected similarity matrix using GDE. The RDP short IDs of all sequences used in building alignments are listed in Table 1. All sequences listed were used in the maximum likelihood calculations but some were pruned from the final tree for clarity. The GenBank database was also searched for closely related sequences which were not included in the RDP, such as environmental isolates and newly published clones. No sequences differing from those in the RDP were found which were closely aligned (similarity of greater than 90%) to any of our clones.
Table 1. Sequences used in phylogenetic analysis
RDP short ID
aSequence used to generate the phylogenetic tree for the domain Bacteria (1), or the sulfate-reducing bacteria (2).
bNo longer included in the RDP alignment.
Anaeroflexus maritimus str. PL12FS (DSM 2831)
Aquifex pyrophilus str. Ko15a
Actinomyces neuii subsp. neuii str. 97/90
Bacillus subtilis str. 168
Bacteroides fragilis (ATCC 25285 (T))
Clostridium botulinum str. D6f
Capnocytophaga gingivalis (ATCC 33624 (T))
Chlorobium limicola str. 8327 (ATCC null)
Chlamydia pneumoniae str. TW 183
Cytophaga diffluens str. Lewin-LIM-1 (ATCC 23140)
Deinococcus radiodurans (ATCC 35073)
Desulfobulbus propionicus str. 1 pr 3, Lindhorst
Desulfobacterium vacuolatum (DSM 3385)
Desulfobacula toluolica str. Tol2 (DSM 7467)
Desulfococcus multivorans str. 1 be 1, Goettingen
Desulfomicrobium baculatus (DSM 1743)
Desulfomonile tiedjei str. DCB-1 (ATTC 49306 (T))
Desulfobacter curvatus str. AcRM3 (DSM 3379 (T))
Desulfobacter hydrogenophilus str. AcRS1 (DSM 3380 (T))
Desulfobacter postgatei str. 2 ac 9 (DSM 2034)
Desulfuromonas acetoxidans str. 11070 (DSM 684)
Desulfosarcina variabilis str. 3 be 13, Montpellier
The enriched consortium has been previously shown to degrade high concentrations (>100 μM) of benzene rapidly (within 2 weeks). Mineralization was confirmed by the release of 14CO2 from radiolabeled benzene. Also, the role of sulfate as the terminal electron acceptor was demonstrated by the stoichiometric balance between benzene loss and sulfate reduction . The degradation of ∼150 μM benzene by the sediment-free culture, and the effect that the addition of sodium molybdate has on degradation are shown in Fig. 1. Benzene loss in control cultures was complete after 15 days, whereas in molybdate-amended cultures, loss was reduced to the same level as in sterile controls after addition of 5 mM sodium molybdate on day 7. Molybdate is a specific inhibitor of bacteria actively reducing sulfate . The observation that its addition interrupts benzene degradation in these cultures further underscores the importance of SRBs in benzene mineralization.
3.2Cloning and sequencing of 16S rRNA genes
Fifty clones from a 16S rRNA gene library were screened and found to contain 12 unique banding patterns by restriction enzyme analysis. Nearly full sequence (positions 27 to 1525 E. coli numbering) was obtained for all of the 12 clones. Partial sequence analysis of clones with similar restriction banding patterns yielded identical sequence and further characterization of the redundant clones was not attempted (data not shown). Sequences of all 12 clones have been deposited in GenBank under the accession numbers AF029039–AF029050.
3.3Diversity of the benzene-degrading consortia
Phylogenetic analysis of the various 16S rRNA clones suggested a broad diversity within the domain Bacteria (see Fig. 2) including γ, δ and ϵ Proteobacteria, Cytophagales, low G+C content Gram-positives and one deeply rooted clone (SB-34) which is not closely related to any other taxa (although a more rigorous analysis of the sequence would be necessary to resolve its specific affiliation). The γ Proteobacteria contained only one clone, SB-3, which matched closely with Thiomicrospira sp. str. L12 (98.4% similarity). Bacteria of this genus are known to be common members of the microbial community around hydrothermal vents such as the area where the original inoculum for these cultures was collected . The clone SB-17 was the only representative of the epsilon subdivision of the Proteobacteria extracted from this culture. Its affinity with other members of this group was very weak (∼85% similarity to both Campylobacter and Wolinella) indicating that this clone may represent a new group within the subdivision.
The six non-proteobacterial clones are widely dispersed throughout the domain Bacteria. Two of these, SB-1 and SB-5, appear to fall within the order Cytophagales of the Bacteroides and Cytophaga group where their closest relative is Anaeroflexus maritimus (90.5% and 89.8% similarity respectively). Three others, SB-15, 22 and 45, are related to the low G+C Gram-positive bacteria. Clone SB-22 has a sequence nearly identical to Exiguobacterium aurantiacum (99.9% similarity). The other two form a pair of closely related genotypes with 95.7% similarity between them, that is deeply rooted within this subdivision. The last clone, SB-34, does not appear to be related to any of the known phyla. The closest match obtained using similarity rank had an S_ab value of 0.373 .
Four clones (SB-9, 21, 29 and 30) fell within the delta Proteobacteria, in the family Desulfobacteriaceae(Fig. 3). One of these showed a close similarity (95.0% of unambiguously aligned bases) to a known aromatic hydrocarbon degrader, Desulfobacula toluolica str. Tol-2. The remaining three sulfate-reducers were all associated with Desulfosarcina variabilis, which has the ability to degrade benzoate . Of these, SB-29, showed 95.1% similarity, while the others were somewhat less related, 90% and 92% similarity for SB-21 and SB-30, respectively.
3.4PCR fingerprinting of 16S rRNA genes
Such a large number of 16S rRNA genes within the highly enriched consortium was not expected. Given these observations, the question arose as to whether we had obtained most or all of the 16S rRNA genes present in the consortia. Therefore, a direct fingerprinting method was employed to identify the number of different 16S rRNA genes present in the sample without the cloning step. The fingerprints from the consortia and two mixtures of the cloned 16S rRNA genes are presented in Fig. 4. Mixture #1 contained the 16S rRNA genes from clones SB-1, 3, 5, 9, 15, and 17. Mixture #2 contained the 16S rRNA genes from clones SB-20, 22, 29, 30, 34, and 45. Six major peaks and a number of minor peaks can be seen in the consortium fingerprint (panel A). Small peaks were ignored as background fluorescence or primer artifact. Alignment of the cloned 16S rRNA genes (panels B+C) with consortium fingerprint indicates the 12 cloned 16S rRNA genes accounted for the majority of peaks seen in the consortium fingerprint with the exception of two bands indicated with the arrows and possibly one other at ∼530 bp. From this, we conclude that the majority of eubacterial 16S rRNA genes amplified from the consortium have been cloned and characterized.
Use of a specific inhibitor of sulfate-reducing bacteria (sodium molybdate) (Fig. 1), along with results from previous experiments , demonstrates that the consortium characterized in this study depends on the activity of SRBs to effect the transformation of benzene. Although we cannot yet determine if the SRBs are involved in the early steps of benzene transformation or in degradation of the end products, it is clear that their activity is necessary for benzene transformation. No pure culture of a benzene-degrading SRB has been isolated to date, nonetheless, several strains from the family Desulfobacteriaceae have been shown to degrade other aromatic compounds [27–29] including toluene [30, 31].
The importance of sulfate reduction in our cultures is reflected in the fact that four of the 12 clones identified belong within the Desulfobacteriaceae (Fig. 3). The close affinity of one of these, SB-9, to a known toluene degrader suggests that this 16S rRNA gene may be from the bacterium responsible for benzene transformation. The other three putative SRB clones, however, are also members of the same proposed family, and there is no direct evidence that a sulfate-reducer carries out the initial transformation.
The presence of a bacterium, clone SB-3, so closely related to a previously described sulfide oxidizer is not surprising despite the strict anaerobic conditions under which the cultures are maintained. Hydrothermal vents such as the one where the original inoculum for this culture was collected  are known to contain large numbers of the genus Thiomicrospira, and not all sulfide oxidizers are strict aerobes . In addition, investigations of the microbial community in anaerobic production waters at oil fields have found significant numbers of sulfide oxidizers [12, 33].
Of the other clones, only SB-17 gives any reliable clues about its possible function in the consortium. As a member of the epsilon Proteobacteria related to Campylobacter and Wolinella it is possible that this bacterium functions as a member of a commensal relationship, possibly scavenging hydrogen produced during fermentation. Co-cultures of anaerobes similar to Syntrophus sp. and Wolinella succinogenes have been shown to degrade several aromatic compounds including phenol and benzoate . The remaining six clones are either distantly related to known genera or are related to poorly described organisms (SB-22) and hence little insight can be gained about their role. Because of the long enrichment period (3 years) and the extent of dilution (∼107) from the original sediment, it is unlikely that any of these bacteria have managed to survive in the culture without growth and thus it must be assumed that they have some function (although not necessarily in the degradation of benzene).
The technique of cloning and sequencing the 16S rRNA genes amplified from environmental samples to study the composition of microbial communities has become well established during the past 10 years (for review, see ). Variations of these techniques have been used to study the microbial communities in hot springs , marine bacterioplankton , soils , marine sediments , oil fields , activated sludge , and freshwater lakes . The information gained by phylogenetically identifying the members of a consortium can be used to make inferences about the microbially mediated processes occurring in an environment [12, 33] or to design isolation schemes for specific bacteria .
Although 16S characterization is now becoming routine in many labs, potential biases to the traditional clone and sequence approach exist. Primarily, DNA extraction procedures can miss entire groups that are difficult to lyse, such as Gram-positive organisms. Additionally, large amounts of template DNA, high cycle numbers in the PCR amplification, and large amounts of transforming DNA are traditionally used to maximize the amount of colonies obtained during cloning. This strategy, however, can lead to PCR [42–45] or cloning biases resulting from asymptotic transformation of E. coli at DNA masses >10 ng for rubidium chloride treatment or >300 ng for electroporation [46, 47] that can exacerbate artifacts. We have taken steps to minimize these biases inherent in the traditional approach. Our extraction procedure can successfully isolate Gram-positive 16S rRNA genes from sediment samples (see Fig. 2). Furthermore, we routinely use minimum template concentration (<10 ng of genomic DNA), low cycle numbers (20–25), and low transforming DNA (<6 ng) to create our 16S rRNA clonal libraries.
In this report, we have presented a qualitative description of a microbial consortium enriched on benzene as the sole carbon and energy source and sulfate as the terminal electron acceptor. Twelve separate genotypes have been identified, four of which belong in the family Desulfobacteriaceae. This information provides insights into the possible role played by the various members of this consortium, and suggests different approaches to isolation of the relevant strains. Among the Gram-negative SRB, only members of the suggested family Desulfobacteriaceae are known to utilize aromatic compounds . The presence of this family in our enrichment culture suggests that they may be the organisms effecting the degradation of benzene, and selective isolation may be possible because the nutritional differences between these and the other sulfate-reducers have been previously catalogued .
It is possible, however, that the initial steps of mineralization are fermentative, and that the role of sulfate-reducers is to degrade the acetate and hydrogen produced as by-products, such as in the methanogenic benzoate-degrading culture described by Ferry and Wolfe . If this is the case, isolation of the organism(s) responsible will be much more difficult. Because the fermentation of benzene to acetate and hydrogen is energetically unfavorable at standard conditions (ΔG=+45.5 kcal/mol) , axenic culture of a benzene fermenting organism could only be achieved by catalytically removing the hydrogen produced in the reaction.
Finally, the results of this study have provided us with information and molecular tools that can be used to manipulate the consortium by adding alternate substrates or changing culture conditions and monitoring the various members without the necessity of retaining benzene as the sole carbon source. These approaches may prove useful for identifying and isolating the strains responsible for benzene mineralization.
This work was supported in part by a grant from ARPA (grant #N00014-92-J-1888) to L.Y.Y. and by funds from Rutgers University to L.J.K.
We are grateful to Dr. Fred Grassle for procuring the Guaymas Basin sediment. We would also like to thank Drs. Mary Voytek and Victoria Knight as well as David Scala, Andy Peek and Oris Sanjur for their invaluable help with sequencing and analysis, and Dr. Max Häggblom for helpful discussion.