Microbial communities in contaminated sediments, associated with bioremediation of uranium to submicromolar levels

Abstract

Microbial enumeration, 16S rRNA gene clone libraries, and chemical analysis were used to evaluate the in situ biological reduction and immobilization of uranium(VI) in a long-term experiment (more than 2 years) conducted at a highly uranium-contaminated site (up to 60 mg/liter and 800 mg/kg solids) of the U.S. Department of Energy in Oak Ridge, TN. Bioreduction was achieved by conditioning groundwater above ground and then stimulating growth of denitrifying, Fe(III)-reducing, and sulfate-reducing bacteria in situ through weekly injection of ethanol into the subsurface. After nearly 2 years of intermittent injection of ethanol, aqueous U levels fell below the U.S. Environmental Protection Agency maximum contaminant level for drinking water and groundwater (<30 microg/liter or 0.126 microM). Sediment microbial communities from the treatment zone were compared with those from a control well without biostimulation. Most-probable-number estimations indicated that microorganisms implicated in bioremediation accumulated in the sediments of the treatment zone but were either absent or in very low numbers in an untreated control area. Organisms belonging to genera known to include U(VI) reducers were detected, including Desulfovibrio, Geobacter, Anaeromyxobacter, Desulfosporosinus, and Acidovorax spp. The predominant sulfate-reducing bacterial species were Desulfovibrio spp., while the iron reducers were represented by Ferribacterium spp. and Geothrix spp. Diversity-based clustering revealed differences between treated and untreated zones and also within samples of the treated area. Spatial differences in community structure within the treatment zone were likely related to the hydraulic pathway and to electron donor metabolism during biostimulation.

FIG. 1.

Map of the Area 3 treatment zone depicting the location of the sampled wells and the control well. FW024, outer-loop injection well; FW104, inner-loop injection well; FW026, inner-loop extraction well; FW103, outer-loop extraction well. Wells FW100, FW101, and FW102 are multilevel sampling wells. FW106 is a control well. The contamination source is approximately 20 m to the right. (Adapted from a map created by Oak Ridge National Laboratory [http://public.ornl.gov/nabirfrc/FRCMaps/Area3_Inset.jpg].)

FIG. 2.

Typical biostimulation of U(VI) reduction by injecting ethanol to the subsurface (days 704 to 714). (A) COD concentrations. (B) Sulfide concentrations. (C) Uranium concentration changes in MLS wells. The U concentration in injection well FW104 was 0.5 μM during this test period and is not shown due to scale. “+E” indicates the start and −E indicates the stopping of ethanol biostimulation.

FIG. 3.

A tracer test performed with Br⁻ on days 801 to 803 shows the hydraulic connection between injection well FW104, extraction well FW026, and MLS wells and biodegradation of ethanol. (A) Changes in bromide concentrations (21). (B) Comparison of measured COD (symbols) and calculated COD (lines) concentrations based on the COD-versus-Br⁻ ratio (2.46 g/g).

FIG. 4.

Rarefaction curves of the 16S rRNA gene libraries constructed. OTUs were defined at 97% sequence identity. The library from the untreated control (FW106) was close to complete sampling (93% coverage).

FIG. 5.

Microbial composition of the clone libraries based on the RDP Classifier. The “other” group category includes the phyla Spirochaetes, Gemmatimonadetes, Verrucomicrobia, Chlamydia, Planctomycetes, Nitrospira, Cyanobacteria, and the proposed phyla OP11, OP10, BRC1, and TM7. Bacteria that could not be assigned with the 80% confidence bootstrap value were included in an artificial “Unclassified Bacteria” (U. Bacteria) taxon. Wells are arranged according to their descending hydraulic connection to the injection well (FW104) based on the tracer studies.

FIG. 6.

Neighbor-joining tree showing the relationship of selected representatives (shown in bold) from the groups similar to known U-reducing bacteria. Metabolic abilities of the clones' closest cultivated relatives are indicated. Nonmodel positions from the 16S rRNA were masked and Jukes-Cantor distance correction used. Type strains have a “(T)” label. Bootstrap values (10,000 repetitions) are displayed if larger than 50%. +, present in closest relatives; −, absent in closest relatives; *, activity found some species of the genus; ?, unknown. The range of relative contribution to the different samples is also shown.

FIG. 7.

Clustering of samples according to diversity patterns. The topology was similar when using the Sørensen index (presence/absence) or the Bray-Curtis index (presence/absence and abundance). The indices were normalized to account for differences in the total number of sequences per library.