Changes in Bacterial Populations and in Biphenyl Dioxygenase Gene Diversity in a Polychlorinated Biphenyl-Polluted Soil after Introduction of Willow Trees for Rhizoremediation

Abstract

The aim of this study was to analyze the structural and functional changes occurring in a polychlorinated-biphenyl (PCB)-contaminated soil ecosystem after the introduction of a suitable host plant for rhizoremediation (Salix viminalis). We have studied the populations and phylogenetic distribution of key bacterial groups (Alpha- and Betaproteobacteria, Acidobacteria, and Actinobacteria) and the genes encoding iron-sulfur protein α (ISPα) subunits of the toluene/biphenyl dioxygenases in soil and rhizosphere by screening gene libraries using temperature gradient gel electrophoresis. The results, based on the analysis of 415 clones grouped into 133 operational taxonomic units that were sequence analyzed (>128 kbp), show that the rhizospheric bacterial community which evolved from the native soil community during the development of the root system was distinct from the soil community for all groups studied except for the Actinobacteria. Proteobacteria were enriched in the rhizosphere and dominated both in rhizosphere and soil. There was a higher than expected abundance of Betaproteobacteria in the native and in the planted PCB-polluted soil. The ISPα sequences retrieved indicate a high degree of catabolic and phylogenetic diversity. Many sequences clustered with biphenyl dioxygenase sequences from gram-negative bacteria. A distinct cluster that was composed of sequences from this study, some previously described environmental sequences, and a putative ISPα from Sphingomonas wittichii RW1 seems to contain greater diversity than the presently recognized toluene/biphenyl dioxygenase subfamily. Moreover, the rhizosphere selected for two ISPα sequences that accounted for almost 60% of the gene library and were very similar to sequences harbored by Pseudomonas species.

Soil contamination by organic compounds is a serious problem in most industrialized countries. Among pollutants, polychlorinated biphenyls (PCBs) are of special importance, and it is estimated that more than 300 million kg has been released to the environment (20). Remediation of PCB-contaminated soil has traditionally been carried out by incineration or burial of the soil in secure landfills. As a less-expensive removal strategy, rhizoremediation (53), the use of dual plant-microorganism systems in which the plant provides nutrients, support, and a greater availability of the substrate and the microorganisms drive the enzymatic remediation, has been proposed (26, 47).

The soil compartment directly under the influence of living roots is designated the rhizosphere (25). It is a niche where complex microbial communities are supported by nutrients released by root exudates, mucilage, and decaying root cells (7), which induces a characteristic change in the distribution of the microorganisms associated with plant roots in comparison with their distribution in the bulk soil, designated the rhizosphere effect. Thus, in order to exploit the potential of the emerging rhizoremediation technology, a better understanding of the diverse microbial populations and catabolic gene polymorphisms that develop in the rhizosphere is required (46). Salix sp., which produces salicylic acid and related compounds that induce the degradation of many xenobiotic molecules (such as polycyclic aromatic hydrocarbons) (41), is a good candidate for rhizoremediation of PCB-contaminated soils and has already been used in PCB rhizoremediation studies (1).

The aerobic degradation of PCBs is initiated by enzymes of the toluene/biphenyl subfamily of the Rieske nonheme iron oxygenases (13). These enzymes are multicomponent complexes composed of a terminal oxygenase iron-sulfur protein (ISP) and various electron transport proteins (4). The ISPs are heteromultimers, composed of a large (α) and a small (β) subunit. The α subunit contains the substrate-binding site (4, 11) and thus is responsible for substrate specificity (13). Recently, various culture-independent studies have shown that the ISPα sequences obtained from cultured strains (from which most available information on ISPα structure and mechanism derives) are not likely to represent the functional gene diversity in the environment, as distinct and numerically dominant (putative) functional genes were obtained from screening various different environments (43, 44, 51). The use of PCR-based genetic-profiling techniques for the study of ISPα diversity in the environment (23, 43, 52) can be employed for monitoring the predominant ISPα's structure and diversity and for detecting enzymes selected under the existing environmental conditions, which should prove highly valuable in future bioremediation strategies.

There are few studies analyzing the bacterial community in PCB-polluted soils, either by culture-dependent (27) or culture-independent methods (31, 32), and to our knowledge, no culture-independent studies of the rhizosphere of plants growing in PCB-polluted soil have been reported. In a previous study (1), we assessed the impact caused by the introduction of genetically engineered bacteria designed for rhizoremediation on the native microflora of a naturally PCB-polluted soil. The aim of this study was to evaluate the structural and functional changes occurring in the bacterial populations of a PCB-contaminated soil ecosystem after the introduction of a suitable host plant for rhizoremediation by screening environmentally derived gene libraries by temperature gradient gel electrophoresis (TGGE) and sequencing.

MATERIALS AND METHODS

Experimental setup.

Willows (Salix viminalis x schwerinii, variety Björn), were pregrown from 20-cm-long cuttings in tap water for three weeks. Three willow plants were planted per galvanized iron pot in 1,900 g of previously homogenized and sieved (2-mm mesh) PCB-contaminated soil. Nonvegetated pots were included as controls. The soil, obtained from Lhenice (Czech Republic), contained Delor 103 (similar to Aroclor 1242), resulting from historic industrial activity, at a concentration of 182 ± 22 mg PCB/kg dry soil (1). All pots were incubated under controlled conditions (1). After 6 months, the plants were removed from three replicate pots, and the roots from each pot were shaken carefully to remove the bulk soil. The soil still adhering to the roots was defined as “rhizosphere” soil from the corresponding replicate. “Soil” samples were taken from three replicate control pots by coring the potted soil using polyvinyl chloride tubes (28-mm inside diameter). All samples were stored at −20 C until DNA extraction.

DNA extraction and clone library construction.

Total DNA was extracted from 0.5-g (wet weight) samples of each of the three rhizosphere and bulk soil fractions following a protocol described by Porteus et al. (35) and then purified using a Microcon YM-100 filtering device (Millipore, Bedford, MA). One μl of pooled DNA (ca. 15 ng) was used to amplify the 16S rRNA genes with group-specific primers targeting Alphaproteobacteria (α-U203 [15] and L1494 [50]), Betaproteobacteria (F948β [15] and L1494), Acidobacteria (31F [3] and L1494), and Actinobacteria (F243HGC and R1378 [19]), and part of the gene encoding the ISPα subunit of the Rieske nonheme oxygenases of the toluene/biphenyl subfamily was targeted by employing primer pair bphAf668-3/bphAr1153-2 (51). The annealing temperatures employed for each primer pair were 56°C, 61°C, 42°C, 63°C, and 58°C, respectively.

The PCR mixture consisted of 2.5 μl of reaction buffer, 5 pmol of primers, 3.75 mM MgCl₂, 0.2 mM deoxynucleoside triphosphates, 5 mg bovine serum albumin, 4% dimethyl sulfoxide, and 1 U Tth DNA polymerase (Biotools, Madrid, Spain) in a final volume of 25 μl. The PCR program was as follows: 5 min of denaturation at 94°C, followed by 25 cycles of 1 min at 94°C, 1 min for primer annealing, 2 min at 72°C for primer extension, and a final cycle at 72°C for 10 min. The products of two consecutive PCRs were then pooled and purified through extraction from agarose gels prior to cloning on pDRIVE vectors (QIAGEN, Germany). The resulting plasmids were transformed in competent Escherichia coli DH5α cells and checked for inserts of the correct size.

In order to infer the percentage of the total bacteria included in the study, the 16S rRNA genes from the Eubacteria present were amplified as well, using universal primers (27f [50] and L1494), and then cloned on pDRIVE and transformed in E. coli DH5α cells. The clones presenting an insert of the correct size were used as templates for group-specific PCRs.

TGGE.

One-microliter amounts of 1:10 dilutions of the amplified products were used for a second round of nested PCR using high-pressure liquid chromatography-purified primers (48). For 16S rRNA gene amplification, primer pair F984GC (19) and R1378 was used, resulting in ∼475-bp fragments suitable for TGGE analysis. For the nested amplification of the ISPα genes, primer bphAf668-3 was modified with a GC clamp (5′-CGC CCG GGG CGC GCC CCG GGC GGG GCG GGG GCA CGG GGG G-3′) at its 5′ end.

For sequence-dependent separation of the PCR products derived from the gene libraries, a TGGE system (Biometra, Göttingen, Germany) was used as specified by the manufacturer. Six-percent polyacrylamide gels were polymerized in 1× Tris-acetate-EDTA buffer, 8 M urea, 20% deionized formamide, and 2% glycerol. Amounts of 5 μl of the PCR samples were resolved by electrophoresis at 130 V for 16 h through a temperature gradient of 44.5°C to 56°C for the 16S rRNA genes and 41°C to 56°C for the ISPα genes. Reference patterns consisting of various group-specific sequences were included in the gels to facilitate the analysis of the gels. The DNA bands were visualized by a routine silver-staining protocol (18). The gels were then digitally documented and analyzed (Quantity One; Bio-Rad, Hercules, CA).

Sequencing, phylogenetic analysis, and statistics.

Clones were grouped according to their corrected electrophoretic mobilities, reported as operational taxonomic units (OTUs), and named with the prefix Lhap, Lhbe, Lhad, Lhac, or LhISP depending on the sequence type (Alphaproteobacteria, Betaproteobacteria, Acidobacteria, Actinobacteria, and ISPα sequences, respectively; Lh stands for Lhenice, the origin of the soil). Their frequencies were used as an indicator of abundance. Representative clones were grown, their plasmidic DNA was extracted (RapidPURE kit; QBIOgene, Irvine, CA), and the insert was fully sequenced (Parque Científico de Madrid, Madrid, Spain). The 16S sequences obtained were submitted to the SEQUENCE MATCH program of the Ribosomal Database Project (RDP-II) (6), and the ISPα sequences were submitted to BLAST to find closely related sequences. The sequence identities between clones and the next related sequences were calculated using BIOEDIT (17). The raw sequences obtained were edited and aligned with ClustalW (45). The online programs CHECK_CHIMERA (7) and Bellerophon (21) were used to rule out the presence of chimeric sequences (hybrid artifact sequences that may arise from the PCR amplification of very similar sequences) (10). The raw data (sequences and frequencies) were used to infer possible differences, at group level, between bulk soil and rhizospheric communities. The diversities of clone libraries were calculated using the Shannon index (40). Richness and diversity analysis was performed using the DOTUR software (38). Maximum-likelihood phylogenetic trees were constructed (for the 16S sequences) using a quartet-puzzling algorithm implemented in the Tree-Puzzle software (nucleotide substitution model HKY, including gamma correction and with the molecular clock assumption enforced) (42) and were visualized with TreeView (33). For each group analyzed, a set of 16S rRNA gene sequences from the closest cultured bacteria were downloaded from the RDP-II database (6) in order to obtain a comprehensive view of the phylogeny of the sequences. The distance matrix and the α value for the gamma correction obtained were used in the subsequent analysis. MEGA software (http://www.megasoftware.net/) was used for the phylogenetic reconstruction of the ISPα sequences by using the neighbor-joining method with the JTT distance model and pairwise deletion of gaps/missing data. A consensus tree was inferred from a total of 1,000 bootstrap trees. The F_ST test (29) was used to compare the genetic diversity within each community to the total genetic diversity of the communities combined, using the equation F_ST = (θ_t-θ_w)/θ_t, where θ_t is the genetic diversity for all samples (based on the mean number of pairwise differences) and θ_w is the genetic diversity in each community. The population differentiation using F_ST was calculated using the ARLEQUIN program (39); statistical significance was evaluated by randomly assigning sequences to populations and calculating the F_ST for 3,034 permutations. Additional estimators were also obtained with this software. Each pair of libraries was compared first in an X/Y fashion and then in the reverse Y/X using the ∫-LIBSHUFF computer program (37) which calculates the integral form of the Cramér-von Mises statistic of the formula for coverage (16) and compares it by a Monte Carlo test procedure. The Library compare program of the RDP-II was used as well. It utilizes the RDP naïve Bayesian classifier (confidence threshold, 95%) to provide classification of library sequences into the bacterial taxonomy and estimates the probability of observing the difference in a given taxon using a statistical test.

Community-level catabolic profiles.

The bacterial fraction from each replicate soil and rhizosphere sample was harvested as previously described (1). Each suspension of bacterial cells was diluted 1:10 and used to inoculate ECOlog plates (5) at 135 μl per well; the plates were then incubated at 28°C in darkness. The optical density (OD) at 595 nm was measured with a microtiter plate reader (Tecan, Zurich, Switzerland) when a maximum of wells showed activity (66 h) (14). The raw OD values were corrected for the background color in the control well without a carbon source. To account for slightly differing initial cell densities, the corrected ODs were divided by the average well color of the plate, giving the standardized OD; thus, standardized patterns rather than absolute values were compared (12). Values smaller than 0.006 were counted as 0 (no catabolic activity) to reduce noise. The resulting 31 variables (corresponding to each carbon source) were reduced to three new principal components (SPSS 12.0) which explained more than 80% of the variance (total eigenvalue sum) using the covariance of the variables. In order to test for differences between the communities, a multivariate analysis was performed (general linear model function; SPSS 12.0) using the new principal components as variables (14).

Nucleotide sequence accession numbers.

The sequence data have been submitted to the GenBank database under accession numbers DQ648899 to DQ649002 and EF565826 to EF565853.

RESULTS

Composition of the eubacterial populations in soil and rhizosphere.

Soil- and rhizosphere-derived eubacterial 16S rRNA gene libraries were constructed and screened for clones belonging to Alpha- and Betaproteobacteria, Acidobacteria, and Actinobacteria by amplifying random clones with each of the group-specific primer pairs. Forty-eight clones from the soil library and 43 clones from the rhizosphere library were screened. Table 1 shows the percentage of clones from each library that was amplified with each of the primer pairs. No clones were amplified with more than one primer pair, indicating that the group specific-primers were targeting nonoverlapping groups. The results also show that the targeted groups represented a significant fraction of the eubacterial populations studied (Table 1). The soil and rhizosphere eubacterial communities appear to be dominated by Proteobacteria, representing 44% and 52% of the soil and rhizosphere amplified clones, respectively. Both Alpha- and Betaproteobacteria were enriched in the rhizosphere compared to their presence in soil.

TABLE 1.

Percentage of the total eubacterial clones screened that were positively amplified with the different group-specific primer sets

Primer pair specificity	% of clones amplified in gene library from:
	Soil	Rhizosphere
Alphaproteobacteria	25.0	32.5
Betaproteobacteria	18.7	25.7
Acidobacteria	14.6	7.0
Actinobacteria	8.3	4.6
Total	66.7	69.8

Analysis of the group-specific 16S rRNA gene libraries.

Group-specific 16S rRNA gene libraries were constructed using soil and rhizosphere DNA samples. Random clones were subjected to nested PCR and TGGE, and OTUs were defined by gel mobility within each library (see Materials and Methods). The number of times that an OTU appeared was scored, and one clone corresponding to each OTU was fully sequenced. A total of 320 clones representing 105 OTUs were analyzed, corresponding to a total of more than 115,000 bp sequenced. Table 2 shows the observed diversity for each library based on OTU abundances. The Shannon index of diversity was higher in the rhizosphere for Alpha- and Betaproteobacteria and higher in the soil for the Actinobacteria, although these differences were not statistically significant. However, a significant difference was found for the Acidobacteria community, which was more diverse in the rhizosphere. Table 2 also shows the genetic diversity, inferred by the nucleotide diversity and θ(π) (average sequence divergence) values, which were very similar in the soil and rhizosphere for the Acidobacteria and the Actinobacteria and slightly higher in the rhizosphere for the Alpha- and Betaproteobacteria. The closest relative (most similar sequence) found in the databases for each OTU, as well as the number of times it was scored, is shown in Table 3.

TABLE 2.

Numbers and diversity indices for gene libraries derived from rhizosphere or soil

Gene library	Zonea	Clones screened	No. of OTUs	Shannon index value ± SD	Nucleotide diversity ± SD	θ(π)b value ± SD
Alphaproteobacteria	S	42	21	3.2 ± 0.2	0.18 ± 0.09	220 ± 110
	R	41	18	3.3 ± 0.2	0.20 ± 0.1	260 ± 130
Betaproteobacteria	S	37	13	2.4 ± 0.2	0.09 ± 0.04	43 ± 21
	R	45	21	2.8 ± 0.3	0.11 ± 0.05	51 ± 25
Acidobacteria	S	30	13	2.3 ± 0.3	0.22 ± 0.11	180 ± 90
	R	41	23	2.9 ± 0.2	0.21 ± 0.10	190 ± 90
Actinobacteria	S	46	19	2.7 ± 0.2	0.31 ± 0.15	370 ± 180
	R	38	17	2.6 ± 0.2	0.32 ± 0.15	380 ± 180
ISPα	S	48	20	2.7 ± 0.2	0.48 ± 0.23	250 ± 120
	R	47	14	2.0 ± 0.3	0.35 ± 0.17	190 ± 90

^aR, rhizosphere; S, soil.

^bθ(π) is a measure of genetic diversity.

TABLE 3.

Bacterial sequences from the RDP-II database with the highest similarity to each OTU (closest relative) and distribution of OTUs in the rhizosphere or soil samples

Clone	Size (bp)	S/Ra	Closest relative; GenBank accession number	Similarity (%)b	Gene library
Lhap1	1,292	3/0	Bacterium Ellin332; AF498714	98.8	Alphaproteobacteria
Lhap2	1,265	4/0	Bacterium Ellin335; AF498717	89.6
Lhap3	1,295	1/0	Pedomicrobium manganicum; X97691	91.4
Lhap4	1,296	1/0	Uncultured alphaproteobacterium; AJ582034	98.8
Lhap5	1,259	2/0	Afipia massiliensis; AY029562	87.9
Lhap6	1,294	2/1	Afipia massiliensis; AY029562	99.2
Lhap7	1,297	2/1	Alphaproteobacterium; AY145553	98.4
Lhap8	1,294	3/2	Magnetospirillum sp. strain CF20; AJ863153	87.7
Lhap9	1,292	4/4	Uncultured alphaproteobacterium; AJ318111	97.0
Lhap10	1,295	6/3	Uncultured bacterium; AY625147	98.6
Lhap11	1,296	1/0	Uncultured bacterium; AY625147	97.0
Lhap12	1,285	1/1	Uncultured bacterium; AY662029	83.6
Lhap13	1,298	3/1	Uncultured forest soil bacterium; AY913244	94.9
Lhap14	1,292	2/0	Uncultured sludge bacterium; AF234730	98.5
Lhap15	1,280	0/1	Bradyrhizobium sp. strain Shinshu-th2; AB121773	83.9
Lhap16	1,285	0/3	Uncultured sponge symbiont; AF186410	92.2
Lhap17	1,299	0/2	Uncultured sludge bacterium; AF234730	97.6
Lhap18	1,275	0/4	Uncultured bacterium; AY212706	99.4
Lhap19	1,330	0/3	Rhizobium etli; AY460185	93.6
Lhap20	1,291	0/2	Rhizobium etli; AY460185	98.8
Lhap21	1,295	0/1	Uncultured bacterium; AY212717	95.8
Lhap22	1,296	0/2	Uncultured bacterium; AF358012	96.7
Lhap23	1,292	0/3	Pedomicrobium australicum; X97693	98.8
Lhap24	1,296	1/6	Uncultured bacterium; AY221056	94.9
Lhap25	1,294	1/0	Uncultured soil bacterium; AY326604	93.0
Lhap26	1,290	1/0	Uncultured bacterium; AY625147	96.9
Lhap27	1,294	1/0	Uncultured bacterium; AY625147	98.2
Lhap28	1,269	1/0	Brevundimonas bullata; AB023428	99.1
Lhap29	1,298	1/1	Afipia massiliensis; AY029562	96.5
Lhap30	1,294	1/0	Uncultured alphaproteobacterium; AJ582029	98.1
Lhbe1	469	3/0	Uncultured soil bacterium; AF423284	99.1	Betaproteobacteria
Lhbe2	465	1/2	Uncultured bacterium; AY752109	99.3
Lhbe3	470	0/3	Uncultured betaproteobacterium; AJ534662	99.7
Lhbe4	470	0/1	Acidovorax delafieldii; AF078764	98.0
Lhbe5	470	0/4	Uncultured bacterium; AY050592	99.3
Lhbe6	466	0/1	Uncultured Comamonadaceae sp.; AF523023	99.3
Lhbe7	465	0/3	Uncultured Hydrogenophaga sp.; AF523011	99.7
Lhbe8	466	0/1	Uncultured Hydrogenophaga sp.; AF52301	99.1
Lhbe9	470	0/1	Uncultured bacterium; AY706434	98.9
Lhbe10	393	4/0	Uncultured Variovorax sp.; AY599725	99.8
Lhbe11	470	0/3	Variovorax sp. strain WDL1; AF538929	98.5
Lhbe12	470	0/1	Uncultured bacterium; AY250108	99.3
Lhbe13	464	0/3	Uncultured betaproteobacterium; AJ58317	96.1
Lhbe14	465	5/3	Uncultured betaproteobacterium; AY622242	99.5
Lhbe15	465	2/0	Uncultured bacterium; AB166773	97.4
Lhbe16	465	2/2	Uncultured bacterium; AY160866	96.5
Lhbe17	465	4/8	Uncultured soil bacterium; AY326597	98.4
Lhbe18	465	2/1	Betaproteobacterium; AJ224618	100
Lhbe19	471	0/1	Uncultured bacterium; AY212629	97.2
Lhbe20	470	4/1	Uncultured bacterium; AY212629	97.0
Lhbe21	470	2/1	Uncultured betaproteobacterium; AF204242	94.8
Lhbe22	470	1/2	Uncultured betaproteobacterium; AF204242	97.2
Lhbe23	470	0/1	Uncultured betaproteobacterium; AJ581593	95.3
Lhbe24	470	4/2	Uncultured bacterium; AY212629	97.2
Lhbe25	469	3/0	Uncultured bacterium; AY212629	96.8
Lhad1	1,516	3/0	Uncultured Acidobacteria; AY922062	97.0	Acidobacteria
Lhad2	1,505	1/1	Uncultured Acidobacteria; AY281352	97.5
Lhad3	1,510	0/1	Uncultured Acidobacteria; AY922096	98.7
Lhad4	1,051	0/2	Uncultured Acidobacteria; AY281352	94.5
Lhad5	1,516	0/2	Uncultured Holophaga sp.; AJ519375	98.3
Lhad6	1,519	1/3	Uncultured Acidobacteria; AY921697	96.6
Lhad7	1,504	0/3	Uncultured Acidobacteria; AY921697	96.9
Lhad8	1,505	0/2	Uncultured soil bacterium; AF423235	98.5
Lhad9	1,504	0/3	Uncultured Acidobacteria; AY921697	98.8
Lhad10	1,502	0/1	Uncultured sludge bacterium; AF234731	97.3
Lhad11	1,511	5/0	Uncultured soil bacterium; AY493932	96.4
Lhad12	1,505	0/3	Uncultured Acidobacteria; AY922121	98.7
Lhad13	1,479	3/1	Uncultured sponge symbiont; AF186413	94.7
Lhad14	1,455	0/1	Uncultured Acidobacteria; AY395390	96.1
Lhad15	1,449	0/2	Uncultured Eubacterium; AJ292582	92.9
Lhad16	1,482	3/1	Uncultured Acidobacteria; AY921970	97.9
Lhad17	1,481	1/0	Uncultured Acidobacteria; AY921970	98.7
Lhad18	1,481	1/2	Uncultured Acidobacteria; AY921727	98.1
Lhad19	1,476	0/2	Uncultured Antarctic bacterium; AF173824	95.9
Lhad20	1,474	0/2	Uncultured Antarctic bacterium; AF173824	96.7
Lhad21	1,519	1/2	Uncultured bacterium; AY456757	97.3
Lhad22	1,520	2/1	Bacteria; Z95711	98.3
Lhad23	1,521	2/0	Uncultured soil bacterium; AY326536	97.9
Lhad24	1,521	1/2	Uncultured Acidobacteria; AY281358	93.2
Lhad25	1,514	0/1	Bacteria; Z95722	96.9
Lhad26	1,534	6/0	Bacteria; Z95730	87.3
Lhad27	982	0/1	Uncultured bacterium; AF371525	95.0
Lhad28	1,505	0/2	Uncultured soil bacterium; AF423235	98.5
Lhac1	1,154	7/0	Uncultured actinobacterium; AJ581630	96.3	Actinobacteria
Lhac2	1,154	1/1	Uncultured actinobacterium; AJ581630	96.6
Lhac3	1,154	2/1	Uncultured actinobacterium; AJ581630	96.4
Lhac4	1,153	1/3	Uncultured actinobacterium; AJ581606	95.8
Lhac5	1,156	3/1	Uncultured bacterium; AJ576407	97.2
Lhac6	1,158	3/1	Mycobacterium sp. strain Ellin151; AF408993	97.8
Lhac7	1,159	4/6	Microbacterium sp. strain V4.BE.51; AJ244679	98.9
Lhac8	1,159	2/4	Curtobacterium sp. strain VKM; AB042093	97.2
Lhac9	1,165	1/4	Streptomyces sp. strain R46S; AY572485	99.5
Lhac10	1,165	1/2	Streptomyces ciscaucasicus; AY508512	99.6
Lhac11	1,165	4/0	Streptomyces sp.; Y10842	99.6
Lhac12	1,157	0/1	Uncultured actinobacterium; AY250884	98.4
Lhac13	1,164	2/0	Uncultured bacterium; AB179534	96.7
Lhac14	1,164	6/4	Uncultured sludge bacterium A21; AF234742	96.0
Lhac15	1,186	1/1	Unidentified eubacterium EA25; U51864	97.9
Lhac16	1,186	0/1	Uncultured Xiphinematobacteriaceae; AY395325	98.4
Lhac17	1,183	0/1	Bacterium Ellin507; AY960770	96.7
Lhac18	1,182	1/2	Uncultured bacterium 139ds10; AY212589	98.4
Lhac19	1,176	1/3	Uncultured Verrucomicrobia; AJ575731	86.2
Lhac20	1,182	2/0	Uncultured bacterium KD4-60; AY218640	98.9
Lhac21	1,186	3/2	Uncultured Verrucomicrobia; AY921923	95.4
Lhac22	871	1/0	Streptomyces phaeochromogenes; AF500071	99.5
LhISP1	481	0/4	Uncultured bacterium; AAZ95312.1	100	ISPα
LhISP2	493	1/1	Uncultured bacterium; AAZ95312.1	96.3
LhISP3	483	1/0	Uncultured bacterium; AAZ95312.1	83.7
LhISP4	484	1/0	Uncultured bacterium, AAZ95312.1	86.8
LhISP5	481	2/1	Uncultured bacterium; AAZ95327.1	99.3
LhISP6	484	1/0	Uncultured bacterium; AAZ95307.1	86.8
LhISP7	481	2/0	Uncultured bacterium; AAZ95307.1	92.5
LhISP8	483	1/19	Pseudomonas sp. strain IC; AAZ95273.1	94.4
LhISP9	484	6/1	Ralstonia eutropha H850; CAD67505.1	95.0
LhISP10	484	2/0	Ralstonia eutropha H850; CAD67505.1	95.0
LhISP11	481	0/9	Pseudomonas putida; AAX45786.1	75.0
LhISP12	469	0/2	Australian soil clone B24; AAF23625.1	55.0
LhISP13	394	3/0	Australian soil clone B24; AAF23625.1	46.8
LhISP14	475	4/0	Australian soil clone B24; AAF23625.1	52.0
LhISP15	478	2/0	Australian soil clone B24; AAF23625.1	44.0
LhISP16	472	0/1	Australian soil clone GV10; AAD53248.1	64.1
LhISP17	466	1/0	Sphingomonas wittichii RW; ZP01607365.1	54.8
LhISP18	494	1/0	Uncultured bacterium; ABB77850.1	37.1
LhISP19	454	8/2	Australian soil clone GV2; AAD53249.1	55.2
LhISP20	545	3/0	Australian soil clone GV2; AAD53249.1	55.9
LhISP21	454	2/0	Australian soil clone GV2; AAD53249.1	56.5
LhISP22	457	0/1	Australian soil clone GV2; AAD53249.1	50.9
LhISP23	451	1/0	Australian soil clone OD18; AAF23624.1	55.3
LhISP24	478	0/1	Australian soil clone OD18; AAF23624.1	60.4
LhISP25	454	4/1	Uncultured bacterium; ABB77848.1	40.1
LhISP26	514	2/0	Australian soil clone H39; AAD53253.1	40.1
LhISP28	500	0/2	Australian soil clone H39; AAD53253.1	37.9

^aNumber of times the OTU was scored in soil (S) versus in rhizosphere (R).

^bIn the case of the ISPα sequences, amino acid-based similarities are shown.

Alphaproteobacteria.

In the case of the Alphaproteobacteria, their closest relatives (similarities of between 83.6 and 99.4%) were usually from uncultured bacteria, although close similarities to several cultured species from the Rhizobiales order were obtained. A wide phylogenetic distribution was observed for this set of sequences (see Fig. S1A in the supplemental material). Many sequences clustered as Rhizobiales, but other orders, including Rhodobacterales, Caulobacterales, and Rhodospiralles, were also represented in the set. Several OTUs showed deep branching in the phylogenetic tree. The group formed by OTUs Lhap1 and Lhap2 shows a close relationship to bacteria that form a novel unnamed order of Alphaproteobacteria described previously (36) and probably belongs to this order. The groups formed by Lhap25 and Lhap8, as well as Lhap15 and Lhap16, do not show a close relationship to any described taxon and may represent novel lineages within Alphaproteobacteria. Sequences from the Rhizobiaceae family were only found in the rhizosphere library, and a slight rise in the population of Rhodobacteraceae was detected in the rhizosphere (P < 0.0588) by using the RDP-II library compare test.

Betaproteobacteria.

Most betaproteobacterial sequences obtained were found to be very similar (94.8 to 100%) to sequences obtained from uncultured organisms already present in the databases. The phylogenetic distribution of the betaproteobacterial sequences obtained (see Fig. S1B in the supplemental material) was clearly dominated by the Burkholderiales order, including members from the Oxalobacteraceae, Alcaligenaceae, and Comamonadaceae families, in both the soil and the rhizosphere libraries. A significant rise in the number of sequences belonging to the Comamonadaceae family (P < 0.032) was observed in the rhizosphere by using the RDP-II library compare test. This result is also clearly observable in the phylogenetic tree (see Fig. S1B in the supplemental material).

Acidobacteria.

Almost all Acidobacteria sequences obtained were most similar to uncultured species (93 to 98.8%). Most sequences, according to the grouping by Barns et al. (3), were from groups III and V (see Fig. S2A in the supplemental material). Group II sequences were detected only in the rhizosphere. No sequences were detected for the other groups. One of the OTUs (Lhad13), which was found both in the soil and rhizosphere libraries, did not belong to any of these groups, as judged from its deep branching. This OTU, together with the previously reported sequence from a sponge symbiont (GenBank accession no. AY703463), may represent a novel group within Acidobacteria. No classification or comparison was possible by using the RDP-II for the Acidobacteria sequences since there is yet no robust phylogeny for this group.

Actinobacteria.

The Actinobacteria sequences were found to be very similar (95.4 to 99.6%) to their closest relatives, which were often cultured species. Surprisingly, two main clusters were found, according to their phylogenetic distribution (see Fig. S2B in the supplemental material), the Verrucomicrobiales and the Actinobacteria clusters, which indicates a lack of true specificity for the Actinobacteria-specific primer pair under the PCR parameters employed. Most actinobacterial sequences were from the Actimomycetales order, including representatives of various families such as Sporichtyaceae, Microbacteriaceae, Mycobacteriaceae, and Microsphaeraceae, many sequences from the genus Streptomyces, and one cluster of unclassified Actinobacteria. No differences between soil- and rhizosphere-derived libraries were observed in the phylogenetic distribution of the sequences. Moreover, no significant difference between libraries was detected according to the RDP-II library compare test.

Statistical comparisons of soil- versus rhizosphere-derived 16S rRNA gene libraries.

A statistical test of differences between soil- and rhizosphere-derived group-specific libraries (Table 4) was performed by comparing the genetic diversity within each community to the total genetic diversity of the communities combined (F_ST test). The test results indicate significant differences between gene libraries for all groups except for the Actinobacteria. The same test results were obtained when analyzing differences between homologous and heterologous coverage curves by a Cramér-von Mises-type statistic and comparing them by a Monte Carlo test procedure (∫-LIBSHUFF).

TABLE 4.

Statistical analyses of the difference between rhizosphere and soil bacterial communities

Gene library	FST test result (P value)a	∫-LIBSHUFF test result P valuea
Alphaproteobacteria	0.03 (<0.01)*	0.05/<0.01*
Betaproteobacteria	0.06 (<0.01)*	0.06/<0.01*
Acidobacteria	0.09 (<0.01)*	0.01/<0.01*
Actinobacteria	0.01 (0.64)	0.42/0.45
ISPα	0.10 (<0.01)*	<0.01/<0.01*

^aEach pair of libraries was analyzed first in an X/Y fashion and then in the reverse Y/X fashion by using the ∫-LIBSHUFF computer program (37), as explained in Materials and Methods. *, significant differences detected at a P value of <0.05.

Analysis of community-level catabolic profiles.

No statistical difference was found between the Shannon diversity indices of the soil and rhizospheric communities based on the use of 31 ecologically relevant carbon sources (5) (3.00 ± 0.05 [mean ± standard deviation] for soil versus 2.92 ± 0.07 for rhizosphere). Nevertheless, a significant statistical difference between the soil and rhizospheric catabolic potentials was found by analyzing the catabolic profiles (P < 0.02).

ISPα gene diversity.

A PCR-TGGE approach was employed for the detection and discrimination of the most abundant ISPα genes of the toluene/biphenyl subfamily of the aromatic ring-hydroxylating dioxygenases present in soil and rhizosphere samples from a PCB-polluted soil. Forty-eight soil-derived clones and 47 rhizosphere-derived clones were analyzed, yielding 28 different OTUs. As expected, all sequences obtained were designated by BLAST searches as Rieske nonheme iron dioxygenases. The inferred peptide sequence of the ISPα sequences obtained ranged between 37% (LhISP18) and 100% (LhISP1) identity to previously described ISPα sequences (Table 3). Most of them contained residues considered characteristic of these enzymes (P. fluorescens IP01 cumene dioxygenase [CumDO] numbering) (8): His234 and His240, involved in the coordination of mononuclear iron at the active site (8, 11); Asp231, with a role in the electron transfer bridge (8, 11); and residues Glu226, Gln227, and Tyr233 (or the equivalent Glu/Asp, Gln/Asn, and Tyr/His), which are thought likely to be involved in electron transfer by connecting the catalytic center to the Rieske center of an adjacent subunit (13, 34).

On the other hand, great variation was observed among the 14 residues proposed to define the inner surface of the substrate-binding pocket in IP01 CumDO (51) (Fig. 1). The phylogenetic analysis of the peptide sequences deduced from the clone sequences (see Fig. S3 in the supplemental material) shows that all of them were members of the toluene/biphenyl subfamily. Many sequences clustered with the gram-negative biphenyl dioxygenase group, while none of them grouped with members of the isopropyl/benzene, toluene/benzene, and gram-positive biphenyl dioxygenase clusters (grouping according to Witzig et al. [51]). Others (LhISP11, LhISP12, LhISP13, LhISP14, and LhISP15) could not be unambiguously assigned to any of the previously described clusters. Surprisingly, many sequences were grouped in a cluster represented previously only by inferred sequences from environmental clones and one putative dioxygenase from Sphingomonas wittichii RW1. As expected, no sequences clustered with members of the other Rieske nonheme oxygenase subfamilies.

FIG. 1.

Residues contributing to the inner surface of the substrate-binding pocket of CumDO from IP01 (CumA1, gray box) and their counterparts in the deduced peptide sequences found in this study. (A) Main cluster sequences. (B) S. wittichii (S.w.) RW1 cluster sequences. The numbers above the panels correspond to the positions in the CumA1 IP01 sequence. Conserved residues are shown in gray, and highly conserved residues are shown in black. Representative sequences from each group are also shown (bold).

Members of the main cluster (including sequences grouped with the gram-negative biphenyl dioxygenases plus LhISP11, LhISP12, LhISP13, LhISP14, and LhISP15) had an average of 5 residues differing with respect to the 14 residues responsible for the substrate-binding pocket in IP01 (Fig. 1). Notably, all of them had fixed Gly321 and Phe384 (instead of Ala and Tyr at such positions in IP01). Surprisingly, some sequences lacked putatively important residues; LhISP4 had apparently no active site, LhISP6 lacked the residue His234, and LhISP9 lacked the residue Asp231. LhISP17 and LhISP16 (grouped in the same subcluster) had the residue His240 at position 238 and Asn227 instead of Gln. Members of the S. wittichii RW1 cluster have all Asp-Asn instead of Glu226-Gln227, and LhISP27 and LhISP28 lack the Asp231 at the active site. These last groups have an average of 11 differences in the pocket residues in comparison with those of IP01.

Comparison of soil- versus rhizosphere-derived ISPα clone libraries.

Comparison of the ISPα clone libraries derived from the soil and rhizosphere revealed differences between their catabolic gene structures; both clone libraries were statistically different according to the F_ST and ∫-LIBSHUFF tests (Table 4), and the Shannon, nucleotide, and θ(π) diversity indexes were higher for the soil library (Table 2). Moreover, there was a strong decrease in the abundance of clones from the S. wittichii RW1 cluster in the rhizosphere library (from 45.76 to 21.2%), especially in the group formed by sequences LhISP19, LhISP20, and LhISP21, which decreased from 27% to 4.2%. On the other hand, there was a rise in the abundance of sequences from the main cluster in the rhizospheric library (54 to 78.6%), driven by the increase in the abundance of sequences LhISP1 (0 to 8.5%), LhISP11 (0 to 19.1%), and, particularly, LhISP8 (2 to 40.4%). Nevertheless, there was a notable decrease in the abundance of other groups, such as those formed by LhISP14-LhISP13 (14.6 to 0%) and LhISP9-LhISP10 (16.6 to 2.1%).

DISCUSSION

The aim of this study was to evaluate the structural and functional changes occurring in the bacterial communities of a PCB-contaminated soil after the introduction of a suitable host plant for rhizoremediation. This was achieved by screening soil- and rhizosphere-derived 16S rRNA and ISPα gene libraries via TGGE (since amplicons with identical length migrate dependent upon their respective primary sequences and base compositions [18]). The clones with identical mobility were defined as OTUs. Felske et al. (9) and Bano and Hollibaugh (2) had previously shown that clone screening by denaturing gradient gel electrophoresis or TGGE is a convenient and efficient way to detect and retrieve the predominant sequences in a sample. Nevertheless, to confirm that the grouping of different OTUs according to TGGE mobilities was reproducible and reliable for discriminating between clones, replicates of five different OTUs (one from each pair of gene libraries) were randomly chosen and sequenced, and no differences in sequence composition were obtained, suggesting that such grouping was efficient for this study.

The selection of the bacterial groups to be studied, which was done depending on the group-specific primers available in the literature and the published information on the groups most commonly present in soils, has allowed the study of approximately 67% to 70% of the bacterial diversity from soil and rhizosphere (Table 1). Furthermore, the long sequences obtained with these primers have given us a great deal of phylogenetic information, allowing a fine clustering of the sequences. Except for the Actinobacteria-specific primers, which also amplified sequences clearly belonging to the unrelated Verrucomicrobia, all the primer pairs seem to be truly group specific.

We have used different approaches to analyze the information obtained, each having particular advantages and limitations (29). In the present study, statistically significant differences were found between soil and rhizospheric communities, using tests based on library coverage (∫-LIBSHUFF) and genetic diversity (F_ST), for all groups studied except for Actinobacteria. In addition, the catabolic profiles of both communities were found to be significantly different.

The relative abundances of the phylogenetic groups studied fall within the ranges previously reported for soil (for a review, see Janssen [22]) and rhizosphere environments (30, 24). Interestingly, the higher-than-expected abundance of Betaproteobacteria found in this PCB-polluted soil was also observed at another PCB-polluted site in Wittenberg, Germany (31, 32), while in most soils studied, this group has been found in relatively small numbers (22). Moreover, an unusual feature of the Wittenberg soil was the abundance of sequence types within the Betaproteobacteria that were related to Variovorax and Burkholderia species, which are highly represented in our Betaproteobacteria libraries as well. Nogales et al. (32) speculated that such high abundance could be due to their ability to thrive in acidic soils, but in our case the soil pH (7.2) cannot be taken as a cause. The enrichment of such species in PCB-contaminated soils of different origins is noteworthy, as it may represent selection due to the presence of PCBs. Nogales et al. (31) postulated that if the PCB concentration was high enough to account for an important part of the soil's carbon and energy reservoir (as is in our case), some of the sequence types detected should indicate the organisms likely to be involved, directly or indirectly, in the utilization of PCBs as a carbon and/or energy source. Following this rationale, it could be that some Burkholderiales species have an important functional role in the bacterial degradation of PCBs in soil.

In a recent study, based on the analysis of the culturable PCB-degrading bacteria associated with the rhizosphere of different tree species established at a PCB-contaminated site (27), Salix sp. trees were found to positively affect the size of the PCB-degrading populations compared to the results for three other tree species and non-root-containing soil. Thus, if we assume that this could also be true for nonculturable PCB-degrading bacteria, which should be the majority of PCB degraders, some of the sequence types enriched in the rhizosphere (mainly Proteobacteria species) could represent PCB degraders.

All putative ISPα sequences obtained in this study seem to be members of the toluene/biphenyl subfamily of the Rieske nonheme dioxygenases, as evidenced by BLAST searches and phylogenetic analysis. The sequences obtained show great diversity with respect to previously described Rieske nonheme oxygenases, either from well-studied cultured species or from environmental clones, with identities ranging from 37% to 100%. Many of them contained all the amino acid residues thought to be involved in the active-site functioning and those thought to play a role in electron transfer, and thus, they probably code for functional enzymes. Those sequences lacking some of the important active-site residues (LhISP4, LhISP6, LhISP9, LhISP27, and LhISP28) may represent inactive (vestigial) enzymes.

Since the samples came from a highly PCB-polluted environment (without significant amounts of other aromatic hydrocarbons), it is not surprising to find that many sequences clustered with the gram-negative biphenyl dioxygenase branch instead of the isopropylbenzene and toluene/benzene groups (since a broad correlation between this grouping and substrate range as been previously described [51]). However, the lack of sequences closely related to described gram-positive biphenyl dioxygenase sequences is noteworthy, as more than 8% of the eubacterial soil clones analyzed were Actinobacteria sequences.

We were surprised to find a previously undescribed cluster composed of many sequences from this study but only a few previously described environmental sequences and a putative ISPα from Sphingomonas wittichii RW1. Such a grouping is clearly separated from the other subfamilies, and yet it seems to contain far greater diversity than the main cluster of the toluene/biphenyl subfamily as judged from its deeper branching (see Fig. S3 in the supplemental material). Moreover, it provides sequence data exploitable for the search of full-length novel dioxygenase genes from a future metagenomic library.

The samples analyzed contained a high degree of catabolic genotype diversity, as evidenced by the low degree of homology of many of the sequences obtained (compared to that of previously described ISPα sequences), the variability found in the residues responsible for active-site functioning, and the great variation observed in the amino acid residues homologous to those defining the substrate-binding pocket in IP01 and other Rieske nonheme dioxygenases (Fig. 1). This suggests that previously described dioxygenases may not play a significant role in the PCB catabolism in the samples investigated.

Comparison of the ISPα clone libraries derived from soil and rhizosphere revealed differences between their catabolic gene structures. They were found to be statistically different, and many sequences were clearly differentially distributed in the two environments. Two ISPα sequences dominated the rhizosphere, LhISP8 and LhISP11, accounting for almost 60% of the gene library. These sequences were most similar to a putative ISPα from Pseudomonas sp. strain IC (identity 94.4%) and an ISPα from Pseudomonas putida (identity 75.5%), respectively. It would not be surprising to find out that these genes belong to Pseudomonas species, as they are known to be very good rhizosphere colonizers (28) and their broad catabolic spectrum is known to include ISPα genes (34). Thus, it would be interesting to identify the bacteria possessing these genes, as well as the substrate spectrum of the putative enzymes, in order to study their applicability for Salix sp.-mediated rhizoremediation.

The different analyses carried out in the present study support the idea that the rhizosphere bacterial community which evolved from the native PCB-polluted soil community during the development of the Salix sp. root system was distinct from its parental community at functional and structural levels. The rhizosphere was enriched in Proteobacteria species and ISPα genes closely related to the gram-negative cluster. It selected and sustained certain ISPα-possessing bacteria (putative PCB degraders), plausibly for their competence in root colonization, which could be used for future rhizoremediation strategies. Furthermore, the discovery of a new cluster containing a high degree of sequence variation opens the road for the detection of novel and useful enzymes selected under existing environmental conditions.