INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

The subject of prokaryotic biodiversity in the sea has received new and substantial attention with the development of molecular techniques to describe and identify the microbial components of natural communities. PCR amplification and the cloning of diagnostic molecules, mostly the 16S rRNA genes, permits extensive studies of the microbial diversity of ecosystems without the bias imposed by pure-culture techniques (or at least with a different one). In any case, molecular techniques represent a new approach to the extremely complex problem of describing microbial diversity. With the proliferation of studies based on cloning and sequencing ribosomal DNA (rDNA) retrieved from ocean samples, it has been recognized that bacterioplankton is dominated by a relatively limited subset of broad phylogenetic groups that are widely distributed and often exhibit clear trends in their vertical distribution in the water column (17, 21, 22, 34, 38).

We previously applied (1) a community-fingerprinting (36, 43) analysis to the water column in the stratified section of a Western Mediterranean station (halfway between Barcelona and the island of Mallorca). Offshore marine waters in tropical and subtropical latitudes often have a typical vertical temperature structure; in the case of temperate waters, this is mostly so during the summer. During this season, temperature and density vary over the first 100 m of depth. Under these conditions there is a characteristic chlorophyll-depth profile with a maximum, frequently sharp, known as the deep chlorophyll maximum (DCM). The DCM corresponds to a layer of maximum primary productivity and phytoplankton concentration. It is located between 40 and 100 m below the surface, where environmental conditions, particularly nutrient concentrations, are apparently optimal for many photosynthetic microorganisms (11, 14, 15, 24, 27). Our previous study by amplified rDNA restriction analysis (ARDRA) community fingerprinting showed that the prokaryotic assemblages at the surface, the DCM, and the deep water mass (400 m) varied (1). Within the bacterial community the main difference found was between the cells that lived in association with large particles (particles over ca. 8 µm, retained by a glass fiber prefilter) and the free-living cells (which passed through the filter).

Attached bacteria are often larger, and are present in higher local concentrations, than those found free living in water (10) (Fig. 1). Although they are relatively few in the open ocean (compared to free-living cells), they could have an important role in carbon cycling (9, 25). There is information in the literature about the community present in relatively large aggregates, such as marine snow (4, 13), but smaller, more widespread aggregates of microscopic size have not been investigated.

View larger version (69K): [in this window] [in a new window]   FIG. 1.   Scanning electron microscopy of attached bacterial communities (A and B) and free-living bacterial communities (C).

Here we have studied in further depth the samples that seemed most promising from our previous ARDRA community fingerprinting (1). We have analyzed both free-living and attached bacterial communities found at three different depths (5 m, DCM, and 400 m) by sequencing 16S rDNA clones. Scanning electron microscopy has been used to examine the morphology of cells collected on each filter. The attached subsample gave very little diversity by the random cloning and sequencing approach used. In an attempt to retrieve more phylotypes from this sample, we have also used a methodology in which the predominant sequences amplified in one environment are characterized by ribosomal internal spacer analysis (RISA) (8).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Sampling. Three samples from different depths were selected from station D (about halfway between Barcelona and Mallorca [1], over a bottom of 2,000 m) in the Western Mediterranean. The work was done during the cruise FRONTS95 of the B/O García del Cid from 16 to 25 June 1995 (1). We analyzed free-living and attached assemblages found at three different depths: the surface (5 m); DCM, located at 52 m in this sample; and 400 m. Water samples were collected with a 30-liter double Van Dorn bottle and dispensed into plastic carboys.

DNA extraction and purification. DNA extraction and purification followed the protocol of Fuhrman et al. (18), with slight modifications as previously described (1).

PCR amplification of 16S rDNA and RISA. 16S rRNA genes were amplified from total DNA by PCR with two bacterial primers: ANT-1 and S (Table 1). The ribosomal internal transcribed spacers plus a stretch of the 16S rDNA (ca. 500 nucleotides) for RISA were amplified with primers B1055 and 23SOR (Table 1). All primers were subjected to CHECK PROBE SSU Prok (Ribosomal Database Project) (31) to confirm their adequacy. PCRs were performed with a Perkin-Elmer 480 thermal cycler. Reaction mixtures contained 50 mM KCl, 10 mM Tris-HCl (pH 9), 1.5 mM MgCl₂, 0.1% Triton X-100, 200 mM of each deoxyribonucleotide triphosphate (dATP, dCTP, dGTP, and dTTP) (Pharmacia Biotechnology LKB, 2 U of TaqI DNA polymerase (Promega Corporation, Madison, Wis.), 0.2 mM (each) oligonucleotide primer, and 100 ng of template DNA in a total volume of 50 µl. The reaction mixtures were overlaid with mineral oil (Light White Oil; Sigma). The following conditions were used for amplification: a cycle of 95°C for 5 min; 35 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min; plus an extension step of 10 min at 72°C. Five microliters of PCR product was analyzed in 1% agarose gels (SeaKem; FMC Bioproducts) in 1× Tris-acetic acid-EDTA (TAE) buffer, stained with 0.5 mg of ethidium bromide/µl, and visualized with UV. Products generated from RISAs were separated in 2% Methaphor agarose gels (FMC Bioproducts) in 1× Tris-borate-EDTA (TBE) buffer.

Purification of RISA products. Major bands (Fig. 2) were cut out of the gel and purified with a Sephaglas BandPrep kit (Pharmacia Biotech), following the manufacturer's instructions. The DNA was recovered in 20 µl of Tris-EDTA (TE) buffer.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Methaphor 2% gel fragments from RISA analyses of free-living and attached (ATT) bacterioplankton communities from surface (SUR), DCM, and 400 m (400). The arrows show the bands that were cloned and sequenced.

Clone library construction. Clone libraries from PCR products were constructed with the TA cloning kit (Invitrogen Corporation, San Diego, Calif.), following the manufacturers' recommendations.

Sequencing. The nucleotide sequences of plasmid inserts were determined by using the ABI PRISM dye terminator cycle-sequencing ready-reaction kit (Perkin-Elmer) and an ABI PRISM 377 sequencer (Perkin-Elmer), according to the manufacturers' instructions. The 16S rRNA genes of 17 clones related to Alteromonas macleodii and chosen to represent the three depths sampled were completely sequenced with M13 Forward (21) and M13 Reverse primers from the TA cloning kit and the internal primers Macle R and Macle F shown in Table 1. The 16S rRNA genes of the other 103 clones were partially sequenced (approximately 350 nucleotides from each end of the gene) with the standard M13 Forward (21) and M13 Reverse primers.

Phylogenetic analysis. Sequences were evaluated by the program CHECK CHIMERA, provided by the Ribosomal Database Project, to check chimerical gene artifacts.

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Phylogenetic tree based on 1,492 nucleotide positions showing relationships of the surface (SUR), DCM, and 400-m (400) clones related to A. macleodii and representative bacterial 16S rRNA genes within the  subdivision of Proteobacteria. An unrooted phylogenetic tree was obtained by performing a neighbor-joining analysis. Bootstrap values over 50% are shown below the segments.

Scanning electron microscopy. Aliquots of 96 ml from each sample were fixed in 1% glutaraldehyde at 4°C overnight. The sample was filtered through a 3-µm-pore-size Millipore filter to recover the attached assemblage. The free-living bacteria that had passed through the 3-µm-pore-size filter were recovered on a 0.22-µm-pore-size filter. The filters were serially dehydrated in 25, 50, 70, and 100% ethanol solutions (three times for 10 min in each stage), critical-point dried, mounted on scanning electron micrograph stubs, sputter coated with gold, and viewed on a JEOL JSM 840 scanning electron microscope.
  Nucleotide sequence accession numbers. GenBank nucleotide sequence accession numbers for completely sequenced clones are from AF114495 to AF114509. Accession numbers for partial sequences of clones recovered at 400 m in the attached fraction are AF114510 to AF114533 and of those in the free-living fraction are AF114534 to AF114577 and AF114654 to AF114657. At the DCM, accession numbers for the attached fraction are AF114578 to AF114598 and AF114643 to AF114644 and those for the free-living fraction are AF114658 to AF114695. At the surface, accession numbers for the attached fraction are AF114599 to AF114622 and AF114623 to AF114642 and those for the free-living fraction are AF114645 to AF114653.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

We sequenced ca. 40 clones from each depth (5 m, DCM, and 400 m). About 20 corresponded to bacterial 16S rDNAs recovered from the particulate fraction, and the other 20 corresponded to those from the free-living population. Each clone contained a nearly complete 16S rRNA gene (8 to 1510 [Escherichia coli numbering]) and was sequenced from both ends, producing an average of ca. 350 nucleotides from each end. Both segments contain hypervariable regions that are often included in environmental studies. The final similarity value was obtained by aligning both ends (ca. 700 nucleotides) to the complete database sequence. The results are shown in Tables 2 to 5, described below.

Sequences recovered at 400 m. At 400 m the assemblages recovered in the attached and free-living fractions were relatively similar. From the attached assemblage we have sequenced 18 clones for an average length of 800 nucleotides (minimum, 578 nucleotides; maximum, 937 nucleotides [Table 2]). This is the sample showing the least sequence diversity, with an average pairwise nucleotide identity of 94.2%. Of the 18 clones, 17 showed a high similarity to the 16S rDNA of A. macleodii; 7 of the clones showed a similarity of 97% or higher, so they very probably belong to a very close taxon, and the lowest similarity found among this cluster was 93.8% (still closely related). The one remaining clone had the best match with Pseudoalteromonas sp. strain SW29 and was therefore also related to Alteromonas (19).

DCM (52 m). The attached assemblage at the DCM is similar to the 400-m subsample. Sixteen of 18 clones had the best match to A. macleodii (Table 3), most with very high similarities, indicating close relationship; 7 of these had similarities over 97%. The two remaining clones, which could be ascribed to the  Proteobacteria, are related to SAR 407, a SAR11 A2 cluster (17) representative.

Surface (5 m). Nineteen clones were sequenced from the attached subsample (Table 4). The results are very similar to those of the rest of the attached samples, with nine clones highly related to A. macleodii. Seven clones had significant similarities to an unidentified marine isolate, E401, from Tanabe Bay, Japan, which is apparently related to the Aeromonas genus of aquatic  Proteobacteria. The free-living subsample (17 clones) was dominated by bacteria related to uncultivated putative organisms of the  Proteobacteria (Table 4). Four clones were similar to the SAR11 A-2 cluster known to be more abundant at the surface (17). Remarkably, one clone had a very high similarity (97.9% over 705 nucleotides) to a sequence retrieved from the Oregon coast and belonging to the SAR116 cluster, a widespread phylotype (34). Among the  Proteobacteria three clones had high similarities to the uncultured Oregon coast sequence OCS44 that belongs to the SAR86 cluster. One clone had 95.6% similarity to uncultured clone SAR7 (cyanobacteria), and another clone (over 83%) belonged to the Cytophagales group.

A. macleodii cluster. The 16S rRNA genes of 17 clones belonging to the A. macleodii cluster were completely sequenced. The relationships among the 17 sequences are shown in Fig. 3. Two clearly defined clusters were found. All of the sequenced clones appeared with A. macleodii in a cluster separated from other genera of the  subdivision of Proteobacteria, such as Pseudoalteromonas and Shewanella. The sequences retrieved from the surface were closely related to A. macleodii IAM 12920^T (within-cluster similarity, 98.4 to 99.1% [Table 6]), whereas DCM and 400-m clones formed a different subcluster (with similarities from 95.8 to 98.5% [Table 6]).

View this table: [in this window] [in a new window]   TABLE 6.   Similarity matrix (based on 1,492 nucleotide positions) among the clones related to A. macleodii and 16S rRNA gene sequences from representatives within the  subdivision of Proteobacteria

Morphology. To assess the morphological types present in the attached assemblage versus those in the free-living assemblage, we examined by scanning electron microscopy the glass fiber filtrate retrieved on a 0.22-µm-pore-size filter (free living) and a small aliquot of the raw sample collected on a 3-µm-pore-size absolute filter (attached) from the surface sample. The differences were quite obvious (Fig. 1). The free-living assemblage was composed of small cells (with diameters well below 1 µm) with very different morphologies (Fig. 1C). On the other hand, the attached assemblage consisted of much larger cells, with diameters around 1 µm and with much less morphological diversity (Fig. 1A and B). Coccobacillary forms (as A. macleodii appears to be in culture) were abundant in the attached sample, although some elongated rods, spirals, and other shapes were observed as well. Aggregates and clustered cells, often bound to detrital material, were also abundant, as expected.

RISA. The RISA technique permits separating different types of 16S rDNAs in a mixed-community DNA sample by simply using a primer located at the 5' end of the 23S rRNA gene so that the spacer between the two genes is amplified together with a section of the 16S gene (8). We have amplified a region spanning from position 1055 (E. coli numbering) in the 16S rDNA to the beginning of the 23S rDNA (position 38), so the expected size range was 600 to 1,600 bp (the 16S rDNA fragment plus the spacer region). The PCR products corresponding to different organisms can be separated by size in an agarose gel, and then different bands can be excised from the gel and the 16S region can be sequenced to identify the organism. Figure 2 shows a Methaphor agarose gel in which the PCR products from the different depths and assemblages are shown. Used in this way, the technique is not very informative as a community-fingerprinting methodology due to the relatively small number of discernible bands. However, in the 400-m attached sample, in which we assume the diversity to be very low, two major PCR products of ca. 890 and 970 bp appeared; we will refer to them as RISA1 and RISA2, respectively. These bands were cloned, and five clones from each one were analyzed. The clones were partially sequenced with the B1055 internal primer (Table 1). The five RISA2 clones were found to belong to the A. macleodii cluster, with similarities between 86.6 and 96.5%. The other five clones sequenced from the band of ca. 890 bp (RISA1 clones) were principally related to the Pseudoalteromonas group of  Proteobacteria. Two clones were similar to P. antarctica, with 89.9 and 98% similarities over 434 and 414 nucleotides, respectively. One clone had a 95.8% similarity to Pseudoalteromonas espejiana over 263 nucleotides. Finally, the last two clones were more distantly related, one giving a 92.9% identity in 282 nucleotides to an unidentified Cytophaga isolate (S23328 environmental sample) and the other giving a 93% identity in 422 nucleotides to Methylobacter sp. strain BB5. The sequence diversity retrieved with RISA was thus higher than that obtained from the sequencing of 18 random clones.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Community structure depth variation. The sequencing results confirm the initial data obtained by ARDRA (1) showing a marked community structure variation with depth in the superficial stratified waters of the Western Mediterranean. There is also abundant information in the literature supporting the thesis that prokaryotic diversity in the open ocean varies with depth (17, 28, 44). Our results support this view, specifically for the first 100 m. The predominance of clones belonging to the  subclass of Proteobacteria at the surface has been described by several authors and seems to be a widespread occurrence (17, 38). As was found previously (17), the abundance of SAR11 A-1 and A-2 clusters decreases sharply from 0 to 40 m, although SAR11 G1 increases slightly with depth. Our clones also belong to a relatively restricted range of phylotypes, perhaps reflecting a clearly predominant ecotype adapted to live in the relatively warm waters of the upper layer of the ocean. However, just about 50 m below, at the DCM, the community structure changes significantly and a much larger representation of  Proteobacteria was found. That could be an effect of the peculiar conditions of the DCM, with a much higher abundance of phytoplankton and perhaps a higher availability of organic nutrients, or it could simply reflect the change in physical conditions, mostly water temperature and light intensity. At 400 m the change is even more dramatic.

Attached versus free living. Our results, as well as some previous reports (1-3, 7, 13, 29, 30), indicated that the bacterial community in aquatic environments is, in terms of species composition, markedly different for cells associated with particles and those that are free living. The attached community shows amazingly little diversity, with most clones belonging to the  Proteobacteria and highly similar to the cultivated marine bacterium A. macleodii IAM 12920^T, a strain isolated in the 1970s from coastal waters near Oahu, Hawaii (6). The pelagic assemblage is dominated by a more heterogeneous population that varies with depth. Here the best matches correspond to uncultivated entries only known by sequence, as shown in many previous studies. Amorphous aggregates that appear in natural aquatic environments can have various origins, e.g., bacteria attached to zooplankton fecal pellets, bacteria attached to each other by polymers, or bacteria attached to animal debris, such as the cast houses of mucous netfeeders (33). Aggregates may form a microhabitat providing protection from some bacteriovores (12) as well as nutrient abundance when advective flow through porous aggregates occurs (33).

Cultured versus uncultured. Our results can shed some additional light on the classical discrepancy between culture- and PCR-based methods to describe biodiversity. On one hand, there is evidence indicating that most bacterial cells living in the ocean are not cultivated on standard marine media and belong to taxa distantly related to those well known from pure-culture studies (41). On the other hand, there are a number of studies in which it is shown that cultured strains of marine bacteria can represent significant fractions of the bacterial biomass in sea water (37, 39). The existence of two largely different assemblages sharing the habitat would be an important factor to consider in this kind of comparison. If a seawater sample is directly plated on an agar medium, the assemblage of large cells, belonging mostly to easily cultivated  Proteobacteria attached to particles, will rapidly grow and override any other microbial group in the "culturable" harvest. The other assemblage, composed of smaller cells that live swimming in the medium, would be very difficult to retrieve if they had to compete. Both groups are probably of similar relevance in terms of biomass (or rRNA present in the sample) since, although in terms of cell numbers the free-living bacteria could be orders of magnitude more abundant, the attached fraction contains much larger cells. Therefore, depending on the technique used to collect biomass and/or detect the presence of one group or another, very different conclusions could be reached. For example, if total biomass is collected and hybridized to a probe for the attached (cultured) representatives (39), a good proportion of the hybridization signal could be accounted for by this fraction, although numerically they represent a very minor component of the community.