INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The past few decades of research in marine microbial ecology have revealed that prokaryotes are important components of the marine plankton. In addition to accounting for bulk biomass and activity, prokaryotes have central roles in mediating a variety of different biogeochemical cycles (2, 13). Determining the specific prokaryote composition of marine water, however, has been hindered by a lack of techniques for studying microbial community structure in situ. Therefore, little is known about which microbial species are responsible for the biomass and activities measured in the field and about the spatial distribution and temporal dynamics of these species. In the last few years, molecular techniques based on the use of 16S rRNA gene sequences as phylogenetic markers have begun to provide information about the identities of microorganisms in natural and complex systems (1, 49). The marine picoplankton assemblage was one of the first assemblages to be investigated, and the results obtained revealed that most marine prokaryotes were undescribed species that had not been cultivated (5, 8, 16, 32). Uncultured and undescribed microorganisms seem to be present and even dominant in many different environments (1, 36).

Of the different uncultivated organisms detected in marine plankton by molecular techniques, new types of archaea were perhaps the most unexpected. The Archaea, the Bacteria, and the Eucarya are the three lineages of life, and the Archaea is composed of the kingdom Crenarchaeota and the kingdom Euryarchaeota (50). Recently, a third kingdom, the Korarchaeota, has been proposed based on sequences retrieved from hot spring environments (3, 46). Cultured crenarchaeotes are extreme thermophiles and are generally obligate anaerobes with sulfur-dependent metabolism. However, many rRNA genes of apparently nonthermophilic crenarchaeotes have been retrieved from different ecosystems (6). Marine crenarchaeotes (referred to as group I archaea) have been found in different geographic areas and at different depths in the plankton community (8, 9, 15, 17, 26, 47) and also in marine and freshwater sediments (23, 24, 31, 46) and marine animals (28, 41). The sequences of group I archaea form a separate cluster related to clusters of sequences recovered from freshwater sediments (43), forest soils (22), and other terrestrial environments (4, 6, 21). The euryarchaeotes that have been cultured are metabolically more diverse than crenarchaeotes and include extreme halophiles, methanogens, and some sulfur-metabolizing thermophiles. Euryarchaeotal sequences have been retrieved from different marine plankton samples (8, 9, 16, 17, 26, 47) and form a cluster referred to as the group II archaea, which also includes sequences obtained from marine sediments (31) and the digestive tracts of fishes (47). Recently, a new euryarchaeotal group (referred to as group III archaea) has been detected in deep-water marine plankton (17). The sequences that are most similar to the group III archaea are sequences recovered in coastal marine sediments (33, 48).

A few studies dealing with the diversity, abundance, and ecology of marine planktonic archaea have been performed. These studies revealed that marine archaea could be important components of the prokaryotic assemblage and could account for up to 20 to 30% of the total picoplankton rRNA in Antarctic coastal waters (9, 27, 34, 35) and in Pacific coastal waters (26) and up to 40 to 60% of the prokaryote counts in California and Mediterranean waters (18). In some studies marine archaea exhibited temporal dynamics, and the level of these organisms decreased during the austral spring in Antarctic coastal waters (27, 34). They also exhibited spatial differences and generally were more abundant below the photic zone (17, 18, 26, 27). In the Santa Barbara Channel, each group dominated the archaeal assemblage in a different region of the water column (26), whereas the group II archaea were generally scarce in Antarctic waters (10, 27, 35). Archaeal composition has been studied by sequencing some clones in genetic libraries (8, 17, 26). However, in these studies the authors examined only a few marine areas, and the genetic libraries have not been systematically analyzed.

The aim of this study was to determine whether the patterns of archaeal distribution that have been observed are a general feature of the world's oceans or a peculiarity of the few samples analyzed to date. In order to do this, we collected samples from eight regions, including the Atlantic Ocean, the Pacific Ocean, the Southern Ocean, and the Mediterranean Sea, and at two depths (the surface and the aphotic zone). Libraries of archaeal 16S rRNA genes were constructed from these samples, thoroughly analyzed, and compared. Overall, we determined the marine archaeal group affiliations for approximately 100 clones per library by performing membrane hybridization experiments and the putative phylogenetic affiliations of approximately 40 clones per library by performing a restriction fragment length polymorphism (RFLP) analysis, as well as 48 new archaeal sequences. In this study we systematically examined the distribution of group I and group II archaea in different marine regions and at different depths (and the presence of other archaeal groups) and determined whether similar types of marine archaea were found in all systems and whether particular types had depth-specific distributions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Sampling and nucleic acid extraction. Samples were collected from different marine locations (Table 1), including the Atlantic Ocean (North Atlantic and Cantabrian Sea), the Mediterranean Sea (Alboran Sea), the Pacific Ocean (Santa Barbara Channel), and the Southern Ocean (three stations across Drake Passage and a coastal station in the Antarctic Peninsula area). Seawater samples from different depths were collected with Niskin bottles attached to a rosette and a conductivity-temperature-depth sensor. The sample from Arthur Harbor, Antarctica, was collected and processed as described by DeLong et al. (10). Chlorophyll a concentrations were determined by fluorometry (37), and prokaryote abundance was determined by epifluorescence microscopy (40) or by flow cytometry (19). Microbial biomass was collected with 0.2-µm-pore-size Sterivex filter units (Durapore; Millipore) by filtering approximately 20 liters of seawater through a prefilter and the Sterivex filter unit in succession with a peristaltic pump. The prefilters used were 0.8-µm-pore-size polycarbonate filters and 1.0- and 1.6-µm-nominal-pore-size glass fiber filters (Table 1). The Sterivex units were filled with 1.8 ml of lysis buffer (40 mM EDTA, 50 mM Tris-HCl, 0.75 M sucrose) and stored at 20°C. Nucleic acids were extracted by digesting preparations with lysozyme, proteinase K, and sodium dodecyl sulfate, extracting the nucleic acids with phenol-chloroform-isoamyl alcohol, and then desalting and concentrating the nucleic acids with a Centricon-100 concentrator (26). The integrity of the DNA extracted was checked by agarose gel electrophoresis. DNA yields were quantified by a Hoescht dye fluorescence assay (38). Nucleic acid extracts were stored at 70°C until they were analyzed.

Ribosomal DNA (rDNA) clone libraries. Archaeal 16S rRNA genes were amplified by PCR by using different combinations of archaeon-specific primers 20f, 21f, and 958r and universal primer 1392r (8, 28). Each PCR mixture (100 µl) contained 10 ng of natural DNA as a template, 10 to 15 pmol of each primer, 20 nmol of each deoxynucleoside triphosphate, 2.5 U of Taq DNA polymerase (GIBCO BRL), and the PCR buffer supplied with the enzyme. PCR were performed with a Genius (Techne) thermocycler by using the following conditions: an initial denaturation step consisting of 94°C for 3 min, 30 cycles consisting of 94°C for 45 s, 55°C for 45 s, and 72°C for 60 s, and a final elongation step consisting of 72°C for 5 min. The products of two to four independent PCR were combined before cloning in order to reduce the potential bias in separate reactions (39). The PCR fragments were cloned by using a TA cloning kit (Invitrogen) as recommended by the manufacturer. Between 100 and 300 putative positive colonies were transferred to multiwell plates (12 by 8 wells) containing Luria-Bertani medium with 7% glycerol and stored at 70°C.

Membrane hybridization experiments. rDNA clones were transferred to three nylon membranes (Amersham) on Luria-Bertani agar and incubated overnight at 37°C, until visible colonies appeared. The colonies were lysed as previously described (26), and the nucleic acids were immobilized on the membranes by UV cross-linking. Each membrane was then hybridized overnight at 45°C with one of the following three probes (26): a universal probe for archaea (probe Arch-915), a probe specific for marine crenarchaeotal group I (probe GI-554), and a probe specific for marine euryarchaeotal group II (probe GII-554). The high-stringency wash temperatures were 56°C for Arch-915 and 40°C for GI-554 and GII-554. For some libraries (the SB and AM libraries [Table 2]), the probes were labeled with ³²P, the hybridization and washing conditions were the conditions described previously (26), and the signal was detected by exposure to autoradiographic film. For the other libraries, the probes were labeled with fluorescein isothiocyanate, and for hybridization and washing we used the reagents and procedures recommended by Boehringer Mannheim. A colorimetric method was used to amplify and detect the hybridized probes. Membranes were incubated with an anti-fluorescein isothiocyanate antibody conjugated with the enzyme horseradish peroxidase (anti-fluorescein-POD Fab fragments; Boehringer Mannheim) and then with a chromogenic substrate (BM blue POD substrate, precipitating; Boehringer Mannheim), which in the presence of the enzyme precipitated and formed a permanent, dark blue spot.

RFLP analysis and sequencing. Archaeal inserts from selected clones were PCR amplified with primers 21f and 958r. The PCR products (length, approximately 915 bp) were subjected to separate enzymatic digestions overnight at 37°C with the tetrameric restriction enzymes HaeIII and RsaI (GIBCO BRL) and subsequently electrophoresed in a 2.5% low-melting-point agarose gel for 2 to 4 h at 60 to 80 V. A 50-bp DNA ladder (GIBCO BRL) was included in the gel, and the sizes of DNA fragments larger than 25 bp were determined; these fragments were used for analysis. RFLP analyses were performed separately for group I and group II clones (affiliations had been determined previously by membrane hybridization analysis). The RFLP patterns obtained with each enzyme were identified by using three-part designations; the first part indicated the enzyme (H, HaeIII; R, RsaI), the second part indicated the archaeal group (I, group I; II, group II), and the third part indicated the actual pattern (patterns A to T). Clones that produced identical patterns with both enzymes were grouped into discrete operational taxonomic units (OTUs), which were also identified by three-part designations; the first part indicated the affiliation of the clone with one of the two groups (I, group I; II, group II), the second part (a letter) indicated the RFLP pattern obtained with HaeIII, and the third part (a letter) indicated the RFLP pattern obtained with RsaI.

Phylogenetic analysis. Phylogenetic analyses were performed with the software GDE 2.2 and Treetool 2.0.1 obtained from the Ribosomal Database Project (25) and the software package Phylip, version 3.5 (11). Sequences were manually aligned by using a GDE file. Distance matrices were calculated with DNAdist (Phylip) by using the Kimura two-parameter model and by assuming that the transition/transversion ratio was 2.0. Trees were inferred by performing a neighbor-joining analysis (Phylip) and were edited with Treetool. A maximum-likelihood analysis was performed with DNAml (Phylip). A bootstrap neighbor-joining analysis was performed with 100 replicates with random taxon addition. A bootstrap maximum-likelihood analysis was performed with 50 replicates by using only 27 significant sequences.

Nucleotide sequence accession numbers. Sequences were deposited in the GenBank database under the following accession numbers: AF223111 for clone AT-5 (1); AF223112 for AT-200 (1); AF223113 for AT-200 (7); AF223114 for CA-15 (P18); AF223115 for ME-450 (5); AF223116 for ME-450 (9); AF223117 for ME-450 (20); AF223118 for ME-450 (P3); AF223119 for ME-450 (P5); AF223120 for SB95-1; AF223121 for SB95-20; AF223122 for DN-5 (1); AF223123 for DN-200 (1); AF223124 for DS-5 (1); AF223125 for DS-5 (P21); AF223126 for AM-20A (101); AF223127 for AM-20A (102); AF223128 for AM-20A (103); AF223129 for AM-20A (104); AF223130 for AM-20A (117); AF223131 for AT-5 (21); AF223132 for AT-5 (P24); AF223133 for AT-200 (29); AF223134 for AT-200 (P25); AF223135 for CA-15 (22); AF223136 for CA-15 (23); AF223137 for CA-15 (27); AF223138 for CA-15 (32); AF223139 for CA-15 (P4); AF223140 for ME-450 (21); AF223141 for ME-450 (30); AF223142 for ME-450 (38); AF223143 for ME-450 (P14); AF223144 for SB95-35; AF223145 for SB95-48; AF223146 for SB95-87; AF223147 for DF-5 (21); AF223148 for AM-20A (122); AF223149 for AM-20A (123); and AF223150 for ME-450 (P9).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sampling sites and genetic libraries generated. We collected samples from eight different marine areas (Table 1), including the Atlantic Ocean, the Mediterranean Sea, the Pacific Ocean, and the Southern Ocean. Stations were located on the continental shelf and the continental slope and offshore. In general, we analyzed two depths, the surface and the aphotic zone. Some physical and biological parameters of the samples are shown in Table 2. The samples collected had wide ranges of temperatures (1.8 to 18°C), salinities (33.37 to 38.50), surface chlorophyll a concentrations (0.10 to 0.98 µg liter¹), and prokaryote concentrations (1.16 × 10⁵ to 6.35 × 10⁵ cells ml¹). The planktonic microbial biomass in the samples was collected on filters, and the total nucleic acids were extracted. We determined (data not shown) that most free-living prokaryotes were present in the size fraction analyzed. The DNA yields ranged from 0.09 to 1.30 µg of DNA liter of seawater¹ (Table 2) and were several times greater in the surface water than in the deeper water at all stations. Nucleic acid extracts were used for PCR amplification of partial 16S rRNA genes in which archaeon-specific primers were used. Amplification was obtained for all of the samples tested, which confirmed that marine archaea were ubiquitous in the plankton. A genetic library of archaeal genes was generated for each sample (Table 2). Of the 16 libraries which we analyzed, 11 were new in this study, whereas the 5 other libraries have been described previously (SB libraries were described by Massana et al. [26]; AM libraries were described by DeLong et al. [10]). These libraries were analyzed and compared by using a hierarchical approach that included membrane hybridization, RFLP analysis, and sequence analysis.

Analysis of archaeal libraries. In the membrane hybridization experiments, most of the archaeal clones hybridized with either group I or group II probes (Table 3), indicating that these groups accounted for the bulk of the marine archaea in our samples. The nine archaeal clones that did not hybridize with these probes were sequenced. Four of these clones were affiliated with group I, and four were affiliated with group II (Table 3) but exhibited between one and three mismatches in the target regions of the probes. The sequence of the remaining clone, ME-450 (P9), was very similar to two sequences belonging to group III marine archaea (17). The fact that only 8 of 2,327 group I or group II clones exhibited mismatches with their respective probes indicates that these probes are well suited for studying marine archaea and detecting new groups in plankton.

View larger version (59K): [in this window] [in a new window]   FIG. 1.   Distribution of group I and group II OTUs among the libraries analyzed. The figure shows the number of clones belonging to each OTU (in italics for deep libraries), the number of clones analyzed, and the coverage values for each library. The last row calculates the same parameters but considers all the clones in each group as belonging to the same assemblage.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Phylogenetic tree for marine archaea inferred from DNAdist and a neighbor-joining analysis by using 631 bp (E. coli positions 45 to 737). One clone of each OTU is shown on the tree. OTUs that appear only in GenBank clones are shaded. The affiliations of the clones of Fuhrman et al. are indicated on the right (the four sequences most similar to OTU I-CD are pB1-47, pB1-80, pB1-124, and pB1-151; the five sequences most similar to OTU I-ED are pB1-22, NH49-8, nH49-14, p712-12, and p712-24; the eight sequences most similar to OTU I-GD are NH25-1, NH25-13, NH49-4, NH49-9, pB1-53, pB1-123, pN1-56, and p712-37; and the four sequences most similar to OTU I-KD are pB1-92, pN1-10, pN1-27, and pN1-43). Bootstrap values (percentages) for the neighbor-joining and maximum-likelihood analyses are indicated for the most significant nodes. Scale bar = 0.10 change per sequence position.

Validation of the RFLP analysis of archaeal libraries. In this study we used a RFLP approach to define archaeal OTUs and to compare genetic libraries. In order to determine the phylogenetic consistency and breadth of the OTUs defined, different clones belonging to the same OTU were sequenced and added to a tree containing all of the clones in our libraries (Fig. 3). For the most abundant OTU, OTU I-AA, we sequenced 11 clones obtained from different places and depths (Fig. 3). All of the sequences obtained were very similar (97.7% similarity), and they were distributed throughout cluster I-, which also included six other OTUs. Four OTU II-CC clones obtained from different places were sequenced, and they grouped together closely (98.1% similarity) (Fig. 3), like the other three examples examined. We estimated that clones belonging to the same OTU had an average level of similarity of 97.7%. This is a very high value, considering that a similarity value greater than 97% for the whole 16S rRNA gene is used to indicate that organisms belong to the same species. Our findings indicate that the RFLP analysis placed clones in the phylogenetic tree very consistently.

View larger version (31K): [in this window] [in a new window]   FIG. 3.   Phylogenetic analysis of all of the marine archaeal sequences in our libraries. The tree was constructed as described in the legend to Fig. 2. Clones belonging to the same OTU are underlined, and the average similarity values are indicated.

View larger version (82K): [in this window] [in a new window]   FIG. 4.   Map of restriction sites for the enzymes HaeIII and RsaI for the RFLP patterns detected in clones from our libraries and retrieved from the GenBank database (patterns exclusive of GenBank clones are shaded). For convenience, the positions of restriction sites are referred to the sequence of clone SB95-57. The DNA fragments that formed each RFLP pattern can be determined by subtracting two consecutive restriction sites (for example, pattern H-I-A was formed by the 215-, 28-, 72-, 187-, 280-, and 132-bp fragments).

View larger version (24K): [in this window] [in a new window]   FIG. 5.   Dendrogram generated by using the restriction sites of each OTU found in our libraries. A distance matrix was calculated from a binary matrix based on the presence of restriction sites, and the dendrogram was inferred by using the unweighted pair group method with arithmetic averages.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Significance and limitations of the methodological approach used. Marine archaea obtained from different regions of the world's oceans were compared by analyzing genetic libraries of 16S rRNA genes. This approach has been widely used to describe the microbial diversity in different habitats (4, 5, 31) and allows workers to study uncultivated prokaryotes, such as the marine archaea (8, 15). However, it is known that clonal representation in genetic libraries can be biased, particularly due to the PCR step (39, 45), and may not reflect the relative levels of organisms in a sample (but see reference 12). The biases can affect the results at two levels; some sequences may be present but not be amplified, and particular types may be over- or underamplified. Using different primer sets (universal primers [17] and different combinations of archaeal primers [Table 3]) resulted in retrieval of similar archaeal phylotypes. This suggests that we amplified all of the archaeal phylotypes present in the samples, since it is unlikely that the different primer sets were biased against the same phylotypes. We studied the quantitative aspect of the libraries by monitoring the relative levels of group I and group II clones. In one case, parallel libraries (SB-200 and SB-200B) were examined with two primer sets, and similar results were obtained (Table 3). The genetic libraries obtained with universal primers revealed that group II clones (6 of 33 archaeal clones) were less prevalent in libraries obtained from Atlantic and Pacific deep-water samples (17). Moreover, the relative levels of group I and group II phylotypes in some samples were determined independently by performing quantitative rRNA hybridization experiments. In one case, both approaches were used for the same samples (26), and there was very good agreement in the relative levels of both groups. Other experiments revealed that the group I signal was dominant throughout the water column in Antarctic waters (27, 35). Therefore, our results based on comparisons of the compositions and levels of particular clone types in the libraries could be somewhat biased, but several pieces of evidence indicate that our main conclusions remain valid.

Distribution of archaea in the water column. The pattern found in California coastal waters (group II phylotypes are the dominant phylotypes at the surface and group I phylotypes are the dominant phylotypes at depth [26]) was also found in the Mediterranean Sea and in the North Atlantic Ocean (Table 3). In contrast, in Southern Ocean samples group I phylotypes were the dominant phylotypes at both depths, and group II phylotypes were almost absent (Table 3). We concluded that group I phylotypes are ubiquitous and are abundant and often the dominant phylotypes in most marine waters, whereas group II phylotypes are the dominant phylotypes only at the surface in temperate regions. The two groups are very distantly related phylogenetically and seem to occupy different ecological niches. At the surface, where group II clones are the predominant clones based on archaeal sequences, the relative level of archaea is lower (18, 26, 27, 35), but the specific activity of the prokaryotic assemblage is higher and there is an active food web based on algal photosynthesis. Recently, it has been suggested that instead of being planktonic, group II archaea could originate from the digestive tracts of very common fishes (47). In deeper, dark waters where group I organisms predominate, the relative archaeal level is higher (18, 26, 35), but the activity of the prokaryotic assemblage is lower (27) and largely dependent on sedimenting material. We still do not know what the role of archaea in the marine plankton is, but defining the distribution of the different groups should help elucidate some aspects of the ecology of these organisms.

Ubiquity and limited diversity of marine planktonic archaea. At the level of resolution of the RFLP analysis of rRNA genes, planktonic archaeal diversity appeared to be remarkably limited. In general, only one or two OTUs dominated each library, and some other OTUs were represented by very few clones (Fig. 1). When the data were investigated more carefully, the less abundant clones were sometimes found to belong to the same phylogenetic unit. This was the case for all OTUs belonging to cluster I-, which were not significantly different from the most abundant OTU, OTU I-AA (Fig. 3). PCR biases probably do not explain the limited diversity, since according to the kinetic model (45) PCR biases tend to favor overrepresentation of rare types, thus increasing the diversity sampled. The relatively low archaeal diversity in worldwide marine plankton assemblages contrasts with the higher archaeal diversity found in other environments, such as marine hydrothermal vent sediments (31), salt marsh sediments (33), soils (4), and hot springs (3). Also, genetic libraries generated with the other microbial components of the marine plankton revealed much greater diversity of both bacteria (5, 16, 32) and eukarya (42; B. Díez, unpublished results).